Testing species limits of New Zealand’s leiopelmatid frogs through morphometric analyses

Testing species limits of New Zealand’s leiopelmatid frogs through morphometric analyses Abstract The taxonomic status of some of New Zealand’s endemic and threatened leiopelmatid frogs has been debated for decades. Clarifying this uncertainty is vital to their conservation, especially given the risk of extinction of cryptic taxa. We re-examined leiopelmatid diversity through multivariate analyses of the skeletal and external morphology of extinct and extant Leiopelma to determine morphological differentiation. Our results suggest that the morphological distinction between extinct taxa is greater than in modern extant taxa. While size ranges of postcranial elements overlapped within extant species, maxillae shape discriminated some extant taxa. We confirm the morphological distinctiveness of the extinct taxa recognized to date but identify latitudinal and altitudinal variation in postcranial element size and shape within the widespread Leiopelma markhami and L. waitomoensis, which suggest possible post-human extinction of cryptic taxa. Furthermore, the lack of morphological and osteological differentiation between L. archeyi and the insular extant L. hamiltoni and L. pakeka leads us to question the taxonomic distinctiveness of these three taxa. Future genetic research using modern and ancient DNA is recommended to enable species limits within Leiopelma to be tested in more detail to provide an evidence-based assessment for their conservation management. extinctions, frogs, Leiopelma, morphology, New Zealand, phylogeography, Quaternary fossils INTRODUCTION Diversification of New Zealand’s endemic biota has been driven by isolation, dynamic tectonic activity and climatic shifts for millions of years. Contrary to the expectation that evolutionarily ‘ancient’ lineages should be more taxon rich than ‘recent’ lineages, given longer evolutionary trajectories, extant Gondwanan vertebrate taxa in New Zealand are not speciose (Daugherty, Patterson & Hitchmough, 1994; Tennyson, 2010). Even so, potential cryptic species diversity within New Zealand’s endemic ‘ancient’ vertebrate lineages such as those of Mystacinidae (short-tailed bats), Sphenodontidae (tuatara) and Leiopelmatidae (frogs) has challenged researchers (Daugherty et al., 1990; Bell, Daugherty & Hay, 1998a; Kennedy et al., 1999; Holyoake, Waldman & Gemmell, 2001; Hay et al., 2003, 2010). Taxonomic uncertainties have major implications for the conservation management of ‘ancient’ herpetofauna as there is the risk of extinction of cryptic taxa (Daugherty et al., 1990). Although Hay et al. (2010) concluded that tuatara is one taxon, taxonomic uncertainties persist in other ‘ancient’ genera such as Leiopelma (Bell, 2010; Bishop et al., 2013; Newman et al., 2013). To date, three extinct Leiopelma spp. have been described from Holocene fossils (the terrestrial L. markhami Worthy, 1987 and L. auroraensis Worthy, 1987, and the semi-aquatic L. waitomoensis Worthy, 1987), along with three extant taxa (the semi-aquatic: L. hochstetteri Fitzinger, 1861 and the terrestrial: L. archeyi Turbott, 1942 and L. hamiltoni McCulloch, 1919). The validity of a fourth extant taxon (L. pakeka Bell, Daugherty & Hay, 1998) is debated (Holyoake et al., 2001; Thurlow, 2015). Worthy (1987a, b) described the palaeoecology and osteology of Leiopelma, but a large proportion of postcranial elements, which differed primarily in size, were taxonomically indeterminate. Cranial elements (e.g. maxillae) show prominent structural differences (Worthy, 1987b) and may distinguish between most Leiopelma taxa. Indeed, Clarke (2007) suggested that the maxilla is the only reliable cranial element for osteological analysis. Within Leiopelma, there are two groups of taxa supported by osteological characters. Leiopelma auroraensis and L. markhami are morphologically similar to L. hochstetteri, whereas L. waitomoensis is similar to the extant terrestrial taxa, which are osteologically indistinguishable (Worthy, 1987b; Worthy et al., 2013). Leiopelma hochstetteri is the only leiopelmatid frog to exhibit sexual dimorphism osteologically, in that the males have more pronounced lateral ridges on the humeri compared with females (Bell, 1978; Worthy, 1987b; Germano, Cree & Bishop, 2011). Of the extinct Holocene taxa, L. waitomoensis and L. markhami were the most geographically widespread; L. waitomoensis was restricted to the North Island, but L. markhami was distributed in both main islands (Figs 1a, b). Leiopelma auroraensis is known only from the holotype (NMNZ S.23413) collected from southern New Zealand (Te Anau, Fiordland) (Worthy, 1987a; Fig. 1a). Worthy (1998) described an incomplete femur from southern New Zealand (Forest Hill, Southland), which was intermediate between L. auroraensis and L. markhami, thus its identity remains unclear (Fig. 1f). Worthy (1987a) suggested that L. auroraensis may just represent the southern extent of L. markhami. The fossil distribution of the extant L. hamiltoni and L. pakeka remains uncertain because these taxa were previously considered as the single species L. hamiltoni (until 1998 when L. pakeka was described) and no morphological characters separate these taxa (Bell et al., 1998a; Bell, 2010; Worthy et al., 2013). Indeterminate Leiopelma fossils indicate a wider pre-human distribution of leiopelmatids (Fig. 1f). Fossils identified by Worthy (1987a, b) as L. hamiltoni and L. hochstetteri have been discovered across most of the North Island and western regions of the South Island (Figs 1c, d), yet fossils identified as L. archeyi have only been found in the northern North Island (Fig. 1e). Late Quaternary fossils of Leiopelma have not been recorded in the eastern South Island, possibly because of dry climates during the Pleistocene and Holocene (Worthy, 1987a). Figure 1. View largeDownload slide Quaternary fossil, historical and current distribution of leiopelmatid frogs in New Zealand: (a) Leiopelma markhami and L. auroraensis, (b) L. waitomoensis, (c) L. hochstetteri, (d) L. pakeka and L. hamiltoni, (e) L. archeyi and (f) indeterminate spp. Figure 1. View largeDownload slide Quaternary fossil, historical and current distribution of leiopelmatid frogs in New Zealand: (a) Leiopelma markhami and L. auroraensis, (b) L. waitomoensis, (c) L. hochstetteri, (d) L. pakeka and L. hamiltoni, (e) L. archeyi and (f) indeterminate spp. Habitat loss and predation from the introduced Polynesian rat (Rattus exulans) after human arrival in the late 13th century are the primary factors identified in the declines and extinctions seen in leiopelmatids (Bell, 1985; Worthy, 1987a; Towns & Daugherty, 1994). All extant Leiopelma taxa are either at risk or nationally threatened (Newman et al., 2013) and rank within the top 100 Evolutionarily Distinct and Globally Endangered amphibians (Zoological Society of London, 2012). Leiopelma spp. are therefore of global conservation concern. Like most New Zealand’s endemic taxa, contemporary Leiopelma populations are highly fragmented and isolated. Leiopelma hochstetteri is the most widespread extant taxon, occurring as phylogeographically structured isolated populations in the northern North Island (Fig. 1c). Leiopelma archeyi occurs naturally in the northern North Island (Thames-Coromandel and Waitomo Districts) (Bishop et al., 2013; Fig. 1e). The two rarest taxa, L. hamiltoni and L. pakeka, are restricted to single insular island populations located off the northern South Island. Based on bathymetry and late Quaternary sea-level reconstructions, these populations have probably been isolated for the past 14000 years (Gibb, 1986; Lewis, Carter & Davey, 1994; Bishop et al., 2013; Thurlow, 2015) (Fig. 1d). An osteological and morphometric reassessment of modern extant and extinct leiopelmatids is warranted given concerns over possible cryptic diversity (Fouquet et al., 2010; Newman et al., 2013). Since Worthy’s (1987a, b) analyses, major additions of modern skeletal material and Pleistocene/Holocene fossils of Leiopelma to museum collections have occurred. Our study re-examined skeletal and external morphology to re-assess morphological differentiation, species limits and the taxonomy of leiopelmatids to provide new insights into the paleogeographic and evolutionary history of this globally significant and threatened anuran lineage. MATERIAL AND METHODS Data collection Skeletal material of Leiopelma was sourced from the University of Otago Department of Zoology (UOZ), Waitomo Caves Museum (WO), Museum of New Zealand Te Papa Tongarewa (NMNZ), Canterbury Museum (CM), Otago Museum (OM) and Landcare Research (LR). Modern material was originally obtained from the Thames-Coromandel region and nearshore islands off the northern South Island. Pleistocene/Holocene fossils of Leiopelma covered the known geographical and temporal range of this genus [regions: (North Island: -) Northland, Waitomo District, Hawkes Bay, Wairarapa (South Island: -) north-west Nelson, West Coast, Te Anau, Southland and Central Otago; age: c. 300–29000 BP] (Worthy, 1987a; Worthy & Holdaway, 1993, 1994; Worthy et al., 2002; Worthy & Roscoe, 2003). Measurements were taken using Vernier callipers to the nearest 0.1 mm. All late Quaternary specimens measured were osteologically mature (i.e. adult). Only unique elements (e.g. only the left femur) from sites/layers were measured to ensure independency of the data. When we could not confidently allocate fossil remains to a particular taxon (e.g. size overlap in L. hamiltoni and L. hochstetteri, Worthy & Holdaway, 1993), we conservatively grouped material from these taxa as ‘indeterminate’. Following Worthy (1987b), we recorded four measurements (proximal width, mid-shaft width, distal width and total length) for each of the five largest postcranial limb bones (humerus, radioulna, femur, tibiofibula and tibiale-fibulare). Only bones where accurate length measurements could be obtained were included in the analyses. However, we did include specimens where ossification of the capituli was absent as, for example, in L. waitomoensis, there is a lack of ossification of the cartilages at either end of the tibiofibula and humerus (Worthy, 1987b). Therefore, any length measurements for relevant specimens will underestimate the equivalent measurements made in other material where the capituli is present. To account for damage caused to the only available modern tibiofibulae of L. hamiltoni (NMNZ AM.292), we performed permutation mean matching for two missing values of total length and proximal width using the ‘mice’ package (van Buuren & Groothuis-Oudshoorn, 2011) in R v. 3.0.2 (R Development Core Team, 2013). To test for morphological variation in maxillae, we used a TG-3 digital camera (Olympus, USA) to photograph lingual views of all available complete modern and fossil maxillae [including specimens from the introduced Australian Ranoidea raniformis (sensuDubois & Frétey, 2016) and Litoria ewingii given that such taxa are sometimes found in fossil deposits (Worthy, 1987b) and some maxilla of Ranoidea spp. have previously been misidentified as those of Leiopelma spp., e.g. CM AM.78]. Digital photographs were downloaded into R, and using the package ‘geomorph’ (Adams & Otárola-Castillo, 2013; Adams et al., 2017), 18 spatial ‘landmarks’ (positions on the maxilla) for each specimen were selected (Supporting Information, Fig. S1). Collection details and measurements of all skeletal material are available in Supporting Information. External morphometric data of extant frogs, which consisted of 19 measurements, were sourced from Bell (1978, 1994). Statistical analyses All analyses on raw data were performed in R v. 3.0.2 (R Development Core Team, 2013). For identifiable specimens that consisted of complete sets of associated morphological elements, we performed a multivariate discriminant function analysis (DFA) on the actual (non-transformed) measurements. We assessed the data on postcranial limb element sizes using a Bayesian information criterion (BIC) model-based clustering and classification approach within the package ‘mclust’ (Fraley et al., 2012). For the element(s) with the most discrete and highest number of clusters, we compared length (size) and proximal/distal width/length ratios (shape) between taxa using non-parametric and univariate Mann–Whitney U (MWU) tests. These tests were not performed for L. auroraensis and L. hamiltoni because for each of these species only one skeleton exists in collections. We tested for inter-specific diagnosability using multivariate principal component analysis (PCA) and DFA on the actual (non-transformed) measurements to determine levels of differentiation and variance explained. We subsequently tested for intra-specific variation using the same three analyses (MWU, DFA and PCA), but only for L. markhami and L. waitomoensis as these taxa were the most widespread prehistorically (Worthy, 1987a, b). Leiopelma auroraensis was included in the PCA for L. markhami to test whether this taxon is the southern extent of a cline that follows Bergmann’s rule. To test for shape variation in maxillae, coordinates of left maxillae were transformed by rotating values on the y-axis so that they aligned with those represented by right maxillae for comparison. Morphological ‘landmark’ coordinates were standardized for size using a generalized Procrustes analysis (see Rohlf & Slice, 1990) in the package ‘geomorph’ (Adams & Otárola-Castillo, 2013; Adams et al., 2017). Using these standardized coordinates, inter- and intra-specific variations in the osteological ‘landmarks’ were assessed using multivariate analysis of variance (MANOVA) with an interaction between species and collection locality as fixed effects. The MANOVA involved 1000 iterations and was followed by post hoc tests. For external morphology, we performed multivariate DFA and PCA comparisons on actual (non-transformed) measurements to test for phenotypic differentiation between extant taxa. Statistical significance was taken at the P < 0.05 level. RESULTS Associated postcranial skeletons For associated postcranial skeletons, 100% of the individual specimens examined were ‘correctly’ assigned to their respective taxon in the DFA analysis (extinct taxa: L. auroraensis, n = 1; L. markhami, n = 10; L. waitomoensis, n = 1; extant taxa: L. archeyi, n = 10; L. pakeka, n = 31; L. hochstetteri, n = 4; L. hamiltoni, n = 1). Key features were that the femur and tibiofibula in L. auroraensis were of equal length, while the tibiofibula in L. waitomoensis was noticeably longer than the femur. Humerus length, in proportion to the tibiofibula, was longer in L. auroraensis than in other taxa. Isolated postcranial elements BIC model-based clustering of isolated postcranial elements estimated three size clusters for tibiofibulae (BIC = −1705.4, d.f. = 26, n = 540), femora (BIC = −1345.4, d.f. = 26, n = 445), humeri (BIC = −1608.3, d.f. = 26, n = 359) and radioulnae (BIC = −707.0, d.f. = 26, n = 258), whereas only two clusters were estimated for the tibiale-fibulare (BIC = −1098.2, d.f. = 20, n = 189). Of the elements with three clusters, tibiofibula was the most morphologically distinct (Supporting Information, Fig. S2); hence, additional analyses were performed only for this element. However, the degree of clustering in tibiofibulae may be due to large sample size. No signals of sexual dimorphism were seen for any element as the morphometric data followed a normal, not bimodal, distribution. The tibiofibulae lengths for extinct taxa [L. markhami: 15.5–30.4 mm (n = 190); L. auroraensis: 18.3 mm (n = 1); L. waitomoensis: 25.1–42.1 mm (n = 50)] were significantly larger than those of extant taxa, which generally overlapped in size, including taxa considered to be indeterminate [L. archeyi: 9.7–14.0 mm (n = 12); L. hamiltoni: 14.9 mm (n = 1); L. pakeka: 11.2–18.5 mm (n = 32); L. hochstetteri: 14.9–17.5 mm (n = 4); indeterminate: 10.1–21.7 mm (n = 250)] (Supporting Information, Table S1). Leiopelma archeyi had significantly smaller tibiofibulae compared with other taxa, but proximal/distal width/length ratios of L. archeyi (0.12–0.16) were relatively higher than L. hochstetteri (0.128–0.135). Indeterminate and proximal/distal width/length ratios of L. pakeka were also significantly broader compared with L. hochstetteri (0.12–0.17 and 0.11–0.16, respectively) and similar in breadth to L. hamiltoni (0.16). Proximal/distal width/length ratios of L. markhami and L. auroraensis were broader than other taxa (0.13–0.21 and 0.21, respectively), while tibiofibulae of L. waitomoensis were the most slender (0.11–0.15) (Supporting Information, Table S1). Inter-specific diagnosability of tibiofibulae was high for all extinct taxa: 97.9% from L. markhami, 98% from L. waitomoensis and 100% from L. auroraensis were assigned to their respective taxon. Only 0.5 and 1.6% of tibiofibulae from L. markhami were assigned as belonging to L. auroraensis and L. pakeka groups, respectively, and 2% of tibiofibulae from L. waitomoensis as L. markhami. For extant taxa, 58.3% of tibiofibulae from L. archeyi and 96.9% from L. pakeka were ‘correctly’ allocated. 