Limited, episodic diversification and contrasting phylogeography in a New Zealand cicada radiation.

Background: The New Zealand (NZ) cicada fauna contains two co-distributed lineages that independently colonized the isolated continental fragment in the Miocene. One extensively studied lineage includes 90% of the extant species (Kikihia + Maoricicada + Rhodopsalta; ca 51 spp.), while the other contains just four extant species (Amphipsalta - 3 spp. + Notopsalta - 1 sp.) and has been little studied. We examined mitochondrial and nuclear-gene phylogenies and phylogeography, Bayesian relaxed-clock divergence timing (incorporating literature-based uncertainty of molecular clock estimates) and ecological niche models of the species from the smaller radiation.
Results: Mitochondrial and nuclear-gene trees supported the monophyly of Amphipsalta. Most interspecific diversification within Amphipsalta-Notopsalta occurred from the mid-Miocene to the Pliocene. However, interspecific divergence time estimates had large confidence intervals and were highly dependent on the assumed tree prior, and comparisons of uncorrected and patristic distances suggested difficulty in estimation of branch lengths. In contrast, intraspecific divergence times varied little across analyses, and all appear to have occurred during the Pleistocene. Two large-bodied forest taxa (A. cingulata, A. zelandica) showed minimal phylogeographic structure, with intraspecific diversification dating to ca. 0.16 and 0.37 Ma, respectively. Mid-Pleistocene-age phylogeographic structure was found within two smaller-bodied species (A. strepitans - 1.16 Ma, N. sericea - 1.36 Ma] inhabiting dry open habitats. Branches separating independently evolving species were long compared to intraspecific branches. Ecological niche models hindcast to the Last Glacial Maximum (LGM) matched expectations from the genetic datasets for A. zelandica and A. strepitans, suggesting that the range of A. zelandica was greatly reduced while A. strepitans refugia were more extensive. However, no LGM habitat could be reconstructed for A. cingulata and N. sericea, suggesting survival in microhabitats not detectable with our downscaled climate data.
Conclusions: Unlike the large and continuous diversification exhibited by the Kikihia-Maoricicada-Rhodopsalta clade, the contemporaneous Amphipsalta-Notopsalta lineage contains four comparatively old (early branching) species that show only recent diversification. This indicates either a long period of stasis with no speciation, or one or more bouts of extinction that have pruned the radiation. Within Amphipsalta-Notopsalta, greater population structure is found in dry-open-habitat species versus forest specialists. We attribute this difference to the fact that NZ lowland forests were repeatedly reduced in extent during glacial periods, while steep, open habitats likely became more available during late Pleistocene uplift.