|Limited, episodic diversification and contrasting phylogeography in a New Zealand cicada radiation.|
|Jump to Full Text|
|PMID: 22967046 Owner: NLM Status: Publisher|
|ABSTRACT: 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.|
|David C Marshall; Kathy B R Hill; Katharine A Marske; Colleen Chambers; Thomas R Buckley; Chris Simon|
Related Documents :
|22052176 - Neurological tests for functional outcome assessment in rodent models of ischaemic stroke.
10840686 - Toward minimalistic modeling of oral drug absorption.
16833686 - Support vector machine and the heuristic method to predict the solubility of hydrocarbo...
16761786 - Spatial angular compounding for elastography without the incompressibility assumption.
20083316 - Correction strategies to manage deformations in head-and-neck radiotherapy.
20040396 - Estimating rainforest biomass stocks and carbon loss from deforestation and degradation...
|Type: JOURNAL ARTICLE Date: 2012-9-11|
|Title: BMC evolutionary biology Volume: 12 ISSN: 1471-2148 ISO Abbreviation: BMC Evol. Biol. Publication Date: 2012 Sep|
|Created Date: 2012-9-12 Completed Date: - Revised Date: -|
Medline Journal Info:
|Nlm Unique ID: 100966975 Medline TA: BMC Evol Biol Country: -|
|Languages: ENG Pagination: 177 Citation Subset: -|
|APA/MLA Format Download EndNote Download BibTex|
Journal ID (nlm-ta): BMC Evol Biol
Journal ID (iso-abbrev): BMC Evol. Biol
Publisher: BioMed Central
Copyright ©2012 Marshall et al.; licensee BioMed Central Ltd.
Received Day: 24 Month: 12 Year: 2011
Accepted Day: 15 Month: 8 Year: 2012
collection publication date: Year: 2012
Electronic publication date: Day: 11 Month: 9 Year: 2012
Volume: 12First Page: 177 Last Page: 177
PubMed Id: 22967046
Publisher Id: 1471-2148-12-177
|Limited, episodic diversification and contrasting phylogeography in a New Zealand cicada radiation|
|David C Marshall1||Email: email@example.com|
|Kathy B R Hill1||Email: firstname.lastname@example.org|
|Katharine A Marske23||Email: KAMarske@bio.ku.dk|
|Colleen Chambers1||Email: email@example.com|
|Thomas R Buckley345||Email: BuckleyT@landcareresearch.co.nz|
|Chris Simon16||Email: firstname.lastname@example.org|
1Department of Ecology and Evolutionary Biology, University of Connecticut, 75 N. Eagleville Rd, Storrs, CT, 06269, USA
2Center for Macroecology, Evolution and Climate, University of Copenhagen, Copenhagen, Denmark
3Landcare Research, Private Bag 92170, Auckland, New Zealand
4School of Biological Sciences, The University of Auckland, Private Bag 92019, Auckland, New Zealand
5Allan Wilson Centre for Molecular Ecology and Evolution, Auckland, New Zealand
6School of Biological Sciences, Victoria University of Wellington, Wellington, New Zealand
The cicadas of New Zealand (NZ) present an example of rapid ecological radiation in response to landscape and climate changes, mainly of the Pliocene (5.3 - 2.6 Ma) and Pleistocene (2.6 - 0.1 Ma) epochs [1-3]. Approximately 55 NZ species form two monophyletic groups, each derived from an overwater colonization event [4,5]. Most belong to a large clade containing three genera: Kikihia Dugdale (ca. 28 taxa), Maoricicada Dugdale (ca. 20 taxa), and Rhodopsalta Dugdale (3 taxa), which are related to New Caledonian cicadas [4,5]. Species of this clade are distributed across all three main islands (North, South, Stewart; see Figure 1 and Figure 2) and several outer islands, and together they inhabit a variety of low- and mid-elevation habitats [6-9]. The diversification of Maoricicada, in particular, is an extraordinary case for the family Cicadidae because its species are largely alpine in distribution. Molecular-clock studies suggest that diversification in both Kikihia and Maoricicada began around 6–7 Ma and accelerated through the Pliocene as mountain-building increased temperature and moisture contrasts throughout NZ and isolated populations geographically [1,2,10].
The remaining NZ cicada species (two genera) belong to the same tribe (Cicadettini) but form a separate monophyletic clade related to Australian taxa [4,5]. This clade is also widely distributed across NZ and includes the familiar wing-clapping Amphipsalta Fleming species. However, these species have received comparatively little study. Amphipsalta zelandica (Boisduval) inhabits low- to mid-elevation forests throughout mainland NZ and nearby islands, while A. cingulata (Fabricius) is found only on North Island (NI), mainly in trees of coastal scrub habitats  (Figure 2). A. strepitans (Kirkaldy) utilizes shrubs (Discaria, Cassinia – ) in dry, open habitat, and it sings on both vegetation and rocks [12,13] from southern NI to southeastern South Island (SI). Notopsalta sericea (Walker) sings from vegetation, volcanic rocks, and clay banks in steep, open country  throughout NI. The small size of the AmphipsaltaNotopsalta radiation is remarkable because its ancestor arrived in NZ at approximately the same time as the KikihiaMaoricicadaRhodopsalta ancestor, and because it likely experienced similar Miocene and Pleistocene landscape and climate changes [14,15].
In this study we use multi-gene molecular-phylogenetic methods, divergence-time analysis, intraspecific sampling, and ecological niche models (ENMs) to examine the evolution of Amphipsalta-Notopsalta. We find that few extant lineages in the clade originated during the late Miocene and Pliocene, either because speciation rates were then low or because most lineages appearing during that time later went extinct. In addition, we find evidence of Pleistocene-age intraspecific diversification with contrasting phylogeographic histories across the species. Species’ modeled geographic ranges at the LGM (Last Glacial Maximum, c. 22,000 cal yrs BP) varied from distributions similar to at present to entirely absent at the LGM refugia.
All sequences were submitted to GenBank under accession numbers JX675238 - JX675427 (COI), JX675449 - JX675465 (Calmodulin), and JX675428 - JX675448Â (EF-1α). Character profiles and partition-specific models selected are shown in Table 1. The two nuclear loci were characterized by low numbers of parsimony-informative sites among ingroup taxa. The absence of a gamma parameter in the Modeltest output for those genes indicates either a simple pattern of among-site rate variation, or insufficient data for estimation of a more complex model. PartitionFinder identified support for a three-partition mtDNA model (with each codon position modeled separately).
COI sequences of the four species formed strongly supported monophyletic groups, and Amphipsalta was recovered as a monophyletic group (91% ML bootstrap), but A. zelandica and A. cingulata were only weakly supported as sister-taxa (Figure 3). COI patristic genetic distances between the four species ranged from 0.21-0.31 substitutions/site (uncorrected p distance 0.061-0.082). The nuclear-gene trees agreed with each other and with the COI topology, and they also supported a monophyletic Amphipsalta (>70%). A. cingulata and A. zelandica were supported as sister-species with >70% ML bootstrap support in both nuclear gene trees (Figure 3).
Phylogeographic structure was evident within all species, but with considerable differences in levels of divergence. Little intraspecific divergence (weak structure) in COI was detected for either A. cingulata or A. zelandica. Maximum patristic genetic distances between haplotypes from the partitioned ML analysis (Figure 3) were 0.005 substitutions/site (s/s) in A. cingulata (0.004 uncorrected, or p distance) and 0.011 s/s (0.009 uncorrected) in A. zelandica. For both species, haplotype patterns suggested a north–south phylogeographic break within NI, especially in A. zelandica, where a distinctive haplotype cluster was limited to northern NI (ML bootstrap 64%).
ML patristic distances within A. strepitans and N. sericea were 3–6 times greater, with obvious geographical patterning (Figure 3). Within A. strepitans, haplotypes from southern SI differed by up to 0.04 s/s (0.024 p distance) from haplotypes in central NZ. The lone N. sericea haplotype from southwestern NI differed from other N. sericea haplotypes by 0.038 s/s in COI (0.025 p distance). N. sericea haplotypes along the eastern coast formed geographically coherent sister-clades, and haplotypes from the northern NI were nearly genetically identical (Figure 3). Otherwise, the phylogenetic relationships between regional haplotype clusters were poorly resolved in both N. sericea and A. strepitans.
The comparatively limited EF-1α dataset mirrored some intraspecific patterns observed in COI. EF-1α sequences for A. zelandica and A. cingulata were nearly identical within species, as for COI. Within A. strepitans, the central Otago sequence was divergent from the rest, and the Wellington (southern NI) and Marlborough (northern SI) samples were sister to one another. The N. sericea EF-1α sequences yielded the same three-way split observed between southwestern, eastern, and northern NI.