41.7% of tibiofibulae from L. archeyi, 100% from L. hochstetteri and 100% from L. hamiltoni were classified as L. pakeka. Only 3.1% of tibiofibulae from L. pakeka were classified as L. archeyi. All tibiofibulae features measured loaded positively and equally on PC1 (axis scores: distal width = 0.51; mid-shaft width = 0.50; proximal width = 0.50; length = 0.49), which explained 94.9% of the variance. Positive values on the PC1 axis represent large tibiofibulae size (e.g. tibiofibulae of L. waitomoensis are generally longer than tibiofibulae of L. markhami, Fig. 2). Tibiofibulae length also loaded negatively on PC2 (−0.83), which accounted for 3.2% of the remaining variance. Negative values on the PC2 axis represent a decrease in overall width/length ratio, indicating tibiofibulae are gracile, whereas positive values indicate robust tibiofibulae (e.g. tibiofibulae of L. waitomoensis are gracile, whereas tibiofibulae of L. markhami are robust, Fig. 2). Figure 2. View largeDownload slide Principal component analysis of fossil and modern tibiofibulae (n = 540) measurements in Leiopelma frogs showing inter-specific size (PC1: small vs. large) and shape (PC2: robust vs. gracile) variation. Taxonomic classifications are as follows: L. auroraensis (solid black circle), L. markhami (hollow triangles), L. waitomoensis (hollow squares), indeterminate spp. (hollow grey circles), L. archeyi (solid squares), L. pakeka (plus signs), L. hamiltoni (solid red circle) and L. hochstetteri (solid triangles). For taxa with three or more data points, different coloured convex hull polygons are shown to highlight separate taxon groups. Figure 2. View largeDownload slide Principal component analysis of fossil and modern tibiofibulae (n = 540) measurements in Leiopelma frogs showing inter-specific size (PC1: small vs. large) and shape (PC2: robust vs. gracile) variation. Taxonomic classifications are as follows: L. auroraensis (solid black circle), L. markhami (hollow triangles), L. waitomoensis (hollow squares), indeterminate spp. (hollow grey circles), L. archeyi (solid squares), L. pakeka (plus signs), L. hamiltoni (solid red circle) and L. hochstetteri (solid triangles). For taxa with three or more data points, different coloured convex hull polygons are shown to highlight separate taxon groups. The tibiofibulae regional lengths of L. markhami, the most widespread of the extinct taxa, were mostly significantly different from each other in the North Island [Northland: 20.6 mm (n = 1); Waitomo: 16.9–22.6 mm (n = 24); Hawkes Bay: 19.1–21.2 mm (n = 6); Coonoor, Manawatu: 20.1 mm (n = 1); Wairarapa: 20.9–22.4 mm (n = 7)] and in the South Island [north-west Nelson: 18.0–30.4 mm (n = 133); West Coast: 15.5–25.1 mm (n = 18)] (Supporting Information, Table S2; Fig. 1a). In contrast, proximal/distal width/length ratios were generally similar throughout (Northland: 0.16; Waitomo: 0.15–0.19; Hawkes Bay: 0.15–0.17; Coonoor, Manawatu: 0.19; Wairarapa: 0.15–0.18; north-west Nelson: 0.13–0.21; West Coast: 0.13–0.18), but tibiofibulae of L. markhami from the South Island were significantly broader than two North Island locations (Supporting Information, Table S2). Tibiofibulae of the widespread North Island endemic L. waitomoensis were significantly smaller in Northland than those from Waitomo [25.1–28.6 mm (n = 4) and 25.5–42.1 mm (n = 40), respectively], whereas tibiofibulae from Hawkes Bay were significantly larger [36.9–40.5 mm (n = 3)]. Coonoor specimens were similar in length to those from Hawkes Bay [35.4–40.4 mm (n = 2)] (Supporting Information, Table S3; Fig. 1b). Proximal/distal width/length ratios were the same across regions (range = 0.11–0.15); thus, tibiofibulae of L. waitomoensis mainly differed in size, not shape (Supporting Information, Table S3). Intra-specific diagnosability of L. markhami and L. waitomoensis was high for some regions. The proportion of tibiofibulae from L. markhami correctly assigned to north-west Nelson, South Island, was 96.2%. For L. waitomoensis, 66.7 and 97.5% of tibiofibulae were correctly assigned to the North Island Hawkes Bay and Waitomo regions, respectively. As above, all tibiofibulae features measured loaded positively and equally on PC1 for both L. markhami (axis scores: distal width = 0.52; mid-shaft width = 0.47; proximal width = 0.52; length = 0.47) and L. waitomoensis (axis scores: distal width = 0.50; mid-shaft width = 0.49; proximal width = 0.50; length = 0.50). PC1 explained 82.0 and 94.9% of the variance for L. markhami and L. waitomoensis, respectively. Large tibiofibulae size was represented by positive values on the PC1 axis for L. markhami (Fig. 3) and L. waitomoensis (Fig. 4). The highest loadings on PC2 were length (−0.71) and mid-shaft width (0.70) for L. markhami, whereas only mid-shaft width (−0.85) loaded highly on PC2 for L. waitomoensis. PC2 explained 10.0 and 2.5% of the remaining variance for L. markhami and L. waitomoensis, respectively. PC2 again refers to the level of robustness, which indicates some intra-specific differentiation in tibiofibulae shape for both taxa (Figs 3, 4). The tibiofibula of L. auroraensis grouped with the most robust tibiofibulae of L. markhami from north-west Nelson, South Island (Fig. 3). Figure 3. View largeDownload slide Principal component analysis of fossil tibiofibulae (n = 191) measurements in the extinct Leiopelma markhami and L. auroraensis (solid black triangle) showing intra- and inter-specific size [PC1: small (negative values) vs. large (positive values)] and shape [PC2: gracile (negative values) vs. robust (positive values)] variation. North Island fossil locations are as follows: Northland (hollow triangles), Waitomo (hollow circles), Hawkes Bay (+ symbols), Coonoor (solid square) and Wairarapa (x symbols). South Island fossil locations are north-west Nelson (abbreviated as NW Nelson, solid grey circles) and West Coast (solid black circles). For regions with three or more data points, different coloured convex hull polygons are shown to highlight separate geographical groups. Figure 3. View largeDownload slide Principal component analysis of fossil tibiofibulae (n = 191) measurements in the extinct Leiopelma markhami and L. auroraensis (solid black triangle) showing intra- and inter-specific size [PC1: small (negative values) vs. large (positive values)] and shape [PC2: gracile (negative values) vs. robust (positive values)] variation. North Island fossil locations are as follows: Northland (hollow triangles), Waitomo (hollow circles), Hawkes Bay (+ symbols), Coonoor (solid square) and Wairarapa (x symbols). South Island fossil locations are north-west Nelson (abbreviated as NW Nelson, solid grey circles) and West Coast (solid black circles). For regions with three or more data points, different coloured convex hull polygons are shown to highlight separate geographical groups. Figure 4. View largeDownload slide Principal component analysis of fossil tibiofibulae (n = 49) measurements in the extinct Leiopelma waitomoensis, showing intra-specific size [PC1: small (negative values) vs. large (positive values)] and shape [PC2: gracile (negative values) vs. robust (positive values)] variation. North Island fossil locations are as follows: Northland (hollow triangles), Waitomo (hollow circles), Hawkes Bay (+ symbols) and Coonoor (solid squares). For regions with three or more data points, different coloured convex hull polygons are shown to highlight separate geographical groups. Figure 4. View largeDownload slide Principal component analysis of fossil tibiofibulae (n = 49) measurements in the extinct Leiopelma waitomoensis, showing intra-specific size [PC1: small (negative values) vs. large (positive values)] and shape [PC2: gracile (negative values) vs. robust (positive values)] variation. North Island fossil locations are as follows: Northland (hollow triangles), Waitomo (hollow circles), Hawkes Bay (+ symbols) and Coonoor (solid squares). For regions with three or more data points, different coloured convex hull polygons are shown to highlight separate geographical groups. Maxillae Osteological shape data of Leiopelma maxillae (Fig. 5) were collected from 161 complete specimens [L. auroraensis, n = 1; L. markhami, n = 84 (West Coast, n = 4; north-west Nelson, n = 72; Wairarapa, n = 3; Waitomo, n = 5); L. waitomoensis, n = 9 (Northland, n = 1; Waitomo, n = 8); L. archeyi, n = 9; L. pakeka, n = 7; L. hochstetteri, n = 3; L. hamiltoni, n = 1; indeterminate, n = 47 (West Coast, n = 4; north-west Nelson, n = 37; Waitomo, n = 5; Northland, n = 1)] and from seven specimens of Ranoidea/Litoria spp. Maxillae morphology of Leiopelma consisted of four main character states: (1) those that had a broad pars facialis and un-notched anterior; (2) those with a deep notch anterior to a narrow preorbital process, (3) those with a subtle preorbital process, elongated anterior section and fairly broad overall; and (4) those with a shallow notch, slightly wide preorbital process and an elongated tip of the posterior process (Fig. 5). The latter two characters were additional to those identified by Worthy (1987b). Maxillae of Ranoidea/Litoria spp. were similar to those from the extinct L. waitomoensis, in that Ranoidea/Litoria spp. had an elongated anterior section and the presence of a small preorbital process. However, in comparison with L. waitomoensis, maxillae of Ranoidea/Litoria spp. were not as broad, notches were distinctive and the interior curvature of the dorsal section was pronounced. The maxilla of L. auroraensis was similar to L. markhami, which typically had a prominent preorbital process and deep anterior notch. The main differences between these two extinct taxa were that the maxilla of L. auroraensis was more compacted posteriorly and that the preorbital process was more vertical with a shallower notch. Maxillae of extant Leiopelma taxa had no notch but a broad pars facialis or had a shallow notch or preorbital process (Fig. 5). Unlike extinct Leiopelma, the osteology of extant Leiopelma maxillae generally overlapped despite some subtle differences (Supporting Information, Table S4). For example, the posterior slope of the preorbital process was slightly shallower in L. archeyi than in other extant taxa. Moreover, the preorbital process was more pronounced in L. pakeka and L. hamiltoni compared with L. archeyi and L. hochstetteri. Osteology of indeterminate maxillae overlapped with extant and extinct taxa (Supporting Information, Table S4). There were significant differences in maxillae osteology between species (d.f. = 8, F = 3.49, P = 0.001) but not between regions (d.f. = 11, F = 0.50, P = 0.08). Intra-specific variation in maxillae was geographically conservative (d.f. = 88, F = 0.03, P = 0.88). The amount of variance explained by PC1 and PC2 was 26.4 and 17.3%, respectively. Figure 5. View largeDownload slide Principal component analysis and deformation mesh grids showing the variation in maxillae shape (n = 168) in New Zealand Leiopelma, Australian Ranoidea and Australian Litoria frogs. Taxonomic classifications are as follows: L. auroraensis (lime green), L. markhami (light blue), L. waitomoensis (purple), L. archeyi (orange), L. pakeka (dark blue), L. hochstetteri (cyan), L. hamiltoni (turquoise), indeterminate Leiopelma spp. (red) and Ranoidea/Litoria spp. (pink). Figure 5. View largeDownload slide Principal component analysis and deformation mesh grids showing the variation in maxillae shape (n = 168) in New Zealand Leiopelma, Australian Ranoidea and Australian Litoria frogs. Taxonomic classifications are as follows: L. auroraensis (lime green), L. markhami (light blue), L. waitomoensis (purple), L. archeyi (orange), L. pakeka (dark blue), L. hochstetteri (cyan), L. hamiltoni (turquoise), indeterminate Leiopelma spp. (red) and Ranoidea/Litoria spp. (pink). External morphology Discrimination of extant Leiopelma taxa was high in the re-analysis of the 19 external morphometric measurements. When all four extant taxa were included, 100% of L. archeyi (n = 33), L. pakeka (n = 27) and L. hochstetteri (n = 16) individuals were ‘correctly’ assigned. 