Pliocene fossils support a New Zealand origin for the smallest extant penguins

Abstract

A late Pliocene (3.36–3.06 Ma) exposure of the Tangahoe Formation on the North Island of New Zealand preserves close fossil relatives of many extant seabird clades. Here, we report an extinct member of the little penguin (Eudyptula Bonaparte, 1856) lineage from the Tangahoe Formation—the smallest extinct crown penguin yet known. Eudyptula wilsonae n. sp. is based on the nearly complete skulls of an adult and a fledged but immature individual. Both skulls show more slender proportions than modern little penguins and precede genome-derived estimates for the divergence between Eudyptula minor minor Forster, 1781 (endemic to New Zealand) and Eudyptula m. novaehollandiae Stephens, 1826 (native to Australia and recently established in New Zealand). This raises the possibility that the fossil taxon represents a lineage directly ancestral to extant little penguins. Our results support a Zealandian origin for little penguins, with subsequent Pleistocene dispersal to Australia and a more recent Holocene range expansion of Eudyptula m. novaehollandiae back into New Zealand.

UUID: https://zoobank.org/a415f70a-e441-4920-85e0-75f6457577ea

Introduction

Aotearoa New Zealand is home to three of the six extant penguin genera, including the little penguins, Eudyptula Bonaparte, 1856. Little penguins occur across southern Australia and the New Zealand archipelago. They are remarkable for being among the smallest obligate marine endotherms yet having breeding colonies distributed across the broadest range of mean annual sea-surface temperatures observed in penguins (Stonehouse, 1967; Borboroglu and Boersma, 2013; see also Supplemental data). Their biogeographic origins and morphological evolution are therefore of great interest.

Eudyptula currently comprises two lineages: the kororā or New Zealand Little Penguin, Eudyptula minor minor Forster, 1781, which is endemic to New Zealand; and the Australian Little Penguin Eudyptula m. novaehollandiae Stephens, 1826. The Australian Little Penguin is native to Australia where the majority of its population occurs, but it also breeds in the Otago region of New Zealand, where it likely arrived after humans first settled in the region (Grosser et al., 2015). The status of these two taxa remains unsettled, with some studies considering them separate species (e.g., Grosser et al., 2017; Cole et al., 2022). Their estimated molecular divergence and relative genetic distance exceed those of other recognized sister species within crown penguins, especially in Eudyptes (Cole et al., 2022, fig. S2). However, they are visually indistinguishable morphologically and so it was only recently recognized that some of the little penguins breeding in Otago in fact belong to the Australian lineage (Grosser et al., 2016). Here, we follow the recently published fifth edition to the checklist of New Zealand birds (Ornithological Society of New Zealand Checklist Committee, 2022) in treating the two taxa as subspecies (see also Miskelly et al., 2023).

The fossil record of Eudyptula is obscured by sparse data and conflicting fossil identifications. Putative records were summarized by Worthy et al. (2009) and include just two occurrences: a small humerus and ulna from late Oligocene or early Miocene Otekaike Limestone of southern New Zealand (Fordyce, 1991, p. 1247–1249), which was reidentified as a stem penguin by Ando (2007), and ‘a few late Pliocene and Recent specimens, not clearly distinct from living Eudyptula,’ which Simpson (1975, p. 21) mentioned in his review of New Zealand penguin fossils without providing specific localities or specimen numbers. We searched several collections to trace this material but found no specimens plausibly assignable to Eudyptula that are older than late Pleistocene. We surmise that Simpson's (1975) reference pertains to material that has since been traced to Pleistocene infill deposits, which means that the only substantiated records of Eudyptula are late Pleistocene and Holocene subfossils (Worthy, 1998; Worthy and Grant-Mackie, 2003).

Here, we describe two Eudyptula fossils recently collected from the Pliocene Tangahoe Formation of New Zealand and, thus, for the first time confidently establish the presence of little penguins in Zealandia during the Neogene.

