TMP 1 BB2
TMP 1 BB2
TMP 1 BB2
Abstract
Forest development following stand-replacing disturbance influences a variety of eco-
system processes including carbon exchange with the atmosphere. On a series of pon-
derosa pine (Pinius ponderosa var. Laws.) stands ranging from 9 to > 300 years in central
Oregon, USA, we used biological measurements to estimate carbon storage in vegetation
and soil pools, net primary productivity (NPP) and net ecosystem productivity (NEP) to
examine variation with stand age. Measurements were made on plots representing four
age classes with three replications: initiation (I, 9±23 years), young (Y, 56±89 years),
mature (M, 95±106 years), and old (O, 190±316 years) stands typical of the forest type in
the region. Net ecosystem productivity was lowest in the I stands (2124 g C m22 yr21),
moderate in Y stands (118 g C m22 yr21), highest in M stands (170 g C m22 yr21), and low
in the O stands (35 g C m22 yr21). Net primary productivity followed similar trends, but
did not decline as much in the O stands. The ratio of fine root to foliage carbon was
highest in the I stands, which is likely necessary for establishment in the semiarid
environment, where forests are subject to drought during the growing season
(300±800 mm precipitation per year). Carbon storage in live mass was the highest in the
O stands (mean 17.6 kg C m22). Total ecosystem carbon storage and the fraction of
ecosystem carbon in aboveground wood mass increased rapidly until 150±200 years,
and did not decline in older stands. Forest inventory data on 950 ponderosa pine plots
in Oregon show that the greatest proportion of plots exist in stands 100 years old,
indicating that a majority of stands are approaching maximum carbon storage and net
carbon uptake. Our data suggests that NEP averages 70 g C m22 year21 for ponderosa
pine forests in Oregon. About 85% of the total carbon storage in biomass on the survey
plots exists in stands greater than 100 years, which has implications for managing forests
for carbon sequestration. To investigate variation in carbon storage and fluxes with
disturbance, simulation with process models requires a dynamic parameterization for
biomass allocation that depends on stand age, and should include a representation of
competition between multiple plant functional types for space, water, and nutrients.
Keywords: biomass, carbon cycle, carbon sequestration, forest development, net ecosystem prod-
uctivity, net primary productivity, Oregon, Pinus ponderosa, soil carbon
Revised version received 1 January 2003
woody detritus left on-site after logging may result in a productivity (NEP). All 12 plots were located within a
relatively large amount of heterotrophic respiration for 100-km2 area in the Metolius River basin of central Oregon.
years to decades depending on decomposition rates. Three of the stands are the Metolius AmeriFlux sites, where
Within a forest type, mean annual fluxes of carbon micrometeorological and automated soil chamber meas-
dioxide to the atmosphere are strongly influenced by urements have been made. This is a semiarid region where
the time since stand replacing disturbance (Cohen et al., most of the precipitation occurs between October and June,
1996; Law et al., 2001b), whereas interannual variation in and the winters are cool. Annual rainfall averages
the net carbon uptake (NEE or NEP) is likely driven by 300±800 mm and the 30-years mean is 550 mm, with 1±2%
the effect of year-to-year variation of climate on short- of the total occurring in the summer.
term carbon pools (foliage, soil microbes) through vari- The 12 stands represent four age classes replicated
ation in photosynthesis and respiration by autotrophs three times including initiation (I, stand age 9±20 years),
and heterotrophs (Schimel et al., 2001). Our sapflow and young (Y, 56±89 years), mature (M, 95±106 years), and old
predawn water potential measurements at young and old (O, 190±316 years). Stands were selected to be representa-
ponderosa pine flux sites showed that the young trees tive of their age-class for the region. Stand age was deter-
suffered more drought stress than the mature and old mined from the mean age of the oldest 10% of trees cored
trees, likely because their roots could not access deep soil on each plot. Ponderosa pine dominates the overstorey in
water (Irvine et al., 2002). Drought stress in young trees all 12 plots with varying amounts of grand fir (Abies
appears to slow reestablishment; young pine stands in grandis), western larch (Larix occidentalis), and incense
this region can take 10±20 years to reestablish, and for cedar (Calocedrus decurrens) occurring in some stands.
carbon uptake to exceed carbon losses through respir- The understorey includes manzanita (Arctostaphylos
ation (Law et al., 2001a). Similarly, Amiro (2001) and patula) and bitterbrush (Purshia tridentata), and is more
Litvak et al. (2003) suggested that carbon uptake offset conspicuous in younger stands. The soil is generally
losses between 10 and 20 years following disturbances sandy loam. Historical records indicate that logging
from fire and harvest in boreal forests where unfavour- rather than wildfire had been the source of stand-
able environmental conditions limit growth. replacing disturbance in all but the O stands in our
In this paper, we use biological measurements in a study (USDA Forest Service, Sisters Ranger District).
chronosequence of ponderosa pine (Pinus ponderosa var. The catastrophic fires of 1911 came close to two of the
Laws.) forests in central Oregon, USA, to quantify carbon stands, and one of the old stands had been underburned
storage and annual fluxes associated with stand develop- in 1990. Site characteristics are summarized in Table 1.
mental stage, and explore the implications to the current
age-distribution of ponderosa pine in Oregon. We hy-
Environmental measurements
pothesize that, in ponderosa pine forests, carbon storage
in live mass does not decrease significantly from mature Carbon dioxide, water vapour, and energy fluxes were
to old stands, and that the old stands, on average, are not estimated from micrometeorological measurements
`carbon neutral'. In addition, we hypothesize that hetero- using the eddy covariance technique at old (OS) and
trophic respiration is primarily from soils in this semiarid initiation stage (YS) flux sites in 2000 (Law et al., 2001b;
system, where decomposition of woody detritus is rela- Anthoni et al., 2002). Flux systems were comprised of
tively slow. We also simulate the historical patterns of three-axis sonic anemometers that measured wind
disturbance and recovery at each of the chronosequence speed and virtual temperature (Solent model 1012 R2,
sites with the process model, Biome-BGC, and test the Gill instruments, Lymington, England; CSAT-3 Campbell
need for dynamic parameterization of some variables. Scientific, Inc., Utah), open-path infrared gas analyzers
We explore the variability in disturbance recovery re- that measured concentrations of water vapour and CO2
sponses related to atmospheric CO2 concentration during (LI-7500, LI-COR, Lincoln, NE), and a suite of software
recovery. This paper builds on earlier flux studies and data processing. Fluxes were averaged half-hourly, and
carbon budget estimates at the young and old Metolius the records in the database were evaluated for data qual-
flux sites (Law et al., 2001b). ity. Details on the instrumentation, flux correction
methods and calculations were reported in Anthoni et al.