81.8% of L. hamiltoni (n = 11) individuals were assigned to this taxon, while the rest were classified as L. pakeka. When only L. pakeka and L. hamiltoni were included in the analysis, all individuals were assigned to their respective taxon. Likewise, frogs from two geographically isolated populations of L. archeyi in the North Island [Thames-Coromandel District (n = 23) and Whareorino in Waitomo District (n = 10)] were all ‘correctly’ assigned when analysed separately from other taxa. In contrast, the PCA yielded major overlap for isolated North Island populations of L. hochstetteri [East Cape (n = 3) and Coromandel (n = 13)], the North Island populations of L. archeyi and for L. pakeka and L. hamiltoni (Supporting Information, Fig. S3). All external features loaded positively on PC1 (data not shown), while radioulna and humerus widths loaded positively on PC2 (0.42 and 0.45, respectively), and the length of the third fore-finger and nostril-eye length had equal negative loadings (−0.35). PC2 loadings indicate that terrestrial Leiopelma have less robust forelimbs than L. hochstetteri, although the former taxa have elongated snouts and fore-fingers (Supporting Information, Fig. S3). PC1 and PC2 explained 79.6 and 9.9% of the variance, respectively. DISCUSSION Ecological correlates of morphology Associated postcranial limb skeletons of known taxa were all ‘correctly’ allocated to their taxon. In particular, L. waitomoensis had lower humerus/tibiofibula and femur/tibiofibula ratios than in other Leiopelma taxa, while L. auroraensis and L. markhami had the highest. Humerus/tibiofibula and femur/tibiofibula length ratios are negatively correlated with jumping ability (Zug, 1972) and, as such, indicate that L. waitomoensis was a strong jumper, unlike L. markhami and L. auroraensis, which are suggested to have primarily walked (Worthy, 1987a). Moreover, physical characteristics of fossil sites suggest that L. waitomoensis was a stream-dweller as its remains are predominantly found in cave deposits associated with streams (Worthy, 1987a). In contrast, L. markhami are predominantly found in pitfall deposits, which disproportionally comprise terrestrial fauna (Worthy, 1987a). Field observations of behaviour in the semi-aquatic L. hochstetteri, and terrestrial L. archeyi, L. pakeka and L. hamiltoni, attest to this association between locomotor repertoire and the environment (Newman, 1990; Reilly et al., 2015). Inter-specific variation in isolated postcranial elements The results of our study, using c. 1000 new specimens, strongly support those of Worthy (1987a, b). Worthy (1987b) estimated snout-vent lengths (SVLs) of extinct L. markhami and L. auroraensis to be c. 50–60 mm, whereas L. waitomoensis was c. 100 mm. In contrast, extant taxa only reach up to c. 50 mm SVL, the smallest being L. archeyi, which is generally between 32- and 40-mm SVL (Worthy, 1987b). Leiopelma markhami and L. auroraensis have been described as morphologically robust, indicated by the broadness of their hindlimb bones. Leiopelma waitomoensis had relatively narrow hindlimb bones and is thus gracile in comparison (Worthy, 1987b). The distinctiveness of these size characteristics is demonstrated by the high diagnosability of tibiofibulae of extinct taxa (Worthy, 1987b; this study). However, while Worthy (1987b) assigned 98.1% of tibiofibulae from L. markhami (n = 101) and 100% of tibiofibulae from L. waitomoensis (n = 43) to their respective taxon, the holotype tibiofibula of L. auroraensis was classified as L. markhami. Worthy’s (1987b) measurements of the holotype differed from our measurements of the same specimen by c. 1 mm, which is why there were contrasting results regarding the diagnosability of L. auroraensis and L. markhami between our studies. Nevertheless, isolated postcranial elements of these taxa are similar. Overall tibiofibulae size range in extant taxa was similar to that reported by Worthy (1987b). However, Worthy (1987b) found proximal/distal widths of L. archeyi to be significantly different from all taxa except L. hochstetteri – in contrast, we found significant differences between all identifiable taxa except L. pakeka (albeit L. pakeka being described as L. hamiltoni in Worthy, 1987b) (Supporting Information, Table S1). Leiopelma archeyi is the only extant taxon that differentiates by its small tibiofibulae size. We thus conclude that size-related differences, but minimal genetic distance between L. archeyi, L. pakeka and L. hamiltoni compared with L. hochstetteri (see Holyoake et al., 2001; Carr et al., 2015), suggest that L. archeyi, L. pakeka and L. hamiltoni represent a single species that has become fragmented into several isolated populations. The overlap of size and lack of diagnosable osteological features between extant taxa and indeterminate fossils support Worthy’s (1987b) contention that such fossils could be either L. hamiltoni or L. hochstetteri [L. archeyi fossils were not described at the time of Worthy’s (1987b) study]. In our study, tibiofibulae from L. hamiltoni and L. pakeka were similar to specimens of indeterminate tibiofibulae, but broader than L. hochstetteri, which implies that indeterminate fossils are more likely to comprise the former taxa. Skeletal morphology of isolated tibiofibulae (and postcranial limb elements in general) is therefore problematic in distinguishing between the Leiopelma taxa. There are some osteological differences in the forelimbs of L. hochstetteri and L. hamiltoni [e.g. lateral and medial crests of the humeri are present in the former and not in the latter (Worthy, 1987b; Worthy et al., 2013)], but from what we observed these characteristics are not always distinct. Temperature-related effects on postcranial elements The two most widespread extinct taxa (Fig. 1a, b), L. waitomoensis and L. markhami, obey Bergmann’s rule in that they show a latitudinal cline in tibiofibulae size with individuals from southern populations larger than those from northern populations (Worthy, 1987a, b). However, tibiofibulae of L. markhami from the north-west South Island were relatively larger than those from the South Island’s West Coast (this was not reported by Worthy, 1987a, b as he did not measure fossils from the latter location). Our observation suggests that an interaction between temperatures associated with altitude and latitude is the primary factor responsible for these size patterns (Berven, Gill & Smith-Gill, 1979). We consider the effect of temporal variation (i.e. the deposition of fossils during warm and cool periods) on Leiopelma size to be minimal as radio-carbon dating of associated avifaunal bones indicates most leiopelmatid bones to be of Holocene age from time-averaged fossil deposits (Worthy, 1987a; Worthy & Holdaway, 1993, 1994; Worthy et al., 2002; Worthy & Roscoe, 2003). Furthermore, warm microclimates driven by the subtropical oceanic current (Alloway et al., 2007) and low elevation may have led to reduced body sizes of L. markhami (and perhaps other sympatric Leiopelma) along the South Island’s West Coast. Non-temperate and latitude-related correlations of postcranial elements Variation in tibiofibulae shape of L. markhami was noticeably different between certain regions (Fig. 3; Supporting Information, Table S2). Tibiofibulae from the north-west South Island were generally more robust than those from the South Island’s West Coast and North Island. Some L. markhami tibiofibulae from the north-west South Island were remarkably similar in appearance to the holotype of L. auroraensis from the southern South Island (Te Anau). This extent of differentiation in tibiofibulae shape within L. markhami was not observed in L. waitomoensis (Fig. 4; Supporting Information, Table S3) although there was high diagnosability of tibiofibulae from the eastern and northern North Island. Overall, these intra-specific size and shape differences suggest that cryptic taxa may be present within these widespread extinct species. Diagnosability of maxillae Worthy (1987b) described two osteological character states in the maxilla: taxa that have a notch anterior to the preorbital process (L. hochstetteri, L. markhami and L. auroraensis) and taxa that do not (L. waitomoensis, L. archeyi, L. hamiltoni and L. pakeka). We identified two additional character states: (1) those that were broad overall, with a small preorbital process and elongated anterior, and (2) those with a shallow notch, expanded preorbital process and an elongated posterior process (Fig. 5). Significant inter-specific differences in maxilla shape were mainly limited to extinct taxa, yet there were some differences between extant taxa (Supporting Information, Table S4). In L. archeyi, the preorbital process was relatively reduced and the maxillae were more compact compared with L. pakeka. Despite these taxa being osteologically identical except for size (Worthy, 1987b), L. archeyi is osteologically neotenic, meaning that the level of ossification is lower compared with all other Leiopelma (Stephenson, 1951; Worthy et al., 2013). Hence, this neotenic condition probably explains the structural differences that we observed. Maxillae osteology did not vary intra-specifically across regions, which means that, based on Worthy’s (1987a) and our study, cranial elements do not change size with latitude and altitude cf. postcranial elements. In introduced Australian Ranoidea and Litoria frogs, maxilla specimens were distinct from most Leiopelma. Maxillae of Ranoidea and Litoria spp. were relatively similar to L. waitomoensis, but shared more features with evolutionarily derived anuran taxa (e.g. Eleutherodactylus spp., Bochaton et al., 2015), such as bearing many teeth and a poorly developed pars facialis. External morphology Differences in external morphological characters among currently recognized Leiopelma species and isolated populations have been discussed and debated (see Stephenson & Stephenson, 1957; Bell, 1978, 1994; Bell et al., 1998a; Bell, Daugherty & Hitchmough, 1998b). For instance, Bell (1994) found no morphometric differences between L. hamiltoni and L. pakeka when comparing all extant taxa. Later multivariate analyses of external morphometric characters that focused on comparisons of just two populations, however, concluded that there were morphometric differences between these taxa and between L. archeyi populations in the Coromandel and Waitomo Districts (Bell et al., 1998a, b). The extant L. hochstetteri is clearly robust in relation to other extant taxa and there is no difference externally between all terrestrial taxa apart from size (Bell, 1978, 1994; Newman, 1990; Bell et al., 1998a, b). While there are some statistically significant phenotypic differences between L. hamiltoni and L. pakeka (see Bell et al., 1998a), such differences are not considered biologically relevant (Newman, 1990; Bell et al., 1998a) – akin to isolated populations of L. archeyi for instance (Bell, 1994; Bell et al., 1998b). Considering there is no genetic, ecological, skeletal or external morphometric evidence to support the distinction between L. hamiltoni and L. pakeka, we therefore suggest the synonymization of L. pakeka Bell, Daugherty & Hay, 1998 and L. hamiltoni McCulloch, 1919, with L. hamiltoni retaining taxonomic priority. Glacier-driven evolution and human-driven extinctions Glaciations during the Pleistocene promoted speciation in alpine taxa (e.g. Aves, Dinornithiformes) by severing their continuous distributions transverse to the Southern Alps, South Island (Bunce et al., 2009; Wallis, Waters, Upton & Craw, 2016). Leiopelma spp. potentially would have been similarly affected. Given the limited number of late Quaternary Leiopelma fossils from the southern South Island, mainly due to the rarity of suitable fossil sites in the region, the distinction between northern and southern Leiopelma assemblages on either side of the central transverse zone of the Southern Alps is unclear, but the presumption that L. auroraensis is a local derivative of L. markhami may support this glacial–speciation hypothesis (Worthy, 1987a). Leiopelma auroraensis should, therefore, be retained as a distinct taxon, as demonstrated by our multivariate analyses and Worthy’s (1987b) osteological descriptions. Likewise, glaciation may have driven the morphological differentiation observed between L. markhami populations in the western and north-western South Island, whereas the geographical separation of the North and South Island L. markhami, L. hamiltoni and L. hochstetteri populations may have occurred more recently when post-glacial sea-level rise fractured land-locked islands of the South Island’s Marlborough Sounds at least 14000 BP (Gibb, 1986; Lewis et al., 1994). The nearshore island populations of L. hamiltoni also became isolated at this time (Thurlow, 2015). Human arrival in the late 13th century (Wilmshurst et al., 2008) led to the extirpation and extinction of many Leiopelma populations (and other herpetofauna), primarily due to habitat loss and predation by introduced mammals (e.g. kiore rat, R. exulans) (Bell, 1985; Worthy, 1987a; Daugherty et al., 1990; Towns & Daugherty, 1994). Large herpetofauna, such as L. waitomoensis, were among the first to disappear, while the remaining smaller species became restricted to isolated, relict distributions (Worthy, 1987a). Leiopelma markhami and L. hamiltoni, which were sympatric with L. hochstetteri (Fig. 1a, c, d), persisted in the north-west South Island until c. 400 BP–southern South Island (Te Anau) fossils of L. markhami come from even younger deposits (c. 300 BP) (Worthy, 1987a; Worthy & Holdaway, 1993, 1994; Worthy & Roscoe, 2003). The extirpation of leiopelmatids from the main islands is therefore recent and human induced. Indeed, Green, Zeyl & Sharbel (1993) and Green (1994) concluded that extirpations of North Island L. hochstetteri populations were accentuated after human settlement. Conservation implications Species translocations are a key tool in managing threatened herpetofauna (Germano & Bishop, 2009), but these are dependent on recognizing species limits and taxonomic units, for which there is high uncertainty in Leiopelma (Newman et al., 2013). Conservation palaeontology and the utilization of fossil data within a contemporary context can, however, provide insights for managing extant taxa (Bochaton et al., 2015). Based on the morphology of fossil and modern skeletal material, our study supports the possibility of extinctions of cryptic taxa in Leiopelma and potential major taxonomic changes, but to test these hypotheses, future research will require a genetic approach using modern and ancient DNA from late Quaternary fossils. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Figure S1. Lingual view of a right Holocene fossil maxilla of Leiopelma markhami (NMNZ S.23121), highlighting the main osteological features and the 18 spatial ‘landmarks’ (positions on the maxilla, solid red circles) that were selected to establish coordinates that described the shape. Scale bar 1 mm (red dotted line). Figure S2. Bayesian information criterion (BIC) model-based clustering of five leiopelmatid frog limb elements: (a) humeri (n &#x003D; 359), (b) radioulnae (n &#x003D; 258), (c) femora (n &#x003D; 445), (d) tibiofibulae (n &#x003D; 540), (e) tibiale-fibulare (n &#x003D; 189), and four respective measurements [distal width, mid-shaft width, proximal width and length (in mm)]. Different colours and symbols represent unique clusters identified by the BIC model-based cluster estimation. Abbreviations are as follows: H &#x003D; humeri, R &#x003D; radioulnae, F &#x003D; femora, T &#x003D; tibiofibulae, Tf &#x003D; tibiale-fibulare, d &#x003D; distal width, s &#x003D; mid-shaft width, p &#x003D; proximal width and l &#x003D; length. For example, ‘Hd’ refers to the distal width of humeri. Figure S3. Principal component analysis of 19 external measurements in extant Leiopelma frogs (see Bell, 1978, 1994 for details) showing inter- and intra-specific shape, and size variation. Different coloured convex hull polygons are shown to highlight separate groups, which are as follows: L. hochstetteri [Coromandel (abbreviated as CORO, n &#x003D; 13), solid circles], L. hochstetteri [East Cape (abbreviated as ECAP, n &#x003D; 3), hollow circles], L. archeyi [CORO (n &#x003D; 23), plus signs], L. archeyi [Whareorino (abbreviated as WHAR, n &#x003D; 10), solid triangles], L. hamiltoni [Stephens Island (abbreviated as STEP, n &#x003D; 11), solid squares] and L. pakeka [Maud Island (abbreviated as MAUD, n &#x003D; 27), hollow squares]. Table S1. Non-parametric Mann–Whitney U test comparisons of tibiofibulae size between Leiopelma taxa. Leiopelma auroraensis and L. hamiltoni were excluded as they had sample sizes of one. Total length and distal width/length ratio comparisons are presented in the upper and lower diagonal sections, respectively. Distal and proximal widths were highly correlated (Pearson’s correlation: t &#x003D; 53.73, d.f. &#x003D; 538, P < 0.001, correlation &#x003D; 0.918), thus only distal width/length ratio comparisons are presented in the table. Mid-shaft width/length ratios were virtually identical between all taxa and were thus excluded from the table. Statistical significance was taken at the P < 0.05 level. Significant P-values are highlighted in bold. Table S2. Non-parametric Mann–Whitney U test comparisons of tibiofibulae size between Leiopelma markhami fossil locations of North Island and South Island, New Zealand. North Island regions include Waitomo, Hawkes Bay and the Wairarapa. South Island regions include north-west Nelson (abbreviated as NW Nelson) and the West Coast. Length and distal width/length ratio comparisons are presented in the upper and lower diagonal sections, respectively. Statistical significance was taken at the P < 0.05 level. Significant P-values are highlighted in bold. Table S3. Non-parametric Mann–Whitney U test comparisons of tibiofibulae size between Leiopelma waitomoensis fossil locations of North Island, New Zealand. North Island regions include Northland, Waitomo, Hawkes Bay and Coonoor. Length and distal width/length ratio comparisons are presented in the upper and lower diagonal sections, respectively. Statistical significance was taken at the P < 0.05 level. Significant P-values are highlighted in bold. Table S4. Multivariate analysis of variance comparisons of maxillae shape in New Zealand Leiopelma and Australian Ranoidea/Litoria frogs. Statistical significance was taken at the P < 0.05 level. Significant P-values are highlighted in bold. ACKNOWLEDGEMENTS We thank the editor Dr. Louise Allcock and one anonymous reviewer for providing helpful comments on the manuscript. We thank the staff of New Zealand museums (in particular Otago Museum, Canterbury Museum, Te Papa Museum and Waitomo Caves Museum) and Jamie Wood from Landcare Research for access to collections. We also thank Ana Dula for helping excavate fossil material from the Wairarapa, Travis Ingram and Scott Jarvie for providing R code and statistical advice, and Benno Kappers from the Department of Conservation for providing distribution data. L.J.E. acknowledges support from a University of Otago Doctoral Scholarship. REFERENCES Adams DC , Collyer ML , Kaliontzopoulou A , Sherratt E . 2017 . Geomorph: software for geometric morphometric analyses, R package version 3.0.5 . Available at: http://cran.r-project.org/package=geomorph (accessed 30 August 2016 ). Adams DC , Otárola-Castillo E . 2013 . Geomorph: an R package for the collection and analysis of geometric morphometric shape data . Methods in Ecology and Evolution 4 : 393 – 399 . Google Scholar CrossRef Search ADS Alloway BV , Lowe DJ , Barrell DJA , Newnham RM , Almond PC , Augustinus PC , Bertler NAN , Carter L , Litchfield NJ , McGlone MS , Shulmeister J , Vandergoes MJ , Williams PW ; NZ-INTIMATE Members . 2007 . Towards a climate event stratigraphy for New Zealand over the past 30000 years (NZ-INTIMATE project) . Journal of Quaternary Science 22 : 9 – 35 . Google Scholar CrossRef Search ADS Bell BD . 1978 . Observations on the ecology and reproduction of the New Zealand Leiopelmid frogs . Herpetologica 34 : 340 – 354 . Bell BD . 1985 . Conservation status of the endemic New Zealand frogs . In: Grigg G , Shine R , Ehmann H , eds. Biology of Australasian frogs and reptiles . New South Wales : Surrey Beatty & Sons , 449 – 458 . Bell BD . 1994 . A review of the status of New Zealand Leiopelma species (Anura: Leiopelmatidae), including a summary of demographic studies in Coromandel and on Maud Island . New Zealand Journal of Zoology 21 : 341 – 349 . Google Scholar CrossRef Search ADS Bell BD . 2010 . The threatened leiopelmatid frogs of New Zealand: natural history integrates with conservation . Herpetological Conservation and Biology 5 : 515 – 528 . Bell BD , Daugherty CH , Hay JM . 1998a . Leiopelma pakeka, n. sp. (Anura: Leiopelmatidae), a cryptic species of frog from Maud Island, New Zealand, and a reassessment of the conservation status of L. hamiltoni from Stephens Island . Journal of the Royal Society of New Zealand 28 : 39 – 54 . Google Scholar CrossRef Search ADS Bell BD , Daugherty CH , Hitchmough RA . 1998b . The taxonomic identity of a population of terrestrial Leiopelma (Anura: Leiopelmatidae) recently discovered in the northern King Country, New Zealand . New Zealand Journal of Zoology 25 : 139 – 146 . Google Scholar CrossRef Search ADS Berven KA , Gill DE , Smith-Gill SJ . 1979 . Countergradient selection in the green frog, Rana clamitans . Evolution 33 : 609 – 623 . Google Scholar CrossRef Search ADS Bishop PJ , Daglish LA , Haigh AJM , Marshall LJ , Tocher MD , McKenzie KL . 2013 . Native frog (Leiopelma species) recovery plan, 2013–2018 . Threatened Species Recovery Plan 63. Wellington : Department of Conservation . Bunce M , Worthy TH , Phillips MJ , Holdaway RN , Willerslev E , Haile J , Shapiro B , Scofield RP , Drummond A , Kamp PJJ , Cooper A . 2009 . The evolutionary history of the extinct ratite moa and New Zealand Neogene paleogeography . Proceedings of the National Academy of Sciences of the United States of America 106 : 20646 – 20651 . Google Scholar CrossRef Search ADS van Buuren S , Groothuis-Oudshoorn K . 2011 . mice: multivariate imputation by chained equations in R . Journal of Statistical Software 45 : 1 – 67 . Carr LM , McLenachan PA , Waddell PJ , Gemmell NJ , Penny D . 2015 . Analyses of the mitochondrial genome of Leiopelma hochstetteri argues against the full drowning of New Zealand . Journal of Biogeography 42 : 1066 – 1076 . Google Scholar CrossRef Search ADS Clarke BT . 2007 . Comparative morphology and amphibian taxonomy: an example from the osteology of discoglossoid frogs . In: Heatwole H , Tyler MJ , eds. Amphibian biology: systematics, Vol. 7 . New South Wales : Surrey Beatty & Sons , 2465 – 2603 . Daugherty CH , Cree A , Hay JM , Thompson MB . 1990 . Neglected taxonomy and continuing extinctions of tuatara (Sphenodon) . Nature 347 : 177 . Google Scholar CrossRef Search ADS Daugherty CH , Patterson GB , Hitchmough RA . 1994 . Taxonomic and conservation review of the New Zealand herpetofauna . New Zealand Journal of Ecology 21 : 317 – 323 . Google Scholar CrossRef Search ADS Dubois A , Frétey T . 2016 . A new nomen for a subfamily of frogs (Amphibia, Anura) . Dumerilia 6 : 17 – 23 . Fouquet A , Green DM , Waldman B , Bowsher JH , McBride KP , Gemmell NJ . 2010 . Phylogeography of Leiopelma hochstetteri reveals strong genetic structure and suggests new conservation priorities . Conservation Genetics 11 : 907 – 919 . Google Scholar CrossRef Search ADS Fraley C , Raftery AE , Murphy TB , Scrucca L . 2012 . mclust: version 4 for R: normal mixture modeling for model-based clustering, classification, and density estimation . Technical Report 597. Washington : Department of Statistics, University of Washington . Germano JM , Bishop PJ . 2009 . Suitability of amphibians and reptiles for translocation . Conservation Biology 23 : 7 – 15 . Google Scholar CrossRef Search ADS Germano JM , Cree A , Bishop PJ . 2011 . Ruling out the boys from the girls: can subtle morphological differences identify sex of the apparently monomorphic frog, Leiopelma pakeka ? New Zealand Journal of Zoology 38 : 161 – 171 . Google Scholar CrossRef Search ADS Gibb JG . 1986 . A New Zealand regional Holocene eustatic sea-level curve and its application to determination of vertical tectonic movements . Royal Society of New Zealand 24 : 377 – 395 . Green DM . 1994 . Genetic and cytogenetic diversity in Hochstetter’s frog, Leiopelma hochstetteri, and its importance for conservation management . New Zealand Journal of Zoology 21 : 417 – 424 . Google Scholar CrossRef Search ADS Green DM , Zeyl CW , Sharbel TF . 1993 . The evolution of hypervariable sex and supernumerary (B) chromosomes in the relict New Zealand frog, Leiopelma hochstetteri . Journal of Evolutionary Biology 6 : 417 – 441 . Google Scholar CrossRef Search ADS Hay JM , Daugherty CH , Cree A , Maxson LR . 2003 . Low genetic divergence obscures phylogeny among populations of Sphenodon, remnant of an ancient reptile lineage . Molecular Phylogenetics and Evolution 29 : 1 – 19 . Google Scholar CrossRef Search ADS Hay JM , Sarre SD , Lambert DM , Allendorf FW , Daugherty CH . 2010 . Genetic diversity and taxonomy: a reassessment of species designation in tuatara (Sphenodon: Reptilia) . Conservation Genetics 11 : 1063 – 1081 . Google Scholar CrossRef Search ADS Holyoake A , Waldman B , Gemmell NJ . 2001 . Determining the species status of one of the world’s rarest frogs: a conservation dilemma . Animal Conservation 4 : 29 – 35 . Google Scholar CrossRef Search ADS Kennedy M , Paterson AM , Morales JC , Parsons S , Winnington AP , Spencer HG . 1999 . The long and short of it: branch lengths and the problem of placing the New Zealand short-tailed bat, Mystacina . Molecular Phylogenetics and Evolution 13 : 405 – 416 . Google Scholar CrossRef Search ADS Lewis KB , Carter L , Davey FJ . 1994 . The opening of Cook Strait: interglacial tidal scour and aligning basins at a subduction to transform plate edge . Marine Geology 116 : 293 – 312 . Google Scholar CrossRef Search ADS Newman DG . 1990 . Activity, dispersion, and population densities of Hamilton’s frog (Leiopelma hamiltoni) on Maud and Stephens Islands, New Zealand . Herpetologica 46 : 319 – 330 . Newman DG , Bell BD , Bishop PJ , Burns R , Haigh A , Hitchmough RA . 2013 . Conservation status of New Zealand frogs, 2013 . New Zealand Threat Classification Series 5. Wellington : Department of Conservation . R Development Core Team . 2013 . R: a language and environment for statistical computing . Vienna : R Foundation for Statistical Computing . Available at: http://www.R-project.org (accessed 10 January 2017). Reilly S , Essner R , Wren S , Easton L , Bishop P . 2015 . Movement patterns in leiopelmatid frogs: insights into the locomotor repertoire of the earliest anurans . Behavioural Processes 121 : 43 – 53 . Google Scholar CrossRef Search ADS Rohlf FJ , Slice DE . 1990 . Extensions of the Procrustes method for the optimal superimposition of landmarks . Systematic Zoology 39 : 40 – 59 . Google Scholar CrossRef Search ADS Stephenson NG . 1951 . Observations on the development of the amphicoelous frogs, Leiopelma and Ascaphus . Journal of the Linnean Society of London, Zoology 42 : 18 – 28 . Google Scholar CrossRef Search ADS Stephenson EM , Stephenson NG . 1957 . Field observations on the New Zealand frog, Leiopelma Fitzinger . Transactions of the Royal Society of New Zealand 84 : 867 – 882 . Tennyson AJD . 2010 . The origin and history of New Zealand’s terrestrial vertebrates . New Zealand Journal of Ecology 34 : 6 – 27 . Thurlow L . 2015 . Deducing the phylogeny of New Zealand’s endemic frog genus, Leiopelma . Unpublished M.Sc. Thesis, University of Otago . Towns DR , Daugherty CH . 1994 . Patterns of range contractions and extinctions in the New Zealand herpetofauna following human colonisation . New Zealand Journal of Zoology 21 : 325 – 339 . Google Scholar CrossRef Search ADS Wallis GP , Waters JM , Upton P , Craw D . 2016 . Transverse alpine speciation driven by glaciation . Trends in Ecology & Evolution 31 : 916 – 926 . Google Scholar CrossRef Search ADS Wilmshurst JM , Anderson AJ , Higham TFG , Worthy TH . 2008 . Dating the late prehistoric dispersal of Polynesians to New Zealand using the commensal Pacific rat . Proceedings of the National Academy of Sciences of the United States of America 105 : 7676 – 7680 . Google Scholar CrossRef Search ADS Worthy TH . 1987a . Palaeoecological information concerning members of the frog genus Leiopelma: Leiopelmatidae in New Zealand . Journal of the Royal Society of New Zealand 17 : 409 – 420 . Google Scholar CrossRef Search ADS Worthy TH . 1987b . Osteology of Leiopelma (Amphibia: Leiopelmatidae) and descriptions of three new subfossil Leiopelma species . Journal of the Royal Society of New Zealand 17 : 201 – 251 . Google Scholar CrossRef Search ADS Worthy TH . 1998 . The Quaternary fossil avifauna of Southland, South Island, New Zealand . Journal of the Royal Society of New Zealand 28 : 537 – 589 . Google Scholar CrossRef Search ADS Worthy TH , Holdaway RN . 1993 . Quaternary fossil faunas from caves in the Punakaiki area, West Coast, South Island, New Zealand . Journal of the Royal Society of New Zealand 23 : 147 – 254 . Google Scholar CrossRef Search ADS Worthy TH , Holdaway RN . 1994 . Quaternary fossil faunas from caves in Takaka Valley and on Takaka Hill, northwest Nelson, South Island, New Zealand . Journal of the Royal Society of New Zealand 24 : 297 – 391 . Google Scholar CrossRef Search ADS Worthy TH , Holdaway RN , Alloway BV , Jones J , Winn J , Turner D . 2002 . A rich Pleistocene-Holocene avifaunal sequence from Te Waka #1: terrestrial fossil vertebrate faunas from inland Hawke’s Bay, North Island, New Zealand. Part 2 . Tuhinga: Records of the Museum of New Zealand Te Papa Tongarewa 13 : 1 – 38 . Worthy TH , Roscoe D . 2003 . Takaka Fossil Cave – a stratified Late Glacial to Late Holocene deposit from Takaka Hill, New Zealand . Tuhinga: Records of the Museum of New Zealand Te Papa Tongarewa 14 : 41 – 60 . Worthy TH , Tennyson AJD , Scofield RP , Hand SJ . 2013 . Early Miocene fossil frogs (Anura: Leiopelmatidae) from New Zealand . Journal of the Royal Society of New Zealand 43 : 211 – 230 . Google Scholar CrossRef Search ADS Zoological Society of London . 2012 . Top 100 Evolutionary Distinct and Globally Endangered (EDGE) amphibians . Available at: http://www.edgeofexistence.org/amphibians/top 100.php (accessed 17 May 2014 ). Zug GR . 1972 . Anuran locomotion: structure and function. I. Preliminary observations on relation between jumping and osteometrics of appendicular and postaxial skeleton . Copeia 1972 : 613 – 624 . Google Scholar CrossRef Search ADS © 2017 The Linnean Society of London, Zoological Journal of the Linnean Society This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Zoological Journal of the Linnean Society Oxford University Press

Loading next page...
