837
Energy Demands for Maintenance, Growth, Pregnancy,
and Lactation of Female Pacific Walruses
(Odobenus rosmarus divergens)
Shawn R. Noren1,*
Mark S. Udevitz2
Chadwick V. Jay2
1
Institute of Marine Science, University of California, Santa
Cruz, California 95060; 2US Geological Survey, Alaska
Science Center, Anchorage, Alaska 99508
Accepted 7/13/2014; Electronically Published 10/27/2014
ABSTRACT
Decreases in sea ice have altered habitat use and activity patterns of female Pacific walruses Odobenus rosmarus divergens
and could affect their energetic demands, reproductive success, and population status. However, a lack of physiological
data from walruses has hampered efforts to develop the bioenergetics models required for fully understanding potential
population-level impacts. We analyzed long-term longitudinal
data sets of caloric consumption and body mass from nine
female Pacific walruses housed at six aquaria using a hierarchical Bayesian approach to quantify relative energetic demands for maintenance, growth, pregnancy, and lactation. By
examining body mass fluctuations in response to food consumption, the model explicitly uncoupled caloric demand
from caloric intake. This is important for pinnipeds because
they sequester and deplete large quantities of lipids throughout their lifetimes. Model outputs were scaled to account for
activity levels typical of free-ranging Pacific walruses, averaging 83% of the time active in water and 17% of the time
hauled-out resting. Estimated caloric requirements ranged
from 26,900 kcal d21 for 2-yr-olds to 93,370 kcal d21 for simultaneously lactating and pregnant walruses. Daily consumption requirements were higher for pregnancy than lactation, reflecting energetic demands of increasing body size
and lipid deposition during pregnancy. Although walruses
forage during lactation, fat sequestered during pregnancy
sustained 27% of caloric requirements during the first month
of lactation, suggesting that walruses use a mixed strategy of
capital and income breeding. Ultimately, this model will aid
*Corresponding author; e-mail: snoren@biology.ucsc.edu.
Physiological and Biochemical Zoology 87(6):837–854. 2014. q 2014 by The
University of Chicago. All rights reserved. 1522-2152/2014/8706-4020$15.00.
DOI: 10.1086/678237
in our understanding of the energetic and population consequences of sea ice loss.
Introduction
Pacific walruses (Odobenus rosmarus divergens) are specialized shallow benthic foragers (Fay 1982), found primarily
in the eastern East Siberian Sea to the western Beaufort Sea
and southward into the Bering Sea from eastern Kamchatka
to Bristol Bay (Fay 1985). This area includes their foraging
grounds along the continental shelves of the Chukchi and Bering Seas (Fay and Burns 1988; Jay et al. 2011). However, their
summer habitat has been drastically altered by global climate
change. The extent of summer sea ice has decreased substantially (Meier et al. 2007), and this trend is expected to
continue (Overland and Wang 2007; Douglas 2010; Wang
et al. 2012). Because sea ice serves as a platform for Pacific
walruses to rest, nurse, and gain access to offshore foraging
grounds (Fay 1982), this environmental change may have
population-level consequences. Reductions of summer sea ice
over the continental shelf in the Chukchi Sea over the past
decade have already resulted in increased use of terrestrial
haul-outs by adult female and young walruses (Kavry et al.
2008; Jay et al. 2012). As summer sea ice continues to decline,
the number of walruses converging on coastal haul-outs and
the time they spend ashore are expected to increase (Jay et al.
2011). The ability of the localized food supply in these coastal
regions to support large numbers of walruses over the long
term is unknown (Ovsyanikov et al. 2008). Changes in access
to prey could impact body condition and ultimately impact
population growth rates (Jay et al. 2011).
Predicting the responses of walruses to changing Arctic
conditions may be facilitated by an understanding of their
energetics and food requirements. Direct measurements of
field metabolic rate (Acquarone et al. 2006) and observations
of foraging (Born et al. 2003) would provide the best information about the energetic requirements and food consumption rates of wild walruses. However, these measurements are
difficult to obtain, in part because walruses are difficult to
capture and remotely distributed. Several other approaches
have been employed to estimate food consumption in marine
mammals, including analyses of stomach contents and scat
from wild animals as well as bioenergetic modeling (Winship
et al. 2006). Bioenergetic modeling has been used to estimate
838
S. R. Noren, M. S. Udevitz, and C. V. Jay
the energy consumption of a variety of marine mammals
(e.g., Lavigne et al. 1982; Øritsland and Markussen 1990; Ryg
and Øritsland 1991; Olesiuk 1993; Hammill et al. 1997; Stenson et al. 1997; Winship et al. 2002; Winship and Trites 2003;
Noren 2011) and was recently applied to female Pacific walruses (Noren et al. 2012).
Noren et al. (2012) constructed a relatively simple bioenergetics model for female walruses because the lack of
physiological data for this species did not support a more
complex model, although this model did include the bioenergetic costs of molting and reproduction, which had largely
been ignored in bioenergetics models for other pinniped species (e.g., Olesiuk 1993; Mohn and Bowen 1996; Stenson et al.
1997). However, the parameters used to characterize the energetic costs of important life-history stages were not derived
from data from walruses. The energetic costs of molting and
reproduction were derived from data from California sea lions (Zalophus californianus; Williams et al. 2007). Likewise,
the parameter used to account for the energetic cost of early
growth was derived from data from phocid seals (Worthy
1987). A validation exercise made it evident that, at least in
some cases, the model overestimated daily energy requirements, which was likely due to the use of data from alternate
species to derive some model parameters (Noren et al. 2012).
For example, the estimated caloric demand of a lactating,
830-kg, 12-yr-old female walrus was not sustainable within
realistic limits for time to forage (Noren et al. 2012). Another
shortcoming of the model in Noren et al. (2012) was that
uncertainty in model-based estimates of energy requirements
was not quantified because the parameters were derived from
previously published values. Noren et al. (2012) recommended
directed research on walruses to determine how caloric intake
and energy stores (body mass) are linked to meet energy requirements, particularly during critical life-history stages such
as lactation, when energetic demand may be partially supported by endogenous energy reserves.
Long-term husbandry records from animals in human care
can be a valuable source of information about daily caloric
intake rates and concurrent fluctuations in body condition
(body mass) in relation to important life-history stages. Data
on the food consumption of captive pinnipeds have the advantage of being direct measurements (Winship et al. 2006).
Nonetheless, in applying studies of captive animals to wild
animals, one must ensure that the effects of age, body size,
season, and energy density of the food are considered, and
it is likely that the energy requirements of captive and wild
animals differ (Winship et al. 2006). Although there are concerns regarding the applicability of captive animal feeding
rates to wild populations, Williams et al. (2007) demonstrated
that data obtained from captive marine mammals can be used
to develop reliable indices of the relative energetic costs of
important life-history stages. In this study, we acquired historical husbandry records of caloric intake and body mass of
female Pacific walruses housed at aquaria and used them in a
Bayesian modeling approach that explicitly decoupled caloric
intake from caloric demand to estimate energetic costs of
maintenance and activity, growth, pregnancy, and lactation
as functions of age and body size. We characterized basal
(maintenance) and activity costs for the captive walruses in
terms of facility and seasonal effects and then rescaled these
to an activity budget characteristic of free-ranging Pacific walruses to estimate energy requirements for free-ranging walruses. This is the first bioenergetic model for walruses that
is based on species-specific data for growth, pregnancy, and
lactation and also accounts for uncertainty in the parameter
estimates.
Methods
Data Acquisition and Processing
Long-term longitudinal husbandry records were obtained for
nine female Pacific walruses (Odobenus rosmarus divergens)
housed in outdoor enclosures that included access to water
and land. Six public display facilities participated in this study,
including Aquarium du Quebec (Quebec City, Quebec, Canada), Brookfield Zoo (Brookfield, IL), Indianapolis Zoo (Indianapolis, IN), New York Aquarium (Brooklyn, NY), Point
Defiance Zoo and Aquarium (Tacoma, WA), and Six Flags
Discovery Kingdom (Vallejo, CA). All walruses were obtained
from the wild as young orphaned calves, and the majority of
the animals were maintained for a minimum of 15 yr. Husbandry records spanned from 0 to 30 yr of age. We considered data for ages ≥2 yr old because data for ages !2 yr old
are being used in a concurrent study on nursing walrus calf
bioenergetics. The data set included three full-term pregnancies (n p 2 individuals) and one 2-yr lactation interval.
The animals were fed a mixed diet consisting primarily
of herring, capelin, clams, and squid. Other prey types were
occasionally included, such as mackerel, sardines, smelt,
shrimp, and trout. Although these prey items may differ
somewhat from what walruses typically eat in the wild (Fay
et al. 1977; Fay and Lowry 1981; Fay and Stoker 1982a,
1982b; Sheffield and Grebmeier 2009), the digestive efficiency
of walruses does not vary with diet (Fisher et al. 1992). The
quantity (kg) of each prey type consumed, which was itemized daily by animal care staff, was multiplied by its specific
energy density (kcal kg21) to estimate the associated calories
consumed. The caloric content of each prey item was based
on the average value obtained from chemical analyses of randomly chosen food items from shipment lots, as determined
by local commercial laboratories hired by each aquarium. In
some cases, caloric content information was provided by the
fish supplier. The ingested calories from all prey types were
summed for each day to provide the value for daily caloric
intake for each animal. Daily caloric intake was not manipulated for the purpose of this study; rather, it was determined
by the husbandry staff based on the perceived requirements
and behaviors of the animals.