(2002). The flux data were used to compare weekly esti-
Methods and materials
mates of net ecosystem production (NEP) by the two flux
sites.
Study sites
Half-hourly measurements of climatic variables made
In 12 ponderosa pine stands ranging from 9 to > 300 years, at the top of the flux towers included air temperature
we measured live and dead carbon pool and net primary (Tair), vapour pressure deficit (D), incident photosynthet-
productivity (NPP) to estimate annual net ecosystem ically active radiation (PAR), and rainfall. Soil moisture
Stump density
Plot Altitude Stand density Mean* Mean* (#Pieces ha 1), Species Stand age,
ID Longitude Latitude (m) (#Trees ha 1) DBH(cm) Height(m) mean diameter{ composition class{ History
25§ 121.568 44.437 1165 431 11.3 (0.49, 78) 4.3 (0.16, 78) 204, 41.6 (24.4, 37) Pipo 23, I Clearcut 1978, natural regeneration, thinned
26 121.691 44.451 918 272 8.0 (0.48, 77) 2.0 (0.12, 77) 219, 57.0 (21.9, 62) Pipo, Pico 9, I Clearcut, natural regeneration, no thinning with
the current crop, 0.25 mile from polygon of 1911 fire
27 121.610 44.433 1113 241 9.6 (0.37, 68) 3.8 (0.11, 68) 134, 38.3 (26.6, 38) Pipo 16, I Clearcut, then thinned 1998, natural regeneration
28 121.585 44.428 1129 349 21.0 (1.08, 74) 8.3 (0.37, 74) 94, 23.7 (14.5, 20) Pipo 69, Y Clearcut, natural regeneration, thinned
29 121.587 44.437 1153 237 30.7 (1.16, 67) 14.5 (0.53, 67) 332, 19.5 (12.0, 94) Pipo, Cade 56, Y Clearcut, natural regeneration (high shrub density),
thinned
30{ 121.558 44.451 1232 325 29.0 (1.31, 69) 14.0 (0.59, 69) 268, 22.9 (13.3, 57) Pipo, Cade 89, Y Clearcut, natural regeneration, thinned
31 121.670 44.434 906 581 27.5 (1.26, 73) 18.9 (0.75, 73) 217, 22.1 (21.1, 46) Pipo 106, M Clearcut, shallow water table, thinned, 0.5 mile from
polygon of 1911 fire
32 121.669 44.453 899 1281 15.4 (0.82, 68) 10.2(0.46, 68) 0 Pipo, Abgr 93, M Clearcut
ß 2003 Blackwell Publishing Ltd, Global Change Biology, 9, 510±524
33 121.663 44.463 891 1070 19.0 (1.53, 86) 13.5 (0.93, 86) 87, 28.3 (20.2, 7) Pipo, Laoc, Abgr 96, M Clearcut, natural regeneration
34** 121.625 44.498 895 46 526 73.8 (6.37, 7), 38.9 (1.48, 7), 0 Pipo 190, O Two layer canopy
14.7 (1.12, 80) 13.5 (1.05, 80)
35 121.624 44.501 887 116 55.0 (3.36, 42) 32.6 (1.45, 42) 6, 79.5 (31.8, 2) Pipo 251, O Underburned 1990, single layer canopy
36 121.607 44.422 1129 103 628 56.5 (4.90, 13), 28.9 (1.64, 7), 56, 64.0 (35.1, 7) Pipo, Cade 316, O Two layer canopy
9.0 (0.90, 79) 8.9 (1.05, 79)
Species codes: Abgr, Abies grandis; Cade, Calocedrus decurrens; Laoc, Larix occidentalis; Pico, Pinus contorta; Pipo, Pinus ponderosa; Psme, Pseudotsuga menziesii.
*Values in parentheses are standard errors (SE) and sample size (n), i.e. (SE, n).
{
Diameter at the stump height.
{
The mean age of the oldest 10% of trees in the stand, and age class of stand (I, initiation; Y, young; M, mature; O, old).
§
YS flux site (initiation stand).
{
IS flux site (young stand).
**OS flux site (old-growth stand).
C H A N G E S I N C A R B O N S T O R A G E 513
was measure at 0±30 cm depth (CS615s, Campbell Scien- Foliage productivity was calculated from leaf mass per
tific, Logan, UT) and soil temperature at 15 cm depth needle age class (LMA) and optical estimates of LAI
(Table 2). The continuous meteorological measurements corrected for clumping and wood interception (LAI-
were used in Biome-BGC simulations. 2000, LICOR, Lincoln, NE, USA; methods in Law et al.,
2001b, 2001c). Fine root turnover rate was assumed to be
0.60 yr 1 based on minirhizotron measurements on two
Carbon budgets from biological measurements
plots (Law et al., 2001b) and fine root production was
In 2001, we measured tree and shrub dimensions, age calculated for each plot as fine root mass multiplied by
and growth increment from wood cores, leaf area index fine root turnover.
(LAI), herbaceous plant biomass, coarse and fine woody Carbon storage was determined on live and dead fine
detritus, forest floor fine litter mass, soil bulk density, soil (< 2 mm diameter) and small coarse roots (2±20 mm). On
C and N concentrations, fine litterfall. On four 10 m three plots at the flux sites representing I, Y, and O
radius subplots within each 100 100 m2 plot, structure stands, we collected 18 soil cores (six cores per subplot
measurements were made of tree height with a laser on three subplots), sampling 0±20 cm, 20±50 cm, and
ranging scope (MapStar and Impulse 200, Laser Tech, 50±100 cm depths in late April 2001. The sampling was
Inc., Englewood, CO), and diameter at breast height repeated with nine cores on each of those three plots in
(DBH, 1.37 m) on all trees > 5 cm DBH. The smaller October. Statistical analysis showed no significant differ-
trees were included in the shrub survey. Increment ence in values between the two sampling periods, so we
cores were taken from five trees per subplot in five pooled the measurements from the two periods to obtain
evenly spaced directions from the subplot centre starting plot average carbon storage in fine and small coarse
from the north for a total of 20 cores per plot to determine roots. For the remaining nine plots, we collected six soil
10-year mean growth increment, age, and wood density. cores to 30 cm depth on each plot in late April and early
Shrub dimensions (length, width, height, diameter at August. The soil core root mass of the surface 30 cm were
shrub base) and herbaceous plant biomass were meas- scaled to 1 m depth assuming 80% of total mass, as ob-
ured on 1±2 m radius microplots at each subplot centre in served in the 1 m profiles. Again, statistical analysis
all stands. In the I stands, dimensions were measured on showed no significant difference in the values for the
all trees (height, diameter at tree base, base of crown, and two periods, and data were pooled to obtain plot aver-
DBH), and shrubs on the subplots. ages.
