Introduction
Analysis of genetic data in modern populations has proven to be a powerful tool for reconstructing crucial aspects of the evolutionary history of threatened species, thus informing conservation management (Willerslev and Cooper Reference Willerslev and Cooper2005, Rohland and Hofreiter Reference Rohland and Hofreiter2007, Leonard Reference Leonard2008, Gangloff et al. Reference Gangloff, Shirihai, Watling, Cruaud, Couloux, Tillier, Pasquet and Bretagnolle2012a). Similarly, the analysis of ancient DNA (aDNA) is valued as a means to compare past and present levels of genetic variation (Shepherd and Lambert Reference Shepherd and Lambert2008, Shepherd et al. Reference Shepherd, Worthy, Tennyson, Scofield, Ramstad and Lambert2012), identify the ranges of morphologically indistinct taxa (Russello et al. Reference Russello, Poulakakis, Gibbs, Tapia, Benavides, Powell and Caccone2010, Steeves et al. Reference Steeves, Holdaway, Hale, McLay, McAllan, Christian, Hauber and Bunce2010, Robertson et al. Reference Robertson, Stephenson and Goldstien2011, Spurgin et al. Reference Spurgin, Wright, van der Velde, Collar, Komdeur, Burke and Richardson2014) and provide accurate species delineation (Boessenkool et al. Reference Boessenkool, Austin, Worthy, Scofield, Cooper, Seddon and Waters2009). Such data are valuable for conservation managers seeking diagnostic tools to accurately assign conservation status and provide guidance for the restoration of species that have undergone extreme range declines (Hofkin et al. Reference Hofkin, Wright, Altenbach, Rassmann, Snell, Miller, Stone and Snell2003, Valentine et al. Reference Valentine, Duffield, Patrick, Hatch, Butler, Hall and Lehman2007, Russello et al. Reference Russello, Poulakakis, Gibbs, Tapia, Benavides, Powell and Caccone2010).
The New Zealand archipelago is a globally significant example of avian biodiversity loss with at least 60 species of endemic birds becoming extinct since human colonisation 750 years ago and many other taxa such as seabirds, experiencing extreme range retractions (Worthy and Holdaway Reference Worthy and Holdaway2002). Cook’s Petrel Pterodroma cookii is a small endemic seabird that has undergone massive population reductions since human colonisation of New Zealand (Imber et al. Reference Imber, West and Cooper2003, Wilmshurst et al. Reference Wilmshurst, Hunt, Lipo and Anderson2011). Today Cook’s Petrels breed only at the northern and southern extremes of their former range on Te Hauturu-o-Toi/Little Barrier Island (LBI, ~286,000 breeding pairs; Rayner et al. Reference Rayner, Clout, Stamp, Imber, Brunton and Hauber2007) and Whenua Hou/Codfish Island (CDF, ~5,000 breeding pairs; Rayner et al. Reference Rayner, Parker and Imber2008a), separated by approximately 1,300 km (Figure 1). Both populations are recovering following introduced predator eradication, the CDF island population notably had declined to less than 100 breeding pairs by predation from introduced Weka Gallirallus australis and Pacific rat Rattus exulans (Bartle et al. Reference Bartle, Hu, Stahl, Pyle, Simons and Woodby1990, Imber et al. Reference Imber, West and Cooper2003). Historically Cook’s Petrels bred throughout New Zealand in vast colonies, as evidenced by extensive Holocene deposits found in cave and dune systems and also supported by records from early European explorers and Māori oral history (Imber et al. Reference Imber, West and Cooper2003).
Median-joining network for 375 bp of mitochondrial COI sequence from modern, historic and ancient samples of Cook’s Petrel (n = 90). Haplotypes are colour-coded by collection location. The size of each circle is proportional to haplotype frequency. Connecting lines represent a single mutation and solid black circles correspond to missing haplotypes. Buller, Christchurch and South Wairarapa haplotypes were found in historic samples TP08, TES05 and TES02, respectively (see Table S5 for details).