Intraspecific relationships observed in the calmodulin dataset generally did not match those found in COI and EF-1α. Notably, no sequence divergence was observed between the four N. sericea specimens even though these individuals were highly divergent at the other two loci. Two divergent haplotypes (mostly differing at a gene section containing tandem repeats) were found for A. cingulata, but little genetic divergence for COI or EF-1α was observed for individuals possessing those haplotypes. Differences in phylogenetic signal between calmodulin and EF-1α/mtDNA have been detected in other NZ cicada species, with calmodulin divergence rates much higher for some species than for their congeners .
The GMYC algorithm identified independent lineages within each of the four described species (Additional file 1: Figure S1) – six in Amphipsalta zelandica, two in A. cingulata, seven in A. strepitans, and four in Notopsalta sericea; the null hypothesis was rejected by a likelihood-ratio test (null likelihood = 238.2016; GMYC likelihood = 246.8606; P < 0.006). However, results were dependent on the dataset used. When A. zelandica and A. cingulata haplotypes were removed to see if their low intraspecific divergence was pulling the between/within-species threshold closer to the present, the GMYC algorithm diagnosed nearly every haplotype of the remaining species as an independent population lineage, although the results were only marginally significant (null likelihood = 141.2776, GMYC likelihood = 145.5174, P < 0.037). AMOVA results (Table 2) indicated considerable geographic-genetic structure, with greater than 30% of the genetic variance explained at the regional level in each species. Amphipsalta strepitans and N. sericea exhibited the highest levels of genetic variance between regions.
The relationships between corrected (patristic) and uncorrected mtDNA distances for Amphipsalta-Notopsalta (this study) and for Kikihia COI + COII sequences  were similar only at low genetic distances (Figure 4). For both datasets, this relationship was curved upward only slightly at low uncorrected genetic distances. At greater distances, the relationships diverged. Despite the fact that maximum uncorrected divergence levels were similar for the two datasets (ca. 8%), the maximum patristic distances were 0.22 s/s in Kikihia and 0.31 s/s in Amphipsalta-Notopsalta (Figure 4). The Amphipsalta-Notopsalta curve was notably ragged at high genetic distances, with substantially different patristic distances reconstructed for some taxa with uncorrected distances greater than 6%.
The BEAST analyses under different tree models showed a strong effect of the prior on interspecific divergence times, but comparatively little effect on intraspecific haplotype divergences and species’ MRCA ages (Table 3). The birth-death model resulted in a mean root age of 24.45 Ma, compared to only 4.82 Ma for the Yule and approximately 17 Ma for each of the two coalescent models. Intraspecific divergences were dated to the Pleistocene in all cases. Analyses run with no data in order to estimate the prior returned divergence times close to zero for all nodes.
While the estimated marginal likelihoods were similar for the birth-death, constant-size coalescent, and expansion-growth coalescent models, the 2 ln Bayes factor score indicated support for the birth-death model (Table 4). Under this model, the Amphipsalta + Notopsalta ancestor diverged from the outgroup approximately 24 Ma, and Notopsalta split from the Amphipsalta common ancestor at around 12.5 Ma. The remaining interspecific divergences were dated to ca. 9 Ma (A. strepitans) and 7 Ma (A. cingulata – A.zelandica split) (Figure 5). The earliest splits within A. strepitans and N. sericea dated to the mid-Pleistocene (ca. 1.2 Ma and 1.4 Ma, respectively), while splits within A. zelandica and A. cingulata dated to the Late Pleistocene (mean ca. 0.4 and 0.2 Ma, respectively). Credible intervals for the intraspecific coalescences were similar in magnitude (relative to the mean estimates) to those observed for the interspecific splits (Table 3). The meanrate estimate (0.012 s/s), reflecting the mean clock rate across the tree, remained close to the fixed ucld.mean rate, as expected because no calibrations other than the fixed values for ucld.mean and ucld.stdev were used in the analysis. Because we fixed ucld.stdev, we could not use this statistic to test for clocklike evolution, but because clocklike data can be accommodated by a relaxed clock model our approach is conservative.
The maximum clade compatibility tree from the BEAST analysis (Figure 5) found different intraspecific root positions for N. sericea and A. strepitans, compared to the ML results. As a result, some geographically coherent haplotype clades that appeared paraphyletic in the ML tree were found to be monophyletic with strong support (> 90% posterior probability) in the BEAST analysis.
Bioclimatic models of the distribution of Amphipsalta zelandica yielded an average AUC of 0.806 (s.d. 0.036) in the 10-fold cross-validation runs and performed significantly better than random (P < 0.05) in the binomial omission tests across all thresholds and runs. Heuristic estimation of relative contributions of the environmental variables indicated that annual temperature (38.9%) and February temperature (20.5%) were the top predictors of the distribution of A. zelandica. Effects of clamping during LGM projection were minor and restricted to montane areas. Under current climate conditions, A. zelandica was projected to be widely distributed throughout NI and northern SI in lowland regions (Figure 6). During the LGM, most of NZ was projected as less suitable for A. zelandica, with the exception of the Coromandel Peninsula and East Cape in northern NI and Karamea, Central Nelson and Kaikoura in northern SI (Figure 6).
Models for the distribution of A. strepitans also performed significantly better than random in the binomial omission tests in all but one run and yielded an average AUC of 0.940 (s.d. 0.054) (note that AUC values tend to be higher for geographically restricted taxa). The top predictor of the distribution of A. strepitans was October vapor pressure deficit (87.1%), with February temperature a distant second (4.4%). Effects from clamping during LGM projection were largely restricted to the central Southern Alps. The species’ projected distribution centered on the northeastern SI, particularly the coastal Kaikoura Ranges (Figure 6), under both climate scenarios, and the projected LGM distribution was slightly larger (more area with suitability >0.5) than that for current conditions. Notably, the model failed to capture the southern extent of the species’ range – several sampled populations fall within the region of 0-10% probability predicted by Maxent. As this may have been due to the high number of localities from the northern SI, resulting in over-fitting of regional conditions, input localities for A. strepitans were thinned to a minimum distance of 5 km apart (removing 15 localities from the northern SI) and modeled again as above. The resulting models were no better at capturing the southern extent of the species’ distribution (data not shown), which is represented by only 5 localities south of central Canterbury.
Results for Notopsalta sericea were also significantly better than random, with only one threshold for one run yielding P > 0.05 in the binomial omission tests, and an average AUC of 0.874 (s.d. 0.031). Top climatic predictors were February temperature (55.9%) and minimum temperature (23.8%), and no effects from clamping were detected during projection. Under modern conditions, northern and coastal NI were projected as climatically suitable, but none of NZ was projected to contain suitable conditions during the LGM (Figure 6).
Models for A. cingulata yielded similar geographic projections, with much of lowland NI projected as climatically suitable under modern conditions, and no refugia projected for the LGM (Figure 6). Results from the binomial omission tests were significantly different from random except for one data partition which yielded multiple thresholds with P > 0.5. The average AUC across ten runs was 0.881 (s.d. 0.043), and annual temperature (73.2%) and minimum temperature (11.0%) were the top predictors of the distribution of A. cingulata.
When model calibration for N. sericea and A. cingulata was restricted to environmental conditions from NI, performance markedly declined, yielding average AUCs of 0.732 (s.d. 0.089) and 0.727 (s.d. 0.102), respectively, and multiple binomial omission tests for multiple runs yielded P > 0.05. Areas of uncertainty (clamping) appeared when projecting onto the full LGM surface, and once these areas were removed, no likely refugia were detected, although some regions had climatic suitability above zero (Figure 6).
Insect molecular-clock studies often apply a mtDNA calibration of 2.3% pairwise distance per my, or 0.0115 substitutions/site/my/lineage, derived by Brower . However, two estimates used by Brower were revised downward during re-analysis [18-20], and the remaining data were derived from rRNA, DNA-DNA hybridization, or RFLP datasets, so their applicability to mtDNA protein coding genes such as COI is unclear. In addition, none of the data used in the Brower study were corrected for multiple hits with data partitioning or the use of gamma-distributed rates, as most data are today [21-24]. If studies that calibrate molecular clocks do not correct for multiple hits as thoroughly as the analyses that employ them, the date estimates will be biased .
The Bayesian relaxed-clock approach in BEAST allows the investigator to conservatively accommodate a range of plausible COI clock values with a single prior probability distribution in each analysis. We calibrated our analyses with fixed values for the mean branch rate and the standard deviation of that mean across the tree (ucld.mean and ucld.stdev parameters), set such that the 95% credible intervals included the published rates cited in Table 5.
