The origin and phylogenetic relationships of the New Zealand ravens

Molecular Phylogenetics and Evolution 106 (2017) 136–143

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev

The origin and phylogenetic relationships of the New Zealand ravens R. Paul Scofield a,⇑, Kieren J. Mitchell b, Jamie R. Wood c, Vanesa L. De Pietri a, Scott Jarvie d, Bastien Llamas b, Alan Cooper b a

Canterbury Museum, Rolleston Avenue, Christchurch 8013, New Zealand Australian Centre for Ancient DNA, School of Biological Sciences, University of Adelaide, North Terrace Campus, South Australia 5005, Australia c Long-term Ecology Laboratory, Landcare Research, Post Office Box 69040, Lincoln 7640, New Zealand d Department of Zoology, Otago University, Dunedin 9054, New Zealand b

a r t i c l e

i n f o

Article history: Received 14 July 2016 Revised 2 September 2016 Accepted 23 September 2016 Available online 24 September 2016 Keywords: New Zealand raven Corvus Ancient DNA Mitogenome Phylogeny Osteology

a b s t r a c t The relationships of the extinct New Zealand ravens (Corvus spp.) are poorly understood. We sequenced the mitogenomes of the two currently recognised species and found they were sister-taxa to a clade comprising the Australian raven, little raven, and forest raven (C.coronoides, C. mellori and C. tasmanicus respectively). The divergence between the New Zealand ravens and Australian raven clade occurred in the latest Pliocene, which coincides with the onset of glacial deforestation. We also found that the divergence between the two putative New Zealand species C. antipodum and C. moriorum probably occurred in the late Pleistocene making their separation as species untenable. Consequently, we consider Corax antipodum (Forbes, 1893) to be a subspecies of Corvus moriorum Forbes, 1892. We re-examine the osteological evidence that led 19th century researchers to assign the New Zealand taxa to a separate genus, and re-assess these features in light of our new phylogenetic hypotheses. Like previous researchers, we conclude that the morphology of the palate of C. moriorum is unique among the genus Corvus, and suggest this may be an adaptation for a specialist diet. Crown Copyright Ó 2016 Published by Elsevier Inc. All rights reserved.

1. Introduction Recent studies have argued that the core Corvoidea (crows, jays, and magpies) originated in what is now Wallacea in the late Oligocene/early Miocene and dispersed via the Indo-Pacific archipelago to the rest of the world (Jønsson et al., 2011). Subsequent work has concluded that the crows (genus Corvus) originated in the Palaearctic in the early Miocene and dispersed to North America and the Caribbean, Africa and Australasia (Jønsson et al., 2012). The colonization of Wallacea by the crows is likely to have taken place in the late Miocene, leading to further colonization of Australo-Papua around 5 Ma (Jønsson et al., 2012, 2016). Evidence for four colonization events of the Pacific from Asia and Australia have been found (Jønsson et al., 2012). Following the work of Jønsson and colleagues, two clades of the genus Corvus are now recognised in Australia: the Australian ravens, comprising the Australian raven (C. coronoides), little raven (C. mellori), and forest raven (C. tasmanicus); and the Australian crows, which include the little crow (Corvus bennetti), Torresian crow (Corvus orru), along with the Bismarck crow (Corvus insularis).

⇑ Corresponding author. E-mail address: [email protected] (R.P. Scofield). http://dx.doi.org/10.1016/j.ympev.2016.09.022 1055-7903/Crown Copyright Ó 2016 Published by Elsevier Inc. All rights reserved.

All the recent genetic work on the evolution of the genus Corvus, however, has been based on the extant species. The status of the recently extinct New Zealand ravens remains unresolved. The New Zealand (NZ) ravens have received little study since two species were described by Henry Ogg Forbes (Forbes, 1892a, c, 1893) based on partial sub-fossil skeletons. The two species, one found on the main islands of New Zealand (Corvus antipodum) and the other on the Chatham Islands (Corvus moriorum), appear to have been slightly smaller than the common raven (Corvus corax) but larger than the Australian raven of southern and eastern Australia. Forbes (1892b) suggested that supposedly unique characters of the cranium warranted separating the NZ ravens into their own genus, which he named Palaeocorax. Although the distinctiveness of some of Forbes’ cranial features were questioned (Pycraft, 1911), no author formally recommended the synonymy of Palaeocorax and Corvus until Brodkorb (1978). However, this recommendation was not widely accepted until Gill (2003) reviewed the New Zealand taxa and concluded that Palaeocorax should be regarded as a junior synonym of Corvus. Further, Gill’s (2003) analyses of overall proportions and osteology led him to conclude that ‘‘New Zealand ravens probably evolved following the invasion of New Zealand by C. coronoides or a population of crows ancestral to C. coronoides.”