Given the historic ubiquity of Cook’s Petrel and increasing recognition of seabirds as nutrient donors in New Zealand terrestrial ecosystems (Worthy and Holdaway Reference Worthy and Holdaway2002, Towns Reference Towns, Moro, Ball and Bryant2018), conservation management efforts have recently focused on re-establishing lost populations on mainland New Zealand through a chick translocation program (Miskelly et al. Reference Miskelly, Taylor, Gummer and Williams2009). However, such actions are limited by a lack of understanding of the evolutionary history and extirpation dynamics of this once widely distributed New Zealand endemic, specifically a lack of information regarding the genetic make-up of extirpated populations across both North and South Islands.
Morphological, behavioural, and genetic studies have identified the two remaining populations of Cook’s Petrels as distinct conservation management units. As reviewed in Rayner et al. (Reference Rayner, Carragher and Hauber2010a), Imber (Reference Imber, West and Cooper2003) identified differences in body mass between smaller birds on LBI and larger birds on CDF, which parallels significant differences in the breeding season foraging habitat use (Rayner et al. Reference Rayner, Hauber, Clout, Seldon, Van Dijken, Bury and Phillips2008b, Reference Rayner, Hartill, Hauber and Phillips2010b). Moreover, the asynchronous breeding and migration schedules of the two populations (approximately six weeks apart), as well as separate nonbreeding distributions, are a recognised barrier to gene flow through local adaptation in combination with strong natal philopatry (Rayner et al. Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011). In the first genetic study of the species, Rayner et al. (Reference Rayner, Carragher and Hauber2010a) identified significant population genetic structure between the extant populations in mitochondrial DNA (mtDNA) cytochrome oxidase 1 (COI) sequences. Five mtDNA haplotypes were detected, four in the LBI population and one in the CDF population. No mtDNA haplotypes were shared between populations and there was a single fixed polymorphism that could be used to separate samples from both contemporary and historic (museum skin) individuals collected in the eastern Pacific Ocean (Rayner et al. Reference Rayner, Carragher and Hauber2010a, Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011). As a result of these studies, it was recommended that the LBI and CDF populations be treated as distinct conservation management units. However, taxonomic designation and an understanding of the level of genetic diversity within each population require further analysis integrating nuclear markers and data from extirpated populations (Rayner et al. Reference Rayner, Carragher and Hauber2010a).
To establish new populations of Cook’s Petrel, more than 500 chicks from the large LBI population have been translocated to predator-free sites in the North Island of New Zealand since 2010. In the absence of information on the historical distribution of Cook’s Petrel genotypes on the New Zealand mainland, the decision to source birds from the northern LBI population was based on geographic proximity alone. Given the availability of Cook’s Petrel fossils and museum skins from throughout the main islands of New Zealand, analysis of aDNA presents an opportunity to understand patterns of genetic diversity in extirpated populations of this taxon.
Lack of time-stamped data in fragmented populations can result in a misleading picture of former connectivity and species boundaries (Haouchar et al. Reference Haouchar, Pacioni, Haile, McDowell, Baynes, Phillips, Austin, Pope and Bunce2016). Accordingly, this study set out to: (i) investigate what, if any, genetic structure was present in Cook’s Petrel populations across its former range, and (ii) further clarify the relationship between the extant populations and thus the taxonomic status of the LBI and CDF populations using both mitochondrial and nuclear loci. By resolving these issues our final aim was to: (iii) integrate ancient and modern DNA to inform conservation management of the species through population restoration.
Methods
DNA extractions, PCR amplification and sequencing
To augment the existing mitochondrial CO1 sequence data for modern and historic Cook’s Petrels collected from CDF (n = 19), LBI (n = 26), North Pacific (n = 9) and South Pacific (n = 10) (Rayner et al. Reference Rayner, Carragher and Hauber2010a, Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011), DNA was extracted from 23 additional blood samples collected by M. Rayner (LBI, n = 12, CDF, n = 11) (Table S1 in the Online Supplementary Material) with an Invitrogen PureLinkTM Genomic DNA kit according to the manufacturer’s protocol for mammalian tissue with the following modifications: the initial incubation step was increased to 18–19 hours, and the proteinase-K was added after the initial incubation step.
