Trait diversity metrics can perform well with highly incomplete datasets

Abstract

1. Characterizing changes in trait diversity at large spatial scales provides insight into the impact of human activity on ecosystem structure and function. However, the approach is often based on trait datasets that are incomplete and unrepresentative, with uncertain impacts on trait diversity estimates. 2. To address this knowledge gap, we simulated random and biased removal of data from a near complete avian trait dataset (9579 species) and assessed whether trait diversity metrics were robust to data incompleteness with and without using imputation to fill data gaps. Specifically, we compared two commonly used metrics each calculated with two methods: trait richness (calculated with convex hulls and trait probabilities densities) and trait divergence (calculated with distance-based Rao and trait probability densities). 3. Without imputation, estimates of global avian trait diversity (richness and divergence) were robust when 30-70% of species had missing data for four out of 11 continuous traits, depending on severity of bias and the method used. However, when missing traits were imputed based on present morphological trait data and phylogeny, trait diversity metrics consistently remained representative of the true value, even when 70% of species were missing data for four out of 11 traits and data were not missing at random (biased with respect to body mass). Trait probability densities and distance-based Rao were particularly robust to missingness and bias when combined with imputation, with convex hull-based trait richness being less reliable. 4. Expanding global morphometric datasets to represent more taxa and traits, and to quantify intraspecific variation, remains a priority. In the meantime, our results show that widely used methods can successfully quantify large-scale trait diversity even when data are missing for two-thirds of species, so long as missing traits are estimated using imputation.

funspace: An R package to build, analyse and plot functional trait spaces

“…funspace provides users with an important tool for conducting functional trait‐space analyses from raw trait data and a way to increase the reproducibility of such analyses. The interconnected modules of the package provide in fact all the necessary steps to streamline sometimes troublesome operations, including imputation of missing data (Stewart et al., 2023) or the definition and analysis of functional spaces using multivariate kernel density (Mammola et al., 2021). funspace is explicitly based on the TPD framework (Carmona et al., 2016, 2019), which is not optimized for plotting.…”

“…multivariate density) (Mammola et al., 2021). This limitation of convex hulls becomes especially important in cases when missing data need to be imputed (Stewart et al., 2023). Additionally, mFD lacks support for external variable mapping within the functional spaces, although this falls outside the package's intended scope.…”

“…For example, across mammals, big species with large range areas are generally better informed than small species or species with small ranges (González‐Suárez et al., 2012); similar biases have been described for plants (Carmona, Bueno, et al., 2021; Sandel et al., 2015). Recent research is increasingly suggesting that imputation can largely correct these biases, so that functional diversity patterns inferred from imputed datasets are much closer to the real ones than those that would be estimated using only species with complete trait information (Penone et al., 2014; Stewart et al., 2023).…”

Contrasting impacts of non‐native and threatened species on morphological, life history, and phylogenetic diversity in bird assemblages

“…This way, we included the evolutionary relationships between species in the imputation process by considering the first ten phylogenetic eigenvectors, as recommended by Penone et al (2014). While phylogenetic diversity is based on the phylogenetic distance between pairs of species, using phylogenetic information in the imputation considers the interaction between all traits and the information, so that the positions imputed in the phylogenetic space are much more accurate (Stewart et al, 2023). We used the imputed traits to project species onto the life-history space, utilizing the complete dataset.…”

funspace: an R package to build, analyze and plot functional trait spaces

“…For example, across mammals, big species with large range areas are generally better informed than small species or species with small ranges (González-Suárez et al, 2012); similar biases have been described for plants (Carmona, Bueno, et al, 2021; Sandel et al, 2015). Recent research is increasingly suggesting that imputation can largely correct these biases, so that functional diversity patterns inferred from imputed datasets are much closer to the real ones than those that would be estimated using only species with complete trait information (Penone et al, 2014; Stewart et al, 2022).…”