Aboveground biomass, coarse root mass, and 10-year Carbon storage in soil was determined from the 12 soil
mean productivity of both were estimated from the in- cores used for fine root mass, plus samples to 100 cm
crement cores and tree dimensions, and local allometric depths below those cores. Live vegetation and roots
equations. The allometric equations for trees and shrubs were removed, and the total C and N were determined
were developed in four of the stands (Law et al., 2001b). (LECO CNS2000 analyzer), as well as bulk density and
Table 2 Annual and seasonal climate at the OS and YS ponderosa pine flux sites, and means for four age classes replicated three times in
year 2001
Annual mean Tsoil (8C) at 15 cm depth 7.6 (5.90) 8.5 (6.30) 9.75 (0.85) 8.48 (1.10) 8.74 (1.54) 9.12 (0.57)
Summer mean Tsoil (8C) at 15 cm depth 15.3 (1.79) 18.8 (2.24) 17.18 (0.20) 16.08 (1.74) 15.51 (1.26) 16.87 (1.44)
Winter maximum soil water content
(m3 m 3; 0±30 cm depth) 0.21 (0.012) 0.22 (0.017) 0.22 (0.02) 0.21 (0.06) 0.14 (0.01) 0.17 (0.02)
Summer minimum soil water content
(m3 m 3; 0±30 cm depth) 0.07 (0.008) 0.06 (0.009) 0.09 (0.01) 0.06 (0.02) 0.08 (0.01) 0.07 (0.01)
Annual mean Tair (8C) 8.7 (9.04) 7.9 (8.73)
Summer mean Tair (8C) 18.0 (6.69) 17.4 (5.85)
Annual diurnal Tair amplitude (8C) 10.4 (1.46) 7.5 (1.20)
Summer diurnal Tair amplitude (8C) 15.1 (0.77) 12.0 (0.66)
Annual rain (mm) 444 476
Summer rain (mm) 34 44
Annual PAR (MJ m 2) 2287 2211
Summer is defined as June through September. Standard deviations are shown in parentheses.
soil texture (hydrometer method, Central Analytical La- significantly different across the 300 years chronose-
boratory, Oregon State University). Soil carbon content quence, consequently DCFR and DCS were assumed to
was calculated for each soil layer from layer depth, bulk be zero.
density adjusted for the volume of coarse fraction
TBCA RS CL CFR CCR CS
2
(> 2 mm), and percent organic carbon data, and summed
for all layers to estimate total soil carbon content to 1 m
depth. NEP
NPPA RhWD
Coarse woody debris (CWD > 10 cm diameter, and
CFR CCR CS CL
3
> 1 m length) and fine woody debris (FWD 1±10 cm
The remaining carbon fluxes, belowground net pri-
diameter) mass were measured following Harmon &
mary production (NPPB), total heterotrophic respiration
Sexton (1996). Downed CWD (logs) were measured on
(Rh), and heterotrophic respiration from soil (RhS), were
four 75 m line transects on each plot and the species and
calculated using Eqns (4), (5), and (6), respectively.
decay class of each piece were identified (5 classes). Dimen-
sions of standing dead trees (snags and stumps) were used NPPB
NPPCR
NPPFR
4
in the allometric equations to determine standing dead
volume. Fine woody debris was measured on four 15 m Rh NEP
NPPA NPPB
5
transects per subplot for all four subplots. We calculated
C mass in CWD and FWD from estimates of wood volume RhS Rh
RhWD
6
for each species and decay class and corresponding wood
density of partially decayed woody detritus (Harmon & Uncertainty estimates were made for each component
Sexton, 1996; M. Harmon, pers. comm.). The annual loss of used for computing NEP by defining upper and lower
carbon from CWD (RhCWD) and FWD (RhFWD) decompos- boundaries based on various estimates of component
ition was calculated from mass per m2 ground, and error. Error estimates in NPPA and DCCR are based on
species±specific decomposition rates for the area error in radial growth estimates propagated through the
(M. Harmon, pers. comm.; Harmon et al., 1990). allometric models. Uncertainties in litterfall and soil
We computed NEP from measurements of carbon carbon estimates were identified as within plot sample
fluxes by the major pools. Net ecosystem production is standard deviations. Maximum boundaries for potential
gross photosynthesis (GPP) minus autotrophic (Ra) and DCFR and DCS were assigned to plots in each age class
heterotrophic respiration. The mass balance approach for based on a maximum range of pool values between age
estimating NEP begins with separation of above and classes. Uncertainty in CWD and FWD is based on esti-
belowground fluxes (Eqn (1)). mates of uncertainty in the sampling methodology
(J. Sexton, Oregon State University, pers. comm.).
NEP
NPPA RhWD
TBCA RS
1 A Monte Carlo model was then used to estimate means
where NPPA is aboveground net primary production and uncertainties in NEP (uncertainties expressed as
(sum of wood and foliage, over- and understorey plants), + 1 SD of 1000 Monte Carlo iterations; Law et al., 2001b).