Ancient specimens from throughout the former range of Cook’s Petrel (Figure 1, Table S2) were processed in dedicated ancient DNA facilities (University of Canterbury, Murdoch University and Museum of New Zealand Te Papa Tongarewa) following established standards for ancient DNA work (Shapiro and Hofreiter Reference Shapiro and Hofreiter2012). For samples processed at the latter, DNA was extracted using the phenol-chloroform protocol detailed in Shepherd and Lambert (Reference Shepherd and Lambert2008). DNA extractions at the other two facilities used previously published column-based extraction protocols for bone (Allentoft et al. Reference Allentoft, Collins, Harker, Haile, Oskam, Hale, Campos, Samaniego, Gilbert, Willerslev, Zhang, Scofield, Holdaway and Bunce2012) and toepad (White et al. Reference White, Dawson, Coghlan, Tridico, Mawson, Haile and Bunce2012). Potential contamination was monitored through extraction and PCR negative controls. A subset of samples extracted at Murdoch were also extracted, amplified, and sequenced at UC for authentication (independent replication), and a subset of samples extracted at Te Papa were also amplified and sequenced at UC for authentication (reproducibility). Sequences were only included in the analyses if they were recovered in a minimum of two independent reactions per sample.
A 692 bp fragment of the mitochondrial CO1 gene was amplified in modern samples using CO1-F1B/CO1-R1B PCR Assay (Gangloff et al. Reference Gangloff, Shirihai, Watling, Cruaud, Couloux, Tillier, Pasquet and Bretagnolle2012a). Two overlapping fragments were amplified in historic and ancient samples to give a total of 375 bp of CO1 sequence using AWCF1/AWCintR2 and LCRintF2/AWCintR3 PCR assays (for details see Patel et al. Reference Patel, Waugh, Millar and Lambert2010, Rayner et al. Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011). For degraded samples that could not be amplified with the above primers, we designed additional overlapping primer pairs with M13 tags (underlined) to amplify an even smaller section (16 bp) of the CO1 region (aDNA_CO1mini_2F CO1 CACGACGTTGTAAAACGACTTGGAAACTGACTAGTCCCC/aDNA_CO1mini_2R CO1 GGATAACAATTTCACACAGGACGTGGGAATGCTATGTCAG and aDNA_CO1mini_3F CO1 CACGACGTTGTAAAACGACGAAACTGACTAGTCCCCCTT/aDNA_CO1mini_3R CO1 GGATAACAATTTCACACAGGGTGGGAATGCTATGTCAGGG). This region specifically included the ‘diagnostic’ SNP reported in Rayner et al. (Reference Rayner, Carragher and Hauber2010a, Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011), between modern individuals from LBI and CDF islands (see Results).
Partial fragments of two nuclear introns were also amplified from the modern samples: a 478 bp region of the PAX interacting protein 1 (PAX) locus using PAX_20F/PAX_21R (Leonard Reference Leonard2008) and a 531 bp fragment of the β-fibrinogen intron 7 (β-fibint7) locus using modified primers from Prychitko and Moore (Reference Prychitko and Moore2003): FIB-B17Umodified β-fibint7 TGGATGGTACGTACTTGCATT/FIB-B17internalR β-fibint7 TCCCCAATCTAAACAATTCCTC.
