Skip to main content
Dryad logo

Pitfalls of ignoring trait resolution when drawing conclusions about ecological processes

Citation

Kohli, Brooks; Jarzyna, Marta (2022), Pitfalls of ignoring trait resolution when drawing conclusions about ecological processes, Dryad, Dataset, https://doi.org/10.5061/dryad.0k6djhb03

Abstract

Aim: Understanding how ecological communities are assembled remains a grand challenge in ecology with direct implications for charting the future of biodiversity. Trait-based methods have emerged as the leading approach for quantifying functional community structure (convergence, divergence) but their potential for inferring assembly processes rests on accurately measuring functional dissimilarity among community members. Here, we argue that trait resolution (from finest-resolution continuous measurements to coarsest-resolution binary categories) remains a critically overlooked methodological variable, even though categorical classification is known to mask functional variability and inflate functional redundancy among species or individuals.

Innovation: We present the first detailed predictions of trait resolution biases and demonstrate, with simulations, how the distortion of signal strength by increasingly coarse-resolution traits can fundamentally alter functional structure patterns and the interpretation of causative ecological processes (e.g., abiotic filters, biotic interactions). We show that coarser trait data impart different impacts on the signals of divergence and convergence, implying that the role of biotic interactions may be underestimated when using coarser traits. Furthermore, in some systems, coarser traits may overestimate the strength of trait convergence, leading to erroneous support for abiotic processes as the primary drivers of community assembly or change.

Main conclusions: Inferences of assembly processes must account for trait resolution to ensure robust conclusions, especially for broad-scale studies of comparative community assembly and biodiversity change. Despite recent improvements in the collection and availability of trait data, great disparities continue to exist among taxa in the number and availability of continuous traits, which are more difficult to acquire for large numbers of species than coarse categorial assignments. Based on our simulations, we urge the consideration of trait resolution in the design and interpretation of community assembly studies and suggest a suite of practical solutions to address the pitfalls of trait resolution biases.

Methods

All data in this project are simulated following previously developed approaches to generate species pools and trait values to serve as the starting data for simulation iterations. The primary method, reported in the main text, follow those of Botta-Dukát & Czúcz, 2016 (Testing the ability of functional diversity indices to detect trait convergence and divergence using individual-based simulation. Methods in Ecology and Evolution, 7, 114–126).  An alternate method based largely on McPherson et al. 2017 (A simulation tool to scrutinise the behavior of functional diversity metrics. Methods in Ecology and Evolution, 9, 200–206) and Kraft & Ackerly, 2010 (Functional trait and phylogenetic tests of community assembly across spatial scales in an Amazonian forest. Ecological Monographs, 80, 401–422) was also used, with results reported primarily in the Supporting Information of the paper.

Usage Notes

Files here include the initial species pool data (abundance matrices and continuous trait values) used for the simulation results reported in the main text (9 files; names starting with "MainSims_") of the article to enable exact replication of our results.  In addition, all simulation data and outputs relevant to Supplementary Methods and Results described in Appendix S1 and S2 (4 files; names starting with "SI_Methods_") of the article are also provided, including the initial regional pools and trait values, summary output of SES values, Wilcoxon tests, and p-values across all simulation iterations. Associated R code for main results and supplementary results can be found at https://github.com/Jarzyna-Lab and on Zenodo: https://doi.org/10.5281/zenodo.4497961.

Funding

National Science Foundation, Award: DEB 1926598