Modeling asymmetric Thermal Performance Curves for proportion data

Abstract

Ectotherms are subject to temperature impacts on a variety of functional traits, such as fecundity and development. Quantifying the shape of the response of traits to temperature involves fitting (usually) non-linear curves to empirically derived data. This study introduces the LINEX Thermal Performance Curve (TPC), a novel non-linear predictor for modeling asymmetric thermal responses in biological traits. Unlike traditional symmetric TPCs, the LINEX function captures a wide range of asymmetric shapes (e.g., left- or right-skewed), making it particularly suitable for traits like survival or fecundity that often exhibit non-normal, bounded distributions. The study implements the LINEX TPC within a Bayesian framework using the bayesTPC R package. We demonstrate the models flexibility and utility through simulations and experimental data from arthropod thermal traits. Results highlight the LINEX TPCs superior fit for asymmetric traits compared to standard quadratic models, as evidenced by improved performance metrics. The approach addresses challenges in estimating uncertainty in bounded data, providing a robust tool for ecological and evolutionary studies that require precise thermal sensitivity modeling.