Body mass was measured at intervals throughout each animal’s lifetime by training the walruses to station on scales
housed at each of the facilities. For each interval spanning
Caloric Requirements of Female Walruses
successive body mass measurements, we calculated average
daily energy consumption per day (kcal d21; fig. 1A), average
mass (kg; fig. 1B), and average daily change in mass. Depending on the facility, intervals between successive body
mass measurements were typically 1 d to 1 wk; data from intervals greater than 3 wk were excluded.
Modeling Energy Consumption
We used a hierarchical Bayesian framework to model caloric
consumption as a function of metabolic costs and changes in
mass. The form of the model was
Ec ðijÞ ∼ Normal(mc (ij); jc2 );
(1)
where Ec(ij) is the average energy (kcal) consumed per day by
walrus i during interval j, with variance jc2 and mean mc(ij),
comprised of a metabolic component (mm(ij)) and a storage
component (ms(ij)), so that mc(ij) p mm(ij) 1 ms(ij). The metabolic component includes metabolic costs due to basal metabolism, activity, and life-history stage (growth, pregnancy,
and lactation) and also includes a random effect to account
for repeated measures on individual walruses. Metabolic costs
were expressed as multiples of Kleiber, which was assumed
to be the resting metabolic rate, equivalent to 70#mass0.75 kcal
d21 (Kleiber 1975). A correction factor to account for digestive
efficiency was also included. Although the heat increment of
feeding may also reduce the amount of energy available from
the diet, we did not consider this factor because we were comparing data from this study to our previous theoretical model
that did not include this factor (Noren et al. 2012). Thus, the
metabolic component is given by
mm ðijÞ p
½AðijÞ 1 Cg Ig ðijÞ 1 Cp Ip ðijÞ 1 Cl Il ðijÞ 1 Wi K1 mKij 2
;
D
(2)
where A(ij) represents base per-day metabolic cost (basal plus
activity) for walrus i during interval j. Cg represents the additional metabolic costs of growth (growth premium) for animals !ag years old, with Ig(ij) as an indicator variable having a
value of 1 for a walrus !ag years old. To determine the maximum age that accrues this additional metabolic cost of growth,
we considered a series of models with ag ranging from 2 (i.e.,
no growth increment) to 14, in increments of 1 yr. The final
value for ag was selected based on a comparison of the deviance information criterion (DIC) for these models (Spiegelhalter et al. 2002). Cp and Cl represent the metabolic costs of
pregnancy and lactation, respectively, with corresponding indicator variables for pregnancy (Ip(ij) p 1) or lactation (Il(ij) p
1). We did not include a metabolic cost for molt because the
molts of the walruses in this study were protracted compared
to those of free-ranging walruses and exploratory analyses did
not indicate additional costs during the molting interval. Wi ∼
Normal(0, jw2 ) is a random effect that accounts for repeated
measurements on walrus i. K1 p 70 and K2 p 0.75 are the
839
Kleiber constants (Kleiber 1975), mij is the average mass (kg)
for the walrus during the interval, and D is digestive efficiency
(0.944; Fisher et al. 1992).
Direct information on activity was not available for the
walruses in this study, but preliminary analyses indicated that
differences in activity could be accounted for by annual cycles
that varied among facilities. Therefore, we represented base
metabolic costs as
2pdj
2pdj
AðijÞ p fi 1 ai cos
1 bi sin
;
365
365
(3)
where A(ij) is the per-day cost at facility i during interval j and
dj is the number of days in the year up to the last day of the
interval. The value 365 was replaced by 366 in leap years.
The storage component of the model represents energy
that is stored in or drawn from body reserves when consumption is greater or less than daily caloric requirements,
respectively. We assumed changes in mass provided an index
to this storage and use of energy, so the storage component is
given by
ms ðijÞ p ½Cd1 Id1 ðijÞ 1 Cd2 Id2 ðijÞdij ;
(4)
where dij is the average daily change in mass (kg) for walrus
i during interval j and Id1 (ij) and Id2 (ij) are indicators for
whether dij is positive or negative, respectively. Multiplying by
Cd1 or Cd2 transforms changes in mass to equivalent amounts
of consumed energy. Adding this term allowed us to estimate
the parameters in the metabolic component of the model while
explicitly accounting for energy used to accumulate mass as a
result of a surplus in consumption and energy released as a
result of a deficit in consumption. Unlike effects in the metabolic component, effects in the storage component depend only
on the magnitude of the change in mass; they do not depend
on the total mass of the animal.
Preliminary analyses indicated that dij values calculated
from a smoothed time series of mass values might provide a
better index to storage and use of energy than values based
directly on observed mass values. Therefore, we considered
a series of models that used dij values calculated from unsmoothed values and from a range of less smoothed values of
mass, differing in the amount of smoothing as specified by
the smoothing degrees of freedom (df; Hastie and Tibshirani
1990). The amount of smoothing did not have a large effect on
other model parameters but primarily affected Cd1 and Cd2 ,
with more smoothing resulting in higher estimates. Preliminary analyses also indicated that, for a fixed set of covariates,
the DIC values for a series of models based on different
amounts of smoothing had a concave pattern with a minimum
that could be located by plotting DIC as a function of trial
values of df. Therefore, we started by fixing the maximum age
for the growth increment at ag p 12 (age at full growth; Fay
1982) and found the smoothing df that gave the minimum DIC
for this value of ag. We then fixed the smoothing df at this
Figure 1. Daily food consumption (A), body mass (B), and mass-specific food consumption (C) in relation to age for nine Pacific walruses
(Odobenus rosmarus divergens) housed at six aquaria. Values are daily averages for each interval spanning successive body mass measurements as described in “Methods.” Data for each individual are denoted by a unique color. Arrows denote pregnancies (all full term), and
the bar denotes a 2-yr lactation interval; colors of the arrows and bar are coordinated with the color of the data for those particular walruses.
Caloric Requirements of Female Walruses
value and found the ag value that minimized DIC for this df.
Finally, we confirmed that this smoothing df still gave the
minimum DIC value for this value of ag.
For a fully grown, nonreproductive walrus, the expected
value of daily change in mass (dij) is 0, so energy requirements
are entirely represented by the metabolic component of the
model, which increases with the mass of the animal. For a
growing walrus, the expected value of dij is positive, and the
storage component accounts for additional energy required to
increase body size. This occurs up to about age 12, when the
asymptote for mature body mass is obtained in wild Pacific
walruses (Fay 1982). These requirements are in addition to the
growth-related metabolic costs accounted for by the growth
premium for walruses !ag yr old (Cg K1 mKij 2 =D). During pregnancy, the expected value of dij is also positive, and the storage
component accounts for the energy required for fetal growth
and storage of energy reserves to be used in the subsequent
lactation period. These requirements are in addition to the
pregnancy-related metabolic costs that accrue during pregnancy (Cp K1 mKij 2 =D). During lactation, energy requirements are
given by the metabolic component of the model, which accounts for the additional lactation-related metabolic costs as
Cl K1 mKij 2 =D. The expected value of dij is negative during this
period, and the storage component of the model accounts for
energy mobilized from stored mass to supplement consumption in fueling lactation.
Estimating Model Parameters
We estimated posterior distributions for each of the parameters in the full model (1) with each combination of smoothing
df and ag values, using Markov chain Monte Carlo (MCMC;
Gelman et al. 1997) and standard, noninformative prior distributions. In each case, we used three separate chains of at
least 30,000 iterations. We assessed convergence by examining the trace for each parameter over the iterations within
chains (Spiegelhalter et al. 2003) and Gelman-Ruben statistics
for comparisons among chains (Brooks and Gelman 1998).
The final 5,000 iterations from each chain were combined to
give 15,000 iterations for estimating posterior distributions
of parameters and derived quantities. Models were compared
with DIC as described above. Final parameter estimates and
subsequent analyses were based on the model with the lowest
DIC value.
We assessed the fit of the model to the data from individual
captive walruses during their growth periods (ages 2–9) by
estimating the model-based expected consumption and comparing this to observed caloric consumption totaled over all
intervals with adequate data (i.e., with consumption data and
no more than 3 wk between successive mass measurements)
for the walrus during the period. We also assessed the fit of
the model to data from individual captive walruses during
pregnancy and lactation periods by comparing model-based
estimates of expected consumption to observed caloric consumption, totaled over all intervals with adequate data for
these periods.
841
Estimating Energy Requirements
In general, we estimated energy requirements as
^ ðijÞ p
E
^ ðijÞ 1 C
^ g Ig ðijÞ 1 C
^ p Ip ðijÞ 1 C
^ l Il ðijÞK1 mK2
½A
ij
D
^ d1 Id1 ðijÞ 1 C
^ d2 Id2 ðijÞdij ;
1 ½C
(5)
using the posterior distributions of the parameter estimates
and either actual or calculated (as described below) values for
mij and dij.