All PCRs for modern samples (25μL) included 1x PCR buffer (BiolineTM), 2mM MgCl2, 0.4μM of each primer, 2.5mM dNTPs, 1U of BIOTAQTM DNA polymerase, and ~10ng of DNA. For all modern samples, PCR thermocycling conditions were: 950C for 90 sec, followed by 30 cycles of 950C for 30 sec, 53–60.70C for 30 sec (530C for CO1, 590C for β-fibint7, 60.70C for PAX), 720C for 45 sec, and a final extension at 720C for 3 min. All PCRs for ancient samples (25μL) included 1x Amplitaq Gold 360 Master Mix, 0.4μM forward and reverse primers, and 2-4μL of DNA. For all ancient samples, PCR thermocycling conditions were: 950C for 10 min, followed by 10 cycles of 950C for 30 sec, 550C for 30 sec, 720C for 30 sec, followed by a further 30 cycles of 950C for 30 sec, 500C for 30 sec, 720C for 30 sec, and a final extension at 720C for 3 min. PCR products were purified for sequencing using AcroPrepTM Advance Filter Plates (PALL). Sequencing was performed in both forward and reverse directions using Big Dye 3.1 chemistry on an Applied Biosystems 3130x1 Genetic Analyser. All sequences were aligned using Geneious 6.1.4 (Biomatters) and all variable sites were confirmed by eye.
Sequence analyses
To address our first aim, we calculated genetic differentiation (ɸST) and standard molecular diversity indices for the mitochondrial CO1 locus for the extant populations of Cook’s Petrel using Arlequin 3.5.1.2 (Excoffier and Lischer Reference Excoffier and Lischer2010). We also constructed median-joining networks (Bandelt et al. Reference Bandelt, Forster and Rohl1999) for modern, historic and ancient mitochondrial CO1 haplotypes using Network 4.613 (www.fluxus-engineering.com) with either transitions (ti) and transversions (tv) equally weighted or using the ti:tv ratio calculated by the AIC in jModeltest 2.1.7 (Darriba et al. Reference Darriba, Taboada, Doallo and Posada2012). We plotted the distribution of the ‘diagnostic’ SNP (nucleotide position 156) across the former range of Cook’s Petrel.
To address our second aim, DnaSP 5.10.1 (Librado and Rozas Reference Librado and Rozas2009) was used to reconstruct haplotype phases for β-fibint7 and PAX according to the algorithms provided by PHASE 2.1 (Stephens et al. Reference Stephens, Smith and Donnelly2001, Stephens and Donnelly Reference Stephens and Donnelly2003). The haplotypes were reconstructed using the following parameters: no recombination, 1,000 MCMC iterations, with 1,000 iterations of burn-in, and an output probability threshold of 0.6 (Garrick et al. Reference Garrick, Sunnucks and Dyer2010). Analyses were repeated three times using different starting seed values and consistency across runs checked by eye. Recombination within β-fibint7 and PAX was tested using TOPALI 2.5 (Milne et al. Reference Milne, Lindner, Bayer, Husmeier, McGuire, Marshall and Wright2008). Genetic differentiation (ɸST) and standard molecular diversity indices were calculated for each locus as per the above. Median-joining networks were constructed for each locus as per the above.
Results
We obtained CO1 sequences for all 23 additional modern samples. Of the 28 historic and ancient samples, only three amplified for the 375 bp fragment of CO1 (Table S2). However, a further 20 samples were successfully sequenced for at least the 16 bp CO1 fragment containing the ‘diagnostic’ SNP at nucleotide position 156 (Table S2).
Among all available modern, historic, and ancient Cook’s Petrel samples (n = 90), there were nine variable sites found within the 375 bp region of the CO1 locus, of which seven were transitions and three were transversions, that defined eight haplotypes (Table S1). There was significant genetic population structure between the extant Cook’s Petrel populations on LBI and CDF (φST = 1.00, P <0.0001). The relationships among mitochondrial CO1 haplotypes are shown in Figure 1. The same network was produced with an equal ti:tv ratio and a ti:tv ratio of 2.83, as calculated in jModeltest. Two of the ancient specimens had novel haplotypes that differed from their most closely related haplotype by one (TES05) or three (TP08) substitutions. The third ancient sample shared a haplotype common in the modern CDF population.
