Two new species of fossil Leggadina (Rodentia: Muridae) from Northwestern Queensland

Univariate analyses

Coefficients of Variation for all measurements of the fossil taxa suggest that only one species is present in each fossil sample, with values ranging from 3.23 to 7.80 in the Leggadina specimens from Rackham’s Roost and 5.50 to 6.06 for the two measurements available for Leggadina specimens from Site 5C (Supplemental Information).

Bivariate analyses

In the bivariate plots, both length and width of M¹ and M² were effective in separating species (Figs. 2 and 3). The M¹ plot shows the Rackham’s Roost Leggadina overlapping with both modern Leggadina species (L. forresti and L. lakedownensis), whereas in the M² plot, the Rackham’s Roost Leggadina groups predominately with the Floraville specimens. The Floraville Leggadina species distinctly separates from other species based on its greater M¹ length. Pseudomys and Notomys group together in both plots, but separate more in the M² plot based on length data. In both plots there is a close association between the fossil specimens and Zyzomys. More detailed morphological evidence effectively separates Zyzomys and the fossil specimens, as detailed in the differential diagnosis.

Differential diagnosis

The fossil species described below refer to the genus Leggadina and display characteristics typical of species of this genus. An accessory cusp on the first upper molar is present on all fossil specimens, all upper molars are inclined posteriorly, molar size is reduced along the row, with M³ often half the size or smaller than M¹, and the anterior edge of the zygomatic plate is relatively straight (Watts & Aslin, 1981). Leggadina gregoriensis differs from other species of the genus in the following combination of features: a greatly anteroposteriorly elongated T6 on M¹; T1-2 and T4-5 complexes are oriented buccolingually with T3 and T6 swept back at right-angles to lean proximally; an accessory cusp is present but small; M¹ is narrow, with M² and M³ being wider than M¹. Leggadina macrodonta differs from other species of the genus in the following combination of characters: M¹ is enlarged, being approximately 18% larger than Leggadina forresti and L. lakedownensis; M² is similarly enlarged, approximately 16% larger than in those species; an anterior cingulum is present and is enlarged with two accessory cuspules that wear to a greatly elongated accessory cusp; T1 and T4 are well-developed and posterolingually aligned; a T1 sup is present on some specimens; the central series of cusps is also enlarged. Both fossil species also differentiate themselves from the two modern Leggadina species through the presence of furrows between the lingual and central series of cusps in M¹ and M². Bivariate analyses determined that L. gregoriensis and L. macrodonta could have been referred to the genus Zyzomys. Shared morphological features and differences between Leggadina and Zyzomys are mentioned here (Fig. 4). Zyzomys species often display an accessory cusp on the first upper molar, have a relatively straight anterior edge to the zygomatic plate, and are of similar size to Leggadina (Watts & Aslin, 1981). A feature clearly distinguishing species of the two genera is a buccal row of cusps present in Leggadina species that is absent in Zyzomys. A distinctive aspect of Leggadina molar morphology, not shared by Zyzomys, is the posterior extension of the lingual series of cusps (Tate, 1951). For these reasons, the fossil species are referred to the genus Leggadina rather than Zyzomys. Character states are unable to be discussed here as relationships between the genus Leggadina and other murines are uncertain and can change depending on methods used. This situation is not helped by the lack of fossil evidence on murines in Australia.

Type species

Leggadina forresti (Thomas, 1906)

Other species

Leggadina lakedownensis Watts, 1976

Leggadina gregoriensis sp. nov.

Holotype QM F57259, partial right maxilla with M^1–3 (Fig. 5).

Type locality and age

Rackham’s Roost Site, Riversleigh World Heritage Area, northwestern Queensland; Pleistocene (Woodhead et al., in press).

Paratypes

QM F57244, partial right maxilla with M¹ and alveoli of M² and M³ (Fig. 6); QM F57258, partial left maxilla including zygomatic plate with M¹ and M² (Fig. 7).