The divergence-time and ML analyses in this study may have suffered from poor branch-length estimation (with “branch lengths” estimated as rate x time products in the case of BEAST), perhaps related to the limited number of deep splits in the tree. Results from the BEAST analyses, specifically the ages of the interspecific splits, were strongly influenced by the choice of tree prior (Table 3). Plots of corrected vs. patristic distances in the Amphipsalta-Notopsalta and published Kikihia datasets showed that corrected distances were considerably greater for Amphipsalta-Notopsalta at the deepest levels examined, although both plots initially followed similar trajectories (Figure 4). Although the two datasets may have evolved under different mechanisms, the magnitude of the difference and the ragged nature of the Amphipsalta-Notopsalta plot are causes for concern. Further examination of the relationships between tree shape, dataset information, and branch lengths in maximum-likelihood and divergence-time analyses would likely prove useful. Notably, the confidence intervals on the intraspecific data estimates from BEAST were much narrower and largely unaffected by the tree prior.
Unlike the NZ Kikihia-Maoricicada-Rhodopsalta cicada radiation, the Amphipsalta-Notopsalta radiation contains no described species with Pleistocene origins. An earlier genus-level phylogenetic analysis including A. cingulata and N. sericea demonstrated a high level of genetic divergence between these species, consistent with a late Miocene to early Pliocene split (ca. 5–6 Ma) . We find here that all interspecific divergences in the clade predate the Pleistocene (i.e., > 2.6 Ma), although these dates are uncertain. In addition, the low number of species in the AmphipsaltaNotopsalta radiation is not due to more recent diversification, since the earliest divergences of both clades appear to have occurred in the mid-Miocene [2,5]. (Due to extensive collecting over multiple decades, undiscovered extant species are unlikely.)
Although species-level diversification apparently preceded the Pleistocene, the genetic data suggest that several independently evolving, morphologically cryptic populations have formed during that period. All four species contain mtDNA phylogeographic structure, i.e., unique haplotypes or haplotype clusters occupying exclusive geographic ranges, with uncorrected genetic distances between clades approaching 2.5%. The nuclear data show some concordant patterns, but they do not confirm all of the mtDNA clades, and there is conflict in some cases (e.g., all four N. sericea individuals have the same calmodulin allele, despite considerable mtDNA structure). Nuclear genes tend to evolve much more slowly than mtDNA, so considerable structure can appear in mitochondrial genes before concurrent changes occur in nDNA. Furthermore, the larger effective population sizes of nuclear genes mean that more time is needed for lineage sorting to become complete , so the lack of detailed structure in calmodulin and EF-1α does not necessarily mean that the mtDNA clades are freely exchanging such genes. However, nuclear genes can also reveal sex-specific gene flow patterns that a uniparental marker like mtDNA can miss. The phylogeographic evidence of intraspecific diversification helps to alleviate the potential concern that, by sampling extensively “within species” for the BEAST analyses, we may be overestimating intraspecific sequence divergence and therefore the MRCA ages of the four species .
In NZ Kikihia and Maoricicada[16,28-30], mtDNA phylogeography and song patterns indicate Pleistocene-age allopatric diversification within currently recognized species. Populations with greater than 2.5% uncorrected mtDNA divergence frequently, although not always, exhibit discrete differences in song [16,29-31] – and in some cases conflicting patterns in EF-1α. The largest intraspecific mtDNA genetic distances within A. strepitans and N. sericea approach this threshold. We have not observed regional differences in the songs of Amphipsalta and Notopsalta during field collection and observation, but detailed examination might uncover minor variation concordant with the mtDNA tree.
The GMYC analysis (Additional file 1: Figure S1), which was intended to identify the shift from between- to within-population branching patterns , yielded results consistent with the inference that some of the geographically coherent mtDNA phylogroups are independently evolving populations. However, the large number of clades inferred for A. zelandica is surprising give the low sequence divergence within that species, and the GMYC results changed substantially when only N. sericea and A. strepitans haplotypes were examined. It is possible that the marginal statistical significance observed in the latter test is a sign of too little data for effective estimation of branching-rate shifts despite the likely existence of several independently evolving populations within these taxa.
If the mtDNA clades represent independently evolving subspecific populations, then at least 2–5 allopatric population-lineages per species have formed in AmphipsaltaNotopsalta during the Pleistocene, a rapid rate compared to the Late Miocene and Pliocene. This contrasts with the results from Kikihia, where the rate of population-lineage formation has remained approximately constant since the early Pliocene .
The low level of intraspecific genetic divergence characterizing Amphipsalta cingulata and A. zelandica is consistent with their dependence on temperate forest habitat. NZ forests progressively declined in extent throughout the Pleistocene, especially during the last million years. During the LGM (~22 ka), the principal forest refugia were located in northern NI and/or northwestern SI [33-35], with smaller fragments located farther south . Roughly similar regions were predicted for A. zelandica refugia (A. cingulata refugia were not successfully reconstructed). Both A. cingulata and A. zelandica contain haplotype clusters consistent with survival during the LGM in as few as two locations, one more northern and one more southern . However, the phylogeographic structure observed within A. zelandica is insufficiently detailed to confirm the ENM predictions of two major refugia in NI and 2–3 in SI, although the genetic data are consistent with the absence of refugia in central or southern SI.
The lack of structure within A. zelandica is remarkable when compared with phylogenetic patterns observed in other forest species believed to have survived in SI [31,38-40]. While NZ forest taxa often show evidence of geographic restriction during the LGM, different species exhibit phylogeographic variation that likely relates to differences in life-history parameters or habitat specialization. For example, forest taxa including beetles [39-41] and a fern  appear to have survived glacial cold phases in small forest fragments in southern NZ and retained considerable phylogeographic structure there. In contrast, the only other widespread NZ forest cicada, Kikihia subalpina, retains considerable geographic structure on NI (where it lives mainly in subalpine scrub), but limited structure on SI where it inhabits lowland forest . Greater Pleistocene stability of NI habitats has been linked to persistence of beetle species (as inferred from fossils; ). It is possible that, for cicadas, additional factors are needed to explain why the forest taxa have been unable to survive in southern NZ during colder climate phases (see also ).
Because the intraspecific lineages within A. zelandica and A. cingulata were inferred to coalesce around 370 ky and 160 ky, respectively, both species were likely limited to a single refugium during the Late Pleistocene. Some support for this idea is observed in climate proxies that suggest, around 430 ky, a glacial cold period that may have been more severe than the LGM [44,45]. Coalescence of regional clades around 430 ky was also observed in Maoricicada campbelli. This hypothesis could be tested more effectively by multi-gene datasets allowing more sensitive estimates of divergence times.
Dispersal ability may also have influenced population genetic structure in the AmphipsaltaNotopsalta species, contributing to lower genetic differentiation in the forest taxa A. zelandica and A. cingulata. Lineage-specific differences in traits affecting dispersal and gene flow should influence relative rates of allopatric divergence, especially in neutral genetic characters [46,47]. The two forest taxa are the largest-bodied cicadas in New Zealand (see photo inset in Figure 5), and they are likely the strongest fliers, suggesting that greater dispersal ability leads to greater gene flow among populations and reduced rates of allopatric differentiation in these species. However, the mtDNA data show that dispersal in A. cingulata and A. zelandica has not completely prevented the formation of intraspecific phylogeographic structure (Table 2, Figure 3 and Figure 5). All cicadas spend multiple years as juveniles underground, a life-history feature that must reduce population vagility compared to similarly sized insects .
In Notopsalta sericea and Amphipsalta strepitans, more intraspecific lineages are evident than in the two forest taxa, and multiple lineages apparently persisted through the hypothesized Pleistocene minimum of 430 ky. In both cases, the credible intervals on mtDNA coalescence include the period from about 0.6-2.0 Ma (mean 1.4 and 1.2 Ma, respectively) (Table 3).
Multiple geological and climatic changes are known to have occurred around 1 Ma that could have affected diversification of these open-habitat species . Pleistocene climates became progressively colder and drier, suggesting the continued replacement of forests by shrub and grassland habitats. Large sections of NI and northern SI began a period of accelerated uplift that has continued to the present [49,50], increasing rain shadow effects across eastern provinces and expanding the dry slope habitats used by A. strepitans and N. sericea. The puzzling mtDNA homogeneity of N. sericea populations across the northern third of NI, a reversal of the “northern richness, southern purity” pattern frequently observed in NZ [51-55], could be explained by very recent acceleration of uplift in that area  leading to northward dispersal from source populations in central NI.
Unlike A. zelandica and other forest insects whose LGM distributions have been modeled [39,40,52,55], for A. strepitans the hypothesized LGM range was not smaller than the present-day range (Figure 6). A higher degree of phylogeographic structuring relative to the other cicadas was also detected, consistent with survival of multiple distinct populations over a longer period, as has been detected in other non-forest invertebrates (e.g., [46,57-59]). A. strepitans was the only cicada for which temperature explained less than 15% of model fit, suggesting its wider LGM distribution may have been related to success in cooler temperatures, although tolerance of aridity could also be involved. Greater success in drier climates was offered to explain a larger LGM distribution implied by microsatellite data for the NZ tree species Pseudopanax ferox. The ENM failed to capture the southern extent of A. strepitans’ current distribution, possibly because of the low number of localities from South Canterbury and Otago relative to the well-sampled northeastern SI, limiting the inferences we can make about the LGM projection. However, the low sample numbers in the south likely indicate genuine scarcity in this region, and the discovery of additional southern localities would likely expand, rather than contract, the projected climatic distribution of this species.
