The radiation of Austral teals (Aves: Anseriformes) and the evolution of flightlessness

Abstract

INTRODUCTION

The Austral teals (see Table 1) are a group of dabbling ducks endemic to Madagascar, the Andaman Islands, Indonesia, New Guinea, Australia, and the New Zealand region. Consisting of the grey-teal species complex, with representatives from all these areas, and the brown-teal species complex from New Zealand and its sub-Antarctic islands (Fig. 1), the origin and evolution of Austral teals are poorly resolved. With data from only a small percentage of the mitochondrial genome available for phylogenetic analysis, this uncertainty encompasses all taxonomic levels in the group, from their relationship to other groups of dabbling ducks (genus Anas), such as green-winged teals, pintails, and mallards and their allies (see Table 1), to the specific relationships within the two species complexes. For example, it is only recently that the Andaman, Mascarene, Sunda, and grey teals have been recognized as separate species (Winkler et al. 2020, Gill et al. 2024; see Table 1). The acknowledged phylogenetic proximity and geographical ranges of these last taxa means that we do not consider them further; their diversification is not pertinent to the questions posed here, and we use the grey teal, Anas gracilis Buller, 1869, as an exemplar for this clade.

Figure 1.

Distribution of Austral teals (Anas spp.) in the Southern Hemisphere, including the Madagascar teal (Anas bernieri), the chestnut teal (Anas castanea) from Australia, and the brown-teal species complex from the New Zealand region. This complex comprises the brown teal (Anas chlorotis) from mainland New Zealand and the Chatham Islands, the extinct (†) Chatham Island teal (Anas chathamica), Auckland Island teal (Anas aucklandica), Campbell Island teal (Anas nesiotis), and the extinct Macquarie Island teal (Anas sp. indet.). The distribution of the grey teal (Anas gracilis) from Australasia (Australia, New Guinea, New Caledonia, and New Zealand) is not shown owing to the overlap with the chestnut and brown teals. We also omit the distributions of species recently separated from A. gracilis, which are not considered here: Sunda teal (Anas gibberifrons), Andaman teal (Anas albogularis), and the extinct Mascarene teal (Anas theodori). Distributions are based off the Cornell Laboratory of Ornithology Birds of the World website.

Open in new tabDownload slide

Based on the mitochondrial Cytb and ND2 genes, Johnson and Sorensen (1999) identified the brown-teal species complex in the New Zealand region as an early diverging sister clade to pintails, green-winged teals, mallards, and the grey-teal species complex. Gonzalez et al. (2009), using the same data but an alternative phylogenetic reconstruction approach, identified brown-teal and grey-teal complexes as sister clades, which, together with pintails and green-winged teals, formed a sister clade to the mallard.

Within the Austral teals, the brown-teal species complex has been of special interest because it includes examples of flightlessness. The New Zealand region is home to one extinct and three living named endemic Anas species (Fig. 1): the extinct Chatham Island teal, the mainland brown teal, the Auckland Island teal, and the Campbell Island teal (see Table 1). The taxonomic status of extinct ‘brown teal’ on the Chatham Islands (nominally Anas chlorotis Gray, 1845) and the sub-Antarctic Macquarie Island teal (Anas sp.) is currently undetermined (Worthy 2002, Miskelly and Powlesland 2013, Birds New Zealand Checklist Committee 2022).

The brown teal is the only fully volant living species within the group. A mere ~2500 individuals of this species are left, with populations on Stewart Island now extinct (Williams 2013) and on the South Island functionally extinct owing to hybridization with mallards (Birdlife International 2015, Cole and Wood 2017). The Campbell Island teal is fully flightless (Livezey 1990), whereas the smaller Auckland Island teal is able to fly, but is apparently not capable of sustained flight. It has been observed to be a weak flyer, capable of only short flights, with feet and wings splashing across the water, and flapping wings to jump onto shoreline ledges and climb stepping stones (Buller 1905, Weller 1975). The Chatham Island teal is the largest of the brown teals and was formally described in a separate genus (Pachyanas) based on its morphological distinctiveness (Oliver 1955). However, recent palaeogenetic research has shown it to be the earliest diverging member of the brown-teal radiation (Mitchell et al. 2014). Furthermore, despite suggestions that it was volant (Mitchell et al. 2014), osteological analysis supports assertions that it was flightless (Millener 1999, Williams 2015a, b).

