Ancient Phenotypes Revealed Through Present Day Species—A Morphological Analysis of Australia's Saw-Shelled Turtles Including the Threatened Myuchelys bellii (Testudines: Chelidae)
Abstract
Understanding of evolutionary history and speciation events to inform phylogeography of extant species can be gleaned from the fossil record. However, the fossil record for Australian freshwater turtles contains many gaps and interpretation of the fossils that are known is difficult because of poor knowledge of the morphology of extant forms. There are no fossils currently assigned to short-necked turtles in the genus Myuchelys. In an alternative approach, morphological characters of the extant species Myuchelys bellii, Myuchelys georgesi, Myuchelys latisternum, and Myuchelys purvisi were mapped against their current molecular phylogeny. The cryptic species pair, M. georgesi and M. purvisi, are morphologically very similar, but distant phylogenetically, and their common ancestor includes M. latisternum and M. bellii in its decendents. This suggests that current shared morphology of M. georgesi and M. purvisi represents a suite of symplesiomorphies (shared primitive characters). These characters include presence of a cervical scute, lack of prominent neck tubercles, a relatively small head size, minor or no serrations of marginal scutes, an oval carapace, and a smooth head shield extending down the parietal arch toward the tympanum. This commonality of characters of this cryptic species pair represents a rare insight to the ancestral phenotype of the Myuchelys. The loss of the cervical scute and presence of enlarged neck tubercles, a large robust head and furrowed head shield, deep serrated shell margins, and compressed shell profile in M. bellii and M. latisternum are synapomorphies (shared derived characters). There is no morphological support for the distinction at the level of species or subspecies between discrete populations of M. bellii. A dichotomous key to species is provided.
Fossils are fundamental in providing morphological data on ancestral lineages (Rougier et al. 1995), clarifying evolutionary relationships among extant taxa (Thomson and Mackness 1999; Thomson 2000) and their biogeography (Noonan 2000; Lieberman 2003). The Australian turtle fossil record includes 5 families of turtles: Meiolaniidae, Trionychidae, Carettochelyidae, Chelonioidea (superfamily), and Chelidae (Gaffney and Bartholomai 1979; Gaffney 1979, 1983, 1991; Legler and Georges 1993; Gaffney et al. 1998). Chelids dominate the Australian freshwater turtle fauna from the Miocene to the present (Legler and Georges 1993; Georges and Thomson 2006), yet the fossil record of Myuchelys is poor. Only 6 species have been described: the extinct forms of Rheodytes devisi, Chelodina insculpta, Chelodina alanrixi, Elseya uberima, and Elseya nadibajagu (deVis 1897; Gaffney 1981; Thomson and Mackness 1999; Thomson 2000; de Broin and Molnar 2001) and the “living fossil” Elseya lavarackorum (Thomson et al. 1997). No chelid fossils to date have been affiliated with Myuchelys (Thomson and Georges 2009).
The fossil record is often incomplete, specimens are seldom entire, or available material has not been examined or has been considered indeterminate (Gaffney 1979, 1981; de Broin and Molnar 2001). For the Chelidae, a lack of foundational knowledge of the morphology of extant forms is a major impediment. Fortunately, morphology and DNA sequence variation of extant forms provides indirect information on evolutionary relationships to complement the incomplete direct evidence in the fossil record.
Recent molecular studies (Georges and Adams 1992, 1996; Georges et al. 1998; Fielder et al. 2012) have yielded a robust phylogeny for extant species of Myuchelys (Fig. 1). Myuchelys georgesi (Cann 1997) and Myuchelys purvisi (Wells and Wellington 1985) were regarded as allopatric populations of the same species, having relatively minor distinguishing features differing overtly only in intensity of coloration (Thomson and Georges 1996, 2009; Cann 1997). Genetically, they are deeply divergent (20% fixed allozyme differences [Georges and Adams 1992, 1996]; 13.5% mtDNA differences [Fielder et al. 2012]), and subsequent examination revealed one major distinguishing feature, presence and absence of neural bones, elsewhere considered to be generic-level characters (Thomson and Georges 1996). Remarkably, these formerly cryptic species (Georges and Adams 1996; Thomson and Georges 1996) are not sister species as they have nested between them in the phylogeny Myuchelys bellii (Gray 1844), Myuchelys latisternum (Gray 1867) and the Emydura (Georges and Adams 1996; Fielder et al. 2012) (Fig. 1).
