While the effects of climate (long-term, prevailing weather) on species abundance, range, and genetic diversity are widely studied, short-term, localised variations in atmospheric conditions (i.e., weather) can also rapidly alter species’ geographic ranges and population sizes, but little is known about how they affect genetic diversity. We investigated the relationship between weather and range-wide genetic diversity in a marsupial, Bettongia gaimardi, using dynamic species distribution models (SDMs). Genetic diversity was lower in parts of the range where weather-based SDM predicted high variability in probability of B. gaimardi occurrence during 1950–2009. This is likely an effect of lower population sizes and extinction-recolonisation cycles in places with highly variable weather. Spatial variation in genetic diversity was also better predicted by mean probabilities of B. gaimardi occurrence from weather- than climate-based SDMs. Our results illustrate the importance of weather in driving population dynamics and species distributions on decadal time-scales and thereby affecting genetic diversity. Modelling the links between changing weather patterns, species distributions and genetic diversity will allow researchers to better forecast biological impacts of climate change.

The genetic data are represented by a genepop file of 5057 filtered SNP loci from 188 Bettongia gaimardi individuals, and a csv file containing metadata for each individual sampled. Ear biopsies were collected from 188 live-trapped or road-killed B. gaimardi individuals at 17 sites across Tasmania in 2006-2007 and 2015-2017. Individuals were genotyped using a reduced representation sequencing technique, DArTseq^TM, and the resulting SNPs were filtered as described in the Supplementary Materials S1 section of the associated paper.

The climate and weather model datasets represent rasters of probability of B. gaimardi occurrence based on climate models, and mean and standard deviation of probability of occurrence based on weather models. As described in the associated paper, we used Maxent to create models of probability of occurrence based on climate and weather, using 1043 occurrence records spanning the years 1961 – 2009, and 8 long-term climatic variables (climate model) or 14 short-term weather variables (weather model). Climate and weather probability of occurrence were then projected for B. gaimardi in each grid cell across Tasmania. A single projection was created for the climate model. For weather, we projected probability of occurrence at monthly intervals from 1950 – 2009, and then calculated the mean and standard deviation of probability in each grid cell across the whole time period. Weather and climate suitability models were developed following the methodology of Bateman et al. (2012)

1. The SNP alleles in the genepop file are represented as ‘03’ and ‘04’. Missing values are represented by ‘0000’. In the metadata file, samples (and populations) are listed in the same order as in the genepop file, and missing data is represented by 0.
2. The No Data value for the raster files is -3.39999995214e+38