Leaner leviathans: Body condition variation in a critically endangered whale population

Abstract

The role of environmental limitation and density-dependent regulation in shaping populations is debated in ecology. Populations at low densities may offer an unobstructed view of basic environmental and physiological interactions that impact individual fitness and thus population productivity. The energy reserves of an organism are reflected in its body condition, a measure linking individual fitness and the environment. From 1997 to 2007, we monitored the critically endangered western gray whale (Eschrichtius robustus) population on its primary summer feeding ground off the northeastern coast of Sakhalin Island, Russia. This effort resulted in a large data set of photo-identification images from 5,007 sightings of 168 individual whales that we used to visually assess western gray whale body condition. We quantified temporal variation in the resulting 1,539 monthly body condition determinations with respect to observations of reproductive status and sex. Western gray whale body condition varied annually, and we identified years of significantly better (2004) and worse (1999, 2006, and 2007) body condition. This study is the 1st to track the within-season body condition of individual whales. Body condition improved significantly as the summer progressed, although results suggest that not all whales replenish their energy stores by the end of the season. The body condition of lactating females was significantly worse than that of other whales at all times and was most often determined to be compromised. The body condition of their weaning calves exhibited no temporal variation and was consistently good. It is possible lactating females provide an energetic buffer to their offspring at the expense of their own body condition and future reproductive success. Findings from the analysis establish a foundation for quantifying links between western gray whale body condition, demographic parameters, and environmental conditions; and provide a baseline for monitoring individual and population condition of an ecosystem sentinel species in a changing environment. Overall, this study highlights the presence of density-independent environmental and physiological mechanisms that affect the abundance and growth of populations.

Full text

Leaner leviathans: body condition variation in a critically endangered
whale population
AMANDA L. BRADFORD,* DAVID W. WELLER,ANDRE
´E. PUNT,YULIA V. IVASHCHENKO,ALEXANDER M. BURDIN,
GLENN R. VANBLARICOM,AND ROBERT L. BROWNELL,JR.
School of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, WA 98195-5020, USA (ALB, AEP, GRV)
Protected Resources Division, Southwest Fisheries Science Center, National Marine Fisheries Service, National
Oceanic and Atmospheric Administration, 3333 North Torrey Pines Court, La Jolla, CA 92037-1022, USA (DWW)
Seastar Scientific, Dzerzhinskogo Street, 5-30, Yaroslavl 150033, Russia (YVI)
Kamchatka Branch of Pacific Institute of Geography, Far East Branch of the Russian Academy of Sciences, Prospect
Rybakov, 19-a, Petropavlovsk-Kamchatsky 683024, Russia (AMB)
Southwest Fisheries Science Center, National Marine Fisheries Service, National Oceanic and Atmospheric
Administration, 1352 Lighthouse Avenue, Pacific Grove, CA 93950-2097, USA (RLB)
Present address of ALB: Protected Species Division, Pacific Islands Fisheries Science Center, National Marine
Fisheries Service, National Oceanic and Atmospheric Administration, 1601 Kapiolani Boulevard, Suite 1000, Honolulu,
HI 96814-4700, USA
* Correspondent: alb992@u.washington.edu
The role of environmental limitation and density-dependent regulation in shaping populations is debated in
ecology. Populations at low densities may offer an unobstructed view of basic environmental and physiological
interactions that impact individual fitness and thus population productivity. The energy reserves of an organism
are reflected in its body condition, a measure linking individual fitness and the environment. From 1997 to
2007, we monitored the critically endangered western gray whale (Eschrichtius robustus) population on its
primary summer feeding ground off the northeastern coast of Sakhalin Island, Russia. This effort resulted in a
large data set of photo-identification images from 5,007 sightings of 168 individual whales that we used to
visually assess western gray whale body condition. We quantified temporal variation in the resulting 1,539
monthly body condition determinations with respect to observations of reproductive status and sex. Western
gray whale body condition varied annually, and we identified years of significantly better (2004) and worse
(1999, 2006, and 2007) body condition. This study is the 1st to track the within-season body condition of
individual whales. Body condition improved significantly as the summer progressed, although results suggest
that not all whales replenish their energy stores by the end of the season. The body condition of lactating
females was significantly worse than that of other whales at all times and was most often determined to be
compromised. The body condition of their weaning calves exhibited no temporal variation and was consistently
good. It is possible lactating females provide an energetic buffer to their offspring at the expense of their own
body condition and future reproductive success. Findings from the analysis establish a foundation for
quantifying links between western gray whale body condition, demographic parameters, and environmental
conditions; and provide a baseline for monitoring individual and population condition of an ecosystem sentinel
species in a changing environment. Overall, this study highlights the presence of density-independent
environmental and physiological mechanisms that affect the abundance and growth of populations.
Key words: density-independence, energy reserves, environmental variability, fitness, Okhotsk Sea, ordinal logistic
regression, photo-identification, western gray whales
E2012 American Society of Mammalogists
DOI: 10.1644/11-MAMM-A-091.1
The degree to which populations are limited by the
environment and regulated by their density is a topic of much
ecological interest and debate (e.g., Berryman 2004; White
2004). Recent mammalian studies have focused on the www.mammalogy.org
Journal of Mammalogy, 93(1):251–266, 2012
251

complex interactions between extrinsic and intrinsic factors
that produce observed population dynamics, particularly in
populations at high densities (e.g., Chamaille´-Jammes et al.
2008). Small or increasing populations are generally not
discussed in this context, although by extension they are often
used as a non–resource-limited reference in comparative
population studies (e.g., Monson et al. 2000). However, it is
clear that populations of all sizes are subject to environmental
and physiological conditions and constraints that impose
physical limits to population productivity. Populations at low
densities may reveal these properties in a more fundamental
form and merit investigation in this regard.
Depending on their metabolic needs and limitations,
individuals in populations employ a variety of life-history
tactics to contend with environmental variability. In pursuing
these strategies, all organisms face some level of temporal
reductions in body mass (Robbins 1993). Variation in stored
energy is a response to current nutritional inputs and demands
or to environmental cues regarding future conditions (Batzli
and Esseks 1992). The dependency on energy reserves for
reproduction varies across taxa, with animals that rely on
endogenous energy stores to sustain reproduction during a
period of fasting (i.e., capital breeders) representing 1 extreme
(Thomas 1990). The body condition of an organism reflects its
energy reserves relative to its size and can serve as a link to its
ecological fitness. In that respect, the influence of a variety of
environmental and physiological factors can be evaluated
using a single metric, assuming an appropriate measure of
body condition is identified (Speakman 2001).
Gray whales (Eschrichtius robustus) are extant only in the
North Pacific Ocean, where they exist as geographically and
genetically differentiated eastern and western populations
(Lang 2010; Weller et al. 2002). Like most other baleen
whales, gray whales feed in seasonally productive waters at
high latitudes, while using warm waters at low latitudes to
calve and breed. The eastern gray whale population migrates
from winter breeding grounds off Baja California, Mexico,
to summer feeding grounds that are encompassed primarily
by the Bering and Chukchi seas. Eastern gray whales are
potentially nearing the current carrying capacity of their
environment at a population size of approximately 20,000
whales (Punt and Wade 2010). The western gray whale
population returns to summer feeding grounds located
principally in the Okhotsk Sea from unknown breeding
grounds suspected to be along the southern coast of China
(Wang 1984). Western gray whales are critically endangered
(International Union for Conservation of Nature 2010) and
number ,150 individuals (Bradford et al. 2008).
Gray whales are in a negative energy balance after leaving
the feeding grounds, relying on stored energy acquired during
roughly 6 months spent foraging at high latitudes (Rice and
Wolman 1971). These reserves are of particular importance to
reproductive females, who have the potential to calve every
other winter. After a 13-month gestation period, pregnant
females give birth to a single calf, which is weaned during the
subsequent feeding season (Rice 1983). In baleen whales, as in
other mammals, energy stores are predominantly composed of
fat (Young 1976), although energy reserves also can include
other components such as carbohydrates and proteins
(Atkinson and Ramsay 1995; Robbins 1993). Collectively,
these sources of energy are catabolized from a variety of
tissues, including blubber, muscle, skeleton, and viscera
(Lockyer 1984; Worthy and Lavigne 1987).
Following a pilot effort in 1995 (Brownell et al. 1997), a
collaborative Russia–United States research program was
established in 1997 to conduct individual monitoring of
western gray whales using photo-identification and genetic
techniques (Weller et al. 1999, 2002). This project takes place
annually during summer months on the primary feeding
ground of the population, which is located in coastal waters off
northeastern Sakhalin Island, Russia, in the western Okhotsk
Sea. This feeding ground is utilized by western gray whales of
both sexes and multiple age classes, including postparturient
females and their weaning calves, and presumably offers
access to most, if not all, of the western gray whale population
(Bradford et al. 2006, 2008).
Several demographic parameters estimated over the course
of this investigation have suggested that western gray whales
are not realizing theorized levels of maximum productivity.
The number of actively reproducing females is small,
postweaning calf survival is low, the calf sex ratio is male-
biased, and calving intervals are prolonged and variable
(Bradford et al. 2006, 2008; Brownell and Weller 2002;
Weller et al. 2002). Undoubtedly, there are anthropogenic
sources of mortality to consider relative to these findings
(Bradford et al. 2009; Weller et al. 2008), and small
population effects cannot be ignored. However, concurrent
with these demographic studies were visual observations of
individuals with notable reductions in body mass (Brownell
and Weller 2001), which appeared to vary in magnitude and
over time, indicating that if western gray whale body condition
could be appropriately measured, environmental links to
individual fitness and thus population productivity could be
explored.
Common methods for evaluating body condition in
terrestrial mammals are generally impractical to apply to
free-ranging whales. Standard techniques such as direct
carcass analysis (e.g., Reynolds and Kunz 2001), mass-based
morphometric indexes (e.g., Jakob et al. 1996), and electrical
conductivity measurement (e.g., Wirsing et al. 2002) require
capturing and handling individuals, lethally in the case of
direct analysis. Given the migratory life cycle of most baleen
whales, seasonal changes in body condition are expected, with
whales reflecting more depleted energy stores while fasting
than while feeding (Lockyer 2007). In fact, studies of whales
killed in whaling operations have demonstrated that relative
body mass increases as the feeding season progresses (e.g.,
Lockyer 1987). Although these studies showed that blubber
thickness also can increase accordingly, particularly in
pregnant females, lipid stores captured in other body
components (e.g., muscle tissue) also are sensitive to changes
in nutritional status and may better explain observed seasonal
252 JOURNAL OF MAMMALOGY Vol. 93, No. 1