R.P. Scofield et al. / Molecular Phylogenetics and Evolution 106 (2017) 136–143

Very little is known about the biology of NZ ravens. Pycraft (1911) found that Chatham Island ravens, in addition to having smaller wings than northern ravens, had a shallower sternal keel and smaller pygostyle, indicating a slightly reduced flight capacity. Brooke (2000) suggested that damage found on the shells of snails of the genus Placostylus in the Far North of New Zealand may have been due to raven predation. Tennyson and Martinson (2007) proposed that the coastal nature of the majority of NZ raven fossil sites and the prevalence of raven bones with the bones of adult and immature sealion (Phocarctos hookeri) may have meant a predator-prey relationship between the two taxa. In the present study, we use molecular, osteological and distributional data to resolve the origins and phylogenetic relationships of these mysterious extinct New Zealand birds and to give further insight into the palaeobiology and distribution of ravens on the mainland of New Zealand. 2. Methods 2.1. Ancient DNA extraction and sequencing We obtained bone samples of both Corvus antipodum (CM Av 12546, Marfells Beach, Marlborough) and Corvus moriorum (CM Av5468, Chatham Island) from the collections of the Canterbury Museum, New Zealand. Both samples are from dune deposits and are of late Holocene age. All DNA extraction and library preparation steps were performed in a purpose-built, physically isolated, ancient DNA laboratory at the Australian Centre for Ancient DNA, University of Adelaide. Digestion of bone samples and DNA extraction were performed exactly as described by Mitchell et al. (2014b). Extracted DNA was enzymatically blunt-ended, and truncated Illumina adapters (see Mitchell et al., 2014b) were ligated according to the protocol of Meyer and Kircher (2010). The custom P5 and P7 adapter sequences each included a 7mer barcode to allow downstream identification of library molecules and exclusion of crosscontamination between DNA libraries. The libraries were amplified, enriched for bird mtDNA, quality controlled, and prepared for sequencing following the procedure outlined in (Mitchell et al., 2014b). We diluted the enriched libraries to 2 nM and ran them on an Illumina MiSeq using 2  150 bp (paired-end) sequencing chemistry. 2.2. Data processing We used ‘sabre’ v1.00 (https://github.com/najoshi/sabre) to identify paired forward and reverse reads that contained the 7mer barcode sequences corresponding to our libraries (no mismatches allowed). We then trimmed adapter sequences and merged the paired forward and reverse reads using AdapterRemoval v2.1.2 (Lindgreen, 2012). Low quality bases were trimmed (Phred < 20 --minquality 4) and merged reads shorter than 25 bp were discarded (--minlength 25). We visualised read quality using fastQC v0.10.1 (http://www.bioinformatics.bbsrc.ac.uk/projects/fastqc) before and after trimming to make sure the trimming of adapters was efficient. We assembled mitochondrial genome sequences for C. antipodum and C. moriorum using an iterative mapping approach (see Mitchell et al., 2014a,b). We mapped successfully merged forward and reverse reads for both taxa against the mitochondrial genome sequence of the rook (Corvus frugilegus, GenBank accession Y18522) using BWA v0.7.8 (Li and Durbin, 2009) with parameters recommended for ancient DNA (aln -l 1024, -n 0.01, -o 2) (Schubert et al., 2012). We excluded reads with low mapping quality scores (Phred <30) using SAMtools v1.4 (Li et al., 2009), and removed

137

duplicate reads using ‘FilterUniqueSAMCons.py’ (Kircher, 2012). A 50% consensus sequence was generated for both C. antipodum and C. moriorum from the remaining mapped reads using Geneious v8.1.6 (Biomatters; http://www.geneious.com), retaining the reference sequence for regions with no read coverage. These consensus sequences were then used for a subsequent round of mapping. This process was iterated until no additional reads could be mapped for either taxon (12 iterations for C. antipodum and nine iterations for C. moriorum). For the final consensus sequences, we called unambiguous bases only for sites with read depth P2 and where P75% of reads agreed, other sites were coded as IUPAC ambiguities. Mean read depth across the 15,065-bp-long C. antipodum mitogenome consensus sequence (GenBank accession KX822154) was 7.9x (standard deviation = 6.2x; coverage = 77.9%), based on 1848 reads of mean length 64.7 bp (standard deviation = 17.0 bp). The C. moriorum mitogenome consensus sequence (GenBank accession KX822153) was 15,593 bp long, covered to a mean depth of 124.6x (standard deviation = 41.2x; coverage = 100%) by 26,626 reads of mean length 73.0 bp (standard deviation = 21.1 bp). We used the MUSCLE algorithm (as implemented in Geneious v8.1.6) to align the CO1, ND2, ND3, and CYTB loci of our new mitogenomes with a previously published, comprehensively sampled matrix of these four loci from corvoid birds (Jønsson et al., 2016). We then used published mitogenome sequences to complete gaps in this alignment for (GenBank accession numbers in parentheses): Corvus brachyrhynchos (KP403809), Corvus frugilegus (Y18522), Corvus hawaiiensis (KP161620), Corvus macrorhynchos (KR072661), Corvus splendens (KJ766304), Cyanopica cyanus (JN108020), Garrulus glandarius (JN018413), Nucifraga columbiana (KF509923), Pica pica (HQ915867), Podoces hendersoni (GU592504), and Urocissa erythrorhyncha (JQ423932). From this alignment we selected taxa from the corvoid sub-clade of interest for downstream phylogenetic analysis (Supplementary Information Tables 1 & 2). 2.3. Phylogenetic analysis and molecular dating RAxML v8.2.0 (Stamatakis, 2006) was used to reconstruct a phylogeny from the dataset under a maximum likelihood framework. We divided the data into five partitions (Supplementary Information Table 3) based on the results of PartitionFinder v1.1.1 (Lanfear et al., 2012). Our partitioned RAxML analysis comprised a maximum likelihood search for the best-scoring tree from 1000 bootstrap replicates (-f a -m MULTIGAMMA -# 1000). We used BEAST v1.8.2 (Drummond and Rambaut, 2007) to simultaneously infer the phylogeny and evolutionary timescale under a Bayesian framework. PartitionFinder was used to determine optimal partitioning and substitution models (Supplementary Information Table 4). Following Jønsson et al. (2016) we constrained the root of the tree to fall between the Cissa/Urocissa magpies and the clade comprising the remaining sampled taxa. We implemented a single lognormal relaxed clock model (with a rate multiplier parameter for each data partition), and constrained the age of two key nodes in accordance with the fossil record (Supplementary Information Table 7; using uniform distributions with hard minima and maxima): (1) We placed a minimum constraint of 5.3 Ma on the divergence between Pica and Corvus based on Miopica paradoxa from Belka and Novaya Emetovka in the Odessa Region of Ukraine. These sites are considered to belong to European fossil stage MN10 (Kurochkin and Sobolev, 2004) and thus be of Messinian (Pontian) age, between 5.3 and 7.2 Ma. M. paradoxa was described based on an incomplete ulna, and although similarities with jays were also noted, it is morphologically more similar to Pica magpies than to crows, ravens,

138

R.P. Scofield et al. / Molecular Phylogenetics and Evolution 106 (2017) 136–143

and jackdaws (Corvus) (Kurochkin and Sobolev, 2004). We placed a maximum constraint of 23.0 Ma (beginning of the Miocene) on this node, based on the paucity of Palaeogene crown-corvid fossils. (2) We placed a minimum constraint of 13.6 Ma on the divergence between Corvus and Cissa/Urocissa based on Miocitta galbreathi Brodkorb, 1972 (see also Becker, 1987), which is known from the Middle Miocene of North America. Although also based on fragmentary material (a distal humerus), osteological features suggest that M. galbreathi is putatively an ‘‘American jay” and as such is more closely related to Corvus than to the Cissa/Urocissa magpies (see Jønsson et al., 2016). So far, this diagnosis has not been challenged. As above, we used the paucity of Palaeogene crown-corvids to place a 23.0 Ma maximum constraint on this node. We ran the BEAST analysis three times with random starting trees in order to ensure convergence on the global optimum. Each independent MCMC chain was run for 108 iterations, sampling trees and parameter values every 105 iterations. The first 10% of samples for each chain were discarded as burnin. Parameter values were monitored and compared between the three independent chains in Tracer v1.6 (http://tree.bio.ed.ac.uk/software/tracer/) to ensure convergence and ESSs >200. We combined sampled trees and parameter values from all BEAST chains before summarising the results. 2.4. Osteology Anatomical terminology follows Baumel and Witmer (1993), where appropriate. We restrict our re-assessment of the osteological features of NZ ravens to the skull, as postcranial material was found to be rather uniform between species of NZ raven and among other species of Corvus (Pycraft, 1911; Gill, 2003). The following skulls at Canterbury Museum were used in this study: C. moriorum – CM AV5454 and CM AV7575; Corvus corone – CM 2016.41.2121. The skulls of other corvids (25 species) were examined at www.skullsite.com. There is currently no cranial material attributable to C. antipodum with an intact palate (see Gill, 2003), so comparisons between the two recognised species are not possible. 2.5. Inferring potential pre-human distribution of Corvus antipodum 2.5.1. Environmental data We used 19 modern bioclimatic variables at a ca. 1 km spatial resolution from WorldClim 1.4 (interpolations representative of 1950s–2000; www.worldclim.org; Hijmans et al., 2005). The bioclimatic variables employed describe annual and seasonal variation in temperature and precipitation. At coarse spatial scales, bioclimatic variables are the primary determinant of species distributions (see Araújo and Peterson, 2012 for a review). We did not aim to understand which variables were driving the pre-human distribution of NZ ravens, thus reducing the number of variables was not necessary. 2.5.2. Ecological niche model A total of 48 Holocene aged Corvus record localities (Supplementary Information Table 5) and the bioclimatic variables were used to make the ecological niche model (ENM). We generated ecological niche predictions of NZ raven’s pre-human distribution using Maxent (v 3.3.3k; Phillips et al., 2006) executed from the dismo package in R (Hijmans et al., 2016; R Core Team, 2016). The Maxent program is a machine learning algorithm that uses environmental covariates to discriminate presence records from background points (Elith et al., 2011). For comparison purposes,

we tested the Maxent model against a null geographic model, Maxent has consistently performed well in comparative studies with other niche modelling methods, including with low number of localities (Elith et al., 2006, 2011). To improve the performance of the ENM for NZ raven, we thinned localities in geographic space using the spThin package in R (Aiello-Lammens et al., 2015) following Radosavljevic and Anderson (2014). We used Maxent with default settings, except that we applied a targeted background sampling to reduce the influence of sample selection bias (Phillips et al., 2009) by using 666 vertebrate fossil site localities (excluding moa bones [Order Dinornithiformes] and swamp sites) throughout New Zealand as background points. Fossil localities with moa bones were excluded, as moa bones are valued by collectors and have often been specifically targeted, i.e. NZ raven bones might have been found at fossil localities, but might not have been recorded as they were not being specifically targeted. All record localities accounted for differences in cell area across latitude (Elith et al., 2011). We evaluated Maxent model performance using five-fold crossvalidation of the area under the curve (AUC) of the receiver operating characteristic curve. The AUC varies between 0.5 for a model that performs no better than random to 1.0 for perfect assignment of presence and absence. We tested the Maxent model against a null geographic model in which the probability of occurrence is inversely proportional to the distance from the occurrence points (Raes and ter Steege, 2007). Model testing was performed using the dismo package in R (Hijmans et al., 2016). We compiled a list of archaeological sites where Corvus remains had been recovered mostly from Gill (2003) (Fig. 2a; Supplementary Information Table 6) to compare post-human and prehuman distributions. We compiled a list of sites where New Zealand sea-lion (Phocarctos hookeri) remains had been recovered mostly from Collins et al. (2014) (Fig. 2b) to compare the overlap between NZ ravens and sea-lions.

3. Results 3.1. Phylogenetic analyses Our phylogenetic analyses (Fig. 1, Supplementary Information Tables 1&2) supported the monophyly of the NZ ravens (Bayesian posterior probability [BPP] = 1.0, maximum likelihood bootstrap [MLB] = 100%), and strongly supported a sister-taxon relationship between this endemic New Zealand lineage and the clade comprising the Australian raven, little raven, and forest raven (BPP = 1.0. MLB = 95%). This combined clade fell within the Australasian/Wallacean radiation of large ravens (‘‘Clade VII”) previously identified by Jønsson et al. (2012). Our inferred relationships among outgroup taxa were largely concordant with previous studies (e.g. Jønsson et al., 2012, 2016), and only differed at nodes that received equivocal support.

