Conservation implications of a new fossil species of hopping-mouse, Notomys magnus sp. nov. (Rodentia: Muridae), from the Broken River Region, northeastern Queensland

Abstract

The Australian hopping-mouse Notomys includes 10 species, eight of which are considered extinct, vulnerable, near-threatened or endangered. Here we report a new fossil species from the Broken River Region, northeastern Queensland. Notomys magnus sp. nov. is represented by craniodental material from late Quaternary cave deposits. It was a relatively large-bodied species of Notomys with a mass estimated to be ca 83 g. Notomys magnus sp. nov. is immediately distinguishable from the spinifex hopping-mouse (Notomys alexis), the northern hopping-mouse (Notomys aquilo), the fawn hopping-mouse (Notomys cervinus), the dusky hopping-mouse (Notomys fuscus), Mitchell’s hopping-mouse (Notomys mitchellii) and the big-eared hopping-mouse (Notomys macrotis) by its larger size (especially its longer upper molar crown length). Notomys magnus sp. nov. differs from the large-bodied Darling Downs hopping-mouse (Notomys mordax), long-tailed hopping-mouse (Notomys longicaudatus), short-tailed hopping-mouse (Notomys amplus) and broad-cheeked hopping-mouse (Notomys robustus) by possessing a unique first upper molar (M¹) morphology including relatively well-developed buccal cusps, cusp T1 prominently isolated from T4, a relatively narrow posterior loph and an incipient anterior accessory cusp. Fossils of N. magnus sp. nov. are found in association with remains of several arid-adapted taxa, including the plains mouse (Pseudomys australis), the northern pig-footed bandicoot (Chaeropus yirratji), and N. longicaudatus, possibly indicating that N. magnus sp. nov. was also arid-adapted. Dating of fossil deposits containing N. magnus sp. nov. demonstrates that it was extant in the mid-Holocene (ca 8.5 ka) so it may have been still extant at the time of European colonization but suffered extinction soon after, mirroring the fate of similarly arid-adapted contemporaneous taxa (Chaeropus yirratji and N. longicaudatus). Historical extinctions in Notomys are biased towards larger species (N. amplus, N. longicaudatus and N. robustus), and the discovery of N. magnus sp. nov. adds further to that list. Given the already high number of extinct and endangered species within Notomys, the discovery of another member that suffered geologically recent extinction has conservation implications for modern critical weight range mammals (including other species of rodents) that are particularly susceptible to extinction. Most historical extinctions of critical weight range mammals were in southern and central Australia, but the discovery of N. magnus sp. nov. suggests that species in the tropical north also were detrimentally affected.

Vikram Vakil [[email protected]], School of Earth and Environmental Sciences, The University of Queensland, Brisbane, Queensland 4072, Australia;

Jonathan Cramb [[email protected]], Queensland Museum, Brisbane, Australia;

Gilbert J. Price [[email protected]], School of Earth and Environmental Sciences, The University of Queensland, Brisbane, Queensland 4072, Australia;

Gregory E. Webb [[email protected]], School of Earth and Environmental Sciences, The University of Queensland, Brisbane, Queensland 4072, Australia;

Julien Louys [[email protected]], Australian Research Centre for Human Evolution, Griffith University, Brisbane, Queensland 4101, Australia.

Keywords:

