Dillon 2004
Dillon 2004
Dillon 2004
http://www.cambridge.org
Preface page xi
1 Introduction 1
2 Bivalve autecology 8
Feeding and digestion 9
Digestive anatomy 9
Particle retention 11
Particle ingestion 14
Assimilation 16
Non-particulate dietary components 19
Feeding period 20
Habitat 22
Unionoids 22
Corbiculoids 28
Dreissena 34
Reproduction 35
Unionoid gonochorism 35
Unionoid hermaphroditism 40
The unionoid larval stage 42
Dreissena 51
Corbiculoids 52
Summary 55
3 Gastropod autecology 57
Feeding and digestion 58
Pulmonate diet and habitat 61
Planorbidae 61
Physidae 66
Lymnaeidae 70
Acroloxidae and Ancylidae 76
viii · Contents
Pulmonate reproduction 79
Outcrossing 79
Selfing 83
Prosobranch diet and habitat 85
Neritidae 85
Pleuroceridae 86
Pomatiopsidae 91
Hydrobiidae 94
Ampullariidae 97
Viviparidae 99
Bithyniidae 100
Valvatidae 101
Prosobranch reproduction 101
Gonochorism 101
Parthenogenesis 109
Hermaphroditism 113
Summary 114
6 Parasitism 227
The digenetic trematodes 228
Prevalence 230
Consequences 239
Schistosomes 240
Fasciola 245
Echinostomes 247
Resistance 248
Fasciola 249
Schistosomes 252
Echinostomes 259
Population regulation 262
Aspidogastrids 268
Unionicolids 269
Summary 270
7 Predation 273
The malefactors 274
Mammals 274
Birds 277
Turtles 279
Salamanders 281
Fish 282
Crustaceans 293
Insects 295
Molluscs 300
Leeches 301
x · Contents
Triclads 304
Defences 305
Behaviour 305
Shell 309
Life history 314
Consequences 316
Community composition 316
Distribution 318
Summary 322
8 Biogeography 326
Environmental calcium 327
Laboratory studies 328
Field studies 332
Area 338
Isolation 349
Other environmental factors 352
Community composition 354
Summary 364
9 Communities 367
Terms and conditions 368
Interspecific ecological overlap 371
Gut content 371
Feeding strategy 377
Depth, temperature, and oxygen 378
Substrate 386
Current 390
Models of species distribution 391
Models of species similarity 407
Models of species abundance 421
The assembly of communities 428
Summary 431
Index 499
Preface
Class Bivalvia
Subclass Paleoheterodonta Subclass Heterodonta
Order Unionoidea Order Veneroidea
Superfamily Unionacea Superfamily Corbiculoidea
Family Margaritiferidae Family Corbiculidae
Margaritifera Corbicula
Cumberlandia Family Pisidiidae
Family Unionidae Pisidium
Subfamily Anodontinae Eupera
Anodonta Sphaerium
Anodontoides Musculium
Cristaria Superfamily Dreissenoidea
Subfamily Ambleminae Family Dreissenidae
Amblema Dreissena
Actinonaias Mytilopsis
Elliptio
Fusconaia
Lampsilis
Unio
Villosa
Family Hyriidae
Diplodon
Hyridella
Superfamily Etheriacea
Family Etheriidae
Family Mutelidae
Family Mycetopodidae
They doubtless represent yet another separate invasion from the sea (their
oldest fossils are Eocene), retaining the overall aspect of edible marine
mussels. By virtue of their ability to spin strong byssal attachment threads,
they have occupied the epifaunal habitat not exploited by unionoids or
corbiculoids. They have also retained a planktonic larval stage in their
development.
Among the bivalves, adaptation to fresh waters does not seem to have
been rare nor restricted to the unionoids, corbiculoids, and dreissenoids.
Freshwater or brackish/freshwater species are to be found among the
Arcidae, Mytilidae, Trapeziidae, Donacidae and Cardiidae, to name but
a few families. Although quite interesting from many points of view, the
Introduction · 3
Class Gastropoda
Subclass Pulmonata Subclass Prosobranchia
Order Basommatophora Order Archaeogastropoda
Family Acroloxidae Superfamily Neritoidea
Acroloxus Family Neritidae
Neritina
Family Lymnaeidae Theodoxus
Lymnaea Order Mesogastropoda
Pseudosuccinea Superfamily Viviparoidea
Galba Family Viviparidae
Myxas Bellamya
Radix Cipangopaludina
Stagnicola Viviparus
Campeloma
Tulotoma
Family Physidae Family Ampullariidae
Physa Marisa
Aplexa Pila
Family Planorbidae Pomacea
Planorbis Superfamily Valvatoidea
Anisus Family Valvatidae
Gyraulus Valvata
Armiger Superfamily Rissoidea
Segmentina Family Hydrobiidae
Biomphalaria Amnicola
Helisoma Hydrobia
Menetus Potamopyrgus
Planorbula Family Bithyniidae
Promenetus Bithynia
Bulinus Hydrobioides
Family Micromelanidae
Pyrgula
Family Pomatiopsidae
Oncomelania
Family Ancylidae Pomatiopsis
Ancylus Tricula
Rhodacmaea Superfamily Cerithioidea
Ferrissia Family Thiaridae
Hebetancylus Melanoides
Laevapex Thiara
Pachymelania
Family Melanopsidae
Melanopsis
4 · Introduction
Family Pleuroceridae
Amphimelania
Goniobasis
Pachychilus
Paludomus
Brotia
Juga
Pleurocera
Semisulcospira
In this chapter we will review a few of the basic attributes of the biology
of freshwater bivalves. Although filter feeding might at first seem a rela-
tively simple process, closer examination shows a wide discrepancy
between the particles in the medium (often largely inorganic) and the
food actually assimilated (diatoms, green and blue-green algal cells, bac-
teria, and organics both dissolved and suspended). Our discussion of
bivalve feeding will be divided into sections on particle retention, inges-
tion, and assimilation. There is some large-scale diet and habitat special-
ization in bivalves; Pisidium seems to have become adapted to filter waters
from within the sediments, sometimes deep in the profundal zone, and
Dreissena has colonized the hard bottoms. But in general, we will see that
all bivalve populations seem to live in about the same habitat and eat
about the same food at the same time. In light of the evidence that large
populations of bivalves may substantially depress the concentration of
suspended particles in even the largest lakes and rivers, the potential for
food limitation and both intra- and interspecific competition must be
acknowledged.
The freshwater bivalves are quite diverse in their modes of repro-
duction. We will see that unionoids are gonochoristic (although their
mechanisms of sex determination are unclear) with widespread her-
maphroditism. Their adaptation to hold developing larvae (‘glochidia’)
and impose them parasitically upon fish hosts constitutes one of the more
interesting natural history sagas of which I am aware. The corbiculoids
are generally hermaphroditic and often self-fertilize, although dioecious
populations are common in some groups, and brooding or delayed
release of juveniles is again the rule. Thus in the freshwater bivalves we
will find examples of both great biological unity and striking biological
diversity.
Feeding and digestion · 9
Digestive anatomy
Although morphological details are quite various in the diverse taxa of
freshwater bivalves, in broad outline the digestive anatomy in typical
unionoids, Corbicula and Dreissena can be discussed together (Morton
1983). The posterior margin on the mantle of these taxa is modified into
an incurrent siphon fringed by sensory tentacles and a smaller, more
dorsal excurrent siphon nearby. The term ‘siphon’may be somewhat mis-
leading, for these structures are very short and sometimes indistinct, more
like holes than tubes. In fact, the posterior mantle of margaritiferid
mussels does not form distinct siphons at all, although the path of water
flow is the same as in other unionoids. Specialized batteries of cilia on
the gills draw water slowly into the mantle cavity, along with any
suspended particles not so large as to irritate the tentacles or other sensory
devices, moving it anteriorly. Most bivalves have a second, longer set of
cilia or cirri on the gills, the ‘laterofrontals’, acting either as mechanical
particle filters or as modulators of fluid mechanical processes capturing
particles (Jorgensen 1983). The movement of particles after their entrap-
ment on the gills has been nicely illustrated by Huca et al. (1982), for the
South American hyriid Diplodon, and by Avelar (1993) for the myceto-
podid Fossula. Filtered water is expelled back posteriorly, often with
increased force, presumably to minimize the likelihood of refiltration.
The pattern of water flow described above is the most common in
bivalves generally. The corbiculoids have a pair of short, fairly elaborate,
well-differentiated posterior siphons. The siphons of Sphaerium are fused
into a proper tube and may extend 50–100% of the animal’s shell length.
But interestingly, Sphaerium and Musculium are apparently also able to take
in water through the mantle margin some distance from the siphons.
The unusual manner by which Pisidium feeds has attracted a good deal
of attention (Mitropolskij 1966a,b, Meier-Brook 1969, Holopainen
1985). Unlike most other bivalves, these little clams often do not main-
tain direct contact with the water column. They bury beneath the surface
and then move through the substrate with umbo down, drawing a
current through the ventral mantle margin (rather than the posterior) and
ejecting excurrent water and pseudofaeces through the posterior siphon.
So although water flow may loop about inside the mantle cavity in a path
at least as elaborate as that seen in typical bivalves (Lopez and Holopainen
10 · Bivalve autecology
1987, Holopainen and Lopez 1989), filtered water is not expelled into
the same region from which unfiltered water is drawn. The adaptive
value of this water flow pattern to a bivalve living below the sediments
in a lake is fairly apparent. Water circulation would be negligible in such
an environment, so that an individual Pisidium would be in danger of
refiltering its own medium if it had two siphons side-by- side, as in the
usual arrangement. By this reasoning, one would predict that in general,
bivalves of any taxon inhabiting more lentic habitats would need some
sort of adaptation to minimize refiltration, for example well-developed
siphons that may be directed in opposite directions. And one can under-
stand why the margaritiferid mussels, with no siphons at all, might be
restricted to areas where good water circulation occurs naturally.
In the last several decades, biologists have become aware that because
of the mechanical properties of water, small aquatic organisms experi-
ence their environments quite differently from larger ones. Water is effec-
tively more viscous to a small organism and much more difficult to move.
In addition to the cilia on its gills, Pisidium apparently has rows of cilia
on its foot to help it move the syrupy water. So even setting aside micro-
habitat choice, one would not be surprised to see unidirectional water
flow in any very small bivalve. Churchill and Lewis (1924) reported that
siphons have not developed in unionid juvenile mussels at 0.2–2.0 mm
length, and that incurrent water passes through the anterior mantle of the
mussel, about where the foot protrudes. Yeager and colleagues (1994)
have described pedal-sweep feeding behaviours in Villosa recently shed
from their hosts.
In all freshwater bivalves, mucus and particles trapped on the gill are
moved forward toward the labial palps by yet a third set of specialized
cilia. The palps convey material to the mouth, dropping any excess onto
the mantle for expulsion as ‘pseudofaeces’. A short oesophagus leads to
an elaborate stomach, with typhlosoles, grooves, ciliated regions, and a
rotating crystalline style regulating the flow of mucus and food. Among
other functions, the style is a primary source of digestive enzymes.
Material is either moved into digestive diverticulae, where digestion
seems to be largely intracellular, or to intestine and anus.
Commonly, then, the particles assimilated are only a subset of those
ingested, and the particles ingested are only a subset of those retained. If
any competition for food occurs between bivalves, it is for particles
retained, but a bivalve’s health and success are a function of the particles
assimilated. In a discussion of bivalve feeding, it is important to keep the
Feeding and digestion · 11
distinction between retention, ingestion, and assimilation in mind. These
topics will be discussed separately.
Particle retention
Recent advances in our understanding of particle retention in bivalves
have come primarily from work with marine species. It is important to
note at the outset that the majority of bivalves have gill filaments
equipped with large laterofrontal cirri. These bivalves, including such
diverse families as the Mytilidae, Veneridae, Carditidae, and Myidae, gen-
erally completely retain all food particles larger than 4 m (Riisgård
1988, Møhlenberg and Riisgård 1978). Efficiency varies somewhat
below this level, with some species retaining as much as 70% of 1 m
particles, and others as little as 20%. A second set of bivalves, including
the oysters (Ostreidae) and scallops (Pectinidae), have reduced or absent
laterofrontal cirri. Most of these species can retain particles of 5–6 m
with 100% efficiency, but show reduced efficiency for small particles, and
may be unable to retain 1 m particles.
The most complete work on particle retention in a freshwater bivalve
is that of M. Sprung on Dreissena polymorpha. Given its similarity to the
marine mussels, one would not be surprised at the finding of Sprung and
Rose (1988) that Dreissena has large laterofrontal cirri, and is capable of
some retention of particles down to 0.7 m. The adult mussel retains
5–35 m particles (at minimum) with 100% efficiency, and is about 60%
efficient below 5 m. The larvae retain particles only between 1 and 4
m in diameter, however – an unusually narrow size range compared
with larvae of marine bivalves (Sprung 1989). In addition to laterofron-
tal cirri, Way and his colleagues (1989) have described ‘frontal cirri’ asso-
ciated with frontal cilia in Corbicula and Musculium.
Unionid laterofrontal cirri have been described by Tankersley and
Dimock (1993), and evidence from several studies suggests that they
retain small particles with admirable efficiency. Charcoal slurries have
been experimentally fed to the European Unio pictorum and Anodonta
anatina (⫽piscinalis) by Bronmark and Malmqvist (1982) and to the Asian
unionid Cristaria discoidea by Dudgeon (1980). The former authors com-
pared particle size distribution in the guts with that in unfiltered slurry,
while Dudgeon compared filtered and unfiltered suspensions. Both
studies found high apparent retention rates of 1–2 m particles, Dudgeon
suggesting that small particles might be preferentially retained over larger
particles. He also reported that the size of the individual mussel affected
12 · Bivalve autecology
the distribution of particle sizes retained. Although selective ingestion of
small particles is fairly commonly observed (discussion to follow), the
selective retention of smaller particles by a bivalve the size of Cristaria is
unprecedented as far as I know. Since the data are relative frequencies,
they may reflect selection against large particles.
Rather than concocting an artificial suspension of particles, Paterson
(1984, 1986) used a Coulter counter to compare the size distribution of
particles naturally occurring in vessels of lake water with and without the
unionid Elliptio complanata. Paterson’s results on Elliptio are consistent with
results from the more efficient marine species. The mussel seems to retain
all particles larger than 1.59 m with high efficiency (maximum effi-
ciency at about 4 m), with a significant drop-off in the 1.26–1.59 m
size class.
A most interesting investigation of particle retention in freshwater
bivalves is the work of Cohen et al. (1984) on Corbicula in the freshwater
tidal region of the Potomac River, Maryland. The river along this 40–50
km stretch produced high concentrations of phytoplankton during every
summer low-flow period of the 1960s and 1970s. But a 40–60% reduc-
tion (or ‘sag’) in one 6 km region was noticed in the summers of 1980
and 1981 (Figure 2.1), corresponding to a recently introduced popula-
tion of Corbicula. Corbicula was first recorded in the freshwater tidal
Potomac in November of 1977. Average densities of the clam increased
from 1.2/m2 to 425/m2 in 1979, expressed as 159 g wet weight/m2 in
Figure 2.1. The wet weights shown in Figure 2.1 translate to average den-
sities of 1400 and 1467 clams/m2 in 1980 and 1981, respectively, after
which a die-off occurred. Cohen and his colleagues effectively ruled out
zooplankton, unionid clams, toxic substances, nutrient limitation, and
peak discharges from dams as explanations for the phytoplankton sags of
1980 and 1981, convincingly demonstrating that filter feeding by
Corbicula can noticeably reduce phytoplankton levels in large rivers. From
their known population density, filtration rate, and water flow, Lauritsen
(1986b) suggested that Corbicula might have a similarly large impact on
the Chowan River in North Carolina.
Using calculations similar to those of Lauritsen, Stanczykowska and
colleagues (1976) estimated that unionids, sphaeriids, and Dreissena com-
bined to remove about 8% of the primary production of a 460 ha Polish
lake during a typical growing season. Madenjian (1995) estimated that
the recently introduced Dreissena population of western Lake Erie may
have removed 26% of the primary production of 1990. The zebra mussel
Feeding and digestion · 13
Figure 2.1. Samples taken at four summer dates over a 60 km length of the Potomac
River, Maryland. Depth-integrated phytoplankton concentration (⫻107 cells/l) is
plotted with open symbols and Corbicula biomass (mean wet weight per transect, in
g/m2) is plotted with closed symbols. (Data of Cohen et al. 1984.)
14 · Bivalve autecology
invasion of the early 1990s seems to have had dramatic effects on the
water column of the Great Lakes. The phytoplankton productivity of
inner Saginaw Bay was lowered approximately 38% by zebra mussel col-
onization, while increased underwater irradiance increased benthic
primary productivity (Fahnenstiel et al. 1995a,b). The mussels seem to
remove diatoms (Cyclotella), autotrophic flagellates (Cryptomonas) and a
variety of protozoans very efficiently, while having no significant effect
on the blue-green Microcystis (Lavrentyev et al. 1995). They also seem to
be able to remove bacterioplankton, although in some cases they may
positively affect bacterial abundance by nutrient or organic carbon excre-
tion (Cotner et al. 1995). Caraco and colleagues (1997) have offered con-
vincing evidence that the spread of zebra mussels to the Hudson River
in 1992–94 resulted in a ‘massive’ decline in phytoplankton biomass
similar to that demonstrated by Cohen for Corbicula. The complex and
far-reaching ecological effects of zebra mussel invasion will be re-exam-
ined in Chapter 5.
Particle ingestion
Because of the bivalve capacity to form pseudofaeces, one cannot be
certain that particles retained will be ingested. And it has often been
found that bivalves remove particles from suspension indiscriminantly,
without regard to their value as food (Jorgensen 1966, Haven and
Morales-Alamo 1970). Especially in silty environments such as one might
find in turbid rivers, it would seem possible that organic particles might
comprise such a small fraction of the total retained that a bivalve might
starve. One mechanism by which bivalves can adjust their intake of par-
ticles is by adjustment of their filtration rate (reviews by Jorgensen 1975,
Kryger and Riisgård 1988). Morton (1971) reported that the rate at
which Dreissena filtered water increased when ambient concentrations of
algae or bacteria were increased, reaching a maximum at a turbidity of
0.075 g collodial graphite/l. Filtration rate was markedly reduced as tur-
bidity reached 0.15–0.2 g/l. Way and colleagues (1990) reported an
inverse relationship between filtration rate and the ambient concentra-
tion of suspended particles in three Corbicula populations from rivers of
the southeast United States.
Early workers assumed that in addition to reducing their filtration rates
when presented with undesirable particles, bivalves could exert some
selection between retention and ingestion, so that undesirable particles
tended to be preferentially rejected as pseudofaeces. Selection by specific
gravity was usually suggested to remove inorganic particles. This view has
Feeding and digestion · 15
been criticized on quite strong grounds in the past. For example, some
of the earliest workers pipetted distasteful particles such as carmine
directly into the incurrent siphons of unsuspecting mussels. As Allen
(1924) wrote, ‘Such . . . material, however neutral, of whatsoever dilu-
tion, or however administered through the respiratory water, was never
found subsequently in the alimentary canal.’ Allen then showed that
mussels fed a mixture of carmine and the known foodstuffs Gleocapsa and
Spirogyra did not sort the mixture, but rejected all particles as pseudo-
faeces.
Nevertheless, it now seems clear that some selective removal of unde-
sirable particles can in fact take place, possibly mediated by the labial palps
(Kiørbe and Møhlenberg 1981, Newell and Jordan 1983). Payne and his
colleagues (1995) have reported increased palp-to-gill area ratios in pop-
ulations of Dreissena and Corbicula from habitats where the concentrations
of suspended particles are especially high. Strong, albeit indirect, evi-
dence of differential sorting has been obtained in laboratory feeding
experiments on Corbicula fluminea, famous for inhabiting silty rivers,
canals, and basins. Foe and Knight (1985) monitored tissue growth in 8
Corbicula cultures, 4 inorganic silt concentrations (up to 150 mg/l) by 2
phytoplankton concentrations (Ankistrodesmus). Clam tissue growth was
independent of silt concentration but significantly affected by phyto-
plankton concentration.
One gets the impression that the contents of a typical freshwater
bivalve gut generally duplicate the naturally occurring population of par-
ticles available (Boltovskoy et al. 1995). The gut contents of the Korean
unionid Lamprotula, for example, included detritus, filamentous algae,
and diatoms mixed in with sand and calcareous debris (Chun 1969).
About 60–70% (by volume) of the gut contents of Corbicula sandai in the
Japanese Lake Biwa was detritus, mud, and sand (Hayashi and Otani
1967). Included among the remainder were 89 species of diatoms, 12
species of green algae, 14 species of blue-green algae, rotifers, protozoans,
sponge spicules, etc.
The most informative of such studies compare gut contents to the par-
ticulate composition of the surrounding water. Wallace et al. (1977)
measured tracings from 1626 particles from the stomachs of 3 individual
Corbicula fluminea collected from the Altamaha River, Georgia. Setting
aside diatoms, which they excluded from analysis, Wallace and colleagues
found that 85% of the stomach content volume (1624 particles) was
detritus, with a mean size of 100 m2 and range of 5 m2 to 20000 m2.
I would not have guessed that the particles would be that large, based on
16 · Bivalve autecology
particle retention experiments. But the mean size of detritus particles in
Altamaha water samples ranged from 157 m2 to 270 m2, a fair match
to the Corbicula gut. Amusingly, the remaining 15% of the gut content
volume comprised just two animal particles (unidentified), one 2000 m2
and the other 30000 m2. One has no way of knowing whether the
removal of such a particle (perhaps approaching 200 m on a side) could
be an accident, or if the little clams have to filter 1624 worthless particles
to get one good one.
Gale and Lowe (1971) compared the monthly gut contents of
Musculium transversum in a Mississippi River pool with water samples
taken at the surface. They found no evidence of selective ingestion
among the various genera of the three major algal groups: greens, blue-
greens, and diatoms. There was some evidence that Musculium preferred
green algae over diatoms. But admittedly, the water sample at the surface
may not reflect the composition of the plankton at the bottom. Ten
Winkel and Davids (1982), on the other hand, compared stomach and
midgut contents of Dreissina collected from the underside of a platform
with the surrounding waters of the Dutch Lake Maarsseveen I. Samples
were taken on five different dates from March to May. The data clearly
suggest that the very large, colony-forming diatom Asterionella formosa,
with clumps of long, slender cells 100 m each, is not readily retained.
It is fairly easy to see how such colonies might be excluded at the incur-
rent siphon. Ten Winkle and Davids also suggest a positive selection for
smaller diatoms on the order of 15–45 m, but again, since the data are
relative frequencies this may be the inevitable consequence of selection
against Asterionella. They also noted an apparent deficit of the smaller
(about 15 m) green algae Phacotus and Trachelmonas in the stomachs of
Dreissena. This is quite at variance with other studies of bivalve feeding,
including Morton (1969a) on Dreissena. Organic particles of this size are
generally retained and ingested readily.
Assimilation
Evidence regarding the actual assimilation of ingested particles comes
from several quarters. For example, assimilation has been inferred by
careful comparison of carbon taken in and lost from individual bivalves.
Walz (1978a) fed Dreissena a pure diet of the diatom Nitzschia and
obtained a value of 40% assimilation. Stanczykowska and co-workers
(1975) monitored how efficiently that naturally occurring foodstuffs
were assimilated by Dreissena suspended in chambers in a Polish lake.
Although the individual elements of the seston were not characterized,
Feeding and digestion · 17
the authors noticed dramatic declines in efficiency on days when the large
(215 m) dinoflagellate Ceratium bloomed.
Radiolabelling, the most direct approach, has been used to investigate
the digestive capabilities of Corbicula by Lauritsen (1986a). Lauritsen fed
individual clams pure cultures of the spherical, unicellular green alga
Chlorella, the sickle-shaped unicellular green Ankistrodesmus, and the fil-
amentous blue-green Anabaena, all labelled with 14C. Because of its shape
and reputation for unpalatibility, I would have predicted very little assim-
ilation for Anabaena. But remarkably, the assimilation rates for Anabaena
(58%) and Chlorella (56%) were almost identical, with slightly reduced
rates for Ankistrodesmus (47%). Intact cells of Chlorella and Ankistrodesmus
were commonly found in clam faeces, apparently protected by their
tough cell walls. Anabaena was never observed in faeces, but was the only
foodstuff to appear in pseudofaeces. So although large, tangled mats of
Anabaena must surely be safe from Corbicula, individual filaments may be
swallowed quite handily. At the other end of the spectrum, Silverman and
colleagues (1995) offered 35S-labelled bacteria of several species (1.3 m
to 4.1 m size range) to Dreissena, reporting assimilation in all cases.
Dreissena seemed more efficient in clearing and assimilating E. coli
(1.7–2.9 m) than did Corbicula or the unionid Carunculina, an ability
perhaps attributable to their laterofrontal cirri, which seem to be larger
and present in greater number (Silverman et al. 1996).
We have already mentioned the unusual feeding method adopted by
tiny clams of the genus Pisidium. Lopez and Holopainen (1987) found
that although diatoms and chloroplasts were commonly observed in the
guts of Finnish Sphaerium corneum and the larger (up to 13 mm) Pisidium
amnicum, the stomachs of the smaller (2 mm) Pisidium casertanum and P.
conventus were always empty. These latter two species live in loose, floc-
culent sediment (water content 60–90%) of high bacterial content in
Lake Paajarvi, Finland: P. casertanum in shallow water and P. conventus in
the profundal zone. Using the 51Cr/14C balance method (Calow and
Fletcher 1972), Lopez and Holopainen showed that these two Pisidium
can assimilate a natural suspension of interstitial bacteria at 40–60% effi-
ciency. Absorption of bacteria on particles was much less efficient. The
authors suggested that although larger sphaeriids quite probably do ingest
and assimilate algae and/or detritus, the smaller Pisidium may feed exclu-
sively on bacteria.
Yet a third method by which assimilation has been verified is by mon-
itoring growth or reproduction in laboratory populations fed controlled
diets. Foe and Knight (1986) measured 30-day tissue growth in cultures
18 · Bivalve autecology
of Corbicula fluminea fed six genera of green algae in various combina-
tions. Evidence suggested that Selenastrum emits some toxic substance
inhibiting Corbicula filtration. But strong positive growth was noted from
all trialgal combinations of Ankistrodesmus, Chlamydomonas, Chlorella, and
Scenedesmus, as well as the dialgal combination of Chlorella and
Chlamydomonas. Growth was slower, and sometimes negative, in other
dialgal combinations. Interestingly, the maximum laboratory growth
obtained on a defined algal diet was still only about 25% of the growth
the authors saw when clams were cultured in water from their own en-
vironment, enriched to stimulate nanoplankton blooms.
The dietary requirements of larval Dreissena contrast strikingly with
those of other freshwater bivalves, including adults of their own species.
Sprung (1989) offered laboratory cultures of Dreissena larvae 47 different
pure cultures of algae, green and blue-green, singly and in combination,
plus two bacterial strains and yeast. Although some of the algal strains
were too large to be ingested (recall Dreissena’s narrow particle size
requirements), most foodstuffs were in fact retained. But larval shell
growth was in all cases poor, and rearing success ‘extremely sporadic’.
Sprung manipulated density of larvae and food, light, temperature, aer-
ation, and vessel, but could not achieve consistent rearing success. So
although the evidence is not conclusive, the suggestion is that the food
requirements of larval Dreissena may be exacting.
As long ago as 1948, Rodina found that Sphaerium, Musculium, and
Pisidium could grow on a diet of pure bacteria at natural concentrations.
To my knowledge, the most thorough laboratory study of freshwater
bivalve diet was completed by Mackie and Qadri (1978) on Musculium
securis from Ontario. These authors showed that M. securis would not
grow in unsupplemented lake sediment or pond-bank soil, nor would
they grow in plain dried leaves, but they would grow and reproduce in a
combination of the two. Importantly, M. securis will not grow under any
condition if the soil is autoclaved. Thus it would appear that nutrients
from the leaves nourish micro-organisms from the soil, and micro-organ-
isms from the soil nourish Musculium. But the little clam lives in regions
where tree foliage is absent, and gut contents typically contain a wide
variety of algae, especially the golden-brown filament Vaucheria and the
blue-green filament Lyngbya. The correlation between the density of the
little clams and the biomass of Vaucheria and Lyngbya on the bottom is
striking. Thus it is clear that M. securis can probably retain, ingest, and
assimilate a great variety of foodstuffs.
Similarly inferential evidence on assimilation in natural populations of
freshwater mussels comes from the work of James (1987) with the New
Feeding and digestion · 19
Zealand hyriid mussel, Hyridella menziesi. He estimated that production
by algae and aquatic macrophytes in a clear, oligotrophic lake might
account for only about 5% of the calculated energy needs of its dense
mussel population, and suggested that the ultimate source of energy for
this population would have to be leaves blown or fallen into the lake, pre-
sumably with associated bacteria.
In spite of all the work done on particle retention and ingestion in
unionoid mussels, and in spite of all the attention devoted to their biology
early this century, surprisingly few controlled studies have been made of
the food they actually assimilate. Early workers were able to rear mussels
only in crates or troughs continually supplied by unfiltered, natural water
(Coker et al. 1921). Imlay and Paige (1972) noted growth on a diet of
commercial trout fry food, although this affords little insight into natural
diet. Hudson and Isom (1984) reared Anodonta imbecillis from mature glo-
chidia to about 5 mm juveniles in 74 days using filtered Tennessee River
water with induced phytoplankton bloom, supplemented with silt (fil-
tered pond substrate). Hudson and Isom listed 16 planktonic taxa
common in their cultured river water, including green algae, blue-green
algae, diatoms, and ciliates. Juvenile mussels would not grow on an arti-
ficial diet composed of three taxa of green algae, nor would they grow
deprived of silt. Gatenby and colleagues (1996) also reported optimum
growth and survivorship for juvenile Villosa and Pyganodon in a trialgal
diet supplemented with sediment.
Feeding period
There has been some work on rhythm, or lack thereof, in unionid activ-
ity, but as Jorgensen (1966:103) has stated, ‘the various authors are not
in agreement’. Jorgensen mentioned three studies showing entirely
Feeding and digestion · 21
different periods of activity for Anodonta. But the evidence is now fairly
strong that any periodicity Anodonta may show in valve activity occurs at
two different frequencies. Barnes (1962) described a rapid rhythm result-
ing in up to 20 partial closures per hour and a slower rhythm, showing
perhaps three complete closures of several hours each per day in A.
cygnea. De Bruin and David (1970) also noted regular interruptions of the
water current in A. cygnea at a frequency of 6–40 per hour, but their
experiment was not designed to detect longer-period cycles. And rapid,
regular valve opening was mentioned yet again by Salbenblatt and Edgar
(1964) in the North American A. grandis, although these authors were
not equipped to quantify such short-period cycles.
Salbenblatt and Edgar reported that although A. grandis spent most of
the time closed under their experimental conditions, different animals
displayed cycles of 7–58 hours, depending upon light and temperature.
Periods of cyclic activity were short-lived, however, and easily disrupted
under laboratory conditions.
The situation with Dreissena seems to be similar. Morton (1969b)
reported regular nocturnal periods of filtration in laboratory populations
of Dreissena, followed by diurnal quiescence. Sprung and Rose (1988)
noted a short-period phenomenon: pauses of some seconds alternating
with several minutes of activity, depending on food concentration. But
no regular activity was noted in the laboratory by Walz (1978a) or in the
the wild by Stanczykowska and colleagues (1975). Many sorts of behav-
iour are extremely sensitive to experimental conditions – feeding in
bivalves may be no exception.
It is the longer-period cycles that are more interesting from an ecolog-
ical standpoint. Thus we are fortunate that the longest-term study of
periodic feeding in unionids to my knowledge is also the study that seems
most likely to be free of observer-induced error. Salànki and Vero (1969)
attached a pair of small electrical devices to each valve of four individual
Anodonta cygnea, allowing them to monitor valve opening after their
release at a depth of 2 m in Hungary’s Lake Balaton. The four Anodonta
displayed irregular periods of filtering and quiescence (about 1–3 cycles
per day). Daily fluctuations of the environment tended to synchronize
the animals only slightly.
Salbenblatt and Edgar found the long-term pattern of valve activity to
be similar among three genera of North American mussels, Anodonta,
Anodontoides, and Lampsilis – usually one or two cycles per day.
Intriguingly, the authors noted in passing that unlike Anodonta, Lampsilis
filters during almost its entire cycle, and that there may be some diel
22 · Bivalve autecology
differences in filtering activity among the species. The laboratory experi-
ments of Chen and colleagues (1999) confirmed that the lentic Pyganodon
grandis tends to gape nocturnally, while the lotic Quadrula pustulosa is a
diurnal gaper. An in situ comparison of feeding schedules in co-occur-
ring freshwater bivalves would be an interesting, if technically demand-
ing, study.
Habitat
Unionoids
Any discussion of the habitats of freshwater molluscs must begin with the
great review work of Boycott (1936). Briefly, Boycott considered that the
four large British unionids (two Anodonta, two Unio) need firm, muddy
bottoms. ‘They will not live either on a hard stony or gravelly bottom or
on one which is covered with a thick layer of soft humus in which they
would have to disappear before they found firm ground’ (pp. 132–133).
Boycott would not hazard any absolute statements about the habitat of
Margaritifera margaritifera on the other hand, but noted that ‘A typical locus
is a quick running river up to 3 or 4 feet deep with a mixture of boul-
ders, stones, and sand; it can burrow in fine gravel and particularly likes
the sand which accumulates behind large stones’ (p. 133). Boycott
referred to a study showing that M. margaritifera occurs only to the north
and west of a line from Scarborough to Beer Head, and footnoted ‘If a
line be drawn from Hull to Gloucester and then on to Plymouth, it can
be said roughly that all the country to the east and south of it was for
Parliament, and everything to the west and north for Charles.’ This I
transmit for the benefit of the next generation of freshwater malacolo-
gists.
Figure 2.2 shows a more contemporary treatment of mussel habitat
choice in Lake Matikko, Finland (Haukioja and Hakala 1974). Four
species co-occur, all showing a pattern that is characteristic of lake-
dwelling unionoids everywhere. From about 1/m2 in the shallows,
mussel density peaks rapidly then falls gradually as depth increases.
Haukioja and Hakala note that the depth distribution of Pseudanodonta
complanata differs somewhat from those of the other three species,
peaking at 2–3 m rather than at 1 m.
Similar patterns have been reported for unionoid populations in
Poland (Lewandowski and Stanczykowska 1975), the United States
(Fisher and Tevesz 1976), England (Stone et al. 1982), and Brazil (Henry
Habitat · 23
Figure 2.2. Average densities (N/m2) of Unio tumidus (Ut.), Unio pictorum (Up.),
Anodonta anatina (Aa.), and Pseudanodonta complanata (Pc.) at 8 depths in Lake
Matikko, Finland. (Data of Haukioja and Hakala 1974.)
and Simão 1985). Because light, temperature, food, and flow rates would
seem to deteriorate with depth, it seems in general that all mussels prefer
to live as close to a lake surface as possible. (This does not seem as uni-
versally true for pisidiids, however, as we shall see.) But because mussels
cannot predict low water, ice, and predation, population densities are
reduced at depths less than 1 m by mortality. Thus as a rule, mussel pop-
ulations of lentic habitats reach maximum abundances about 0.5 m to
2.0 m, varying only in the rates at which they disappear with depth there-
after.
This effect may not be as pronounced in clear, oligotrophic lakes. For
example, Burla (1972) tabulated the abundance of Anodonta in
transects collected over a period of 1 year from Lake Zurich, Switzerland.
The maximum density in any sample, about 50 individuals per square
metre, occurred at 1 m in a June/July sample. But I would not judge that
any significant decrease in abundance occurred in most samples until
depths of 9–12 m. In some transects a (probably non-significant)
maximum abundance occurred as deep as 8 m. Another unusual case is
presented by the dense population of Hyridella menziesi inhabiting clear,
oligotrophic Lake Rotokawau, New Zealand (James 1987). This lake was
24 · Bivalve autecology
formed when an explosion crater filled with water 4000 years ago, and
has very steep sides. Maximum mussel densities were not reached until
depths of 5–10 m in some of James’ transects, apparently because the
bottom levelled off to a shelf at those depths. But once again, the
maximum population density recorded in the lake (814/m2!) was at 1 m
in a transect taken where the sediment on the steep sides was exception-
ally stable.
There would seem to be potential for space competition among lake-
dwelling mussels at their optimum depths, in a fashion analogous to that
well-documented in rocky intertidal environments. In addition to the
high densities recorded by James, Fisher and Tavesz (1976) reported 15
Elliptio complanata/0.1 m2 in a Connecticut Lake, and Bronmark and
Malmqvist (1982) recorded about 100 mussels/m2 in a Swedish lake
outlet. A density of 161 mussels/m2 was the average at 1 m in Lake
Matikko (Figure 2.2). Haukioja and Hakala did not mention the possibil-
ity of competition between their four species of mussels, but I suggest
that it may be intense, and that Pseudanodonta may be excluded from shal-
lower water because of it. We shall return to this subject in Chapter 9.
Often it is the composition of the substrate that determines the depth
to which mussel populations extend. Although modal mussel abundance
again occurs at 1 m, the transects of Cvancara (1972) found
Anodonta grandis and Lampsilis radiata to depths of 8 m and 9 m on a side
of Long Lake, North Dakota, where sand and gravel were prevalent.
Cvancara speculated that the depth to which Long Lake mussels extend
on coarse substrate may be a function of the metalimnion, the top of
which varies from 4 m to 12 m in Long Lake in the summer. Mussels
ranged only to a depth of 1 m on another side of the lake where mud
and aquatic plants predominated.
Figure 2.3 shows the depth distribution of Elliptio complanata and
Anodonta grandis in Lake Bernard, Ontario, as recorded by Ghent et al.
(1978) over 11 circular plots using . The authors noted that the
substrate in Lake Bernard is typically fine, hard sand down to about 2.5
m, where it is generally replaced by soft silty mud between 3 m and 6m,
so that the substrate is overlain by a 10 cm silt ‘blanket’ below 6 m. The
distribution of E. complanata seems to correspond closely to firm bottom,
with the usual density reduction nearshore. Fisher and Tavesz (1976) have
noted that in life, E. complanata is found half-buried, anterior downward,
and suggested that it cannot maintain this life position in the mud found
on the bottoms of deeper lakes. Elliptio complanata is very compressed lat-
erally, and I have rarely seen it lying on its side as the fatter Anodonta often
Habitat · 25
Figure 2.3. The bar graphs show population densities (N/m2) of Elliptio complanata
and Anodonta grandis at 11 circular plots down a depth gradient in Lake Bernard,
Ontario. The line shows the cumulative growth to age 9 years (shell height) of A.
grandis for combined samples from five depth ranges. (Data of Ghent et al. 1978.)
do, for example. Lying on its side, the siphons of E. complanata might not
be 10 mm above the bottom. So taken with the results of Kesler and
Bailey (1993) and the mark-recapture study of Kat (1982a), it appears that
substrate may be more critical than depth per se, at least to this mussel.
Returning to Figure 2.3, one’s first impression of Anodonta grandis is
that it selects deeper, muddier sites in Lake Bernard. But on the right
scale of the figure one finds recorded a striking inverse relationship
between A. grandis growth rate and depth. Ghent et al. viewed the dis-
tribution of A. grandis in Lake Bernard as evidence that it is adapted to a
soft, muddy bottom. This is true, but maladaption to a hard, sandy
bottom may describe the situation just as well. Perhaps the 12 A. grandis
from 0.5–2.0 m depth upon which the figure is based were collected
from little pockets of softer substrate behind obstructions in an otherwise
hard, sandy region.
The effects of depth and substrate on lake-dwelling unionids have been
examined experimentally on several occasions. Narrow Lake, Alberta, is
another lake in which Anodonta grandis occurs to some depth. Hanson
and his co-workers (1988) could not detect any difference in the annual
26 · Bivalve autecology
growth rates of mussels collected from 1 m to 7 m in depth, even though
the deeper waters are significantly cooler. Samples of clams measured and
stocked in small enclosures at depths of 1 m, 3 m, 5 m, and 7 m (using
uniformly sandy substrate) did, however, show significant variance in
growth rate, suggesting that migration between depths may be sufficient
to keep population growth rates uniform. Inexplicably, no such migra-
tion seems to occur in Lake Bernard.
Hinch et al. (1986) found that a population of Lampsilis radiata from a
sandy site in Long Point Bay, Lake Erie, grew faster than a population
from a muddy area in the same bay about 10 km distant. They marked
and transplanted samples of Lampsilis from sand to mud and mud to sand
for a growing season. Somewhat surprisingly, no improvement was noted
in the growth rate of the clams transferred from mud to sand over those
remaining in mud. (Recapture rates may have been low for some cate-
gories, however.) Clams originally from the sand habitat seemed to be
faster growers regardless of where they spent their four months, suggest-
ing perhaps some genetic divergence between populations occupying
different regions of Lake Erie. Further complicating the picture, a sub-
strate-choice experiment performed by Bailey (1989) showed that all
Inner Long Point Bay L. radiata prefer mud over sand. Individuals from
the sand-dwelling subpopulation actually showed greater preference for
mud than did mud-collected individuals, at least in artificial ponds. One
wonders about the generality of these unusual findings.
Cvancara (1972) found a third unionid population in Long Lake,
Anodontoides ferrusacianus, a species ‘generally characteristic of the shallow
water in creeks and rivers’. Although Lampsilis and Anodonta extended to
8–9 m, Anodontoides was not collected below 2 m anywhere in the lake.
The species was not collected at all in a 1969 survey, suggesting that its
existence in Long Lake may be tenuous. It seems clear that freshwater
mussels show varying degrees of adaptation to the lentic environment.
Lewis and Riebel (1984) did not notice any substantial differences
between A. grandis, L. radiata, and E. complanata in burrowing speed or
efficiency in sand, clay, mud, or gravel. But we have noted that the
heavier, compressed shell of Elliptio complanata does not seem as stable in
soft substrate as the rounder shells of Anodonta grandis.
Turning now to more lotic environments, Norelius (1967) and
Bronmark and Malmqvist (1982) have studied populations of Anodonta
anatina and Unio pictorum inhabiting Swedish lake outlets. The latter
authors took 64 samples of 0.05 m2 each, recording distance from the
lake, distance from the margin, depth, current velocity, vegetation, per-
Habitat · 27
centage organic content of the sediment, and particle size, and performed
several discriminant analyses. Unio tended to be found more toward the
centre of the river, away from vegetation, and on coarser sediment than
Anodonta. Unio was strikingly more common nearer the lake than
Anodonta, and showed increased growth rates. Norelius found the trend
just as strong but in the opposite direction: Anodonta more common near
the lake and Unio increasing downstream. Bronmark and Malmqvist
suggested that food quality (e.g. chlorophyll a content) may be improved
immediately below their study lake.
Our discussion of mussel habitat choice thus far may give the impres-
sion that unionoid mussels prefer lakes and ponds. They do not.
Worldwide, the great majority of mussel species are never found in lentic
habitats. But ecological studies of truly riverine unionoids are fewer than
those directed toward lake and pond populations. Huehner (1987) mod-
ified a substrate-choice technique developed by Meier-Brook, placing
mussels on edge at an interface between sand and gravel. Most of
Huehner’s experiments involved two populations of A. grandis and three
populations of L. radiata, lake-dwellers with which we are already famil-
iar. These mussels always showed highly significant preference for sand.
But Huehner included one truly riverine species, Elliptio dilatata, in his
trials. Elliptio dilatata preferred gravel over sand by a margin of 42 to 32,
which though not significantly different from 1:1 is certainly significantly
different from Huehner’s five populations of mussels more characteristic
of slackwater.
Stern (1983) collected 28 species of unionids in his 14 tran-
sects of the Lower Wisconsin and St Croix Rivers, Wisconsin. He divided
his samples into four different depth categories from 1 m to 3.3 m and
four different substrate categories (sand, mud–sand, mud–sand–gravel,
and sand–gravel–boulder). Stern found mussels at all depths and all sub-
strates except pure sand. The most striking feature of his data, however,
was that only six of the mussel species were ever found on mud–sand:
60% of all Anodonta grandis, 40% of Lampsilis radiata, and four riverine
species at trace frequencies. Here we see our two familiar lentic union-
oids inhabiting the muddy parts of the Wisconsin and St Croix Rivers,
while 26 other unionoid populations co-occur on coarser sediments.
One of the saddest ‘natural experiments’ we have witnessed in North
America has been the impoundment of most of our free-flowing rivers
during the last 50–60 years. Lake Ashtabula, North Dakota, was formed
in 1950 by the impoundment of the Sheyenne River. Eight mussel
species are known from the river above the lake, and 11 species below.
28 · Bivalve autecology
Cvancara and Freeman (1978) collected belt transects by at 6
stations on Lake Ashtabula from 1 m to 6 m in depth at 1 m increments.
They found alive 81 A. grandis and 12 L. radiata, both at a modal depth
of 3 m. This seems odd in light of the fact that substrate at this depth is
generally soft, organic mud. But two drawdowns of over 2 m had
occurred at Lake Ashtabula in the five years previous to Cvancara and
Freeman’s study. The density of L. radiata was significantly higher in the
Sheyenne River above the lake than in the lake and below. But popula-
tion density of A. grandis in the lake is comparable to that in the river
above and significantly greater than that below the dam. Of the river
mussels, Cvancara and Freeman found two individual Amblema plicata,
one Lasmigona compressa, and none of the other seven species.
One might speculate that the habitat choice of juvenile unionacean
mussels might be different from that of adult mussels, and possibly more
similar to that of the smaller pisidiids (a discussion of which follows
shortly). But this does not seem to be the case, at least for riverine mussels
of the North American interior, such as Fusconaia, Pleurobema,
Medionidus, and Villosa. Neves and Widlak (1987) took 573 cm2 circular
bucket samples in Big Moccasin Creek, Virginia, a small stream aver-
aging only 7 m wide and 0.2 m deep at low flow. They took 15 samples
from each of three ‘major habitats’: pool (slow flow, silty bottom), run
(moderate, laminar flow, coarse substrate), and riffle (swift, turbulent
flow, coarse substrate), and two ‘microhabitats’: behind boulders (which
occurred in riffle and run only) and stream bank (which occurred in all
three major habitats). The age-0 mussels collected by Neves and Widlak
averaged only 2.7 mm shell length, and the age-1 individuals but 6.4 mm,
certainly within the size range of pisidiid clams. But Table 2.1 shows that
while the pisidiids (Sphaerium striatinum, Pisidium compressum, and P. caser-
tanum) preferred silty pools, unionids preferred riffle and run regardless
of age. Juvenile mussels seem to differ from adults only when very young
in their significant preference for the sandy areas behind boulders.
Corbiculoids
Boycott (1936) counted 4 species of Sphaerium and 15 species of Pisidium
in Britain. He stated that ‘As the species are as much (or more) affected
by the nature of the bottom in which they live as by the character of the
whole habitat, their preferences cannot always be described in ordinary
terms . . . The species may therefore be extremely localised in a habitat,
and collecting them satisfactorily is something of a special art.’ He specif-
ically noted substrate or depth preferences for only five species. Ordinary
Habitat · 29
Table 2.1. Number, age group, and habitat of unionoid mussels and pisidiid
clams collected in 75 quantitative samples from Big Moccasin Creek,Virginia,
on nine sampling dates
Unionids, by age
Pisidiids,
Habitat 0 1 2 3 4⫹ all ages
Pool 0 4 3 1 0 1218
Run 4 3 3 1 7 717
Riffle 5 6 3 3 13 189
Boulder 15 6 7 7 8 664
Bank 1 1 0 1 8 557
Figure 2.5. The density of Dreissena polymorpha in four Polish lakes as a function of
depth. (Data of Stanczykowska et al. 1975.)
(2.5–4.5 mm), and any substrate was preferable to none at all. Field dis-
tributional data from a South Carolina stream corroborate this: maximum
abundance on sand, lesser in ‘sand/muck’, and absent from gravel habi-
tats (Leff et al. 1990). Thus the substrate preference of Corbicula seems
similar to that of P. casertanum in the Ottawa River, or M. securis in the
experiments of Mackie and Qadri. Its substrate choice seems quite differ-
ent from that of M. transversum as determined by Gale (1971), however.
Dreissena
This is clearly a subject quite separate from habitat choice by other fresh-
water bivalves. After perhaps 1–3 weeks in the plankton, Dreissena must
find a suitable surface upon which to settle and attach a byssus.
Stanczykowska (1978) noted that the following regions have generally
been found unsuitable: the hypolimnion (where oxygen is insufficient),
the shore and shallow littoral (where strong water currents may not allow
settlement), and areas of inappropriate substrate (e.g. flocculent silt).
Figure 2.5 shows some typical data on the depth distribution of Dreissena,
extracted from a larger study of 13 Polish lakes by Stanczyskowska and
colleagues (1975). Maximum abundances are usually about 4 m, some-
what deeper than our experience with unionacean and sphaeracean
Reproduction · 35
bivalves has led us to expect. Ranges are much deeper in the exception-
ally clear waters of lakes Geneva and Zurich (Stanczykowska 1977).
Dreissena depth distribution is probably more a function of current
speed and substrate than depth per se. Once settled, Dreissena shows
optimum growth in shallower regions with at least some natural water
movement, as we have seen in the other bivalves (Smit et al. 1992).
Although its larvae tend to settle most frequently in regions of calm,
Stanczykowska noted in her 1977 review that larvae cannot settle where
fine silt may bury them or clog their gills, preferring macrophytes, stones,
shells, and other hard items uncommon in deep water. Thus low density
populations are found in patches on solid substrates, not in the shallow-
est water but not especially deep, and high density populations often tend
to form continuous belts. Stanczykowska notes that populations migrate
from shallower areas in the winter, not significantly extending their lower
depth ranges but rather condensing.
Settlement in regions of otherwise soft, inhospitable bottom is no
doubt facilitated by a significant tendency for Dreissena individuals to
aggregate (Stanczykowska 1964). Lewandowski (1976) placed a large
number of small (4–10 mm) individuals in an aquarium with large, living
mussels, dead mussel shells, and pebbles. After 20 days, the small Dreissena
were significantly more common on the living individuals of their own
species.
Reproduction
A large portion of our knowledge regarding unionoid reproduction
derives from the U.S. Bureau of Fisheries which, under the influence of
an expanding pearl button fishery, maintained an active program of
mussel propagation for many years (Lefevre and Curtis 1912, Coker et al.
1921). More recent reviews have been offered by Fuller (1974) and Kat
(1984). The reproduction of spharaeceans has received considerably less
attention, although Heard (1965, 1977) has compiled a fair summary for
Pisidium, Sphaerium, and Musculium. Burky (1983) and Mackie (1984a)
have reviewed bivalve reproduction generally.
Unionoid gonochorism
Sexes are separate in most unionoids. In some taxa the posterior shell
margin of the female becomes enlarged to accommodate bulging mar-
supia, resulting in one of the most striking examples of sexual dimor-
phism in the Mollusca. But as is true for bivalves generally, in most
36 · Bivalve autecology
unionoids gender can be distinguished only by histological examination
of the gonads, which lie in close proximity to the digestive gland in the
central visceral mass. One might imagine, given the commercial impor-
tance of such marine bivalves as oysters, clams, and scallops, that bivalve
mechanisms of sex determination might be understood. Alas, although
some progress has been made recently (Guo et al. 1998), we are as ignor-
ant of these matters for most commercial bivalve species as we are for the
relatively obscure species of fresh waters. The discovery that mitochon-
dria show ‘double uniparental inheritance’ in unionoids, as previously
reported for marine mytilids, has led to some speculation that cytoplas-
mic factors may be involved in bivalve sex determination (Skibinski et al.
1994, Hoeh et al. 1996, Liu et al. 1996).
Large random samples from unionoid populations usually contain
ratios of male:female not significantly different from 50:50. Table 2.2
shows a sample of nine studies (20 populations) involving collections of
at least 30 adult unionoids made without conscious regard to size or sex.
The sex ratio in 18 of these populations is not significantly different from
equality, and rejection of the 1:1 hypothesis in a nineteenth (Lampsilis
radiata, Kat 1983) may be a type I statistical error. The data of Cruz and
Villalobos (1984) on sex ratios in the Limon population of the
Neotropical mycetopodid Glabaris luteolus are anomalous. They reflect a
highly significant male bias, even after Bonferroni correction. The
authors were hard put to explain this phenomenon. But it may reflect
phenotypic plasticity of sexual expression, as we shall see.
Samples of unionoids often show some variation in sex ratio when
sorted by size or apparent age class. Smaller individuals seem more likely
to be male. This phenomenon has been interpreted as evidence for pro-
tandry; some individual mussels may pass through a male stage in their
development to female reproductive function, while other individuals
remain male (Tudorancea 1969, Kat 1983, Downing et al. 1989). Other
workers have interpreted such data not as evidence for protandry, but
simply a result of sexual dimorphism with respect to size (Smith 1979,
Cruz and Villalobos 1985). There is some anecdotal evidence of proto-
gyny in a captive population of the Malayan unionid Contradens ascia
(Berry 1974). Clearly our understanding of sex determination in the
unionoids remains in its infancy.
Unionoid populations vary widely in the timing of their gametogenic
cycles. Gametogenesis of both sorts proceeds year round in Chilean
Diplodon (Peredo and Parada 1986) and in four Virginia unionid popula-
Reproduction · 37
Note:
a
indicates a significant difference from 1:1, by goodness-of-fit 2.
38 · Bivalve autecology
tions surveyed by Zale and Neves (1982b). Although active spermatogen-
esis occurs year round in Michigan Actinonaias and Hong Kong Anodonta,
oogenesis is reduced during the colder months (van der Schalie and van
der Schalie 1963, Dudgeon and Morton 1983). Both oogenesis and sper-
matogenesis decrease during the colder months in Florida populations of
Anodonta (Heard 1975) and Alabama populations of Pleurobema (Yokley
1972). In any of these cases, gametes are ultimately released into the
suprabranchial chamber, through which external water is being pumped
for the purpose of feeding.
As is true for bivalves generally, the fairly lengthy gametogenic period
displayed by individual unionoids is followed by episodic spawning. In a
turn quite different from typical marine bivalves, however, it is only the
male that ‘spawns’ in the ordinary sense, releasing sperm into the en-
vironment. I am not aware of any direct experimentation regarding the
‘trigger’ for spawning in unionoids, but observational evidence suggests
that temperature is the primary factor, just as in the better studied marine
bivalves. (This subject will be pursued in Chapter 4.) Sperm morulae
(multinucleated masses) have often been reported, although their signif-
icance is not clear (Edgar 1965, Heard 1975, Jones et al. 1986). Sperm
are taken through the incurrent siphon of the females, along with food
particles and miscellanea. It is not clear how the female transports her
eggs from her gonad to her gills. Fertilization may occur in the supra-
branchial chamber itself, or perhaps in the water tubes of the gills
(Mackie 1984a).
Unionoids are distinguished from all other bivalve species by the adap-
tation of their larvae to parasitic development, usually on fish, and the
modifications of the female reproductive system that this entails. Embryos
develop in special chambers in the female’s gills, called marsupia. There
they are held for varying lengths of time, developing into a dispersal stage
called a glochidium in the unionids and hyriids, a haustorium in the
mutelids, and a lasidium in the mycetopodids. The period of larval devel-
opment and holding (during which the female is said to be ‘gravid’) as
well as the phenomenon of ectoparasitism on fish, will be examined in
greater detail both later in this chapter, and again in Chapter 4.
It has been speculated that the likelihood of fertilization in unionoid
populations is exceptionally low. Downing and his colleagues (1993) used
and a fine system of co-ordinates to take detailed samples of the
Elliptio complanata population inhabiting a 6 m⫻7 m portion of Lac de
l’Achigan, Quebec. Their effort was concentrated on two July days after
Reproduction · 39
Figure 2.6. The result of Bauer’s (1987) experiment involving 84 Margaritifera mar-
garitifera. The arrows indicate net sex change after 7 weeks of displacement upstream.
the population had spawned, but prior to the release of mature glochidia.
The marsupium of each female was examined to estimate percentage fer-
tilization. Mussels were collected by hand, and as such ‘endobenthic’ (i.e.
buried) individuals were not tallied. Nevertheless, their data seem to
suggest almost complete fertilization failure where local mussel densities
fell below about 15 individuals per square metre. This is much lower than
typical mussel densities observed in the developed world today, and bodes
ill for the future of our unionoid faunas.
Bauer (1987) reported the results of a remarkable experiment involv-
ing the sex characteristics of a population of Margaritifera margaritifera
inhabiting a rapid, mountainous stream in Germany. He hypothesized
that the displacement of a small sample of mussels upstream from a large
bed prior to spawning ought to adversely impact their fecundity, by
reason of fertilization failure. Such was not the case; glochidia developed
abundantly even in his most upstream individual. Bauer repeated his
experiment the following year, marking each of 84 mussels and assessing
the sex of each via a needle-punch biopsy prior to displacement
upstream. He was surprised to discover that a substantial fraction (15%)
appeared to be hermaphrodites. He was even more surprised when he
reassessed the sexuality of his 84 animals seven weeks later and found that
28 had changed sex (Figure 2.6). Only 3 of 45 males seemed to have
changed, a result that Bauer thought might be attributable to initial errors
in sexing. But Bauer considered his observation that 69% of the displaced
females became hermaphroditic an adaptive response to a low concentra-
tion of exogenous sperm. It is not clear how general this phenomenon
may be over the Unionoidea; many aspects of the biology of Margaritifera
are unusual. But broader subject of hermaphroditism in unionoids will
be pursued in the section that follows.
40 · Bivalve autecology
Unionoid hermaphroditism
As is true for bivalves generally, simultaneous hermaphrodites seem
present at low frequencies in most unionoid populations. In addition, a
few species appear predominantly hermaphroditic. The phenomenon
was first reported in the unionoids by Sterki (1898). Van der Schalie
(1970) surveyed 97 North American unionid species (1871 individuals),
finding four species ‘dominantly hermaphroditic’and detecting ‘sporadic’
hermaphroditism in 22 additional species, generally those with large
sample sizes. Hermaphroditism is also well documented in Eurasian uni-
onids (Bloomer 1939, Nagabhushanam and Lohgaonker 1978) and in the
hyriids and mycetopodids of South America (Avelar et al. 1991, Avelar
1993, Avelar and de Mendonca 1998). The gonads of unionoid her-
maphrodites typically show either monoecious acini (producing both
eggs and sperm) or closely intermingled male and female acini (Heard
1975). Sometimes testicular and ovarian tissues seem regionally distinct,
but gonads are never entirely separate within a single hermaphrodite.
Sporadically occurring hermaphrodites usually show differential devel-
opment of the two gonadal types, such that the functionality of their her-
maphroditism is dubious. This is not always the case in populations where
hermaphrodites predominate.
During the 1930s, Bloomer (review 1939) performed an influential
survey of hermaphroditism in English populations of Anodonta cygnea. A
sample of 95 individuals from Bracebridge Pool included 20% males, 59%
females, and 21% hermaphrodites. But a sample of 66 individuals from
Highams Park included 38% hermaphrodites, with all the remainder
apparently normal females. And populations from Beam Brook,
Holmwood, Eardisland, and the River Frome appeared to be composed
almost entirely of hermaphrodites. It is important to note that unionoids
manifest their gender in two capacities: the gametes being produced in
the gonad determine the ‘visceral sex,’ and the differentiation of the gill
into a repository for glochidia the ‘marsupial sex.’ Bloomer found all
combinations of visceral and marsupial sex among English A. cygnea.
Most commonly, both eggs and sperm were made in the viscera, and the
marsupium was female. But the occasional individuals showing male vis-
ceral sex and female marsupial sex prompted Bloomer to hypothesize that
sex-reversal may occur in A. cygnea.
The situation is similar in several species of North American Anodonta.
But Heard (1975) recognized only two sorts of hermaphrodites: ‘female
hermaphrodites’ with ovarian tissue exceeding testicular tissue and gills
Reproduction · 41
modified into marsupia, and ‘male hermaphrodites’ lacking marsupia,
with a prevalence of testicular tissue. Two populations of A. imbecillis
comprised substantial fractions of both pure females and female her-
maphrodites, and Heard was able to confirm the simultaneous presence
of both egg and sperm in the gonads of the latter. A population of A.
couperiana contained 31% male hermaphrodites, 6% female hermaphro-
dites, and 63% females. Heard also reported about 10% male hermaph-
rodites in an otherwise bisexual population of A. peggyae. The complete
correlation between visceral and marsupial sex displayed by North
American Anodonta led Heard to question Bloomer’s hypothesis of sex-
reversal.
Heard (1979) characterized the phenomenon of sporadic hermaph-
roditism in nine species of North American Elliptio. Among 1178 indi-
viduals examined, Heard discovered 22 hermaphrodites. Most of these
data are shown in Table 2.2. There were 17 male hermaphrodites, 1
female hermaphrodite, and 4 individuals in which visceral sex and marsu-
pial sex were not in agreement. As had been reported by Bloomer, Heard
found two Elliptio with predominantly testicular gonads but gills differ-
entiated into marsupia, and two Elliptio with predominantly ovarian
gonads but undifferentiated gills. He did not notice any parasitism,
although Kat (1983) reported an association between digenean trema-
todes and sporadic hermaphroditism in Lampsilis radiata and Alasmidonta
undulata. None of Heard’s 22 animals contained both mature eggs and
sperm simultaneously. That observation, together with the rarity of her-
maphrodites in most unionoid populations and their occasional tendency
to display parasitic castration, suggested to Heard that the phenomenon
of sporadic hermaphroditism might be of minor importance. But in an
evolutionary sense, the capability for hermaphroditic reporduction is
unquestionably available for selection in a large fraction of unionoid pop-
ulations.
Mackie (1984a) suggested that protandric hermaphroditism might be
the ancestral condition in the molluscan Class Bivalvia. The unionoids
appear to be primitively dioecious, however, with predominant simulta-
neous hermaphroditism arising independently on several occasions
(Hoeh et al. 1995). Van der Schalie (1970) credits the early German
worker H. Weisensee with the first observation that hermaphroditism
seems more common in the mussel populations of lentic waters than in
lotic environments. Kat (1983) found that several especially high-density
Anodonta imbecillis populations from lentic environments tended to have
strikingly low numbers of males and/or lower proportions of testicular
42 · Bivalve autecology
tissues. Bauer’s (1987) results suggest to me that sexuality in the
Unionoidea may be more plastic than previously suspected. Might the
striking equality of sex ratios shown in Table 2.2 hide a non-genetic
component? As we shall see in Chapter 4, many attributes of the life
history of freshwater molluscs have a large environmental component;
perhaps the production of eggs, sperm, or both is no exception.
Figure 2.7. The top figure shows the persistence of Margaritifera margaritifera glo-
chidia (initial exposure 15000 glochidia) on the gills of two salmon species (Fustish
and Millemann, 1978). The bottom figure shows the impact of varying levels of glo-
chidial infection on the survivorship of the fish. (Data of Meyers and Millemann,
1977.)
Figure 2.8. The top figure shows number of fish hosts reported in Appendix A of
Watters (1994), for 96 unionoid species. The bottom figure shows number of fish
hosts listed in Appendix B of Watters (1994) combining species within the 7 most
commonly reported fish families. Shaded regions in the lower graph indicate fish
species hosting both amblemine and anodontine mussels. Margaritiferid hosts are
counted separately.
50 · Bivalve autecology
ship between glochidial morphology, or any other aspect of the biology
of these disparate groups of mussels, and known host range.
Unionoid host range may not be a function of the number of fish
species parasitized, however, but the number of higher taxa. Appendix B
of Watters (1994) lists all known unionoid parasites for 196 species of fish,
162 of which host amblemines, anodontines, and/or margaritiferids.
These 162 fish species represent 23 families. So within each family of fish,
I counted the number of fish species known to host amblemines but not
anodontines, known to host anodontines but not amblemines, known to
host both unionid subfamilies, and known to host margaritiferids (separ-
ately, regardless of other unionoid parasites). Only seven of the 23 fish
families had five species or more known to host unionid and/or marga-
ritiferid glochidia (Figure 2.8 bottom). A relationship between margarit-
iferids and salmonids is obvious. Ten salmonid species are known to host
margaritiferids. Three of these fishes are known also to host anodontines,
and one also hosts an amblemine species. No salmonid species is known
to host unionids but not margaritiferids. In addition to the margariti-
ferid/salmonid relationship, an obvious relationship was also evident
between mutelids and cichlid fish in Watters’ Appendix B, although these
data were not tabulated.
However, in spite of the differences in glochidial morphology between
amblemines and anodontines, and many other biological differences as
well, the lower half of Figure 2.8 reveals no difference in their known
fish hosts or host ranges. Both unionid subfamilies parasitize at least a few
species of all seven commonly reported fish families. Combining data
from the four less-represented fish families (Catostomatidae, Cottidae,
Ictaluridae, and Salmonidae) into a single category, the value of chi-
square from a test for independence was 2.35, not significant with three
degrees of freedom.