3.2. Molecular dating Molecular dating estimates in the present study were highly concordant with those of Jønsson et al., 2016, with substantially overlapping 95% Highest Posterior Densities (95% HPDs) for the posterior ages of equivalent nodes (Fig. 1, Supplementary Information Tables 1&2). Our posterior estimate for the mitochondrial rate of divergence among the sampled taxa was 2%–3.6% million years 1 (95% HPD). We estimated that the divergence between the NZ ravens and the clade comprising the Australian raven, little raven, and forest raven occurred 1.73 Ma (1.14–2.41 Ma 95% HPD). The divergence between the currently recognised New Zealand

139

Corvus moriorum Chathams raven

1.0 / 100%

Corvus antipodum NZ raven 1.0 / 95%

Corvus mellori Little raven

1.0 / 100% 1.0 / 93%

Corvus coronoides Australia raven Corvus bennetti Little crow

0.60 / 50%

Corvus orru Torresian crow

1.0 / 100%

Australia / PNG

Corvus tasmanicus Forest raven

0.63 / 69%

0.99 / 90%

NZ

R.P. Scofield et al. / Molecular Phylogenetics and Evolution 106 (2017) 136–143

Corvus insularis Bismarck crow Corvus tristis Grey crow

0.96 / 61% 0.65 / 48%

Corvus fuscicapillus Brown-headed crow Corvus violaceus Violaceous crow

0.85 / 44%

Corvus moneduloides New Caledonian crow

0.47 / 41%

Corvus meeki Bougainville crow

1.0 / 100%

Corvus woodfordi White-billed crow

0.98 / 48% 1.0 / 94%

Corvus validus Long-billed crow Corvus unicolor Banggai crow

1.0 / 97%

Corvus typicus Piping crow 0.76 / 29%

Corvus enca Slender-billed crow

Pliocene 5

4

Pleistocene 3

2

1

0

Time (millions of years before present) Fig. 1. Jønsson et al.’s (2012) Corvus ‘‘Clade VII” extracted from our time-calibrated BEAST maximum clade credibility tree. Full trees - including outgroups - are available in the Supplementary Information. Scale is in millions of years before the present. Node heights represent mean age estimates, while node bars represent 95% Highest Posterior Densities (HPDs). Branch support values (BEAST posterior probability / maximum likelihood bootstrap%) are displayed for all clades.

taxa C. antipodum and C. moriorum occurred 0.15 Ma (0.05–0.27 Ma 95% HPD). 3.3. Cranial osteology Forbes and Pycraft’s descriptions of the skull of Corvus moriorum, were limited by a lack of a comprehensive sample of comparative specimens. In light of the phylogenetic hypotheses here presented, we re-examined the distinctiveness of the cranial features assessed by Forbes (1892a,b), Pycraft (1911) and Gill (2003). In his diagnosis of Palaeocorax, Forbes (1892b) described some features of the skull characterising Corvus (Palaecorax) moriorum, namely: (1) the presence of rudimentary processus basipterygoidei; (2) an ossified nasal septum (septum nasale) fused rostrally

to the premaxillae (see also Pycraft, 1911); (3) an ossified mesethmoid projecting into the concavity of the dorsal surface of the vomer, and (4) a flatter and broader vomer compared to other species of Corvus, having three ‘‘points” rostrally (Fig. 3c). Of these features mentioned by Forbes, and among species of Corvus, we found only the latter to possibly be autapomorphic for C. moriorum, although we found that a three-pointed vomer is also present in C. hawaiiensis (Olson & James 1991; Fig. 5). Although processus basipterygoidei are actually absent in all passerines (Huxley, 1867; Mayr and Clarke, 2003), there are indeed marked attachment scars for the insertion of the pterygoids both sides of the rostrum parasphenoidale (see also Pycraft, 1911). These vestigial attachments have been reported for corvoid lineages (Schodde and Mason, 1999). A rostrally expanded and ossified nasal septum

140

R.P. Scofield et al. / Molecular Phylogenetics and Evolution 106 (2017) 136–143

Fig. 2. (a) Predicted pre-human distribution of the mainland New Zealand Raven Corvus based on ecological niche models. Black dots represent fossil records and blue triangles represent archaeological records (see Supplementary Information Table 5 & 6); (b) Recorded pre-human distribution of the extinct clade of New Zealand sea lion (Phocarctos hookeri) after Collins et al., 2014; (c) Pleistocene records of Corvus (see Supplementary information Table 5) mapped onto are construction of the New Zealand coastline and vegetation at c. 22,000ybp (after Alloway et al., 2007) demonstrating the more inland distribution during this epoch. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3. Dorsal views of Corvus skulls demonstrating variation in the degree of ossification of the palate (ossa maxillae et palati): (a) Australian Raven C. coronoides (from Gill, 2003); square indicates the area enlarged in images b, c & d); (b) rook C. frugilegus (CM Av15384); (c) NZ raven C. moriorum Chatham Islands (CM Av6744); (d) extinct Hawaiian crow C. impluviatus (after James and Olson, 1991); (e) NZ raven C. moriorum, lateral view of skull (from Pycraft, 1911). Key: om os maxillare; os septum nasale; op os palatinum (pars lateralis); pm processus maxillopalatinus; vo vomer (highlighted in mauve). Scale bar = 10 mm.

R.P. Scofield et al. / Molecular Phylogenetics and Evolution 106 (2017) 136–143

is also present in the white-necked raven Corvus albicollis, fantailed raven C. rhipidurus, and the extinct C. impluviatus from Hawaii (Fig. 3d). The skull of C. moriorum was described in detail by Pycraft (1907, 1911), who mentioned that the distinguishing features of this taxon occurred in the region of the palate (Fig. 3c) and included: (1) the extensive ossification of the roof of the palate (which includes the premaxillae, ossa palati, and ossified septum nasale and conchae nasales) so that only a narrow space occurs between the ossa palati and the vomer (compare Fig. 3b with Fig. 3c); (2) large and broad processus maxillopalatini (also related to the extensive ossification of the palate region); and, as noted by Forbes (1892b), (3) an unusually large vomer with a peculiar shape (broad, flat, with a ‘‘double-scalloped” edge; Fig. 3c). The processus maxillopalatini are indeed broader (Fig. 3c) than in most species of Corvus we examined, with the exception of C. rhipidurus. The roofed-in palate and the broad and flat vomer do, however, appear to be autapomorphic for C. moriorum. Gill’s (2003) analysis was the first to compare the NZ ravens to the Australian ravens. Although species of Corvus are notoriously conservative in their osteology, he found that the morphology of the palate was indeed unique among the Corvus specimens he studied, which is further supported by our assessment of additional species of Corvus. Gill (2003) hypothesised that the partially closed palate in the Australian raven C. coronoides and the Torresian crow C. orru was intermediate between that of C. moriorum and North American species of Corvus (C. corax and C. brachyrhynchos), concluding that there was a likely relationship between the New Zealand and Australian taxa. Gill (2003 contra Pycraft 1911) found that NZ ravens only showed slightly reduced ratios of ulna to tarsometatarsus length compared to northern hemisphere and Australian ravens, suggesting they flew strongly. This conclusion was slightly cofounded by the finding that compared to Australian Corvus, the tarsometatarsus of NZ ravens was elongated suggesting more terrestrial habits (in a similar manner to laughing owl and the Chatham Island kaka Nestor chathamensis (Wood et al., 2014, 2016).