The distribution of the ‘diagnostic’ CO1 SNP at site 156 is mapped in Figure 2. The SNP defining the CDF population (C) was found in ancient samples from throughout the South Island and also in samples from the North Island south of Mahoenui, the SNP defining the LBI population (T) was identified in five ancient populations and co-occurred with the CDF SNP in two North Island locations (Featherston and Mahoenui).
Distribution of the 'diagnostic' Cook’s Petrel mitochondrial CO1 SNP (Rayner et al. Reference Rayner, Carragher and Hauber2010a, Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011). Extant populations represented by squares and extirpated populations by circles. Pink and purple denote a T and a C at site 156, respectively (see text for details). The diamond indicates the holotype specimen of P. cookii (Gray, Reference Gray1843) and the star indicates a specimen from an unknown provenance that was collected after crashing into a streetlight.
Nuclear PAX interacting protein 1 (PAX) and β-fibrinogen intron 7 (β-fibint7)
For the two extant populations of Cook’s Petrel, there were 19 variable sites, 10 of which were transitions and nine of which were transversions. The phased haplotypes were consistent across runs and defined 20 haplotypes (Table S2). The networks of relationships between the PAX haplotypes resulting from the different PHASE runs were identical, as were the networks using an equal ti:tv ratio or a ti:tv ratio of 0.78, as calculated in jModeltest (Figure 3). Haplotype diversity (h) was 0.92 and nucleotide diversity (π) was = 0.0068. Twelve haplotypes were found at LBI, seven of which were unique (h = 0.90, π = 0.0058) and 13 haplotypes were found at CDF, eight of which were unique (h = 0.95, π = 0.0076). There was no significant genetic structure for the PAX locus between the two extant populations (ɸST <0.001, P = 0.44).
Median-joining networks for modern samples of Cook’s Petrel (A) PAX phased haplotypes and (B) β-fibint7 phased haplotypes. Haplotypes are colour-coded by location. The size of each circle is proportional to haplotype frequency. Connecting lines represent a single mutation and solid black circles correspond to missing haplotypes.
For the β-fibint7 locus, there were 12 variable sites, nine of which were transitions and three were transversions. The phased haplotypes were consistent across runs and defined 15 haplotypes (Table S3) (h = 0.86, π = 0.0056). Identical networks of haplotype relationships were reconstructed from all three PHASE runs and using an equal ti:tv ratio or a ti:tv ratio of 3.17, as calculated in jModeltest (Figure 3). Twelve haplotypes were found at LBI, eight of which were unique (h = 0.79, π = 0.0046). Seven haplotypes were found at CDF, three of which were unique (h = 0.81, π = 0.0058). There was moderate genetic structure for the β-fibint7 locus between the two extant Cook’s Petrel populations (ɸST = 0.12, P = 0.0029).
Discussion
Phylogeography of Cook’s Petrel across its historic range
Our phylogeographic analysis of Cook’s Petrel across New Zealand supports existing research showing that genetic structuring can occur in populations at the ends of a broad geographic range (Stephens et al. Reference Stephens, Smith and Donnelly2001). There is little evidence for a distinct phylogeographic break in historic mainland populations as observed in studies of other New Zealand avifauna (Shepherd et al. Reference Shepherd, Worthy, Tennyson, Scofield, Ramstad and Lambert2012, Rawlence et al. Reference Rawlence, Scofield, Spencer, Lalas, Easton, Tennyson, Adams, Pasquet, Fraser, Waters and Kennedy2016, Reference Rawlence, Till, Easton, Spencer, Schuckard, Melville, Scofield, Tennyson, Rayner, Waters and Kennedy2017). Instead, our results demonstrate a zone of co-occurring mitochondrial lineages, particularly in the North Island, more reflective of a haplotype cline.
