Polar Biology

, 34:1751 | Cite as

Using body mass dynamics to examine long-term habitat shifts of arctic-molting geese: evidence for ecological change

  • Tyler L. Lewis
  • Paul L. Flint
  • Dirk V. Derksen
  • Joel A. Schmutz
  • Eric J. Taylor
  • Karen S. Bollinger
Original Paper

Abstract

From 1976 onward, molting brant geese (Branta bernicla) within the Teshekpuk Lake Special Area, Alaska, shifted from inland, freshwater lakes toward coastal wetlands. Two hypotheses explained this redistribution: (1) ecological change: redistribution of molting brant reflects improvements in coastal foraging habitats, which have undergone a succession toward salt-tolerant plants due to increased coastal erosion and saltwater intrusion as induced by climate change or (2) interspecific competition: greater white-fronted geese (Anser albifrons) populations increased 12-fold at inland lakes, limiting food availability and forcing brant into coastal habitats. Both hypotheses presume that brant redistributions were driven by food availability; thus, body mass dynamics may provide insight into the relevance of these hypotheses. We compared body mass dynamics of molting brant across decades (1978, 1987–1992, 2005–2007) and, during 2005–2007, across habitats (coastal vs. inland). Brant lost body mass during molt in all three decades. At inland habitats, rates of mass loss progressively decreased by decade despite the increased number of greater white-fronted geese. These results do not support an interspecific competition hypothesis, instead suggesting that ecological change enhanced foraging habitats for brant. During 2005–2007, rates of mass loss did not vary by habitat. Thus, while habitats have improved from earlier decades, our results cannot distinguish between ecological changes at inland versus coastal habitats. However, we speculate that coastal forage quality has improved beyond that of inland habitats and that the body mass benefits of these higher quality foods are offset by the disproportionate number of brant now molting coastally.

Keywords

Body mass dynamics Brant geese Habitat change Interspecific competition Molt Teshekpuk Lake 

Introduction

All avian species undergo a regular molt of their flight feathers (wing molt), whereby old, worn flight feathers are completely replaced. Molting is an energetically costly event, requiring substantial investment in the synthesis of both the feather structures and the biological tissues required for feather production (Dolnik and Gavrilov 1979; Heitmeyer 1985; Guillemette et al. 2007). In addition to the physiologic costs associated with feather production, birds may also incur costs associated with an absent or reduced flight capacity, including a limited ability to escape predators or access more favorable habitats and food supplies (Swaddle et al. 1999; Fox and Kahlert 2000; Lind and Jakobsson 2001). Given the considerable energetic costs and behavioral constraints, many birds significantly alter their allocation of time and energy during the wing molt (Fox et al. 1998; Lind and Jakobsson 2001; Fox and Kahlert 2005).

Northern Hemisphere goose species undergo a simultaneous wing molt, rendering them flightless for a 3- to 5-week period (Hohman et al. 1992). Many geese molt at high latitudes, where seasonal constraints favor rapid completion of molt (Jenni and Winkler 1994). Accordingly, such a rapid wing molt may subject molting geese to elevated daily energetic and nutritional demands relative to species that molt sequentially or at lower latitudes. Indeed, basal metabolic rate, a measure of energy expenditure while at rest, is higher for geese during wing molt than at other stages of their annual cycle (Portugal et al. 2007). Further, significant losses of body mass during wing molt have been documented in several goose species (Williams and Kendeigh 1982; Taylor 1993; Fox and Kahlert 2005; Portugal et al. 2007). Multiple theories have been proposed for mass loss during molt in geese, the primary being (1) geese are unable to meet the energetic requirements of wing molt solely from their diet (Fox and Kahlert 2005) and (2) mass loss is an adaptive strategy to decrease wing loading and thereby shorten the duration of flightlessness (Owen and Ogilvie 1979; Brown and Saunders 1998). Mass loss during wing molt, however, is not a consistent trait of all Northern Hemisphere geese. At some sites, molting geese maintain a constant body mass by relying on exogenous sources to meet their energetic and nutritional requirements (Ankney 1979, 1984; Fox et al. 1998). Nor is mass loss consistent within species; greylag geese (Anser anser) molting in Iceland lost little or no body mass while their conspecifics in Denmark lost 12–26% of body mass during wing molt (Fox et al. 1998; Fox and Kahlert 2005). Such results suggest that body mass dynamics of molting geese are highly influenced by local conditions, including food availability and quality, and may therefore serve as a sensitive indicator of habitat quality.

Each year, thousands of Pacific black brant geese (Branta bernicla nigricans; hereafter brant) migrate to the Teshekpuk Lake Special Area (TLSA), Alaska, to undergo wing molt (Derksen et al. 1982). Thirty years of aerial surveys in the TLSA (1976–2005) has documented a distributional shift of molting brant from large, inland lakes toward coastal, salt-influenced lakes, and sloughs (Flint et al. 2008). Consequently, the proportion of brant molting coastally has increased from ≤35 to ≥60% over this 30-year period, while the overall number of molting brant has not changed. Flint et al. (2008) concluded that the habitat-specific shift was driven by changes in foraging conditions, not predation pressure, and used this concept as the basis for two non-exclusive hypotheses to explain the distributional shift of brant. (1) Ecological change hypothesis: Warming of permafrost and shrinking of summer pack ice extent have increased rates of coastal erosion, which in turn has increased the frequency and extent of saltwater intrusion into coastal lakes and tundra (Mars and Houseknecht 2007; Jones et al. 2008). Consequently, coastal plant communities have undergone a succession toward salt-tolerant species that may be more favorable food items for molting brant (Person et al. 2003; Ward et al. 2005), leading to a redistribution of brant toward the coast. (2) Interspecific competition hypothesis: Concurrent with the distributional shift of molting brant, numbers of molting greater white-fronted geese (A. albifrons) have increased 12-fold at inland lakes, from approximately 1,300 individuals during the late 1970s to >16,000 individuals 30 years later. As a result, increased interspecific competition may have reduced food availability and/or excluded molting brant from their preferred inland habitats, forcing them to redistribute to coastal habitats.

Both hypotheses for long-term shifts in brant distributions should be discernible via patterns of body mass change because it is presumed that energy acquisition by molting brant is limiting. Furthermore, both hypotheses are predicated on spatial and temporal differences in habitat quality, of which molting body mass dynamics may serve as an effective indicator because of its dependence on habitat and food conditions. Therefore, we compared body mass dynamics of molting brant, measured as initial mass and rate of mass change, across habitats (inland, coastal) and time periods (1978, 1987–1992, 2005–2007) to lend insight into the relevance of distributional shift hypotheses. Because we have historic (1978, 1987–1992) body mass data for inland lakes only, our body mass comparisons will provide strong evidence for or against the interspecific competition hypothesis. Specifically, the following comparative outcomes would support the interspecific competition hypothesis: (1) greater mass loss at inland habitats now versus historically and (2) greater contemporary mass loss at inland versus coastal habitats. The lack of historic body mass data from coastal habitats prevents us from definitively testing the ecological change hypothesis. Nonetheless, a rejection of the interspecific competition hypothesis, coupled with lower mass loss at coastal versus inland sites, would strongly suggest that coastal habitat conditions have improved and thereby support the ecological change hypothesis.