No LGM refugia were projected for N. sericea or A. cingulata, despite apparently successful model performance, implying that no counterpart of either species’ current realized niche existed during the LGM. Yet genetic patterns suggest that both species survived on NI during that period, and N. sericea contains considerable mtDNA phylogeographic structure on NI. Three hypotheses (which are not mutually exclusive) can explain these surprising results: 1) the species have only recently occupied their current climatic niche (niche evolution), 2) they survived in isolated microrefugia undetectable using these methods, or 3) their current realized niches inadequately represent the fundamental niche of each species (environmental disequilibrium). The first hypothesis is less likely; existing evidence for insects, while largely based on beetles, offers little evidence of post-LGM shifts in climate exploitation [61,62].
The second and third hypotheses are both plausible, but difficult to disentangle. Amphipsalta cingulata and Notopsalta sericea survived somewhere in NZ, and it is likely that local habitat attributes created small refugia undetectable using our elevation-scaled meteorological data . Past climates may have had different combinations of environmental variables than exist today, yielding predictions of past ranges that are under- or overestimates [31,64]. In addition, ENMs as applied here are a representation of each species’ climatic niche, but if a species is restricted from exploiting its full climatic niche due to other factors, projections onto past climates may exclude potentially habitable areas . Several cicada species are present in the southernmost NI but are absent from SI (see maps in ), indicating that Cook Strait is an important dispersal barrier even for flighted species , and the absence of A. cingulata and N. sericea from cooler and wetter conditions in SI would result in underrepresentation of suitable LGM refugia. Excluding these potentially habitable, unoccupied regions from the training data reduces the risk of over-fitting but increases model uncertainty (clamping; ) and thus, transferability of the model , and for both species, restricting the training data resulted in clamping (extensively so for N. sericea) where none was previously detected. Interestingly, neither species has extended its range into the cooler mid- or high elevations of NI, and restricting the training data failed to yield additional refugia, suggesting that spurious over-fitting of the climate data was not the key driver of this absence, although it cannot be conclusively ruled out. Projections of ENMs onto past climates must be interpreted with caution, particularly where their congruence with external data (e.g., DNA) is low.
This study and others (e.g., ) show that closely related and regionally co-occurring organisms can experience dramatically different diversification rates over a substantial period (here, mid-Miocene to present), a conclusion presaged by earlier literature on species distributions (e.g., ). Because the two NZ cicada lineages colonized the landscape independently, the difference in the number of extant species is not simply a side-effect of the geography of initial diversification, as is possible with sister-clades . Apparently, many physiological, ecological, and historical factors interact to influence diversification. In the case of the two NZ cicada radiations, we speculate that differences in climatic tolerances/preferences of the two ancestors could have persistently influenced the geographic ranges of the two radiating lineages. Climatic refugia were not detected for two of the four species, suggesting that Pleistocene survival and lineage diversification were moderated by factors other than broad regional climatic conditions. This study also adds to a growing number of phylogeographic analyses implicating mountain-building and Pleistocene habitat fragmentation as drivers of diversification in NZ insects.
Amphipsalta and Notopsalta specimens (typically male) were collected during surveys from 1993–2010. The species Kobonga apicata (Ashton) (see ), from central and western Australia, was selected as an outgroup based on phylogenetic analyses at the tribe level (unpublished data). Tissue collections were sought from sites distributed as uniformly as possible within each species' range. When a species was detected but not collected, the presence of its diagnostic song was noted and digitally recorded if possible. Localities used for ecological niche modeling were based on collection and song data from this study and historical records  (Figure 2; see Additional file 2: Appendix for details).
Genomic DNA was extracted from 1–2 legs using the Qiagen DNeasy Tissue Kit and protocol (Qiagen, Valencia, California, USA), except that the digestion was conducted over 3–18 hours at 54 °C. Approximately 1350 base pairs of the mitochondrial cytochrome oxidase I (COI) gene were amplified in one of two ways: (1) in two sections; the 5’ (barcoding) section with primers C1-J-1490 + C1-N-2198 , annealing temperature 50°C, and the 3’ section with primers C1-J-2195 + TL2-N-3014 , annealing temperature 56°C; (2) in one piece using C1-J-1490 + TL2-N-3014, annealing temperature 45°C. Amplified products were cleaned using the Clontech Extract II Kit (Clontech, Mountain View, CA, USA) or ExoSAP-IT (USB Corp., Cleveland, OH). For examination of the species-level AmphipsaltaNotopsalta relationships, a subset (see Additional file 2: Appendix) of the individuals sampled were sequenced for two nuclear genes, Calmodulin (763 bp) and elongation factor-1 alpha (EF-1α; ca. 1400 bp). Calmodulin was amplified using the primers Cal60For  and Cal2Rev (UBC insect primer kit). A touchdown cycle was used with the following parameters: (1) a 94°C hold for 2 min.; (2) a touchdown PCR sequence of 94°C for 45 sec., 60-50°C for 45 sec., and 72°C for 1 min. – decreasing by 1°C every cycle and repeated for 10 cycles; (3) a steady PCR sequence of 94°C for 45 sec., 50°C for 45 sec., and 72°C for 1 min., repeated for 25 cycles; and (4) a 72°C hold for 10 min. For A. cingulata, internal calmodulin primers were created from the sequence of the other cicadas. EF-1α was amplified in two overlapping sections using the primers EF1-F001-cicada + EF1-R752 , annealing temperature 53°C, and EF1-PA-f650ambig  + EF1-N-1419 , annealing temperature 58°C. Cycle sequencing was conducted using the Applied Biosystems Big Dye Terminator v1.1 cycle sequencing kit at 1/8- to 1/4-scale reaction volume, and the product was cleaned by Sephadex filtration (GE Healthcare, Piscataway, NJ, USA) and visualized on an Applied Biosystems ABI 3100 capillary sequencer. Sequences were analyzed using ABI Prism Sequencing Analysis software v3.7 (Applied Biosystems), edited in Sequencher v3.1 (Gene Codes Corp., Ann Arbor, MI), aligned in MacClade  and checked by eye.
Individual gene-trees were estimated using partitioned maximum-likelihood (ML) in Garli version 2.0 , using datasets that contained all unique haplotypes. Each nuclear gene was modeled using a single partition because of the low number of parsimony-informative sites available for estimating parameters (see Results); substitution models for these genes were selected using Modeltest v3.7 . For the mtDNA data, alternative combinations of partition schemes and partition-specific substitution models, based on the three codon-position categories, were tested using the “greedy” algorithm of the program PartitionFinder v1.0  and the BIC criterion, with the branch lengths of alternative partitions linked and with the software set to evaluate the full substitution model set. Default settings were used for Garli analyses, except that the number of generations required for termination (genthreshfortopoterm) was increased to 40,000 to improve the search. ML analyses in Garli offered estimates of branch lengths that were not influenced by Bayesian priors [78,79]. The number of discrete gamma categories was set to 8 to improve resolution of site rates over the default setting of 4 categories. For each analysis, 100 bootstrap pseudoreplicates were conducted in Garli, and these were summarized on the ML tree with the program SumTrees v3.3.1, part of the DendroPy package .
Because “saturation” of genetic data by multiple hits has been observed to influence divergence time estimates [81,82], we constructed plots of uncorrected genetic distances calculated in Mesquite v. 2.75  vs. ML model-corrected patristic (tree-based) distances for the AmphipsaltaNotopsalta COI dataset and a published, combined COI + COII ingroup dataset of about the same size (bp) from Kikihia containing 30 species. Maximum uncorrected genetic distances from the Kikihia dataset are similar to those in AmphipsaltaNotopsalta.
To place the Amphipsalta-Notopsalta tree in an approximate time frame, Bayesian relaxed-clock phylogenetic analyses of the full COI dataset, with identical haplotypes and the outgroup included, were conducted using BEAST v1.6.1 . Because no fossil or geological calibrations are available for dating these cicadas, the analysis was calibrated using COI molecular clock estimates from the literature. While many insect studies have used the 0.0115 substitutions/site/my/lineage clock rate estimated by Brower , more recent studies have suggested both lower (e.g., [19,85]) and higher rates  (Table 5), in part because of the use of highly partitioned sequence-evolution models  that tend to reconstruct more change along branches. In each analysis, the mean and standard deviation of the relaxed clock were fixed to reflect the literature estimates from Table 5 by the following settings: (1) we set ucld.mean (the mean of the lognormal prior distribution on branch rates) to 0.0115 substitutions/site/my, the Brower  rate, (2) we set ucld.stdev, the standard deviation of the lognormal prior on branch rates, to 0.3, so that the 95% confidence interval of the lognormal distribution included rates from 0.006 to 0.0188 substitutions/site/my, and (3) we deactivated the operators for ucld.mean and ucld.stdev, so that the software did not attempt to estimate these parameters. We suggest that the approach of combining all empirical knowledge into a single prior is superior to the use of separate analyses under alternative rates an approach that was used in earlier work on Kikihia.