Predictions regarding the hosts or host ranges of unionid mussels seem
to be frustratingly difficult to make. For example, Weiss and Layzer
(1995) surveyed both the fishes and mussels inhabiting a five kilometre
length of the Barren River, Kentucky, at monthly intervals over the
course of a year. Inhabiting the river were at least 43 species of fish (15
families) and 27 species of unionids. The authors found glochidia on 25
species of fish, 14 of which had no previously documented unionid par-
asites, while 7 species of fish predicted to host Barren River mussels nev-
ertheless appeared uninfected. A key factor may be Weiss and Layzer’s
observation that only 4.1% of their 2510 fish were infected with glo-
chidia, and that most of these 4.1% carried only 1–5 glochidia each.
The selection pressure exerted by unionid glochidia on fish popula-
Reproduction · 51
tions may generally be negligible. The reason the other 2407 Barren
River fish were uninfected with glochidia is almost certainly that their
behaviours, habitat choices, and various aspects of their biology did not
bring them into contact with mussels, and/or that their non-specific
defences, such as tough integument or general hyperplastic, leukocytic,
and humoral responses, rejected all glochidia with equal efficiency. Most
host responses resulting in rejection of one glochidial species with appar-
ently increased efficiency are likely individual, and acquired.
From the parasite’s perspective, I suggest that whatever host specificity
may be displayed by unionoids may primarily be a function of physical,
ecological, and/or behavioural factors. Some mussel populations are
clearly adapted for lotic habitats and others for lentic, some release glo-
chidia in the spring and others in the autumn, and so tend to come into
contact with subsets of potential host fish. Some mussels seem to be
adapted to offer baits that may tempt some fish but not others.
Nevertheless, such laboratory experiments as those of Zale and Neves
(1982c) have demonstrated clear evidence of cellular or molecular adap-
tation by certain mussel parasites to specific fish hosts. Zale and Neves
reported that all 50 banded sculpins they exposed to the glochidia of
Villosa vanuxemi carried infections to metamorphosis, but that 30 sculpins
shed allV. nebulosa, M. conradicus, and L. fasciola glochidia with 100% effi-
ciency. One might predict that some sort of ecological or behavioural
association might also exist between sculpins and V. vanuxemi, such as a
temporal relationship involving glochidial release. We will return to Zale
and Neves’ mussel fauna in Chapter 4.
Watters (1997) has suggested that, to the extent that mussels become
behaviourally and/or ecologically associated with particular subsets of
hosts, cellular and molecular specializations may follow to counter special
subsets of defences. If a mussel population has evolved a special lure that
is only attractive to large basses, we might also predict special cellular and
molecular adaptations to counter the basses’ defences after glochidial
attachment. An illustration of the Watters hypothesis may be provided by
Leptodea leptodon and Potamilus capax, whose glochidia seem to success-
fully transform only on drum (of 24 and 29 potential fish hosts tested,
Barnhart 1999). Both of these mussels have unusually large thin shells,
and it has been speculated that glochidial infection occurs when drum
crush gravid parents upon ingestion.
Dreissena
Dreissena polymorpha is gonochoristic, with sexes generally found in
approximately equal proportions (Stanczykowska 1977). Antheunisse
52 · Bivalve autecology
(1963) reported 56% female, 40% male, and 4% hermaphrodite in a river
near Amsterdam. Walz’s (1978b) sample of 306 adult Dreissena from Lake
Constance also included 56% female, with the remainder male, and no
hermaphroditism in evidence. A ratio of 3 female to 2 male was reported
by Mackie (1993) in North American Lake St. Clair. Genetic polymor-
phism is high, and although some heterozygote deficiency is occasionally
reported, Hardy–Weinberg equilibrium generally prevails (Marsden et al.
1996).
Gametogenesis in Lake Constance begins in August and proceeds
through the winter, terminating with May spawning. Although the
Dreissena population introduced into Lake St. Clair between Michigan
and Ohio also spawns in May, gametogenesis is not initiated until late
winter or early spring. Some German populations spawn in mid-May and
August, while others show a single summer spawn, dependent on tem-
perature (Borcherding 1991). Males and females spawn into the water
column, the ova containing species-specific chemoattractants (Ram et al.
1996). Fertilized eggs develop into planktonic veliger larvae, which may
settle in as few as five days or as many as five weeks, with about two weeks
of planktonic life commonly observed (Sprung 1993). A chemical cue
may be released by adults to trigger larval settlement (Chase and Bailey
1996).
Corbiculoids
Most corbiculoids are simultaneously hermaphroditic, with gonads
divided into ovarian and testicular regions. In some groups (Pisidium,
Corbicula) the gonad shows medio-lateral specialization, while in other
groups (Sphaerium) the specialization is from anterior to posterior. Testes
mature before ovaries in pisidiids and in some Asian populations of
Corbicula, but the reverse is true in the best-studied North American pop-
ulations of C. fluminea. In any case, after maturity at least some oogene-
sis seems to occur year round, with seasonal peaks. Spermatogenesis
seems to be more periodic, especially in North American C. fluminea.
Sperm morulae dissociate into biflagellate sperm of poor mobility
(Kraemer and Galloway 1986, Kraemer et al. 1986). Gametes are
exported from the gonad through common gonoducts, showing no
apparent specialization to separate egg from sperm. Araujo and Ramos
(1997) observed self-fertilization within the ovary itself in a Spanish pop-
ulation of Pisidium amnicum.
Self-fertilization is generally believed to be widespread in the
Corbiculacea. The Swedish biologist N. Odhner seems to have been the
Reproduction · 53
first to demonstrate asexual reproduction in the group. His report that
Pisidium conventus could reproduce in isolation was expanded by Thomas
(1959), who reared six generations of Musculium partumeium isolating all
individuals at birth. Protein electrophoretic surveys have uncovered low
levels of polymorphism and highly significant heterozygote deficiencies
at almost all variable loci examined in wild populations of Sphaerium
(Hornbach et al. 1980), Musculium partumeium (McLeod et al. 1981) and
Corbicula (Hillis and Patton 1982, McLeod 1986, Kijviriya et al. 1991).
The levels of heterozygosity at polymorphic loci are never zero,
however, suggesting that some outcrossing may at least occasionally take
place.
I am unaware of any description of spawning in either Pisidium or
Sphaerium. King and colleagues (1986) were able to induce sperm release
in North American Corbicula by a combination of thermal, mechanical,
and salinity shock. Kraemer and her colleagues have also reported mid-
summer observations of sperm being extruded between individual
Corbicula on mucous strands. On the other hand, developing embryos are
periodically observed within Corbicula gonads, suggesting intrafollicular
fertilization. Such embryos appear to be successfully transported from the
gonad via the ‘gonoducts’. It has been suggested that North American
Corbicula may produce both an outcrossed spring generation and a self-
fertilized autumn generation (Kraemer and Galloway 1986, Kraemer et
al. 1986).
In North American C. fluminea, the water tubes of the inner pair of
gills are swollen and modified into marsupia for the brooding of devel-
oping embryos. All the ontogenetic stages one normally expects to
observe in molluscs are demonstrated within the marsupium. Blastula,
gastrula, trochophore, veliger and pediveliger stages are typically passed
in 6–12 days, and early straight-hinged juveniles shed to the environ-
ment. Kraemer and Galloway noted that stressed animals may release
trochophore and veliger larvae, which do not seem especially viable
outside their mother’s marsupium. Mothers may also retain their off-
spring for over two months, should the environment dictate.
A wide range of sexuality is displayed by the Corbicula populations of
Asia. The Chinese population identified by Morton (1982) as C. cf. flu-
minalis was dioecious, with 4.5% hermaphrodites. Sex ratios in the dio-
ecious fraction varied, however, from about 60% female among the
younger individuals to 60% male in samples of older individuals. This
particular population did not seem to brood its young, but rather
spawned gametes openly into the water column. Asian populations of
54 · Bivalve autecology
C. fluminea may also contain dioecious individuals, although the situation
here has been confused by inadequate sampling.
Smith and his colleagues (1979) reported high levels of genetic vari-
ability at allozyme loci in one Japanese Corbicula population of unknown
species. I might speculate that this population may be related to the dio-
ecious C. sandai, apparently reproducing sexually in Lake Biwa (Komaru
et al. 1997). In contrast, Komaru and his colleagues reported that Japanese
‘Corbicula aff. fluminea’, Japanese C. leana, and Taiwanese C. fluminea were
all hermaphroditic, producing ‘non-reductional’ gametes identical in
DNA content to somatic cells. They hypothesized that sperm may be
necessary to initiate development in these species, but that reproduction
might be effectively parthenogenetic.
In pisidiids the semi-enclosed spaces between the lamellae of the inner
gill demibranchs serve as the marsupia. Fertilized eggs pass into these
spaces, become enclosed by gill tissues, and form sacs in the marsupial
wall. The typical ontogenetic stages demonstrated by Corbicula seem to
be suppressed in pisidiids. Workers find it convenient to recognize
primary, secondary, and tertiary brood sacs, containing embryos, foetal
larvae, and prodissoconch larvae, respectively. Embryonic mortality (or
‘intramarsupial suppression’) is often substantial, and it has been suggested
that some developing larvae may be sacrificed to yield space or perhaps
even direct nourishment for others (Meier-Brook 1977). Young released
from tertiary sacs remain in the marsupium for some period, either free
or attached by a byssal stalk, as ‘extramarsupial larvae’. Extramarsupial
larvae may grow quite large, relatively speaking: typically 20–25% of the
size of their parents and sometimes 50%. Heard (1977) confirmed early
reports of J. Groenewegen and M. Thiel that extramarsupial larvae of
many Sphaerium species have in fact begun active (‘precocious’) gameto-
genesis. It seems odd to apply the term ‘larvum’ to an individual much
beyond its sexual maturity. The mother ultimately releases her young via
the excurrent siphon.
Pisidium develop no more than one brood sac per demibranch at any
time, all embryos enclosed are substantially the same size and age, and
all are released simultaneously. The elapsed time from fertilization to
birth may be no more than one month. Adult Sphaerium and Musculium
commonly carry multiple brood sacs through much of the year, in
various stages of development, and the young within each sac may be of
different sizes. Newborn Sphaerium and Musculium are typically released
a few at a time, even in populations considered semelparous. The total
time elapsed from fertilization to release in Sphaerium and Musculium is
Summary · 55
typically several months, although exceptions have been noted (Morton
1985).
Total lifetime fecundities of most pisidiids are spectacularly low by
comparison with other molluscs; e.g. 12 offspring over two years for
Sphaerium simile (Avolizi 1976), and 10 over a single year for S. striatinum
(Hornbach et al. 1982). The number, size, and developmental stage of
offspring brooded by the various populations and species of corbicula-
ceans has been the object of considerable interest, as has been the overall
timing of their reproductive cycles. These topics will be pursued in
Chapter 4.
Summary
The unionoids and Dreissena are gonochoristic, with sexes generally
found in equal proportions. Yet most populations show some frequency
of hermaphroditism, and several species are predominantly hermaphro-
ditic. The unionoid populations of ponds and other possibly temporary
habitats seem more likely to display hermaphroditism. The likelihood
that sperm finds its way into the siphon of a mature female unionoid may
be extrordinarily low, and some evidence suggests that sexuality may
respond plastically to fertilization failure.
Dreissena produces planktonic larvae unique among the freshwater
bivalves. The life history of the Unionoidea includes a remarkable dis-
persal stage, the glochidium, adapted to parasitize fish. Some specific
associations have evolved between particular unionoid populations and
populations of co-occurring fishes, while in other cases, little host
specificity has been noted. Margaritiferid mussels generally seem to par-
asitize salmonid fishes. But no similar relationship seems to have evolved
between any taxon of unionids and any group of fish, nor is any relation-
ship apparent between unionid glochidial morphology and host range.
Close host–parasite associations may only arise in exceptional cases,
where cellular and molecular adaptations evolve as a consequence of an
initial physical or ecological relationship between fish and mussel.
The corbiculoids, which include many taxa famously adapted to tem-
porary habitats, are almost all hermaphroditic, and apparently self-fertil-
izing. Williams (1975) pointed out that across all kingdoms of life, asexual
reproduction is often associated with colonizing ability. It would seem
that bivalves are no exception. Corbiculoids are brooders; their larval
stage is passed internally, and offspring (sometimes quite large relative to
their mothers) are released directly to the environment. The connection
56 · Bivalve autecology
between asexual reproduction, ovoviviparity, and polyploidy will be
explored in chapters to follow.
The recurring theme of this chapter has been overlap, not just of
unionid fish host, but of bivalve diet and habitat generally. In lentic en-
vironments, the ideal habitat for almost all bivalves seems to be as close
to the surface as predation and disturbance will allow, perhaps 0.5–2.0 m
deep. One might speculate that the differences in the life forms of the
broad, light Anodonta, the narrow, hatchet-shaped Elliptio, and the epi-
benthic Dreissena have resulted from adaptation to different substrates in
these same, shallow waters. Species of Pisidium may select sediments of
different grain sizes, and possibly different depths in the sediment. A few
Pisidium species seem to have become physiologically adapted to deeper
water, apparently exchanging reduced competition for reduced lifetime
reproduction. But for most lentic bivalve populations, adapted as they are
for the shallows, both intra- and interspecific competition for space seems
a real possibility.
Lotic waters typically offer a relatively greater proportion of suitable
habitat for bivalves and, perhaps as a consequence, bivalve diversity is
increased. There is little evidence of divergence among the species in
microhabitat preference, however. Most species find a gravel substrate in
a region of moderate current to be superior. So to the extent that the size
of individual freshwater bivalve populations generally escapes the vagar-
ies of the physical environment, the potential for space competition would
seem substantial. This theme will be developed in Chapters 5 and 9.
3 · Gastropod autecology
Here we broadly review the diet, habitat, and reproduction of the fresh-
water gastropods, starting with general observations on their feeding
mechanism. Snails are able to exert more control over their immediate
environment and the materials entering their mouths than bivalves, and
hence there is greater opportunity for ecological specialization. The diets
and habitats of 13 freshwater gastropod families are discussed separately,
compared and contrasted. Members of the eight prosobranch families
tend to grow more slowly than those of the five pulmonate families, and
to be found more commonly in lotic environments. Substrate preferences
are striking and widespread. Among the groups we catalogue special
adaptations and dietary preferences for suspended particles, detritus, bac-
teria, algal filaments, diatoms, macrophytes, and even carrion.
Differential gastropod grazing has been demonstrated to impact algal and
macrophyte community structure, as well as the macroinvertebrate com-
munity as a whole. Yet against the numerous striking examples of eco-
logical specialization in freshwater gastropods, there is a general
uniformity of body plan and habit that suggests the potential for inter-
specific, as well as intraspecific competition.
We will review a large literature on the complexities of pulmonate
reproduction. Adults are generally simultaneous hermaphrodites,
although they cannot mate reciprocally at the same time, a situation that
may lead to mating conflicts of at least two sorts. Pulmonates can also
self-fertilize, although usually at a sacrifice of fitness to both parent and
offspring. Long-term sperm storage, multiple insemination, ‘sperm
sharing’, mixed mating and aphally all contribute to the diverse repro-
ductive strategies available to the pulmonate snails. By contrast, most
prosobranchs are gonochoristic, with reproductive behaviours inconspic-
uous and/or poorly documented. Sometimes sex ratios are balanced, but
often they are not, and dimorphism in growth or survivorship often does
58 · Gastropod autecology
not seem to account for the bias. The actual mechanism of sex determi-
nation in the diverse prosobranch taxa is called into question.
Parthenogenesis has arisen three times in the freshwater prosobranchs. In
all three cases females produce apomictic ova, which are brooded and
released viviparously. Males may be present (although rare) and sexual
reproduction occurs at least occasionally in many parthenogenic popula-
tions. The valvatids are hermaphrodites, completing the general picture
of reproductive diversity we will survey in the pages that follow.
Planorbidae
More is known of the medically important planorbids than any other
freshwater mollusc taxon. The large literature on factors affecting planor-
bid distribution has been reviewed on several occasions (Malek 1958,
Appleton 1978, Brown 1994). The following quotations, patched
together from Malek, serve as a fair summary: ‘The type of substrata
almost always associated with bilharziasis intermediate hosts is a firm mud
bottom, rich in decaying matter. The current should not be strong. Water
plants . . . affect the number of snails, but are not essential for their occur-
rence.’
Helisoma anceps is rather unusual among the planorbids in that it seems
to occur more often in river pools than in lakes or ponds. Cummins and
62 · Gastropod autecology
Lauff (1969) have reported an unusually complete investigation of sub-
strate particle size selection in 10 macroinvertebrates inhabiting a small,
shallow Michigan stream, among which they included H. anceps. The
authors combined a quantitative field survey with artificial stream experi-
ments on substrate choice over eight particle sizes (fine sand to pebble).
Although strong particle-size selectivity was demonstrated in a number
of aquatic insects, evidence was much less striking in the case of Helisoma.
It is a shame that no investigation as thorough as that of Cummins and
Lauff seems to have been completed for the more typical, lentic-water
planorbids.
Liang (1974) has reviewed, I would estimate, approximately 100 pub-
lished accounts of pulmonate diet in culture, the great majority involv-
ing medically important planorbids. Although most workers have used
lettuce or fish foods of various complex formulation, Liang’s review
includes ten papers reporting at least some evidence that an algal diet may
contribute to the growth and fecundity of planorbids. For his experi-
ments, Liang cultured both Biomphalaria and Bulinus on pure diets of
blue-green algae (Nostoc or Fischerella), innoculated into a mud substrate.
In a more recent study, Jantataeme and colleagues (1983) found that
Indoplanorbis exustus fed lettuce plus Nostoc, or lettuce plus the diatom
Achnanthes, showed improved fecundity over cultures fed any of the three
dietary items separately.
The only demonstrations that planorbids can be cultured on aquatic
macrophytes appearing in Liang’s (1974) review were six applications of
watercress (Nasturtium, e.g. Chernin and Michelson 1957a,b). But more
recently Thomas and colleagues (1983) have studied Biomphalaria feeding
on duckweed (Lemna paucicostata) and two pondweeds (Groenlandia densa
and Potamogeton crispus). Not surprisingly, assimilation rates for these
macrophytes, calculated as the difference between dry weight consumed
and dry weight of faeces over 24 hours, were all much lower than the
assimilation rate for lettuce. But Thomas and colleagues did demonstrate
short-term growth in Biomphalaria fed on these macrophytes.
Thomas (1987) expanded this research programme to include 33
genera of macrophytes, representing 23 higher taxa. Snails were fed two
1.7 cm discs (or the equivalent) of each of these plants daily and weighed
after four days (experiment A). Clearly, more macrophyte tissue is typi-
cally available for consumption in the wild than was supplied to these
snails. Nevertheless, snail growth was significant on most of the diets,
even surpassing lettuce in the case of Groenlandia. Example data are
shown in Figure 3.1. Thomas offered other snails homogenized plant
tissues in a calcium alginate gel, again measuring four-day growth
Pulmonate diet and habitat · 63
Figure 3.1. Four-day growth rates of Biomphalaria glabrata fed lettuce and samples
from 10 aquatic macrophytes. Snails were offered leaf disks (Exp. A) or calcium algi-
nate homogenates (Exp. B). Growth rate is expressed as (100⫻) the difference
between ending weight and starting weight, as a proportion of (4⫻) starting weight.
(Data of Thomas 1987.)
Physidae
Although unimportant from a medical standpoint, snails of this family are
very common and widespread in the parts of the world most frequented
by freshwater biologists. Thus there is little information on physid
culture, but a fairly rich literature on their diet and habitat in the wild.
Crowl and Schnell (1990) reported that Oklahoma populations of
Physa virgata reached maximum size and abundance in the reaches of
streams with reduced flow and fine substrate. Bickel (1965) accomplished
the difficult trick of sampling quantitatively (or at least semi-quantita-
tively) three different substrates over one year in the Ohio River near
Louisville, Kentucky: mud bottom, macrophytes, and wood/dock. Physa
integra, by far the most common snail, was found almost entirely on plants
and other solid substrate in the summer, strongly preferring Potamogeton
pectinatus. Snails may have been excluded from the bottom by low dis-
solved oxygen. As the macrophytes died in the autumn and winter (and
levels of dissolved oxygen increased), Physa appeared on the mud. But
Bickel suggested that spring floods would tend to carry such individuals
away, so that the population may be maintained by individuals overwin-
tering on docks and submerged wood.
The fondness of P. integra for solid substrate was confirmed by
Clampitt (1970) in an interesting comparative study of several adjoining
Pulmonate diet and habitat · 67
north Iowa lakes and ponds. Clampitt surveyed the region in a thorough
but non-quantitative fashion. He observed that P. integra was found in
lakes on solid substrates, either on rocks nearshore or on offshore vege-
tation. Physa gyrina was never found offshore, in contrast, but was
common in the shallows on mud and sand bottom as well as rock. Unlike
P. integra, P. gyrina was also often found in nearby shallow ponds. Aplexa
hypnorum is similar in some respects to P. gyrina in that it commonly
inhabits shallow ponds and ditches subject to seasonal drying. Den
Hertog (1963) surveyed 29 Dutch populations of A. hypnorum, finding a
striking positive relationship with certain sandy- clay soil types. The
adaptation of populations of such species as P. gyrina and A. hypnorum to
marginal and transient habitats will comprise a major theme of discussion
in Chapter 4.
Clampitt (1970) qualitatively analysed the gut contents of individuals
from both P. integra and P. gyrina populations collected on rocky shores
where they co-occurred. No interspecific differences were apparent.
Detritus was always the most common item in the guts of Physa, followed
by algae of all sorts, and some sort of gelatinous material (mucus?), with
animal and vascular plant remains of minor importance. The only strik-
ing differences were between lakes, not between species: a preponder-
ance of diatoms in the gut of snails from one lake, colonial blue-green
algae from another. Among the animal remains were rotifers, ostracods,
amphipods, dipteran larvae and the chaetae of oligochaetes.
Nor was any difference apparent between P. gyrina and P. integra in
food-choice experiments. Clampitt found that both species accumulated
on algae-covered stones or dead Carex stems rather than on clean stone
controls, although there was no apparent preference between algae and
dead plant material. (Clampitt reared his laboratory populations on a
mixed diet of lettuce and dead maple leaves.) Neither Physa species was
attracted to living Ceratophyllum demersum or Myriophyllum exalbescens,
two common offshore macrophytes upon which Physa seemed to accu-
mulate. This agrees well with the rarity of vascular plant remains in the
guts of wild-caught snails.
Clampitt also performed several laboratory studies of comparative
physiology. He concluded that the difference in the habitats of the two
species is due to a higher temperature tolerance and drought tolerance of
P. gyrina, which allow it to colonize small ponds, and a lower metabolic
rate of P. integra, which allows it to survive in deeper water without
surface breathing. I would speculate that their dietary overlap makes
competition quite likely on those occasions where they co-occur.
68 · Gastropod autecology
Although not typically found in lotic environments, Physa gyrina was
the primary grazer in a set of laboratory streams established by Kehde and
Wilhm (1972). Only six genera of algae developed in the streams, includ-
ing greens and blue-greens but apparently no diatoms, possibly due to
artificially high temperatures. In any case, Kehde and Wilhm found that
Physa reduced the standing crop of algae, without noticably affecting pro-
ductivity or diversity. Patrick (1970) mentioned a study showing that
Physa heterostropha will feed on most species of diatoms, seeming to have
difficulty only with the tightly cemented Cocconeis. Physa sp. was men-
tioned by Hambrook and Sheath (1987) in their list of freshwater grazers
known to ingest red algae. Coffman (1971) examined the guts of 75
macroinvertebrates inhabiting a small Pennsylvania woodland stream,
recognizing three categories of contents: animal remains, detritus, and
algae. Expressed in calories (an unusual approach that has much to rec-
ommend it), ‘Physa sp.’ was a herbivore, with algae comprising about 75%
of its gut content, even in the autumn, when detritus would be expected
to be plentiful.
Carman and Guckert (1994) made an effort to assess the relative assim-
ilation efficiencies displayed by Physa virgata grazing on microbiological
and algal components of the periphyton. Their study was complicated by
the fact that one of their radiotracers (3H-amino acids) tended to adsorb
abiotically. Carmen and Guckert calculated that Physa assimilated
adsorbed organic carbon (such as their label) at 78% efficiency, bacterial
carbon at 50% efficiency, and algal carbon at only 32% efficiency. This
suggests substantial adaptation for detritivory in Physa.
Lowe and Hunter (1988) allowed algae to colonize glass microscope
slides held in screened enclosures for 26 days. They then introduced Physa
integra for 21 days at either two or ten snails per enclosure, holding one
set of enclosures as controls. Physa reduced overall biovolume from a
mean of 3.4⫻104 m3/mm2 in the control to 1.8⫻103 at low density
and 9.5⫻102 at high density. Physa also caused a significant reduction in
diversity (Figure 3.2). Declines in relative abundance were suffered by six
taxa of diatoms, six taxa of green algae, three taxa of blue-greens, and
three taxa of golden-browns. The relative increases came to the green
Oedogonium, which Lowe and Hunter attribute to a tightly adhering
holdfast, and the prostrate diatom Eunotia. Swamikannu and Hoagland
(1989) reported a similar effect from high grazing pressure by Physa, but
found evidence that low-level grazing increases periphyton diversity.
Among the most remarkable studies of macrophytes and freshwater
snails published to date has been that of Sheldon (1987) on Physa gyrina
Pulmonate diet and habitat · 69
Figure 3.2. Relative abundances of selected algal taxa on glass slides held in enclo-
sures with ten Physa (‘High Grazer’) and no snails (‘Control’). Algal biovolume on
grazed slides was only 3% of that on the control. (Data of Lowe and Hunter 1988.)
Lymnaeidae
The richness of the literature on lymnaeid culture is due both to the
medical and veterinary importance of lymnaeids and to an unexplainable
Pulmonate diet and habitat · 71
interest of basic researchers in the biology of unimportant species.
Noland and Carriker (1946) list 40 studies of lymnaeids involving at least
some culture work, starting in 1854. Among the successful diets have
been dried leaves, algae, and the higher plants Chara, Elodea, Lemna, and
Cabomba. More recent workers with the Fasciola host Lymnaea natalensis
have raised snails in shallow aquaria with mud bottoms, into which cul-
tures of Oscillatoria have been innoculated (Kendall and Parfitt 1965). The
algae serve as food for younger snails, until they have grown to a size at
which an artificial diet can be substituted. An algal diet has also been
found suitable for the culture of L. truncatula under similar conditions
(Taylor and Mozley 1948, Kendall 1953). Boray (1964a,b) reported that
cultures of lymnaeids, in this case Lymnaea tormentosa (the host of Fasciola
hepatica in Australia), can grow and reproduce fed only filter paper or cel-
lulose powder. Boray also reported snail growth on diets of softer plants,
such as the watercress (Nasturtium) and duckweed (Lemna), but felt that
most macrophytes presented cell walls too difficult for the snail to pene-
trate.
Among the most complete culture studies of freshwater snails of which
I am aware is the work of Skoog (1978) on L. peregra. The author mon-
itored 14-day shell growth on 24 juveniles fed four different diets, and
40-day growth and egg production on 10 individuals fed five diets. Skoog
found diatoms to be best for juvenile growth, followed by mixed blue-
green algae (dominated by Calothrix), then the green alga Cladophora. But
mixed blue-green algae were significantly better than any other diet
(diatoms, spinach, Rivularia, or Cladophora) for promoting 40-day growth
in adult L. peregra. And the egg production promoted by spinach was sig-
nificantly better than all others.
Bronmark (1985) presented laboratory evidence that grazing by L.
pereger may stimulate growth in Ceratophyllum. He placed terminal,
unbranched portions of the plant in 28 aquaria, half of which received
two L. pereger and two Planorbis planorbis. After nine weeks the
Ceratophyllum in the aquaria with snails had grown significantly longer
than controls. Bronmark attributed this phenomenon to the removal of
epiphytes by the snails, reducing shading and competition for nutrients.
Olfactometer tests showed that L. pereger was significantly attracted to
Ceratophyllum but could not distinguish grazed and ungrazed plants, and
that epiphytes had no attraction by themselves. Bronmark made the sur-
prising suggestion that macrophytes may excrete chemical cues as an
adaptation to attract snails. But we have noted that Ceratophyllum is
among the most unpalatable of all macrophytes. Lymnaea pereger can in
72 · Gastropod autecology
fact grow and reproduce quite well on a diet of other sorts of macrophy-
tes exclusively. For his study of the importance of mineral particles in L.
pereger guts, Storey (1970) maintained healthy cultures over periods of
months on a variety of rations. He found good long-term growth on the
starwort Callitriche, unspecified coccoid algae, and even filter paper.
Surprisingly, growth was not as good on a diet of the filamentous alga
Cladophora. Storey noted some growth on a diet of bark and rotting
leaves.
Before turning to more field-oriented studies, some laboratory evi-
dence of bacterial feeding in lymnaeids should be noted. Underwood and
Thomas (1990) placed growing shoots of previously grazed Ceratophyllum
in beakers with either four Planorbis planorbis or four L. pereger (a total of
eight replications) for 35 days. (It may be recalled that Calow, 1974, found
excellent evidence of bacterial feeding in P. contortus, a snail quite similar
to P. planorbis.) Underwood and Thomas sampled plant tissues at the
beginning, middle, and end of the experiment, applied a stain, and exam-
ined for bacteria and algae. They observed that L. pereger made as large
an impact on attached bacteria as did Planorbis. All of this effect could be
due to mechanical disturbance, and the authors emphasized that the
standing crops of epiphytic bacteria are low and unlikely to constitute a
significant food source. Nevertheless, the possibility of bacterial feeding
in Lymnaea is certainly not ruled out.
Turning now to observations in the field, Bovbjerg (1965, 1968, 1975)
noted that the natural diets of four local species, Lymnaea stagnalis,
Stagnicola exilis, S. elodes, and S. reflexa, most visibly included a variety of
filamentous algae and macrophytes (for example, Utricularia, Ranunculus,
Lemna, and Potamogeton). Gut contents contained the usual collection of
‘organic debris’, ‘substratum materials’, vascular plant materials, algal
strands, blue-green cells, and miscellaneous small animal fragments. He
placed 30 S. reflexa in a tank with alternating patches of the filamentous
green Spirogyra and any one of several macrophytes, and found after 12
hours that the snails were always more common on the algae. Bovbjerg
recorded an average of 80% of the snails on algae over the despised
Ceratophyllum, 77% on algae over Myriophyllum, and 70% on algae over
Ranunculus. But all other dietary preferences paled when compared with
carrion. Prompted by an observation of several snails feeding on a dead
insect, Bovbjerg performed a series of chemoreception experiments in a
large Y-shaped chamber, 30 cm⫹2(15) cm long. At these distances, none
of the lymnaeids showed any significant ability to orient toward chopped
Potamogeton, but 76–86% of the snails could find chopped crayfish.
Pulmonate diet and habitat · 73
Calow (1970) failed to find any higher plant material in the crop con-
tents of 350 L. pereger collected from Elodea. Rather, he identified about
70% diatoms, 25% green filamentous algae, and 5% other algae, an almost
exact match to the epiphytic flora of Elodea. Choice experiments showed
that L. pereger was highly and equally attracted to Elodea extract and fila-
mentous green algae, followed by normal Elodea (with attached epiphy-
tes), then diatoms, then clean-scrubbed Elodea. (This result does not
match that of Bronmark especially well.) The attraction of the Elodea
extract may seem unexpected, but it may materialize that crushed vascu-
lar plant material is extremely strong but mildly attractive, while such
material as diatom cells, although highly attractive, may emit almost no
chemical cue.
The experiments of Lowe and Hunter (1988) on grazing by Physa
were in some measure patterned on earlier experiments by Hunter (1980)
involving Lymnaea elodes. The 1980 work was performed in a shallow
Michigan lake where about 82% of the snails were L. elodes, the remain-
der being Physa gyrina and Helisoma trivolvis. Hunter suspended a series of
enclosed slide racks in the lake, half of the enclosures without snails and
half with L. elodes at natural proportions and densities. (There was no
initial period for periphyton colonization as in the 1988 work.) Results
were very similar, however: the snails reduced algal standing crop, pro-
ductivity, and diversity, removing all filamentous algae and fungal hyphae,
and nearly all diatoms, especially the larger ones. Remaining but in
reduced numbers were smaller diatoms such as Gomphonema, and again,
apparently unaffected by grazing was the adpressed diatom Cocconeis.
Cuker (1983a) also examined grazing in a population of L. elodes, this
inhabiting a lake in arctic Alaska. He placed rocks covered with algae in
10 dark plastic tubs of lake water, maintaining all at lake temperature with
a continuous bath. Cuker introduced snails at nine densities (keeping an
ungrazed control) and sampled epilithic algae periodically for the month
of July. Once again algal standing crop and primary productivity were
severely reduced by the snails. In this particular algal flora, the diatoms
and filaments were reduced by the grazing while the absolute densities of
coccoid green and blue-green algae, said to have mucus sheaths, were
unaffected. Using a radiolabelling technique, Cuker estimated that the
snails actually assimilated 68% of algal carbon grazed, incorporating 25%
into the tissues and respiring 43%.
Cuker (1983b) performed a second set of experiments to assess the
impact of L. elodes on other grazers in his arctic lake, especially chirono-
mids and microcrustaceans. Algal-coated rocks and snails at several
74 · Gastropod autecology
densities were again placed in chambers, these suspended in the lake for
one month. Tube-dwelling chironomids were reduced an order of mag-
nitude at the highest snail densities, free-living scrapers reduced 85%, and
predatory chironomids by 66%. Part of this effect may be due to mechan-
ical disturbance by the snails, but competition for periphyton must also
play a role. Among the other organisms, ostracods and cladocerans were
reduced by high snail densities, while copepods, trichopterans, and mites
were unaffected. This is the first occasion we have mentioned that the
guild to which freshwater snails belong may not be exclusive. Others will
follow in Chapter 9.
Knecht and Walter (1977) and Walter (1980) performed a thorough
comparison of diet and habitat in two Lake Zurich lymnaeid populations,
L. auricularia and L. pereger. Walter took 256 samples of 1 m2 from the
bottom at depths of 3–9 m using and a suction device. If food
was abundant, the snails tended to aggregate on stones, while if food was
less abundant, they aggregated where the organic content of the mud was
greatest. Walter’s impression was that the total density of snails was more
constant than the densities of the two species considered separately,
implying that they replaced each other. The only minor difference
between the species was that L. pereger was more common on stones.
Knecht and Walter analysed gut contents of individuals from the two
species collected from submerged structures, comparing them with each
other and with substrate samples. They detected no significant difference
between the diets of L. auricularia and L. pereger. By volume, the 20 most
common items identified from the guts (amounting to 97% combined)
included 47% green algae (8 taxa), 4% diatoms (5 taxa), 3% blue-green
algae (2 taxa), and a remarkable 43% protozoans (4 taxa of flagellates, 1
ciliate). There were significant differences between the gut contents of
both snails and the substrate upon which they grazed. But the absence of
any material differences in the diet or habitat of these two snail species
suggested to the authors that food was not limiting and that population
sizes were perhaps being regulated by fish predation.
We now turn from the lymnaeids of deep lakes to those spending the
greatest portion of their lives in the open air, on mud. Two such amphi-
bious lymnaeids of veterinary importance are common in New Zealand
marshes, the native Lymnaea tormentosa and the North American Lymnaea
(Pseudosuccinea) columella. Harris and Charleston (1977) took 489 quadrat
samples from sites where one or the other snail occurred. (Interestingly,
they do not seem to occur together.) They counted snails and collected
a variety of environmental data, including proportion of substratum
Pulmonate diet and habitat · 75
Figure 3.3. Path diagrams showing the hypothesized effects of quadrat proportion
under water, quadrat proportion exposed mud, and flocculence on population
density of Lymnaea tormentosa and L. columella in New Zealand. (Harris and
Charleston 1977.)
Figure 3.4. Algal contents of 30 individual Ancylus fluviatilis guts, compared with
the Lake Windermere substrate from which they were collected. (Data of Geldiay
1956.)
‘indiscriminately on the algal felt that covers the rocks and stones where
they live’, although to my eye, some of the differences look significant.
The North American ancylid Ferrissia rivularis, very similar ecologi-
cally to A. fluviatilis, was among the most herbivorous of all 75 macroin-
vertebrates in Coffman’s (1971) Pennsylvania stream study. Less than 10%
of the calories in its gut contents were due to detritus or animal material.
Streit (1975) found that A. fluviatilis assimilated the diatom Nitzschia and
the green Scenedesmus equally well, on the order of 40–60% efficiency.
Interestingly, Streit found that A. fluviatilis could not assimilate the blue-
green Anabaena, a result that dovetails well with the findings of Calow
(1973b).
One of the nice aspects of Calow’s (1973b) work on the diet of
Planorbis contortus (previously discussed) was that he simultaneously exam-
ined A. fluviatilis using similar techniques. For example, as Calow
observed that P. contortus was limited to stone surfaces with detritus but
without algae, he noticed that A. fluviatilis occurred only ‘on the stone
sides where there is an abundance of epilithic algae’. Algae were by far
the most common items in the gut content of Malham Tarn limpets, fol-
lowed by detritus, with fungi a distant third. In the same chambers where
P. contortus aggregated on detritus-covered filter paper and ignored algae,
Pulmonate reproduction · 79
A. fluviatilis aggregated on algae and ignored detritus. Previous work by
another author (and his personal observations) led Calow to offer the
limpet a choice between algae and the lichen Verrucaria. Ancylus did eat
the lichen, but still preferred algae.
In other studies, Calow (1973a) concentrated only upon the algal
component of the diet of A. fluviatilis. He compared the gut contents of
limpets from Malham Tarn with the algal flora of grazed stones, finding
broad agreement while noting some evidence of selection for diatoms
and against blue-green algae. He then performed a series of food choice
tests, discovering significant differences in every comparison. Ancylus
prefers mixed diets of diatoms over all other offerings. The next most
preferred is the filamentous green algae (Cladophora), then unicellular
greens (Scenedesmus), and finally the blue-greens (Rivularia). Among the
diatoms, Calow found a significant preference for Gomphonema over
Navicula and Achnanthes. One might speculate that such strong preference
among the herbivores might affect the algal flora of Malham Tarn. But
Calow found little evidence of this using the (admittedly weak) method
of comparing stones with snails with stones without snails. Among the
complicating variables here are the population size of A. fluviatilis, which
may not be large enough to produce an effect on the Malham Tarn flora,
and Calow’s observation that the limpets generally manifest dietary pref-
erences best when satiated, a rare condition in the wild.
Pulmonate reproduction
Outcrossing
The freshwater pulmonates are hermaphroditic, both sperm and egg
being manufactured in a single regionally differentiated gonad, the ovo-
testis. Spermatogenesis seems to precede oogenesis by at least a few weeks
in Physa fontinalis (Duncan 1959), Lymnaea stagnalis (Duncan 1975),
Laevapex fuscus (Russell-Hunter and McMahon 1976), Bulinus globosus
(Rudolph 1983) and Biomphalaria glabrata (Trigwell and Dussart 1998). In
contrast, Richards (1962) reported that female fertility precedes male fer-
tility in Gyraulus parvus. Judging by reproductive function (rather than
histological section), Wethington and Dillon (1993) reported that male
maturity preceded female maturity in 46 of 50 Physa heterostropha, female
preceded male in one case, and three individuals matured in both func-
tions during the same seven-day period.
The genital systems of the several families have been described,
80 · Gastropod autecology
compared and figured by Duncan (1960, 1975), Geraerts and Joosse
(1984), and Visser (1981, 1988). Eggs and sperm are passed together
through a narrow ‘hermaphrodite duct’, with a ‘seminal vesicle’ region,
where the sperm may mature, be stored, or ultimately degraded
(Hodgson 1992). Egg and sperm then enter a rather elaborate region,
usually called the ‘carrefour’. Opening into the carrefour are fertilization
pocket(s), an albumen gland, a female duct and a male duct. Most fresh-
water pulmonates rarely self-fertilize. They seem to have mechanisms that
favour fertilization by allosperm (donated by a partner) over autosperm
(produced endogenously). But the physiological and/or anatomical adap-
tations of the hermaphrodite duct and carrefour to separate eggs from
autosperm in inseminated individuals are at present poorly understood.
The male and female reproductive systems are separate from the carre-
four onward in all freshwater pulmonates except the Acroloxidae.
Autosperm is generally transported through the male duct into the vas
deferens, along a prostate gland, to the penis. The penis is held in a sheath
behind the head and, at mating, everted through a sack called the prae-
putium. A muscular ‘preputial organ’ may serve as a holdfast during cop-
ulation. Pulmonate copulation has been well described (Rudolph
1979a,b, van Duivenboden and ter Maat 1988, Wethington and Dillon
1996, Trigwell et al. 1997). There is no evidence of ‘courtship’ behavi-
our; snails seeking to mate as male do not seem to behave in any special
manner to increase the receptivity of a prospective partner. Nor is there
evidence that pulmonates have any behaviours or chemical cues to induce
insemination from a prospective partner who might otherwise be unwill-
ing. (It might seem that inductive behaviours would be especially adap-
tive for snails whose reproduction has been delayed for want of an
allosperm donor, but such is not the case.) Rather, the likelihood that
mating is initiated between a pair of snails encountering each other is a
function of the maximum autosperm store (sometimes called the ‘mas-
culinity’) of either (Figure 3.5). The energetic cost of mating as a male
is not negligible (DeVisser et al. 1994). So if both snails have recently
mated as males, they generally pass each other by, even though neither
may have ever mated as female.
A snail carrying autosperm in storage (a ‘male’), upon encountering a
second individual, typically mounts the shell of that individual (the
‘female’) and crawls across her shell to orient himself in parallel fashion.
The male positions himself along the edge of the female’s shell aperture,
everts his penis, and inserts it through the female’s vagina a considerable
distance into the oviduct. Sperm may apparently be conducted all the
Pulmonate reproduction · 81
Figure 3.5. Model showing the probability of mating in pulmonate snails as a func-
tion of life history stage. The abscissa is divided into juvenile, reproductively mature
but not self-fertilizing adult, and (only if uninseminated) self-fertilizing adult stages.
Solid lines trace mating probabilities if no prior copulation has taken place, as male
and as female separately. Note that the initial likelihood of female copulation is Pm,
the likelihood that a male initiates copulation. Dashed lines show mating probabil-
ities subsequent to a successful copulation. Based on studies with Physa heterostropha
(Wethington and Dillon 1996).
Selfing
All freshwater pulmonates seem to be able to reproduce successfully by
self- fertilization. Usually there are adverse fitness consequences, both to
the selfing parent directly (termed self-fertilization depression) and to the
selfed offspring (inbreeding depression, in its strict sense.) But as pointed
out by Doums et al. (1994), early workers (who tended to report an
advantage for selfing) biased their estimates by comparing the fitness of
84 · Gastropod autecology
isolated snails with the mean fitness of mass-cultured (and almost cer-
tainly crowded) outcrossers. DeWitt and Sloan (1958, 1959) were the
first to compare the fitness of isolates with uncrowded pairs, showing
clear overall advantages to outcrossing in Pseudosuccinea columella and
Physa heterostropha. Reductions in both the lifetime fecundity and the off-
spring viability of isolated snails have been documented in populations
of Lymnaea peregra from Lake Geneva (Jarne and Delay 1990), Bulinus glo-
bosus from Niger ( Jarne et al. 1991, Njiokou et al. 1992), Physa heterostro-
pha from South Carolina (Wethington and Dillon 1997) and Lymnaea
peregra from France (Coutellec-Vreto et al. 1998).
Copulation certainly does, however, have some negative effects on
fitness (van Duivenboden et al. 1985). It interferes with egg laying, is
energetically expensive, and exposes the copulants to predation. Njiokou
and his colleagues (1992) reported that B. globosus from an Ivory Coast
population showed improved fertility and offspring survivorship when
reared in isolation. And in fact, that population does seem to reproduce
at least partly by self-fertilization in nature, to judge from the poor fits to
Hardy–Weinberg expectation observed at allozyme loci. Jarne (1995) has
reviewed allozyme data collected from 24 studies (16 species, 260 popu-
lations) of pulmonates worldwide. Self-fertilization seems to be the ‘pre-
dominant’ mating system in Biomphalaria pfeifferi (2 studies), Bulinus
forskalii (2 studies), Bulinus senegalensis (1 study), and Bulinus truncatus (3
studies). To this list could be added Ancylus fluviatilis (Stadler et al. 1995).
Bulinus truncatus and A.fluviatilis are allotetraploids, although the chromo-
somal situation is not abnormal in the other species, as far as is known.
Bulinus truncatus populations often contain aphallic individuals, lacking
the terminal regions of their male reproductive system. Phally polymor-
phism seems to have both genetic and environmental components.
Development in a warm climate tends to promote aphally (Schrag and
Read 1992, Schrag et al. 1994). But the breeding experiments of Doums
et al. (1996a) demonstrated significant correlations in aphally both
between parent and offspring and within sibships. Aphally may be an
adaptation to lessen the consequences of trematode parasitism (Schrag
and Rollinson 1994). Regardless of the origin of the phenomenon, it
seems clear that self-fertilization should be more likely in populations
where aphally occurs in high frequency. And in fact, inbreeding depres-
sion in B. truncatus generally seems negligible (Doums et al. 1996b).
Jarne (1995) characterized 11 of the 16 pulmonate species in his tab-
ulation ‘predominantly’ cross-fertilizing. This does not mean that the fits
to Hardy–Weinberg expectation have been uniformly good, even in this
Prosobranch diet and habitat · 85
subset. In fact, heterozygote deficits are very commonly reported even
among outcrossing species. One likely factor must certainly be the inci-
dence of outcross sterility (but self-fertility) noted in P. heterostropha by
Wethington and Dillon (1993). Our large laboratory populations of P.
heterostropha showed roughly 6% male outcross sterility, 2% female out-
cross sterility, 2% double-outcross sterility, and 19% self-sterility. Another
factor is likely to be ‘mixed mating’, the production of selfed progeny
along with outcrossed progeny that has occasionally been detected even
in snails demonstrably holding stores of allosperm. Wethington and
Dillon (1997) monitored the lifetime reproductive output of 26 P. hetero-
stropha known to have been inseminated, and characterized 10 as ‘mixed-
maters’, producing offspring by self-fertilization at a frequency greater
than 1%. The absence of any fitness differences between the mixed
maters and the more purely outcrossing individuals suggested that the
phenomenon may be accidental, resulting from imperfect suppression of
autosperm.
Neritidae
Work on north European Theodoxus fluviatilis provides a useful compar-
ison with the pulmonates. Skoog (1978) performed a series of laboratory
growth studies with Norwegian Theodoxus along with the Lymnaea peregra
work we have discussed previously. His experimental design was similar:
10 adults per food item, maintained individually in filtered water, with
food and water refreshed every other day. Skoog carried his Theodoxus
experiment 80 days, rather than the 40 days of data he gathered for adult
Lymnaea. Lymnaea grew much faster, averaging 11.7 m of shell growth
86 · Gastropod autecology
per day on their preferred diet, as opposed to a maximum of 0.09 m
per day for Theodoxus. Lymnaea also generally laid eggs in culture, while
Theodoxus did not, but there was some mortality among the former,
while no Theodoxus died over a period twice as long.
Skoog observed some growth on all four diets offered Theodoxus. A
mixed culture of diatoms gave the best growth, followed by mixed blue-
green algae, Cladophora with diatoms, and an animal diet (crushed Jaera).
But the differences in growth rate were not significant. (It may be recalled
that mixed blue-green algae was significantly better than any other diet
for adult Lymnaea growth.) Skoog noted that his results agreed well with
those of Neumann (1961), who concluded that Theodoxus is not selec-
tive. Theodoxus seems in fact to have even less dietary specificity than an
extreme generalist among the pulmonates. In a second set of experi-
ments, however, Skoog found that the animal diet was significantly worse
in promoting juvenile growth in Theodoxus.
One much more commonly encounters studies performed in lotic
waters when surveying the prosobranch literature. Jacoby (1985)
anchored 12 aluminium-framed chambers in the rocky outfall of Swedish
Lake Erken, using vaseline-coated plastic strips as snail barriers. She
placed 50 uniformly sized rocks from a local quarry in each, and allowed
two weeks for colonization by periphyton. She then allowed Theodoxus
to graze at 5 per chamber, 40 per chamber (about their natural density of
450/m2) and 70 per chamber. After another 2 weeks, Jacoby found sig-
nificantly lower periphyton biomass (as chlorophyll a) in the low density
chambers, and much greater abundances of the green filamentous
Cladophora. Theodoxus guts contained 65% diatoms, 30% detritus and
bacteria, and 5% algal filaments, by volume. We shall have more to say
about prosobranch grazing in riffles in the next section.
Pleuroceridae
Pleurocerids are common and conspicuous in quite a few of the little
North American streams that freshwater benthic ecologists have found
convenient for study. For snails in such an environment, current and sub-
strate are the critical habitat variables. In eddys and pools, one may find
snails on any substrate where food is available, but in moving water,
clearly snails will require a solid substrate upon which to hold. Houp
(1970) mentioned this phenomenon in a population of Pleurocera acuta
from a small Kentucky stream, and Krieger and Burbanck (1976) make
similar observations in the Goniobasis suturalis population of a larger
Georgia river. Goniobasis livescens seems to require a solid substrate in
Prosobranch diet and habitat · 87
exposed regions of Lake Erie, but may be found on softer substrates
where protected (Krecker 1924).
Big Sandy Creek, Alabama, is typical of many streams in the piedmont
and coastal regions of the United States where pleurocerids find solid
substrate less than common. The creek is inhabited by a large population
of Goniobasis cahawbensis. As it passes Big Sandy Spring, a population of
G. carinifera is also introduced, and the two species co-occur for about
1 km before G. carinifera disappears. The temperature and dissolved
oxygen are much more constant in the spring water; temperatures are
generally cooler and dissolved oxygen generally lower. Thus it seemed
possible to Hawkins and Ultsch (1979) and Ross and Ultsch (1980) that
the distribution of these two pleurocerids might have a physiological basis.
They measured oxygen consumption at several temperatures, ‘heat
coma temperature’, ‘critical thermal maximum’, ‘maximal exposure
temperature’, and ‘ultimate lethal temperature’ for upstream G. cahawben-
sis, downstream G. cahawbensis, and G. carinifera. The results were quite
reminiscent of those obtained by Berg using limpets 20 years earlier: very
few differences. G. carinifera had lower lethal temperatures, but these were
still much in excess of field temperatures. Temperature preference tests
showed that both species seemed to prefer water somewhat warmer than
ambient in the wild. The authors therefore turned their attention away
from physiology and toward depth, current, and substrate.
Ross and Ultsch sampled 18 transects from 2.5 km above the spring to
3.3 km downstream, recording snail occurrence in two categories of
depth, two categories of current, and 17 substrate types. The data only
indirectly address the question of G. carinifera’s restricted range, and the
absolute preferences of the two species cannot be determined. But the
relative preferences of G. cahawbensis and G. carinifera in sympatry are
summarized in Figure 3.6. The spring-dwelling G. carinifera significantly
prefers water shallower and faster than G. cahawbensis, and hence is found
hanging onto logs and vegetation at streamside significantly more often.
Goniobasis cahawbensis was less common in the shallow rapids, but seemed
to prefer the other three habitat types, shallow/slow, deep/slow, and
deep/fast, equally well. (The authors measured current speed at the
surface, so the ‘deep/fast’ category is problematic.) Goniobasis cahawbensis
is more common on sand. The authors speculated that feeding prefer-
ences may play a role in the differing habitat choice. I would add that if
Big Sandy Creek becomes progressively slower and sandier downstream,
such habitat choice might explain the disappearance of G. carinifera 1 km
from the spring.
88 · Gastropod autecology
Figure 3.6. Habitat selection in two species of Goniobasis sympatric in Big Sandy
Creek, Alabama. (Data of Ross and Ultsch 1980.)
Pomatiopsidae
Although they have gills, as do all prosobranchs, many pomatiopsid
species are truly amphibious, or possibly even terrestrial. A well-studied
92 · Gastropod autecology
example is Oncomelania, the intermediate host of Schistosoma japonicum.
The young and newly hatched Oncomelania are generally aquatic, but
adults spend the majority of their time on moist soil, and can survive 129
days of desiccation (Komiya 1961). Komiya lists typical habitats in Japan
as ‘on the soil in river beds’, ‘above the edge of the water in irrigation
ditches’, and ‘uncultivated marshy areas’. Several analyses have detected
little effect of soil chemistry or texture (Pesigan et al. 1958), but moist
organic soil is clearly the preferred site for feeding and egg laying. One
might think that an amphibious habit could provide some insulation from
the effects of pollution and low dissolved oxygen, but Garcia (1972) has
found otherwise.
Van der Schalie and Davis (1968) reviewed 11 reports on Oncomelania
culture from 1933 to 1965, and notable subsequent contributions have
come from Davis (1971), French (1974), and Iwanaga (1980). It became
clear very early that these snails require a mud substrate in culture, usually
formed into banks or islands, and that, under some conditions, enough
algae, diatoms, detritus and/or bacteria may be present in the mud to
make additional food supplements unnecessary. Davis (1971), however,
found that innoculating the substrate with a 5% suspension of the diatom
Navicula luzonensis greatly increased average fecundity. Following Liang
(1974), French (1974) showed that the addition of small pieces of algal
mat (Fisherella, Nostoc, Schizothrix, etc.) resulted in higher average fecun-
dities than the artificial dietary supplements commonly offered
Oncomelania. These techniques recall some of the methods used to
culture the lymnaeid hosts of Fasciola.
Based on a detailed analysis of stomach and faecal contents from five
snails collected at each of 18 sites in the Philippines, Dazo and Moreno
(1962) concluded that Oncomelania was a herbivore. The most common
identifiable food items included 5 genera of green algae, 9 genera of
diatoms, 5 of desmids, 2 of euglenoids, and 4 of blue-green algae. The
authors reported a correlation between population density and the abun-
dance of green algae and diatoms, but offered no statistics, and the rela-
tionship is less than striking judging from the raw data. Protozoans and
nematodes were fairly common among the gut contents. And actually,
the most common item in the guts at all 18 sites was ‘vegetative matter’.
Among the most remarkable results obtained from the study of any
gastropod diet are those of E. D. Wagner and colleagues in the mid-1950s
(Ritchie 1955, Wagner and Wong 1956, Winkler and Wagner 1959). It
was discovered that Oncomelania survival was greater on a diet of ordinary
filter paper than on any of nine other items, including plain soil, fish
Prosobranch diet and habitat · 93
foods, liver extract, or leaves. Using radioactively labelled cellulose,
Winkler and Wagner demonstrated fairly convincingly that the cellulase
originates in the snail’s crystalline style. In summary, 50 years of research
has shown that the diet of Oncomelania may include just about anything
one could find on mud.
The North American pomatiopsid genus Pomatiopsis has received
some attention due, at least partly, to its biological similarity with
Oncomelania. Pomatiopsis lapidaria can on occasion be found in moist en-
vironments metres from any permanent water body. Pomatiopsis cincinna-
tiensis is often restricted to fairly narrow strips of muddy river banks, more
after the fashion of Oncomelania. Van der Schalie and Getz (1963) have
shown that the two Pomatiopsis species are more tolerant of subzero tem-
peratures and less tolerant of high temperatures than the four species of
Oncomelania. The substrate moisture preference and resistance to desic-
cation of P. lapidaria are similar to that of Oncomelania, while P. cincinna-
tiensis seems better adapted to drier conditions.
Van der Schalie and Getz (1962) noticed a relationship between soil
moisture and the abundance of P. cincinnatiensis within sites, the snails
apparently preferring 75–92% saturation. The authors felt that they might
be systematically overestimating soil moisture in the field, however, since
their sampler dug several centimetres into the substrate. So they con-
structed a test chamber with compartments containing a uniform soil
over a range of moisture contents, placed 20 snails in each compartment,
and allowed them to move about in darkness for 24 hours. The prefer-
ence for 70% saturation was striking, while moisture ranges of 0–40% and
90–100% seemed unacceptable to the snails. Since the authors detected
little variance in temperature and soil texture at their several study sites,
they concluded that moisture was the primary determinant of Pomatiopsis
microdistribution.
Van der Schalie and Getz also compared the soil textures at 24 sites
inhabited by P. cincinnatiensis in Ohio and Michigan with 54 sites from
which the snail was absent, and could find no differences in the relative
proportions of sand, silt, and clay. Their laboratory results on substrate
choice were quite interesting, however. The authors placed 25–50 snails
in the centre of a dish which had been equally divided into zones of
(rather unnatural) sand and (quite familiar) loam, and noticed that after
24 hours, the snails significantly preferred loam, by greater than 2:1. As
P. cincinnatiensis is not generally found on sand, this was not surprising.
They then repeated their experiment comparing loam with pure clay,
another substrate upon which the snail is not typically found. The snails
94 · Gastropod autecology
preferred clay 10:1. Van der Schalie and Getz do not offer an explanation
for this odd result, but I would speculate that the apparent preference for
clay was a temporary one, and would have changed as soon as the snails
got hungry.
Pomatiopsids of the Asian subfamily Triculinae are much more
aquatic, often living on rocks, sticks, and mud in streams with rapid
current. Tricula aperta, the intermediate host of Schistosoma mekongi, has
been cultured by several workers (Liang and van der Schalie 1975, Liang
and Kitikoon 1980, Lohachit et al. 1980, Kitikoon 1981a). The number
and quality of the experiments does not approach that available for other
medically important snails, for various reasons not under the control of
the investigators. Tricula is generally cultured in Petri dishes with at least
some mud, after the fashion of Oncomelania, although the mud is not
vital. (Mud may be more important as a refuge for food algae.) The life
cycle can be completed on provisions of the blue-green algae Nostoc mus-
corum or the diatoms Navicula and Achnanthes, and Tricula will also ingest
both maple leaves and filter paper. Kitikoon found over an order of mag-
nitude improvement in the fecundity of Tricula in an artificial stream,
compared with the more traditional Petri-dish method. It seems to be a
general result that animals from lotic waters are difficult to rear in the
laboratory, for reasons of temperature, flow, dissolved oxygen, or all three.
Hydrobiidae
The best studied hydrobiid is the European Potamopyrgus, whose habitat
Boycott (1936) described as ‘Running waters, rivers, canals, brooks,
ditches, running ponds [Boycott’s emphasis], often in the meanest little
trickles’. The species also commonly inhabits brackish water, and can
survive for months in full sea water, although apparently reproduction is
affected (Todd 1964, Duncan 1966).
Much unlike the outwardly similar pomatiopsids Oncomelania and
Pomatiopsis, Potamopyrgus may be found on rocks, mud, or macrophytic
vegetation, with little sign of substrate specificity. Heywood and Edwards
(1962) estimated that about 30% of the individuals in one reach of the
River Ivel, England, were on macrophytes and the remainder on the
muddy river bed, but it is difficult to know if this represents any prefer-
ence without estimates of total surface area available.
In a circular arena, Haynes and Taylor (1984) found that Potamopyrgus
was attracted to homogenized, filtered extracts of the macrophyte Apium
nodiflorum, fresh or decayed, the large alga Ulva, miscellaneous micro-
scopic algae, and crushed amphipod. The snail was indifferent to an
Prosobranch diet and habitat · 95
extract of fresh Elodea, but attracted by decayed Elodea, and was repelled
by an extract of the watercress Nasturtium. In paired choice experiments,
Potamopyrgus preferred (in this order) tissue extracts from (1) the amphi-
pod Gammarus pulex, (2) the snail Lymnaea peregra and well-decayed
Elodea tied, (3) decayed Apium, (4) miscellaneous green algae, (5) partly
decayed Elodea, and (6) diatoms. When interpreting the results of this sort
of experiment, it is important to remember that a particular foodstuff
may appear to be highly desirable while entirely unavailable to the snails
(e.g. crushed amphipod). It should also be kept in mind that two vari-
ables are being tested simultaneously: the actual preference for a foodstuff
and the ability to locate it. It is quite possible, for example, that diatoms
are the optimum diet for Potamopyrgus in the wild, but that diatoms
excrete little chemical cue when compared with rotten macrophyte.
Haynes and Taylor did not offer their snails any terrestrial leaf litter,
but such a potential food resource could be quite important, especially in
the ‘mean little trickles’. The experiments of Hanlon (1981) fill this gap
admirably, including both food preference and measurements of growth
rate. The author placed 200 Potamopyrgus in tanks with two 900 mm2
squares of a fresh-fallen beech leaf standard diametrically opposite two
900 mm2 squares of one of 11 other tree species, left these in the dark for
24 hours, and counted the individual snails on each leaf. Hanlon per-
formed five replicates and then repeated his experiments with a lime-leaf
standard. His results (Figure 3.7) show a striking preference for the ‘softer’
leaves (poplar, willow, and elm) over the leaves with harder cuticles, such
as beech and oak. The order of preference with the lime-leaf standard
was similar.
Note in Figure 3.7 that mechanically abraded beech leaves were sig-
nificantly preferred over unabraded controls. Hanlon performed some
choice experiments comparing freshly fallen leaves with those that had
been submerged in a stream for three months. Potamopyrgus significantly
preferred the partly decayed leaves over the fresh leaves, perhaps both
because they were softer and because such leaves have certainly been col-
onized by bacteria and fungi.
Hanlon placed 3 g of each leaf species in a litre of filtered pond water,
added 60 Potamopyrgus, and allowed the snails to feed for 10 weeks at con-
stant temperature. The top axis of Figure 3.7 shows the growth in shell
height of a sample from each of these 13 experiments. Clearly
Potamopyrgus can assimilate dead leaf litter and grow. Moreover, the cor-
relation between the two halves of the figure is gratifying. It is com-
monly taken for granted, especially by researchers who do not have small
96 · Gastropod autecology
Figure 3.7. The upper axis (hatched bar) shows the mean number of 200
Potamopyrgus attracted to 13 types of leaves (5 trials of 24 hours), and the lower axis
(solid bar) shows the mean ten-week increase in shell height for 30 snails fed on each
sort of leaf. (Data of Hanlon 1981.)
children, that animals tend to prefer the diets most beneficial to them.
This is the main rationale for choice experiments. But to my knowledge,
Hanlon’s data are the only demonstration that freshwater snails actually
grow better on diets they tend to prefer.
Two nicely complementary experiments have been performed with
lake populations of the North American hydrobiid Amnicola. The experi-
ments of Kesler (1981) were directed specifically to the population of
Amnicola limosa inhabiting Nonquit Pond, Rhode Island, and its effect on
the periphyton. Kesler exposed glass slides for one week at four sampling
periods to allow periphyton to colonize, then placed them for a second
week either in an enclosure containing A. limosa at normal density
(45–107 snails/m2) or an exclosure containing no snails. Generally
diatoms were the only algae colonizing these slides at densities sufficient
to report. At all periods, Kesler found that A. limosa grazing reduced the
abundances of all diatoms except our friend, the adnate Cocconeis.
Cattaneo and Kalff (1986) anchored 10 cubical frames in mesotrophic
Prosobranch diet and habitat · 97
Lake Memphremagog, on the Quebec/Vermont border, containing arti-
ficial macrophytes colonized by periphyton. They screened two each
with mesh sizes 0.1 mm, 0.25 mm, and 0.5 mm (effectively excluding
snails), and 2.0 mm (admitting Amnicola). Two frames were left
unscreened as controls. Between 31 July and 29 August there was a
marked decline in diatom biomass in the 2 mm and unscreened contain-
ers, and an increase in blue-green algae, particularly striking when com-
pared with more finely meshed chambers. The author’s inventory of all
grazers in all chambers convinced me that Amnicola almost exclusively
accounted for this effect. Small diatoms (⬍100 m3) such as Achnanthes
minutissima suffered especially significant declines, while the abundance
of the blue-green algae Gleotricha pisium (colonies up to 2 mm diameter)
showed a strong increase. Diatoms over 100 m3 and the green algae
(really very few colonies smaller than 1000 m3) were unaffected by
Amnicola. It is interesting to note that total grazer biomass remained fairly
uniform in all enclosures, with oligochaetes, chironomids, and cladoce-
rans replacing snails behind fine mesh sizes. None of these taxa seemed
to have an effect on the periphyton community as strong as Amnicola,
however. Thus in Lake Memphremagog, snails have some effect on other
grazers as well as on the grazed.
Ampullariidae
The capacity of these snails to enclose air within their mantle cavities
confers buoyancy unique among the prosobranchs. Their proboscis has
been modified into a pair of palps used to manipulate bulky items, such
as leaves. Thus in spite of their large size and apparently ponderous shells,
many species of ampullariids are characteristically observed in floating
mats of vegetation and among emergent macrophytes. We left our dis-
cussion of macrophytes and pulmonates with the impression that the
match was, more or less, a draw. In some places the pulmonates seemed
to be nibbling meekly about the edges of a plant population too large to
notice, while in others they seemed entirely frustrated by tough cell walls.