Materials and methods

Skull measurements and specimen age

We collected the following measurements from the skulls of Eudyptula minor minor (N = 68), Eudyptula m. novaehollandiae (N = 19), and the fossils described here (N = 2) (Tables 1, S1): (1) maximum skull length, i.e., the distance between the bill tip and caudal margin of supraoccipital; (2) width of internarial bar, i.e., the maximum distance between the lateral edges of premaxillae in dorsal view; (3) interorbital width, i.e., the minimum distance between the orbits excluding the salt glands; and (4) postorbital width, i.e., the distance between the lateral edges of the postorbital processes. Not all measurements could be collected from all specimens owing to differences in preservation or preparation. Measurements (1) and (4) correspond to greatest length and greatest breadth as reported by Grosser et al. (2017) from the skulls of 84 and 82 individuals, respectively.

Animal age at time of death was inferred from suture closure and supported by museum record data. Individuals were considered adult if the internasal, frontal-nasal, and rostral portion of the interpremaxillary sutures were obliterated, and immature if any of these sutures remained visible. Especially young individuals retain visible sutures in their skull roof. Data from specimens for which all four measurements were available (N = 60 for Eudyptula m. minor, N = 19 for Eudyptula m. novaehollandiae, and two fossils) were summarized via principal component analysis using stats::prcomp in R ver. 4.1.2 (R Core Team, 2021).

Phylogenetic analysis

We scored the newly-described fossils into a combined morphological and molecular matrix modified from Ksepka et al. (2023). Anatomical terminology is consistent with terminology used in other recent descriptions of fossil penguins (e.g., Ksepka et al., 2023) and originally adapted from Baumel and Witmer (1993). We added one morphological character: (98) lacrimal, descending process: (0) smoothly contacts jugal bar along dorsal edge of latter, (1) ends in a flange formed by laterally deflected and widened ventral extremity. We further modified the scoring for eight taxa for one character: (117) mandible, length of the dorsal edge of the dentary relative to the mandibular ramus length in lateral view. Upon reassessment, we found that Megadyptes antipodes antipodes Hombron and Jacquinot, 1841, Megadyptes a. richdalei Tennyson and Cole in Cole et al., 2019, Pygoscelis papua Forster, 1781, Spheniscus demersus Linnaeus, 1758, S. humboldti Meyen, 1834, S. magellanicus Forster, 1781, and S. mendiculus Sundevall, 1871 had been incorrectly scored as state 117:1, dorsal edge of the dentary approximately half the length of the ramus, and changed this scoring to 117:0, dorsal edge of the dentary markedly more than half the length of the ramus. Madrynornis mirandus Acosta Hospitaleche et al., 2007 had previously not been scored for this character, but was scored as 117:0 upon re-examining a cast of the holotype.

The revised matrix comprises 281 morphological characters scored for 70 penguin taxa (20 extant, 3 subfossil, and 47 fossil taxa; see Supplemental data). The morphological dataset originated with the extant penguin matrix of Bertelli and Giannini (2005), which was extended to included fossil taxa by Ksepka et al. (2006) and expanded iteratively over many analyses. Studies that contributed characters include O'Hara (1989), Giannini and Bertelli (2004), Bertelli and Giannini (2005), Ksepka et al. (2006, 2012, 2023), Acosta Hospitaleche et al. (2007), Ando (2007), Clarke et al. (2007, 2010), Ksepka and Clarke (2010), Ksepka and Thomas (2012), Chávez Hoffmeister et al. (2014), Blokland et al. (2019), Thomas et al. (2020), and Giovanardi et al. (2021).

Phylogenetic analyses were performed in PAUP*4.0a168 (Swofford, 2003) using a heuristic search strategy with 10,000 replicates of random taxon addition (holding 10 trees per replicate), with tree bisection and reconnection branch swapping limited to 10 million rearrangements per replicate. All characters were equally weighted. Multistate codings were considered to represent polymorphism. Zero-length branches were collapsed.

Repositories and institutional abbreviations

The specimens described here are lodged at the Museum of New Zealand Te Papa Tongarewa, Wellington, New Zealand (NMNZ). Specimens used for comparison were from: Department of Ornithology, American Museum of Natural History, New York, USA (AMNH); Canterbury Museum, Christchurch, New Zealand (CM); Institute of Biology, University of Białystok, Poland (IB/P/B); Museo Egidio Feruglio, Trelew, Argentina (MEF); Museo de La Plata, Argentina (MLP); Muséum National d'Histoire Naturelle, Paris, France (MNHN); Museum of San Marcos University, Lima, Peru (MUSM); The Natural History Museum, London, UK (NHMUK); Otago Museum, Dunedin, New Zealand (OM); Geology Museum, University of Otago, Dunedin, New Zealand (OU); Iziko South African Museum, Cape Town (SAM); Sección de Paleontología del Museo Nacional de Historia Natural, Santiago, Chile (SGO-PV); University of California Museum of Paleontology, Berkeley, USA (UCMP); and Western Australian Museum, Perth (WAM). An unnumbered specimen from the teaching collections at Massey University, Auckland, New Zealand, was also studied.