Materials and methods

The TLSA, located on the Arctic Coastal Plain of Alaska (Fig. 1), is characterized by tundra vegetation, continuous permafrost, and large permafrost thaw lakes (Markon and Derksen 1994). Molting brant occur primarily in the northeast section of the TLSA (70°30′N, 153°30′W) and are comprised of sub-adults, non-breeders, and failed breeders from breeding colonies in Alaska, Russia, and Western Canada (Bollinger and Derksen 1996). Molting brant use low aspect shorelines of lakes and sloughs and feed on grasses and sedges associated with shoreline habitats (Weller et al. 1994). Flint et al. (2008) divided the TLSA molting area into three broad strata based on distinct habitat and landscape characteristics. Our research was conducted solely in strata two (hereafter inland stratum) and three (hereafter coastal stratum; Fig. 1). In general, the coastal stratum is characterized by the occurrence of saltwater intrusion and elevations less than 1.5 m, while the inland stratum lacks saltwater intrusion and has slightly greater elevations (4 m or less).
Fig. 1

Molting goose region of the Teshekpuk Lake Special Area, Alaska. Lakes at which brant were sampled are darkened, and strata boundaries are depicted with bold lines

Molting brant were captured in the TLSA in three distinct time periods: 25–28 July 1978, 10–27 July 1987–1992, and 16–23 July 2005–2007. We encircled flocks of flightless brant on lakes using float-equipped aircraft and slowly herded them into shore-based corrals (Bollinger and Derksen 1996). Lakes were generally shallow enough that personnel could herd flocks by wading, but inflatable boats were also used on occasion. During 1978, brant were captured only on lake 62, located in the inland stratum. From 1987–1992, brant were primarily captured on five lakes: 62, 99, 106, 107, and 145, all of which are located in the inland stratum. From 2005–2007, brant were captured on six lakes: 62, 104, 106, and 145 in the inland stratum and lakes 175 and 193 in the coastal stratum (Fig. 1). A parallel study of brant marked with GPS transmitters confirmed that brant rarely move between lakes while flightless (Lewis et al. 2011); thus, the influence of lake on body mass is restricted to the lake on which individuals were captured.

All captured birds were fitted with U.S. Fish and Wildlife Service stainless-steel bands and sexed by cloacal inversion. Birds were aged as second year (SY) or after second year (ASY) based on plumage characteristics (Bollinger and Derksen 1996). We measured culmen and tarsus to the nearest tenth of a millimeter using calipers and weighed each bird to the nearest gram using an electronic scale (1987–1992, 2005–2007) or spring scale (1978). Ninth primary length (the longest feather) was measured by inserting a ruler between the 8th and 9th primaries, then measuring from the skin surface to the distal edge of the feather. Additionally, during 1987–1992, some brant (n = 1,379) had their tenth primary measured in addition to their ninth primary, or solely their tenth primary measured. For brant with both measurements, tenth primary length was regressed against ninth primary length to derive the following relationship: 9th primary length = 10th primary length*1.12–2.94 (r2 = 0.99). Given this strong linear relationship, we were able to produce accurate ninth primary lengths for birds in which only the tenth primary was measured.

Statistical analyses

To examine factors affecting body mass dynamics of brant during the flightless wing molt, we conducted our analyses in three functionally separate stages: (1) analysis of data from years 2005–2007, (2) combined analysis of 1987–1992 and 2005–2007 datasets, and (3) combined analysis of 1978, 1987–1992, and 2005–2007 datasets. We analyzed the 2005–2007 dataset separately because it was the only time period with data from both coastal and inland strata lakes. For the two analyses involving multiple time periods, we used only banding records from lakes which were sampled within each period. Only one banding record per individual per year was included in all analyses. Aerial surveys indicated that the three time periods (1978, 1987–1992, and 2005–2007) used in our analyses each offered widely differing spatial distributions of brant and numbers of greater white-fronted geese (see Flint et al. 2008 for survey methodology). In 1978, 65% of brant molted in the inland stratum and 1,337 greater white-fronted geese were in the inland stratum (estimates based on 1976–1978 survey data). In 1987–1992, 54% of brant molted inland and greater white-fronted geese numbered 1,734 per year in the inland stratum. By 2005–2007, 41% of brant molted inland and greater white-fronted geese had increased to 16,530 per year in the inland stratum.

A complete understanding of body mass dynamics during molt requires knowledge of both the initial body mass at the onset of molt (intercept) and the subsequent rate of mass change during molt (slope). Accordingly, using body mass (g) as our response variable, we fitted two candidate sets of general linear models (GLMs) to our 2005–2007 dataset using PROC GLM in SAS: one model set to determine the appropriate intercept parameters and another model set to determine the slope parameters. We began by constructing our model set for the intercept, which consisted of 13 models and was based on various combinations of the following explanatory variables: size, sex, age (ASY or SY), year (2005–2007), lake (62, 104, 106, 145, 175, 193), and ninth primary length. The size variable, based on a mm scale, accounted for structural size differences among birds and was derived from the following formula: (log(tarsus) + log(culmen))/100. Ninth primary length served as our index of molt stage. With the exception of lake, all the aforementioned explanatory variables were included as main effects in every model and together constituted our simplest model. To this simple model, we next added lake as a main effect, producing the second of two models which lacked an interaction. The remaining 11 models differed solely by their interaction terms, which consisted of all possible combinations of sex, age, year, and lake. Because all the interaction terms are restricted to categorical variables, each model in our set will have the same slope value, differing only by their intercept.

We used an information-theoretic approach to guide model selection (Burnham and Anderson 2002). Akaike’s information Criterion (AIC) was calculated for each model in the candidate set, and ΔAIC and AIC weights were used to infer the relative support of each model. To examine the importance of stratum (inland or coastal), we substituted stratum for lake in the best-supported model (ΔAIC = 0). By grouping lakes into stratum, the parameter “penalty” in the AIC values is significantly reduced, providing a test of the relevance of spatial habitats to body mass dynamics of molting brant at the stratum versus lake scale.

We next fitted a candidate set of 16 GLMs to determine the optimal parameterization of the slope term. Each model in this candidate set contained identical intercept terms and differed solely by their slope term. The intercept term was the best-supported model (ΔAIC = 0) from the intercept candidate model set. The slope term was comprised of solely ninth primary length or an interaction term that included ninth primary length and all interaction combinations of sex, age, year, and lake. Ninth primary length was included in every interaction because it served as our time-index of molt stage, and it is a continuous variable, thus allowing each model’s slope to vary by the multiple age, sex, year, and lake interaction combinations. Finally, stratum was substituted for lake in the best-fitting model to test for the importance of a stratum effect.

