A New Model to Estimate Daily Energy Expenditure for Wintering Waterfowl

2005·8.3 MB·English

by  Richard A Mckinney

Wilson Bulletin 117(1);44-55, 2005
A NEW MODEL TO ESTIMATE DAILY ENERGY EXPENDITURE
EOR WINTERING WATEREDWL
RICHARD A. McKinney^ 3 and SCOTT R. MCWILLIAMS^
—
ABSTRACT. Current models to estimate daily energy expenditure (DEE) forfree-living birds are limitedto
either those that use fixed thermoregulatory costs orthose that more accurately estimate thermoregulatorycosts,
but require extensive and often logistically difficult measurements. Here, we propose a model based on basal
metabolic rate (BMR), activity budgets, and site-specific energetic costs ofthermoregulation that requires only
simple measures of ambient temperature and wind speed to provide estimates of DEE. We use the model to
calculate the DEE of Buffleheads (Bucephala albeola) wintering at six habitats that afford differing degrees of
protectionfromexposure withinNarragansettBay, RhodeIsland. Buffleheadactivitybudgetdatacollectedduring
the winters of 2001-2002 and 2002-2003, along with average temperatures and wind speeds at the sites, were
used to calculate DEE that ranged from 46.9 to 52.4 kJ/hr and increased with increasing wind speed. The
energetic cost ofthermoregulation composed as much as 28% oftotal DEE and increased with wind speed. Our
DEE values were 13.4% higher, and thermoregulatory costs were up to 2X higher than those calculated using
an existing model that incorporates fixed thermoregulatory costs. We also saw an increase in feeding activity
with increasing wind speed; sensitivity analysis ofthe effects ofwind speed and feeding activity showed that a
1 m/sec increase in wind speed at oursites increased DEE by 2.5%, whereas acorresponding increase infeeding
activity increased DEE by 4.5%. This suggests that in temperate winter habitats, increased feeding activity may
have a greater impact on Bufflehead DEE than wind exposure. Site-specific model estimates ofDEE could also
provide additional insight into the relative contribution of environmental conditions and changes in waterfowl
behavior to DEE. Received 27 May 2004, accepted 12 January 2005.
The daily energy expenditure (DEE) of a from the influence of habitat characteristics.
species is the sum of basal metabolic rate For example, increased exposure to cold and
(BMR), thermoregulatory requirements, and wind may increase thermoregulatory energy
the energetic cost of daily activities such as costs, and therefore require increased feeding
feeding, locomotion, and social behaviors. to offset higher energetic costs (Bennett and
Quantitative assessments ofthe daily activities Bolen 1978, Hickey and Titman 1983). Mod-
of wintering waterfowl have been used both els that allow comparison between the ener-
to identify important habitats for these species getic costs of thermoregulation and specific
and to assess their response to changes in hab- waterfowl behaviors could be used to deter-
itat quality (Fredrickson and Drobney 1979, mine the relative magnitude of these costs,
Brodsky and Weatherhead 1985a, Baldassarre and may also provide insight into the effects
et al. 1988, Paulus 1988). Waterfowl activity of habitat quality on the DEE of resident wa-
budgets may be influenced by habitat type terfowl.
(Turnbull and Baldassarre 1987, Rave and Traditional measures ofDEE for birds from
Baldassarre 1989) and site characteristics such time-activity budgets use multiples of BMR
as food abundance, protection from exposure, to estimate energetic costs of activities, but
and level of disturbance (Nilsson 1970, Jorde may differ in how the thermoregulatory com-
et al. 1984, Paulus 1984, Quinlan and Baldas- ponent of DEE is estimated (Weathers et al.
sarre 1984, Brodsky and Weatherhead 1985b, 1984). Early estimates of DEE included either
Miller 1985). Changes in waterfowl activity a fixed cost of thermoregulation or one based
may also be tied to changes in DEE that result solely on ambient temperature (Kendeigh
1949, Schartz and Zimmerman 1971, Koplin
' U.S. Environmental Protection Agency, Office of et al. 1980). Models subsequently evolved to
Research and Development, National Health and En- include a means to more accurately estimate
vironmental Effects Research Lab., Atlantic Ecology thermoregulatory costs, but only by the exten-
Div., 27 Tarzwell Dr., Narragansett, RI 02882, USA. sive measurement of many variables (e.g.,
^Dept, of Natural Resources Science, Univ. of
Rhode Island, Kingston, RI 02881, USA. whole-body thermal resistance, forced-con-
Corresponding author; e-mail; vective resistance), some of which may be lo-
[email protected] gistically difficult to obtain for free-living
44
McKinney andMcWilliams • WATERFOWL DEE MODEL 45
wildlife (Pearson 1954, Stiles 1971, Walsberg local conditions, the model may be useful in
1977). Weathers et al. (1984) proposed the use evaluating habitats that provide differing de-
of standard operative temperature, or indices grees of protection from high winds and ex-
that allow single-number representations of treme temperatures. Model estimates could
complex thermal environments, to overcome also be used to provide insight into the rela-
some of these difficulties. However, while tive contribution of environmental conditions
providing a much more rigorous estimate of and differences in waterfowl behavior to
thermoregulatory costs, this approach is lim- changes in DEE.
