Model-averaging results for fecundity linear models.

Model-averaged beta coefficients (β), bootstrapped unconditional standard errors (SE), 95% confidence interval lower and upper limits (LCL and UCL), and summed AIC weights (ω+) of parameters for models of group fecundity, individual fecundity (juveniles/adult female), and litter size of hoary marmots in the Ruby Range, Yukon Territory. SEs and 95% confidence limits were bootstrapped with 1000 replications. Except for litter size, analyses were based on data from 1999–2004 and 2007–2009. Due to highly significant poisson overdispersion, juveniles per group were modeled using negative binomial linear models. Juveniles per female were modeled using Poisson mixed models with social group random effects. Litter size data were approximately normally distributed, and were modeled as such. Only models that were within 7 AICc of the top model, and with smaller AICc values than an intercept-only (null) model were used for model averaging and for calculating AICc weights [32].Model-averaging results for fecundity linear models.