Challenges and advances in measuring phenotypic convergence

Abstract

Tests of phenotypic convergence can provide evidence of adaptive evolution, and the popularity of such studies has grown in recent years due to the development of novel, quantitative methods for identifying and/or measuring convergence. Two commonly used methods include (i) distance-based methods that measure morphological distances between lineages in phylomorphospace and (ii) fitting evolutionary models to morphological datasets to test whether lineages have evolved toward adaptive peaks. Here, we demonstrate that both types of convergence measures are influenced by the position of putatively convergent taxa in morphospace such that morphological outliers are statistically more likely to exhibit convergence by chance. A more substantial issue is that some methods will often misidentify divergent lineages as being convergent. These issues likely influence the results of many studies, especially those that focus on morphological outliers. To help address these problems, we developed a new distance-based method for measuring convergence that incorporates distances between lineages through time and minimizes the possibility of divergent taxa being misidentified as convergent. We advocate the use of this method when the phylogenetic tips of putatively convergent lineages are of the same or similar geologic ages (e.g., extant taxa), meaning that convergence among the lineages is expected to be synchronous. We conclude by emphasizing that all available convergence measures are imperfect, and researchers should recognize the limitations of these methods and use multiple lines of evidence when inferring and measuring convergence.

Carrion converging: Skull shape predicts feeding ecology in vultures

“…Using the C measure of Stayton (2015), within-group phenotypic convergence was found to be significant, and relatively consistent within all feeding groups, with an average of 39% convergence in gulpers, 33% in rippers and 35% in scrappers. This is contradicted by the results of the Ct measure of phenotypic convergence of Grossnickle et al (2023), with no groups showing significant convergence overall. Examples of significantly converging taxa, however, can be found by focusing on smaller numbers of taxa that appear to show convergence within morphospace, rather than across whole groups.…”

“…This is contradicted by the results of the Ct measure of phenotypic convergence of Grossnickle et al. (2023), with no groups showing significant convergence overall. Examples of significantly converging taxa, however, can be found by focusing on smaller numbers of taxa that appear to show convergence within morphospace, rather than across whole groups.…”

“…Mean phenotypic convergence is statistically significant within all feeding groups with the C measure of phenotypic convergence (Stayton, 2015; gulper C 1 = 0.39, P = <0.001; ripper C 1 = 0.33, P = <0.001; scrapper C 1 = 0.35, P = <0.001). Convergence is not significant across any group when implementing the Ct measure (Grossnickle et al., 2023; ripper, Ct 1 = −0.296, P = 0.47; gulper, Ct 1 = −0.978, P = 0.55; scrapper, Ct 1 = −0.105, P = 0.05). When examining putatively converging taxa rather than all taxa within feeding groups, however, convergence was significant and high for gulper ( Vultur gryphus , Gymnogyps californianus , Gyps tenuirostris ; Ct 1 = 0.63, P = 0.00, Fig.…”

“…This method has been shown to overestimate convergence in outlying taxa, and so we also applied the method developed by Grossnickle et al. (2023), Ct , which measures phenotypic distance at equivalent points on a time‐scaled phylogeny. This was implemented in the R package convevol (Stayton, 2015), using the same parameters as the C measures calculated above.…”

“…Results were compared with a set of 100 simulations under a BM null model of evolution to provide a significance value for each cluster, using the first 14 principal components, accounting for a cumulative 95% of total shape disparity to reduce dimensionality, and implemented with the 'convratsig' function in the R package convevol (v. 2.0.0, Stayton, 2015). This method has been shown to overestimate convergence in outlying taxa, and so we also applied the method developed by Grossnickle et al (2023), Ct, which measures phenotypic distance at equivalent points on a time-scaled phylogeny. This was implemented in the R package convevol (Stayton, 2015), using the same parameters as the C measures calculated above.…”