ited by the need for the construction and cal- In this study, we used our model to estimate
ibration of taxidermic mounts, and may be the DEE of Buffleheads (Bucephala cilbeola)
best suited for aviary or well-controlled field at six wintering habitats in Narragansett Bay,
applications. To date, researchers estimating Rhode Island, that afford differing degrees of
DEE
for free-living birds using published ac- protection from exposure to wind and cold
tivity-based models are limited to either those temperatures. Our specific objectives were to
that use fixed thermoregulatory costs or those (1) compare estimates of DEE obtained using
that more accurately estimate thermoregula- our model with those obtained using a previ-
tory costs, but at the expense ofextensive and ously published model that incorporates a
often logistically difficult measurements of fixed cost of thermoregulation, and (2) ex-
many variables. amine changes in DEE across the sites and
Previous studies estimating DEE for win- determine the relative contribution of wind
tering waterfowl have employed models that speed and waterfowl feeding behavior to
BMR
use factorial increases of and that as- changes in DEE.
sume a fixed cost of thermoregulation (Wool-
METHODS
ey and Owen 1978, Albright et al. 1983, Mor-
ton et al. 1989, Parker and Holm 1990). For DEE site-specific thermoregulation mod-
—
wintering waterfowl in northern areas exposed el. Our model incorporating site-specific
to low temperatures and high winds, thermo- thermoregulatory costs into DEE for winter-
regulation may compose as much as 80% of ing Buffleheads (hereafter, SST model) con-
daily energetic costs (Ettinger and King 1980, sists of (1—) a thermoregulatory component
Walsberg 1983). These costs may vary be- (EEjhermoreg) cstimatc ofthc mctabofic heat
tween wintering habitats because of differing production required to balance heat loss from
degrees of protection from exposure to wind the bird to the environment through conduc-
and cold (Porter and Gates 1969, Goldstein tion and convectio—n, and (2) an activity com-
1983, Bakken 1992). If estimates of DEE are ponent (EEAetivity) at! estimate of additional
to be useful in assessing habitat quality for energetic costs resulting from specific daily
wintering waterfowl, they need to include activities ofwintering Buffiehead expressed as
some measure of the energetic cost of ther- multiples of basal metabolic rate (BMR). We
moregulation based on local environmental sum these components to arrive at an esti-
conditions. mated DEE. In our model, metabolic heat pro-
Here, we present an activity-based model duction includes resting energy expenditure in
that includes habitat-specilic measures ofther- a thermoneutral environment (i.e., BMR) and
moregulatory costs to estimate DEE of water- the additional energy expenditure required to
fowl in different habitats. Our model requires maintain thermal equilibrium. The model uses
only simple measures of ambient temperature average temperatures and wind speeds that co-
and wind speed, along with waterfowl activity incide with activity budget sampling at the
budgets and morphological measurements. sites: DEE is reported in k.I/hr.
Thermoregulatory costs are calculated by us- Basal metabolic rates were estimated from
ing heat loss via conduction and convection those ol’ 16 North American thick species
as a function of temperature anti wiiul spcetl summari/cd in McNab (2003). A plot of BMR
tt) estimate the metabolic heat protluctioii re- versus botly mass for these species gave the
quired to maintain body temperature (Birkc- relation: BMR 4.05M"^‘f where BMR is
bak 1966, Goldstein 1983). Because of the basal metabolic rate in ml (),/hr. aiitl M is
ability to estimate site-specific D1T{ based on body mass in g. L!stimales of BMR were con-
46 THE WILSON BULLETIN • Vol. 117, No. 1, March 2005
verted to kJ/hr using a conversion factor of M is body weight in g and AT is the difference
18.8 kJ/L O derived from the average com- between lower critical temperature and ambi-
position of2,the Bufflehead’s winter diet ent temperature in ° C. The coefficient a is de-
(Schmidt-Nielsen 1997). Body mass was ap- termined under conditions of free convection
proximated at 450 g for males and 325 g for (u = 0.06 m/sec) by the relation:
females (Gauthier 1993). a = Hj- foVa06,
Before calculating metabolic heat produc-
tion, we first determined when this component where Hj is an adjusted metabolic rate in
of a Bufflehead’s DEE is necessary by com- watts at ambient temperature (Goldstein
paring ambient temperature with their lower 1983). We estimated Hj using a heat transfer
critical temperature, or the temperature below model proposed by Birkebak (1966) that cal-
which metabolic heat production is required culates conductive heat loss from different an-
to maintain body temperature (Schmidt-Niel- atomical regions of the bird to the environ-
sen 1997). Lower critical temperature (LCT) ment using geometrical representations (e.g.,
was estimated by the empirical relation: LCT head represented as a sphere, body represent-
M
= 47. where LCT is in ° C, and is ed as a cylinder) and heat transfer theory (Ap-
body mass in g (Kendeigh 1977). We com- pendix; Birkebak 1966). Morphological mea-
pared effective ambient temperature (T^^ or sures of body dimensions (Eig. 1) can be ob-
the ambient temperature corrected for the ef- tained from the literature (e.g., Belrose 1980,
fect of wind speed; Siple and Passel 1945) to Gauthier 1993) or from measurements of mu-
LCT to determine whethermetabolic heat pro- seum specimens. Average values for live Buf-
duction would be required to maintain the fleheads {n = 4, obtained from the Connecti-
duck’s body temperature. If T^/ was less than cut Waterfowl Trust, Earmington, Connecti-
=
LCT, we assumed that metabolic heat produc- cut) and Bufflehead study skins {n 16, ob-
tion was required to maintain body tempera- tained from the Harvard Museum of
ture; we then calculated this energy require- Comparative Zoology, Cambridge, Massachu-
ment and included it in the final DEE. On the setts) are summarized in the Appendix. Also
other hand, if T^^ was greater than the lower summarized in the appendix are the equations
critical temperature, we did not include met- drawn from Birkebak (1966), which were
abolic heat production. Effective temperature used to calculate metabolic heat production.
