Molecular Phylogenetics and Evolution xxx (2018) xxx-xxx

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier.com

Ancient DNA of crested : Testing for temporal genetic shifts in the world’s most diverse clade

Theresa L. Colea⁠ ,⁠ b⁠ ,⁠ ⁎⁠ , Nicolas J. Rawlencea⁠ , Nicolas Dussexc⁠ ,⁠ d⁠ , Ursula Ellenberge⁠ ,⁠ f⁠ , David M. Houstong⁠ , Thomas Matterna⁠ ,⁠ f⁠ , Colin M. Miskellyh⁠ , Kyle W. Morrisoni⁠ , R. Paul Scofieldj⁠ , Alan J.D. Tennysonh⁠ , David R. Thompsonk⁠ , Jamie R. Woodb⁠ , Jonathan M. Watersa⁠ a Department of , University of Otago, PO Box 56, 9054, b Manaaki Whenua Landcare Research, PO Box 69040, Lincoln, Canterbury 7640, New Zealand c Department of Bioinformatics and Genetics, Swedish Museum of Natural History, Box 50007, Stockholm, Sweden PROOF d Department of Anatomy, University of Otago, PO Box 56, Dunedin 9054, New Zealand e Department of Ecology, Environment and Evolution, La Trobe University, Melbourne, f Global Penguin Society, University of Washington, Seattle, USA g Group, Department of Conservation, Auckland, New Zealand h Museum of New Zealand Te Papa Tongarewa, PO Box 467, 6140, New Zealand i 964 Oriole Dr, Peterborough, Ontario K9H 6K7, Canada j Canterbury Museum, Rolleston Avenue, 8001, New Zealand k National Institute of Water and Atmospheric Research Ltd., Private Bag 14901, Kilbirnie, Wellington 6241, New Zealand

ARTICLE INFO ABSTRACT

Keywords: Human impacts have substantially reduced avian biodiversity in many parts of the world, particularly on iso- Ancient DNA lated islands of the Pacific Ocean. The New Zealand archipelago, including its five subantarctic island groups, Eudyptes holds breeding grounds for a third of the world’s penguin , including several representatives of the diverse crested penguin Eudyptes. While this species-rich genus has been little studied genetically, recent pop- ulation estimates indicate that several Eudyptes taxa are experiencing demographic declines. Although crested Demographic history Genetic diversity penguins are currently limited to southern regions of the New Zealand archipelago, prehistoric and archae- ological deposits suggest a wider distribution during prehistoric times, with breeding ranges perhaps extending to the North Island. Here, we analyse ancient, historic and modern DNA sequences to explore two hypotheses regarding the recent history of Eudyptes in New Zealand, testing for (1) human-driven of Eudyptes lineages; and (2) reduced genetic diversity in surviving lineages. From 83 prehistoric bone samples, each ten- tatively identified as ‘Eudyptes spp.’, we genetically identified six prehistoric penguin taxa from mainland New Zealand, including one previously undescribed genetic lineage. Moreover, our Bayesian coalescent analyses indi- cated that, while the range of Fiordland crested penguin (E. pachyrhynchus) may have contracted markedly over the last millennium, genetic DNA diversity within this lineage has remained relatively constant. This result con- trasts with human-driven biodiversity reductions previously detected in several New Zealand coastal vertebrate taxa.

1. Introduction larly those endemic to the Pacific Islands (Ceballos et al., 2015), in- cluding New Zealand (Worthy and Holdaway, 2002). The unique evo- Humans have had major impacts on island biota (Duncan et al., lutionary ’s vertebrate biota has long been the 2013; Wood et al., 2017). , habitat loss and the introduction subject of particular scientific intrigue (Fleming 1979, Diamond 1990). of invasive predators accompanied human settlement and caused de- With prolonged geographic isolation, and in the absence of mammalian clines and of many island species (Spatz et al., 2017; Wood predators, many native avian taxa evolved distinctive ecological adap- et al., 2017), particu tations and lost anti-predator defences. New Zealand was the world’s last major landmass to be settled by humans, with Polynesian arrival UNCORRECTEDapproximately 750 ago (Wilmshurst et