3.4. Distribution Our model predicts that the ecological niche for the pre-human distribution of NZ raven on the mainland of New Zealand was largely confined to coastal areas on the North Island, South Island and Stewart Island of New Zealand (Fig. 2a). Highly suitable areas were found in clusters in the north and the southeast of the North Island, the east coast of the South Island and throughout Stewart Island approximating the pre-human distribution of the extinct clade of New Zealand sea-lion (Fig 2b; Collins et al., 2014; Rawlence et al., 2014). It is notable that all archaeological sites with Corvus remains are from the Early Period of Polynesian settlement in New Zealand (the period identified by the presence of moa bone and therefore occurred less than 150 years after settlement (Perry et al., 2014)). The Maxent model showed very good performance with AUC (0.96) demonstrating a better performance than the null geographic models (AUC = 0.87). Examining the ecological niche map showed that archaeological deposits containing Corvus mostly fell within an area identified as highly suitable (Fig. 2a). Plotted position of the locations of Pleistocene fossils give limited support for the contention (Tennyson and Martinson, 2007) that Corvus occupied a larger area during this epoch (Fig 2c). Average distances from the current coastline in the Holocene (3.06 km, n = 47, 0–25 km; Supplementary Information Table 5) are significantly less (one-tailed, heteroscedastic t-test, t = 00,134, p = 2.01) than those for the small Pleistocene sample (13.45, n = 11, 0– 131 km).

141

4. Discussion 4.1. Size variation Due to the similarity of the postcranial skeleton of. C. antipodum and C. moriorum, and the absence of intact cranial material attributable to C. antipodum, distinguishing between the two species based on osteological features is not possible (Gill, 2003). Forbes (1893) considered that the smaller size of North Island birds warranted their separation as a different species. The separation of mainland NZ raven and Chatham Island raven was not questioned until they were lumped into C. moriorum, the NZ raven, without comment by Kinsky (1970). Subsequently only Gill (2003) has examined, in detail, the status of these two taxa. It was Gill’s opinion that the difference between measurements of mainland and the larger Chatham Island raven was sufficiently great to consider the two populations separate species. Gill also considered that the larger South Island raven was sufficiently distinct to allow it to be afforded recognition as a separate sub-species status, C. antipodum pycrafti. Gill’s decision was made despite a documented increase in size as one moved further south, consistent with Bergman’s rule (Gill, 2003, Figs. 6 & 7). The Chatham Island taxon is, on average, larger than mainland taxa however, there is considerable overlap (Gill, 2003, Figs. 2 & 3). There is a tendency for gigantism in most bird taxa on the Chatham Island but that those that are considered full species do not overlap significantly in either bone measurements [i.e. the Chatham Islands and New Zealand pigeon (Hemiphaga spp.; Millener and Powlesland, 2001)] or osteology [i.e. New Zealand and Chatham Island kaka (Nestor spp.; Wood et al., 2014)].

4.2. Taxonomic recommendations The results of our morphological analyses support the taxonomic recommendation of Brodkorb (1978) and Gill (2003) that Palaeocorax should be considered a junior synonym of Corvus. While hybridisation or incomplete lineage sorting can potentially mislead phylogenetic analyses of single loci such as the mitochondrial genome (e.g. Maddison, 1997), our molecular results are consistent with the synonymy of Corvus and Palaeocorax. We estimate that the divergence between the NZ raven mitochondrial lineage and that of the clade comprising the Australian raven, little raven, and forest raven occurred in the latest Pleistocene c.1.73 Ma (1.14–2.41 Ma 95% HPD). Our genetic data do not support the separation of the New Zealand taxa as species, as the divergence between C. antipodum and C. moriorum is estimated to have occurred only 0.15 Ma (0.05–0.27 Ma 95% HPD). The likelihood that the close molecular affinity between these taxa is an artifact of either incomplete lineage sorting or hybridisation is very low given that comparison of the post-cranial skeleton of each taxon by other authors (Pycraft, 1911; Gill, 2003) resulted in no identifiable osteological differences. Furthermore, there also appears to be considerable overlap in morphometrics between the two taxa (Gill, 2003). Consequently, we recommend that Corvus antipodum (Forbes, 1893) be considered a subspecies of Corvus moriorum Forbes, 1892. The status of the two subspecies defined by Gill (2003), C. antipodum antipodum and C. antipodum pycrofti, was not examined genetically in this study. We do consider that the manner in which measurements vary is consistent with Bergman’s rule and based on the South Island taxon’s (C. antipodum pycrofti) recent separation from Corvus moriorum, we recommend that Corvus antipodum pycrofti Gill, 2003 be considered a junior synonym of Corvus antipodum (Forbes, 1893).

142

R.P. Scofield et al. / Molecular Phylogenetics and Evolution 106 (2017) 136–143

4.3. A single invasion The analysis of the complete mitochondrial genome of the two putative species of NZ raven indicate that the diversity of corvids in prehistoric New Zealand came from a single invasion, unlike several diverse genus-level avian taxa in New Zealand – e.g. Porphyrio (see Trewick, 1997; Garcia-R and Trewick, 2015) and Gallirallus (see Kirchman, 2012), which came from three and at least five invasive events, respectively. Molecular dating indicates that the dispersal event that led to the speciation of the NZ ravens is likely to have occurred in the late Pliocene (New Zealand time stage Castlecliffian). The invasion of the ancestor of this taxon at this time coincides with the inferred arrival of several other notable taxa including the ancestors of Haast’s eagle (Aquila moorei; Bunce et al., 2005) and the black stilt (Himantopus novaezelandiae; Wallis, 1999). These invasions coincide with the onset of the Quaternary Glaciations and a significant decrease in global temperature worldwide. Such a series of glaciation events altered the composition of the New Zealand flora and fauna and undoubtedly opened up new niches for dispersive species to exploit (Atkinson and Millener, 1991). Although there are limited data available, there is some indication that the NZ raven was more widespread during the last Glacial maximum (Tennyson and Martinson, 2007 Fig. 2c; Supplementary Information Table 5). 4.4. A raven or a crow? The common raven Corvus corax has been shown by Jønsson et al., 2012, 2016 to be part of a clade of large bodied, heavy billed species primarily from Africa. All members of this clade weigh over 500 g and most have heavier bills with dorsally protuberant nares. Three species in Australia have historically been called ravens (C. coronoides, C. tasmanicus, C. mellori) due to having larger and heavier bills than their Australian relatives (the Australian crows), but genetic data do not support any relationship with Corvus corax (Jønsson et al., 2012). The New Zealand Corvus taxon weighed between 833 g and 955 g. These estimations are based on measurements in Gill (2003) and calculated using equations based on the femoral circumference of passerines (Campbell and Marcus, 1992). The New Zealand taxa did have a larger heavier bill than the Australian crows but did not have prominently raised nares. Whilst we support the continued usage of the name raven for the large-billed Australasian clade of Corvus (including the New Zealand taxa) we stress that this name does not imply any close relationship between Corvus corax and its close relatives. 4.5. Adapted to carrion eating? Examination of the sub-fossil record of NZ raven (Corvus moriorum antipodum; Fig. 2a; Supplementary Info Tables 5) indicates that since the last glacial maximum this taxon was primarily coastal and not found more than c. 25 km from the sea (see also Gill et al., 2010). For this reason it has been suggested that this species was a coastal scavenger (Atkinson and Millener, 1991). The greatest concentrations of bones of this species are found in association with fossil sea-lion rookeries (Fig. 2b; e.g. Delaware Bay: Worthy, 1994, Collins 2014; Long Bay, Chatham Islands, RPS pers obs.) and elephant seals (Mirounga leonina) haul out (or rookery) sites (e.g. Pounawea: Smith and James-Lee, 2010). Amongst Holocene sites Lake Poukawa (c. 25 km from the sea) is the only inland site where there is a significant number of individuals recorded (MNI = 12; Horn, 1983 c.f. Worthy and Holdaway, 1993: 228). The distinctiveness of aspects of the cranial morphology of the NZ raven, compared to other members of the genus Corvus, suggests a specialist feeding strategy. Amongst birds, an ossified septum and conchae nasales are features of species that rely on their

bills for crushing or thrusting motions (i.e. the seed eating Passerines; Herons Ardeidae and Woodpeckers Picidae) suggesting that ossification of these features may give species a stronger construction necessary for these foraging methods. It is notable that the New Caledonian crow Corvus moneduloides (Matsui et al., 2016), a species apparently adapted to delicate manipulation of objects with its bill, lacks an ossified septum. An ossified or enlarged bony palate is a feature that may in many groups be phylogenetic (i.e. Ciconiiformes) but that may have evolved in other taxa to allow the ingestion of large hard carrion items (e.g. in bearded vulture Gypaetus barbatus; this study). The presence of an ossified palate in the NZ raven suggests it may have been adapted for feeding on carrion, most probably from seal and sea-lion rookeries. The suggested utilisation of the snail Placostylus ambagiosus in the far North of New Zealand (Brooke, 2000) does not necessarily negate the hypothesis that seal and sea lion rookeries were the niche defining food source, as there would be a lean period for food for these taxa over the winter when pups and seal carrion was not available. 5. Conclusions Our phylogenetic analyses supports the monophyly of the NZ ravens and that the NZ ravens likely arrived in New Zealand as recently as the late Pliocene. The distribution of remains of this species and ecological niche modelling suggests coastal habitats during the Holocene. Unique features of the skull compared to other species of Corvus imply a unique dietary specialisation and we suggest that this species may have been a facultative scavenger at the vast sea-lion rookeries that once lined the beaches of the North Island and eastern and southern South Islands and Chatham Islands. NZ ravens may have been more widespread during the forest-free inter-glacial period of the Pleistocene. Our genetic analyses only support a single species with one or two subspecies, as opposed to the currently recognised two species and two subspecies. Acknowledgements We thank the Canterbury Museum for access to specimens, to Trevor Worthy for data on New Zealand sub-fossil Corvus distribution and N. Zelenkov and A. Mangold for discussion on fossil status. Grid computing facilities were provided by CIPRES (Cyberinfrastructure for Phylogenetic Research). Funding was provided by the Royal Society of New Zealand’s Marsden Fund and the Australian Research Council. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ympev.2016.09. 022. References Aiello-Lammens, M.A., Boria, R.A., Radosavljevic, A., Vilela, B., Anderson, R.P., 2015. spThin: an R package for spatial thinning of species occurrence records for use in ecological niche models. Ecography 38, 541–545 (ver. 0.1.0). Araújo, M.B., Peterson, A.T., 2012. Uses and misuses of bioclimatic envelope modeling. Ecology 93, 1527–1539. Atkinson, I.A.E., Millener, P.R., 1991. An ornithological glimpse into New Zealand’s pre-human past. Proceedings of the 20th International Ornithological Congress, vol. 1, pp. 129–192. Alloway, B.V., Lowe, D.J., Barrell, D.J.A., Newnham, R.M., Almond, P.C., Augustinus, P. C., Bertler, N.A.N., Carter, L., Litchfield, N.J., McGlone, M.S., Shulmeister, J., Vandergoes, M.J., Williams, P.W.NZ-INTIMATE members, 2007. Towards a climate event stratigraphy for New Zealand over the past 30 000 years (NZINTIMATE project). J. Quatern. Sci., 9–35