Systematic paleontology

Aves Linnaeus, 1758 (sensu Gauthier, 1986)
Sphenisciformes Sharpe, 1891 (sensu Clarke et al., 2003)
Spheniscidae Bonaparte, 1831
Eudyptula Bonaparte, 1856

Type species

Eudyptula minor Forster, 1781.

Eudyptula wilsonae new species
 Figures 1–5

Figure 1. Skull of Eudyptula wilsonae n. sp., specimen NMNZ S.048854, presented alongside a skull from Eudyptula m. minor Forster, 1781, specimen NMNZ S.000863, for comparison. Left lateral view: (1) NMNZ S.048854; (3) NMNZ S.000863. Caudal view: (2) NMNZ S.048854; (4) NMNZ S.000863. Right lateral view: (5) NMNZ S.048854. Dorsal view: (6) NMNZ S.048854; (8) NMNZ S.000863. (7) Detail of nasal region of NMNZ S.048854 identifying fused frontal-nasal suture (compare with specimen NMNZ S.048855 in Fig. 1.2).

Figure 2. Skull of Eudyptula wilsonae n. sp., specimen NMNZ S.048855: (1) left lateral view; (2) cranial view; (3) right lateral view; (4) caudal view; (5) dorsal view. (6) Detail of nasal region of skull identifying unfused sutures between premaxillae and nasals.

Figure 3. Skull of Eudyptula wilsonae n. sp., specimen NMNZ S.048855, with emphasis on right quadrate, compared to right quadrate from extant Eudyptula minor minor Forster, 1781. (1) Right lateral view of NMNZ S.048855 with emphasis on quadrate; image taken before acid preparation to expose right quadrate. (2) Right lateral view of right quadrate from an immature Eudyptula m. minor (unnumbered specimen from Massey University teaching collection). (3) Otic process of right quadrate from NMNZ S.048855 in oblique dorsolateral view; image taken after acid preparation. (4) Right quadrate from immature Eudyptula m. minor in similar orientation to quadrate in (3) (unnumbered specimen from Massey University teaching collection).

Figure 4. Analysis of measurements from modern and fossil Eudyptula skulls. (1) Principal component biplot from analysis performed on four measurements; biplot shows scores and loadings for principal component one (PC1) and PC2. (2) Four measurements used for analyses and shown on Eudyptula wilsonae n. sp. paratype, specimen NMNZ S.048855. i-i’ = distance between bill tip and caudalmost surface of the supraoccipital (i.e., maximum length); ii-ii’ = maximum distance between lateral edges of premaxillae in dorsal view (i.e., internarial width); iii-iii’ = minimum distance between the left and right frontals between the orbits in dorsal view without including the salt gland depressions (i.e., interorbital width); iv-iv’ = distance between lateral edges of postorbital processes (i.e., postorbital width). (3) Maximum length of skull shown against distance between lateral edges of postorbital processes. (4) Maximum length of skull shown against distance between lateral edges of postorbital processes; specimen age inferred from suture closure.

Figure 5. Skulls of extant Eudyptula and Eudyptula wilsonae n. sp., comparisons using placement of homologous points. (1) Dorsal and right lateral views of Eudyptula m. minor skulls with features identified. (2) Solid and wireframe views of modern and fossil skulls in dorsal and right lateral views. (3) Homologous points from all three skulls overlain in dorsal and right lateral views, aligned on point demarking the nasofrontal flexure. yellow = Eudyptula m. minor Forster, 1781 (adult, NMNZ S.000863); purple = Eudyptula wilsonae n. sp. (adult, NMNZ S.048854); blue = Eudyptula wilsonae n. sp. (immature, NMNZ S.048855).