variations in body mass (Lockyer 1987; Vı
´kingsson 1990).
Further, there are many biases associated with measurements
of blubber, suggesting that blubber thickness may not be the
most reliable index of whale body condition (Aguilar et al.
2007).
In this regard, Rice and Wolman (1971) found that girth is a
better indicator of body condition than blubber thickness in
eastern gray whales and concluded that weight loss occurs
primarily because of utilization of internal fat depots as
opposed to blubber. Therefore, metrics of girth should
theoretically be able to reflect the nutritional status of
individuals. Indeed, aerial photogrammetry of eastern gray
whales found that changes in body condition associated with
fasting periods and reproductive status were reliably detected
from measurements of width relative to length (Perryman and
Lynn 2002). It is thus reasonable to assume such changes in
body mass could be observed sidelong during boat-based
photo-identification efforts and consequently recorded in the
photographic record.
This assumption formed the premise of a recent evaluation
of body condition in free-ranging North Atlantic right whales
(Eubalaena glacialis). In a retrospective analysis of photo-
identification data, Pettis et al. (2004) visually assessed the
relative amount of subcutaneous fat in the postcranial area
of whales in this population. This index accurately tracked
known changes in body condition associated with the
reproductive cycle (Pettis et al. 2004). Follow-up studies
demonstrated that the visual assessment of body condition
captured general trends in estimates of girth and blubber
thickness (Angell 2006). Thus, this type of assessment
represents a viable method of measuring right whale body
condition, which was the main premise of the work of Pettis
et al. (2004). The authors did not consider temporal variation
in individual body condition.
The western gray whale photo-identification project pro-
duced a large data set of digital, film, and video images suitable
for a visual assessment of western gray whale body condition.
Therefore, our objectives in the present study were to develop a
protocol to measure the body condition of western gray whales
from photo-identification images; and to quantify temporal
variation in the resulting determinations of body condition with
respect to observations of reproductive status and sex.
Specifically, we evaluated the relative amount of subcutaneous
fat for individual whales and tested for interannual and within-
season differences, with a particular interest in lactating females
and their weaning calves and in males and females. Our
overarching aim was to establish a baseline for monitoring
western gray whale body condition that can ultimately be used
to detect demographic and environmental linkages, in antici-
pation of future ecosystem changes.
MATERIALS AND METHODS
Whale sighting data.—From 1995 to 2007, during months
ranging from June to October, we conducted 336 small-boat
photo-identification surveys off the northeastern coast of
Sakhalin Island, Russia. Weller et al. (1999) contains detailed
information about the study area and the photo-identification
data collection and analysis protocols. We obtained biopsy
samples (for genetic studies, including sex determination and
relatedness analyses) in coordination with photo-identification
efforts, following animal care guidelines in accordance with
the American Society of Mammalogists (Sikes et al. 2011).
These surveys produced 5,159 sightings of 169 individual
western gray whales, which include 24 known reproductive
females, 72 whales 1st identified as calves, and 142
individuals of known sex (59 females and 83 males). We
considered a given female to be reproductive overall and
lactating for the field season when genetic or behavioral
observations, or both, linked her to a calf of the year. A
sighting is represented by at least 1 high-quality photo-
identification image, although we usually collected several
photo and simultaneous video images during each sighting.
We acquired 14 additional sightings of 12 of the 169
individuals during a survey of an ephemeral feeding area
approximately 60 km southeast of the nearshore feeding area.
In total, we examined more than 34,000 film and digital
images and 38 h of digital video from 5,173 sightings of 169
photo-identified individuals to assess western gray whale body
condition. However, we utilized only data collected during
July–September of 1997–2007 in the statistical analysis of
body condition, so that we could make temporal comparisons.
The analysis subset consisted of 5,007 sightings of 168
individual whales, which include the same individual
composition as above less 1 male 1st identified as a calf.
Body condition assessment.—We expanded the protocol
developed for North Atlantic right whales (Pettis et al. 2004)
to assess the body condition of western gray whales. We
measured western gray whale body condition using a similar
scoring approach, but along with the postcranial area, we
evaluated 2 additional body regions also regularly captured
during photo-identification. That is, we visually assessed the
relative amount of subcutaneous fat in 3 body regions: the
postcranial area, the scapular region, and the lateral flanks.
Apparent reductions in body mass in these regions lead to 3
diagnostic features, respectively: a postcranial depression, a
subdermal protrusion of the scapula, and a depression along
the dorsal aspect of the lateral flanks (Brownell and Weller
2001). Although the underlying physiological mechanisms are
not well understood, whales exhibiting these features are
considered to be in compromised body condition (Brownell
and Weller 2001).
We examined all available digital, film, and video images of
individual western gray whales in the assessment of body
condition. Specifically, for each survey sighting of a whale,
we assigned a numerical score to the 3 body regions of
interest, with higher values corresponding to better condition
(Figs. 1–3). We scored the postcranial condition on a 3-point
scale (Fig. 1), but scored the scapular and lateral flank
conditions on a 2-point scale (Figs. 2 and 3) because
distinguishing varying degrees of these characteristics would
have been highly subjective. If we could not assign a reliable
February 2012 BRADFORD ET AL.—VARIATION IN GRAY WHALE BODY CONDITION 253

FIG.1.—Example images depicting the 3-point scale we used to assess the postcranial condition of western gray whales (Eschrichtius
robustus). We assigned a) a score of 3 to whales with a flat or rounded back; b) a score of 2 to whales with a slight to moderate postcranial
depression, indicated by an arrow; and c) a score of 1 to whales with a significant postcranial depression such that a pronounced ‘‘hump’’ is
visible posterior to the blowholes, noted by large and small arrows, respectively.
FIG.2.—Example images showing the 2-point scale we utilized to evaluate the scapular condition of western gray whales (Eschrichtius
robustus). We assigned a) a score of 2 to whales with rounded sides over the shoulder blades; and b) a score of 1 to whales with a subdermal
protrusion of the scapula, identified by an arrow.
254 JOURNAL OF MAMMALOGY Vol. 93, No. 1

numerical score to a body region (e.g., we did not take images
of the body region or whale body position confounded body
region condition), we coded the region as X. One analyst
(ALB) performed all scoring to maintain consistency in the
image review (Pettis et al. 2004). However, we demonstrated
through an interrater agreement study that the western gray
whale body region scoring protocol can be used by more than
1 trained researcher to achieve comparable results (Appendix I).
To maximize the use of irregular sightings with incomplete
image coverage of the 3 body regions, we collapsed the scored
data for each sighting into monthly composites of postcranial (P),
scapular (S), and lateral flank (L) condition for each whale. We
conducted sensitivity analyses to confirm that month was an
appropriate and feasible scale at which to aggregate these data
(Appendixes II–IV). We then needed a scheme to classify these
composites to produce overall categorical determinations of
individual body condition (i.e., good, fair, or poor) for use in the
statistical analysis. Given that the postcranial condition is
presently the standard visual measure of cetacean body condition
(e.g., Pettis et al. 2004), we assumed this region to be most
indicative of overall body condition. That is, we classified a
composite of 3SL as good body condition, 2SL as fair body
condition, and 1SL as poor body condition, unless we scored both
the scapular and lateral flank conditions as poor. In those cases
(i.e., composites of 311 and 211), we brought the body condition
rating down a level (i.e., to fair and poor, respectively). Any other
combination after the postcranial condition score (e.g., 3XX or
21X) did not change the rating suggested by the postcranial
condition. If we coded the postcranial condition as X, then we
considered the body condition as unknown and unusable for
analysis. In summary, the possible composites within each body
condition category are:
Ngood—322, 321, 32X, 312, 31X, 3X2, 3X1, 3XX;
Nfair—311, 222, 221, 22X, 212, 21X, 2X2, 2X1, 2XX;
Npoor—211, 122, 121, 12X, 112, 111, 11X, 1X2, 1X1, 1XX;
and
Nunknown—X22, X21, X2X, X12, X11, X1X, XX2, XX1,
XXX.
This classification system is a conservative rating approach
that allows composites with X entries to be incorporated
into the 3 categories of known body condition and therefore
utilized in the analysis.
Note that we use the body condition descriptors good, fair,
and poor to refer to the amount of energy reserves available
to a given whale and not to imply a prognosis of survival.
Further, a determination of poor body condition does not
equate to starvation, although if starving individuals were a
part of this study, presumably we would have classified
their body condition as poor. Instead, we regard whales
in compromised body condition (i.e., fair or poor) as not
completely buffered against the full suite of demands
associated with their extreme life history, which could lead
to a behavioral, physiological, or reproductive response.
Statistical analysis.—We employed multinomial logistic
regression for ordinal responses to analyze variation in
western gray whale body condition. Specifically, we used
the proportional odds model (e.g., Agresti 2002) to evaluate
the effect of 4 categorical variables (year, month, reproductive
class, and sex) on body condition as a multinomial response
(good, fair, or poor), where year is 1997–2007; month is July,
August, or September; reproductive class is lactating female,
calf, or other whale; and sex is male, female, or unknown. We
specified these temporal and demographic covariates given
their relevance to the study objectives. Further, preliminary
univariable analyses suggested that these variables are each
significant predictors of body condition. We utilized likeli-
hood ratio tests to determine the most-parsimonious model,
which we found by singly dropping each of the covariates
FIG.3.—Example images showing the 2-point scale we employed to rate the lateral flank condition of western gray whales (Eschrichtius
robustus). We assigned a) a score of 2 to whales with rounded sides from the postcranial area to the start of the caudal peduncle; and b) a score of
1 to whales with a depression along the dorsal aspect of the lateral flanks, indicated by an arrow.
February 2012 BRADFORD ET AL.—VARIATION IN GRAY WHALE BODY CONDITION 255

