Marmosops, Matschie, 1916

Díaz-Nieto, Juan F., Jansa, Sharon A. & Voss, Robert S., 2016, DNA sequencing reveals unexpected Recent diversity and an ancient dichotomy in the American marsupial genus Marmosops (Didelphidae: Thylamyini), Zoological Journal of the Linnean Society 176 (4), pp. 914-940 : 927-928

PHYLOGENETICS IN MARMOSOPS

A recent overview of opossum systematics ( Voss & Jansa, 2009) recognized a monophyletic Marmosops containing 15 species, and subsequent descriptions of new taxa ( Voss et al., 2013; García et al., 2014) raised the total to 17 species currently recognized as valid. However, Voss & Jansa (2009: 138) noted that ‘few of the currently recognized species have received critical revisionary attention, and it seems likely that several widespread taxa (e.g., M. fuscatus , M. impavidus , and M. noctivagus ) will prove to be composite’. The present study is the first to assess intra- and interspecific variation for Marmosops by sequencing representatives of all currently recognized species and by including multiple individuals from many widespread taxa. Our findings suggest that the diversity of Marmosops is underestimated by the currently accepted taxonomy, and that the genus might contain as many as 37 species. However, GMYC analyses are based on a number of assumptions about evolutionary processes that need to be considered.

The GMYC model operates in a coalescent-based framework that identifies the point(s) at which the phy- logeny shifts from interspecific (Yule) to intraspecific (coalescent) processes; therefore, differences in branching rate within and between species are crucial for methodological success. In particular, when the coalescent branching rate is much higher than the Yule branching rate, GMYC is likely to be reasonably accurate ( Reid & Carstens, 2012). However, in clades with large population sizes and high speciation rates, the coalescent and Yule processes tend to have similar rates and, in such situations, GMYC has proven to be less accurate ( Esselstyn et al., 2012; Fujisawa & Barraclough, 2013). When implementing the likelihood version of GMYC with our data, the shift from interspecific to intraspecific branching processes was only marginally significant (P = 0.0512), suggesting that the rates in question are not very different, perhaps because speciation rates are high and population sizes are large in Marmosops . Unfortunately, there are currently no independent data with which to evaluate these possibilities.

The likelihood version of the GMYC model has, additionally, several other potential sources of error. Notably, the model does not take into account uncertainty in the evaluated parameters (e.g. coalescent and Yule processes) nor does it account for phylogenetic error ( Reid & Carstens, 2012). By contrast, the Bayesian implementation of GMYC takes uncertainty of the parameters and phylogenetic error into account. Additionally, in our application, BGMYC suggests that the coalescent branching rate is substantially larger than the Yule rate: the mean values that we obtained across 10 000 generations suggest that the rate of branching for the coalescent process is an order of magnitude larger than that for the Yule process (by about 44.1 to 4.4). Therefore, BGMYC could be providing a better estimate of species-level diversity within Marmosops than the corresponding likelihood implementation. However, because the single discrepancy between the two models is nested (Fig. 4), their results are not incongruent; in fact, they provide a scenario that can be further tested with additional evidence (see below).

It is important to highlight that the GMYC model was originally devised to delimit species from singlelocus gene trees in the absence of additional information ( Fujisawa & Barraclough, 2013). Whenever other information – such as sequences from multiple loci, morphological data, or geography – is available, however, that information should be used to inform the results from GMYC ( Fujisawa & Barraclough, 2013). Although our current data set does not include relevant genetic data from other loci, we consider morphology and geographical distributions in the following taxonomic accounts, which discuss the possibility that some putative species delimited by GMYC methods might actually be evolutionarily independ- ent lineages (valid species). In effect, our results provide, for the first time, a set of testable hypotheses based on methodologically explicit data analyses that can serve as the basis for future revisionary work.

Our second principal result, the discovery of a strongly supported basal dichotomy in the genus, implies an ancient speciation event that gave rise to two speciose lineages with broadly overlapping geographical distributions. Based just on the samples analysed for this report ( Figs 1–3 ), members of subgenera I and II are found together throughout much of western Amazonia, in the northern Andes, and in eastern Panama. Apparently, only members of subgenus I occur in northern Venezuela, in the Guianas, and in eastern Amazonia, whereas only subgenus II occurs in south-eastern Brazil. These distributions, together with an estimated divergence age of about 9 000 000 years – based on the time tree in Jansa, Barker & Voss (2014) – and a consistent difference in mean body size between members of the two subgenera where they occur sympatrically (e.g. Patton et al., 2000; Díaz-N et al., 2011; Hice & Velazco, 2012), suggest a long independent history of geographical dispersion and ecological adaptation.

Figures

Figure 1. Collection localities for sequenced specimens of subgenus II of Marmosops. Progressively darker shading indicates the following elevations: pale grey ≥ 500 m, medium grey ≥ 1000 m, dark grey ≥ 2000 m, and darkest grey ≥ 3000 m.

Figure 2. Collection localities for sequenced specimens of subgenus I of Marmosops. Progressively darker shading indicates the following elevations: pale grey ≥ 500 m, medium grey ≥ 1000 m, dark grey ≥ 2000 m, and darkest grey ≥ 3000 m.

Figure 3. Collection localities for sequenced specimens of Atlantic Forest species included in subgenus II of Marmosops. Progressively darker shading indicates the following elevations: pale grey ≥ 500 m, medium grey ≥ 1000 m, and dark grey ≥ 2000 m.