RhWD is the respiration from woody debris (decompos-
ition of coarse and fine woody debris, stumps and snags),
TBCA is total belowground carbon allocation, and RS is
Physiological measurements
total soil surface CO2 flux. Following Raich & Nadelhof-
fer (1989), TBCA can be determined by subtracting We measured assimilation rates in relation to intercel-
annual litterfall carbon from RS plus any detectable lular CO2 concentration (A±Ci curves) on 1-year-old-
changes in belowground C pools (Eqn (2)). By substitut- needles at mid-canopy of 15- and 50-year-old-trees at
ing the TBCA equation into Eqn (1), RS cancels, allowing two of the chronosequence sites (plots 25 and 34) using
NEP to be calculated according to Eqn (3), where DCFR a portable infrared gas analyzer (LI-6400; LI-COR,
and DCCR are the changes in fine and coarse root carbon Lincoln, NE, USA). Measurements were made in June
with stand age, DCS is the net change in mineral soil C, and August. Foliar carbon and nitrogen (LECO
and DCL is annual litterfall carbon. Coarse root carbon CNS2000 analyzer), and specific leaf area (SLA; m2 one-
was calculated as the difference between the net growth sided leaf area g 1 dry mass) were determined on the
of live coarse roots and the decomposition of coarse roots same needles collected immediately after photosynthesis
attached to snags and stumps. The change in C stored in measurements. These data were necessary for calculating
roots, litter and soil with stand age accounted for non- Biome-BGC model parameters. We follow the methods of
steady state conditions when using a carbon balance Wilson et al. (2000) and Thornton et al. (2002; their
approach to estimating TBCA (Giardina & Ryan, 2002). Appendix A) to estimate leaf nitrogen in the Rubisco
Measurements of fine root mass and soil C were not enzyme ( flnr) from variables derived from A±Ci curves,
given concurrent measurements of leaf temperature, SLA, variable comparisons to observations. We used these
and foliar C : N. In this analysis we assume that the specific model results to develop an additional hypothesis
activity of Rubisco (act) varies only with temperature regarding variation in leaf nitrogen allocation to the
(60 mmol g 1 Rubisco s 1 at 25 8C, from Woodrow & Rubisco enzyme as a function of time since disturbance.
Berry, 1988), and that the mass fraction of nitrogen in the We also explored the variability in disturbance recovery
Rubisco molecule is constant (0.1397, calculated from responses related to atmospheric CO2 concentration
amino acid composition of Rubisco given in Kuehn & during recovery. The complete computer code and in-
McFadden, 1969). structions for running the model to reproduce the results
in this manuscript are available from P. E. Thornton on
request.
Data analysis
The effects of stand development (stand age classes) on Results and discussion
various carbon pools and carbon fluxes were examined
Air temperature averaged 8 8C in 2001 at the flux sites
by performing one-way analysis of variance using the
(Table 2). The summer maximum soil temperature was
anova procedure of SAS (Version 8; SAS Institute Inc.,
greatest in the more open I stands averaging 20.9 8C at
1999) based on our experimental design of four age
15 cm depth, and 17.5 8C in the M stands with higher leaf
classes with three replications. If the effect of stand age
area. Soil water content over 0±30 cm depth reached
classes were significant on a variable, means were further
a maximum of 0.22 m3 m 3 in the winter months for the
compared by Duncan's multiple range tests with a confi-
I stands and 0.14 m3 m 3 for the M stands, which on
dence level of P < 0.05. Non-linear curves of various
average had the highest leaf area (Table 3). In the summer
carbon pools and carbon fluxes as function of stand age
months, soil water content was similar between sites,
were fitted using the built-in statistical program of
reaching a mean low value of 0.08 m3 m 3.
SigmaPlot (Version 8.0, SPSS Science, IL, USA).
The YS flux site (plot 25 of I stands), with half the leaf
area of the OS flux site (plot 34 of O stands) suffered more
drought stress during the growing season than the O
Modeling
stand, and had consistently lower gross ecosystem prod-
We used a daily time step model of coupled carbon, uctivity (GEP) and less net carbon uptake through the
nitrogen, and water cycles (Biome-BGC, version 4.1.1, year, and lower annual NPP (Table 4). Weekly NEP(YS)
Thornton et al., 2002; with modifications as noted average 20% less than NEP(OS) through the year in
below) to simulate the historical patterns of disturbance 2000 (Fig. 1), with the greatest differences in summer
and recovery at each of the chronosequence sites. Daily (e.g. 40 and 30 g C m 2 week 1 at OS and YS). A study
surface weather drivers for the period 1980±1997 from the on hydraulic redistribution of water suggested that the
Daymet database (www.daymet.org, Thornton et al., old trees with more developed root systems were able to
1997; Thornton & Running, 1999; Thornton et al., 2000) access deep soil water during drought (Brooks et al.,
were used to drive the model, repeating the sequence as 2002).
necessary to obtain a weather record of the required Available soil moisture, particularly in spring and
length for each site. We followed the simulation ensem- early summer, is critical to survival of ponderosa pine
bling and disturbance simulation methods described in seedlings (Barrett, 1979). Competition for water is
Thornton et al. (2002) to isolate the disturbance recovery greatest in the 0±20 cm soil zone (Riegel et al., 1992),
signal from variability driven by interannual climate fluc- within which the pines will root in the first year. The
tuations. We followed the age-class ensembling methods process of stand development is function of shade in-
described in Law et al. (2001b) to represent mixed age tolerance, conditions favorable for seedling establishment
class stands (plots 34 and 36). associated with years of above-average precipitation, and
The default model parameterization uses static alloca- frequent fire (Cooper, 1960). Gaps created by mortality
tion ratios to relate new growth in fine root and woody allow seedlings to become established when good seed
tissue to new growth of leaves (Thornton et al., 2002). An crop years coincide with above-average precipitation.
earlier study demonstrated the effects on modeled NEP This may explain the slow reestablishment following log-
of changes in these ratios as a function of stand age (Law ging (8 year difference between logging date and mean
et al., 2001b). With the observations from the current stand age in the initiation stands), and combined with
study we were able to construct simple relationships logging intensity and the catastrophic fires that occurred
describing the variation of these allocation parameters in 1911 across the Pacific Northwest, it may explain the
with stand age, and we tested the hypothesis that this high frequency of stands in the 90±130 age classes on
variation reduces model bias in both storage and flux private and public lands (Fig. 2).
