20.

The Rule and the accompanying Memorandum of Agreement (MOA) developed by the states for managing
Yellowstones grizzly bears post-delisting are inadequate in provisions for calculating and managing total
allowable mortality. Moreover, both the MOA and the Rule are deficient in terminology and provisions for
managing grizzly bear distribution.

20.1. The methods described in the Rule to account for background mortality are not only discrepant with
methods described in the MOA, but also fail to account for unknown-unreported grizzly bear deaths, thus
constituting a major methodological threat to the Yellowstone grizzly bear population.
On page 13203 of the Rule the Service walks the reader
through an example of how discretionary mortality will
be calculatedof which presumably all could be
provisionally allocated by the states at the beginning of a
given year for sport hunting. The example calculations
purport to account for background mortality, which is
defined in the previous paragraph as including various
causes, including unknown/unreported calculations.
Yet the example given by the Service, fails, in fact to
account for unknown/unreported mortalities, leading
to an inflated estimate of the number of bears candidate
for discretionary mortality. This is a major error given
that discretionary mortality for independent females and
males ends up being over-estimated by approximately
75% and 200%, respectively. In other words, this is a nontrivial even fatal mistake on the part of the Service.
The Service needs to account for unknown-unreported
mortalities in its calculations. Given the current IGBST
method used to estimate unknown-unreported deaths,
this foreseeable cause needs to be subtracted before
other foreseeable non-discretionary deaths are
deducted. Given that the unknown-unreported fraction
during 2010-2014 has averaged 39% of the total knownprobable deaths for independent females and 37% of the
same for independent males, these fractions need to be
the first to be subtracted from the initial calculation of
allowable mortality. After that, other non-discretionary
causes can be deducted.

Figure 20.1.1. An illustration of reductions in allowable

mortality that need to happen if unknown-unreported and
other non-discretionary deaths are to accounted for, in
addition to discretionary deaths foreseeably occurring on
NPS jurisdictionsprior to any determinations regarding
bears available for discretionary management on non-NPS
lands.

assumed to go to jurisdictions managed by the states of

Wyoming, Montana, and Idaho, and none to National
Park Service-managed lands. Recent history would
suggest that this NPS fraction is roughly 10% of the
discretionary total. In other words, explicit provision
needs to be made for this foreseeable mortality
occurring on NPS jurisdictions, with corresponding

The other major missing piece in the Services proposed

method for estimating and allocating discretionary
mortality is its lack of provision for the National Park
Service. This is especially glaring in the MOA, which
presumably further codifies methods described in the
Rule. In other words, all of the discretionary mortality is

69

(or at least transparent methods for calculating exact

percentages) for each stepdown need to be calculated
and presented by the Service

deductions from the states share of discretionary up

for grabs as a prospective sport hunt.
By way of clarification, figure 20.1.1, illustrates the
nature of the stepdown that needs to occur in the
Services calculations; first accounting for unknownunreported, then other non-discretionary, then the NPS
sharethis for a population of roughly 717. The
percentages I use are approximations. Exact percentages

Finally, the methods described in the MOA for calculating

discretionary mortality are considerably discrepant
from those described in the Rule. This discrepancy needs
to be corrected, along with other corrections that I have
outlined here.

20.2. A guideline for total mortality rate, with the

intent of producing population growth, is needed in
both the Rule and the MOA for an estimated
population size of 600 bears.

mortality once 600 is reachedtantamount to a drop in

total mortality rate. This is not enough. An explicitly
named target rate would provide incentive to reduce
what the agencies are currently calling Nondiscretionary mortality once the population is at or near
600; and this mortality rate should be several percentage
points below the 7.6% and 15% caps applied to males
and females when the population is <675. Given that
there are a 1.4% and 5% drops in the benchmark
mortality rates for females and males, respectively, once
the population drops below 675, it would be logical to
apply a commensurate drop in benchmark rates once the
population drops below 600, to around 6% for females
and 10% for males. (Also, see my 20.6, below, for more
on the problems of using Discretionary and Nondiscretionary).

The rationale for this recommendation is pretty straightforward. Managers should set a goal, expressed in terms
of total mortality rates, that will likely produce growth in
the Yellowstone grizzly bear population should it drop
below the basement number of 600. Even taken at face
value, the mortality guidelines currently set for any
population less than or equal to 674 are, by the Services
own admission, rates associated with a stable population
at best.
The agencies may argue that they already have a
provision for eliminating virtually all Discretionary

20.3. Review of current management approaches

should be mandated whenever mortality guidelines
are exceeded during any two consecutive years, for
any of the three specified cohorts of bears, rather
than the standard of three consecutive years
specified in the current Rule and MOA.

taking all of the other elements of the MOA at face value,

would be a recipe for producing a declining population
and without provision for introducing a timely change in
management. Under the current approach, a check
would only be introduced if estimated population size
dropped below 600, at which point, options for reversing
course would be intrinsically limited.