• Muridae
• extinct
• Quaternary
• critical weight range
• Australia

AUSTRALIA’S critical weight range (CWR) species include many small to medium-sized (35–3500 g) (Burbidge & McKenzie 1989) mammalian taxa that have undergone localized or total extinction as a result of introduced predators (e.g., the feral cat Felis catus Linnaeus, 1758 and the red fox Vulpes vulpes Linnaeus, 1758) (Burbidge & McKenzie 1989, Cardillo et al. 2005), habitat loss (Murphy & Davies 2014, Woinarski et al. 2015) and other anthropogenic impacts (Johnson et al. 1989, 2002, Johnson & Isaac 2009). Amongst the factors responsible for the current biodiversity crisis (Pereira et al. 2010, Barnosky et al. 2011), predation by non-native predators and habitat destruction (Woinarski et al. 2015) are major threats faced by these faunas. Declines of CWR taxa have been greatest in arid/open regions and relatively low in closed habitats, such as rainforests (McKenzie et al. 2007). Anthropogenic climate change is predicted to compound the threats in the near future (Williams et al. 2003, Williams 2006, Solomon et al. 2007, Seebacher & Post 2015), and is already directly implicated in the recent extinction of one CWR rodent, the Bramble Cay melomys, Melomys rubicola Thomas, 1924 (Fulton 2017, Waller et al. 2017).

A better understanding of the response of CWR faunas to long-term environmental change can be achieved by examining the relatively recent (late Quaternary) fossil record, prior to European colonization (1788). Fossils of CWR taxa are abundant in late Quaternary fossil sites (e.g., Bilney et al. 2010) and the late Quaternary is known for high magnitude climatic oscillations (Kershaw et al. 2011). Consequently, CWR faunas were subjected to various climate-related environmental stresses, possibly compounded by impacts of human arrival in Australia during the Late Pleistocene. Late Quaternary mammal communities thus faced challenges similar, in part, to those they face today and those predicted to occur in the near future such as habitat loss, predation, and ongoing altering of fire regimes (Flannery 1990, 1994, Johnson 2006). Unless their pre-European presence in the geological record is examined for long-term trends through deep-time, modern conservation efforts risk being biased by distribution data acquired from either depauperate faunas or relict populations that may have only formed post-1788 (Fusco et al. 2016), thereby making it difficult to understand the actual environmental factors that may support or impact the survival of a given species.

There are two major groups of Australian murid rodents—the Old Endemics, belonging to the Tribe Hydromyini, and New Endemics, belonging to Tribe Rattini (Aplin 2006, Breed et al. 2020). A recent molecular study (Roycroft et al. 2022) suggested that the ancestors of the former arrived on the Australian mainland in the late Miocene–early Pliocene; that estimate is fairly consistent with fossil evidence that indicates arrival ca 4.46 Ma (Piper et al. 2006). The ancestors of the New Endemics probably arrived ca 0.5–1 Ma (Breed et al. 2020). Within the Old Endemics, the hopping-mouse, Notomys Lesson, 1842, comprises a monophyletic group (Roycroft et al. 2022). Notomys is a CWR taxon that includes 10 species, at least one-half of which are extinct (IUCN 2019, Alhajeri 2021). Two members of the genus are widespread with stable populations—Notomys alexis Thomas, 1922 and Notomys mitchellii (Ogilby, 1838) (IUCN 2019, Alhajeri 2021). Three species—Notomys aquilo Thomas, 1921, Notomys cervinus (Gould, 1853) and Notomys fuscus (Wood Jones, 1925)—are endangered, near threatened and vulnerable respectively (Musser & Carleton 2005, Breed & Ford 2007, IUCN 2019). Species of Notomys are characterized by their bipedal hopping gait, very long hind legs, and elongated, tufted tails (Tate 1951, Nowak & Paradiso 1983, Mahoney et al. 2007, Alhajeri 2016, 2021). These characters have evolved in parallel in multiple groups of small mammals in arid environments (e.g., Dipodomyinae, Dipodinae, Antechinomys Krefft, 1867), suggesting that these features are adaptations to arid habitats (Alhajeri 2021).

Notomys includes five small-bodied species, ranging in weight from 30 to 60 g (N. alexis, N. aquilo, N. cervinus, N. fuscus and N. mitchellii), two intermediate-sized species for whose masses have not been recorded directly (Notomys macrotis Thomas, 1921 and Notomys mordax Thomas, 1922), and three comparatively large-bodied species (ca 100 g; Notomys amplus Brazenor, 1936 and Notomys longicaudatus [Gould, 1844]); Notomys robustus Mahoney et al., 2007 is likely to have had a similar weight). Extant hopping-mice are omnivorous, with diets consisting of seeds, insects and plant material (Watts & Kemper 1989, Turner 2004, Alhajeri 2021). The rapid and catastrophic decline of some species of Notomys is incompletely understood (Mahoney et al. 2007) and this may hamper conservation efforts for surviving species.

