Answers are grounded in this paper and cite their sources.
Explores trait-environment relationships in native and exotic species, suggesting implications for biodiversity models.
Functional traits allow ecologists to synthesise general rules of community assembly. In joint species distribution modelling, traits are increasingly used to explain species–environment relationships, with the intention of being able to generalise predictions to other species in other systems. However, recent studies have challenged the transferability of trait‐based species distribution models across systems, by showing that not all trait–environment relationships are broadly applicable. There is therefore a need to reassess the utility of functional traits by examining when trait‐based models can be generalised and when they cannot. Using a dataset of tropical post‐agricultural secondary forests of two different land‐use histories (‘abandoned‐land forests' in Singapore, where former plantations were abandoned with the plantation trees still standing, and ‘waste‐woodlands', where trees were clear‐cut before the land was abandoned), we analysed the abundances of 80 woody‐plant species in 97 plots as a function of trait–environment interactions. To quantify model transferability, we split the dataset into factorial combinations of native versus exotic species and the two land‐use histories, cross‐validated them against one another, and then examined their prediction accuracies against the environmental and trait dissimilarities between datasets. We found that models trained with native species and/or abandoned‐land forests generally transferred well onto exotic species and/or waste‐woodlands, as native species and abandoned‐land forests encompassed broader, better sampled environmental and trait spaces. In contrast, the reverse transfers had less accurate predictions due to both environmental and trait dissimilarities, which led to disparate trait–environment coefficients. These findings suggest that species (not necessarily native) and more heterogeneous systems that span wider trait and environmental ranges are useful for building predictive models of biological invasions. More broadly, our study demonstrates that some ecological groups or systems yield more general insights than others, so it is crucial to identify the bounds within which each trait–environment relationship operates.
No takes yet. Share an insight, caveat, or question.
Lai et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: