Prions (Pachyptila) are small seabirds with a Southern Hemisphere breeding distribution. Antarctic prion (Pachyptila desolata) and Salvin’s prion (P. salvini) are two species that are regularly recorded in New Zealand as beach-wrecks but they are difficult to distinguish morphologically. Salvin’s prion is restricted to breeding on the Prince Edward Islands and Crozet Islands in the Indian Ocean but Antarctic prions have a circumpolar breeding distribution on numerous sub-Antarctic and Antarctic islands in the Southern, South Atlantic and Indian Oceans. Our aim was to examine the level of mitochondrial DNA (mtDNA) structuring within Antarctic prion and Salvin’s prion colonies, to test whether this technique can determine the provenance of beach-cast birds. The Auckland Islands Antarctic prion population exhibited distinct mtDNA haplotypes from all other populations, supporting the suggestion that these islands may have been an ice-free refugium during the Last Glacial Maximum. All other sampled breeding populations shared haplotypes, limiting the use of these sequences for determining the provenance of beach-cast birds. None of our museum specimens of Salvin’s prion collected from breeding colonies produced DNA sequences. This result indicates that the method by which these specimens, which were collected in the 1960s and 70s, were preserved, or subsequent treatments, has resulted in the loss of their DNA.

cytochrome b, cytochrome oxidase I, phylogeography, population structure, Procellariiformes, seabird

Prions (Pachyptila) are small seabirds with a Southern Hemisphere breeding distribution. Eight extant species of prions are currently recognised (Checklist Committee (OSNZ) (2022); Shepherd et al. 2022) and they are closely related and estimated to have diverged in the last 6 million years (Masello et al. 2019). The species are similar in appearance and behaviour, making their identification challenging, especially at sea (Harper 1980). Prions mainly differ in the structure and size of their bills (Warham 1990), which reflects differences in prey selection and feeding strategies.

New Zealand is the centre of prion diversity, with five species breeding within the region (Checklist Committee (OSNZ) (2022); Shepherd et al. 2022), mostly on remote, predator-free islands. Prions are also regularly found beach-cast on the New Zealand mainland, particularly during mass mortality events (wrecks) when many thousands of individuals may die, often following stormy weather (Harper 1980; Powlesland 1989; Warham 1996). Such beach-cast birds are particularly difficult to identify because immature birds of larger-billed species can look like adults of smaller-billed species and shrinkage in bill dimensions can occur as dead specimens dry out (Harper 1980).

In this paper we focus on Antarctic prion (Pachyptila desolata (Gmelin, 1789)) and Salvin’s prion (P. salvini (Mathews, 1912)), two species that are commonly confused morphologically (Harper 1980) and that have been regularly recorded in New Zealand wrecks (Harper 1980; Powlesland 1989). Salvin’s prion appears to have arisen through hybridisation between broad-billed prion (P. vittata (G. Forster, 1777)) and Antarctic prion and has a bill intermediate in size between these two species (Massello et al. 2019). The breeding distribution of Salvin’s prion is restricted to the Prince Edward Islands and Crozet Islands in the Indian Ocean (Fig. 1) (Marchant and Higgins 1990). In contrast, Antarctic prions have a much wider circumpolar distribution, breeding on numerous sub-Antarctic and Antarctic islands in the Southern, South Atlantic and Indian Oceans (Fig. 1) (Marchant and Higgins 1990). Morphological variation is present within Antarctic prions that is geographically based. Historically, the taxonomy of Antarctic prions has been unstable with various authors recognising up to six subspecies (e.g. Mathews 1912, 1934; Falla 1940; Tickell 1962; Bretagnolle 1990) or no subspecies (Harper 1980). Most recent treatments consider the species monotypic (Marchant and Higgins 1990; Dickinson and Remsen 2013; Checklist Committee (OSNZ) (2022)).

Figure 1. 

Distribution map of Antarctic (Pachyptila desolata) and Salvin’s (Pachyptila salvini) prion breeding colonies. The symbol ‘?’ indicates a likely but unconfirmed breeding colony and ‘†’ denotes a colony thought to be extinct. Populations included in this study are underlined.

A number of population genetic studies on prions have recently been published and these have shown that some species exhibit little differentiation between populations whilst others exhibit considerable structuring. The fairy and fulmar prion clade (P. turtur, P. crassirostris and P. pyramidalis) demonstrated a high level of genetic structuring with both genomic SNPs and mitochondrial DNA (mtDNA) sequences, resulting in further taxa being recognised (Shepherd et al. 2022). In contrast, populations of broad-billed prion from the Atlantic Ocean and the New Zealand region could not be distinguished with either mtDNA sequences (cytochrome b and cytochrome oxidase I (COI)) or 18 microsatellite loci (Masello et al. 2021). Quillfeldt et al. (2017) examined cytochrome b sequences plus genotypes from 25 microsatellite loci of Indian and Atlantic Ocean populations of Antarctic prion and Thin-billed prion P. belcheri (Mathews, 1912) and found no population structuring within either species. Masello et al. (2021) subsequently genotyped the same 25 microsatellite loci from Macquarie Island Antarctic prions and found them indistinguishable from Indian and Atlantic Ocean birds. There are no published mtDNA sequences (or microsatellite data) from New Zealand Antarctic prions. For Salvin’s prions there are published COI and cytochrome b sequences published from Marion Island but only a single sequenced individual from the Crozet Islands (Masello et al. 2021), which hold the largest populations of this species.

Our aim for this study was to determine the level of mtDNA genetic structuring within Antarctic prion and Salvin’s prion colonies. In particular, is the morphological variation observed in Antarctic prions supported by genetic differences? Determining the level of connectivity between populations of these species is important for tracking threats on their populations and to assess whether DNA sequencing can be used to determine the provenance of beach-wrecked birds.

Both modern samples and historical skins were included in this study in order to cover the geographic spread of Antarctic and Salvin’s prions (Fig. 1, Table 1, Suppl. material 1), when combined with published sequences. Thirty-six Antarctic prion and six Salvin’s prion specimens were selected for sampling. Footpad or skin tissue was sampled from study skins from the Museum of New Zealand Te Papa Tongarewa collection using a new sterile scalpel blade for each specimen. These study skins dated from between 1912 and 1992 and most were collected from breeding colonies (Table 1). Additionally, blood samples of Antarctic prions from the Auckland Islands were collected under a permit from the Department of Conservation (permit number 97330-FLO). Sampling methods followed approved ethical procedures as required under this permit.

We targeted the mitochondrial cytochrome oxidase (COI) and cytochrome b loci because these markers have previously been used successfully to differentiate prion populations (Shepherd et al. 2022) and primers are available, including short internal primers for amplifying the degraded DNA typically found in museum specimens. Also, cytochrome b sequences for Antarctic prions from South Georgia and the Kerguelen Islands are available on Genbank for comparison.

DNA extraction, PCR amplification and sequencing of COI and cytochrome b loci followed Shepherd et al. (2022), except that cytochrome b from the historical specimens was sequenced by amplification of five short overlapping fragments. These fragments were amplified and sequenced with the internal primers provided in Masello et al. (2021).

Sequences were edited in Sequencer 5.4.6 (Gene Codes Corporation) and, because they contained no insertion/deletion events (indels), were aligned manually to sequences available in GenBank (Table 1). Thin-billed and Salvin’s prion sequences were used as outgroups (Masello et al. 2019).

The relationships between the mtDNA sequences were examined by constructing median-joining networks (Bandelt et al. 1999) in PopART (Leigh and Bryant 2005) and phylogenetic trees using maximum likelihood (ML) and Bayesian Inference (BI). Separate networks were produced for each locus because only cytochrome b was available for some individuals on GenBank. For the phylogenetic analyses only the samples with both loci sequenced were included and the two loci were concatenated.

ML analyses were performed with the PhyML v3.0 (Guindon et al. 2010) web server (http://www.atgcmontpellier.fr/phyml/). Heuristic searches were performed with SPR branch-swapping and 10 random addition sequence replicates, with the best fit model of sequence evolution (HKY85) determined by Smart Model Selection (Lefort et al. 2017) and the Akaike Information Criterion. Branch support was assessed with 1000 bootstrap (BS) pseudoreplicates.

MrBayes v3.2.7 (Ronquist et al. 2012) was used to perform BI with the dataset partitioned by locus and substitution model parameters unlinked between the loci. Two concurrent analyses, each with four Markov chains of fifty million generations were run, sampling every 1000 generations and with nst = 6, rates = invgamma and default parameters. The first 20% of samples were discarded as ‘‘burn-in”, after this point the standard deviation of split frequencies was below 0.01. Tracer v.1.71 (Rambaut et al. 2018) also confirmed that stationarity had been reached.

Population differentiation between colonies was estimated by calculating the global fixation index (F_ST) and two parameters of population subdivision (G_ST and N_ST) from the cytochrome b data (this locus had the least missing data) in SPADS v1.0 (Dellicour and Mardulyn 2014). Statistical significance of these values was assessed by 1000 random permutations. N_ST considers the relationships between haplotypes, whereas G_ST is calculated using only haplotype frequency data. G_ST and N_ST were compared with a permutation test with 10 000 permutations.

