A null hypothesis in biogeography is that different taxon groups will show similar patterns of distribution and phylogeny if their evolution has responded to the same historic processes. Some biogeographers have used such a proposal as the basis of a putative test of the role of vicariance in biogeography, under the assumption that patterns associated with dispersal would not be coincident [1–3]. However, early in the application of phylogeographic approaches it was expected that dispersal could also be expected to yield congruent biogeographic patterns, where the taxa involved were responding to a common cause [4] and in many contexts the perceived distinction between vicariance and dispersal processes in biogeography is illusional [5]. In continental systems the past location of species' refugia has been identified using phylogeographic data and found to coincide for many taxa, as does the general trend of expansion from refugia [6,7]. On oceanic islands a general trend in the phylogeographic history of taxa colonizing successive islands as they emerged, has been revealed (e.g., [8,9]). However, discordance is also found and whilst this might be attributed to an overwhelming influence of stochastic events, it might also reflect, at least in part, differences associated with mobility, population size and reproductive strategy as these might influence establishment success [10,11]. Distinguishing between random effects and those linked to species traits is very difficult, especially because traits associated with dispersal might be selected against after colonization [10,12,13].

The study of New Zealand biogeography has in the past focused on the role of plate tectonic vicariance in the origins of New Zealand lineages, but more recently molecular data have shown that many lineages have arrived relatively recently and that diversification is young in many taxon groups [14,15]. Furthermore, it is increasingly evident that there is no uniform pattern in the phylogeographic structuring of the biota within New Zealand [16], with so far, no consistent linkage between lineage formation, landscape history or distribution of taxa being evident. Although New Zealand was influenced by Pleistocene climate cycling, this does not by itself explain the phylogeographic structure of taxa we see today. Not every lineage or population in New Zealand is younger than the last glacial maximum (LGM) [14,16], even though every species and population must have been affected by climate cycling. Although some similarity can be found in lineage ages for different taxa, the pattern of spatial structuring is not consistent among lineages [17]. This was a plausible expectation of species that have similar ecological requirements (e.g., habitat, microclimate, diet) and have experienced range retraction and expansion in a similar time frame (e.g., since the LGM). Instead, what we observe among the extant biota in New Zealand is that some endemic taxa have retained high genetic and taxonomic diversity more consistent with pre-Pleistocene events. On the other hand there are endemic taxa that are widespread with low genetic diversity and in many cases comprising a single mainland New Zealand species (e.g., [18,19]; and see [20]) or a combination of both patterns with widespread species showing low genetic diversity and localized populations/species that retained high genetic diversity within small ranges (e.g., Phaulacridium grasshopper [21]; weta, [15,22]). Understanding why different taxon groups have responded so differently requires comparison of taxa across a variety of diverse ecological and evolutionary backgrounds. Their genetic, taxonomic and spatial structuring will help development of more robust hypotheses about the relative role of intrinsic, extrinsic and stochastic processes in the present day biota of New Zealand. Maximising geophysical heterogeneity in such studies allows contrasts to be made between those intrinsic and extrinsic influences and this can be achieved in New Zealand by examining taxa distributed across the mainland and the largest offshore islands, the Chathams.

The Chatham Islands are located on the Pacific Plate (S 44°03′47.16″ and W 175°57′35.73″) approximately 850 km east of mainland New Zealand (Figure 1) at the eastern end of the Chatham Rise, a submerged ridge-structure extending from mid-South Island. The ridge, together with the Chathams, is part of the same continental crust on which New Zealand is located, and only 10% of the Chathams landmass is above sea level nowadays [23]. The present land above sea level consists of two inhabited islands (Main or Chatham and Pitt Island) and several smaller islets and rock-stacks (Figure 1). Paleomagnetic studies provide evidence that the position of the islands on the eastern tip of the Chatham Rise has been more or less fixed since the break up of Gondwanaland [24]. The geology of the Chathams is unlike most of the other small Pacific island groups because their basement consists of old metamorphic rocks (Chatham schist), similar to the old schists known from the foot of the Southern Alps of Otago and Canterbury, New Zealand. Today the Chatham schist is only exposed in the northern part of the Chathams and on the Forty Fours and its metamorphosis is dated to 160 million years (Myr). On top of this old layer are several younger deposits from the Cretaceous and Cenozoic, which contain the oldest fossils on the Chathams. These fossils indicate a sustained isolation of the Chatham area for the last 65 Myr [24]. In contrast to its settled behavior today, the Chatham area has a long history of volcanic activity, mainly between 60–40 million years ago (Ma) (Eocene to Oligocene) and 5–1.6 Ma (Pliocene). In the latter period most of the smaller islands were formed.