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0939.1
In Australia, reptile taxonomy is confused by the proliferation of names attached to taxonomic concepts with little or no support (see Georges and Thomson 2010; Oliver and Lee 2010). Names have been erected for taxa or suspected taxa outside the mainstream scientific literature, often with scant morphological analysis and without the benefit of peer review, and this is certainly true of many species of Myuchelys (Wells 2007a, Wells 2007b, 2009). In this article, I present an analysis of the morphological characters that distinguish the species of Myuchelys, identify diagnostic characteristics and present a key to species. These characters are mapped against the molecular phylogeny for the genus, which offers a unique window into the phenotype of the Myuchelys common ancestor. I provide an assessment of the assignment of the holotype for the threatened M. bellii, assess the status of the Bald Rock Creek population regarded by Cann (1998) as a separate species, and speculate on the origins of the radiation of the group across eastern and northern Australia as revealed through a greater understanding of plesiomorphic (retained) and synapomorphic (derived) traits.
METHODS
Turtles were captured by hand with mask and fins, in cathedral traps (Legler 1960; as modified by Georges et al. 2006) baited with meat, liver, and sardines and by use of a long-handled dip net and spotlight from a boat at night. Traps were checked at intervals of 2–12 hrs, rebaited after approximately 24 hrs, and left in place for 24–72 hrs at each location. Turtles were generally processed and returned to the place of capture within 24 hrs of being caught, or otherwise held in polyethylene tubs of water to prevent desiccation until their release (< 72 hrs; infrequently, isolated individuals were held for longer periods).
Myuchelys bellii was collected from the Namoi River (New South Wales), Gwydir River (New South Wales), and Bald Rock Creek (tributary of the Border Rivers, New South Wales and Queensland); M. georgesi from the Bellinger River (New South Wales); M. latisternum from the Brisbane and Albert river catchments (Queensland); and M. purvisi from the Barnard River (tributary of the Manning River, New South Wales) (Fig. 2). Several specimens representing M. bellii from Bald Rock Creek and the McDonald River (n = 6), M. georgesi paratypes from the Bellinger River (n = 2) and M. latisternum from the Pascoe, Lynd, and Ross rivers (n = 13) were examined from the Queensland Museum. Additional data for M. latisternum from the Albert River were obtained from C.J. Limpus and D.J. Limpus and for M. georgesi from the Bellinger River from M. Welsh, A. Georges, and R.-J. Spencer. Measurements and carapace diagram for the M. bellii holotype specimen were taken from Cann (1998, p. 210).
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0939.1
Shell, head, and tail measurements were recorded to the nearest 0.1 mm using vernier calipers. Mass was recorded to the nearest 2 g using digital scales (< 2.5 kg) or to the nearest 25 g using a series of spring balances. Nomenclature of the bones of the shell and overlying scutes follows that of Zangerl (1969). Morphometric measurements: carapace length (CL), maximum straight midline length of carapace from the anterior margin of the cervical scute or juncture of the first marginal scutes to the marginal junction of the two most posterior marginal scutes; carapace width (CW), maximum straight-line width of the carapace perpendicular to the axis of CL; plastron length (PL), maximum midline distance from the anterior margin of the intergular shield to the anal notch; anterior plastron width (PW), width immediately anterior to the bridge (distance between the auxiliary notches); posterior plastron width (PWP), width immediately posterior to the bridge (distance between inguinal notches); head width (HW), maximum straight-line width of skull at tympanum; head length (HL), maximum straight-line length between the anterior junction of the premaxillae and posterior extent of the supraoccipital; tail length (TL), straight-line length of tail from tip to the most anterior point of the anal notch of the plastron with the tail held at its maximum extent; tTail to carapace (TC), straight-line length of tail from tip to most posterior point of carapace; tail-to-vent length (TVL), straight-line distance of tail from tip to anterior edge of the vent; shell depth (SD), maximum straight-line depth of the shell; and body mass (WT).