The geographic structure of the cytochrome b variation in Antarctic prions was examined by spatial analysis of molecular variance (SAMOVA, Dupanloup et al. 2002), implemented in SPADS 1.0. SAMOVA partitions groups of populations by maximising the proportion of the total genetic variance due to differences between groups of populations (F_CT). The number of groups (K) was set to vary between 2 and 5, with SAMOVA run with 10 000 iterations and 10 repetitions. For calculating the population differentiation measures and running the SAMOVA only birds from known breeding colonies were included and the single specimen from the South Sandwich Islands was grouped with the larger population sample from nearby South Georgia (they shared a haplotype).

Twenty-three of the thirty-six Antarctic prions amplified for at least one locus, but none of the Salvin’s prions produced any PCR products (Table 1, Suppl. material 1). Newly-generated sequences have been deposited in GenBank (Accession numbers provided in Table 1).

The COI and cytochrome b alignments of Antarctic prion sequences were 702 bp and 812 bp in length, respectively. Neither locus contained internal stop codons when translated. For the COI locus we recovered 8 haplotypes, defined by 9 variable sites and for cytochrome b there were 22 haplotypes, defined by 21 variable sites.

The relationships between Antarctic prion haplotypes at the COI and cytochrome b loci are shown in the median-joining networks (Fig. 2A, B). The outgroups (thin-billed prion and the Salvin’s prion) both joined the COI network by connecting to different haplotypes from the Auckland Islands. In the cytochrome b network both outgroups connected to the same haplotype that was sequenced in Antarctic prions from the Auckland Islands and one of the Campbell Island specimens.

Figure 2. 

Median-joining networks of (A) cytochrome b and (B) COI for Antarctic prions. Sampling locations are colour-coded and the size of each circle is proportional to frequency. Mutational steps between haplotypes are shown as hatch marks.

For both COI and cytochrome b, no haplotypes were shared between the Auckland Islands and elsewhere, except for one of the two specimens from Campbell Island. However, the genetic differences were small, with only a single substitution separating the most closely-related haplotypes from the Auckland Islands and South Georgia, Kerguelen and Antarctica. There was no clustering by location for the remaining sequences (Kerguelen, Heard, South Georgia, South Sandwich and Antarctica).

The ML and BI phylogenies had similar topologies and the ML phylogeny is presented in Fig. 3 with the support values from both analyses. The deeper relationships within the phylogeny were largely unresolved. However, the sequences from Kerguelen, Heard, South Sandwich, Antarctica and one from Campbell Island formed a well-supported clade (1.00 PP/85 BS). Within the Auckland Islands two clades received support in the BI analysis (1.00 PP/61 BS and 0.95 PP/65 BS) and the second Campbell Island sequence was sister to one of these clades (1.00 PP/61 BS).

Figure 3. 

Maximum likelihood phylogeny of concatenated COI and cytochrome b sequences for Antarctic prions. Support values for nodes are as follows: Bayesian posterior probability (PP)/maximum likelihood bootstrap (BS). Only PP values over 0.9 and BS values over 60% are shown.

Measures of population differentiation between Antarctic prion colonies were moderate. G_STwas 0.287 (P < 0.05), whereas N_ST was 0.447 (P < 0.001). N_ST was significantly higher than G_ST (P < 0.01), which suggests a phylogeographic component to the structuring with haplotypes in close geographic proximity more likely to also have a close genetic relationship. The global F_ST was also moderate (0.321; P < 0.0001).

The SAMOVA (Suppl. material 2) indicated a lack of higher-level structuring in Antarctic prions. For all values of K, including K = 2, single populations were partitioned into exclusive groups rather than groups of populations. At K = 2 the Auckland Islands was split from the remaining populations.

LS - Writing - Investigation, Formal analysis, Writing - Original draft. AT - Conceptualisation, Investigation, Writing - Review and Editing. CM - Conceptualisation, Investigation, Writing - Review and Editing

Fieldwork to the Auckland Islands was funded from the Museum of New Zealand Te Papa Tongarewa acquisition fund. LDS received support from a Rutherford Discovery Fellowship from the Royal Society of New Zealand (contract number RDF-MNZ1201). CMM received support from Charles-Andre Bost and the Institut Polaire Français Paul-Émile Victor during fieldwork on the Kerguelen Islands. We thank Graeme Taylor, Rodrigo Salvador and an anonymous reviewer for helpful comments during the review process.