Mechanistic distribution models: Incorporating demography and physiology

Background/Question/Methods
Statistical niche models estimate parameters phenomenologically by relating current distributions to underlying environmental conditions. In contrast, mechanistic models incorporate parameters based on organisms’ traits that describe survival and fitness in given environmental conditions. Mechanistic modeling approaches include translating environmental conditions into biologically-relevant metrics such as potential activity hours; estimating thermal sensitivity curves for survival and development; and using foraging energetics to link environmental conditions and demography. We use two focal species- a skipper butterfly and a fence lizard- to compare an array of statistical and mechanistic models.

Results/Conclusions
Statistical models require less biological data to implement, but offer only a limited potential to consider how variation between individuals influences species’ ranges. Mechanistic models successfully predict how a species’ response to environmental change depends on its physiology and morphology. Biological processes such as species’ interactions and movement limitations are implicitly incorporated in statistical models. While mechanistic models do not generally include these biological processes, the models can be extended to explicitly include them. We find that the relative performance of mechanistic and statistical models varies with the type of question asked.