We then explored the importance of period, defined as 1978, 1987–1992, or 2005–2007, for explaining variation in body mass of molting brant. First, we analyzed data from the 1987–1992 and 2005–2007 datasets, using only data from lakes that were sampled during both time periods (n = 3). We did not include 1978 data in this analysis because only one lake was measured during that sampling period, and its inclusion would limit our inference to one lake. We fitted a candidate set of five GLMs, using the results from 2005–2007 to parameterize the intercept (sex*lake) and slope (wing*lake*age*sex) interaction terms (wing = ninth primary length). In addition to the interaction terms, sex, age, and lake were included as main effects in all five models. Of the five models, the simplest model did not contain period as an explanatory variable and was thus composed solely of the two interaction terms and the main effects. The remaining four models included period as either a main effect or as a main effect and an extra term in the intercept interaction, the slope interaction, or both interaction terms. Size was not included because culmen and tarsus were not measured at lake 145 during 1987–1992, and its exclusion would limit the analysis to two lakes: 62 and 106. Exclusion of size introduces more variability into our analysis. Thus, should our analysis detect a period effect, the exclusion of size is of no consequence, because the period effect remained strong enough for detection despite the increased variability. Year was not included as an explanatory variable because it would be confounded with period. Furthermore, while we could have nested year within period, the inclusion of year would have reduced the analysis to two lakes.

Finally, we conducted an analysis of body mass variation by period using data from all three periods. During 1978, all brant captured were on lake 62 or nearby interconnected wetlands. Thus, we restricted this analysis to banding records from this one lake, which precluded our ability to use lake as an explanatory variable. Furthermore, because the earliest sampling period (1978) includes just a single year, we could not include year as an explanatory variable. We fitted a candidate set of five GLMs, again using the results from our prior analyses to build the foundation of the model set. The interaction terms for the intercept and slope portions of each model were parameterized from the best-fitting models from the 2005–2007 analyses. However, because lake and year could not be included as explanatory variables, the intercept interaction term was limited to sex*period, and the slope interaction term was limited to wing*age*sex or wing*age*sex*period. Of the five models, the simplest model did not contain period as an explanatory variable. The remaining four models included period as solely a main effect or as a main effect and a term in the intercept interaction, the slope interaction, or both interactions.

Results

Body mass dynamics 2005–2007

A total of 5,066 banding records were used in the 2005–2007 analysis. The best-supported model to describe variation in body mass of brant at the beginning of wing molt contained sex*year*lake as the intercept interaction term (Table 1). No other models received substantial support, including all that lacked an interaction term. Thus, at the commencement of wing molt, body mass of brant varied across lakes within the TLSA, and between-lake variation was not consistent across years and sexes (Fig. 2). Within a given year and sex, body mass variation across the six sampled lakes did not exceed 117 g at the onset of molt. The explanatory variable of stratum received no support; substituting stratum for lake in the top model resulted in a ΔAIC value of 231.37 (Table 1). Parameter estimates (±SE) from the top model for the main effect of sex indicated that males were 119.03 ± 10.06 g larger than females (Fig. 2).
Table 1

Candidate general linear models evaluating variation in body mass at the start of wing molt for brant in the Teshekpuk Lake Special Area, Alaska, 2005–2007

Model

Number of parameters

ΔAIC

AIC weight

Lake, wing, sex*year*lake

40

0

0.95

Lake, wing, sex*year*age*lake

75

7.29

0.02

Lake, wing, year*lake

23

7.97

0.02

Lake, wing, year*age*lake

40

8.84

0.01

Lake, wing, sex*age

14

231.48

0

Lake, wing, sex*year*age

20

233.69

0

Lake, wing, sex*age*lake

29

238.75

0

Lake, wing, sex* lake

18

243.41

0

Lake, wing, sex*year

15

246.90

0

Lake, wing

13

249.97

0

Lake, wing, age*lake

18

250.67

0

Lake, wing, year*age

15

250.85

0

Wing

8

467.23

0

Stratum, wing, sex*year*stratum

16

231.37

0

Size, sex, age, and year were included as main effects in every model and are not shown above. Models are listed in order of ΔAIC, with one exception: stratum was substituted for lake in the top model, and the results are presented separately at the bottom of the table. Abbreviations: wing = ninth primary length

Fig. 2

Body mass (g) at the start of wing molt by lake and year for female and male brant in the Teshekpuk Lake Special Area, Alaska, 2005–2007. Values are least square means ± SE

Variation in the rate of body mass change of brant throughout the molting period was best described by the model that contained the slope interaction term of wing*lake*age*sex (Table 2). No other models were well supported by the data. Substituting stratum for lake in the top model increased our ΔAIC by 24.24, indicating that patterns of mass change are better explained at the finer-scale level of individual lakes. With the exception of SY females at lake 62, parameter estimates for the slope term (wing*lake*age*sex + wing) were all negative, indicating that molting brant lost body mass at every lake, irrespective of sex or age (Fig. 3). Rates of body mass decline, grouped by sex and age and measured as grams of body mass per millimeter of ninth primary growth (g mm−1 hereafter), ranged from −0.29 to −1.46 g mm−1 across the six sampled lakes (Fig. 3). These rates of mass loss translate to overall mass losses of 43–228 g during the flightless molt when assuming that molting brant lose mass at a constant rate and regain flight at ninth primary lengths of 156 mm for males and 150 mm for females (Taylor 1995). For each age and sex cohort, estimates of body mass at the end of the flightless molt did not converge at a similar value across lakes (Fig. 4). This suggests that molting brant are not targeting an ideal body mass at which flight could be regained without over-depleting body reserves (see discussion).
Table 2

Candidate general linear models evaluating variation in rates of body mass change during wing molt for brant in the Teshekpuk Lake Special Area, Alaska, 2005–2007

Model

Number of parameters

ΔAIC

AIC weight

Intercept, wing*lake*age*sex

63

0

0.91

Intercept, wing*lake*age

51

4.71

0.09

Intercept, wing*lake

45

14.87

0

Intercept, wing*lake*sex

51

19.13

0

Intercept, wing*year*lake

57

23.57

0

Intercept, wing*year*age*lake

75

23.64

0

Intercept, wing*age*sex

43

35.03

0

Intercept, wing*year*age*sex

51

35.86

0

Intercept, wing*year*age*sex*lake

111

37.46

0

Intercept, wing*year*sex*lake

75

38.46

0

Intercept, wing*year*sex

45

55.09

0

Intercept, wing*year*age

45

57.17

0

Intercept, wing*sex

41

57.49

0

Intercept, wing*age

41

57.70

0

Intercept, wing*year

42

58.43

0

Intercept

40

60.51

0

Intercept wing*stratum*age*sex

47

24.24

0

The best-fitting intercept model (size sex age year lake wing sex*year*lake) was included in every model and is referred to as “intercept” in the table. Models are listed in order of ΔAIC, with one exception: stratum was substituted for lake in the top model, and the results are presented separately at the bottom of the table. Abbreviations: wing = ninth primary length