 
/lp/ou_press/testing-species-limits-of-new-zealand-s-leiopelmatid-frogs-through-Cuzh7OSCLS
Publisher
Oxford University Press
Copyright
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society
ISSN
0024-4082
eISSN
1096-3642
D.O.I.
10.1093/zoolinnean/zlx080
Publisher site
See Article on Publisher Site

Abstract

Abstract The taxonomic status of some of New Zealand’s endemic and threatened leiopelmatid frogs has been debated for decades. Clarifying this uncertainty is vital to their conservation, especially given the risk of extinction of cryptic taxa. We re-examined leiopelmatid diversity through multivariate analyses of the skeletal and external morphology of extinct and extant Leiopelma to determine morphological differentiation. Our results suggest that the morphological distinction between extinct taxa is greater than in modern extant taxa. While size ranges of postcranial elements overlapped within extant species, maxillae shape discriminated some extant taxa. We confirm the morphological distinctiveness of the extinct taxa recognized to date but identify latitudinal and altitudinal variation in postcranial element size and shape within the widespread Leiopelma markhami and L. waitomoensis, which suggest possible post-human extinction of cryptic taxa. Furthermore, the lack of morphological and osteological differentiation between L. archeyi and the insular extant L. hamiltoni and L. pakeka leads us to question the taxonomic distinctiveness of these three taxa. Future genetic research using modern and ancient DNA is recommended to enable species limits within Leiopelma to be tested in more detail to provide an evidence-based assessment for their conservation management. extinctions, frogs, Leiopelma, morphology, New Zealand, phylogeography, Quaternary fossils INTRODUCTION Diversification of New Zealand’s endemic biota has been driven by isolation, dynamic tectonic activity and climatic shifts for millions of years. Contrary to the expectation that evolutionarily ‘ancient’ lineages should be more taxon rich than ‘recent’ lineages, given longer evolutionary trajectories, extant Gondwanan vertebrate taxa in New Zealand are not speciose (Daugherty, Patterson & Hitchmough, 1994; Tennyson, 2010). Even so, potential cryptic species diversity within New Zealand’s endemic ‘ancient’ vertebrate lineages such as those of Mystacinidae (short-tailed bats), Sphenodontidae (tuatara) and Leiopelmatidae (frogs) has challenged researchers (Daugherty et al., 1990; Bell, Daugherty & Hay, 1998a; Kennedy et al., 1999; Holyoake, Waldman & Gemmell, 2001; Hay et al., 2003, 2010). Taxonomic uncertainties have major implications for the conservation management of ‘ancient’ herpetofauna as there is the risk of extinction of cryptic taxa (Daugherty et al., 1990). Although Hay et al. (2010) concluded that tuatara is one taxon, taxonomic uncertainties persist in other ‘ancient’ genera such as Leiopelma (Bell, 2010; Bishop et al., 2013; Newman et al., 2013). To date, three extinct Leiopelma spp. have been described from Holocene fossils (the terrestrial L. markhami Worthy, 1987 and L. auroraensis Worthy, 1987, and the semi-aquatic L. waitomoensis Worthy, 1987), along with three extant taxa (the semi-aquatic: L. hochstetteri Fitzinger, 1861 and the terrestrial: L. archeyi Turbott, 1942 and L. hamiltoni McCulloch, 1919). The validity of a fourth extant taxon (L. pakeka Bell, Daugherty & Hay, 1998) is debated (Holyoake et al., 2001; Thurlow, 2015). Worthy (1987a, b) described the palaeoecology and osteology of Leiopelma, but a large proportion of postcranial elements, which differed primarily in size, were taxonomically indeterminate. Cranial elements (e.g. maxillae) show prominent structural differences (Worthy, 1987b) and may distinguish between most Leiopelma taxa. Indeed, Clarke (2007) suggested that the maxilla is the only reliable cranial element for osteological analysis. Within Leiopelma, there are two groups of taxa supported by osteological characters. Leiopelma auroraensis and L. markhami are morphologically similar to L. hochstetteri, whereas L. waitomoensis is similar to the extant terrestrial taxa, which are osteologically indistinguishable (Worthy, 1987b; Worthy et al., 2013). Leiopelma hochstetteri is the only leiopelmatid frog to exhibit sexual dimorphism osteologically, in that the males have more pronounced lateral ridges on the humeri compared with females (Bell, 1978; Worthy, 1987b; Germano, Cree & Bishop, 2011). Of the extinct Holocene taxa, L. waitomoensis and L. markhami were the most geographically widespread; L. waitomoensis was restricted to the North Island, but L. markhami was distributed in both main islands (Figs 1a, b). Leiopelma auroraensis is known only from the holotype (NMNZ S.23413) collected from southern New Zealand (Te Anau, Fiordland) (Worthy, 1987a; Fig. 1a). Worthy (1998) described an incomplete femur from southern New Zealand (Forest Hill, Southland), which was intermediate between L. auroraensis and L. markhami, thus its identity remains unclear (Fig. 1f). Worthy (1987a) suggested that L. auroraensis may just represent the southern extent of L. markhami. The fossil distribution of the extant L. hamiltoni and L. pakeka remains uncertain because these taxa were previously considered as the single species L. hamiltoni (until 1998 when L. pakeka was described) and no morphological characters separate these taxa (Bell et al., 1998a; Bell, 2010; Worthy et al., 2013). Indeterminate Leiopelma fossils indicate a wider pre-human distribution of leiopelmatids (Fig. 1f). Fossils identified by Worthy (1987a, b) as L. hamiltoni and L. hochstetteri have been discovered across most of the North Island and western regions of the South Island (Figs 1c, d), yet fossils identified as L. archeyi have only been found in the northern North Island (Fig. 1e). Late Quaternary fossils of Leiopelma have not been recorded in the eastern South Island, possibly because of dry climates during the Pleistocene and Holocene (Worthy, 1987a). Figure 1. View largeDownload slide Quaternary fossil, historical and current distribution of leiopelmatid frogs in New Zealand: (a) Leiopelma markhami and L. auroraensis, (b) L. waitomoensis, (c) L. hochstetteri, (d) L. pakeka and L. hamiltoni, (e) L. archeyi and (f) indeterminate spp. Figure 1. View largeDownload slide Quaternary fossil, historical and current distribution of leiopelmatid frogs in New Zealand: (a) Leiopelma markhami and L. auroraensis, (b) L. waitomoensis, (c) L. hochstetteri, (d) L. pakeka and L. hamiltoni, (e) L. archeyi and (f) indeterminate spp. Habitat loss and predation from the introduced Polynesian rat (Rattus exulans) after human arrival in the late 13th century are the primary factors identified in the declines and extinctions seen in leiopelmatids (Bell, 1985; Worthy, 1987a; Towns & Daugherty, 1994). All extant Leiopelma taxa are either at risk or nationally threatened (Newman et al., 2013) and rank within the top 100 Evolutionarily Distinct and Globally Endangered amphibians (Zoological Society of London, 2012). Leiopelma spp. are therefore of global conservation concern. Like most New Zealand’s endemic taxa, contemporary Leiopelma populations are highly fragmented and isolated. Leiopelma hochstetteri is the most widespread extant taxon, occurring as phylogeographically structured isolated populations in the northern North Island (Fig. 1c). Leiopelma archeyi occurs naturally in the northern North Island (Thames-Coromandel and Waitomo Districts) (Bishop et al., 2013; Fig. 1e). The two rarest taxa, L. hamiltoni and L. pakeka, are restricted to single insular island populations located off the northern South Island. Based on bathymetry and late Quaternary sea-level reconstructions, these populations have probably been isolated for the past 14000 years (Gibb, 1986; Lewis, Carter & Davey, 1994; Bishop et al., 2013; Thurlow, 2015) (Fig. 1d). An osteological and morphometric reassessment of modern extant and extinct leiopelmatids is warranted given concerns over possible cryptic diversity (Fouquet et al., 2010; Newman et al., 2013). Since Worthy’s (1987a, b) analyses, major additions of modern skeletal material and Pleistocene/Holocene fossils of Leiopelma to museum collections have occurred. Our study re-examined skeletal and external morphology to re-assess morphological differentiation, species limits and the taxonomy of leiopelmatids to provide new insights into the paleogeographic and evolutionary history of this globally significant and threatened anuran lineage. MATERIAL AND METHODS Data collection Skeletal material of Leiopelma was sourced from the University of Otago Department of Zoology (UOZ), Waitomo Caves Museum (WO), Museum of New Zealand Te Papa Tongarewa (NMNZ), Canterbury Museum (CM), Otago Museum (OM) and Landcare Research (LR). Modern material was originally obtained from the Thames-Coromandel region and nearshore islands off the northern South Island. Pleistocene/Holocene fossils of Leiopelma covered the known geographical and temporal range of this genus [regions: (North Island: -) Northland, Waitomo District, Hawkes Bay, Wairarapa (South Island: -) north-west Nelson, West Coast, Te Anau, Southland and Central Otago; age: c. 300–29000 BP] (Worthy, 1987a; Worthy & Holdaway, 1993, 1994; Worthy et al., 2002; Worthy & Roscoe, 2003). Measurements were taken using Vernier callipers to the nearest 0.1 mm. All late Quaternary specimens measured were osteologically mature (i.e. adult). Only unique elements (e.g. only the left femur) from sites/layers were measured to ensure independency of the data. When we could not confidently allocate fossil remains to a particular taxon (e.g. size overlap in L. hamiltoni and L. hochstetteri, Worthy & Holdaway, 1993), we conservatively grouped material from these taxa as ‘indeterminate’. Following Worthy (1987b), we recorded four measurements (proximal width, mid-shaft width, distal width and total length) for each of the five largest postcranial limb bones (humerus, radioulna, femur, tibiofibula and tibiale-fibulare). Only bones where accurate length measurements could be obtained were included in the analyses. However, we did include specimens where ossification of the capituli was absent as, for example, in L. waitomoensis, there is a lack of ossification of the cartilages at either end of the tibiofibula and humerus (Worthy, 1987b). Therefore, any length measurements for relevant specimens will underestimate the equivalent measurements made in other material where the capituli is present. To account for damage caused to the only available modern tibiofibulae of L. hamiltoni (NMNZ AM.292), we performed permutation mean matching for two missing values of total length and proximal width using the ‘mice’ package (van Buuren & Groothuis-Oudshoorn, 2011) in R v. 3.0.2 (R Development Core Team, 2013). To test for morphological variation in maxillae, we used a TG-3 digital camera (Olympus, USA) to photograph lingual views of all available complete modern and fossil maxillae [including specimens from the introduced Australian Ranoidea raniformis (sensuDubois & Frétey, 2016) and Litoria ewingii given that such taxa are sometimes found in fossil deposits (Worthy, 1987b) and some maxilla of Ranoidea spp. have previously been misidentified as those of Leiopelma spp., e.g. CM AM.78]. Digital photographs were downloaded into R, and using the package ‘geomorph’ (Adams & Otárola-Castillo, 2013; Adams et al., 2017), 18 spatial ‘landmarks’ (positions on the maxilla) for each specimen were selected (Supporting Information, Fig. S1). Collection details and measurements of all skeletal material are available in Supporting Information. External morphometric data of extant frogs, which consisted of 19 measurements, were sourced from Bell (1978, 1994). Statistical analyses All analyses on raw data were performed in R v. 3.0.2 (R Development Core Team, 2013). For identifiable specimens that consisted of complete sets of associated morphological elements, we performed a multivariate discriminant function analysis (DFA) on the actual (non-transformed) measurements. We assessed the data on postcranial limb element sizes using a Bayesian information criterion (BIC) model-based clustering and classification approach within the package ‘mclust’ (Fraley et al., 2012). For the element(s) with the most discrete and highest number of clusters, we compared length (size) and proximal/distal width/length ratios (shape) between taxa using non-parametric and univariate Mann–Whitney U (MWU) tests. These tests were not performed for L. auroraensis and L. hamiltoni because for each of these species only one skeleton exists in collections. We tested for inter-specific diagnosability using multivariate principal component analysis (PCA) and DFA on the actual (non-transformed) measurements to determine levels of differentiation and variance explained. We subsequently tested for intra-specific variation using the same three analyses (MWU, DFA and PCA), but only for L. markhami and L. waitomoensis as these taxa were the most widespread prehistorically (Worthy, 1987a, b). Leiopelma auroraensis was included in the PCA for L. markhami to test whether this taxon is the southern extent of a cline that follows Bergmann’s rule. To test for shape variation in maxillae, coordinates of left maxillae were transformed by rotating values on the y-axis so that they aligned with those represented by right maxillae for comparison. Morphological ‘landmark’ coordinates were standardized for size using a generalized Procrustes analysis (see Rohlf & Slice, 1990) in the package ‘geomorph’ (Adams & Otárola-Castillo, 2013; Adams et al., 2017). Using these standardized coordinates, inter- and intra-specific variations in the osteological ‘landmarks’ were assessed using multivariate analysis of variance (MANOVA) with an interaction between species and collection locality as fixed effects. The MANOVA involved 1000 iterations and was followed by post hoc tests. For external morphology, we performed multivariate DFA and PCA comparisons on actual (non-transformed) measurements to test for phenotypic differentiation between extant taxa. Statistical significance was taken at the P < 0.05 level. RESULTS Associated postcranial skeletons For associated postcranial skeletons, 100% of the individual specimens examined were ‘correctly’ assigned to their respective taxon in the DFA analysis (extinct taxa: L. auroraensis, n = 1; L. markhami, n = 10; L. waitomoensis, n = 1; extant taxa: L. archeyi, n = 10; L. pakeka, n = 31; L. hochstetteri, n = 4; L. hamiltoni, n = 1). Key features were that the femur and tibiofibula in L. auroraensis were of equal length, while the tibiofibula in L. waitomoensis was noticeably longer than the femur. Humerus length, in proportion to the tibiofibula, was longer in L. auroraensis than in other taxa. Isolated postcranial elements BIC model-based clustering of isolated postcranial elements estimated three size clusters for tibiofibulae (BIC = −1705.4, d.f. = 26, n = 540), femora (BIC = −1345.4, d.f. = 26, n = 445), humeri (BIC = −1608.3, d.f. = 26, n = 359) and radioulnae (BIC = −707.0, d.f. = 26, n = 258), whereas only two clusters were estimated for the tibiale-fibulare (BIC = −1098.2, d.f. = 20, n = 189). Of the elements with three clusters, tibiofibula was the most morphologically distinct (Supporting Information, Fig. S2); hence, additional analyses were performed only for this element. However, the degree of clustering in tibiofibulae may be due to large sample size. No signals of sexual dimorphism were seen for any element as the morphometric data followed a normal, not bimodal, distribution. The tibiofibulae lengths for extinct taxa [L. markhami: 15.5–30.4 mm (n = 190); L. auroraensis: 18.3 mm (n = 1); L. waitomoensis: 25.1–42.1 mm (n = 50)] were significantly larger than those of extant taxa, which generally overlapped in size, including taxa considered to be indeterminate [L. archeyi: 9.7–14.0 mm (n = 12); L. hamiltoni: 14.9 mm (n = 1); L. pakeka: 11.2–18.5 mm (n = 32); L. hochstetteri: 14.9–17.5 mm (n = 4); indeterminate: 10.1–21.7 mm (n = 250)] (Supporting Information, Table S1). Leiopelma archeyi had significantly smaller tibiofibulae compared with other taxa, but proximal/distal width/length ratios of L. archeyi (0.12–0.16) were relatively higher than L. hochstetteri (0.128–0.135). Indeterminate and proximal/distal width/length ratios of L. pakeka were also significantly broader compared with L. hochstetteri (0.12–0.17 and 0.11–0.16, respectively) and similar in breadth to L. hamiltoni (0.16). Proximal/distal width/length ratios of L. markhami and L. auroraensis were broader than other taxa (0.13–0.21 and 0.21, respectively), while tibiofibulae of L. waitomoensis were the most slender (0.11–0.15) (Supporting Information, Table S1). Inter-specific diagnosability of tibiofibulae was high for all extinct taxa: 97.9% from L. markhami, 98% from L. waitomoensis and 100% from L. auroraensis were assigned to their respective taxon. Only 0.5 and 1.6% of tibiofibulae from L. markhami were assigned as belonging to L. auroraensis and L. pakeka groups, respectively, and 2% of tibiofibulae from L. waitomoensis as L. markhami. For extant taxa, 58.3% of tibiofibulae from L. archeyi and 96.9% from L. pakeka were ‘correctly’ allocated. 