Alternative models for the relative abundance of crested tits in juniper woodlands of Spain during winter, and results of multimodel inference using all predictor variables.

Sample size is 26 juniper woodlands. X: variables included in the a priori models. Multimodel inference averages the first eight models with ΔAICci). Values presented for variables in multimodel inference are weighted averages of standardized beta regression coefficients obtained in generalized linear models considering model weights Wi (beta values inform about the magnitude and sign of the partial relationships of the predictor variables with the relative abundance of crested tits). AICc: AIC corrected for small sample sizes. ΣWi: sum of Akaike weights for each variable considering those models where they were selected in AIC multimodel inference. K: number of predictor variables+intercept. Habitat variables (HB): hJ - average height of juniper trees; #J - number of juniper trees per 5,000 m2. Tm: minimum night temperature. Food variables: AR - arthropod abundance; FR - fruit abundance index. DL: day length at winter solstice; Landscape variables: ALT - altitude; L1 - multivariate gradient of increasing cover of agricultural landscape around each juniper woodland plot; L2 - multivariate gradient of increasing cover of coniferous and mixed forests dominated by Pinus nigra; L3 - multivariate gradient of increasing cover of holm oak, Quercus ilex, forests.