Skip to main content

Can we implement mechanistic niche models without measuring traits?

Cite this dataset

Radomski, Tom (2021). Can we implement mechanistic niche models without measuring traits? [Dataset]. Dryad.


Despite increasing demand for more mechanistic models of species’ distributions, they remain difficult to implement due to their intense data requirements. Here, we suggest that three existing tools might help researchers estimate missing trait data that can fill necessary requirements: phylogenetic imputation (incorporating evolutionary relationships into the imputation process), allometry, and biophysics. These tools can bridge the gap between species for which we have fairly little data and ecological models of their distributions. We focus on ecophysiological traits of salamanders; a simple theoretical model of water loss showed that changing body mass results in a decrease in mass-specific water loss, driven by both increasing total size and increasing the thickness of the epidermis. Yet this has two costs – mass-specific oxygen absorption decreases with the surface area-to-volume ratio, and a thicker epidermis slows the rate of respiration. Using a mechanistic distribution model, we modeled the ranges of six species of lungless, terrestrial salamanders. The model performed well for two species of Plethodon and one species of Desmognathus, but poorly for three other species; this discrepancy likely reflects biases in the data available to build the imputation model and biological assumptions on which the mechanistic distribution model is based. Though overall model performance was mediocre, the parameter estimates used in this study may be useful when fewer total parameters are missing. We provide relevant R functions for the mechanistic niche model in the package ‘ManderGander’.


Body mass, resting metabolic rate, and epidermal thickness were all collated from published sources.