Pterodroma petrels are among the most mobile birds on earth, routinely travelling a thousand or more kilometres in a single day during oceanic foraging trips (Boessenkool et al. Reference Boessenkool, Austin, Worthy, Scofield, Cooper, Seddon and Waters2009, Rayner et al. Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011, Reference Rayner, Taylor, Gummer, Phillips, Sagar, Shaffer and Thompson2012). Thus, it is unsurprising that historic populations breeding on the North Island of New Zealand do not show the same level of phylogeographic structuring as other New Zealand seabirds, such as some shag species, which by comparison are sedentary and remain in close proximity year round to breeding colonies (Rawlence et al. Reference Rawlence, Scofield, Spencer, Lalas, Easton, Tennyson, Adams, Pasquet, Fraser, Waters and Kennedy2016). We acknowledge that the lack of a distinct phylogeographic break in Cook’s Petrel, compared with other New Zealand seabirds, could be attributed to sampling of ancient bones not related to local breeding colonies. However, we consider this unlikely given the strong natal philopatry of Pterodroma petrels in general (Brooke Reference Brooke2004), which has been shown to impact genetic structure between populations breeding on differing islands at close geographic proximity (Welch et al. Reference Welch, Fleischer, James, Wiley, Ostrom, Adams, Duvall, Holmes, Hu, Penniman and Swindle2012a).
Our results are consistent with previous ancient DNA studies demonstrating character fixation in extant populations supported by human-mediated extinctions across a species’ historic range (Stephens and Donnelly Reference Stephens and Donnelly2003, Rohland and Hofreiter Reference Rohland and Hofreiter2007). In the case of Cook’s Petrel our study suggests that human-mediated extinction of intermediary populations across a latitudinal cline has reinforced the differentiation of northern and southern Cook’s Petrel lineages. This is demonstrated by character fixation in the mitochondrial CO1 locus, and additional differentiation in the nuclear β-fibint7 locus. However, in the case of Cook’s Petrel, anthropogenic character fixation has supported an existing evolutionary process already under way, resulting in the differentiation at the northern and southern ends of the species range (Stephens et al. Reference Stephens, Smith and Donnelly2001). Ecological isolating mechanisms including breeding and migration timing (allochrony), foraging habitat selection and natal philopatry provide a robust framework for explaining the divergence of seabird populations in both allopatry and sympatry (Friesen et al. Reference Friesen, Burg and McCoy2007a, Reference Friesen, Smith, Gomez-Diaz, Bolton, Furness, Gonzalez-Solis and Monteiro2007b, Boessenkool et al. Reference Boessenkool, Austin, Worthy, Scofield, Cooper, Seddon and Waters2009, Rayner et al. Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011, Wilmshurst et al. Reference Wilmshurst, Hunt, Lipo and Anderson2011, Friesen Reference Friesen2015). However, how these mechanisms contributed to a prehistoric phenotypic and genotypic cline across intermediate (locally) extinct populations of Cook’s Petrel requires further study.
In the meantime, when only a single locus, such as mtDNA, is used for phylogeographic reconstruction then the effects of selection, sex-biased dispersal or the stochastic nature of the coalescence process could result in a biased view of species history. Studies have shown that mtDNA is a useful marker, compared with nuclear markers, for detecting recent population differentiation in avian taxa because of its rapid coalescent time (Zink and Barrowclough Reference Zink and Barrowclough2008, Hung et al. Reference Hung, Drovetski and Zink2016). However, including additional independently evolving nuclear markers improves the power of molecular data to test phylogeographic hypotheses (Toews and Brelsford Reference Toews and Brelsford2012).