Table 3 Carbon pools of ponderosa pine stands in the Metolius area of central Oregon
25 0.89 (0.10) 0.57 1.46 (0.15) 802 633 587 459 1696 (698) 6918 (388)1 11 513
26 0.14 (0.02) 0 0.14 (0.02) 687 109 1568 1115 1153 (552) 12 686 (1912) 16 899
27 0.29 (0.07) 0.25 0.54 (0.10) 389 215 587 450 841 (271) 5380 (961) 7861
28 1.46 (0.12) 0.39 1.85 (0.13) 3653 1063 404 186 735 (185) 4060 (675) 10 102
29 1.53 (0.11) 0.4 1.93 (0.13) 5261 1816 320 373 1101 (217) 7634 (1617) 16 506
30 2.96 (0.12) 0.16 3.12 (0.12) 6898 2315 1459 382 1147 (310) 7868 (326)1 20 069
31 2.18 (0.14) 0.09 2.28 (0.13) 10 806 3489 510 339 1800 (561) 9699 (835) 26 643
32 2.47 (0.10) 0.01 2.48 (0.10) 5459 1603 3801 187 2073 (676) 6906 (590) 20 019
33 3.48 (0.11) 0.2 3.69 (0.12) 14 254 4662 1366 433 2070 (805) 7361 (801) 30 147
34 1.99 (0.14) 0 2.00 (0.12) 12 228 3990 524 451 1127 (348) 5556 (317)1 23 875
35 1.83 (0.07) 0.04 1.86 (0.09) 12 326 3829 510 136 1195 (471) 4122 (905) 22 117
36 2.12 (0.14) 0.39 2.50 (0.14) 15707 4837 1177 202 1954 (370) 6496 (2155) 30 372
Values are in grams carbon per m2 ground basis unless indicated otherwise. Standard deviations are shown in parentheses.
*LAI calculated as m2 half-surface area foliage per m2 ground, and corrected for clumping and wood interception; standard errors were
computed from grid-point measurements at 39 locations on each plot.
{
LMA is aboveground live mass, including understorey woody and herbaceous plants.
{
LMB is belowground live mass.
§
Coarse woody debris.
{
Fine woody debris.
**Forest floor fine litters, n 6.
{{
Soil carbon to 1 m depth; n 12, otherwise 1n 18.
{{
Ecosystem carbon is the sum of live and dead carbon pools and soil carbon.
Table 4 Annual carbon fluxes of ponderosa pine stands in the Metolius area of central Oregon
Plot ID Age class* NPPA NPPB RhS RhCWD RhFWD DCCR DCFR DCS NEP
25 I 158 (49) 150 (37) 368 (64) 7 (2) 16 (5) 20 (4) 0 (2) 0 (97) 83 (17)
26 I 109 (53) 45 (23) 299 (69) 56 (14) 43 (11) 17 (2) 0 (2) 0 (97) 244 (29)
27 I 75 (22) 87 (75) 173 (79) 16 (4) 18 (5) 15 (2) 0 (2) 0 (97) 44 (6)
28 Y 251 (45) 119 (39) 179 (66) 6 (2) 7 (2) 49 (9) 0 (5) 0 (129) 177 (28)
29 Y 200 (34) 175 (54) 294 (72) 8 (2) 14 (4) 45 (9) 0 (5) 0 (129) 60 (33)
30 Y 242 (57) 212 (42) 290 (78) 33 (9) 14 (4) 46 (14) 0 (5) 0 (129) 117 (32)
31 M 325 (71) 222 (69) 333 (107) 12 (3) 12 (4) 71 (18) 0 (1) 0 (17) 190 (39)
32 M 172 (81) 78 (36) 264 (98) 19 (5) 7 (2) 20 (20) 0 (1) 0 (17) 40 (41)
33 M 473 (118) 186 (80) 245 (155) 39 (10) 16 (4) 88 (34) 0 (1) 0 (17) 359 (58)
34 O 243 (77) 167 (35) 213 (94) 12 (3) 18 (5) 53 (22) 0 (2) 0 (22) 168 (41)
35 O 103 (36) 112 (48) 189 (64) 17 (5) 6 (2) 17 (10) 0 (2) 0 (22) 4 (23)
36 O 194 (84) 176 (57) 408 (110) 21 (6) 8 (2) 33 (24) 0 (2) 0 (22) 67 (41)
Values are in grams carbon per m2 ground basis unless indicated otherwise. Uncertainties are shown in parentheses: Uncertainty
estimates were determined by using various estimates of component errors and propagating this error through the computation of
each flux either arithmetically or through Monte Carlo simulation of 1000 iterations (See Methods and materials).
*I, initiation; Y, young; M, mature; O, old.
NPPA, Aboveground net primary productivity; NPPB, Belowground net primary productivity; RhS, Heterotrophic respiration in soil;
RhCWD, Respiration of coarse woody detritus; RhFWD, Respiration of fine woody detritus; DCCR , Net change in coarse root C pool; DCFR ,
Net change in fine root C pool; DCS, Net change in soil C pool; NEP, Net ecosystem productivity.
30 25 000
1:1 line
y = −2.07 + 0.816x
r 2 = 0.86 20 000
20
15 000
10 10 000
5000
y = 18433 (1 − exp( − 0.0215 x ))3.2793
0 Adj r 2 = 0.79
0
0 100 200 300
−5 0 5 10 15 20 25 30
Stand age (yr)
NEP (OS ) (g C m−2 wk−1)
Fig. 3 Total live biomass (above and belowground) of ponder-
Fig. 1 Weekly net ecosystem productivity (NEP) from micro- osa pine stands as a function of stand age in the Metolius area of
meteorological measurements made at the old pine flux site (OS) central Oregon.
and initiation pine flux site (YS).
014 35 000
25 000
Fraction of total plots
0.10
0.08 20 000
0.06 15 000
0.04 10 000
Fig. 2 Frequency distribution of ponderosa pine on federal Fig. 4 Total ecosystem carbon, including live and dead carbon
(CVS survey plots) and private (FIA survey plots) lands in pools and soil C, of ponderosa pine stands in the Metolius area of
Oregon as of 1998, representing stands where 80% or more of central Oregon.
the trees are ponderosa pine.
