An extinct Austral snipe (Aves : Coenocorypha) from New Caledonia

CSIRO PUBLISHING

Emu, 2013, 113, 383–393 http://dx.doi.org/10.1071/MU13019

An extinct Austral snipe (Aves : Coenocorypha) from New Caledonia Trevor H. Worthy A,D, Atholl Anderson B and Christophe Sand C A

School of Biological Sciences, Flinders University, Adelaide, GPO 2100, SA 5001, Australia. Archaeology and Natural History, College of Asia and the Pacific, Australian National University, Canberra, ACT 0200, Australia. Email: [email protected] C Institut d’Archéologie de la Nouvelle-Calédonie et du Pacifique, 65 Rue T. de Laubarède, BP 11423, 98802 Nouméa Cedex, New Caledonia. Email: [email protected] D Corresponding author. Email: trevor.worthy@flinders.edu.au B

Abstract. A new and extinct species of Austral snipe (Aves : Scolopacidae : Coenocorypha) is described from late Holocene cave deposits in New Caledonia. The new species is larger than all congeners in the New Zealand archipelago, including its subantarctic islands, but is slightly smaller than C. miratropica from Viti Levu, Fiji. Better developed ligamental sulci and attachment processes on the humerus suggest that this new species was perhaps the strongest flier in the genus. Nevertheless, it went extinct within c. 1000 years of human arrival in New Caledonia, probably as a result of depredation by the rats that arrived with humans. Additional keywords: extinction, Grand Terre, owl deposits, prehuman avifauna, Scolopacidae. Received 14 March 2013, accepted 29 May 2013, published online 27 November 2013

Introduction The Austral, or New Zealand, snipes (Scolopacidae : Coenocorypha) have been considered to be a New Zealand endemic genus with taxa historically inhabiting virtually the entire archipelago. Coenocorypha species form a distinct group of rather unusual Charadriiformes that have nocturnal aerial displays and a preferred habitat of forests and shrublands where they probe soils for invertebrates, rather than foraging along aquatic margins (Higgins and Davies 1996). Until fairly recently, only three species were recognised (Turbott 1990; Higgins and Davies 1996; Piersma 1996): the extinct C. chathamica (Forbes’ Snipe) and the extant C. pusilla (Chatham Island Snipe) on the Chatham Islands, and subspecies of C. aucklandica elsewhere in New Zealand. More recent research has resulted in the elevation of several of these subspecies to specific status. Six species are now recognised, three of which are extinct, with C. aucklandica (Subantarctic Snipe) retaining three allopatric subspecies (Worthy et al. 2002; Miskelly and Baker 2010; Baker et al. 2010; Gill et al. 2010). The genus, however, has a low public profile in mainland New Zealand owing to its extinction on the North and South Islands. Coenocorypha is usually considered sister to Gallinago and both are included in the subfamily Gallinagininae, together with Lymnocryptes (Jack Snipe (L. minimus)) and, sometimes, Limnodromus (dowitchers) (Higgins and Davies 1996; Piersma 1996; Christidis and Boles 2008; Gill et al. 2010). The Gallinagininae is itself usually considered sister to Scolopax (Scolopacinae, the Journal compilation
BirdLife Australia 2013

woodcocks). However, recent analyses of molecular data suggest that Coenocorypha renders Gallinago paraphyletic when Gallinago imperialis (Imperial Snipe) is included in analyses, and provides strong support for Limnodromus being sister to Gallinagininae plus Scolopax (Baker et al. 2007; Gibson and Baker 2012). Gallinago imperialis and G. jamesoni (Andean Snipe) are Andean species sometimes included in Chubbia Mathews, 1913 (type-species Gallinago stricklandii (G.R. Gray)), whose nocturnal behaviour and preferred inland forested habitats are similar to those of Coenocorypha. As neither the type-species for Chubbia nor G. jamesoni were examined by Gibson and Baker (2012), additional investigations are required to resolve the composition of Gallinago, the generic validity of Chubbia and the relationships of Coenocorypha to both. With most extant taxa existing on subantarctic islands, owing to the extinction of species on the North and South Islands, Coenocorypha could have been considered southern cool-climate specialists. But any such conclusion was found to be illusory when the fossil record of the genus began to be unveiled. The possibility that the genus may have existed outside of New Zealand in tropical climes was first presented by Balouet and Olson (1989) when they tentatively identified some bones from New Caledonia. Shortly afterwards, Meredith (1991) reported the genus from Norfolk Island, based on material that remains undescribed. More recently the genus was reported from tropical Fiji, with C. miratropica Worthy, 2003 (Fiji Snipe), the largest species known, being described from a few fossils from inland www.publish.csiro.au/journals/emu