⁎ Corresponding author at: Department of Zoology, University of Otago, PO Box 56, Dunedin 9054, New Zealand. Email address: [email protected] (T.L. Cole) https://doi.org/10.1016/j.ympev.2018.10.025 Received 29 September 2018; Received in revised form 19 October 2018; Accepted 19 October 2018 Available online xxx 1055-7903/ © 2018. T.L. Cole et al. Molecular Phylogenetics and Evolution xxx (2018) xxx-xxx al., 2008, 2011; Anderson 2017), followed 550 years later by Euro- Chilvers, 2014). In contrast, Warham (1974), Mattern (2013) and pean colonisation. Subsequently, New Zealand’s biota has experienced a Hiscock and Chilvers, (2016) suggested that E. robustus populations well-documented decline that is clearly associated with human activities have remained stable. By contrast, the extent to which any such demo- (Worthy 1999, Worthy and Holdaway, 2002) over a period of relative graphic shifts might have impacted the genetic composition of Eudyptes climatic stability (Wanner et al., 2008; Rawlence et al., 2016; Waters et populations remains unclear. al., 2017). As a result of hunting (Rawlence et al., 2015a), rapid land- Here, we investigate the prehistoric distribution and population his- scape modification (McWethy et al., 2010) and the introduction of ex- tory of Eudyptes species in New Zealand by comparing genetic informa- otic species that followed both colonisation events, a substantial propor- tion derived from prehistoric bones, historical museum , and blood tion of New Zealand’s endemic vertebrates have become extinct. Those from extant populations. Specifically, we test two hypotheses regarding species unable to fly, including many , were particularly suscep- the effects of human colonisation on Eudyptes diversity. Specifically, we tible to these rapid ecological changes. Since initial human settlement, test for: (1) human-driven extinction of Eudyptes lineages; and (2) tem- 61 endemic birds have gone extinct (Tennyson and Martinson, 2007; poral declines in genetic diversity within extant Eudyptes lineages. Boessenkool et al., 2009; Rawlence et al., 2017a,b), and 71 are now threatened with extinction, while an additional 23 taxa are classified as Nationally Critical (Robertson et al., 2017). 2. Materials and methods Recently, a number of studies have combined ancient and contem- porary DNA with morphological analyses and , to ex- 2.1. Source of specimens plore the impact of human settlement on population size in a range of New Zealand taxa (see Cole & Wood, 2017), including sea lions Sub-fossil and archaeological bones (n = 83) tentatively identified (Phocarctos hookeri; Collins et al., 2014; Rawlence et al., 2016), fur seals as Eudyptes spp. based on size, and historical museum skins (n = 27 (Arctocephalus spp; Salis et al., 2016), (Apteryx spp; Shepherd and and n = 21 sequences obtained from GenBank (Tables S1 and S2)) were Lambert, 2008; Shepherd et al., 2012), (Cygnus spp; Rawlence sourced from collectionsPROOFat the Museum of New Zealand Te Papa Ton- et al., 2017a), kākāpō (Strigops habroptilus; Bergner et al., 2016; Dussex garewa, Canterbury Museum, Auckland Museum and the Australian Na- et al., 2018), (Nestor notabilis; Dussex et al., 2015), shags (Leuco- tional Wildlife Collection. To ensure independence of prehistoric sam- carbo spp; Rawlence et al., 2015b, 2017b) and penguins ( ples, only common elements of the left or right orientation were sam- and spp.; Boessenkool et al., 2009; Rawlence et al., 2015a; pled from an individual deposit, or else bones were sampled from differ- Grosser et al., 2015, 2016, 2017). Several of these analyses have dis- ent stratigraphic units within a site. In addition, whole blood or tissue covered cryptic extinctions and population declines in coastal and ma- (n = 87) was collected from wild individuals across the breeding range rine fauna following human settlement, and subsequent range expan- of E. pachyrhynchus, E. robustus and E. sclateri (except the Bounty Is- sions of congeneric species. For example, the presence of bones attrib- lands) (Table S2). Our final sample size consisted of 268 specimens (in- uted to yellow-eyed penguin (Megadyptes antipodes) in natural and ar- cluding additional specimens obtained from GenBank) collected across chaeological bone deposits around the New Zealand coast led to the New Zealand and the subantarctic region, over a temporal scale encom- assumption that this species had been present on the mainland before passing Polynesian and European arrival (contemporary to > 1000 years human settlement. By combining ancient and contemporary DNA with BP). Ages were determined based on stratigraphic context (archaeologi- morphological analyses, Boessenkool et al. (2009) overturned this hy- cal details and radiocarbon dates provided in Table S2). pothesis, revealing that a previously unknown species of Megadyptes, the (M. waitaha) had inhabited New Zealand coast prior 2.2. DNA extraction, amplification and sequencing to human arrival. Following human settlement M. waitaha was likely hunted to extinction, and was subsequently rapidly replaced by an ex- Genomic DNA was extracted following several techniques, optimised panded population of M. antipodes from the subantarctic (Rawlence et for the sample type used. Ancient DNA extraction and Polymerase Chain al., 2015a). Moreover, Grosser et al., (2015, 2017) identified that two Reaction (PCR) set-up for sub-fossil bones, archaeological bones and species of now breed around the New Zealand coast; the historical museum skins was carried out in two purpose-built ancient endemic Eudyptula minor and the Australian E. novaehollandiae. Grosser DNA laboratories that were physically isolated from other molecular et al., (2016) demonstrated that the presence of E. novaehollandiae in laboratories, following rigorous ancient DNA protocols (as advocated by New Zealand represents a cryptic colonisation event from Australia, that Cooper and Poinar, 2000). DNA extractions were mostly performed in probably followed the localised extirpation of the New Zealand lineage. the Otago Palaeogenetics Laboratory at the Department of Zoology (Uni- However, little is known about whether similar extinction/colonisation versity of Otago, Dunedin) following Rohland et al, (2010) for sub-fos- events occurred within prehistoric crested penguins (Eudyptes spp.) in sil/archaeological material and Rawlence et al. (2015b) for historical New Zealand, even though skeletal remains belonging to this genus have museum skins. Three prehistoric specimens from Hunter Island, Aus- been collected from natural and subfossil and mid- tralia, were extracted at Manaaki Whenua Landcare Research (Lincoln) den deposits around New Zealand (Scarlett 1979; Leach 1979; Millener following Thomson et al., (2014), as outlined in Cole et al., (2017). DNA 1990; Worthy 1997). from blood was extracted following the Qiagen DNeasy® Tissue Extrac- New Zealand provides important breeding grounds for half of the tion Kit (Qiagen Inc., Chatsworth, California, USA) at Manaaki Whenua world’s Eudyptes penguins. Endemic Eudyptes include the erect-crested Landcare Research. penguin (E. sclateri), which breeds on the Antipodes and Bounty Is- To verify and assign the historical museum skins and sub-fossil/ar- land groups, the Snares crested penguin (E. robustus), on the Snares Is- chaeological bones to a species we followed the methods described by lands (including the Western Chain islets); and the Fiordland crested Cole et al., (2017) to amplify up to 500 bp of the mitochondrial cy- penguin, (Tawaki; E. pachyrhynchus) which breeds in dense forest along tochrome oxidase 1 (COI), using four overlapping primer pairs, appro- the coast of southern Westland and Fiordland (), Stew- priate for ancient DNA. To supplement the dataset, we sequenced COI art Island and smaller offshoreUNCORRECTEDislands (see Fig. S1). Several studies from two samples of M. waitaha and three of M. antipodes, which have have demonstrated major declines in crested penguins during the late previously been differentiated based on their mitochondrial control re- 20th century (Crawford et al., 2009, 2010; Davis 2013; Morrison et gion (CR) sequences. As the COI markers could only amplify 1-2 frag- al., 2015; Otley et al., 2018). Taylor (2000) and Otley et al. (2018) ments for Megadyptes spp., we assigned samples that were identified suggest E. pachyrhynchus populations have declined significantly since as Megadyptes spp. to species level using CR primers of Boessenkool this time. Surveys of E. sclateri suggest that populations were stable et al. (2009). To test the demographic history of Eudyptes spp., we until 2011, but have subsequently declined by 19% (Hiscock and also designed Eudyptes specific primers (Eud_CR2F 5′GGGTTTAGTC- CTCTCTATTGGG and Eud_CR1R 5′GCTTACTGCTGTGTTTCAGGGA)

2 T.L. Cole et al. Molecular Phylogenetics and Evolution xxx (2018) xxx-xxx to amplify a 131 bp CR fragment that contains 23 variable sites, 16 2.5. Estimating temporal fluctuations of effective population size of which are parsimoniously informative, within Eudyptes (E. pachyrhynchus n = 12 variable/n = 6 informative; E. robustus n = 7 vari- We reconstructed the coalescent history of E. pachyrhynchus by cre- able/n = 3 informative; E. sclateri n = 10 variable/n = 7 informative). ating a Bayesian Skyline Plot (Drummond et al., 2005) in BEAST 2. Tip PCRs (12.5 μL) were performed on sub-fossil/archaeological and histor- dates, in mean calibrated years BP (where present is 2016) were deter- ical material using 2 mg/mL BSA (Sigma), 1 x PCR buffer, 2 mM mined for all samples based on (1) dates of associated biological ma-