We estimated daily and average daily energy requirements
for an 830-kg, static-mass female at each facility by assuming
mij p 830 and dij p 0 and equated this to the proportion of
time a free-ranging walrus would spend active in water (vs.
hauled-out and resting) based on the relation
AðijÞ p 6.0Pw ðijÞ 1 2.2½1 2 Pw ðijÞ;
(6)
where Pw(ij) is the proportion of the time walrus i spends in
water during interval j. Equation (6) is based on the assumption used by Noren et al. (2012) that periods active in water are
associated with a metabolic cost of 6 # Kleiber (Acquarone
et al. 2006) and periods hauled-out resting are associated with
a metabolic cost of 2.2 # Kleiber (Williams et al. 2007). This
gives the proportion of time in water as Pw(ij) p [A(ij) 2 2.2]/
3.8.
To estimate energy requirements for free-ranging walruses,
we set Pw p 0.83 based on observed activity budgets of freeranging walruses in the Bering Sea (Udevitz et al. 2009),
which was also assumed by Noren et al. (2012). This gave a
value of A(ij) p 5.4 for free-ranging walruses. We estimated
average mass-at-age values for a free-ranging female walrus,
using least squares regression to fit a logistic model of the
form
mij p
v1
1 1 exp v2 1 v3 aij
(7)
to mass (mij) at age (aij) values obtained from Fay (1982,
fig. 11). These values were used for mij to calculate corresponding daily changes in mass (dij) and to estimate caloric requirements using equation (5). We partitioned requirements into
the component due to basal plus activity costs (associated with
A(ij)), the component due to the metabolic cost of growth (associated with Cg), and the component due to mass gain (associated with Cd1 ).
We estimated the pattern of mass gain during pregnancy
by fitting a hierarchical logistic model to data from the two
pregnancies with consistently recorded data. We defined
pregnancy as the time from implantation of the blastocyst
to birth of the calf (assumed to be 334 d, approximately
11 mo). The model had the form
Mij ∼ Normal mij ; jM2 ;
(8)
where Mij is the mass gain of walrus i on day j of the pregnancy
and mij is the same as equation (7) except that aij now represents the number of days into the pregnancy (aij p 1, . . . ,
334), v1 ∼ Normal(m1, j12 ), v2 ∼ Normal(m2, j22 ), and v3 ∼
842
S. R. Noren, M. S. Udevitz, and C. V. Jay
Normal(m3, j32 ). Posterior estimates of the model parameters
were obtained using MCMC as described above for the consumption model. We used the hierarchical median of Mij added
to a starting mass of 830 kg for the mij to calculate the corresponding dij and to estimate the associated caloric requirements using equation (5) with A(ij) p 5.4 for a free-ranging
12-yr-old walrus during pregnancy. We partitioned requirements into components due to basal plus activity costs (associated with A(ij)), metabolic cost of pregnancy (associated with
Cp), and mass gain (associated with Cd1 ). Energy requirements
during pregnancy included mass that was stored to be used
during the subsequent lactation period.
We had lactation data for only a single walrus. The additional weight gained by this walrus during pregnancy was lost
by the end of the first year of the lactation period. Based on
this, we approximated the mass of a lactating walrus as
mij p 830 1 195exp 20.017aij ;
(9)
where aij represents number of days into lactation (aij p 1, . . . ,
730), 830 kg is the mass of a mature female walrus from Fay
(1982, fig. 11), and 195 kg is the model-based median weight
gain during pregnancy for the two pregnancies in our data set
less the mean weight of a newborn calf and placenta (based on
mass measurements from two newborns and one placenta
from the study animals). This function exponentially declines
from the initial postpartum weight of 1,025 kg to an asymptotic weight of 830 kg approximately 1 yr after lactation commences. We used the values from equation (9) for mij to calculate the corresponding dij and to estimate caloric requirements
(based on eq. [5] with A(ij) p 5.4) for a free-ranging 12-yrold walrus during lactation. We partitioned requirements into
components due to basal plus activity costs (associated with
A(ij)), metabolic cost of lactation (associated with Cl), and
mass loss (associated with Cd2). Energy consumption requirements for lactation excluded energy that was gained from mass
loss because this represents caloric intake that was already
accounted for during the pregnancy period.
We also estimated energy requirements for the case where
a free-ranging 12-yr-old walrus becomes pregnant while she
is still nursing a calf and gives birth at the end of the 2-yr
lactation interval for the first calf. Values for mij were obtained by combining model (8) and equation (9), and these
were then used to calculate the corresponding dij and to estimate caloric requirements based on equation (5). Requirements were partitioned into components as described above for
pregnancy and lactation.
Validation
For our model to be plausible, walruses must be able to meet
estimated energetic requirements within realistic limits for
foraging time and ingestion capacity. The proportion of time
required to forage must be less than the assumed activity level
(83% of the time active in water) because the time in water
must also include transit and search time. The required ingestion rate should be approximately 5%–7% of body mass
per day to be in agreement with consumption estimates based
on stomach contents (Fay 1982), and observations of foraging
(Born et al. 2003) for walruses but could be as high as 15%–
20% of body mass per day, based on observed upper limits to
food consumption by Steller sea lions (Eumetopias jubatus;
Rosen 2009).
Because of a lack of data for Pacific walruses, we estimated
foraging and consumption rates based on observations of
free-ranging Atlantic walruses Odobenus rosmarus rosmarus
(Born et al. 2003). The approach is described in detail by
Noren et al. (2012). Briefly, we estimated that walruses consume eight clams per minute of dive cycle (dive duration
plus subsequent surface duration), which is equivalent to
92.57 kcal or 87.68 shell-free grams wet weight per minute of
dive, or 11.57 kcal and 10.96 shell-free grams wet weight
per bivalve consumed (energy density p 1.06 kcal g21 or
4.42 kJ g21; Born et al. 2003). Admittedly, we oversimplified
this system. First, the feeding rates from Born et al. (2003)
are for adult walruses, but young walruses, like other immature pinnipeds, could have limited foraging capabilities
due to naïveté, small body size, and underdeveloped diving
physiology (for review, see Noren et al. 2005). Thus, our estimates of foraging times for immature walruses likely represent minimum values. Second, the diets of Pacific walruses
are more diverse than strictly clams (Sheffield and Grebmeier
2009). Nonetheless, walruses are highly selective for bivalves
(Fay et al. 1977; Fay and Lowry 1981; Fay and Stoker 1982a,
1982b), and the energy contents of diverse taxa of marine
benthic invertebrates from the Canadian Arctic fall within a
relatively narrow range (Wacasey and Atkinson 1987). Even
with these limitations, we feel that this approach provides
adequate approximations for assessing model plausibility.
Results
For comparison to Noren et al. (2012), the energetic requirements of walruses determined in this study are reported in
kilocalories, but these values can be converted to kilojoules
according to the following conversion factor: 1 kcal p
4.184 kJ. The daily caloric intake for the female Pacific walruses in this study varied with body size, facility, and season
and ranged from 0 to 56,298 kcal d21 for 210–1,085-kg, 2.0–
30.4-yr-old walruses. Older animals tended to have greater
daily caloric intake and lower mass-specific caloric intake,
consistent with the scaling of metabolism with body mass
(fig. 1). Body mass increased during pregnancy, as did caloric intake, while during lactation, body mass decreased, as
did caloric intake (fig. 1A, 1B). Interestingly, mass-specific
food consumption changed little during reproductive events
(fig. 1C), suggesting that a large portion of the elevated caloric intake during pregnancy and early lactation (fig. 1A)
was due to the larger body sizes of the animals (fig. 1B) because of the scaling of metabolism with body mass (Kleiber
1975).
Caloric Requirements of Female Walruses
Model Parameters and Base Energy Requirements
The lowest DIC model for energy consumption used d values
calculated from a 46 df loess smooth of observed mass values
and applied the metabolic cost of growth (growth increment)
for ages !ag p 10 yr. Estimated metabolic costs for growth and
pregnancy were each about half of the estimated metabolic cost
of lactation (table 1). The relative effects of these costs on required consumption were moderated by the additional costs of
mass accumulation during growth and pregnancy and by energy gained from mass depletion during lactation. During periods of weight gain, mass was acquired at an estimated rate
of 1 kg per Cd1 p 6,813 kcal consumed (table 1). During periods of weight loss, required consumption was offset at an
estimated rate of Cd2 p 8,103 kcal per kg lost (table 1).
Estimated daily energy requirements varied across facilities, tending to be higher and more constant at the northernmost facilities and lower with stronger seasonal patterns at
the more southerly and coastal facilities (fig. 2). Mean daily
energy requirements at all facilities corresponded to activity
levels that were low relative to observed averages for freeranging walruses (Udevitz et al. 2009). Estimated energy requirements for an 830-kg female at the northernmost facilities were equivalent to requirements for a free-ranging
843
walrus active only 11%–15% of the time, while requirements
at the southern and coastal facilities were equivalent to requirements for a free-ranging walrus active only up to 3% of
the time.