Taxonomic status of extant Cook’s Petrel populations
Typical of nuclear intron loci, extant Cook’s Petrels exhibit less phylogeographic structure for PAX and β-fibint7 than for the mitochondrial CO1 locus (Zink and Barrowclough Reference Zink and Barrowclough2008, Edwards and Bensch 2009). Similarly, it is typical for some nuclear introns, particularly β-fibint7 in birds, to exhibit more phylogeographic structure than others due to stochastic lineage sorting (Gangloff et al. Reference Gangloff, Zino, Gonzalez-Solis, Couloux and Bretangolle2012b, Morris-Pocock et al. Reference Morris-Pocock, Anderson and Friesen2016). Detecting relatively weak or no phylogeographic structure using nuclear markers may indicate male-mediated gene flow (e.g. Burg and Croxall Reference Burg and Croxall2001, but see Karl et al. Reference Karl, Toonen, Grant and Bowen2012), but there is limited evidence of male-biased dispersal in seabirds, particularly for Pterodroma petrels (Gangloff et al. Reference Gangloff, Zino, Gonzalez-Solis, Couloux and Bretangolle2012b, Iglesias-Vasquez et al. Reference Iglesias-Vasquez, Gangloff, Ruault, Ribout, Priddel, Carlile, Friesen, Cibois and Bretagnolle2017). Further, the differences observed in this study are consistent with the four-fold difference in effective population size between mitochondrial and nuclear markers (Zink and Barrowclough Reference Zink and Barrowclough2008, Edwards and Bensch 2009). Thus, the β-fibint7 and CO1 networks, combined with respective estimates of moderate to strong population genetic structure, are congruent with Rayner et al. (Reference Rayner, Carragher and Hauber2010a, Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011) and indicate that the two extant populations of Cook’s Petrel are genetically isolated and thus warrant taxonomic revision.
Taxonomic justification
The current analysis complements the morphological, behavioural, and ecological traits outlined in the introduction that separate the extant Cook’s Petrel populations (Imber et al. Reference Imber, West and Cooper2003, Rayner et al. Reference Rayner, Parker and Imber2008a, Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011). The extent of these differences is comparable or greater than levels of differentiation between seabird lineages considered to be distinct subspecies (Bandelt et al. Reference Bandelt, Forster and Rohl1999, Brooke and Rowe Reference Brooke and Rowe1996, Allentoft et al. Reference Allentoft, Collins, Harker, Haile, Oskam, Hale, Campos, Samaniego, Gilbert, Willerslev, Zhang, Scofield, Holdaway and Bunce2012, Welch et al. Reference Welch, Wiley, James, Ostrom, Stafford and Fleischer2012b, Iglesias-Vasquez et al. Reference Iglesias-Vasquez, Gangloff, Ruault, Ribout, Priddel, Carlile, Friesen, Cibois and Bretagnolle2017). For example in comparison with our study, Iglesias-Vasquez et al. (Reference Iglesias-Vasquez, Gangloff, Ruault, Ribout, Priddel, Carlile, Friesen, Cibois and Bretagnolle2017) found less genetic differentiation between the two subspecies of Australian and New Caledonian Gould’s Petrel Pterodroma leucoptera for the CO1 locus and similar levels of differentiation for nuclear loci. These results suggest the two extant Cook’s Petrel populations are sufficiently divergent to warrant recognition as distinct subspecies.
Accordingly, we recognise CDF breeding Cook’s Petrels as belonging to the taxon P. cookii orientalis (Murphy, Reference Murphy1929). Murphy (Reference Murphy1929) described this taxon based on body size and plumage characteristics of birds collected off the coast of Peru, which is now known to be within the core of the CDF Cook’s Petrel non-breeding habitat (Rayner et al. Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011). Moreover, Rayner et al. (Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011) generated CO1 sequences from several of these specimens held in the American Museum of Natural History (AMNH), including a paratype (AMNH 446060). All specimens, including AMNH 446060, were collected at the same time and place as the holotype, all were identified by the same expert (Rollo Beck), and share the same ‘diagnostic’ SNP identified as southern Cook’s Petrel in Rayner et al. (Reference Patel, Waugh, Millar and Lambert2010a). Based on these combined data, AMNH declined our request to destructively sample the holotype. We advocate that LBI breeding Cook’s Petrel should be recognised as P. cookii cookii (G. R. Gray, Reference Gray1843), as per the results of Matthews (Reference Matthews1912). This taxon was originally described by George Gray from the holotype (which has a haplotype typical of the northern birds) that has recently been shown (Medway Reference Medway2004) to have been collected at the Mangaoraka Stream near Kaimiro, north Taranaki, New Zealand (see Figure 1).