Holotype

NMNZ S.048854: cranium and upper beak, mandible, and right quadrate of an adult individual (Figs. 1, 5; Table 1).

Table 1. Skull measurements (in mm) of Eudyptula wilsonae sp. nov. compared to summaries of those from other Eudyptula taxa; expressed as mean ± standard deviation (N), measured with digital calipers. See Supplemental Data for individual measurements and other data, including age class when known.

Paratype

NMNZ S.048855: cranium and upper beak, right lacrimal, and right quadrate of immature individual (Figs. 2–5; Table 1).

Diagnosis

Differs from extant Eudyptula minor in having more slender skull proportions, including proportionally narrower transverse distances between the caudal margins of the nares, the lateral edges of the salt-gland fossae, and the dorsal rim of the temporal fossae (Figs. 4, 5).

Occurrence

Late Pliocene (Piacenzian) Tangahoe Formation, exposed in the southern Taranaki region of the North Island of New Zealand (Naish et al., 2005). Local Waipipian stage, with exposure constrained to 3.36–3.06 Myr based on oxygen isotope stage and magnetic polarity data (Naish et al., 2005; Raine et al., 2015). Specimens were surface collected as boulders from the base of a nearby exposure of the Tangahoe Formation and do not have an exact location lodged in the Fossil Record Electronic Database (https://fred.org.nz/, but see Q21/f0002 for a nearby location).

Description

The holotype (adult) skull falls within the size range of adult modern little penguins, whereas the paratype (immature) specimen is smaller than any of the individuals examined here or by Grosser et al. (2017) (Tables 1, S1). The parietal region of the holotype has been slightly compressed dorsoventrally as shown by the apex of the nuchal crest nearly aligning with the dorsalmost regions of the temporal fossae, and the temporal crests and lateral margins of the postorbital processes have been eroded in both skulls. There is, however, no sign of transverse compression, indicating that the slenderness of the skull is a genuine feature.

The tip of the premaxilla is weakly hooked in both skulls, as in extant Eudyptula. The tip is deeply pitted with neurovascular foramina and bears a shallow groove that is continuous with the anteroventral margin of the nares. The internarial bar is slender and slightly swollen near its midpoint in the adult skull (but not in the immature skull); in lateral view, its dorsal edge curves smoothly to the tip of the beak. In the adult skull, the premaxilla-nasal suture is largely obliterated and only visible caudally as in adults of extant Eudyptula and Spheniscus Brisson, 1760. In other extant penguins, this suture remains visible along its entire length. In the immature skull, both the premaxilla-nasal and the interpremaxillary sutures remain visible, the premaxillae are elevated dorsally above the nasals and frontal (further indicating an open premaxilla-nasal suture), and the left and right premaxillae are visibly separated along the caudal section of internarial bar.

The salt gland fossae are relatively wide, rugose, and oriented dorsolaterally. They are strongly excavated and each terminates in a sharp edge without a raised lateral margin as in extant Eudyptula (but not in Eudyptes Vieillot, 1816, Megadyptes Milne-Edwards, 1880, and Pygoscelis Wagler, 1832). The postorbital processes are directed ventrally and taper to a narrow point. The temporal fossae are deep, as in most extant penguins (shallower in Pygoscelis and Aptenodytes Miller, 1778); and widely separated, as in all crown penguins except Spheniscus. The jugal lacks a prominent dorsal process (present in Eudyptes) and is only slightly curved in lateral view, as in extant Eudyptula, Spheniscus, and Madrynornis. A small portion of the right lacrimal is preserved in the immature skull. As in most extant penguins, the dorsal portion of the lacrimal is exposed in dorsal view (hidden in Spheniscus).

The adult mandible has a short bony symphysis and slender proportions resembling those of extant Eudyptula. The dentary makes up markedly more than half the total length of the ramus. The mandible is dorsoventrally narrow across the first third of the ramus and then modestly deepens near the midpoint, but it lacks the strong expansion characterizing Eudyptes and some species of Pygoscelis. The apparent presence of a large rostral fenestra is likely an artifact of matrix filling in the area between the angular, surangular, and dentary. There is a small, ovoid caudal mandibular fenestra. The retroarticular process is moderately elongate, as in extant Eudyptula.