MgSO4⁠ , 80 μM dNTP, 0.4 μM each primer, 0.625 U HiFi Platinum Taq terial from the same archaeological sites, excavation squares or strati- (Invitrogen) and 1 μL DNA extract on a BIO-RAD MyCycler thermal cy- graphic units, (2) associated faunal/cultural assemblages, and (3) col- cler with an initial denaturation of 94 °C for 3 min, followed by 55 cy- lection date for historical skins (see Rawlence et al., 2015a,b, Waters et cles of 94 °C for 30 s, 55 °C for 30 s and 68 °C for 45 s, and a final 10 al., 2017; Table S2) We removed the sample AD357 (from North Cape, extension at 68 °C for 10 min. PCRs for amplifying CR from blood were Northland) from the analysis because we were not confident about its performed in 1 x PCR Buffer, 250 μM dNTP, 1.25 U i-Taq (iNtRON), calibrated age (between ca. 6000 BP to 1450 CE) and collection locality 0.25 μM each primer, and 1uL DNA extract, following the same PCR (an outlier to all other locations). We found no phylogeographic struc- ture within E. pachyrhynchus (see Fig. S2), so the assumption of popula- conditions while lowering the number of cycles to 42 and raising the an- tion panmixia was not violated (Drummond et al., 2005). We estimated nealing temperature to 58 °C. PCR products were purified using SPRIs- the timing of coalescent events of E. pachyrhynchus using a constant, ex- elect (Beckman Coulter, Inc., Indianapolis, IN, USA) or exoSAP-ITTM⁠ ponential increase, Bayesian Skyline coalescent prior in BEAST2 using (ThermoFisher Scientific), for sub-fossil/archaeological/historical mu- the Jukes Cantor model of nucleotide substitution and a strict molecu- seum skins and blood samples respectively, and sequenced at the Man- lar clock model of evolution. Each run was done in triplicate, consisting aaki Whenua Landcare Research sequencing facility (Auckland) on an of 50 million MCMC generations, sampling tree parameters every 1000 Applied Biosystems 3500xL Genetic Analyzer. generations, with a pre-burn-in of 10 percent. An extended Bayesian Contiguous sequences of both COI and CR were constructed and Skyline coalescent prior was also attempted, extending the MCMC gen- aligned using Geneious 8.1.8 (Biomatters) from high quality bi-direc- PROOF erations to 100, 150, 300 and 500 million, however for those analyses tional sequence reads and checked by eye. Due to post-mortem DNA the ESS values did not converge. See supplementary information for de- damage, when inconsistency between sequences from a given individual tails of the subsequent model comparisons performed on the data. was observed (e.g. G-A and C-T transitions), additional PCRs and bidi- rectional sequencing were conducted, and a majority rule consensus was 2.6. Testing demographic scenarios applied (Brotherton et al., 2007).

We used an Approximate Bayesian Computation (ABC) approach (Beaumont et al., 2002) implemented in DIYABC 2.1.0 (Cornuet et al., 2.3. Phylogenetic analysis 2014) to test for population bottlenecks and estimate values of key demographic parameters. Specifically, we designed models to test for To verify and assign the samples to a species, we created two the potential demographic declines in Eudyptes pachyrhynchus associated Bayesian maximum credibility phylogenies using COI, and one Bayesian with post-glacial climate change, and human arrival in New Zealand. maximum credibility phylogeny using Megadyptes specific CR. As is We assumed a generation time of 5 years based on the age of first breed- typical with ancient DNA, our data contained some partial sections ing (Otley et al., 2017) and tested four models: (1) a model of con- (i.e. missing sequence data). We therefore constructed the first phy- stant population size since the Holocene with N uniformly distributed logeny which contained all COI data, and a second phylogeny that con- e⁠ through time (0–3000 generations; 0–15,000 years BP), (2) a model tained only samples with 3-4 of the 4 overlapping fragments. In addi- tion, we included one sequence representative of all extant sphenisci- of Holocene decline (2000–3000 generations; 10,000–15,000 years BP), form species, and wandering (Diomedia exulans) as outgroup (3) a model of human-induced (2–160 generations, 10–800 years BP) from GenBank. For the Megadyptes phylogeny, we obtained an addi- decline and (4) a model of Holocene (2000–3000 generations; tional 15 and 17 M. waitaha and M. antipodes sequences, and one E. 10,000–15,000 years BP) and human-induced (2–160 generations, chrysocome as outgroup from GenBank (Table S1). In addition, we cre- 10–800 years BP) declines (Fig. S3). Recent surveys suggest that the ated a CR phylogeny to test for phylogeographic structure among E. current population numbers are at least 5500–7000 mature individu- pachyrhynchus samples. All phylogenetic trees were created using BEAST als (BirdLife International, 2016), and there is no prior information on 2.4.7 (Bouckaert et al., 2014), with a relaxed log-normal clock for COI pre-human population size in the species available. We therefore used and a strict clock for CR, with a yule model of speciation, for 100 mil- wide priors (Table S3) for pre-glaciation, pre-human and modern female lion MCMC generations, sampling tree parameters every 1000 genera- effective population size (Nef⁠ ). For this reason, we set up as condition tions with a burn-in of 10 percent. Each tree was replicated in triplicate of the ABC analysis the pre-glaciation and pre-human Nef⁠ to be larger and combined using Log Combiner. We implemented the Akaike Infor- than the modern Nef⁠ in all the scenarios tested in to simulate a mation Criterion in JModelTest2 (Darriba et al., 2012) to determine the population bottleneck (constrains applied to population effective size: most appropriate model (Jukes Cantor for all genetic markers). Nef⁠ modern < Nef⁠ pre-glaciation, and Nef⁠ modern < Nef⁠ pre-human). For each model, 1 million datasets were simulated with the defined demo- graphic and marker parameters (Table S3). Samples were grouped into 2.4. Temporal haplotype network three temporal samples corresponding to a modern (0–50 years BP), a historical (50–100 years BP) and a prehistorical period (500–1000 years We created temporal haplotype networks using CR for E. BP). In our models, the collection times of these samples were thus set at pachyrhynchus, E. sclateri and E. robustus, by employing the TempNet t = 0, t = 20 and t = 200 generations. We used a Jukes Cantor substitu- R script (Prost and Anderson, 2011) to visualise haplotypic changes tion model with a mean rate uniformly distributed between 1.00 × 10−⁠ 8 over a temporal scale (sub-UNCORRECTEDfossil/archaeological bones, historical mu- −⁠ 7 and 1.00 × 10 substitutions/generations. For summary statistics, we seum skins and blood). We amplified 111 bp of CR from a Eudyptes taxon used the number of haplotypes, number of segregating sites, mean of which fell separate to E. pachyrhynchus, E. sclateri and E. robustus, so pairwise differences, and Tajima’s D. we created additional networks that included a representative of all Eu- For model selection and parameter estimation, normalized Euclid- dyptes species, except the Northern rockhopper penguin (E. moseleyi) be- ean distances were calculated between the observed dataset and each cause no sequence was available. of the simulated datasets using the local linear regression method of Beaumont et al.