BEAST analyses of the COI dataset used the same three-partition scheme as the ML analysis. The Amphipsalta sequences were constrained to form a monophyletic clade, as supported by the ML analysis (see Results), because initial runs of some models suggested a tendency for BEAST to place the ingroup root between clades containing A. cingulata + A. zelandica and N. sericea + A. strepitans. Model parameters and partition rate multipliers were estimated separately for each partition. Four tree priors were tested in separate analyses and compared using the Bayes factor scores calculated in Tracer v1.5 : Yule, birth-death, a constant-size coalescent, and an expansion-growth coalescent. The criterion 2 ln Bayes factor > 10  was used to decide if a given model was superior. Models were run to 40 million generations, and stationarity and adequate sample sizes (i.e., > 200) were confirmed using Tracer  and by ensuring that multiple runs found the same solution. Other priors were left at the default values specified by BEAUti v 1.6.1 (provided with the BEAST package). Statistically improper priors, such as intervals extending to infinity, were converted to proper priors with a large, finite upper bound. Prior distributions were estimated by running the analysis with no data.
Because phylogeographic patterns in the mtDNA dataset suggested ongoing intraspecific allopatric diversification, we tested for the threshold between interspecific and intraspecific lineage dynamics using the generalized mixed Yule-coalescent (GMYC) model of Pons et al.  as implemented in the R package SPLITS [88,89]. If the samples of each species are drawn from a single panmictic population, the method should identify a branching-rate shift at the base of each species caused by the change from coalescent (intraspecific) to birth-death (interspecific) dynamics. A tree containing only unique haplotypes was analyzed in BEAST to obtain a chronogram under the same model used in the divergence-time analysis above. The chronogram was tested in the GMYC analysis using an interval of 0,10 and a single-threshold model.
Geographic structure in the mtDNA dataset was also tested using an analysis of molecular variance (AMOVA) as implemented in Arlequin v 3.1 . Populations were combined into groups according to a six-region scheme defined by Cook Strait and latitude lines -37° S, -39° S, -43° S, and -45° S.
Combining ecological niche models with paleoclimatic data to hindcast species ranges offers a means of testing refugia hypotheses developed from genetic information. For comparison with intraspecific phylogeographic patterns, the current and LGM distributions of each species were estimated using ecological niche models. Climate data used to construct the ENMs included mean annual rainfall (mm), mean February rainfall (mm), mean annual solar radiation (kJ/day/m2), mean annual temperature (°C), mean February temperature (°C), minimum temperature of the coldest month (°C, July), and October vapor pressure deficit (kPa), all at 100 m resolution . These layers, derived from meteorological data (1950–1980, ), were fitted to the LGM based on estimates of temperature depression from marine isotope stages and estimates of LGM topography based on depression of the sea level by 120 m (J.R. Leathwick, unpublished data, see ).
Ecological niche models were generated with Maxent 3.3.1 [92,93]. Models were built using 341 A. zelandica, 62 A. strepitans, 86 A. cingulata and 87 Notopsalta sericea localities. Maximum iterations were increased as necessary to allow the algorithm to converge, with default settings used otherwise. To ensure consistency of model predictions among repeated runs, we performed a 10-fold cross-validation, in which a different 90% of localities were used to train the model and 10% were used to test it for each of 10 runs, such that each locality was used to test the model once. Estimated models for each species were projected onto LGM climate surfaces for all 10 runs, using the "fade by clamping" setting for output grids, to visually inspect for differences in regions projected as suitable habitat between replicate runs. "Clamping" refers to the handling of LGM grid sections with climatic values outside the range observed during training – Maxent sets these to the corresponding maximum or minimum observed training value. The "fade by clamping" method attempts to correct for the effect of this uncertainty on the projected distribution by reducing the estimated climatic suitability proportional to the level of clamping. Clamping grids were also inspected for differences in model output under different data partitions. The final geographical projections represent the mean point-wise prediction over ten model runs. Model performance was evaluated using threshold-dependent binomial omission tests and the Area Under the (Receiver Operating Characteristic) Curve (AUC) calculated by Maxent.
Regions which contain suitable climate conditions for a species may be unoccupied due to other factors, such as dispersal barriers . Inclusion of these regions during model calibration can result in overfitting to conditions found near the collection localities . When hindcasting to LGM climate conditions indicated no refugia for the two taxa limited to NI (A. cingulata and N. sericea; see below), we re-ran these analyses using NI climate data only for training and testing to examine whether the model for these species was being overfitted by the inclusion of SI data. The model was then projected onto LGM surfaces for all of NZ. While Cook Strait is not likely to be an absolute dispersal barrier to cicadas, it could delay colonization of SI by species restricted to northern NI refugia.
All authors declare that no competing financial or non-financial interests exist.
DM collected specimens, conceived the study, conducted phylogenetic and phylogeographic analyses, and drafted the manuscript. KH collected specimens, conceived the study, conducted genetic sequencing, and drafted the manuscript. KM conceived the study, conducted the ENM analyses and drafted the manuscript. CC conducted genetic sequencing and assisted with phylogenetic and phylogeographic analyses. TB collected specimens, assisted with the analyses, and drafted the manuscript. CS collected specimens, conceived and coordinated the study and its funding, assisted with the analyses, and drafted the manuscript. All authors read and approved the final manuscript.
Figure S1. Population-lineages diagnosed within the four described species using the generalized mixed Yule-coalescent algorithm. Connected red branches indicate haplotypes contained within a single diagnosed population. Results varied considerably depending on the particular tree used (e.g., all sequences, haplotypes only, A. strepitans-N. sericea only) but multiple lineages were always detected within the four described species.
Additional file 2
Appendix Specimen locality data. Asterisked latitude indicates that GPS was estimated from a printed map. Elevations were recorded in meters.
Click here for additional data file (1471-2148-12-177-S2.doc)
The authors thank Craig Briggs, Robbie Price, Gary Barker, and David Nogués-Bravo for assistance with the climatic data and ecological niche models. Field work assistance and local knowledge were contributed by Peter Arensburger, Steve Chiswell, John Cooley, John Dugdale, George Gibbs, Kees Green, Chris Owen, David Lane, Ricardo Palma, Dan Vanderpool, and Beth Wade. NZ Dept. of Conservation (Ian Millar, Bev Freer and numerous field staff), the National Museum of NZ (Te Papa) and the National Insect Collections contributed permits and expertise. Anne Firnan, Peter Hill, and Barbara Hill allowed access to private land. The shaded elevation map was provided by Roger Smith of http://www.geographx.co.nz. KAM acknowledges the Danish National Research Foundation for support to the Center for Macroecology, Evolution and Climate. TRB and KAM were funded by the Royal Society of New Zealand Marsden Fund (LCR0502) and the Ministry of Science and Innovation through the Defining New Zealand’s Land Biota programme. Computational resources were provided by Bioportal at the University of Oslo (http://www.bioportal.uio.no). This material is based upon work partially supported by the National Science Foundation under Grants No. DEB 04–22386, DEB 05–29679, DEB 07–20664, and DEB 09–55849. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the NSF.