There have been several attempts to resolve the phylogenetic relationships and colonization history of the brown-teal radiation using comparative morphology (Livezey 1990, 1991), allozymes (Daugherty et al. 1999), and partial mitochondrial DNA sequence data (Young et al. 1997, Johnson and Sorenson 1998, 1999, Kennedy and Spencer 2000, Mitchell et al. 2014). The majority of studies have suggested that the Campbell Island teal and Auckland Island teal could be sister taxa, most closely related to the brown teal (Livezey 1991, Young et al. 1997, Johnson and Sorenson 1998, 1999, Mitchell et al. 2014). Alternatively, Kennedy and Spencer (2000) suggested that the Campbell Island teal and brown teal could be sister taxa to the exclusion of the Auckland Island teal, although their data were equivocal, whereas Daugherty et al. (1999) suggested that the brown teal and Auckland Island teal are sister taxa.

There is also disagreement over the closest relatives of the Madagascar teal. Gonzalez et al. (2009) and Mitchell et al. (2014) found that it was sister to the brown-teal complex, in contrast to Johnson and Sorenson (1998, 1999), who recovered it as a member of the grey-teal complex, where it has traditionally been placed.

The unresolved nature of the phylogenetic relationships has implications for our understanding of the evolution of the Austral teals in general and for the colonization history of New Zealand and outlying islands, in addition to the evolution of flightlessness in the brown-teal radiation. This study presents the first whole-mitogenome approach to resolve the phylogeny and evolutionary history of Austral teals.

MATERIALS AND METHODS

Specimens

Tissue (N = 6) and feather (N = 1) samples or genomic DNA extracts (N = 12) from a previous study (Kennedy and Spencer 2000) were sourced from the Department of Zoology (University of Otago), Auckland Museum, and the Durrell Wildlife Conservation Trust in Jersey, including: brown teal (N = 4); Auckland Island teal (N = 2); Campbell Island teal (N = 2); Madagascar teal (N = 4); grey teal (N = 2); chestnut teal (N = 2); grey duck (N = 2); and mallard (N = 1) (Supporting Information, Table S1).

DNA extraction

Existing DNA extracts were produced by Kennedy and Spencer (2000) following a modified phenol–chloroform extraction protocol by Kocher et al. (1989). To test for DNA degradation, fragmentation was checked on a 1% agarose gel using GelRed (Biotium) and a 1 kbp HyperLadder (Bioline). Genomic DNA extraction from new tissue samples was conducted using the MagJet gDNA kit (Thermo Fisher Scientific) with a modified protocol based on the manufacturer’s instructions for manual genomic DNA purification (see Supporting Information). Successful extractions were verified by visualizing DNA using gel electrophoresis as above and quantified using a NanoDrop 2000 (Thermo Fisher Scientific).

Whole-mitogenome amplification and sequencing

Four overlapping primer pairs were designed for complete mitogenome amplification (Supporting Information, Tables S2 and S3) using published complete anatid mitogenomes: mandarin duck, falcated duck, Baikal teal, mallard, Indian spot-billed duck, tufted duck, Muscovy duck, ruddy shelduck, common shelduck, black swan, and the lesser whistling duck. Mitogenome sequences were downloaded from GenBank and aligned in BioEdit (v.7.2.5; Hall 1999). The four primer pairs (F16–R8, F3–R10, F5043–R3, and F8–R12734) amplified fragments with lengths ranging from 4 kb to >5 kb (F5043–R3, 3757 bp; F3–R10, 4627 bp; F8–R12734, 4854 bp; and F16–R8, 5346 bp).

Long-range PCR was conducted using a KAPA LongRange HotStart Kit (KAPA Biosystems) in 25 µL volumes with 24 µL Master-mix (ultra-pure MilliQ water, 1× KAPA LongRange Buffer, 1.75 mM of MgCl₂, 0.3 mM of dNTPs, 0.5 mM of each primer, and 0.025 units of DNA polymerase) and 1 µL DNA. PCR thermocycling conditions followed the KAPA LongRange HotStart DNA Polymerase kit protocol for amplification of long targets and/or low concentrations of template DNA (Supporting Information, Table S4). A negative control was included in each PCR run to test for contamination. PCR amplification was verified by visualization on a 1% agarose gel with GelRed (Biotium) and 1 kbp Hyperladder (Bioline). The PCR products were purified using Mag-Bind RXNPurePlus beads (OMEGA Biotek) (for preparation of magnetic bead size selection buffer, see the Supporting Information) following the manufacturer’s instructions with the following modification: 1:1 volume of magnetic beads, 85% ethanol wash, and 30 μL elution buffer. Concentrations of DNA were quantified using a NanoDrop 2000 (Thermo Fisher Scientific).