Fig. 3

Rates of body mass change (g per mm ninth primary growth) by lake, sex, and age for brant during their flightless wing molt in the Teshekpuk Lake Special Area, Alaska, 2005–2007. Values are parameter estimates ± SE from the best-fitting model describing variation in rates of mass change

Fig. 4

Predicted body mass values (g) from 2005–2007 for age and sex classes of brant at the start (9th primary = 0 mm) and near the end of the flightless molt (9th primary = 150 mm). Each predicted body mass value, represented by circle points, is the least square mean for a lake and year group (e.g., lake 62, year 2005) as predicted from our best-fitting model explaining variation in body mass of molting brant. Least square means accounted for structural size differences among individuals. The error bars (±SE) represent the total range of variation for all lake-by-year groups within each age and sex class

Changes in body mass dynamics through time

Our first model set explored changes in body mass dynamics at three lakes (62, 106, and 145) over two time periods (1987–1992 and 2005–2007), and a total of 4,116 banding records were used in this analysis. The only model to receive substantial support contained the explanatory variable of period, defined as either 1987–1992 or 2005–2007, in both the intercept (sex*lake*period) and slope interaction terms (wing*lake*age*sex*period; Table 3). Body mass of both sexes was greater at the start of molt at lakes 62 and 106 during 2005–2007, while time period differences were not evident at lake 145 (Fig. 5). To compare rates of body mass loss by time period, we subtracted the 2005–2007 top-model slope values from the 1987–1992 top-model slope values, producing estimates of rate change (g mm−1) through time. Thus, a positive rate change indicates that rates of body mass loss decreased from 1987–1992 to 2005–2007, while a negative rate change indicates that rates of body mass loss increased through time. At lake 62, rate changes were positive for all sex and age cohorts, indicating that brant lost body mass at a slower rate during the 2005–2007 wing molt (Fig. 6). At lakes 106 and 145, with the exception of SY males, rate changes were again positive for all sex and age cohorts, suggesting that molting brant lost less body mass during 2005–2007 (Fig. 6).
Table 3

Candidate general linear models evaluating variation in body mass of molting brant in the Teshekpuk Lake Special Area, Alaska, 1987–1992 and 2005–2007

Model

Number of parameters

ΔAIC

AIC weight

Period, sex*lake*period, wing*lake*age*sex*period

38

0

0.92

Period, sex*lake, wing*lake*age*sex*period

33

5.89

0.05

Period, sex*lake*period, wing*lake*age*sex

26

6.83

0.03

Lake, period, sex*lake, wing*lake*age*sex

21

91.28

0

Sex*lake, wing*lake*age*sex

20

439.47

0

Sex, age, lake, and wing were included as main effects in every model and are not shown above. Models are listed in order of ΔAIC. Period is a categorical variable with two levels (1987–1992 and 2005–2007). Abbreviations: wing = ninth primary length

Fig. 5

Body mass (g) at the start of wing molt by lake and period (1987–1992, 2005–2007) for female and male brant in the Teshekpuk Lake Special Area, Alaska. Values are least square means ± SE

Fig. 6

Difference in rate of body mass change (g per mm ninth primary growth) from 1987–1992 to 2005–1907 for brant during their flightless wing molt in the Teshekpuk Lake Special Area, Alaska. Values are parameter estimates ± SE from the best-fitting models describing variation in rates of mass change. Because brant lose mass during molt, positive values indicate that rates of mass loss have decreased from 1987–1992 to 2005–2007, and negative values indicate that rates of mass loss have increased from 1987–1992 to 2005–2007

Our next model set contained data from lake 62 over three time periods: 1978, 1987–1992, and 2005–2007. A total of 1,408 banding records were used for this analysis. The only model to receive substantial support contained period within the slope interaction term (wing*age*sex*period); inclusion of period in the intercept interaction term received only moderate support (Table 4). We subtracted the 2005–2007 slope values from the 1978 and 1987–1992 slope values to estimate body mass rate changes (g mm−1) through time. For all age and sex cohorts, brant experienced their greatest rates of body mass loss in 1978, intermediate rates of body mass loss in 1987–1992, and their lowest rates of body mass loss in 2005–2007 (Fig. 7).
Table 4

Candidate general linear models evaluating variation in body mass of molting brant in the Teshekpuk Lake Special Area, Alaska, 1979, 1987–1992, and 2005–2007

Model

Number of parameters

ΔAIC

AIC weight

Period, wing*age*sex*period

18

0

0.86

Period, sex*period wing*age*sex*period

20

3.67

0.14

Period, sex*period wing*age*sex

12

14.80

0

Period, wing*age*sex

10

15.85

0

Wing*age*sex

8

408.62

0

Sex, age, and wing were included as main effects in every model and are not shown above. Models are listed in order of ΔAIC. Period is a categorical variable with three levels (1979, 1987–1992, 2005–2007). Abbreviations: wing = ninth primary length

Fig. 7

Difference in rate of body mass change (g per mm ninth primary growth) from 1978 to 2005–2007 and from 1987–1992 to 2005–2007 for brant during their flightless wing molt at lake 62, Teshekpuk Lake Special Area, Alaska. Values are parameter estimates ± SE from the best-fitting models describing variation in rates of mass change. All values are positive, indicating that rates of mass change have decreased over time

Discussion

The interspecific competition hypothesis

From 1976 onward, molting brant steadily changed their habitat use within the TLSA, redistributing from inland, freshwater lakes (inland stratum) toward coastal, brackish lakes (coastal stratum; Flint et al. 2008). Our data demonstrate that, over this same time period, rates of body mass loss for molting brant in the TLSA have decreased, irrespective of stratum or lake. Flint et al. (2008) proposed two hypotheses to explain the long-term shifts in molting distributions of brant from inland to coastal habitats. One of these, the interspecific competition hypothesis, states that the 12-fold population increase of greater white-fronted geese in the inland stratum has reduced food availability, displacing brant into the coastal stratum. This hypothesis presumes that (1) prior to the increase in greater white-fronted geese, inland stratum habitats were preferred by molting brant, as historically evidenced by the greater number of brant inland and (2) coastal stratum habitats, due to their lower frequency of historic use by molting brant, were less optimal. Accordingly, under this hypothesis, brant are predicted to have greater mass loss at inland habitats now versus historically. We observed the opposite, however, as the rate of mass loss at three inland stratum lakes generally decreased from 1987–1992 to 2005–2007, concurrent with the distributional shift of brant away from the inland stratum. At inland stratum lake 62, at which molting brant were sampled across three time periods, rates of mass loss progressively decreased from 1978 to 1987–1992 to 2005–2007, yet the long-term distributional shift of molting brant has resulted in fewer brant at this lake (Mallek 2007). Our results suggest that foraging conditions have improved at inland stratum lakes for molting brant, despite the increased abundance of greater white-fronted geese.

