Contrasting environments shape thermal physiology across the spatial range of the sandhopper Talorchestia capensis

Integrating thermal physiology and species range extent can contribute to a better understanding of the likely effects of climate change on natural populations. Generally, broadly distributed species show variation in thermal physiology between populations. Within their distributional ranges, populations at the edges are assumed to experience more challenging environments than central populations (fundamental niche breadth hypothesis). We have investigated differences in thermal tolerance and thermal sensitivity under increasing/decreasing temperatures among geographically separated populations of the sandhopper Talorchestia capensis along the South African coasts. We tested whether the thermal tolerance and thermal sensitivity of T. capensis differ between central and marginal populations using a non-parametric constraint space analysis. We linked thermal sensitivity to environmental history by using historical climatic data to evaluate whether individual responses to temperature could be related to natural long-term fluctuations in air temperatures. Our results demonstrate that there were significant differences in the thermal response of T. capensis populations to both increasing/decreasing temperatures. Thermal sensitivity (for increasing temperatures only) was negatively related to temperature variability and positively related to temperature predictability. Two different models fitted the geographical distribution of thermal sensitivity and thermal tolerance. Our results confirm that widespread species show differences in physiology among populations by providing evidence of contrasting thermal responses in individuals subject to different environmental conditions at the limits of the species’ spatial range. When considering the complex interactions between individual physiology and species ranges, it is not sufficient to consider mean environmental temperatures, or even temperature variability; the predictability of that variability may be critical.