The PCR products for each sample were pooled at equimolar ratios and sonicated using a Picorupter (Diagenode) to shear the DNA into small fragments (seven cycles of 15 s on and 45 s off). Successful fragmentation was checked on a 2% agarose gel. After sonication and in between each of the steps in double-stranded DNA library construction, DNA purification was conducted as above using Mag-Bind RXN Pure Plus beads.

Indexed double-stranded barcoded libraries for high-throughput sequencing were prepared from the purified sheared long-range PCR products following a modified Illumina TruSeq protocol (see Supporting Information). Quantification of libraries was conducted with a Qubit 2.0 fluorometer (Invitrogen) following the manual assay preparation instructions. Libraries were pooled at equimolar concentrations, and a final concentration was determined using the Qubit 2.0 (Invitrogen). High-throughput sequencing was conducted at the Otago Genomic and Bioinformatics Facility (University of Otago) on the Illumina MiSeq platform.

Bioinformatic and phylogenetic analysis

Mitogenomes were assembled in Geneious (v.10.1.3; Kearse et al. 2012) using the standard mapping algorithm. We used the mallard as a reference genome (EU009397.1). No duplicates were removed after assembly because sonication of long-range PCR products produces ‘pseudo-unique’ reads (i.e. even if two reads were derived from amplicons of the same molecule, the sonication process would result in different start and end coordinates, making them indistinguishable from truly unique reads). Coverage for our mitogenomes ranged between 95× and 422×. The assembled mitogenomes were aligned, annotated, and manually edited to exclude evident errors in BioEdit (v.7.2.5; Hall 1999). Our phylogenetic analysis used complete mitogenomes supplemented with published mitogenome sequences of Anas spp., including two each of the Indian spot-billed duck, mallard, falcated duck, northern shoveler, northern pintail, Baikal teal, Eurasian teal, and one Chatham Island teal, with outgroup mitogenomes including two each of the redhead, the ruddy shelduck, and the common pochard (see Supporting Information, Table S2). We used the same fossil calibration points as Mitchell et al. (2014). Sequences were aligned in BioEdit (v.7.2.5; Hall 1999) using the default settings and checked by eye.

Phylogenetic analyses were conducted using a number of different approaches and dataset treatments, including an unpartitioned alignment, an alignment partitioned according to PartitionFinder2 (Lanfear et al. 2016), and an alignment partitioned into individual genes, tRNAs, and rRNAs. Genes were partitioned further into first, second, and third codon positions (i.e. genes first, second, third; tRNA; rRNA partitioning scheme). In all cases, the control region/D-loop was excluded from the analyses owing to ambiguous alignment. The PartitionFinder2 analyses split the alignment into 19 different partitions but assigned a version of the HKY or GTR model of nucleotide substitution to all but two of these partitions (both F81). In order to prevent over-parameterization, we disregarded this highly split model. The impact of partitioning on the dataset was evaluated by conducting all analyses with an unpartitioned dataset, and with the genes first, second, third; tRNA; rRNA partitioning scheme. The best-fitting models of nucleotide substitution were identified using the Akaike information criterion as implemented in jModeltest2 (Darriba et al. 2012). Downstream phylogenetic and molecular clock analyses were run with the following substitution models: rRNA, TPM1uf+I; tRNA, GTR+I; and genes (all codon positions), TIM2+I+G. For the unpartitioned dataset, we used GTR+I+G.

