Multiple regression models predicting contemporary specialization in plant-hummingbird networks.

Predictor estimates are for each model given as standardized regression coefficients. Predictors included in the best-fit multiple regression models are: network size, i.e, species richness in the network (SIZE); mean annual precipitation (MAP); Quaternary climate-change velocity (VELOCITY). None of the other predictors included in the analysis - length of study period (DAYS); mean annual temperature (MAT); precipitation seasonality (SEASP); temperature seasonality (SEAST) - were included in any of the best-fit models, and are therefore not included here. Moran's I and VIF/CN show that neither positive spatial autocorrelation nor multicollinearity was a problem in our models. See Tables S2, S3 and Materials and Methods for modelling approach.**P*PNSP>0.05.