Metabolic response to wind of downy chicks of Arctic-breeding shorebirds(Scolopacidae)

Many shorebird species (Scolopacidae and Charadriidae) migrate long distances to breed on the Arctic tundra(Piersma et al., 1996). On arrival, energy reserves of individuals are small, and consequently the energy needed to provide for eggs and chicks comes from the brief summer abundance of arthropod food items (Järvinen and Väisänen, 1978; Klaassen et al., 2001). Shorebird chicks leave the nest soon after hatching to follow their parents and forage for food (Nice,1962). Success requires chicks to grow to maturity quickly while food is available, and to do this while exposed to the frequently cold and windy Arctic microclimate.

Because downy chicks are poorly insulated compared to adults(McNabb and McNabb, 1977; Taylor, 1986; Visser, 1998), foraging and thermoregulation costs are high, accounting for 25-30% of the total energy budget (Schekkerman and Visser,2001). Moreover, foraging requires physiological maturity of muscle tissue for effective locomotion and thermoregulation, while rapid growth is typically associated with physiological immaturity(Ricklefs et al., 1994; Visser and Ricklefs, 1995; Starck and Ricklefs, 1998; Krijgsveld et al., 2001). This conflict limits the peak metabolic rate of young chicks.

Thus, foraging chicks often cannot, or at least do not, maintain a stable body temperature (T_b). Chicks of some species can continue foraging until T_b falls as low as 30°C(Norton, 1973), but nevertheless parental brooding eventually becomes necessary. Because brooding prevents foraging by both the chicks and the brooding parent, adverse weather can depress growth (Beintema and Visser,1989a; Schekkerman et al.,1998; Schekkerman and Visser,2001). To understand the behavior, physiology and energetics of Arctic breeding one must therefore relate foraging time and thermoregulatory costs to thermal conditions. Given the scarcity of shelter on the tundra, wind is likely to be a significant factor in the cost of Arctic shorebird thermoregulation (Piersma and Morrison,1994; Wiersma and Piersma,1994). Although some authors have remarked that down seems rather sensitive to wind (Taylor,1986; Visser,1998), the effect of wind on downy chicks has received but limited attention (Bakken et al.,1999), and no data are available on shorebird chicks.

We measured oxygen consumption in relation to temperature and wind speed in downy scolopacid chicks of different ages varying in mass by an order of magnitude. Chicks were from species with small, medium and large adults: least sandpipers Calidris minutilla (approx. 20-30 g), shortbilled dowitchers Limnodromus griseus (approx. 100-130 g) and whimbrels Numenius phaeopus (approx. 300-400 g). Heat loss from the legs and bill during cold stress is minimized by countercurrent heat exchange in a wide range of species (Ederstrom and Brumleve,1964; Steen and Steen,1965; Kilgore and Schmidt-Nielsen, 1975; Hagan and Heath, 1980; Midtgård, 1980). Consequently, the variation in relative leg and bill length of growing shorebird chicks of different species is probably unimportant. As their morphology is otherwise similar, we describe thermoregulatory responses by a single allometric model. We also use our data to define a standard operative temperature (T_es) scale that may be useful in field studies of shorebird chick behavior and energetics. Standard operative temperature is the temperature of a standard respirometer with essentially no wind, within which the same animal would have the same net heat production(Bakken, 1976, 1992).

We conducted this study at the Churchill Northern Studies Centre,Churchill, Manitoba, Canada, during the summer of 1997. We collected eggs from different clutches of least sandpiper Calidris minutilla Vieillot,short-billed dowitcher Limnodromus griseus Gmelin and whimbrel, Numenius phaeopus L., and transported them to the laboratory, where they were incubated at 37°C and 80% relative humidity. After hatching,chicks were kept in boxes with an elevated wire mesh floor over sawdust. A light bulb in one corner of each box allowed chicks to thermoregulate behaviorally until they were fully homeothermic at room temperature. The lights were then removed. Food (turkey starter, canned tuna, boiled eggs and fresh insects) and water were provided ad libitum (for details, see Krijgsveld et al., 2001). Eggs were collected under permit from the Canadian Wildlife Service, and laboratory procedures were approved by the Institutional Animal Care and Use Committee of the University of Missouri — St Louis.