We used BEAST (v.2.7.4; Bouckaert et al. 2014) for phylogenetic reconstructions and molecular clock analyses. For the partitioned dataset, each of the five partitions was introduced as an individual alignment. We followed Mitchell et al. (2014) to calibrate our dataset with fossils under a Bayesian framework. We modelled the stem age of Anas using a lognormal prior distribution (mean = 1, standard deviation = 1, offset 11.2) such that 95% of the prior probability fell between 11.6 and 30.5 Mya. A very broad, uniform substitution rate prior from 0.0001 to 0.1 substitutions per site/Myr was set for the unpartitioned dataset and for all partitions of the partitioned dataset to avoid the Markov chain Monte Carlo algorithm converging on extreme values. We applied a strict molecular clock and an optimized relaxed molecular clock using the Yule prior and birth–death priors, respectively, with a 10 000 000 generation chain. Results were visualized in Tracer (v.1.5.0) to evaluate whether they adhered to a strict or a relaxed molecular clock model and which tree prior best described the data. Node strength was assessed using Bayesian posterior probabilities. The different settings of (i) strict and relaxed molecular clock with the Yule prior and (ii) strict and relaxed molecular clock with the birth–death prior were then compared using Bayes factors as implemented in Tracer (v.1.5.0), and their suitability was interpreted following Kass and Raftery (1995). After Bayes factor analysis, a relaxed clock (three independent 20 000 000 generation chains) with a Yule prior was applied to the dataset to estimate the substitution rate and divergence times. After checking for Markov chain Monte Carlo convergence for each run, all three runs were combined using LogCombiner (v.2.7.4) to ensure that all parameters had reached effective sample sizes of ≥200. TreeAnnotator (v.2.7.4) was used to combine the independent trees with a 25% burn-in, and FigTree (v1.4.4) was used to construct the chronogram.

In addition to our BEAST analyses, we used RAxML (Stamakatis 2014) to reconstruct our teal phylogeny using the partitioned and unpartitioned datasets described above. The program was run through the Geneious plugin (v.10.1.3; Kearse et al. 2012), with substitution models independently estimated for each partition or set to GTR+I+G for the unpartitioned dataset. We used the rapid bootstrapping algorithm with a parsimony random seed of one. Branch support was estimated with 10 000 bootstrap pseudo-replicates.

RESULTS

Mitogenome sequencing

Long-range PCR amplification of the mitogenome was successful for all 19 samples. High-throughput sequencing resulted in a mean depth of coverage for the mitogenome of 95× to 422×, with the number of mapped reads ranging from 5524 to 22 194, producing mitogenomes of ~16 850 bp with 99.8%–100% coverage of the mallard reference genome.

Phylogenetic analysis

Independent of the data partitioning used, the time-calibrated Bayesian phylogenetic analysis (Fig. 2; Supporting Information, Fig. S1) shows that the Austral teals, in addition to the widely distributed pintails (here represented by Anas acuta Linnaeus, 1758) and green-winged teals (represented by A. crecca Linnaeus, 1758), form a clade that diverged from the mallards in the late Miocene (node 1; Table 2), ~7–8 Mya. Approximately 1 Myr later, the Austral teals, pintails, and green-winged teals diversified into four major clades: the pintails, green-winged teals, brown-teal complex, and grey-teal complex, including the Madagascar teal. The resolution of this divergence (node 2; Fig. 2) is poor, with different datasets and approaches (partitioned, unpartitioned, maximum likelihood, and Bayesian; Fig. 2; Supporting Information, Figs S2, S3) yielding different, but consistently poorly supported topologies.

Figure 2.

A, time-calibrated Bayesian phylogeny of Austral teal mitochondrial DNA created using BEAST (for full phylogeny, see Supporting Information, Fig. S1). Branch lengths are proportional to time (in millions of years before present). Posterior probability support values are shown by the nodes subtending the branches they belong to if less than one. Divergence times and 95% highest posterior densities of node age estimates are shown in Table 2. Nodes mentioned in the main text are numbered in red next to the node. Tips are labelled with species names (see Table 1) and GenBank accession numbers. B, schematic diagram of colonization routes of the New Zealand region by the brown-teal complex. Divergence times are mean estimates based on our phylogenetic analysis in A and Table 1. Solid arrows represent known dispersal events, and dashed arrows represent inferred dispersal events.

Open in new tabDownload slide

The topology of the brown-teal complex is consistent across all datasets and approaches. Divergence time estimates differ slightly depending on whether a partitioned or unpartitioned dataset was used (Table 2). The differences have no influence on the conclusions of this study, and here we report divergence time estimates only for the partitioned dataset.

Within the brown-teal complex, the ancestors of the Chatham Island teal diverged first [node 3; 1.4 Mya, 95% highest posterior density (HPD) 0.9–1.8 Mya], followed by those of the brown teal (node 4; 0.9 Mya, 95% HPD 0.7–1.3 Mya). The Auckland Island teal and Campbell Island teal lineages diverged from each other soon after (node 5; 0.9 Mya, 95% HPD 0.6–1.2 Mya; Table 2). The sister relationship between the Auckland Island and Campbell Island teal is highly supported (node 5; Fig. 2; Supporting Information, Figs S1–S3; posterior probability: partitioned, 0.99; unpartitioned, 0.93; bootstrap support: partitioned, 89; unpartitioned, 88), despite the short internodal distances between branches in the brown-teal clade.