Table 5 Summary of carbon pools and carbon fluxes of ponderosa pine stands by age classes in the Metolius area of central Oregon
Mean age, class Total LAI* LMA{ LMB{ Fine litter C§ Ecosystem C{ NPP** NEP{{
20, I 0.71 (0.68) b 626 (213) b 283 (277) c 1230 (433) b 12 055 (4547) b 208 (87) b 124 (106) b
70, Y 2.30 (0.71) a 5271 (1623) b 1685 (630) bc 994 (226) b 15 513 (5051) b 400 (48) ab 118 (59) ab
100, M 2.82 (0.76) a 10 173 (4432) a 3251 (1543) ab 1981 (157) a 25 603 (5143) a 485 (212) a 170 (200) a
250, O 2.12 (0.34) a 13 420 (1981) a 4187 (542) a 1425 (459) ab 25 423 (4348) a 332 (103) ab 35 (120) ab
5
0.6
0.5
0.3
2
0.2
1
0.1 y = 0.5262 (1 − exp( − 0.0183 x ))1.7478
Adj r 2 = 0.91
0
0.0
0 100 200 300 0 100 200 300
Fig. 5 Ratio of aboveground wood mass to total ecosystem Fig. 6 Ratio of fine root mass to foliage mass as a function of
carbon reaches an asymptote at about stand age 100 year in stand age, indicating a change in carbon allocation patterns with
ponderosa pine stands in the Metolius area of central Oregon. stand developmental stage in ponderosa pine stands in the
Metolius area of central Oregon.
low in the O stands (35 g C m 2 yr 1) (Fig. 8, Tables 4 and observed in old forests, leads to an overall decrease in
5). The mean NEP of the 12 stands was 50 + 160 g wood production per unit resource used because non-
C m 2 yr 1. The variation in NEP of the M stands was dominant trees may decrease total ecosystem growth
high because one stand had been thinned about 10 year due to lower resource acquisition and resource-use effi-
previously and another stand includes very productive ciency than the dominant trees. On the other hand,
grand fir (33% difference in NPPA and 54% difference in canopy gaps form in old forests due to a variety of
RS between the stands with the lowest and highest NEP small-scale disturbances, allowing dense patches of
values in the M stand age class). Such structural and trees to develop. At the OS flux site, NPPA of a 50-year-
compositional heterogeneity is typical for mature stands old-stand (225 g C m 2 yr 1) was greater than that of a
in this region. Statistically, the only significant differences mixed age stand (126 g C m 2 yr 1), and the lowest NPPA
in NEP were between I and M stands; NEP values of the was observed in a pure old stand (96 g C m 2 yr 1) which
M stands were significantly higher than the I stands. The also had the lowest stand density (Law et al., 2001b). Pure
O stands, ranging from 190 to 316 year were mostly a net stands of old ponderosa pine trees in eastern Oregon are
sink for atmospheric CO2, albeit a much weaker sink than typically small patches (e.g. 1±2 ha), thus at the spatial
M stands (Table 4). Binkley et al. (2002) hypothesized that scale of flux measurements, the complex structure of the
differentiation in stand structure, as is commonly old forest may lead to greater net carbon uptake than
Heterotrophic respiration (g C m−2 yr −1) 600 of about 70 g C m 2 yr 1. This is a relatively low value
R hT compared with other temperate coniferous forests, but
R hS
500 reasonable for a semiarid region. For example, annual
NEP at flux sites in coniferous forests in Howland
400
Maine and Duke Forest North Carolina averaged 250
300
and 595 g C m 2 yr 1, respectively (Hollinger et al., 1999;
Law et al., 2002). Flux studies in 15 European forests show
200 a large range in NEP, from a net source of 100 to a sink of
660 g C m 2 yr 1 across forest types and latitudes (Valen-
100 tini et al., 2000). Disturbance effects were attributed to
some of this variation (e.g. high respiration in a Swedish
0
100 200 300 forest that is losing carbon because of past soil drainage).
Stand age (yr)
100 Thornton et al., 2002) forced the annual turnover rate for
fine roots and leaves to be the same, so in evergreen
0 vegetation the fine roots persisted for more than one
year. Observations of fine root biomass and production
−100
from this study show that the fine root lifespan is much
y = −239.4 + 399.9 exp(−0.5 (In(x /95.63)/1.098)2)
−200 Adj r 2 = 0.61 shorter than that for foliage, being approximately one
year compared to an average of about four years for
−300 foliage. The effect of this change in the model is to reduce
0 100 200 300
Stand age (yr) the standing fine root biomass and the associated main-
tenance respiration costs, increasing productivity. Prior
Fig. 8 Net ecosystem production (NEP) as function of stand age to this change we found it impossible to simulate the
in ponderosa pine stands in the Metolius area of central Oregon. large ratios of fine root production to foliage production
observed in the young stands ± the model would predict
would be expected from models based on homogenous unreasonably low leaf area due to high maintenance
stands. respiration costs.
Net primary production followed generally the similar Preliminary model simulations using fixed allocation
trends as NEP, but did not decline as much in the O ratios showed that the accumulation of woody biomass
stands (Table 5) which was driven mainly by the trends was too slow and that the ratios of fine root mass to
of NPPA (Table 4). Belowground net primary production foliage mass for young stands were too low, compared
was low in the I stands (averaged 94 + 53 g C m 2 yr 1), to observations. Based on the observed annual produc-
and high in other three stand age classes that did not tion of wood and fine root vs. foliage across the chron-
differ from each other (averaged between 169 + 47 g osequence, we established two simple linear
C m 2 yr 1 in Y stands and 152 + 35 g C m 2 yr 1 in O relationships that forced the wood and fine root alloca-
stands). tion ratios to start at relatively high values and decrease
Forest inventory data from the Forest Inventory Analy- with time since disturbance, remaining at a constant
sis (FIA) and Current Vegetation Survey (CVS) in Oregon value after 150 year. The allocation ratio for aboveground
provided age distributions for 5400 plots, 950 of which wood production : leaf production was set to vary from
had 80% or more ponderosa pine trees (probability-based 6 : 1 to 2 : 1, and the allocation ratio for fine root produc-
designs). The greatest proportion of plots is in the tion : leaf production from 7 : 1 to 3 : 1.
