For common community phylogenetic analyses, go ahead and use synthesis phylogenies

Abstract

Should we build our own phylogenetic trees based on gene sequence data, or can we simply use available synthesis phylogenies? This is a fundamental question that any study involving a phylogenetic framework must face at the beginning of the project. Building a phylogeny from gene sequence data (purpose-built phylogeny) requires more effort, expertise, and cost than subsetting an already available phylogeny (synthesis-based phylogeny). However, we still lack a comparison of how these two approaches to building phylogenetic trees influence common community phylogenetic analyses such as comparing community phylogenetic diversity and estimating trait phylogenetic signal. Here, we generated three purpose-built phylogenies and their corresponding synthesis-based trees (two from Phylomatic and one from the Open Tree of Life, OTL). We simulated 1,000 communities and 12,000 continuous traits along each purpose-built phylogeny. We then compared the effects of different trees on estimates of phylogenetic diversity (alpha and beta) and phylogenetic signal (Pagel's λ and Blomberg's K). Synthesis-based phylogenies generally yielded higher estimates of phylogenetic diversity when compared to purpose-built phylogenies. However, resulting measures of phylogenetic diversity from both types of phylogenies were highly correlated (Spearman's $\rho$ > 0.8 in most cases). Mean pairwise distance (both alpha and beta) is the index that is most robust to the differences in tree construction that we tested. Measures of phylogenetic diversity based on the OTL showed the highest correlation with measures based on the purpose-built phylogenies. Trait phylogenetic signal estimated with synthesis-based phylogenies, especially from the OTL, was also highly correlated with estimates of Blomberg's K or close to Pagel's λ from purpose-built phylogenies when traits were simulated under Brownian motion. For commonly employed community phylogenetic analyses, our results justify taking advantage of recently developed and continuously improving synthesis trees, especially the Open Tree of Life.

Keywords: alpha diversity; beta diversity; community phylogenetic structure; open tree of life; phylogenetic diversity; phylogenetic signal; trait.

Figure 1

Workflow to assess effects of commonly used synthesis phylogenies on community phylogenetic diversity and trait phylogenetic signal estimations. Boxes with light yellow background are related to community phylogenetic diversity; boxes with light blue background are related to trait phylogenetic signal. APG, Angiosperm Phylogeny Group; OTL, Open Tree of Life; PD, Faith's phylogenetic diversity; MPD, mean pairwise distance; MNTD, mean nearest taxon distance; Unif, Unifraction; PCD, phylogenetic community dissimilarity; λ, Pagel's lambda; K, Blomberg's K.

Figure 2

Median correlations of phylogenetic alpha diversity values based on different phylogenies.

Figure 3

Median correlations of phylogenetic beta diversity values based on different phylogenies.

Figure 4

Estimated Pagel's λ for traits simulated with divergence rate ${\mathrm{\sigma }}^2$ of 0.2. When traits were simulated under BM and weak OU models, estimated Pagel's λ values based on tree_otl were the closest to those estimated based on tree_purpose in most cases and had smaller variances than other synthesis‐based phylogenies. Note that we allow λ to be larger than 1 in all estimates.

Figure 5

Estimated Blomberg's K for traits simulated with divergence rate ${\mathrm{\sigma }}^2$ of 0.2. Because for Blomberg's K, it is the mean, not the median, value that has the expected value of 1, we did not use boxplots as in Fig. 4. Instead, we added the average values (red points) on top of jittered raw estimated values.

Figure 6

Spearman's rank correlations of estimated Blomberg's K values between tree_purpose and the three synthesis‐based phylogenies.