R.P. Scofield et al. / Molecular Phylogenetics and Evolution 106 (2017) 136–143 Baumel, J.J., Witmer, L.M., 1993. Osteologia. In: Baumel, J.J., King, A.S., Breazile, J.E., Evans, H.E., Vanden Berge, J.C. (Eds.), Handbook of Avian Anatomy: Nomina Anatomica Avium, Publications of the Nuttall Ornithological Club, No. 23, pp. 45–132. Becker, J.J., 1987. Neogene Avian Localities of North America. Smithsonian Institution Press, Washington, DC. Brodkorb, P., 1972. Neogene fossil jays from the Great Plains. The Condor 74, 347– 349. Brodkorb, P., 1978. Catalogue of fossil birds. Part 5 (Passeriformes). Bullet. Florida State Museum 23, 139–228. Brook, F.J., 2000. Prehistoric predation of the landsnail Placostylus ambagiosus Suter (Stylommatophora: Bulimulidae), and evidence for the timing of establishment of rats in northernmost New Zealand. J. Royal Soc. New Zealand 30, 227–241. Bunce, M., Szulkin, M., Lerner, H.R.L., Barnes, I., Shapiro, B., Cooper, A., Holdaway, R. H., 2005. Ancient DNA provides new insights into the evolutionary history of New Zealand’s extinct giant eagle. PLoS Biol. 3 (1), e9. http://dx.doi.org/ 10.1371/journal.pbio.0030009. Campbell Jr., K.E., Marcus, L., 1992. The relationship of hindlimb bone dimensions to body weight in birds. Natural History Museum Los Angeles County Sci. Series 36, 395–412. Collins, C.J., Rawlence, N.J., Worthy, T.H., Scofield, R.P., Tennyson, A.J.D., Smith, I., Knapp, M., Waters, J.M., 2014. Pre-human New Zealand sea lion (Phocarctos hookeri) rookeries on mainland New Zealand. J. Royal Soc. New Zealand 44 (1), 1–16. Drummond, A.J., Rambaut, A., 2007. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evol. Biol. 7. Elith, J., Graham, C.H., Anderson, R.P., Dudík, M., Ferrier, S., Guisan, A., Hijmans, R.J., Huettmann, F., Leathwick, J.R., Lehmann, A., Li, J., Lohmann, L.G., Loiselle, B.A., Manion, G., Moritz, C., Nakamura, M., Nakazawa, Y., Overton, J., Townsend Peterson, A., Phillips, S.J., Richardson, K., Scachetti-Pereira, R., Schapire, R.E., Soberón, Williams S., Wisz, M.S., Zimmermann, N.E., 2006. Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29, 129–151. Elith, J., Phillips, S.J., Hastie, T., Dudík, M., Chee, Y.E., Yates, C.J., 2011. A statistical explanation of MaxEnt for ecologists. Divers. Distrib. 17, 43–57. Forbes, H.O., 1892a. Aphanapteryx and other remains in the Chatham Islands. Nature 46, 252–253 (January). Forbes, H.O., 1892b. Report of meeting at which H.O. Forbes exhibited specimens of Chatham Islands birds. Bullet. British Ornithologists’ Club 1, 21–22 (March). Forbes, H.O., 1892c. Preliminary notice of additions to the extinct avifauna of New Zealand. Trans. Proc. N Z Inst. 24, 185–189 (May). Forbes, H.O., 1893. A list of the birds inhabiting the Chatham Islands. Ibis 5 (5th Series), 521–546. Garcia-R, J.-C., Trewick, S.A., 2015. Dispersal and speciation in purple swamphens (Rallidae: Porphyrio). Auk 132 (1), 140–155. Gill, B.J., Bell, B.D., Chambers, G.K., Medway, D.G., Palma, R.L., Scofield, R.P., Tennyson, A.J.D., Worthy, T.H., 2010. Checklist of the Birds of New Zealand, Norfolk and Macquarie Islands, and the Ross Dependency, Antarctica. Te Papa Press and Ornithological Society of New Zealand, Wellington. Gill, B., 2003. Osteometry and systematics of the extinct New Zealand ravens. J. Syst. Paleontol. 1, 43–58. Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A., 2005. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965–1978. Hijmans, R.J., Phillips, S., Leathwick, J., Elith, Jane, 2016. dismo: Species Distribution Modeling. R package version 1.0-12. . Horn, P.L., 1983. Subfossil avian remains from Poukawa, Hawke’s Bay, and the first record of Oxyura australis (Blue-billed duck) from New Zealand. J. Royal Soc. New Zealand 13, 67–78. Huxley, T.H., 1867. On the classification of birds: and on the taxonomic value of the modifications of certain of the cranial bones observable in that class. Zool. Soc. Lond. 1867, 415–472. James, H.F., Olson, S.L., 1991. Descriptions of thirty-two new species of birds from the Hawaiian Islands: Part II. Passeriformes. Ornithol. Monogr. 46, 1–88. Jønsson, K.A., Fabre, P.H., Ricklefs, R.E., Fjeldså, J., 2011. Major global radiation of corvoid birds originated in the proto-Papuan archipelago. Proc. Natl. Acad. Sci. USA 108, 2328–2333. Jønsson, K.A., Fabre, P.-H., Irestedt, M., 2012. Brains, tools, innovation and biogeography in crows and ravens. BMC Evol. Biol. 12, 72. Jønsson, K.A., Fabre, P.-H., Kennedy, J.D., Holt, B.G., Borregaard, M.K., Rahbek, C., Fjeldså, J., 2016. A supermatrix phylogeny of corvoid passerine birds (Aves: Corvides). Mol. Phylogenet. Evol. 94 (Part A), 87–94. Kinsky, F.C., 1970. Annotated Checklist of the Birds of New Zealand (convener). A.H. and A.W. Reed, Wellington. Kircher, M., 2012. analysis of high-throughput ancient DNA sequencing data. Ancient DNA: Methods Protocols, 197–228. Kirchman, J.J., 2012. Speciation of flightless rails on islands: a DNA-based phylogeny of the typical rails of the Pacific. Auk 29 (1), 56–69. Kurochkin, E.N., Sobolev, D.V., 2004. Miopica paradoxa gen. et sp. n. – Hodse pol B Bbl Mboweyodsx Copor. Vestn. Zool. 38, 87–90.