Interspecific competition can act via depletion (sensu Schoener 1983, Alatalo et al. 1987), whereby numerically dominant greater white-fronted geese deplete food resources shared by brant, or via interference (sensu Park 1962, Creswell 1997), in which aggression or territoriality of greater white-fronted geese excludes brant from preferred foraging areas. Importantly, although the mechanism differs, the main effect of depletion versus interference competition is largely the same: less food for molting brant. Thus, irrespective of the type, interspecific competition appears insufficient to explain the decreased rate of mass loss over time. Alternatively, brant body mass may be negatively impacted by interspecific competition, but the effects are offset by reduced intraspecific competition in inland habitats. However, under an interspecific competition hypothesis, it is not the number of brant which is important, but the total number of geese (brant + greater white-fronted geese) since both species would hypothetically reduce food availability. Indeed, the total number of geese at inland habitats has significantly increased over time (Flint et al. 2008), yet rates of mass loss declined, further marginalizing the interspecific competition hypothesis.

The lack of an effect of greater white-fronted geese on brant body mass dynamics at inland lakes suggests that dietary overlap between the two species is minimal. Low relief moss/peat zones along edges of lakes are the preferred feeding habitats of molting brant at inland TLSA lakes (Derksen et al. 1982; Weller et al. 1994). While greater white-fronted geese feed in these same areas, they also show greater flexibility in their habitat use, commonly feeding in upland tundra vegetation and molting on lakes devoid of moss/peat habitats (Derksen et al. 1979; Markon and Derksen 1994). Accordingly, dietary overlap between the two species may not be great enough to negatively affect food intake of molting brant, although we lack actual dietary data from inland lakes to quantify the degree of overlap. Future studies should assess dietary intake of both species at inland lakes, as well as interspecific antagonistic behaviors. Finally, while decreased rates of mass loss suggest that interspecific competition is minimal, these results also suggest that food abundance and/or quality has improved at inland habitats. Improvements may be driven by a warming Arctic climate, which directly affects the net primary production and growth rate of Arctic plants (explained in detail below).

The ecological change hypothesis

The second hypothesis put forward by Flint et al. (2008) was that warming climates and coastal erosion have increased the frequency and extent of saltwater intrusion into the coastal stratum (Mars and Houseknecht 2007; Jones et al. 2008), causing a community succession toward salt-tolerant plant species. Such plants are used extensively by brant in other coastal tundra habitats in Alaska (Sedinger et al. 2001; Person et al. 2003) and may be a more optimal food source than inland stratum plants, inducing brant to shift toward coastal stratum habitats. This hypothesis presumes that brant preferred inland stratum habitats prior to the changes in coastal plant communities, but that coastal stratum habitats, which currently support larger brant populations than inland stratum habitats, are now at least of equal quality, if not preferred by molting brant. While the 1978 and 1987–1992 data are strictly from the inland stratum, not allowing us to test for historic stratum effects, a comparison of rates of body mass loss across stratums during 2005–2007 indicated no difference between coastal and inland strata. Rather, rates of mass loss varied by lake, a much smaller spatial scale than stratum, given that each stratum is composed of many lakes. These results suggest that, from the energetic perspective of molting brant, foraging conditions across the two stratums are not significantly different. Assuming a distribution based primarily on food resources, molting brant would distribute themselves into habitat patches throughout the TLSA in approximate proportion to the amount of resources available (Kacelnik et al. 1992; Bautista et al. 1995). Based on such reasoning, we speculate that coastal food resources have improved, perhaps even exceeding the quality of inland food resources. However, these benefits are offset by the increased intraspecific competition coastally, leading to the lack of a detectable stratum-level effect on body mass dynamics.

Over the same time period that rates of mass loss decreased, the population size of molting brant in the TLSA has not changed, as indicated by population trajectories that account for inter-annual variation (Flint et al. 2008). Hence, at a TLSA-wide scale, decreased rates of mass loss cannot be attributed to an overall reduction in intraspecific competition. Rather, TLSA-wide body mass patterns strongly suggest that ecological change has occurred in the TLSA, enhancing foraging habitats for molting brant and thereby decreasing their rates of mass loss. Furthermore, data from radio-marked brant indicate that most failed breeding birds from the Arctic Coastal Plain of Alaska use coastal wetlands prior to and during the molt migration to the TLSA (Lewis et al. 2010). Assuming that ecological change has improved foraging conditions in coastal habitats, the use of such habitats immediately preceding molt may explain our result that brant now arrive at the TLSA molting grounds at a heavier body mass than during previous decades.

While distributional and body mass evidence strongly suggests that ecological change has occurred at coastal stratum habitats in the TLSA, we cannot rule out such change at inland habitats as well. Northern latitudes have experienced significant increases in surface temperatures, especially over the last 30 years (Walther et al. 2002). As a result, arctic plants have exhibited earlier growth of new shoots, increased growing season, increased biomass, and significant range shifts (Keeling et al. 1996; Parmesan and Yohe 2003; Root et al. 2003; Hudson and Henry 2009), all of which may alter foraging conditions for brant across both inland and coastal stratums. Ultimately, while our body mass results provide evidence for ecological change in the TLSA, validation of this hypothesis will require a detailed study of forage type, quality, and abundance in both coastal and inland stratums.

Function and shape of mass loss

Mass loss during wing molt has been broadly explained by two fundamentally different hypotheses: (1) flightless wing molt is energetically demanding, and geese are unable to meet the high energetic requirements solely from their diet; hence, mass loss acts as a constraint (e.g., Hohman et al. 1992) and (2) mass loss during molt enables geese to reduce wing loading on incompletely grown flight feathers and thereby regain flight sooner; hence, mass loss acts as an adaptive strategy (Brown and Saunders 1998). As our objectives herein are to explain the long-term shift in habitat use of molting brant via body mass dynamics, the adaptive hypothesis would appear to challenge our assumption that body mass dynamics can explain patterns of habitat quality. Were mass loss strictly an adaptive strategy to regain flight earlier, then rates of mass loss would depend largely upon body mass at the start of molt and much less on habitat conditions experienced during molt.