from the full model and, if required, from selected reduced
models until we identified the most-parsimonious set of
covariates. Individual whales were represented by a body
condition category in as many months as the individual
was sighted. To account for the correlation between these
observations, we modeled individual whales as normally
distributed random effects in the analysis, which we conducted
using the Ordinal Package (Christensen 2010) within the
program R (R Development Core Team 2010).
RESULTS
We collapsed the 5,007 survey sightings between July and
September of 1997–2007 into 1,539 monthly body region
condition composites of 168 photo-identified western gray
whales. The distribution of composites within each body
condition category is: good—658 (42.8%), fair—317 (20.6%),
poor—158 (10.3%), and unknown—406 (26.4%). Fig. 4
shows the frequencies of the possible composites within each
body condition category. Within each category of known body
condition, most of the composites are comprised of non-X
entries (i.e., 322, 321, 312, 311, 222, 221, 212, 211, 122, 121,
112, and 111): good—583 (88.6%), fair—243 (76.7%), and
poor—136 (86.1%). More information on the scored and
collapsed body region data that led to the monthly deter-
minations of body condition is available in Appendixes
II–IV. Whales with known body condition determinations
are represented by 165 individuals, with a median of 5
determinations per whale (range 1–24 determinations).
Table 1 summarizes the distribution of these individuals and
observations within the analytical framework.
Likelihood ratio tests indicated that sex is not a significant
predictor of body condition in the presence of the covariates
year, month, and reproductive class (Table 2). Because there
was not support for dropping additional covariates, we
selected the mixed model incorporating year, month, and
class as the most parsimonious. When fit to the body condition
determinations, this model revealed that, compared to the
reference year of 1997, whales were in significantly better
body condition in 2004 and in significantly worse body
condition in 1999, 2006, and 2007 (Table 3). Moreover,
whales were in significantly better body condition in August
and September than in July, with the magnitude of the
predictor coefficients pointing toward an improvement in body
condition as the season progressed. Finally, lactating females
were in significantly worse body condition relative to other
whales, while weaning calves that were in significantly better
condition than other whales.
The estimated predictor coefficients in Table 3 can be
exponentiated and expressed as odds ratios. For example, a
specific whale had about 3 times the odds of being in better
body condition in 2004 (all other factors being equal) than in
FIG.4.—Frequencies of monthly body region condition composites
of those possible within each of the 4 body condition categories
(good, fair, poor, and unknown) for 1,539 composites of 168 western
gray whales (Eschrichtius robustus). Frequencies are shown accord-
ing to reproductive class (lactating females, calves, and other whales),
with individual whales represented in as many months as the
individual was sighted.
TABLE 1.—Summary of observations used in the quantitative analysis of western gray whale (Eschrichtius robustus) body condition.
Individual whales are represented once in the annual numbers of whales in known body condition and within each month, but are represented
in as many months and years as the individual was sighted. Further, individual whales can be represented multiple times in the annual
numbers within each reproductive class and sex category, depending on the number of known monthly body condition determinations for the
individual. BC 5body condition; LF 5lactating female.
Year
Whales in
known BC
Month Reproductive class Sex
Jul. Aug. Sep. LF Calf Other Male Female Unknown
1997 37 16 24 22 5 5 52 29 28 5
1998 48 33 28 22 16 15 52 35 41 7
1999 64 42 54 35 4 7 120 70 46 15
2000 54 7 50 38 3 5 87 58 34 3
2001 63 42 53 46 16 17 108 78 59 4
2002 70 38 47 50 16 21 98 68 62 5
2003 65 16 50 41 18 20 69 56 51 0
2004 55 22 50 1 11 13 49 31 37 5
2005 67 18 41 36 9 8 78 48 42 5
2006 61 33 45 0 7 5 66 41 28 9
2007 75 32 62 39 18 21 94 77 51 5
256 JOURNAL OF MAMMALOGY Vol. 93, No. 1

1997, had approximately 14 times the odds of being in better
body condition in September than in July, and had roughly 124
times the odds of being in worse body condition when a lactating
female than when classified as an other whale. However, when
considering effect size, the calf coefficient and associated values
merit particular attention. Of the 137 calf determinations of
known body condition, 136 (99.3%) of them are classified as
good. Thus, this covariate level perfectly predicts the outcome,
which can create numerical problems when fitting a logistic
regression model, typically manifested as a lack of convergence,
a large estimated coefficient, and a large estimated standard
error (Hosmer and Lemeshow 2000). In the present case, the
model converged and although the estimated coefficient is large,
the resulting standard error is not large enough to lead to a
paradoxically small Wald test statistic (Hauck and Donner
1977). Further, compared to the original model formulation (see
Table 2), eliminating the calf covariate level (residual degrees
of freedom [d.f.]51,117, log-likelihood 52765.565) was not
supported by a likelihood ratio test (chi-square distributed
likelihood ratio statistic [LR]
1
5106.794, P,0.001). Also,
completely removing calf observations from the statistical
analysis did not produce appreciable differences in the estimates
corresponding to the other covariates. Therefore, we retained the
calf observations and covariate for illustrative purposes, but the
resulting effect size should be interpreted with caution.
The predicted probabilities of an average whale (i.e., a
random effect of zero) being in good, fair, and poor body
condition according to various combinations of the covariates
are shown in Fig. 5. The random effect estimates conformed
to a normal distribution as intended. A random effects model
allowed for the appropriate statistical treatment of the
correlation between observations of individual whales.
Additionally, compared to the same covariate model without
random effects (d.f. 51,117, log-likelihood 52798.273), the
random effects model (see Table 2) provided a significantly
better fit to the data (LR
1
5172.209, P,0.001).
DISCUSSION
Photo-identification of gray whales involves the comparison
of natural and unique pigmentation patterns and generally
focuses on the dorsal flank region of individual whales
(Darling 1984; Weller et al. 1999). Thus, this body region
was the primary target during western gray whale photo-
identification efforts, but was not used to measure individual
body condition, which explains the large number of unknown
body condition determinations in the current retrospective
assessment (Fig. 4). However, it is important to note that these
unknown determinations are random with respect to the
examined covariates.
Although the western gray whale body condition assess-
ment protocol allows composites with X entries to be
incorporated into the known body condition categories, most
of the resulting known body condition determinations are
composed of non-X entries (Fig. 4). These non-X composites
offer 2 potential insights into patterns of weight loss in gray
whales. First, there is individual variation in where on the
body declines in mass occur, as evidenced by the frequencies
of the various types of non-X composites (Fig. 4). Second,
despite this variation, there is evidence that of the 3 body
regions evaluated, the postcranial area is the most sensitive to
reductions in subcutaneous fat. That is, the relatively high
frequency of the composite 222 compared to frequencies of
composites with normal postcranial and compromised scap-
ular and lateral flank conditions (i.e., 321 and 312) suggests
that discernible mass loss occurs 1st in the postcranial area and
then in the other 2 body regions (Fig. 4), although unknown
differences in the ease of mass loss detection between body
regions also may have contributed to the perceived order of
mass loss.
The decision to differentiate between the body condition
categories of good, fair, and poor for western gray whales was
primarily dictated by our ability to visually discern 3 levels of
TABLE 2.—Results of comparing the full proportional odds mixed
model of western gray whale (Eschrichtius robustus) body condition
to reduced models formed by singly dropping each of the 4
covariates. We used likelihood ratio tests to compare the full model
(top row) with each reduced model as a means to evaluate the
significance (P,0.05) of the dropped covariate. The selected model
is shown in boldface type. d.f. 5degrees of freedom; LR 5likelihood
ratio statistic (chi-square distributed).
Model predictors
Residual
d.f.
Log-
likelihood LR LR d.f. P-value
Year +month +class +sex 1,114 2710.991
Month +class +sex 1,124 2750.663 79.342 10 ,0.001
Year +class +sex 1,116 2778.171 134.360 2 ,0.001
Year +month +sex 1,116 2892.766 363.549 2 ,0.001
Year
+
month
+
class 1,116 2712.168 2.354 2 0.308
TABLE 3.—Maximum-likelihood estimates resulting from fitting
the proportional odds mixed model to determinations of western gray
whale (Eschrichtius robustus) body condition, given year, month, and
reproductive class. The first 2 rows represent model intercepts and the
rest predictor coefficients. Note that year 51997, month 5July, and
class 5other whale served as the reference categories. Significant
predictor coefficients (P,0.05) are shown in boldface type. SE 5
standard error; LF 5lactating female.
Variable Estimate SE Wald zP-value
Yfair 2.736 0.413 6.618 ,0.001
Ygood 20.243 0.395 20.616 0.538
Year 51998 0.647 0.477 1.358 0.175
Year =1999 21.071 0.404 22.655 0.008
Year 52000 20.511 0.435 21.175 0.240
Year 52001 20.568 0.411 21.381 0.167
Year 52002 0.094 0.427 0.219 0.826
Year 52003 0.116 0.450 0.258 0.796
Year =2004 1.128 0.507 2.225 0.026
Year 52005 20.526 0.441 21.192 0.233
Year =2006 20.971 0.447 22.171 0.030
Year =2007 21.706 0.422 24.044
,
0.001
Month =Aug. 1.235 0.188 6.553
,
0.001
Month =Sep. 2.615 0.246 10.622
,
0.001
Class =LF 24.821 0.365 213.195
,
0.001
Class =calf 5.694 1.073 5.309
,
0.001
February 2012 BRADFORD ET AL.—VARIATION IN GRAY WHALE BODY CONDITION 257