384

Emu

Viti Levu (Worthy 2003). This new expanded distribution is, however, with the exception Fiji, fundamentally restricted to the continent of Zealandia, which is a continent the size of India that is now largely submerged. It encompasses the area from New Caledonia to the Chatham Islands to Campbell Island and its emergent lands share many biogeographical links (Wallis and Trewick 2009 and references therein). The first report of Coenocorypha from New Caledonian was based on three fragmentary bones from Gillies Cave on the western side of Grande Terre (Balouet and Olson 1989; Balouet 1991). These bones were obtained from levels of the deposit lacking rodents and thus dated to a period before human arrival, and are likely of Holocene age. The site had a fauna characterised by a high abundance of lizard bones and a significant contribution from the leavings of owl predation was inferred (Balouet 1991). Our investigations of the avifauna of New Caledonia stem from fieldwork in the Pindai Caves on the Nepoui Peninsula in 2003, with site investigations and a preliminary analysis of the fauna presented by Anderson et al. (2010). Our research was primarily archaeological and directed at resolving ambiguity about the overlap of human occupation at the caves and in their vicinity with changes in the faunal record, especially the extinction of birds and reptiles. We excavated fossil material from several discrete deposits in Caves A, B and C in the Pindai Caves complex, see figs 2 and 3 in Anderson et al. (2010). In Cave A, richly fossiliferous microvertebrate faunas were recovered from test pits TP3 and TP4 between the surface and 110–120 cm deep. These fossil faunas were dominated numerically by fossils of lizards (analysis in preparation) but many hundreds of bird bones were also found, representing at least 43 species even with passerines excluded (Anderson et al. 2010). The fossil fauna in these test pits was mainly accumulated as a result of predation by owls, with the extinct Tyto letocarti contributing to the lower layers and the Australian Barn Owl (T. delicatula) to the upper layers (Anderson et al. 2010). Based on the mineralised state of the bones and two dates on bones of Sylviornis (a giant megapode), deposition began a few thousand years before human arrival, resulting in a dense bed of bones ~20 cm thick at 85- to 110–115-cm depth on the cave floor. Below this, the sediment was sparsely fossiliferous and its depth was not determined. Throughout the entire sequence a small proportion of the fossil fauna likely resulted from pitfall into the cave. The entrance to the cave is a hole in the roof through which animals may fall and their remains preferentially accumulate at the foot of the slope along a wall of the chamber where the test pits were located. Also, a significant contribution to the fossil fauna was derived from other cave inhabitants, in the form of various bats, as yet unstudied, and swiftlets (Collocalia spp.). Over the period 2800–2500 calendar years before present, human occupation in and around the cave began to contribute elements to the fossil fauna, including the extinct taxa Sylviornis and the crocodilian Mekosuchus, but simultaneously, it is likely that this human activity resulted in the abandonment of the site by T. letocarti, thereby eliminating a major fossil contributor from the system (Anderson et al. 2010). Associated with the human occupation, from ~85-cm depth in test pit TP3 and 95-cm depth in TP4, the rate of sedimentation rapidly increased and the

T. H. Worthy et al.

density of fossils declined markedly, and extinct taxa became rare or absent (Anderson et al. 2010). However, in this humanmediated upper part of the section where Tyto delicatula is the presumed primary vector for bone accumulation, bones of the Pacific Rat (Rattus exulans) rise to numerical dominance. However, it is in the lower levels from ~85- to 110–115-cm depth, with its highly diverse fossil fauna, that an indication of the pre-human fauna is obtained. Apart from having huge numbers of lizard bones, this fauna includes numerous small birds, with the Painted Button-quail (Turnix varius) by far the most abundant and the New Caledonian Owlet-nightjar (Aegotheles savesi) next most common (Anderson et al. 2010). However a relatively abundant component of the avifauna in both test pits was the Coenocorypha snipe, whose bones were concentrated in the lower levels. As expected from their origin in the site having been mainly via owl predation, many specimens are fragmentary and, in particular, the sterna are represented only by the robust parts of the anterior carina. The material includes good examples of most post-cranial limb elements, particularly the humeri, which is important to facilitate comparison with C. miratropica of Fiji. In this contribution, we describe this species to reveal better the original range of this genus.