|Marshall DC,Slon K,Cooley JR,Hill KBR,Simon C,Steady Plio-Pleistocene diversification and a 2-million-year sympatry threshold in a New Zealand cicada radiationMol Phylogenet EvolYear: 2008481054106610.1016/j.ympev.2008.05.00718590969|
|Buckley TR,Simon C,Evolutionary radiation of the cicada genus Maoricicada Dugdale (Hemiptera: Cicadoidea) and the origins of the New Zealand alpine biotaBiol J Linn SocYear: 20079141943510.1111/j.1095-8312.2007.00807.x|
|Simon C,Using New Zealand examples to teach Darwin's "Origin of Species:" Lessons from molecular phylogenetic studies of cicadasNZ Sci RevYear: 200966102112|
|Buckley TR,Arensburger P,Simon C,Chambers GK,Combined data, Bayesian phylogenetics, and the origin of the New Zealand Cicada generaSyst BiolYear: 20025141810.1080/10635150275347584411943089|
|Arensburger P,Buckley TR,Simon C,Moulds M,Holsinger KE,Biogeography and phylogeny of the New Zealand cicada genera (Hemiptera: Cicadidae) based on nuclear and mitochondrial DNA dataJ BiogeogrYear: 20043155756910.1046/j.1365-2699.2003.01012.x|
|Fleming CA,A new species of cicada from rock fans in southern Wellington, with a review of three species with similar songs and habitatNZ J SciYear: 197114443479|
|Fleming CA,Adaptive radiation in New Zealand cicadasProc Am Phil SocYear: 1975119298306|
|Fleming CA,The cicada genus Kikihia (Dugdale) (Hemiptera, Homoptera). Part 1. The New Zealand green foliage cicadasNat Mus NZ RecYear: 19842191206|
|Dugdale JS,Fleming CA,New Zealand cicadas of the genus Maoricicada (Homoptera: Tibicinidae)NZ J ZoolYear: 19785295340|
|Arensburger P,Simon C,Holsinger K,Evolution and phylogeny of the New Zealand cicada genus Kikihia Dugdale (Homoptera: Auchenorrhyncha: Cicadidae) with special reference to the origin of the Kermadec and Norfolk Islands’ speciesJ BiogeogrYear: 2004311769178310.1111/j.1365-2699.2004.01098.x|
|Dugdale JS,Fleming CA,Two New Zealand cicadas collected on Cooks' Endeavour voyage, with description of a new genusNZ JYear: 196912929957|
|Hudson GV,Fragments of New Zealand EntomologyYear: 1950Wellington, New Zealand: Ferguson and Osborn|
|Myers JG,Insect singers: a natural history of the cicadasYear: 1929London: George Routledge and Sons|
|Cooper RA,Millener PR,The New Zealand biota: Historical background and new researchTr Ecol EvolYear: 1993842943310.1016/0169-5347(93)90004-9|
|Wallis GP,Trewick SA,New Zealand phylogeography: evolution on a small continentMol EcolYear: 2009183548358010.1111/j.1365-294X.2009.04294.x19674312|
|Buckley TR,Cordeiro M,Marshall DC,Simon C,Differentiating between hypotheses of lineage sorting and introgression in New Zealand alpine cicadas (Maoricicada Dugdale)Syst BiolYear: 20065541142510.1080/1063515060069728316684720|
|Brower AVZ,Rapid morphological radiation and convergence among races of the butterfly Heliconius erato inferred from patterns of mitochondrial DNA evolutionProc Natl Acad Sci USAYear: 1994916491649510.1073/pnas.91.14.64918022810|
|Knowlton N,Weight LA,Solorzano LA,Mills DK,Bermingham E,Divergence in proteins, mitochondrial DNA, and reproductive compatibility across the Isthmus of PanamaScienceYear: 19932601629163210.1126/science.85030078503007|
|Farrell BD,Evolutionary assembly of the milkweed fauna: cytochrome oxidase I and the age of Tetraopes beetlesMol Phylogenet EvolYear: 20011846747810.1006/mpev.2000.088811277638|
|Knowlton N,Weigt LA,New dates and new rates for divergence across the Isthmus of PanamaProc Roy Soc Biol Sci BYear: 19982652257226310.1098/rspb.1998.0568|
|Buckley TR,Simon C,Chambers GK,Exploring among-site rate variation models in a maximum likelihood framework using empirical data: effects of model assumptions on estimates of topology, branch lengths, and bootstrap supportSyst BiolYear: 200150678612116595|
|Sullivan J,Joyce P,Model selection and phylogeneticsAnnu Rev Ecol SystYear: 20053644546610.1146/annurev.ecolsys.36.102003.152633|
|Simon C,Buckley TR,Frati F,Stewart JB,Beckenbach AT,Incorporating molecular evolution into phylogenetic analysis, and a new compilation of conserved polymerase chain reaction primers for animal mitochondrial DNAAnnu Rev Ecol Evol SystYear: 20063754557910.1146/annurev.ecolsys.37.091305.110018|
|Venditti C,Meade A,Pagel M,Detecting the node-density artifact in phylogeny reconstructionSyst BiolYear: 20065563764310.1080/1063515060086556716969939|
|Papadopoulou A,Anastasiou I,Vogler AP,Revisiting the insect molecular clock: The mid-Aegean trench calibrationMol Biol EvolYear: 2010271659167210.1093/molbev/msq05120167609|
|Moore WS,Inferring phylogenies from mtDNA variation: mitochondrial-gene trees versus nuclear-gene treesEvolutionYear: 19954971872610.2307/2410325|
|Ho SYW,Saarma U,Barnett R,Haile J,Shapiro B,The effect of inappropriate calibration: three case studies in molecular ecologyPLoS OneYear: 20083e161510.1371/journal.pone.000161518286172|
|Buckley TR,Simon C,Chambers GK,Phylogeography of the New Zealand cicada Maoricicada campbelli based on mitochondrial DNA sequences: Ancient clades associated with Cenozoic environmental changeEvolutionYear: 2001551395140711525463|
|Hill KBR,Simon C,Marshall DC,Chambers GK,Surviving glacial ages within the biotic gap: phylogeography of the New Zealand cicada Maoricicada campbelliJ BiogeogryYear: 20093667569210.1111/j.1365-2699.2008.02036.x|
|Marshall DC,Hill KBR,Cooley JR,Simon C,Hybridization, mitochondrial DNA taxonomy, and prediction of the early stages of reproductive isolation: Lessons from New Zealand cicadas of the genus KikihiaSyst BiolYear: 20116048250210.1093/sysbio/syr01721471306|
|Marshall DC,Hill KBR,Fontaine KM,Buckley TR,Simon C,Glacial refugia in a maritime temperate climate: Cicada (Kikihia subalpina) mtDNA phylogeography in New ZealandMol EcolYear: 2009181995200910.1111/j.1365-294X.2009.04155.x19434813|
|Pons J,Barraclough TG,Gomez-Zurita J,Cardoso A,Duran DP,Hazell S,Kamoun S,Sumlin WD,Vogler AP,Sequence-based species delimitation for the DNA taxonomy of undescribed insectsSyst BiolYear: 20065559560910.1080/1063515060085201116967577|
|Marra MJ,Shulmeister J,Smith EGC,Reconstructing temperature during the Last Glacial Maximum from Lyndon Stream, South Island, New Zealand using beetle fossils and maximum likelihood envelopesQuat Sci RevYear: 2006251841184910.1016/j.quascirev.2006.01.016|
|Alloway BV,Lowe DJ,Barrell DJA,Newnham RM,Almond PC,Augustinius PC,Bertler NAN,Carter L,Litchfield NJ,McGlone MS,et al. Towards a climate event stratigraphy for New Zealand over the past 30,000 years (NZ-INTIMATE project)J Quat SciYear: 20072293510.1002/jqs.1079|
|Marra M,Leschen RAB,Late Quaternary paleoecology from fossil beetle communities in the Awatere Valley, South Island, New ZealandJ BiogeogrYear: 20043157158610.1046/j.1365-2699.2003.00998.x|
|McGlone MS,Newnham RM,Moar NT,The vegetation cover of New Zealand during the Last Glacial Maximum: do pollen records under-represent woody vegetation?Terra AustralisYear: 2010324968|
|Provan J,Bennett KD,Phylogeographic insights into cryptic glacial refugiaTr Ecol EvolYear: 20082356457110.1016/j.tree.2008.06.010|
|Boyer SL,Baker JM,Giribet G,Deep genetic divergences in Aoraki denticulata (Arachnida, Opiliones, Cyphophthalmi): a widespread 'mite harvestman' defies DNA taxonomyMol EcolYear: 2007164999501610.1111/j.1365-294X.2007.03555.x17944852|
|Marske KA,Leschen RAB,Barker GM,Buckley TR,Phylogeography and ecological niche modelling implicate coastal refugia and trans-alpine dispersal of a New Zealand fungus beetleMol EcolYear: 2009185126514210.1111/j.1365-294X.2009.04418.x19900173|
|Marske KA,Leschen RAB,Buckley TR,Reconciling phylogeography and ecological niche models for New Zealand beetles: Looking beyond glacial refugiaMol Phylogenet EvolYear: 2011598910210.1016/j.ympev.2011.01.00521262367|
|Marske KA,Leschen RAB,Buckley TR,Concerted versus independent evolution and the search for multiple refugia: comparative phylogeography of four forest beetlesEvolutionYear: 2011661862187722671552|