Throughout this history, the Chatham Islands have been submerged and emerged several times in response to regional tectonism affecting the entire eastern end of the Chatham Rise, culminating in the last emergence event no more than 4 million years ago [25,26]. Since then, phases of climate cooling in the Pleistocene (<2 Ma) have influenced the connectivity and land area of the islands [24,27] (Figure 1). Connection of islands during lowered sea level may have enabled population extension and gene flow within the Chathams and mainland New Zealand separately, but could not have influenced New Zealand–Chatham connectivity.

The Chathams therefore provide a convenient context for studying New Zealand biogeography in terms of both spatial (two land areas separated by ∼800 km of ocean) and temporal (emergence ∼4 Ma) scale [28]. Mainland New Zealand is about 270-times larger than the Chathams, and the geographic distance from the mainland to Chathams is about half the length of the main islands (∼1600 km), and 4 Myr is an appropriate time period for exploring species evolution [29]. Furthermore, this Plio/Pleistocene period was one of significant geophysical activity in New Zealand [30]. Thus, there is a nice contrast of highly disjunct versus near continuous habitat. Mitochondrial data from taxa in several insect orders present on the Chatham Islands and New Zealand were examined. These data allow assessment of the degree of congruence among genetic diversity, taxonomic status and spatial distribution of mitochondrial lineages within this ecologically diverse set of invertebrate taxa.

This study reports on the mtDNA phylogeography of eight New Zealand invertebrate taxa (Figure 2). As such, each component does not assess variation that might exist among genes. While these data represent just one heritable unit within each taxon group, our focus has been on large-scale differences among ecologically diverse taxa. Taxonomic diversity, distribution and abundance of each taxon differs considerably and the scale of analyses varies accordingly, but our sampling reflects approximately their abundance and the strategy was to sample each genus as intensely as possible to include possible sister lineages to the Chatham taxa. The taxa compared share the characteristics of relative large size and flightlessness (with one partial exception), but they differ in their foraging habits, abundance, and landscapes that they occupy. Some of them are subjects of ongoing research, incorporating non-New Zealand outgroup taxa and additional genetic markers, the results of which are in peer-review at the moment and will be published separately elsewhere. Here, we first report data and phylogenetic structure within each genus, we then summarize and compare phylogeographic structure across New Zealand/Chathams.

The invertebrate specimens for this study were collected mainly by hand and preserved in 95% ethanol. They are housed and catalogued in the collection of the Phoenix Lab at Massey University, Palmerston North, with unique voucher numbers. In some cases data were supplemented by specimens from other collections. DNA extraction used leg tissue and a standard salting-out method [31], with whole genomic DNA extractions stored at −20 °C.

The polymerase chain reaction (PCR) was used to target the mitochondrial cytochrome oxidase I (COI) gene region, using in most cases the published oligonucleotide primers C1-J-1718, C1-J2195 and L2-N-3014 [32]. PCR amplifications were performed in a total volume of 10 μL using Red Hot Taq (ABgene). After an initial denaturation at 94 °C for 3 min. DNA was amplified during 35 cycles of 30 s at 94 °C, 45 s at 50 °C and 30 s at 72 °C, followed by a final extension step at 72 °C for 4 min. The PCR cycle conditions varied from above as follows: 35 cycles of 1 min. at 94 °C, 1 min. at 42 °C and 1.5 min. at 72 °C, followed by a final extension step at 72 °C for 5 min. The amplified products were checked on 1% Agarose gels and purified using SAP/EXO1 (USB Corporation) enzyme digest following the manufacturer's instructions. Purified DNA fragments were used for cycle sequencing with Big Dye terminators under standard conditions and read on an ABI 377 sequencer (ABI).