The ancestors of the Madagascar teal diverged from the rest of the grey-teal complex 4.1 Mya (node 6; 95% HPD 2.7–5.5 Mya; Table 2). Within the species we sampled from the recent crown group grey-teal complex, chestnut teals and grey teals are not reciprocally monophyletic.

The estimated substitution rates (in substitutions per site/Myr) were as follows: genes first and second codon positions, 0.0044 (95% HPD 3.2 × 10⁻³ to 5.6 × 10⁻³ for first codon position, 3.2 × 10⁻³ to 5.8 × 10⁻³ for second codon position); genes third codon position, 0.0048 (95% HPD 3.4 × 10⁻³ to 6.2 × 10⁻³); rRNA, 0.0014 (95% HPD 1.0 × 10⁻³ to 1.7 × 10⁻³); and tRNA, 0.0013 (95% HPD 9.0 × 10⁻⁴ to 1.6 × 10⁻³).

DISCUSSION

Evolution of the Austral teals

Our phylogenetic analysis of whole mitogenomes (Fig. 2; Table 2; Supporting Information, Fig. S1) shows that Austral teals, along with the more widespread green-winged teals (here represented by A. crecca) and pintails (here represented by A. acuta), diverged during the late Miocene from a common ancestor with the widespread Northern Hemisphere mallard (Anas platyrhynchos Linnaeus, 1758) (node 1). However, we were unable to determine whether there was: (i) a single pulse of dispersal into the Southern Hemisphere, then subsequent divergence into the grey-teal and brown-teal complexes, or (ii) two independent dispersal events, one leading to the brown teals and the other to the grey teals (including the Madagascar teal). The rapid radiation at this point (node 2) was too fast for the evolution of informative single nucleotide polymorphisms that would have resolved the branching patterns and phylogenetic relationships.

Although mostly consistent with the results of Mitchell et al. (2014), our analyses are in contrast to this earlier study with regard to the relationship of Madagascan and brown teals. Mitchell et al. (2014) (and Gonzalez et al. 2009) found that the brown-teal radiation was closest to the Madagascan teal, whereas our findings strengthen the hypothesis that the Madagascan, grey, and chestnut teals are more closely related to each other (see Johnson and Sorenson 1999) (Fig. 2; Supporting Information, Fig. S1; node 6). The paraphyletic nature of grey and chestnut teals in Australia (forming a single intermixed clade rather than two reciprocally monophyletic clades; Fig. 2; Supporting Information, Fig. S1) indicates high levels of hybridization (Donne-Goussé et al. 2002). Future genomic research, involving nuclear data and additional samples of other pintails and green-winged teals (e.g. Spaulding et al. 2023), is needed to help resolve deeper nodes in our phylogeny and the evolution of Austral teals.

Phylogeography of brown teals

Within Austral teals, the phylogeography and evolution of flightlessness in the brown-teal complex also remains unresolved. Our analyses support previous assertions that the Auckland and Campbell Island teals are sister taxa (node 5) (Young et al. 1997, Johnson and Sorenson 1999, Mitchell et al. 2014).

Our phylogenetic analyses imply that over a short period of time the brown-teal ancestors dispersed from mainland New Zealand to the Chatham Islands, with a subsequent independent stepping-stone dispersal to the Auckland Islands, then the Campbell Island (Fig. 2; Supporting Information, Fig. S1). However, we cannot fully exclude the parallel colonization scenario, with the Auckland and Campbell Islands being populated independently from mainland New Zealand. For parallel colonization, two scenarios are possible: (i) the founding populations of the Auckland and Campbell Islands originated from the same ancestral population on mainland New Zealand, which later became extinct (Mitchell et al. 2014); or (ii) the divergence occurred on mainland New Zealand in combination with subsequent dispersal to each of the sub-Antarctic islands and their dying out on the mainland. For the first scenario, it is conceivable that the extinct South Island and Stewart Island mitochondrial brown teal lineages (Williams 2013, Cole and Wood 2017) represent the missing ancestral population on mainland New Zealand. The second scenario seems less parsimonious, however. Independent colonization events of the Chatham Islands, mainland New Zealand, and the Auckland and Campbell Islands (as suggested by Williams et al. 1991) from Australia can be ruled out, given the monophyly of the brown-teal radiation.