3 T.L. Cole et al. Molecular Phylogenetics and Evolution xxx (2018) xxx-xxx

(2002). The posterior probabilities of each model were estimated by tak- gions, a subset of sub-fossil/archaeological specimens failed to amplify ing a logistic regression approach on the closest 1% of data sets to the for one or the other region (n = 13 amplified COI but not CR, and n = 4 observed data, providing both point estimated and 95% confidence in- amplified CR but not COI; details in Table S2). In total, the data ob- tervals (Cornuet et al., 2010; Fagundes et al., 2007). Posterior parame- tained in our study captured a total of 58 new COI sequences, 145 Eu- ter distribution was estimated by retaining the 10,000 datasets for the dyptes CR sequences and 9 new Megadyptes CR sequences (Table S2). All preferred model (1%) with the smallest Euclidean distances to the ob- sequences were deposited on GenBank (details provided in Table S2). served dataset. To check model performance, we first empirically evalu- ated the power of the model to discriminate among models (confidence in model choice). We simulated pseudo-observed data sets with para- 3.1. Phylogenetic relationships meters drawn from the posterior parameter distribution of the preferred model and positioned the summary statistics of the observed data in the We amplified COI from 41 sub-fossil/archaeological bones that had summary statistic distribution of these pseudo-observed data. The model previously been tentatively identified as Eudyptes or Megadyptes based was then considered suitable if the observed data summary statistics on morphological assessment (Figs. 1 and S4) (including five known were included in the confidence interval drawn from pseudo-observed Megadyptes penguins). Almost a third of the COI sequences (n = 16) data. Secondly, we calculated statistical measures of performance and fell within Eudyptes’ sister genus Megadyptes, although the posterior Type I and Type II error rates as a means of model checking (Cornuet et probabilities for these phylogenetic assignments using COI was some- al., 2010; Excoffier et al., 2005). times low, possibly due to some specimens having limited amplifica- tion success (average 175 bp) and/or the relatively small genetic di- 3. Results vergence values among some of these lineages. Use of Megadyptes-spe- cific CR primers similarly identified nine of these samples as repre- We successfully amplified COI and/or CR fragments from all blood senting the extinct Waitaha penguin (M. waitaha) (Fig. S5). We could samples and historical museum skins. Specifically, from a total of 83 not amplify CR for thePROOFremaining seven samples so we only tenta- sub-fossil/archaeological bones we amplified partial COI fragments tively refer to them as Megadyptes spp. When we removed the short se- from 41 samples, Eudyptes specific CR from 22 samples, and Megadyptes quences (including all but two attributed to Megadyptes), the COI se- specific CR from 9 samples. While the majority of samples (n = 32, quences successfully differentiated every Eudyptes species (except the E. including those obtained from GenBank (see Tables S1 and S2)) am- schlegeli/E. chrysolophus complex), including the occasionally merged E. plified for both mitochondrial re

UNCORRECTED

Fig. 1. Bayesian phylogenetic tree of up to 500 bp mitochondrial COI gene, of all extant sphenisciform taxa. Eudyptes robustus samples from the Western Chain are indicated with WC. AD numbers with the species name indicated were from historical museum skins, whereas samples without the species name in bold were prehistoric sub-fossil or archaeological bone. The sample size has been reduced from Fig. S4 to include only samples with 3–4 of 4 COI sequence fragments.

4 T.L. Cole et al. Molecular Phylogenetics and Evolution xxx (2018) xxx-xxx pachyrhynchus/E. robustus (Fig. 1) (Christidis and Boles, 2008). Of the When testing for competing demographic scenarios for E. remaining 25 samples, one is potentially attributable to E. sclateri, E. pachyrhynchus, a model of constant population size over the last mil- schlegeli or E. chrysolophus, one is either E. schlegeli or E. chrysolophus, lennium was supported with a posterior probability of 67% (95% CI: three are the erect-crested penguin (E. sclateri), two are the Snares 0.66–0.69) using the logistic regression approach (Tables S3 and S4). crested penguin (E. robustus), and 11 are the Fiordland crested penguin This finding is consistent with the Bayesian Skyline Plot (Fig. S8). By (E. pachyrhynchus) (Figs. 1 and 2; see Figs. S1 and S4 for alternative contrast, alternative models of post-glacial, human and or post-glacial placements of AD258). Seven samples formed a separate clade within and human-induced declines obtained a posterior probability of 10%, Eudyptes, that is sister to E. sclateri, yet contains no extant samples (Figs. 13% and 10%, respectively. The observed data summary statistics for 1 and 2, and S4). We refer to this hereafter as Eudyptes clade X. the preferred model were included in the confidence interval drawn from pseudo-observed data (Fig. S9). Type I and II errors were of 0.053 and 0.14, respectively for the preferred model of constant population 3.2. Population declines and range contractions size.

We amplified CR in E. robustus (n = 31; contemporary n = 21, his- 4. Discussion toric n = 7, prehistoric n = 3), E. sclateri (n = 30; contemporary n = 18, historic n = 10, prehistoric n = 2), E. pachyrhynchus (n = 72; contem- The present study presents genetic evidence for the formerly wide- porary n = 34, historic n = 23, prehistoric n = 15) and Eudyptes clade spread distribution of Eudyptes pachyrhynchus throughout mainland New X (n = 1; 111 bp). Snares Island and Western Chain islets E. robustus Zealand, which is consistent with previous morphological studies populations (which breed six weeks apart [Sagar 1971; Miskelly et al., (Worthy 1997). This is in contrast to our coalescent models, which de- 1997]) were genetically indistinguishable from one another. Moreover, tected little change in population size of the species, despite the ap- our data revealed only six haplotypes in E. robustus, but indicate there parent range reductions. Additionally, the detection of a previously un- may have been no loss of genetic diversity in these species through time. known Eudyptes clade inPROOFprehistoric New Zealand raises further ques- Our data revealed only ten haplotypes in E. sclateri, and there was no tions about the recent evolutionary history of this clade. loss of genetic diversity in this species through time. Of the 12 hap- lotypes that were observed in E. pachyrhynchus, the temporal network 4.1. Prehistoric penguin fauna (Figs. 3, and S6 and S7) showed only a slight downward trend of ge- netic diversity over the last 1151 years; sub-fossil/archaeological bones Our study genetically confirms the presence six large (Eudyptes and (9 haplotypes), historical museum specimens (6 haplotypes) and con- Megadyptes) penguin taxa from mainland New Zealand’s prehistoric temporary blood (6 haplotypes). We did not find any phylogeographic record (see Worthy, 1997), a result which contrasts with the pres- structure when comparing CR sequences across E. pachyrhynchus speci- ence of just two contemporary breeding mainland representatives of mens (Fig. S2). these genera. The novel finding of a previously unknown (and pre- sumed extinct) clade of Eudyptes (Eudyptes clade X, represented by 7 archaeological and subfossil samples)

UNCORRECTED

Fig. 2. Penguin taxa that were identified from prehistoric New Zealand natural and midden sites using mitochondrial COI and CR markers. The map labelled ‘Unidentified’ indicates those samples that could not be amplified. Megadyptes samples in grey could not be identified to species level, so could be M. waitaha or M. antipodes. Megadyptes identification is based on CR.