Etymology

Named for the Gregory River which flows next to the Rackham’s Roost Site.

Diagnosis

Leggadina gregoriensis is characterised by a small accessory cusp, greatly anteroposteriorly elongated T6 on M¹; T3 and T6 swept back at right-angles to lean proximally; M¹ narrow, M² and M³ wider.

Referred specimens

QM F57240, right M¹; QM F57241, left M¹ in partial maxilla; QM F57242, right M¹; QM F57243, right M¹; QM F57245, left M¹; QM F57246, right upper molar row in partial maxilla; QM F57247, left M² in partial maxilla; QM F57248, right M¹ and M² in partial maxilla; QM F57249, right M¹ and M² in partial maxilla; QM F57250, right M¹ and M²; QM F57251, right M¹ and M²; QM F57252, left M¹ and M²; QM F57253, right M^1–3 in partial maxilla; QM F57254, left M¹; QM F57255, left M¹ in partial maxilla; QM F57256, right M¹ and M² in partial maxilla; QM F57257, left M¹ and M² in partial maxilla; QM F57260, right M^1–3; QM F57261, right M¹; QM F57262, right M^1–3; QM F57263, right M¹ and M²; QM F57264, left M¹ in partial maxilla; QM F57265, right M¹ in partial maxilla; QM F57283, left upper molar row; QM F39958, left M^1–3 (Table 1).

Description

M¹ large and elongated. M² approximately two-thirds the size of M¹. M³ smaller again, approximately half the size of M² (Table 1). Tooth row exhibits spiral torsion, M¹ straight with M² and M³ twisted slightly to the buccal edge. Furrow present between lingual series of cusps and central series of cusps in M¹ and M². Buccal series of cusps reduced along tooth row, central series of cusps enlarged. All cusps inclined posteriorly with tooth crown overlap.

M¹: Elongated and narrow. Anterior cingulum with a single and small elliptical accessory cusp sweeping backwards along lingual edge. Accessory cusp small in all specimens, almost indistinguishable in QM F57244. T1 very small and circular, connected to T2 at early stages of wear. T2 posteriorly inclined, large and elliptical. It is the highest cusp at early stages of wear but becomes uniform with the other M¹ cusps after wear. T1-2 complex buccolingually aligned. T3 positioned to posterior of T1-2 complex, at mid-point of tooth. T3 elliptical, directed proximally and connected to T2 by an enamel rim in the holotype. At early stages of wear it is entirely distinct but merges completely with T1-2 complex after extreme wear. T4 small, circular and merged with T5 at most stages of wear. It sweeps posteriorly from T5 so anterior edge of T4 is in line with the posterior edge of T5. T5 large, subtriangular in occlusal outline and leans posteriorly. Enamel rim connects T5 to both T4 and T6. T6 positioned posterior to T5, elongated anteroposteriorly and directed proximally, similar to T3. T6 merges with T4-5 complex after extreme wear. T6 also distinct from T9 at early stages of wear but merges quickly. Posterior edge of cusps T4–T6 arcs anteriorly to enclose T8. T7 barely discernible in holotype but is present in other specimens at early stages of wear before merging completely with T8. In these specimens it is small and directed posteriorly. T8 very large and circular, directed posteriorly. T9 incorporated at all stages of wear with T8. Enamel rim around cusps uniform throughout tooth but becomes slightly wider with extreme wear. Has three roots, all of which directed somewhat anteriorly.

Anterior root largest of the three, circular in shape and positioned under accessory cusp and T1–T3. Lingual root anteroposteriorly stretched, narrow and positioned under T6 and T9. Posterior root smallest of the three, circular and positioned under T8.