But if macrophytes have nightmares, it would be of ampullariids. Some
taxa may be restricted to the tropics and subtropics at least partly because
the growth rates of temperate macrophyte populations are insufficient to
support viable population sizes.
Ampullariids have attracted some attention as biological control agents
for both aquatic weeds and medically important pulmonates, and are not
difficult to culture. Santos and colleagues (1987), for example, fed Elodea
to their experimental population of Pomacea lineata, the ‘staple food in
98 · Gastropod autecology
the pond’ in Brazil. Meenakshi (1954) noted that Indian populations of
Pila virens generally eat aquatic vegetation such as Vallisnaria and Hydrilla,
but consumed filter paper ‘readily and in large quantities’. Experimental
populations of the Indian Pila globosa grow quite well on pure diets of
Ceratophyllum demersum, which as we have seen is nearly invulnerable to
attack by pulmonates (Haniffa and Pandian 1974, Haniffa 1982).
Salvinia is a large, aquatic fern that has become a pest in many tropical
regions. Thomas (1975) found that although Lymnaea and Indoplanorbis
could consume the tender leaves of young plants, the hispid texture of
the older fronds discouraged herbivory. But he reported that Indian Pila
globosa consumed Salvinia with no apparent difficulty, 20 g snails eating
an average of 2 g of plant tissue per day. The most extensive experiments
in ampullariid weed control have taken place with Marisa cornuarietis in
the United States. Seaman and Porterfield (1964) performed a series of
large-scale experiments, involving hundreds of snails and pounds of
macrophytes. Of seven Florida aquatic weeds, they found that Marisa
totally ingested first Najas, followed by Potamogeton illinoensis, Salvinia,
and Ceratophyllum over 25 days. The snails ingested somewhat more
Alternanthera tissue than was produced, retarded Pistia growth somewhat,
and had no effect on water hyacinth. Seaman and Porterfield attribute the
resistance of Pistia and water hyacinth to their habit of floating on the
surface, with leaf surfaces aerial.
The food-choice experiments of Estebenet (1995) suggested that
Pomacea canaliculata prefers macrophytes in the following order:
Zannichellia ⬎Myriophyllum ⬎Chara ⬎Rorippa ⫽Potamogeton ⬎Elodea.
Growth recorded over 120 days on these diets matched the preference
data exactly. It is thus quite clear that ampullariids can process and eat
large food items, such as macrophytes. They may even attack other snails,
as Chapter 7 will document. But nothing I have written here should be
construed as evidence that they cannot graze on small particles as well.
In fact, African populations of Pila and Lanistes may be abundant where
no macrophytes are available, on the stony shores or sandy bottoms of
lakes.
Ampullariids have evolved an unusual method for harvesting particles
from the surface film as well. In calm waters such as may prevail in labor-
atory aquaria, Pomacea may climb to the surface, fold its foot into a funnel,
and initiate a current with its pedal cilia (Johnson 1952, Cheesman 1956,
McClary 1964). A very shallow layer of surface water is drawn down this
funnel, from which the snail collects food particles with its foot.
Periodically the snail turns its head into the funnel and ingests the
Prosobranch diet and habitat · 99
entrapped particles. Both Johnson and McClary describe this behaviour
as primarily a response to the placement of food on the water surface, but
Cheesman suggested that the snails may be collecting and ingesting the
proteinaceous surface film itself.
Viviparidae
The most striking modification of the basic gastropod feeding method in
fresh water has been the evolution of filter feeding. The gills of Viviparus
are characterized by unusually large, triangular lamellae whose tips hang
over a ciliated gutter or ‘food groove’ running across the floor of the
mantle cavity (Cook 1949). Cilia propel water currents through the
mantle cavity of all prosobranchs, but in Viviparus they also direct mucus
and entrapped particles to the food groove. Particles collected on the gill
filaments are carried to the tip, where they also fall into the groove. A
food/mucus string forms, which is carried forward and collected into a
ball or ‘sausage’. Periodically the snail will turn its head and eat the col-
lected food.
The evolution of a new feeding apparatus does not necessarily imply
the forfiture of the old. Just as was the case with ampullariids, viviparids
have retained their ability to graze. In fact, Campeloma are sometimes
baited with carrion. But viviparids are more benthic than the closely
related ampullariids. Duch (1976) collected V. contectoides from a New
York river and placed 20 individuals in a levelled, darkened aquarium,
half with a sterile rock bottom and half with sterile silt-mud. The snails
initially moved about randomly, but after six hours all 20 had collected
on the mud side. Gut analysis of 50 wild-caught snails showed primarily
diatoms, some filamentous green algae, and very little blue-green algae.
Duch placed mixed cultures of diatoms in random areas of a sterile silt
substrate, and once again released 20 V. contectoides. The animals again
appeared to move about randomly until encountering a patch of diatoms,
until at one point 80% of them were aggregated on diatom patches.
Studier and Pace (1978) also noticed V. contectoides most commonly
burrowing in fine silt. They compared oxygen consumption in 16 snails
(half male, half female) enclosed for 2 hours in 250 ml Erlenmeyer flasks
containing sterilized natural substrate, fine sand, coarse sand, aquarium
gravel, or no substrate. The authors found no significant differences in
oxygen consumption among all these trials, concluding with the sugges-
tion that Viviparus distribution may be more a function of the food in the
substrate than its texture.
The dominant snail among the reeds fringing Japan’s Lake Biwa is the
100 · Gastropod autecology
viviparid Sinotaia historica, reaching summer densities of about three indi-
viduals per 10 cm of stem. Higashi and colleagues (1981) enclosed small
PVC rods previously colonized by algae with 5 mm mesh and introduced
Sinotaia into half of them, leaving the other half as controls. The results
were familiar: snails significantly reduced algal biomass and primary pro-
ductivity. Diatoms seem to have constituted the bulk of the algae on the
rods. From a comparison of grazed and ungrazed rods, it appears that the
Navicula population suffered more from grazing pressure than did either
Synedra or Cocconeis, although no estimates of significance were fur-
nished.
Bithyniidae
Filter feeding has also evolved in Bithynia, in a manner remarkably similar
to, but entirely independent of, its occurrence in the Viviparidae
(Tsikhon-Lukanina, cited in Monakov 1972, Meier-Brook and Kim
1977, Tashiro and Colman 1982). The apparent facility by which snails
of this family switch from grazing to filter feeding should have attracted
much more attention for them from ecologists than it has. Lilly (1953)
generally observed English B. tentaculata grazing actively over macrophy-
tes or hard substrates during the growing season, and buried in the mud
during winter. She did not notice any filter feeding, identifying the struc-
ture generally called a ‘food groove’ as a channel to clear silt from the gills.
But Tsikhon-Lukanina cultured laboratory populations of this species on
suspensions of the green algae Scenedesmus, and considered Soviet popu-
lations of this species primarily filter feeders.
Bithynia siamensis is the first intermediate host of a human liver fluke
in Thailand. Kruatrachue and colleagues (1982) have made an effort at
laboratory culture, using small round containers with a thin layer of mud
on the bottom. (It is not certain that the mud is required.) They offered
B. siamensis three diets: a complex artificial one, a suspension of unspec-
ified diatoms, and diatoms⫹artificial. Survival was poor on diatoms
alone, but addition of diatoms to the artificial diet significantly improved
fecundity. Brendleberger and Jurgens (1993) reported no difference in
rates at which B. tentaculata filtered green algal cells of three taxa: the
6 m (flagellated) Chlamydomonas, 6 m (non-motile) Chlorella, and the
10⫻20 m (flagellated) Chlorogonium. Although Bithynia could grow on
a diet of Chlamydomonas alone, the laboratory populations of
Brendleberger (1995) seemed to require solid food (lettuce) for repro-
duction.
The most thorough studies of the biology of Bithynia to date have been
Prosobranch reproduction · 101
those of Tashiro on the population of Oneida Lake, New York (Tashiro
1982, Tashiro and Colman 1982). Here the snails seem to graze only
during spring blooms of attached algae and sometimes during the
autumn, when the concentration of suspended particles is much reduced.
At all other times, they remain stationary under rocks and on pilings, filter
feeding. In the laboratory, Tashiro showed good assimilation both for sus-
pensions of Chlorella and for collections of natural ‘aufwuchs’ (unspec-
ified) offered on slides.
Valvatidae
This small group of small snails has attracted small attention. North
American species are primarily inhabitants of lentic waters, occuring at
greater depths and colder temperatures than most other gastropods. But
the European V. piscinalis is common in streams. Tsikhon-Lukanina has
also reported filter feeding in Valvata, and has reared these little snails on
a diet of suspended algae. Valvata do often live in very turbid environ-
ments, and their bodies are covered with cilia which constantly cleanse
them of silt. But Cleland (1954) and Fretter and Graham (1962) have not
reported any special devices for collecting particles. Instead, Valvata has a
long, heavily ciliated extension of the mantle, the pallial tentacle, which
these authors believe directs silt (as well as faeces) away from the body.
Valvata’s gill is not made of triangular filaments, but is bipectinate, and
often projects at least partly out of the mantle cavity, open to the water.
Thus both Cleland and Fretter and Graham believe that Valvata is strictly
a grazer.
Prosobranch reproduction
Gonochorism
The first reports of sex chromosomes in molluscs were made by Jacob
(1959a,b) working with the Indian Thiara crenulata and Paludomus tans-
chaurica (a pleurocerid). Based on the older technique of paraffin section-
ing, Jacob’s observations have subsequently been questioned (Patterson
1973). More reliable documentation of sex chromosomes was obtained
by Burch (1960) and Patterson (1963) from North American Pomatiopsis.
The largest pair of chromosomes in the complement of male P. cincinna-
tiensis are unmatched with regard to centromere position, but such is not
the case in the females. The largest chromosome in the complement of
male P. lapidaria seems to have no homologue at all (2N⫽33), although
102 · Gastropod autecology
again this is not the case in females (2N⫽34). Thus P. cincinnatiensis seems
to show an XY system of sex determination, while P. lapidaria shows XO.
Patterson (1969) also reported a large, dimorphic pair of chromosomes
peculiar to males of the North American viviparid Tulotoma angulata.
The above constitutes almost the entirety of our current knowledge
regarding sex determination in putatively gonochoristic molluscs.
(Marine species are even less known than the freshwater fauna.) In light
of their oft-skewed sex ratios, and the unexplored potentials of sequen-
tial hermaphroditism, parthenogenesis, and environmental influence,
much additional research is needed.
Although the gonochoristic prosobranchs of fresh water are diverse in
their evolutionary origin, some generalization regarding their mode of
reproduction is possible. The males generally possess a large, muscular
penis that is not retractable. It arises from the head behind the right ten-
tacle, and lies curled under the mantle when not in use in Theodoxus
(Fretter and Graham 1962), bithyniids (Lilly 1953), pomatiopsids
(Dundee 1957, van der Schalie and Getz 1962, Davis et al. 1976), and
most hydrobiids. The penis of the ampullariids arises from the edge of
their mantle, and that of the viviparids is a modified right (copulatory)
tentacle. The thiarids, melanopsids, and pleurocerids have no penis at all,
nor any external clue to their gender. Thus it would seem that sexual
reproduction has arisen at least four separate times in prosobranch groups
that have exploited fresh waters.
Female ampullariids of most species retain a vestigial penis, leading
Keawjam (1987) to speculate that Thai Pila may be protandrous her-
maphrodites. But careful study has uncovered no evidence of hermaph-
roditism throughout the life history of Pomacea (Andrews 1964, Lum-
Kong and Kenny 1989, Estebenet and Cazzaniga 1998). The absence of
a penis in male pleurocerids and thiarids also seems to indicate potential
hermaphroditism, although again, several complete studies of their mat-
uration have detected no evidence (Jewell 1931, Magruder 1935, Dazo
1965).
Male prosobranchs generally achieve copulation by mounting the shell
of a female, orienting in the same direction as their mate, and extending
their penis into a vaginal duct or opening on the female’s right. Very little
courtship or special behaviour prior to mating has been documented for
any freshwater prosobranch. Mating cues may be chemical. Many popu-
lations of freshwater prosobranchs manifest great genetic and morpho-
logical similarity with one another, yet co-occur with little or no
hybridization. It is also possible that prosobranchs do show mating beha-
Prosobranch reproduction · 103
viours which we have yet to observe. Opportunity for detailed observa-
tion of fleeting behaviours simply may not present itself in the wild, and
unlike the pulmonates, most prosobranchs are not easily kept in culture.
There is considerable indirect evidence that female prosobranchs can
store sperm or fertilized eggs for months before egg deposition.
Typically, prosobranchs cement small (1 mm) eggs singly or in small
clusters on solid substrates and cover them with sand grains, mud, or
faeces. Female ampullariids, pleurocerids, and melanopsids are endowed
with egg-laying grooves or ‘ovipositors’ on the right, dorsal side of their
foot for this purpose. Eggs hatch directly into crawling juveniles after
several weeks, depending on temperature. A minor, but nonetheless
striking exception is provided by the tropical neritid Neritina, whose pop-
ulations typically inhabit small rocky streams flowing directly into the sea.
Neritina spawns planktonic larvae, which are carried to the ocean and
develop into marine juveniles. Juveniles migrate back into fresh waters.
Ampullariid reproduction differs in quite a few respects from the
general prosobranch pattern outlined above. In the more amphibious
genera of ampullariids (including Pila and Pomacea), the male achieves
copulation by gripping the margin of the female’s shell and rolling her
over so that her aperture faces his. The penis is then inserted down the
length of the female’s pallial oviduct. In some circumstances, both snails
may withdraw until their opercula are almost closed, while in other
circumstances, the female may continue actively crawling (Andrews
1964, Albrecht et al. 1996). In any case, copulating pairs remain nearly
inseparable for many hours. Female ampullariids may emit some chemi-
cal or behavioural cue to indicate receptivity to insemination. In her
laboratory colony of Pomacea canaliculata, Andrews (1964) observed
‘Several males are usually attracted to the female at the same time, and
two or more may be attached to her, with the sheath in her mantle cavity,
but only one inserts his penis into the oviduct. Even after copulation has
begun other males remain in the vicinity.’ It would be most interesting to
learn how the successful male is determined.
Ampullariid eggs are exceptionally large, packed with albumen and
ultimately measuring 3–5 mm in diameter. In most ampullariids, the egg
is covered with a calcareous shell and laid above the water level on veg-
etation (Pomacea) or in excavated pits (Pila). Pomacea urceus inhabits neo-
tropical marshes and savannahs liable to de-water entirely during the dry
season. As the rains end, it burrows into the mud and deposits very large
eggs (10 mm diameter) just inside the aperture of its shell. It then with-
draws deeply inside its shell and aestivates. The young hatch during the
104 · Gastropod autecology
dry season, but remain under their mother’s shell until the water rises
(Burky 1974, Lum-Kong and Kenny 1989). Marisa cornuarietis is more
strictly aquatic than Pomacea or Pila. Marisa and Lanistes lay eggs (2–3 mm
diameter) on vegetation or other solid support below water level, unpro-
tected by a calcareous shell.
Pleurocerids and melanopsids also differ from the general prosobranch
pattern in that no behaviour distinguishable even as mating has ever been
reported. But spermatophores (2–4 mm) have been reported in their
culture water as eggs are being deposited. Pleurocerids lay eggs of the
ordinary prosobranch type, while melanopsids produce gellatinous spawn
containing several hundred eggs (Bilgin 1973). Dazo (1965) observed
that the pleurocerid pallial oviduct is saclike, lined with epithelial cells
‘into which spermatozoa, that have entered the female tract, insert their
heads and remain embedded for an indefinite time’. Both Dazo and I
(Dillon 1986) have observed the production of fertile eggs by North
American pleurocerids isolated in the winter. Thus it seems likely that
insemination may occur as early as the late summer or autumn, and that
females may store sperm over winter. Further, I have some indirect evi-
dence that female Goniobasis proxima mate but a single time in their entire
3–4 year lifespan (Dillon 1986, 1988).
Ovoviviparity is demonstrated by the viviparids, the thiarids, the pleu-
rocerid Semisulcospira, and the hydrobiid Potamopyrgus. Fertilized ova are
brooded in a modified pallial oviduct, sometimes called a ‘uterus’ in viv-
iparids (Vail 1977), or a ‘brood pouch’ in Semisulcospira or Potamopyrgus
(Fretter and Graham 1962, Davis 1969). The thiarids have a special
‘cephalic brood pouch’ in the back of their heads (Berry and Kadri 1974,
Muley 1977). At least one species of thiarid, the Indian Melania crenulata,
seems to release young from its brood pouch while they are still at the
free-swimming veliger stage (Seshaiya 1940). Many of the ovoviviparous
groups reproduce parthenogenically, as will be developed in the next
section.
In some cases, sex ratios in prosobranch populations seem to be
remarkably balanced. For example, on 13 June 1987 I hand-collected all
the Goniobasis proxima from a small rock ledge (about 1 m2 area) in the
Elk River of western North Carolina (population Elkp of Dillon 1984a).
The site, submerged under about 8 cm of clear, gently flowing water, was
especially chosen to yield a large sample of snails unbiased with regard to
size. The 487 individuals collected were cracked with pliers and sexed by
direct examination of gonad under low magnification. I counted 62
(generally large) parasitically neutered snails, making no attempt to estab-
Prosobranch reproduction · 105
Figure 3.8. Goniobasis proxima sampled from a single square metre of rock substrate
in the Elk River, western North Carolina, in June of 1987. The shell length is of the
last three whorls, due to apical erosion (Dillon 1984b). Size categories are minima.
Figure 3.9. Size distribution of Pomatiopsis cincinnatiensis from the Tecumseh Station
on the River Raisen, Michigan. (Data of van der Schalie and Getz 1962.)
Table 3.1. Sex ratios (as proportion male) for populations of gonochoristic freshwater prosobranchs
Notes:
a
indicates a significant difference from 1:1, by goodness-of-fit 2.
Prosobranch reproduction · 109
reliable estimates of sex ratio available for a variety of populations from
this group. I have excluded studies where the sampling technique was
unspecified, the sample size unknown, or the methods were likely to
yield size bias. Excluded also are studies where sex ratios have been cal-
culated across generations without first verifying that such a practice does
not introduce bias. (This criterion is more likely to exclude studies
reporting skewed sex ratios than those reporting equality, e.g. Calow and
Calow 1983.) Where sex ratio is demonstrated to vary by age (as in P.
cincinnatiensis) I have recorded only the estimate at the earliest age.
The techniques used by the various authors contributing to Table 3.1
have been quite diverse. Demian and Ibrahim (1972) used external shell
and body pigmentation characters to sex large samples of Marisa just as
they reached maturity (25–30 mm) from both field and laboratory colo-
nies. The west African bithyniid Sierraia sets a thickened aperture lip
upon reaching sexual maturity, a feature that facilitates survey of its sex
ratio. D. Brown (1988, personal communication) sexed samples of Sierraia
evenly selected to represent the size range displayed by adults from each
of three species. Lack of a penis makes pleurocerids especially difficult to
sex. Since both males and juveniles are missing an egg-laying groove, the
only reliable method of sexing (especially small) pleurocerids is direct
examination of their gonads. But Dazo’s (1965) data on laboratory-reared
cohorts, together with Mancini’s (1978) data on ‘mature’ G. semicarinata
from the field (summed over 13 months) convince me that the excess of
females observed by these authors is a genuine phenomenon.
In the end, the data of Table 3.1 remain enigmatic. I suspect that the
environment may play a role in the determination of prosobranch sex,
although again, established theory on the evolution of environmentally
determined sex does not seem to fit the particulars of prosobranch natural
history (Charnov and Bull 1977, Bull 1983, Head et al. 1987). Answers
await the day when clever and patient workers, at last, reliably rear a
variety of prosobranchs under appropriately controlled conditions.
Parthenogenesis
Parthenogenesis is known to have arisen only four times in the phylum
Mollusca, three of these in the freshwater prosobranchs. Honours for the
initial discovery go to my hero, A. E. Boycott (1919) working with the
hydrobiid Potamopyrgus ‘jenkinsi’. (It is now generally accepted that
European populations described under this nomen in the late nineteenth
century are Potamopyrgus antipodarum, described earlier in New Zealand.)
Parthenogenesis was subsequently recognized in the North American
110 · Gastropod autecology
viviparid Campeloma rufum (van Cleave and Altringer 1937) and in Indian
thiarids (Seshaiya 1936, Jacob 1957a,b). It now seems clear that the phe-
nomenon is quite widespread in the Thiaridae and Campeloma, although
it has not subsequently been documented in other hydrobiids. In any
case, the biological parallels in these three situations are most striking.
Parthenogenesis is apomictic in all three groups, as far as can be told.
Parthenogenetic females produce ova of the same ploidy as themselves,
without a reduction division. Males may be entirely absent from some
populations, rare in others, and common in yet others. Sexually repro-
ducing females do not seem to have been distinguished within bisexual
populations of thiarids, Potamopyrgus, or Campeloma, although there is
good evidence that sexual reproduction does take place under some
circumstances in all three groups. In no case has the female anatomy lost
its capacity to receive sperm. (The sperm channel, bursa copulatrix, and
receptaculum seminalis all remain.) Parthenogenesis does not seem to
originate from interspecific hybridization in any of these taxa. In neither
Potamopyrgus, nor Campeloma, nor the thiarids have putative ancestors
been identified from among the sexually reproducing mollusc fauna cur-
rently extant. Rather, allozyme studies seem to suggest multiple origins
of parthenogenesis from within sexually reproducing conspecific popu-
lations (Johnson 1992, Dybdahl and Lively 1995).
Wallace (1992) has made some observations suggesting that partheno-
genetic Potamopyrgus are polyploid, perhaps autotriploid, but a consistent
chromosome count remains elusive. Potamopyrgus males seem to be
diploid. Jacob’s (1957a) discovery of several 2N⫽32 populations of
Melanoides tuberculata prompted him to suggest polyploid origins for M.
lineatus (71–73 chromosomes), M. scabra (76–78 chromosomes) and most
other Indian populations of M. tuberculata (90–94 chromosomes). The
karyotypic situation in Campeloma is unclear.
Heaven knows that more work is needed in this area; I have already
emphasized that we are frighteningly ignorant regarding sex determina-
tion in gonochorists, and the situation in parthenogens must be compli-
cated yet further. However, it is tempting to speculate that the karyotype
of parthenogenetic females is unbalanced, yielding viable ova by mitosis
but generally inviable meiotic products. Males (and sexually fertile
females?) may arise from the rare reduction divisions that may be fortui-
tously viable. The incidence of viable diploid progeny in each popula-
tion would, of course, depend on the genetic background of the clone(s)
present. So too will the reproductive success of the sexuals. Sexual males
would seem more likely to contribute to the next generation if they can
Prosobranch reproduction · 111
find diploid, sexual females (not as yet humanly distinguishable), rather
than relying upon parthenogenetic females to furnish fertilizable gametes.
Wallace has gathered extensive data on the sex ratio in Potamopyrgus
from both its Australasian source of origin and its currently impressive
European range extension. Her (1992) review of 207 populations from
outside New Zealand found males to be absent or very rare (less than 1%)
in 183, and present at frequencies of 1–9% in 22 additional. Only in two
Welsh populations did the frequency of males exceed 10%. The situation
is quite different back home in New Zealand, however, where Wallace’s
survey of 58 populations found 18 with less than 1% males but 24 pop-
ulations with 10% males or greater. In fact, some New Zealand popula-
tions of Potamopyrgus approach 50% male. Wallace recognized ‘no
obvious relationship between sex ratio and habitat’ in her New Zealand
populations, although to my eye, her 27 ‘stream/drain’ populations seem
to hold especially high proportions of males. Sex ratio seemed fairly stable
in three New Zealand populations that Wallace monitored over a 15-year
period, although apparently significant trends were noted in two others.
Male Potamopyrgus are smaller than females, with a shell perhaps not
quite as elongated (Wallace 1979). Hermaphroditism has been described
(Wallace 1985). Although copulation has not been observed to my
knowledge, females from high-male populations ‘regularly’ hold sperm
in their receptaculum seminalis (Wallace 1992). Mother–offspring analy-
sis of allozyme inheritance has demonstrated at least occasional outcross-
ing in several bisexual Potamopyrgus populations from New Zealand
(Phillips and Lambert 1989). Wallace (1979) designed a rather heroic
‘breeding trial’ involving several hundred sets of one to several female
Potamopyrgus with or without males, held together in meshed containers
over four years. While the potential for sperm storage rendered some of
his data ambiguous, Wallace’s results with a ‘semi-albino’ line convince
me that outcrossing did, in fact, occur in her experiment. She reported
that females from high-male populations were much less likely than
females from low-male populations to reproduce successfully without
male contact. Given the reproductive success of the former group of
females when males were subsequently added, Wallace’s data convince
me that Potamopyrgus populations indeed vary in the degree to which
they reproduce sexually.
Data on sex ratios in Campeloma are neither as common nor as reliable,
both because of sexual dimorphism in size and body, and because of col-
lecting difficulties. It is clear, however, that both unisexual and bisexual
populations are found in the southern United States, while the north
112 · Gastropod autecology
seems to be inhabited only by unisexual populations (Karlin et al. 1980,
Johnson 1992). Again it appears that parthenogenesis is associated with a
range extension, in this case the natural colonization of higher latitudes
as the glaciers receded. Karlin and his colleagues performed a
mother–offspring analysis of allozyme genotypes in Campeloma from
both unisexual and bisexual populations. The offspring being brooded by
11 mothers from two bisexual populations did occasionally show out-
crossed genotypes, although results were far from Mendelian expectation.
But seven mothers from each of two unisexual populations, each hetero-
zygous at a minimum of one allozyme locus, bore entirely heterozygous
offspring, identical to themselves.
The incidence of sexual reproduction in thiarids parallels that of
Potamopyrgus and Campeloma in many ways. Both entirely female popu-
lations and bisexual populations of M. tuberculata are common in Israel,
and the frequency of males can be rather high (Livshits et al. 1984).
Livshits and Fishelson (1983) found that the esterase and acid phospha-
tase zymograms shown by the offspring of individual M. tuberculata from
some of these bisexual populations often did not match those of their
mothers. This constitutes considerable evidence of sexual reproduction.
But the Melanoides tuberculata populations introduced to Malaysia and
Hong Kong are entirely female (Berry and Kadri 1974, Dudgeon 1986),
as is the endemic thiarid fauna of Australia (Stoddart 1985). Males are
quite rare among the Indian thiarids, where they appear chromosomally
sterile (Jacob 1957b). Again, as was the case for Potamopyrgus and
Campeloma, some thiarids seem to be exceptionally well adapted for col-
onization (Madsen and Frandsen 1989).
One hypothesis to account for parthenogenesis in Potamopyrgus has
commonly been offered in reverse fashion, that sexual reproduction may
be an adaptation to a worsened environment, such as an increased level
of trematode parasitism. But sexual reproduction is generally believed to
have evolved with the Phylum Mollusca itself, in Precambrian times, and
it seems most likely that all three groups of freshwater parthenogens
evolved from sexual ancestors. Thus more recently this hypothesis has
been rephrased to the effect that parthenogenesis has evolved as a
response to environmental improvement. In either case, Lively (1989,
1992) has reported that in New Zealand bisexual populations of
Potamopyrgus tend to be found where parasite loads are especially high.
Lively noted that his data are consistent with the ‘Red Queen’ hypothe-
sis that sex remains adaptive to promote genetic variation as an escape
Prosobranch reproduction · 113
from parasites, while parthenogenesis evolves because of the twofold
advantage of asexual reproduction where parasite pressure is relaxed.
(That Potamopyrgus does in fact incur a ‘cost of sex’ has been verified by
Jokela et al., 1997.) Lively has, at times, offered data which suggest to him
that the ‘temporal variation’, ‘spatial variation’, and ‘reproductive assur-
ance’ hypotheses for the origin of reproductive mode do not fit his data
as well as ‘Red Queen’. More recently, however, it has become apparent
that parthenogenesis has not led to a reduction in genetic diversity in
Potamopyrgus (Phillips and Lambert 1989, Dybdahl and Lively 1995). And
no relationship is apparent between parasitism and the frequency of males
in the M. tuberculata populations of Israel (Heller and Farstey 1990).
If parasitism has been the driving force in the evolution of reproduc-
tive mode in Potamopyrgus, what is the role of other life history traits? Is
it a coincidence that the only parthenogenetic hydrobiid is the only
hydrobiid that broods its young? Is it a coincidence that the only parthe-
nogenetic hydrobiid is the only hydrobiid that has successfully colonized
a new continent in historic times? It should be hoped that any hypothe-
sis to account for the origin of parthenogenesis in freshwater molluscs
would offer some explanation for ovoviviparity and all the other life
history phenomena that seem so closely associated with it.
Hermaphroditism
The valvatids are hermaphrodites; a condition very unusual among pro-
sobranch snails. The male system seems to develop slightly before the
female, but mature Valvata manufacture egg and sperm simultaneously in
a single ‘hermaphrodite gland’ (Cleland 1954). The large penis (folded
behind the head) and the female opening are both located to the right
side of the body, as is normal for prosobranchs. Copulation has not been
described.
As there is only a single duct leading from the hermaphrodite gland,
self-fertilization seems quite possible in Valvata. There are no data on this,
as far as I know. Cleland never found ova in any region but the hermaph-
rodite gland, and suggested that copulation may be necessary to stimu-
late egg release, as in other prosobranchs. Hence the opportunity for
self-fertilization may be reduced. Some Valvata species seem to group
their eggs into larger masses than is typical for freshwater prosobranchs
(Heard 1963). Masses are cemented onto vegetation or other hard sur-
faces. This may necessitate periodic migration, since Valvata populations
often inhabit deeper bottoms characterized by soft mud substrate.
114 · Gastropod autecology
Summary
The trophic machinery mounted by typical gastropods seems somewhat
more selective than that of the bivalves. Snails are better able to select the
surface onto which they place their mouths, and their radular ribbon and
associated musculature seem finely adapted. Yet the overall impression
one takes from a review of gastropod diet is that the animals will stuff
anything into their mouths they can rip off the substrate.
The gastropod literature contains many studies of artificial culture
showing growth and reproduction on a bewildering variety of organics,
from fish meal to filter paper. The most interesting of these studies
compare growth and reproduction on varieties of natural diets, and here
we are not surprised to see differences both between species and between
age categories within species. There is no question that snails may be
attracted by chemical cues emanating from food items, and that feeding
may be stimulated by these cues. And there is no question that snails of
different age, size, and species vary in their preferences. I simply stipulate
that as they feed upon a dead worm they have travelled two hours to
search out, they are just as likely to eat the chaetae as the flesh.
Pulmonate snails are primarily adapted for calmer waters, while the
more heavily bodied prosobranchs are more commonly encountered in
lotic environments. Planorbids and ampullariids seem especially adapted
for softer substrates and floating vegetation by virtue of their shells, which
are rendered light by the air bubbles they enclose. Ancylids and neritids
prefer firm substrates, by virtue of their limpet forms, and lymnaeids and
pomatiopsids seem especially amphibious. Gastropod diets may include
any small organic particle: macrophytic vegetation (living or dead), litter
from the bank of the river or pond, algal filaments, fungal hyphae, detri-
tus, bacteria, and even carrion. Larger ampullariids, lymnaeids, and pla-
norbids are especially liable to consume tougher foodstuffs. Filter feeding
has evolved in the Viviparidae and the Bithyniidae.
Of all the aspects of freshwater mollusc biology, the reproductive func-
tion of pulmonates is certainly among the best known. Perhaps as a con-
sequence of this attention, we can say for certain that the phenomenon
is bewilderingly complex. The snails are best considered simultaneously
hermaphroditic, but usually the male function matures first, and some-
times the reverse is true. Given that a pair of pulmonates meet, the gender
role(s) they may assume is a function of prior reproductive history, among
other variables. And pulmonates also have the ability to self-fertilize,
which is preferred in some populations, and engage in ‘mixed mating’,
Summary · 115
the production of outcrossed and selfed progeny simultaneously. All this
reproductive diversity strikes me as ‘weedy’. The pulmonate reproduc-
tive system seems adapted for colonization and spread into disturbed or
newly opened habitat, or recolonization after a population crash. A single
colonizer can carry stored allosperm for thousands of fertilizations, or
self-fertilize if necessary. Then as the population grows and expands, its
members can cross-fertilize to preserve the genetic diversity remaining.
Most prosobranchs are gonochoristic; the particulars of their repro-
ductive biology being as diverse as their origins. Sex chromosomes have
been reported in several taxa. Sexes may be dimorphic with respect to
the speed and size at which they mature, yet even factoring out such con-
siderations, prosobranch sex ratios often seem highly skewed in the wild.
Hermaphroditism has evolved once in freshwater prosobranchs (the val-
vatids) and parthenogenesis at least three separate times. One striking par-
allel between the three parthenogenetic groups is that all are
ovoviviparous. While parthenogenesis seems to be a common feature of
reproduction in freshwater molluscs that brood their young, no molluscs
lay parthenogenetic eggs. I would not be so bold as to predict that an
oviparous parthenogen will never be discovered; on the contrary, I think
that unrecognized parthenogenesis is quite likely among the prosob-
ranchs. However, males seem to be at least moderately common in all the
populations of oviparous gonochorists thus far examined. Williams
(1975) has also observed that the offspring of parthenogens generally tend
to be large compared with the size of their mothers, continuously pro-
duced, liberated into an environment essentially the same as that experi-
enced by their mothers, and subjected mainly to intraspecific
competition. All these generalizations seem to hold quite nicely for the
thiarids, Campeloma, and Potamopyrgus, as will be seen in Chapters 4 and
5.
A second striking parallel among the three parthenogenetic groups is
that all have recently expanded their ranges by (sometimes spectacularly
successful) colonization. I view parthenogenesis as one of many life
history traits that have evolved in a set of freshwater molluscs I will term
‘R-selected’ (in a specially narrowed sense, to be detailed in the next
chapter). Individuals from R-selected populations may produce excep-
tionally large numbers of offspring, or offspring of exceptionally large
size, as is typical of such ovoviviparous groups as Potamopyrgus,
Campeloma, and the thiarids. R-selection also favours asexual reproduc-
tion, including parthenogenesis and self-fertilization, as may be demon-
strated by the pulmonates. Thus I would class my own theory regarding
116 · Gastropod autecology
the evolution of hermaphroditism and parthenogenesis in freshwater
molluscs as ‘Reproductive Assurance’, that asexual reproduction of any
type facilitates colonization by increasing the likelihood of reproduction
in a small number of colonists (Heath 1977). It further seems possible to
me that ovoviviparity may increase the physiological likelihood of par-
thenogenesis, especially as a consequence of polyploidy, independent of
its adaptive value. It is to the larger issues of life history evolution that we
now turn.
4 · Life history
Environment
Between 1969 and 1972, van der Schalie and Berry (1973) performed an
extensive series of experiments upon the effects of temperature on fresh-
water snails. Their apparatus enabled them to assess the effects of differ-
ences as small as 2°C simultaneously over large numbers of snail
populations held in vessels connected by a flow-through system, mini-
mizing variation due to water chemistry. Thirty newborn snails were
introduced into each vessel and reared for about 50 days beyond earliest
maturity, holding food and photoperiod constant. Example results are
shown in Figure 4.1.
First, it is clear that small differences in ambient temperature may have
profound effects on snails in culture. As little as 4°C may translate into
an order-of-magnitude difference in 50-day production of viable eggs.
Note that variation in total production of young was variously attribut-
able to differing survivorships, ages at first reproduction, fecundity, and
egg viability. For example, although 30 juveniles were placed in all vessels,
in the 14°C L. stagnalis experiment first egg laying occurred at day 92,
and only 12 individuals were alive when the experiment was terminated
at day 111. Thus the (standardized) 334 eggs counted by van der Schalie
120 · Life history
Figure 4.1. Viable eggs laid by four species of pulmonate snails, Lymnaea emargin-
ata, L. stagnalis, Helisoma trivolvis, and Physa gyrina, as a function of temperature
(van der Schalie and Berry 1973). Data are standardized to 50 days of reproduction.
N⫽30 individuals to start, although mortality occurred in all vessels.
Population studies
One might expect that the levels of phenotypic plasticity typically
observed in laboratory studies of life history traits, when combined with
the environmental variability and genetic diversity typical of freshwater
122 · Life history
mollusc populations, would result in an immense variety of life history
pattern. Kilgour and Mackie (1991) sampled 17 Ontario populations of
the cosmopolitan pill clam Pisidium casertanum, visiting each population
12 times during the course of one year. They recorded broods per year
(BPY), proportion of the population with shelled larvae (PPSL), adult
length at which shelled larvae first appeared (LSL), number of shelled
larvae per reproductive adult of standard size (NSL), lifespan (LS),
maximum adult length (MAL), and weight (MAW).
The authors observed two-fold or three-fold ranges in most variables:
BPY 1–2, LSL 1.7–2.6 mm, NSL 2.2–7.1, LS 12.5–36 months, and MAL
2.4–5.0 mm. Greater variation was observed in PPSL (2.6–16.7%) and
especially maximum adult weight (0.9–16.0 mm). The author’s principal
component analyses suggested that the populations of warmer environ-
ments tended both to reproduce earlier and die earlier. Highly significant
negative univariate correlations were obtained between temperature and
LSL, MAL, and MAW. Both MAL and MAW showed high positive cor-
relations with the particle size of the substrate, particularly the coarse
organic fraction. There was also evidence of a relationship between water
quality (calcium concentration, conductivity) and demography (MAW,
PPSL), a topic which will be pursued in Chapter 8.
What fraction of such life history variation among populations of
freshwater molluscs can be attributed to additive genetic variance? One
approach, pioneered by Forbes and Crampton (1942), is to compare pop-
ulations in a constant environment. Forbes and Crampton noticed that,
cultured separately in quart jars, individual Lymnaea elodes from a New
York population matured much earlier and at a smaller size than individ-
uals from a Connecticut population. The New York population laid sig-
nificantly fewer eggs, over a significantly shorter time, and had a
significantly shorter life span. Since these differences persisted over three
generations in a uniform laboratory environment, Forbes and Crampton
were justified in concluding that ‘their genetic causation seems probable’.
Calow (1981) noticed that the Lymnaea peregra populations of sheltered
habitats (a canal and a small pond) matured at a substantially larger size
than snails from exposed habitats (a lakeshore and a stream). He reared
snails from both habitat types in the laboratory at constant temperature,
ration, and water quality for three generations. Although no differences
were noticed in growth rate, snails from sheltered environments consis-
tently delayed reproduction until a larger size was reached. Their delay in
reproduction was compensated for, however, by an increased number of
eggs per individual per week once maturity was reached. We will return
Reproductive effort · 123
to further experiments with L. peregra along these lines later in this
chapter.
Another approach occasionally taken by those interested in the origin
of life history variation is reciprocal transplantation. McMahon (1975)
found considerable evidence of a heritable component to variation in
growth and fecundity comparing two New York Laevapex fuscus popula-
tions, while Geldiay (1956) could only distinguish environmental vari-
ance in growth among Ancylus fluviatilis from different sites in Lake
Windermere. Hornbach and colleagues (1991) were impressed by the
non-genetic component of life cycle variation among their native and
transplanted populations of Musculium partumeium. The experiments of
Hunter (1975) provide a cautionary note in this regard. He compared the
growth and fecundity of Lymnaea elodes (⫽L. palustris) in an abandoned
feeder canal (‘MAR’) and a swamp (‘FAY’) in upstate New York. His
reciprocal transplant experiments showed the FAY environment to be
much richer. The fecundity of MAR animals improved from 34.9
eggs/adult caged in their home environment to 255.7 eggs/adult caged
in FAY, while that of the FAY animals decreased from 371.3 eggs/adult
caged in their home environment to only 1.4 eggs/adult caged in MAR.
Hunter interpreted his findings to demonstrate that L. elodes ‘does not
occur as a series of populations each tactically suited to specific fresh-
water habitats.’ But I should note that each population remained strik-
ingly superior when caged in its home environment, and might argue
that the opposite is the case.
In sum, variance for life history traits will not be treated in this chapter
as though it arises either from phenotypic plasticity or from genetic adap-
tation. Traits as complex as those controlling demography will doubtless
have genetic, environmental, and interaction components. Experiments
to date suggest that some component of freshwater mollusc life history
variation will indeed generally be available for selection to act upon.
Reproductive effort
I am aware of several definitions of this term, falling into three catego-
ries. Sometimes reproductive effort is taken to mean the clutch size or
the total number of newborn offspring produced, or their value in calo-
ries or grams of carbon. On some occasions, the yolk volume or propor-
tion of carbohydrate, lipid and protein contained in the eggs or newborn
is estimated. Defined in this general fashion, reproductive effort is
expected to increase as a function of adult size. By a second category of
124 · Life history
definition, an attempt is made to remove the effect of adult size.
Reproductive effort is estimated as the ratio clutch weight:body weight,
carbon channelled to offspring:carbon contained in the female, egg
volume:parent volume, or yolk volume:adult growth. By the third cate-
gory of definition, reproductive effort is based on rates, as for example
the percentage of non-respired, assimilated energy allocated to reproduc-
tion or the more elaborate cost indicies of Calow (1979) or Hughes and
Roberts (1980). But a problem with these last two categories of defini-
tion is that linear relationships are assumed between adult and offspring
weight or between rate of assimilation and allocation to reproduction. A
non-linear relationship between whatever is in a numerator and whatever
is in a denominator will result in a statistic that may be misleading. So for
my purpose here, I will define the term ‘reproductive effort’ simply to
mean the carbon content of the offspring or eggs produced by a female.
Pisidium
Because of the difficulty of collecting eggs in the wild, more detailed data
are generally available regarding natural life history variation in brooding
molluscs than in oviparous ones. The pisidiid clams have attracted con-
siderable attention in this regard.
In 1986, Holopainen and Hanski published an extensive review of life
history variation in Pisidium. Their Table 1 summarized data on 17 var-
iables in 31 populations of Pisidium (9 species), including body size,
clutch, life span, and timing of reproduction. In almost all respects,
Pisidium populations display the range of life history variation we have
come to expect. For example, it will be recalled from Kilgour and
Mackie’s (1991) survey of 17 Ontario populations of P. casertanum that
all combinations of 1 to 2 broods per year and lifespans of 1 to 3 years
were observed. In the Holopainen and Hanski tabulation, all 10 P. caser-
tanum populations produced one clutch per year, but the northern
European populations seem to be much longer lived, up to five years. In
fact, over the entire data set, only one population produced two genera-
tions per year – the P. subtruncatum population of the River Tarrent,
England.
Although populations were sampled over two continents, from lotic
and lentic environments, temporary or permanent, Holopainen and
Hanski were struck by at least one constant. It appears that across all pop-
ulations, embryo size is rather uniformly about 1.0 mm shell length at
birth. Exceptions only include the unusually large P. amnicum (about 1.8
mm) and one or two unusually tiny species (e.g. P. moitessierianum, 0.6
Reproductive effort · 125
General survey
Pisidium are quite homogeneous reproductively, and range in adult size a
little over two orders of magnitude. One wonders to what extent the
relationship in equation (1) might apply across the broad diversity of
freshwater molluscs. Browne and Russell-Hunter (1978) have reviewed
the subject of reproductive effort in molluscs generally, including in their
literature survey one population each of eight freshwater species. Among
other attributes, the authors totalled the mean number of young per
average female, the average weight of young, and the weight of the
average female (in g C, hence my earlier conversion of Holopainen and
Reproductive effort · 127
Hanski’s data from AFDW). Included separately were data on the two
annual generations of the limpets Ferrissia rivularis and Laevapex fuscus,
plus first- and second-year data on the iteroparous Corbicula and three
year’s data on Viviparus georgianus, for a total of 13 observations, seven
semelparous and six iteroparous.
Browne and Russell-Hunter measured reproductive effort using two
coefficients: percentage non-respired assimilation annually diverted to
reproduction, and the ratio of carbon channelled into reproduc-
tion:carbon in the average female. Their analysis of the entire data set,
including marine as well as freshwater species, suggested that reproduc-
tive effort is higher in semelparous than in iteroparous populations, and
that in iteroparous populations, reproductive effort increases with succes-
sive breeding seasons. Oviparous species seemed to allot more energy to
reproduction than viviparous species.
From reports generally subsequent to 1978 I have collected data on
reproductive effort in five additional species, amounting to 12 observa-
tions beyond those of Browne and Russell-Hunter (Table 4.1). For the
pleurocerid Leptoxis carinata, May (pre-reproductive) mean adult weight
(mg C) was obtained for three New York populations from Aldridge
(1982, table 1). Adult weight was then substituted into population-spe-
cific regressions to obtain total egg carbon (Aldridge’s table 3, obtained
from cage experiments). Data relating mean individual size to fecundity
were also obtained from cage experiments for three New York popula-
tions of the pleurocerid Goniobasis livescens from Payne (1979, appendix
C). Adult weights (mg C) were calculated from Payne’s population-spe-
cific length/weight regressions and average egg weight read from Payne’s
appendix D. Reproduction was primarily semelparous in all six of these
pleurocerid populations.
For Musculium partumeium, Burky and colleagues (1985b) provided
selection ratios (average number of juveniles born per average parent) for
the spring generation of population AM (total), autumn generation of
AM, and population DW, as well as average weights of newborns. I esti-
mated average parent weight by back-calculating from stated reproduc-
tive efficiencies. Reproduction in the permanent pond (AM) was
iteroparous, while in the temporary DW reproduction was semelparous.
Haukioja and Hakala (1978, table 1 and equation 1) provided equa-
tions to estimate dry tissue weight from the shell length of reproductive
Anodonta anatina (⫽A. piscinalis) individuals from 13 populations in
Finland. They also provided a general equation (2) to estimate the dry
weight of glochidia as a function of shell length. Into these equations I
128 · Life history
Mean total
Mean annual
adult wt. offspring wt.
(mg C) (mg C)
Leptoxis carinata, New York (Aldridge 1982)
River TIO 7.30 1.151
River SUS 11.20 3.820
River UNA 9.76 3.260
Goniobasis livescens, New York (Payne 1979)
Lake GRL 153.0 0.278
Reservoir DER 104.0 0.238
Reservoir JAS 89.0 0.278
Musculium partumeium, Ohio (Burky et al. 1985b)
Pond AM, spring gen. 9.406 3.292
Pond AM, autumn gen. 4.876 0.829
Pond DW 5.139 0.925
Anodonta anatina (⫽piscinalis), Finland (Haukioja and Hakala 1978)
50 mm, population 95 200.0 28.84
75 mm, population 380 1044 220.4
Anodonta anatina, England (Negus 1966)
River Thames 290.0 12.50
Size at birth
In the preceding section we focused on the (rather unsurprising) fact that
the total weight of young produced annually is a function of adult body
size, in most cases. Our interest now turns to questions regarding the allo-
cation of a parent’s reproductive effort to juveniles. In particular, the
trade-off called ‘a few large vs. many small young’ has been identified by
Sterns (1976) as one of central interest.
136 · Life history
R S U
Reproductive effort, High Low Normal
relative to body size
Asexual reproduction More likely (–) (–)
(cf. Chapters 2 and 3)
Growth and/or maturation Rapid Slow (–)
(cf. section on maturity)
Lifespan Short Long (–)
Survivorship after Often semelparous (–) (–)
reproduction
Common in rich but Yes No No
unpredictible environments
Common in poor but No Yes No
stable environments
Adaptation to competition Intraspecific Intraspecific Interspecific
(cf. Chapter 5)
Note:
(–) indicates that a trait is undifferentiated, i.e., that no special prediction is made.
Pisidiids
We have previously touched on the observation of Holopainen and
Hanski (1986) that the size of Pisidium larvae at birth seems to be rather
fixed at about 1.0 mm shell length, over 31 populations of 10 species.
Although this is generally true, the smallest species (P. moitessierianum)
does give birth to the smallest juvenile (0.6 mm), while the unusually
large P. amnicum bears the largest juvenile (up to 2.2 mm). From the 19
data graphed as circles on Figure 4.4, Holopainen and Hanski suggested
that to a limited extent, size at birth might be constrained by parent size.
Pill clams of the genus Musculium tend to mature slightly larger than
Pisidium, perhaps 5–14 mm. Heard (1977) included data on the
maximum length of shelled larvae for populations of Musculium lacustre,
M. partumeium, M. securis, and M. transversum. This would be roughly
equivalent to Holopainen and Hanski’s ‘birth size’. I have graphed these
data as triangles in Figure 4.4, taking typical adult sizes from the scaled
illustrations in Burch (1975). It appears that Musculium birth sizes are also
rather invariant, here around 2.0 mm. Heard also included data on the
largest shelled larvae found in 10 species of Sphaerium, ranging in adult
Size at birth · 137
Figure 4.4. Shell size of the largest shelled larva as a function of the shell size of a
typical adult in the three genera of pill clams. The regression (r⫽0.80) is (larval size)
⫽0.325(adult size) ⫺0.063. (Data from Heard 1977 and Holopainen and Hanski
1986.)
size from 6mm to 18 mm in shell length (Burch 1975). Their larvae seem
more variable, ranging from 2mm to 6 mm, and a relationship between
birth size and adult size seems fairly clear.
Summing over all 33 pisidiid species, the relationship between the size
of a typical adult and the maximum size of the larvae it bears goes from
‘fairly clear’ to ‘quite convincing’. The simple linear regression:
y⫽0.325x⫺0.063 (3)
fits the data with a value of r⫽0.80. Evidently larger pill clams tend to
have larger offspring. But two points need to be emphasized. First, the
entire range of size at birth shown in Figure 4.4 is 0.6–6.2 mm, one order
of magnitude, while the range in reproductive effort, just for the genus
Pisidium alone, is three orders of magnitude (Figure 4.2). Clearly variance
in offspring size is not as important as offspring number in accounting for
variance in pisidiid reproductive effort. Second, much of the variation in
offspring size seems to be related to higher levels of phylogeny. Pisidium
138 · Life history
Figure 4.5. Bars show the range of size at release for 10 freshwater bivalve taxa. The
number of species surveyed for each taxon is given at right, followed by the number
of populations, if different. Size was measured across shell at maximum dimension.
General survey
As brooders, one might not be surprised to learn that the three pisidiid
genera give birth to the largest offspring among the freshwater bivalves.
In unionaceans, for example, glochidia are released at sizes an order of
magnitude smaller. From Chapter 2 it may be recalled that glochidia are
classified into three types: the ‘hooked’ type characteristic of the subfam-
ily Anodontinae, the ‘hookless’ type of most Ambleminae, and the ‘axe-
head’ type of the North American genus Proptera. I have surveyed the
works of Lefevre and Curtis (1912), Coker et al. (1921) and Kat (1984)
and obtained glochidial sizes (maximum shell dimension) for 61 taxa of
freshwater mussels, primarily from scale drawings. Figure 4.5 shows that
hooked (anodontine) and axe-head (Proptera) glochidia tend to be
released at larger sizes, up to 0.47 mm and 0.41 mm, respectively. Among
amblemine genera, Truncilla is distinguished by an unusually small adult
and juvenile size, the two species surveyed releasing glochidia of 0.05 mm
and 0.08 mm. Mussels of the family Margaritiferidae also produce excep-
tionally small glochidia. Thus a large phylogenetic component to off-
spring size is as apparent in unionaceans as it was in pill clams. But
Size at birth · 139
Figure 4.6. Bars show the range of size at birth for nine taxa of freshwater gastro-
pods. The number of species surveyed for each taxon is given at right, followed by
the number of populations if different. Size was measured as maximum shell dimen-
sion.
combining all 61 mussel species surveyed, glochidial sizes range only from
0.05 mm to 0.47 mm, not quite an order of magnitude.
Figure 4.5 also shows data on size at release for Dreissena, Corbicula, and
enlarged samples of Pisidium, Musculium, and Sphaerium. Again, most of
the Pisidium data, expanded from 19 to 28 populations, come from
Holopainen and Hanski (1986), while Sphaerium data (expanded from 10
to 19 populations) are primarily from Heard (1977). The effect of
enlarged data sets are negligible, however. Judged generally by largest
shelled larva, size at birth still ranges only from 0.6 to 2.2 mm for Pisidium
populations, while increasing slightly to 1.8–7.0 mm for Sphaerium and
1.2–2.2 mm for Musculium.
Figure 4.6 shows the result of a fairly thorough literature search for data
on shell size at first release in freshwater gastropod taxa. For inclusion in
this data set, I required direct observation of eggs and hatchlings (or live-
born) in the laboratory. About 55 references were ultimately discovered
containing sufficiently detailed information. Size was taken as the
maximum shell dimension at hatch, or release in the case of the brood-
ers (Potamopyrgus, Thiaridae, Viviparidae). Note that in some cases, juve-
nile gastropods have shells shaped so differently from those of adults that
their maximum dimension will be measured in a different plane.
140 · Life history
Over all 74 freshwater gastropod populations, the shell size of juveniles
at release seems to range not quite two orders of magnitude, from 0.3 mm
in Valvata piscinalis to 15 mm in the ampullariid Pomacea urceus. But just
as was the case with the bivalves, Figure 4.6 shows that most of the var-
iance is attributable to levels of classification higher than species. In no
family does offspring size vary more than fourfold. Especially striking in
their constancy are the pulmonates, where hatching size in 36 popula-
tions, taken about evenly from four families, ranges only from 0.36 mm
to 1.4 mm.
There seems to be some weak tendency for taxa with large-bodied
adults (ampullariids, viviparids) to produce large offspring, and small taxa
(Valvata, pomatiopsids) to produce smaller offspring. But pleurocerids
have rather large adults and small juveniles. One can also detect a posi-
tive relationship within taxa (such as pulmonates) between adult size and
size at hatching, although perhaps not as marked as that shown in Figure
4.6. The pulmonate with the smallest adult size surveyed (the planorbid
Armiger crista) had the smallest hatchling size, while the largest adults
(Lymnaea stagnalis) have among the largest hatchlings (1.3 mm).
So in summary, over all freshwater molluscs, shell size at release ranges
from 0.04 mm to 15 mm. The great majority of this variance seems to
be attributable to modes of reproduction that differ strikingly among
diverse higher taxa. One would not expect Dreissena, with planktonic
larvae, to produce young comparable in size to parasitic glochidia, much
less egg-laying pulmonates or brooding thiarids. Within higher taxa, var-
iance is generally much less than an order of magnitude, and seems to be
positively related to adult size. The ‘many small vs. few large young’
trade-off does not seem as important in freshwater molluscs as it may be
in other groups.
Maturity
We have now seen that both reproductive effort (as total weight of young)
and individual offspring size (setting aside a phylogenetic component) are
functions of the size of the mother in freshwater mollusc populations.
What determines the size at which a snail or bivalve matures? Our obser-
vations regarding offspring size are reminiscent of those for the much
better studied mammals (Charnov 1991). In Charnov’s view, the positive
correlation observed between parent size and newborn size in mammals
may be an adaptation to speed juveniles through an especially perilous
childhood. And a view prevailing in the literature of life history evolu-
Maturity · 141
tion holds that delayed maturation may also be an adaptation to similar
selection pressures (Reznick et al. 1990). If the survivorship schedule is
deeply concave, selection will favour those individuals allotting all their
energetic resource to growth through the most hazardous age classes,
deferring reproduction. But if early survivorship is high relative to late
survivorship, selection will act to promote reproduction at the earliest
possible date.
Somatic growth rates in freshwater molluscs are commonly shown to
be sensitive functions of temperature, diet, water quality, and other en-
vironmental variables. Thus size, rather than age, is generally found to be
a better predictor of reproductive maturity. The heritable component of
size at maturity may be large. In the laboratory, individual Biomphalaria
glabrata vary in shell diameter at onset of (selfed) egg laying from 5 mm
to 16 mm. Richards and Merritt (1975) produced an outcrossed B. glab-
rata line and subjected it to bidirectional selection. In four generations the
mean size at onset of selfed-egg production for the low-selected line
was 8.6 mm (range 7–11 mm), and for the high it was 12.1 mm (range
11–16 mm), the trait apparently showing almost no overlap. F1 hybrids
between snails of these two lines were intermediate in shell diameter at
onset of egg laying, although nearer the large-diameter parent. Although
Richards and Merritt did not monitor age, they stated that there was ‘no
indication that age at onset of egg laying was consistently correlated with
diameter.’
Wethington and Dillon (1993) have also documented significant
differences in size at first reproduction between isofemale lines of Physa
reared in a constant, laboratory environment, although we have not per-
formed any selection experiments thus far. As in Biomphalaria, age at first
onset of selfed egg laying in Physa seems to be a function of ambient tem-
perature, among other things. But size at onset of egg laying seems to be
genetically controlled, and I have an extensive data set showing growth
rate in my laboratory populations to be highly heritable.
Lam and Calow (1989a) quantitatively sampled the Lymnaea peregra
populations at sites on three rivers near Sheffield, England (Sheaf,
Rivelin, and Don) from January 1985 to December 1986, noting several
quite striking life history differences. Survivorship over the first few
months of life was similar in all three populations, but much greater for
the Don population during the autumn, winter, and early spring. (The
Don site was protected from strong current, and artificially warmed by
pipes from an adjacent steel factory.) That population began breeding
earlier than the Sheaf and Rivelin populations, at a much smaller mean
142 · Life history
shell length (6.2 mm in the 1985 cohort, as compared with 10.9 mm and
11.6 mm). Maximum lifespan for individuals of all populations was about
one year. Data on survivorship and shell growth at two of Lam and
Calow’s populations are shown in Figure 4.7.
The authors considered the possibility that Don snails may have been
induced to breed earlier in the season by the warmer water temperatures
they enjoyed. But they noted that the mean growth rate in the cooler
Sheaf waters actually exceeded that observed in the Don. In late March
1985, 24 weeks after hatching, the Don snails were 6.2 mm in shell length
and laying eggs, while Sheaf snails were 7.6 mm and not. Why did Sheaf
snails continue to grow, postponing reproduction until May? It does
appear that Sheaf snails ultimately produced more eggs, on average, as
would be expected from their larger sizes at reproduction. Although there
were no differences between the two 1985 cohorts in mean egg volume
or percentage hatchability of the eggs they laid, 43 egg capsules collected
from the Sheaf site contained on average 57.2 eggs, while 40 from the
Don showed a mean of only 15.9. Lam and Calow indirectly estimated
a mean of 521 eggs per individual Sheaf snail, and only 50 eggs per snail
in the Don. Thus in spite of the lower survivorship to reproduction at
Sheaf, snail population density there in the summer of 1985 far exceeded
that observed in the Don.
In June, 1985, Lam and Calow (1989b) brought 50 adult snails from
each of their three study sites into the laboratory, collected a sample of
their newborn, and initiated both controlled-breeding and mass-culture
experiments. In the former, F1 individuals were reared in small contain-
ers at 12 h light:12 h dark at 20°C for 100 days, paired randomly with
another of its own population until onset of breeding, then re-isolated.
F2 individuals were collected and reared in a similar fashion. In the latter
experiments, 50 F1 individuals from each of the three populations were
reared in large, continuously flowing tanks at natural light and ambient
temperatures (mean⫽16°C, s.d.⫽4°C). A great variety of life history
variables were measured on all snails, including growth rate, shell length
at reproduction, egg volume, and total number of eggs produced.
From the controlled-breeding experiments, Lam and Calow observed
significant intrapopulation variance for most life history traits. Between
populations, however, results were equivocal. The authors did observe a
highly significant difference in shell length at onset of breeding in the
mass-culture experiments, consistent with field observations. The Don
culture began laying eggs at a mean shell length of 7.5 mm (10 weeks),
while the Sheaf postponed to a mean shell length of 10.2 mm (13 weeks).
Maturity · 143
Figure 4.7. Growth and survivorship data of Lam and Calow (1989a) on 1985
cohorts of Lymnaea peregra from the Rivers Sheaf and Don, England. Time is meas-
ured in four-week intervals from the May date of maximum juvenile density. The
left axis (and closed circles) show age-specific survivorship (lx, standardized to 1000
individuals at start), while the right axis (and open circles) show mean shell length
(SL). The dashed lines mark the size and approximate date of the onset of egg laying.
144 · Life history
There was no difference in the total number of eggs produced, perhaps
because the Don ultimately bred for 26 weeks, Sheaf only 22. Significant
among-population variance in shell length at maturity was not observed
in the controlled-breeding experiments, either at F1 or at F2 generations.
My reading of these results would be that the Sheaf and Don populations
do, in fact, show significant genetic divergence in their mean sizes at
reproduction as shown in Figure 4.7, but that this effect may be extin-
guished if reared at a constant temperature of 20°C. And I would suggest
that the selection agent has been differential survivorship in the vicinity
of reproductive age, quite possibly in the form of a steel mill.
The contrast between the L. peregra populations of England and the
Bulinus globosus populations of Zimbabwe studied by Shiff (1964c) is in
some respects jarring. Zimbabwean Bulinus mature in about 10 weeks
and reproduce year-round, except in temporary ponds during the dry
season. But one can see precisely the same relationships between sur-
vivorship and age at reproduction within a single Bulinus population as
were manifest between populations of English Lymnaea or Oklahoma
Physa.
The absence of readily distinguishable cohorts introduces an added
challenge to the study of Bulinus demography. Shiff (1964b) preceded his
field studies with careful observations on the relationships between shell
growth, age, and temperature in the lab. He estimated lx in the field by
establishing arbitrary ‘cohorts’ as groups of snails newborn on each of five
dates during the year. He followed the survivorship of each ‘cohort’ from
month to month by counting the number of snails in their expected size
range. Shiff’s (1964c) ‘Table II’ reported cohort ages in 12-day ranges
(since 12 days are required for egg hatch) and included estimates of egg
fertility. For my Figure 4.8, I standardized viable embryos (not total eggs)
to 1000 and age to the last of the 12 days for each period.
Shiff provided a thorough description of both his study population and
the climate of Zimbabwe, doubtless with those of us at home in temper-
ate regions of the northern hemisphere in mind. His B. globosus popula-
tion inhabited two small, interconnected ponds (area 448 m2, maximum
depth 80 cm). Rains came in mid-November, 1962, and on November
22 the habitat started to fill. Shiff took his first population sample on
November 27, by which time the (really very few) Bulinus that had sur-
vived the dry season had already begun to lay eggs. Water temperatures
continued to warm through the next several months, with flooding (and
much reduced survivorship) in January and February. The upper half of
Figure 4.8 shows that the November 27 ‘cohort’ reproduced in February
Maturity · 145
Figure 4.8. Growth and survivorship data of Shiff (1964c) on two cohorts of Bulinus
globosus from a small pond in Zimbabwe. Time is marked in days from hatching;
remainder of the details are as in Figure 4.7.
Figure 4.9. Life history data on two populations of Lymnaea elodes from ponds in
northern Indiana (Brown et al. 1988). Survivorship (left axis, solid lines) is standard-
ized to 1000 individuals at date of maximum egg production. The upper survivor-
ship curve shows uncorrected ‘head count’ while the lower curve is corrected for
parasitic castration. Mean fecundity (right axis, dashed lines) is given in eggs per indi-
vidual per month, corrected for parasitism.
148 · Life history
rates were greater in pond F, so that this minimum size was reached in 11
months rather than 12.
By far the more spectacular life history difference in the two popula-
tions was their reproductive effort. Although adults were of comparable
sizes, individual population F snails laid on average 242⫹715⫹637⫽
1594 eggs (corrected for parasitism), over three months of reproduction,
then died. Brown and colleagues estimated that pond A snails produced
a corrected average of just 163 eggs in the single month of reproduction
available to them in their first year, before dry-out. At least some,
however, survived to breed again.
It would appear that the pond F L. elodes population studied by Brown
et al. (1988) displays the attributes of R-selection (Table 4.2). The en-
vironment seems to be very rich, promoting good growth and high sur-
vivorship (9.8% survive to age 9 months, as compared with 0.7% at
pond A). But the environment is also unpredictable. (Brown reported
that pond F did dry occasionally, although not in the years of his study.)
In response, an annual, semelparous life cycle seems to have evolved, pop-
ulation F individuals producing an order of magnitude more offspring
than the iteroparous population A. But how much of this phenomenon
is additively genetic?
Brown (1985a) used transfer experiments of two different designs to
answer this question. All snails were held in small, floating chambers with
screened openings, either two or four per chamber. In his ‘reciprocal
transfer’ experiment, Brown reared juvenile snails in all combinations of
two ponds, two populations, and two densities for 50 days, until pond A
dried. He recorded age and shell length at maturity, shell growth incre-
ment, clutch size, and total fecundity. For his ‘parent–offspring’ experi-
ment, Brown collected eggs from pond F chambers laid by parents of
both populations. He held eggs and newborns in the lab at 5°C over
winter, then returned the juveniles to pond F chambers until the last died
in mid-November. He recorded the same five life history traits as in the
reciprocal transfer experiment, plus age and shell length at death.
Brown clearly demonstrated significant genetic divergence in life
history traits between populations A and F of a remarkable (and, to me,
unexpected) nature. The tendency for population F snails to grow faster
is heritable. Reciprocal transplant experiments demonstrated that, reared
in either environment at either density, population F snails showed a sig-
nificantly greater growth increment. Their ages at maturity were not
different from population A snails, but of 45 values of the F-statistic, the
Maturity · 149
highest was a population difference in shell length at maturity (F⫽78.7,
with 1,88 d.f.). Reared at either density in pond A, population F snails
reached maturity at a mean shell length of 20 mm while population A
snails were reaching maturity at the (more normal) 15 mm.