Both the Rule and the MOA specify that a review of

management leading to potential changes in harvest
would occur only if recommended mortality rates were
exceeded in 3 consecutive years. If the sequencing was
right (e.g., 2 years of excess followed by one year within
bounds, followed by two more sequences of this nature),
this protocol would allow the states to kill bears in excess
of recommended guidelines for 7 out of 10 years, which,

With the change recommended here (review after 2

rather than 3 successive years of violated mortality rate
targets), timely review would be triggered much more
often and with the prospect of actually reversing course
prior to excessive declines in the bear population.
Certainly, the current proposed approach is not
precautionary nor in any other way conservative.

70

20.4. Both the Rule and the MOA need to commit to

resetting or recalibrating all aspects of the methods
used to monitor trend, calculate allowable total
mortality, and trigger various outside reviews if and
when new methods are adopted for estimating total
population size. Without this provision, the existing
approach constitutes a methodological threat to the
population

Allowance for such a scenario not only introduces

substantial risk, but also, even more importantly,
emasculates and otherwise renders immaterial all of the
presumed safeguards against over-exploitation described
by the Service in the current Rule. A population that the
Service currently represents as numbering around 675
could suddenly be inflated to over 925, thereby allowing
for a potential sport harvest of, not 15, but rather nearer
25, a 67% increase. Likewise, a population at a threshold
of 600 that would debar all sport harvest could be
inflated to over 800, with instantaneous allowance for
harvest of 15 or so bears. And, even more problematic, a
population at the threshold of 500, that the Rule claims
might trigger a status review by the Service, could be
suddenly increased to 700, well above such a trigger.

The Rule and MOA describe methods for calculating total

allowable mortality (Discretionary plus Nondiscretionary) that are highly sensitive to estimates of
total population size. Given that different estimators of
total population size can yield numbers that vary by as
much as 40%even using the same inputs (e.g., MarkResight versus Chao2; Higgs et al. [2013] and IGBST
[2015])the Service needs to include language in the
Rule that explicitly guards against state agencies
introducing a new method that dramatically increases
estimates of total population size without any
commensurate adjustments in reckonings of trend,
methods for estimating total allowable mortality, or
thresholds that trigger outside reviewsall of which is
currently allowed in both the Rule and the MOA.

Whether such scenarios came to pass intentionally or

unintentionally, they would unambiguously pose a
serious threat to the population embedded in methods
currently contained in the Rule. The Service needs to
remedy this unacceptable risk. There are several options,
amongst which the least ambiguous and straight-forward
would be to commit to continued use of the Chao2
method for estimating total population size, but with an
accompanying commitment to rigorously standardize
search effort and distribution (see my comment 20.5
below). This would help control for the bias that besets
the Chao2 method. Another option would be to commit
in the Rule and MOA to use the lower bound of
uncertainty intervals for estimates of total population
size generated by the Mark-Resight method, should it be
adopted. This would presumably mitigate for the major
short-coming of this method identified by IGBST (2015),
which is the large uncertainty in annual estimates.

As is, state management agencies could (for example)

adopt the existing Mark-Resight method for estimating
total population size as soon as Yellowstones grizzly
bears are delisted, and produce a dramatic purported
increase in the population. This would instantaneously
translate into a markedly positive increase in putative
population trend along with numbers of bears available
for discretionary mortalitywithout any change
whatsoever in the underlying population or on-theground conditions.

20.5. The MOA and the Rule need to explicitly

specify that population monitoring will continue
indefinitely at the same intensity (neither more nor
less) and according to the same design as occurred
during the 5 years prior to delisting.

under 19.1), the Rule and MOA both need to contain a

commitment to maintaining the current exact intensity
and distribution of search effort devoted to documenting
the presence of females with COY. Such a commitment
would help curb any tendencies on the part of
management agencies to temporarily inflate population
estimates through increased search effort, especially if
such an increase were coupled with greater orientation
towards areas where grizzly bears are most easily seen.

Given vulnerabilities of the Chao2 methodor any other

foreseeable methodto bias introduced by search effort
and intrinsic sightability of bears (see my comments

71

20.6. Terminology for referring to bear mortality

should be changed in the Rule and MOA from
Discretionary versus Non-discretionary to
Management versus Other.