M²: Tooth is mostly circular in holotype but shape variable, with other specimens more elongate. Elongation is affected by size of T3 and T8, with the anterior of M² developing a bulge with increase in T3, similarly, posterior developing a bulge with increase in T8. T1 and T2 absent. T3 distinct and elliptical, directed proximally. T3 and T5 are the highest cusps at early stages of wear but T3 wears faster than T5 to become uniform with the other cusps. T4 small, circular and leans posteriorly. It is incorporated into T5, but also sweeps posteriorly from T5, with anterior edge of T4 in line with posterior edge of T5. T5 subtriangular and directed posteriorly. T6 positioned posterior to T5, elongated anteroposteriorly and oriented proximally. At later stages of wear T6 merges with T4-5 complex. Posterior edge of T4-5 complex and posterior edge connecting T6 with T9 forms anterior arc to enclose T8, similar to M¹. T7 absent. T8 large, circular in occlusal outline and directed posteriorly. At extreme stages of wear T8 merges with elongated T6. T9 merges with T8 at all stages of wear, similar to M¹. Enamel rim surrounding the cusps of uniform width, becoming thicker with wear.

With three roots, all directed vertically. The anterobuccal and posterobuccal roots of equal size and circular. Anterobuccal root extends from underneath T4 and T5, while posterobuccal root positioned beneath T8. Lingual root large and elongated, extending from T3 to T6.

M³: Tooth circular with a bulge on anterolingual edge for T3, cusp height uniform. T1 and T2 absent. T3 small, circular and distinct, directed proximally. Furrow between T3 and T4-6 complex ensures T3 distinct in all but very late stages of wear. T4 completely incorporated into T5. It sweeps posteriorly markedly from T5, directed posterobuccally. T5 subtriangular in occlusal outline, large and directed posteriorly. T6 small and subtriangular. It merges with T5, slightly sweeping posteriorly from T5 with enamel rim connecting to T8-9 complex. Posterior edge of T4-5 complex curves anterobuccally, with posterior edge of T6 curving anterolingually. T7 absent. T8 large, elliptical and orientated vertically. Anterior edge of T8 curves posteriorly. Anterior edge of T8 combined with posterior edge of T4-6 complex creates elliptical furrow. T9 entirely incorporated into T8. Enamel rim uniform in width and connecting all cusps except T3 in holotype which only connects at very late stages of wear.

With three roots all directed vertically. Anterobuccal root small and circular, extending from beneath T5. Anterolingual root slightly larger and more elongated than anterobuccal root and positioned under T3 and T6. Posterior root largest of the three, supporting approximately half tooth length and extending from T8.

Attachment node for the origin of the superficial masseter is of moderate size and well defined in some specimens, positioned anterior to M¹. Posterior extent of anterior palatal foramen lies at anterior root of M¹. Zygomatic plate of QM F57258 wide with posterior edge convex (Fig. 7).

Leggadina macrodonta sp. nov.

Holotype

QM F57276, partial left maxillary including zygomatic plate with M¹ (Fig. 8).

Type locality and age

Site 5C, Floraville Station, northwestern Queensland; Plio-Pleistocene (Rich et al., 1991).

Paratypes

QM F57273, partial left maxillary with M² (Fig. 9); QM F57268, left M¹ (Fig. 10); QM F57275, partial left maxillary with M¹ and alveoli of M² (Fig. 11).

Etymology

Named for the distinctively large size of the first upper molar.

Diagnosis

Leggadina macrodonta is characterised by a greatly enlarged M¹ and M²; enlarged anterior cingulum with two accessory cuspules that wear to a greatly elongated accessory cusp; well-developed T1 and T4 posterolingually aligned; enlarged central series of cusps.

Referred specimens

QM F57266, right M¹; QM F57267, left M¹; QM F57269, left M¹; QM F57270, right M¹; QM F57271, left M¹; QM F57272, right M¹; QM F57274, left M¹; QM F57277, left M¹; QM F57278, right M¹; QM F57279, right M¹; QM F57280, right M¹; QM F57281, right M¹; QM F57282, left M¹ (Table 2).