Sex and sexual maturity were determined either by laparoscopy (Limpus et al. 2002) or using secondary tail characteristics. Mature males were distinguished from mature females based on tail length (Georges et al. 2006), and the minimum size of visually identifiable males was used as the upper size limit for all juveniles. In some cases, males were identified and classified as mature by penis eversion (Georges 1983).
Principal components analyses (PCA) were carried out on variance–covariance matrices to summarize the data and reveal the morphometric variables that are best able to distinguish among taxa. Ratios of variables (e.g., CL/HW) were used to reduce size biases in statistical analyses, and for WT, a log transformation was completed prior to analysis. Excluded from PCA were those specimens without a full set of morphometric data, hatchlings with CL < 70 mm and specimens with known shell or body deformities that compromised recorded morphometrics.
For each PCA, variables most highly correlated with the first two principal components were examined in greater detail, since these two components accounted for most of the variation among taxa. An analysis of similarities (ANOSIM) between the samples was conducted on a resemblance matrix for each PCA performed. Multivariate statistical analyses were carried out using PRIMER (Version 6, Plymouth Marine Laboratories) and univariate analyses, including linear regression and analysis of variance (ANOVA), were carried out using Statistix (Version 7, Analytical Software). Bonferroni corrections were applied to all pairwise multiple comparisons following a significant result in the ANOVA. All results are presented as mean ± standard error unless otherwise stated.
I follow the advice of Fritz and Havas (2007), Turtle Taxonomy Working Group (2010), and Georges and Thomson (2010) and do not regard the documents circulated by Wells under the banner Australian Biodiversity Record as publications for purposes of nomenclature. I follow the taxonomy of Georges and Thomson (2010) with the exception of Elseya novaeguineae, which I regard as remaining within the genus Elseya.
RESULTS
Sexual Maturity
The smallest sexually mature male M. bellii examined by laparoscope (n = 26) had a CL of 168 mm, and the largest immature male M. bellii had a CL of 187 mm. For M. latisternum (n = 119), the corresponding range was CL 125–135 mm. Sexual maturity of male M. georgesi and M. purvisi was based on secondary sexual characteristics, with maturity occurring at CL > 140 mm. The smallest mature female M. bellii (n = 27) had a CL of 222 mm, and the largest immature female had a CL of 212 mm; for M. latisternum (n = 100) the corresponding values were 182–189 mm. Sexually mature female M. georgesi and M. purvisi were identified as having a CL > 140 mm in the absence of male sexual dimorphic characters.
Morphometric Analyses of Myuchelys Species
PCA undertaken separately for mature males and females (Fig. 3a, b) reveals M. latisternum to be the most distinct morphologically among the 4 species. PC 1 accounted for 62.3% of variation for males and 67.1% for females. Along this axis, M. latisternum is clearly discriminated from its congeners, with variables most correlated with PC1 being CL/HW (males, r = −0.80; females, r = 0.74), CW/HW (males, r = −0.53; females, r = 0.51), PWP/HW (males, r = −0.22; females, r = −0.26) and CL/HL (females, r = 0.35) (Table 1). PC 2 accounted for 22.9% of variation for males and 15.2% for females. The variables most correlated with this axis were CL/WTlog (males, r = 0.46; females, r = 0.57), CL/TL (males, r = 0.37; females, r = −0.79), CL/SD (males, r = 0.56; females, r = 0.16), CL/PWP (males, r = −0.27), CW/HW (males, r = 0.30; females, r = 0.03) and CL/HW (males, r = −0.22; females r = −0.05) (Table 1). These variables along the second axis identify differences between the four Myuchelys species in WT, SD, TL, plastron width, and head size. Male and female M. bellii were separated along PC 2 for their relatively shorter and longer tails respectively, and comparatively lighter body weight for their size than either M. georgesi or M. purvisi.
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0939.1
An ANOSIM on the PCA resemblance matrix supports significant overall differences (0.05%) between the taxa (global R for males = 0.83 and females = 0.80) (Table 2). The ANOSIM reveals M. latisternum and M. bellii as the most morphologically distinct taxa of Myuchelys. Myuchelys georgesi and M. purvisi were the most similar (males r = 0.71; females r = 0.41).