was calculated using the relationship derived For these equations, a heat transfer coefficient
by Siple and Passel (1945): {k) of0.102 cal/cm/° C was used for the entire
body surface (Calder and King 1974). The
Tef = T, - {T, - rj thermal conductance of Common Eider {So-
X (0.474 + 0.239 X Vu - 0.023 X w), cmaotnedruicatamnoclel)isshiamsa)bienenwatsehro(wi.ne.,twoetbtehe5rm7a%l
where T^j is the effective temperature (° C) greater than it is in the air (Jenssen et al.
used for comparison with the lower critical 1989); therefore, we used a heat transfer co-
temperature, T;, is body temperature (° C), T^, efficient of 0.160 cal/cm/° C to calculate heat
is ambient temperature (° C), and u is wind loss from the ventral body surface to the wa-
speed (m/sec). ter. Metabolic heat production was calculated
If T^f was less than LCT, we used an em- as: BMR + Qhead Qneck Qbreast Qbody
BMR
pirical model to estimate metabolic heat pro- Q,e„.r3i surface, where is basal metabolic
duction as a function oftemperature and wind rate and Q is the heat loss term for each body
speed (Goldstein 1983): component.
Estimates of additional energetic costs re-
sulting from specific daily activities (EEActivity)
where u is wind speed (m/sec) and is were calculated by multiplying the proportion
metabolic heat production (watts). The coef- of time spent in a particular activity by the
ficient b is determined empirically from data energetic cost of that activity. We used pre-
summarized by Goldstein (1983) on seven viously reported multiples of BMR, summa-
species of birds (body size 13.5-3,860 g) by rized in Table 1, to calculate the energetic
the relation: h = 0.0092M°^^ X where costs of activities by multiplying the propor-
McKinney and McWilliams • WATERFOWL DEE MODEL 47
FIG. 1. Body dimension measurement points required for input into the SST model to estimate DEE (see
Appendix). A = head length, B = head height, C = head width, D = body width, F = body length, G = body
height, H = neck length, I = neck width, J = neck height.
TABI.E 1. Energetic costs as a multiple ofbasal metabolic rate (BMR) ofactivities used in the site-specific
and fixed-cost thermoregulation DEE models.
Multipleof
Activity Opcratioiuilcletinition BMR Retorenee
Dive Diving for food 5.\ de Leeuw I9db
Surface Surface and pause between dives .^.1 de Leeuw 199b
Look Peering through the water at the cove bottom I.S Wooley and Owen I97S
Courtship .Social tlisplay toward indi\idual of the opposite gender 2.4 Albright et al. I9S.^
Agonistic Hostile interaction between two itidi\idiuiK I.S Wooley and Ovsen I97S
Switn I.ocomotion Butler 2()()()
Fly Locomotion \2.5 Wooley and Owen I97S
Preen Maintenance of feathers 2.1 Albright et al. I9S.^
Alert Not moving, but actively observitig surroundings I.S Wooley and Owen I97S
Rest Not moving with bill tucked in feathers 1.4 Wooley and Owen I97S
—
48 THE WILSON BULLETIN • Vol. 117, No. I, March 2005
tion of time spent in that activity by the cor- 965 observations on individual birds, resulting
responding multiple of BMR. The contribu- in over 80 hr of activity budget data. Obser-
tion of physical activity to DEE (Table 2; vations were randomly distributed over sam-
EE^^^ti^jjy) was then calculated by summing the ple sites and time during the daytime through-
energetic costs of all activities in which Buf- out the winterperiod when ducks were present
fleheads engaged. (November-April). We chose individual
DEE fixed-cost thermoregulation model. ducks at random (i.e., observations began with
Estimates of DEE were calculated using a the ith duck from the left in each flock, where
method that incorporates a fixed cost of ther- i was a randomly generated number) and ob-
moregulation (fixed-cost model; Morton et al. served through a 32-60X spotting scope or
1989). In this model, the thermoregulatory through 10 X 50 binoculars for 5 min; behav-
component (EE^hermoreg) is fixed and estimated iors were categorized as dive, surface, look
at 5.9 kJ/hr (Morton et al. 1989). Additional (i.e., peering through the water at the cove
energetic costs resulting from specific daily bottom), courtship, agonistic, swim, fly, preen,
activities (EE^^-tj^ity) were calculated as in the alert, and rest (Table 1). Preening included
SST model by multiplying the proportion of wing flapping, stretching, and scratching.