Brown’s parent–offspring experiments confirmed the genetic nature of
the population difference in growth rate. The six other values of F from
his reciprocal transfer experiment that I would judge significant at the
(Bonferroni-corrected) 0.05 level included pond (i.e. environmental)
effects on shell length at maturity and shell growth increment, as well as
age at maturity and total fecundity. This is unsurprising, since these two
ponds were initially selected for study because of their striking environ-
mental differences.
At first these results would seem to directly contradict the thesis I have
advanced in this section, that differences in survivorship schedules such
as displayed in these two populations (Figure 4.9) should promote matu-
ration at a smaller size in population F, not larger. But the key is to note
that reproduction may be accelerated in two fashions: lowering the size
at maturity or raising the growth rate. Even though their size at matura-
tion may be larger, population F L. elodes mature a month earlier than
pond A snails. Although maturity is best considered a function of size in
freshwater molluscs, survivorships generally seem more time specific than
size specific. Thus age at maturity may be more critical, in an evolution-
ary sense, than size at maturity.
Lymnaea elodes, it may be recalled, was one of the two snail species in
the dietary dilution studies of Rollo and Hawryluk (1988). Individuals
from their Canadian population of L. elodes reared in the lab were not
growing or reproducing at maximal rates, but rather at submaximal rates
homeostatically maintained. I suggest that the Canadian population of L.
elodes studied by Rollo and Hawryluk was, like most mollusc populations,
Undifferentiated with respect to life history. I further suggest that R-
selection may have worked to bring the growth rate displayed by Brown’s
pond F L. elodes much closer to maximal.
In Table 4.2 I have included a line labelled ‘growth and/or matura-
tion’, indicating my prediction that growth rate will be more rapid in R-
populations, and/or that maturity will come early. This I would take to
be an adaptation to environmental conditions which are generally good,
rich in nutrients and safe, but which may unpredictably turn hostile. S-
populations inhabit poor environments, where juvenile survivorship may
be predictably low, and hence maturation delayed.
150 · Life history
Unionoids
All unionoid populations are iteroparous and perennial, displaying life
cycle patterns classifiable into a fairly manageable number of discrete cat-
egories. Sterki (1895) recognized two groups: short term (‘tachytictic’)
breeders gravid only in the summer, and long term (‘bradytictic’) breed-
ers remaining gravid through the winter for glochidial release the follow-
ing spring. ‘Breeding’ may not be the best term for holding glochidia in
the marsupium, but its use has historical precendent. Ortmann (1909)
counted 13 tachytictic and 36 bradytictic species among the fauna of
Pennsylvania. The tabulation of Coker et al. (1921) included (to my eye)
36 tachytictic North American species and 18 bradytictic, with data on
another 12 species too fragmentary to permit classification.
Spawning time did not enter into the original tachytictic/bradytictic
system used by the early workers to classify unionoid life history. But with
the recognition that some populations may spawn in the autumn came
the realization that animals gravid in the winter might be short term
Life cycle pattern · 151
Figure 4.10. Example life cycles for the unionoids, illustrated by northern hemi-
sphere examples given in the text. ST, summer tachytictic, WT, winter tachytictic,
WB, winter bradytictic, SB, summer bradytictic, M, multicyclic.
that this population of A.peggyae produces two generations per year seems
inescapable. ‘Multicyclic’ is included as a fifth category of unionoid life
history in Figure 4.10.
Over the last hundred years, a large volume of information has been
published regarding the life cycles of unionoids. Early reports tended to
be fragmentary; based on small samples or short term observations. I have
resurveyed the literature to extract only those studies involving samples
over periods of a year or more. Neither the number of samples, nor the
number of individuals per sample, could be small. Data on gravidity were
required; data on gametogenesis desirable but not necessary. The 45 pop-
ulations (22 studies) for which adequate information is available are
graphed by their life history category in Figure 4.11.
It appears safe to conclude that most unionoid populations spawn on
rising, rather than falling, temperatures. The summer tachytictic cycle is
by far the most commonly reported life history in the literature, followed
by the summer bradytictic. The seven populations thus far known to
spawn on falling temperatures include five Anodonta: two from Florida
(WT, Heard 1975), A. piscinalis from Poland (WB, Tudorancea 1972), and
A. cygnea from Portugal (WB, Galhano and Da Silva 1983), in addition
to the Italian A. cygnea mentioned previously. Other than the Japanese
Inversidens used to illustrate WT, the only non-Anodonta known to spawn
in cooling water is the hyriid Cucumerunio from New South Wales (WT,
Jones et al. 1986).
154 · Life history
Multicyclic life histories are not commonly reported. They seem to
occur more frequently in lower latitudes, including in Louisiana (Parker
et al. 1984) and Zimbabwe (Kenmuir 1981), as well as New South Wales
( Jones et al. 1986). Their comparative rarity may reflect both the diffi-
culty of recognizing M life cycles in the field, and the obscurity of trop-
ical faunas in general. It also seems clear from Figure 4.11 that tachytictic
cycles are more commonly reported than bradytictic. I can see little evi-
dence of either a systematic or a latitudinal trend, however. Both SB and
ST populations seem equally common in Wisconsin, Minnesota,
Alabama and Florida.
At least some life history diversity may reflect adaptation for the par-
titioning of fish hosts. The examples illustrated in Figure 4.10 include a
WB population releasing glochidia in the spring, an ST releasing in the
summer, and a WT releasing in the autumn. Available fish hosts over
mixed mussel beds may well differ among seasons, or in any case, the same
individual fish would seem less likely to be multiply infected. Host fish
partitioning may, at least in some cases, be even finer. As may be recalled
from Chapter 2, Villosa nebulosa, V. vanuxemi, and Medionidus conradicus are
very similar unionids co-occurring in Big Moccasin Creek, Virginia.
Zale and Neves (1982b) reported that all three show SB life cycles,
spawning on the rising water temperatures. But their ‘set points’ seem to
differ, such that in 1979 M. conradicus spawned July 8–16, V. vanuxemi
spawned July 23–30, and V. nebulosa spawned August 13–20. This may be
a coincidence. It may reflect an adaptation to minimize hybridization. Or
it may, in later months, translate to slightly differing windows for glo-
chidial release, yielding in turn the preferential infection of three differ-
ent glochidial hosts, a darter, a sculpin, and a bass, respectively. In any
case, I am intrigued.
Gastropods
The first formal classification of freshwater gastropod life cycles was that
of Russell-Hunter (1961b, 1964, 1978), based primarily on the number
of generations per year and survival after egg laying. He reviewed several
dozen populations of snails, mostly European pulmonates, and recog-
nized seven life cycles: (1) one generation per year, reproducing semel-
parously in the spring, (2) one generation per year, reproducing
semelparously in the late summer, (3) two generations per year, the
spring-born iteroparous and the summer-born semelparous, (4) two gen-
erations per year, both semelparous, (5) three semelparous generations per
year, (6) one generation per year, reproducing iteroparously, and (7) a
two-year generation time.
Russell-Hunter documented great interpopulation variability in fresh-
water gastropod life cycles. British populations of Lymnaea peregra, for
example, may display five of the seven categories of pattern (1–4 and 6),
while Ferrissia rivularis and Physa fontinalis may display any of four. He
stated (1961b:165) ‘In most cases, these interpopulation variations could
be determined phenotypically by trophic or temperature differences in
the environment.’ But he clearly did not mean to imply no role for genet-
ics. Regarding number of generations per year, he documented life cycle
(5) only in the physically small Lymnaea truncatula and life cycle (7) in
some populations of the large Lymnaea stagnalis. And regarding semelpar-
ity/iteroparity, he stated ‘The selective advantages of an annual cycle, in
which the adults do not survive long after breeding to compete for food
with their offspring, are obvious.’ Russell-Hunter’s primary conclusion
in 1961 was that ‘selection has produced genotypes which can show phe-
notypic plasticity.’
Russell-Hunter’s classification was expanded and modified by Calow
(1978). Calow combined (1) and (2) into category A, and called catego-
ries (3), (4), and (5) B, C, and F, respectively. He added two rather rare
life cycle types: (D) three generations per year, spring and summer gen-
erations iteroparous, autumn generation semelparous, and (E) three gen-
erations per year, spring generation semelparous, summer iteroparous,
autumn semelparous. Calow combined Russell-Hunter’s categories (6)
and (7) into a single category G.
Life cycle pattern · 157
Calow then surveyed the literature and gathered life cycle data on 52
populations (or population groups) of pulmonates and 14 populations of
prosobranchs, representing 37 species combined. He de-emphasized the
non-genetic component of the variation catalogued, focusing instead on
its adaptive significance. Semelparity, in Calow’s view, might not be an
adaptation to minimize parent/offspring competition, as suggested by
Russell-Hunter. Instead he suggested that it may represent ‘reproductive
recklessness’, an adaptation to maximize reproductive effort at the
expense of adult survival.
In reviewing data collected during the last 15–20 years on the life
cycles actually displayed by freshwater snail populations, it has become
clear to me that the classification system of Russell-Hunter and Calow
requires some clarification regarding semelparity and iteroparity. Note,
for example, that if two generations per year are the norm, both the first
(usually spring) generation and the second (summer or autumn) genera-
tion could be semelparous or iteroparous. That would imply four types
of cycles: abbreviated by their initials ss, si, is, and ii. But Russell-Hunter
and later Calow described and figured only two types of such cycles: B
showing is and C showing ss. What is one to do with the Indian Bellamya
population described by Khan and Chaudhuri (1984), where the animals
reach reproductive age in half a year but typically reproduce at age 1 and
age 11⁄2 years as well? I believe it is in the spirit of the currently-existing
classification system to recognize types si, is, and ii as subtypes of of B,
reserving C for complete semelparity. This system is illustrated in Figure
4.12.
Similarly, given three generations per year, there are eight possible per-
mutations of s and i. Russell-Hunter recognized sss in the wild, which
became Calow’s F. Calow’s D would be iis, and his E would be sis. Since
the distinction between D and E is in the reproduction of the first
(spring) generation, it would seem consistent to also include iss, isi, and
iii as subtypes of D and ssi and sii as subtypes of E, again reserving F for
complete semelparity. This system is illustrated in Figure 4.13.
I agree with Russell-Hunter that it is important to distinguish between
snails that mature in one year and those that require two years to mature.
Figure 4.12 shows as category G only those populations that mature in
12–23 months and reproduce iteroparously. New category Hi includes
populations that require 24–35 months to mature and reproduce iterop-
arously, with new category Hs describing those reproducing semelpar-
ously at age 24–35 months. The iteroparous categories G and Hi include
those populations that reproduce twice, surviving a spring reproduction
158 · Life history
Figure 4.12. Schematic diagrams to illustrate eight types of freshwater snail life
cycles involving two generations per year (B and C), one generation per year (A and
G) and a two-year generation time (H). The ordinate for each graph is a measure of
size. Each curling line represents a cohort, born at size 0 and growing with time until
maturity. Cohorts are semelparous when they are mature for a single period of repro-
duction, and iteroparous when they are mature over two or more periods of repro-
duction.
but not a second, summer or autumn reproduction, and hence not sur-
viving to an additional year.
Note that the diagrams shown in Figures 4.12 and 4.13 are intended
only as examples, not as absolute patterns. Each of the iteroparous gen-
erations for cycle types B and D–F is shown surviving to contribute to
exactly two cohorts, but could in reality contribute to any number
greater than one. Life cycle types G and H are shown with one, spring-
born cohort per year, but additional cohorts per year are certainly pos-
Life cycle pattern · 159
Figure 4.13. Schematic diagrams to illustrate eight types of freshwater snail life
cycles involving three generations per year. See Figure 4.12 for further details.
sible. For example, the Hong Kong thiariid Brotia hainanensis ordinarily
requires 21⁄2 years to mature, reproduces twice a year, and lives 31⁄2 years
(Dudgeon 1989). By definition this is an Hi cycle, but is far from match-
ing anything in Figure 4.12.
Since 1978, several populations of freshwater snails have been reported
to display three-year generation times. Two such populations of
Goniobasis livescens were discovered by Payne (1979), and one each of
Theodoxus fluviatilis and Viviparus georgianus were reported by McMahon
(1980) and Jokinen et al. (1982), respectively. In addition, McKillop
(1985) suggested that Canadian populations of Bulimnea megasoma might
be triennial, although Gilbertson et al. (1978) reported type A in
Minnesota. Thus I suggest a new (ninth) category of freshwater snail life
160 · Life history
Figure 4.14. The life cycles displayed by 120 populations of freshwater pulmonates,
combining Calow’s pre-1977 data set with 16 studies published subsequently. See
Figures 4.12 and 4.13 for an explanation of life cycle types.
Figure 4.15. The life cycles displayed by 57 populations of prosobranch snails, com-
bining Calow’s pre-1977 data set with 21 studies published subsequently. See Figure
4.12 for an explanation of life cycle types.
simply become absorbed by a more recently born cohort, one must rely
on the judgement of the individual researchers involved.
Second, even if the real life history of all snail populations were
known, considerable judgement would be called for in sorting them into
neat categories. For example, Brown et al. (1989) concluded that most
individuals of a Louisiana population of Viviparus subpurpureus repro-
duced semelparously at age two years, although some were reproductive
at age one, and that although most individuals of a co-occurring popu-
lation of Campeloma decisum were semelparous, some survived to repro-
duce a second time at age three. Both these populations appear as Hs data
in my Figure 4.15. The Connecticut population of Viviparus georgianus
that I have mentioned earlier among the T data comprises males that first
mate at age two years and die, while females reproduce semelparously at
age three. The large range of life cycles reported for V. georgianus (G –
Browne 1978, Hi – Buckley 1986, both from New York) is interesting
to note in passing.
Calow’s (1978) observations are generally supported by the enlarged
data sets of Figures 4.14 and 4.15. There is a very striking difference
between the life cycles displayed by pulmonates and prosobranchs. Life
cycle A is now the most common type in both groups, not just pulmo-
162 · Life history
nates. But while 32% of pulmonate populations display generation times
of less than one year, only two such prosobranch populations are known
(one of the Canadian Bithynia tentaculata populations studied by Pinel-
Alloul and Magnin 1971, and the Indian Bellamya bengalensis of Khan and
Chaudhuri 1984). While 39% of prosobranch populations have genera-
tion times of 24 months or more, the phenomenon is rare in pulmonates.
McKillop (1985) reported two-year generation times for Lymnaea stag-
nalis and L. palustris, while Morris and Boag (1982) obtained the same
result for Helisoma trivolvis. Since pulmonates and prosobranchs often co-
occur, and since there is no systematic difference in their sizes at repro-
duction, one might infer a phylogenetic component to growth rates, with
prosobranchs slower.
Semelparity seems to be more common among pulmonates. Of the 77
pulmonate populations that reproduce first at age one year, 66 do so and
die (type A). I have collected records for 33 prosbranch populations
maturing at age one, of which only 18 show life cycle type A and 15
show G, that is, living to reproduce again. Brown and colleagues (1998)
have observed that such broad differences between pulmonates and pro-
sobranchs may extend to their competitive abilities, dispersal capabilities,
and risk from predation as well. Such clear-cut differences along phylo-
genetic lines comprise important evidence that variation in life cycle and
autecology, although perhaps not always available for selection to act
upon, are nevertheless not entirely the result of phenotypic plasticity.
Figure 4.16. Schematic diagram showing the life cycle types (classified according
to Figures 4.12 and 4.13) displayed by model populations varying two parameters:
survivorship (m) and maximum size (k, in mm). Two levels of environmental fluctu-
ation were examined, high (Iowa) and low (England). Varying k above 10 mm did
not change the outcome; values of m less than those shown resulted in the extinc-
tion of the model population.
Summary
Fundamental aspects of the demography of freshwater snail populations
are very sensitive to environmental fluctuation. Varying ambient temper-
ature over a few degrees, for example, may result in order of magnitude
differences in survivorship and reproduction. Little wonder, then, that
comparisons of the life histories displayed by allopatric populations,
which generally inhabit different environments and display some genetic
divergence, uncover bewildering diversity. Much of this variance seems
due to phenotypic plasticity. But a heritable component of interpopula-
tion life history variation can usually be distinguished, especially with
laboratory studies. And life history variation between freshwater mollusc
species, which often ranges much greater than an order of magnitude for
any character, clearly has a genetic basis.
In freshwater mollusc populations, reproductive effort is a power func-
tion of adult weight. In Pisidium, the annual weight of offspring seems to
be determined by adult weight alone, such that individuals surviving to
a second year reproduce according to their new, larger size. There is no
evidence that Pisidium populations sacrifice survivorship for reproduc-
tion; they do not apparently allocate extra energetic resource to repro-
duction and die as a consequence. The same could be said for freshwater
molluscs generally.
Exceptionally, however, populations of freshwater snails have been
described whose adults, on the average, have reproductive efforts an order
of magnitude greater than expected for their body size. These popula-
tions inhabit rich environments but reproduce semelparously. Also
Summary · 169
exceptionally, populations have been described with an order of magni-
tude lower reproductive effort than would be expected from body size,
inhabiting poor but stable environments. Inspired by the CSR model of
plant strategy, I here propose the ‘USR’ model of life history variation in
freshwater molluscs. Mollusc populations of the former type are here des-
ignated Reproduction-adapted, those of the latter type Stress-adapted,
and all other populations are considered Undifferentiated.
Juvenile size at hatch or parental release varies from about 0.4 mm shell
size up to 15 mm in freshwater molluscs. The great majority of this var-
iance is attributable to phylogeny – genus, family, or higher level classifi-
cation. Within taxon, variation in juvenile size is generally much less than
an order of magnitude, and positively correlated with parent size. Thus
it seems that the ‘many small vs. few big offspring’ trade-off is not a major
variable in the life history strategies of freshwater molluscs.
Given, then, that both newborn size and overall reproductive effort
generally seem to be functions of adult body size, the determinants of age
and size at reproduction take special importance in freshwater molluscs.
This seems to be largely a function of age-specific survivorship. Setting
aside (rather immediate) newborn mortality (following Charnov 1991),
we note that if early survivorship is relatively high, freshwater snails
mature rapidly, while if early survivorship is low, reproductive maturity
will be delayed. Size at reproduction usually seems to be the critical var-
iable, although I am aware of at least one case where a (apparently R-
selected) population achieved early maturity by improved growth rate.
Interpopulation differences in size and age at reproduction, and in sur-
vivorship after reproduction, manifest themselves strikingly in the vari-
able life cycle patterns displayed by freshwater mollusc populations. Five
distinct life cycle patterns can be distinguished in unionid mussel popu-
lations. Although previous authors have recognized seven categories of
freshwater snail life cycles, here I recommend an expanded and clarified
system involving nine types, some with subtypes. Among both pulmo-
nate and prosobranch populations, the annual semelparous type A life
cycle is most common. Pulmonates often show two generations per year
and rarely three generations per year or two years per generation.
Prosobranchs very rarely show two generations per year and much more
commonly show generation times of two or even three years.
A modelling effort suggests that at least some of the observed variation
in pulmonate life cycle pattern may be explained as a consequence of
varying size at reproduction and environmental harshness. Life cycle
pattern may often best not be viewed as directly adaptive. Subjectively,
170 · Life history
however, the observed distribution of pulmonate life cycles does seem to
include both more semelparity and more examples of lengthened gener-
ation time than my model would suggest. So once again, we find some
evidence of both R- and S-adaptation in data taken from the wild.
5 · Population dynamics and
competition
Population growth
Discussions of population dynamics conventionally begin with an exam-
ination of the form of exponential growth taken by populations unlim-
ited by resources and unchallenged by predators, parasites, or pathogens.
Such data are generally obtained in the laboratory. DeWitt (1954) seems
to have been the first researcher to examine the growth of a freshwater
mollusc population growing exponentially in a laboratory setting.
Although the techniques in use today are not substantially different from
those employed by DeWitt, the years have seen considerable variation in
terminology and convention. For the purpose of this review, a popula-
tion growing exponentially increases as:
Nt ⫽N0 ert
where N0 is the population size at start and Nt is the population size t
units of time in the future. Curiously, the standard unit of time in lab-
oratory studies of freshwater snail demography has come to be the fort-
night, 14 days. The constant e is the base of the natural logarithms, and
r is the intrinsic rate of increase, which will be the object of our interest.
For convenience, let er ⫽, the ‘finite rate of increase’, the factor by
which a population increases per unit time.
Population growth · 173
Values of r are generally obtained by solving Euler’s equation:
兺l
X
x
mx e⫺rx ⫽1
Laboratory studies
The experiments of Pointier and colleagues (1991) nicely illustrate the
application of demographic methodology to laboratory populations of
freshwater snails. The authors collected Biomphalaria glabrata from
Celigny Pond, Guadeloupe, and removed 81 of their apparently viable
eggs to a separate container for observation. Hatchlings were transferred
to larger containers to minimize crowding, and four adults were ulti-
mately cultured per 1 litre aquarium. Snails were held at 25°C and fed
dried lettuce, with weekly water change. Pointier and colleagues
recorded the proportion surviving and the mean viable eggs they pro-
duced every two weeks (‘fortnightly’) for 21 fortnights, by which time
65% of the adults had died, and the mean fecundity of the survivors had
decreased substantially (m21 ⫽25.2, l21m21 ⫽8.7).
Pointier et al.’s results are shown graphically in Figure 5.1. Note that
since the experiment was terminated before the population had
exhausted its reproductive capacity (either by mortality or senescence),
the value of R⫽1419 obtained by the authors is a slight underestimate.
Solving Euler’s equation, the authors obtained a value of r⫽0.84 per fort-
night, and hence ⫽2.32. Considered objectively, a mean reproductive
rate of 1419 viable offspring per adult per lifetime, and a population that
increases by a factor of 2.32 every 14 days, are both quite remarkable sta-
tistics. But as we shall see, such statistics are not unusual for freshwater
snails.
Pointier and colleagues performed similar analyses on a total of seven
snail populations – two each of the medically important Biomphalaria
glabrata, B. alexandrina, and B. straminea, plus one population of Melanoides
174 · Population dynamics and competition
Figure 5.1. Proportion surviving (lx, closed circles) and average per capita produc-
tion of viable offspring (mx, open circles) for a cohort of Biomphalaria glabrata
(Pointier et al. 1991).
tuberculata, which has been nominated as a possible agent for the compet-
itive control of Biomphalaria. The Melanoides experiment started with 40
juveniles newly liberated from adults collected on Guadeloupe. All snails
surviving to reproductive age seem to have been females. Although raised
under conditions identical to those previously described for B. glabrata,
the life history of Melanoides proved strikingly different (Figure 5.2).
Reproduction commenced at fortnight 9, but did not peak until the
second year of Pointier et al.’s study, at fortnight 56, and was still ongoing
when the experiment was terminated at fortnight 133. In the conclud-
ing fortnight, after over five years of observation, 25% of 40 Melanoides
were still alive, bearing 15 offspring among them. Although the net
reproductive rate of Melanoides (R⫽493) does not seem as impressive as
the R⫽1419 of B. glabrata, newborn Melanoides are much larger than pul-
monate hatchlings, and presumably show much greater survivorship. The
intrinsic rate of increase r⫽0.24 obtained by Pointier and colleagues is
one of only two such values available for a non-pulmonate freshwater
mollusc.
Population growth · 175
Figure 5.2. Proportion surviving (lx, closed circles) and average per capita produc-
tion of viable offspring (mx, open circles) for a cohort of Melanoides tuberculata. These
data are extracted from the much larger (fortnightly) data set of Pointier et al. (1991).
Figure 5.3. Intrinsic rate of increase (r, per fortnight) as a function of temperature
for eight pulmonate populations. From de Kock and Joubert (1991): crosses,
Helisoma duryi; open triangles, Biomphalaria pfeifferi (South Africa). From Shiff and
Garnett (1967): upright closed triangles, Biomphalaria pfeifferi (Rhodesia). From
Sturrock and Sturrock (1972): inverted closed triangles, Biomphalaria glabrata. From
Prinsloo and van Eeden (1969): squares, Lymnaea natalensis; closed circles, Bulinus
tropicus. From Shiff (1964a): open circles, Bulinus globosus. From Brackenbury and
Appleton (1991a): diamonds, Physa acuta.
Prinsloo and van Eeden 1969 Bulinus tropicus 1.98 890 15 0 Circulating system 21°C (opt.)
S. Africa
Lymnaea natalensis 0.55 382 15 0 Circulating system 18°C (opt.)
S. Africa
Rankin and Harrison 1979 Physa marmorata 1.76 458 ? ? 2/200 ml, 5/week Dasheen leaves
St Vincent Sta. 71
Physa marmorata 1.52 425 ? ? 2/200 ml, 5/week Dasheen leaves
St Vincent Sta. 51
Physa marmorata 2.70 1712 ? ? 2/200 ml, 5/week Dasheen leaves
St Vincent Sta. 77
Physa marmorata 0.80 36 ? ? 2/200 ml, 5/week Dasheen leaves
St Vincent Sta. 69
Shiff 1964a Bulinus globosus 1.93 467 30.5 3.2 6/‘large aquarium’,
Zimbabwe 2 months
Shiff and Garnett 1967 Biomphalaria pfeifferi 0.48 111 17.5 11.2 2/L
Zimbabwe
Sturrock 1966a Biomphalaria pfeifferi 0.86 182 15 0 10/10 L, 2 months
Tanzania
Sturrock and Sturrock 1972 Biomphalaria glabrata 0.47 324 15.5 63.7 9–12/10 L,
St Lucia biweekly
Wethington and Dillon 1997 Physa heterostropha 1.41 808 30 0 1/200 ml, weekly FF
South Carolina
Notes:
Values of r and study duration (xmax ) are expressed in (14-day) ‘fortnights.’ All studies at optimum temperature (if known) or ‘room
temperature’ (23–25°C) unless specified. Food was lettuce unless specified otherwise; ‘FF’ designates commercial fish food.
Population growth · 181
Figure 5.4. Values for the intrinsic rate of increase (r, per fortnight) from 36 labor-
atory populations of freshwater snails. The data are from Table 5.1.
Field studies
The quantity r is a parameter of the logistic growth equation, and as such
has primarily found use to describe populations under ideal circum-
stances, perhaps those newly colonizing a rich habitat or growing unlim-
ited in an artificial environment. A stable age distribution is assumed. But
r has also been calculated as a statistic to describe the growth (or decrease)
of a natural population or cohort with some (inevitably fluctuating)
schedule of survivorship and fertility (Marti 1986, Woolhouse and
Chandiwana 1990a,b). The assumptions involved have been reviewed by
Loreau and Baluku (1987b).
Especially where reproduction is continuous, as it may be in tropical
populations of medically important planorbids, construction of complete
life tables is a daunting task. O’Keeffe (1985) sampled populations of
Bulinus globosus from two Kenyan reservoirs weekly over the course of
two years using ‘snail traps’: plastic sheeting framed in wire, dangling
from a float. He estimated size-specific fecundity from snails reared in
large, floating plastic boxes with mesh windows. O’Keeffe was able to
relate age and size in wild-caught snails using a series of cohort growth
curves specific to month of birth, which he derived from growth in his
boxes (for juveniles) and from mark–recapture studies (for larger snails).
His aim was to assign all snails in every sample to a week of birth and a
fecundity.
Population growth · 183
Figure 5.5. Monthly mean number of Bulinus globosus per trap (closed circles, left
axis) in Benes reservoir, Kenya, compared with the intrinsic rate of increase (per
week) ultimately posted by cohorts hatched on each date (open circles, right axis).
(Data are from O’Keeffe 1985.)
xmax
Data source Population R r (months)
Browne 1978 Viviparus georgianus,
New York
Pop. JAM 2.61 ⫺0.57b 36
Pop. SON 2.17 ⫺0.47b 36
Pop. TUL 1.38 ⫺0.20b 24
Pop. CAZ 0.34 ⫺0.68b 24
Brown et al. Lymnaea elodes, Indiana
1988 Pop. A (parasitized/not) 0.19/0.23 ⫺0.71/⫺0.62b 24
Pop. B 0.61/0.70 ⫺0.24/⫺0.021b 36
Pop. F 10.2/11.8 ⫺1.33/1.36a 13
Avolizi 1971 Sphaerium, New York
S. striatinum 0.78 ⫺0.10a 6
S. simile 0.24 ⫺0.90a 6
Notes:
a
per month. b per year.
Crowding
In the previous sections of this chapter we have seen that the reproduc-
tive potential of freshwater molluscs is generally quite high when grown
under conditions as benign as can be engineered by human hand, but that
in the wild, population growth rates are often zero or less. One explana-
tion for this phenomenon might be intraspecific competition.
It has long been evident that the effects of crowding on freshwater
mollusc populations are many and complex. In addition to the physical
competition for food and other resources one would ordinarily expect as
density increases, and beyond the general degradation of the environ-
ment by waste products and oxygen depletion, freshwater molluscs inter-
act chemically in ways that we humans only dimly understand. From his
observations that crowding effects in Bulinus forskalii cultures could be at
least partly extinguished by activated charcoal, Wright (1960) proposed
that crowded snails produce a ‘growth-inhibiting pheromone’. This was
but one year after P. Karlson and A. Butenandt coined the term
‘pheromone’ for use in insects. The 1970s and 1980s saw an explosion of
interest in chemical interactions among pulmonates, driven largely by the
Crowding · 187
Figure 5.6. Mean length (lines, left axis) and weekly fecundity (bars, right axis) of
15 Bulinus tropicus raised in three water treatments: unchanged (closed bars and
symbols), dialysed (crossed bars and symbols) and changed weekly (open bars and
symbols). (Data of Chaudhry and Morgan 1987.)
search for a new compound that might be useful against the intermedi-
ate hosts of trematode parasites. Inhibitory pheromone(s) do seem to
exist, and stimulatory ones as well, but their use against economically
important snail populations has not materialized.
Chaudhry and Morgan (1987) reared three groups of 15 juvenile
Bulinus tropicus in vessels containing 900 ml of gently aerated water, pre-
viously ‘conditioned’ in a fish tank, with chalk initially added as a source
of calcium carbonate. Snails were fed scalded lettuce ‘in excess’, and any
individual dying was replaced with a snail of similar size from a stock
tank. One group of snails received a weekly water change, the second
group was held entirely without exchange of medium, and in the third
group, the culture was dialysed against fresh (fish conditioned, carbonate
enriched) water.
Chaudhry and Morgan’s results (Figure 5.6) well illustrate the crowd-
ing phenomenon in freshwater pulmonates. Even at a (fairly low) con-
stant density, fed in excess, the Bulinus seem to be most adversely affected
by the absence of water exchange. Water analysis showed a dramatic
decline in the concentration of Ca⫹2 over 10 weeks in the unchanged
188 · Population dynamics and competition
tank (from 0.5 to 0.1 mmole/l), an increase in ammonia from 0.09 ppm
to 4.1 ppm, and a drop in pH from 7.2 to 6.5. There were small increases
in the concentrations of Na⫹, K⫹, and Mg⫹2. These various water quality
problems were corrected in the dialysed treatment, and apparently as a
consequence, both growth and reproduction showed significant
improvement.
The 15 Bulinus living in dialysed water nevertheless grew more slowly
than the 15 individuals whose water was entirely refreshed weekly. Some
material unable to pass through a dialysis membrane would seem to be
responsible. Levy et al. (1973) began the characterization of such a sub-
stance, from water in which stocks of Lymnaea cubensis had been cultured
for one month. They evaporated the culture water almost to dryness,
dialysed it against distilled water, then gathered its protein component
with ammonium sulphate precipitation. Levy and his co-workers called
the precipitate (redissolved, with ammonium sulphate removed) ‘inhib-
iting substance’ (IS), and demonstrated its striking negative effects on
growth and egg laying in both L. cubensis and Biomphalaria glabrata.
Inhibiting substance had no effect on planarians or fish, however. The
activity of IS appeared stable for some weeks at room temperature, but
was abolished by heating in solution at 85°C or trypsin digestion.
Thomas and Benjamin (1974) examined growth and reproduction in
B. glabrata over an unusually complete experimental design. They man-
ufactured their own ‘standard snail water’ (‘SSW’) from distilled, deion-
ized water and five reagent-quality salts, treated with pressurized air and
CO2. Into volumes of 25, 50, 100, 200, and 400 ml of SSW they placed
1, 2, 4, 8, or 16 young snails of uniform size in 15 combinations, gener-
ally 10 replicates of each combination. Water was changed every three
days, growth was monitored over 12 days, and reproduction (egg masses)
were totalled over 15–21 days. A clear reduction was apparent in both
growth and reproduction in the five highest (0.04) density cultures: 1/25,
2/50, 4/200, and 16/400. A proteinaceous inhibiting substance may have
been at least partly responsible. Thomas and Benjamin preferred to
hypothesize, however, that some chemical factors (particularly Ca⫹2)
were being rapidly depleted at densities of 0.04. The further experiments
of Thomas and his colleagues on Ca⫹2 dynamics in pulmonate cultures
will be featured in Chapter 8.
In addition to growth inhibition at high densities, Thomas and
Benjamin also felt that they saw some evidence of growth inhibition at
very low densities (1/400 to 2/200 snails per ml). They hypothesized that
the growth of (especially) juvenile snails might be promoted by some
Crowding · 189
chemical factor secreted by adults. In follow-up studies, Thomas et al.
(1975) found evidence of such growth-promoting factors in SSW con-
ditioned for three days with moderately high densities of adult B. glab-
rata (40 snails per 4 litres), fed lettuce. They called this ‘heterotypically
conditioned medium’, HetCM. Thomas and colleagues showed signifi-
cant improvement in the growth of juvenile (20 mg) B. glabrata reared for
12 days in HetCM, over that observed in plain, unconditioned SSW.
They fractionated HetCM using a series of membrane filters, and were
able to attribute improved growth to several substances of greater than
500 molecular weight units. Interestingly, some of the authors’ sub-
stance(s) also seemed to inhibit growth at some concentrations.
Vernon (1995) has found evidence that substances secreted by one B.
glabrata individual may augment the hatching success of eggs laid by
another, suggesting that the absence of such ‘social facilitation’ may at
least partially explain the poor reproductive performance of isolated
snails. One gets the impression that humankind’s chemical ignorance is
profound.
In an effort to remove food competition as a variable among treat-
ments, the authors of the crowding studies we have thus far discussed
sought to provide lettuce ‘in excess of requirements’ in all cultures.
Ensuring that food is always present does not, however, seem necessarily
to forestall competition for it. Mooij-Vogelaar and van der Steen (1973)
monitored two-week growth in sets of Lymnaea stagnalis cultures with
equal densities (two adults per 500 ml) but different quantities of lettuce.
The authors measured lettuce offered and consumed in units equal to 9
mm2, as judged on a transparent, gridded surface. Their ‘Experiment V’
returned an average growth increment of 3.22 mm at 551 units of lettuce
per snail per day, approximately the minimum ration totally consumed.
At four times minimum ration (2204 units daily), snails consumed an
average of 814 units (leaving uneaten 1390 units) and grew 6.73 mm.
And at 20 times minimum ration, snails consumed an average of 1069
units daily, left uneaten 9951 units, and grew 9.57 mm. It would appear
that large snails feeding on a single lettuce leaf cannot avoid interfering
with one another. Larger food volumes seem to result in less jostling, and
hence more feeding. A similar phenomenon was reported by Brown and
his colleagues (1994), rearing Physa virgata at various densities on peri-
phyton-covered tiles. The authors attributed reductions in grazing rates
to increased behavioural interference in crowded cultures.
It will be recalled that pulmonate snails have large capacities for sperm
storage, a single insemination typically furnishing allosperm sufficient to
190 · Population dynamics and competition
fertilize weeks of egg laying. There is evidence that the frequency at
which mating occurs in crowded cultures is far from adaptive, constitut-
ing rather a substantial interference to growth and reproduction. Van
Duivenboden et al. (1985) showed that oviposition rates in sets of 12
(adult, previously inseminated) Lymnaea stagnalis declined when grouped
and rose when isolated. Oviposition rates remained uniformly high,
however, when surgery was performed to cut the vas deferens of 12 snails
before grouping. Controls subjected to ‘sham surgery’ – operated upon
but vas deferens left intact – continued to display reductions in fecundity
associated with grouping. Van Duivenboden and his colleagues offered
some evidence suggesting that excessive copulation as female, rather than
as male, was more responsible for the reduction in observed oviposition
rates. Bohlken et al. (1987) and Bayomy and Joosse (1987) have also doc-
umented rapid increases in oviposition rate when adult pulmonates are
removed to isolation, and likewise attributed the phenomenon to a
reduction in copulation interference.
Crowding effects are not unique to pulmonates. They have been doc-
umented in most of the freshwater molluscs successfully cultured to date,
including the prosobranch Tricula (Liang and Kitikoon 1980) and the
pisidiid Musculium (Mackie 1976b). Paterson (1983) introduced sets of 1,
3, 5, 7, 9, and 12 Elliptio complanata into a respirometer of 20 litre volume,
monitoring respiratory rates over 12 hours. All animals exhibited elevated
respiratory rates initially, then equilibrated long before dissolved oxygen
was reduced to any level that might cause an adverse impact. The average
equilibrium rate was, however, significantly higher for single mussels than
for groups of three, and significantly higher for groups of three mussels
than for groups of five. Groups of 7, 9 or 12 mussels all equilibrated at a
uniformly low respiratory rate (about 0.2–0.3 mg O2/g/hr) that Paterson
characterized as ‘comfortable’. And when single mussels were placed in
chambers containing water previously conditioned by higher numbers of
individuals, they equilibrated at the low, ‘comfortable’ rate. Paterson
hypothesized that Elliptio release ‘some detectable substance’ which may
influence the metabolism of their neighbours.
Paterson preceded his physiological studies with an Ekman grab survey
of the natural distribution of his animals in Morice Lake, New
Brunswick. They were, in fact, aggregated; they were absent from about
59% of his (522.6 cm2) grabs but present in numbers as great as 13 in the
remainder. Paterson translated natural density to a ‘standard lower
density’ of about 3.4 mussels per his respirometer chamber. Hence he
considered that the lower respiratory rates he measured in crowded
Population regulation · 191
Elliptio a reflection of ‘comfort’, and the higher rates shown by isolated
individuals some indication of stress. But Paterson noted that just the
reverse might well be true; higher respiratory rates may indicate health.
The assumption that natural unionid aggregation is an adaptive response,
rather than the potentially deleterious consequence of microhabitat pref-
erences that must inevitably overlap, warrants further study.
We have thus far discussed crowding effects as though they were pri-
marily phenomena of artificial culture. But while crowding certainly is
most easily recognized and studied in the laboratory environment, it is
clear that natural populations of freshwater molluscs can, on occasion,
become crowded to such levels that growth and reproduction are strik-
ingly impacted. Berrie (1968) monitored growth rates and population
dynamics in a population of Biomphalaria sudanica tanganyicensis inhabit-
ing a small, closed pool near Uganda’s Lake Victoria. No more than two
metres across, the pool suffered a dramatic volume reduction in May
when local drainage operations seem to have lowered the water table.
Berrie estimated that its resident 1500 Biomphalaria were reduced to a
volume of 150 litres for five months, during which time there was vir-
tually no shell growth or reproduction. He extracted a large sample of
this water with activated charcoal, stripped the charcoal with various sol-
vents, and via bioassay attributed growth inhibition to a chloroform-
soluble ‘toxin’, perhaps an ester of molecular weight 360 (Berrie and
Visser 1963). In mid-October, Berrie (1968) removed 1490 snails from
the pool, returning but 53 individuals. Rain in the amount of 25 cm fell
during the next three weeks, and when Berrie resampled in early
November, he found that his much-reduced population of Biomphalaria
had posted substantial shell growth. Reproduction was well under way
by December.
Population regulation
Berrie’s studies of Biomphalaria (initiated in 1960, although not published
until 1968) were among the first in which experimental manipulations
were used to establish what factors (if any) may regulate the size of natural
freshwater mollusc populations. Subsequent investigators have taken a
variety of approaches: experimental manipulation of population density
or food supply in situ or in facilities designed to duplicate the natural en-
vironment. Our discussion here leads from such experimental studies to
more analytical treatments, where the question of population regulation
is examined through long-term observation. We will discuss the results
192 · Population dynamics and competition
of experiments especially designed to measure predation effects in
Chapter 7.
Experimental approaches
Researchers have found unionid clams to be especially suitable candidates
for direct manipulation of population density in the field. We have pre-
viously mentioned the studies of Hanson et al. (1988) involving Anodonta
grandis caged at various depths in Narrow Lake, Alberta, over the course
of a growing season. Although depth does seem to affect growth rate,
Hanson et al. could see no effect of stocking densities that varied an order
of magnitude, over the range of depths naturally occupied by the animal.
Kat (1982a) established square-metre plots on the bottom of Norwich
Creek, Maryland, and placed marked, measured Elliptio complanata at
densities ranging from 25% to four times normal. His recovery rates were
quite good a year later, generally in the 70–80% range. Analysis was hin-
dered by small sample sizes, especially given the complication that mussels
of different ages are expected to grow at different rates. A series of sep-
arate one-way analyses of variance (rather than analysis of covariance)
nevertheless suggested that growth rates were inversely related to stock-
ing density.
Snail population densities are generally manipulated using cages.
Aldridge (1982) could not see an effect of adult density on the fertility
of Leptoxis carinata caged in a New York river, although artificial removal
of all Leptoxis eggs prior to their hatch did stimulate increased egg laying
in both low and high density treatments. Aloi and Bronmark (1991)
monitored three-week shell growth in Lymnaea emarginata held at four
densities in flow-through chambers located by Trout Lake, Wisconsin.
The chambers were opened to colonization by periphyton two weeks
prior to the beginning of the experiment. Both periphyton biomass and
primary productivity decreased rapidly upon the introduction of the
snails, to their effective detection limits. The growth rates ultimately
achieved by the snails appeared to be a direct function of their stocking
densities. A limitation of both the studies of Aldridge and Aloi and
Bronmark is that it is difficult to relate their experimental densities to the
density of snails actually observed in the field.
Turning to experiments involving both density manipulation and
resource augmentation, we come to one of the few studies involving
freshwater molluscs that may truly be considered influential to the study
of ecology as a whole. Eisenberg (1966) constructed large (4.23 m2) pens
on the exposed, frozen soil surrounding a small Michigan pond. He
Population regulation · 193
Figure 5.7. Demographic data from the cage experiments of Eisenberg (1966). The
left axis is the average density (per 4.23 m2 pen) of adult Lymnaea elodes in the spring
(several June estimates) and in the final autumn census. The right axis shows the
average number of young-of-the-year juveniles (per 80 cm2 sampler) counted over
2–3 July weeks.
enclosed his pens in May, by which time pond levels had risen and a large,
resident population of Lymnaea elodes had emerged from aestivation.
Using several techniques in tandem, he estimated the numbers of snails
enclosed in each of 12 pens, and in June, adjusted densities such that four
pens held approximately five times their normal density, and four pens
held only 25% normal density. To two other pens he added packages
of frozen spinach at three or four-day intervals. (No initial estimates of
density were made in these two pens.) He estimated the numbers of
young-of-the-year snails in all 14 pairs on several dates in July, and re-
estimated the densities of old adults in September.
Eisenberg found very little mortality in his enclosed populations of
adult snails from May to September, even at high density (Figure 5.7).
But regardless of the size of the resident population of adults, he found
roughly the same numbers of juveniles born in all 12 pens of his density
experiments. Average adults were thus much less fecund at high densities
than they were at low. Fecundities showed further improvement in the
pens to which Eisenberg added spinach. It would appear that the size of
the Lymnaea elodes population in this small pond is regulated by food,
acting through fecundity. In a series of follow-up experiments, Eisenberg
(1970) confirmed his earlier results, adding observations on somatic
194 · Population dynamics and competition
growth. He reported that snails living at low density grow more rapidly
than those at high density, and those whose normal diet is augmented
with spinach grow more rapidly still.
Gilbert et al. (1986) also varied both density and feed in their cage
study of the Bithynia tentaculata population inhabiting the St Lawrence
River, Quebec. Their experiments involved three starting densities (10,
30, and 90 adults per cage) and two levels of food – unsupplemented and
supplemented to excess with a mixture of lettuce, spinach, and chicory
leaves. Assessed at the end of a three-month season, the authors could
distinguish no effect of food level on the numbers of offspring produced
per parent. The 10 adult treatment yielded high numbers of relatively
large offsping, the 30 adult treatment yielded lower numbers of large off-
spring, and 90 adults yielded high numbers of small offspring. These
results seem to reflect crowding in both parental and offspring genera-
tions.
‘Cage effects’ are a general problem with studies of this kind. In spite
of the best efforts of researchers, the environment inside a cage may bear
little resemblance to the environment outside. It is odd, for example, that
adult mortality seemed to have been nearly negligible inside Eisenberg’s
cages over four months, while outside the cages it seems to have been
over 90% (to judge from Eisenberg’s figure 1). Cages certainly provide
unusual protection from predation, as we shall see in Chapter 7, but
Eisenberg’s own efforts to assess this effect experimentally proved incon-
clusive. Thus the resource augmentation study of Osenberg (1989),
involving both caged and uncaged controls, provides an especially
welcome insight.
Lawrence Lake, Michigan, is home to eight species of freshwater snails:
three populations of small, numerically dominant prosobranchs (Amnicola
limosa, Marstonia lustrica and Valvata tricarinata) and five populations of pul-
monates (two Gyraulus, two Helisoma, and a Physa). Osenberg (1989)
established eight 2.5 m2 stations at about one metre depth and applied
four treatments: caged (surrounded by a wood frame supporting 1 mm
mesh), fertilized (with resin-coated phosphate pellets on wooden
dowels), caged-and-fertilized, and control. His experiment ran from
mid-August to mid-September, after which Osenberg quantitatively
sampled both snails and macrophytes. Although the experiment may have
been too brief to return significant density effects for most snail popula-
tions, fertilized treatments did show greater numbers of Gyraulus parvus
and newborn Physa. Snails of most populations also seemed to grow more
rapidly where nutrients were applied. Cage effects generally were not sig-
Population regulation · 195
nificant. But the highest total snail biomasses (and macrophyte biomasses)
were recorded in the caged, fertilized treatments, an effect Osenberg
attributed to higher nutrient retention where water circulation was
reduced. Here it should be kept in mind that macrophytes may serve as
habitat, cover, and substrate for egg laying as well as food; treatments that
benefit macrophytes benefit snails in more ways than one.
Among the most effective methods of eliminating cage effects (but not
problems of interpretation) is not to use cages. Repeated drainage and
refilling of a fish pond in northeastern Tanzania had complex and unpre-
dictable influences on its snail fauna (Pringle and Msangi 1961, Pringle
and Raybould 1965). Hershey (1992) reported the effects of five years of
artificial fertilization upon the benthos of an entire (1.8 ha) lake in arctic
Alaska. The four snails present, Lymnaea elodes, Valvata lewisi, a Physa and
a Gyraulus, typically ‘live for several years and appear to take 2–3 yr to
reach adult size’ in this extreme environment. In late June, 1985, a fibre-
glass-reinforced plastic curtain was used to divide the lake in half. Then
from 1985 to 1990, the western sector was continuously fertilized (with
both N and P) using a solar-powered pump, to about five times its normal
annual nutrient loading. Midsummer Valvata densities were estimated
from quantitative samples taken at 4–5 m on both sides of the partition,
and adult Lymnaea were counted by divers swimming measured transects.
Results are shown in Figure 5.8.
One must not expect rapid responses from arctic molluscs. By
1987–88, however, the density of adult Lymnaea was significantly greater
in the fertilized half of the lake than in the unfertilized control. Data on
water quality and productivity are not at hand, but it appears that fertil-
ization effects were building; divers were unable to swim transects in 1989
because of reduced visibility due to an algal bloom. Although adult
Lymnaea densities had dropped in the fertilized half of the lake by the
summer of 1990, Hershey began to notice increased densities of juve-
niles. The Valvata population did not post a significant increase until
1989, as the Lymnaea population seemed to be falling. Although artificial
fertilization of their habitat will clearly affect freshwater snails in many
ways, both populations certainly reacted as though they were food-
limited. Hershey noted that the density response of Valvata ‘is consistent
with the cascade concept’, as increased resources are first expected to
benefit the Lymnaea grazing the shallows, then later the detritivorous
Valvata of deeper water.
It is certainly not true, however, that the addition of fertilizer to any
freshwater habitat will necessarily benefit resident molluscs. Daldorph
196 · Population dynamics and competition
Figure 5.8. Densities of Valvata and adult Lymnaea on the fertilized side of an arctic
lake (closed circles), compared with densities on the unfertilized control side (open
circles). (Data of Hershey 1992.)
Perturbations
Over against whatever evidence might be marshalled for density depen-
dence in mollusc populations, the literature contains numerous and
sometimes spectacular examples of density-independent effects.
We noted in Chapter 3, for example, that the relationship between
certain snail populations and macrophytes seems to be an especially close
one. Lodge and Kelly (1985) documented a ‘summerkill’ event in the
small (1 ha) Radley Pond, associated with warm weather and anoxic con-
ditions. The submerged macrophyte population suffered greatly, as did
populations of two snails living upon them; 99% of the Lymnaea peregra
population and 35% of the Valvata piscinalis population died. Heywood
and Edwards (1962) were on hand when macrophytes seem to have been
eliminated from England’s River Ivel. They attributed a 78% drop in the
summer density of Potamopyrgus jenkinsi to this calamity. Drought is
another widely cited perturbation. During 1975–76, only about half the
normal precipitation fell on the River Roding catchment in Essex,
England. Flows were reduced to negligible at some sites, insufficient to
mix and dilute urban sewage. Extence (1981) reported a significant
decrease in the mean annual density of the Roding Potamopyrgus popu-
lation, which he attributed both to stranding and to organic pollution.
Effects upon other species were mixed; pisidiids (apparently more than
one species) seem to have significantly benefited from the drought.
All too often, it is simply impossible to separate fluctuations in popu-
lation size due to natural exigencies from those due to the interference
of man. Most of the long-term declines in populations of freshwater
molluscs documented to date have been attributed to human alteration
of the environment (e.g. Covich 1976, Eckblad and Lehtinen 1991,
Lewandowski 1991, Nalepa et al. 1991). Shaw and Mackie (1989, 1990)
200 · Population dynamics and competition
reported that the decline (or extinction) of Amnicola limosa in a variety of
Canadian lakes seemed to be a function of pH. They found some popu-
lations inhabiting lakes so close to their physiological tolerance limits for
acidity that adult size (and hence fecundity) seemed to be affected. But
whether the apparent extinctions are natural, or due to the phenomenon
of acid rain, cannot be determined.
Payne and Miller (1989) offer an example of a perturbation (of uncer-
tain nature) in a freshwater mollusc population that gives every appear-
ance of being natural. Using , the authors took 17–24 samples of
0.25 m2 each from a gravelly shoal on the lower Ohio River. Although
the 256 Fusconaia ebena they collected in 1983 ranged from 1 mm to 90
mm in standard shell length, 71% of all individuals were clustered
between 12.8 mm and 19.5 mm. The authors identified this as a single
cohort born in 1981, distinguishing much smaller cohorts born in 1979
(6%) and 1982 (3%). They were easily able to follow the 1981 cohort in
samples from similar dates in subsequent years. It again comprised 71%
of 269 mussels sampled in 1985, and 74% of 219 mussels sampled in
1987, due to lack of strong subsequent recruitment. Payne and Miller
offered no explanation for the exceptional reproductive success of their
Fusconaia population in 1981. Perhaps such dynamics are not unusual in
populations of individuals where the average longevity is 11–18 years.
Chief among perturbations must be the new colonization of a previ-
ously unoccupied habitat. On a number of famous occasions, such ‘col-
onizations’ have come at the hand of man, and the consequences have
been unfortunate from our perspective. Worldwide zebra mussel inva-
sions have been reviewed by Kerney and Morton (1970), Stanczykowska
(1977), Mills et al. (1996), Ram and McMahon (1996) and Karatayev et
al. (1997), Corbicula invasion by McMahon (1983b) and Counts (1986),
Potamopyrgus by Bondesen and Kaiser (1949) and Ponder (1988a), and the
invasions of other snails by Madsen and Frandsen (1989). For the present
purposes, the artificially mediated colonizations of such pests as Dreissena
and Corbicula may serve as models of processes which must, at some lower
frequency, occur naturally.
The people at Glen Lyn Power Plant, on the New River at the
Virginia/West Virginia border, were not surprised to discover that
Corbicula had ascended the Ohio and Kanawha Rivers to threaten their
water works. Their first quantitative samples (October, 1976) contained
no individuals greater than 10 mm (standard shell length), clearly all
young-of-the-year (Rodgers et al. 1977). Population densities of
18–29/m2 suggested to the authors that (unsampled) parents may have
Population regulation · 201
Figure 5.9. Densities of Corbicula fluminea at Glen Lyn Power Plant, Virginia
(station 4, thermally influenced). (Data of Graney et al. 1980.)
Long-term trends
It should be apparent from the previous discussion that freshwater
mollusc populations are often subject to density-independent perturba-
tions, and that the existence of such perturbations does not disprove
density-dependent control. The key is to examine population sizes over
sufficient scales of time and space.
The best available record of freshwater mollusc population density
comes from the Pisidium fauna of Denmark’s large (17.3 km2), temper-
ate, eutrophic Lake Esrom. Beginning in 1953, and extending irregularly
to 1973, P. M. Jonasson collected 5 or 10 bottom hauls (250 cm2 each)
using an Ekman grab in the deepest spot in the lake (22 m – the ‘Endrup’
station). An additional set of samples was taken in 1977, and an account
of the entire data set was offered by Holopainen and Jonasson (1983). As
one might expect, the environment at 22 m seems to be fairly constant
among years, although within years there are regular thermal cycles
(about 2–15°C) and cycles of summer oxygen depletion. Yet Holopainen
and Jonasson reported that the annual abundances of the three Pisidium
species at Endrup seemed to vary substantially – P. subtruncatum and P.
casertanum swapped dominance twice as P. henslowanum was reduced to
(local) extinction.
Figure 5.10 shows summer densities for the three Pisidium species at
the Endrup station. To construct this figure I averaged the (1–3) records
collected for June, July, or August from the larger data set kindly provided
by I. J. Holopainen. Summer data were not available for 1953 or 1968.
Holopainen and Jonasson could offer no explanation for the large popu-
lation explosions in 1955, nor for the 1956–1959 declines. The only sub-
stantial perturbation identified by the authors over the entire 24-year
Population regulation · 203
Figure 5.10. The average summer densities of three Pisidium species at the ‘Endrup’
station of Lake Esrom, Denmark. (Data of Holopainen and Jonasson 1983.)
Notes:
* P⬍0.05
**P⬍0.01
Sources: The values of t are from PBLR tests for density dependence (Dennis and
Taper 1994). The ‘Endrup’ and ‘Tumlinghus’ data were taken in the summer at
two different stations in Denmark’s Lake Esrom (Holopainen and Jonasson 1983,
1989b). Ancylus data are from Russell-Hunter (1961a).
Figure 5.11. Annual estimates of density in five freshwater snail populations, stan-
dardized by statistics cited in the text. Pomacea paludosa data are from Kushlan (1975),
and Hydrobioides manchouricus data are from Chung et al. (1980). Lymnaea peregra,
Physa fontinalis and Ancylus fluviatilis data are from Russell-Hunter (1961a).
Interspecific competition
In our review of the diet and habitat of freshwater molluscs through
Chapters 2 and 3 we found both extraordinary diversity and remarkable
uniformity. Certainly the habitats of planorbids and lymnaeids often
differ; for example, many populations of the latter are much more amphi-
bious. But their diets are quite general, and broadly overlap. And if some-
times lymnaeids and planorbids may be spied jostling each other on the
same blade of grass, surely the same regulatory phenomena we have doc-
umented within species in the present chapter thus far ought to apply
between species. In the following sections we will review the evidence
of competition between populations of freshwater molluscs in natural sit-
uations, as well as where artificial introductions may have brought the
phenomenon into more vivid focus.
Natural situations
Some pairs of closely related species seem to show strikingly discordant
local distributions, as though they exclude one another. For example,
Lymnaea peregra and L. auricularia have similar geographic ranges in Great
Britain but are rarely found together. In those unusual habitats were they
do co-occur, there seems to be an inverse relationship between their tem-
poral abundances. Adam and Lewis (1992) reported a striking example
of L. auricularia replacing (outcompeting?) L. peregra in an English gravel-
pit lake.
It is perhaps not surprising that among the schemes for the biological
control of medically important planorbids are various ideas for the pro-
motion of interspecific competition. Biomphalaria straminea, for example,
is not as effective a vector of schistosomiasis in tropical America as is
Biomphalaria glabrata. The former species has replaced the latter on several
occasions, both in long-term laboratory studies and in the wild (Barbosa
208 · Population dynamics and competition
1973, Barbosa et al. 1984, Guyard et al. 1986). It is not clear whether the
phenomenon is due to competition or hybridization, but the end result
seems to be desirable.
The potential for competition between quite diverse freshwater mol-
luscs is well illustrated by the relationship between the Alaskan popula-
tions of Lymnaea elodes and Valvata lewisi we have touched on previously.
The tiny, delicate prosobranch, with a penchant for feeding on fine
deposited detritus, reaches mean densities of 1000/m2 on the deeper sed-
iments of Toolik Lake, but may also be found (at about 80/m2) in the lake
shallows (Hershey 1990). The large pulmonate Lymnaea, with its strong
jaws and formidable radula, grazes at about equal densities (80/m2)
throughout the lake. It will be recalled that the artificial fertilization study
of Hershey (1992) suggested that populations of both these species
seemed to be food limited (Figure 5.8). Hershey (1990) caged four
Valvata either with or without a Lymnaea on both sides of her artifically
fertilized lake. After 12 days, Valvata enclosed on the fertilized side had
laid significantly more eggs than those enclosed on the unfertilized side,
and those without Lymnaea significantly more than those with. The latter
result is unsurprising, since the addition of a single Lymnaea probably
more than doubled the snail biomass depending on the resources of an
800 cm3 cage. Hershey also conducted a more extensive set of laboratory
experiments (at lake temperatures) confirming the negative impact of
Lymnaea on Valvata fecundity, although again, aquaria with both species
supported higher biomass than aquaria with Valvata alone. Interestingly,
the effect of Valvata on Lymnaea fertility seemed to be positive. Hershey
concluded by suggesting that trout predation may have greater impact
upon the Lymnaea populations of arctic lakes than upon the Valvata pop-
ulations, facilitating their continued coexistence.
Competition experiments are more easily performed if the species
compared are of roughly equal size. Four larger pulmonates are com-
monly found in the ponds of the Crooked Lake Field Station in north-
ern Indiana: Helisoma trivolvis, Aplexa hypnorum, Physa gyrina, and
Lymnaea elodes. Brown (1982) calculated pairwise MacArthur–Levins
niche overlap statistics in this community of four, based on the duration
of their reproductive periods, their habitat preferences (quantitative
samples of eight lakes varying in permanence and cover) and their diet
(laboratory experiments). We will explore Brown’s assumption that high
habitat overlap implies increased competition in Chapter 9. In any case,
his analysis suggested that the greatest likelihood of competition might
be between Physa gyrina and Lymnaea elodes.
Interspecific competition · 209
Figure 5.12. Nine-week production of egg masses for Physa gyrina and Lymnaea
elodes in containers floating in a permanent pond (‘F’) and a temporary pond (‘A’).
The treatments within each pond (two of each, four of each, mixed two-and-two)
are explained in the text. (Data of Brown 1982.)
Figure 5.13. Densities of bivalves (N/m2) in the Altamaha River, Georgia, October
1971 to October 1975 (Gardner et al. 1976).
Among the best observations are those of Boozer and Mirkes (1979) on
the co-occurrence of Corbicula and Musculium partuneium in a sedimenta-
tion basin at a nuclear plant in South Carolina. They reported that a dra-
matic increase in the number of Corbicula on the floor of the basin was
associated with a small decrease in the abundance of Musculium. At the
same time, the population of Musculium attached by byssal thread to the
wall of the basin increased dramatically. (Corbicula was very rarely found
bysally attached to the wall.) Although the evidence is anecdotal, it is cer-
tainly consistent with displacement of Musculium by Corbicula.
We have the Georgia Power Company to thank for the wealth of
information available on the bivalves of the middle Altamaha River. The
quantitative unionid samples taken by Sickle in 1968 (but not published
until 1980) will be discussed in Chapter 9. The first report of Corbicula
in the region was made in 1971. Beginning in October of that year, and
continuing for four years thereafter, Gardner et al. (1976) collected quan-
titative samples at six-week intervals from a number of sites with a
Petersen dredge. At the start, densities of native bivalves (primarily pisid-
iids, but with at least 4 or 5 unionid species) were about 200/m2. But
Figure 5.13 shows that the results of the 1972 and 1973 samples (months
3–27) were spectacular, even on a log scale. In the space of three years,
Corbicula densities rose to about 10000/m2, while native bivalves disap-
peared. Gardner’s descriptions of subsequent qualitative samples indicate
212 · Population dynamics and competition
that the disappearance of unionids, at least, is not a sampling artifact.
The authors mention no environmental variation, such as pollution or
development, that might have influenced these results. These data stand
as, admittedly circumstantial, evidence of competition.
To my knowledge, however, there are no additional data or observa-
tions supporting competition between North American bivalves and
Corbicula in the fifty-year history of the introduction. From measured fil-
tration rates, water flow rates, and population size, Leff et al. (1990) esti-
mated that Corbicula could significantly reduce the concentration of food
particles suspended in a small South Carolina stream. This is very much
in accordance with the earlier findings of Cohen et al. (1984), Foe and
Knight (1985), and Lauritsen (1986b). But their extensive field survey
failed to show any apparent affect of Corbicula on the native Elliptio com-
planata population, in distribution, abundance, or individual size. Miller
and Payne (1994) found no significant correlations (negative or positive)
between Corbicula density and the density of native unionids in 403
quadrat samples taken over several years in the gravelly shoals of the Ohio
River. In general, it has been concluded that Corbicula is not a threat to
native North American bivalves (McMahon 1983b).
Helisoma duryi
Interest in the potential of the large North American Helisoma duryi as a
competitor against medically important planorbids dates to 1941. As may
be gathered from Figure 5.3, Helisoma populations have greater intrinsic
rates of increase than typical Biomphalaria populations at warm tempera-
tures, at least under the favourable food and water conditions provided
by De Kock and Joubert (1991). Joubert and De Kock (1990) recorded
values of r⫽2.34 for H. duryi and 1.69 for B. pfeifferi in large aquaria at
26°C , fed and changed daily. The authors saw no competitive effects
between the two species when reared together (r⫽2.12 and r⫽1.60,
respectively), as long as water was changed constantly and eggs removed.
However, it has been well established that H. duryi has a strikingly neg-
ative effect on the growth and reproduction of African Biomphalaria and
Bulinus when the environment is more restricted (Frandsen and Madsen
1979). Detailed experiments involving different densities of Helisoma and
Biomphalaria, reared together, separately, in partitioned aquaria and in
variously conditioned waters, have tended to rule out allelopathic chem-
icals (Madsen 1979a,b, 1982, 1984). Such chemical inhibition as has been
described appears to be attributable to non-specific pollutants and
perhaps competition for inorganic ions (Madsen 1987).
Nor in fact is the markedly negative effect of H. duryi due exclusively
to competition for food. Lassen and Madsen (1986) reared Helisoma and
Bulinus together and separately in small cages immersed in a single large
aquarium, agitating and changing water regularly to remove any possible
effect of fouling or inhibiting chemicals. Food (dog food plus lettuce) was
supplied at three levels: less than the amount consumed daily, daily con-
sumption, and in apparent excess of daily consumption. As expected,
they found strikingly lower growth and fecundity in Bulinus caged
together with Helisoma than in Bulinus caged with an equal number of
conspecifics. In contrast, eight Helisoma showed improved growth and
fertility caged with eight Bulinus than with eight other conspecifics. The
negative effect of Helisoma on Bulinus remained striking even when
excess food was left floating on the water all day. Excess food did seem
Interspecific competition · 215
to relieve a little of what appeared to be egg predation by Helisoma,
although its negative effect on Bulinus was still significant. Further, con-
firmatory experiments of Meyer- Lassen and Madsen (1989) also pointed
strongly to egg predation.
Such results prompted Madsen (1986) to perform a number of behav-
ioural experiments comparing the two species. Helisoma seems generally
to be more active, running into other snails more often and able to find
food faster. But perhaps as importantly, Bulinus seem to be more adversely
affected by collisions with Helisoma than with conspecifics, withdrawing
into their shells rather than simply turning aside. It is possible that
Helisoma is directly antagonistic.