postcranial condition. However, during analyses associated
with such categorical assessments, it may ultimately be
difficult to statistically reconcile the resulting body condition
determinations with particular covariates of interest, which
may require restructuring the analysis or even revising the
protocol. In the present case, the calf covariate level perfectly
predicted the outcome of good body condition, which could
have led to numerical problems when fitting the logistic
regression model to the data set as a whole. Although this
situation did not occur, the estimated effect size should
nevertheless be considered unreliable. To a lesser degree,
there also is reduced contrast in the body condition
determinations for lactating females, with 121 (98.4%)of
the 123 determinations of known body condition classified as
fair or poor. If model instability had ensued, it might have
been necessary to treat lactating females separately, perhaps
in a binomial instead of a multinomial context. In general,
flexibility in the treatment of covariate and response levels
may be needed when statistically analyzing visual determina-
tions of body condition.
Gray whale calves are weaned at approximately 7 months of
age (Rice and Wolman 1971), a relatively short period of
time given their body size, implying they rely on maternally
derived energy for some period of time postweaning (Costa
and Williams 1999). The comparatively good body condition
of western gray whale calves (Table 3) is in line with findings
from previous cetacean studies (e.g., Angell 2006) and reflects
the significant energetic investment and high milk fat provided
to them by lactating females (Rice and Wolman 1971). The
consistently good body condition of calves (Fig. 5b) suggests
that differences in maternal condition and environmental
factors affecting calf development are not expressed in the
stored energy of weaning calves, at least as measured by the
present protocol. These differences may instead be manifested
in the overall size of calves as opposed to their body condition.
Perryman and Lynn (2002) found a positive correlation
between the length of northbound migrating eastern gray
whale females and the length of their calves, although the
authors did not compare calf length to a metric more
indicative of the body condition of associated females.
Maternal and environmental effects also likely influence the
growth of calves in ways that are not immediately apparent
(Bernardo 1996), but that have long-term fitness consequences
(Lindstro¨m 1999).
Still, the lack of variation in the body condition of calves is
striking, particularly in light of the pronounced variation
exhibited by noncalves (Figs. 5a and 5c). Perhaps reproduc-
tive females that are not nutritionally prepared to wean a calf
with complete energy reserves remain anestrous or lose their
calves prematurely (Lockyer 1984; Rice and Wolman 1971).
FIG.5.—Predicted monthly probabilities of an average (i.e., a random effect of zero) western gray whale (Eschrichtius robustus) being in
good, fair, and poor body condition during 1997 as compared to years in which body condition was significantly better (2004) and significantly
worse (1999, 2007; 2006 not shown) for a a) lactating female (LF), b) calf, and c) other whale.
258 JOURNAL OF MAMMALOGY Vol. 93, No. 1

Calving intervals of eastern and western gray whales are
variable (Bradford et al. 2008; Jones 1990), and there is
evidence of high neonatal mortality in eastern gray whales
(Swartz and Jones 1983). Further, it is conceivable that
lactating females are able to energetically buffer their calves
from adverse environmental conditions, although such an
energy transfer would come at the expense of the body
condition and future reproductive success of the female
(Lindstro¨m 1999).
The high energetic costs of mammalian lactation (Young
1976) are particularly considerable for whales, who are fasting
during much of this period (Lockyer 1984). As expected, the
body condition of lactating female western gray whales was
relatively worse than that of other whales (Table 3) and was
most often determined to be compromised (Figs. 4 and 5a).
Although there was some degree of monthly improvement
in the body condition of lactating females, probabilities of
complete within-season recovery were to a greater or less
extent low in all years (Fig. 5a), indicating that postparturient
females usually have not fully replenished their energy stores
by the time of the next breeding opportunity (i.e., the
subsequent winter).
There is a well-established correlation between body
condition and reproductive success in female mammals
(Loudon et al. 1983), with maternal body condition potentially
impacting all aspects of the reproductive process, including
the timing of reproduction (Hickling et al. 1991), probability
of pregnancy (Cook et al. 2004), embryonic absorption
(Belonje and van Niekerk 1975), fetal growth (Lockyer
2007), offspring mass (Atkinson and Ramsay 1995), offspring
survival (Cameron et al. 1993), and progeny sex ratio
(Wauters et al. 1995). Although the relationship between
body condition and reproduction is not well understood for
whales, a few basic scenarios have been proposed. It is
generally presumed that if a reproductive female has
insufficient energy reserves, she may either fail to ovulate,
fail to conceive, fail to give birth, or fail to nurse.
Alternatively, she may direct her own maintenance reserves
into producing and weaning a calf (Lockyer 1986). Further,
these mechanisms are thought to be regulated by environ-
mental conditions, such that ovulation and conception rates are
likely linked to 1 feeding season and abortion and calving
rates to the next, although a series of good or bad years could
mitigate or confound these connections (Lockyer 1987).
Given that western gray whale calving intervals do vary
(Bradford et al. 2008), it is possible that some form of
environmental regulation through maternal body condition is
occurring (Brownell and Weller 2002). There are conflicting
ideas about the primary method of nutritional control in gray
whale reproduction. Rice and Wolman (1971) suggested
that females are likely to suppress ovulation when in poor
condition and unable to carry a pregnancy to term. However,
Perryman et al. (2002) found that calf production in eastern
gray whales was positively correlated with the length of the
previous feeding season (as determined by ice cover), with no
significant correlation when a 1-year lag was introduced,
implying that existing pregnancies were affected, rather than
ovulations or conceptions. The latter scenario is consistent
with the differential costs of pregnancy in whales, which only
become substantial during the last one-half or one-third of
gestation (Lockyer 1984). Other baleen whale studies point to
the importance of the feeding season prior to (e.g., Lockyer
1986) and during (e.g., Lockyer 2007) pregnancy, but it is
unlikely that these links are mutually exclusive, particularly
when interactions between years and other factors (e.g.,
previous calf production or maternal age) are considered.
Regardless, it appears that western gray whale females do
fully invest in their calves at a certain point, potentially even
providing environmental amelioration. As the energetic costs
of lactation are much greater than those of pregnancy (Millar
1977), there are likely to be consequences of this investment
for female reproductive success.
Sex is not an important predictor of western gray whale
body condition given the incorporation of reproductive class
in the mixed model (Table 2), suggesting that lactating
females were responsible for the significant differences
detected during preliminary univariable analyses. However,
the presence of pregnant females and juveniles of both sexes
in the sample might have confounded the sex comparison. A
variety of measures (e.g., blubber thickness, body girth, and
lipid content) have shown that pregnant female whales have
the highest energy stores relative to other whales (e.g.,
Lockyer 1986). With the exception of lactating females,
juvenile whales generally have the lowest fat reserves (e.g.,
Vı
´kingsson 1990), although some studies have found juvenile
males to be leaner than all other whales (e.g., Lockyer 1987).
Additionally, comparing the body condition of males and
females at the same time may be inherently problematic
because these whales may have been on the feeding ground for
varying durations, since there appears to be some degree of
temporal segregation by age, sex, and reproductive status in
migrating gray whales (Rice and Wolman 1971). In any case,
energy deposits are clearly needed by male and female gray
whales for maintenance activities during the fasting period and
by females to sustain substantial reproductive demands.
The monthly improvement in the body condition of western
gray whales (Table 3; Fig. 5) demonstrates the significance of
the feeding period for accumulating energy stores and is
consistent with findings from previous whale research (e.g.,
Lockyer 1987; Perryman and Lynn 2002; Rice and Wolman
1971; Vı
´kingsson 1990), although the current study is the 1st
to monitor the within-season body condition of individual
whales. The predicted probability that a nonlactating, noncalf
(i.e., other whale) was in good condition at the end of the field
season was generally, but not always, very high (Fig. 5c).
Given patterns of seasonal sea-ice formation in the Okhotsk
Sea, western gray whales presumably have access to the
northeastern Sakhalin feeding area for at least 2 months
beyond the monitoring period of the present assessment.
Whales have been observed in the study area as late as mid-
November, but in considerably reduced numbers, suggesting
that most whales have left the region by that time (Blokhin
February 2012 BRADFORD ET AL.—VARIATION IN GRAY WHALE BODY CONDITION 259