Previous research on body mass dynamics of molting geese has offered support for both hypotheses, despite their apparently conflicting theoretical basis. Captive barnacle geese (B. leucopsis) fed ad libitum spent less time feeding and lost 25% of their body mass during wing molt (Portugal et al. 2007). Because these birds had unrestricted access to food, these results appear to support the adaptive mass loss hypothesis. Specifically, mass loss is adaptive because it hastened the return of flight, the most important escape behavior, while concomitantly freeing more time for vigilance (Panek and Majewski 1990). Alternatively, body mass loss during the flightless wing molt is not a consistent trait in Northern Hemisphere goose populations, as some molting goose populations do not lose mass (Fox and Kahlert 2005). For example, greylag geese molting in Iceland lost little or no body mass while their conspecifics in Denmark lost 12–26% of body mass during the wing molt (Fox et al. 1998, Fox and Kahlert 2005). Similarly, rate of mass change varied widely for molting Barrow’s goldeneye, including both positive and negative rates of change, even though all individuals originated from the same molting location (van de Wetering and Cooke 2000). These results suggest that extent of mass loss depends on local habitat conditions and would appear to support the energetic constraint hypothesis.

Our body mass data for molting brant in the TLSA appear to conflict with the inherent presumptions of the adaptive mass loss hypothesis. Were the re-attainment of flight the sole purpose of mass loss, then individuals with body masses that varied at the start of molt, after controlling for structural body size differences, would be predicted to have similar body masses near the end of the flightless period (Portugal et al. 2007). That is, at the end of molt brant should converge on the ideal mass at which flight could be regained without an over-depletion of necessary body reserves. Thus, after subdividing our data into age and sex groups, we would expect the average body mass variance among lakes and years to become smaller as the return of flight drew near. However, predicted body mass values at the end of the flightless period were equally or more variable than predicted values at the onset of molt (Fig. 4), indicating that mass loss is flexible, and molting brant are not losing mass strictly to regain flight sooner. Rather, our results suggest that temporal and spatial variation in habitat conditions directly affects the energetic balances of molting brant (Fox and Kahlert 2005), resulting in body masses that vary among lakes and years. Furthermore, we documented a change in rates of mass loss over time, with brant gradually losing body mass at a slower rate from 1978 to 2005–2007. If rates of mass loss were strictly adaptive and under selective pressure, we would not have expected rates of mass loss to detectably change over such a relatively short period of evolutionary time (i.e., a few generations).

Finally, an assumption of our linear models is that ninth primary growth rate and mass loss are constant throughout the molt. Based on the recapture (n = 226) of wild brant in the TLSA, Taylor (1995) determined that ninth primary growth rate does not significantly differ throughout the molt period. Thus, our assumption for ninth primary growth seems warranted. Similar recapture data are not available for body mass measurements. If rates of mass loss are non-linear and individuals were systematically weighed at different stages of molt across decades, then our estimates of decadal change may be biased. For our 1978, 1987–1992, and 2005–2007 datasets, the mean ± SE ninth primary lengths (i.e., stage of molt) were 56.21 ± 0.32, 76.24 ± 0.51, and 84.99 ± 0.42 mm, respectively. While decadal differences in stage of molt are apparent, the average length difference between the 1978 and 2005–2007 datasets represents a small portion of the total range of ninth primary length measurements (0–206 mm). Accordingly, the shape of non-linearity would have to be very sharp during a small portion (28.8 mm) of the total growth range to strongly bias our estimates of mass loss rate. Finally, to compare rate of mass loss across decades, we subtracted the current slope coefficient (g mm−1) from the historic coefficient, giving us an estimate of change relevant throughout the duration of molt. Non-linear slope coefficients would require unique estimates of decadal mass change for each mm of feather growth, considerably complicating our ability to detect and interpret decadal change.

Sex, age, and lake effects

Body mass at the start of molt varied by the interaction term of sex*year*lake. Inter-annual variation in body mass at the start of molt may reflect carry-over effects from different portions of the annual cycle, such as breeding or wintering (Harrison et al. 2011). Inger et al. (2010) found that successful breeding brant finish the subsequent winter in lower body condition than non-breeders because of the prolonged costs of parental care. Accordingly, body mass at the start of molt may reflect the prior year’s breeding success, potentially explaining a portion of our inter-annual variation. Similarly, sex differences may be explained by the differential costs of reproduction, although the current year’s breeding success may be more pertinent than carry-over effects from prior years. In general, molting brant populations are comprised of either failed or non-breeding individuals (Derksen et al. 1982). Years with a larger proportion of failed breeders may result in lighter-weight females, relative to males, since they incurred the costs of egg production less than a month before arrival in the TLSA. Finally, lake-by-lake variation in body mass at the start of molt does not appear to reflect recent habitat conditions and may thus be influenced by recent reproductive efforts as well. Brant in the TLSA typically spend 1–3 days on their molt wetland prior to going flightless (Lewis et al. 2010), which is too short a time for lake-specific habitat conditions to significantly influence body mass at molt initiation. As well, because molting brant in the TLSA do not segregate by origin of nesting colony (Bollinger and Derksen 1996), lake-by-lake variation in body mass at the start of molt does not reflect conditions at nesting colonies or distance of the molt migration. Rather, non-breeding brant may arrive in the TLSA sooner and at a heavier body mass because of their decision to forego the energetic and time demands of egg production. Because brant are gregarious and typically molt in large flocks (Derksen et al. 1982; Reed et al. 1998), these early-arriving brant likely group together during molt migration and upon arrival in the TLSA, leading to a disproportionate grouping of heavy birds on select lakes. Once settled on a lake for the 3- 4-week flightless period, subsequent variation in rates of mass change of molting brant likely reflects lake-specific differences in habitat conditions, such as food availability and quality.

Age was the only variable excluded from the intercept interaction term, indicating that age-specific differences at the onset of molt are not influenced by differential conditions experienced across years or habitats. A comparison of time periods supports this idea, as the body mass difference between SY birds and adults during 1987–1992 (SY birds 83.37 ± 9.99 g smaller) was almost identical to the difference during 2005–2007 (SY birds 89.96 ± 9.84 g smaller). Conversely, rates of mass change varied by age but not sex, as SY brant lost mass at a slower rate than adults. This was unexpected because young birds are typically inefficient foragers relative to adults, potentially increasing their rate of mass loss (Stahl et al. 2001, Heise and Moore 2003). In the TLSA, brant typically molt in flocks ranging from several hundred to several thousand birds (Derksen et al. 1982). These large flocks may allow SY birds to rely on experienced adults for vigilance and predator warnings, freeing more time for foraging and thereby slowing mass loss. Alternatively, young brant may have a higher tendency to explore, encountering profitable lakeside habitats that are under-used by adults. Black et al. (1991) found that immature barnacle geese were pushed into peripheral habitats by adult birds loyal to old sites, yet peripheral habitats were actually higher quality. Finally, although rates of body mass loss generally decreased across decades, this pattern was not observed for SY males at two lakes. While these results are perplexing, we hypothesize that changes in age and sex distributions at individual lakes may have altered patterns of intraspecific competition and aggression, potentially producing lake-specific body mass dynamics for SY males that differ from other age and sex cohorts.

