Cryptic species diversity in a widespread neotropical tree genus: The case of Cedrela odorata

Abstract

Premise: Reconciling the use of taxonomy to partition morphological variation and describe genetic divergence within and among closely related species is a persistent challenge in phylogenetics. We reconstructed phylogenetic relationships among Cedrela odorata (Meliaceae) and five closely allied species to test the genetic basis for the current model of species delimitation in this economically valuable and threatened genus.

Methods: We prepared a nuclear species tree with the program SNPhylo and 16,000 single-nucleotide polymorphisms from 168 Cedrela specimens. Based on clades present and ancestral patterns ADMIXTURE, we designed nine species delimitation models and compared each model to current taxonomy with Bayes factor delimitation. Timing of major lineage divergences was estimated with the program SNAPP.

Results: The resulting analysis revealed that modern C. odorata evolved from two genetically distinct ancestral sources. All species delimitation models tested better fit the data than the model representing current taxonomic delimitation. Models with the greatest marginal likelihoods separated Mesoamerican C. odorata and South American C. odorata into two species and lumped C. angustifolia and C. montana as a single species. We estimated that Cedrela diversified in South America within the last 19 million years following one or more dispersal events from Mesoamerican lineages.

Conclusions: Our analyses show that the present taxonomic understanding within the genus obscures divergent lineages in C. odorata due in part to morphological differentiation and taxonomic distinctions that are not predictably associated with genetic divergence. A more accurate application of taxonomy to C. odorata and related species may aid in its conservation, management, and restoration efforts.

Keywords: CITES; Meliaceae; SNPs; conservation; illegal logging; multispecies coalescence; neotropical forests; next generation sequencing; target capture; taxonomy.

Figure 1

Maximum likelihood nuclear species tree, species identifications, and results of ancestry estimation for K = 3, 4, 6, and 8. Tips, or individual specimens, are color‐coded by latitude listed on the herbarium specimen label, and correspond to the map, showing a portion of South America and the geographic positions of all specimens included in the species tree. The colors in the column labeled “Spp” indicate the species identification listed on the specimen label and correspond to the legend labeled “Spp.” The bar plots show ancestry models indicated by K = 3, 4, 6, and 8. For ease of interpretation, we have used a similar color palette for ancestral populations such that proportions correlate with species identifications where possible, but the colors in these bar plots do not indicate species identity. Numbers at nodes indicate bootstrap support values for 1000 replicates with the same data and RAxML. Specimens in red were selected for species delimitation with BFD*.

Figure 2

Maps show the geographic distribution of ancestral populations as pie charts. The legend corresponds to the ADMIXTURE models presented in Figure 1: (A) K = 3, (B) K = 4, (C) K = 6, (D) K = 8.

Figure 3

Species delimitation models tested with BFD* and their marginal likelihood scores shown in the scatter plot. The phylogeny is a reduced version of the maximum likelihood species tree generated with one specimen from each taxon and RAxML via the CIPRES server with 1000 bootstrap replicates and Swietenia mahagoni as the outgroup. Cedrela odorata M represents C. odorata s.s., and C. odorata SA represents C. odorata from southeastern Costa Rica, Panama, and South America. Colored blocks show different taxonomic concepts for C. odorata and allied species. The scatterplot shows estimated mean marginal likelihood estimates (MLE) across BFD* replicates for each species delimitation model (Table 1). Numbers near mean MLE estimates are model rankings with number 1 indicating the model that we determined to be the best representation of the data (but also see Table 2).

Figure 4

Maximum clade credibility tree for 30 Cedrela and Swietenia mahagoni specimens via SNAPP and BEAST2. Numbers on the branches indicate posterior probabilities for species tree topology. Numbers at the right of nodes indicate median node ages. The 95% HPD interval around node ages are bracketed and indicated by node bars. Triangles on branches indicated time calibration positions 51.9 Ma corresponding to the age of the stem node of the Cedreloideae, and 31.05 Ma corresponding to the stem age of Cedreleae, and 24.25 Ma constraining constrained the stem node leading to the outgroup, S. mahagoni.