Energetic Cost of Growth
The logistic growth model fit the mass-at-age values from Fay
(1982) very closely. Least squares estimates of the growth
model parameters were v1 p 833 (SE p 3.3), v2 p 1.78 (SE p
0.062), and v3 p 20.45 (SE p 0.014). Energy requirements
were mainly associated with base (basal and activity) costs,
though a smaller but important portion was attributable to
the metabolic cost of growth and the energetic demand of
adding mass for ages !10 yr (fig. 3). The combined costs of
growth (metabolic and mass accumulation) increased to a
maximum of 7% of total requirements at about 5.5 yr and
then declined, becoming negligible (!1% of total requirements)
at age 10, when the metabolic cost of growth no longer accrued (fig. 3). The posterior mean estimate of the total energy requirement for the early growth period (ages 2–5) was
57,830,000 kcal (95% credibility interval [CI] p 57,430,000–
58,250,000). The posterior mean estimate of the total energy
Table 1: Parameter estimates for the model of daily caloric consumption by female Pacific walruses
95% CI
Parameter
f1
f2
f3
f4
f5
f6
a1
a2
a3
a4
a5
a6
b1
b2
b3
b4
b5
b6
Cg
Cp
Cl
Cd2
Cd1
Description
Base metabolic cost at facility 1a
Base metabolic cost at facility 2a
Base metabolic cost at facility 3a
Base metabolic cost at facility 4a
Base metabolic cost at facility 5a
Base metabolic cost at facility 6a
Cosine component of seasonal metabolic cost
Cosine component of seasonal metabolic cost
Cosine component of seasonal metabolic cost
Cosine component of seasonal metabolic cost
Cosine component of seasonal metabolic cost
Cosine component of seasonal metabolic cost
Sine component of seasonal cost at facility 1a
Sine component of seasonal cost at facility 2a
Sine component of seasonal cost at facility 3a
Sine component of seasonal cost at facility 4a
Sine component of seasonal cost at facility 5a
Sine component of seasonal cost at facility 6a
Metabolic cost of growthb
Metabolic cost of pregnancyb
Metabolic cost of lactationb
Energy mobilization coefficientc
Energy storage coefficientc
Note. CI p credibility interval.
a
From equation (3).
b
From equation (2).
c
From equation (4).
at
at
at
at
at
at
facility
facility
facility
facility
facility
facility
1a
2a
3a
4a
5a
6a
Mean
SD
Lower limit
Upper limit
2.62
1.99
2.79
2.30
2.20
2.63
.01
2.13
.03
2.05
2.14
2.01
.06
.10
2.07
2.10
2.03
.08
.23
.26
.60
8,103
6,813
.12
.17
.13
.10
.12
.18
.02
.03
.02
.02
.03
.02
.02
.03
.02
.02
.04
.02
.02
.04
.06
293
231
2.37
1.65
2.55
2.09
1.95
2.24
2.04
2.19
2.01
2.09
2.21
2.05
.02
.04
2.10
2.14
2.10
.04
.19
.18
.48
7,530
6,361
2.88
2.36
3.06
2.49
2.46
2.97
.05
2.07
.07
2.01
2.07
.04
.10
.16
2.03
2.07
.05
.13
.27
.34
.72
8,678
7,268
844
S. R. Noren, M. S. Udevitz, and C. V. Jay
Figure 2. Map showing the locations of the facilities (A) and graph showing the estimated total energetic demand per day for 1 yr for a
hypothetical, nonreproductive, 830-kg, static-mass, 12-yr-old female at each of the six facilities (B). The color for the facility and that for the
modeled animal from that facility are coordinated. Energetic demands tended to be higher in regions that experience lower air temperatures.
requirement for the later growth period (ages 6–9) was
83,810,000 kcal (95% CI p 83,230,000–84,410,000). Differences between model-based estimates of expected total consumption during the entire growth phase (12 to !10 yr of age)
ranged from 1% to 10% of the observed value, but 75% of the
differences were less than 4% and consumption was not consistently over- or underestimated. By age 10, requirements were
primarily due to basal and activity costs for nonreproductive
animals (fig. 3) and achieved an asymptote of about 61,500 kcal
d21 (fig. 3; table 2). At a consumption rate of 92.57 kcal min21
Caloric Requirements of Female Walruses
845
Figure 3. Estimated energetic demand throughout the life of a nonreproductive, free-ranging female Pacific walrus over 20 yr, assuming an
activity budget of 83% of the time active in water and 17% of the time hauled-out resting. The height of the shaded area represents the total
energetic demand on a given day, partitioned into calories required for early growth, adding body mass, and base (basal plus activity).
(Born et al. 2003), 2–5-, 6–9-, and ≥10-yr-old walruses must
forage 20%–35%, 38%–45%, and 44%–46% of the day, respectively, to meet these requirements. Assuming a clam diet,
where a gram of shell-free wet weight has an energy density
of 1.06 kcal (calculated from Born et al. 2003), 2–5-, 6–9-,
and ≥10-yr-old walruses could support energetic requirements
by consuming 9%–10%, 7%–8%, and 7% of body mass d21,
respectively.
Energetic Costs of Reproduction
There were two full-term pregnancies with consistently recorded data for estimating the pattern of weight gain during
pregnancy. Estimates of the posterior medians of the weight
gain model parameters were m1 p 281 (95% CI p 258–318),
m2 p 5.63 (95% CI p 2.58–9.03), and m3 p 0.0247 (95% CI p
20.1016 to 0.1457). This gave a pattern of weight gain that
began increasing rapidly about halfway through the pregnancy
to a total of 256 kg by the end of the period (fig. 4). Energy
requirements increased during pregnancy as a result of the
increasing mass of the walrus and the metabolic cost associated with Cp, which also increased with mass (table 3; fig. 5A).
However, there was also a substantial energy requirement for
accumulating mass (associated with Cd1 ), some of which went
to growth of the fetus and placenta but most of which (195 kg)
was apparently stored for use during the subsequent lactation
(fig. 5B). The energy required for accumulating mass increased
to a maximum of 14% of total requirements at about day 220 of
the pregnancy and then began decreasing, accounting for 3% of
total requirements by the end of the pregnancy (fig. 5A). The
additional metabolic requirements of pregnancy accounted for
a relatively constant 4%–5% of total requirements throughout
the pregnancy period (fig. 5A). The posterior mean estimate
of total energy required during the 11-mo pregnancy period
was 24,950,000 kcal (95% CI p 24,640,000–25,260,000). Differences between model-based estimates of expected total consumption during the two pregnancy periods were 3% less than
the observed value in one case and 13% greater than the observed value in the other case. Based on consumption rates for
walruses consuming clams (from Born et al. 2003), the estimated peak energy requirements during pregnancy (month 9
of pregnancy) could be met by a walrus foraging for 64% of
the day and consuming 8% of body mass per day.
Total energy requirements during the lactation interval
decreased throughout the first year (table 4; fig. 5B), which
was associated with the decreasing mass of the walrus; mass
decreased exponentially from an initial weight of 1,025 kg to
an asymptotic weight of approximately 830 kg. Total energy
requirements leveled off after the first year of lactation to a
value about 11% higher than that required for a nonlactating walrus of comparable size. The posterior mean estimate
of total energy consumption required during the 2-yr lactation period was 48,930,000 kcal (95% CI p 47,930,000–
49,940,000). This does not include the estimated 1,580,000
kcal (95% CI p 1,470,000–1,690,000) derived from utilizing
mass during lactation, which was acquired and stored during
the preceding pregnancy period. The estimated peak energy
consumption period for the 2-yr lactation interval occurred
during months 6–24, once onboard energy reserves were exhausted (table 4). The model-based estimate of expected to-
846
S. R. Noren, M. S. Udevitz, and C. V. Jay
Table 2: Estimated daily energy requirement on day 1 for each age class for nonreproductive, freeranging female Pacific walruses, assuming an activity budget of active in water for 83% of the time
and hauled-out resting for 17% of the time
95% CI
Age class
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Body mass (kg)
Daily requirement (kcal)
Lower limit
Upper limit
243
326
418
509
592
661
714
753
780
798
810
818
824
827
829
831
832
832
833
26,900
33,410
39,960
45,990
51,080
55,060
58,000
60,060
59,010
59,900
60,490
60,870
61,120
61,280
61,390
61,450
61,490
61,520
61,540
26,700
33,170
39,670
45,660
50,720
54,680
57,590
59,640
58,980
59,880
60,480
60,870
61,120
61,280
61,380
61,450
61,490
61,520
61,540
27,100
33,650
40,250
46,320
51,440
55,450
58,410
60,490
59,030
59,920
60,500
60,880
61,130
61,290
61,390
61,450
61,500
61,520
61,540
Note. CI p credibility interval.
tal consumption was within 1% of observed consumption for
the single lactation period in this study. Assuming a consumption rate of 92.57 kcal min21, the peak energy consumption period during lactation could be met by a walrus
foraging for 51% of each day at a rate of 92.57 kcal min21.