143

Lanfear, R., Calcott, B., Ho, S.Y., Guindon, S., 2012. PartitionFinder: combined selection of partitioning schemes and substitution. Mol. Biol. Evol. 29, 1537– 1719. Li, H., Durbin, R., 2009. Fast and accurate short read alignment with BurrowsWheeler transform. Bioinformatics 25, 1754–1760. Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., Marth, G., Abecasis, G., Durbin, R., Subgroup, G.P.D.P., 2009. The Sequence Alignment/Map (SAM) format and SAMtools. Bioinformatics 25, 2078–2079. Lindgreen, S., 2012. AdapterRemoval: easy cleaning of next-generation sequencing reads. BMC Res. Notes 5, 337. Maddison, W.P., 1997. Gene trees in species trees. Syst. Biol. 46, 523–536. Matsui, H., Hunt, G.R., Oberhofer, K., Ogihara, N., McGowan, K.J., Mithraratne, K., Yamasaki, T., Gray, R.D., Izawa, E.-I., 2016. Adaptive bill morphology for enhanced tool manipulation in New Caledonian crows. Sci. Reports 6, 22776. Mayr, G., Clarke, J., 2003. The deep divergences of neornithine birds: a phylogenetic analysis of morphological characters. Cladistics 19, 527–553. Meyer, M., Kircher, M., 2010. Illumina sequencing library preparation for highly multiplexed target capture and sequencing. Cold Spring Harbor Protocols 2010, 1–10. Mitchell, K.J., Llamas, B., Soubrier, J., Rawlence, N.J., Worthy, T.H., Wood, J., Lee, M.S. Y., Cooper, A., 2014a. Ancient DNA reveals elephant birds and kiwi are sister taxa and clarifies ratite bird evolution. Science 344, 898–900. Mitchell, K.J., Wood, J.R., Scofield, R.P., Llamas, B., Cooper, A., 2014b. Ancient mitochondrial genome reveals unsuspected taxonomic affinity of the extinct Chatham duck (Pachyanas chathamica) and resolves divergence times for New Zealand and sub-Antarctic brown teals. Mol. Phylogenet. Evol. 70, 420–428. Millener, P.R., Powlesland, R.G., 2001. The Chatham Islands pigeon (parea) deserves full species status; Hemiphaga chathamensis (Rothschild 1891); Aves: Columbidae. J. Royal Soc. New Zealand 31 (2), 365–383. Perry, G.L., Wheeler, A.B., Wood, J.R., Wilmshurst, J.M., 2014. A high-precision chronology for the rapid extinction of New Zealand moa (Aves, Dinornithiformes). Quatern. Sci. Rev. 105, 126–135. Phillips, S.J., Anderson, R.P., Schapire, R.E., 2006. Maximum entropy modeling of species geographic distributions. Ecol. Model. 190, 231–259. Phillips, S.J., Dudík, M., Elith, J., Graham, C.H., Lehmann, A., Leathwick, J., Ferrier, S., 2009. Sample selection bias and presence-only distribution models: implications for background and pseudo-absence data. Ecol. Appl. 19, 181–197. Pycraft, W.P., 1907. Mr. W. P. Pycraft exhibited a skull, and briefly described the skeleton, of Palaeocorax moriorum, a fossil Raven from the Chatham Is., New Zealand. Bullet. British Ornithologists Club 23, 95. Pycraft, W.P., 1911. On the skeleton of Palaeocorax moriorum. Novitates Zoologicae 18, 123–128. R Core Team, 2016. R: a Language and Environment for Statistical Computing. Version 3.2.4. Radosavljevic, A., Anderson, R.P., 2014. Making better Maxent models of species distributions: complexity, overfitting and evaluation. J. Biogeogr. 41, 629–643. Raes, N., ter Steege, H., 2007. A null-model for significance testing of presence-only species distribution models. Ecography 30, 727–736. Schodde, R., Mason, I.J., 1999. Directory of Australian Birds: Passerines: Passerines. CSIRO Publishing. Schubert, M., Ginolhac, A., Lindgreen, S., Thompson, J.F., Al-Rasheid, K.A., Willerslev, E., Krogh, A., Orlando, L., 2012. Improving ancient DNA read mapping against modern reference genomes. BMC Genom. 13 (1), 1. Smith, I., James-Lee, T., 2010. Data for an Archaeozoological Analysis of Marine Resource Use in Two New Zealand Study Areas. Otago Archaeological Laboratory Report No. 7. University of Otago, Dunedin. Stamatakis, A., 2006. RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 22, 2688– 2690. Tennyson, A.J.D., Martinson, P., 2007. Extinct Birds of New Zealand. Te Papa Press, Wellington. Trewick, S.A., 1997. Flightlessness and phylogeny amongst endemic rails (Aves: Rallidae) of the New Zealand region. Philos. Trans. R Soc. B 352, 429–446. Wallis, G., 1999. Genetic Status of New Zealand Black Stilt (Himantopus novaezelandiae) and Impact of Hybridisation. Conservation Advisory Science Notes No. 239. Department of Conservation, Wellington. Wood, J.R., Mitchell, K.J., Scofield, R.P., Tennyson, A., Fidler, A.E., Wilmshurst, J.M., Llamas, B., Cooper, A., 2014. An extinct nestorid parrot (Aves, Psittaciformes, Nestoridae) from the Chatham Islands, New Zealand. Zool. J. Linnean Soc. 172 (1), 185–199. Wood, J.R., Mitchell, K.J., Scofield, R.P., De Pietri, V.L., Rawlence, N.J., Cooper, A., 2016. Phylogenetic relationships and terrestrial adaptations of the extinct laughing owl, Sceloglaux albifacies (Aves: Strigidae). Zool. J. Linnean Soc. Early View. http://dx.doi.org/10.1111/zoj.12483. Worthy, T.H., Holdaway, R.N., 1993. Quaternary fossil faunas from caves in the Punakaiki area, west coast, South Island, New Zealand. J. Royal Soc. New Zealand 23, 147–254. Worthy, T.H., 1994. Holocene breeding of sea lions (Phocarctos hookeri) at Delaware Bay, Nelson, South Island, New Zealand. New Zealand Nat. Sci. 21, 61–70.