Untangling Sampling Bias From Lemur Dietary Specialization.

Identifying the drivers of wildlife dietary specialization is fundamental to understanding trophic interactions and species' functional roles in their environment. However, sampling bias is pervasive in trophic interaction research, especially in biodiverse areas with cryptic species. We aim to investigate the role of sampling bias in mediating the estimated effects of functional traits on dietary specialization. Specifically, we improve estimates of observed lemur dietary richness by analyzing trait-based biases in lemur-plant ecological interactions. First, we quantified undersampling of plants and their interactions with lemurs. Next, we tested the inclusion of sampling effort on the estimated effects of trait predictors by comparing three negative binomial regression models with the following predictors: (i) five lemur traits (body mass, litter size, group size, diurnality, frugivory), (ii) sampling effort, and (iii) the five traits together with sampling effort. We then assessed the influence of sampling bias on lemur traits and considered the uncertainty of these mediation effects on general linear model predictions. Sampling effort was a better predictor of observed dietary richness than lemur traits. By accounting for sampling bias and its effects on functional traits, we can better understand the ecological roles of species in their environments and uncover critical mechanisms underlying trophic interactions.