On the right quadrate of the immature individual, the attachment of m. adductor mandibulae externus pars profunda forms a well-developed ridge (Fig. 3). This origin varies between and within extant penguin species, ranging from a very weak ridge (in all examined Megadyptes, nearly all individuals of Aptenodytes) to a strongly projected and rounded tubercle (in most individuals of Spheniscus and Eudyptes). We observed that a ridge is present in most extant Eudyptula specimens, but a more tubercle-like morphology occurs in few individuals of Eudyptula m. minor.

Etymology

The species name honors New Zealand ornithologist Kerry-Jayne Wilson, in recognition of her many contributions to seabird conservation, particularly her cofounding of the West Coast Penguin Trust.

Remarks

In extant penguins, most cranial sutures (e.g., frontal-parietal, interfrontal, parietal-supraoccipital, laterosphenoid-parietal, squamosal-laterosphenoid, and supraoccipital-exoccipital) are obliterated only after individuals attain adult size (unpublished data, DTK, 2022). The combination of nearly-adult skull length and open sutures leads us to infer that NMNZ S.048855 was fully fledged but not yet sexually mature, and thus a young-of-the-year. The cerebellar prominence is more ventrally and cranially oriented in the juvenile than the adult (Fig. 5). A more ventral orientation of this structure has been observed in juvenile penguins (Sosa and Acosta Hospitaleche, 2018; Plateau and Foth, 2021) (e.g., in the Emperor Penguin, Aptenodytes forsteri Gray, 1844). However, this difference is only observable in very young individuals, and the immature fossil skull is closely approaching adult size. Therefore, we believe it is more likely that the difference in orientation of the cerebellar prominence between the two skulls is due to postburial alteration of the immature skull.

Phylogenetic analysis

The strict consensus of our analysis places Eudyptula wilsonae n. sp. inside a polytomy alongside extant Eudyptula, living and extinct Spheniscus, and the extinct Inguza predemersus Simpson, 1971 and Madrynornis mirandus (Fig. 6). This lack of resolution reflects both the paucity of cranial characters supporting the monophyly of Eudyptula and the partial or complete lack of skull data for several species within the polytomy (e.g., Spheniscus muizoni Göhlich, 2007 and Inguza predemersus). Nonetheless, the assignment of Eudyptula wilsonae n. sp. to Eudyptula is well supported based on their extremely similar skull shape, small size, and virtually identical codings. The only exception was character 108 (quadrate, attachment scar for m. adductor mandibulae externus, par profunda), which was coded ‘1’ (present as a tubercle) for Eudyptula wilsonae n. sp. but shows intraspecific variation within Eudyptula m. minor and so coded ‘0/1’ (present as a ridge / present as a tubercle).

Figure 6. Strict consensus of most parsimonious trees. Evolutionary relationships for Eudyptula wilsonae n. sp. described in this study were not fully resolved. Instead, Eudyptula wilsonae n. sp. is placed in a clade with extant and fossil Spheniscus, Inguza predemersus and Madrynornis mirandus. Location information for species in this clade are identified with colored symbols: green = New Zealand; purple = Australia; blue = South America, Falkland Islands, and Galápagos Islands; yellow = South Africa.

Our results suggest that most synapomorphies uniting extant Eudyptula are soft-tissue characters, including: (20:3) iris bluish gray in color; (34:1) preoccular region blue; (40:1) throat pattern white; (43:2) feathers of dorsum light bluish gray; (57:1) flipper with small circular dot near tip of underside; and (65:3) feet white-pink.

In contrast, we identified only three osteological synapomorphies for Eudyptula: (98:1) lacrimal with ‘flange’ formed by laterally deflected and widened ventral extremity; (128:1) 12 synsacral vertebrae (also present in the stem penguin Palaeospheniscus and thus likely a reversal); and (154:1): coracoid fenestra incomplete (convergent in Aptenodytes and Pygoscelis). None of these characters can currently be scored for Eudyptula wilsonae n. sp., but they provide a benchmark to test its placement via future discoveries from the Tangahoe Formation.