time spent in a particular activity by the en- Gender for each individual was identified
ergetic cost of that activity. These two com- when possible, except in rare instances when
ponents were then summed to arrive at fixed- we were unable to distinguish between fe-
cost model estimates of DEE. males and first-year males that had not yet de-
—
DEE-habitat correlations. We identified veloped breeding plumage (Carney 1992).
six Bufflehead wintering habitats within well- Therefore, we report results for “males”
defined coves or embayments of the Narra- (showing breeding plumage) and “females”
gansett Bay estuary. Included were two me- (includes first year males). Activity data were
sotrophic, rocky- and sandy-bottom embay- collected using an observational software pro-
ments (Sheffield Cove: 41° 29' 41" N, 71° 22' gram installed on a laptop computer (JWatch-
89"W; and Mackeral Cove: 41° 29' 28" N, er. Animal Behaviour Laboratory, Macquarie
71° 20' 86" W), two mesotrophic soft-bottom University, Australia; http://www.jwatcher.
coves (Coggeshal Cove: 41° 39' 32" N, 71° ucla.edu/). Prior to analysis, data were aggre-
20' 52" W; and Brush Neck Cove: 41°41' 47" gated into the following categories: feeding
N, 71° 24' 48" W), and two eutrophic soft-bot- (dive, surface, look), social (courtship, ago-
tom coves (Apponaug Cove: 41°41' 40" N, nistic), locomotion (swim, fly), maintenance
71° 28' 58" W; and Watchemoket Cove 41° (preen, alert), and resting. Each sampling
48' 00" N, 71° 22' 75" W). Cove areas ranged event at a site consisted of 20-30 five-min
from 18.6 to 86.1 ha, with an average of42.2 observations; final data were averaged by
ha. Each cove supported consistent numbers sampling event and by—site.
ofBuffleheads throughout the winter (Novem- Sensitivity analysis. We used linear re-
ber through April); the median flock size at gression analysis of SST model estimates of
the six sites (determined by bimonthly cen- DEE versus wind speed and feeding behavior,
suses during the winters of 2001-2002 and respectively, to assess the relative contribution
2002-2003) was 18, ranging from 13 to 41. of each to DEE. First, we estimated DEE us-
In winter, Buffleheads spend the majority of ing average values of feeding activity across
their time on the water and tend to favor shal- all sites, and plotted DEE versus wind speed
low water habitats (<3 m) in protected coves over the range ofwind speeds recorded during
(Stott and Olson 1973, Gauthier 1993). They the study (i.e., feeding activity held constant,
feed by diving to the cove bottom where they wind speed varied; regression equation: DEE
consume benthic invertebrates including crus- = [1.1 X wind speed] + 42.1). Second, we
taceans, gastropods, and bivalves (Yocum and estimated DEE using average wind speed and
Keller 1961, Wiemeyer 1967, Gauthier 1993). temperature across the sites and plotted DEE
We used focal animal sampling to quantify versus the proportion of feeding activity (i.e.,
activities of Buffleheads at each of the study wind speed held constant, feeding activity var-
sites during the winters of 2001-2002 and ied; regression equation: DEE = [43.1 X pro-
2002-2003 (Altmann 1974). We completed portion oftime spent feeding] + 17.3). In each
• • <
McKinney and McWilliams • WATERFOWL DEE MODEL 49
al.
Buffle- temperature et 0(0N On C>Nn 00 in OiNn od cnon' OONN d rN-O q00 (qN 00 ioni dq dq 00 d qd q00
of Morton QUPJJ +1 O+N1 i+n1 +1 +1 0+01 do+1 O+N1 q+1 d(+N1 d+1 qd+1 dq+1 q+1 dq+1 dq+1 q+1 dq+1 dq+1 dq+1 dq+1 11 1 11
(d rd (N (N T— (N ri
(DEE) 'It 'it 'it 'It 'it 'it 'It of of
model; c
site-specific
3
expenditure E 00 d00 dO din 0i0n dOn do 0r-0; d(N dq dq dq do 0q0 dq dq do dq dq qd
on
thermoregulation -C W< (+N1 od+1 N+O1 (d+Ni i+n1 O+n1 +1 o+1 +1 q+1 (+N1 q+1 q+1 +1 qd+1 Nd+O1 qd+1 0+01 qd+1 f+-1 qd+1 11 11 11
based ON m NO NO NO NO NO 0m0 mON m00 0m0 NmO m m m NO m NO
energy E*s C) r<-) CO C) cn fO ro ro Cove.