Experiments were begun when body mass was equal to or greater than 9 g(approx. 7 days post-hatching) for least sandpipers (N=4), 20 g (5 days) for dowitchers (N=8), and 63 g (7 days) for whimbrels(N=4). Studies were continued until contour feathers began to appear,at which time body mass was less than 18 g (14 days) for least sandpipers, 56 g (19 days) for dowitchers and 109 g (11 days) for whimbrels.

We measured T_b at 1 min intervals to within±0.1°C with a 36AWG (0.13 mm conductors, 0.14 mm×0.69 mm overall) Teflon®-insulated duplex thermocouple (type TT, Omega Engineering, Stamford, CT, USA). We looped one full turn of the wire through the thread holes of a clothing button, cemented it in place with the sensing junction extending 1-2.5 cm beyond the button, and rounded the sensing junction with a drop of epoxy cement. We matched button diameter and probe length to the size of the chick. At the time of the experiment, we lubricated the sensing junction with petroleum jelly, and inserted it into the cloaca. We then secured the assembly by folding several nearby feathers over the button and gluing them to the button with cyanoacrylate adhesive. At the end of the experiment, the glue was dissolved with acetone to release the feathers so that the thermocouple could be removed. Because some down was lost each time,we limited body temperature measurements to approximately 40% of our runs to avoid increasing normal rates of heat loss. We monitored T_b during the first experiments in each size class to avoid conditions leading to hypothermia. We also monitored T_b on a random sample of chicks thereafter.

Our experimental conditions excluded solar radiation and simulated the cold, windy conditions experienced by chicks during storms, heavy overcast days, or the brief Arctic night. Our small-volume wind tunnel metabolism chamber was limited to winds of speeds less than 3.5 ms^-1(Bakken et al., 1989). However,this range of wind speed is sufficient to determine the quantitative trend of wind effects. Further, wind speed varies logarithmically with height(Campbell et al., 1980), such that wind measured at 10 cm height over tundra is only 20-35% of that measured 3 m above the surface (G.S.B., unpublished data from four tundra sites near the Churchill Northern Studies Centre). Thus, our data can be used directly with standard wind speeds up to 8-15 ms^-1 (30-50 km h^-1). Finally, internal temperature gradients and peak metabolic rates vary with body temperature, which complicates accurate measurements of thermal conductance on cooling animals(Eppley, 1994; O'Conner, 1999). Consequently,we used experimental air temperatures chosen such that chicks were normothermic but below their lower critical temperature at all wind speeds.

We measured oxygen consumption and evaporative water loss at four wind speeds (u; 0.1, 0.8, 1.8 and 3.0 ms^-1) and at least two air temperatures (T_a) between 15° and 30°C, the exact value of T_a depending on the age and mass of the chick. Procedures generally follow those detailed in earlier studies (Bakken et al., 1991, 1996), with some changes in instrumentation. We measured oxygen consumption using an open-circuit system supplied with dry, CO₂-free, outdoor air at 1.11 min^-1,measured using a precision rotameter (Brooks Full View, Emerson Electric,Hatfield, PA, USA) calibrated on-site against a bubble meter(Levy, 1964). We measured the oxygen content of a dry, CO₂-free subsample (200 ml min^-1) of chamber air with an electrochemical oxygen analyzer(FC-1, Sable Systems, Henderson, NV, USA) and the water vapor density of a separate subsample (200 ml min^-1) using a dewpoint hygrometer (880,EG&G, Waltham, MA, USA). We corrected all values to standard temperature and pressure (STP) using an electronic readout barometric pressure transducer(Model 2014-27/31.5, YSI Instruments, Yellow Springs, OH, USA). We calibrated fan-shaft revs min^-1, measured with a dedicated phototachometer,against wind speed, measured with a Prandtl-design Pitot tube and electronic hook gauge (Dwyer Instruments, Michigan City, IN, USA). We measured various temperatures with copper-constantan thermocouples (type TT, Omega Engineering,Stamford, CT, USA). A digital data logger (CR-21X, Campbell Scientific, Logan,UT, USA) recorded T_a, T_b and rotameter temperature, analyzer outputs and fan revs min^-1. Control outputs were used to automatically step through a predetermined randomized sequence of wind speeds and switch a solenoid valve at 1 min intervals to alternate sample and reference gas flow to the FC-1 for continuous baseline correction. The FC-1 equilibrated within 40s of switching, including washout time for the analyzer gas scrubbers. We used Z-transform methods to compute `instantaneous'oxygen consumption and evaporative water loss for each 2 min sample interval(Bartholomew et al., 1981; Bakken, 1991a).