2004). Therefore, in addition to lactating females, other
noncalf western gray whales have the potential to leave the
study area with less than optimal energy stores.
The body condition of western gray whales varied annually,
but was significantly better in 2004 and significantly worse
in 1999, 2006, and 2007 (Table 3). Eastern gray whales
experienced a high-mortality event in 1999 and 2000 that
may have been caused by reductions in prey productivity
brought on by short- and long-term climate effects in the
North Pacific (Le Boeuf et al. 2000; Moore et al. 2001),
leading Brownell and Weller (2001) to suggest an oceano-
graphic link between the eastern gray whale mortality event
and concurrent observations of western gray whales in
relatively worse body condition. However, assuming males
are most reflective of annual environmental conditions (Pettis
et al. 2004), a post hoc analysis of the body condition of
western gray whale noncalf males revealed that only the 2004
and 2007 year effects were maintained. This difference
resulting from the exclusion of females implicates the
interactions that can occur between the reproductive cycle
and environmental variability (Lockyer 1987). Overall, the
characteristics (e.g., prey quantity and quality and ice cover)
of the years identified as significant in this study have not been
evaluated and warrant additional attention.
Interannual variation in the energy reserves of whales has
been previously detected and correlated with both prey
availability (e.g., Ichii et al. 1998) and whale fecundity (e.g.,
Lockyer 1986). In that regard, a primary way environmental
and associated foraging conditions affect population demog-
raphy is by influencing maternal body condition and
subsequent reproductive success (Le Boeuf and Crocker
2005), as has been demonstrated for North Atlantic right
whales (Greene et al. 2003). The body condition of lactating
female western gray whales was estimated to vary by year
(Fig. 5a). However, the body condition of reproductive female
whales likely exhibits complex and asynchronous dynamics
that are a function of previous calf production and environ-
mental factors. Thus, interannual variation in the body
condition of reproductive female western gray whales merits
a more thorough investigation.
Although a recent development in whale research, the use of
visual body condition assessment methods is not new to
animal ecology (e.g., Riney 1960; Robinson 1960). Visual
determinations of body condition have been shown to
successfully correlate with quantitative measures of energy
stores for a variety of mammalian species (Kistner et al. 1980;
Prestrud and Pond 2003; Stephenson et al. 2002; Stirling et al.
2008), including whales (Angell 2006), an important valida-
tion for any index of body condition (Schulte-Hostedde et al.
2005). Further, in cases such as free-ranging baleen whales,
where a variety of components in a wide range of tissues
reflect long-term energy reserves that cannot be comprehen-
sively quantified, a more holistic assessment of relative body
mass might offer some advantage over enumerating a specific
measure of energy storage. That is, because energy is
deposited in a number of forms and locales, which can vary
according to age, sex, and reproductive status (Lockyer 1987),
it could be limiting and potentially problematic to focus on
1 measurable attribute (e.g., blubber thickness—Aguilar et al.
2007).
In our study, differences in the relative amount of
subcutaneous fat were detected collectively for the postcranial
area, scapular region, and lateral flanks and presumed to
reflect individual body condition. It is possible that body mass
in these areas does not in fact correspond to important energy
reserves, a problematic lack of correlation prevalent in the use
of body condition indexes (Hayes and Shonkwiler 2001).
However, findings of the analysis, particularly the compro-
mised body condition of lactating females and the monthly
improvement in noncalf body condition (Table 3; Figs. 5a and
5c), are consistent with well-supported patterns of mammalian
and baleen whale life history, suggesting that the present
assessment protocol can indeed measure western gray whale
body condition. Results of research associated with whaling
operations have indicated that for some balaenopterid species,
the most substantial and variable, and therefore most useful,
site of lipid storage is the dorsal tail region (e.g., Lockyer et al.
1985). Thus, the anterior body regions evaluated here may not
represent the most sensitive or temporally precise gauge of
internal fat depots, although it is also plausible that mass loss
by body region may vary by species. Nevertheless, in addition
to exhibiting meaningful variation, these areas are routinely
documented during photo-identification efforts, providing an
informative and practical means to infer body condition.
This assessment quantified temporal variation in western
gray whale body condition given confirmed observations of
reproductive class and sex. We suggest the next steps in the
examination of western gray whale body condition are to
evaluate the effect of inferred reproductive states (e.g.,
pregnant, resting, or immature) on body condition, and to
explore the relationship between body condition and calving
interval, calf sex ratio, and other life-history parameters; and
between body condition and environmental indicators of food
availability and access, such as indexes of sea-ice and
oceanographic conditions. Assessing the body condition of
free-ranging eastern gray whales also is recommended because
it would allow for interpopulation comparisons, as well as
illustrate the impact of density feedback mechanisms on the
relationship between gray whale body condition and environ-
mental variability. Finally, dead eastern gray whales strand in
some numbers each year throughout their range (Le Boeuf et
al. 2000). An anatomical and biochemical evaluation of these
whales could be used to better understand subcutaneous fat
deposition in the postcranial, scapular, lateral flank, and other
body regions of gray whales.
The endogenous energy stores of mammalian capital
breeders such as baleen whales allow individuals to sustain
reproduction as well as survive periods of poor feeding,
although trade-offs are involved (Lockyer 2007). Our study
highlights linkages between the environmental conditions,
physiological constraints, and reproductive costs of western
gray whales. Further, we introduce a robust method for
260 JOURNAL OF MAMMALOGY Vol. 93, No. 1

monitoring an aspect of individual condition, which will
facilitate both following and elucidating population responses
to a changing environment. Given that gray whales can track
productivity changes at local scales and ecosystem alterations
at ocean-basin scales, they have been referred to as bio-
indicators of environmental variability (Moore et al. 2003) and
ecosystem sentinels (Moore and Huntington 2008), respec-
tively. Incorporating the role of individual fitness is important
for achieving a mechanistic view of these paradigms.
Fowler (1984) reported that cetacean populations are
regulated through density-dependent changes in reproduction
and survival that are a function of food resources. Others have
argued that populations are not regulated by density-
dependence but are limited by environmental capacity (e.g.,
White 2004) or that the 2 perspectives are indistinguishable
(e.g., Berryman 2004). Whether ecologists will agree on
regulation or limitation as the driver of population dynamics,
it is clear that more effort is needed to identify ecological
factors and mechanisms that affect individuals and ultimately
population abundance and growth rate (Krebs 2002). The
observed variation in western gray whale body condition
indicates fundamental environmental and physiological inter-
actions that will influence the productivity of the population
regardless of its size, although a consideration of size is
critical in a conservation context. That is, environmental
variability can increase extinction risk in small populations
(Stacey and Taper 1992), as well as compound the impact of
demographic stochasticity and other small-population effects
that may be contributing to the dynamics of the critically
endangered western gray whale population.
ACKNOWLEDGMENTS
We thank the many individuals who have provided field assistance
over the years, especially S. Blokhin, H. W. Kim, A. Lang,
S. Rickards, and G. Tsidulko. Participants in the 2006 NOAA Large
Whale Health Assessment Workshop contributed valuable feedback
on the body condition assessment protocol. N. Ellis, P. Heagerty, and
R. Christensen offered useful advice on the statistical analysis. The
edits of W. Perrin and 2 anonymous reviewers improved the
manuscript. Support for ALB was funded in part by a grant from
the Washington Sea Grant Program, University of Washington,
pursuant to National Oceanic and Atmospheric Administration
Graduate Fellowship Program in Population Dynamics and Marine
Resource Economics. The views expressed herein are those of the
authors and do not necessarily reflect the views of National Oceanic
and Atmospheric Administration or any of its subagencies. Support
and funding for the Russia–United States western gray whale research
project were provided (in alphabetical order) by Alaska SeaLife
Center, Exxon Neftegas Limited, the International Fund for Animal
Welfare, the International Whaling Commission, the United States
Marine Mammal Commission, the Marine Mammal Research
Program at Texas A&M University at Galveston, the National Fish
and Wildlife Foundation, the United States National Marine Fisheries
Service, Ocean Park Conservation Foundation, Sakhalin Energy
Investment Company, the School of Aquatic and Fishery Sciences at
the University of Washington, the United States Environmental
Protection Agency, and the Washington Cooperative Fish and
Wildlife Research Unit. The project was conducted as part of the
Marine Mammal Project under Area V: Protection of Nature and the
Organization of Reserves within the United States–Russia Agreement
on Cooperation in the Field of Environmental Protection.
LITERATURE CITED
AGRESTI, A. 2002. Categorical data analysis. John Wiley & Sons, Inc.,
Hoboken, New Jersey.
AGUILAR, A., A. BORRELL,AND E. GO
´MEZ-CAMPOS. 2007. The reliability
of blubber thickness as a measure of body condition in large
whales. Paper SC/59/O17 presented to the International Whaling
Commission Scientific Committee, June 2007, Anchorage, Alaska.
ANGELL, C. M. 2006. Body fat condition of free-ranging right whales,
Eubalaena glacialis and Eubalaena australis. Ph.D. dissertation,
Boston University, Boston, Massachusetts.
ATKINSON, S. N., AND M. A. RAMSAY. 1995. The effects of prolonged
fasting of the body composition and reproductive success of female
polar bears (Ursus maritimus). Functional Ecology 9:559–567.
BATZLI, G. O., AND E. ESSEKS. 1992. Body fat as an indicator of
nutritional condition for the brown lemming. Journal of Mammal-
ogy 73:431–439.
BELONJE, P. C., AND C. H. VAN NIEKERK. 1975. A review of the
influence of nutrition upon the oestrous cycle and early pregnancy
in the mare. Journal of Reproduction and Fertility 23:167–169.
BERNARDO, J. 1996. Maternal effects in animal ecology. American
Zoologist 36:83–105.
BERRYMAN, A. A. 2004. Limiting factors and population regulation.
Oikos 105:667–670.
BLOKHIN, S. A. 2004. Terms of grey whales’ (Eschrichtius robustus)
appearance near the northeastern Sakhalin, their abundance and
behavior in the beginning and in the end of their stay in the area
of Piltun Bay. Pp. 78–82 in Marine mammals of the Holarctic,
collection of scientific papers (V. M. Belkovich, ed.). 3rd
International Conference, Koktebel, Crimea, Ukraine, 11–17
October 2004.
BRADFORD, A. L., D. W. WELLER,Y.V.IVASHCHENKO,A.M.BURDIN,
AND R. L. BROWNELL,JR. 2009. Anthropogenic scarring of western
gray whales (Eschrichtius robustus). Marine Mammal Science
25:161–175.
BRADFORD, A. L., D. W. WELLER,P.R.WADE,A.M.BURDIN,AND
R. L. BROWNELL,JR. 2008. Population abundance and growth rate
of western gray whales Eschrichtius robustus. Endangered Species
Research 6:1–14.
BRADFORD, A. L., ET AL. 2006. Survival estimates of western gray
whales Eschrichtius robustus incorporating individual heterogene-
ity and temporary emigration. Marine Ecology Progress Series
315:293–307.
BROWNELL, R. L., JR., AND D. W. WELLER. 2001. Is the ‘‘carrying
capacity hypothesis’’ a plausible explanation for the ‘‘skinny’’
gray whale phenomenon? Paper SC/53/BRG20 presented to the
International Whaling Commission Scientific Committee, July
2001, London, United Kingdom.
BROWNELL, R. L., JR., AND D. W. WELLER. 2002. Prolonged calving
intervals in western gray whales: nutritional stress and pregnancy.
Paper SC/54/BRG12 presented to the International Whaling
Commission Scientific Committee, April 2002, Shimonoseki, Japan.
BROWNELL, R. L., JR., ET AL. 1997. Observations on Okhotsk–Korean
gray whales on their feeding grounds off Sakhalin Island. Report of
the International Whaling Commission 47:161–162.
CAMERON, R. D., W. T. SMITH,S.G.FANCY,K.L.GERHART,AND R. G.
WHITE. 1993. Calving success of female caribou in relation to body
weight. Canadian Journal of Zoology 71:480–486.
February 2012 BRADFORD ET AL.—VARIATION IN GRAY WHALE BODY CONDITION 261