El-Emam and Madsen (1982) found that laboratory populations of H.
duryi, B. truncatus, and Biomphalaria alexandrina all showed optimal growth
at 26–28°C and optimal reproduction on the same artificial diet. They
noticed only minor differences in tolerance of extreme temperatures,
starvation, and darkness. Unfortunately, their habitat preference may not
be especially similar in the field. In 1972, H. duryi was introduced into a
sugar plantation in Tanzania, where Biomphalaria, Bulinus, and Lymnaea
were widespread. But Madsen (1983) reported that it was subsequently
restricted to a very small area where (perhaps not coincidentally)
Biomphalaria was not present. Madsen noted that the canals are regularly
treated with molluscicide, and one wonders how much of the problem
may be Helisoma’s ability to disperse and colonize.
Joubert and his colleagues (1992) were unable to introduce viable pop-
ulations of H. duryi into small ponds on the South African lowveld. A few
wild populations of Helisoma have become established about Africa (De
Kock and Joubert 1991, Brown 1994), but by comparison with the
immediate and rapid spread through Africa shown by such exotics as
Physa acuta and Pseudosuccinea columella, this is unimpressive. It is also
important to note that H. duryi shares the New World with Biomphalaria
glabrata, with little evidence of competition between the species. Perera
et al. (1986) noted that Cuban populations of H. duryi do not seem to
multiply as quickly as do Biomphalaria, nor do they seem to survive sea-
sonal drying. Thus the prospects for H. duryi as a biological control agent
do not seem promising.
The ampullariids
In Chapter 3 we noted that ampullariid populations can, on occasion, be
prodigious consumers of macrophytes. The diet of Biomphalaria is more
typical of freshwater snails generally: certainly including macrophytes but
216 · Population dynamics and competition
Figure 5.14. Mean percentage of plant tissue eaten (with standard error) in macro-
phyte choice tests involving ten each Biomphalaria glabrata (‘Biomph.’), juvenile
Marisa cornuarietis (‘J Marisa’), and adult M. cornuarietis (‘A Marisa’). See text for key
to the seven macrophyte species. (Data of Cedeno-Leon and Thomas 1982.)
The thiarids
Of all the biological agents offered for the control of medically impor-
tant pulmonates, Thiara granifera and Melanoides tuberculata have in recent
years displayed the most promise for lotic waters. The former species, a
native of the Far East, began to be widely reported in Caribbean regions
in the 1950s and 1960s. The latter species, initially an inhabitant of the
milder climates of the Middle East and Africa, seems to have spread
through the New World tropics about ten years later. As the two thiar-
ids have spread, workers have noted striking declines and even extinc-
tions of Biomphalaria and other pulmonates. Pointier and McCullough
220 · Population dynamics and competition
(1989) have reviewed independent reports of this phenomenon from
Puerto Rico, Venezuela, the Dominican Republic, Cuba, Martinique,
Dominica, and Guadeloupe.
Pointier et al. (1994) provide a good set of example data from 29 small,
rocky streams running north into the Gulf of Mexico on the coast of
Venezuela. Prior to the arrival of T. granifera and M. tuberculata in 1975,
B. glabrata inhabited seven of these rivers. But occasional, qualitative
surveys conducted in 1975–90 have documented the spread of the thiar-
ids to all 29 streams, and the disappearance of Biomphalaria. The compli-
cating factor here (as throughout the Caribbean) is that all seven of the
Biomphalaria populations have been treated with molluscicides.
By the 1980s Pointier et al. (1989) reported that B. glabrata had been
all but extinguished from the island of Guadeloupe, coincident with the
spread of Biomphalaria straminea. But a few populations remained in
shallow, flowing waters managed for watercress culture. Pointier and his
colleagues collected 60–70 quadrat samples of 0.05 m2 each from each of
two watercress beds, two or three times yearly from 1981 to 1987. Both
B. glabrata and B. straminea were common, at least initially. So in January
1983, Pointier introduced 700 Melanoides into each pond, for a calculated
initial density of just over 1/m2. Within two years, Melanoides had peaked
at 9941/m2 in one pond and 13388/m2 in the other, and both
Biomphalaria populations had disappeared. Figure 5.16 (upper) shows
population density data from one of these watercress ponds. A larger set
of follow-up studies confirmed the negative impact of Melanoides on B.
glabrata in Guadeloupe watercress beds, but yielded no evidence of an
effect in smaller cattle ponds or in Colocasia fields subject to irregular
drying (Pointier et al. 1993).
At this point the reader might well thumb back to compare Figure 5.1
and Figure 5.2, note the scales on their abscissas, and marvel that a pop-
ulation with r⫽0.024 might outcompete a population with r⫽0.84.
However, Melanoides seems to be better adapted to stable environments
than Biomphalaria. The Guadeloupe watercress beds afforded constant
water supply, constant food, and very small temperature fluctuations
(3°C annually). Pointier notes that Melanoides ‘is able to maintain very
high densities for a long time’, that is, intraspecific competition does not
seem especially severe. And as long as the water is well oxygenated,
Melanoides shows remarkable resistance against pollution, such as one
might expect in rich beds of watercress. Finally, it is my guess that at least
occasionally, these beds of watercress were harvested, and that
Interspecific competition · 221
Figure 5.16. The upper graph shows the densities of Melanoides tuberculata and
Biomphalaria glabrata in a watercress bed on Guadeloupe from the beginning of 1982
until mid-1987 (data of Pointier et al. 1989). The lower graph shows the relative
abundance of M. tuberculata and Biomphalaria pfeifferi in Kenya’s Matingani Stream
from early 1987 until late 1989. (Data of Mkoji et al. 1992.)
222 · Population dynamics and competition
Biomphalaria populations suffered greatly, but that the more benthic
Melanoides tended to be left undisturbed.
I am aware of only one experiment matching a thiarid against
Biomphalaria under controlled conditions. Gomez et al. (1990) designed
an elaborate experiment with four controls and nine treatments, involv-
ing 3 gallon or 1.5 gallon aquaria, 5 or 10 snails, and B. glabrata and Thiara
granifera either together or partitioned by a fine screen. The snails were
fed powdered rat food, which must have rapidly settled and tended to
foul the water. Oddly, the authors make no mention of aeration or water
change over the 12-month duration of the experiment. I should assume
that the tanks were aerated, and that water must have been changed occa-
sionally, perhaps irregularly. Neither Thiara nor Biomphalaria reproduced
during the first five months of the experiment, quite normal for the
former but rather unusual for the latter. Juveniles remained in the aquaria,
competing against their parents and each other. At the end of 12 months,
populations had grown about tenfold, probably to the carrying capacity
of the tanks. In all cases, even when the two populations were separated
by screens, the densities of Thiara were strikingly greater than those of
Biomphalaria. Gomez attributed the superiority of Thiara to ‘chemical
factor(s), possibly water-soluble pheromones.’ A second interpretation
would be that Thiara does better than Biomphalaria in aerated water fouled
with rotten rat food.
Finally, it is important to note that Melanoides populations native to
Africa peacefully coexist with an extensive pulmonate fauna, including
Biomphalaria pfeifferi. In fact, Mkoji et al. (1992) reported positive associ-
ations between the abundances of Melanoides and Biomphalaria in the
Sudan and Nigeria. They monitored snail population densities at four
sites in Kenya over three years. At each site, Mkoji and his colleagues
established six stations, at which three collectors searched for ten minutes.
The densities shown in the lower half of Figure 5.16 are counts over
6⫻10⫻3⫽180 person-minutes at one of these four sites, Matingani
Stream. Their results at the other three sites were the same. There was no
tendency for native Melanoides populations to replace or outcompete
established pulmonate populations. In fact, Melanoides and Biomphalaria
(as well as Bulinus tropicus and Lymnaea natalensis, not shown) seemed to
fluctuate together. Mkoji pointed out that native African populations will
have, by now, had the opportunity to adapt to one another, so that any
competitive displacement will have occurred long ago. Also, the sites they
studied were all what one would consider typical pulmonate habitat:
subject to considerable fluctuation in volume and shallow only at narrow
Summary · 223
littoral zones. Pointier and McCullough (1989) offer the following
summary of the prospects for biological control with thiarids: favourable
in the neotropics if waters are permanent, shallow, well oxygenated, and
free of floating plants.
Summary
The intrinsic rate of increase (r) has been estimated under benign labor-
atory conditions for a broad range of planorbids, physids, and lymnaeids.
Although comparisons are complicated by differing details of culture,
populations of all three pulmonate families typically demonstrate r⫽0.5
to r⫽2.0 per fortnight, with values ranging to r⫽3.0 on occasion. This
is to say that given suitable physical conditions, an excess of food, and
protection from competition and predation, typical pulmonate popula-
tions may be expected to increase by factors of 1.6 to 7.4 every 14 days.
Observed intrinsic rates of increase show a good deal of variance. They
seem very sensitive to temperature for some species, while other species
seem unaffected over broad ranges. It is difficult to know whether the
populations in this latter situation are reproducing at their physiological
limit, or whether they are limited by some aspect of their environment
unrelated to temperature. Although the (probably low) levels of inbreed-
ing experienced under typical laboratory culture do not seem to influ-
ence values of r, an artificially imposed mating system (whether selfing
or outcrossing) may have a profound effect.
I am aware of only two reports of the intrinsic rate of increase for pro-
sobranch populations. Both these values are much lower than those typ-
ically posted by pulmonates. No such data are available for any bivalve.
Although r was originally conceived as a parameter of the logistic
growth equation, it has found use as a statistic to describe the instantane-
ous status of natural populations under conditions neither constant nor
benign. Tropical cohorts of pulmonates typically exhibit r⬍1.0 per fort-
night in the wild, and often show r⬍0.0. Fewer data are available for tem-
perate populations (surprisingly), but they seem comparable to values
from tropical pulmonates. Among the many candidate explanations for
the lower values of r observed in the wild, one must include the effects
of competition.
Competition for food seems to occur as jostling and disturbance in
crowded laboratory cultures, even when food is made available ‘in
excess’. Excessive mating may also constitute a measurable disturbance in
dense snail cultures, while waste products build up and available calcium
224 · Population dynamics and competition
is rapidly depleted. In addition, workers have been most impressed with
the evidence of chemical interactions among snails cultured at various
densities. Crowded pulmonate snails seem to produce substances that
inhibit growth and egg laying in conspecifics, and growth-promoting
factors may be produced in uncrowded situations. Although best docu-
mented in pulmonates, there is evidence of chemical messages passing
through crowded populations of prosobranchs and bivalves, in the wild
as well as in culture.
Experiments on the regulation of freshwater mollusc populations in
the wild have generally involved the use of pens or cages. Workers have
manipulated population density, directly adjusted food availability, or
indirectly augmented food by the use of fertilizers. Of approximately ten
such studies reviewed in this chapter, about half returned at least some
evidence of food limitation in freshwater molluscs. The famous work of
Eisenberg (1966) on Michigan Lymnaea, followed by the large-scale fer-
tilization study of Hershey (1992) on Alaskan Lymnaea and Valvata and
the thorough food and density manipulations of Hill et al. (1992, 1995)
involving Tennessee Goniobasis, leave little doubt that many natural pop-
ulations of freshwater molluscs are under density-dependent control.
Workers also often report striking (presumably natural) fluctuations in
mollusc population densities (Brown 1994: 487–95). These are often
associated with identifiable perturbations of the environment, such as
drought, spate, or vegetation summerkill, although sometimes the cause
remains obscure. Among the most spectacular fluctuations have been
demonstrated by invading populations of Corbicula and Dreissena in North
America, which may follow initial logarithmic growth rates with popu-
lation declines, and then perhaps stability.
The literature does not contain a great wealth of long-term data on
the density of freshwater mollusc populations. An inspection of five snail
population data records of greater than five years duration (2 proso-
branchs, 3 pulmonates) leaves the impression of some stability, however.
Density dependence can be distinguished in a series of annual censuses
using parametric bootstrap likelihood ratio (PBLR) tests. I applied such
tests to seven data sets from the Pisidium fauna of Denmark’s Lake Esrom,
two records of 20 years, one of 9 years, and four of 7 years. One of the
gastropod records was also of sufficient length to analyse using the PBLR
technique. The three significant results obtained from these eight tests
constitute a success rate comparable to those posted by workers with
much better bird or insect data sets. This again constitutes some evidence
of density dependence.
Summary · 225
If natural populations rarely grow to the carrying capacity of their
environments, and if they are generally adapted to exploit widely varying
resources, one might expect competition to be a rare phenomenon in the
wild. But workers using the same approaches as used for intraspecific
competition, including the design of enclosure experiments and the
examination of long-term trends, often report evidence of the interspe-
cific phenomenon. In situ cage experiments have returned evidence of
competition between North American populations of Musculium trans-
versum and M. securis (Mackie et al. 1978), Lymnaea and Physa (Brown
1982) and even Lymnaea and Valvata (Hershey 1990).
Competition is a widely reported consequence of invasion or artificial
introduction. There is some evidence of competition between Corbicula
and bivalves native to North America, the American pisidiids perhaps
most adversely affected. Much stronger evidence has accumulated regard-
ing the adverse affect of Dreissena upon unionids. Fouling by zebra
mussels lowers the glycogen and lipid stores of host unionids, resulting in
decreased growth rates and increased mortality. Large ampullariids, vora-
cious consumers of macrophytes (and perhaps pulmonate eggs and neo-
nates), have reliably and efficiently eliminated Biomphalaria from larger
ponds and lakes in the neotropics. Thiarids have been similarly effective
in better oxygenated and less weedy environments, by virtue of the high
population densities they achieve, and their environmental tolerance.
The failure of Helisoma duryi to outcompete Biomphalaria in the wild,
given its promise in laboratory trials, warrants further examination.
Traditionally, invertebrate population density has been viewed as a
function of the time elapsed since the last calamity. But here we have
reviewed a not inconsequential store of evidence for density-dependent
regulation in freshwater mollusc populations, and for both intraspecific
and interspecific competition. Thus the questions become: under what
circumstances may competition be expected, and what shall be the
outcome.
In the previous chapter, it was suggested that freshwater mollusc pop-
ulations may profitably be divided into categories according to their
reproductive effort: R-adapted, S-adapted, and Undifferentiated.
Because they are adapted to special environments, I would hypothesize
that intraspecific competition at least occasionally becomes an important
factor in the regulation of R and S populations, but that interspecific
competition would be less usual. One might thus expect adaptations to
minimize intraspecific competition in R and S populations, as for
example the broad generalization of diet and substrate use manifested by
226 · Population dynamics and competition
the Physa population studied by Brown (1982). But interspecific compe-
tition would seem more likely to impact U populations. Hence U pop-
ulations may be more likely to evolve specializations to minimize
ecological overlap with other species: the unusual size of the ampullar-
iids, for example, or the epibenthic habit of Dreissena. It may be recalled
from Chapter 4 that unusual life cycles may evolve in large communities,
such as those reported by Jokinen (1985) among the 14 snail populations
of her Connecticut pond. I have added a row indicating the predicted
relative importance of intra- and interspecific competition in U, S, and
R populations to Table 4.2.
Successful molluscan invaders almost certainly include many R-
adapted populations in their number. I would speculate that the repro-
ductive efforts of Physa acuta, Pseudosuccinea columella, and Helisoma duryi
would rank them among the R-adapted, for example. But it seems
unlikely to me that an R-adapted or an S-adapted population could com-
petitively eliminate a reproductively Undifferentiated population in the
wild. Although R populations may replace U populations (such as B.glab-
rata appears) in restricted or artificial environments, in the wild typical U
populations probably display sufficient adaptations and habitat specializa-
tions to coexist with any generalist. Thus H. duryi seems ineffective
against Biomphalaria in field trials.
It may be recalled from Chapter 4 that the single population of
Corbicula for which adequate data are available seems to be
Undifferentiated in its reproductive effort. I speculate that such success-
ful invaders as Dreissena, Corbicula, certain thiarids, and Potamopyrgus may
not be R-selected. They may all prove to be reproductively
Undifferentiated species that have evolved special adaptations to mini-
mize the impact of intraspecific competition in environments that are
neither unpredictable nor poor. When introduced (typically by human
hand) into a new region, they may exploit previously unused resources,
and/or outcompete resident populations. The interactions among the
ecologically diverse U populations of predictably rich environments will
be the focus of Chapter 9.
6 · Parasitism
Prevalence
It is unfortunate that more is not known about the great majority of the
trematodes of which a direct economic consequence is not immediately
obvious. Mackie (1976a) tabulated data from 27 separate studies on di-
Prevalence · 231
genetic trematode infections in 19 pisidiid species. Pisidiids are known to
host 13 species from the family Allocreadiidae and 7 from the
Gorgoderidae, almost all of which are autogenic, requiring a second
intermediate host (usually an aquatic insect) before maturing in a fish or
an amphibian. Data on prevalence for most of these are rare; data on con-
sequence to the molluscan host nearly non-existent.
For three years Mackie took monthly quantitative samples from the
Musculium securis population of Britannia Bay near Ottawa, Canada, ulti-
mately examining 1764 individuals for trematode parasites. Observable
parasitism was limited to adults greater than 3.5 mm in length during the
months of July, August, and September. Mackie noted, however, that the
40–50 day intramolluscan development times normal for trematodes of
this sort would imply that infection often occurs among the juveniles.
Prevalence peaked at 27.8% in July 1970, 25.0% in August 1971, and
21.8% in August 1972. Although uninfected adult Musculium generally
carried 3–6 brood sacs with 2–8 larvae per sac throughout the summer,
parasitized individuals appeared to be entirely sterile.
Clearly trematode parasitism may be of considerable consequence to
populations of freshwater bivalves (Taskinen et al., 1997). The gorgoderid
Phyllodistomum has drawn some attention for its potential to infect (and
perhaps help control) the zebra mussel Dreissena. Martell and Trdan
(1994) reported a 61% Phyllodistomum infection rate in a Michigan pop-
ulation of the unionid Venustachoncha ellipsiformis. Infected individuals
appeared to be entirely castrated. The authors implicated the trematode
in the very low replacement rate demonstrated by this mussel population.
Ismail and Arif (1993) reported infections of at least seven different
cercarial types in a dense population of Melanoides tuberculata from the
United Arab Emirates. The overall infection rate increased from about
50% to about 90% over a year of observation. The total of 3737 snails
examined included several hundred double- and even triple-infections,
for an overall prevalence of 74%. Such levels of parasitism cannot fail to
have a profound impact on the dynamics of a snail population. How
commonly may populations of freshwater molluscs be infected at these
rates?
Studies of human schistosomiasis afford the richest source of data on
trematode infection rates in natural populations of snails. Anderson and
May (1979) tabulated such data from about 30 studies conducted world-
wide, involving all three major hosts of human schistosomes (Figure 6.1).
The authors concluded that average prevalence of infection ‘tends to lie
in the range 1–5%, irrespective of the species of snail or schistosome, or
232 · Parasitism
A B C D E F G H I J K L M N
Ben Hur (18 months)
Hebetancylus 6 10 1 12
Ferrissia 1 2 2 2 1 1 1
Beaver Pond Branch (17 months)
Hebetancylus 2 5
Ferrissia 1 1 3 1 1 3
Ramah (14 months)
Laevapex 5 2 1 2 6
Hebetancylus 1
Ferrissia 1 1 2
Sorrento (14 months)
Laevapex 12 1 12 1 1 2
Hebetancylus 2 1
Ferrissia 4
Head of Island Pond (15 months)
Hebetancylus 2 1
Ferrissia 1 1 1
Notes:
The total number of months sampled is given by the name of each station. Letters
A–N correspond to Turner and Corkum’s taxa I–XVII, disregarding the rare types
VIII, XII, XVI, XVIII, and XIX.
Source: Turner and Corkum (1979).
Consequences
In the preceding section we were left with the impression that the overall
prevalence of trematode infection in freshwater mollusc populations
240 · Parasitism
generally seems low, but that both temporal and spatial variation may be
extraordinarily high. Some freshwater snail populations, at some times
and places, may show very high percentage infection rates. What is the
impact of trematode infection on a snail?
Holmes (1983) outlined ‘three contrasting models of the way hosts and
parasites may coevolve’: mutual aggression, prudent parasite, and incipi-
ent mutualism. The key distinction among these models is the observed
effect of the parasite on its normal host. Holmes’ mutual aggression
model (embracing the gene-for-gene and red queen hypotheses of other
workers) predicts that parasites evolve to maximize their total reproduc-
tive output, even at the expense of their host’s viability. His prudent par-
asite model predicts that parasites evolve to minimize adverse effects on
their hosts, and his incipient mutualism model proposes that, at least
occasionally, parasites may actually benefit their hosts. The theme of this
section will be to demonstrate that trematodes infecting mollusc hosts are
not prudent, much less mutualistic. They are always, insofar as I can
determine, aggressive.
Schistosomes
Among the earliest and most thorough descriptions of the effects of
trematode infection on snails was the work of Pan (1965) on S. mansoni
and B. glabrata. Pan’s experimental snails were 5.5–9.0 mm in shell diam-
eter, approaching sexual maturity. He exposed 150 individuals to
‘numerous’ freshly hatched schistosome miracidia for three hours,
keeping 100 snails as unexposed controls. The onset of egg laying
occurred within the week in both treatments, but declined to zero among
infected snails as the first cercariae began to emerge, about five weeks
post-infection. Survivorship was also affected; mortality to week 18 was
77% in infected snails but only 11% in uninfected controls. Mean shell
growth was significantly greater in the control population, giving no evi-
dence of ‘parasitic gigantism’ in host snails. Pan continued his observa-
tions to week 30, by which time only two of his 150 infected snails
survived, but did not see any recovery from infection. His B. glabrata had
been entirely castrated by the parasites.
In his (apparently successful) effort to eliminate the confounding
effects of snails exposed to schistosomes but not infected, Pan used an
unspecified but doubtless high concentration of miracidia. Evidence that
the actual number of miracidia infecting an individual snail may be of
profound consequence is shown in Figure 6.3. Chu and his co-workers
Consequences · 241
Figure 6.3. Survivorship and fecundity (eggs per survivor per week) in Bulinus trun-
catus infected with Schistosoma haematobium (data of Chu et al. 1966c). An uninfected
control is compared with treatments of 2, 5 and 20 miracidia per snail. Survivorship
is standardized to 100 at week 8 post-infection, the onset of cercarial shedding.
(Actual N⫽20–30.)
242 · Parasitism
(1966c) exposed individual Bulinus truncatus to 1, 2, 5, 10, and 20 S. hae-
matobium miracidia per snail (50 snails per treatment), keeping a sixth set
of 50 snails uninfected as a control. The snails were approaching sexual
maturity as were Pan’s Biomphalaria, four weeks old at the beginning of
the experiment. Snails began shedding cercaria on day 38 of the experi-
ment, and on day 53 (roughly eight weeks post-exposure) the authors
began to monitor survival and reproduction in a successfully infected
subset.
Figure 6.3 is based on Chu et al.’s table 4, with survivorship standard-
ized to 100, days (and fecundities) converted to weeks, and all data log
transformed. Survivorship to week 22 was about 53% for the control,
10% for the two-miracidia treatment, and nil for five- and 20-miracidia.
But while the effect of parasitism on reproduction was striking, only in
the 20-miracidium/snail treatment did castration approach 100%. In the
two-miracidium snails, fertility generally rose throughout most of the
study, peaking at a (corrected) 76 eggs laid per six snails surviving to 17
weeks post-infection. It should be noted that the entire 76-egg effort
probably did not come equally from all six animals. In the 20-miracid-
ium treatment, the single survivor to 17 weeks quite suddenly laid 24
eggs, apparently affecting a ‘self cure’. The occasional ability of host snails
to out-last schistosome infection has also been noted in B. glabrata by
Etges and Gresso (1965).
The age of the snail host is another important variable influencing the
severity of schistosome attack. Sturrock (1966b) collected 140
Biomphalaria pfeifferi hatched on the same day and exposed groups of 30
individuals to five S. mansoni miracidia at ages one, three, five, and seven
weeks, keeping the remaining 20 snails as an uninfected control. The top
half of Figure 6.4 shows that considerable mortality is associated with
schistosome attack, regardless of age. While the control snails suffered no
mortality over 21 weeks of observation, only 18 of 30 snails exposed at
age one week survived, including 8 uninfected (subtracted from Figure
6.4), and but 13/30 of the three-week treatment survived, including a
single uninfected individual. Mortality was most severe in the seven-
week treatment, where only a single individual survived to age 21 weeks.
Figure 6.4 also shows that parasitic castration was 100% only in the
group infected at seven weeks, already mature when infected.
Interestingly, these snails actually laid more eggs in weeks 9 and 10 while
their infections were prepatent than uninfected controls. Similar phe-
nomena have been reported by Minchella and Loverde (1981), Minchella
Consequences · 243
Figure 6.4. Survivorship and fecundity (eggs per survivor per week) in Biomphalaria
pfeifferi infected with Schistosoma mansoni (data of Sturrock 1966b). An uninfected
control is compared with snails infected at ages 1, 3, and 7 weeks. Survivorship is
standardized to 100 at start (actual N⫽30).
244 · Parasitism
(1985), and Thornhill et al. (1986) in the B. glabrata/S. mansoni system,
and have generally been interpreted as a compensatory response for
expected future suppression of egg laying.
Although their fecundity was reduced, Biomphalaria challenged with
five miracidia per individual at an early age often did not seem to lose
their reproductive capabilities entirely. All the individuals included in the
fecundity estimates shown in Figure 6.4 were actively shedding cercar-
iae. Mean fecundities of snails infected at one week varied around 10–20
eggs/individual/week, while the three-week exposed snails averaged
about 5–7. Sturrock did note a significantly high proportion of sterile
and inviable eggs among those laid by infected snails. Such data from Chu
and colleagues would have been welcome. However, Sturrock reported
that by week 26, four of his one-week and three-week individuals had
stopped shedding cercariae and raised their fecundities to about 80
eggs/week, apparently cured. One might speculate that, although five
miracidia may attack and penetrate a one-week-old snail, not all five may
develop. The effective parasite loads in these snails may have been lower.
All four infected groups of snails (including a five-week group I have
neglected here thus far) showed accelerated shell growth during their pre-
patent periods. Mean shell sizes at week 19 ranged from 10 mm to 12 mm,
significantly greater than the 9 mm averaged by uninfected controls. In a
second, larger and longer-term experiment involving three-week
exposed snails only, Sturrock showed significantly greater mean size and
greater size variance in parasitized B. pfeifferi. This constitutes consider-
able evidence for ‘parasitic gigantism’, at least under some circumstances.
The effect of S. mansoni infection has also been examined on B. glab-
rata of varying age by Meier and Meier-Brook (1981). Although their
experimental design was similar to Sturrock’s, they obtained results strik-
ingly different in some respects. The authors exposed groups of 90 snails
to three miracidia each at five different ages (one, two, four, six, and eight
weeks post-hatch) keeping a sixth group of 90 snails unexposed as a
control. Snails exposed at age one week showed 81% mortality during
the five-week prepatent period before the emergence of the first cercaria.
In fact, five of the 17 snails surviving to age six weeks proved not to have
been successfully infected. Snails exposed to schistosomes at age two
weeks showed somewhat greater mortality than controls. But quite in
contrast to Sturrock’s findings, survivorship in snails infected at four, six
and eight weeks was not strikingly different from control.
In another contrast with Sturrock, Meier and Meier-Brook observed
no egg laying by any of their B. glabrata infected at age one week, nor by
Consequences · 245
those infected at age two weeks. In snails infected at age four weeks and
older, fecundity was reduced during the five-week prepatency, declining
steeply in all treatments as cercariae began to emerge. All three older
groups (exposed at 4, 6, and 8 weeks) were still laying eggs as the experi-
ment ended at week 14, perhaps too early to observe total castration. In
a final contrast with Sturrock, the authors found evidence that schisto-
some infection retards shell growth in B. glabrata; the effect was most pro-
nounced in snails infected in their youth.
The maturity of the host snail at time of schistosome infection unques-
tionably influences the consequences of that infection (Fryer et al. 1990).
But Meier and Meier-Brook’s results strongly indicate that S. mansoni
attack weighs most severely on small, juvenile B. glabrata, while Sturrock’s
results suggest just as clearly that schistosome infection is of more pro-
found consequence to older, reproductively mature B. pfeifferi. Perhaps
this is a function of host specificity; the relationships between schisto-
somes and their snail hosts are tuned at the population level. Meier and
Meier-Brook used an S. mansoni strain derived from Liberian stock 50
years previous and B. glabrata from Puerto Rico. Sturrock’s snails were
Tanzanian, schistosome unspecified. The consequences of host specific-
ity will be explored further in the section that follows.
Fasciola
The effect of Fasciola on English Lymnaea truncatula has been described
by Hodasi (1972). He exposed three batches of 20 individual snails each
to five newly hatched miracidia at ages two weeks (juvenile), four weeks
(‘adolescent’), and six weeks (adult), keeping a fourth batch of 20 unin-
fected as a control. He achieved successful infection in 18, 18, and 19
individuals, respectively. In a separate experiment, Hodasi exposed a
second (apparently rather large) group at age two weeks to obtain a finer
estimate of the effect of Fasciola on snail survivorship.
Several striking differences in the biology of the snail hosts of Fasciola
and Schistosoma are evident from a comparison of Figure 6.5 with Figures
6.3 and 6.4. In the laboratory, Bulinus and Biomphalaria seem to reach
reproductive age at about 6–7 weeks, while L. truncatula begins to lay eggs
at week four. And while uninfected laboratory populations of Bulinus and
Biomphalaria show very little mortality and uninterrupted reproduction
through 20 weeks, control laboratory populations of L. truncatula seem
to expire naturally.
The top half of Figure 6.5 shows a negligible effect of Fasciola on snail
survivorship up to about week 12, at which time the decline ordinarily
246 · Parasitism
Figure 6.5. Survivorship and fecundity (eggs per survivor per week) in Lymnaea
truncatula infected with Fasciola hepatica (data of Hodasi 1972). Survivorship data
compare uninfected controls with snails infected at age two weeks. The fecundity of
an uninfected control is compared with snails infected at ages 2, 4 and 6 weeks.
Consequences · 247
observed in L. truncatula is postponed. All uninfected snails were dead by
week 20, while 38% of the snails infected at age two weeks remained.
The last infected snail did not die until week 25 in Hodasi’s survivorship
experiment. The effect of parasitic infection on fecundity seems to be
more severe than ordinarily observed from schistosomes. Snails infected
at weeks two, four, and six were 100% castrated at weeks six, eight, and
ten, respectively. But note that by week ten, individual L. truncatula will
have already laid a large number of eggs. Given the four-week prepatent
period typical of many trematodes, rapid maturation would seem to have
considerable adaptive value as a defence against parasitism.
Hodasi showed that Fasciola infection may induce gigantism in L. trun-
catula, at least under some circumstances. His control animals reached a
size plateau of about 7.5 mm at onset of egg laying, age 6–8 weeks. Snails
infected with Fasciola at age two or four weeks evinced a growth spurt
about 4–5 weeks post-infection, just as cercariae were beginning to
emerge and reproduction was being suppressed. They reached a size
plateau at about 9.5 mm, strikingly larger than the controls. But snails
infected at age six weeks, after reaching maturity, remained at their fixed
adult size and showed no evidence of gigantism.
Echinostomes
The echinostome Ribeiroia marini is reported to castrate Biomphalaria glab-
rata completely, as does Echinostoma togoensis its host B. pfeifferi (Combes
1982). Kuris (1980) introduced five miracidia of Echinostoma liei per snail
into aquaria containing 50 Biomphalaria glabrata of three size classes: 1
mm, 2 mm, and 5 mm. He kept a second set of all three size classes unex-
posed as a control, and monitored survivorship over 30 days. While unex-
posed snails of all sizes suffered negligible mortality, Kuris did detect
significantly increased mortality among 1 mm and 2 mm size classes
exposed to Echinostoma. For unexposed snails initially 1 mm, mean shell
diameter at the end of the experiment was 11.15 mm, significantly larger
than the 9.15 mm averaged for similar snails successfully infected with
Echinostoma.
Echinostomes have attracted considerable attention because of their
especially aggressive interactions with other trematodes. In the event a
single snail is doubly infected by an echinostome and, for example, a
schistosome, the echinostome rediae often outcompete or even consume
the larvae of the second species (Lim and Heyneman 1972). The effect
of multiple trematode infections upon the molluscan host does not seem
248 · Parasitism
to have been well characterized, however. Interestingly, previous infec-
tion with Echinostoma paraensei seems to cause resistant strains of B. glab-
rata to lose much of their resistance to schistosome infection (Loker et al.
1986). Further discussion of the resistance of snails to trematode attack
follows.
Resistance
Successful infection by a trematode parasite inevitably lowers the fitness
of a snail host. Whether that fitness equals zero or some small fraction
above zero depends on the age of the snail at time of attack, the number
and species of the miracidia involved, and many other factors. In any case,
it is clear that any defences a snail might mount to protect itself from
invading trematodes would be most adaptive.
Our understanding of the mechanisms by which molluscs defend
themselves from invasion by foreign bodies has progressed steadily for
over 30 years, but details remain far from clear. Trematode miracidia are
usually encapsulated and destroyed by mollusc haemocytes shortly after
penetration. Only on those (exceptional) occasions when miracidia find
a host for which they have evolved some special counter-measures can
further development proceed (Coustau and Yoshino 1994). Mollusc
plasma contains a protein (an ‘opsonin’) that makes foreign bodies more
susceptible to attack by haemocytes (Bayne and Yoshino 1989, Yang and
Yoshino 1990, Lodes et al., 1991). Opsonins may belong to the general
class of defence proteins found in invertebrate plasma, the lectins (Harris
et al. 1993). It has been suggested that ‘resistant’ trematode sporocysts may
have developed a mechanism to avoid or abort the process of opsoniza-
tion. Lie and colleagues (1987) present evidence that resistant trematodes
may have evolved the ability to modulate host haemocyte function by
misdirecting haemocytes from the site of infection. But should they
attract the attention of haemocytes, resistant miracidia and sporocysts
seem to have evolved a variety of counter-measures (Connors et al. 1991).
Given that primary sporocyst development proceeds normally, however,
it is by no means certain that secondary sporocysts or rediae will find hos-
pitable conditions in the host digestive gland. Less is known about the
inhibition and even elimination of later larval stages, such as often seems
to be observed in laboratory infections.
At this point we might profitably return to the three models of
host/parasite coevolution outlined by Holmes (1983). Under the incip-
ient mutualism model, where the host may accrue some benefit from par-
Resistance · 249
asitism, there would be selection to minimize host resistance to the par-
asite. This does not seem to be the case in trematodes and snails. Under
the mutual aggression and prudent parasite models, selection acts to max-
imize host resistance, although selection is much stronger in the former
than in the latter case. Should a parasite be able to evade the defences of
a new host, with which it is not coevolved, the effect of a prudent par-
asite is predicted to be exceptionally pathological. But the effect of a par-
asite that has evolved by mutual aggression with its ordinary host may be
lessened when first infecting a new host. We begin with data suggesting
that Fasciola seems prudent from this viewpoint, but conclude with evi-
dence for the aggression of Schistosoma and Echinostoma. This last situa-
tion is, again, usually observed in experiments with trematodes and snails.
Fasciola
In temperate regions of the old world, the usual intermediate host for
Fasciola hepatica is the amphibious Lymnaea truncatula, while in the new
world Lymnaea bulimoides is the most commonly reported host. In
Australia, where the trematode has been introduced with livestock, its
usual host is the slightly more aquatic L. tormentosa. Fasciola gigantica is
more common in the tropics, hosted by L. auricularia in Asia and the
related L. natalensis in Africa.
Boray (1966) exposed large numbers of lymnaeids collected world-
wide to 25 miracidia from three F. gigantica strains (Kenya, Malaya and
Java). Figure 6.6 shows selected results from experiments with the Kenyan
strain. The author did not keep snails unexposed as controls, nor in fact
are data available on snail survivorship. But as the normal host for African
Fasciola, the N⫽30 adult L. natalensis serve as a control for comparisons
of infectivity.
Boray recovered viable metacercariae from 83% of the L. natalensis he
exposed. It would also appear that his 100 adult L. tormentosa served as
fairly efficient hosts for F. gigantica, even though this trematode is
unknown in Australia. (The L. tormentosa strain used, from New South
Wales, does, however, host F. hepatica.) Interestingly, such resistance as is
shown by L. tormentosa seems to be post-redial. Of the 86 snails surviv-
ing exposure, 37 did develop rediae but never shed cercariae. The three
other prospective hosts shown in Figure 6.6 are all European – L. peregra
from Austria and L. stagnalis and L. palustris from Lower Saxony,
Germany. All three of these were exposed as juveniles, aged less than 21
days in the first two cases and 21–70 days in the case of L. palustris. Only
L. palustris proved entirely refractory. Eighteen L. peregra became infected
250 · Parasitism
Figure 6.6. Proportion of five Lymnaea populations infected with Fasciola gigantica
in laboratory experiments of Boray (1966). Snails showing redial infection (only) are
indicated by shaded bars, and blackened regions show those from which viable meta-
cercariae were recovered.
Figure 6.7. The infectivity of Fasciola hepatica from Australia (A) or Germany (G)
on Australian Lymnaea tormentosa (data of Boray 1966). ‘Pass 1’ is at the start of the
experiment (N⫽200 snails for both types of miracidia) and ‘Pass 3’ is at the third
generation of culture in Australian snails (N⫽39–47). The height of each bar shows
snail survivorship after exposure to miracidia, with shaded portion developing rediae
(only) and metacercarial cysts recovered from the blackened portion.
Schistosomes
Much in contrast to the situation with Fasciola, Schistosoma haematobium
is generally observed to display great specificity to individual populations
of snail hosts. Sturrock (1967) exposed five groups of 50 Bulinus nasutus
productus aged two, four, six, eight, and 14 weeks to five S. haematobium
miracidia each. Although B. nasutus is in fact the normal host for S. hae-
matobium in Tanzania, the author began his study with the knowledge
that the snail strain he was using in this experiment was at least partially
resistant. Ten snails aged two weeks at exposure did eventually shed cer-
cariae, as well as a single individual exposed at four weeks. But to judge
from cercarial shed, no infections developed in any of the remaining 239
snails.
Infections in the eleven snails progressed in a variety of fashions.
Although Sturrock recorded no deaths among control snails or among
snails exposed but not infected, seven of the infected snails died prior to
week 28. One of these proved to have been entirely castrated – never pro-
ducing an egg while shedding thousands of cercariae. Six infected snails
did produce eggs, sometimes at the rate of hundreds per week, before
their reproductive systems were finally overwhelmed and shut down.
Two snails continued to produce viable eggs to week 28, although not at
the rates of control snails, while shedding cercariae. Finally, two snails
affected self-cures, shedding only a few cercariae early in their infections
and ultimately regaining normal fecundity.
Sturrock’s results illustrate three phenomena commonly observed in
experiments involving trematodes and resistant snail hosts. First, as we
also noted in Boray’s F. gigantica data (Figure 6.6), it seems clear that snails
mount multiple lines of defence. The 239 animals never shedding cer-
cariae apparently defeated infection at some earlier stage than those shed-
ding cercariae and subsequently affecting a cure. Second, the defences of
resistant snails may not be 100% effective, but may be overwhelmed when
the snails are young and/or exposed to large numbers of miracidia. I
would speculate that all five miracidia successfully developed in the most
Resistance · 253
severely affected snails among Sturrock’s 11 hosts, and that snails able to
recover were successfully penetrated by fewer miracidia. Finally, it is
evident that the consequences of infection were not as severe as ordinar-
ily observed in the 11 resistant snails ultimately hosting trematodes. This
matches expectation from the mutual aggression model of host/parasite
coevolution.
It is not entirely clear that in the arms race between parasite and host,
the parasite will hold the upper hand. And in fact, it is not uncommonly
observed that trematode populations show reduced infectivity to the snail
population with which they are evolving. For example, the pomatiopsid
Tricula aperta (or Neotricula aperta) is the (normally listed) host of
Schistosoma mekongi. Three ‘races’ of N. aperta are recognized, called alpha,
beta, and gamma, showing some morphological and chromosomal differ-
ences and quite possibly some reproductive isolation as well. The infec-
tivity experiments performed by Kitikoon (1981b) were exceptional
because they involved both wild-collected snails and wild-collected
schistosomes, rather than laboratory-reared (and hence possibly inbred)
lines.
Kitikoon collected his gamma snails at Khong Island near the
Laos/Cambodia border. Here he also collected snails of a second poma-
tiopsid genus very similar to Neotricula, Manningiella. His miracidia were
hatched from eggs collected from naturally infected Khong Island dogs.
Kitikoon collected alpha N. aperta from the Mekong River near
Khemmarat and beta snails from the Mune River near Phibun, both local-
ities in Thailand about 200 km upstream from Khong Island. He held all
snails in the laboratory for 60 days, checking for shed cercariae, and
cracked a large sample of each to verify the absence of pre-existing trem-
atode infection. He then exposed 200–250 snails of each of the four types
individually to five S. mekongi miracidia, keeping about 100 unexposed
controls, and monitored cercarial shed for another 60 days.
Pomatiopsids seem to be more difficult to culture than pulmonates.
The survivorships to 60 days post-exposure shown in Figure 6.8, ranging
from 79% to 95%, were comparable to those observed in unexposed con-
trols. None of the controls shed cercariae, but all four exposed groups
were successfully infected to some degree. About 13% of surviving
gamma T. aperta, the nominal intermediate host for the Mekong schisto-
some at its Khong Island focus, did in fact shed cercariae. Remarkably,
this figure was not significantly different from the 12% infection rate
observed for co-occurring Manningiella, not ordinarily considered a host.
The Khong island schistosome is better adapted to its local pomatiopsid
254 · Parasitism
Figure 6.8. The infectivity of Khong Island Schistosoma mekongi to four populations
of Mekong River pomatiopsids: the alpha, beta, and gamma races of Neotricula aperta
and Manningiella conica. (Data of Kitikoon 1981b.) The total length of each bar shows
the survivorship of (N⫽200–250) snails to 40–60 days post-exposure, with black-
ened portions showing the survivors shedding cercariae.
populations than to the alpha snails of Khemmarat. But the 60% infectiv-
ity observed for beta-race snails collected 200 km upstream at Phibun is
significantly greater than the Khong values. Clearly, the Khong Island
schistosome could be doing worse, but it could also be doing better.
The relationship between schistosomes and their host populations may
be as dynamic as that observed by Boray in Fasciola (Figure 6.7). Kagan
and Geiger (1965) reported the results of three years of infectivity experi-
ments involving laboratory lines of both B. glabrata and S. mansoni from
both Puerto Rico and Brazil. The snails were mass-cultured. Every few
months for 2–3 years, the authors exposed 100–200 snails 6–8 mm indi-
Resistance · 255
vidually to 6-12 miracidia. Cercariae recovered 6 weeks post-exposure
were used to infect mice and hamsters, which were necropsied 8 weeks
subsequently to recover a new generation of eggs. It is not clear how
many generations of schistosomes passed per year, but over a three-year
period, the infectivity of Kagan and Geiger’s Puerto Rican schistosome
to its Puerto Rican host improved from 36% to 64%. The results from
other combinations were mixed, and less dramatic. But at least occasion-
ally, the mechanisms of trematode adaptation to a new host seem more
efficient than one would expect from selection on available genetic vari-
ation. Again one might wonder about the relative heritabilities of trem-
atode adaptation and snail defence.
The heritability of resistance to S. mansoni infection in B. glabrata cer-
tainly may be high (Webster and Woolhouse 1998). Richards and Merritt
(1972) designed an elaborate breeding programme directed toward an
understanding of the heritable basis for juvenile resistance to S. mansoni
in Biomphalaria glabrata. Their schistosome was a strain that had been col-
lected from Puerto Rico about ten years previously and maintained in
mice. Several laboratory lines of snails were involved initially, carrying
pigmentation markers to verify outcross. The authors had available
Newton’s (1955) ‘M’ B. glabrata line of mixed Puerto Rican/Brazilian
origin, susceptible to infection and homozygous for the entirely recessive
albinism allele. Richards and Merritt employed a subline of Newton’s M
developed from a spontaneous ‘black-eye’ mutant. (Black-eye is domi-
nant over albinism but recessive to wild-type pigmentation.) The authors
also used a pigmented B. glabrata strain from Puerto Rico resistant to
schistosome infection, which they designated ‘H’.
Richards and Merritt’s first goal was to obtain lines as pure breeding
for juvenile susceptibility or resistance as possible. Snails 1–4 mm in diam-
eter were initially exposed to five miracidia each. Since the shells of the
albinos and black-eyed individuals are translucent, individuals could be
examined non-destructively after one week for signs of infection. Those
not developing parasites were re-exposed to five miracidia each. Snails
remaining uninfected after two exposures were isolated and allowed to
self-fertilize, as founders of a new H-line ultimately proving to show 0%
infectivity to the Puerto Rican strain of schistosome employed. The
unexposed progeny of snails from which test exposures yielded high
infection rates were allowed to self-fertilize to found a new M-line
showing essentially 100% juvenile infectivity.
Results of hybridization studies are shown in Figure 6.9. Seven of 15
progeny produced by the M-line parent after outcross (47%) proved
256 · Parasitism
new world may render schistosomes less infective back in the old world.
I further suggest that new world S. mansoni populations may have suffered
a bottleneck upon their introduction into the new world, so that they
have lost whatever general adaptation they may have had for old world
snails through reduced selection and genetic drift. Thus old/new data are
indistinguishable from new/new or new/old.
Echinostomes
We began this section with a demonstration that a fluke with a Type I life
cycle, Fasciola, seems to show broad adaptation to amphibious lymnaeid
populations worldwide. We then found substantial host specificity in the
Type II schistosomes, but noted that the host–parasite relationship seems
dynamic, with both heritable and non-heritable components in both
260 · Parasitism
fluke and snail. What sort of relationship is be be expected in the Type
III echinostomes?
Echinostoma liei was first described by Jeyarasasingam et al. (1972) par-
asitizing Biomphalaria alexandrina in the Cairo area. The authors consid-
ered that it showed an ‘extreme degree of primary host specificity’,
apparently unable to infect Brazilian (Minas Gerais) B. glabrata, Brazilian
B. straminea, or even a second ‘intractable’ Egyptian strain of B. alexan-
drina. Yet they did report successful infection of an ‘NIH strain’ of B. glab-
rata, more specific details on strain unavailable.
Lim and Heyneman (1972) summarized a series of single-trematode
infectivity studies in B. glabrata (unspecified NIH strain) in preparation
for echinostome double-infection experiments. They exposed snails of
five size classes to five different numbers of miracidia and estimated rates
of successful infection, to cercarial shed. No data are available on the sur-
vivorship or fecundity of the snails. Lim and Heyneman obtained fairly
low rates of successful infection with single S. mansoni miracidia, perhaps
around 20% for size classes 2–10 mm, dropping to 5% for their 20 mm
snails. Their E. liei strain seems to have been slightly more successful than
their schistosome in infecting smaller B. glabrata. Exposed to single E. liei
miracidia, infection rates for the 2 mm snails were 26% and 5 mm snails
31%, with a drop-off to 6% in 20 mm snails comparable to the schisto-
some data. As might be expected, the Brazilian echinostome
Paryphostomum segregatum was more infective to B. glabrata than either old
world trematode. From single miracidia Lim and Heyneman achieved
40–64% infectivity in the size 2–15 mm, the rate dropping to 15% only,
once again, in their largest 20 mm snails.
Christensen et al. (1980) exposed a wide variety of freshwater pulmo-
nates (3–4 mm) to eight E. liei miracidia and evaluated infection rates 35
days post-exposure. Figure 6.11 shows results from five species: B. alex-
andrina (the normal Egyptian host), B. camerunensis (from quite distant
Kinshasa), B. glabrata (an NIH strain from Puerto Rico), Helisoma duryi
(Florida), and Physa acuta (Egypt). Although the authors did not keep
unexposed controls, Figure 6.11 seems to reflect some evidence of
decreased survivorship to day 35 among exposed snails. The effect is
especially apparent in B. camerunensis; surely 11 of 43 snails 3–4 mm
would not ordinarily die in five weeks of culture. Christensen and col-
leagues obviously achieved higher infection rates in B. alexandrina and B.
glabrata with their eight miracidia than Lim and Heyneman achieved with
their one. No individual snails from the remaining three species shed cer-
cariae by day 35, although a single B. camerunensis was found to be har-
Resistance · 261
Figure 6.11. The infectivity of Echinostoma liei miracidia and cercariae to five
species of pulmonate snails. (Data of Christensen et al. 1980.) The total length of
each upper bar shows the survivorship of (N⫽variable) individual snails 35 days post-
exposure to miracidia, with blackened portions showing the survivors shedding cer-
cariae. From a second experiment, the shaded bars show the mean percentage of 20
or 25 cercariae exposed to each of six snails recovered as metacercarial cysts.
bouring rediae. The observation that new world B. glabrata is a better host
than old world B. camerunensis to an old world echinostome is interest-
ing, but not surprising in light of our experiences with schistosomes.
As has been observed in trematodes generally, the infectivity of E. liei
is considerably less specific to its second intermediate host than to its first.
Christensen and colleagues exposed six 3–4 mm snails of each species to
20 or 25 cercariae per snail and counted metacercarial cysts via dissection
several days later (Figure 6.11). Data on snail survivorship are not avail-
able, and it is not clear whether complete resistance was demonstrated by
262 · Parasitism
any individual. But Christensen and colleagues considered P. acuta, har-
bouring a mean of 70% of the metacercaria to which it was exposed, a
‘high susceptibility’ second intermediate host. The remaining species
were considered to show ‘low susceptibility’.
The echinostome Echinoparyphium recurvatum successfully utilizes the
entire mollusc community of Harting Pond, West Sussex, as second
intermediate hosts: pulmonates, prosobranchs, and pisidiids (Evans et al.
1981). There seems to be little evidence of ‘preference’. With the excep-
tion of two planorbids (Planorbis planorbis and Gyraulus albus), ranking by
mean number of metacercaria per individual mollusc appears to follow
their ranking by individual size: Sphaerium corneum, Lymnaea peregra,
Valvata piscinalis, planorbids, Pisidium subtruncatum and Potamopyrgus jen-
kinsi. The authors note that Pisidium and Valvata might be considered the
most important second intermediate hosts for Echinoparyphium, by virtue
of their high population densities in the pond. Thus on the limited evi-
dence available, it would appear that the relationship between echinos-
tomes and their first molluscan host may be as close as that seen in
schistosomes, but their relationship with their second intermediate host
is non-specific.
Population regulation
To prospect for evidence of an effect of parasitism on snail density, one
might begin where prevalence rates are unusually high. Rates ranging up
to 60% have been reported by Fernandez and Esch (1991a,b) working
with the Helisoma anceps population of a small North Carolina pond. The
authors sampled snails at five localities along a 15 m shoreline, on ten
occasions from April of one year to June of the next. Snails were hand
picked, marked, assayed for parasitism, and returned to the water.
Although no data were available on that portion of the population less
than 3–4 mm, this approach gave an excellent picture of parasitism in the
adult fraction.
The authors inferred that their population of H. anceps was primarily
annual and semelparous, with a reproductive peak in April. Survival
analysis was complicated by their observation that a small portion of
the spring cohort grew sufficiently to reproduce in August. In any case,
the prevalence of parasitism in adult snails (almost exclusively due to the
hemiurid Halipegus occidualis) seemed positively correlated with the
adult population density, offset by some months, as one might expect
from predator–prey interaction. Adult Helisoma density was low in
Population regulation · 263
April–June and peaked in October, while their proportion parasitized
peaked in May.
But as is usually the case, H. occidualis infection generally leads to the
castration of its host (Crews and Esch 1986), not its immediate demise.
The rising densities of adult pulmonates generally observed in the
summer, and their dramatic decline through winter and early spring, are
conventionally ascribed to season and the life cycle of the snail. It is pos-
sible to view Halipegus prevalence as passively following Helisoma density,
with no reference to an effect of parasite on host. Clearly more than a
single generation of data will be required to detect an effect of a parasite
on the density of its host.
Total trematode infection rates also occasionally top 50% in the
Lymnaea populations of an English gravel-pit pond. Adam and Lewis
(1993) monitored snail densities on 25 occasions over 21⁄2 years, docu-
menting along their way the sporadic occurrence of twelve trematode
species. Their snail-collecting technique was more conventional (hand
net) and more likely to include juvenile snails, although some size bias
doubtless remained. The wonderfully complicating factor in Adam and
Lewis’ data is that during their study period, the Lymnaea peregra popula-
tion more typical of their small body of water was completely replaced
by the similar but large-lake adapted L. auricularia (see Chapter 5).
Since the L. auricularia population seems to have passed through a bot-
tleneck upon first colonizing this gravel-pit pond, it seems unlikely that
the snails could have carried their own trematode fauna with them. Thus
one might expect a reduction in their parasite load. Yet almost all the local
trematodes seem to have successfully switched hosts, and in fact the intro-
duction of L. auricularia seems to have brought several new trematodes
into the system. Figure 6.12 shows the combined prevalence of all twelve
parasites together with the relative population densities of their two lym-
naeid hosts. Prevalence rates appear to have increased since the replace-
ment of L. peregra by L. auricularia. But it is difficult to detect a
relationship between monthly snail densities and rates of parasitism. Snail
populations peaked in August in all three years, corresponding directly to
peak infection rates in years one and three. Parasitism lagged behind the
snail population by a month or two in year two, when L. auricularia
appeared. One is tempted to speculate that this delay may have repre-
sented a period of adaptation on the part of the trematode community.
But again, the appearance is of a snail population responding to season,
and of a parasite population responding to snail.
If seasonality could be factored out, perhaps an effect of parasite upon
264 · Parasitism
Aspidogastrids
Ogden Nash’s observation that ‘Big fleas have little fleas upon their backs
to bite ’em’ here finds some illustration. For unionacean mussels, which
as larvae parasitize fish, are themselves the targets of a variety of parasites
upon reaching adulthood. But the long-term culture of even healthy,
unparasitized freshwater mussels presents a considerable technical chal-
lenge. We might not be surprised to learn that the consequences of par-
asitism to their survivorship and reproduction are not, as yet, well
investigated.
Aspidogastrid trematodes are among the most common internal para-
sites of unionaceans, and may infect freshwater snails as well (Huehner
and Etges 1981). The mixture of life cycles these worms display, involv-
ing both single hosts and multiple hosts, has prompted workers to recog-
nize a small trematode subclass or order (the Aspidobothrea) to contain
them (Cheng 1986). The eggs of Aspidogaster conchiola are shed into the
water and carried by currents from a mussel host to the incurrent siphons
of its neighbours, where they are directly infective. Cotylaspis insignis
eggs, by contrast, develop into free-swimming larvae and enter the
siphons of their hosts actively. Thus Huehner (1984) observed that
Aspidogaster infections tend to be more common in populations of uni-
onids inhabiting lotic waters, and Cotylaspis infections more common in
the lentic.
Huehner examined 590 mussels from nine Missouri locations, repre-
senting 32 species. He found Aspidogaster infections in 213 individuals (23
species), C. insignis in 50 individuals (16 species), and Cotylogaster occiden-
talis in four individuals (three species). Duobinis-Gray et al. (1991)
reported aspidogastrids in 30% of 219 unionids collected from Kentucky
Lake, including 9 of the 10 mussel species present. Mussels usually con-
tained only a few aspidogastrids each; the maximum intensity was 20
worms.
Adult worms (2–10 mm long) may be found throughout a bivalve host,
especially in the intestines, digestive gland, and pericardial and nephrid-
ial cavities. They have been characterized as ‘epithelial grazers’ (Huehner
Unionicolids · 269
et al. 1989). But even the short-term consequences of aspidogastrid infec-
tions to their hosts are poorly documented, and deserve additional atten-
tion. Unionids mount a defence by encapsulation (Huehner and Etges
1981).
Unionicolids
The life cycles of unionicolid mites are quite diverse. Although associated
with a variety of freshwater invertebrates (sponges, chironomid midge
larvae, gastropods), these mites are most commonly encountered in
unionid mussels. Unionicola aculeata requires a mussel for oviposition but
spends its adult life as a free-living planktivore. Unionicola intermedia, on
the other hand, appears to be parasitic on mussels (primarily Anodonta
anatina) at all life history stages (Baker 1976, 1977). Mites of this species
(both nymphs and adults) seem to induce an inflammatory response in
host tissues, then feed on haemocytes that congregate at the site of infes-
tation. Mites locate mussel hosts by a combination of phototaxis and sen-
sitivity to host-elaborated compounds, then enter through the siphons.
One or both sexes may display aggressive or territorial behaviours, a
phenomenon that contributes to considerable variation in the intensity
of mite infection. Edwards and Dimock (1988) reported that the entire
adult population of Anodonta imbecillis inhabiting a small North Carolina
pond appeared to be infected with Unionicola formosa year round. The
mean density of female mites was 53.4/mussel, with but a single male
generally in attendance. Approximately 90% of a large population of
Lampsilis siliquoidea from a Michigan lake were infected with Unionicola
fossulata (Mitchell 1965). The great majority of these mussels contained,
however, only one or two females with their single males. At an average
infection intensity of 1.96 mites/mussel, clearly this Lampsilis population
is in no danger of succumbing to mite infection.
Unionicolids may show considerable host specificity (Edwards and
Dimock 1995). Four species of mussels are common in the St. Marks
River, Florida: Elliptio icterina, Villosa villosa, Uniomerus declivus, and
Anodonta imbecilis (Downes 1986, 1989). Unionicola abnormipes is found
most commonly in Villosa, U. serrata is almost entirely restricted to
Uniomerus, and U. formosa is restricted to Anodonta. Unionicola fossulata is
found in both Villosa and Uniomerus. (U. formosa aggressively excludes all
other mite species from Anodonta.) Interestingly, Downes reported that
no mites were ever found in Elliptio, and in fact mites would not enter
her Elliptio in choice tests.
Vidrine (1990) examined 1000 individual mussels from a single site on
270 · Parasitism
Village Creek in Hardin County, Texas, representing 20 species. In six
visits during the course of a year he detected infection by one or more
of 12 mite species in 499 mussels. Most of the unionicolids appeared to
show striking preferences for individual unionid taxa. There has been
some concern, however, that certain characters used for the identifica-
tion of mite species may show host-induced phenotypic plasticity, con-
founding mite taxonomy (Downes 1990).
Some anecdotal evidence suggests that heavy mite infestation may
weaken mussels to the point that they succumb to relatively minor en-
vironmental perturbation, such as handling (Majumder and Pal 1992). In
any case, the occurrence of unionicolid mites would seem significantly
widespread to warrant more detailed study of their effect on mussel hosts.
Summary
The same attributes that make freshwater molluscs attractive as organisms
for ecological study, their predictable abundance and the ease by which
they may be located also make them attractive hosts for parasites. Mollusc
parasites are quite various; we concluded this chapter with brief reviews
of the aspidogastrid trematodes and unionicolid mites that parasitize uni-
onaceans. But the host/parasite relationship finds greatest development
in the Digenea, an order of 40000 fluke species entirely dependent on
molluscs at some stage in their life cycle.
Some digenetic flukes produce eggs or miracidia that infect their
mollusc host upon ingestion. In other cases, fluke eggs hatch into swim-
ming miracidia that actively seek host molluscs, burrowing into body
tissues. Through successive stages of development, trematode larvae
convert their host’s metabolic machinery to the production of cercariae,
which typically begin emerging several weeks post-infection to seek
second hosts (vertebrates, aquatic insects, or in some cases, other mol-
luscs).
Host snails may be entirely castrated by digeneans of the genus
Schistosoma, especially if infected as they approach adulthood and if multi-
ple miracidia are involved. Castration may not be complete, however,
depending on the age of the snail at infection. If infected as adults, snails
sometimes demonstrate increased egg laying during the several weeks
prior to the first emergence of cercariae, in an apparent effort to com-
pensate for castration to come. Snails may also (perhaps rarely) outlive
schistosome infection, regaining fertility after many weeks of shedding
cercariae. But schistosome infection generally seems to depress the sur-
Summary · 271
vivorship of host snails many times below that observed in uninfected
controls.
Infection by Fasciola seems more likely to lead to castration than infec-
tion by Schistosoma, although the little lymnaeids hosting the former seem
to mature more rapidly than most planorbid hosts of the latter, and thus
still manage to reproduce. Fasciola infection may improve the survivor-
ship of a host snail.
Trematode miracidia are usually encapsulated and destroyed shortly
after penetration by the molluscs they attack. Generally, only in those
circumstances where miracidia penetrate a host for which they have
evolved special countermeasures can development proceed. Host speci-
ficity varies greatly; S. haematobium populations seem to form very close
associations with particular Bulinus populations, while Fasciola miracidia
generally seem infective to a broad range of lymnaeids. The infectivity of
cercariae to molluscs as second intermediate hosts is considerably less spe-
cific than the infectivity of miracidia.
In the last century, Fasciola hepatica successfully invaded Australia, and
Schistosoma mansoni the Americas. Once initially introduced to a new
population of snails, trematode populations may show dramatic improve-
ment in infectivity after just two or three worm generations. The
defences mounted by snail hosts develop more slowly, but seem more
heritable. Susceptible and resistant strains of Biomphalaria can be obtained
by selective breeding, as well as strains showing different ages at onset of
resistance. At least occasionally, the defences of coevolved hosts have
reached such efficiency that their trematodes are more infective to closely
related (but heretofore unchallenged) snail populations.
The introduction of Schistosoma mansoni into the new world has been
characterized as a ‘gigantic though unplanned experiment’ in host/para-
site coevolution. It appears that challenging a new world Biomphalaria
population with an old world schistosome generally results in reduced
infectivity. Results are approximately equivalent to challenging an old
world Biomphalaria with a new world worm or even a new world
Biomphalaria population with a new world worm adapted to a second
snail population. Lessened resistance to infection is seen, however, when
an old world Biomphalaria is challenged with an old world schistosome
from a second population. This suggests that snails have a ‘head start’ in
the race between host and parasite, but that this head start lessens when
(as in the old/old comparison) trematodes become generally adapted to
the range of potentially suitable hosts by interpopulation dispersal.
The relationship between trematodes and their mollusc hosts may be
272 · Parasitism
summarized as ‘mutually aggressive’, the parasites having evolved to max-
imize their total reproductive output, even at the expense of host viabil-
ity. Trematodes usually show lesser effects upon hosts with which they
have not coevolved.
Although trematode parasitism is widespread in freshwater mollusc
populations generally, data on percentage prevalence are by far most com-
monly available for the hosts of human schistosomes. Considering dozens
of published surveys from low latitudes worldwide, it appears that only
about 1–5% of the individuals of an average snail population are infected
with schistosomes. Occasionally, however, populations 20%, 50%, or
even 90% infected may be encountered. Scattered data sets involving
other taxa of snails and worms tend to confirm the schistosome figures.
Sampling a snail population for trematode parasites presents a consid-
erable challenge in design. Multiple worm species may be involved, each
of which will likely show a very patchy distribution in both temporal and
spatial dimensions, influenced not only by the physical environment but
by the biology of at least two hosts. The prevalence of trematode infec-
tion will generally be a function of the size and age of the host snails
sampled.
The prevalence of parasitic infection may be related to snail popula-
tion density. Schistosoma japonicum infection in ten Philippine Oncomelania
populations ranged from 1.9% to 9.3%, averaged over 18 months. A time
series analysis designed to prospect for a relationship between monthly
percentage prevalence and snail population density returned some evi-
dence that Schistosoma infection rate was positively correlated to
Oncomelania density two months previous. The relationship was not quite
significant, but reminiscent of predator–prey oscillation. Thus even
apparently low parasitic infection rates may contribute to the regulation
of freshwater mollusc populations.
7 · Predation
Molluscs are edible, and easily subdued by predators of even the most
modest ferocity. Thus most aquatic vertebrates seem to eat molluscs, at
least under some circumstances. I include under this sweeping general-
ization not only fish, but semi-aquatic vertebrates such as amphibians,
certain reptiles, and a few mammals. In a charming review entitled
‘Enemies of the land and freshwater Mollusca of the British Isles’, Wild
and Lawson (1937) listed 20 vertebrate predators of Planorbis alone,
excluding fish but including such eye-catchers as the natterjack toad,
pheasants, and bats. In addition, a large fraction of the macroinvertebrate
predators of fresh waters recognize molluscs among their prey. Among
the fauna of Oneida Lake, New York, F. C. Baker (1918) recognized six
species of insect potentially dining on molluscs, one crayfish, eight
leeches, and two of the molluscs themselves (large Lymnaea). To the
Oneida rogue’s gallery Baker added 46 species of fish, 8 amphibians, 7
reptiles, 6 birds, and 3 mammals. Michelson’s (1957) review of possible
biological control agents for pulmonates included single paragraphs on
predatory flatworms, leeches, crustaceans, predatory molluscs, and
mammals, plus two paragraphs on birds, three on reptiles and amphib-
ians, and four on insects. Molloy and colleagues (1997) catalogued 176
predators of zebra mussels.