|Shepherd LD,Perrie LR,Brownsey PJ,Fire and ice: volcanic and glacial impacts on the phylogeography of the New Zealand forest fern Asplenium hookerianumMol EcolYear: 2007164536454910.1111/j.1365-294X.2007.03451.x17877716|
|Marra MJ,Leschen RAB,Persistence of New Zealand Quaternary beetlesN Z J Geol GeophysYear: 20115440341310.1080/00288306.2011.599399|
|Carter RM,A New Zealand climatic template back to c. 3.9 Ma: ODP Site 1119, Canterbury Bight, south-west Pacific Ocean, and its relationship to onland successionsJ Roy Soc N ZYear: 20053594210.1080/03014223.2005.9517776|
|Zachos J,Pagani M,Thomas E,Billups K,Trends, rhythms, and aberrations in global climate 65 Ma to presentScienceYear: 2001292689693|
|McCullough GA,Wallis GP,Waters JM,Do insects lose flight before they lose their wings? Population genetic structure in subalpine stonefliesMol EcolYear: 2009184073408710.1111/j.1365-294X.2009.04337.x19754508|
|Gillespie RG,Roderick GK,Arthropods on islands: colonization, speciation, and conservationAnnu Rev EntomolYear: 20024759563210.1146/annurev.ento.47.091201.14524411729086|
|de Boer AJ,Duffels JP,Historical biogeography of the cicadas of Wallacea, New Guinea and the West Pacific: a geotectonic explanationPalaeogeogr, Palaeoclimat, PalaeoecolYear: 199612415317710.1016/0031-0182(96)00007-7|
|Trewick SA,Bland KJ,Fire and slice: palaeogeography for biogeography at New Zealand's North Island/South Island junctureJ Roy Soc N ZYear: 2011131 In press.|
|Bunce M,Worthy TH,Phillips MJ,Holdaway RN,Willerslev E,Haile J,Shapiro B,Scofield RP,The evolutionary history of the extinct ratite moa and New Zealand Neogene paleogeographyProc Natl Acad Sci USAYear: 2009106206462065110.1073/pnas.090666010619923428|
|Gardner RC,De Lange PJ,Keeling DJ,Bowala T,Brown HA,Wright SD,A late Quaternary phylogeography for Metrosideros (Myrtaceae) in New Zealand inferred from chloroplast DNA haplotypesBiol J Linn SocYear: 20048339941210.1111/j.1095-8312.2004.00398.x|
|Buckley TR,Marske K,Attanayake D,Phylogeography and ecological niche modelling of the New Zealand stick insect Clitarchus hookeri (White) support survival in multiple coastal refugiaJ BiogeogrYear: 20103768269510.1111/j.1365-2699.2009.02239.x|
|Chapple DG,Daugherty CH,Ritchie PA,Comparative phylogeography reveals pre-decline population structure of New Zealand Cyclodina (Reptilia: Scincidae) speciesBiol J Linn SocYear: 20089538840810.1111/j.1095-8312.2008.01062.x|
|Morgan-Richards M,Trewick SA,Wallis GP,Chromosome races with Pliocene origins: evidence from mtDNAHeredityYear: 20018630331210.1046/j.1365-2540.2001.00828.x11488967|
|Buckley TR,Marske KA,Attanayake D,Identifying glacial refugia in a geographic parthenogen using palaeoclimate mdeling and phylogeography: the New Zealand stick insect Argosarchus horridus (White)Mol EcolYear: 2009184650466310.1111/j.1365-294X.2009.04396.x19840262|
|Claessens L,Veldkamp A,Broeke EM,Vloemans H,A Quaternary uplift record for the Auckland region, North Island, New Zealand, based on marine and fluvial terracesGlobal Planet ChangeYear: 20096838339410.1016/j.gloplacha.2009.03.001|
|Pons J,Fujisawa T,Claridge EM,Savill RA,Barraclough TG,Vogler AP,Deep mtDNA subdivision within Linnean species in an endemic radiation of tiger beetles from New Zealand (genus Neocicindela)Mol Phylogenet EvolYear: 20105925126221338699|
|O'Neill SB,Chapple DG,Daugherty CH,Ritchie PA,Phylogeography of two New Zealand lizards: McCann's skink (Oligosoma maccanni) and the brown skink (O. zelandicum)Mol Phylogenet EvolYear: 2008481168117710.1016/j.ympev.2008.05.00818558496|
|O'Neill SB,Buckley TR,Jewell TR,Ritchie PA,Phylogeographic history of the New Zealand stick insects Niveaphasma annulata (Phasmatodea) estimated from mitochondrial and nuclear lociMol Phylogenet EvolYear: 20095352353610.1016/j.ympev.2009.07.00719596452|
|Shepherd LD,Perrie LR,Microsatellite DNA analyses of a highly disjunct New Zealand tree reveal strong differentiation and imply a formerly more continuous distributionMol EcolYear: 2011201389140010.1111/j.1365-294X.2011.05017.x21366745|
|Marra MJ,Quaternary fossil beetles from New ZealandN Z EntomolYear: 20083151610.1080/00779962.2008.9722160|
|Elias S,Quaternary beetles and the environmentsYear: 1994Washington, D.C.: Smithsonian Institution Press|
|Ashcroft MB,Identifying refugia from climate changeJ BiogeogrYear: 20103714071413|
|Varela S,Rodríguez J,Lobo JM,Is current climatic equilibrium a guarantee for the transferability of distribution model predictions?J BiogeogrYear: 2009361645165510.1111/j.1365-2699.2009.02125.x|
|Anderson RP,Raza A,The effect of the extent of the study region on GIS models of species geographic distributions and estimates of niche evolution: preliminary tests with montane rodents (genus Nephelomys) in VenezuelaJ BiogeogrYear: 2010371378139310.1111/j.1365-2699.2010.02290.x|
|Thuiller WL,Brotons L,Araújo MB,Lavorel S,Effects of restricting environmental range of data to project current and future species distributionsEcographyYear: 20042716517210.1111/j.0906-7590.2004.03673.x|
|Davis MB,Quaternary history of deciduous forests of eastern North American and EuropeAnn Missouri Bot GardenYear: 19837055056310.2307/2992086|
|Pigot AL,Phillimore AB,Owens IPF,Orme CDL,The shape and temporal dynamics of phylogenetic trees arising from geographic speciationSyst BiolYear: 20105966067310.1093/sysbio/syq05820952757|
|Ashton JH,Catalogue of the Cicadidae in the South Australian museum with descriptions of several new speciesTrans Proc R Soc S AustYear: 191438345358|
|Folmer O,Black M,Hoeh W,Lutz R,Vrijenhoek R,DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebratesMol Mar Biol BiotechnolYear: 199432942977881515|
|Simon C,Frati F,Beckenbach A,Crespi B,Liu H,Flook P,Evolution, weighting, and phylogenetic utility of mitochondrial gene sequences and a compilation of conserved PCR primersAnnals of the Entomological Society of AmericaYear: 199487651701|
|Lee YJ,Hill KBR,Systematic revision of the genus Psithyristria Stål (Hemiptera: Cicadidae) with seven new species and a molecular phylogeny of the genus and higher taxaSystematic EntomologyYear: 20103527730510.1111/j.1365-3113.2009.00509.x|
|Sueur J,Vanderpool D,Simon C,Ouvrard D,Bougoin T,Molecular phylogeny of the genus Tibicina (Hemiptera, Cicadidae): rapid radiation and acoustic behaviourBiol J Linn SocYear: 20079161162610.1111/j.1095-8312.2007.00823.x|
|Maddison DR,Maddison WP,MacClade 4: Analysis of phylogeny and character evolutionBook MacClade 4: Analysis of phylogeny and character evolutionYear: 2000ed.^edsCity: Sinauer Associates|
|Zwickl DJ,Genetic algorithm approaches for the phylogenetic analysis of large biological sequence datasets under the maximum likelihood criterionYear: 2006The University of Texas: Ph. D|
|Posada D,Crandall KA,Modeltest: testing the model of DNA substitutionBioinformaticsYear: 19981481781810.1093/bioinformatics/14.9.8179918953|
|Lanfear R,Calcott B,Ho SYW,Guindon S,PartitionFinder: combined selection of partitioning schemes and substitution models for phylogenetic analysesMol Biol EvolYear: 2012291695170110.1093/molbev/mss02022319168|
|Marshall DC,Cryptic failure of partitioned Bayesian phylogenetic analyses: lost in the land of long treesSyst BiolYear: 20105910811710.1093/sysbio/syp08020525623|
|Brown J,Hedtke SM,Lemmon AR,Lemmon EM,When trees grow too long: Investigating the causes of highly inaccurate Bayesian branch-length estimatesSyst BiolYear: 20105914516110.1093/sysbio/syp08120525627|
|Sukumaran J,Holder MT,DendroPy: a Python library for phylogenetic computingBioinformaticsYear: 2010261569157110.1093/bioinformatics/btq22820421198|
|Brandley MC,Wang Y,Guo X,de Oca ANM,Feria-Ortiz M,Hikida T,Ota H,Accommodating heterogeneous rates of evolution in molecular divergence dating methods: an example using intercontinental dispersal of Pleistodon (Eumeces) lizardsSyst BiolYear: 20116031510.1093/sysbio/syq04520952756|
|Phillips MJ,Branch-length estimation bias misleads molecular dating for a vertebrate mitochondrial phylogenyGeneYear: 200944113214010.1016/j.gene.2008.08.01718809474|