Materials and methods Abbreviations The following abbreviations are used: IANCP, Institut d’Archéologie de la Nouvelle-Calédonie et du Pacifique (Nouméa, New Caledonia) (all IANCP catalogue numbers cited in this paper have the prefix ‘IANCP.PN/WNP011.PA/1.2003/’, which is abbreviated to ‘IANCP’ herein); CM, Canterbury Museum (Christchurch, New Zealand); NMNZ S, Fossil Vertebrate Collection, Museum of New Zealand Te Papa Tongarewa (Wellington, New Zealand); SAM B, Ornithology Collection, South Australian Museum (Adelaide, South Australia).

Nomenclature Names for specific bone landmarks follow Baumel and Witmer (1993). Measurements were made with Tesa dial callipers and rounded to 0.1 mm. We follow the taxonomic nomenclature in Gill et al. (2010). The fossils were compared with the following taxa using the specimens and data listed in Worthy et al. (2002), Worthy (2003) and the additional CM specimens listed here († indicates species is extinct): Coenocorypha aucklandica aucklandica (G. R. Gray, 1845) (Auckland Island Snipe); C. a. meinertzhagenae Rothschild, 1927, (Antipodes Island Snipe): CM Av5190; C. huegeli (Tristram, 1893) (Snares Island Snipe); C. pusilla (Buller, 1869) (Chatham Island Snipe): CM Av6574, 28590, 30012; †C. iredalei Rothschild, 1921 (South Island Snipe): CM Av14088, 18007, 25331, 36393; †C. barrierensis Oliver, 1955 (North Island Snipe): CM Av25009, 25332, 31592, 31593, 33641; †C. chathamica (Forbes, 1893) (Forbes’ Snipe): CM Av 28470; †C. miratropica Worthy, 2003 (Fiji Snipe); and Gallinago gallinago (Linnaeus, 1758) (Common Snipe); The fossils were also compared to G. hardwickii (J. E. Gray, 1831) (Japanese Snipe): SAM B.48079.

An extinct Austral snipe from New Caledonia

Systematic palaeontology Order CHARADRIIFORMES Family SCOLOPACIDAE Rafinesque, 1815 Coenocorypha G. R. Gray, 1855 The fossil humerus is assigned to a Coenocorypha rather than related scolopacids owing to the combination of the following features. In most birds, the tendon of musculus supracoracoideus inserts only on the tuberculum dorsale, but in scolopacids, as in a few other birds (e.g. galliforms and columbids), an accessory tendon of the musculus supracoracoideus inserts on the crista m. supracoracoidei, an elongate tubercle or impressio extending distally of the tuberculum dorsale. In Coenocorypha, this impressio is not elevated off the shaft except in a minor way at its most distal point in some taxa. The crista deltopectoralis is low and recurved over the large flat-bottomed impressio musculi pectoralis on the cranial facies, the impressio coracobrachialis is small, the processus supracondylaris dorsalis is closer to the distal end than the width across the condyli ventralis et dorsalis and is directed more cranially than dorsally (Worthy 2003). Coenocorypha neocaledonica, sp. nov.

Emu

385

from the dorsal condyle by a space narrower than the tuberculum width; (5) the fossa musculi brachialis is larger and extends proximad of the processus supracondylaris dorsalis and, distomedially, extends beside the condylus dorsalis where it is deeply excavated into the base of the tuberculum supracondylare ventrale, rather than ending proximal to both the tuberculum and the condylus dorsalis; and (6) the processus flexorius is more distally elongate making the condylus ventralis appear less protuberant. Coenocorypha neocaledonica is similar to all the New Zealand Coenocorypha species in features 1, 2, 4 (except in C. pusilla where the space between the tuberculum and dorsal condyle is wider than the tuberculum), and 5, but differs from all by its larger size, the more prominent ridge ventrally adjacent to the sulcus scapulotricipitalis (3), and a more elongate processus flexorius (6). Type-locality Test pit 3, spit 17, 80–85-cm depth, Cave A, Pindai Caves, located at 21
210 00.500 S, 164
570 47.800 E, Nepoui Peninsula, north-eastern coast, Grande Terre, New Caledonia. Measurements (mm) of holotype Length 39.7, proximal width 9.6, mid-shaft width 3.2, distal width 6.8, depth at distal end 4.0.