Morphometrics

Maximum skull length and postorbital width were the major sources of variation in the dataset (Fig. 4.1). Principal component 1 explained 85.1% of the variation, with strong negative loadings for both maximum length and postorbital width. Principal component 2 explained 12.2% of the variation, with maximum length strongly positively loaded and postorbital width strongly negatively loaded. Eudyptula m. novaehollandiae skulls tend to be short and narrow, whereas those of Eudyptula m. minor tend to be longer and wider; however, this could partly reflect the immature status of some Eudyptula m. novaehollandiae skulls that were included in the study (Fig. 4.3, 4.4). The holotype (adult) skull of Eudyptula wilsonae n. sp. is proportionally narrower than any extant Eudyptula specimens measured here (Fig. 4.3). The skull of the paratype immature specimen is even narrower and most closely resembles the proportions of immature Eudyptula m. novaehollandiae (Fig. 4.2, 4.4).

Discussion

Zealandian origins of little penguins

The Pliocene age of Eudyptula wilsonae n. sp. supports a Zealandian origin of little penguins as proposed by recent biogeographic analyses (Thomas et al., 2020; Cole et al., 2022). The age of these fossils also precedes the proposed molecular divergence between Eudyptula minor minor and Eudyptula m. novaehollandiae at 1.34 Ma (95% HPD = 0.48–2.26 Ma; Cole et al., 2022). Provided that the estimated divergence date is accurate, Eudyptula wilsonae n. sp. could plausibly be ancestral to both of the living little penguin species inhabiting Australia and New Zealand today.

Whereas a Zealandian origin for crown Eudyptula is increasingly well supported, the dispersal of stem Eudyptula to Zealandia remains an open question, obscured in part by the unstable phylogenetic positions of Inguza predemersus from the Pliocene of Africa and Madrynornis from the Miocene of South America. Previous analyses (Thomas et al., 2020; Cole et al., 2022) favored dispersal from Africa to New Zealand due to Inguza Simpson, 1971 being recovered as sister taxon to Eudyptula. Intriguingly, a single vertebra supports the presence of a small penguin within the size range of Eudyptula in the Miocene of South Africa (Thomas and Ksepka, 2013). In this scenario, the South Atlantic Gyre could plausibly have facilitated the eastward dispersal of the ancestors of the Inguza-Eudyptula clade from South America to Africa (Ksepka and Thomas, 2012), followed by a further eastward dispersal from Africa to Zealandia via the Antarctic Circumpolar Current. Regardless of the pathway, Eudyptula wilsonae n. sp. minimally dates the arrival of little penguins in Zealandia to the Pliocene.

Beak evolution in seabirds

Apart from a slightly narrower skull, Eudyptula wilsonae n. sp. is almost indistinguishable from its extant relatives. In this, the species notably contrasts with some other Pliocene seabirds from the Tangahoe Formation, which show skull proportions that are substantially different from those of closely related extant taxa. For example, the stem crested penguin Eudyptes atatu Thomas, Tennyson, Scofield, and Ksepka (2020) in Thomas et al. (2020) has a substantially slenderer mandible than all extant Eudyptes and might have been less reliant on planktonic prey (Thomas et al., 2020). Likewise, the probable stem albatross Aldiomedes angustirostris Mayr and Tennyson, 2019 is smaller and has a more mediolaterally compressed beak than any living albatrosses and was likely more piscivorous (Mayr and Tennyson, 2019). Future studies might explore the coevolution of the marine environment and seabird skull structure in a Zealandian context.

Body size

Eudyptula wilsonae n. sp. was likely the same size as extant Eudyptula (~1 kg; Stonehouse, 1967). It is thus the smallest crown penguin fossil ever described and shows that the lower body size bound for crown penguins had been reached by at least the Pliocene, when sea surface temperatures were several degrees warmer than today (Burke et al., 2018). This, in turn, suggests a deep-time thermal tolerance that mirrors the latitudinal temperature range occupied by little penguins today.

Conclusions

We describe fossils of Eudyptula from the late Pliocene of New Zealand, which show that little penguins have been part of this globally famous seabird biodiversity hotspot for at least three million years. Further discoveries and better-resolved phylogenies are needed to establish when and from whence the ancestors of Eudyptula first arrived in the New Zealand region. The fossils described here show that both the cranial morphology and body size of little penguins have remained largely unchanged since the Pliocene, despite substantial environmental changes in the region over the past three million years.