5 T.L. Cole et al. Molecular Phylogenetics and Evolution xxx (2018) xxx-xxx

(Waters et al., 2017) may explain why E. pachyrhynchus has persisted in the southern mainland. While our genetic data suggest that E. pachyrhynchus experienced anthropogenic range reductions in mainland New Zealand, alternative possibilities exist. The prehistoric presence of Eudyptes beyond its cur- rent range could instead represent occasional vagrant indi- viduals, rather than northern breeding colonies (Keogan et al., 2018). Moreover, Cole et al, (2018) suggested there may be a bias towards large vagrant species in midden deposits, as such individuals may have attracted the attention of hunters and be more docile and easier to catch (e.g. Bried 2003; Lees and Gilroy, 2009). Moreover, trading of penguins by Māori is unlikely, as no South Island endemic avian species has been found in fossil or archaeological deposits in the North Island (e.g. Worthy 1997, 1998; Scofield et al. 2003), compared to cultural items (e.g. Davidson, 1988). Nevertheless, by the time Europeans ar- rived in New Zealand, E. pachyrhynchus was breeding only in the south- west of the South Island, an isolated region that has remained relatively inaccessible, with Māori abandoning the area following the extinction of terrestrial and coastal (Waters et al., 2017). These factors may help explain the apparently temporally stable genetic history of this Fig. 3. Temporal haplotype network for three Eudyptes species based on CR sequences. species. Coloured circles represent different haplotypes, with the size of the circle and number within the circle corresponding to the number of samples that had that haplotype. Small PROOF 4.3. Conservation implications for modern populations white circles represent those haplotypes that are present in one or two-time periods, but are absent in the time period displayed. Black circles represent haplotypes that were ab- sent from the dataset. Each layer represents a different time period, based on the sam- The findings of our study might be consistent with the suggestion of ple type used (Prehistoric = sub-fossil/archaeological bone, Historic = historical museum Mattern and Long (2017), who commented that E. pachyrhynchus popu- skins, Modern = contemporary blood). The coloured cylinders connect those haplotypes lations may not be declining as rapidly as previously indicated. Despite that are present through multiple time periods. Green circles are E. pachyrhynchus, yellow circles are E. robustus and orange circles are E. sclateri. our finding of little change in mitochondrial DNA diversity across sev- eral centuries of human occupation, ongoing monitoring and a thorough assessment of nuclear genetic diversity are critical steps to ensure future management and appropriate conservation decisions are made for Eu- provides intriguing new evidence of a cryptic coastal vertebrate extinc- dyptes penguins. tion in the New Zealand region, potentially adding to the findings of In conclusion, our ancient genetic analysis confirms the presence recent ancient DNA studies (e.g. Megadyptes, Phocarctos and Leucocarbo of substantial penguin species diversity in prehistoric mainland New extinctions (Boessenkool et al., 2009; Collins et al., 2014; Rawlence et Zealand, including the finding of Eudyptes clade X that has no extant al., 2015a, 2016, 2017a,b, 2018)). It is unclear, however, whether these equivalents. Despite a possible range contraction, the findings also sug- unique Eudyptes specimens represented a mainland breeding colony gest that the extant mainland species E. pachyrhynchus has persisted (e.g. from northern South Island or the lower North Island), or were within its core range, with relatively stable population size, likely ow- vagrants from elsewhere. Notably, vagrant Eudyptes are not uncommon ing to its presence in habitats that have experienced relatively limited to New Zealand (Robertson et al., 2017), and we therefore also suspect human pressure. These findings highlight the potential for differential that the relatively ‘rare’ mainland records of E. sclateri, E. robustus, and human impacts across species ranges. E. schlegeli specimens, and the single E. pachyrhynchus from Northland, likely represent vagrants. Financial support

This work was supported by an Australasian Group grant 4.2. No loss of genetic diversity despite range contractions and Manaaki Whenua Landcare Research. TLC was supported by an Otago Postgraduate Scholarship. Our data supports a model of constant population size in E. pachyrhynchus prior to human arrival in New Zealand, which suggests Declaration of interest that the population has remained relatively stable, at least over the last millennium. Based on our analyses, it seems unlikely that hunting by None. early Polynesian settlers had detectable species-wide effects on genetic diversity. These findings strongly contrast with recent studies on New Acknowledgements Zealand coastal taxa such as the Otago shag (Rawlence et al., 2015a,b) and New Zealand fur seals (Salis et al., 2016), which experienced both The authors thank Alastair Judkins ( Ocean Research Insti- geographic and genetic declines. Importantly, in cases where popula- tute) for providing a prehistoric penguin skull for analysis. Leo Joseph tions are genetically well connected, range contractions do not necessar- and Robert Palmer (Australian National Wildlife Collection), and Brian ily result in substantially reduced population sizes and genetic diversity Gill (Auckland War Memorial Museum) assisted with arranging sam- (as seen in Matocq & Villablanca, 2001, Hoffman & Blouin, 2004, and pling from museum collections. Fieldwork would not have been pos- Dussex et al. 2015). UNCORRECTEDsible without Hotte Mattern, Paul Sagar, Jo Hiscock, Chris Rickard, Based on these findings, we hypothesise that the core population Sarah Fraser, Kath Walker and Graeme Elliott. We are grateful to Tom of E. pachyrhynchus has remained stable in the remote southern and Hart, Gary Miller, Paul Nolan, Petra Quillfeldt and Lara Shepherd for southwestern regions of New Zealand over recent centuries. Similarly, providing blood samples that were used as the outgroups in Fig. S7, Rawlence et al. (2015a,b) and Salis et al. (2016) demonstrated that, and to Barbara Wienecke for tracking down sample information. We following extirpation of northern lineages, endemic genetic lineages of thank several anonymous reviewers and Gustavo Cabbane for improv- Leucocarbo shags and New Zealand fur seals have persisted in south- ing the manuscript. We consulted the Ngāi Tahu Research Consulta- ern New Zealand throughout human settlement. The relatively low tion Committee about this project. The research was approved by the human pressure in this remote southern region