A second set of PCA that excluded M. latisternum and M. bellii specimens was better able to discriminate between M. georgesi and M. purvisi (Fig. 4a, b). PC 1 accounts for 56.8% of variation for mature males and 66.2% for mature females. PC1 provides discrimination between male and female M. georgesi and M. purvisi; variables most correlated with PC 1 are CL/PWP (males, r = 0.45; females, r = 0.54), CW/PWP (males, r = 0.47), CL/SD, (males, r = −0.44; females, r = −0.28), PW/HW (males, r = −0.41; females, r = −0.46), and PWP/HW (females, r = −0.54) (Table 3). PC 2 accounts for 22.1% of variation for males and 23.4% of females, but does not provide further differentiation between species. SD, CW, PWP, and head size are significant variables that distinguish M. georgesi and M. purvisi. The ANOSIM of a resemblance matrix revealed significant differences (males R = 0.82 and females R = 0.83, 0.05%).
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0939.1
Correlated morphometric variables identified through PCA were selected for univariate analyses. HW and HL (CL/HW and CL/HL) between all four species were found to be significantly different (N = 748, p < 0.001; Fig. 5). The larger relative HW (CL/HW = 4.66 ± 0.0147; n = 349) and HL (CL/HL = 3.34 ± 0.012; n = 349) for M. latisternum separates it from its congeners and was a primary morphometric character for describing variation among species. Additionally, M. bellii has a broader head than M. purvisi, which in turn has a broader head than M. georgesi. TL was also found to separate each species in univariate space with mature M. bellii males having relatively shorter TLs (CL/TC = 3.28 ± 0.039, n = 65) than their male congeners, and mature M. bellii females having correspondingly longer TLs (CL/TC = 6.61 ±0.094, n = 82; Fig. 6) than other Myuchelys species. In contrast, male M. georgesi and M. purvisi have the longest relative TL and female M. purvisi the shortest relative TL of the Myuchelys taxa.
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0939.1
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0939.1
Shell morphology was also diagnostic between species. The CW of males and females were found to have a similar pattern where M. georgesi was the most elongate (narrow), followed by M. purvisi and then M. latisternum. Myuchelys bellii exhibited an oval carapace that was generally wider than all other taxa for all size classes. Mean anterior plastron width (CL/PW) was significantly different between male M. bellii, M. georgesi, M. latisternum, and M. purvisi (F3,283 = 74.77, p < 0.0001). Myuchelys latisternum males were found to have the broadest anterior plastron and M. purvisi males the narrowest (Fig. 7). In contrast, female M. latisternum and M. bellii possessed statistically similar PWs, but were significantly different than M. georgesi and M. purvisi (F3,332 = 45.98, p < 0.0001). PW means were significantly different between female M. georgesi and M. purvisi, which in turn were significantly different from female M. bellii and M. latisternum. PWP (CL/PWP) means were significantly different between male M. georgesi and M. purvisi, which were significantly different from M. bellii and M. latisternum (F3,186 = 76.66, p < 0.0001). Mean CL/PWP for female M. purvisi was significantly different from M. bellii, M. georgesi, and M. latisternum (F3,259 = 54.34, p < 0.0001).
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0939.1
Myuchelys latisternum possess the broadest plastron (anterior and posterior) followed by M. bellii and then M. georgesi; M. purvisi has the narrowest plastron of all 4 taxa (Fig. 7). Both male and female M. purvisi have a much greater SD than any other species and M. bellii has the most compressed shell profile of all species.
Morphometric Analysis of M. bellii
PCA applied to the 3 M. bellii populations revealed some segregation between the Bald Rock Creek population and the populations from the Gwydir and Namoi rivers (Fig. 8). The morphometric variables used for the PCA and their correlation scores are listed in Table 4. Bald Rock Creek males were most dissimilar to Gwydir River males (r = 0.53) and Bald Rock Creek females were most dissimilar to Namoi River females (r = 0.56). Male and female Gwydir and Namoi river populations were the most similar (males r = 0.09, females r = 0.06). Although there was some distinction in these variables, there were also large overlaps of values revealed in each PCA. None of the ANOSIM results for either male or female were significant at the 0.05% level.