(Fig. 1) Description and comparisons Material examined Holotype. b, g, h).

IANCP.PN/WNP011.PA/1.2003/500, L humerus (Fig. 1a,

Paratypes. IANCP/502, distal + shaft L humerus; IANCP/503, proximal + shaft L humerus; IANCP/505, proximal R humerus; IANCP/510, proximal + shaft L humerus; IANCP/511, distal + shaft L humerus; IANCP/513, L humerus; IANCP/514, p+sL humerus. Referred specimens.

See Tables 1 and 2.

Coenocorypha neocaledonica, with humeri length 37.6–39.9 mm and carpometacarpi length 23.2–25.2 mm (see Appendix 1), was slightly smaller than C. miratropica from Fiji (humerus 40.2 mm, n = 1; carpometacarpi 26.0–26.8 mm, n = 3; Worthy 2003). It was, however, based on lengths of humeri, ulnae, carpometacarpi, femora, tarsometatarsi and coracoids, slightly larger than all Coenocorypha species in the New Zealand region, although it had similar proportions (see data in Worthy et al. 2002).

Diagnosis A large species of Coenocorypha that differs from the similarly sized C. miratropica and all other Coenocorypha species by the humerus having a larger and more prominent processus supracondylaris dorsalis with a notch between its tip and the shaft proximally. Its humeri further differ from those of C. miratropica by: (1) the crista m. supracoracoidei is dorsoventrally narrow, offset dorsally and distally from the tubercle, and is separated from the margo caudalis by a shallow fossa, whereas in C. miratropica, the crista m. supracoracoidei is broader dorsoventrally and extends along the whole side of the tubercle to join the margo caudalis at a point adjacent to the caput; (2) the linea musculus latissimus dorsi pars cranialis is weakly formed rather than forming an obvious ridge; (3) distally, the ridge bounding the sulcus scapulotricipitalis ventrally is more prominent and extends proximally to a point level with the processus supracondylaris dorsalis; (4) the tuberculum supracondylare ventrale is narrower, and is separated from the condylus dorsalis by a space equivalent to its width, rather than being broader and separated

Humerus Apart from being larger than all New Zealand species of Coenocorypha and slightly smaller than the Fijian C. miratropica, humeri of C. neocaledonica are most easily distinguished from those of all other species of Coenocorypha by their relatively larger and more protuberant processus supracondylaris dorsalis. In addition to the features listed in the diagnosis, the following observations further distinguish humeri of C. neocaledonica from those of the New Zealand species of Coenocorypha: (1) the margo caudalis is lower and more rounded caudally than the sharper crest in New Zealand species of Coenocorypha with the exception of C. chathamica, in which it is similarly rounded; (2) distally the sulcus humerotricipitalis is relatively deeper (except for C. chathamica) with the resultant effect that the fossa olecrani is comparatively shallower and more poorly defined proximally; the sulcus humerotricipitalis is similarly deep in C. miratropica and C. neocaledonica; (3) the fossa musculi

386

Emu

T. H. Worthy et al.

(a)

(b)

(c) (f )

(e)

(d )

(g)

(i)

(k)

(j) (h)

Fig. 1. Selected elements of Coenocorypha neocaledonica, n. sp. All catalogue numbers carry the prefix ‘IANCP. PN/WNP011.PA/1.2003/’. Left humerus 500, holotype, in cranial (a, h) and caudal (b, g) views; right ulna 504 in cranial view (c); right carpometacarpus 520 in ventral view (d); right coracoid 504 in dorsal view (e); left tarsometatarsus 512 in dorsal view (f); proximal right tarsometatarsus 515 in plantar (i) and dorsal (j) views; proximal left tarsometatarsus 518 in plantar view (k). Scale bars = 10 mm. Abbreviations: am, angulus medialis; c, caput; cd, condylus dorsalis; cm, crista medianoplantaris; dc, crista deltopectoralis; ei, eminentia intercotylaris; etd, elongate tuberculum dorsale; fhl, tendinal canal/sulcus for m. flexor hallucis longus; fmb, fossa musculus brachialis; ims, impressio of crista m. supracoracoidei; irm, impressio retinaculum medialis; mc, margo caudalis; pf, processus flexorius; psd, processus supracondylaris dorsalis; pII–ppII, sulcus for tendons for musculus flexor perforatus digiti II and musculus flexor perforans et perforatus digiti II; pIII–IV, ppIV, sulcus for the tendons of musculus flexores perforati digitorum III and IV and musculus flexor perforans et perforatus digiti III; r, ridge bounding sulcus scapulotricipitalis; tsv, tuberculum supracondylare ventrale; ttc, tuberositas musculus tibialis cranialis; TII, trochlea metatarsi II.