costs
daily Watchemoket
and (fixed-cost HcB vOiN IOTN) dOn dOn OiNn OdN Odn OdN qd qd qd qd qd qd qd qd qd qd qd qd qd 11 11 11 =
UUP
thermoregulatory
WATCV
(EEAc,,vity)» costs
0(N0 ON <dN 0d0 idn din od NO OO)n 'dt cN-D q00 qri 00 ioni qd dq 00 d qd q00 ONOfO iionn m(ON Cove;
activities usingthermoregulatory QUUJP Oc+n-1 0—+01 0O+0n1 Ni+On1 d(+N1 NO+ON1 Od+N1 Or+-n-1 0q+01 dO+n1 dq+1 dq+1 0q+01 dq+i qd+1 or+ri-j1 0q+01 00+001 00+001 o—+i1 oo+i1 dII dII dII She=ffield
't in in in of of of Ck ct. A.
calculated
daily
SHFCV
3
3
specific were fixed '2tUij i- 00 0d0 qd idn 0i0n Od) qd r0-0 rdsj qd qd qd qd 0q0 dq dq od dq d(N dtn Cove;
from Values using S*£ UP< (o+Ni1 os+1d Nd+O1 m(d+N1 iN+nO1 ON+nO1 mN—+O1 moN+O1 mN—+O1 q0+01 m(O+NN1 mq0+01 mqo+61 mmN+O1 mqd+1 qd+1 qd+1 0N+0O1 qd+i rn+-C1 mmd+1 1 1 1 Mackerel
1 r^i tn <n fO ro ro =
method
1-2003. 1
resulting MAKCV
o 00 o c e't (N (N m ON o <N O fN r-
costs I2sl0an0d, praevious UuPu51 '+C1) o—+ii —+1 O+n'1 r+-1; o+1 0+01 d+i rq+i1 qn+i1 qd+1 dq+1 oN+iO1 q+1 q-+1 q+1 dt+o1 ro+ii q+1 di+n1 q+1 1 dAqnI.IC 1 CCooggveesh;al
energetic and
-
Rhode
model), o O o
(EEnK-rmoreg)-Bay. dS —r3 uEuuu 'UcEVE 1 UuEVL dEcOP 1 uEI-P UE•c<ou _C4ZP 7Z uEOP u5PO-P UicIoP X<Eu Ui•cIooP 1 tIEi!P UitcIcJP 1jp ajOE.pP cE-OOP NCCOoeGvcCekV;
Narragansettthermoregulation
costs
Brush
SD.
in ± V CV
V
aA. BRI
Thermoregulatoryhabitats (site-specifickJ/hr
Cove;
in
£
winter are
Apponaug
2. speed J'.
six Values
TABLE at wind > u>u O> >d U> u> o7X \VPFX
heads and 10S9). a. aX; u/-N < lIcln 7? o
50 THE WILSON BULLETIN • Vol. 117, No. 1, March 2005
case, we used average values of temperature
and all other activities in the model. Regres-
sion equations generated from each analysis
were used to estimate the relative contribution
of wind speed and feeding behavior to DEE.
For wind speed, we calculated the average
percent increase in DEE per 1 m/sec increase
in wind speed. For proportion of time spent
feeding, we calculated the average percent in-
crease in DEE per 5% increase in the propor-
tion of time spent feeding.
—
Statistical analyses. Differences in the
proportion oftime spent on different activities
by males versus females were investigated us-
ing two-tailed Student’s Utests on data aver-
aged across all sample sites. Site-specific time
budgets were calculated by averaging individ-
ual observations by sampling event and then
by averaging sampling events by site. Propor-
tions were arcsine-square-root transformed
prior to regression analysis (Fowler et al.
1998:87-88). Wind speed and temperature 0.0 1.0 2.0 3.0 4.0 5.0
were averaged by sampling event and by site. Windspeed(m/sec)
Regression analysis and analysis of varianee EIG. 2. Correlation of wind speed with (A) DEE
were used to assess the influence of environ- and (B) time spent feeding for Buffleheads (males and
mental eonditions on DEE and feeding behav- females combined) wintering at six coastal habitats in
ior. Statistieal analyses were performed using Narragansett Bay, Rhode Island, 2001-2003. Wind
SAS (SAS Institute, Inc. 2001). speeds are means of all sampling sessions conducted
at a site. Error bars are ± SE.
