Molecular phylogeny and divergence times of Malagasy tenrecs: influence of data partitioning and taxon sampling on dating analyses

Abstract

Background: Malagasy tenrecs belong to the Afrotherian clade of placental mammals and comprise three subfamilies divided in eight genera (Tenrecinae: Tenrec, Echinops, Setifer and Hemicentetes; Oryzorictinae: Oryzorictes, Limnogale and Microgale; Geogalinae:Geogale). The diversity of their morphology and incomplete taxon sampling made it difficult until now to resolve phylogenies based on either morphology or molecular data for this group. Therefore, in order to delineate the evolutionary history of this family, phylogenetic and dating analyses were performed on a four nuclear genes dataset (ADRA2B, AR, GHR and vWF) including all Malagasy tenrec genera. Moreover, the influence of both taxon sampling and data partitioning on the accuracy of the estimated ages were assessed.

Results: Within Afrotheria the vast majority of the nodes received a high support, including the grouping of hyrax with sea cow and the monophyly of both Afroinsectivora (Macroscelidea + Afrosoricida) and Afroinsectiphillia (Tubulidentata + Afroinsectivora). Strongly supported relationships were also recovered among all tenrec genera, allowing us to firmly establish the grouping of Geogale with Oryzorictinae, and to confirm the previously hypothesized nesting of Limnogale within the genus Microgale. The timeline of Malagasy tenrec diversification does not reflect a fast adaptive radiation after the arrival on Madagascar, indicating that morphological specializations have appeared over the whole evolutionary history of the family, and not just in a short period after colonization. In our analysis, age estimates at the root of a clade became older with increased taxon sampling of that clade. Moreover an augmentation of data partitions resulted in older age estimates as well, whereas standard deviations increased when more extreme partition schemes were used.

Conclusion: Our results provide as yet the best resolved gene tree comprising all Malagasy tenrec genera, and may lead to a revision of tenrec taxonomy. A timeframe of tenrec evolution built on the basis of this solid phylogenetic framework showed that morphological specializations of the tenrecs may have been affected by environmental changes caused by climatic and/or subsequent colonization events. Analyses including various taxon sampling and data partitions allow us to point out some possible pitfalls that may lead to biased results in molecular dating; however, further analyses are needed to corroborate these observations.

Figure 1

Phylogenetic tree as inferred by maximum likelihood analysis of the concatenated 4,287-bp dataset. Phylogenetic relationships of the investigated mammalian species were reconstructed using ADRA2B, AR, vWF and GHR sequences. Bayesian analyses result in an identical topology. Nodes receiving high support (BP ≥ 90% and PP ≥ 0.99) are marked with filled circles; open circles indicate that nodes received such high support with only one phylogenetic method (either BP or PP). Although the overall phylogenetic relationships as deduced from the present tree are consistent with the broadly accepted branching pattern of the mammalian tree [1], the phylogenetic position of the Eulipotyphla, displaying a high PP node support value, deviates from this consensus. The length of the branch connecting eutherians to the marsupial outgroup was reduced six times. Taxa not indicated by species name are represented by different species in the concatenated dataset, and the higher taxonomic unit is indicated (Table 1).

Figure 2

Phylogenetic relationships of tenrecs as inferred by maximum likelihood analysis of the four separate datasets. DNA matrix lengths were 1,101 bp for ADRA2B, 1,161 bp for AR, 852 bp for GHR and 1,173 bp for vWF. Bayesian analyses result in identical topologies. Nodes receiving high support (BP ≥ 90% and PP ≥ 0.99) are marked with filled circles; open circles indicate that nodes received a high support with only one phylogenetic method (either BP or PP). M. talazaci sequences were only available for GHR.

Figure 3

Timing of tenrec speciation events and Madagascar colonization. Tree topology as in Figure 1. Divergence times were estimated from the concatenated dataset by a Bayesian relaxed molecular clock method, with six time constraints from fossil calibrations (see Material and Methods). One of them, the paenungulate radiation is represented on the chronogram. Black circles indicate the divergence from the non-Malagasy sister group (node 2) and the initial divergence of Malagasy tenrecs (node 3). Standard deviations are indicated by grey bars, and 95% credibility intervals by open bars. The period of a putative land bridge between Madagascar and Africa at 45–26 Mya [53] is shaded.

Figure 4

Congruence of divergence time estimates (A) and their associated standard deviations (SD) (B) when calculated with different partition types. The X-axis represents the estimates without partitioning and the Y-axis the ones with 5, 9 or 12 partitions (see Methods). The age estimates increase with the number of partitions (A) and the SDs are larger for extreme numbers of partitions (none and 12 partitions) (B). For clarity purpose only the age estimates relative to tenrecs are displayed in these graphs; however the estimated ages and SDs in the rest of the tree give the same results.