Based on a clam diet (energy density p 1.06 kcal g21), this
translates to consuming 8% of body mass per day. By utilizing onboard stores to support 27% of the energetic demands
of lactation, females during the first month of lactation could
meet energetic needs by foraging for only 42% of the day and
consuming only 5% of body mass per day (based on consumption rates for walruses consuming clams from Born
et al. 2003). Surprisingly, this level of consumption and foraging effort is lower than what we estimated for mature,
nonreproductive females.
The additional energy requirements due to pregnancy were
added to those of lactation for a walrus assumed to become
pregnant during the lactation interval (table 5; fig. 5C). The
maximum energy requirement occurred when lactation overlapped the last half of the pregnancy period. The posterior
mean estimate of total energy required during a 2-yr lactation
period that included an 11-mo pregnancy was 53,560,000 kcal
(95% CI p 52,440,000–54,710,000). The peak energy consumption period across all age classes and reproductive statuses occurred for females that are simultaneously lactating
and are 9 mo into pregnancy. This peak energy consumption
period for a simultaneously pregnant and lactating female
walrus could be met by foraging for 70% of each day, which
translates to consuming 9% of body mass per day (based on
consumption rates for walruses consuming clams from Born
et al. 2003).
Discussion
The only previously available bioenergetics model for walruses was developed by Noren et al. (2012), but that model
had to be relatively simple because the lack of physiological
data for this species did not support a more complex model.
For example, the energetic cost of adding mass was accounted
for only by including a growth multiplier that accounted for
the elevated metabolic rate of immature animals !6 yr old,
even though mature body size is not attained in female Pacific walruses until they are 12 yr old (Fay 1982). In addition,
body mass was kept static during reproductive events (pregnancy and lactation), though blubber thickness (and hence
body mass) has been observed to vary with the reproductive
condition of female walruses (Fay 1985). Perhaps more important, Noren et al. (2012) did not attempt to explore the
linkage between caloric intake and energy stores (body mass),
which can allow energy consumption and demand to be uncoupled during critical life-history stages. These types of
limitations led, for example, to the unreasonable implication
that a lactating walrus would need to forage 95%–101% of
its time to meet energetic requirements (Noren et al. 2012).
By explicitly accounting for accumulation and depletion of
energy stores and by conducting focused research on walruses
Caloric Requirements of Female Walruses
847
Figure 4. Gain in body mass from the onset of pregnancy for two Pacific walruses. Observed and model-based estimates of mass for the two
individual walruses are denoted by light gray and dark gray, respectively. The median body mass gain function across the two walruses is
represented by the black line.
in aquaria to develop walrus-specific estimates for model parameters, we have been able to address many of the limitations of the model developed by Noren et al. (2012).
Base (Basal and Activity) Requirements
The amount of food consumed by captive animals is generally
lower than what bioenergetic models predict for wild animals
(Winship et al. 2006). Compared to free-ranging walruses in
the Arctic, walruses in aquaria are likely to have relatively
low base (basal and activity) energetic demands because in
aquaria, at least in temperate regions, thermoregulatory demands are lower. Also, because the captive walruses do not
need to forage for their food or avoid predators, their activity demands are undoubtedly lower. Caloric intakes observed
for the nonreproductive captive walruses in this study ranged
Table 3: Estimated mean daily energy requirements by month of pregnancy after implantation of
the blastocyst for a free-ranging female Pacific walrus, assuming an activity budget of active in water for
83% of the time and hauled-out resting for 17% of the time
95% CI
Month
1
2
3
4
5
6
7
8
9
10
11
Mean body mass (kg)
Mean daily requirement (kcal)
Lower limit
Upper limit
831
833
836
843
856
881
919
969
1,018
1,057
1,080
64,720
65,010
65,720
67,250
69,950
74,370
79,580
83,680
85,400
84,090
81,810
63,820
64,110
64,830
66,370
69,080
73,450
78,580
82,620
84,340
83,050
80,750
65,610
65,900
66,610
68,130
70,840
75,290
80,580
84,750
86,470
85,140
82,860
Note. CI p credibility interval.
Figure 5. Estimated daily energy demand during reproductive events for a 12-yr-old free-ranging female walrus. The animal’s initial body
mass was assumed to be 830 kg, with a weight gain of 256 kg during the pregnancy period. Estimates for a walrus that is pregnant, lactating,
and simultaneously lactating and pregnant are shown in A, B, and C, respectively. The height of the shaded area represents the total caloric
demand on a given day, which is partitioned into calories associated with the energy of adding body mass or the energy released from
depleting body mass and the energy required for base (basal plus activity) and reproduction (pregnancy, lactation, or both). Dotted shading
represents periods when a portion of the calories required to support energetic demands is coming from depletion of body mass, as during
lactation.
Caloric Requirements of Female Walruses
849
Table 4: Estimated mean daily energy requirements, via consumption and from body stores, by month of lactation for a
free-ranging female Pacific walrus, assuming an activity budget of active in water for 83% of the time and hauled-out resting
for 17% of the time
95% CI
Month
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
95% CI
Mean body
mass (kg)
Mean daily requirement
from consumption (kcal)a
Lower
limit
Upper
limit
Mean daily requirement
from stores (kcal)
Lower
limit
Upper
limit
983
921
884
862
849
841
837
834
832
831
831
831
830
830
830
830
830
830
830
830
830
830
830
830
56,330
61,140
64,030
65,740
66,750
67,360
67,710
67,930
68,050
68,130
68,180
68,200
68,220
68,230
68,230
68,240
68,240
68,240
68,240
68,240
68,240
68,240
68,240
68,240
54,210
59,440
62,520
64,330
65,380
66,000
66,360
66,580
66,710
66,780
66,830
66,860
66,870
66,880
66,890
66,890
66,890
66,900
66,900
66,900
66,900
66,900
66,900
66,900
58,510
62,880
65,560
67,180
68,150
68,730
69,070
69,280
69,410
69,480
69,530
69,560
69,570
69,580
69,590
69,590
69,590
69,590
69,590
69,590
69,590
69,590
69,590
69,590
21,050
12,640
7,530
4,480
2,670
1,590
950
570
340
200
120
70
40
30
20
10
10
0
0
0
0
0
0
0
19,580
11,760
7,000
4,170
2,480
1,480
890
530
310
190
110
70
40
20
10
10
10
0
0
0
0
0
0
0
22,530
13,530
8,060
4,800
2,860
1,700
1,020
610
360
220
130
80
50
30
20
10
10
0
0
0
0
0
0
0
Note. CI p credibility interval.
from 0 to 56,298 kcal d21 (fig. 1A), which corresponded to
estimated activity levels of up to only 15% of the time active
in water. This activity level is dramatically lower than the
average 83% of the time active in water observed for freeranging walruses in the Bering Sea (Udevitz et al. 2009).
Based on this assessment, we used estimates from Noren
et al. (2012) for base (basal and activity) energy requirements
of wild walruses, and we recommend that these continue to
be used in the absence of new data. The estimates were based
on the average proportions of time spent active and resting by
free-ranging Pacific walruses (Udevitz et al. 2009), with periods of activity and rest requiring a metabolism of 6 times
basal metabolic rate (BMR; based on measured field metabolisms of free-ranging male Atlantic walruses Odobenus rosmarus rosmarus; Born et al. 2003) and 2.2 times BMR (measured for resting California sea lions Zalophus californianus;
Williams et al. 2007), respectively. These estimates are in the
range of field (5–6 times BMR; Costa et al. 1991; Costa and
Williams 1999; Costa 2002) and maintenance (1.4–2.8 times
BMR; for review, see Williams et al. 2001) metabolisms measured directly from a range of pinniped species. As with the
captive walruses (fig. 2), there may be seasonal variations in
energy requirements and consumption in wild walruses, but
a lack of physiological data from wild walruses precluded
us from characterizing these. The simplification of maintaining base energetic costs at a consistent mean level throughout
the year should not affect estimates of annual energy requirements (Noren et al 2012) but should be considered when
inferring the magnitude of seasonal variations in energy requirements and consumption in wild walruses.
Growth Requirements
Consistent with the mass-related scaling of metabolism
(Kleiber 1975), observed caloric intake increased with walrus
age (fig. 1A), as did estimated caloric intake requirements
(fig. 3). However, our estimates for 2–5-yr-old immature
walruses were lower than the estimates of Noren et al. (2012).
As a result, newly weaned 2-yr-old female walruses require
1.3 h less foraging time per day than previously estimated by
Noren et al. (2012). Overall, the posterior mean estimate of
the total energy requirement for this 4-yr early growth period was 6% less than the estimate based on the Noren et al.