6 T.L. Cole et al. Molecular Phylogenetics and Evolution xxx (2018) xxx-xxx

Otago University Animal Ethics Committee 61/2016 and conducted un- Dussex, N., von Seth, J., Robertson, B.C., Dalén, L., 2018. Full mitogenomes in the criti- cally endangered kākāpō reveal major post-glacial and anthropogenic effects on neu- der NZ Department of Conservation authority numbers: OT-25557-DOA tral genetic diversity. Genes 9, 220. (Bruce Robertson), 32202-FAU, 35682-FAU, 37312-LND, 50437-DOA Fleming, C.A., 1979. The Geological History of New Zealand and its Life. Auckland Uni- (David Thompson), 50436-FAU, 50464-DOA (Theresa Cole), and versity Press, Auckland, New Zealand, 141p. 38882-RES (Thomas Mattern and Philip Seddon). Excoffier, L., Estoup, A., Cornuet, J.M., 2005. Bayesian analysis of an admixture model with mutations and arbitrarily linked markers. Genetics 169, 1727–1738. Fagundes, N.J., Ray, N., Beaumont, M., Neuenschwander, S., Salzano, F.M., Bonatto, S.L., Excoffier, L., 2007. Statistical evaluation of alternative models of human evolution. Appendix A. Supplementary material Proc. Natl. Acad. Sci. USA 104, 17614–17619. Grosser, S., Burridge, C.P., Peucker, A.J., Waters, J.M., 2015. Coalescent modelling sug- gests recent secondary-contact of cryptic penguin species. PLoS ONE 10, e0144966. Supplementary data to this article can be found online at https:// Grosser, S., Rawlence, N.J., Anderson, C.N.K., Smith, I.W.G., Scofield, R.P., Waters, J.M., doi.org/10.1016/j.ympev.2018.10.025. 2016. Invader or resident? Ancient-DNA reveals rapid species turnover in New Zealand little penguins. Proc. R. Soc. B 283, 20152879. Grosser, S., Scofield, R.P., Waters, J., 2017. Multivariate skeletal analyses support a taxo- nomic distinction between New Zealand and Australian Eudyptula penguins (Sphenis- References ciformes: Spheniscidae). – Austral. Ornithol. 177, 176–283. Hiscock, J.A., Chilvers, L.B., 2014. Declining eastern rockhopper (Eudyptes filholi) and erect-crested (E. sclateri) penguins on the Antipodes Islands. N. Z. J. Ecol. 38, Anderson, A.J., 2017. Changing perspectives upon Māori colonisation voyaging. J. R. Soc. 124–131. N. Z. 47, 222–231. Hiscock, J.A., Chilvers, L.B., 2016. Snares crested penguins Eudyptes robustus population , . estimates 2000 2013. N. Z. J. Ecol. 40, 108 113. Beaumont, M.A., Zhang, W.Y., Balding, D.J., 2002. Approximate Bayesian computation in – – population genetics. Genetics 162, 2025–2035. Hoffman, E.A., Blouin, M.S., 2004. Historical data refute recent range contraction as cause of low genetic diversity in isolated frog populations. Mol. Ecol. 13, 271 276. Bergner, L.M., Dussex, N., Jamieson, I.G., Robertson, B.C., 2016. European colonization, – not Polynesian arrival, impacted population size and genetic diversity in the critically Keogan, K., Daunt, F., Wanless, S., Phillips, R.A., Walling, C.A., Agnew, P., Ainley, endangered New Zealand kākāpō. J. Hered. 107, 593–602. D.G., Anker-Nilssen, T., Ballard, G., Barrett, R.T., Barton, K.J., Bech, C., Becker, P., Berglund, P.-A., Bollache, L., Bond, A.L., Bouwhuis, S., Bradley, R.W., Burr, Z.M., BirdLife International, 2016. Eudyptes pachyrhynchus. IUCN Red List of Threatened Camphuysen, K., et al., 2018.PROOFGlobal phenological insensitivity to shifting ocean tem- Species. http://doi.org/10.2305/IUCN.UK.2016-3.RLTS. T22697776A93638571.en. peratures among . Nat. Clim. Change 8, 313 318. Boessenkool, S., Austin, J.A., Worthy, T.H., Scofield, P., Cooper, A., Seddon, P.J., Waters, – J.M., 2009. Relict or colonizer? Extinction and range expansion of penguins in south- Leach, B.F., 1979. In: Birds of a : Osteological and Archaeological Papers from the ern New Zealand. Proc. R. Soc. B 276, 815–821. South Pacific in Honour of R. J. Scarlett. New Zealand Archaeological Association Monograph II. BAR International Series 62. BAR, Oxford, pp. 103 122. Bouckaert, R., Heled, J., Kühnert, D., Vaughan, T., Wu, C.-H., Xie, D., Suchard, M.A., Ram- – baut, A., Drummond, A.J., 2014. BEAST 2: a software platform for Bayesian evolu- Lees, A.C., Gilroy, J.J., 2009. Vagrancy mechanisms in and near-passerines. In: tionary analysis. PLoS Comput. Biol. 10, e1003537. Slack, R. (Ed.), Rare birds, Where and When: An Analysis of Status and Distribution Bried, J., 2003. Impact of vagrant predators on the native fauna: a short-eared (Asio in Britain and Ireland. Volume 1: to New World Orioles. Rare Bird Books, flammeus) preying on Madeiran storm petrels (Oceanodroma castro) in the Azores. York, pp. 3–4. Life Mar. Sci. 20A, 57–60. , . , Brotherton, P., Endicott, P., Sanchez, J.J., Beaumont, M., Barnett, R., Austin, J., Cooper, . A., 2007. Novel high-resolution characterisation of ancient DNA reveals C > U-type , . , base modification events as the sole cause of post mortem miscoding lesions. Nucl. Matocq, M.D., Villablanca, F.X., 2001. Low genetic diversity in an endangered species: re- Acids Res. 35, 5717–5728. cent or historic pattern?. Biol. Conserv. 98, 61–68. Ceballos, G., Ehrlich, P.R., Barnosky, A.D., García, A., Pringle, R.M., Palmer, T.M., 2015. Mattern, T., 2013. Snares penguin (Eudyptes robustus).. In: Garcia Borboroglu, P., Accelerated modern human-induced species losses: entering the sixth mass extinction. Boersma, P.D. (Eds.), Penguins: Natural History and Conservation. Washington Uni- Sci. Adv. 5, e1400253. versity Press. Christidis, L., Boles, W.E., 2008. Systematics and of Australian Birds. CSIRO Mattern, T., Long, R., 2017. Survey and population size estimate of Publishing, Collingwood, Australia. (tawaki; Eudyptes pachyrhynchus) in Milford Sound/Piopiotahi, New Zealand. Notor- Cole, T.L., Wood, J.R., 2017. The ancient DNA revolution: the latest era in unearthing New nis 64, 97–101. Zealand’s faunal history. N. Z. J. Zool. 45, 91–120. Mattern, T., Housten, D.M., Lalas, C., Setiawan, A.N., Davis, L.S., 2009. Diet composition, Cole, T.L., Waters, J., Shepherd, L.D., Rawlence, N.J., Joseph, L., Wood, J.R., 2017. An- continuity in prey availability and marine habitat – keystones to population stability cient DNA