RESULTS
Estimates ofDEE forwintering Buffleheads males and females combined (P = 0.76, P =
generated using the SST model averaged 49.0 0.023; Fig. 2A). The proportion of time spent
± 8.4 kJ/hr, or 1,176 ± 202 kJ/day, and dif- feeding by Buffleheads also inereased with in-
fered by up to 12% among sites (Table 2). The creasing wind speed (r- = 0.67, P = 0.047
mean thermoregulatory component of DEE Fig. 2B). Estimates of DEE that were gener-
(EExhermoreg; Table 2) was 11.7 ± 1.1 kJ/hr or ated using the fixed-cost model showed no re-
23.9% of total DEE; EEyhermoreg increased with lationship between DEE and wind speed.
increasing wind speed (r- = 0.61, P = 0.067). Buffleheads spent 75.7 ± 4.3% oftheirtime
DEE
did not differ between males and fe- feeding during daylight hours, and females fed
males; however, thermoregulatory costs were more often than males (77.1 ± 5.4% versus
higher for females (mean = 12.5 ± 1.2 versus 74.2 ± 6.9%; ^545 = -2.6, P = 0.004; Table
10.9 ± 1.0 kJ/hr for males; t^ = —7.2, P < 3). Males, however, spent more of their time
0.0T0h1e).mean DEE (all sites) calculated using 0e.n4g3a%g;ed^5i45n =cou7r.t4s,hiPp<act0i.v0i0t1i)es. (M2a.l3e9s%avnedrsfues-
the SST model was 13.4% higher than that males (combined) averaged 16.8% of their
calculated using the fixed-cost model (Table time engaged in loeomotion and maintenance,
2). The thermoregulatory component of DEE, and 4.5% of the time resting (Table 3). Buf-
5.9 kJ/hr, composed 13.7% of total DEE cal- fleheads at Mackerel Cove spent the greatest
eulated with the fixed-cost model. proportion of time feeding and the least in all
Daily energy expenditures of Buffleheads other activities, whereas those at Coggeshal
ealculated with the SST model increased with Cove spent the least time feeding and the most
increasing wind speed for males (P = 0.67, P in all other activities, except resting. Overall,
= 0.046), females (U = 0.64, P = 0.055), and Buffleheads spent between 0.3 and 2.6% of
' i
McKinney andMcWilliams • WATERFOWL DEE MODEL 51
•CO*a0 (CNN X <XN Xq dq dq rqi
aEj 3 Xin XrN dX dX On Xr- qin
<
i
0o—0 mdX rP- XXr- q0d0 OdN idqn qd XXd idqn qod mXq qXX oq qdX XXd, qod qX— q— Xq
1cOSX) XXi+1n (X+N1 roi+1nq qi+1n (qxX+N'1 XrX+-1 XXq+1 q(i+Nn1 mXi+n1 d(q+N1 dq+1 (dX+N1 XXi+1n qdX+1 xXXX+’,1 (qd+N1 0Xm+0i1 (Xm+N,1 qoX+o1 qXx+’1 rXX+i1 III III WCatochvemeok.et
ai
'XX7. (cU isno XOdN OXrNn XXrn <qdN 0Xd0 _qx' (xqN' oXr-- dr-- XqX _qd iqinn Xd XqX (mdN 0Cd0- OX qd_ 0q—0 rodi —
Occr>J3 1sc q+1 XX+1 0Od+0N1 O(i+NNn1 d+1 00+001 O0O+n0N1 OOX+N1N OOX+nN1 q+1 miq+1n Cdq+M1 do+1 dqm+1 CXX+M1 00q+001 Xrn+i-,,1 md+1 0qX+01 XrX+f1 qd+1 I 1 I I WCAoTvCeV;
flj
qcE- ’aCr>J3 -S iXn X00 (rN- iOn x0X'0 Xq qin dq qx’ 0di0n iqdn dOrN- imn qin OdqN ir<n~, OXXN OqXN dX rdin-, irqni, She=ffield
XV ECC3Q3 ECQJ (r+-N1 dXO+N1 _X+1 0X+01 0O+01n d0+01 Xq+1 dq'+1' dX+1 iX+n1 N+O1 qi+1n oXX+1 XiO+nN1 XXtl dq+1 X0d+01 Oo—+n<1 Xd+1 q0_+01 idi+nn..1 I I I 1 I SHFCV
tJj Cove;
Cr>3 in d— — o 0i0n X00 Xrd- Xd XOdN Xdr- <OdNn dm(N 0dq0 <drNf qod dXX dqX od r0-0, X0d0 rdq'. Mackerel
OcJ. •3 (N (N
X51) 0C0 Xq+1 dii+1nn (—o+N1 O(0+s0d1 qr—+-1 (mX+N1 iXq+n1 d0—+01 d—+1 Xd+1 0d—+01 odr+r-)1 0q+01 dX+1 qX+1 rrC+<'~i,,1 0qo+01 r—X+j1 ro+i1 rXO+~,1 X—.+—1 I I 1 I MAK—CV
OJ
_5Jj
X1) Xin qd (rNn dX dX q00 dq X— rX^, dX q00 X— (i'nI',, X_— d Q— rdj 0X0 SX' iXn xr-'. C'ove;
Ho£ uEi XXX+1 0X—+01 dmr+--1 rrX+i'1 do+1 nqr+-i-1 0qr+-01- oqX+o1' dqr+--1 0qi+0n1 d0—+01 d0+01 dr0+-01 (0q+^01i r0C+^0'.1 0XX+01 rrd+j-1 rro+i-1 Xrr+j~1 drq+~1 iqr+-n1 I I C'ogg—eshal
COCiCV
~ojj:O;J “(uX;(U ~1)x.1E) XojX.lEi XD X.E. X5jX.ur XO XX Cove;
1) }:i O ^
^^U5-j UP ^Su5-UP 52 uS-PU ^ 5 P pi o p i^oxOp Neck
Brush
cD 0-2 BKl'CV
C.