The present review will proceed from mammals through the verte-
brates and into the invertebrates, as predator body sizes decrease and den-
sities rise. We will focus on the evidence that predation may impact the
size of freshwater mollusc populations, or their life histories, prospecting
for community effects later in the chapter. We will find that the effects
of large vertebrate predators (mammals, birds, reptiles and amphibians)
are generally minor and patchy. Enclosure/exclosure experiments have,
on the other hand, sometimes returned clear evidence of substantial pre-
dation effects due to fish and larger invertebrates such as crayfish and
274 · Predation
waterbugs. The impacts of the smaller but numerically more common
invertebrates (sciomyzid flies, leeches, triclads, etc.) must also be impor-
tant, at least on occasion, although more difficult to document.
Immediate responses to predation threat may be characterized as ‘resist or
retreat’: crawl out, burrowing, shell shaking, clamping, or enclosure. On
a more evolutionary time scale, we will see that various modifications of
shell shape and weight have led to great variability in the protection that
shells offer their inhabitants. One might expect that all this variability in
predation pressure and prey vulnerability might lead to a substantial effect
on freshwater mollusc community structure. Such an effect, we shall see,
is surprisingly difficult to establish.
The malefactors
Mammals
Mammalian predators of freshwater molluscs include mink, otter,
raccoon, nutria, and various shrews, voles, and mice. All are dietary gen-
eralists, typically present in densities rather low to have substantial impact
on mollusc populations. Exceptional circumstances arise, however. Wild
Norwegian rats inhabiting the Po River valley, Italy, seem to prey heavily
upon Lymnaea stagnalis, Planorbis corneus, Viviparus ater, V. contectus,
Anodonta cygnaea, and Unio pictorum (Parisi and Gandolfi 1974). Nieder et
al. (1982) observed the behaviour of rats in enclosures along the Po for a
period of about four months. Although provided commercial pellet food,
the rats seem to have hunted Viviparus ater rather avidly. The number of
snails consumed rose along with the rat population, from about 5
snails/week for the founding pair of rats to 60 snails/week as the rat pop-
ulation increased to 16. The authors felt that the total of 502 broken shells
ultimately recovered over the four months (mean size about 30 mm)
would have been greater but for a ‘dearth of V. ater available in the
channel.’
Although primarily herbivorous, the North American muskrat
(Ondata zebithicus) will under some circumstances prey heavily upon
unionid mussels. Muskrats carry individual mussels to safe spots onshore
where, after a number have been consumed, shells accumulate in iden-
tifiable ‘middens’. They insert their incisors between the valves of their
victims and pry upwards, typically breaking one of the valves of thin-
shelled mussels but not of more heavily shelled species. As is true of pred-
ators generally, muskrats do not randomly sample all size categories of
The malefactors · 275
prey. Large mussels may be difficult to extract from the substrate, carry to
the shore, and open, while small mussels may escape notice, or prove
unworthy of the effort.
The effect of size-specific muskrat predation on mussel populations has
been well documented by Hanson et al. (1989) and Jokela and
Mutikainen (1995). Narrow Lake in central Alberta is home to large pop-
ulations of both muskrats and Anodonta grandis simpsoniana. Hanson and
his colleagues used an Ekman dredge to sample all depths of the north
basin of Narrow Lake, in May through August of two successive years.
They sieved samples through a coarse (6 mm) screen, forfeiting data on
the young-of-the-year and 1⫹ cohorts which, in any case, were too small
of size to be affected by muskrat predation. Hanson was able to estimate
the age of the remaining mussels by reference to annual growth-stoppage
lines on the shell. This technique seems to be especially reliable in
Canadian populations of anodontines, where smooth and continuous
shell growth during the summer is abruptly terminated by severe winter.
Hanson standardized all data (both age and size) to the 1986 annulus, then
combined to obtain the distribution shown in the lower half of Figure
7.1. I have converted the authors’ original figures from percentage to
absolute numbers, using their overall estimate of 1.23 million mussels in
the study area.
Hanson and colleagues removed all old shells from the muskrat
middens on the shores of Narrow Lake, then periodically resurveyed all
middens during the remainder of the season. Surveys resumed when the
ice cover opened the following spring. Shells collected during this 12-
month period were measured and aged to their 1986 annulus. The
authors considered that their total sample of 36771 individual mussel
remains (about 3% of the population aged 2⫹) underestimated the annual
predatory impact of muskrat. Although there is some danger of double-
counting should both valves remain intact but become separated, both
valves may become lost or broken at greater frequency (Convey et al.
1989). The top half of Figure 7.1 shows the age distribution of the
mussels consumed by muskrats in the north basin of Narrow Lake during
the course of a single year.
The overall age distribution of the Narrow Lake Anodonta population
was rather uneven; 1984 and 1981 seem to have been good years for
recruitment, and 1982 and 1983 rather poor. Comparison of the two dis-
tributions in Figure 7.1 shows that muskrats made very little use of that
large fraction of the Anodonta population younger than five years.
Selection was doubtless more directly by size than by age, however. The
276 · Predation
Figure 7.1. The 1986 age distribution of Anodonta grandis in Narrow Lake, Alberta.
(Data of Hanson et al. 1989.) Mussels aged 0⫹ or 1⫹ were excluded. The fraction
eaten by muskrats is graphed alone in the upper graph and cast onto the total pop-
ulation (as solid bars) in the lower graph.
authors noted that 86% of the mussels eaten were greater than 55 mm
long, while only 37% of the natural population fell into this category.
They felt that such predation pressures might result in selection for slow
growth in this unionid population. One might argue, however, that a
better life history ‘strategy’ would be to mature at a smaller size, and never
reach 55 mm at all. (Mussels from the Narrow Lake population typically
mature at age 4–5, or about 50 mm, and grow to about 70 mm.) We shall
return to the effects of predation on life history later in this chapter.
The malefactors · 277
Birds
Under some circumstances, ducks, geese, and swans (Anatidae) may
consume substantial numbers of molluscs. For example, the blue-winged
teal (Anas discors) is a surface- feeding, or ‘dabbling’ duck (subfamily
Anatinae). Mabbott (1920) reported the contents of 319 blue-winged
teal digestive tracts, collected in 29 states and four Canadian provinces,
every month except June. All contained some vegetable matter, especially
pondweeds and bullrushes. At most, 29 individuals contained any iden-
tifiable gastropod remains (Valvata, Amnicola, Physa, lymnaeids, planor-
bids) and at most seven contained identifiable bivalves (Sphaerium,
Pisidium). However, there is evidence that during the nesting season,
blue-winged teal may prefer invertebrate prey over vegetation, perhaps
for its high protein content. Swanson and Meyer (1977) collected 44
blue-winged teal from nesting grounds in North Dakota, 24 of which
were ‘laying’ (as judged by inspection of the reproductive tract) and the
remainder of which were ‘breeding’. Invertebrate remains constituted
99% of the gut volume of the former category, and 97% of the latter. The
molluscan component of the diet was substantial, comprising 23% of the
content of breeding blue-winged teal and 40% of the diet of laying
females. Swanson and Meyer mention Stagnicola caperata and S. palustris
as particularly important, with physids, planorbids, and sphaeriids also
part of the diet.
Invertebrate prey also seem to be especially preferred over aquatic veg-
etation by juvenile ducklings at some developmental stages. Reinecke
(1979) examined the diets of 41 juvenile black ducks collected from the
Penobscot River valley in Maine. There was a pattern of decreased inver-
tebrate food use with age, from 95% of the gut content dry weight for
downy juveniles to 84% for partially feathered to 34% for fully feathered
juveniles. Downy ducklings were surface feeders, and did not often
capture molluscs. But molluscs (unspecified clams and snails) comprised
24% of the gut content dry weight of 12 partially feathered juvenile black
ducks.
Diving ducks (subfamily Aythyinae) also may eat substantial numbers
of molluscs as juveniles, and as egg laying females. In southwestern
Manitoba, Bartonek and Hickey (1969) reported that snails were the
second most common food item in the guts of female and juvenile can-
vasbacks (Aythya valisineria) and lesser scaups (A. affinis) collected during
the breeding season. Males were much more vegetarian in diet. Cottam
278 · Predation
(1939) summed data on gut contents from 1051 adult lesser scaup from
30 states of the United States and 5 Canadian provinces, over every
month of the year. A great diversity of molluscs (both freshwater and
marine) combined for 24.9% of the volume. Freshwater taxa mentioned
specifically were Amnicola (1.7%), Helisoma newberryi (1.6%), and
‘Planorbis’ (probably other Helisoma, 1.6%). The data of Rogers and
Korschgen (1966) suggest that lesser scaup molluscivory may be a sea-
sonal phenomenon. Few molluscs were found in the guts of their samples
of lesser scaups breeding in Manitoba or overwintering in Louisiana. But
the guts of 88 migrating individuals collected on the Keokuk Pool (Pool
19) of the Mississippi River between Iowa and Illinois contained, by
volume, 85% mollusc remains.
Thompson (1973) identified nine taxa of gastropods and four of
bivalves in the guts of 270 lesser scaup he collected from Pool 19. Of
these, 76% had eaten Sphaerium striatinum, 40% had eaten Musculium
transversum, and 10% had eaten unionids. Hydrobiids seem to have been
the most important gastropod food, including Somatogyrus (33%),
Fontigens (19%), and Amnicola (5%). Campeloma (21%), Lioplax (10%), and
Pleurocera (8%) were also commonly eaten. Pulmonates were oddly rare
in the diet of this sample of lesser scaup from the Mississippi, but may
have fallen among the ‘unidentified gastropods’ found in 68% of the guts.
Large flocks of diving ducks visit Lake Erie in the vicinity of Point
Pelee from late October until mid-December. Individuals are frequently
observed surfacing with Dreissena in their beaks, which they manipulate
and swallow. Hamilton et al. (1994) anchored exclosures of an unusual
design (covered tops, open on all four sides) at ten sites in 5–7 m of water.
Two separate rock beds were involved, 250–350 m offshore, surrounded
by soft substrate largely uninhabited by Dreissena. Rocks were removed
from each exclosure by divers from September to November, and
again in May. November samples did in fact reflect significantly greater
mussel biomass (g/cm2 rock surface area) under exclosures than in
uncaged control areas. This seemed to be due to the selective removal of
the larger size classes of mussels. But such losses as may have been inflicted
on the largest mussels seem to have been compensated for by increased
substrate availability for smaller individuals. Even in November, the
differences in mussel density between exclosure and control were not sig-
nificant, and overall population size distributions had readjusted by the
spring. Hamilton concluded that ‘ducks had little lasting impact on
mussel populations, but mussel abundance may have determined duck
concentration in the area.’
The malefactors · 279
Among the birds there are several rather spectacular examples of diet
specialization on freshwater molluscs. The limpkin (Aramus guarauna) is a
wading bird of the Order Gruiformes (rails, coots, and cranes) with a
narrow, hooked bill. It ranges from Florida to Argentina. Although some
small quantities of insect larvae and vegetation (especially seeds) are occa-
sionally found in stomach analyses, its normal diet seems to be almost
entirely composed of large viviparid and ampullariid snails. According to
Cottam (1942), it lifts its prey (typically Pomacea paludosa) from the sub-
strate, turns it, and seats it back firmly in the mud aperture up. It then
extracts the meat neatly from the shell, sometimes casting the operculum
aside.
More famous than the limpkin is the snail kite (Rostrhamus sociabilis), a
diurnal bird of prey (family Accipitridae) also ranging from Florida to
Argentina. It has an oddly long and hooked beak (for a raptor) which it
uses to great effect on Pomacea. Snyder and Snyder (1969) describe how
it courses over the marshes until it spots a snail at the surface or on marsh
vegetation. It then dives on the snail, siezes it in a talon, and carries it to
a low tree. The bird perches on one foot, holds the snail aperture-up with
its other foot, and extracts the meat cleanly with its unusual bill. For many
years it was thought that the kite was a strict specialist on Pomacea. It is
now clear that South American kites commonly attack Marisa, although
their extraction efficiency is not as great (Snyder and Kale 1983). They
may also infrequently attack small vertebrates, especially turtles.
Turtles
Returning to the subject of omnivory, four species of kinosternid turtles
are common in Oklahoma: the yellow mud turtle (Kinosternon flavescens),
the Mississippi mud turtle (K. subrubrum), the common musk turtle
(Sternothaerus odoratus) and the Mississippi musk turtle (S. carinatus).
Mahmoud (1968) examined the gut contents of several hundred individ-
uals of all four species, collected statewide May to October. Mollusc
remains were found in over 90% of the stomachs examined, ranging only
from 24% of the gut volume in (N⫽121) K. flavescens to 32% in (N⫽
178) K. subrubrum. The pie diagrams in Figure 7.2 show example data for
the Mississippi musk and Mississippi mud turtles, combining crustaceans
and insects as arthropods and carrion and aquatic vegetation as ‘other’.
Mahmoud considered all four turtle species ‘euryphagic’ bottom feeders,
eating a wide variety of foodstuffs as available in their habitats.
Laboratory feeding trials suggested some differences in the predatory
capabilities of mud and musk turtles, however. Mahmoud performed 47
280 · Predation
Figure 7.2. The bar graphs show the frequency at which musk turtles (S. carinatus)
and mud turtles (K. subrubrum) fed on four types of prey in 235 feeding trials, and
the pie diagrams compare wild-collected gut contents by volume. (Data of
Mahmoud 1968.)
feeding trials, each involving 20 turtles (five of each species) and a bucket
of mixed prey items: snakes, tadpoles, beetles, fly larvae and snails (Physa).
The bar graphs of Figure 7.2 show the number of occasions (in 47⫻5⫽
235 opportunities) upon which each of four prey items was eaten by two
of Mahmoud’s turtle species. Mud turtles fell first upon the tadpoles,
beetles, larvae, and other squirming prey, while musk turtles seemed to
prefer snails. So although their diets in the wild are not strikingly differ-
ent, Mahmoud characterized mud turtles as ‘more aggressive, fast-
moving, and alert’, suggesting that they may be less likely to pick up snails
when other prey are available.
Given their omnivory, one might expect turtle populations to vary in
diet as different prey becomes available in different environments. The
yellow mud turtle is most common in Mexico and the southern United
States; it reaches the northern limit of its range in Missouri. Kofron and
Schreiber (1985) reported that the gut contents of 34 yellow mud turtles
inhabiting a small pond in northeastern Missouri (51 captures) comprised
primarily mollusc remains. Of 325 prey items recognized in faeces, 268
were Physa, Helisoma, pisidiids, or unidentified gastropods. The remain-
der were insects, fish, and crayfish. The predominance of molluscs in the
faeces of this population does not seem to be an artifact of the
authors’ sampling method. Kofron and Schreiber found that the faeces of
The malefactors · 281
the co-occurring Blanding’s turtle (Emydoidea blandingii) contained cray-
fish, insects, fish, and amphibian remains, but no mollusc shells at all.
In southern Mexico’s Laguna Escondida, the mud turtle Kinosternon
leucostomum eats very few molluscs (Vogt and Guzman 1988). By volume,
freshwater snails comprised only 1.3% of the stomach flushings of 140
mud turtles trapped during the course of one year; by item count they
comprised 10%. But unidentified molluscs were the majority prey item
in the guts of 13 Staurotypus triporcatus from the Laguna, as judged by item
count, second only to seeds and fruits in volume. In light of its ‘large
alveolar surfaces and massive jaw musculature’, Vogt and Guzman con-
sidered Staurotypus to be a ‘mollusc specialist’.
Salamanders
The term ‘euryphagic’ has found great use in reviews of salamander
biology. Workers generally report that amphibians of the order Caudata
will ingest any animal in their environment suitable to their mouth gape.
For example, five species of salamanders inhabit the Hubbard Brook
experimental forest in New Hampshire: the terrestrial Plethodon cinereus,
terrestrial subadults of the newt Notophthalmus viridescens, and three
species of more aquatic habit, the dusky salamander (Desmognathus fuscus),
the brook salamander (Eurycea bislineata), and the spring salamander
(Gyrinophilus porphyriticus). Burton (1976) reported at least occasional
occurrence of gastropod and sphaeriid shells in the guts of adults of all
three of these more aquatic species, and in the guts of larval Desmognathus
as well.
Larval tiger salamanders (Ambystoma tigrinum) and larval newts (N. vir-
idescens) are the only abundant aquatic vertebrates in McGuire’s Pond,
south of Carbondale, Illinois. Brophy (1980) reported that 57% of the
larval newts and 21% of the larval Ambystoma he collected between
March and September contained Physa in their guts. He recorded
Sphaerium in 9% and 27%, respectively. By weight, freshwater molluscs
were the second most important food category for both species, after
ostracods, increasing in proportion through the season, as the larvae grew.
It would seem reasonable that the larger the salamander, and the more
aquatic its habit, the greater danger it would pose to freshwater molluscs.
A siren (Siren lacertina) measuring 369 mm from snout to vent, collected
in a stagnant ditch near Norfolk, Virginia, contained 120 individual
Musculium, 8 Helisoma, 5 Gyraulus, and 2 Pseudosuccinea (Burch and
Wood 1955). The remainder of the ingesta were incidental. A 325 mm
282 · Predation
specimen collected in Orange Lake, Alachua County, Florida, defecated
the shells of 323 Planorbella scalare (up to 9.1 mm diameter), 1 Physa and
1 Musculium (Moler 1994).
Fish
The vertebrates we have discussed thus far, mammals, birds, reptiles and
amphibians, unquestionably do on occasion consume large numbers of
freshwater molluscs. But these predators themselves are typically at low
density and/or patchy in their impact. It is difficult to imagine that turtle
or duck predation, for example, could measurably affect a freshwater
mollusc population, simply because there are relatively few turtles or
ducks per pond. Fish predation would seem to be a different matter.
Among the more thorough comparative surveys of fish diet is that of
Maitland (1965), working in tributaries of the River Endrick, Scotland.
The study streams were generally shallow, rocky, riffle-and-pool environ-
ments, somewhat acidic, with smallish mollusc faunas: Ancylus, Lymnaea,
and Pisidium (species not identified). They comprise less than 1% of the
benthos. The five most common fishes of this region are salmon (Salmo
salar), trout (Salmo trutta), minnows (Phoxinus phoxinus), stone loach
(Nemacheilus barbatula), and three-spined stickleback (Gasterosteus aculea-
tus). Maitland examined gut contents from several hundred individuals of
most species, netted monthly at six sites. He reported some dietary spe-
cialization: trout ate most commonly at the surface, loach and stickleback
rarely. But all five fishes fed heavily on the most common aquatic inver-
tebrates (insects and crustaceans) and all except the loach ate molluscs, at
least occasionally.
Hartley (1948) examined the stomach contents of all 11 fishes inhab-
iting the River Cam and one of its small tributaries in Barrington,
England. He collected his fish by a variety of methods over several years,
and his sample sizes ranged from 8 perch to 391 gudgeon. Molluscs seem
to be much more common in the Cam and its tributaries (both quite
alkaline) than in the rather acidic Endrick. And setting aside perch and
loach for their small sample sizes, six of the nine fishes inhabiting the
Cam can fairly be considered molluscivorous (brown trout, stickleback,
eel, gudgeon, dace and roach). Indeed, Potamopyrgus jenkinsi was (by item
count) the single most common food identified in the stomachs of the
dace (Leuciscus leuciscus) and the roach (Rutilus rutilus). Numerical super-
iority for mollusc prey disappeared when Hartley summed the categories
of insect larvae in his data set. But clearly the combination of fish abun-
dance, their omnipresence, and their general willingness to consume
The malefactors · 283
Figure 7.3. Frequencies of four common food items in the guts of cutthroat and
rainbow trout captured April–October in sequential sections of Marshall Creek,
British Columbia. (Data of Bryan and Larkin 1972.)
clams and snails, increases the likelihood that they may substantially
impact mollusc populations.
The large literature on the diets of various trout species paints them as
quite general predators. Brook, cutthroat, and rainbow trout are all
common in Marshall Creek, British Columbia, downstream from a small
trout hatchery. Bryan and Larkin (1972) divided a study section of the
creek into six subsections of about 10–20 m each and sampled the trout
population using electroshocking gear. Fish were anaesthetized and their
stomach contents ‘flushed out’, then tagged and released. Figure 7.3
shows the proportions of four major food types observed in trout stom-
achs which Bryan and Larkin sampled down their transect between April
and October, as item counts. Prey categories not shown included non-
chironomid dipterans, fish and their eggs, tubificid worms, and non-tubi-
ficid oligochaetes. In any case, the spatial variation over this small study
area is striking. ‘Gastropoda’ (unspecified) predominated in the two most
upstream subsections, while insects of various sorts became more
common in trout diet downstream. Trout population densities were
greater in Bryan and Larkin’s upstream sites, as were their fish capture
rates. It is possible that the dietary shifts shown in Figure 7.3 are partially
explained by severe cropping of the insect resource upstream.
284 · Predation
Differing availability of prey types does not seem to be the entire story
in Marshall Creek, however. For fish caught more than once, Bryan and
Larkin calculated a correlation coefficient between first and second gut
contents and converted this to its standard normal deviate ‘within fish’,
(Zw). The authors compared values of Zw with similar statistics calculated
between trout of the same size and species, caught at the same time (Zb).
In some circumstances they felt that their data indicated food specializa-
tion in individual trout, estimated as Zw ⫺Zb. Individual specialization
was most commonly on gastropod prey. The degree of food specializa-
tion in trout recaptured in the same subsection was not significantly
greater than that shown by trout recaptured in different subsections. It
would appear that at least part of the dietary variation displayed by fish
populations may be due to differing individual foraging strategy or search
image among individual fish.
Large, bottom-dwelling fish with powerful jaws seem especially dan-
gerous to freshwater molluscs. At 16 cm standard length, roach begin to
switch from diets of algae and macrophytes to zebra mussels in Poland’s
Lake Sniardwy (Prejs et al. 1990). By the time the fish reach 28 cm, they
prey upon mussels almost exclusively. The massive jaws mounted by
roach, freshwater drum, and round gobies have brought these species to
the attention of researchers prospecting for controlling agents for zebra
mussels in North America (French and Bur 1993, Nagelkerke and
Sibbing 1996, French and Love 1995, Ghedotti et al. 1995). Jude (1973)
examined the anterior guts of 344 gizzard shad netted from Pool 19 of
the Mississippi River in Iowa. (This may be remembered as the site of
several studies of duck diet.) Jude sampled stations many kilometres apart,
over the course of 18 months, and collected fish 10–44 cm in standard
length. Shad smaller than 15 cm seemed to eat little but algae. Older fish
primarily ate pisidiids less than 5 mm long, which they crushed in their
powerful gizzards. The hydrobiid Fontigens also appeared among shad gut
contents occasionally, along with smaller numbers of microcrustaceans.
Cinetodus froggatti is an ariid catfish inhabiting the Fly River of New
Guinea. The gut contents of 16 specimens, ranging in length from 23 cm
to 42 cm, were reported by Turner and Roberts (1978). Three guts con-
tained no food, but from the remainder Turner and Roberts identified
14 species of molluscs, representing 13 genera and 8 families. The catfish
had eaten, without crushing, heavily shelled thiarids, delicate planorbids,
viviparids up to 22 mm, and brackish-water erodonid bivalves. No non-
molluscan food items were observed.
Stein et al. (1975) found molluscs in the guts of 31.1% of 415 carp
The malefactors · 285
(Cyprinus carpio) netted from Skadar Lake in the former Yugoslavia. On
occasion some of these carp ‘had their gut literally packed with shells and
shell fragments’. The relative proportions of the molluscs in the diets of
these carp did not match their proportions in Ekman grab samples of lake
bottom. Stein and colleagues identified eight mollusc taxa, in order of
abundance: Pyrgula annulata (from the minor rissoidean family
Micromelanidae) ‘form 1’ (with carinae), Valvata piscinalis, Dreissena,
Pisidium (perhaps several species), P. annulata form 2 (smooth), Viviparus
viviparus, Radix auricularia, and the pleurocerid Amphimelania holandri.
Carp seemed to prefer Valvata by a large margin. Though representing
only a little over 14% of the mollusc fauna of Skadar Lake (by direct
count), Valvata shells comprised about 82% of the mollusc remains iden-
tified in carp guts. Stein et al. attributed the high ‘electivity’ of Valvata to
its convenient size, its thin shell, and the high proportion of living snails
to dead Valvata shells on the lake bed. We will return to predator discrim-
ination among mollusc species in a discussion of community effects later
in this chapter.
The large and diverse family Cichlidae includes among its number
quite a few fishes with specializations for molluscivory (McKaye et al.
1986, Witte et al. 1990). Macropleurodus bicolor is an ‘oral sheller’, seizing
its prey and jerking violently to tear the mollusc’s body from its shell.
Haplochromis ishmaeli is a ‘pharyngeal crusher’, generally breaking its
prey’s shell entirely. Some interest has been directed toward such animals
as possible biological controls against the pulmonate hosts of schistoso-
miasis (Slootweg et al. 1993).
Slootweg (1987) performed a series of laboratory feeding trials involv-
ing these species (and two others) wild-caught from Tanzanian Lake
Victoria and cultured populations of Biomphalaria glabrata. Typical results
from two individual fish are shown in Figure 7.4. Here fish isolated in
tanks were offered 130 Biomphalaria per experiment; ten snails from each
of 13 single-millimetre size classes. (Slootweg presented evidence else-
where that food input at this level was more more than sufficient to satiate
his study animals.) Experiments were performed four times over a study
lasting several days.
Figure 7.4 shows that a larger (120 mm) H. ishmaeli preferred snails
ranging from 7 mm to 9 mm in diameter, while an (88 mm) M. bicolor
preferentially selected snails in the 4–5 mm range. These preferences fit
Slootweg’s predictions from a simple optimal foraging model based upon
two functions of shell size: tissue dry weight of the snail and the handling
time required by the two fish. Somewhat unexpectedly, the ‘oral sheller’
286 · Predation
M. bicolor entirely crushed all prey smaller than 8 mm. Slootweg noted
that previous observations had been of attacks on natural prey (usually
Melanoides and Bellamya) with shells much stronger than Biomphalaria.
Oral shelling may be specialization of the sort catalogued by Liem (1980):
its inclusion in the repertoire does not involve trading off simpler, more
generalized behaviours such as crushing. Indeed, by resort to oral shell-
ing M. bicolor was able to eat a few of the largest Biomphalaria offered on
occasion.
In the last twenty years a great deal of research has been directed
toward the influences of predatory fishes upon benthic community struc-
ture (Gilinsky 1984, Power 1990, Diehl 1992). Among the fishes com-
monly involved in such studies have been sunfishes with demonstrated
molluscivorous preference. Sadzikowski and Wallace (1976) compared
the stomach contents of three size classes of the bluegill (Lepomis macro-
chirus), the green sunfish (L. cyanellus) and the pumpkinseed (L. gibbosus)
wild-collected from two interconnected Michigan lakes over 40 summer
days. Example data from (N⫽166) bluegill and (N⫽122) pumpkinseeds
are shown in Figure 7.5.
The malefactors · 287
Figure 7.5. Frequencies of three common food items in the guts of bluegill and
pumpkinseed sunfish, of three size classes. (Data of Sadzikowski and Wallace 1976.)
288 · Predation
When young (29–70 mm fork length), both species of sunfishes pri-
marily eat cladocerans and copepods, with a substantial chironomid com-
ponent. Upon reaching the ‘medium’ size class, however, the
pumpkinseed begins to switch to molluscs. Sadzikowski and Wallace
reported 44.3% of the items in medium-sized pumpkinseeds were
unspecified ‘gastropods’, and 0.7% unspecified bivalves. The distinction
between the two sunfishes is even more striking in the 111–160 mm
‘large’ size class, where 50% of the items in pumpkinseed guts were mol-
luscs, but only a ‘trace’ of such material was detectable in the bluegill diet.
It should be noted that adult bluegills have been found to eat Physa,
Helisoma, and ancylids in other situations (Crowder and Cooper 1982).
But pumpkinseeds have exceptionally wide pharyngeal arches and round,
molar-like teeth or plates which are generally interpreted as specializa-
tions for molluscivory. They tend to inhabit weedier areas than other sun-
fishes. Molluscs seem to comprise a large portion of the pumpkinseed
diet throughout its range, consistently (Laughlin and Werner 1980).
There is some evidence of a positive relationship between pumpkinseed
density and snail density (Mittelbach 1984).
Our attention now turns from studies of fish diet and gut content to
experimental demonstrations of the impact fish may have upon mollusc
populations. Since the 1970s we have been blessed with a wealth of
enclosure/exclosure experimentation bearing on this question (Table
7.1). Some studies have been simple tests to see whether mollusc popu-
lation densities may rise when predatory fish are excluded, while others
have involved artificial enclosures of fish as well as exclosures. The enclo-
sure of predators clearly increases the likelihood that a researcher will
detect a predation effect, since even if overall natural predator densities
are mimicked, the temporal and spatial patchiness of the predator popu-
lation will be constrained, to the ultimate detriment of the prey. But
enclosure of predators is nevertheless valid. Clearly if no effect is detected
under such conditions, a genuine natural effect would seem quite
unlikely.
Brown and DeVries (1985) investigated a phenomenon as well known
in freshwater molluscs as elsewhere: high population density in a subop-
timal environment. Although the author’s field rearing experiments
showed that Lymnaea elodes (of any origin) have better growth and fecun-
dity in their larger, permanent pond (‘F’), observed adult population den-
sities were greater in their small, temporary pond (‘A’). Their quantitative
survey of mollusc predators suggested that predation by the mudminnow
Umbra lima upon the egg and juvenile Lymnaea might be sufficient to
The malefactors · 289
Table 7.1. Exclosure tests of fish predation effects upon freshwater mollusc
populations
Effect on
Workers Population Location density
Exclosure (only)
Robinson and Wellborn Corbicula Texas reservoir Y
(1988)
Osenberg (1989) Snail community Michigan lake N
Bronmark (1988) Snail community Swedish lake N
Walter and Kuiper (1978) Pisidiid community Swiss lake Y
Thorp and Bergey (1981) Mollusc community S. Carolina reservoir N
Keller and Ribi (1993) Viviparus Lake Zurich Y
Enclosure/exclosure
Brown and DeVries (1985) Lymnaea elodes 2 Indiana ponds Y
Bronmark et al. (1992) Snail community 2 Wisconsin lakes Y
Martin et al. (1992) Snail community Tennessee pond Y
Gilliam et al. (1989) Pisidiid community New York stream N
Date 1 2 3 4 5 6
Open controls
May 0.0 0.0 0.0 1.0 1.5 1.0
July 0.5 1.5 0.0 1.5 0.0 1.0
Sept 1.0 1.0 0.0 1.0 2.0 1.0
Exclosures
May 3.5 1.0 19.0 1.0 71.0 246.5
July 0.5 4.5 12.5 0.0 87.5 10.0
Sept 6.5 5.0 9.0 8.5 56.0 50.5
Note:
Only the September difference is significant.
Sources: data from Robinson and Wellborn (1988).
densities and predator effects are temporally variable and spatially patchy,
a researcher’s difficulties are compounded. Where macrophytes add
habitat structure to some sites on some dates but not to others, where
mollusc populations reproduce, migrate and colonize, and where inver-
tebrate predators and parasites complicate matters, it is a wonder that any
significant effects of exclosure have been detected under any circum-
stances at all. The demonstration of such effects in the six studies cited in
Table 7.1 stands as testimony to the potential for predation to regulate
freshwater mollusc populations.
Crustaceans
Crayfish are omnivorous; their diet in the wild generally matches the
availability of plants and animals (living and dead) in their immediate
environment. If their immediate environment contains freshwater snails,
crayfish seem to manifest a taste for them. Covich (1977) wild-collected
ten Procambarus acutus from Massachusetts ponds and offered them Physa
gyrina, leaves from Elodea, or both. In 60 one-hour trials, the crayfish
never ate fewer than five (of ten) Physa offered, and often ate the entire
lot. No individual crayfish preferred Elodea in mixed offerings, but several
seem to have strongly preferred snails. Hanson et al. (1990) demonstrated
a strong inverse relationship between the density of Orconectes virilis and
294 · Predation
that of snails (Stagnicola elodes and Physa gyrina) in small, artificially stocked
ponds. Interestingly, the effect of crayfish density upon macrophyte
growth was stimulatory in some circumstances, perhaps because the cray-
fish tended to reduce snail grazing pressure (Chambers et al. 1990).
Crayfish species may differ in their threat to snail populations.
Orconectes rusticus consumed an average of about 150 Amnicola lustrica per
day in the laboratory feeding trials of Olsen et al. (1991), significantly
greater than the numbers consumed by similar-sized O. virilis (about 90
snails/day) or O. propinquus (about 60 snails/day). The large North
American Procambarus clarkii has been implicated in the disappearance of
Biomphalaria and Bulinus from regions of Kenya (Harper et al. 1990,
Hofkin et al. 1991, 1992).
A number of laboratory feeding experiments have been directed
toward understanding the potential impact of crayfish predation on pop-
ulations of nuisance bivalves (Love and Savino 1993, Martin and Corkum
1994). Orconectes limosus are commonly observed eating Dreissena from
steelon netting used to protect various works in Poland. Piesik (1974)
found that crayfish in the 34–47 mm size range consumed 20–40 mussels
per day, modally of the smallest (1–3 mm) size class. A 67 mm crayfish
ate 48 mussels/day, modally of the 4–5 mm class, and a 90 mm crayfish
ate 72 mussels/day, modally 6–7 mm. Most of Piesik’s experiments
reflected a much greater consumption of mussels by female crayfish than
by males, a result he attributed to the transiently poor condition of males
in mating season. MacIsaac (1994) performed similar experiments using
larger mussels and Orconetes propinquus from the North American Great
Lakes. Feeding rates were a more modest 4–5 mussels/day in his 3–5 mm
size class, reduced to about a single mussel per day in size classes 8–14
mm. Crayfish size and sex were again important. MacIsaac found that
predation rate upon (especially large) mussels could be reduced by the
inclusion of alternative macrophyte foods, but that the crayfish contin-
ued to prefer mussels. Covich et al. (1981) reported that Procambarus clarkii
consumed 11 Corbicula per day, in the size range 4–6 mm. Although
Procambarus seemed unable to open Corbicula larger than 6 mm, Cambarus
bartonii successfully attacked clams up to 9 mm.
Some of the increase in mollusc diversity observed in the exclosure
experiments reviewed in the previous section (Table 7.1) may be attrib-
utable to reduced crayfish predation, particularly where the mesh size
employed was fine. Enclosure/exclosure experiments involving the
introduced Orconectes rusticus suggest that the crayfish may significantly
reduce macrophyte and snail abundance in northern Wisconsin lakes
The malefactors · 295
(Lodge and Lorman 1987), and snail and zebra mussel abundance in
Michigan streams (Perry et al. 1997). The quantity of periphyton also
seems to be negatively related to crayfish abundance, since periphyton is
dependent on macrophyte surface area, but its quality (as chlorophyll a)
increases as snail grazing pressure is reduced (Lodge et al. 1994).
Insects
Large predaceous hemipterans are fairly common in fresh waters, those
hunting in the benthos constituting a natural threat to whatever molluscs
they encounter. The North American water bug Belostoma flumineum
(often exceeding 20 mm in length) seizes pulmonate snails between its
front tibia and femur, plunges its beak into the exposed foot, and sucks
all tissues cleanly from the shell. Crowl and Alexander (1989) reported
laboratory predation rates of 3.5 Physa/day for male Belostoma encum-
bered with eggs, up to 5.5 Physa/day for unencumbered females. The
laboratory observations of Kesler and Munns (1989) suggested 1.8
Physa/day for 4.5 mm Belostoma, up to 4.5 Physa/day for 18 mm bugs.
Kesler and Munns cited other reports ranging from 4.1 small Physa and
Gyraulus/day for 12–15 mm Belostoma to 0.3 Physa/day for the smaller
belostomatid Abedus.
There is good reason to expect natural predation rates to be much
lower than those observed in the laboratory. Snails will be more difficult
to find in the wild, and other foodstuffs will be available to the bugs.
Kesler and Munns implanted four pairs of plastic cans (40 cm diameter)
in the grassy bottom of Peckham Pond, Kingston, Rhode Island. Into all
eight they introduced four Pseudosuccinea columella, and into one of each
pair they introduced a single adult Belostoma. (The enclosures also con-
tained natural densities of Pseudosuccinea and Physa vernalis, as well as an
occasional resident Belostoma.) After nine days the authors thoroughly
censused the contents of all cans for bugs and freshly dead shells, con-
cluding that a more natural estimate of Belostoma predation rate might be
about 0.5 snails/bug/day. The authors could detect no predator prefer-
ence for either snail species.
Kesler and Munns monitored the density of Belostoma in Peckham
Pond, along with the densities of its two pulmonate prey Pseudosuccinea
and Physa, biweekly over four years. Samples were taken principally by
dip net along a 22 m transect. The authors felt that their method system-
atically underestimated Belostoma density, since the insects ‘could dive
below the volume swept by the net’. I should think the underestimate of
snail density would be worse, as all snails inhabiting the bottom sediments
296 · Predation
Figure 7.6. The line shows the combined density of Physa and Pseudosuccinea (all
age classes) in Peckham Pond, Rhode Island, averaged over 1982–84 and 1986 . Bars
indicate the potential snail consumption rate (per m3) of the Belostoma population
standing in 1983. (Data of Kesler and Munns 1989.)
will pass underneath a dip net as well. In any case, the Physa population
appeared to be annual and semelparous, adults reproducing and dying in
June. Two semelparous generations of Pseudosuccinea pass per year in
Peckham Pond, one born in early May and reproducing in late July, the
other born in July and reproducing in May.
Figure 7.6 shows the combined densities of Physa and Pseudosuccinea,
over all age classes, averaged over four years of observations. The early
May peak reflects the birth of the summer Pseudosuccinea cohort, added
to overwintering Physa. The late May decline in snail density is due to
the initially high Pseudosuccinea mortality, as the Physa population is laying
eggs and expiring. Water bugs seemed to be rare during the early
summer; Kesler and Munns recorded none at all in May and early July
1983, and only 1.3/m3 in June. But the Belostoma density increased dra-
matically in late July to 41 bugs/m3, and remained high in late August
(about 15/m3), dropping back to zero in late October.
I derived the estimates of the ‘Belostoma demand’ for snails shown in
Figure 7.6 by integrating between the (only six) observations of Belostoma
density provided by Kesler and Munns. The average bug density between
The malefactors · 297
each pair of observations was multiplied by the number of days in the
interval and author’s field predation estimate of 0.5 snails/bug/day.
During the late summer and autumn, it appears that Belostoma could eat
more snails than are present in Peckham Pond. Clearly snails are not the
only prey item recognized by Belostoma. And it should immediately be
noted that the snail densities are four-year averages, while the bug density
is for 1983 only. Kesler and Munns state that the density of bugs in late
July 1983 was exceptionally high. It is not clear why bug densities were
not reported for other years, but one might expect a great deal of spatial
and temporal variation in the density of such a mobile predator.
(Discussion of habitat refuges in Peckham Pond will be postponed until
later in the chapter.) In any case, the data of Kesler and Munns suggest
that, at least occasionally, predaceous water bugs have the potential to
exert substantial pressure on mollusc populations.
Diving beetles of the family Dytiscidae, numbering about 3000
species, are common inhabitants of ditches, lakes, and quiet streams
worldwide. They are voracious predators, typically including on their
menu all manner of aquatic insects and crustaceans, tadpoles and fish fry
much larger than themselves. Their benthic larvae (often called ‘water
tigers’) may pose an even greater threat to freshwater snails than the adult
beetles, which primarily hunt in the water column. I am aware of several
anecdotal reports of dytiscid predation upon freshwater snails (Bequaert
1925, Michelson 1957) but no formal studies.
There are about 1900 species in the coleopteran family Lampyridae
(fireflies, lightning bugs), distributed worldwide. Larvae typically burrow
in moist soils and are predatory upon soft-bodied invertebrates. Species
adapted to especially muddy substrates along stream and pond banks may
pose some threat to more amphibious freshwater snails. For example,
lampyrids common in central New York include Photinus consanguineus,
P. scintillans, Photurus pennsylvanica, and Pyropyga fenestralis (Hess 1920).
Photurus larvae roam actively at night through moist, grassy fields, and the
two Photinus species are generally subterranean, perhaps primarily
hunters of earthworms. But according to Hess, Pyropyga larvae ‘can be
found wandering about on the ground at the edge of streams apparently
in search of food.’ Hess reported that Pyropyga will eat snails in laboratory
feeding trials, although whether the victims were aquatic or terrestrial is
not clear.
The larvae of the Asian lampyrid Luciola are semi-aquatic, and have
been investigated for their potential as biological control agents for
Oncomelania and the lymnaeid hosts of livestock flukes (Michelson 1957).
298 · Predation
Kondo and Tanaka (1989) demonstrated that the larvae of L. lateralis may
be reared on both Semisulcospira libertina and newborn Pomacea canaliculata
(up to 11 mm) with good growth and survivorship. Second, third, and
fourth instar larvae consumed an average of 0.7, 2.3 and 3.2 snails/day.
Luciola lateralis was among four lampyrid species released in the Hawaiian
islands during the 1950s, in unsuccessful efforts to control both Lymnaea
ollula and the giant African land snail Achatina, an agricultural pest
(Chock et al. 1961).
Dipteran flies of the family Sciomyzidae (‘marsh flies’) have also been
the focus of considerable interest as possible biological control agents for
pulmonates of medical and veterinary importance. Berg (1964, 1973)
and his co-workers examined the life histories of a large fraction of the
approximately 600 species of sciomyzids worldwide, concluding that the
great majority are, as larvae, strictly predatory on freshwater snails. Adults
typically lay eggs on emergent vegetation standing in very still, shallow
water. The larvae (1 mm to 10 mm) have gut air-bubbles and hang from
the surface film by posterior tufts of hydrophobic hairs, occasionally fa-
stening onto emergent vegetation or crawling over moist soils at the
water’s edge. They sink their mouth hooks into the exposed foot of any
snail they may encounter, including snails substantially larger than them-
selves. Prey typically draw the fly larvae into their shells by muscular
contraction, where the larvae engorge themselves. In the laboratory, a
single sciomyzid larva commonly ingests more than a dozen snails before
forming a floating puparium.
An interesting exception is provided by the European and North
American sciomyzid genus Renocera, whose newly hatched larvae swim
to the bottom hunting pisidiid clams (Foote 1976). They pass their first
instar entirely submerged, eating one or two clams. Second- and third-
instar Renocera larvae hang inverted from the surface film, as usual,
actively probing the bottoms of shallow pools for the 11–13 pisidiid clams
they will need prior to pupation. Another class of exceptions are the scio-
myzids, especially those relying on amphibious or terrestrial prey, that
complete their larval development within a single snail. These are more
accurately described as parasitoids, rather than predators.
Most sciomyzid larvae maintain contact with the atmosphere at all
times. They do not typically dive to attack snails in the water below their
grasp, and generally suspend their dying prey from the surface film while
they feed. Eckblad (1973b) examined the effects of predator density, prey
density, and water depth on the feeding rate of the North American scio-
myzid Sepedon fuscipennis. Normal feeding trials were run at a water depth
The malefactors · 299
of 5 mm, under which conditions the larvae consumed averages of 14 to
24 small (2–4.5 mm) Lymnaea palustris, depending on densities. But both
larval survival and kill rate were adversely affected when depth in the
experimental chambers was increased to 50 mm. In field cages, larval sur-
vival halved when water depth was increased from 5 mm to 100 mm.
Sciomyzids occasionally manifest marked prey preferences (Manguin
et al. 1988a,b). Neff (1964) exposed 13 species of medically important
snails (Biomphalaria, Bulinus, and Oncomelania) to 10 species of sciomy-
zids. All larvae seemed to be able to attack almost all the pulmonates suc-
cessfully. But the larger Bulinus seemed to repel or even suffocate smaller
sciomyzids with copious mucus in some circumstances. Neff noted that
in rapidly withdrawing from the initial bite, Oncomelania often pinched
or injured its sciomyzid attacker with its operculum.
During the late 1950s and early 1960s, sciomyzids were widely intro-
duced in efforts to control the lymnaeid intermediate hosts of the live-
stock fluke, Fasciola. The Central American Sepedon macropus did in fact
become established on four of the Hawaiian Islands, and was subse-
quently observed eating Lymnaea ollula (Chock et al. 1961, Berg 1964).
Sepedon was also introduced to Guam, although its fate is unknown.
Researchers were unsuccessful in establishing a breeding population of
the European Pherbellia (or Sciomyza) dorsata into Hawaii (Davis et al.
1961), nor (apparently) was the North American Sepedon praemiosa able
to adapt to Australia (Berg 1964). It is unfortunate that no data seem to
have been gathered regarding fly or snail population densities in any of
these cases, either before or after introduction.
In addition to his laboratory studies, Eckblad (1973b) performed an
enclosure/exclosure experiment similar to those more commonly
attempted in studies of vertebrate predation. He erected 24 quadrats (0.6
m⫻4.6 m) on the dry margins of Bool’s Backwater in Ithaca, New York,
enclosing 18 with very fine (110 m) mesh. He then closed a small dam
to flood the study area, and stocked his quadrats with natural densities of
Lymnaea palustris, Physa gyrina, and Gyraulus parvus. The tops of six quad-
rats were left open to allow natural colonization by egg laying sciomy-
zids, six were enclosed entirely to prevent any sciomyzid predation, and
six were enclosed and stocked with artificially reared Sepedon larvae above
natural densities. Eckblad used an Ekman grab to sample snail densities
in each quadrat on eight occasions from June through September.
Although very carefully designed and executed, Eckblad’s experiments
did not reveal especially convincing effects of sciomyzid predation of
freshwater snail populations. No predation effect was detected on Physa
300 · Predation
or Gyraulus densities at any sampling date, possibly because these snails
inhabited water too deep to be accessible to sciomyzid larvae. Analysis of
variance did detect a predation effect on overall density of Lymnaea at one
of the eight dates (14 July), as well as an effect on the density of small
(less than 4.5 mm) Lymnaea on 29 July. These two results might at first
appear spurious, seen together with the large number of non-significant
results Eckblad also obtained. But only early in his experiments (6/28 and
7/13) was Eckblad especially successful in maintaining increased densities
of sciomyzids in his high-predation treatment pens. There may also have
been caging effects. Data are unavailable on the six unenclosed quadrats,
but snail populations declined steeply through most of the summer in all
three enclosed treatments. Eckblad concluded, in any case, that ‘the
overall effect of the predator on the L. palustris population by the end of
the summer, either in reducing population density or changing its
size–frequency distribution, appeared to have been minimal.’
Parashar and Rao (1988) used Ochterlony immunodiffusion tests in an
effort to identify potential predators of the medically important
Indoplanorbis exustus. They induced antibody production in mice by a
series of injections with Indoplanorbis extract, then screened a variety of
predatory insects co-occurring with the snail for immunological reaction
with mouse antiserum. They obtained positive results only with the flesh
fly Sarcophaga misera, a member of a family of dipterans not known pre-
viously as molluscan predators (although other sarcophagids have long
been known as parasites or parasitoids of land snails; Bequaert 1925).
Laboratory feeding experiments verified that larval S. misera killed
Indoplanorbis, at an average rate of 6 snails/day. Parashar and Rao also
reared S. misera on an artificial diet, repeated their Ochterlony test, and
found no immunological reaction with mouse antisera, further confirm-
ing that S. misera is a natural predator of Indoplanorbis. One wonders how
many more freshwater mollusc predators remain unrecognized in the vast
diversity of the Class Insecta, if a fly from such a typically saprophagic
and terrestrial family as the Sarcophagidae is a participant.
Molluscs
The strikingly negative impact that populations of large ampullariid snails
may have upon pulmonates was reviewed in Chapter 5. They are prodig-
ious consumers of macrophytic vegetation, and as such, their relationship
with pulmonates may be described as competitive. There is evidence,
however, that their impact as predators may not be inconsequential.
On many mornings, Demian and Lufty (1965a) arrived at their lab to
The malefactors · 301
find the bottom of their Marisa tank strewn with the shells of adult
Bulinus. I quote the grisly details: ‘The remains of the soft parts left in the
shells showed no signs of decomposition, and when they were cut or
injured by the scissors during dissection red blood frequently oozed out,
indicating that death had occurred only a short time previously . . . Each
day the aquaria were cleared of dead snails and empty shells, but a new
crop of victims was found the following morning.’ Suspecting foul play,
Demian and Lufty posted a watch, and with patience it was discovered
that Marisa will in fact prey upon Bulinus of all sizes nocturnally. Marisa
do not seem to seek prey actively, but upon contacting a Bulinus, they
sieze the shell of the smaller snail, invert it, and bite out such soft parts as
they can reach. Very young snails are eaten whole. The rate of predation
was highest in the absence of other food, but some predatory activity
occurred regardless. Demian even observed his Marisa eating Bulinus eggs
from Elodea leaves, leaving the Elodea untouched.
Lymnaea may also be included in the diet of Marisa, but Biomphalaria
over about 5 mm are protected by the narrow aperture of their shells. Chi
et al. (1971) extended the known diet of Marisa to include the mud-
covered eggs of another medically important snail, Oncomelania, and
Paulinyi and Paulini (1972) demonstrated that Pomacea will also eat
Biomphalaria eggs. Indeed, the very large Pomacea canaliculata has been
observed gnawing through the shell of adult (13 mm) Biomphalaria to
consume the inhabitant (Cazzaniga 1990).
However, subsequent studies of Marisa by Demian and Lufty (1965b,
1966) showed that predatory behaviour seems to be learned; snails from
the wild seem to be much less bloodthirsty. Cedeno-Leon and Thomas
(1983) reported that such behaviour seems to be highly plastic, increas-
ing with age, maturity, and (surprisingly) the presence of plant foods.
Some individual Marisa do not seem to eat pulmonate eggs under any
circumstances, even if previously conditioned by culture with
Biomphalaria. Cedeno-Leon and Thomas concluded that Marisa might be
effective as a predator of Biomphalaria only at high densities. That ampul-
larids have reliably eliminated Biomphalaria in Puerto Rican field trials
suggests competition as the greater component of the interaction
between these two groups.
Leeches
The leeches are a group of about 400 carnivorous annelid species com-
pletely oblivious to prey size. They attack the entire range of animal life
inhabiting fresh waters. If the victim is large relative to the leech itself,
302 · Predation
observers generally describe the phenomenon as parasitism, otherwise it
is predation; the leech itself seems to be unconcerned. Among the best-
studied enemies of freshwater molluscs are ‘snail leeches’ of the family
Glossiphoniidae, small animals with a muscular proboscis they can thrust
into the tissues of their prey. Jawed leeches of the family Hirudinidae may
also attack molluscs, as may erpobdellid leeches, which are characterized
by a muscular, sucking pharynx (Klemm 1975). Even placobdellid
leeches, generally considered ectoparasites of turtles, may feed upon mol-
luscs (Moser and Willis 1994).
Among the earliest detailed descriptions of leech predation upon snails
were derived from the laboratory observations of Chernin et al. (1956b)
involving Biomphalaria and the glossiphoniid Helobdella fusca (a culture
originating from New York). Adult leeches (about 10 mm) typically
attach first to the shell of their victim with their posterior sucker, then
probe exposed parts of the snail’s body, ultimately inserting their probos-
cis. The leech then transfers its attachment from shell to substrate, and
ingests the snail’s tissues and body fluids while holding the shell suspended
in the water column. Chernin was able to demonstrate predation of adult
H. fusca on juvenile Biomphalaria up to about 5–6 mm shell diameter.
Larger snails were by no means safe from attack, however. Helobdella
broods its young (as do glossiphoniid leeches generally), and when their
mother attacks a prey item, the juveniles rush forward on their mother’s
back to join the meal. Chernin reported that juvenile leeches may under
some circumstances crawl entirely inside larger (12 mm) Biomphalaria and
take up at least temporary existence as parasites.
As leeches typically ingest only liquefied meals, researchers have found
serological techniques indispensable for the analysis of their gut content
in the wild. Wrona et al. (1979) developed antisera against four putative
prey groups for Glossiphonia complanata in Alberta, Canada: chironomids,
oligochaetes, amphipods, and gastropods. While a locally common
Lymnaea served as antigen for Wrona’s approach, the resulting serum
cross-reacted to Physa and Helisoma as well. Wrona tested for reactions in
the stomachs of approximately 150 G. complanata collected through the
course of a year at each of two sites near Calgary. Gastropods were in
both cases the most common food item, at all times of the year except
January and February. Across all months, mollusc reactions were detected
in 42 of the 95 G. complanata guts testing positive for food of any sort at
a temporary pond, and in 64 of 113 positive guts at a lake.
Glossiphonia complanata is among the more common macroinvertebrate
predators (of any sort) inhabiting temperate fresh waters. Young (1981a)
The malefactors · 303
performed a serological study of G. complanata diet in Britain similar in
many respects to that of Wrona in Canada. He tested each leech with a
larger battery of antisera (ten in all) including an ‘anti-mollusc’ developed
from a mixture of Lymnaea peregra and Potamopyrgus jenkinsi. His primary
interest was in the difference between leech diets in unproductive lakes
of northern Wales and productive lakes of Cheshire and Shropshire,
England. Molluscs were once again the prevalent dietary item identified
in the eight productive lakes, accounting for 31–58% of all positive reac-
tions. Oligochaetes and chironomids were also commonly ingested,
while the contributions of the seven other prey types were negligible.
Molluscs did not, however, seem to figure heavily in the diets of G. com-
planata inhabiting the five unproductive lakes surveyed by Young, pos-
sibly because molluscan prey were unavailable.
Certainly among the most thorough studies focusing on the parasitic
character of the snail–leech association was that of Klemm (1975). He
took monthly samples of snails and pisidiids at ten sites in southern
Michigan over the course of two years, carefully examining each indi-
vidual snail or bivalve for leeches, both external and internal. Ultimately
he discovered leech parasitism in 14 species of pulmonates, plus
Sphaerium simile. Large Helisoma seemed especially vulnerable, small pla-
norbids and amphibious lymnaeids somewhat less so, and seven proso-
branch species appeared entirely invulnerable. Five glossiphoniid species
were responsible: three Helobdella, Glossiphonia heteroclita, and
Marvinmeyeria. Glossiphonia complanata was not among the leeches Klemm
found parasitizing southern Michigan molluscs, but he verified its status
as a mollusc predator with immunological tests. Klemm could not verify
the molluscivory of Hellobdella stagnalis in southern Michigan, either as
parasite or as predator, although Young (1980) subsequently demon-
strated that H. stagnalis occasionally consumes molluscs in England.
Klemm’s data set on the Helisoma trivolvis population of Bert Pond was
especially large (Figure 7.7). Over all dates, 328 of 885 snails collected
contained one or more leeches, for a summed prevalence of 37%. Quite
a few snails were multiply parasitized, with 74 hosting two leeches, 23
hosting three, etc., up to one snail which contained 12 leeches. Klemm
identified two species of leech among the 523 leeches recovered: 379
Helobdella fusca and 144 H. lineata. He found most individuals of both
leech species between the snail’s shell and mantle, although quite a few
H. fusca occupied the kidney. Figure 7.7 indicates that the (combined)
abundance of leeches in snails generally seemed lowest in the spring
(especially May) when the new snail generation was born. Abundance
304 · Predation
Triclads
Young (1981b) extended his serological survey to include the gut con-
tents of predatory triclads, as well as those of leeches. As in his earlier
work (1980, 1981a), he sampled both productive English lakes and
unproductive Welsh ones over the course of a year, testing the four triclad
and three leech species he discovered with antisera to ten potential prey
types. Summing over all dates and lakes, the triclad Dugesia polychroa
proved to be a more specialized mollusc predator than Glossiphonia com-
planata. Young’s samples of D. polychroa ranged from N⫽49 to N⫽244
over seven productive lakes, with from 70% to 91% testing positive for at
least one food item. No less than 52% of all positive reactions were to
Young’s mollusc antiserum, as against 6–14% for his Asellus antiserum,
the runner-up. The diet of Polycoelis (two species, P. nigra and P. tenuis)
recalls that of Helobdella stagnalis, occasionally including molluscs but
Defences · 305
evincing a preference for oligochaetes. The two final species in Young’s
survey, the triclad Dendrocoelum lacteum and the macrophagous leech
Erpobdella octoculata, did not seem to consume molluscs.
The studies of triclad gut content authored by Reynoldson and Piearce
(1979a,b) are interesting from our perspective because they involved four
separate mollusc antisera. The authors’ anti-Potamopyrgus and anti-
Bithynia sera were entirely specific, their anti-Planorbis serum more
broadly reactive to four planorbids, and their anti-Lymnaea cross-reacting
only slightly with Physa. (One wonders what fraction of positives might
be missed with a general ‘anti-mollusc’ serum such as Young’s, even if of
complex origin.) Reynoldson and Piearce sampled four Shropshire meres
and one Scottish loch over 6–24 months, collecting Dugesia polychroa, the
very closely related D. lugubris, Polycoelis tenuis, and the (possibly exotic)
Planaria torva. Three Welsh ponds were added for Polycoelis nigra. The
diets of the two Dugesia were found to be indistinguishable with respect
to their gastropod prey. Both triclads commonly seemed to ingest all four
snail species in similar proportions, the overall percentages positive
ranging from 38% to 71% across five lakes. At Cole Mere, however, P.
torva appeared to prefer Potamopyrgus over the other three prey classes
available. Again the two Polycoelis appeared less dangerous to molluscs
than Dugesia, anti-snail serum of any sort accounting for only 2–14% of
all positive reactions over eight sites.
Defences
Behaviour
Threatened with attack, the behavioural options available to most animals
fall into the two categories, fight or flight. Snails differ from humans in
this regard only by matters of scale; ‘resist or retreat’ may perhaps be more
descriptive. In the ‘resistance’ category, Townsend and McCarthy (1980)
have described a striking behaviour displayed by Physa fontinalis upon
contact of its mantle with a small section of leech tissue (anterior sucker
plus four segments). Typical responses recorded include vigorous shell
shaking, followed by detachment of the foot from the substratum. Since
the mantle cavity often happens to enclose an air bubble of some size at
the time of leech contact, threatened snails commonly float to the
surface. Alternatively, shell shaking occasionally spills the air bubble held
by snails at the surface, causing them to sink. Such rising or falling
through the water column are better counted as the consequences of
306 · Predation
‘resistance’ than as the consequences of ‘retreat’, which in snails might be
reserved to describe active, foot-to-the-substrate movement.
Townsend and McCarthy tested 30 Physa for their responses to each
of 15 leech species (four families) and eight flatworm families (two fam-
ilies). Almost all 23 of these test predators, including large blood-sucking
vertebrate parasites (such as Hirudo medicinalis) not considered especially
dangerous to snails, at least occasionally elicited shell shaking from the
snail contacted. But known snail predators, such as Glossiphonia complan-
ata, elicited strong reactions from 100% of the snails tested, while the
weakest response was evoked by Erpobdella octoculata, which is generally
considered harmless to snails.
Bronmark and Malmqvist (1986) recorded similar resistance responses
to G. complanata attack in Lymnaea peregra and Planorbis planorbis, although
‘the reaction was not as vigorous as in P. fontinalis’. Two prosobranchs
tested (Bithynia tentaculata and Theodoxus fluviatilis) simply withdrew into
their shells, pulling their opercula behind them. It seemed to Bronmark
and Malmqvist that the little limpet Ancylus fluviatilis had no behavioural
responses available to it, and was consequently much more likely to be
eaten by G. complanata. Wilken and Appleton (1991) compared four pul-
monates (Physa acuta, Aplexa marmorata, Bulinus tropicus and Lymnaea nat-
alensis) in their behavioural response to direct leech contact, using
methods similar to those of Townsend and McCarthy. Neither Bulinus
nor Lymnaea displayed any responses different from that evoked by
contact with a blunt dissecting needle. But especially if the leech contact
was made about the head or mantle fringe, Wilken and Appleton con-
firmed distinctive shell shaking behaviour for their two physids. Juvenile
Physa may be less likely to fall to leech predation than juvenile Lymnaea
or Helisoma for similar reasons (Bronmark 1992).
In passing, it might be noted that the reaction evoked when a snail is
poked with an artificial instrument is also, in its own simple way, a resis-
tance behaviour. Pulmonates in particular seem likely to withdraw their
heads and pull their shells tightly to the substrate. This reaction has been
called ‘clamping’ (Brown and Strouse 1988), and we will return to it
before this chapter is concluded.
Over 24 hours, Bronmark and Malmqvist (1986) noticed an inverse
relationship between the activity of G. complanata and the activity of
Bithynia tentaculata in combined containers. They used time-lapse video
to record movement in seven vessels containing ten Bithynia and an
uncaged leech, seven containing ten snails and a caged leech, and six con-
trols with no leech. Snail activity in the caged-leech treatment did not
Defences · 307
differ from control, but was much reduced in the unrestricted-leech
treatments. Leeches seem to hunt, and snails seem to respond, via tactile
cues almost exclusively.
Our review of the behavioural response to predation has thus far
focused primarily on physical contacts between specific predators and
their molluscan prey. The responses we have catalogued as ‘resistance’
include shell shaking, enclosure, and clamping, but not as yet active
movement away from the threat. We now turn to ‘retreat’, the behavi-
oural response most commonly observed when threat is received as a
chemical cue.
Among the simpler manifestations of ‘retreat’ would be avoidance.
Haynes and Taylor (1984) included several experiments relevant to the
present discussion in their larger investigations of food finding in
Potamopyrgus. They introduced over 100 snails into the centre of a circu-
lar ‘choice chamber’, recording the number of individuals ultimately
entering five radially arranged rooms. Snails significantly avoided rooms
containing filter-paper disks soaked with water from a leech tank
(perhaps not entirely incompatible with the observations of Bronmark
and Malmqvist), or with extracts of squashed Potamopyrgus. McCollum
and colleagues (1998) reared Physa in aquaria with redear sunfish separ-
ated by permeable partitions. Although a few Physa were fed to the
sunfish daily, the fish was not allowed into the main aquarium itself.
Nevertheless, the chemical cues emitted by the sunfish seem to have
prompted the snails to spend much more time under leaves and in dark
corners, eating less and reproducing less than control snails reared
without sunfish neighbours.
Snyder (1967) constructed a flowing-water apparatus that allowed
fairly rapid and uniform introduction of varying concentrations of fil-
tered crushed-snail extract to a chamber containing ten previously equili-
brated test snails. Ultimately his experiments involved six concentrations
of extract, 17 pulmonate species (3 families), and 10 freshwater proso-
branchs (6 families), as well as a few marine species. Snyder characterized
16 snail species as ‘reactive’, 7 as ‘questionably’ or ‘inconsistently’ reac-
tive, and 4 as ‘nonreactive’. Prosobranchs typically dropped from the tank
walls (which we have classified as ‘resistance’) and buried themselves, a
retreat response that seemed to be extinguished by darkness. Pulmonates
either dropped to the bottom (resistance) or crawled out of the water
entirely (retreat).
Snyder also performed two sets of interspecific tests, one comparing
representatives of six different families, the other four different species of
308 · Predation
Helisoma. Although no reaction was seen in most of the former experi-
ments, Pseudosuccinea did react to extracts from both Physa and Viviparus,
and Pomacea reacted to Pseudosuccinea extract. At least occasional alarm
reaction was elicited from any Helisoma by extract from any congener.
One reads something of a sense of frustration in Snyder’s account of
his experiments. Even among his ‘reactive’ species at high concentrations
of extract, responses were often not displayed by substantial fractions of
the animals tested. Snyder and Snyder (1971) showed that snail age should
be figured among the many variables influencing the various behaviour-
al responses to predation threat. Newborn Pomacea paludosa do not react
to crushed intraspecific extracts; some sensitivity seems to develop after
a week or two. Interestingly, reactions to turtle odour are present in
newborn Pomacea, becoming extinguished at different rates for two
different turtle species. Snyder and Snyder reported that Pomacea lose
their sensitivity to the odour of the musk turtle (Sternotherus minor) when
snails reach 3 g wet weight, while retaining sensitivity to the odour of
snapping turtle (Chelydra serpentina) up to 20 g. Snapping turtles are
indeed able to crush much larger snails than are musk turtles.
Triggers for the behavioural reactions displayed by freshwater gastro-
pods in response to threat are probably much more complex than typical
human observers recognize. Alexander and Covich (1991a) performed a
series of experiments on the ‘crawl out’ retreat response manifested by
Physa virgata when threatened by the crayfish Procambarus simulans. Physa
did not seem to leave the water (at a frequency greater than control
expectation) when crushed conspecifics were added to their vessel, nor
when a caged crayfish was added, nor even when a caged crayfish was
allowed to feed on crushed Biomphalaria. Only in response to the com-
bination of a crayfish and crushed Physa did a significant number of snails
(a mean of perhaps 25%) crawl out. Crawl-out duration seemed to
extend for about two hours.
Helisoma trivolvis larger than about 8–9 mm in shell diameter are effec-
tively protected from attack by Procambarus, while smaller Helisoma are
vulnerable. Alexander and Covich (1991b) demonstrated a crawl-out
response in 4–6 mm Helisoma quite similar to that displayed by Physa,
lesser responses in 6–12 mm Helisoma, and no crawl out in 12–16 mm
animals. The survey experiments of Covich et al. (1994) demonstrated
significant crawl-out responses in three Physa species and Lymnaea emar-
ginata, but not in the planorbids Helisoma anceps or Gyraulus parvus or the
prosobranchs Amnicola limosa or Campeloma decisa (which burrowed).