reveals that the ‘extinct’ Hunter Island penguin (Tasidyptes hunteri) is not in the Snares penguin (Eudyptes robustus). Emu 109, 204–213. a distinct taxon. Zool. J. Linn. Soc. 182, 459–464. , . Collins, C.J., Rawlence, N.J., Prost, S., Anderson, C.N.K., Knapp, M., Scofield, R.P., Robert- McWethy, D.B., Whitlock, C., Wilmshurst, J.M., McGlone, M.S., Fromont, M., Li, X., Dief- son, B.C., Smith, I., Matisoo-Smith, E.A., Chilvers, B.L., Waters, J.M., 2014. Extinction fenbacher-Krall, A., Hobbs, W.O., Fritz, S.C., Cook, E.R., 2010. Rapid landscape trans- and recolonization of coastal megafauna following human arrival in New Zealand. formation in South Island, New Zealand, following initial Polynesian settlement. Proc. Proc. R. Soc. B 281, 20140097. Natl. Acad. Sci. 107, 21343–21348. Cooper, A., Poinar, H.N., 2000. Ancient DNA: do it right or not at all. Science 289, 1139. Millener, P.R., 1990. Evolution, extinction and the subfossil record of New Zealand’s avi- Cornuet, J.-M., Ravigné, V., Estoup, A., 2010. Inference on population history and model fauna. In: Gill, B.J., Heather, B.D. (Eds.), A Flying Start. Random Century, Auckland, checking using DNA sequence and microsatellite data with the software DIYABC pp. 93–100. (v1.0). BMC Bioinf. 11, 401. Miskelly, C.M., 1997. Biological notes on the Western Chain, Snares Islands, 1984-85 and Cornuet, J.M., Pudlo, P., Veyssier, J., Dehne-Garcia, A., Gautier, M., Leblois, R., Marin, 1985-86. Conservation Advisory Science Notes No. 144. Department of Conservation, J.M., Estoup, A., 2014. DIYABC v2.0: A software to make approximate Bayesian com- Wellington. putation inferences about population history using single nucleotide polymorphism, Morrison, K.W., Battley, P.F., Sagar, P.M., Thompson, D.R., 2015. Population dynamics of DNA sequence and microsatellite data. Bioinformatics 30, 1187–1189. eastern rockhopper penguins on Campbell Island in relation to sea surface tempera- Crawford, R.J.M., Whittington, P.A., Upfold, L., Ryan, P.G., Petersen, S.L., Dyer, B.M., ture 1942–2012: current warming hiatus pauses a long-term decline. Polar Biol. 38, Cooper, J., 2009. Recent trends in numbers of four species of penguins at the Prince 163–177. Edward Islands. Afr. J. Mar. Sci. 31, 419–426. Otley, H., Tansell, J., Scofield, P., 2017. A comprehensive demographic assessment of the Crawford, R.J.M., Cooper, J., Dyer, B.M., Greyling, M.D., Klages, M.T.W., Nel, D.C., Nel, endangered Fiordland crested penguin Eudyptes pachyrhynchus. N. Z. J. Zool. 44, J.L., Petersen, S.L., Wolfaardt, A.C., 2010. Decrease in numbers of the eastern rock- 144–162. hopper penguin Eudyptes chrysocome filholi at Marion Island, 1994/95-2002/03. Afr. Otley, H., Edmonds, H., Hiscock, J., Newton, G., Tansell, J., van Klink, P., Wilson, R., J. Mar. Sci. 25, 487–498. Westbrooke, I., 2018. Assessing the population trend and threats to New Zealand’s Darriba, D., Taboada, G.L., Doallo, R., Posada, D., 2012. jModelTest 2: more models, new Fiordland crested penguin using counting and demographic modelling approaches, N. heuristics and parallel computing. Nat. Methods 9, 772. Z. J. Ecol.. Davidson, J., 1988. The coming of the Maori. In: Baldwin, O. (Ed.), The Celebration His- Prost, S., Anderson, C.N.K., 2011. TempNet: a method to display statistical parsimony net- tory of the Kapiti District: 100 Years Plus. Kapiti Borough Council, pp. 31–36. works for heterochronous DNA sequence data. Methods Ecol. Evol. 2, 663–667. Davis, L.S., 2013. Erect-crested penguin (Eudyptes sclateri). In: García Borboroglu, P.G., , . , Boersma, P.D. (Eds.), andUNCORRECTEDConservation of the World’s Penguins. University of Robertson, H.A., Baird, K., Dowding, J.E., Elliott, G.P., Hitchmough, R.A., Miskelly, C.M., Washington Press, Seattle U.S.A. McArthur, N., O’Donnell, C.F.J., Sagar, P.M., Scofield, R.P., Taylor, G.A., 2017. Con- Diamond, J., 1990. New Zealand as an archipelago: an international perspective. In: servation Status of New Zealand birds, 2016. New Zealand Threat Classification Series Towns, D.R., Daugherty, C.H., Atkinson, I.A.E. (Eds.), Ecological Restoration of New 19. New Zealand Department of Conservation. Zealand islands. Conservation Sciences Publication No 2. Department of Conservation, Rawlence, N.J., Perry, G.L.W., Smith, I.W.G., Scofield, R.P., Tennyson, A.D., Mati- Wellington, New Zealand, pp. 3–8. soo-Smith, E.A., Boessenkool, S., Austin, J.J., Waters, J.M., 2015. Radiocarbon-dating Drummond, A.J., Rambaut, A., Shapiro, B., Pybus, O.G., 2005. Bayesian coalescent in- and ancient DNA reveal rapid replacement of extinct prehistoric penguins. Quat. Sci. ference of past population dynamics from molecular sequences. Mol. Biol. Evol. 22, Rev. 112, 59–65. 1185–1192. Rawlence, N.J., Kennedy, M., Anderson, C.N.K., Prost, S., Till, C.E., Smith, I.W.G., Scofield, Duncan, R.P., Boyer, A.G., Blackburn, T.M., 2013. Magnitude and variation of prehistoric R.P., Tennyson, A.J.D., Hamel, J., Lalas, C., Matisoo-Smith, E.A., Waters, J.M., 2015. bird extinctions in the Pacific. Proc. Natl. Acad. Sci. USA 16, 6436–6441. Geographically contrasting biodiversity reductions in a widespread New Zealand Dussex, N., Rawlence, N.J., Robertson, B.C., 2015. Ancient and contemporary DNA reveal seabird. Mol. Ecol. 24, 4605–4616. a pre-human decline but no population bottleneck associated with recent human per- Rawlence, N.J., Collins, C.J., Anderson, C.N.K., Maxwell, J.J., Smith, I.W.G., Robertson, secution in the kea (Nestor notabilis). PLoS ONE 10, e0118522. B.C., Knapp, M., Horsburgh, K.A., Stanton, J.-A.L., Scofield, R.P., Tennyson, A.J.D., Matisoo-Smith, E.A., Waters, J.M., 2016. Human-mediated extirpation of the unique sea lion and implications for the conservation management of re- maining New Zealand sea lion populations. Mol. Ecol. 25, 3950–3961. Rawlence, N.J., Kardamaki, A., Easton, L.J., Tennyson, A.J.D., Scofield, R.P., Waters, J.M., 2017. Ancient DNA and morphometric analysis reveal extinction and replacement of New Zealand’s unique black swans. Proc. R. Soc. B: Biol. Sci. 284, 20170876.