One prediction widely made by biogeographers is that distribution patterns and phylogeographic structure of different taxon groups are likely to be congruent where they are the product of the same geophysical history. In the present case, emergence of the Chatham Island archipelago from the southern ocean about 4 Ma might be expected to have yielded a set of Chatham Island endemic taxa that were similarly divergent in terms of genetics and morphology, from neighboring New Zealand populations from which they were derived. Although the signature of Chatham diversity is consistently youthful, in keeping with the young age of the islands [28], we found that similarity of mainland New Zealand and Chatham sister lineages ranged from sharing identical haplotypes (e.g., Phaulacridium grasshoppers) to having distinct haplotypes differing by as much as 14.1% (Geodorcus beetles). Such results are in keeping with previous studies of taxa that incorporate New Zealand and Chatham Island sampling, where divergence estimates range from populations with identical cpDNA haplotypes in the fern Asplenium hookerianum [52], to 2% mtDNA sequence divergence in a skink Oligosoma nigriplantare nigriplantare [53], and plants with up to 6.4% sequence divergence [54]. Among these examples are spiders [55], insects (cicada [56]; cockroach [51]; damselfly [57]; stick insect [58]), amphipods [59], isopods [60], and birds (rails [61]; robin [62]; parakeet [63,64]; pigeon [18]). Given the wide taxonomic and ecological diversity represented in these studies, perhaps this wide range in divergence estimates reflects differences in biology rather than history, or a combination of the two.

In reality, there are numerous plausible reasons for the variation observed among studies of different taxa, including different rates of lineage extinction; different times of arrival in the Chatham Islands; differences in effective population size; and different rates of molecular evolution. As far as genetic distance is concerned the sampling effect of lineage extinction will always tend to lengthen the inferred distance to a common ancestor from surviving taxa, and the same can result from accidental failure to include close relatives in analyses [5,14,20]. In Amychus beetles, for example, biogeographic inference is likely to be strongly influenced by extinction as their present narrow range appears to be a remnant of former diversity [50]. Similarly, high observed divergence in Geodorcus was likely exacerbated by gaps in sampling. Some traction is gained in distinguishing these effects in our study when taxonomy (as a proxy for morphological variation) is taken into account, providing an opportunity for comparison that is not available in the majority of published studies that deal with single taxa. Three parameters together contribute information, taxonomy, genealogy and geography, and we note there is no readily discernable concordance of these traits among the taxon groups we studied. Unlike single taxon studies or those with few Chatham samples, here we are able to demonstrate how different the biogeographic patterns expressed by separate taxon groups are (Figure 3). While spatial population structuring within the Chathams might not be expected for taxa that are not also structured between New Zealand and Chathams (e.g., Phaulacridium and Anisolabis), those that are extensively partitioned into species at the source (NZ) might be expected to show equivalent spatial structure at a finer scale (e.g., Celatoblatta and Mecodema). The Chatham Islands do contain endemic species and sometimes endemic sister species, but these are not consistently associated with genera that are speciose in New Zealand (e.g., Geodorcus, Talitropsis). Additionally, there is no consistent geographic relationship between New Zealand and Chatham taxa in each group; New Zealand lineages sister to those in the Chathams are in South Island (Mecodema, Celatoblatta, Hadramphus, Amychus) or the North Island (Geodorcus)—although this is sensitive to taxon sampling, or in the case of the widespread species (Anisolabis, Phaulacridium) in both of the main islands of New Zealand.

Given the young geological age for the Chatham Islands (Pliocene) and much younger connection among them (Pleistocene low sea level), the incidence of multiple endemics such as Talitropsis is intriguing, especially where this contrasts strongly with the situation of their sibling taxa in New Zealand. This is borne out by the respective distribution of genetic diversity, which in the case of Talitropsis is indicative of recent southwards range expansion throughout New Zealand (i.e., less genetic diversity in southern parts of the range), whereas in the Chathams, genetic diversity and morphological traits are tightly partitioned over a narrow spatial scale across 50 km of small islands. This is despite the fact that New Zealand and Chathams would each have existed as continuous landscapes during the LGM. Evidently, spatial and temporal effects are not consistent even among sister lineages, and this suggests a strong influence of stochastic processes.