CHAMAILLE
´-JAMMES,S.,H.FRITZ,M.VALEIX,F.MURINDAGOMO,AND
J. CLOBERT. 2008. Resource varia bi li ty, a ggre gation and direct density
dependence in an open context: the local regulation of an African
elephant population. Journal of Animal Ecology 77:135–144.
CHRISTENSEN, R. H. B. 2010. Ordinal: regression models for ordinal
data. R package version 2010.06-12. cran.r-project.org/web/
packages/ordinal/index.html. Accessed 30 June 2010.
CICCHETTI, D. V., AND A. R. FEINSTEIN. 1990. High agreement but low
kappa: II. Resolving the paradoxes. Journal of Clinical Epidemi-
ology 43:551–558.
COHEN, J. 1960. A coefficient of agreement for nominal scales.
Educational and Psychological Measurement 20:37–46.
COHEN, J. 1968. Weighted kappa: nominal scale agreement with
provision for scaled disagreement or partial credit. Psychological
Bulletin 70:213–220.
COOK, R. C., J. G. COOK,AND L. D. MECH. 2004. Nutritional condition
of northern Yellowstone elk. Journal of Mammalogy 85:714–722.
COSTA, D. P., AND T. M. WILLIAMS. 1999. Marine mammal energetics.
Pp. 176–217 in Biology of marine mammals (J. E. Reynolds III and
S. A. Rommel, eds.). Smithsonian Institution Press, Washington, D.C.
DARLING, J. D. 1984. Gray whales off Vancouver Island, British
Columbia. Pp. 267–287 in The gray whale Eschrichtius robustus
(M. L. Jones and S. L. Swartz, eds.). Academic Press, Orlando,
Florida.
FEINSTEIN, A. R., AND D. V. CICCHETTI. 1990. High agreement but low
kappa: I. The problems of the two paradoxes. Journal of Clinical
Epidemiology 43:543–549.
FOWLER, C. W. 1984. Density dependence in cetacean populations.
Report of the International Whaling Commission Special Issue
6:373–379.
GREENE, C. H., A. J. PERSHING,R.D.KENNEY,AND J. W. JOSSI. 2003.
Impact of climate variability on the recovery of endangered North
Atlantic right whales. Oceanography 16:98–103.
HAUCK, W. W., JR., AND A. DONNER. 1977. Wald’s test as applied to
hypotheses in logit analysis. Journal of the American Statistical
Association 72:851–853.
HAYES, J. P., AND J. S. SHONKWILER. 2001. Morphometric indicators of
body condition: worthwhile or wishful thinking? Pp. 8–38 in Body
composition analysis of animals: a handbook of non-destructive
methods (J. R. Speakman, ed.). Cambridge University Press,
Cambridge, United Kingdom.
HICKLING, G. J., J. S. MILLAR,AND R. A. MOSES. 1991. Reproduction
and nutrient reserves of bushy-tailed wood rats (Neotoma cinerea).
Canadian Journal of Zoology 69:3088–3092.
HOSMER, D. W., AND S. LEMESHOW. 2000. Applied logistic regression.
John Wiley & Sons, Inc., Hoboken, New Jersey.
ICHII, T., N. SHINOHARA,Y.FUJISE,S.NISHIWAKI,AND K. MATSUOKA.
1998. Interannual changes in body fat condition index of minke
whales in the Antarctic. Marine Ecology Progress Series 175:1–12.
INTERNATIONAL UNION FOR CONSERVATION OF NATURE. 2010. IUCN Red
list of threatened species, version 2010.4. www.iucnredlist.org.
Accessed 1 December 2010.
JAKOB, E. M., S. D. MARSHALL,AND G. W. UTEZ. 1996. Estimating
fitness: a comparison of body condition indices. Oikos 77:61–67.
JONES, M. L. 1990. The reproductive cycle in gray whales based on
photographic resightings of females on the breeding grounds from
1977–82. Report of the International Whaling Commission Special
Issue 12:177–182.
KISTNER, D. R., C. E. TRAINER,AND N. A. HARTMANN. 1980. A field
technique for evaluating physical condition of deer. Wildlife
Society Bulletin 8:11–17.
KREBS, C. J. 2002. Two complementary paradigms for analyzing
population dynamics. Philosophical Transactions of the Royal
Society of London, B. Biological Sciences 357:1211–1219.
LANG, A. R. 2010. The population genetics of gray whales
(Eschrichtius robustus) in the North Pacific. Ph.D. dissertation,
University of California, San Diego.
LEBOEUF, B. J., AND D. E. CROCKER. 2005. Ocean climate and seal
condition. BMC Biology 3:1–10.
LEBOEUF, B. J., M. H. PE
´REZ-CORTE
´S,R.J.URBA
´N,B.R.MATE,AND
U. F. OLLERVIDES. 2000. High gray whale mortality and low
recruitment in 1999: potential causes and implication. Journal of
Cetacean Research and Management 2:85–99.
LINDSTRO
¨M, J. 1999. Early development and fitness in birds and
mammals. Trends in Ecology & Evolution 14:343–348.
LOCKYER, C. 1984. Review of baleen whale (Mysticeti) reproduction
and implications for management. Report of the International
Whaling Commission Special Issue 6:27–50.
LOCKYER, C. 1986. Body fat condition in northeast Atlantic fin
whales, Balaenoptera physalus, and its relationship with repro-
duction and food resource. Canadian Journal of Fisheries and
Aquatic Sciences 43:142–147.
LOCKYER, C. 1987. The relationship between body fat, food resource
and reproductive energy costs in north Atlantic fin whales
(Balaenoptera physalus). Symposia of the Zoological Society of
London 57:343–361.
LOCKYER, C. 2007. All creatures great and smaller: a study in cetacean
life history energetics. Journal of the Marine Biological Associ-
ation of the United Kingdom 87:1035–1045.
LOCKYER, C., L. C. MCCONNELL,AND T. D. WATERS. 1985. Body
condition in terms of anatomical and biochemical assessment of
body fat in North Atlantic fin and sei whales. Canadian Journal of
Zoology 63:2328–2338.
LOUDON, A. S. I., A. S. MCNEILLY,AND J. A. MILNE. 1983. Nutrition
and lactational control of fertility in red deer. Nature 302:145–147.
MILLAR, J. S. 1977. Adaptive features of mammalian reproduction.
Evolution 31:370–386.
MONSON, D. H., J. A. ESTES,J.L.BODKIN,AND D. B. SINIFF. 2000. Life
history plasticity and population regulation in sea otters. Oikos
90:457–468.
MOORE, S. E., AND H. P. HUNTINGTON. 2008. Arctic marine mammals
and climate change: impacts and resilience. Ecological Applica-
tions 18:S157–S165.
MOORE, S. E., ET AL. 2001. Are gray whales hitting ‘‘K’’ hard? Marine
Mammal Science 17:954–958.
MOORE, S. E., J. M. GREBMEIER,AND J. R. DAVIES. 2003. Gray whale
distribution relative to forage habitat in the northern Bering Sea:
current conditions and retrospective summary. Canadian Journal of
Zoology 81:734–742.
PERRYMAN, W. L., M. A. DONAHUE,P.C.PERKINS,AND S. B. REILLY.
2002. Gray whale calf production 1994–2000: are observed
fluctuations related to changes in seasonal ice cover? Marine
Mammal Science 18:121–144.
PERRYMAN, W. L., AND M. S. LYNN. 2002. Evaluation of nutritive
condition and reproductive status of migrating gray whales
(Eschrichtius robustus) based on analysis of photogrammetric
data. Journal of Cetacean Research and Management 4:155–
164.
PETTIS, H. M., R. M. ROLLAND,P.K.HAMILTON,S.BRAULT,A.R.
KNOWLTON,AND S. D. KRAUS. 2004. Visual health assessment of
North Atlantic right whales (Eubalaena glacialis) using photo-
graphs. Canadian Journal of Zoology 82:8–19.
262 JOURNAL OF MAMMALOGY Vol. 93, No. 1