7 T.L. Cole et al. Molecular Phylogenetics and Evolution xxx (2018) xxx-xxx

Rawlence, N.J., Till, C.E., Easton, L.J., Spencer, H.G., Schuckard, R., Melville, D.S., Thomson, V.A., Lebrasseur, O., Austin, J.J., Hunt, T.L., Burney, D.A., Denham, T., Raw- Scofield, R.P., Tennyson, A.J.D., Rayner, M.J., Waters, J.M., Kennedy, M., 2017. Spe- lence, N.J., Wood, J.R., Gongora, J., Girland Flink, L., Linderholm, A., Dobney, K., ciation, range contraction and extinction in the endemic New Zealand king shag com- Larson, G., Cooper, A., 2014. Using ancient DNA to study the origins and dispersal plex. Mol. Phylogenet. Evol. 115, 197–209. of ancestral Polynesian chickens across the Pacific. Proc. Natl. Acad. Sci. USA 111, Rawlence, N.J., Tennyson, A.J.D., Cole, T.L., Verry, A., Scofield, R.P., 2018. Evidence for 4826–4831. breeding of Megadyptes penguins in North Island at the time of human arrival. N. Z. , . J. Zool. 1-9. Waters, J.M., Fraser, C.I., Maxwell, J.J., Rawlence, N.J., 2017. Did interaction between Rohland, N., Siedel, H., Hofreiter, M., 2010. A rapid column-based ancient DNA extraction human pressure and Little Ice Age drive biological turnover in New Zealand?. J. Bio- method for increased sample throughout. Mol. Ecol. Resour. 10, 677–683. geogr. 44, 1481–1490. Sagar, P.M., 1971. Birds of the Western Chain, Snares Islands, New Zealand. Notornis 24, Wanner, H., Beer, J., Bütikofer, J., Crowley, T.J., Cubasch, U., Flückiger, J., Goosse, H., 178–183. Grosjean, M., Fortunat, J., Kaplan, J.O., Küttel, M., Müller, S.A., Prentice, C., Solomin, Salis, A.T., Easton, L.J., Robertson, B.C., Gemmell, N., Smith, I.W.G., Weisler, M., Waters, O., Stocker, T.F., Tarasov, P., Wagner, M., Widmann, M., 2008. Mid- to Late Holocene J.M., Rawlence, N.J., 2016. Myth or relict: does ancient DNA detect the enigmatic up- climate change: an overview. Quat. Sci. Rev. 27, 1791–1828. land seal?. Mol. Phylogenet. Evol. 97, 101–106. Warham, J., 1974. The Fiordland crested penguin Eudyptes pachyrhynchus. Ibis 116, Scarlett, R.J., 1979. Avifauna and man. In: Anderson, A. (Ed.), Birds of a feather: Osteo- 1–23. logical and archaeological papers from the South Pacific in honour of R. J. Scarlett. , . New Zealand Archaeological Association Monograph II. BAR International Series 62. Wilmshurst, J.M., Anderson, A.J., Higham, T.F.G., Worthy, T.H., 2008. Dating the late BAR, Oxford, pp. 75–90. prehistoric dispersal of to New Zealand using the commensal Pacific . Scofield, P., Worthy, T.H., Schlumpf, H., 2003. What birds were New Zealand’s first peo- Proc. Natl. Acad. Sci. USA 105, 7676–7680. ple eating? – ’s avian remains re-examined. Rec. Canterbury Museum 17, Wilmshurst, J.M., Hunt, T.L., Lipo, C.P., Anderson, A.J., 2011. High-precision radiocarbon 17–35. dating shows recent and rapid initial human colonization of East Polynesia. Proc. Natl. Shepherd, L.D., Lambert, D.M., 2008. Ancient DNA and conservation: lessons from the en- Acad. Sci. USA 108, 1815–1820. dangered kiwi of New Zealand. Mol. Ecol. 17, 2174–2184. . Shepherd, L.D., Worthy, T.H., Tennyson, A.J.D., Scofield, R.P., Ramstad, K.M., Lambert, , . D.M., 2012. Ancient DNA analyses reveal contrasting phylogenetic patterns amongst Wood, J.R., Alcover, J.A., Blackburn, T.M., Bover, P., Duncan, R.P., Hume, J.P., Louys, J., kiwi (Apteryx spp.) and a recently extinct lineage of spotted kiwi. PLoS ONE 7, Meijer, H.J.M., Rando, J.C., Wilmshurst, J.M., 2017. Island extinctions: processes, pat- e42384. and potential for restoration. Envrion. Conserv. 4, 348–358. Spatz, D.R., Zilliacus, K.M., Holmes, N.D., Butchart, S.M.H., Genovesi, P., Ceballos, G., Ter- Worthy, T.H., 1997. The identification of fossil Eudyptes and Megadyptes bones at Mar- shy, B.R., Croll, D.A., 2017. Globally threatened vertebrates on islands with invasive fells Beach, Marlborough, SouthPROOFIsland. N. Z. Nat. Sci. 23, 71–85. species. Conserv. Biol. 10, e1603080. Worthy, T.H., 1998. A remarkable fossil and archaeological fauna from Marfell’s Beach, Taylor, G.A., 2000. Action Plan for Seabird Conservation in New Zealand. Part A: Threat- Lake Grassmere, South Island, New Zealand. Rec. Canterbury Museum 12, 79–176. ened Seabirds. Threatened Species Occasional Publication No. 16. Department of Con- Worthy, T.H., 1999. What was on the menu- avian extinction in New Zealand. N. Z. J. Ar- servation, Wellington. chaeol. 19, 125–160. Tennyson, A., Martinson, P., 2007. Extinct , revised ed. Te Papa Worthy, T.H., Holdaway, R.N., 2002. The Lost World of the , Prehistoric Life of New Press, Wellington, New Zealand. Zealand. Canterbury Univ Press, Christchurch, New Zealand

UNCORRECTED

8