When you look at the more concrete categories of bear

deaths that the Rule and MOA are allocating to
Discretionary versus Non-discretionary, it turns out
that the distinction is fairly straight-forward distinction.
Discretionary deaths are simply those that will be
directly sanctioned by managers and meted out by either
uniformed employees of a state agency, by Wildlife
Services, or by those licensed to act on a state agencys
behalf (e.g., licensed hunters). Non-discretionary deaths
are simply all others resulting from the actions of those
(including other animals) who are not explicitly and
directly authorized, in any immediate sense, to kill grizzly
bears.

The semantics of the current distinction between

Discretionary and Non-discretionary mortality in the
Rule and the MOA lead to confusion. Moreover, the
distinction is disingenuous. By using these terms, the
Service and state managers lead both themselves and
their readers to assume that they have no control or
influence over so-called Non-discretionary mortalities
that this category of mortalities needs to happen or is
the result of some act of God. This is not the case.
History has shown (as the current Rule would claim) that
managers do, in fact, have substantial influence over the
so-called Non-discretionary mortalities through
activities such as law enforcement, education, and
sanitation. The Rule even strongly implies that managers
have influence over natural mortalities to the extent
that sport harvest of specific cohorts of bears can amplify
or dampen levels of mortality caused by conspecifics
especially infanticide (e.g., Swenson et al. 2001a, 2001b; ,
Bellemain et al. 2006; Bischof et al. 2009; Gardner et al.
2014). In other words, Non-discretionary mortalities
can, in fact, be Discretionary.

In fact, the current category of Discretionary correlates

closely with historical deaths of grizzly bears caused by
managers responding to conflict situations, including
threats to human safetyin other words, management
removals. Non-discretionary correlates with all of the
other historical categories. For the sake of clarity and in
service of reducing ambiguity, I recommend that the
Service revise the Rule so as to refer to Discretionary
kills as Management kills and Non-discretionary kills as
simply Other.

21. Occupancy provisions for adult females need to apply to all portions of the DMA, not just the PCA, stratified
on the basis of what are currently called Flight areas.
less than 33% of the adult females in the population (say,
25%), all concentrated on the ecosystem periphery, will
be subject to most of the planned killing each year. And,
importantly, the current approach essentially uses
females inside National Parks to subsidize calculations of
allowable sport harvest outside.

The current approach outlined in the Rule and the MOA

essentially loads all of the Discretionary mortality
allotted for independent females on those without
dependent young (lone females) outside of National
Parks. Under current provisions, no sport harvest of
females accompanied by dependent young would be
allowed. This amounts to the brunt of discretionary
human-caused deaths among females being borne by
lone bears on the periphery of the DMA.

The end result will be patently unsustainable killing of

females on the periphery. Source-sink population
dynamics would also certainly be amplified which,
according to Doak (1995), could lead to increased
vulnerability of the population to unintended and longlagged declines.

On average, only 1 of 3 adult females will be without

young during a given year. Moreover, some percentage
of these lone females will be inside National Parks where
they will not be subject to hunting. As a result, something

72

the ecosystem periphery and less predictability regarding

the consequences of management actions.

Preferentially killing females that would otherwise have

given birth to cubs the following year could introduce yet
other unpredictable amplifications of population trends.
On the face of it, amplified oscillations might be curbed
by the fact that, with fewer females giving birth to cubs
any following year, estimates of total population size
based on observations of females with COY would be
smaller, which might then lead to a lower Discretionary
kill the year after. But there would then be a pulse of
cubs from females that were subsequently subject to a
light harvest, which would lead to an inflated estimate of
population size and a resulting inflated sport harvest the
year afterand so on. In short, the approach described in
the Rule and the MOA will lead to fewer females living on

A requirement by the Service for occupancy of all

management units by reproductive females, including
flight areas outside the DMA (see figure 21.1 below),
would introduce a curb on excesses built into the Rule
and MOA, which currently have punitive implications for
females outside National Parks. Alleviating these current
excesses would result in proportionately greater
numbers of females on the periphery, which would
foster eventual connectivity between Yellowstone and
the NCDE.

Figure 21.1. Map A, above, shows the PCA (delineated by red, as shown in map B) relative to the full extent and partitioning of the
DMA (in yellow). The Rule and current MOA only provide for insurance of occupancy by adult females within the units contained by the
PCA and leaves occupancy of all the other units (Flight Areas) in the DMA up in the air when it comes to presence of reproductive
females.

73