Projected climate reshuffling based on multivariate climate-availability, climate-analog, and climate-velocity analyses: implications for community disaggregation

Abstract

There is a need for biologically relevant metrics of climate risk for regional- to global-scale climate vulnerability assessments and adaptation planning. Here, we develop, combine, and compare univariate and multivariate forms of several metrics (climate-avail- ability, climate-analog, and two forms of climate-velocity) used to assess the risks arising from future climate change, using downscaled climate projections for Wisconsin (USA) as a case study. Climate-availability and climate-analog analyses show little or no overlap between late-20th-century and projected late-21st-century climates for Wisconsin, and large differences among variables in the distance, bearing, and velocity of projected climate change. There is a strong negative correlation between geographic and climatic distances to closest analogs, creating a tradeoff when climate velocity is assessed using multivariate analog-based approaches: some locations have no good analogs anywhere in future climate space and so analog-based methods pick nearby locations, resulting in low velocity esti- mates. local velocities projected for Wisconsin are higher than global means. In this region, lake effects, not topographic heterogeneity, exert the strongest influences on regional patterns of climate-velocity and analogs. The multivariate analog-based velocities are correlated with univariate velocity measures that are scaled to local spatial heterogeneity, with the magnitude and correlation analog-based velocities estimates most similar to those of the intervariable mean of climate velocities. Because species are differentially sensitive to particular dimensions of climate change, and vary in their dispersal capacity, the strong differences among climate variables in the spatial direction, distance, and rate of projected climate change provide a powerful mechanism for community restructuring.