Description

Complete tooth row not known. M¹ and M² are isolated specimens, no specimen of M³ found to date. M¹ large, M² approximately half length of M¹ (Table 2). Furrow between lingual series and central series of cusps in M¹ and M². Buccal series of cusps reduced in M¹, all cusps inclined posteriorly.

M¹: Tooth elliptical with thin and uniform enamel rim around all cusps. Two small accessory cusplets present on anterior cingulum in holotype. With wear they become one very large accessory cusp, elongated posterolingually, sweeping back along lingual edge. T1 large and elongated, becoming more elongated with wear. Anterior edge of T1 sits posterior to T2, at half-way point of tooth. T1 orientated posteriorly with axis of cusp stretching posterolingually, parallel to single accessory cusp in specimens other than holotype. It merges with T2 at late stages of wear. T1 sup present on some specimens, situated on posterolingual edge of T1. It is small and circular, merging into T1 with wear. T2 of moderate size and subtriangular in occlusal outline. T3 very small and circular, sweeping slightly posteriorly from T2 in some specimens. T3 often connected to T2 by enamel rim, later merging with wear. T4 large and tear-shaped, increasing in size posteriorly with wear but never merging with T7 or T8. It only barely merges with T5, even at late stages of wear. Large size of T4 together with similarly sized T1 creates a bulge on lingual edge of tooth, enlarging width of otherwise slender tooth. Anterior edge of T4 sits posterior to the posterior edge of T5. T4 higher at posterior edge than anterior edge. Cusp posteriorly inclined, with axis running almost parallel to main axis of tooth. T5 large and subtriangular, orientated posteriorly. T6 circular, elongating anteroposteriorly with wear and merged with T5 at most stages of wear. Posterior edge of T6 sweeps posteriorly slightly from T5 in most specimens. Posterior edge of T4-6 complex mostly arcuate anteriorly, enclosing T7-9 complex, especially on lingual side. T7 indistinguishable from T8 in the holotype but very small and completely incorporated into T8 in other specimens. T8 large and circular, orientated posteriorly. It is the highest cusp with all others roughly uniform in height. T9 small and elliptical. Lower half of T9 connects to T8 at early stages of wear, becoming fully incorporated with further wear.

With three roots. Anterior root the largest of the three. It is circular and directed anteriorly from the accessory cusp and T2. Posterolingual root narrow and plunges vertically from T1 and T4. Posterior root of equal size with posterolingual root, is more circular and is elongate vertically from T8 and T9.

M²: Triangular in shape with broadest point along anterior edge. T1 circular and distinct, cusp directed posteriorly with occlusal surface inclined proximally. Deep furrows on buccal and posterior side of T1 separate it from other cusps and retains identity through wear. T2 and T3 absent. T4 large, elongated and tear-shaped, stretching posterolingually. Anterior edge of T4 sits posterior to posterior edge of T5. T4 posteriorly inclined, with occlusal surface facing proximally, similar to T1. T5 only slightly larger than T4 and subtriangular, connecting to T4 by its enamel rim and directed posteriorly. T6 absent. Posterior edge of T4-5 complex arcuate anteriorly, enclosing T8. T7 almost indistinguishable from T8 but indicated by a small bulge on the lingual edge of T8. T8 large and circular, directed posteriorly. Posterior edge arcuate posteriorly and delineates the most posterior edge of the tooth. No obvious indication of presence of T9. Remnant of furrow that marked its position present, indicating it has been wholly incorporated into T8. Enamel rim of cusps is variable, with T5 and T8 thicker than other cusps. All cusps of equal height and incline posteriorly at varying degrees, with T5 and T8 leaning posteriorly more than T1 and T4.

Roots not visible on only available specimen of M². Description has been gathered from alveoli in a specimen also preserving M¹ (QM F57275). M² has three roots. Lingual root very large and elongated, directed vertically. Anterobuccal root is circular, extends anteriorly, and is smaller than the lingual root. Posterobuccal root smallest of the three, elongated and extends vertically.

M³: no specimen known.