brachialis is deeper; and (4) the tuberculum supracondylare ventrale is more elevated cranially. Ulna The ulna in C. neocaledonica (e.g. those in IANCP/506, IANCP/504; Fig. 1c), is typical of Coenocorypha as follows:

the proximal end is dorsoventrally much wider than its craniocaudal depth; the olecranon is rounded proximally and prominent; the impressio brachialis is very deep and undercuts the cotyla ventralis by more than half of the shaft width so that it faces cranially rather than ventrally (shared with all Charadriiformes); the tuberculum ligamentosa collateralis ventralis is robust and

An extinct Austral snipe from New Caledonia

Emu

387

Table 1. Material of Coenocorypha neocaledonica from Pindai Cave A test pit TP3, listed by excavation depth All catalogue numbers have the prefix ‘IANCP.PN/WNP011.PA/1.2003/’; Spit is the excavation unit of depth within an excavation, with each spit usually being 5 cm; NISP is number of specimens. Abbreviations: ant, anterior; cmc, carpometacarpus; cor, coracoid; fem, femur; hum, humerus; mand, mandible; pmx, premaxilla; pt, part; scap, scapula; tib, tibiotarsus; tmt, tarsometatarsus. Prefix to elements: L, left, or R, right; p, d or s (as prefix to L and R) indicate proximal, distal and shaft parts Catalogue number

Spit; depth (cm)

Specimens

NISP

500 501 502 503 504

17; 80–85 17; 80–85 18; 85–90 18; 85–90 18; 85–90

505 506

19; 90–95 19; 90–95

507 508

20; 95–100 20; 95–100

509

21–22; 100–110

510 511 525

21–22; 100–110 21–22; 100–110 26–27; 125–135

L hum (Holotype) 1pL 1pR hum, 1dR ulna, 1dL 1dR tmt 1d+sL hum (Paratype) 1p+sL hum (Paratype) 2pL1dR hum; 1R1pR ulna; 1pR cmc; 1 cranial pt L, 2 sternal pt L, 1R, 1 sternal pt R cor; 1p+sR fem; 1d+sR tib 1pR hum (Paratype) 1pL1pR hum, 1pL1dL ulna, 1pL1dL1R fem, 2dL1pL tib, 1L2dL1pL1R tmt, 1L ramus mand, 1L1R cor, 1R scap 1R tmt 1R1pR ulna, 1pR cmc, 2L1R scap, 1 pmx tip, 1 ant sternum, 1dL1pR1dR tmt 1L1pL1dL2pR hum, 3pL3dL1pR1dR ulna, 2pL1pR cmc; 1L, 3 cranial pt L, 1 sternal pt L cor; 2L1R scap, 5 ant sternum, 2pL2dL4dR fem, 2dL tib, 1R1dL tmt, 1 pmx, 1L side mand, 1L quadrate 1p+sL hum (Paratype) 1d+sL hum (Paratype) pL ulna

1 5 1 1 13 1 19 1 11 44

1 1 1

Table 2. Material of Coenocorypha neocaledonica Pindai Cave A test pit TP4, by excavation depth All catalogue numbers have the prefix ‘IANCP.PN/WNP011.PA/1.2003/’; NISP is number of specimens. Spit is the excavation unit of depth within an excavation, with each spit usually being 5 cm. Abbreviations: As Table 1 plus fur, furcula Catalogue number

Spit; depth (cm)

Specimens

522 523 521 520 519 518 517 516 515 514 513

–; 45–50 –; 50–55 19; 90–95 20; 95–100 21; 100–105 21; 100–105 22; 105–110 23; 110–115 23; 110–115 23; 110–115 23; 110–115