(2012) model. This discrepancy between the models arises
850
S. R. Noren, M. S. Udevitz, and C. V. Jay
Table 5: Estimated mean daily energy requirement, via consumption and from body stores, by month of lactation for a
free-ranging female Pacific walrus, assuming an activity budget of active in water for 83% of the time and hauled-out resting
for 17% of the time
95% CI
Month
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
95% CI
Mean body
mass (kg)
Mean daily requirement
from consumption (kcal)
Lower
limit
Upper
limit
Mean daily requirement
from stores (kcal)a
Lower
limit
Upper
limit
981
921
884
862
849
841
837
834
832
831
831
831
830
831
833
836
843
856
880
919
968
1,018
1,057
1,079
56,330
61,140
64,030
65,740
66,750
67,360
67,710
67,930
68,050
68,130
68,180
68,200
68,220
71,490
71,860
72,590
74,150
76,900
81,350
86,890
91,320
93,370
92,340
90,180
54,210
59,440
62,520
64,330
65,380
66,000
66,360
66,580
66,710
66,780
66,830
66,860
66,870
69,820
70,170
70,900
72,450
75,180
79,580
85,020
89,370
91,360
90,310
88,140
58,510
62,880
65,560
67,180
68,150
68,730
69,070
69,280
69,410
69,480
69,530
69,560
69,570
73,190
73,580
74,310
75,880
78,660
83,190
88,810
93,340
95,440
94,420
92,280
21,050
12,640
7,530
4,480
2,670
1,590
950
570
340
200
120
70
40
2310
2480
21,000
22,120
24,010
26,940
210,040
211,420
210,400
26,990
23,470
19,580
11,760
7,000
4,170
2,480
1,480
890
530
310
190
110
70
40
2340
2510
21,070
22,260
24,280
27,400
210,710
212,170
211,090
27,460
23,700
22,530
13,530
8,060
4,800
2,860
1,700
1,020
610
360
220
130
80
50
2290
2450
2930
21,980
23,750
26,480
29,380
210,660
29,710
26,530
23,240
Note. In this case, at month 14, the animal becomes pregnant, so she is simultaneously lactating and pregnant. CI p credibility interval.
Positive values indicate energy coming from stores; negative values indicate energy going into stores.
a
primarily because, lacking specific information about growth
in odobenids, or more closely related otariids (Schröder et al.
2009; Agnarsson et al. 2010; Fulton and Strobeck 2010), Noren et al. (2012) based their estimates for the energetic demand of growth on data from phocids (Worthy 1987). However, compared to odobenids, phocids have a greater rate of
mass gain early in life (for review, see Ferguson 2006), which
is sustained by relatively high caloric intake, as the milk produced by phocids is higher in fat and energy content than
that of odobenids (for review, see Riedman 1990; Ferguson
2006). This difference in life-history patterns results in lower
energy requirements for walruses during the early growth
period.
Although our estimates of daily caloric requirements of
young walruses are lower than estimates in Noren et al. (2012),
2–5-yr-olds still must consume proportionately greater quantities of food than 6–20-yr-olds. Our estimates suggest that
the youngest animals must consume 9%–10% of their body
mass on a daily basis, while older animals need to consume
only 7%–8% of their body mass. The need for immature ani-
mals to consume proportionately more food than larger conspecifics is in agreement with the empirical data collected
from the walruses in aquaria (fig. 1C) and observations from
other pinniped species (for review, see Winship et al. 2006).
In addition, our estimate of the quantity of food consumed
per day by older walruses is in agreement with previous estimates for free-ranging adult walruses (5%–7% of body mass
per day; Fay 1982; Born et al. 2003). The difference in massspecific caloric requirements with age increases the vulnerability of immature pinnipeds during food-limited periods
because they must acquire proportionally greater amounts
of prey. This may in part explain the disproportionate deleterious effects on juveniles during prey-limited periods (DeLong et al. 1991).
Our approach to explicitly account for the cost of accumulating mass (storage component in the model) and our
data-based model selection indicated that both the mass accumulation and metabolic costs associated with growth extend beyond the 4-yr growth period (2–5-yr-olds) considered
by Noren et al. (2012). Our estimate of the total energy re-
Caloric Requirements of Female Walruses
quirement for the later growth period for 6–9-yr-olds was
6% greater than the estimate based on the Noren et al. (2012)
model, which did not include energy requirements of growth
for animals ≥6 yr old. By age 10, as female walruses approach
the asymptote for mature body size (Fay 1982), caloric consumption requirements of nonreproductive females in our
model and that of Noren et al. (2012) are essentially the same
since they are primarily due to basal and activity costs. By
using empirical data from walruses in aquaria, estimates for
the energetic demands of growth presented here more closely
reflect the life-history patterns of free-ranging walruses than
those presented in Noren et al. (2012).
Reproduction Requirements
The walruses in this study showed marked increases in daily
caloric intake during pregnancy and lactation (fig. 1A). This
is consistent with previous studies on captive female walruses,
where the animals consumed 30%–40% more food when
pregnant, 50%–101% more food when lactating, and 90%–
130% more food while simultaneously lactating and pregnant
(Gehnrich 1984; Kastelein et al. 2000). Elevated food consumption has also been observed in wild and captive pregnant and lactating otariids (Costa et al. 1989, 1991; Williams
et al. 2007), yet these potentially significant reproductive costs
have been largely ignored in previous bioenergetic models
for pinnipeds (Olesiuk 1993; Mohn and Bowen 1996; Stenson
et al. 1997). Although the model in Noren et al. (2012) incorporated costs of reproduction for female Pacific walruses,
it used the oversimplifying assumption that body mass was
static during reproductive events.
We found large increases in body mass during pregnancy
and subsequent mass loss during lactation (fig. 1B). Our
bioenergetics model enabled us to account for alterations in
caloric requirements associated with changes in body size
due to the scaling of metabolism with size (Kleiber 1975) and,
importantly, allowed us to uncouple caloric intake and demand during reproductive events (fig. 5; tables 3–5). This is
important for pinnipeds because they sequester and deplete
large quantities of lipids throughout their lifetimes, particularly during reproduction. Some of the energetic demand of
lactation in walruses is undoubtedly met by utilizing endogenous energy reserves (e.g., blubber) accumulated during pregnancy, which is consistent with the observation that blubber
thickness varies with reproductive condition in female walruses (Fay 1985). Our estimate of the amount of energy recovered per kilogram of body mass is slightly lower than the
energy found in lipid (9,386.805 kcal kg21; Schmidt-Nielsen
1997), suggesting that walruses likely also metabolize some
protein during periods of energy deficit, which for Steller sea
lions (Eumatopias jubatus) can represent up to 31.2% of the
caloric contribution from body reserves (Rea et al. 2007).
Our more comprehensive approach to modeling energy
requirements during pregnancy differed from that of Noren
et al. (2012) in several respects. For example, based on data
851
from an otariid, Noren et al. (2012) only ascribed a cost of
pregnancy during the last trimester. However, the empirical
data from the walruses in this study indicated that, even
though the per-day metabolic cost for pregnancy (Cp p 0.26)
was lower than the otariid-based value (Cp p 0.92) used by
Noren et al (2012), the cost accrued throughout the entire
pregnancy. In addition, there was a cost due to the scaling
of metabolism with the increasing mass of the walrus, and
there was a substantial energy requirement for accumulating
mass, some of which went to growth of the fetus and placenta but most of which (76%) was apparently stored mass
(e.g., lipid) for use during the subsequent lactation (fig. 5B).
Neither of these requirements was accounted for by Noren
et al. (2012). As a result, the posterior mean estimate of total energy required during the 11-mo pregnancy period was
15% greater than the estimate based on the Noren et al.
(2012) model. Nonetheless, our estimated daily caloric requirements of pregnant walruses are plausible. At the height
of their energetic demand, pregnant walruses could meet caloric requirements by consuming 7,831 clams per day, which
would require them to forage for only 64% of the day (based
on foraging efficiency estimates in Born et al. 2003).
A substantial proportion of the energetic requirements
for lactation is supported by consumption and storage of
lipid during the pregnancy period. Our estimate of total energy consumption required during the 2-yr lactation period
was 47% less than the estimate based on the Noren et al.
(2012) model, partly because we accounted for an estimated
1,580,000 kcal derived from mass that was acquired and stored
during the preceding pregnancy period, though this was offset somewhat by the cost of maintaining the additional mass
until it was depleted. Most of the difference, however, was
due to our walrus-specific estimate of the metabolic cost of
lactation (Cl p 0.6), which was substantially lower than the
otariid-based value (Cl p 5.72) used by Noren et al (2012).
Our posterior mean estimate of total energy consumption
required during the 2-yr lactation period and 11-mo pregnancy was 43% less than the estimate based on the model in
Noren et al. (2012). During the most energetically taxing lifehistory stage (simultaneously pregnant and lactating), caloric
requirements could be met by foraging 70% of the day. This
coarse estimate of foraging requirements is more reasonable
than the estimates of Noren et al. (2012), which indicated
caloric requirements during lactation could not be sustained
by foraging.