Conservation implications

Along with brant, the TLSA also supports large numbers of molting greater white-fronted, Canada, and snow geese. The number of molting geese counted in the TLSA in recent years has ranged from 70,000 to 90,000, collectively making it one of the most significant goose molting areas in the circumpolar arctic (Flint et al. 2008). Although currently undeveloped, the TLSA contains known oil and gas deposits and has often been proposed as an area for future development. Planning to minimize the effects of oil and gas development on molting brant populations requires a clear understanding of patterns of habitat use and how such patterns may change through time. Our body mass results suggest that, over the last 30 years, habitat conditions in the TLSA have improved for brant, as molting birds now lose less mass and utilize a broader range of habitats. Accordingly, we believe habitat change is the primary driving force behind the long-term distributional shift of molting brant. In the future, an ongoing knowledge of body mass dynamics may improve our ability to predict distributions of molting brant in the face of ecological changes in arctic ecosystems, as well as alleviate the potential for overlap between molting brant distributions and resource development.

Notes

Acknowledgments

The Bureau of Land Management and US Geological Survey, Alaska Science Center, provided funding and logistic support. U.S. Fish and Wildlife Service, Region 7, Division of Migratory Bird Management, provided aerial support and assisted with brant captures. J.S. Sedinger, S.J. Portugal, and an anonymous reviewer thoroughly examined and improved the manuscript. Use of trade, product, or company names is solely for descriptive purposes and does not imply endorsement or criticism by the U.S. government. All procedures were approved by Alaska Science Center’s Animal Care and Use Committee, U.S. Geological Survey, under protocol 06-SUP-02, and were authorized by U.S. Fish and Wildlife Service and Bureau of Land Management under permit number BLM AK FF094979.