PRESTRUD, P., AND C. M. POND. 2003. Fat indices of arctic foxes
Alopex lagopus in Svalbard. Wildlife Biology 9:193–197.
PUNT, A. E., AND P. R. WADE. 2010. Population status of the eastern
North Pacific stock of gray whales in 2009. United States
Department of Commerce, National Oceanic and Atmospheric
Administration Technical Memorandum NMFS-AFSC-207:1–43.
RD
EVELOPMENT CORE TEAM. 2010. R: a language and environment for
statistical computing. R Foundation for Statistical Computing.
www.r-project.org. Accessed 30 June 2010.
REYNOLDS, D. S., AND T. H. KUNZ. 2001. Standard methods for
destructive body composition analysis. Pp. 39–55 in Body
composition analysis of animals: a handbook of non-destructive
methods (J. R. Speakman, ed.). Cambridge University Press,
Cambridge, United Kingdom.
RICE, D. W. 1983. Gestation period and fetal growth of the gray whale.
Report of the International Whaling Commission 33:539–544.
RICE, D. W., AND A. A. WOLMAN. 1971. The life history and ecology
of the gray whale (Eschrichtius robustus). Special Publication 3,
The American Society of Mammalogists.
RINEY, T. 1960. A field technique for assessing physical condition of
some ungulates. Journal of Wildlife Management 24:92–94.
ROBBINS, C. T. 1993. Wildlife feeding and nutrition. Academic Press,
New York.
ROBINSON, W. L. 1960. Test of shelter requirements of penned white-
tailed deer. Journal of Wildlife Management 24:364–371.
SCHULTE-HOSTEDDE, A. I., B. ZINNER,J.S.MILLAR,AND G. J. HICKLING.
2005. Restitution of mass-size residuals: validating body condition
indices. Ecology 86:155–163.
SIKES, R. S., W. L. GANNON,AND THE ANIMAL CARE AND USE COMMITTEE
OF THE AMERICAN SOCIETY OF MAMMALOGISTS. 2011. Guidelines of
the American Society of Mammalogists for the use of wild
mammals in research. Journal of Mammalogy 92:235–253.
SIMONOFF, J. S. 2003. Analyzing categorical data. Springer-Verlag,
New York.
SPEAKMAN,J.R.(ED.). 2001. Body composition analysis of animals:
a handbook of non-destructive methods. Cambridge University
Press, Cambridge, United Kingdom.
STACEY, P. B., AND M. TAPER. 1992. Environmental variation and the
persistence of small populations. Ecological Applications 2:18–29.
STEPHENSON, T. R., V. C. BLEICH,B.M.PIERCE,AND G. P. MULCAHY.
2002. Validation of mule deer body composition using in vivo and
post-mortem indices of nutritional condition. Wildlife Society
Bulletin 30:557–564.
STIRLING, I., G. W. THIEMANN,AND E. RICHARDSON. 2008. Quantitative
support for a subjective fatness index of immobilized polar bears.
Journal of Wildlife Management 72:568–574.
SWARTZ, S. L., AND M. L. JONES. 1983. Gray whale (Eschrichtius
robustus) calf production and mortality in the winter range. Report
of the International Whaling Commission 33:503–507.
THOMAS, V. G. 1990. Control of reproduction in animal species with
high and low body fat reserves. Pp. 27–41 in Adipose tissue and
preproduction. Progress in reproductive biology and medicine
(R. E. Frisch, ed.). S. Karger AG, Basel, Switzerland.
Vı
´KINGSSON, G. A. 1990. Energetic studies on fin and sei whales
caught off Iceland. Report of the International Whaling Commis-
sion 40:365–373.
WANG, P. 1984. Distribution of the gray whale (Eschrichtius
gibbosus) off the coast of China. Acta Theriologica Sinica 4:21–26.
WAUTERS, L. A., S. A. DE CROMBRUGGHE,N.NOUR,AND E. MATTHYSEN.
1995. Do female roe deer in good condition produce more sons than
daughters? Behavioral Ecology and Sociobiology 37:189–193.
WELLER, D. W., A. M. BURDIN,B.WU
¨RSIG,B.L.TAYLOR,AND R. L.
BROWNELL,JR. 2002. The western gray whale: a review of past
exploitation, current status, and potential threats. Journal of
Cetacean Research and Management 4:7–12.
WELLER, D. W., ET AL. 1999. Gray whales (Eschrichtius robustus) off
Sakhalin Island, Russia: seasonal and annual patterns of occur-
rence. Marine Mammal Science 15:1208–1227.
WELLER, D. W., ET AL. 2008. Photographic match of a western gray
whale between Sakhalin Island, Russia and Honshu, Japan: first
link between feeding ground and migratory corridor. Journal of
Cetacean Research and Management 10:89–91.
WHITE, T. C. R. 2004. Limitation of populations by weather-driven
changes in food: a challenge to density-dependent regulation.
Oikos 105:664–666.
WIRSING, A. J., T. D. STEURY,AND D. L. MURRAY. 2002. Noninvasive
estimation of body composition in small mammals: a comparison
of conductive and morphometric techniques. Physiological and
Biochemical Zoology 75:489–497.
WORTHY,G.A.J.,AND D. M. LAVIGNE. 1987. Mass loss, metabolic
rate, and energy utilization by harp and gray seals during the
postweaning fast. Physiological Zoology 60:352–364.
YOUNG, R. A. 1976. Fat, energy and mammalian survival. American
Zoologist 16:699–710.
Submitted 14 March 2011. Accepted 9 August 2011.
Associate Editor was William F. Perrin.
APPENDIX I
Report of an interrater agreement study that evaluated the
comparability of results produced by 2 trained researchers using the
body region scoring protocol developed to assess the body condition
of western gray whales (Eschrichtius robustus).
We conducted an interrater agreement study to assess if the body
region scoring protocol used to determine western gray whale body
condition can be utilized by more than 1 qualified researcher to
achieve similar results. To this end, a 2nd trained analyst (YVI)
reviewed images from a subset of 300 randomly selected sightings
and scored the 3 body regions of interest (i.e., postcranial area,
scapular region, and lateral flanks) for the 98 individual whales
represented in the subset. We then compared scored data from the 2
researchers in 2 separate tests of interrater agreement for each of the 3
body regions. First, we evaluated the decision to code each body
region as visible (non-X) or not visible (X) for each of the 300
sightings. Next, when both analysts coded the body region as visible
for a sighting, we assessed agreement in the assigned numerical
postcranial (P), scapular (S), and lateral flank (L) condition scores.
We measured interrater agreement using the kappa (k) coefficient
(Cohen 1960), where k.0.75 is strong agreement, 0.75 .k.0.40
is good-to-moderate agreement, and k,0.40 is fair-to-poor
agreement (e.g., Simonoff 2003). Given that we scored the
postcranial condition using a 3-point ordinal scale, we employed a
weighted kappa (k
w
) coefficient (Cohen 1968) with linear weighting
in this case. For each of the 6 tests, we expected the underlying
prevalence of the observed entities to be imbalanced (e.g., more P
scores of 3 than 2 and 1), which can lead to low values of kdespite
relatively high values of total agreement (Feinstein and Cicchetti
1990). Therefore, for each test, we computed the proportional
agreement (p) of each unit (e.g., p
P3
,p
P2
, and p
P1
) along with kas a
February 2012 BRADFORD ET AL.—VARIATION IN GRAY WHALE BODY CONDITION 263