90±130 year age classes (Fig. 2). A curve fitted to our We found that high allocation to both fine roots and
NEP data and scaled according to proportion of total wood relative to foliage in the young stands caused the
survey plots in each age class results in a mean net sink model to significantly under-predict the rate of increase
in leaf area over time, resulting in low productivity for for both fine and small roots, covering a range in root size
older stands when compared to observations, so we ex- up to 2 cm in diameter, and the model results agreed
plored the possibility that per-unit-leaf area photosyn- more closely with the measurements for the fine root
thetic potential, held constant in the default model class than the small root class. Simply assuming the life-
parameterization, was in fact varying with stand age. span for fine roots is shorter than that for foliage pro-
There was no trend in measured foliar C : N with duces a pattern similar to that shown in Figs 6 and 10,
stand age, suggesting that increases in mass-based because the fine root mass reaches a steady state value
N concentration could not be invoked to produce vari- sooner than the foliage mass, but this mechanism alone
ation in area-based photosynthetic rates (Fig. 9). can explain only about half of the observed variation in
Similarly, there was no trend in canopy average specific mass ratio with stand age. Decreasing fine root allocation
leaf area (SLA) with stand age, suggesting that increases with age was required to simulate the full amplitude of
in area-based N concentration could also not be invoked the change in the mass ratio. Compared to the default
to produce this variation (Fig. 9). Analysis of A±Ci curves model parameterization, which kept the fine root alloca-
collected for fully developed leaves from 15- and 50-year- tion constant at its mature value, this has the effect of
old-trees at two of the chronosequence stands (plots 25 delaying the period of rapid leaf area development by
and 34) showed a decrease in the apparent fraction of leaf
nitrogen in the Rubisco enzyme ( flnr) with tree age
(Table 6). We refer to flnr as an apparent fraction since
60
variation between stands in specific activity of Rubisco 11
and/or actual fractional allocation of leaf nitrogen to
Rubisco could cause variation in A±Ci responses. We
Table 6 Summary of analysis of A±Ci curves to estimate flnr in two age classes
T range{
Trees SLA{ Foliar Vcmax (25 8C)§ MAE{ Bias{
Age Yearday (curves)* min max (m2 kg 1
C) C:N (mmol m 2 s 1) flnr (mmol m 2 1
s ) (mmol m 2
s 1)
*Number of trees sampled and A±Ci curves collected at each site on each date.
{
Range in measured leaf temperature during A±Ci curve collection.
{
One-sided specific leaf area.
§
Corresponding to the fitted values for flnr, measured values for SLA and foliar C : N, and values for Rubisco activity (at 25 8C) and mass
fraction of nitrogen in the Rubisco molecule as given in the text.
{
Mean absolute error (MAE) and bias in fitted model estimate of assimilation across all A±Ci points collected at a given site on a given
date.
about 5 year, also delaying the peaks in GPP, NPP, and The variation in total ecosystem carbon (the year-to-
NEP. year integral of NEP) agrees well with observations
The simulated leaf area over the chronosequence across the chronosequence (Fig. 13), although there is a
agrees well with observations for young and old stands, tendency for accumulation of live mass to be too rapid
but is significantly lower than observed for several stands during the first 50±100 year (Fig. 14), consistent with the
with ages around 100 year (Fig. 11). These results high- overestimation of NEP during this period. Comparing
light the difficulty in inferring time trajectories from the the simulated and observed NEP for the two oldest
chronosequence ± variations in species mixtures, ground stands (plots 35 and 36), it appears that the two-layer
water availability, and management practices between canopy (and two distinct age classes) for plot 36 is re-
stands are likely to play an important role in the observed sponsible for it having a higher NEP.
differences, none of which have been included in the
simulations.
The modeled and observed patterns of net carbon ex- 4
One-sided LAI
are similar, but the model predicts a more rapid shift
from source to sink during the early stages of stand
2
development following disturbance, with the modeled
peak sink strength occurring around stand age 25 year,
compared to the observed peak around 70±100 year.
A possible explanation for this bias is that during this 1
period the trees are competing with established shrubby
and herbaceous vegetation for water and nutrient re-
sources, an effect not accounted for in the model. These 0
results suggest that an accurate representation of early 0 50 100 150 200 250 300
Stand age (yr)
recovery dynamics in this system might require a more
sophisticated model that can explicitly represent compe- Fig. 11 Simulated one-sided leaf area index (LAI) as a function
tition between multiple plant functional types (Bachelet of stand age (squares, vertical bars show one standard deviation
et al., 2001). range of climate-driven interannual variability) and observed
overstorey LAI (d).
400
4
Fine root mass : foliage mass
200
NEP (g C m−2 yr−1)
0
2
−200
1
0 −400
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Stand age (yr) Stand age (yr)
Fig. 10 Simulated ratio of fine root mass to foliage mass as a Fig. 12 Simulated net ecosystem production (NEP) as a function
function of stand age (&) and observed ratios for two different of stand age (squares, vertical bars show one standard deviation
root size classes: fine roots (< 2 mm diameter, black circles), and range of climate-driven interannual variability), and observed
small roots (2 mm 2 cm diameter, grey circles). NEP (d).
average become a net source for CO2 , because of lower Giardina CP, Ryan MG (2002) Total belowground carbon alloca-
NPP but relatively small changes in heterotrophic respir- tion in a fast-growing Eucalyptus plantation estimated using
ation from soil. Accurate simulation of these dynamics a carbon balance approach. Ecosystems, 5, 487±499.
Harmon ME, Ferrell WK, Franklin JF (1990) Effects on carbon
requires a dynamic parameterization for biomass alloca-
storage of conversion of old-growth forests to young forests.
tion that depends on stand age, and should also include a
Science, 247, 699±702.
representation of competition between multiple plant
Harmon ME, Sexton J (1996) Guidelines for Measurements of Woody
functional types for space, water, and nutrients. Detritus in Forest Ecosystems. Publication No. 20. US LTER
Network Office, University of Washington, Seattle, WA.
Acknowledgements Hollinger DY, Goltz SM, Davidson EA et al. (1999) Seasonal
patterns and environmental control of carbon dioxide and
This study was funded by US EPA National Center for Environ- water vapour exchange in an ecotonal boreal forest. Global
mental Research (NCER) Science to Achieve Results (STAR) Pro-
Change Biology, 5, 891±902.
gram (Grant # R-82830901-0), and NASA (grant # NAG 5-11231),
IPCC (2001) Climate Change 2001: Synthesis Report (eds Watson
DOE (Grant # FG0300ER63014). Funding for model evaluation
was provided by NASA's Carbon Cycle Science Program (Grant R T, The Core Writing Team), Geneva, Switzerland, pp. 184.