Information on dental arcade is limited. Large posterior palatal foramen extends distally from posterior of M¹. Zygomatic plate wide with posterior edge appearing almost straight but is slightly convex.

Taphonomy

Even though Riversleigh’s Rackham’s Roost Site and Floraville’s Site 5C represent vast differences in both mode of death and environment of preservation, similar skeletal elements have been preserved. Rackham’s Roost Site is interpreted to have been a Ghost Bat (Macroderma gigas) roost during the Early Pleistocene (Hand, 1996; Woodhead et al., in press) and specimens of Leggadina found there are thought to be the result of bat predation (Godthelp, 1997). Floraville’s Site 5C specimens are more likely to have come from marsupial predators, fossils of which have also been found at the site (Rich et al., 1991). There have been no complete skulls found at either site. The fractured cranial and post cranial elements found cannot be attributed to individual murine taxa due to overlaps in size and a lack of features known to separate them (H Godthelp, 2013, unpublished data).

The Rackham’s Roost assemblage contains only upper molars (upper = 28, lower = 0) which are all identifiable as belonging to the same species of Leggadina. They include complete molar rows, dental arcades and zygomatic plates. Site 5C specimens are dominated by lower molars but lack any molar rows (upper = 17, lower = 20). Upper molars all belong to the same species of Leggadina. Lower molars are observed to belong to the genus Leggadina but since they are not associated at all with the upper molars found, it is not possible to confidently identify them as belonging to the same species. Dental arcade and zygomatic plate information is fragmentary. An increased preservation of upper molars over lower molars is expected since the lower molars, attached to the mandible, have a greater chance of early disarticulation before preservation, whereas the upper molars are more likely to be retained in situ with the skull and post cranial bones for a longer period of time (Behrensmeyer, 1984). Nevertheless, it is important to note that the mandible tends to be stronger than the cranium, suggesting the large number of lower molars at Site 5C is the result of the lowers surviving the preservation process more readily than the uppers (Behrensmeyer, 1984). It is possible that sampling could have played a part in these results. The question then is whether further sampling at numerous places at Site 5C would increase the number of upper molars found. The only way to test this is through continued sampling. The Rackham’s Roost specimens on the other hand would have suffered little disturbance during the process of fossilisation as specimens would have been protected inside the cave until it eroded. This is the likely reason more complete molar rows have been found at this site, however this does not explain why so few lower molars have been found. Again this could be due to sampling (Lundelius, 2006).

The occlusal surface of molars from specimens found at Rackham’s Roost Site provides additional information on the age of individual animals through the degree of wear present on molars. The specimens collected from Rackham’s Roost are dominated by largely unworn occlusal features, indicating a large number of the specimens were juveniles. Macroderma gigas moves to different feeding roosts to take advantage of seasonal resources, and it is likely they followed the breeding cycles of its prey, explaining the dominance of juveniles in the sample (Tidemann et al., 1985).

Environmental impact

The early and middle Miocene in Australia was characterised by high levels of rainfall and the dominance of rainforest communities (Martin, 2006). As Australia moved from ‘greenhouse’ to ‘icehouse’ conditions in the later Miocene (10-5mya) the environment became increasingly arid and the biota needed to adapt (Dawson & Dawson, 2006). Environmental communities also changed during the Pliocene from rainforest dominated areas to mosaics of grassland and open woodland (Archer, Hand & Godthelp, 1991). The changing distribution and diversity of mammals in the Riversleigh World Heritage Area fossil deposits is evidence of these changes (Archer et al., 1989; Travouillon et al., 2009). It is likely that as these changes occurred, arid-type responses were produced in much of its fauna (Archer et al., 1998), as seen in the Alcoota assemblage in the Northern Territory which shows a marked change in biota present in the late Miocene and early Pliocene (Black et al., 2012). By the end of the Pleistocene period animals of all types were forced to adapt their diet and behaviour where possible in order to survive because great climatic fluctuations caused by over 20 cycles of glacial and interglacial periods resulted in unpredictability of resources (Archer et al., 1998; Martin, 2006).