In sum, resistance and retreat in their various forms may both be
Defences · 309
among the responses manifested by freshwater snails to threat of attack.
Resistance seems to occur in response to physical contact, and retreat in
response to chemical cue, although both may occur in tandem under
some situations (e.g. ‘drop and burrow’). The triggers for these behavi-
ours are complex, very much dependent upon the nature of the cue and
the snail receiving it.
Shell
The protection afforded by the molluscan shell is of two effects. Most
obviously, especially large and strong shells may protect their wearers
from crushing and/or engulfment by powerful, hard-bodied predators.
Second, even the thinnest and lightest layer of calcium carbonate will be
sufficient to insulate its wearer from at least some environmental fluctu-
ations, pathogenic micro-organisms, trematode cercaria, calanoid copep-
ods (Liebig and van der Ploeg 1995), and such soft-bodied predators as
leeches, triclads, and other snails. These two effects will be examined in
turn.
Regarding the likelihood of crushing and/or engulfment, some selec-
tion by predators on the size of their prey seems to be universal. We have
already touched upon the size preferences displayed by muskrats foraging
on mussels (Hanson et al. 1989, Figure 7.1), cichlids offered Biomphalaria
(Slootweg 1987, Figure 7.4) and crayfish offered Dreissena (Piesik 1974).
Other clear demonstrations of size-selective predation include Keller and
Ribi’s (1993) work on fish predation of Viviparus, Dudgeon and Cheung
Pui Shan’s (1990) laboratory experiments with a freshwater crab and five
Hong Kong gastropods, and de Leeuw and van Eerden’s (1992) data on
diving duck predation of Dreissena.
It may be recalled that Covich et al. (1981) found crayfish able to eat
only the smallest Corbicula offered them, typically ranging to no more
than 6 mm standard length. Kennedy and Blundon (1983) crushed N⫽
70 Corbicula of varying size in an industrial instrument designed to
measure the pressure applied to a surface. The regression of (log) shell
length against (log) crushing force was quite significant (r 2 ⫽0.73) and is
converted to a semi-log scale in Figure 7.8 (line ‘C’). Kennedy and
Blundon were quite impressed with the strength of Corbicula shells; their
resistance to crushing seems to be significantly better than any of eight
estuarine bivalves previously examined with similar methods.
The dashed line running obliquely across the lower corner of Figure
7.8 roughly traces the crushing strength of an average-sized adult cray-
fish (Procambarus clarkii). Force data, from Brown et al. (1979), reflect
310 · Predation
decreasing strength but increasing gape from the base to the tip of the
chiliped. The maximum gape shown, about 12 mm, is my own rough
estimate. Figure 7.8 seems to indicate that even the smallest (5 mm)
Corbicula should be invulnerable to crayfish attack. But in practice,
Covich and colleagues report that crayfish typically chip at the edges of
the clams with their mandibles, rather than their chilipeds, ultimately
opening a hole large enough to consume the contents. The dashed lines
shown in Figure 7.8 should be viewed as minima for refuge, not guaran-
tees.
When they are offered snails with greatly different resistance to crush-
ing, it is perhaps not surprising to observe that predatory sunfish consume
the weakest. In laboratory choice tests, Stein et al. (1984) observed that
redear sunfish slightly preferred Physa integra over Helisoma (two species)
and that Goniobasis (⫽‘Oxytrema’) semicarinata were either entirely
Defences · 311
ignored or rejected. Although no regression statistics were offered,
approximations of the author’s figures on crushing resistance as a func-
tion of shell size for these three snail genera are shown in Figure 7.8.
Goniobasis (Go) seems to be indistinguishable from Corbicula in shell
strength. Helisoma (H) is much weaker, and Physa (P) weaker still, reca-
pitulating the preferences manifest by their fish predator.
Osenberg and Mittelbach (1989) performed tests similar to those of
Stein et al., involving 11 species of freshwater snails co-existing with
pumpkinseed sunfish in three small Michigan lakes. The relationship
between crushing resistance and shell length closely fit a semi-log linear
model in all cases. I have selected data from Gyraulus parvus (N⫽20,
r 2 ⫽0.93), Amnicola limosa (N⫽43, r 2 ⫽0.92) and Viviparus georgianus
(N⫽20, r 2 ⫽0.82) for inclusion on Figure 7.8 to reflect a range of shell
morphology. (I have taken the liberty of extrapolating the Viviparus
regression in Figure 7.8, as the authors tested only juveniles to 13 mm.)
Although Viviparus seems comparable to most other species we have
examined thus far, Osenberg and Mittelbach’s data suggest that the
smaller species demonstrate a much steeper increase in shell strength with
increase in size than we have previously noted.
The figure also features approximate minimum boundaries for shell
size and strength required for snails to obtain refuge from pumpkinseed
sunfish. The 19 mm minimum size was calculated by substituting the
length of the largest pumpkinseed examined into Osenberg and
Mittelbach’s Figure 7, solving for a 50% probability of escape, and con-
verting snail mass to shell length using the regressions supplied.
Minimum refuge in crush resistance (10 N) was estimated from the 50%
survival rate in the author’s Figure 8.
Clearly the prey populations will vary in their availability to the pump-
kinseeds – larger snails escaping by virtue of the size and strength of their
shells, smaller snails escaping notice. The authors predicted that pump-
kinseeds ought to impose a ‘hump-shaped’ selection curve upon their
prey population, with differing modes determined by the size of individ-
ual fish. But because of the rarity of large prey in the study lakes, the
author’s observations generally suggested that the likelihood that a snail
was selected increased monotonically with size. The most important
determinant of pumpkinseed diet in small Michigan lakes seems to have
been the population dynamics of the snails upon which they feed.
Considerable attention has been directed toward the question of the
vulnerability of Dreissena to predation by a variety of fishes. Nagelkerke
312 · Predation
and Sibbing (1996) reported that the maximum mussel size taken by
common bream and white bream seems to be determined by their pha-
ryngeal crushing power, but that the (somewhat greater) maximum size
taken by roach is constrained by its oral gape. The common bream dis-
played a much lesser feeding efficiency on zebra mussels than either the
white bream or roach. Redear sunfish can also feed on smaller Dreissena
(French and Morgan 1995). When offered a choice among large Helisoma
(12 mm), small Helisoma (8 mm) or zebra mussels (same sizes), adult
sunfish always preferred the more energetically profitable Helisoma.
French and Morgan’s data on crushing resistance of Helisoma shells closely
matched that previously published by Stein and colleagues (1984). Their
data on the resistance of 5–15 mm Dreissena shells (N⫽50, r 2 ⫽0.59) are
added to Figure 7.8.
The thiarids of Lake Tanganyika are remarkable for more than their
diversity of species and genera. The heavy calcification and armour of
the fortresses carried by many of the denizens of that ancient lake remind
all beholders of the seashells of the tropics (West and Cohen 1996). West
et al. (1991) compared the crushing strength of six Tanganyika thiarids
with that of five thiarids from elsewhere in Africa. All measurements
were taken on about 25 adults fairly uniform in size, and tabulated as aver-
ages. The data from four Tanganyika species (T1 – Spekia zonata, T2 –
Lavigera sp. A, T3 – Paramelania imperialis, T4 – Paramelania crassigranulata)
and two more cosmopolitan species (T4 – Potadoma ignobilis, T6 –
Melanoides tuberculata) are plotted in Figure 7.8. Although the latter two
are unremarkable in shell strength, the four in the former category are
indeed exceptional.
West and colleagues noted that predators, especially freshwater crabs
of great size and strength, seem to have coevolved with the thiarids of
Lake Tanganyika. In particular, the endemic crab Platytelphusa armata
carries a crushing chiliped with broad, molariform dentition averaging
about 6 cm long and opening perhaps 3 cm wide. No direct measures on
the strength of this formidable armament are available, but the crabs do
successfully crush Lavigera shells of 480 N average load strength, consum-
ing the contents.
Entirely independent of their strength, shells provide protection from
the generally softer and smaller predators that may invade. We have earlier
mentioned the value of an operculum for the repulse of an invader, as
witness Klemm’s (1975) data on leech parasitism in Michigan proso-
branchs and Neff’s (1964) observations on Oncomelania under sciomyzid
Defences · 313
attack. But certain aspects of shell shape, particularly relative reduction of
the shell aperture, seem to provide some protection from invasion even
when an operculum is absent or ineffective.
Demian and Lufty (1966) performed a series of laboratory feeding
trials involving four Marisa and 20 prey snails of varying size, either
Bulinus or Biomphalaria, monitored over five days. Each day the authors
inspected the tanks, counted surviving prey, and replaced any missing
with snails of similar size. In a 24-hour period, four Marisa were gener-
ally able to eat most of the juvenile Bulinus offered, up to the largest size
tested by Demian and Lufty (3–4 mm). Smaller prey were eaten whole,
while larger snails were eaten from within their shells. But Marisa only
seemed to prey effectively on Biomphalaria up to to 2–3 mm, the
maximum size they could break or engulf whole. Biomphalaria as large as
5 mm in diameter seemed invulnerable to Marisa attack by virtue of both
the strength of their shells and the narrowness of their apertures.
Snyder and Kale (1983) reported a similar phenomenon with an
amusing turn-around. Here the predator is the snail kite and the prey,
none other than Marisa itself. It will be recalled that the ordinary prey
of the kite is Pomacea, possessing a globose shell with a wide aperture.
But in Colombia, Snyder and Kale noted common attacks on Marisa,
whose ability to pull its operculum deeply within a narrow aperture
seems to afford considerable protection. Kites succeeded in about 92%
of their extractions of Pomacea, but only 62% of Marisa extractions. As
the birds often perch over water, dropped snails may be awarded a
‘second chance’.
Freshwater molluscs are generally drab. Through most of its range,
Israel’s populations of Theodoxus jordani bear shells that are plain black.
Yet on the shores of Lake Kenneret one will encounter populations of
Theodoxus whose shells are decorated with genuinely festive stripes,
checks, and zebra patterns. Heller (1979) reported a significant correla-
tion between the relative frequency of dark-patterned shells in the lake
and the distribution of black (basalt) background. His survey of fish gut
contents identified two predators of Theodoxus: Barbus longiceps and (espe-
cially) the blenny Blennius fluvialis, neither of which occurs outside Lake
Kenneret. Heller suggested that the polymorphic colour patterns may
result from visually hunting predators within the lake, while outside the
lake solid black is favoured by virtue of the superior insulation it affords
against solar radiation.
Vermeij and Covich (1978) have pointed out that, setting aside the
314 · Predation
spectacular faunas endemic to ancient rivers and lakes, the shells of fresh-
water molluscs seem to show substantially fewer adaptations to thwart
predators than do marine forms. Shells are generally thinner in fresh
water, and more rarely show spines, knobs and other shell sculpturing.
The weak, planispiral gastropod form is more frequently observed in fresh
waters than in marine. Adaptations against invasive predators, such as the
narrow, elongated apertures or aperture teeth shown by marine gastro-
pods or the interlocking shell margins common in marine bivalves, also
seem to be quite rare in freshwater molluscs. Vermeij and Covich felt that
physiological limitations on shell deposition, such as lower temperatures
and calcium availability, ‘may be in part responsible’. They also noted that
the predators of freshwater molluscs may be weaker and less diverse than
those encountered by marine molluscs.
But I would add, by way of clarification, that predation pressure (the
likelihood of being eaten) does not seem any less in freshwater than in
the marine environment. If predation has had more impact on the evo-
lution of shell form in marine environments, its impact on life history
evolution may have been greater in freshwater. Rapid growth, large
reproductive output and efficient dispersal may be seen as adaptations
both to ephemeral environment and to predator escape. To this subject
we now turn.
Life history
It may be recalled from previous discussions (Chapter 4) that selection
seems to favour early maturation time if mortality in a critical window
(early, but not newborn) is low, but a later maturation if such mortality
is high (Figures 4.7 to 4.9). Crowl (1990, Crowl and Covich 1990)
studied populations of Physa (Physella) virgata inhabiting four Oklahoma
streams, designated PCF (permanent, with crayfish), PNCF (permanent,
no crayfish), TCF (temporary, with crayfish) and TNCF (temporary, no
crayfish). His temporary streams dried for three months in the late
summer, inflicting heaviest mortality on the largest size classes of Physa.
The crayfish common in his study streams was a known snail predator,
Orconectes, preferring smaller (but not newborn) Physa and unable to
handle the largest animals.
In the two streams from which crayfish were absent (PNCF, TNCF),
Physa reached maturity at 5.4 mm (47.5 days) and 5.3 (42.5 days), respec-
tively, and generations overlapped to the extent that Crowl could not dis-
tinguish cohorts. Time to maturity was similarly brief at the TCF site
Defences · 315
Figure 7.9. The effect of water from four different treatments on the mean size at
first reproduction and death in Physa virgata (Crowl and Covich 1990). Each of the
eight measures is a grand mean of 72 data, four populations⫻18 snails per popula-
tion.
(4.6 mm, 50.0 days), with the periodic nature of the environment impos-
ing two distinct generations per year. But maturity was much delayed at
the PCF site (9.3 mm, 117.9 days), so that a single generation passed
annually. Crowl hypothesized that the larger size at reproduction
observed in the PCF population might be a response to crayfish preda-
tion, a response precluded from the TCF site by abbreviated growing
season.
Differences in age and size at maturity disappeared when snails from
the four populations were reared in a common laboratory environment.
Six groups of three snails were reared over their lifetimes for each of four
populations⫻four treatments. All four populations matured at about 5
mm (45 days) when reared in water circulating from a reservoir that con-
tained other snails, or that contained crayfish, or that contained snails and
crayfish separated by meshwork. But most remarkably, all four popula-
tions delayed maturity to an average of about 8 mm (70 days) when
reared in water circulating from a reservoir where crayfish were allowed
to eat snails (Figure 7.9). Some chemical cue, apparently released by a
316 · Predation
crayfish attack on a conspecific, seems to delay maturity in Oklahoma
Physa virgata. Clearly selection has not acted directly to fix adult size
optima in these populations. But just as clearly, a trait that can delay
maturity in response to an environmental cue, effectively promoting
rapid growth through especially vulnerable size classes, has obvious adap-
tive value.
Consequences
Community composition
By this juncture the reader ought to be fairly well convinced that popu-
lations of molluscs commonly differ in their vulnerability to predation.
The freshwater molluscan fauna is characterized by great diversity of size,
shell strength, habitat and behaviour, and all of these variables may have
demonstrable impacts on the likelihood of falling victim to a mob of
predators that ranges from mink to flatworm. For example, it may be
recalled that the carp inhabiting Skadar Lake, former Yugoslavia, seemed
to select Valvata piscinalis over seven other molluscs available, including
more common species of comparable size (Stein et al. 1975). Carp appar-
ently selected Valvata for its shell, which was rather more fragile than
those of other molluscs available.
The North Fork of the Holston River in southwestern Virginia is
home to about 16 species of unionid mussels and a large population of
muskrats. Neves and Odom (1989) surveyed the shell middens left behind
by these rapacious rodents from 1979 to 1986, comparing their contents
with quantitative samples taken from the stream bed in 1981. Although
most mussel species seem to be taken according to their relative abun-
dances, Neves and Odom recognized some preference for Pleurobema ovi-
forme and the endangered Fusconaia cor, two species of ‘medium’ size
(50–70 mm). Figure 7.10 compares the estimated abundance of the ten
most common unionid populations in the North Holston Ford study area
with their occurrence in muskrat middens 1981–86. (I have excluded
midden data taken prior to the 1981 quantitative samples.) Overall, the
six-year predation rate appears to have been about 3.8%.
My value of chi-square from the 2⫻10 contingency table correspond-
ing to Figure 7.10 was 173.8, which with 9 degrees of freedom is signif-
icant at the 0.001 level. Combined by size class, Neves and Odom
collected the shells of 408 individuals belonging to their two ‘small’
species, 573 individuals belonging to their five ‘medium’ species, and 135
Consequences · 317
Figure 7.10. The ten most common unionid mussel populations inhabiting the
upper North Fork Holston River, together with an estimate of their 1981 popula-
tion sizes in the Neves and Odom (1989) study area. The top two species were cat-
egorized as ‘small’ by Neves and Odom, the middle five were ‘medium sized’, and
the bottom three were ‘large’. The fractions known to have fallen to muskrat preda-
tion 1981–86 are darkened.
Distribution
Although it certainly seems reasonable to expect some sort of relation-
ship between the distribution of predators and their prey, the form of
relationship is far from clear. Sih (1984) has outlined circumstances under
which negative correlations, positive correlations, and no relationship at
Consequences · 319
all may be expected between predator and prey densities. Negative cor-
relations, to be expected either when prey outmanoeuvre their predators
or when prey are effectively sessile, seem to me most likely to apply to
freshwater molluscs. It is simply my impression that snails may indeed
outmanoeuvre at least some of their invertebrate predators (ambushing
leeches and flatworms, perhaps crayfish hunting from home burrows) but
simultaneously are so slow relative to their vertebrate predators as to be
effectively sessile. I should imagine that given the many other factors
influencing distributions of both levels of the food chain, however, any
sort of ‘signal’ would be extremely difficult to detect.
Most reports do, in fact, seem to suggest negative correlations. Such a
relationship has been reported between the density of Anodonta and
muskrat foraging range in a Finnish lake outlet (Jokela and Mutikainen
1995), and between the density of Neritina latissima and the molluscivor-
ous fish Sphoeroides annulatus in a Costa Rican stream (Schneider and
Lyons 1993). McKaye and his colleagues (1986) hypothesized that the
absence of Bulinus from open, sandy areas of Lake Malawi might be
attributable to predation by cichlids. Erection of fences to exclude the
fish resulted in 40–60% increase in Bulinus density within the week.
Even though most species of snails inhabiting Trout Lake, Wisconsin,
demonstrably preferred rocks with thick covers of periphyton, the
quadrat samples of Weber and Lodge (1990) contained higher snail den-
sities on more barren rocks. The authors suggested that this incongruous
situation might arise from crayfish predation, since crayfish were more
common among the densely packed, periphyton-covered rocks of the
exposed shores than in the scattered, barren rocks found on protected
shores. It is possible that crayfish may have been excluded from the richer
snail-hunting grounds of Trout Lake by locally low dissolved oxygen.
Covich (1981) reported increased density of Physa gyrina at the head-
waters of an Oklahoma mineral spring, from which crayfish seemed
excluded by water quality considerations.
At a larger scale, Crowl and Schnell (1990) reported a significant neg-
ative correlation betwen crayfish abundance and the density of Physa
virgata over 11 streams in Oklahoma. Hofkin and colleagues (1991) sur-
veyed 53 freshwater habitats in Kenya selected because they seemed suit-
able for Biomphalaria. They reported that 21 of these sites were inhabited
by the large North American crayfish Procambarus clarkii but no pulmo-
nate snails, 19 were inhabited by pulmonate snails but no crayfish, and
only 4 sites contained both crayfish and snails, a highly significant nega-
tive correlation.
320 · Predation
Moving to positive correlations, it may be recalled that Bronmark
(1988) collected exhaustive samples from nine 1 m2 predator exclosures
and nine uncaged control plots of the same size at his small study pond
in southern Sweden. Although his experiments were designed to esti-
mate the importance of vertebrate predation on a diverse gastropod
fauna, he also recorded data on the distribution of leeches (Glossiphonia
complanata) in his benthic samples. He found leeches in ten of his 18
samples (5 exclosures and 5 controls). Over all sites, the median gastro-
pod density (combining all 12 snail species) was between 445 and 402
individuals/m2. Then Bronmark’s data seem to reflect a significant posi-
tive association between snails and leeches; eight of the samples with high
snail densities harboured leeches, while seven of the samples with low
snail densities did not (Fisher’s exact probability was 0.015). Both leeches
and snails may here have been responding more directly to macrophyte
densities, which were high in the snail-rich sites.
In addition to the enclosure/exclosure experiments previously dis-
cussed, Kesler and Munns (1989) took dip net samples from four habitat
types in their small Rhode Island pond. Even given the apparently large
impact that predatory bugs (Belosoma) seem to have on pulmonates at this
locality, their single-date survey showed no correlation between bug
abundance and the densities of either Physa or Pseudosuccinea. Their
survey did seem to indicate an inverse correlation between bug abun-
dance and the size of the potential gastropod prey, however.
At this stage it would be fun to introduce some data from Esrom Lake,
Denmark, collected by Berg (1938). One of Berg’s primary goals was to
examine the relationship between depth and the bottom fauna of all sorts.
He established a transect perpendicular to the shore and took seven
samples of two Ekman grabs each on 12 dates from November 1932 to
January 1934. He collected and identified a great variety of invertebrates,
including Dreissena, one species of Sphaerium, several species of Pisidium
(tallied together), and 12 species of snails (Table 7.3). I have graphed some
of Berg’s snail data in the upper half of Figure 7.11. The ‘summer’ plot
is based on the samples of May 30, July 7, and August 2 combined, and
shows the abundance of the four most common snail species as a func-
tion of depth. The ‘winter’ plot is identical except that the sample dates
were February 15, December 5, and January 23. Although the snail
species vary in the depths to which each extends, all reached maximum
abundance in the shallowest (2 m) sample, regardless of season (as will be
developed in Chapter 9).
In addition to molluscs, Berg tabulated the abundances of a great
Consequences · 321
Species Total
Lymnaea ovata 85
Physa fontinalis 7
Planorbis planorbis 23
P. carinatus 4
P. albus 74
P. contortus 14
P. crista 47
Theodoxus fluviatilis 17
Bithynia tentaculata 641
B. leachii 66
Valvata piscinalis 740
V. cristata 51
Figure 7.11. Abundance of the four most common snail species in Esrom Lake,
Denmark, and two classes of their predators, as a function of depth (Berg 1938).
Abundance is given as the total counts of individuals summed from two 225 cm2
grab samples taken on each of three dates in the summer, or three dates in the winter.
Summary
The majority of the predators whose feeding habits have been touched
upon here could safely be categorized as ‘generalists’. And the feeding
habits of the majority of the predators whose diets are touched upon here
are difficult to observe directly. Thus most of the studies reviewed in the
first section of this chapter were directed toward establishing the mollus-
Summary · 323
can component of a wide sample of broad and changing diets. Indirect
methods, especially gut content analysis and laboratory feeding trials
involving suspected predators, have been favoured.
The molluscan fraction of the gut contents of such omnivores as turtles,
amphibians, and many fishes seems quite sensitive to prey availability.
Groups as diverse as ducks, sunfish, and crayfish seem to vary their mollusc
consumption with age, gender, and reproductive condition. But by com-
parison with the gut contents of similar predators in similar habitats, it is
possible to establish that some fish (e.g. large bottom-dwellers, some cich-
lids, pumpkinseed sunfish) as well as some invertebrates (particular species
of leeches and triclads) specialize on freshwater mollusc prey.
While gut contents strongly reflect prey availability and weakly reflect
preference, the data gathered from laboratory feeding trials leave just the
opposite impression. Where both sorts of data may be compared, they
may not agree. For example, under laboratory conditions crayfish may
evince a preference for molluscs, as in fact may large ampullariid snails,
even though both are typically considered herbivores. Predation rates are
much more easily obtained in the laboratory, and are available for cray-
fish, sciomyzid and lampyrid larvae, and belostomatid bugs. Such figures
are generally viewed as maxima, only weakly indicative of reality in the
wild. Natural densities of Belostoma may peak at such levels as to consume
all snails available, if one accepts experimentally estimated predation rates.
The importance of predation is much more difficult to establish than
its widespread occurrence. Freshwater mollusc population densities seem
positively affected by cages designed to exclude predators in about half
the studies thus far published. Workers enclosing fish and crayfish pred-
ators with mollusc populations at natural densities have more reliably
detected predation effects, although an enclosure of sciomyzid flies pro-
duced no result. Given their dependence on the coincidence of popula-
tions of two living creatures, predation effects would seem likely to be
both temporally and spatially patchy. Artificial enclosure of a predator
with its potential prey decreases such patchiness, perhaps increasing the
apparent impact. For this reason, Robinson and Wellborn’s (1988) dem-
onstration of a (very patchy) predation effect in a Texas Corbicula popu-
lation, using exclosures (but not enclosures) seems especially significant,
as does Keller and Ribi’s (1993) work with Viviparus.
Direct, especially long-term studies of freshwater mollusc densities,
and those of their potential predators, are desperately needed. We know,
for example, that as many as 52% of the Dugesia polychroa in productive
English lakes may have eaten a freshwater mollusc recently, but we have
324 · Predation
no idea how many molluscs or Dugesia there may be in an English lake.
I am aware of only one study, that of Hanson et al. (1989) on the muskrat
predation of Anodonta, that begins to assess the impact of predation on
freshwater mollusc populations directly. Programmes of this nature are
not cheap, nor lightly entered into. Thus the artificial introductions of
sciomyzids and lampyrids in biological control efforts, with no data taken
on target snail population densities before or after, were opportunities
lost.
The first line of defence for the typical freshwater mollusc is, and ever
will be, its shell. Researchers report good correlations between shell size
and (log) crushing resistance. Compared with the strength and gape of
their typical crushing predators (fish, crayfish), many freshwater molluscs
carry shells that afford excellent protection indeed. The shell strength of
Lake Tanganyika thiarids seems to be greater than that displayed by other
freshwater molluscs, and in fact, they are hunted by larger and more pow-
erful crabs. It should be remembered, however, that crushing predators
may be able to overcome shell defences by manipulation of the shell to
attack its weak points, or by simple patience. There is some evidence that
the shell markings carried by a few species (Theodoxus) may provide cam-
ouflage. There are also many freshwater molluscs, especially pulmonates
and juveniles of all sorts, for which the shell constitutes no defence at all
against crushing predators. But even the smallest and lightest shells may
provide afford some protection against invading predators, such as insects,
leeches, and other snails. The planispiral shell form so common in fresh
water (planorbids, Marisa) presents an invading predator with a narrow
opening, behind which the shell’s inhabitant can withdraw deeply.
Withdrawing into the shell, or clamping tightly to the substrate, are
two of the most obvious defensive behaviours displayed by freshwater
molluscs. These we have categorized as ‘resistance’ responses, along with
more spectacular shell shaking, dropping and rising reactions often dis-
played by freshwater snails. They seem to be elicited by direct contact
with slow-moving predators, such as leeches, or sometimes by chemical
cues, such as filtered extracts from crushed conspecifics. Snails may also
display ‘retreat’ responses to chemical cues: avoidance, burrowing, or (for
some pulmonates) crawling out of the water entirely. The triggers for all
these defensive behaviours may be complex, depending certainly upon
the species of snail and its age. The perception of chemical cues may be
especially fine; responses in some circumstances seem to require active
feeding on conspecifics by specific predators.
In at least one well-documented case, pulmonate snails seem to
Summary · 325
respond to chemical cues of predation by life history shifts. Growth is
extended and reproduction delayed under long-term threat by crayfish to
Physa at a young (but not newborn) age.
It is difficult to imagine the circumstances under which a predator
population might take a random sample of a population of prey. For our
purposes, it is immaterial whether predators actually choose among
potential mollusc prey, or whether they attack everything but only ulti-
mately prevail over some subset. Size-selective predation is a rule amply
demonstrated by muskrats, cichlid fish, and leeches, among studies
reviewed here. Other variables affecting ‘handling time’ besides overall
size (including behaviour, shell shape, opercula, and defences such as
mucus) combine to yield varying vulnerability to predation at the spe-
cific level. It is thus quite surprising to see how infrequently researchers
have detected predation effects on freshwater mollusc communities.
There is a little experimental evidence of a community effect due to scio-
myzid flies and crayfish, and some analytical data on the varying impacts
of muskrats to mussels. But review volumes on predation and aquatic
communities (e.g. Kerfoot and Sih 1987) do not typically feature refer-
ences to freshwater molluscs in this regard.
Turning to the potential for predators to affect freshwater mollusc dis-
tribution, the problem may be too much data, rather than too little. Most
studies seem to suggest negative correlations between mollusc predators
and their prey, although positive correlations, or no correlations at all, are
often reported. Quadrat data from Esrom Lake, Denmark, can be viewed
as supporting either positive or negative correlations, depending on one’s
selection of specific predator type (leech or flatworm) and prey type
(Bithynia or Valvata). There seems to be no temporal correlation between
predator and prey density in either direction. Distributional data are
probably better interpreted as arising more directly from the environ-
ment. So although significant impacts of predation have been demon-
strated upon specific molluscan populations, both directly and indirectly,
by many methods on many occasions, the community effects of preda-
tion remain surprisingly elusive.
Although but briefly touched upon in the preceding chapter, the
ability to disperse to new habitats may constitute some defence against
predation, in an evolutionary sense. The many variables that impinge
upon successful colonization of a new habitat, as well as the benefits that
may accrue to successful colonists, will be discussed in Chapter 8.
8 · Biogeography
We begin this chapter with a review of the nominally abiotic factors that
may contribute to the distribution of freshwater bivalves and gastropods
on a regional scale. At least a dozen studies over the last 30 years have
returned strong positive correlations between the abundance and/or
diversity of freshwater molluscs and calcium concentration or related
water quality variables. Laboratory experiments show both that calcium
concentration in normal ranges can affect growth, survivorship, and
fecundity, and that the various species offen differ strikingly in their
calcium optima. Some workers have suggested that calcium concentra-
tion (and correlated variables) may not act on abundance in the field,
however, but rather serve as a ‘filter for colonists’ only. Others have sug-
gested that the effects of calcium on natural abundances of molluscs are
real but indirect, calcium concentration influencing the abundance of
food.
We next review a relationship with analogues throughout biology, the
number of mollusc species as a power function of habitat area. About a
dozen studies worldwide have returned species/area regression coeffi-
cients somewhat less than or equal to the ‘canonical’ value of 0.25.
Species vary in the mean areas of the habitats they occupy, just as they
vary in calcium range. And an interaction is sometimes noticeable
between area and calcium in their effects on the freshwater mollusc com-
munity, such that sites with larger areas tend to have more species than
might be predicted from their calcium concentrations. This is again con-
sistent with the hypothesis that the effects of calcium are indirect,
through food abundance.
We review several other factors (isolation, current, substrate, etc.) for
their regional effects on freshwater mollusc communities, generally
finding them of lesser consequence than calcium and area. In the final
section of this chapter, we examine evidence that variance in these two
Environmental calcium · 327
respects has its origin in life history adaptation. It will be recalled from
Chapter 4 that, although most populations of freshwater molluscs seem to
be undifferentiated with regard to reproductive effort, occasionally pop-
ulations we have characterized as ‘R-adapted’ seem to allocate an order of
magnitude greater energetic resource to reproduction and ‘S-adapted’
populations an order of magnitude less. It will also be recalled that, as a
consequence of their resource allocation, both R-adapted and S-adapted
populations were hypothesized to inhabit special environments. The
present chapter concludes with a demonstration that such a model may in
fact have some explanatory power in mollusc biogeography.
Environmental calcium
The influence of calcium availability on the distribution of freshwater
molluscs has been recognized since well back into the nineteenth century.
One should immediately note that the intercorrelations among total
hardness, calcium concentration, pH, alkalinity, and conductivity are typ-
ically very high in natural fresh waters, and that workers have differed in
their choice of variable(s) to emphasize. But regardless of the units in
which this particular aspect of water chemistry is measured, whether it
be acidity, buffering capacity, or the concentration of divalent cations, it
has repeatedly been reported that freshwater mollusc species richness is a
function of it.
The key seems to be substantial interspecific variance in tolerance for
low calcium levels. For example, Boycott (1936) divided the British fauna
into 30 ‘calciphile’ species ‘that ordinarily . . . need water with at least
20 mg calcium per litre and some of them more’, and 26 species ‘which
we can find in soft water without surprise’. Boycott could not identify
any ‘certainly calcifuge’ freshwater mollusc species among the British
fauna; all 26 softwater species are also found in hard waters. (Extremely
hard water seems uniformly detrimental.) Thus species richness would be
expected to increase with calcium concentration, at least to a point.
Scattered through Boycott (1936) one can find 12 lists of species
(bivalves and gastropods combined) from individual British lakes, accom-
panied by estimates of calcium concentration. In Figure 8.1 their number
of species is graphed by calcium concentration on a double-log scale. If
multiple estimates of calcium were provided by Boycott, I selected the
lowest. Lakes Windermere and Leane are large and have exceptional
mollusc faunas; Boycott chose these to make a point to which we shall
return. Excluding them for the moment, over the other 10 lakes the
328 · Biogeography
Figure 8.1. Number of mollusc species in British lakes as a function of calcium con-
centration (data of Boycott 1936). The larger lakes Leane (L) and Windermere (W)
were excluded from the regression analysis.
Laboratory studies
It appears that calcium concentrations within the normal ranges observed
in fresh waters may have direct metabolic affects on molluscs. Greenaway
(1971) held Lymnaea stagnalis in a medium loaded with 45Ca tracer, coated
their shells with wax, and measured the minimum equilibrium concen-
tration of calcium – the concentration at which the net movement of
tracer into and out of the animal balanced. Several snails failed to achieve
calcium balance, continuing to show a net loss of calcium to the medium
at low concentrations. Eight successful measurements of minimum equi-
librium concentration ranged from 1.2 mg/l to 3.8 mg/l calcium, with
a mean about 2.5 mg/l calcium.
Interestingly, L. stagnalis was among Boycott’s ‘calciphile’ species, not
normally found reproducing in waters less than 20 mg/l calcium.
Greenaway notes that L. stagnalis are ordinarily not in calcium balance
with their medium, but rather show net uptakes. Thus in the environ-
mental concentrations of calcium where the snails are typically found,
concentration gradients are absent or even slightly favourable for uptake.
Environmental calcium · 329
This contrasts strikingly with the situation for sodium, for example,
which L. stagnalis maintains in balance at considerable energetic cost
(Greenaway 1970). It is also interesting to note that L. stagnalis seems to
be able to use calcium as a cue for orientation, an ability more weakly
expressed in the softwater L. peregra (Piggott and Dussart 1995).
Thomas and Lough (1974) placed individual Biomphalaria glabrata in
variable volumes of water (down to 6.25 ml/snail) at calcium concentra-
tions initially ranging from 0 mg/l to 80 mg/l. No attempt was made to
control calcium input from shell dissolution or food, which may have
been substantial (Young 1975). After 15 days, they found that final
calcium concentrations in some of the smaller-volume experiments had
decreased from 2.5 mg/l to about 0.5 mg/l calcium. They suggested that
0.5 mg/l might represent a minimum equilibrium calcium concentration
for Biomphalaria, a figure substantially lower than Greenaway reported for
Lymnaea. As the natural habitats of B. glabrata do often have much lower
calcium concentrations than those of L. stagnalis, it seems likely that the
calcium requirements of the two species may genuinely differ.
Thomas and colleagues (1974) reported some snail weight gain in
almost all of these culture volumes and treatments, including ‘0’ mg/l
calcium, over their brief study. In larger volumes of water (400 ml/snail),
growth seems to have been significantly lower at ‘0’and 2.5 mg/l calcium,
but constant in 5–80 mg/l calcium. Egg production seemed to be more
sensitive to external calcium concentration, on the other hand. No eggs
were laid by any snail at ‘0’ mg/l, very few were laid at 2.5 mg/l, and pro-
ductivity increased steadily through the concentration series up to over 5
eggs/snail/day at 80 mg/l calcium.
The approach taken by Meier-Brook (1978) to measure the calcium
requirements of the large ampullariid Marisa cornuarietis was more indi-
rect. He prepared a ‘standard fresh water’ series based on mean world
ratios of ions, ranging from 150 mg/l calcium to 10 mg/l. He then held
4–7 Marisa in individual beakers of water from this series for periods of
several hours, measuring calcium concentrations before and after. Meier-
Brook found that net calcium uptake increased with increasing external
concentration up to about 50–75 mg/l. An especially striking peak
uptake at about 75 mg/l was obtained from ‘calcium-starved’ snails –
those which had been acclimated for 40 days at 25 mg/l calcium. Thus
it appears that a calcium concentration as high as 25 mg/l may still be
suboptimal for Marisa. Meier-Brook noted that the snails may be reared
in a 25 mg/l medium, although their survivorship is initially much lower
than observed at 100 mg/l, and shells are thinner.
330 · Biogeography
Figure 8.2. Average size (below) and net reproductive rate (above) for Helisoma duryi
and Biomphalaria camerunensis reared 10 weeks in water of varying calcium concen-
tration (Madsen 1987).
Field studies
Hornbach and Cox also took monthly quantitative samples of their two
P. casertanum populations in the wild over a period of one year. The most
striking difference was that individuals from the harder water population
reached a much greater maximum size. But individuals of the softwater
population seemed capable of reproduction at a smaller size, so both pop-
ulations gave birth on the same schedule, summer and autumn. They
counted a greater number of embryos per adult in the hardwater popu-
lation, however, suggesting that fecundities in hard water may be greater.
Probably because of their larger adult sizes, birth rates in caged individ-
uals from the hardwater population were greater than those from the soft-
water population, even when transferred to soft water. The reverse
experiment, softwater populations transferred to hard water, resulted in
some minor improvement in birth rates. Thus it would appear that envi-
ronmental calcium influences both the growth rates and the fecundity of
Pisidium in the wild.
Kilgour and Mackie (1991) compared 10 demographic variables to 11
Environmental calcium · 333
environmental variables in 17 Ontario populations of P. casertanum.
Samples of clams, water, and sediment were collected on 12 occasions,
April–December. It is difficult to assess the significance of the matrix of
10⫻11⫽110 correlation coefficients their analysis provided. (For ‘table-
wide’ significance at the 0.05 level, Bonferroni correction would require
an apparent P value of 0.00045.) Kilgour and Mackie performed two
principal component analyses, reducing the environment to 4 variables
and the demography to 3 variables, the interpretations of which were
problematic. They then noted two ‘significant’ values in their new matrix
of 3⫻4⫽12 correlations among principal components.
Inspecting the original matrix of 110 correlations directly, one notes
that the highest correlations are positive ones between particle size or
percentage coarse substrate lost on ignition (i.e. leaf fragments) and the
mean adult length or weight of Pisidium. This would seem a local phe-
nomenon, attributable to the food and habitat preferences we noted for
Pisidium in Chapter 2. There are several large negative correlations
between temperature and maximum adult shell size or weight or the adult
length at which shelled larvae appear, as well as a positive correlation
between temperature and abundance. This suggests that P. casertanum
grows larger and reproduces less in cold waters, an effect that could be
due both to local differences in the depths at which samples were taken
and to larger climatic differences over 2° of latitude. Finally, maximum
adult weight or proportion of the population with shelled larvae seem to
be positively correlated with calcium hardness, alkalinity, or conductiv-
ity. This tends to substantiate the results of Hornbach and Cox.
Kilgour and Mackie reported no significant correlation between the
abundance of P. casertanum and any water chemistry variable. However,
the environments selected by them ranged in mean calcium hardness
from 4 mg/l to 225 mg/l. We have noted that freshwater mollusc popu-
lations may be adversely affected by hardnesses that are too high as well
as too low. Although there is considerable scatter, Figure 8.3 shows that
mean P. casertanum abundances seemed to increase with hardness to a
point, then decline. No monotonic relationship would seem expected
over such a range, without data transformation.
If the optimum value for some water chemistry variable is Wopt, then
the quantity
W⫽cos (Wopt ⫺Wobs)
will equal 1.0 when the observed value (Wobs) matches Wopt. Values of
W will be symmetrical around Wopt, decreasing to ⫺1.0 as observed
334 · Biogeography
Figure 8.4. Total (usually over 12 months) individual snails counted for 5 species
collected at 12 sites in Zimbabwe (Williams 1970a). Two pairs of sites (4 and 5, 6
and 7) were located in the same water bodies and hence had identical calcium values.
These pairs are combined and their average snail densities graphed. B.g., Bulinus glo-
bosus; Bio, Biomphalaria pfeifferi; Gy, Gyraulus spp.; Ly, Lymnaea natalensis; B.t.,Bulinus
tropicus.
Area
Thus a second major factor influencing species richness, habitat area,
comes to the fore. One of the primary goals Boycott (1936) set for
himself was to establish the lower limits of calcium concentration toler-
ated by elements of the British mollusc fauna. As such, he concentrated
on two softwater lakes with unusually diverse communities: Windermere
(‘exceptionally good . . . mature, sheltered with plenty of plants, silt and
mud and varied subhabitats in different parts’) and Leane (‘particularly
productive’). Figure 8.1 shows that these two lakes had many more
species than one would predict from their water chemistry. The impor-
tant factor seems to be that they are, with surface areas about 15 km2 and
24 km2 respectively, by far the largest lakes examined by Boycott.
At least four explanations have been advanced for the universal obser-
vation that larger habitats or islands have more species than smaller ones.
They provide larger ‘targets’ for immigrants, their greater habitat com-
plexity may facilitate coexistence once immigrants arrive, and the larger
populations they support are less likely to go extinct stochastically.
Trivially, a large sample from a fauna would be expected to contain more
species than a small sample. Early work on the species–area relationship
was reviewed by MacArthur and Wilson (1967), and more recently by
Lomolino (1989). Workers have generally settled on a power function to
model species-area relations, which together with its log transformation
is given below:
S⫽cAz
ln(S)⫽ln(c)⫹z ln(A)
where S is the number of species on a habitat island, A is the area of the
island, and c and z are fitted constants, typically obtained from a least-
Area · 339
Figure 8.5. Number of snail species as a function of surface area in Danish lakes.
(Data of Lassen 1975.)
N r C z (SE) F
Denmark (data of Lassen 1975)
Oligotrophic 19 0.96 ⫺1.6 ⫺0.23 (0.015) 218***
Eutrophic 68 0.83 ⫺6.1 ⫺0.16 (0.013) 149***
Finland (Aho 1978a)
43 – ⫺1.1 ⫺0.15 –
New York snails and unionids (Browne 1981)
37 0.93 ⫺1.2 ⫺0.23 –
Sweden (Bronmark 1985a)
115 0.40 ⫺2.8 ⫺0.11 –
Southern New England (Jokinen 1987)
Low calcium 23 0.85 ⫺4.0 ⫺0.19 (0.026) –
High calcium 41 0.65 ⫺5.6 ⫺0.12 (0.022) –
Northeastern New York (Jokinen 1987)
Low calcium 14 0.11 ⫺4.6 ⫺0.02 (0.043) –
High calcium 35 0.47 ⫺4.1 ⫺0.11 (0.034) –
Connecticut (data of Jokinen 1983)
Area only 90 0.48 ⫺3.8 ⫺0.14 (0.027) 26.6**
Calcium only 90 0.26 ⫺3.5 ⫺0.21 (0.085) 6.2**
Multiple 90 0.56 ⫺0.9 ⫺0.14 (0.025) a 20.0***
⫺0.24 (0.073) c
English Lake District (data of Macan 1950)
Area only 53 0.63 ⫺2.5 ⫺0.22 (0.042) 26.1***
Calcium only 53 0.20 ⫺0.26 ⫺0.34 (0.26) 1.65
Multiple 53 0.69 ⫺0.03 ⫺0.23 (0.040) a 17.33**
⫺0.47 (0.20) c
Distance only 41 0.31 ⫺0.79 ⫺0.011 (0.005) 4.15*
Africa (data of Brown 1994)
Area only 18 0.79 ⫺1.1 ⫺0.27 (0.052) 27.0***
Conductivity only 18 0.73 ⫺0.47 ⫺3.32 (0.78) 18.1***
Multiple 18 0.86 ⫺0.65 ⫺0.19 (0.055) a 21.2***
⫺1.85 (0.74) c
Unionacea, rivers of Eastern US (data of Sepkoski and Rex 1974)
Area only 41 0.67 ⫺0.0 ⫺0.32 (0.058) 31.9***
pH only 41 0.19 ⫺2.44 ⫺0.10 (0.085) 1.51
Distance only 41 0.27 ⫺2.73 ⫺0.14 (0.082) 2.99
Multiple 41 0.79 ⫺0.19 ⫺0.37 (0.049) a 32.4***
⫺0.24 (0.054) d
Area · 341
N r C z (SE) F
Notes:
N, number of habitat ‘islands’; r, correlation coefficient; C, y intercept, expressed
in hectares for simple area regressions (back-transformed from the log scale);
z, regression coefficient (and standard error); a, area; c, calcium or conductivity
depending on context; d, distance; F, value of F ratio from (P⬍0.05*,
P⬍0.01**, P⬍0.001***). All data are for snail faunas inhabiting lakes and ponds,
unless otherwise noted.
correlations between (log) number of snail species and (log) lake area are
extremely high in Denmark. The slope of the regression over all
eutrophic lakes combined seems to be about 0.16, a fair compromise
between the 0.09 and 0.37 reported by Lassen. The difference between
the 0.25 reported by Lassen for oligotrophic lakes and the 0.23 calculated
here is doubtless due to small errors in data transcription.
Figure 8.5 shows striking effects due to general trophic level, probably
equivalent to hardness or pH here. Oligotrophic lakes generally have
fewer species than eutrophic lakes, but the slopes of the two species–area
relationships are different. The y-intercepts reported in Table 8.1 (anti-
logs) show that 1 hectare oligotrophic ponds have fewer than a third of
the species expected in eutrophic lakes of similar size. Yet the species
richness of large oligotrophic lakes approaches that of large eutrophic
lakes. Here clearly displayed in Scandinavian lakes is the same interaction
reported by Dillon and Benfield in a southern Appalachian river. Again
the suggestion is that both water chemistry and habitat area affect the
food supply for freshwater molluscs, and it is to the food supply that
diversity responds.
The best fit to a mollusc species–area regression (r⫽0.93, Table 8.1) of
which I am aware was obtained by Browne (1981). His 37 sites in upstate
New York were distributed fairly evenly from springs less than 0.001 km2
up to Lake Erie, at over 25000 km2, and included from 1 to 78 species
of snails and mussels. In contrast, the 115 Swedish ponds sampled by
Bronmark (1985a) ranged only from 0.003 ha to a little over 2.3 ha, with
from 2 to 14 snail species. Table 8.1 shows that Bronmark’s species–area
342 · Biogeography
correlation was much lower, although still highly significant (r⫽0.40,
P⬍0.001).
Bronmark did not measure any water chemistry variables, noting that
his ponds were all eutrophic and underlain by a ‘homogeneous geologi-
cal background’. He did, however, have access to previously published
lists of macrophyte species at 45 of his ponds. The number of macrophyte
species ranged from 4 to 30, and Bronmark showed a significant
species–area relationship for the plants as well, fitting the equation 0.129
log A⫹0.629 with r⫽0.43 (P⬍0.001). It would seem that both habitat
complexity and food resources, at least for snails able to exploit macro-
phytes, can be significantly correlated with lake size. Bronmark obtained
a very nice multiple regression (r⫽0.902) using log number of snail
species as the dependent variable and pond area, number of macrophyte
species, and a measure of isolation to be discussed presently, all log-trans-
formed, as independent variables.
Clearly a more realistic representation of the effect of area on species
will be obtained if the effects of water quality can be taken into consid-
eration. Analytically, one might perform a multiple regression including
both area and calcium content, or some other water chemistry variable
correlated with calcium, as independent variables. Aho (1978b) surveyed
the gastropods in four ‘subareas’ of the lake district of southern Finland.
Although diversity in his subareas I and II was very low (5 species), his
regions III and IV had enough variance in snail species number to
warrent examination (12 and 15 species, respectively). The 20 lakes of
subarea III were all small, ranging from 0.003 km2 to 0.79 km2, with
calcium ranging from about 3 mg/l to 10 mg/l (converting from German
degrees of hardness). Here Aho reported a high correlation between
number of species and hardness (r⫽0.740, P⬍0.001) and a lesser one
with area (r⫽0.482, P⬍0.05). A multiple regression (not on a log scale,
unfortunately) found hardness and area together accounting for 73% of
the variance in species number. Lakes were larger in subregion IV,
ranging from 8.4 km2 to 119 km2, with calcium ranging from about 5
mg/l to 16 mg/l. Aho considered 11 of these lakes polluted and 11 ‘in
more or less natural condition’. There was a highly significant correla-
tion between hardness and species number among the 11 unpolluted
lakes (r⫽0.848, P⬍0.001), although none with area.
Aho (1978a) reported the results of a simple log-log regression analy-
sis of the species–area relationship over 43 lakes; essentially his combined
samples from subareas III and IV. Water chemistry was not included in
this model, although the fit would clearly have benefited, and a value of
Area · 343
r was not given. Nevertheless, such statistics as were supplied by Aho are
added to Table 8.1.
Lodge et al. (1987) examined previously published data on snail species
richness and physico- chemical variation in 64 northern Wisconsin lakes.
Two independent variables were included in their stepwise regression of
species number, area (F⫽16.2, P⬍0.001) and alkalinity (F⫽14.7,
P⬍0.001), the result being a very good fit (r⫽0.65). These results
compare very well to those of Aho. All variables were untransformed,
however, and thus it is difficult to compare Lodge’s findings to previous
species–area regressions as collected in Table 8.1. Jokinen (1987) also per-
formed stepwise multiple regressions of species number on area and seven
other independent variables. Her analysis was of two data sets: 46 lakes
in southern New England and 29 lakes in northeastern New York, all
greater than 10 ha. Lake area (F⫽7.25, P⬍0.01) and Na/Ca ratio
(F⫽5.88, P⬍0.05) were selected in the former analysis, and lake area
(F⫽38.4, P⬍0.001) and altitude (F⫽8.44, P⬍0.01) in the latter. The
stepwise procedure may have been misleading, however, because Na/Ca
ratio was highly correlated with calcium concentration in New England
(r⫽⫺0.42, P⬍0.01), as might be expected, as well as with altitude in
New York (r⫽⫺0.59, P⬍0.01). Again neither species number nor lake
area was transformed before analysis.
However, Jokinen also performed more conventional double-log
species–area analyses of a larger data set of ponds and lakes, further divid-
ing each of her two regions into high-calcium (⬎5 mg/l) and low-
calcium sets. Statistics for these four regressions are shown in Table 8.1.
The situation in New England is reminiscent of that reported by Lassen
in Denmark 10 years previously (Figure 8.5). Jokinen found excellent fits
to the (log)linear model for both hard (eutrophic) and soft (oligotrophic)
New England lakes, the former showing a higher intercept but lower
slope than the latter. Thus one expects more species in small hardwater
lakes than in small softwater lakes, but the difference disappears with
increasing size. Jokinen’s high-calcium regression for New York matches
her high-calcium New England regression rather well. The low-calcium
New York regression is non-significant, perhaps reflecting ‘the isolation
of the Adirondack lakes, with their depauperate fauna’.
Jokinen’s southern New England data deserve a more detailed exam-
ination. Data on a large fraction of the 64 lakes she analysed in 1987 are
available in her (1983) monograph on the freshwater snails of
Connecticut. The appendices of that earlier work list 230 sites, 92 of
which are lakes or ponds with a least one snail species present, with
344 · Biogeography
measurements of area (in hectares) and calcium (mg/l) available. Number
of species ranged from 1 to 17, area from 0.1 ha to 370 ha, and calcium
from 0.8 mg/l to 35 mg/l. (There is clearly little danger that calcium con-
centrations below 35 mg/l will affect freshwater snails adversely.) Setting
aside two outliers, Table 8.1 shows that the simple regressions of log area
on log species and log calcium on log species were both significant. The
area-only simple regression seems a fair compromise between the two
New England regressions of Jokinen (1987) previously tabulated,
showing an intermediate slope of 0.14 and a substantially lower r. It is
interesting to note that this slope of 0.14 is preserved in the multiple
regression, while r increases substantially, to 0.56 (probably as significant
with n⫽90 as the overall New England values).
So far in discussing the effects of lake area, we have proceded as though
all species were equally likely to inhabit lakes of any size. Indeed to the
extent that the species–area relationship is due to non-deterministic
factors such as sampling bias or ‘target size’, this would be true. But
judging from actual data, it appears that species vary in their ability to
colonize small, possibly unstable habitat islands, just as they appear to vary
in their tolerance for low calcium concentrations. One tends to find the
same subset of species in the smallest ponds, and as pond size increases,
the addition of species does not appear to be entirely random. The data
of Macan serve as an illustration.
Over 10 years, Macan (1950) made a series of qualitative and quanti-
tative samples of molluscs from the English Lake District. His studies
centred about the large, soft Lake Windermere, (1500 ha, about 5 mg/l
calcium). Macan catalogued 11 snail species in Windermere, the same
number Boycott obtained from the literature. (Note that Figure 8.1
includes bivalves as well.) Interestingly, Macan found four species not
reported by Boycott: two planorbids, the lymnaeid Myxas glutinosa, and
Potamopyrgus jenkinsi. He failed to collect two Ancylus species (he felt his
equipment was inappropriate for limpets): the deep water Valvata cristata,
and Planorbis crista, whose occurrence in Windermere he questioned.
Macan also collected snails in 12 other nearby lakes, in 36 of 41 bodies
of water he considered ‘tarns’ (14 ha to 0.08 ha), and in 4 of 15 ‘pools’
(less than 0.08 ha). Taken altogether Macan’s data set included 53 bodies
of water harbouring 16 species.
Macan measured calcium concentration (mg/l) on three occasions at
most tarns and pools, discounting the spotty data from August 1947. Data
were available for only two sampling periods in some cases, and only
single estimates were presented for the 12 lakes. Such data as were avail-
Area · 345
able I have averaged. The maximum average calcium concentration (22.4
mg/l) seems to be well below any reasonable estimate of Wopt, and thus
I judge cosine transformation to be unnecessary. Macan provided esti-
mates of surface area for all tarns, but not for lakes. I estimated lake areas
by clipping and weighing an enlargement of his Figure 1.
Table 8.1 shows statistics for the simple regressions of area and calcium
on number of snail species in 53 bodies of water from the English Lake
District. The species–area regression was highly significant and rather
canonical (r⫽0.63, F⫽26.1, P⬍0.001). Although the simple regression
of species number on log mean calcium concentration was not signifi-
cant, calcium did contribute significantly to a multiple regression when
entered as a second independent variable along with log area, as we have
seen in previous analyses. Table 8.1 shows that the multiple r improved
from 0.63 to 0.69, while the species–area regression coefficient remained
unaffected.
Of Macan’s 16 species, Myxas, Potamopyrgus, and P. carinatus were
found only in Windermere, at least as of 1950. Valvata cristata and
Lymnaea stagnalis were found in but two sites. Then setting these aside,
Figure 8.6 shows the mean, SEM, and range for the log surface area of
the bodies of water inhabited by the remaining 11 species. The error bars
are rather large for many of these, especially those only rarely collected
by Macan. However, it is clear from Figure 8.6 that significant differences
do exist in the mean lake sizes inhabited by the various taxa. The mean
size of the 6 tarns inhabited by Lymnaea glabra was aln (⫺1.53)⫽0.22 ha,
while the maximum size was only 1.6 ha. Boycott states that it ‘has been
generally recognized that L. glabra characteristically occurs in mean places
where few if any other Mollusca care to live – shallow grassy ditches and
ponds which usually dry in the summer.’ It is interesting to note that the
first eight species listed in Figure 8.6 are lymnaeids and planorbids, and
that generally only in the larger (and more diverse) lakes does one add
other families. At the other end of the spectrum we see Physa fontinalis,
found in 13 lakes averaging about aln (4.34)⫽77 ha, although ranging
down to 1.3 ha. Boycott says it is ‘not particular about volumes’ but is
‘never found in closed ponds or static water.’ It is also significant that
ranges, with just a couple of exceptions, are restricted at the low end, not
at the high. The snails all generally seem to be able to inhabit large bodies,
but vary in their capacities to colonize smaller, perhaps more unstable
habitats successfully.
The number of species on a habitat island is usually, today, thought of
as a function of immigration and extinction. But from a community
346 · Biogeography
Figure 8.6. Means, standard errors of the mean, and ranges for the surface area of
the water bodies inhabited by the 11 most widespread gastropods in the English Lake
district (data of Macan 1950). Lgl, Lymnaea glabra (N⫽6); Pcr, Planorbis crista (4); Pal,
P. albus (35); Lpe, Lymnaea pereger (45); Lpa, L. palustris (22); Ltr, L. truncatula (8);
Pcm, Planorbis complanatus (8); Psp, P. spirorbis (4); Vpi, Valvata piscinalis (14); Pcn,
Planorbis contortus (15); Pfo, Physa fontinalis (13).
Figure 8.7. (Top) Regression of species number in African lakes on lake area.
(Middle) Regression of species number in African lakes on log10 conductivity (cosine
transformed). (Bottom) Residual variation in the number of snail species in African
lakes not explained by a multiple regression on area and conductivity. The lakes are
arranged by the number of species they contain, left to right. The abbreviations for
the individual lakes are: Ab, Abaya (Margherita); Al, Albert; Aw, Awasa; B,
Bangweulu; C, Chad; E, Edward; G, George; K, Kivu; L, Lungwe; Ma, Malawi
(Nyasa); Mw, Mweru; Nb, Nabugabo; Nv, Naivasha; R, Rudolf (Turkana); Tg,
Tanganyika; Tn, Tana; V, Victoria; Z, Zwai. The number ‘4’ marks lakes E, Mw, Ab,
and Aw. (Data of Brown 1994.)
348 · Biogeography
33 mhos, respectively), while conductivity is very high in Kivu and
Rudolf (1240 and 2750 mhos).
The results of a simple regression of species number on transformed
median conductivity are also given in Table 8.1 and shown in Figure 8.7
(middle). Values of conductivity were log10 transformed (rather than loge,
given their range) and cosine transformed, as detailed in the previous
section. Again the value of Wopt was obtained iteratively. The best fit to
a linear regression (r⫽0.73) was obtained when Wopt ⫽2.5, that is, when
optimum conductivity is taken as about 316 mhos. In the middle
section of Figure 8.7 one especially notices that high- conductivity lakes
K and R fit the regression much more closely.
Finally, Table 8.1 shows the results of a multiple regression including
both area and transformed conductivity as independent variables. Both
variables contributed significantly to the r⫽0.86 fit, with the log–area
slope reduced from ‘canonical’ to 0.19, a value more consistent with the
other values in Table 8.1. The residual variance from this multiple regres-
sion, shown in the lower section of Figure 8.7, is in many ways the most
interesting component of the data. Here the ordinate has been converted
to read in number of species directly by subtracting the anti-loge of the
regression estimate from the value observed. One might be surprised,
judging from the scales of the upper graphs in Figure 8.7, that Lake
Tanganyika has 34 species more than the regression estimate. Also
showing an unusual ‘excess’ of species, above what one might expect
from their sizes and conductivities, were Lakes Victoria (⫹6.5), Awasa
(⫹5.5), and Malawi (⫹3.7). Showing unexpected deficits of species were
lakes Naivasha, Tana, and the two high-conductivity lakes Kivu and
Rudolf.
These 18 lakes range in median latitude from about 13° N for L. Chad
to 12° S for L. Malawi. Given the generality of the observation that diver-
sity increases toward the equator, one might speculate that latitude could
account for some of the residual variation shown in Figure 8.7. But
simple regressions of species number on latitude, both log-transformed
and untransformed, were not significant, nor did latitude contribute sig-
nificantly to a multiple regression with area and conductivity. In fact the
mean latitude of the four negative-residual lakes is lower than the mean
latitude of the four positive-residual lakes.
Noting that the deeper lakes are generally the oldest and most perma-
nent, Brown speculated that snail diversity might be a function of depth.
Lake Tanganyika is in fact the deepest of the African lakes, at 1435 m
maximum. The simple regression of log depth on log species number in
Isolation · 349
the 18 African lakes was slightly more significant than the untransformed
regression (F⫽5.75, P⬍0.05, r⫽0.53). But because log depth is highly
correlated with log area in this data set (r⫽0.63, P⬍0.01), its contribu-
tion to a multiple regression was not significant. As a third independent
variable, log depth had a regression coefficient of 0.078⫾0.070, improv-
ing the multiple r only from 0.86 to 0.90.
D. Brown (pers. comm.) has further suggested a role for altitude. Lakes
Tanganyika, Victoria, Awasa, and Malawi are low enough to enjoy truly
tropical climates, while Lake Tana, for example, is quite high. But it
seems unlikely to me that any single variable acting on a biogeographi-
cal scale can be found to contribute much beyond the 0.862 ⫽74% of the
variance explained by area and water quality here. For an explanation of
the variance shown in Figure 8.7, one might better look within-lakes
than among-lakes. Lake Naivasha, for example, dried out completely in
fairly recent times.
It is interesting to note that Brown counted 30 endemic species among
the Lake Tanganyika snail fauna, almost exactly the residual number in
Figure 8.7. The fit is not as good elsewhere in Africa. Lake Malawi has
3.7 extra species but 17 of its 27 species are endemic, while Victoria has
6.5 extra species with 13 of its 28 endemic. Lake Awasa, with its residual
of ⫹5.5 species, has no endemics at all. Gorthner (1992) has suggested
that some fraction of the speciation seen in ancient lakes may be non-
adaptive. Regardless of the explanation one may favour for the data of
Figure 8.7, it is fair to speculate that if competition is to be found, it is
in lakes such as Awasa, Victoria, Malawi, and Tanganyika.
Isolation
Among the contributions made by MacArthur and Wilson (1967) to the
theory of island biogeography was their analysis of the effects that dis-
tance from a mainland source area ought to have on the equilibrium
number of species. In addition to his observations on the effects of lake
area and water chemistry, Aho (1978a,b) examined his data on the gas-
tropods of southern Finland for evidence of distance effects. He chose to
concentrate on the 20 small lakes of his ‘subarea III’ in the vicinity of
large, diverse, Lake Pyhäjärvi. Aho detected a significant negative corre-
lation between species number in these lakes and their distances to
Pyhärjärvi, just as would be expected if the larger lake were serving as a
source of colonists. But he also found a highly significant positive cor-
relation between distance from Pyhäjärvi and water hardness. Since no
350 · Biogeography
distance-diversity effect was apparent when the effects of water chemis-
try were partialled out, Aho did not pursue this line of inquiry further.
Aho did report, however, a significant positive correlation between
species number in his low-diversity ‘subarea I’ and the percentage of the
total drainage basin occupied by lakes, which might be taken as an indi-
rect measure of the importance of isolation. Bronmark (1985a) also
included several variables designed to estimate indirectly the significance
of isolation in his analysis of Swedish gastropod faunas discussed previ-
ously. For about 100 ponds he measured distance to the closest pond
(nearest-neighbour distance), mean distance to the five closest ponds, and
number of ponds in a 50 ha square centred about the pond under con-
sideration. Two additional variables, number of species in the nearest
pond and in the five nearest ponds, could be taken either as measures of
isolation or as measures of the general suitability of the local environ-
ment. Of these five variables only the last, mean number of species in the
five nearest ponds, was entered (along with pond area and macrophyte
diversity) into Bronmark’s stepwise multiple regression of gastropod
species number.
The most influential examination of the effects of isolation on the
number of freshwater mollusc species to date may have been that of
Sepkoski and Rex (1974). They analysed lists of unionid species for 49
‘coastal rivers as biogeographic islands’, ranging from the Penobscot
River of Maine to the Escambia River on the Gulf Coast of Florida.
Estimating area as land drained (rather than land inundated, as customary
for lakes), they obtained a double-log fit (r⫽0.60***) to the equation
S⫽0.78A0.316.
Sepkoski and Rex obtained values of 10 environmental variables for
their 49 rivers: temperature (both mean and variance), dissolved solids,
and seven water chemistry measures. They also estimated degree of iso-
lation as the number of stepping-stones to each of four presumed source
rivers: the St Lawrence (north of the study area), the Alabama-Coosa
(south of the study area), and two internal sources, the Appalachicola and
the Savannah, chosen by their diversity. The number of steps to nearest
source area was included as well. Thus Sepkoski and Rex had in their
quiver 16 independent variables (area, 5 stepping-stone measures, 10
environmental measures) with which to predict the dependent variable,
number of unionid species. At the end of 12 multiple regressions using
various combinations of these, both log-transformed and untransformed,
they obtained a multiple r⫽0.89 with an ‘interactive model’ involving
hydrogen ion concentration (antilog10 of ⫺pH), dissolved solids, nitrate
Isolation · 351
concentration, and A/AC and A/SV, drainage area divided by stepping
stone distances to the Alabama-Coosa and the Savannah, all untrans-
formed.
I would prefer a simpler approach. From the author’s appendix 1 I
selected area and hydrogen ion concentration (data were unavailable for
calcium or alkalinity), two variables which we have good reason to expect
will influence freshwater mollusc diversity. In addition, over the distances
encompassed by this survey (15 degrees of latitude) one might expect
climate or coastal distance from the tropics to influence diversity. Thus I
selected stepping-stone distance from the Alabama-Coosa River (ranging
from 1 in the south to 33 in the north) not as a measure of isolation but
as a measure of latitude. These three variables were log-transformed
(effectively reconverting hydrogen ion concentration to pH) and used as
independent variables in an analysis of unionid diversity.
