Résumé
Despite many studies illustrating the perils of utilising mitochondrial
DNA in phylogenetic studies, it remains one of the most widely used
genetic markers for this purpose. Over the last decade, nuclear introns
have been proposed as alternative markers for phylogenetic reconstruction.
However, the resolution capabilities of mtDNA and nuclear introns have
rarely been quantified and compared. In the current study we generated a
novel ∼5 kb dataset comprising six nuclear introns and a mtDNA fragment.
We assessed the relative resolution capabilities of the six intronic
fragments with respect to each other, when used in various combinations
together, and when compared to the traditionally used mtDNA. We focused on
a major clade in the horseshoe bat family (Afro-Palaearctic clade;
Rhinolophidae) as our case study. This old, widely distributed and
speciose group contains a high level of conserved morphology. This
morphological stasis renders the reconstruction of the phylogeny of this
group with traditional morphological characters complex. We sampled
multiple individuals per species to represent their geographic
distributions as best as possible (122 individuals, 24 species, 68
localities). We reconstructed the species phylogeny using several
complementary methods (partitioned Maximum Likelihood and Bayesian and
Bayesian multispecies-coalescent) and made inferences based on consensus
across these methods. We computed pairwise comparisons based on
Robinson–Foulds tree distance metric between all Bayesian topologies
generated (27,000) for every gene(s) and visualised the tree space using
multidimensional scaling (MDS) plots. Using our supported species
phylogeny we estimated the ancestral state of key traits of interest
within this group, e.g. echolocation peak frequency which has been
implicated in speciation. Our results revealed many potential cryptic
species within this group, even in taxa where this was not suspected a
priori and also found evidence for mtDNA introgression. We demonstrated
that by using just two introns one can recover a better supported species
tree than when using the mtDNA alone, despite the shorter overall length
of the combined introns. Additionally, when combining any single intron
with mtDNA, we showed that the result is highly similar to the mtDNA gene
tree and far from the true species tree and therefore this approach should
be avoided. We caution against the indiscriminate use of mtDNA in
phylogenetic studies and advocate for pilot studies to select nuclear
introns. The selection of marker type and number is a crucial step that is
best based on critical examination of preliminary or previously published
data. Based on our findings and previous publications, we recommend the
following markers to recover phylogenetic relationships between recently
diverged taxa (<20 My) in bats and other mammals: ACOX2, COPS7A,
BGN, ROGDI and STAT5A.