Are Species Distribution Models That Use Climate Variables Suitable to Forecast the Changing Distribution of Freshwater Organisms?

Species distribution models can be used to identify locations with environmental conditions suitable for populations in the past, present and future. However, climate variables commonly applied in species distribution models may be unsuitable for freshwater species because their explanatory power is difficult to assess. We examined the suitability of such models using a pair of extant mayfly species, Ichthybotus hudsoni and I. bicolor , that are allopatric in the main islands of Aotearoa–New Zealand. These mayflies were confirmed as distinct species by concordance of hindwing colour, forewing shape and mtDNA. Using present‐day species ranges and climate variables, we built models to predict the distribution of each during the climate of the last glacial maximum. Genetic data were used to validate predictions derived from hindcasted species distribution models. Species distribution models suggested that climatically suitable habitat for the two mayfly species was widespread in lowland New Zealand during the last glacial maximum, from which we predicted a genetic signature of large, stable populations for both. High haplotype diversity and a signature of isolation by distance were detected as predicted for I. hudsoni in North Island, but models did not predict the observed association between latitude and nucleotide diversity. A small I. bicolor sample size in South Island limited our ability to distinguish patterns, but the low genetic diversity found suggests poor performance of species distribution models based on climate variables. We conclude that species distribution models based solely on climate variables are unlikely to accurately predict changes in the distributions of freshwater organisms.