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0939.1
The largest individuals recorded during this study from each population were as follows: female 300 mm (3250 g) and male 213 mm (1205 g) from Bald Rock Creek; female 278 mm (2560 g) and male 219 mm (1195 g) from the Gwydir River; and female 298 mm (3240 g) and male 213 mm (1025 g) from the Namoi River.
Morphometric analyses of Myuchelys bellii holo-type
A PCA of mature males and the holotype was performed using the limited number of available variables for the holotype (Fig. 9). The holotype is clearly discriminated being closest to M. bellii. PCA variables that most influence discrimination along PC1 were CL/HW (r = 0.82) and CW/HW (r = 0.55); along PC2 they were CW/SD (r = −0.72), CW/HW (r = −0.43), and CL/CD (r = −0.43) (Table 5). Thus, HW, CW, and SD are key morphological features enabling the assignment of the holotype.
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0939.1
From univariate analyses, the HW of the holotype relative to its CL of 5.8 (Fig. 5) is closest to the geometric mean of M. bellii (5.6 ± 0.016; n = 228) and M. purvisi (5.8 ± 0.03; n = 58) and is consequently too narrow to be M. latisternum. The CL/CW (1.08) is intermediate between M. latisternum and M. bellii. The CL/PW (2.33; Fig. 7) place it between the geometric mean of M. bellii (2.3 ± 0.006; n = 224) and M. georgesi (2.37 ± 0.008; n = 113). Additionally, the holotype has a very compressed dorsolateral shell profile (like individuals of M. bellii), although it is a statistical outlier in comparison to all Myuchelys (CL/SD = 3.56). A possible explanation for this may be that the 150-yr-old dry preserved specimen has developed an artificial compression of the shell during its storage. Other characteristics of the holotype appear to place it most closely to M. bellii (Fig. 10). These characteristics include the specimen retaining a cervical scute (98% of M. bellii possess one, and 15% of M. latisternum), moderately deep serrations on the posterior margin of the carapace, coloration patterns on the plastron, and the soft tissue parts of the limbs and head.
![Figure 10. . Carapace outlines for (a) Myuchelys bellii (126 mm carapace length [CL]) Gwydir River, (b) Myuchelys bellii holotype (114 mm CL), (c) Myuchelys latisternum (120 mm CL) Albert River, (d) Myuchelys georgesi (137 mm CL) Bellinger River, and (e) M. purvisi (102 mm CL) Barnard River (tributary of the Manning).](/view/journals/ccab/12/1/i1071-8443-12-1-101-f10.png)
![Figure 10. . Carapace outlines for (a) Myuchelys bellii (126 mm carapace length [CL]) Gwydir River, (b) Myuchelys bellii holotype (114 mm CL), (c) Myuchelys latisternum (120 mm CL) Albert River, (d) Myuchelys georgesi (137 mm CL) Bellinger River, and (e) M. purvisi (102 mm CL) Barnard River (tributary of the Manning).](/view/journals/ccab/12/1/full-i1071-8443-12-1-101-f10.png)
![Figure 10. . Carapace outlines for (a) Myuchelys bellii (126 mm carapace length [CL]) Gwydir River, (b) Myuchelys bellii holotype (114 mm CL), (c) Myuchelys latisternum (120 mm CL) Albert River, (d) Myuchelys georgesi (137 mm CL) Bellinger River, and (e) M. purvisi (102 mm CL) Barnard River (tributary of the Manning).](/view/journals/ccab/12/1/inline-i1071-8443-12-1-101-f10.png)
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0939.1
Additional Observations
Most M. bellii (98%, n = 247) had a cervical scute, contrasting with only 15% of M. latisternum (n = 246) in the Albert and Brisbane rivers, 77% of M. georgesi (n = 382) in the Bellinger River, and 96% of M. purvisi (n = 74) from the Manning River. All M. bellii specimens examined had an olive-grey iris and a black pupil encircled by a thin pale-white ring that becomes more pronounced with pupil dilation. This distinguishes it from M. latisternum, which have marked variation in eye color from catchment to catchment but always with a leading and trailing dark spot (Georges and Thomson 2010), and M. georgesi and M. purvisi, which have a golden brown iris with specks. Both M. bellii and M. latisternum have well-defined dorsal neck tubercles (Legler and Winokur 1979) and prominent scales in the temporal region behind the eyes. The numerous cornified tubercles distinguish them from M. purvisi and M. georgesi which have a reduced presence of neck tubercles. The head shield for M. bellii and M. latisternum is large and prominent, nearly extending down to the tympanum, and is ridged, including a distinct central furrow. Myuchelys purvisi and M. georgesi head shields are smoother, lacking a well-defined central furrow. Moderately deep double serration of the carapace for M. bellii and M. latisternum separate them from M. purvisi and M. georgesi, which exhibit minor or no serrations on rear marginal scutes.
