a For SESA and LALO, age was log10-transformed in all models to control for heteroscedasticity.
b Females are the reference level; sexes unknown for SESA and LALO.
The effect of the environmental covariates on chick growth varied by species (Table 4). For brant and snow geese, age-specific body mass of goslings increased with earlier nest timing (nest incubation date – date of 50% snow cover; Table 4). For brant, lower resource abundance (g m-2 subspathacea biomass at 15 days of age; Table 4) was associated with increased gosling mass, and the body mass of snow goose goslings declined as thaw-degree days increased (Table 4). For semipalmated sandpipers, higher age-specific chick masses were associated with warmer temperatures and nests that were initiated early with respect to the date of 50% snow cover (Table 4). For longspurs, all model-averaged parameters for environmental variables were small and with confidence intervals that broadly overlapped zero (Table 4), indicating that none of the environmental variables that we included in our models meaningfully predicted the mass of chicks of this species. Finally, wind speed (all species) and arthropod abundance (insectivores) did not meaningfully influence chick mass as indicated by confidence intervals on parameter estimates that overlapped zero (Table 4).
For environmental covariates that influenced chick growth (Table 4), we estimated age-specific chick masses for each species using values of the 25th and 75th quartiles of these predictors to assess the effect of optimal and sub-optimal (see Methods: Analysis) values of these variables on chick mass. Across the ages and sexes of chicks for which we modeled growth, the difference between optimal and sub-optimal chick-growth conditions resulted in body mass differences of 5.0–8.5% for brant, 9.1–14.3% for snow geese, and 24.4% for semipalmated sandpipers. For snow geese (Figs 4c, 4d) and semipalmated sandpipers (Fig. 4e), the 95% confidence intervals on estimates derived under optimal conditions did not overlap those derived under sub-optimal conditions, but those for brant overlapped slightly (Figs 4a, 4b).
FIGURE 4 Model-averaged predictions of chick growth of female (a ) and male (b ) black brant, female (c ) and male (d ) snow geese, semipalmated sandpipers (e ), and Lapland longspurs (f ) at the Colville River, Alaska. Brant and snow goose goslings measured from 2012–2017, semipalmated sandpiper and Lapland longspur chicks from 2015–2017. Measured values are represented by black circles, and black lines represent estimates under average environmental conditions. Red and blue lines for brant, snow geese, and semipalmated sandpipers represent body-mass estimates of chicks under optimal (red lines) and sub-optimal (blue lines) environmental conditions based on variation in biologically meaningful predictor variables (see Table 3 for species-specific variables). For all colors, heavy lines represent model-averaged estimates, and fine lines represent the associated 95% confidence intervals