The observation that land area and age might not be accurate predictors of genetic or taxonomic diversification is significant because these are the very traits that are routinely used in biogeographic analyses [5]. Similarly, the demonstration that not only are taxa with “good” dispersal characteristics (e.g., flying rails [61]; pigeons [18]) and windblown fern spores [52] able to reach the Chathams, but so are taxa that lack obvious dispersal attributes. In our study, the presence of flightless insects in the Chathams that are sister to flightless species in New Zealand shows that such animals are not prevented from oversea dispersal. This suggests that passive mechanisms are influential. It might be relevant, for instance, that all the taxa studied spend a substantial amount of their lives in logs, and logs are known to be transported down rivers and to the sea, and to arrive on beaches having drifted in the ocean sometimes carrying animals and plants with them [65]. Records of logs and other flotsam arriving in the Chatham Islands from New Zealand indicate this passage is frequent and rapid. New Zealand Anisolabis earwigs are most readily found in and around drift wood on beaches, stag beetle larvae and pupa live and feed in logs, and Talitropsis cave weta of all ages occupy holes in wood. Interestingly, in the Chathams the opportunities for log dwelling are fewer on some islands (Forty Fours and The Sisters) where trees are absent. Here, Talitropsis occupy cavities in the peaty soil formed between rocks and Geodorcus and Amychus must complete their life cycles in the humus-rich soil. If passive dispersal is as influential as it appears, we predict that biotic assembly may be mostly contingent upon opportunities for establishment instead.

From examination of eight insect taxon groups comprising many species we conclude that lack of congruence is likely the result of observing evolution at different stages. Anisolabis littorea appears to be in or just completing a wave of expansion around the New Zealand coastline including Chathams. There is no indication from the distribution of genetic diversity, that colonizing the Chathams was any more problematic than expansion along the mainland coast; several separate matrilines have made the 800 km trip successfully. The situation in Phaulacridium marginale is similar with many genealogical lineages shared across New Zealand and across 800 km of ocean. However, Phaulacridium history in New Zealand is intriguing with an endemic species in South Island that contains, in a small geographic area, more genetic diversity than the widespread P. marginale. In Talitropsis, morphogenesis matches mtDNA partitioning, but genetic distance is not correlated with geographic distance. Talitropsis sedilotti appears to have expanded through New Zealand since arrival of the genus on the Chathams, perhaps following local extinction of prior diversity as indicated by distinct localized New Zealand lineages in our data (Figure 3A).

In contrast to this, we see that Celatoblatta has maintained taxonomic diversity in New Zealand that may date from before colonization of the Chathams. Whatever process removed most Talitropsis diversity (possibly Pleistocene climate cycling) appears to have left Celatoblatta relatively intact. We note that in New Zealand, many cockroach species are allopatric but some, especially those associated with forest, have come to occupy relatively wide overlapping ranges. Perhaps even further along in this process, Mecodema reveals a well developed New Zealand radiation, apparently retained over protracted time (since Miocene- [21]). Extensive range shifting resulting in sympatry suggests well developed ecological partitioning of the species concerned. In this case arrival in the Chatham Islands appears to have been relatively late in the diversification of the genus. Thus, colonization and speciation appear to have occurred throughout the short geological lifetime of the Chatham Islands, and extinction in New Zealand has influenced the resulting biogeographic patterns.

Finding biogeographic congruence among taxa, at small or large geographic distances, with or without obvious habitat distinctions and in a time frame that encompasses substantial geogphysical changes is difficult [5]. For the Chatham Islands, the frailty of biogeographic analysis is starkly evident in the former assertion that biological evidence of taxon distribution in the New Zealand/Chatham system was consistent with a process of ancient vicariance [67], whereas it is now clear that neither geology nor biology support this view or are necessary [28,52,54,68]. Two important conclusions arise from this. The first is that if hypotheses about the influence of the size, age and distribution of land on the partitioning of biodiversity cannot be upheld when data about land history are available, what confidence can there be in inferences made about the biogeography of taxa where such data are missing? Secondly, if such landscape traits are not the most important predictors of phylogeographic and phylogenetic partitioning as appears to be the case, then considerable work is needed to describe ecological and behavioral traits of plants and animals so that tests for their biogeographic influence can be made [10,69]. Furthermore, these data are required to refute the hypothesis that stochasticism is the primary force in biogeography.

Future, more intensive sampling of populations and loci may allow assessment of population size and even rates of molecular evolution operating in this system. Natural history studies will contribute to knowledge of population size, behavior and important traits such as longevity and fecundity; all of which influence underlying population genetics. Quantifying these ecological traits will be an important step in elucidating the extent of stochastic processes including extinction and colonization, which are likely to exert strong influence on biogeographic outcomes.