References

  1. Alatalo RV, Eriksson D, Gustafsson L, Larsson K (1987) Exploitation competition influences the use of foraging sites by tits: experimental evidence. Ecology 68:284–290CrossRefGoogle Scholar
  2. Ankney CD (1979) Does the wing molt cause nutritional stress in lesser snow geese? Auk 96:68–72Google Scholar
  3. Ankney CD (1984) Nutrient reserve dynamics of breeding and molting brant. Auk 101:361–370Google Scholar
  4. Bautista LM, Alonso JC, Alonso JA (1995) A field test of ideal free distribution in flock-feeding common cranes. J Anim Ecol 64:747–757CrossRefGoogle Scholar
  5. Black JM, Deerenberg C, Owen M (1991) Foraging behaviour and site selection of barnacle geese Branta leucopsis in a traditional and newly colonised spring staging habitat. Ardea 79:349–358Google Scholar
  6. Bollinger KS, Derksen DV (1996) Demographic characteristics of molting black brant near Teshekpuk Lake, Alaska. J Field Ornithol 67:141–158Google Scholar
  7. Brown RE, Saunders DK (1998) Regulated changes in body mass and muscle mass in molting blue-winged teal for an early return to flight. Can J Zool 76:26–32CrossRefGoogle Scholar
  8. Burnham KP, Anderson DR (2002) Model selection and multi-model inference: a practical information-theoretic approach. Springer, New YorkGoogle Scholar
  9. Creswell W (1997) Interference competition at low competitor densities in blackbirds Turdus merula. J Anim Ecol 66:461–471CrossRefGoogle Scholar
  10. Derksen DV, Weller MW, Eldridge WD (1979) Distributional ecology of geese molting near Teshekpuk Lake, National Petroleum Reserve-Alaska. In: Jarvis RL, Bartonek JC (eds) Management and biology of pacific flyway geese. Oregon State University Bookstores, Corvallis, pp 189–207Google Scholar
  11. Derksen DV, Eldridge WD, Weller MW (1982) Habitat ecology of Pacific black brant and other geese moulting near Teshekpuk Lake, Alaska. Wildfowl 33:39–57Google Scholar
  12. Dolnik RD, Gavrilov VM (1979) Bioenergetics of molt in the chaffinch (Fringilla coelebs). Auk 96:253–264Google Scholar
  13. Flint PL, Mallek EJ, King RJ, Schmutz JA, Bollinger KS, Derksen DV (2008) Changes in abundance and spatial distribution of geese molting near Teshekpuk Lake, Alaska: interspecific competition or ecological change? Pol Biol 31:549–556CrossRefGoogle Scholar
  14. Fox AD, Kahlert J (2000) Do moulting greylag geese Anser anser forage in proximity to water in response to food availability and/or quality? Bird Study 47:266–274CrossRefGoogle Scholar
  15. Fox AD, Kahlert J (2005) Changes in body mass and organ size during wing moult in non-breeding greylag geese Anser anser. J Avian Biol 36:538–548CrossRefGoogle Scholar
  16. Fox AD, Kahlert J, Walsh AJ, Stroud DA, Mitchell C, Kristiansen JN, Hansen EB (1998) Patterns of body mass change during moult in three different goose populations. Wildfowl 49:45–56Google Scholar
  17. Guillemette M, Pelletier D, Grandbois J, Butler PJ (2007) Flightlessness and the energetic cost of wing molt in a large sea duck. Ecology 88:2936–2945PubMedCrossRefGoogle Scholar
  18. Harrison XA, Blount JD, Inger R, Norris DR, Bearhop S (2011) Carry-over effects as drivers of fitness differences in animals. J Anim Ecol 80:4–18PubMedCrossRefGoogle Scholar
  19. Heise CD, Moore FR (2003) Age-related differences in foraging efficiency, molt, and fat deposition of gray catbirds prior to autumn migration. Condor 105:496–504CrossRefGoogle Scholar
  20. Heitmeyer ME (1985) Protein costs of the prebasic molt of female mallards. Condor 90:263–266CrossRefGoogle Scholar
  21. Hohman WL, Ankney CD, Gordon DH (1992) Ecology and management of postbreeding waterfowl. In: Batt DJ, Afton AD, Anderson MG, Ankney CD, Johnson DH, Kadlec JA, Krapu GL (eds) Ecology and management of breeding waterfowl. University of Minnesota Press, Minneapolis, pp 128–189Google Scholar
  22. Hudson JMG, Henry GHR (2009) Increased plant biomass in a high arctic heath community from 1981 to 2008. Ecology 90:2657–2663PubMedCrossRefGoogle Scholar
  23. Inger R, Harrison XA, Ruston GD, Newton J, Colhoun K, Gudmundsson GA, McElwaine G, Pickford M, Hodgson D, Bearhop S (2010) Carry-over effects reveal reproductive costs in a long-distance migrant. J Anim Ecol 79:974–982PubMedCrossRefGoogle Scholar
  24. Jenni L, Winkler R (1994) Moult and ageing of European passerines. Academic Press, LondonGoogle Scholar
  25. Jones BM, Hinkel KM, Arp CD, Eisner WR (2008) Modern erosion rates and loss of coastal features and sites, Beaufort Sea coastline, Alaska. Arctic 61:361–372Google Scholar
  26. Kacelnik A, Krebs JR, Bernstein C (1992) The ideal free distribution and predator-prey populations. Trends Ecol Evol 7:50–55PubMedCrossRefGoogle Scholar
  27. Keeling CD, Chin JFS, Whorf TP (1996) Increased activity of northern vegetation inferred from atmospheric CO2 measurements. Nature 382:146–149CrossRefGoogle Scholar
  28. Lewis TL, Flint PL, Schmutz JA, Derksen DV (2010) Pre-moult patterns of habitat use and moult site selection by brent geese Branta bernicla nigricans: individuals prospect for moult sites. Ibis 152:556–568CrossRefGoogle Scholar
  29. Lewis TL, Flint PL, Schmutz JA, Derksen DV (2011) Fine scale movements and habitat use of black brant during the flightless wing molt in Arctic Alaska. Waterbirds: in pressGoogle Scholar
  30. Lind J, Jakobsson S (2001) Body building and concurrent mass loss: flight adaptations in tree sparrows. Proc R Soc B 268:1915–1919PubMedCrossRefGoogle Scholar
  31. Mallek EJ (2007) Teshekpuk Lake area molting goose survey 2007. US Fish Wildlife Service, FairbanksGoogle Scholar
  32. Markon CJ, Derksen DV (1994) Identification of tundra land cover near Teshekpuk Lake, Alaska using SPOT satellite data. Arctic 47:222–231Google Scholar
  33. Mars JC, Houseknecht DW (2007) Quantitative remote sensing study indicates doubling of coastal erosion rate in past 50 years along a segment of the Arctic coast of Alaska. Geology 35:583–586CrossRefGoogle Scholar
  34. Owen M, Ogilvie MA (1979) Wing molt and weights of barnacle geese in Spitsbergen. Condor 81:42–52CrossRefGoogle Scholar
  35. Panek M, Majewski P (1990) Remex growth and body mass of mallards during wing molt. Auk 107:255–259Google Scholar
  36. Park T (1962) Beetles, competition, and populations. Science 138:1369–1375PubMedCrossRefGoogle Scholar
  37. Parmesan C, Yohe G (2003) A globally coherent fingerprint of climate change impacts across natural systems. Nature 421:37–42PubMedCrossRefGoogle Scholar
  38. Person BT, Herzog MP, Ruess RW, Sedinger JS, Anthony RM, Babcock CA (2003) Feedback dynamics of grazing lawns: coupling vegetation change with animal growth. Oecologia 135:583–592PubMedGoogle Scholar
  39. Portugal SJ, Green JA, Butler PJ (2007) Annual changes in body mass and resting metabolism in captive barnacle geese (Branta leucopsis): the importance of wing moult. J Exp Biol 210:1391–1397PubMedCrossRefGoogle Scholar
  40. Reed A, Ward DH, Derksen DV, Sedinger JS (1998) Brant (Branta bernicla). In: Poole A, Gill F (eds) The birds of North America, no 337. Academy of Natural Sciences, PhiladelphiaGoogle Scholar
  41. Root TL, Price JT, Hall KR, Schneiders SH, Rosenzweig CR, Pounds JA (2003) Fingerprints of global warming on wild animals and plants. Nature 42:57–60CrossRefGoogle Scholar
  42. Schoener TW (1983) Field experiments on interspecific competition. Am Nat 122:240–285CrossRefGoogle Scholar
  43. Sedinger JS, Herzog MP, Person BT, Kirk MT, Obritchkewitch T, Martin PP, Stickney AA (2001) Large-scale variation in growth of black brant goslings related to food availability. Auk 118:1088–1095CrossRefGoogle Scholar
  44. Stahl J, Tolsma PH, Loonen MJJE, Drent RH (2001) Subordinates explore but dominants profit: resource competition in high Arctic barnacle goose flocks. Anim Behav 61:257–264PubMedCrossRefGoogle Scholar
  45. Swaddle JP, Williams EV, Rayner JMV (1999) The effect of simulated flight feather moult on escape take-off performance in starlings. J Avian Biol 30:351–358CrossRefGoogle Scholar
  46. Taylor EJ (1993) Molt and bioenergetics of Pacific black brant (Branta bernicla nigricans) on the Arctic Coastal Plain, Alaska. Dissertation, Texas A&M UniversityGoogle Scholar
  47. Taylor EJ (1995) Molt of black brant (Branta bernicla nigricans) on the Arctic Coastal Plain, Alaska. Auk 112:904–919Google Scholar
  48. van de Wetering D, Cooke F (2000) Body weight and feather growth of male Barrow’s goldeneye during wing molt. Condor 102:228–231CrossRefGoogle Scholar
  49. Walther G, Post E, Convey P, Menzels A, Parmesan C, Beebee TJC, Fromentin J, Hoegh-Guldberg O, Bairlein F (2002) Ecological responses to recent climate change. Nature 416:389–395PubMedCrossRefGoogle Scholar
  50. Ward DH, Reed A, Sedinger JS, Black JM, Derksen DV, Castelli PM (2005) North American brant: effects of changes in habitat and climate on population dynamics. Glob Change Biol 11:869–880CrossRefGoogle Scholar
  51. Weller MW, Jensen KC, Taylor EJ, Miller MW, Bollinger KS, Derksen DV, Esler D, Markon CJ (1994) Assessment of shoreline vegetation in relation to use by molting black brant Branta bernicla nigricans on the Alaska Coastal Plain. Biol Conserv 70:219–225CrossRefGoogle Scholar
  52. Williams JE, Kendeigh SC (1982) Energetics of the Canada goose. J Wildl Manage 46:588–600CrossRefGoogle Scholar

Copyright information

© US Government 2011

Authors and Affiliations

  • Tyler L. Lewis
    • 1
    • 2
  • Paul L. Flint
    • 1
  • Dirk V. Derksen
    • 1
  • Joel A. Schmutz
    • 1
  • Eric J. Taylor
    • 3
  • Karen S. Bollinger
    • 4
  1. 1.U.S. Geological SurveyAlaska Science CenterAnchorageUSA
  2. 2.Department of Biology and WildlifeInstitute of Arctic Biology, University of Alaska FairbanksFairbanksUSA
  3. 3.U.S. Fish and Wildlife ServiceMigratory Bird ManagementAnchorageUSA
  4. 4.U.S. Fish and Wildlife ServiceMigratory Bird ManagementFairbanksUSA

Personalised recommendations