41.7% of tibiofibulae from L. archeyi, 100% from L. hochstetteri and 100% from L. hamiltoni were classified as L. pakeka. Only 3.1% of tibiofibulae from L. pakeka were classified as L. archeyi. All tibiofibulae features measured loaded positively and equally on PC1 (axis scores: distal width = 0.51; mid-shaft width = 0.50; proximal width = 0.50; length = 0.49), which explained 94.9% of the variance. Positive values on the PC1 axis represent large tibiofibulae size (e.g. tibiofibulae of L. waitomoensis are generally longer than tibiofibulae of L. markhami, Fig. 2). Tibiofibulae length also loaded negatively on PC2 (−0.83), which accounted for 3.2% of the remaining variance. Negative values on the PC2 axis represent a decrease in overall width/length ratio, indicating tibiofibulae are gracile, whereas positive values indicate robust tibiofibulae (e.g. tibiofibulae of L. waitomoensis are gracile, whereas tibiofibulae of L. markhami are robust, Fig. 2). Figure 2. View largeDownload slide Principal component analysis of fossil and modern tibiofibulae (n = 540) measurements in Leiopelma frogs showing inter-specific size (PC1: small vs. large) and shape (PC2: robust vs. gracile) variation. Taxonomic classifications are as follows: L. auroraensis (solid black circle), L. markhami (hollow triangles), L. waitomoensis (hollow squares), indeterminate spp. (hollow grey circles), L. archeyi (solid squares), L. pakeka (plus signs), L. hamiltoni (solid red circle) and L. hochstetteri (solid triangles). For taxa with three or more data points, different coloured convex hull polygons are shown to highlight separate taxon groups. Figure 2. View largeDownload slide Principal component analysis of fossil and modern tibiofibulae (n = 540) measurements in Leiopelma frogs showing inter-specific size (PC1: small vs. large) and shape (PC2: robust vs. gracile) variation. Taxonomic classifications are as follows: L. auroraensis (solid black circle), L. markhami (hollow triangles), L. waitomoensis (hollow squares), indeterminate spp. (hollow grey circles), L. archeyi (solid squares), L. pakeka (plus signs), L. hamiltoni (solid red circle) and L. hochstetteri (solid triangles). For taxa with three or more data points, different coloured convex hull polygons are shown to highlight separate taxon groups. The tibiofibulae regional lengths of L. markhami, the most widespread of the extinct taxa, were mostly significantly different from each other in the North Island [Northland: 20.6 mm (n = 1); Waitomo: 16.9–22.6 mm (n = 24); Hawkes Bay: 19.1–21.2 mm (n = 6); Coonoor, Manawatu: 20.1 mm (n = 1); Wairarapa: 20.9–22.4 mm (n = 7)] and in the South Island [north-west Nelson: 18.0–30.4 mm (n = 133); West Coast: 15.5–25.1 mm (n = 18)] (Supporting Information, Table S2; Fig. 1a). In contrast, proximal/distal width/length ratios were generally similar throughout (Northland: 0.16; Waitomo: 0.15–0.19; Hawkes Bay: 0.15–0.17; Coonoor, Manawatu: 0.19; Wairarapa: 0.15–0.18; north-west Nelson: 0.13–0.21; West Coast: 0.13–0.18), but tibiofibulae of L. markhami from the South Island were significantly broader than two North Island locations (Supporting Information, Table S2). Tibiofibulae of the widespread North Island endemic L. waitomoensis were significantly smaller in Northland than those from Waitomo [25.1–28.6 mm (n = 4) and 25.5–42.1 mm (n = 40), respectively], whereas tibiofibulae from Hawkes Bay were significantly larger [36.9–40.5 mm (n = 3)]. Coonoor specimens were similar in length to those from Hawkes Bay [35.4–40.4 mm (n = 2)] (Supporting Information, Table S3; Fig. 1b). Proximal/distal width/length ratios were the same across regions (range = 0.11–0.15); thus, tibiofibulae of L. waitomoensis mainly differed in size, not shape (Supporting Information, Table S3). Intra-specific diagnosability of L. markhami and L. waitomoensis was high for some regions. The proportion of tibiofibulae from L. markhami correctly assigned to north-west Nelson, South Island, was 96.2%. For L. waitomoensis, 66.7 and 97.5% of tibiofibulae were correctly assigned to the North Island Hawkes Bay and Waitomo regions, respectively. As above, all tibiofibulae features measured loaded positively and equally on PC1 for both L. markhami (axis scores: distal width = 0.52; mid-shaft width = 0.47; proximal width = 0.52; length = 0.47) and L. waitomoensis (axis scores: distal width = 0.50; mid-shaft width = 0.49; proximal width = 0.50; length = 0.50). PC1 explained 82.0 and 94.9% of the variance for L. markhami and L. waitomoensis, respectively. Large tibiofibulae size was represented by positive values on the PC1 axis for L. markhami (Fig. 3) and L. waitomoensis (Fig. 4). The highest loadings on PC2 were length (−0.71) and mid-shaft width (0.70) for L. markhami, whereas only mid-shaft width (−0.85) loaded highly on PC2 for L. waitomoensis. PC2 explained 10.0 and 2.5% of the remaining variance for L. markhami and L. waitomoensis, respectively. PC2 again refers to the level of robustness, which indicates some intra-specific differentiation in tibiofibulae shape for both taxa (Figs 3, 4). The tibiofibula of L. auroraensis grouped with the most robust tibiofibulae of L. markhami from north-west Nelson, South Island (Fig. 3). Figure 3. View largeDownload slide Principal component analysis of fossil tibiofibulae (n = 191) measurements in the extinct Leiopelma markhami and L. auroraensis (solid black triangle) showing intra- and inter-specific size [PC1: small (negative values) vs. large (positive values)] and shape [PC2: gracile (negative values) vs. robust (positive values)] variation. North Island fossil locations are as follows: Northland (hollow triangles), Waitomo (hollow circles), Hawkes Bay (+ symbols), Coonoor (solid square) and Wairarapa (x symbols). South Island fossil locations are north-west Nelson (abbreviated as NW Nelson, solid grey circles) and West Coast (solid black circles). For regions with three or more data points, different coloured convex hull polygons are shown to highlight separate geographical groups. Figure 3. View largeDownload slide Principal component analysis of fossil tibiofibulae (n = 191) measurements in the extinct Leiopelma markhami and L. auroraensis (solid black triangle) showing intra- and inter-specific size [PC1: small (negative values) vs. large (positive values)] and shape [PC2: gracile (negative values) vs. robust (positive values)] variation. North Island fossil locations are as follows: Northland (hollow triangles), Waitomo (hollow circles), Hawkes Bay (+ symbols), Coonoor (solid square) and Wairarapa (x symbols). South Island fossil locations are north-west Nelson (abbreviated as NW Nelson, solid grey circles) and West Coast (solid black circles). For regions with three or more data points, different coloured convex hull polygons are shown to highlight separate geographical groups. Figure 4. View largeDownload slide Principal component analysis of fossil tibiofibulae (n = 49) measurements in the extinct Leiopelma waitomoensis, showing intra-specific size [PC1: small (negative values) vs. large (positive values)] and shape [PC2: gracile (negative values) vs. robust (positive values)] variation. North Island fossil locations are as follows: Northland (hollow triangles), Waitomo (hollow circles), Hawkes Bay (+ symbols) and Coonoor (solid squares). For regions with three or more data points, different coloured convex hull polygons are shown to highlight separate geographical groups. Figure 4. View largeDownload slide Principal component analysis of fossil tibiofibulae (n = 49) measurements in the extinct Leiopelma waitomoensis, showing intra-specific size [PC1: small (negative values) vs. large (positive values)] and shape [PC2: gracile (negative values) vs. robust (positive values)] variation. North Island fossil locations are as follows: Northland (hollow triangles), Waitomo (hollow circles), Hawkes Bay (+ symbols) and Coonoor (solid squares). For regions with three or more data points, different coloured convex hull polygons are shown to highlight separate geographical groups. Maxillae Osteological shape data of Leiopelma maxillae (Fig. 5) were collected from 161 complete specimens [L. auroraensis, n = 1; L. markhami, n = 84 (West Coast, n = 4; north-west Nelson, n = 72; Wairarapa, n = 3; Waitomo, n = 5); L. waitomoensis, n = 9 (Northland, n = 1; Waitomo, n = 8); L. archeyi, n = 9; L. pakeka, n = 7; L. hochstetteri, n = 3; L. hamiltoni, n = 1; indeterminate, n = 47 (West Coast, n = 4; north-west Nelson, n = 37; Waitomo, n = 5; Northland, n = 1)] and from seven specimens of Ranoidea/Litoria spp. Maxillae morphology of Leiopelma consisted of four main character states: (1) those that had a broad pars facialis and un-notched anterior; (2) those with a deep notch anterior to a narrow preorbital process, (3) those with a subtle preorbital process, elongated anterior section and fairly broad overall; and (4) those with a shallow notch, slightly wide preorbital process and an elongated tip of the posterior process (Fig. 5). The latter two characters were additional to those identified by Worthy (1987b). Maxillae of Ranoidea/Litoria spp. were similar to those from the extinct L. waitomoensis, in that Ranoidea/Litoria spp. had an elongated anterior section and the presence of a small preorbital process. However, in comparison with L. waitomoensis, maxillae of Ranoidea/Litoria spp. were not as broad, notches were distinctive and the interior curvature of the dorsal section was pronounced. The maxilla of L. auroraensis was similar to L. markhami, which typically had a prominent preorbital process and deep anterior notch. The main differences between these two extinct taxa were that the maxilla of L. auroraensis was more compacted posteriorly and that the preorbital process was more vertical with a shallower notch. Maxillae of extant Leiopelma taxa had no notch but a broad pars facialis or had a shallow notch or preorbital process (Fig. 5). Unlike extinct Leiopelma, the osteology of extant Leiopelma maxillae generally overlapped despite some subtle differences (Supporting Information, Table S4). For example, the posterior slope of the preorbital process was slightly shallower in L. archeyi than in other extant taxa. Moreover, the preorbital process was more pronounced in L. pakeka and L. hamiltoni compared with L. archeyi and L. hochstetteri. Osteology of indeterminate maxillae overlapped with extant and extinct taxa (Supporting Information, Table S4). There were significant differences in maxillae osteology between species (d.f. = 8, F = 3.49, P = 0.001) but not between regions (d.f. = 11, F = 0.50, P = 0.08). Intra-specific variation in maxillae was geographically conservative (d.f. = 88, F = 0.03, P = 0.88). The amount of variance explained by PC1 and PC2 was 26.4 and 17.3%, respectively. Figure 5. View largeDownload slide Principal component analysis and deformation mesh grids showing the variation in maxillae shape (n = 168) in New Zealand Leiopelma, Australian Ranoidea and Australian Litoria frogs. Taxonomic classifications are as follows: L. auroraensis (lime green), L. markhami (light blue), L. waitomoensis (purple), L. archeyi (orange), L. pakeka (dark blue), L. hochstetteri (cyan), L. hamiltoni (turquoise), indeterminate Leiopelma spp. (red) and Ranoidea/Litoria spp. (pink). Figure 5. View largeDownload slide Principal component analysis and deformation mesh grids showing the variation in maxillae shape (n = 168) in New Zealand Leiopelma, Australian Ranoidea and Australian Litoria frogs. Taxonomic classifications are as follows: L. auroraensis (lime green), L. markhami (light blue), L. waitomoensis (purple), L. archeyi (orange), L. pakeka (dark blue), L. hochstetteri (cyan), L. hamiltoni (turquoise), indeterminate Leiopelma spp. (red) and Ranoidea/Litoria spp. (pink). External morphology Discrimination of extant Leiopelma taxa was high in the re-analysis of the 19 external morphometric measurements. When all four extant taxa were included, 100% of L. archeyi (n = 33), L. pakeka (n = 27) and L. hochstetteri (n = 16) individuals were ‘correctly’ assigned. 