After weighing the chick, we installed the cloacal thermocouple and placed it in the chamber, before allowing 30 min for it to recover from handling. We then conducted measurements for 30 min at each wind speed. At the end of the run, we removed the thermocouple, reweighed the chick, and returned it to its box no more than 3.5 h after it was first removed. Because we needed to maintain normal growth patterns while making maximum use of a limited number of chicks, we made no attempt to ensure that chicks were post-absorptive. Consequently, the respiratory exchange ratio may have varied during the run. The order of wind speeds was randomized during each series to minimize any resulting bias in our estimates of the effect of wind on heat loss. Total chick metabolism included basal, thermoregulatory and active metabolism, the heat increment of feeding (SDA) and growth. We assumed full thermoregulatory substitution of all forms of metabolic heat production, which may or may not be correct (Hart, 1952; Paladino and King, 1984; Klaassen et al., 1989; Webster and Weathers, 1990; Zerba and Walsberg, 1992; Zerba et al., 1999).

Data were converted to heat production M (W) and evaporative cooling E (W), assuming 20.08 J ml^-1 O₂consumed and 2427 J g^-1 H₂O evaporated. We discarded all data from chicks that became hypothermic during the experiment(T_b<37°C). Also, data points identified as outliers(studentized residuals >3) in a preliminary multiple-regression model were further scrutinized, and the entire run was discarded if there was evidence of hyperactivity or abnormal experimental conditions during the measurements.

We used SYSTAT 7.0 (SPSS, Chicago, IL, USA) for all statistical analyses. Because of the considerable inter- and intraspecific variation in mass, we conducted allometric analyses using log₁₀-transformed data when appropriate to ensure homogeneity of variances. Homogeneity of variance was then verified by examining residual plots. To allow statistical analysis, we assumed that rapid growth justified treating each measurement as independent. This assumption is not strictly correct, and consequently our results apply only to the specific animals studied. In any event, the greatest source of uncertainty in applying our data to wild chicks is systematic rather than statistical, namely the effects of captive rearing. For example, captive shorebird chicks may have lower total energy requirements and higher average growth rates than wild ones (Beintema and Visser, 1989b; Schekkerman and Visser, 2001).

The variation in T_b values in an individual chick during a run generally exceeded the variation among birds. Therefore, we treated the mean T_b value at each wind speed as an independent measurement. For each run, this yielded four temperature measurements spaced at 30 min intervals. Nearly all such T_b measurements fell within the 38-40°C range. The main exceptions were for a few of the smallest sandpipers and dowitchers near the end of measurements at the highest wind speed, 3 ms^-1. Generally, if T_b fell below 37°C it would continue to decrease. We excluded these cases as incipiently hypothermic. For the remaining data, average T_b values were 39.24±0.45°C (mean ± S.D., N=15) for whimbrels,38.73±0.56°C (N=38) for dowitchers, and 38.56±0.80°C (N=9) for least sandpipers. These values are consistent with other data on chicks of the same species (J.B.W., unpublished data), and were assumed in calculations for experiments when T_b was not measured.

Heat is stored or released when chick T_b changes, and this needs to be included in the overall energy budget(Eppley, 1994). We estimated the maximum rate of heat storage or release as c_pm(ΔT_b/Δt),where m is body mass (g), c_p is tissue heat capacity (3.48 J g^-1) and t is time (s). However, for the T_b variation observed in our normothermic chicks, heat storage was negligible relative to the average energy expended during a 30 min interval.

Wind speed u affects both boundary layer convection processes,which are proportional to u^0.5, and dynamic processes(wind displacement of and penetration among insulating fibers), which are proportional to u²(Bakken, 1991b). The net result is that overall thermal conductance is related to u raised to an intermediate exponent, the value reflecting the weighting of convective and dynamic processes by the structure of the insulation. Therefore, we first fitted a preliminary model of the form a+bu^c+dm^f to our data, where a, b, c, d and f are fitted coefficients. The best-fit wind speed exponents were c=1.04-1.09. The exact value of c was not critical, and consequently u^1.0 was used in the final allometric models (Table 1).

