Climate and urbanization data
We included four climate and one urbanization variable in our modeling framework. The climate variables were year-specific and included mean annual temperature, annual precipitation, temperature seasonality, and precipitation seasonality. We obtained annual mean values of the maximum 2-m air temperature data and annual precipitation values for North America at a 1-km spatial resolution for the five years of our study from Daymet (Thornton et al., 2016a). We generated annual temperature and precipitation seasonality using the monthly maximum temperature and precipitation summaries provided by Daymet at a 1-km spatial resolution (Thornton et al., 2016b). Temperature seasonality was calculated as the standard deviation of the monthly maximum temperature values for the corresponding year, and precipitation seasonality was the coefficient of variation of the monthly precipitation values for the corresponding year.
We used estimated human population density for the year 2020 as a proxy for urbanization and obtained this data from the Center for International Earth Science Information Network, which provides global estimates of population density at a 0.25 degree resolution (~27-km; CIESIN, 2017). Global estimates of human population density are only available on five-year intervals, so we choose to only use the 2020 estimates as a proxy for urbanization given our sampling temporal extent. Year-specific changes in human population density are minimal compared to the variation across space, and therefore should have minimal impact on statistical models.