81.8% of L. hamiltoni (n = 11) individuals were assigned to this taxon, while the rest were classified as L. pakeka. When only L. pakeka and L. hamiltoni were included in the analysis, all individuals were assigned to their respective taxon. Likewise, frogs from two geographically isolated populations of L. archeyi in the North Island [Thames-Coromandel District (n = 23) and Whareorino in Waitomo District (n = 10)] were all ‘correctly’ assigned when analysed separately from other taxa. In contrast, the PCA yielded major overlap for isolated North Island populations of L. hochstetteri [East Cape (n = 3) and Coromandel (n = 13)], the North Island populations of L. archeyi and for L. pakeka and L. hamiltoni (Supporting Information, Fig. S3). All external features loaded positively on PC1 (data not shown), while radioulna and humerus widths loaded positively on PC2 (0.42 and 0.45, respectively), and the length of the third fore-finger and nostril-eye length had equal negative loadings (−0.35). PC2 loadings indicate that terrestrial Leiopelma have less robust forelimbs than L. hochstetteri, although the former taxa have elongated snouts and fore-fingers (Supporting Information, Fig. S3). PC1 and PC2 explained 79.6 and 9.9% of the variance, respectively. DISCUSSION Ecological correlates of morphology Associated postcranial limb skeletons of known taxa were all ‘correctly’ allocated to their taxon. In particular, L. waitomoensis had lower humerus/tibiofibula and femur/tibiofibula ratios than in other Leiopelma taxa, while L. auroraensis and L. markhami had the highest. Humerus/tibiofibula and femur/tibiofibula length ratios are negatively correlated with jumping ability (Zug, 1972) and, as such, indicate that L. waitomoensis was a strong jumper, unlike L. markhami and L. auroraensis, which are suggested to have primarily walked (Worthy, 1987a). Moreover, physical characteristics of fossil sites suggest that L. waitomoensis was a stream-dweller as its remains are predominantly found in cave deposits associated with streams (Worthy, 1987a). In contrast, L. markhami are predominantly found in pitfall deposits, which disproportionally comprise terrestrial fauna (Worthy, 1987a). Field observations of behaviour in the semi-aquatic L. hochstetteri, and terrestrial L. archeyi, L. pakeka and L. hamiltoni, attest to this association between locomotor repertoire and the environment (Newman, 1990; Reilly et al., 2015). Inter-specific variation in isolated postcranial elements The results of our study, using c. 1000 new specimens, strongly support those of Worthy (1987a, b). Worthy (1987b) estimated snout-vent lengths (SVLs) of extinct L. markhami and L. auroraensis to be c. 50–60 mm, whereas L. waitomoensis was c. 100 mm. In contrast, extant taxa only reach up to c. 50 mm SVL, the smallest being L. archeyi, which is generally between 32- and 40-mm SVL (Worthy, 1987b). Leiopelma markhami and L. auroraensis have been described as morphologically robust, indicated by the broadness of their hindlimb bones. Leiopelma waitomoensis had relatively narrow hindlimb bones and is thus gracile in comparison (Worthy, 1987b). The distinctiveness of these size characteristics is demonstrated by the high diagnosability of tibiofibulae of extinct taxa (Worthy, 1987b; this study). However, while Worthy (1987b) assigned 98.1% of tibiofibulae from L. markhami (n = 101) and 100% of tibiofibulae from L. waitomoensis (n = 43) to their respective taxon, the holotype tibiofibula of L. auroraensis was classified as L. markhami. Worthy’s (1987b) measurements of the holotype differed from our measurements of the same specimen by c. 1 mm, which is why there were contrasting results regarding the diagnosability of L. auroraensis and L. markhami between our studies. Nevertheless, isolated postcranial elements of these taxa are similar. Overall tibiofibulae size range in extant taxa was similar to that reported by Worthy (1987b). However, Worthy (1987b) found proximal/distal widths of L. archeyi to be significantly different from all taxa except L. hochstetteri – in contrast, we found significant differences between all identifiable taxa except L. pakeka (albeit L. pakeka being described as L. hamiltoni in Worthy, 1987b) (Supporting Information, Table S1). Leiopelma archeyi is the only extant taxon that differentiates by its small tibiofibulae size. We thus conclude that size-related differences, but minimal genetic distance between L. archeyi, L. pakeka and L. hamiltoni compared with L. hochstetteri (see Holyoake et al., 2001; Carr et al., 2015), suggest that L. archeyi, L. pakeka and L. hamiltoni represent a single species that has become fragmented into several isolated populations. The overlap of size and lack of diagnosable osteological features between extant taxa and indeterminate fossils support Worthy’s (1987b) contention that such fossils could be either L. hamiltoni or L. hochstetteri [L. archeyi fossils were not described at the time of Worthy’s (1987b) study]. In our study, tibiofibulae from L. hamiltoni and L. pakeka were similar to specimens of indeterminate tibiofibulae, but broader than L. hochstetteri, which implies that indeterminate fossils are more likely to comprise the former taxa. Skeletal morphology of isolated tibiofibulae (and postcranial limb elements in general) is therefore problematic in distinguishing between the Leiopelma taxa. There are some osteological differences in the forelimbs of L. hochstetteri and L. hamiltoni [e.g. lateral and medial crests of the humeri are present in the former and not in the latter (Worthy, 1987b; Worthy et al., 2013)], but from what we observed these characteristics are not always distinct. Temperature-related effects on postcranial elements The two most widespread extinct taxa (Fig. 1a, b), L. waitomoensis and L. markhami, obey Bergmann’s rule in that they show a latitudinal cline in tibiofibulae size with individuals from southern populations larger than those from northern populations (Worthy, 1987a, b). However, tibiofibulae of L. markhami from the north-west South Island were relatively larger than those from the South Island’s West Coast (this was not reported by Worthy, 1987a, b as he did not measure fossils from the latter location). Our observation suggests that an interaction between temperatures associated with altitude and latitude is the primary factor responsible for these size patterns (Berven, Gill & Smith-Gill, 1979). We consider the effect of temporal variation (i.e. the deposition of fossils during warm and cool periods) on Leiopelma size to be minimal as radio-carbon dating of associated avifaunal bones indicates most leiopelmatid bones to be of Holocene age from time-averaged fossil deposits (Worthy, 1987a; Worthy & Holdaway, 1993, 1994; Worthy et al., 2002; Worthy & Roscoe, 2003). Furthermore, warm microclimates driven by the subtropical oceanic current (Alloway et al., 2007) and low elevation may have led to reduced body sizes of L. markhami (and perhaps other sympatric Leiopelma) along the South Island’s West Coast. Non-temperate and latitude-related correlations of postcranial elements Variation in tibiofibulae shape of L. markhami was noticeably different between certain regions (Fig. 3; Supporting Information, Table S2). Tibiofibulae from the north-west South Island were generally more robust than those from the South Island’s West Coast and North Island. Some L. markhami tibiofibulae from the north-west South Island were remarkably similar in appearance to the holotype of L. auroraensis from the southern South Island (Te Anau). This extent of differentiation in tibiofibulae shape within L. markhami was not observed in L. waitomoensis (Fig. 4; Supporting Information, Table S3) although there was high diagnosability of tibiofibulae from the eastern and northern North Island. Overall, these intra-specific size and shape differences suggest that cryptic taxa may be present within these widespread extinct species. Diagnosability of maxillae Worthy (1987b) described two osteological character states in the maxilla: taxa that have a notch anterior to the preorbital process (L. hochstetteri, L. markhami and L. auroraensis) and taxa that do not (L. waitomoensis, L. archeyi, L. hamiltoni and L. pakeka). We identified two additional character states: (1) those that were broad overall, with a small preorbital process and elongated anterior, and (2) those with a shallow notch, expanded preorbital process and an elongated posterior process (Fig. 5). Significant inter-specific differences in maxilla shape were mainly limited to extinct taxa, yet there were some differences between extant taxa (Supporting Information, Table S4). In L. archeyi, the preorbital process was relatively reduced and the maxillae were more compact compared with L. pakeka. Despite these taxa being osteologically identical except for size (Worthy, 1987b), L. archeyi is osteologically neotenic, meaning that the level of ossification is lower compared with all other Leiopelma (Stephenson, 1951; Worthy et al., 2013). Hence, this neotenic condition probably explains the structural differences that we observed. Maxillae osteology did not vary intra-specifically across regions, which means that, based on Worthy’s (1987a) and our study, cranial elements do not change size with latitude and altitude cf. postcranial elements. In introduced Australian Ranoidea and Litoria frogs, maxilla specimens were distinct from most Leiopelma. Maxillae of Ranoidea and Litoria spp. were relatively similar to L. waitomoensis, but shared more features with evolutionarily derived anuran taxa (e.g. Eleutherodactylus spp., Bochaton et al., 2015), such as bearing many teeth and a poorly developed pars facialis. External morphology Differences in external morphological characters among currently recognized Leiopelma species and isolated populations have been discussed and debated (see Stephenson & Stephenson, 1957; Bell, 1978, 1994; Bell et al., 1998a; Bell, Daugherty & Hitchmough, 1998b). For instance, Bell (1994) found no morphometric differences between L. hamiltoni and L. pakeka when comparing all extant taxa. Later multivariate analyses of external morphometric characters that focused on comparisons of just two populations, however, concluded that there were morphometric differences between these taxa and between L. archeyi populations in the Coromandel and Waitomo Districts (Bell et al., 1998a, b). The extant L. hochstetteri is clearly robust in relation to other extant taxa and there is no difference externally between all terrestrial taxa apart from size (Bell, 1978, 1994; Newman, 1990; Bell et al., 1998a, b). While there are some statistically significant phenotypic differences between L. hamiltoni and L. pakeka (see Bell et al., 1998a), such differences are not considered biologically relevant (Newman, 1990; Bell et al., 1998a) – akin to isolated populations of L. archeyi for instance (Bell, 1994; Bell et al., 1998b). Considering there is no genetic, ecological, skeletal or external morphometric evidence to support the distinction between L. hamiltoni and L. pakeka, we therefore suggest the synonymization of L. pakeka Bell, Daugherty & Hay, 1998 and L. hamiltoni McCulloch, 1919, with L. hamiltoni retaining taxonomic priority. Glacier-driven evolution and human-driven extinctions Glaciations during the Pleistocene promoted speciation in alpine taxa (e.g. Aves, Dinornithiformes) by severing their continuous distributions transverse to the Southern Alps, South Island (Bunce et al., 2009; Wallis, Waters, Upton & Craw, 2016). Leiopelma spp. potentially would have been similarly affected. Given the limited number of late Quaternary Leiopelma fossils from the southern South Island, mainly due to the rarity of suitable fossil sites in the region, the distinction between northern and southern Leiopelma assemblages on either side of the central transverse zone of the Southern Alps is unclear, but the presumption that L. auroraensis is a local derivative of L. markhami may support this glacial–speciation hypothesis (Worthy, 1987a). Leiopelma auroraensis should, therefore, be retained as a distinct taxon, as demonstrated by our multivariate analyses and Worthy’s (1987b) osteological descriptions. Likewise, glaciation may have driven the morphological differentiation observed between L. markhami populations in the western and north-western South Island, whereas the geographical separation of the North and South Island L. markhami, L. hamiltoni and L. hochstetteri populations may have occurred more recently when post-glacial sea-level rise fractured land-locked islands of the South Island’s Marlborough Sounds at least 14000 BP (Gibb, 1986; Lewis et al., 1994). The nearshore island populations of L. hamiltoni also became isolated at this time (Thurlow, 2015). Human arrival in the late 13th century (Wilmshurst et al., 2008) led to the extirpation and extinction of many Leiopelma populations (and other herpetofauna), primarily due to habitat loss and predation by introduced mammals (e.g. kiore rat, R. exulans) (Bell, 1985; Worthy, 1987a; Daugherty et al., 1990; Towns & Daugherty, 1994). Large herpetofauna, such as L. waitomoensis, were among the first to disappear, while the remaining smaller species became restricted to isolated, relict distributions (Worthy, 1987a). Leiopelma markhami and L. hamiltoni, which were sympatric with L. hochstetteri (Fig. 1a, c, d), persisted in the north-west South Island until c. 400 BP–southern South Island (Te Anau) fossils of L. markhami come from even younger deposits (c. 300 BP) (Worthy, 1987a; Worthy & Holdaway, 1993, 1994; Worthy & Roscoe, 2003). The extirpation of leiopelmatids from the main islands is therefore recent and human induced. Indeed, Green, Zeyl & Sharbel (1993) and Green (1994) concluded that extirpations of North Island L. hochstetteri populations were accentuated after human settlement. Conservation implications Species translocations are a key tool in managing threatened herpetofauna (Germano & Bishop, 2009), but these are dependent on recognizing species limits and taxonomic units, for which there is high uncertainty in Leiopelma (Newman et al., 2013). Conservation palaeontology and the utilization of fossil data within a contemporary context can, however, provide insights for managing extant taxa (Bochaton et al., 2015). Based on the morphology of fossil and modern skeletal material, our study supports the possibility of extinctions of cryptic taxa in Leiopelma and potential major taxonomic changes, but to test these hypotheses, future research will require a genetic approach using modern and ancient DNA from late Quaternary fossils. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Figure S1. Lingual view of a right Holocene fossil maxilla of Leiopelma markhami (NMNZ S.23121), highlighting the main osteological features and the 18 spatial ‘landmarks’ (positions on the maxilla, solid red circles) that were selected to establish coordinates that described the shape. Scale bar 1 mm (red dotted line). Figure S2. Bayesian information criterion (BIC) model-based clustering of five leiopelmatid frog limb elements: (a) humeri (n &#x003D; 359), (b) radioulnae (n &#x003D; 258), (c) femora (n &#x003D; 445), (d) tibiofibulae (n &#x003D; 540), (e) tibiale-fibulare (n &#x003D; 189), and four respective measurements [distal width, mid-shaft width, proximal width and length (in mm)]. Different colours and symbols represent unique clusters identified by the BIC model-based cluster estimation. Abbreviations are as follows: H &#x003D; humeri, R &#x003D; radioulnae, F &#x003D; femora, T &#x003D; tibiofibulae, Tf &#x003D; tibiale-fibulare, d &#x003D; distal width, s &#x003D; mid-shaft width, p &#x003D; proximal width and l &#x003D; length. For example, ‘Hd’ refers to the distal width of humeri. Figure S3. Principal component analysis of 19 external measurements in extant Leiopelma frogs (see Bell, 1978, 1994 for details) showing inter- and intra-specific shape, and size variation. Different coloured convex hull polygons are shown to highlight separate groups, which are as follows: L. hochstetteri [Coromandel (abbreviated as CORO, n &#x003D; 13), solid circles], L. hochstetteri [East Cape (abbreviated as ECAP, n &#x003D; 3), hollow circles], L. archeyi [CORO (n &#x003D; 23), plus signs], L. archeyi [Whareorino (abbreviated as WHAR, n &#x003D; 10), solid triangles], L. hamiltoni [Stephens Island (abbreviated as STEP, n &#x003D; 11), solid squares] and L. pakeka [Maud Island (abbreviated as MAUD, n &#x003D; 27), hollow squares]. Table S1. Non-parametric Mann–Whitney U test comparisons of tibiofibulae size between Leiopelma taxa. Leiopelma auroraensis and L. hamiltoni were excluded as they had sample sizes of one. Total length and distal width/length ratio comparisons are presented in the upper and lower diagonal sections, respectively. Distal and proximal widths were highly correlated (Pearson’s correlation: t &#x003D; 53.73, d.f. &#x003D; 538, P < 0.001, correlation &#x003D; 0.918), thus only distal width/length ratio comparisons are presented in the table. Mid-shaft width/length ratios were virtually identical between all taxa and were thus excluded from the table. Statistical significance was taken at the P < 0.05 level. Significant P-values are highlighted in bold. Table S2. Non-parametric Mann–Whitney U test comparisons of tibiofibulae size between Leiopelma markhami fossil locations of North Island and South Island, New Zealand. North Island regions include Waitomo, Hawkes Bay and the Wairarapa. South Island regions include north-west Nelson (abbreviated as NW Nelson) and the West Coast. Length and distal width/length ratio comparisons are presented in the upper and lower diagonal sections, respectively. Statistical significance was taken at the P < 0.05 level. Significant P-values are highlighted in bold. Table S3. Non-parametric Mann–Whitney U test comparisons of tibiofibulae size between Leiopelma waitomoensis fossil locations of North Island, New Zealand. North Island regions include Northland, Waitomo, Hawkes Bay and Coonoor. Length and distal width/length ratio comparisons are presented in the upper and lower diagonal sections, respectively. Statistical significance was taken at the P < 0.05 level. Significant P-values are highlighted in bold. Table S4. Multivariate analysis of variance comparisons of maxillae shape in New Zealand Leiopelma and Australian Ranoidea/Litoria frogs. Statistical significance was taken at the P < 0.05 level. Significant P-values are highlighted in bold. ACKNOWLEDGEMENTS We thank the editor Dr. Louise Allcock and one anonymous reviewer for providing helpful comments on the manuscript. We thank the staff of New Zealand museums (in particular Otago Museum, Canterbury Museum, Te Papa Museum and Waitomo Caves Museum) and Jamie Wood from Landcare Research for access to collections. We also thank Ana Dula for helping excavate fossil material from the Wairarapa, Travis Ingram and Scott Jarvie for providing R code and statistical advice, and Benno Kappers from the Department of Conservation for providing distribution data. L.J.E. acknowledges support from a University of Otago Doctoral Scholarship. REFERENCES Adams DC , Collyer ML , Kaliontzopoulou A , Sherratt E . 2017 . Geomorph: software for geometric morphometric analyses, R package version 3.0.5 . Available at: http://cran.r-project.org/package=geomorph (accessed 30 August 2016 ). Adams DC , Otárola-Castillo E . 2013 . Geomorph: an R package for the collection and analysis of geometric morphometric shape data . Methods in Ecology and Evolution 4 : 393 – 399 . Google Scholar CrossRef Search ADS Alloway BV , Lowe DJ , Barrell DJA , Newnham RM , Almond PC , Augustinus PC , Bertler NAN , Carter L , Litchfield NJ , McGlone MS , Shulmeister J , Vandergoes MJ , Williams PW ; NZ-INTIMATE Members . 2007 . Towards a climate event stratigraphy for New Zealand over the past 30000 years (NZ-INTIMATE project) . Journal of Quaternary Science 22 : 9 – 35 . Google Scholar CrossRef Search ADS Bell BD . 1978 . Observations on the ecology and reproduction of the New Zealand Leiopelmid frogs . Herpetologica 34 : 340 – 354 . Bell BD . 1985 . Conservation status of the endemic New Zealand frogs . In: Grigg G , Shine R , Ehmann H , eds. Biology of Australasian frogs and reptiles . New South Wales : Surrey Beatty & Sons , 449 – 458 . Bell BD . 1994 . A review of the status of New Zealand Leiopelma species (Anura: Leiopelmatidae), including a summary of demographic studies in Coromandel and on Maud Island . New Zealand Journal of Zoology 21 : 341 – 349 . Google Scholar CrossRef Search ADS Bell BD . 2010 . The threatened leiopelmatid frogs of New Zealand: natural history integrates with conservation . Herpetological Conservation and Biology 5 : 515 – 528 . Bell BD , Daugherty CH , Hay JM . 1998a . Leiopelma pakeka, n. sp. (Anura: Leiopelmatidae), a cryptic species of frog from Maud Island, New Zealand, and a reassessment of the conservation status of L. hamiltoni from Stephens Island . Journal of the Royal Society of New Zealand 28 : 39 – 54 . Google Scholar CrossRef Search ADS Bell BD , Daugherty CH , Hitchmough RA . 1998b . The taxonomic identity of a population of terrestrial Leiopelma (Anura: Leiopelmatidae) recently discovered in the northern King Country, New Zealand . New Zealand Journal of Zoology 25 : 139 – 146 . Google Scholar CrossRef Search ADS Berven KA , Gill DE , Smith-Gill SJ . 1979 . Countergradient selection in the green frog, Rana clamitans . Evolution 33 : 609 – 623 . Google Scholar CrossRef Search ADS Bishop PJ , Daglish LA , Haigh AJM , Marshall LJ , Tocher MD , McKenzie KL . 2013 . Native frog (Leiopelma species) recovery plan, 2013–2018 . Threatened Species Recovery Plan 63. Wellington : Department of Conservation . Bunce M , Worthy TH , Phillips MJ , Holdaway RN , Willerslev E , Haile J , Shapiro B , Scofield RP , Drummond A , Kamp PJJ , Cooper A . 2009 . The evolutionary history of the extinct ratite moa and New Zealand Neogene paleogeography . Proceedings of the National Academy of Sciences of the United States of America 106 : 20646 – 20651 . Google Scholar CrossRef Search ADS van Buuren S , Groothuis-Oudshoorn K . 2011 . mice: multivariate imputation by chained equations in R . Journal of Statistical Software 45 : 1 – 67 . Carr LM , McLenachan PA , Waddell PJ , Gemmell NJ , Penny D . 2015 . Analyses of the mitochondrial genome of Leiopelma hochstetteri argues against the full drowning of New Zealand . Journal of Biogeography 42 : 1066 – 1076 . Google Scholar CrossRef Search ADS Clarke BT . 2007 . Comparative morphology and amphibian taxonomy: an example from the osteology of discoglossoid frogs . In: Heatwole H , Tyler MJ , eds. Amphibian biology: systematics, Vol. 7 . New South Wales : Surrey Beatty & Sons , 2465 – 2603 . Daugherty CH , Cree A , Hay JM , Thompson MB . 1990 . Neglected taxonomy and continuing extinctions of tuatara (Sphenodon) . Nature 347 : 177 . Google Scholar CrossRef Search ADS Daugherty CH , Patterson GB , Hitchmough RA . 1994 . Taxonomic and conservation review of the New Zealand herpetofauna . New Zealand Journal of Ecology 21 : 317 – 323 . Google Scholar CrossRef Search ADS Dubois A , Frétey T . 2016 . A new nomen for a subfamily of frogs (Amphibia, Anura) . Dumerilia 6 : 17 – 23 . Fouquet A , Green DM , Waldman B , Bowsher JH , McBride KP , Gemmell NJ . 2010 . Phylogeography of Leiopelma hochstetteri reveals strong genetic structure and suggests new conservation priorities . Conservation Genetics 11 : 907 – 919 . Google Scholar CrossRef Search ADS Fraley C , Raftery AE , Murphy TB , Scrucca L . 2012 . mclust: version 4 for R: normal mixture modeling for model-based clustering, classification, and density estimation . Technical Report 597. Washington : Department of Statistics, University of Washington . Germano JM , Bishop PJ . 2009 . Suitability of amphibians and reptiles for translocation . Conservation Biology 23 : 7 – 15 . Google Scholar CrossRef Search ADS Germano JM , Cree A , Bishop PJ . 2011 . Ruling out the boys from the girls: can subtle morphological differences identify sex of the apparently monomorphic frog, Leiopelma pakeka ? New Zealand Journal of Zoology 38 : 161 – 171 . Google Scholar CrossRef Search ADS Gibb JG . 1986 . A New Zealand regional Holocene eustatic sea-level curve and its application to determination of vertical tectonic movements . Royal Society of New Zealand 24 : 377 – 395 . Green DM . 1994 . Genetic and cytogenetic diversity in Hochstetter’s frog, Leiopelma hochstetteri, and its importance for conservation management . New Zealand Journal of Zoology 21 : 417 – 424 . Google Scholar CrossRef Search ADS Green DM , Zeyl CW , Sharbel TF . 1993 . The evolution of hypervariable sex and supernumerary (B) chromosomes in the relict New Zealand frog, Leiopelma hochstetteri . Journal of Evolutionary Biology 6 : 417 – 441 . Google Scholar CrossRef Search ADS Hay JM , Daugherty CH , Cree A , Maxson LR . 2003 . Low genetic divergence obscures phylogeny among populations of Sphenodon, remnant of an ancient reptile lineage . Molecular Phylogenetics and Evolution 29 : 1 – 19 . Google Scholar CrossRef Search ADS Hay JM , Sarre SD , Lambert DM , Allendorf FW , Daugherty CH . 2010 . Genetic diversity and taxonomy: a reassessment of species designation in tuatara (Sphenodon: Reptilia) . Conservation Genetics 11 : 1063 – 1081 . Google Scholar CrossRef Search ADS Holyoake A , Waldman B , Gemmell NJ . 2001 . Determining the species status of one of the world’s rarest frogs: a conservation dilemma . Animal Conservation 4 : 29 – 35 . Google Scholar CrossRef Search ADS Kennedy M , Paterson AM , Morales JC , Parsons S , Winnington AP , Spencer HG . 1999 . The long and short of it: branch lengths and the problem of placing the New Zealand short-tailed bat, Mystacina . Molecular Phylogenetics and Evolution 13 : 405 – 416 . Google Scholar CrossRef Search ADS Lewis KB , Carter L , Davey FJ . 1994 . The opening of Cook Strait: interglacial tidal scour and aligning basins at a subduction to transform plate edge . Marine Geology 116 : 293 – 312 . Google Scholar CrossRef Search ADS Newman DG . 1990 . Activity, dispersion, and population densities of Hamilton’s frog (Leiopelma hamiltoni) on Maud and Stephens Islands, New Zealand . Herpetologica 46 : 319 – 330 . Newman DG , Bell BD , Bishop PJ , Burns R , Haigh A , Hitchmough RA . 2013 . Conservation status of New Zealand frogs, 2013 . New Zealand Threat Classification Series 5. Wellington : Department of Conservation . R Development Core Team . 2013 . R: a language and environment for statistical computing . Vienna : R Foundation for Statistical Computing . Available at: http://www.R-project.org (accessed 10 January 2017). Reilly S , Essner R , Wren S , Easton L , Bishop P . 2015 . Movement patterns in leiopelmatid frogs: insights into the locomotor repertoire of the earliest anurans . Behavioural Processes 121 : 43 – 53 . Google Scholar CrossRef Search ADS Rohlf FJ , Slice DE . 1990 . Extensions of the Procrustes method for the optimal superimposition of landmarks . Systematic Zoology 39 : 40 – 59 . Google Scholar CrossRef Search ADS Stephenson NG . 1951 . Observations on the development of the amphicoelous frogs, Leiopelma and Ascaphus . Journal of the Linnean Society of London, Zoology 42 : 18 – 28 . Google Scholar CrossRef Search ADS Stephenson EM , Stephenson NG . 1957 . Field observations on the New Zealand frog, Leiopelma Fitzinger . Transactions of the Royal Society of New Zealand 84 : 867 – 882 . Tennyson AJD . 2010 . The origin and history of New Zealand’s terrestrial vertebrates . New Zealand Journal of Ecology 34 : 6 – 27 . Thurlow L . 2015 . Deducing the phylogeny of New Zealand’s endemic frog genus, Leiopelma . Unpublished M.Sc. Thesis, University of Otago . Towns DR , Daugherty CH . 1994 . Patterns of range contractions and extinctions in the New Zealand herpetofauna following human colonisation . New Zealand Journal of Zoology 21 : 325 – 339 . Google Scholar CrossRef Search ADS Wallis GP , Waters JM , Upton P , Craw D . 2016 . Transverse alpine speciation driven by glaciation . Trends in Ecology & Evolution 31 : 916 – 926 . Google Scholar CrossRef Search ADS Wilmshurst JM , Anderson AJ , Higham TFG , Worthy TH . 2008 . Dating the late prehistoric dispersal of Polynesians to New Zealand using the commensal Pacific rat . Proceedings of the National Academy of Sciences of the United States of America 105 : 7676 – 7680 . Google Scholar CrossRef Search ADS Worthy TH . 1987a . Palaeoecological information concerning members of the frog genus Leiopelma: Leiopelmatidae in New Zealand . Journal of the Royal Society of New Zealand 17 : 409 – 420 . Google Scholar CrossRef Search ADS Worthy TH . 1987b . Osteology of Leiopelma (Amphibia: Leiopelmatidae) and descriptions of three new subfossil Leiopelma species . Journal of the Royal Society of New Zealand 17 : 201 – 251 . Google Scholar CrossRef Search ADS Worthy TH . 1998 . The Quaternary fossil avifauna of Southland, South Island, New Zealand . Journal of the Royal Society of New Zealand 28 : 537 – 589 . Google Scholar CrossRef Search ADS Worthy TH , Holdaway RN . 1993 . Quaternary fossil faunas from caves in the Punakaiki area, West Coast, South Island, New Zealand . Journal of the Royal Society of New Zealand 23 : 147 – 254 . Google Scholar CrossRef Search ADS Worthy TH , Holdaway RN . 1994 . Quaternary fossil faunas from caves in Takaka Valley and on Takaka Hill, northwest Nelson, South Island, New Zealand . Journal of the Royal Society of New Zealand 24 : 297 – 391 . Google Scholar CrossRef Search ADS Worthy TH , Holdaway RN , Alloway BV , Jones J , Winn J , Turner D . 2002 . A rich Pleistocene-Holocene avifaunal sequence from Te Waka #1: terrestrial fossil vertebrate faunas from inland Hawke’s Bay, North Island, New Zealand. Part 2 . Tuhinga: Records of the Museum of New Zealand Te Papa Tongarewa 13 : 1 – 38 . Worthy TH , Roscoe D . 2003 . Takaka Fossil Cave – a stratified Late Glacial to Late Holocene deposit from Takaka Hill, New Zealand . Tuhinga: Records of the Museum of New Zealand Te Papa Tongarewa 14 : 41 – 60 . Worthy TH , Tennyson AJD , Scofield RP , Hand SJ . 2013 . Early Miocene fossil frogs (Anura: Leiopelmatidae) from New Zealand . Journal of the Royal Society of New Zealand 43 : 211 – 230 . Google Scholar CrossRef Search ADS Zoological Society of London . 2012 . Top 100 Evolutionary Distinct and Globally Endangered (EDGE) amphibians . Available at: http://www.edgeofexistence.org/amphibians/top 100.php (accessed 17 May 2014 ). Zug GR . 1972 . Anuran locomotion: structure and function. I. Preliminary observations on relation between jumping and osteometrics of appendicular and postaxial skeleton . Copeia 1972 : 613 – 624 . Google Scholar CrossRef Search ADS © 2017 The Linnean Society of London, Zoological Journal of the Linnean Society This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Zoological Journal of the Linnean SocietyOxford University Press

Published: Dec 4, 2017

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off