Three outliers were identified in the initial regressions: the Blackstone
River with an extremely small drainage area and the Satilla and
Waccasassa Rivers with very low numbers of unionid species (two each).
The results of my regression analyses based on 41 rivers are shown in
Table 8.1. My simple species–area regression was not substantially differ-
ent from that obtained by Sepkoski and Rex, with an identical estimate
of 0.32 for the slope and an improved r. (Note that all values of C in Table
8.1 have been converted to hectares, here from the mi 2 used by Sepkoski
and Rex.) The simple regression of log species number on pH was not
significant, nor did pH add significantly to subsequent multiple regres-
sions. The simple regression on log stepping-stone distance from the
Alabama-Coosa River approached significance, and when included in a
multiple regression with log area improved the multiple r from 0.67 to
0.79. Thus a regression with two independent variables, area and an esti-
mate of climate or latitude, is only slightly less predictive than that of
Sepkoski and Rex, with five independent variables. One must judge the
evidence weak that isolation and dispersal have played a role in the diver-
sity of eastern United States unionid faunas.
We now return to the data of Macan (1950) on the gastropods of the
English Lake District. In the previous section it was demonstrated that
the number of species in 53 bodies of water in this region seemed to be
a function of habitat area and, to some extent, calcium concentration.
Five of the 12 lakes surveyed by Macan were located within 10 km of
large, diverse Lake Windermere, as were all 36 of the tarns. Among this
subset of 41 water bodies, Macan found five gastropod species absent
from his Lake Windermere list, as might have been guessed from Figure
352 · Biogeography
8.6. Subtracting these species from the several sites where they occurred,
it is reasonable to hypothesize that Windermere has served as the source
for the remaining 11 elements of the Lake District gastropod fauna.
The relationship between species numbers and distance-from-source
is customarily modelled logarithmically (e.g. Lomolino 1989):
S⫽e⫺kI ** 2
where I is the measure of isolation (generally distance), e is the base of
natural logs, and k is a fitted constant. Taken together with the effects of
habitat size and water chemistry, species richness can now be modelled:
S⫽cAz Wm e⫺kI**2
ln(S)⫽ln(c)⫹z ln(A)⫹m ln(W) – kI 2
where area is A, the measure of water chemistry is W, and c, z, m, and k
fitted constants. In its log-transformed version, a multiple regression may
be used to fit the constants by least squares.
I estimated dispersal distance as the minimum linear distance from
Windermere to each of the 41 smaller lakes and tarns, taken from an
enlargement of Macan’s figure 1. Before including these data in a multi-
ple regression with area and calcium as above, I performed a simple
regression analysis. I found a significant positive regression of log species
on squared distance from Lake Windermere (Table 8.1)! It happens that
by chance of geology, regions outlying Windermere tend to have harder
water than those nearby. The correlation between calcium concentration
and distance (squared) to Windermere is 0.55 (P⬍0.001). Thus diversity
tends to increase away from the putative ‘source’, a reverse of Aho’s
findings.
Clearly these data show no more evidence of isolation effects than any
other we have examined. We must conclude this section by noting that
it has generally proved difficult to identify source areas from which to test
the effects of isolation on the diversity of freshwater mollusc commu-
nities. The small number of tests that have been made to date have yielded
unspectacular results.
Community composition
R. H. Green’s (1971) ‘multivariate approach to the Hutchinsonian niche’
must be numbered among the more influential of all studies involving
freshwater molluscs. In this first use of multiple discriminant analysis in
community studies, Green analysed the 10 most widespread bivalves of
central Canada: 2 Sphaerium species, 6 Pisidium, Lampsilis radiata and
Anodonta grandis. His environmental variables were about half of the
between-lake and half of the within- lake variety.
For each of 32 lakes surveyed, Green took one water sample above the
thermocline and one below, or only a single such sample if stratification
was absent. From these samples he measured pH, alkalinity, calcium, total
hardness less calcium, and sodium chloride. These last four were log
transformed. He then took a large number of qualitative bottom samples
from each lake. For each sample containing bivalves he measured depth
(log transformed), mean sediment particle size (also log transformed to
‘phi units’), the standard deviation of the particle size (in phi units – the
‘sorting coefficient’), and percentage loss on ignition (arcsine trans-
formed). Knowing depth, Green could accord each sample estimated
values for the five water chemistry variables.
The most widespread species, Anodonta, was found in 62 samples (18
lakes), next was Pisidium casertanum in 52 samples (12 lakes), and so on,
down to Sphaerium lacustre in 11 samples (5 lakes). Green’s aim was to use
discriminant analysis to distinguish samples containing the 10 species as
far as possible: to separate the 62 Anodonta samples from the 52 P. caser-
tanum samples, and so forth. Some overlap would seem inevitable,
however, since many of Green’s samples doubtless contained more than
one species. Further, his 62 Anodonta samples, for example, did not
contain 62 different values of the five water chemistry variables, but at
Community composition · 355
most 2(18)⫽36. Since many of the 18 lakes containing Anodonta were
surely also among the 12 lakes containing P. casertanum, the potential for
overlap on the water chemistry ‘axes’ of niche dimension would seem
great.
Five discriminant functions (DF) were required to account for 95% of
the among-species variance. Interestingly, only two variables showed
high coefficients in the first DF (45%): total alkalinity (⫺0.77) and
calcium (0.61). The Pisidium species scored low on this DF, and the uni-
onids high. Regarding the opposite symbols of these two variables, Green
noted that in small, closed basins with high hardness and little water turn-
over, calcium carbonate precipitates before other salts. A number of such
ponds were included among Green’s sample sites, and they apparently
contained Pisidium but no unionids. Thus Green interpreted his first DF
as a function of lake size or water turnover. It is interesting that lake area,
neglected in the introduction and methods sections of Green’s paper,
came to figure so prominently in the discussion.
Most of the remainder of the discrimination was due to within-lake
factors. Depth showed a very high coefficient (0.822) in Green’s second
DF (24% of the variance), and (decreasing) particle size a rather smaller
one. Here the two unionids seem to inhabit shallower water with coarser
substrate, with the Pisidium species (especially P. conventus) deeper.
Green’s DF 3 was also primarily interpretable in terms of sediment char-
acteristics. Such within-lake variables were the subject of our discussion
in Chapter 2.
In studies of this sort it is customary to furnish statistics regarding the
power of the discriminant functions obtained. One often sees classifica-
tion matrices giving percentage of correct assignments or scatter plots
with species means and confidence limits. The absence of such informa-
tion in Green’s work leads me to speculate that he uncovered a phenom-
enon well known by freshwater mollusc ecologists since the nineteenth
century. Across any typical sample as environmentally diverse as Green’s
32 lakes seem to have been, the general trend may have been for bivalves
to occur together, not apart.
Boycott felt it so unlikely that British species of freshwater molluscs
typically interact in any way that he declined to use the term ‘associa-
tion’, preferring ‘concurrence’ instead. Nevertheless he was quite inter-
ested in ‘concurrences’. Boycott tabulated data on the distribution of
Pisidium from 466 sites in Britain provided by C. Oldham. For each of
15 species he calculated the ‘average number of companions’: the average
number of other species of Pisidium in habitats occupied by the species
356 · Biogeography
in question. These figures varied from 1.1 and 1.3 for P. obtusale and P.
personatum up to 6.1 and 6.3 for P. moitessierianum (‘torquatum’) and P.
tenuilineatum. He then proposed the ‘hierarchy of gentility’ shown at the
top of Figure 8.8, from P. personatum (‘which lives in the slums’) up to
‘henslowanum and amnicum which insist on the best places with 4 or 5
other species and supinum, tenuilineatum, and ‘torquatum’ which are even
more particular.’
This is a different (but not less valid) approach from the one I have
pursued in this chapter thus far. Rather than inferring the suitability of a
water body by its size and water quality, Boycott effectively let the clams
themselves tell him which habitats were superior. In so doing, he reduced
our multiple dimensions (habitat size, water quality, and perhaps other
variables as well) down to a single one: ‘average number of companions’.
Some 40 years later, Bishop and Hewitt (1976) confirmed Boycott’s
judgments. They took quantitative samples of Pisidium from 182 localites
in eastern England. Three rare species, P. hibernicum, P. moitessierianum,
and P. pulchellum were found less than five times. They then performed
simple chi-sqare tests for association between all pairs of the remaining
nine species. The result (bottom of Figure 8.8) is quite strikingly similar
to Boycott’s ‘hierarchy’. Three species, P. milium, P. personatum, and P.
obtusale were not significantly associated with any other. A putatively sig-
nificant association was detected between P. nitidum, P. subtruncatum, and
Community composition · 357
Figure 8.9. Three example incidence functions compiled by Jokinen (1987) from
the freshwater snail faunas of New York and New England. Gyraulus deflectus is a
‘high-S’ species, Amnicola limosa a ‘C–D tramp’, and Pseudosuccinea columella a ‘super-
tramp’.
Table 8.2. Summary statistics for loge surface area and loge calcium
concentration at the lakes and ponds inhabited by the 24 most widespread snail
species in Connecticut
Species are ranked by the average area of their habitats
Stagnicola elodes 6 ⫺2.53 0.62 (1.07) 4.40 1.16 2.64 (0.31) 3.14
Helisoma trivolvis 12 ⫺2.12 1.28 (0.66) 4.39 0.87 2.28 (0.19) 2.98
Planorbula armigera 13 ⫺3.91 1.34 (0.73) 4.76 0.79 1.67 (0.16) 2.77
Physa vernalis 5 ⫺3.00 1.38 (1.67) 5.33 1.22 1.78 (0.24) 2.49
Pseudosuccinea
columella 41 ⫺3.91 2.15 (0.34) 5.33 0.09 1.66 (0.11) 3.08
Lyogyrus granum 15 ⫺3.91 2.23 (0.57) 4.70 0.26 1.54 (0.16) 2.48
Ferrissia fragilis 37 ⫺4.61 2.33 (0.38) 5.33 0.10 1.53 (0.12) 3.14
Physa heterostropha 36 ⫺3.22 2.38 (0.40) 5.91 0.79 2.47 (0.10) 3.56
Micromenetus
dilatatus 45 ⫺3.91 2.59 (0.35) 5.62 0.09 1.46 (0.11) 3.08
Gyraulus parvus 30 ⫺2.53 2.70 (0.36) 5.92 1.10 2.27 (0.13) 3.56
Helisoma anceps 41 ⫺3.91 2.78 (0.32) 5.41 0.47 1.85 (0.12) 3.56
Lyrogyrus pupoidea 7 ⫺1.81 2.82 (0.29) 3.91 0.47 1.57 (0.32) 2.76
Physa ancillaria 24 ⫺3.91 2.88 (0.41) 5.43 0.47 1.72 (0.17) 3.56
Physa gyrina 13 ⫺1.72 3.01 (0.52) 4.79 0.69 1.16 (0.11) 2.10
Promenetus
exacuous 20 ⫺0.49 3.05 (0.36) 5.91 0.99 2.21 (0.15) 3.56
Fossaria modicella 14 ⫺0.53 3.10 (0.34) 5.43 0.99 1.98 (0.18) 3.56
Laevapex fuscus 23 ⫺1.17 3.24 (0.17) 4.40 0.47 1.74 (0.16) 3.08
Campeloma decisum 22 ⫺0.92 3.47 (0.25) 5.62 0.09 1.73 (0.14) 3.56
Cipangopaludina
chinensis 8 ⫺0.60 3.52 (0.53) 5.43 1.96 2.65 (0.20) 3.56
Amnicola limosa 45 ⫺1.31 3.57 (0.16) 5.62 0.09 1.74 (0.14) 3.56
Ferrissia parallela 5 ⫺1.55 3.65 (0.64) 5.43 0.69 2.54 (0.49) 3.56
Gyraulus deflectus 23 ⫺1.81 3.74 (0.21) 5.43 0.47 2.00 (0.17) 3.56
Helisoma
campanulata 16 ⫺1.41 3.74 (0.30) 5.92 0.26 2.10 (0.26) 3.56
Valvata tricarinata 6 ⫺2.24 4.16 (0.49) 5.43 2.48 3.03 (0.15) 3.56
Note:
N, lakes and ponds inhabited by the 24 most widespread snail species.
Source: data from Jokinen (1983).
362 · Biogeography
Figure 8.10. Twenty-four Connecticut snail species, plotted by the mean area and
calcium concentration of their habitats (Table 8.2; data of Jokinen 1983). Closed
symbols are prosobranchs: diamond Valvata, square vivipariids, circle hydrobiids.
Open symbols are pulmonates: squares lymnaeids, circles physids, diamonds ancyl-
ids, erect triangles small planorbids, and inverted triangles Helisoma. Dashed lines
divide the species into three groups referred to in the text as U, S, and R.
example, that the 3 smallest mean habitat areas, plus 2 of their standard
errors, do not overlap the 3 largest mean habitat areas minus 2 standard
errors. This phenomenon does not seem attributable to variation in the
maximum lake size inhabited; almost all species seem to have been found
in large lakes. Rather, the species vary strikingly in the minimum size of
the lake or pond in which they were found. Although not as obvious in
Table 8.2, mean calcium concentration varies as widely as mean habitat
area, again the result of differing species minima. The 3 smallest mean
calcium concentrations (those of Physa gyrina, Micromenetus dilatatus, and
Ferrissia fragilis), plus 2 standard errors, do not overlap the 3 largest (Valvata
tricarinata, Cipangopaludina chinensis, and Stagnicola elodes) minus 2 standard
errors.
The 24 species are plotted in Figure 8.10 by their mean habitat areas
and mean calcium concentrations. I have drawn a dashed line through an
apparent gap in the distribution on the abscissa, while the ordinate is
divided by a dashed line at 5 mg/l calcium. These lines are not meant to
imply absolute boundaries. Rather, they are drawn to divide the graph
into four regions for the sake of further discussion.
The region I have labelled R is occupied by the first four species listed
Community composition · 363
in Table 8.2. These species seem to be adapted to habitats of small surface
area, probably subject to freezing and drying and possibly temporary. The
mean calcium concentration in these habitats is, on the other hand, at
least fairly high, suggesting that they are productive. Notice for example
that Stagnicola elodes inhabits ponds of the smallest mean surface area, but
with the third highest mean calcium concentration. I therefore suggest
that, in the main, Connecticut populations of S. elodes, Helisoma trivolvis,
Planorbula armigera, and Physa vernalis are R-adapted, in the sense of
Chapter 4. We have previously reviewed a considerable body of evidence
suggesting that at least some Indiana populations of Lymnaea (Stagnicola)
elodes show R-adaptation (population ‘F’ of Brown 1982). A New York
population of Helisoma trivolvis was, in fact, one of two populations
known to have demonstrated an order of magnitude greater reproductive
effort than predicted for body size (see Figure 4.3). So for the four species
in region R of Figure 8.10 I predict a high reproductive effort, a short
lifespan, and semelparity (see Table 4.2).
The species occupying the region I have labelled S in Figure 8.10
inhabit regions low in calcium, perhaps relatively unproductive.
However, these sites are larger, and in no case temporary. S-adapted pop-
ulations seem to tolerate low food and perhaps osmoregulatory stress in
exchange for environmental predictability, and as a result their overall
reproductive capacity may be reduced. Notice that the species with the
lowest mean calcium concentration, Physa gyrina, has a mean habitat area
above average for the 24 species. It may be recalled that Indiana popula-
tions of P. gyrina are unusually broad in their diet and habitat use (Brown
1982) and that laboratory populations show unusual temperature toler-
ance (van der Schalie and Berry 1973, figure 4.1). Thus I suggest that
Connecticut populations of Physa gyrina and to a lesser extent
Micromenetus dilatatus, Ferrissia fragilis, and the two Lyrogyrus species tend
to be S-adapted in the sense of Chapter 4. I hypothesize that their repro-
ductive effort, relative to their body size, may be reduced.
Region U includes species typically found in habitats good in all phys-
ical respects. U- populations are undifferentiated with regard to life
history traits, and as such can mature slowly or quickly and reproduce
semelparously or iteroparously. But because gastropod diversity will be
expected to reach a maximum in large, hard water habitats, U-popula-
tions would seem most likely to encounter interspecific competition.
This theme will be pursued in the next chapter.
We have noted that in an ‘incidence function’ researchers effectively
collapse the two axes graphed in Figure 8.10 to a single axis. Scoring
364 · Biogeography
lowest on such axes one finds ‘supertramps’, followed by ‘A–B tramps’,
‘C–D tramps’, and finally ‘high-S species’. One might imagine a diago-
nal line running through Figure 8.10 and expect correspondence
between projections of the 24 species on it and Jokinen’s (1987) classifi-
cation of them into incidence categories. To offer some anecdotal evi-
dence, the square symbol very close to the intersection of the two dashed
lines plots the average environment of Pseudosuccinea columella. One might
argue that Pseudosuccinea occupies the worst average environment occu-
pied by any of the 24 species; both rather soft and rather small. Of all the
incidence functions plotted by Jokinen, that of P. columella (Figure 8.9)
best fits the description of ‘supertramp’. Similarly, the species with the
most strikingly ‘high-S’ incidence function identified by Jokinen, Valvata
tricarinata, would appear to score highest on a diagonal axis drawn in
Figure 8.10.
Summary
The number of mollusc species present in a lake or river tends to increase
as a function of a set of variables normally intercorrelated in natural fresh
waters: calcium concentration, hardness, alkalinity, pH, and conductiv-
ity. Laboratory studies have confirmed years of field observations
showing that calcium concentration in particular may directly affect both
the growth and the reproduction of freshwater molluscs. Importantly, it
appears that populations vary in their adaptation to calcium stress. Thus
as the field biologist samples lakes and rivers of increasing hardness, the
species count generally increases, up to a point. But laboratory studies
have also shown that water of very high hardness (for example, greater
than 40 mg/l calcium) may be detrimental to freshwater molluscs. Cosine
transformation of the dependent variable can be employed to analyse the
diversity or abundance of molluscs as a function of water quality ranging
into very high hardnesses.
There is evidence, however, that the effects of calcium and related
water chemistry variables on mollusc populations may be largely indirect.
Harder waters are generally richer, supporting more primary productiv-
ity, greater biomass, and higher levels of decomposition. Perhaps it is to
the greater availability of food, rather than to the improved osmotic envi-
ronment, that natural populations of freshwater molluscs are responding.
Habitat area, measured as surface area for lakes and ponds or drainage
area for rivers and streams, is a second variable well known for its influ-
ence on freshwater mollusc diversity. Table 8.1 shows the results of 14
Summary · 365
double-log species–area regressions, including four data sets from
Scandinavia, one from England, one from Africa, and the remainder from
the eastern United States. The correlation coefficients of 13 are highly
significant. Values of the slope (z) approach a ‘canonical’ value of 0.25 in
four of these cases, seem substantially less in seven cases and greater for
the two unionacean data sets. Just as the case with calcium, individual
mollusc species seem to vary in their adaptation for smaller rivers or
ponds, so that species are not added randomly as area increases.
Table 8.1 also shows that trophic level or water quality often signifi-
cantly affect the relationship between habitat area and freshwater mollusc
diversity. Harder, more eutrophic ponds have more species, but the slope
of their species–area regressions are generally shallower. Hence the diver-
sity of a large oligotrophic lake may approach that of a large eutrophic
lake. The inclusion of a measure of water quality along with habitat area
in a multiple regression analysis of mollusc species number may improve
the fit significantly.
These two independent variables, water chemistry and habitat area, are
as useful for explaining the diversity of the ancient African lakes as they
are for the recently colonized lakes of Scandinavia. In general, other
important variables are difficult to identify. Isolation, measured as distance
from presumed sources of colonists, would seem a likely candidate for
freshwater molluscs. Little evidence of an isolation effect was obtained in
an analysis of the snail faunas in small lakes and ponds near England’s Lake
Windermere or in the unionid faunas of eastern United States coastal
rivers. Some latitude effect may be present in the latter data set. Most of
the data sets reviewed here were collected at scales too small to reflect
substantial influence from climatic variation, and large enough to remove
the influence of depth, substrate, and current.
Workers have often found that freshwater mollusc species do not
appear to assort independently on a geographic scale, but rather seem to
form ‘clusters’, ‘concurrences’ or ‘associations’. There seems to be little
evidence that such associations reflect any sort of interaction among the
species, however. Rather, it appears that the distribution of each species
is an independent consequence of the environment. I suggest that not
only may the number of species inhabiting a particular lake or river be
modelled as a function of water chemistry and habitat area, the compo-
sition of the molluscan community may be as well.
To illustrate this approach, 24 species from the freshwater snail fauna
of Connecticut were plotted by the mean size and mean calcium con-
centration of their habitats (Figure 8.10). Three categories were recog-
366 · Biogeography
nizable: R-populations adapted to small (i.e. ephemeral) but productive
habitats, S-populations adapted to soft (i.e. unproductive) but permanent
habitats, and U-populations adapted to permanent, productive habitats.
In the final chapter of this work, the potential for interactions among
such mollusc populations as may co-occur in permanent, productive hab-
itats will be examined in additional detail.
9 · Communities
Gut content
Dudgeon and Yipp (1985) collected samples of nine Hong Kong gastro-
pods: the prosobranchs Melanoides tuberculata, Bithynia spp., and Sinotaia
quadrata, and the pulmonates Physa acuta, Lymnaea ollula, Radix plicatulus,
Biomphalaria straminea, Gyraulus sp., and Hippeutis cantonensis (a planor-
bid). Ten sites were visited, with most species collected at more than one
site. They dissected stomachs from approximately 20 individuals for each
species per site, and estimated percentage volume from microscopic
examination. Intersite differences were striking. Dudgeon and Yipp
compiled a series of tables comparing B. straminea gut contents over all
pairs of the five sites it was collected, L. olivula over all pairs of the seven
sites it was collected, and so forth, noting significant intraspecific differ-
ences in almost all comparisons. The only gut content data published
were pooled over all sites, however, rendering interspecific comparison
difficult.
By volume, detritus comprised the majority of the average gut content
in all nine Hong Kong species (Figure 9.1, upper). The most striking
contrast is the high proportion of green algae in the stomachs of P. acuta,
B. straminea, and L. ollula. This may certainly be due at least partly to food
availablity at different sites. These three species were found together in at
least four of five to seven sites, so the difference between their gut con-
tents and those of the other six species may be an artifact. Fortunately,
three species (R. plicatulus, Gyraulus, and Bithynia) were found only at a
single site, so their comparison is a fairer one. While Bithynia and
Gyraulus contained 95–100% detritus, Radix guts contained a strikingly
large fraction of macrophyte tissue. So at least some of the variation seen
in Figure 9.1 seems to reflect dietary differentiation in the snails.
Reavell (1980) surveyed the gut contents of 20 British species of fresh-
water gastropods collected from over 20 sites. He estimated percentage
total squashed area for each component of the diet in a fashion similar to
that of Dudgeon and Yipp, but presented data on the gut contents of
snails collected at individual sites. The Shropshire Union Canal is
reported to have both high nutrient levels and high hardness, while the
upland Tannyrallt stream has low levels of both. Nevertheless the main
organic fraction of gastropod gut contents at both sites was detritus, as in
Hong Kong (Figure 9.1, lower). Shropshire species as a whole seemed to
372 · Communities
Figure 9.1. Organic gut contents (percentage volume) for freshwater gastropods
from three areas. Most data from Hong Kong are pooled for several sites (Dudgeon
and Yipp 1985). Data from the hardwater canal and softwater upland stream in
Britain are from Reavell (1980). Data with known month of collection are super-
scripted: 1April, 2August. Results are given for diatoms (Diat.), green algae (both
unicellular and filaments combined, Gre.), macrophytes (Mac.), and Detritus (Detr.).
Table 9.1. Abundance of the most common diatoms found on substrate and in
guts of adult Goniobasis proxima, adult Physa, and juvenile Leptoxis in two
samples of the Mitchell River, NC
Riffle Pool
Substrate Goniobasis Leptoxis Physa Substrate Goniobasis
Achnanthes deflexa 1090 1117 1090 1114 853 900
A. microcephala 42 41 62 43 0 0
A. minutissima 0 1 0 1 183 141
Cocconeis placentula 35 34 31 37 8 9
Cymbella tumida 21 1 6 7 9 11
Gomphonema parvulum 13 13 16 13 10 11
Navicula cryptocephala 17 24 25 26 78 53
N. decussis 31 30 51 33 53 41
Nitzschia acicularis 16 19 24 15 4 13
Synedra fasciculata 15 21 31 25 66 67
S. rumpens 156 135 95 113 126 90
S. ulna 61 54 52 62 75 106
Feeding strategy
Daldorph and Thomas (1988) compared six British freshwater snails in
their responses to a variety of short-chain carboxylic acids and maltose,
known to be released from decaying plant tissues. All the chemicals
attracted and/or arrested Lymnaea peregra, all but the smallest carboxylic
acid (C2) attracted Planorbis contortus and P. planorbis, and all but the larger
carboxylic acids (C5–C8) attracted Physa fontinalis. Very little response was
noticed in Planorbis vortex or Bithynia tentaculata.
I am intrigued by the available data on diurnal variation in feeding rates
or activity rates (the distinction is a fine one). A heroic study was made
of a Biomphalaria glabrata population inhabiting a St Lucia marsh by
McKillop and Harrison (1982). They followed individual snails using a
square grid made of thread and a flashlight, and observed nine times more
feeding activity at night than during daylight. This same phenomenon
was noted in a laboratory population of B. glabrata by Rotenberg et al.
(1989). A related phenomenon, also demonstrated in both laboratory and
field populations, is a significant tendency for Biomphalaria to float to the
surface during dark hours (Pimentel and White 1959b, Bao 1985). The
phenomenon may be seasonally reversed, however (Pimentel-Souza et al.
1984).
Williams and Gilbertson (1983) verified nocturnal increases in the
378 · Communities
heartbeat, locomotion, and feeding rates of healthy Biomphalaria but
noted that these differences were extinguished in parasitized snails.
Infected snails ate continually, presumably in an effort to keep up with
the increased metabolic demands of a parasite load. In a similar fashion,
McDonald (1973) found two to three times more feeding at night in a
laboratory population of Lymnaea stagnalis at a comfortable 20°C.
Feeding periodicity was extinguished at 30°C, however, as the snails
endeavoured to balance their energy budgets. Health, food, or tempera-
ture stress may explain Veldhuijzen’s (1974) failure to note any feeding
rhythm in his population of L. stagnalis. We have previously noted what
a fragile thing is behaviour.
In direct contrast, laboratory populations of Bulinus are much more
active and feed at increased rates during daylight hours. This has been
shown for B. africanus by Morgan and Last (1982a) and for B. tropicus by
Chaudhry and Morgan (1983). To my knowledge, the only other fresh-
water snail that has been well studied in this regard is Melanoides tubercu-
lata (Beeston and Morgan 1979a,b; Morgan and Last 1982b). This snail
generally spends daylight hours buried and feeds at night, but has two
peaks of activity, one at dawn and the other at dusk, the latter perhaps
more associated with return to hiding. It would be fascinating to put
Biomphalaria, Bulinus, and Melanoides together and observe if they con-
tinue to display three such different cycles of activity when under iden-
tical conditions.
Depths (m)
0.25 0.36a 3.0 6.0
Total
Mean particle (mm) 0.17 0.18 0.15 0.29 (all 8 depths)
Musculium securis 2 0 9 4 39
M. transversum 50 24 0 0 80
Pisidium casertanum 106 165 16 0 437
P. dubium 2 1 0 0 8
P. ferrugineum 0 2 1 0 8
P. henslowanum 0 17 0 0 37
P. lilljeborgi 14 54 4 4 230
P. nitidum 26 85 2 3 291
P. punctatum 15 38 1 1 75
P. variabile 12 9 0 0 44
P. walkeri 0 1 0 0 1
Sphaerium striatinum 0 0 26 2 64
Notes:
a
Only four grab samples. Multiply N by 1.25 to compare with other depths.
Source: Kilgour and Mackie (1988).
Table 9.3. The 14 pisidiid species collected in quantitative samples from Lake
Zurich, categorized by depth of occurrence
Note:
A few Valvata were found at depths below 5.0 m.
Source: Økland (1964).
Substrate
Our discussion of gastropod habitat has been facilitated in no small
measure by the happy chance that the University of Michigan Biological
Station is situated on Douglas Lake, inhabited by healthy populations of
a number of snail species. Clampitt (1973) quantitatively sampled four
transects from shore to 30–150 m offshore, some transects uniformly
shallow throughout, others falling to depths of 10 m, carefully noting
substrate. Example results from transects ‘a’ and ‘b’ are shown in Figures
9.3 and 9.4, for two different sampling dates. Figure 9.3 shows that in the
summer of 1969, Physa integra seemed to show a slight preference for
algae-covered cobbles near shore, while Helisoma antrosa strongly pre-
ferred the detritus-covered sand found 10–20 m offshore. All of these
samples were taken at a depth of about 1 m. The preference of P. integra
for cobbles was much more pronounced upon resampling in May 1970.
That the distribution of H. antrosa in transect ‘a’ was not a function of
distance to shore can be seen in transect ‘b’, where in 1969 the H. antrosa
population clearly preferred sandy substrate close to shore.
The instability of freshwater snail distributions is an important second
message from these data. Although it is not clear why H. antrosa seems to
have disappeared in 1970, both sets of observations at transect ‘a’ are in
accord with Clampitt’s (1972) results on seasonal migration in these two
species previously discussed. But the contrast between the July 1969 and
July 1970 data collected at transect ‘b’ is striking. Further, the collections
of Laman et al. (1984) shown in Figure 9.2 were made very near
Clampitt’s transect ‘b’, in June through August 12 years later. Laman et
al. noted many discrepancies between his data and Clampitt’s, but could
offer no explanation other than ‘populations of snails in Douglas Lake are
not stable over long periods’. The lesson to be learned from this is that
surveys of freshwater snail distribution should never be repeated.
Interspecific ecological overlap · 387
Figure 9.3. Clampitt’s (1973) transect ‘a’. Density of Physa integra and Helisoma
antrosa (N/m2) as a function of distance to shore, depth, and substrate on two dates
in South Fishtail Bay, Douglas Lake, Michigan.
Figure 9.4. Results of Clampitt’s (1973) transect ‘b’. Density of Physa integra and
Helisoma antrosa (N/m2) as a function of distance to shore, depth, and substrate on
two dates in North Fishtail Bay, Douglas Lake, Michigan.
Current
Most of the studies discussed in the previous two sections have taken
place in ponds and lakes. But in lotic environments, although it is diffi-
cult to separate the effects of depth, substrate, and current, I would
suggest that current velocity may be of primary importance. In my expe-
rience, few pulmonate species (in particular) can be said to inhabit lotic
environments at all; their occurrence in rivers being restricted to calm
backwaters and eddys. I am unaware of any studies on the effect of
current as rigorous as many of the studies on substrate we have discussed:
for example, combining field observation with laboratory experiments.
But the phenomenon may be so striking and so obvious as to discourage
study. One rarely sees study on the importance of water to fish.
There have been many good laboratory investigations showing that
Models of species distribution · 391
most pulmonates can be dislodged by fairly low current speeds, typically
0.6–0.9 m/sec, depending on snail size and substrate (Jobin and Ippen
1964, Moore 1964, Dussart 1987). But the measurement of actual micro-
current speeds in situ is difficult, especially on the irregular surfaces where
snails reside. Thus most authors who have considered current speed crit-
ical in the distribution of freshwater snail populations have relied on
rather indirect evidence (e.g. Appleton 1975, 1976, Thomas and Tait
1984). Pimentel and White (1959a) measured current velocity at two-
thirds depth in midstream for over 1000 square yard (0.91 m2) samples
from a small tributary of the Quebrada Sabana Llana, Puerto Rico. They
also measured shade, turbidity, depth, and substrate, relating each of these
factors to the density of Biomphalaria glabrata. Pimentel and White sug-
gested that stream velocity and depth were among the more important
determinants of Biomphalaria distribution, and that bottom material was
secondary.
As we have seen in Chapter 5, striking differences are often noted
between the life histories of snail populations in the field and laboratory.
Loreau and Baluku (1987b) found a much lower intrinsic rate of increase
in the population of Biomphalaria pfeifferi of the Virunga Stream at Lwiro,
Zaire, than had been observed in laboratory populations previously. They
offered good evidence that neither temperature, water hardness, nor food
limitation could be responsible for the lower survivorship and fecundity
seen in the stream, and considered current velocity (as affected by rain-
fall) likely to be the critical determinant of population dynamics.
To be complete, it should be noted that preference for current may
vary seasonally, just as depth preference. Rowan (1966) noticed autumn
migration in a population of Helisoma trivolvis inhabiting a small pond in
Montana. The snails did not move to the deeper regions of the pond, but
rather to the pond inlet, where increased current would provide addi-
tional protection from freezing.
coefficient due to Mountford (1962) between all pairs of sites, and then
used the symmetric matrix of site similarites in a cluster analysis.
Both Storey and Calow found that the presence of gastropod species
can be used to classify freshwater sites into distinct clusters well correlated
with major environmental features. Storey could distinguish four zones
in the River Hull drainage. Zone ‘A’ sites were unstable and subject to
drying, and were characterized by Lymnaea peregra, while ‘B’ sites were
small but permanent, being characterized by Potamopyrgus jenkinsi and
Ancylus fluviatilis. High-diversity zone ‘C’incorporated sites in the middle
reaches of the River Hull, inhabited by these three species plus as many
as 11 others. No snails were found at zone ‘D’ sites, admittedly hetero-
geneous environmentally. Calow’s analysis of Malham Tarn resulted in
two very distinct clusters – one of weed beds and the other of rocky
shores, as may be seen in Figure 9.5.
The studies in French were also similar to each other methodologi-
cally. Leveque and Gaborit (1972) found seven mollusc taxa in their
quantitative samples of Lake Chad: the viviparid Bellamya, the thiarids
Melania and Cleopatra, the unionid Caelatura, and corbiculoids
Byssanodonta, Corbicula, and Pisidium. Rather than cluster analysis, the
authors used factor analysis to characterize their approximately 100 sites
by the abundances of these seven taxa. They noted striking regional
differences in the mollusc fauna, and recognized clusters of sites that
appeared related to substrate. Mouthon (1979) identified 28 species of
gastropods and 12 bivalve species at his 17 sites on the River Aube. He
used the abundances of these species as variables in a factor analysis,
plotted his sites on the first two factors, and recognized five major groups
of sites.
394 · Communities
Such studies as those of Storey, Calow, Leveque and Gaborit, and
Mouthon are certainly informative regarding environmental conditions.
But from the standpoint of freshwater mollusc ecology, there may be
more interest in associations of species than associations of sample sites.
Since distinct regions of rivers or lakes seem to be inhabited by fairly dis-
tinct mollusc faunas in these studies, I would expect the results of a
reanalysis of molluscan associations to be similar (but certainly not iden-
tical) to the original site analyses of the authors. So I have reanalysed
Mouthon’s data, using sites as the variables and species as the objects to
be clustered. Further, I would prefer not to perform principal compo-
nent analysis, factor analysis, or indeed any sort of technique that involves
the calculation of a variance/covariance matrix from data sets where most
observations take the value 0.0. So I set aside the five species Mouthon
found only at his site 13 and converted the abundance data on his remain-
ing 35 species to the presence/absence matrix shown in Figure 9.6.
Choice of ordination technique has always been a matter of taste.
Jackson and colleagues (1989) have reviewed the properties of eight sim-
ilarity coefficients that may be applied to presence/absence data, the
coefficient of Mountford used by Calow and Storey not among them.
(Mountford’s similarity coefficient can range to infinity, whereas all the
coefficients conventionally used for ordination range from 0.0 to 1.0 or
⫺1.0 to 1.0). I prefer the simplest coefficient in the list of Jackson and
his colleagues, the Jaccard coefficient, calculated as the number of species
shared by two sites divided by the total species in the two sites considered
together. ‘Negative matches’, sites where neither species is present, are
not considered in the calculation of the Jaccard coefficient. This might
seem undesirable at first. We are assuming no barriers to dispersal in these
communities, so the observation that two species are both absent at site
A would seem to contain as much information as their mutual presence
at site B. But while presence can be reported with certainty, absence has
a large error term. Negative matches seem much more likely to reflect
sampling error than joint occurrences. So from the River Aube data I cal-
culated a 35⫻35 symmetric matrix of Jaccard matching coefficients
using the CORR module of (Wilkinson 1988).
Although cluster analysis as applied by Calow and Storey is certainly a
useful way to examine data for groups, the conversion of a complex
matrix of similarities to a linear diagram necessarily involves considerable
distortion. A representation in more dimensions is obviously preferred.
Again, many techniques have been employed for this purpose over many
years, but the review of Kenkel and Orloci (1986) persuades me that
Models of species distribution · 395
Notes:
Ni, number of individuals collected; Ns, number of sites where each population occurred; abs, a species found in qualitative samples
only. Synonyms in parentheses.
Source: data of Ellis (1931a,b), Ahlstedt (1986), Havlik and Marking (1980) and van der Schalie (1948).
402 · Communities
Figure 9.8. Abundance of selected unionid mussels in regional zones of the upper
Mississippi River (Ellis 1931a,b; van der Schalie and van der Schalie 1950).
Models of species distribution · 403
and others, it is clear from Figure 9.7 that the majority of the mussel
species of the upper Mississippi River tend to vary independently, scat-
tered in clumps about the river in what appears to us to be a random
fashion. To illustrate this point I have also plotted in Figure 9.8 the abun-
dances of two outliers from Figure 9.7 – Anodonata imbecillis (6) and
Leptodea fragilis (36). The former fluctuates narrowly, from 0 to 10,
through the first 13 sites, then increases tenfold at site 14. The latter fluc-
tuates widely through all 14 sites (dramatic even on a log scale), with no
trend apparent.
Freshwater mollusc populations may aggregate together, they may
aggregate independently, and they may also aggregate apart. We now turn
to a remarkable data set collected by F. C. Baker (1918) from the Lower
South Bay of Oneida Lake, New York (Dillon 1981). Baker made 162
quantitative benthic samples using a dredge, a long-handled dipper, or
direct inspection of measured stones, recording depth, substrate, and dis-
tance to the shore and counting all macroinvertebrates and macroalgae.
A total of 21 of his samples either contained no snails or were omitted
from the report. Baker recognized 37 species and subspecies in the
remaining 141 samples, although most workers today would recognize
somewhat fewer. In their reanalysis of this fauna 50 years later, Harman
and Fourney (1970) revised the number of snail species to the 29 I have
listed in Table 9.6. (Omitted are several species Baker recorded from the
lake but did not collect in quantitative samples.) The table shows clear
evidence of the aggregation that is so common in benthic samples
(Caquet 1990). One might hypothesize that each Oneida Lake snail pop-
ulation may be aggregating in a preferred habitat type of some sort, just
as Calow found in Malham Tarn.
Baker’s 141 samples included 23 with just one snail species present, 24
with two species present, and so on, up to a single site with ten different
species of snails. This ‘So’ distribution (the number of species observed)
is shown in Figure 9.9. The question is whether the many separate snail
populations of Oneida Lake generally tend to find the same habitat
favourable, or whether they tend to aggregate in different habitats. Is the
So distribution approximately what one would expect from randomly
casting the 29 species into 141 samples?
I designed a Monte Carlo simulation of Baker’s sampling process,
creating a species pool in which each species was represented according
to its abundance in the lake overall. For example, the probability of select-
ing Campeloma decisum from this pool was 17/5582, since 17 individual
C. decisum were collected among the total of 5582 snails (Table 9.6).
404 · Communities
Table 9.6. Snail populations collected by Baker (1918) from Oneida Lake,
New York
‘Sites expected’ column derived from a Monte Carlo simulation, as described in
the text
Sites expected,
Species Individuals Sites observed mean (s.d.)
1) Campeloma decisum 17 10 13.3 (3.2)
2) Viviparus georgianusa 1 1
3) Valvata bicarinataa 4 3
4) V. sincera 49 6 31.9 (3.5)
5) V. tricarinata 545 36 97.5 (5.0)
6) Amnicola limosa 39 2 24.5 (3.6)
7) A. lustrica 2841 78 133.6 (1.9)
8) Cincinnatia binneyanaa 13 5
9) Gillia altilisa 11 4
10) Somatogyrus subglobosusa 3 3
11) Bithynia tentaculata 220 15 70.4 (5.3)
12) Goniobasis livescens 82 11 42.8 (4.6)
13) Physa integra 210 22 69.5 (5.0)
14) P. sayii 160 38 62.6 (4.1)
15) Lymnaea catascopium 455 72 95.0 (4.4)
16) L. columella 17 2 13.4 (3.1)
17) L. haldemani 6 2 5.1 (2.0)
18) L. humilis 3 1 2.8 (1.4)
19) L. stagnalis 20 1 16.3 (4.3)
20) Gyraulus deflectus 1 1 1.2 (1.0)
21) G. hirsutus 47 18 30.7 (4.3)
22) G. parvus 581 60 99.1 (4.6)
23) Helisoma anceps 44 23 28.8 (4.0)
24) H. campanulata 67 24 37.9 (5.0)
25) H. trivolvis 50 12 29.6 (6.9)
26) Planorbula jenksii 1 1 1.1 (1.0)
27) Promenetes exacuous 82 34 41.8 (4.3)
28) Ferrissia parallela 45 17 29.1 (6.9)
29) Laevapex fuscusa 2 1
Notes:
a
Omitted from morphological analysis.
Models of species distribution · 405
Figure 9.9. The points show the number of snail species observed in Baker’s (1918)
samples from Oneida Lake (the So distribution), and the histrogram shows the
number expected (the Se distribution) with 95% confidence limits. (Analysis of
Dillon 1981.)
Figure 9.10. The abundance of Actinonaias carianta and Actinonaias pectorosa at 141
sites on the Clinch River, Virginia and Tennessee. (Data of Ahlstedt 1986.) Curves
(solid for A. carinata, dashed for A. pectorosa) fit using a polynomial.
200.0 (277 km total – his Table 14). The Elk River data have been pub-
lished separately by Ahlstedt (1983).
Table 9.5 summarizes the distribution and abundances of unionid
mussels in these two rivers. Both mussel faunas showed at least some
zonation along the lengths of stream sampled. For example, the two
species of ‘muckets’ (genus Actinonaias) appeared to be almost equally
common in the Clinch River. These are equally large, heavy, conspicu-
ous mussels, distinguished only by minor details of shell shape and pig-
mentation. There should be no bias to collect one over the other. Figure
9.10 shows, however, that A. pectorosa is much more common upstream
and A. carinata is more common downstream. How general is this phe-
nomenon? It is clear from the raw data that at least some zonation occurs
throughout the unionid communities of the Clinch and Elk, just as we
saw in the data of Storey and Mouthon. Do the most similar mussel
species tend to assort along the stream gradient to minimize competition
present or potential?
I have categorized 128 Clinch River samples simultaneously by the
number of species and the number of genera each contained (6 samples
had no mussels and 7 samples had but one species) in Table 9.7. The
number of samples containing two species, three species, and so forth, is
Table 9.7. Ahlstedt’s (1986) samples of the Clinch River mussel fauna, categorized by the number of species and genera they
contained
Number of genera
Number of
species 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 So
2 7 7
3 1 1 2
4 3 3
5 1 1
6 2 2 4
7 4 3 7
8 2 2 4
9 2 7 2 11
10 1 2 5 4 12
11 1 6 1 3 11
12 4 1 2 7
13 1 1 2 4
14 4 1 3 1 9
15 2 2 1 5
16 1 1 3 1 6
17 1 3 2 6
18 2 1 1 4
19 1 4 5
20 1 1
21 1 2 3
22 1 1 1 3
23 2 2
24 1 1 2
25 1 1 2
26 0
27 1 1
28 1 1
29 1 1
30 1 1
31 1 1
32 0
33 1 1
34 1 1
To 0 8 1 4 2 7 10 19 13 11 5 13 7 5 7 3 4 1 2 1 2 1 2
412 · Communities
totalled along the right margin of the table – this is the So distribution
we encountered previously. In a similar fashion, summing the columns
of this table to the bottom margin results in the To distribution, the
higher taxa (genera in this case) observed. Is this To distribution approx-
imately what one would expect given 128 samples of this species number
from this fauna?
If there is no tendency for co-occurring mussels to be more or less
similar to one another taxonomically, a random sample of species from
the Clinch River fauna using the So distribution should give a distribu-
tion of genera (Te, higher taxa expected) indistinguishable from To. But
if co-occurring mussels tend to be taxonomically dissimilar, for example,
the To distribution will tend to be higher than the randomly generated
Te distribution.
I used a Monte Carlo simulation to generate Te distributions over
12800 samples. Two simulations were performed, under different
assumptions. In the ‘weighted’ simulation, the probability of selecting
each species was proportional to the number of sites at which it was col-
lected. For example, Table 9.5 shows that in the Clinch River, Actinonaias
carinata was found at 104 sites, A. pectorosa at 101 sites, Alasmidonta mar-
ginata at 37 sites, and so on. The total occurrences of all mussels at all sites
were 104⫹101⫹37⫹. . .⫽1694. So the probability of selecting A.
carinata in the weighted simulation was 104/1694⫽0.061. (Contrast this
with the simulation used to generate the Se distribution for Oneida Lake
(Figure 9.9), which was weighted by the number of individuals collected,
not the number of sites.) But the degree to which each species is wide-
spread may be a function of the community processes that this analysis is
trying to uncover. So a second ‘unweighted’ simulation was performed,
where the probability of selecting each of the 43 species was equal. These
two null hypotheses are analogous to those commonly derived from sim-
ulations used to test island biotas (e.g. Simberloff 1978).
I repeated this entire analysis using Ahlstedt’s data from the Elk River.
Table 9.5 shows many contrasts between the Clinch and the Elk faunas.
Collections from the Elk included 38 species, only 26 of which appear
among the 43 species listed from the Clinch. Abundances were lower in
the Elk. The largest number of different species at any site on the Elk was
19, compared with the 34-species samples of the Clinch (Table 9.7).
Using one-sample Kolmogorov–Smirnov tests, I found no differences
between the observed distributions of genera in the Clinch River and
either weighted or unweighted simulation. From the weighted simula-
tion, for example, one would expect 0.3 samples with one genus, 6.9
samples with two genera, 2.4 samples with three genera, etc. The fit to
Models of species similarity · 413
To (Table 9.7) was very good. Nor was any difference apparent between
the Elk To and either weighted or unweighted Te. So the mussel species
do not generally seem to sort by genera along stream gradients. It should
be noted that this mussel fauna includes a large proportion of monospe-
cific genera, rendering taxonomic tests such as mine somewhat insensi-
tive. But as far as can be told, Ahlstedt’s mussel samples seem to be
taxonomically random.
It is certainly possible, however, that any competition between mussel
species may be for fish hosts for their glochidia. Although most mussels
seem to be able to use a variety of fish hosts, as we saw in Chapter 2,
some species seem to show considerable host specificity. But since the fish
hosts are known for only about half of the 43 Clinch River species, any
prospecting for host competition might more profitably be directed
toward other communities of unionids. Barr and colleagues (1986)
reported another activity of the Cumberlandian Mollusc Conservation
Program, a survey of macrofauna at 15 sites in nine rivers potentially har-
bouring endangered mussel species. They used a variety of techniques
(snorkelling, seining, electrofishing) to collect both qualitative and quan-
titative samples of the fish fauna at each site during the three-month
period glochidia are believed released. They also took quantitative
samples of the mussel faunas to supplement those of Ahlstedt. By good
fortune, the fish hosts have been described for 6 of the 8 mussel species
(74% of the individuals) collected by Barr and colleagues in quantitative
samples from a site (river mile 27.8) on the Nolichucky River, Tennessee.
What sort of host overlap is apparent, given the relative abundances of
both mussels and fish available? Barr obtained 15 fish species in quanti-
tative samples at the Nolichucky River site of interest to us, and recorded
an additional 8 rarer species in qualitative samples only. Not a single one
of the 15 most common fish species has ever been implicated as a host
for any of the 6 mussel species. Watters (1994) lists 13 known hosts for
Actinonaias ligamintina (⫽carinata), 12 fish not collected by Barr and the
smallmouth bass, found in qualitative samples only. None of the 13
known fish hosts for Amblema plicata (⫽costata) seems to be found in this
section of the Nolichucky. The problem is doubtless due in no small part
to the diversity of fish. Simply for reasons of availability, researchers have
generally experimented with larger, more common fish such as catfish,
bass, sunfish, and crappie. Until some brave soul selects a place like the
Nolichucky and quantitatively tests all 8⫻23 combinations, the question
of host competition in unionid communities will remain open.
There is a another method of measuring the similarity of co-occur-
ring species, simpler and more practical than trying to estimate resource
414 · Communities
use and more direct than taxonomic relatedness. One could take direct
morphological measurements. This practice is as old as the method of
congeneric pairs, originating with Huxley (1942) and Lack (1947), and
being brought to the fore in a seminal paper by Hutchinson (1959).
Hutchinson noted that congeneric species of mammals and birds tended
to be of unexpectedly diverse sizes when occurring sympatrically, and
went so far as to suggest that the ratio of the larger to the smaller species
tends to be about 1.3. He considered this effect an adaptation to mini-
mize competition. Although some attention has continued to centre on
body size ratios, many workers (e.g. Ricklefs et al. 1981, Travis and
Ricklefs 1983) have measured a greater variety of morphological attrib-
utes, especially including limb sizes and aspects of the trophic apparatus.
The assumptions are the same as those that are made for taxonomic anal-
ysis, and the question remains the same. Do co-occurring animals tend
to be morphologically similar, morphologically different, or random in
this regard?
Let us return to Baker’s (1918) data on the snails of Oneida Lake to
address this question. Of the dozens of morphological attributes one
could measure in an effort to characterize the ecological requirements of
a snail, aspects of the radula would seem to be among the most impor-
tant. (Some caveats may be recalled from Chapter 3, however.) It also
seems reasonable to expect that shell morphology might give clues about
a snail’s ecological requirements. Diet studies lead us to expect that the
ten-fold size range seen in the Oneida Lake gastropod fauna may have
ecological importance. I was able to obtain living or preserved specimens
of the 23 species not marked with a footnote superscript in Table 9.6.
(The 6 species excluded from this analysis accounted for only 35 individ-
uals in total.) For each species I measured shell length, body whorl length,
shell width, aperture length, aperture width, spire angle, radula length,
and radula width, and counted radula rows, teeth per row, and cusps per
30 teeth. Further details regarding measurement techniques, as well as
means for all 11 morphological variables measured over all 23 species, are
available in Dillon (1981).
The data set included variables that were skewed, measured in all
manner of units with all manner of variances, and highly intercorrelated.
So I performed a principal component analysis on the between-species
correlation matrix of log-transformed character means. I accepted the six
principal components that together accounted for the first 95% of the
variance. I then calculated two statistics to measure overall morphologi-
cal dissimilarity within each of Baker’s 111 samples containing two to
seven species present. (Samples of eight, nine, and ten species were
Models of species similarity · 415
冤冢兺 冥
1
S ⁄2
R⫽ f 2
i
⫺1/S )/(1⫺1/S )
i⫽1
and provided a ‘lookup table’ relating any value of R from S⫽1 to 50 to
a ‘score’. With some minor reservations, a score of 20 is equivalent to a
significance level of 1/20⫽0.05.
We have discussed thus far three large, quantitative data sets on snail
abundance, Berg’s on Lake Esrom (Chapter 7), Økland’s on Lake
Borrevann (Table 9.4), and Baker’s on Oneida Lake (Table 9.6). We have
also examined a fair number of bivalve data sets in this chapter: those of
Kilgour and Mackie (1988) on the abundance of pisidiid species in the
Ottawa River (Table 9.2), Walter and Kuiper (1978) on pisidiids in Lake
Zurich (Table 9.3), Ellis on the unionids of the upper Mississippi, and
422 · Communities
Species Individuals
Alasmidonta arcula 13
Anodonta couperiana 4
A. gibbosa 10
Canthyria spinosa 88
Elliptio dariensis 6
E. hopetonensis 237
E. shepardianus 8
Lampsilis dolabraeformis 495
L. splendida 15
Figure 9.13. Number of unionid mussel species in log2 categories (octaves) of species
abundance. Category ‘1’ includes species with from 0 to 21 ⫽2 individuals; category
‘2’ are species with from 2 to 22 ⫽4 individuals, etc. (Data are from Sickel 1980, van
der Schalie 1948, Ellis 1931a,b, Havlik and Marking 1980, and Ahlstedt 1986.)
424 · Communities
Anodonta grandis (Table 9.5). Then Figure 9.13 shows 1.5 species in inci-
dence category 1: the Villosa and half the Anodonta.
Table 9.10 shows values of R and ‘score’ from bull’s eye tests of
Hopf and Brown. (The tiny discrepancy between the value of R I have
calculated from Baker’s data here and that calculated by Hopf and Brown
from the same data comes from my use of the entire So ⫽29 data set, not
So ⫽23.) The data of Ellis, van der Schalie, Kilgour and Mackie, and
Økland all had scores less than 20, suggesting that the relative abundances
of molluscs in these four communities might be modelled as random pro-
cesses. The figures show that in spite of the log transformation, all four
of these distributions have long tails to the left and high modal abun-
dances. We now turn our attention to the six data sets that do not seem
as likely to be random: Baker, Berg, Walter, Sickel, Havlik and Ahlstedt.
Two mathematical distributions developed to describe species abun-
dances have attained prominence because some ecological meaning has
been accorded them – the lognormal and the log-series. Although there
is some controversy, the lognormal is generally viewed as fitting abun-
dances in large, stable, equilibrium communities, while species abun-
Models of species abundance · 425
Figure 9.15. Number of snail species in octaves of species abundance (data of Baker
1918) Berg 1938 and Økland 1964.) Abscissa scaled as in Figure 9.13.
Table 9.10. Species–abundance statistics for the data sets shown in Figures
9.13 to 9.15
So R Score Se o e
Oneida snails (Baker) 29 0.512 20 29 2.95 3.19
Esrom snails (Berg) 12 0.501 20 12 2.19 2.85
Zurich pisidiids (Walter) 14 0.477 20 14 3.91 2.91
Altamaha mussels (Sickel) 9 0.573 20 10 2.82 2.73
Mississippi mussels (Havlik) 36 0.329 20 42 3.50 3.26
Mississippi mussels (Ellis) 38 0.184 19 – – –
Ottawa pisidiids (Kilgour) 12 0.361 19 – – –
Grand mussels (van der Schalie) 17 0.275 17 – – –
Borrevann snails (Økland) 9 0.323 13 – – –
Clinch mussels (Ahlstedt) 34 0.339 20 – – –
Notes:
So, number of species observed; R and ‘score’, bull’s eye statistics of Hopf and
Brown (1986); Se, number of species expected given a log normal model; o,
observed variance; e, variance expected for canonical log normal model fit to
observed.
consider that more than half of the species in a community are missing
from a sample hypothesized to fit the distribution, such that the abun-
dance of no species can be considered modal. Hughes (1986) argued con-
vincingly, however, that it is unwise to even attempt to fit a lognormal to
any distribution that does not reveal a mode.
Hughes collected 222 data sets on species abundances (no freshwater
molluscs) and compared them for their fit with each of these three
models. He plotted number of species against log2 number of individu-
als and examined for a mode. Hughes found modes in 16 of his 31 data
sets for vertebrates, but only 24% of his 182 invertebrate data sets. Over
all 222 sets, only 28% could thus be considered lognormal. To distinguish
between data fitting log-series and dynamics models, Hughes examined
the upper part of their dominance-diversity curves to see if they were
deeply concave. He found that the log-series distribution proved a better
prediction than the dynamics model in only 4% of the samples. Thus his
dynamics model, based from the start on a simulation of the behaviour
of natural populations, emerged as the clear victor over the two older
models based on theory.
In apparent contrast to the majority of invertebrate communities, five
of the six non-random sets of abundance data gathered in Figures 9.13
to 9.15 are strongly modal. The lognormal model would seem to fit these
Models of species abundance · 427
freshwater mollusc abundances rather well. The exception is Ahlstedt’s
Clinch River mussel community, which seems to have relative abun-
dances fitting a log-series. (The dynamics model would predict more
very abundant species, and thus a longer tail on the right of the Clinch
River distribution.) A traditional interpretation is that such a log-series
may in fact be a lognormal with a mode at the veil line, and that the mode
would be revealed if more individuals were collected. This may be the
case – Ahlstedt’s samples come from a much larger fauna than Sickel’s but
were much less extensive than those of Havlik and Marking.
I hesitate to fit a lognormal distribution to Ahlstedt’s data or to the
abundance data that appear to be random. But Table 9.10 shows variance
fits to the other five data sets. Notice in Figure 9.15 that more snail
species occur to the left of the modes than to the right. The same obser-
vation could be made for Walter and Kuiper’s pisidiid data (Figure 9.14).
If we assume that these distributions are in fact lognormal, this implies
that all species have been collected. So the observed variances given in
Table 9.10 for the Baker, Berg, and Walter data were calculated directly
from the log2 abundances. In contrast, some of the species seem to be
obscured behind a ‘veil line’ in the collections of Havlik and Sickel. I
assumed the true mean of these two distributions to be the average of the
(transformed) modal abundances, and thus inferred that the true number
of mussel species in the Altamaha would be 10 and in Havlik and
Marking’s Mississippi sample 42. I then plotted the known portions of
these two distributions on probability paper and estimated the variances
graphically (Sokal and Rohlf 1995:116).
Preston (1962) pointed out that the variance of lognormal abundance
distributions seems to be ‘canonically’ related to the number of species in
communities. The standard deviation of a canonical lognormal distribu-
tion, e, is related to S by the following relationship (Sugihara 1980):
冤 冥
(e ln 2)2
S⫽e兹/2 exp
2
May (1975) thought that this canonical relationship might be a trivial
mathematical property of all lognormal distributions for large S, but it
has since been shown fairly convincingly that natural communities seem
to fit the canonical relationship more closely than can be explained by
mathematical generalities (Sugihara 1980, May 1986).
The last column of Table 9.10 shows the standard deviation expected
from a canonical lognormal distribution for S the same as observed by
Baker, Berg, Walter and Kuiper, Sickel, and Havlik. In general, the fit
428 · Communities
between observed and canonically expected standard deviation is very
good. Judging by the rather broad error bars associated with such esti-
mates of variance (Sugihara 1980, fig. 7), I am sure that four of the five
data sets do not differ significantly from canonical. The pisidiid data set
of Walter and Kuiper seems to show rather more variance than would be
predicted from a canonical lognormal.
To explain the canonical lognormal distribution of species abundances,
Sugihara proposed the ‘sequential breakage model’. This differs from the
‘broken stick’ null hypothesis that we have rejected for these data in an
important respect. Rather than breaking up the community resources
(the ‘stick’) randomly, Sugihara suggests that the first species takes a
random portion, and then the second species takes a random portion of
the remainder, and then the third species takes a random portion of the
resources not taken by the first two, and so on.
I am prepared to accept this model as a generally applicable to fresh-
water mollusc communities. Two of the three snail communities seem to
fit the sequential breakage model, and two quite different mussel com-
munites as well. I consider it likely that Ahlstedt’s abundance data would
have been found to be canonically lognormal if more samples were taken.
The pisidiid community of Lake Zurich is anomalous. Abundances
appear to be more evenly distributed than expectation, much as
MacArthur’s original ‘broken stick’. This requires further study, but I
have no idea what sort.
It is certainly possible that species abundances in the other two mussel
communities, Økland’s snails, and Kilgour and Mackie’s Ottawa pisidiids
are genuinely random. But in light of the fact that Havlik’s clear lognor-
mal distribution was obtained from a sample taken at one of the 14 zones
Ellis collected, it also seems possible that the Ellis data, at least, is an arti-
ficial admixture of several distinct communities. Disturbance may have
played a role. The mussels of the Grand River were, like those of the
Mississippi, exploited commercially, and fishermen remove certain heavy,
pearly species at disproportionate freqencies. The northerly populations
of snails and pisidiids may have been founded only recently, and the com-
munities not in equilibrium. Thus we are brought back to the subject of
variation in species composition between communities, first discussed in
Chapter 8.
Table 9.11. Lakes and ponds (total 81) from Connecticut, categorized
simultaneously by the number of snail species and number of snail ‘families’ they
contained ( planorbids divided – see text)
Number of Families
Number
of species 2 3 4 5 6 7 8 S
2 10 10
3 2 3 5
4 1 8 9
5 1 3 6 10
6 1 4 4 3 12
7 1 2 6 9
8 2 1 3 6
9 1 5 3 9
10 4 1 5
11 1 1 2
12 1 1
13 1 1 2
16 1 1
To 12 6 16 15 20 11 1 81
Figure 9.16. The To distribution from Jokinen’s (1983) Connecticut snail data
(Table 9.11) compared with Te from Monte Carlo simulation.
Summary
The distinction between diet and habitat is a fine one for snails (and to
some extent, for the bivalves too). Researchers generally report broad
similarities in the gut contents of diverse gastropod species, detritus
almost always the primary component by volume. Yet specific differ-
ences, especially regarding the proportions of macrophytes and filamen-
tous algae in the diet, are occasionally striking. This seems to arise not
from selective retention of foodstuffs, but from differing habitat choice
and differing size and strength of various gastropod jaws. Some snails are
large and have strong mouth parts, some are not, and some macrophytes
are tough while others are tender. But if a snail crawling over a macro-
phyte can rip up some plant tissue, the snail will not stop to consider
whether this constitutes ‘high quality food’. The snail will eat it. Thus
little snails, such as the ancylids and smaller planorbids, do not eat aquatic
vegetation, and big snails, such as the ampullariids, do.
432 · Communities
Most diatoms fit in typically sized snail mouths, bigger diatoms often
probably not as well as small diatoms. But some diatoms are stuck so
tightly to rocks that they cannot be prised loose, and typical snails find
that entire rocks do not fit in their mouths at all, and pass on. As much
as I would like to, I cannot find any evidence that the actual morphol-
ogy of the radula matters. The overall size of the trophic apparatus does,
certainly. But if two snails have radular ribbons that are in the same order
of magnitude, whether they have 7 teeth or 100 teeth per row seems to
be immaterial. Dietary overlap is a likely consequence.
Substrate preference has been demonstrated in a wide variety of snails,
both in the field and in the laboratory. Populations have been found pre-
ferring vegetation, cobble bottom, or sand. Some snails manifest prefer-
ences for macrophytes of particular species or life forms, in response to
the varying architectures of the leaves, the palatabilities of the plants
themselves, or the palatibility of their epiphyton. Such relationships are
very dynamic in the wild. Other snail/macrophyte associations may result
as much from the protection such cover affords. The distribution of snails
in lotic environments seems to be strongly affected by current. Again,
snail populations differ at least somewhat in the current speed they can
tolerate. In lakes, however, almost all mollusc populations reach
maximum densities in the shallows. There are just a few systematic
exceptions; Valvata among the gastropods and Pisidium conventus among
the clams. Evidence suggests that it is the substrate/current variable that
seems to render the shallows more attractive to most species, rather than
dissolved oxygen, distance to shore, or depth per se. Seasonal depth migra-
tion has occasionally been documented, and snail distributions in general
seem quite volatile. But again, the overall impression is one of general
habitat overlap, with particular exceptions.
Just like most other organisms, freshwater molluscs have aggregated
distributions. If an ecologist siezes a sampler and rushes down to the
nearby river or lake with intent to sample a heterogeneous environment
or an environmental gradient, mollusc populations will appear clumped
together. I think this is because, with some exceptions of course, fresh-
water molluscs have broadly similar ecological requirements, at least on
the scale a human might notice, and they perceive their environment
more discretely than we do. But if the ecologist declares, instead, an
intention to sample in an environment that seems homogeneous, he or
she will find the species of molluscs clumped without much regard to
each other, or (possibly even significantly) clumped apart from each
other. Again, the molluscs perceive their environment in what appears to
Summary · 433
us to be a discrete fashion, so what a human would judge to be grossly
uniform will constitute a patchwork to the mollusc community.
In at least one unusually large and diverse snail community of which
we have record, the snail species that occur together tend to be unexpect-
edly diverse morphologically, no matter how this is measured. The phe-
nomenon seems to be due to the dominance of three very different taxa,
a fairly large lymnaeid, a small planorbid, and a small prosobranch, and
to their tendency to occur together. I do feel that this is fairly convinc-
ing evidence of community structure, most likely to minimize competi-
tion. But no such evidence of competition is apparent in the smaller snail
faunas of two Scandinavian lakes, nor in lists of snail species when com-
pared among Connecticut lakes, nor in the large unionid faunas of the
North American interior.
A survey of 10 freshwater mollusc communities (three of snails, two
of pisidiids, and five of unionids) suggests that the relative abundances of
the species in most are not random. They tend to be distributed in a
‘canonical lognormal’ fashion, which Sugihara has attributed to ‘minimal
community structure’. With the caveat that random processes may often
combine to leave communities far from equilibrium, ‘minimal com-
munity structure’ has great appeal as a model for freshwater molluscs
generally.
Literature cited