Continent-wide climatic shifts during the Pliocene and Pleistocene were very fast in terms of evolutionary response time, requiring taxa to either adapt quickly, be resilient enough to survive, or to be lost entirely (Archer et al., 1998). One of the factors that characterises the success of rodents in Australia is their rapid speciation (Bush et al., 1977). Modern Leggadina species inhabit arid-environments in northeastern Queensland (L. lakedownensis) and a variety of areas through inland Australia (L. forresti) (Watts & Aslin, 1981). However, the environment of Southeast Asia during the Miocene, thought to be the originating point of Australian murids, was characterised by tropical rainforests which were slowly beginning to contract (Heaney , 1991). It is therefore likely that the genus Leggadina evolved from an ancestor which was not arid-adapted.

Species of Leggadina have reasonably complex upper molars in comparison to closely related taxa, for example, both Leggadina gregoriensis and L. macrodonta have an additional occlusal structure (furrows) that allows for increased precision during mastication, indicating the evolution and specialisation of their teeth for a predominantly granivorous diet (Herring, 1993; Evans et al., 2007). Similarly, the width of the zygomatic plate is a useful indicator of the kinds of food eaten by rodents, because width of the zygomatic plate increases with an increase in the size of the anterior deep masseter muscle used for pulverising food (Watts & Aslin, 1981; Satoh, 1997). The zygomatic plate in both fossil species is quite wide suggesting further specialisation for a predominately granivorous diet. When the fossil Leggadina species evolved these adaptations cannot be determined at the moment due to the lack of knowledge on both the timing and method of their dispersal to and within Australia, as well as appropriate morphological evidence for other Australian fossil species.

One particularly interesting feature distinguishing Leggadina macrodonta is the size of its teeth, particularly M¹ which is up to 18% larger than the M¹ of L. gregoriensis or the two modern forms. The increase in size of the teeth and occlusal structures could be due to a number of different factors. Larger teeth would be a useful adaptation for taking advantage of a wider variety of resources necessary for survival in a changeable climate. Alternatively, increased tooth size could represent specialisation for a more selective diet, again resulting from a changing environment. It is also possible the increase in size of the molars of L. macrodonta was due to an increase in overall body mass, with this particular species growing larger in order to compete against larger animals for resources, as well as becoming able to process low nutrient foods more easily and reduce water loss (Archer et al., 1998; Dawson & Dawson, 2006). Unfortunately it is not possible to calculate body mass of this species currently due to the absence of adequate lower molar data and a lack of long bones in the fossil assemblage relatable to this species (Hopkins , 2008). However, the question that remains is why did L. macrodonta develop exceptionally large molars, while molar size in L. gregoriensis remained more closely aligned with its modern relatives. Broader ecological evidence needs to be presented on changes in tooth structure in other species during the Plio-Pleistocene to make a more informed determination on tooth variation between L. macrodonta and L. gregoriensis.

Future work

Molar morphology has been an important tool for understanding the evolution of the Murinae and other rodent groups for over 100 years. At this point in time it is still essential for the description of new fossil species of Australian murids. However, to date there has been no comprehensive phylogenetic analysis based on morphology including both fossil and modern species. The leading analysis on morphological relationships using molar morphology was conducted over 40 years ago (Misonne, 1969). On the other hand, advances in molecular assessment of murid relationships have proliferated over the past 30 years (Baverstock et al., 1981; Pascale, Valle & Furano, 1990; Catzeflis, Aguilar & Jaeger, 1992; Watts et al., 1992; Jansa & Weksler, 2004; Steppan et al., 2005; Rowe et al., 2008; Nilsson et al., 2010; Schenk, Rowe & Steppan, 2013). An updated morphological phylogeny combined with molecular phylogenies would give a much more cohesive picture of Australian murid evolutionary history than using either alone (Wiens, 2004; Aplin, 2006).