We also plotted conductance as a function of body mass(Fig. 2A,B), using LOWESS(locally weighted scatterplot smoother) regressions(Cleveland, 1985) to data at each of the four wind speeds. The LOWESS lines suggest that the correlations between both log₁₀K_o and log₁₀K_ow and log₁₀m are nonlinear, and that wind has a relatively greater effect on the smallest chicks, a result expected from convective transport theory. LOWESS does not assume a particular model, and thus cannot provide statistical information on goodness-of-fit that might test the reality of, or hypotheses about, the curvature. (Indeed, the goodness of fit is set a priori by the LOWESS tension parameter). We tested polynomial regressions on body mass, as well as regressions where each species was assigned a dummy variable equal to 1 if the data applied to that species and 0 otherwise(Draper and Smith, 1981). These models explained significantly more variance than the linear model,indicating the relation is indeed nonlinear. However, we cannot discriminate between phylogenetic and mass-related physical effects such as convection because mass and taxon are confounded.

For downy chicks that respond to u^1.0, we used Equations 3a and 3b, plus a T_es scale developed for 1-day-old mallard ducklings (equations 4 and 5 in Bakken et al., 1999). The junco and mallard T_es scales involve recursive relations,and were solved numerically (Bakken et al.,1991).

The only clear conclusion about the effect of different wind speed exponents is that the correlation using u^½ is too sensitive to wind speeds from 0.5 to 1.5 m s^-1, and that the discrepancy is most significant for downy chicks. Otherwise, the shorebird chick scales based on wet and dry thermal conductance bracket the mallard scale and overlap the junco scale. These differences are within experimental error.

During the experiments, humidity was generally higher than has been recommended for making standardized evaporative water loss measurements(Lasiewski et al., 1966). The high chamber humidities were unexpected, as we were examining the response of animals below their lower critical temperature, where evaporative water loss is generally low. The dewpoint increased significantly (P<0.001)with the size of the bird. Consequently, the E of small birds may be exaggerated compared to the larger ones, and this may account for the shallower slope of the shorebird chick regression compared to the adult correlation (Fig. 1). High chamber humidity should have decreased evaporation, but nevertheless our measured values of evaporative water loss from shorebird chicks, adjusted to T_a=25°C using Equation 2(Fig. 1), are substantially higher than comparable values for adult birds under thermally neutral conditions (Williams,1996).

Calder and King (1974)suggested that, to a first approximation, the ratio E/M should be mass-independent. Briefly, they argued that metabolism is proportional to respiratory minute volume, as is respiratory water loss. Further, metabolism is proportional to mass, as is surface area, and cutaneous water loss is proportional to surface area. Indeed, their review of published data demonstrated that, at a given temperature, E/M had a consistent relationship to ambient temperature, across a wide range of body mass. It is particularly interesting that while E and M are positively correlated with each other and with mass at one temperature, they are negatively correlated across temperatures, such that E/M=0.05+0.0148exp(0.087T_a). Bartholomew(1972) suggested that the remarkable ability of birds to decrease E when M, and thus minute volume, was increasing was because colder inspired air chilled the respiratory tract, and thus water condensed on the respiratory passages when exhaled. Respiratory passages were increasingly chilled as ambient air temperature decreased, and consequently the temperature of exhaled air decreased, and therefore its water content. This mechanism is not affected by wind, and E should thus remain constant or even increase when convective cooling increases M. Consequently, one expects E/M should track T_a as in Calder and King's equation, and perhaps even increase when wind increased M at a given T_a.

We suggest that reduced skin temperature and relatively high cutaneous evaporative water loss may account for the wind effect. Because down is relatively poor thermal insulation, a significant part of the overall resistance to heat loss is provided by tissue. Thus, a wind-induced increase in heat loss reduces skin temperature with no change in air temperature. Reduced skin temperature may reduce E via two mechanisms unrelated to M, and thus decrease E/M: (1) a low skin temperature causes vasoconstriction and reduces water movement to the skin by increasing the diffusion distance for liquid water; (2) the vapor pressure of the water that does reach the stratum corneum decreases with air temperature.

This wind effect on E/M was not seen in a study of adult juncos(Bakken et al., 1991), possibly because they are better insulated so that skin temperature is less affected by wind (thermal conductance of shorebird chicks is 30-60% greater than that of juncos of similar mass).