C’nvc;
.\pp«>n;iiig
z
r^:
2<C w- >O >ud >uo od o> AI'K'V
H '£ S<I a: u0 q
X. j'.
52 THE WILSON BULLETIN • Vol. 117, No. 1, March 2005
their time in social activities, and this time haviors, as is the fixed-cost model. We applied
decreased as the ducks increased feeding (r^ the model to Buffleheads, but were restricted
== 0.71, P = 0.043). Similarly, the amount of to using literature-based energetic equivalents
=
time spent in maintenance activities (range that were not specific to that species. There-
3.7-9.7%) decreased as time spent feeding in- fore, the DEE estimates presented here, while
creased (r^ = 0.96, P = 0.001). higher than those calculated from the fixed-
Sensitivity analysis of model estimates of cost model and from body mass alone, fall
DEE versus wind speed at constant proportion well within the probable errorof20-40% pro-
of time spent feeding showed that a 1 m/sec posed by Weathers et al. (1984) for models
increase in wind speed resulted in a 2.5% in- that rely on generic energetic equivalents.
crease in DEE. Analysis of DEE versus the However, while it would be difficult to argue
proportion of time spent feeding at constant that our model estimates are more accurate
wind speed showed that a 0.05 increase in than those calculated from fixed-cost or body
proportion of time spent feeding resulted in a mass models, the utility of our model lies in
4.5% increase in DEE. the ability to determine the relative contribu-
DISCUSSION tion of wind speed, temperature, and specific
Our estimates ofDEE using the SST model waterfowl behaviors to DEE across sites with
different environmental conditions and levels
for Buffleheads at the Narragansett Bay win- of activity.
tering sites (1,175 ± 202 kJ/day) are higher Wintering waterfowl may incur substantial
than those predicted from the fixed-cost model
thermoregulatory costs depending on ambient
(1,036 ± 202 kJ/day), which uses a single en-
temperatures and the combined effect ofwind
ergetic cost of thermoregulation. Thermoreg- and cold, and these may lead to increases in
ustliattuotreyucposttos p2r8ed%icotfedtbhey tahneimSaSl’TsmtoodtaellDcoEnE- DEE. Changes in the relative amounts of ac-
tivities exhibited by wintering Buffleheads
and are approximately twice as high on av- may also alter DEE. In our study, estimates of
eDrEagEe as that used in the fixed-cost model. DEE calculated with the SST model were cor-
estimates for Buffleheads at our sites
related with wind speed (Fig. 2A). However,
were also higher than a field metabolic rate
feeding activity also increased with increasing
predicted by an allometric relation of energy
wind speed (Fig. 2B), which could also con-
expenditure based on empirical studies (606
kJ/day, non-passerines; Nagy et al. 1999). tribute to an increase in DEE. Sensitivity anal-
rHeolwaetivoenr,wamsandyeroifvetdhewsetruediecsarfrrioedm wouhticihntthhies yasnids foefedthiengefafcetcitvsitoyfoinncDreEaEsesshionwweidndthastpeiend-
breeding season in warm ambient tempera- creases in feeding activity resulted in a rela-
tures, so our higher DEE estimates may be tive increase in waterfowl DEE nearly twice
attributed in part to environmental conditions that of a corresponding increase in wind
and the inclusion of thermoregulatory energy speed. Feeding activity may increase because
of decreased prey abundance, or because of
costs.
Our model does not include the contribu- changes in the availability orenergetic content
tion ofheat gained from solarradiation orheat of prey. Further studies at our sites have
lost through evaporative water loss because shown that feeding activity increased with de-
these effects are likely relatively small (less creasing prey abundance, and also with de-
than 10% ofheat loss; Scholander et al. 1950, creasing prey energy density resulting from
Strunk 1971, Wolf and Walsberg 2000), and changes in available prey species at a site and
were likely similar between our study sites. inter-specific differences in the energetic con-
Nonetheless, these constraints limit the appli- tent of prey (RAM and SRM unpubl. data).