Noren et al. (2012), lacking information specific to walruses, derived estimates of energetic costs for reproduction
using data from an otariid (California sea lions Zalophus
californianus; Williams et al. 2007) because odobenid phylogeny (Schröder et al. 2009; Agnarsson et al. 2010; Fulton
and Strobeck 2010) and life-history patterns (Fay 1982; Kovacs and Lavigne 1992) are similar in some respects to those
of otariids. Odobenids, like otariids, forage throughout their
prolonged lactation interval (for review, see Riedman 1990;
Bowen 1991; Costa 1991). In addition, the milk of odobenids
852
S. R. Noren, M. S. Udevitz, and C. V. Jay
and otariids is comparatively lower in energy content than
the milk of phocids (for review, see Reidman 1990). However,
odobenids are much larger than otariids, and body size has a
large impact on lactation traits (Ferguson 2006). According
to Costa (1991), a larger female can devote a greater proportion of body stores to its offspring because metabolic
overhead (the energy expended to support the basal and activity costs of the female; Fedak and Anderson 1982) scales
with body size (M0.75) at a lower rate than the amount of
energy stored as adipose tissue (M1.19; Calder 1984). This
suggests that it may be more energetically efficient for walruses to undergo weight gain during pregnancy to build up
energy stores to support lactation, as has been observed in
phocids (Chabot et al. 1996). Although the lactation intervals
of otariids are partially supported by lipid reserves (i.e., California sea lions Zalophus californianus; Williams et al. 2007),
the larger walrus can theoretically support a greater proportion of the energetic demands of lactation through lipid reserves than the comparatively smaller otariid.
Much of the accumulated mass of the captive walruses
during pregnancy was probably lipid, since the combined
average mass for a full-term fetus and placenta was only 24%
of the estimated mass gained during pregnancy. The observation that blubber thickness varies with reproductive condition in free-ranging female walruses (Fay 1985) supports
the assumption that the energetic demands of lactation in
walruses are partially met by utilizing endogenous energy
reserves (fig. 5B). This attribute is a hallmark of capital breeders (provision offspring using energy stores accumulated at
an earlier time), but, except for perhaps several days during
estrus and a week or so at parturition, walruses feed during
lactation (Fay 1985), which would classify them as income
breeders (provision offspring using energy gained concurrently). Thus, it seems that walruses straddle these two lactation strategies. This is consistent with the recommendation
of Houston et al. (2007) that capital and income breeding
should not be thought of as dichotomous strategies because
some pinnipeds may adopt a mixture of the two lactation
strategies. By shifting some of the energetic costs of lactation
into the pregnancy period, walruses are able to reduce caloric
intake requirements during the first few months of lactation. This may be especially important for walruses, because
unlike otariids, walruses are accompanied by their offspring
while they forage (Fay 1985), which could impact the female’s
foraging success.
Conclusion
Our extension of the Noren et al. (2012) model, incorporating
new walrus-specific data, represents a substantial improvement in quantifying energy requirements of key life-history
stages (growth and reproduction) for Pacific walruses. More
generally, this work demonstrates the utility of food consumption and body mass data acquired from animals in
aquaria for elucidating the bioenergetics of wild conspecifics. By decoupling daily caloric intake from daily caloric de-
mand, we were able to improve our understanding of how
female walruses sequester and deplete body reserves throughout their lifetime. This approach revealed new information
about the basic biology of walruses, including the apparent
use of a mixed reproductive strategy of capital and income
breeding. Moreover, our bioenergetics model provides a basis
for quantifying how energy deficits can manifest in mass loss
and reduced body condition in female walruses. It also provides the linkage required for understanding energetic consequences of the changes in walrus behavior and prey accessibility that are resulting from changes in sea ice availability.
This type of information will be essential for predicting Pacific walrus population responses to the changing Arctic environment.
Acknowledgments
We thank the staff, trainers, and animals at Aquarium du
Quebec (particularly F. Couture), Brookfield Zoo (particularly R. Stacey), Indianapolis Zoo (particularly L. Oland),
New York Aquarium (particularly S. Mitchell), Point Defiance Zoo and Aquarium (particularly A. Shaffer), and Six
Flags Discovery Kingdom (particularly J. Paschke) for providing the invaluable data for this study. We are indebted to
M. Goguen and K. Heuer, who assisted with data entry, and
C. Reichmuth, H. Muraco, and M. Muraco for introducing us
to the walrus in human care community. We thank B. S.
Fadely and the laboratory group of T. M. Williams for helpful
comments on previous versions of this manuscript. Funding
for this study was provided by the US Geological Survey,
Ecosystems Mission Area, for the Changing Arctic Ecosystems Initiative. Any mention of trade names is for descriptive
purposes only and does not constitute endorsement by the
federal government.
Literature Cited
Acquarone M., E.W. Born, and J.R. Speakman. 2006. Field
metabolic rates of walrus (Odobenus rosmarus) measured
by the doubly labeled water method. Aquat Mamm 32:
363–369.
Agnarsson I., M. Kuntner, and L.J. May-Collado. 2010. Dogs,
cats, and kin: a molecular species-level phylogeny of Carnivora. Mol Phylogenet Evol 54:726–745.
Born E.W., S. Rysgaard, G. Ehlmé, M. Sejr, M. Acquarone,
and N. Levermann. 2003. Underwater observations of foraging free-living Atlantic walrus (Odobenus rosmarus rosmarus) and estimates of their food consumption. Polar Biol
26:348–357.
Bowen W.D. 1991. Behavioural ecology of pinniped neonates.
Pp. 66–127 in D. Renouf, ed. Behaviour of pinnipeds.
Chapman & Hall, London.
Brooks S.P. and A. Gelman. 1998. Alternative methods for
monitoring convergence of iterative simulations. J Comput
Graph Stat 7:434–455.
Caloric Requirements of Female Walruses
Calder W.A. 1984. Size, function, life history. Harvard University Press, Cambridge, MA.
Chabot D., G.B. Stenson, and N.B. Cadigan. 1996. Short- and
long-term fluctuations in the size and condition of harp
seal (Phoca groenlandica) in the Northwest Atlantic. Northwest Atl Fish Organ Sci Counc Stud 26:15–32.
Costa D.P. 1991. Reproductive and foraging energetics of
pinnipeds: implications for life history patterns. Pp. 300–
344 in D. Renouf, ed. Behaviour of pinnipeds. Chapman &
Hall, London.
———. 2002. Energetics. Pp. 387–394 in W.F. Perrin, B.
Würsig, and J.G.M. Thewissen, eds. Encylopedia of marine
mammals. Academic Press, San Diego, CA.
Costa D.P., G.A. Antonelis, and R.L. Delong. 1991. Effects of
El Niño on the foraging energetics of the California sea
lion. Pp. 156–165 in F. Trillmich and K.A. Ono, eds. Pinnipeds and El Niño: responses to environmental stress.
Springer, Berlin.
Costa D.P., J.P. Croxall, and C.D. Duck. 1989. Foraging energetics of Antarctic fur seals in relation to changes in prey
availability. Ecology 70:596–606.
Costa D.P. and T.M. Williams. 1999. Marine mammal energetics. Pp. 176–217 in J.E. Reynolds III and S.A. Rommel,
eds. Biology of marine mammals. Smithsonian Institution
Press, Washington, DC.
DeLong R.L., G.A. Antonelis, C.W. Oliver, B.S. Stewart, M.C.
Lowry, and P.K. Yochem. 1991. Effects of the 1982–83 El
Niño on several population parameters and diet of California sea lions on the California Channel Islands. Pp. 166–
172 in F. Trillmich and K.A. Ono, eds. Pinnipeds and El
Niño: responses to environmental stress. Springer, Berlin.
Douglas D.C. 2010. Arctic sea ice decline: projected changes
in timing and extent of sea ice in the Bering and Chukchi
Seas. Open-File Report 2010-1176. US Geological Survey,
Reston, VA.
Fay F.H. 1982. Ecology and biology of the Pacific walrus,
Odobenus rosmarus divergens Illiger. N Am Fauna 74:1–279.
———. 1985. Odobenus rosmarus. Mamm Species 238:1–7.
Fay F.H. and J.J. Burns. 1988. Maximal feeding depth of
walruses. Arctic 41:239–240.
Fay F.H., H.M. Feder, and S.W. Stoker. 1977. An estimation
of the impact of the Pacific walrus population on its food
resources in the Bering Sea. Final report, US Marine Mammal Commission, PB-272-505. National Technical Information Service, Springfield, VA.
Fay F.H. and L.F. Lowry. 1981. Seasonal use and feeding
habits of walruses in the proposed Bristol Bay clam fishery
area. Council Document 18. North Pacific Fishery Management Council, Anchorage, AK.
Fay F.H. and S.W. Stoker. 1982a. Analysis of reproductive
organs and stomach contents from walruses taken in the
Alaskan native harvest, spring 1980. Final report, contract
14-16-007-81-5216. US Fish and Wildlife Service, Anchorage, AK.
———. 1982b. Reproductive success and feeding habits of
walruses taken in the 1982 spring harvest, with comparisons
853
from previous years. Final report. Eskimo Walrus Commission, Nome, AK.
Fedak M.A. and S.A. Anderson. 1982. The energetics of lactation: accurate measurements from a large wild mammal, the
gray seal (Halichoerus grypus). J Zool (Lond) 198:473–479.
Ferguson S.H. 2006. The influences of environment, mating
habitat, and predation on evolution of pinniped lactation
strategies. J Mamm Evol 13:63–82.