# W-19953). Additional support for P. E. Thornton was provided (http://www.ipcc.ch).
by the National Center for Atmospheric Research (NCAR). Na- Irvine J, Law BE, Anthoni PM et al. (2002) Water limitations to
tional Center for Atmospheric Research is sponsored by the carbon exchange in old-growth and young ponderosa pine
National Science Foundation. We gratefully acknowledge the stands. Tree Physiology, 22, 189±196.
Sisters Ranger District of the US Forest Service for permission Janisch JE, Harmon ME (2002) Successional changes in live and
and assistance to access to the study sites. We thank Mark Har- dead wood carbon stores: impolications for net ecosystem
mon and Jay Sexton for help and useful discussion on sampling productivity. Tree Physiology, 22, 77±89.
coarse and fine woody detritus, and Darrin Moore for assisting
Kuehn GD, McFadden BA (1969) Ribulose 1, 5-diphosphate car-
sample collections. Acknowledgements of field sampling and
boxylase from Hydrogenomonas eutropha and Hydrogenomonas
data collection and root sample processing to: Jesse Bablove,
Jason Barker, Aaron Domingues, David Dreher, Marie Duch- facilis. II. Molecular weight, subunits, composition, and sulf-
arme, Isaac Emery, Nathan Gehres, Angie Hofhine, Erica hydryl groups. Biochemistry, 24, 359±371.
Lyman-Holt, Adam Pfleeger, Lucia Reithmaier, Matthew Shep- Law BE, Falge E, Baldocchi DD et al. (2002) Environmental con-
herd, Nathan Strauss, Vernon Wolf. trols over carbon dioxide and water vapor exchange of terres-
trial vegetation. Agricultural and Forest Meteorology, 113,
97±120.
References
Law BE, Kelliher FM, Baldocchi DD et al. (2001a) Spatial and
Amiro BD (2001) Paired-tower measurements of carbon and temporal variation in respiration in a young ponderosa pine
energy fluxes following disturbance in the boreal forest. forest during a summer drought. Agricultural and Forest
Gobal Change Biology, 7, 253±268. Meteorology, 110, 27±43.
Anthoni PM, Unsworth MH, Law BE et al. (2002) Seasonal dif- Law BE, Thornton P, Irvine J et al. (2001b) Carbon storage and
ferences in carbon and water vapor exchange in young- and fluxes in ponderosa pine forests at different developmental
old-growth ponderosa pine ecosystems. Agricultural and Forest stages. Global Change Biology, 7, 755±777.
Meteorology, 111, 203±222. Law BE, van Tuyl S, Cescatti A et al. (2001c) Estimation of leaf
Bachelet D, Lenihan JM, Daly C et al. (2001) MC1: A dynamic area index in open-canopy ponderosa pine forests at different
vegetation model for estimating the distribution of vegetation successional stages and management regimes in Oregon.
and associated ecosystem fluxes of carbon, nutrients, and Agricultural and Forest Meteorology, 108, 1±14.
water. USDA Forest Service General Technical Report PNW- Litvak M, Miller S, Wofsy FC et al. (2003) Effect of stand age on
GTR-508. whole ecosystem CO2 exchange in the Canadian boreal forest.
Barrett JW (1979) Silviculture of ponderosa pine in the Pacific Journal of Geophysical Research, 108.
Northwest: the state of our knowledge. USDA Forest Service Monleon VJ, Cromack K, Landsberg J (1997) Short and long-term
General Technical Report PNW-97. effects of prescribed underburning on nitrogen availability in
Binkley D, Stape JL, Ryan MG et al. (2002) Age-related decline in ponderosa pine stands in Central Oregon. Canadian Journal of
forest ecosystem growth: an individual tree, stand-structure Forest Research, 27, 369±378.
hypothesis. Ecosystems, 5, 58±67. Raich JW, Nadelhoffer KJ (1989) Belowground carbon allocation
Brooks JR, Meinzer FC, Coulombe R et al. (2002) Hydraulic in forest ecosystems: global trends. Ecology, 70, 1346±1354.
redistribution of soil water during summer drought in two Riegel GM, Miller RF, Krueger WC (1992) Competition for re-
contrasting Pacific Northwest coniferous forests. Tree sources between understorey vegetation and overstorey Pinus
Physiology, 22, 1107±1117. ponderosa in northeastern Oregon. Ecological Applications, 2,
Cohen WB, Harmon ME, Wallin DO et al. (1996) Two decades of 71±85.
carbon flux from forests of the Pacific Northwest. Bioscience, SAS Institute Inc. (1999) SAS OnlineDoc, Version 8. Cary, NC,
46, 836±844. USA.
Cooper CF (1960) Changes in vegetation, structure, and growth Schimel DS, House JI, Hibbard KA et al. (2001) Recent patterns
of southwestern pine forests since while settlement. Ecological and mechanisms of carbon exchange by terrestrial ecosystems.
Monographs, 30, 129±164. Nature, 414, 169±172.
Thornton PE, Hasenauer H, White MA (2000) Simultaneous Thornton PE, Running SW, White MA (1997) Generating sur-
estimation of daily solar radiation and humidity from ob- faces of daily meteorological variables over large regions of
served temperature and precipitation: an application over complex terrain. Journal of Hydrology, 190, 214±251.
complex terrain in Austria. Agricultural and Forest Valentini R, Matteucci G, Dolman AJ et al. (2000) Respiration as
Meteorology, 104, 255±271. the main determinant of carbon balance in European forests.
Thornton PE, Law BE, Gholz HL et al. (2002) Modeling and Nature, 404, 861±865.
measuring the effects of disturbance history and climate on Wilson KB, Baldocchi DD, Hanson PJ (2000) Spatial and seasonal
carbon and water budgets in evergreen needleleaf forests. variability of photosynthetic parameters and their relationship
Agricultural and Forest Meteorology, 113, 185±222. to leaf nitrogen in a deciduous forest. Tree Physiology, 20,
Thornton PE, Running SW (1999) An improved algorithm for 565±578.
estimating incident daily solar radiation from measurements Woodrow IE, Berry JA (1988) Enzymatic regulation of photosyn-
of temperature, humidity, and precipitation. Agricultural and thetic CO2 fixation in C3 plants. Annual Reviews of Plant
Forest Meteorology, 93, 211±228. Physiology and Plant Molecular Biology, 39, 533±594.