KEY TO MYUCHELYS SPECIES
A key to Australian Myuchelys is provided here. The species included are the western saw-shelled turtle, M. bellii; the common saw-shelled turtle, M. latisternum; the Bellinger River helmeted turtle, M. georgesi; and the Manning River helmeted turtle, M. purvisi. The key excludes the New Guinea spotted turtle, E. novaeguineae, which was only tentatively included in Myuchelys pending further genetic studies by Georges and Thomson (2010). Common names follow that of Georges and Thomson (2010). The key is applicable to animals > 70 mm carapace length.
- 1
Head broad with deeply furrowed head shield; prominent conical neck tubercles; carapace exhibits deep, double serrations to posterior marginal scutes 8–12 (except if worn smooth in older individuals); carapace oval with or without expanded marginal scutes 7–12; plastron broad . . . 2
- –
Head slender; head shield smooth; neck tubercles small and rounded or nonexistent; carapace oval and smooth edged, sometimes exhibiting shallow serration of posterior marginal scutes; plastron narrow . . . 3
- 2
CL to HW dimensions < 5.2 (except as rare variant); iris with leading and trailing dark spots; cervical scute absent or narrow when present (15% of population); body size moderate with relatively narrow carapace width; plastron color varies between catchments; shell profile moderately deep . . . M. latisternum
- –
CL to HW dimensions > 5.2 (except as rare variant); iris clear with a uniform olive-grey color; cervical scute present (except as rare variant); carapace oval with expanded posterior marginals; plastron becoming predominantly black with age; shell profile laterally compressed . . . M. bellii
- 3
CL to PWP dimensions < 2.62 females and < 2.72 males (except as rare variant); shell profile moderately shallow; cervical scute narrow (absent in 23% of population); olive coloring on ventral surface of carapace and plastron; neural bones absent . . . M. georgesi
- –
CL to PWP dimensions > 2.62 females and > 2.72 males (except as rare variant); shell profile moderately deep; broad cervical scute present (except as rare variant); bright yellow coloring of the ventral surfaces of the carapace and plastron (except in the oldest individuals); a bright yellow lateral stripe on the ventral surface of each limb, extending from the plastron to the first toe; central, bright yellow tail stripe from the plastron anal notch to the cloaca with two lateral yellow tail stripes meeting at the cloaca; ventral tail tip yellow; neural bones present . . . M. purvisi
DISCUSSION
From the morphological comparisons presented in this article, M. georgesi and M. purvisi share a suite of characteristics including the following: medium size, having a relatively narrow oval carapace exhibiting minor serrations to the posterior margins, retaining a cervical scute (considered to be a primitive character state, Gaffney 1977), possessing a relatively small head for their body size, having small rounded neck tubercles, and exhibiting a smooth head shield (not deeply furrowed). The sister species pair of M. bellii and M. latisternum, which are nested between M. georgesi and M. purvisi in the phylogeny (Fig. 1), have a larger overall size, laterally expanded carapace (particularly M. bellii), prominent conical neck tubercles (Legler and Winokur 1979), a significantly larger head (particularly M. latisternum), a furrowed head shield, expanded posterior carapace margins with moderately deep serrations, and no cervical scute (M. latisternum only). This means either character convergence in allopatry between M. georgesi and M. purvisi, or that their shared characters are plesiomorphic. This latter interpretation is more likely and thus, the phenotype of the Myuchelys common ancestor would have expressed the plesiomorphic characters of the present-day M. georgesi and M. purvisi. In the same way, the shared characters of the sister species pair M. bellii and M. latisternum are considered to be synapomorphies, although, neck tubercles are thought to be primitive traits based on fossil chelids from South America (Georges et al. 1998).