recommended supplemental diagnostic (Cicchetti and Feinstein
1990).
For the decision to code the postcranial area as visible or not
visible, kindicates good agreement and both values of pare high
(k50.67, p
Pnon-X
50.85, p
PX
50.83, n5300). For the postcranial
condition score assigned when both raters coded the region as visible,
k
w
denotes good agreement, while pis high for scores of 3 and only
somewhat reduced for scores of 2 and 1 (k
w
50.65, p
P3
50.88, p
P2
50.64, p
P1
50.63, n5135). For the choice to code the scapular
region as visible or not visible, kshows moderate agreement and both
p-values are high (k50.58, p
Snon-X
50.74, p
SX
50.83, n5300).
For the scapular condition score assigned when both researchers
coded the region as visible, kdemonstrates good agreement and p-
values are high, particularly for scores of 2 (k50.69, p
S2
50.94, p
S1
50.76, n589). For the judgment to code the lateral flanks as visible
or not visible, kreveals moderate agreement and both values of pare
high (k50.59, p
Lnon-X
50.82, p
LX
50.76, n5300). For the lateral
flank condition score assigned when both analysts coded the region as
visible, kspecifies strong agreement and p-values are high, especially
for scores of 2 (k50.83, p
L2
50.98, p
L1
50.85, n5141).
Observed entities were imbalanced only in the 3 tests involving
numerical condition scores. In these cases, values of pare higher for
the more prevalent observation (i.e., the score indicating best
condition) within each test.
Interrater agreement within the 6 tests was strong to moderate as
measured by k. Agreement was weakest for the choice to code each
of the 3 body regions as visible or not visible. A closer examination
of the decisions made by each rater revealed that 1 analyst reliably
coded each body region as visible more frequently than the other
analyst, suggesting slightly different, but consistent, interpretations of
the body position and photographic extent and quality needed to
assess body region condition. Determining the visibility of the
scapular region and lateral flanks can be challenging, requiring the
additional consideration of how much of the body is submerged,
which is possibly reflected in the reduced kcoefficients for those
regions. Agreement was strongest when assigning a numerical
condition score to mutually visible body regions. Further, kin these
cases is likely biased low given the imbalance in prevalence of the
observed entities (Feinstein and Cicchetti 1990), a suggestion that is
generally supported by values of the pdiagnostic. Unsurprisingly,
agreement was highest when assigning the scapular and lateral flank
condition scores, because these regions were scored on a 2-point
scale. Overall, findings of the interrater agreement study suggest that
although the sets of sightings with visible body regions identified by
multiple researchers may vary marginally in size, the numerical
scores assigned to these regions will be similar. Thus, the western
gray whale body region scoring protocol can be used by more than 1
trained researcher to achieve comparable results.
APPENDIXES II–IV OVERVIEW
Overview of sensitivity analyses conducted to confirm that month
was an appropriate and feasible scale at which to collapse the
numerical body region condition scores for the body condition
assessment of western gray whales (Eschrichtius robustus).
In general, each survey sighting of a western gray whale did not
result in a comprehensive set of images that allowed us to assign a
numerical score to each of the 3 body regions of interest (i.e.,
postcranial area, scapular region, and lateral flanks). Consequently, we
could not produce an overall individual determination of body
condition on a per-sighting basis. Additionally, a body condition
determination made for a single sighting might be too sensitive to the
effects of body position and other factors that can confound the body
region scoring process. Thus, it was necessary to collapse the scored
data for each sighting so that we could generate robust composites of
postcranial, scapular, and lateral flank condition. Specifically, we
needed to collapse the scored data at a scale that would be large enough
to maximize the use of intermittent sightings lacking the full suite of
body region condition scores, be small enough to minimize detectable
transitions from one score to the next, and allow for temporal
comparisons between annual field seasons. A preliminary assessment
of the scored body region data suggested that month would be a useful,
appropriate, and feasible scale at which to aggregate these data. We
conducted 2 sensitivity analyses to evaluate this decision.
Analysis 1: body region score transitions.—The objective of the 1st
analysis was to determine if the numerical scores assigned to each of
the 3 body regions changed within each month of the study (July,
August, and September). Accordingly, we used logistic regression to
model the effect of the interaction between the categorical variable
month and the continuous variable date on the body region condition
score as a categorical response. We treated individual whales as
random effects. Given the 3-point ordinal scale applied to the
postcranial area, we utilized the proportional odds model formulation
(e.g., Agresti 2002) in this case. From the 5,007 survey sightings of 168
western gray whales photo-identified between July and September of
1997–2007, 2,337 numerical (i.e., non-X) postcranial condition scores
from 165 individual whales (median of 9 scores per whale, range 1–72
scores), 2,091 scapular scores from 165 whales (median of 8, range 1–
62 scores), and 2,790 lateral flank scores from 167 whales (median of
11, range 1–75 scores) were available for this analysis.
Consistent with findings from the analysis of the body condition
determinations (see Table 3), results from the 3 model runs indicate
that condition in each body region improved as the field season
progressed (i.e., by month), although significant recovery in the
scapular region and lateral flanks was not detected until September
(Appendix II). However, significant improvements in body region
condition were not observed within each month, with the exception of
the scapular region in September (Appendix II). Given that the
overall body condition determinations were based primarily on the
postcranial condition (see ‘‘Materials and Methods: Body condition
assessment,’’ for explanation), and that the within-September
recovery of the scapular region was not highly significant, we
concluded that month was an appropriate scale to aggregate the
scored body region data for interannual comparisons, because it was
robust to detectable transitions between consecutive condition scores.
Collapsing the scored body region data.—We established a set of
hierarchical decision rules to guide the process of collapsing the scored
data into monthly determinations of postcranial, scapular, and lateral
flank condition for each whale. Appendix III presents these decision
rules and the frequency of their use. The 2,337 numerical postcranial
scores resulted in 1,133 monthly determinations of postcranial
condition, of which 1,010 (89.1%) are based on sightings that shared
the same numerical score during the month (i.e., decision rule A was
applied; Appendix III). Similarly, we collapsed the 2,091 scapular
scores into 1,035 monthly scapular determinations, with 953 (92.1%)
based on decision rule A, and the 2,790 lateral flank scores into 1,1,67
monthly lateral flank determinations, with 1,099 (94.2%) based on
decision rule A (Appendix III). In other words, most of the monthly
determinations of body region condition reflect no variation in
numerical scores assigned within the month, which likely is at least
partially explained by the aforementioned lack of detectable transitions
between adjacent scores at a monthly scale, but could also be a function
264 JOURNAL OF MAMMALOGY Vol. 93, No. 1

of the timing and number of monthly sightings. Therefore, we formed
most of the overall body condition determinations from body region
composites characterized by no within-month variation.
Analysis 2: effect of within-month variation.—The 2nd sensitivity
analysis examined whether incorporating body condition determina-
tions made from body region composites with possible within-month
variation (i.e., we applied decision rules B–F; Appendix III) would
refine or confound the statistical analysis of western gray whale body
condition. To this end, we compared the analysis of the full set of
body condition determinations (described in the main text, see
Tables 2 and 3 for results) to an identical analysis performed using
only the body condition determinations resulting from body regions
composites based on decision rule A. Note that composites where we
coded the scapular region or lateral flanks, or both, as X were
included in the sensitivity analysis, as long as we made the associated
postcranial condition determinations using decision rule A. Of the
1,133 determinations of known body condition (i.e., good, fair, or
poor), 929 (82.0%) met the composite specification criteria for the
sensitivity analysis. Identical to the full analysis, we employed
ordinal logistic regression (in the form of the proportional odds
model) in the sensitivity analysis to evaluate the effect of year,
month, reproductive class, and sex on the body condition of
individual whales, which we regarded as random effects.
As in the model selection procedure of the full analysis (Table 2),
likelihood ratio tests revealed that the model incorporating year, month,
and class as covariates is the most-parsimonious. Results of fitting this
model to the reduced set of body condition determinations (Appendix
IV) are equivalent to findings from the full analysis (Table 3).
Specifically, body condition was significantly worse in 1999, 2006,
and 2007 and significantly better in 2004 relative to the reference year of
1997. Further, body condition improved significantly with each month
of the field season, whereas lactating females were in significantly worse
body condition and calves in significantly better body condition as
compared to other whales. The comparable results of the 2 analyses and
the smaller standard errors associated with the predictor coefficients of
the full analysis indicate that utilizing the full set of known body
condition determinations refined the statistical analysis. Additionally,
the complementary nature of the analyses suggests that our method of
handling within-month variation when aggregating the scored body
region data (Appendix III) was reasonable. In summary, most
observations of body region condition did not vary within the month,
but we found a suitable means for collapsing scores when there was
variation. Thus, month offered a feasible, in addition to appropriate,
scale at which to aggregate the numerical body region condition scores
for the western gray whale body condition assessment.
APPENDIX II
Maximum-likelihood estimates resulting from fitting logistic
regression mixed models to numerical scores of western gray whale
(Eschrichtius robustus) postcranial, scapular, and lateral flank
condition, given an interaction between month and date. We used
the proportional odds formulation to model condition in the
postcranial area. For each body region, the 1st row(s) represents the
model intercept(s) and the rest predictor coefficients, with month 5
July serving as the reference category. Significant predictor
coefficients (P,0.05) are shown in boldface type. SE 5standard
error.
Body region Variable Estimate SE Wald zP-value
Postcranial Y2 2.845 0.351 8.112 ,0.001
Y320.544 0.341 21.594 0.111
Month =Aug. 0.904 0.321 2.818 0.005
Month =Sep. 2.169 0.346 6.266
,
0.001
Date 0.021 0.012 1.780 0.075
Aug.:Date 0.007 0.015 0.498 0.618
Sep.:Date 0.022 0.019 1.186 0.236
Scapular Y2 1.180 0.436 2.706 0.007
Month 5Aug. 0.222 0.439 0.505 0.614
Month =Sep. 1.302 0.467 2.791 0.005
Date 0.003 0.017 0.150 0.880
Aug.:Date 0.032 0.020 1.577 0.115
Sep.:Date 0.053 0.026 2.068 0.039
Lateral flank Y2 2.501 0.493 5.077 ,0.001
Month 5Aug. 0.397 0.425 0.935 0.350
Month =Sep. 1.649 0.473 3.483
,
0.001
Date 0.009 0.016 0.589 0.556
Aug.:Date 0.035 0.020 1.788 0.074
Sep.:Date 0.048 0.027 1.764 0.078
APPENDIX III
Hierarchical decision rules (DRs) used to collapse the scored western gray whale (Eschrichtius robustus) body region data into monthly
determinations of postcranial (P), scapular (S), and lateral flank (L) condition for each whale. Decision rules were hierarchical in the sense that
we did not consider a rule unless the previous rule(s) did not provide a resolution. An uncertain score refers to instances in which an image(s)
within a sighting suggested a different score than that indicated by the majority of images.
DR Description of numerical score selected from those available
Frequency
PSL
A Only score assigned during the month 1,010 953 1,099
B Majority score assigned when there were no uncertain scores 30 39 34
C Majority score assigned after removing any uncertain scores 57 17 14
D Score from 1st one-half of month when 10 days separate conflicting scores 8 3 4
E Score that was not .score from next month or ,score from previous month 3 2 3
F Score that was most conservative (i.e., reflected better condition) 25 21 13
February 2012 BRADFORD ET AL.—VARIATION IN GRAY WHALE BODY CONDITION 265

APPENDIX IV
Maximum-likelihood estimates resulting from fitting the
proportional odds mixed model to western gray whale (Eschrichtius
robustus) body condition determinations (formed from body region
composites reflecting no within-month variation), given year, month,
and reproductive class. The first 2 rows represent model intercepts
and the rest predictor coefficients. Note that year 51997, month 5
July, and class 5other whale served as the reference categories.
Significant predictor coefficients (P,0.05) are shown in boldface
type. SE 5standard error; LF 5lactating female.
Variable Estimate SE Wald zP-value
Yfair 3.173 0.470 6.755 ,0.001
Ygood 0.050 0.435 0.116 0.908
Year 51998 0.471 0.519 0.907 0.364
Year =1999 21.261 0.459 22.749 0.006
Year 52000 20.764 0.522 21.463 0.143
Year 52001 20.801 0.483 21.660 0.097
Year 52002 20.163 0.492 20.332 0.740
Year 52003 0.453 0.521 0.869 0.385
Year =2004 1.117 0.565 1.976 0.048
Year 52005 20.873 0.471 21.852 0.064
Year =2006 20.989 0.494 22.004 0.045
Year =2007 21.881 0.477 23.947
,
0.001
Month =Aug. 1.267 0.221 5.729
,
0.001
Month =Sep. 2.841 0.297 9.577
,
0.001
Class =LF 25.480 0.467 211.728
,
0.001
Class =calf 5.533 1.086 5.094
,
0.001
266 JOURNAL OF MAMMALOGY Vol. 93, No. 1