Age and evolution of flightlessness in brown teals

An upper limit for the age of a (flightless) island species is the age of island formation, although the real age can be much younger (Slikas et al. 2002). The Auckland and Campbell Islands are all of late Miocene/early Pliocene volcanic origin (Adams et al. 2008, Scott and Turnbull 2019) and therefore considerably older than the estimated divergence times in the brown-teal species complex. Flightlessness in this complex is likely to have evolved within the past 1.4 Myr, during the Pleistocene. Although this estimate provides a maximum time frame for the loss of flight, it does not allow for any conclusions about the time it took for the sub-Antarctic teals to lose their ability to fly. Loss of flight can be rapid and has occurred frequently in other bird lineages. The extinct New Zealand Finsch’s duck, Chenonetta finschi (Van Beneden, 1875), is thought to have become flightless in a mere 9000 years (Worthy 1988). In the genus Tachyeres, three species became flightless within the past 1.4–0.015 Myr (Fulton et al. 2012): the Fuegian steamer duck, Tachyeres pteneres (Forster, 1844), the Chubut steamer duck, Tachyeres leucocephalus Humphrey & Thompson, 1981, and the Falkland steamer duck, Tachyeres brachypterus (Latham, 1790). Flightlessness evolved independently in a number of Porzana rails within the last 0.125–0.5 Myr (Slikas et al. 2002). On the sub-Antarctic islands, numerous species became (near) flightless, including the Auckland Island rail, Lewinia muelleri (Rothschild, 1893), Chatham Island coot, Fulica chathamensis Forbes, 1892, Dieffenbach’s rail, Gallirallus dieffenbachii (G. R. Gray, 1843), and Chatham Island rail, Cabalus modestus (Hutton, 1872).

Flightlessness in the brown-teal species complex is likely to have evolved as an adaptation to new environmental conditions (e.g. isolated islands, extreme prevailing westerly winds) and a lack of predators (James and Olson 1983, Worthy 1988, Livezey 1990, Williams 1995, Trewick 1997, Kennedy and Spencer 2000) but probably not for energy conservation, although flight is energetically expensive (McNabb 1994, 2003).

It is possible that the insular island lineages within the brown-teal species complex were pre-adapted towards flightlessness (i.e. a shift towards flightlessness was already occurring), given morphological characteristics in the brown teal compared with the grey teal, such as shortening of the wing bones and coracoid, relative wing length, wing loading, reduced strength in pectoral girdle elements, and reduced depth of sternal carina (Worthy 1988, Livezey 1990). However, these changes could also result from adaptive responses to selective pressures such as a terrestrial lifestyle through arrested development in adults (Lack 1970, Weller 1975, 1980, Livezey 1990, Trewick 1997).

CONCLUSION

The origin and evolution of the Austral teals has long interested scientists, especially given the incidence of flightlessness within this group. Our data suggest that the radiation of Austral teals dates to the late Miocene. Within the brown-teal species complex, we were able to infer a basal divergence of the extinct Chatham Island teal and the likely stepping-stone colonization of the sub-Antarctic islands from mainland New Zealand within the past 1.4 Myr. Our data therefore provide a phylogenetic and temporal framework for the evolution of the group. These results can serve as a foundation for more in-depth functional genomic studies focusing on the molecular basis of flightlessness in insular lineages.

ACKNOWLEDGEMENTS

Thank you to Glyn Young (Durrell Wildlife Conservation Trust), Andrew Excley, Murray Williams, Brent Evans, and Charles Daugherty for providing specimens for genetic analysis. Māori are kaitiaki (guardians) of the fauna and flora of Aotearoa (New Zealand), with which they are interconnected through shared whakapapa (genealogy). We would like to thank Ngāi Tahu for their kind support of this study. Two anonymous referees and the editor made helpful comments and suggestions on the manuscript.

CONFLICT OF INTEREST

None declared.

FUNDING

This research was supported by funding to M.K. and N.J.R. from the Royal Society of New Zealand.

DATA AVAILABILITY

Mitochondrial consensus sequences produced as part of this study are available on GenBank (PP929786-PP929804).

REFERENCES

Author notes