These plesiomorphic and synapomorphic characters mapped against the molecular phylogeny for the genus (Fig. 1) provide the first insights into the likely sequence of events in the Myuchelys dispersal across eastern and northern Australia. The distinctive characters possessed by M. bellii and M. latisternum—broader heads, conical neck tubercles, and diagnostic shell morphology—developed prior to their radiation over a wide geographic area from coastal catchments in northern New South Wales to tropical Australia in Queensland and the Northern Territory and into the Murray Darling system. Furthermore, it is inferred that the M. latisternum and M. bellii split from their ancestral state occurred prior to M. latisternum developing a significantly larger head (considered here to be an apomorphy), and M. bellii developing a relatively more compressed shell profile and wider carapace. Myuchelys bellii retained a cervical scute and a relatively smaller head than M. latisternum after its dispersal westward into the Murray-Darling Basin system. Conversely, M. latisternum developed a robust head and an absence of a cervical scute (narrow when present) prior to its rapid radiation along the coastal drainages of eastern and northern Australia.
The rapid dispersal by M. latisternum contrasts to the biogeography of the cryptic species pair M. georgesi and M. purvisi. These species are endemic to single catchments of the Manning and Bellinger rivers from which they apparently have never dispersed. The plesiomorphic traits retained in allopatry by M. georgesi and M. purvisi may have resulted from the stability of the environments that they now occupy. The Manning and Bellinger river systems, separated by 2 coastal river catchments, are similar in their geological features and climatic conditions. Hence, their habitats are similar, possibly allowing for particular plesiomorphic traits to be retained independently of one another over the millennia.
An aged M. georgesi specimen captured as part of a genetic study of the Bellinger River Emydura was mistakenly identified in the field as an M. latisternum but later confirmed as having an M. georgesi mtDNA haplotype (Georges et al. 2011). It had features that resembled M. latisternum, including more prominent neck tubercles, robust head size, and larger body size. This leads me to speculate that the M. latisternum phenotype may have arisen by gerontomorphy, the early expression of character states not expressed in the ancestral phenotype, represented today by M. georgesi and M. purvisi, except in aged individuals. In contrast, Emydura, displaying minor tubercles, small head size, lack of a prominent head shield, and smooth shell margins, may be the reverse expression of a neotenous phenotype derived from the ancestral phenotype of M. georgesi and M. purvisi. Neoteny, the maintenance of juvenile characteristics in adulthood, is a common phenomenon proposed to be responsible for major morphological changes (Gould 1977). Myuchelys latisternum may be an example of selection for early expression of traits normally expressed only in the very oldest, largest M. georgesi. This proposal, involving the evolution in the timing of development of character states already present in the ancestor, would allow rapid change of form and new opportunities. Herbivory in the case of Emydura and carnivory in the case of M. latisternum would have driven and consolidated the morphology shift.
Despite M. georgesi and M. purvisi previously being cryptic species, I was able to identify subtle, yet significant, morphological features to separate them. External differences between the two species included M. purvisi having a comparatively deeper shell for both sexes (the deepest of all Myuchelys species examined), a significantly narrower plastron for both sexes, and a wider head and shorter tail for mature females than M. georgesi. Together, these previously undocumented characters are diagnostic for the 2 species.