Fisher K.I., R.E.A. Stewart, R.A. Kastelein, and L.D. Campbell. 1992. Apparent digestive efficiency in walrus (Odobenus rosmarus) fed herring (Clupea harengus) and clams
(Spisula sp.). Can J Zool 70:30–36.
Fulton T.L. and C. Strobeck. 2010. Multiple fossil calibrations,
nuclear loci and mitochondrial genomes provide new insight into biogeography and divergence timing for true seals
(Phocidae, Pinnipedia). J Biogeogr 37:814–829.
Gehnrich P.H. 1984. Nutritional and behavioral aspects
of reproduction in walruses. Master’s thesis, University of
Alaska, Fairbanks.
Gelman A., J.B. Carlin, H.S.Stern, and D.B. Rubin. 1997.
Bayesian data analysis. Chapman & Hall, New York.
Hammill M.O., C. Lydersen, K.M. Kovacs, and B. Sjare. 1997.
Estimated fish consumption by hooded seals (Cystophora
cristata), in the Gulf of St. Lawrence. J Northwest Atl Fish
Sci 22:249–257.
Houston A.I., P.A. Stephens, I.L. Boyd, K.C. Harding, and
J.M. McNamara. 2007. Capital or income breeding? a theoretical model of female reproductive strategies. Behav Ecol
18:241–250.
Jay C.V., A.S. Fischbach, and A.A. Kochnev. 2012. Walrus
areas of use in the Chukchi Sea during sparse sea ice cover.
Mar Ecol Prog Ser 468:1–13.
Jay C.V., B.G. Marcot, and D.C. Douglas. 2011. Projected
status of the Pacific walrus (Odobenus rosmarus divergens)
in the twenty-first century. Polar Biol 34:1065–1084.
Kastelein R.A., N.M. Schooneman, and P.R. Wiepkema. 2000.
Food consumption and body weight of captive Pacific
walrus (Odobenus rosmarus divergens). Aquat Mamm 26:
175–190.
Kavry V.I., A.N. Boltunov, and V.V. Nikiforov. 2008. New
coastal haul-outs of walrus (Odobenus rosmarus): response
to the climate changes. Pp. 248–251 in Collection of scientific papers from the Marine Mammals of the Holarctic
V Conference, Odessa, Ukraine.
Kleiber M. 1975. The fire of life: an introduction to animal
energetics. Krieger, Huntington, NY.
Kovacs K.M. and D.M. Lavigne. 1992. Maternal investment in
otariid seals and walruses. Can J Zool 70:1953–1964.
Lavigne D.M., W. Barchard, S. Innes, and N.A. Øritsland.
1982. Pinniped bioenergetics. Pp. 191–235 in FAO Advisory Committee on Marine Resources Research, Working
Party on Marine Mammals. Mammals in the seas. Vol. 4.
Small cetaceans, seals, sirenians and otters. FAO Fisheries
Series no. 5. Food and Agriculture Organization, Rome.
Meier W.N., J. Stroeve, and F. Fetterer. 2007. Whither Arctic
sea ice? a clear signal of decline regionally, seasonally and
854
S. R. Noren, M. S. Udevitz, and C. V. Jay
extending beyond the satellite record. Ann Glaciol 46:428–
434.
Mohn R. and W.D. Bowen. 1996. Grey seal predation on the
eastern Scotian Shelf: modeling the impact on Atlantic cod.
Can J Fish Aquat Sci 53:2722–2738.
Noren D.P. 2011. Estimated field metabolic rates and prey
requirements of resident killer whales. Mar Mamm Sci 27:
60–77.
Noren S.R., S.J. Iverson, and D.J. Boness. 2005. Development
of the blood and muscle oxygen stores in grey seals
(Halichoerus grypus): implications for juvenile diving capacity and the necessity of a terrestrial postweaning fast.
Physiol Biochem Zool 78:482–490.
Noren S.R., M. Udevitz, and C. Jay. 2012. Bioenergetics
model for estimating food requirements of female Pacific
walruses (Odobenus rosmarus divergens). Mar Ecol Prog
Ser 460:261–275.
Olesiuk P.F. 1993. Annual prey consumption by harbor seals
(Phoca vitulina) in the Strait of Georgia, British Columbia.
Fish Bull 91:491–515.
Øritsland N.A. and N.H. Markussen. 1990. Outline of a
physiologically based model for population dynamics. Ecol
Model 52:267–288.
Overland J.E. and M. Wang. 2007. Future regional Arctic sea
ice declines. Geophys Res Lett 34:L17705
Ovsyanikov N.G., I.E. Menyushina, and A.V. Bezrukov. 2008.
Unusual Pacific walrus mortality at Wrangel Island in 2007.
Pp. 413–416 in Collection of scientific papers from the
Marine Mammals of the Holarctic V Conference, Odessa,
Ukraine.
Rea L.D., D.A.S. Rosen, and A.W. Trites. 2007. Utilization
of stored energy reserves during fasting varies by age and
season in Steller sea lions. Can J Zool 85:190–200.
Riedman M. 1990. The pinnipeds seals, sea lions, and walruses. University of California Press, Berkeley.
Rosen D.A.S. 2009. Steller sea lions, Eumetopias jubatus, and
nutritional stress: evidence from captive studies. Mammal
Rev 39:284–306.
Ryg M. and N.A. Øritsland. 1991. Estimates of energy expenditure and energy consumption of ringed seals (Pusa
hispida) throughout the year. In E. Sakshaug, C.C.E. Hopkins, and N.A. Øritsland, eds. Proceedings of the Pro Mare
Symposium on Polar Marine Ecology, Trondheim, May
12–16, 1990. Polar Res 10:595–601.
Schmidt-Nielsen K. 1997. Animal physiology: adaptation and
environment. Cambridge University Press, Cambridge.
Schröder C., C. Bleidorn, S. Hartmann, and R. Tiedemann.
2009. Occurrence of can-SINEs and intron sequence evo-
lution supports robust phylogeny of pinniped carnivores
and their terrestrial relatives. Gene 448:221–226.
Sheffield G. and J.M. Grebmeier. 2009. Pacific walrus (Odobenus rosmarus divergens): differential prey digestion and
diet. Mar Mamm Sci 25:761–777.
Spiegelhalter D.J., N.G. Best, B.P. Carlin, and A. Van der
Linde. 2002. Bayesian measures of model complexity and
fit. J R Stat Soc B 64:583–616.
Spiegelhalter D.J., A. Thomas, N.G. Best, and D. Lund. 2003.
WinBUGS Version 1.4 User Manual. MRC Biostatistics
Unit, Cambridge.
Stenson G.B., M.O. Hammill, and J.W. Lawson. 1997. Predation by harp seals in Atlantic Canada: preliminary consumption estimates for Arctic cod, capelin, and Atlantic
cod. J Northwest Atl Fish Sci 22:137–154.
Udevitz M.S., C.V. Jay, A.S. Fischbach, and J.L. GarlichMiller. 2009. Modeling haul-out behavior of walruses in
Bering Sea ice. Can J Zool 87:1111–1128.
Wacasey J.W. and E.G. Atkinson. 1987. Energy values of
marine benthic invertebrates from the Canadian Arctic.
Mar Ecol Prog Ser 39:243–250.
Wang M., J.E. Overland, and P. Stabeno. 2012. Future climate
of the Bering and Chukchi Seas projected by global climate
models. Deep-Sea Res Part II 65–70:46–57.
Williams T.M., J. Haun, R.W. Davies, L. Fuiman, and S.
Kohin. 2001. A killer appetite: metabolic consequences of
carnivory in marine mammals. G.L. Kooyman Symposium.
Comp Biochem Physiol A 129:785–796.
Williams T.M., M. Rutishauser, B. Long, T. Fink, J. Gafney,
H. Mostman-Liwanag, and D. Casper. 2007. Seasonal variability in Otariid energetics: implications for the effects of
predators on localized prey resources. Physiol Biochem
Zool 80:433–443.
Winship A.J., A.M.J. Hunter, D.A.S. Rosen, and A.W. Trites.
2006. Food consumption by sea lions: existing data and
techniques. Pp. 177–191 in A.W. Trites, S. Atkinson, D.P.
DeMaster, L.W. Fritz, T.S. Gelatt, L.D. Rea, and K. Wynne,
eds. Sea lions of the world. Alaska Sea Grant College
Program. University of Alaska, Fairbanks.
Winship A.J. and A.W. Trites. 2003. Prey consumption of
Steller sea lions (Eumetopias jubatus) off Alaska: how
much prey do they require? Fish Bull 101:147–167.
Winship A.J., A.W. Trites, and D.A.S. Rosen. 2002. A bioenergetic model for estimating the food requirements of
Steller sea lions Eumetopias jubatus in Alaska, USA. Mar
Ecol Prog Ser 229:291–312.
Worthy G.A.J. 1987. Metabolism and growth of young harp
and grey seals. Can J Zool 65:1377–1382.
Copyright of Physiological & Biochemical Zoology is the property of University of Chicago
Press and its content may not be copied or emailed to multiple sites or posted to a listserv
without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.