Taxonomic recommendation
ORDER: Procellariiformes Fürbringer, 1888
FAMILY Procellaridae Leach, 1820
GENUS Pterodroma Bonaparte, 1856
Pterodroma cookii cookii Gray, Reference Gray1843
Procellaria cooki Gray, Reference Gray1843 in Dieffenbach, Travels in New Zealand 2:199. Type locality: Mangaoraka Stream near Kaimiro, North Taranaki (fide Medway Reference Medway2004).
Formerly bred throughout the mountains of the northern North Island of New Zealand. Breeding now restricted naturally to Little Barrier Island with mainly unsuccessful breeding attempts on Great Barrier Island in the presence of introduced mammalian predators. Outcome of two translocations of chicks to the eastern north Island (Hawkes Bay) is awaiting confirmation. Shown to spend the non-breeding season in the north-eastern South Pacific Ocean with arrival on breeding grounds in late September and egg-laying in November (Rayner et al. Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011).
Pterodroma cookii orientalis Murphy, Reference Murphy1929
Pterodroma cookii orientalis Murphy, Reference Murphy1929: American Museum Novitiates 370: 5 Type locality: 200 miles west of Callao, Peru.
Formerly bred throughout the mountains of the southern North Island and the South Island of New Zealand and Stewart Island and its surrounding islets. Breeding now restricted naturally to Codfish Island. Shown to spend the non-breeding season in the central-eastern South Pacific Ocean arriving on breeding grounds in November and laying in December (Rayner et al. Reference Rayner, Hauber, Steeves, Lawrence, Thompson, Sagar, Bury, Landers, Phillips, Ranjard and Shaffer2011).
Conservation implications
Despite apparent past geographic overlap on the North Island (not taking age of fossil sites into consideration), available genetic and non-genetic data support the subspecies status of the extant populations of Cook’s Petrel which has implications for conservation management. There are currently management efforts attempting, or planning to attempt, the restoration of Cook’s Petrel populations on the New Zealand mainland to gain ecological benefits of seabird based nutrient subsidies. To date, these efforts, which include chick translocations and acoustic attraction, have been focused on the larger LBI population as a source population for geographically proximate translocation sites as recommended by Rayner et al. (Reference Rayner, Carragher and Hauber2010a). However, we argue that all available data, genetic and non-genetic, should be taken into consideration when planning future mainland translocations. We acknowledge that some might suggest that individuals should be sourced from both CDF and LBI for future translocation to the North Island given that the genetic data in this study show historic co-occurrence of northern and southern Cook’s Petrel mitochondrial lineages on the North Island. However, we argue that sourcing individuals from two populations that exhibit substantially different migratory behaviour and asynchronous breeding times/seasons would introduce a risk of translocation failure and future translocations to the North Island should continue to be sourced from Cook’s Petrel from LBI only. Accordingly, these data advocate any future translocation to the South Island be sourced from Cook’s petrel from CDF only to maximise recovery of this relict population.
Acknowledgements
We acknowledge mana whenua for Te Hauturu-o-Toi/Little Barrier Island, Ngāti Manuhiri, and Whenua Hou/Codfish Island, Ngāi Tahu, who are kaitiaki for Cook’s Petrel. We thank Mireia Casas Marcé, Jayne Houston, Selina Patel and Maggie Tisch for technical advice and support during this research and the Natural History Museum (Tring) for supplying the toepad sample of the Pterodroma cookii holotype. Funding for this research was provided by the Brian Mason Scientific Technical Trust and the University of Canterbury. LDS acknowledges support of the Royal Society of New Zealand Rutherford Discovery Fellowship. MJR was supported by the University of Auckland Vice Chancellor’s Faculty Research Development Fund during this research and acknowledges support of Wendy Rayner during its completion. This research was conducted under permit from the New Zealand Department of Conservation (permit numbers AK-14677-RES and SO-17501-LND). We thank the four anonymous reviewers whose comments improved this manuscript.
Supplementary Materials
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0959270920000350.