1R cmc 1R scap 1pL1dL hum, 1L cmc, 1R scap, 1dR tib,1dL tmt 1dR hum, 1pL ulna, 1R cmc, 1 ant sternum, 1L scap, 1L cor, 1dL1dR tib 1pL1dL1R hum, 1pL ulna, 2R cor, 1 ant stern, 1 pt fur 1pL tmt 1pL1R hum, 1pL1dL ulna, 1L cor, 1 fur, 1R tmt, 1pR1R ulna, 1L1R cmc, 1L1R cor, 1 ant sternum, 1pR fem, 1dL tib 1pR tmt 1p+sL hum (Paratype) 1L hum (Paratype)

markedly prominent ventrally; and in cranial aspect, the shaft width attenuates continuously to the distal end. The proximal ulna is not known for C. miratropica but compared to more southern species of Coenocorypha that of C. neocaledonica has a dorsoventrally broader impressio brachialis adjacent to the processus cotylaris dorsalis. In addition, the ligament attachment scar that extends proximoventrally to a point under the cranial margin of the cotyla ventralis and bounds the incisura radialis distally is not as protuberant over the impressio brachialis. Immediately distal to the cotyla ventralis, this same scar is smaller than it is in New Zealand Coenocorypha, where it is generally larger and contributes to a markedly

NISP 1 1 6 8 8 1 7 9 1 1 1

overhanging ridge extending along the proximodorsal margin of the impressio brachialis. Coracoid In C. neocaledonica, the facies articularis sternalis is dorsoventrally thick, especially towards the angulus medialis, which in consequence is more robust than it is in New Zealand species of Coenocorypha and thicker dorsoventrally. The angulus medialis tends to have a convex medial profile, but its shape varies within the species, with some specimens more pointed than others (e.g. see IANCP/504 specimens; Fig. 1e), whereas it is

388

Emu

always concave and highly acute in the New Zealand taxa. In C. neocaledonica, the crista located medially and extending from about mid-length to the point of inflection proximal to the angulus medialis (but not along the angulus medialis) is more robust than in all the New Zealand Coenocorypha. The impressio musculus sternocoracoidei is deeply concave dorsally and the processus lateralis is strongly directed dorsally. Femur, tibiotarsus and tarsometatarsus The leg bones differ in no significant way from those of other Coenocorypha species. A few near-complete tarsometatarsi (IANCP/512 (Fig. 1f), IANCP/507) and a well-preserved proximal fragment with hypotarsus near complete (IANCP/518; Fig. 1k) is typical of Coenocorypha with the cotyla medialis displaced markedly proximad of the cotyla lateralis and in having a globose eminentia intercotylaris that projects proximally of both cotylae. The impressio retinaculum medialis is an acute, elongate, medially convex, arc-like crista, that extends distad to the tuberositas musculus tibialis cranialis. The tuberositas musculus tibialis cranialis has two facets, a larger medial one and a smaller lateral one located distal to the foramina vascularia proximalia slightly mesad of the midline in a deep sulcus extensorius that extends to about mid-length. The hypotarsus is short, wider than long, with one deep enclosed canal adjacent to the crista medianoplantaris for the tendinal canal for m. flexor digitorum longus (fdl), and three more plantarly (shallowly) located canals. The most lateral of these more shallowly located canals is probably the tendinal canal/sulcus for m. flexor hallucis longus (fhl) (it is open in specimens IANCP/515 (Fig. 1i, j) and IANCP/ 518 (Fig. 1k) but closed in IANCP/507). The shallow, most medial canal is open plantarly in all specimens and carries the tendons for m. flexor perforatus digiti II and m. flexor perforans et perforatus digiti II (pII–ppII). The centrally located shallow canal is presumed to carry the tendons of m. flexores perforati digitorum III and IV and m. flexor perforans et perforatus digiti III (pIII–IV, ppIII) and is partly closed (over the mid-part of the canal length) plantarly in IANCP/518 but in the left tarsometatarsus in IANCP/506 it appears to have been open plantarly as the hypotarsal ridge separating this canal and that medial to it is entire and unbroken. Distally, trochleae metatarsi II and IV had similar divergence from the shaft, trochlea metatarsi II is short, barely creating an incisura intertrochlearis medialis, and is only slightly retracted plantarly. Trochlea metatarsi III is longer than wide. Trochlea metatarsi IV extends distally more than half the length of trochlea metatarsi III. The foramen vasculare distale lies at the end of an elongate sulcus on the dorsal facies, close to the incisura intertrochlearis lateralis. The fossa metatarsi I is shallow and on the plantar facies rather than somewhat on the medial facies. In New Zealand Coenocorypha, the canal for fhl was open plantarly in all species, but variation was observed in the plantar closure of the medially located shallow canal for pII–ppII. This was usually open plantarly but, in C. iredalei specimen CM Av36393, was closed, and was nearly closed in C. barrierensis specimen CM Av31593. Considerable variation was observed in the centrally located shallow canal for pIII–IV, ppIII: in C. barrierensis, it was variably open or closed plantarly, yet closed plantarly in all examples examined of other taxa (e.g. C. iredalei and C. aucklandica).