cation of our model to comparative, single- However, other factors, such as intra- and in-
species studies between habitats that are lo- ter-specific competition and increased ener-
cated in a similar geographic region. It is also getic demands, may also influence the amount
important to note that the SST model is lim- offeeding activity. Although we are uncertain
ited by the availability of empirically derived as to the cause ofincreases in time spent feed-
energetic equivalents ofspecific waterfowl be- ing at our sites, ourresults show that increased
McKinney and McWilliams • WATERFOWL DEE MODEL 53
feeding activity may have a greater impact butes such as prey abundance, or degree of
than wind exposure on DEE of Buffleheads. protection from high winds and extreme tem-
Increased feeding activity may also affect peratures. However, further studies will be
the short- and long-term survival of Buffle- needed to establish the independence of be-
heads. For example, if wintering Buffleheads havioral responses to environmental condi-
need to spend more time feeding, time foroth- tions from the primary effect ofthe conditions
er activities such as courtship and pair for- themselves on the DEE of resident waterfowl
mation may be limited (Drent and Daan 1980, before model estimates can be used in habitat
Meijer and Drent 1999). Although they exhib- assessment.
it long-term pair bond formation and a high ACKNOWLEDGMENTS
degree of flock synchrony, which results in a
relatively small proportion oftime spent in so- We would like to thank K. Bannick and B. Timm
cial behaviors, courtship and maintenance ac- for assistance collecting field data. Access to Buffle-
tivities are still important for their overall re- hateiavde sZtouodlyogskyinwsasatotbhteaiHnaerdvawridthMuthseeukimndofasCsoimsptaanrc-e
productive success (Robertson et al. 1998). of J. Trimble and A. Pirie. We also thank K. Appier
Our results indicate that as Buffleheads at our and the Connecticut Waterfowl Trust foraccess to live
sites spent more time feeding, they had less birds, and D. T. Blumstein, C. S. Evans, and J. C.
time available for maintenance and social be- Daniel for assistance with behavioral data analysis. S.
haviors, which may have an impact on both Walters, S. A. Ryba, C. Wigand, and T. R. Gleason
their short- and long-term survival. This, cou- provided insightful comments on the manuscript. We
also thank M. R. Miller and two anonymous referees
pled with the greater increases in energetic
for thoughtful comments on an earlier version of this
costs due to feeding activity predicted from paper. Mention oftrade names orcommercial products
model sensitivity analysis, suggests that DEE in this report does not constitute endorsement or rec-
of wintering waterfowl in harsh climates ommendation. Although the research described in this
would be lower in habitats with both high article has been funded wholly by the U.S. Environ-
prey density and adequate protection from ex- mental Protection Agency, it has not been subjected to
posure. For example, sites such as Brush Neck Agency-level review. Therefore, itdoesnot necessarily
reflect the views ofthe Agency. This paper is the Of-
C(oRvAe,MwahnidchShRaMd tuhnepuhbilg.hesdtatap)reaynadbualnsdoantchee afincde EonfviRreosnemaerncthalanEdffDeectvselRoepsmeeanrtc,h NLaatbioornaatloryH,ealAtth-
lowest thermoregulatory costs for Buffleheads lantic Ecology Division contribution numberAED-04-
(Table 2), may be better candidate sites for 074.
protection as waterfowl wintering habitats LITERATURE CITED
compared with sites such as Mackerel Cove,
which had low prey abundance and high ther- Albright,J.J., R. B. Owen,Jr.,and P. O.Corr. 1983.
The effects ofwinter weatheron the behaviorand
moregulatory costs.
In summary, our SST model estimated DEE eacnteirognysroefsetrhveesNoorfthBelaa.cstk SDeucctkisoninofMatihnee.WiTlrdalnisf-e
as the sum of basal metabolic rate and site- Society 40:1 18-128.
specific energetic costs of activity and ther- Altmann, j. 1974. Observational study of behavior:
moregulation. The primary benefits of the sampling methods. Behaviour 49:227-26.‘i.
SST model compared to other approaches in- Barken, G. S. 1992. Measurement and application of
operative and standard operative temperature in
clude its ability to (1) evaluate the effect of ecology. American Zoologist 32:194-216.
thermoregulatory costs on DEE of wintering Baldassarre, G. A.. S. L. Pallus, A. Ta.smisii;r, and
waterfowl using simple measurements ofwind R. D. Titman. 1988. WorkslK>p summary: tech-
speed and ambient temperature, (2) predict the niques for timing activity ofwintering waterfowl.
extent to which the behavior ofwaterfowl dur- Pages 181-188 in Waterfowl in winter: selected
ing winter affects DEE, and (3) track changes Gpaalpveersstofnr.omTexsaysm,p7o-sIiOu.ImanaunatirywIo9r8k.Ssh(oMp. Wh.elWdeli-n
in DEE over different time scales (i.e., hourly, ler, lid.). University ol Minnesota Press. Minne-
daily, or seasonally) if the corresponding ac- apolis.
tivity and environmental data are available. Bi i ROM. p; 1980. Ducks, geese, and swans of North
Also, because of its ability to estimate site- America, 3rd ed. .Stackpole Books. Harrisburg.
Pennsylvania.
smpoedceillicmaDyEEb^ebuasseefudloinn elvoaclaulatcionngditthieonqsu,alitthye Benniinrwi.in.1t.eWr.inagnGdrelie.nG-.wBionlgeetnl.T1ea9l7.8.Jo.uSrtrneaslsroefsWpioknls-e
of waterfowl habitats that have different attri- life Management 42:81 86.