|Maddison WP,Maddison DR,Mesquite: a modular system for evolutionary analysis. Version 2.75. http://mesquiteproject.orgBook Mesquite: a modular system for evolutionary analysisYear: 2011275City Version 2.75. http://mesquiteproject.org.|
|Drummond AJ,Rambaut A,BEAST: Bayesian evolutionary analysis by sampling treesBMC Evol BiolYear: 2007721410.1186/1471-2148-7-21417996036|
|Sota T,Hayashi M,Comparative historical biogeography of Plateumaris leaf beetles (Coleoptera: Chrysomelidae) in Japan: interplay between fossil and molecular dataJ BiogeogrYear: 20073497799310.1111/j.1365-2699.2006.01672.x|
|Rambaut A,Drummond AJ,Tracer v1.3 Available from http://evolve.zoo.ox.ac.uk/.|
|Kass RE,Raftery AE,Bayes factorsJ Am Stat AssocYear: 19959077379510.1080/01621459.1995.10476572|
|SPLITSSPecies Limits by Threshold Statistics, v. 1.0-11 Available from http://barralab.bio.ic.ac.uk/downloads.html.|
|R Development Core TeamR: A language and environment for statistical computingYear: 2011Vienna, Austria: R Foundation for Statistical Computing|
|Excoffier L,Laval G,Schneider S,Arlequin (ver. 3.0): an integrated software package for population genetics data analysisEvol Bioinform OnlineYear: 20051475019325852|
|Leathwick J,Morgan F,Wilson G,Rutledge D,McLeod M,Johnston K,Land Environments of New Zealand Technical GuideYear: 2003Auckland, New Zealand: David Bateman Ltd.|
|Phillips SJ,Anderson RP,Schapire RE,Maximum entropy modeling of species geographic distributionsEcol ModelYear: 200619023125910.1016/j.ecolmodel.2005.03.026|
|Phillips SJ,Dudík M,Modeling of species distributions with Maxent: new extensions and a comprehensive evaluationEcographyYear: 20083116117510.1111/j.0906-7590.2008.5203.x|
|Pearson RG,Dawson TP,Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful?Glob Ecol BiogeogrYear: 20031236137110.1046/j.1466-822X.2003.00042.x|
|Caccone A,Sbordoni V,Molecular biogeography of cave life: a study using mitochondrial DNA from Bathysciine beetlesEvolutionYear: 20015512213011263733|
|Juan C,Oromi P,Hewitt GM,Mitochondrial phylogeny and sequential colonization of Canary Islands by darkling beetles of the genus PimeliaProc Royal Soc Biol Sciences Series BYear: 199526117318010.1098/rspb.1995.0133|
|Morrison CL,Ríos R,Duffy JE,Phylogenetic evidence for an ancient rapid radiation of Carribean sponge-dwelling snapping shrimpsMol Phylogenet EvolYear: 20043056358110.1016/S1055-7903(03)00252-515012939|
|Schubart CD,Diesel R,Hedges SB,Rapid evolution to terrestrial life in Jamaican crabsNature (London)Year: 199839336336510.1038/30724|
[Figure ID: F1]
The mainland New Zealand islands showing major locations referenced in the text. The dotted line in eastern NI approximates the major SW-NE mountain axis (which lies on the Alpine Fault together with the Southern Alps in SI). The latitudinal reach of the mainland islands extends from approximately 34° S to 47° S.
[Figure ID: F2]
Distributions of the Amphipsalta and Notopsalta species. Records are based on C. Simon lab (circles) and Dugdale and Fleming  (triangles) observations, plotted on a shaded elevation map. Filled symbols indicate collected specimens; open symbols represent song records only.
[Figure ID: F3]
Maximum-likelihood gene trees estimated in Garli v2.0. Bootstrap support values from 100 pseudoreplicates are shown on major branches. Colors indicate membership in groups shown on Figure 5.
[Figure ID: F4]
Plots of mtDNA uncorrected distance vs. ML patristic distance.A) Plot from a well-sampled Kikihia COI + COII mtDNA ingroup dataset containing 30 species . B) Plot from the AmphipsaltaNotopsalta COI dataset from this study (light green dots), superimposed on the plot from A. The datasets span a nearly identical range of uncorrected distance values, but many corrected pairwise distances are over twice as large for AmphipsaltaNotopsalta at uncorrected distances > 6%. The trendline in B was fitted to the Kikihia data points under 4% uncorrected distance
[Figure ID: F5]
BEAST v 1.6.1 relaxed-clock chronogram (maximum clade compatibility tree, with mean node ages calculated), with outgroup taxon and branches pruned. Tree was constructed using the Amphipsalta-Notopsalta mtDNA COI dataset and calibrated using literature estimates of per lineage molecular clock rates in insects. Asterisks indicate major branches with > 90% posterior probability. Horizontal blue bars indicate 95% credible intervals on the mean divergence time estimates. Major haplotype groups are shown, by color, on the accompanying maps of the NZ landscape. Shading represents climatic deterioration during the Pleistocene, and the vertical dotted line represents accelerated uplift in northern NZ around 1 Ma.
[Figure ID: F6]
Projected current and Last Glacial Maximum distributions for each Amphipsalta and Notopsalta species, modeled in Maxent. For A. cingulata and N. sericea, which are found only on NI, two projected LGM distributions are shown – one with the model trained on background localities from all of NZ, and the other trained only on NI localities. Scale shows climate suitability values; warmer colors indicate higher suitability.
Nuclear and mtDNA dataset statistics, with outgroup excluded, and including partition-specific substitution models from Modeltest; asterisked partitions were used for the final analyses
|Gene/partition||# Sites||# Variable sites||# Parsimony informative||Model selected|
|COI (1st pos.)*
||TrN + I
|COI (2nd pos.)*
|COI (1st + 2nd pos.)
||HKY + G
|COI (3rd pos.)*
||TIM + G
|COI (all sites)
||GTR + I + G
||TIM + I
|EF-1a||1304||69||56||HKY + I|
Analysis of molecular variance (AMOVA) showing partitioning of mtDNA genetic variation within and among regions1
|Source of variation||d.f.||Sum of squares||Variance components||Percentage of variation|
||Fst = 1.00000
|Among populations within regions
||Fst = 0.53778
|Among populations within regions
||Fst = 0.98929
|Among populations within regions
||Fst = 0.83766
|Among populations within regions
1 Regions were defined by a six-way, north-to-south partition of the NZ mainland. Asterisks indicate statistically significant tests (p < 0.05).
Mean divergence-time estimates (in Ma) and 95% confidence intervals from BEAST analyses using different tree models
|Tree model||root||Amphipsalta-Notopsalta||Amphipsalta||N. sericea MRCA||A. strepitans MRCA||A. zelandica MRCA||A. cingulata MRCA|
|coalescent (expansion)||17.00 (10.70-23.42)||8.13 (5.64-10.35)||5.85 (4.20-7.59)||1.26 (0.77-1.89)||1.07 (0.68-1.52)||0.39 (0.23-0.59)||0.18 (0.08-0.30)|
Estimates of interspecific divergence times were strongly influenced by the tree prior, while ages of species’ MRCAs and intraspecific divergences varied much less.
Bayes factor scores (from Tracer v1.5) from comparisons of alternative tree models for the BEAST divergence time analyses
|Tree model||estimated mean -lnL||S. E.||
2 (ln Bayes factor score)
|vs. Yule||vs. birth-death||vs. coalescent (constant size)||vs. coalescent (expansion)|
|coalescent (constant size)
Strong support for the birth-death model was indicated, according to the criterion 2 ln Bayes factor > 10.
Published arthropod substitution rate estimates for the mitochondrial gene cytochrome oxidase subunit I (COI)
|Reference||Taxon||Corrected pairwise distance (%)||Per lineage rate, (subst./site/my)|
|Caccone and Sbordoni 
Ovobathysciola cave beetles
|Caccone and Sbordoni 
Speonornus cave beetles
|Juan et al. 
Pimelia beetles, Canary Is.
|Knowlton and Weigt , Morrison et al. 
Alpheus snapping shrimp
|Papadopoulou et al. 
||Tenebrionid beetle genera (6)
|Schubart et al. 
Sesarma land crabs #1
|Sota and Hayashi ||Plateumaris leaf beetles||1.6||0.008|
Although these distances were reported as percentages in the papers referenced, a common practice, this is not ideal for corrected distances since the values can exceed 100%.
Previous Document: The impact of workplace factors on evidence-based speech-language pathology practice for children wi...
Next Document: Photochemical behavior of Polylactide/ZnO nanocomposite films.