T. H. Worthy et al.

Mandible In the catalogued lot IANCP/506, a fragment of left mandible preserves the region from just anterior to the processus coronoideus to the end of the processus retroarticularis, although the processus mandibulae medialis is broken at the foramen pneumaticum. The element is large and robust (length processus coronoideus to caudal extreme processus retroarticularis 16.3 mm, depth at processus coronoideus 5.7 mm), as is typical of Coenocorypha, and is absolutely much larger and more robust, with a proportionally more elongate processus retroarticularis, than that of the much larger (based on size of bones of both leg and wing) Gallinago hardwickii. Etymology The new species is named for its origin on New Caledonia. Discussion The extinct snipe from New Caledonia described here as Coenocorypha neocaledonica was one of the largest species in the genus, with only C. miratropica of Fiji larger. The New Zealand snipe closest to C. neocaledonica in size is the extinct C. chathamica, with all other species being smaller. The humeri of the three larger taxa all shared a more rounded margo caudalis and a deeper sulcus humerotricipitalis, which correlate with larger body mass and the requirement for greater musculature in flight, and so are likely not of phylogenetic significance. Most of the features that distinguish the new taxon appear to be related to greater musculature associated with flight, for example, the larger processus supracondylaris dorsalis, longer processus flexorius, deeper sulcus scapulotricipitalis, deeper sulcus humerotricipitalis and large and deep fossa musculi brachialis. Given that it is slightly smaller than C. miratropica, it seems likely that C. neocaledonica was perhaps the strongest flyer in the genus. The description here of the snipe Coenocorypha neocaledonica, sp. nov. confirms the earlier report of the genus from New Caledonia (Balouet and Olson 1989; Anderson et al. 2010) and reveals that this largely Austral genus penetrated the tropical South Pacific with at least two species. The two tropical species, C. neocaledonica from New Caledonia and C. miratropica from Viti Levu in Fiji, were the largest species in the genus, an observation that runs counter to Bergmann’s Rule, which predicts that the larger forms would be in the cooler climate habitats of the subantarctic islands. This size trend is, however, paralleled to some extent by the teals (Anas spp.), whose forms on subantarctic islands are also smaller than their lower latitude relatives as evidenced by the extinct teal Anas marecula of Amsterdam Island and Anas on the New Zealand subantarctic islands (Livezey 1990; Olson and Jouventin 1996). This trend might be explained at least partly by a limited diversity and abundance of invertebrate food in the low floral diversity tussock grasslands and shrublands of the subantarctic island habitats of these species. Whereas the Brown Teal (Anas chlorotis), the New Zealand sister taxon of the New Zealand subantarctic teal, is substantially larger than its coolclimate relatives, there is in contrast essentially no difference in size between the Coenocorypha species on the New Zealand subantarctic islands and those on North or South Island of New Zealand (Worthy et al. 2002). Thus, although there may be some trade-off between nutrient availability and size for both the teals