Wind has a substantial effect on thermal conductance. For a 25 g bird (the approximate geometric mean for our study), increasing wind speed from 0.1 to 3 m s^-1 increased K_o by 50% and K_ow by 30%. Data from heated taxidermic mounts(Bakken, 1991b) indicated that the thermal conductance of contour feather insulation increases with wind speed as u^0.7—u^0.8. Fur is more affected by dynamic processes and responds as u^1.0, with some very soft furs on strongly curved surfaces responding as u^1.5. The best exponent for downy shorebird chicks was 1.04-1.09, which indicates a fur-like response to wind matching the fur-like appearance of down.

Because the chicks of Arctic-breeding shorebirds are often exposed to cold,one might expect their down to provide unusually effective insulation. In Fig. 2, we compare our data with other chicks and regressions to data on adult birds with contour feathers. The downy insulation of shorebird chicks was slightly better than that of day-old mallard ducklings (Anas platyrhynchos), but inferior to the extremely dense down of notably cold-hardy capercaillie (Tetrao urogallus; Pis, 2002) and Xantus' murrelet (Synthliboramphus hypoleucus; Eppley, 1984) chicks. The down of the latter two approaches the insulating value of summer adult plumages(Fig. 2B). Comparison with various regressions for adults indicates that shorebird chicks are rather poorly insulated compared to adults. The regression coefficients for published adult regression models as well as a regression that we constructed from published data for adult shorebirds(Kendeigh et al., 1977; Kersten and Piersma, 1987) are listed in Table 1.

Energetic considerations are probably paramount in determining the use of down rather than contour feathers in chicks. The mass of contour feathers needed to give the same amount of insulation is approximately 3 times the mass of down (McNabb and McNabb,1977), and the cost of synthesizing feathers appears to be considerably greater than for other proteins(Lindstrom et al., 1993). Apparently, the selective tradeoff between the thermal benefits of improved insulation and the cost of synthesizing contour feathers and keeping them properly preened favors the use of down by chicks.

Some data are available to compare the overall thermal conductance K_o of downy shorebirds to that of ducklings breeding in the same area. In another study at Churchill, we measured three long-tailed(oldsquaw) 4-day-old ducklings (Clangula hyemalis) averaging about 35 g, using the same equipment and protocol. These long-tailed ducklings had an average K_o=0.040 W °C^-1, increasing by 13%to 0.045 W °C^-1 at u=1 m s^-1. The conductance of shorebird chicks with the same body mass as the ducklings, 35 g, is similar, K_o=0.039 W °C^-1, increasing by 14% to 0.045 W °C^-1 at u=1 m s^-1(Table 1).

Mallards (Anas platyrhynchos) typically breed at lower latitudes and have higher thermal conductance than Arctic-breeding shorebirds(Fig. 2). However, 1- to 2-day-old mallard ducklings are physiologically more mature, having greater thermogenic capacity than 1- to 2-week-old shorebirds of similar size(Visser and Ricklefs, 1995). Consequently, they are able to tolerate T_a values at least as low as 10°C for over an hour(Koskimies and Lahti, 1964; Bakken et al., 1999).

In conclusion, the insulation of Arctic-breeding shorebird chicks thus appears somewhat, but not greatly, better than that of many other downy chicks. However, shorebird chicks are considerably less well insulated than capercaillie or Xantus' murrelet chicks of similar mass. Wind can increase heat loss and metabolic rate by 30-50% under windy conditions (30-50 km h^-1). Thus, the success of Arctic breeding shorebirds depends on efficient parental care, the ability to use stored heat to enable foraging with falling body temperature, and perhaps behavioral adjustments such as foraging in such sheltered areas as may be available.

List of symbols
 
• C_p

  tissue heat capacity

 
• E

  evaporative cooling

 
• K_o

  thermal conductance in actual environment

 
• K_os

  thermal conductance in convection-free enclosure

 
• K_ow

  wet thermal conductance

 
• m

  body mass

 
• M

  heat production

 
• t

  time

 
• T_a

  air temperature

 
• T_b

  body temperature

 
• T_e

  operative temperature

 
• T_es

  standard operative temperature

 
• u

  wind speed

We thank the Churchill Northern Studies Centre for housing and logistic support. Quynh Luong assisted with animal care, and Jacqueline de Fritas assisted with the measurements. Indiana State University provided a van for operational travel. This research was supported by NSF grant OPP 94-23522 to R.E.R. The authors wish to dedicate this article to the memory of Bill Calder in honor of his work in avian allometry. The personal encouragement and inspiration he gave G.S.B. is greatly appreciated.