In the case of M. bellii I found no morphological support from analyses of over 233 individuals across all populations for a cryptic species as asserted by Cann (1998). Evidence presented by Fielder et al. (2012) show a shallow genetic structure, which in combination with a lack of phenotypic variation among M. bellii populations (this study), preclude the existence of a cryptic taxon at the species or subspecies level. Morphological assessment of the M. bellii holotype specimen (Phrynops bellii, OUM 8460) confirmed earlier statements by Cann (1998) that its origins were M. bellii. Recommendations by Georges and Thomson (2010) for genetic testing of the 150-yr-old dry preserved specimen would provide independent data to this question. Based on this morphometric analysis taxonomic validity of the binomial name for the biological entity of M. bellii is maintained.
The melding of morphology in this present study with the molecular phylogeny of Myuchelys provided a unique insight into the ancient phenotype of the common ancestor to Myuchelys. In the absence of direct evidence from fossils, this research has advanced our understanding of the morphological traits, evolution, and dispersal events of the Myuchelys species of freshwater turtles in Australia. An area for future research would be the examination of morphometrics from northern Australia populations of M. latisternum, which may reveal greater levels of intraspecific morphological variation as suspected by Cann (1998). In addition, testing the proposed hypothesis of Emydura showing neoteny in comparison with the ancestral phenotype of M. georgesi and M. purvisi, and M. latisternum gerontomorphy, would be worthy of further investigation to elucidate the selection processes that influenced the present-day morphology of Myuchelys.
The most recent phylogeny for the Myuchelys based on mtDNA sequence variation (Fielder et al. 2012).
A map showing the capture localities of the Myuchelys specimens used for morphological analysis (excluding Queensland Museum specimen locations). Drainage basin names are those of “Australia's River Basins 1997” (Geoscience Australia). Circles represent Myuchelys bellii, squares are Myuchelys latisternum, triangles are Myuchelys georgesi, and inverted triangles are Myuchelys purvisi.
Principal components anaylysis (PCA) of (a) mature males and (b) mature females: Myuchelys bellii (+), Myuchelys georgesi (•), Myuchelys latisternum (□), and Myuchelys purvisi (△). The morphometric ratios included in each PCA are listed in Table 1.
Principal components anaylysis (PCA) of (a) mature males and (b) mature females: Myuchelys georgesi (•) and Myuchelys purvisi (△). The morphometric ratios included in each PCA are listed in Table 3.
Means ± SE for carapace length to head width (CL/HW, solid line and diamond) and carapace length to head length (CL/HL, dashed line and x icons): ML, Myuchelys latisternum; MB, Myuchelys bellii; MBh, Myuchelys bellii holotype; MG, Myuchelys georgesi; and MP, Myuchelys purvisi. Mean CL/HW were significantly different for all taxa (F3,742 = 1282.85, p < 0.0001).
Means ± SE for carapace length to tail-to-carapace length (CL/TC) for sexually mature males (dashed line) and females (solid line): ML, Myuchelys latisternum; MB, Myuchelys bellii; MG, Myuchelys georgesi; and MP, Myuchelys purvisi. Male M. bellii were significantly different from M. georgesi, M. latisternum, and M. purvisi (F3,232 = 16.03, p < 0.0001). All 4 means for females were significantly different (F3,243 = 42.28, p < 0.0001).
Means ± SE for carapace length to plastron width anterior (CL/PW) and carapace length to plastron width posterior (CL/PWP) for male and female Myuchelys. ML, Myuchelys latisternum; MB, Myuchelys bellii; MG, Myuchelys georgesi; and MP, Myuchelys purvisi.
Principal components analysis (PCA) of M. bellii for the catchments of Bald Rock Creek (•), Gwydir River (▽), and Namoi River (□): (a) males and (b) females. The morphometric ratios included in each PCA are listed in Table 4.
Principal components analysis (PCA) of mature male Myuchelys latisternum (□), Myuchelys bellii (+), Myuchelys georgesi (•), and Myuchelys purvisi (△), including the M. bellii holotype (⧫). The morphometric ratios included in each PCA are listed in Table 5.
Carapace outlines for (a) Myuchelys bellii (126 mm carapace length [CL]) Gwydir River, (b) Myuchelys bellii holotype (114 mm CL), (c) Myuchelys latisternum (120 mm CL) Albert River, (d) Myuchelys georgesi (137 mm CL) Bellinger River, and (e) M. purvisi (102 mm CL) Barnard River (tributary of the Manning).