An extinct Austral snipe from New Caledonia

and snipes in the tussock-grassland and shrubland habitats of the Subantarctic, this does not entirely explain the range in body size observed over the entire range of Coenocorypha. It is not known when the Fijian form became extinct, because the sole deposit bearing its bones remains undated (Worthy 2003; Worthy and Anderson 2009a). C. neocaledonica, however, survived until after the arrival of humans in its environment on Nepoui Peninsula, on the north-eastern coast of Grande Terre, New Caledonia. Extensive dating of samples from the test pits in Cave A, in the Pindai Caves, suggested that human occupation of the site began c. 2600–2800 calendar years before present (~80–85-cm depth), with the 40–45-cm level in test pit TP3 c. 2000–2300 calendar years before present and 40–65 cm in TP4, c. 1800–2000 calendar years before present (Anderson et al. 2010). Most Coenocorypha bones were at depths below 80 cm in test pit TP3 and below 90 cm in TP4 and hence likely of prehuman age but a few bones were found higher in the sections in both test pits. One bone in test pit TP3 found at a depth of 20–25 cm and two at 45–55 cm in TP4 attest to survival of Coenocorypha perhaps more than 1000 years after human arrival in the area. The extinction of C. neocaledonica is probably related to the introduction of the Pacific Rat (Rattus exulans), which arrived in New Caledonia with the earliest human colonists c. 3000 years ago (Roberts 1991). This suggestion is supported by the history of extinctions of Coenocorypha in New Zealand. Populations only survive on islands where there are no rats, nor any other mammalian predators, because, although they are strong fliers, Coenocorypha species nest on the ground and lay only two eggs, making them extremely susceptible to predation by such mammals (Holdaway 1999; Miskelly et al. 2006; Miskelly 2012). The populations of Coenocorypha on the North and South Islands were not recorded as living birds historically. Remnants of these species were restricted historically to offshore islands: C. barrierensis (North Island Snipe) to Little Barrier Island, where a unique specimen was collected in 1870 (Miskelly 1988) and where the Pacific Rat was likely to have been a late arrival; and C. iredalei (South Island Snipe) to islets around Stewart Island. The last population of C. iredalei to go extinct was on Big South Cape Island, which Black Rats (Rattus rattus) colonised in the early 1960s, and shortly afterwards this population of snipe was extinct (Miskelly 2012). Given the lack of historical records, the New Zealand mainland Coenocorypha populations were likely driven to extinction in just a few hundred years after the arrival of humans and Pacific Rats in the 13th century (Wilmshurst et al. 2008). The recent history of C. aucklandica perseverance (Campbell Island Snipe) is also telling. This form was discovered on the tiny Jacquemart Island off Campbell Island in 1997, where it likely had been confined as a breeding species for the last 170 years. After the eradication of Brown Rats (Rattus norvegicus) from Campbell Island in 2001, Snipe rapidly recolonised the island (Miskelly and Fraser 2006). Thus, on both Fiji and New Caledonia, where the arrival of humans and rats was c. 3000 years ago, Coenocorypha had little chance of surviving into historical times. That Coenocorypha species forage on the ground and also are nocturnally active makes them susceptible to predation by owls. In New Zealand, Coenocorypha bones are common in deposits attributed to the endemic large strigid Laughing Owl (Sceloglaux

Emu

389

albifacies) (e.g. Worthy and Holdaway 1994, 1996; Holdaway and Worthy 1996; Worthy 1997, 2001). The New Zealand Falcon (Falco novaeseelandiae), although not nocturnal does have crepuscular habits and, on the Auckland Islands, Coenocorypha was a common prey item (Hyde and Worthy 2010). In Fiji, there were no owls until sometime after human arrival when Tyto delicatula colonised the archipelago (Worthy and Anderson 2009b), and the fossils of C. miratropica were discovered only at a single site, where the bones were interpreted to have been deposited by falcons (Worthy 2003). No bones of Coenocorypha have been found in fossil sites attributed to pitfall origin in New Caledonia (Balouet and Olson 1989; Anderson et al. 2010), so the discovery of Coenocorypha bones only at sites where accumulation is attributed to the predatory activities of owls exemplify the importance sites with this taphonomy have for reconstructing the former avifauna of a region or country. It is interesting that C. neocaledonica was not found in the deposits at Mé Auré Cave (Boyer et al. 2010), in the western sclerophyll forested region of New Caledonia, despite its presence in the Pindai Caves on the Nepoui Peninsula to the north of this site and in Gillies Cave a little to the south. Accumulation in Mé Auré Cave was primarily post-human in age and deposited by Tyto delicatula (see Boyer et al. 2010), and given that C. neocaledonica was already very rare in deposits