The Quaternary fossil record provides crucial information on the past distribution of species of Notomys and the timing of their decline. Fossils and subfossils of multiple species of Notomys have been found in various cave surficial deposits, usually in owl pellets under rocky overhangs or within crevices (Smith 1977, Lundelius 1983, Baynes 1984, Copley et al. 1989, Robinson et al. 2000). In Queensland, the oldest fossils of the genus are from a Chibanian deposit (170–205 ka) at Mount Etna (Hocknull et al. 2007). Phylogenetic analysis indicates that recent extinctions amongst various CWR murine taxa were not restricted to certain clades but are instead strongly correlated with body size (Roycroft et al. 2021). Here, we investigate body size in Notomys and describe a new large-bodied species from late Quaternary cave deposits of the Broken River region of northeastern Queensland.

Geological setting

The Broken River region is located just south of Greenvale in northeastern Queensland (Fig. 1). Deposits in the area have traditionally been sampled for Palaeozoic fossils (see Henderson et al. 2013 and references therein), with more recent investigations centred on the Quaternary vertebrate deposits in local caves (Hocknull 2005, Cramb & Hocknull 2010, Price et al. 2017, 2020, Travouillon et al. 2019).

Figure 1. Locality map, highlighting the study site and other sites mentioned in text (modified from Breed & Ford, 2007 and the Atlas of Living Australia (https://www.ala.org.au/). the type locality for Notomys longicaudatus is the Moore River in southwestern Western Australia.

The Broken River Province is divided into two sub-provinces—the Camel Creek Sub-province in the east and the Graveyard Creek Sub-province in the west. The Graveyard Creek Sub-province consists of folded Silurian-Mississippian siliciclastic and carbonate rocks that occupy the southwestern region of the Broken River Province (Henderson et al. 2013). White (1959, 1965), Withnall (1989) and Withnall & Lang (1993) recognized two subdivisions within the Graveyard Creek Sub-province: the siliciclastic Graveyard Creek Group and the carbonate-rich Broken River Group. Fossil specimens described here come from deposits in two caves located in limestones of the Graveyard Creek Sub-province: Beehive Cave and Tripot Cave. Beehive Cave occurs in the southern part of the extensively karstified upper Silurian to Lower Devonian Jack Formation in the Graveyard Creek Group and the deposits sampled here were dated to the Holocene (ca 8.5 ka) (Price et al. 2020). The outcrop of Jack Formation is tilted ca 90° and contains extensive rillenkarren (Price et al. 2020). Beehive Cave is joint-controlled, containing tall, narrow passages and caverns (Price et al. 2020). Vertebrate fossils were previously described from Beehive Cave by Price et al. (2020). Tripot Cave is also joint-controlled (see also Chillagoe Caving Club 1988) and occurs in the Middle Devonian Dosey Limestone within the Wando Vale Subgroup of the Broken River Group. The Dosey Limestone consists of bioclastic limestone, calcareous quartzose sandstone and siliciclastic mudstone (Henderson et al. 2013). Some vertebrate fossils from Tripot Cave (previously, but erroneously, referred to as ‘Dodgey’s Cave’) were mentioned by Hocknull (2005) and Cramb & Hocknull (2010), although the stratigraphic relationships to the Notomys-bearing deposits considered here are unclear. The present samples were recovered from unlithified sands in a small passage immediately to the left of the main cave entrance. Preliminary U-series dating of associated flowstones within the deposit suggests that the Tripot Cave specimens described here are Middle Pleistocene in age; these are currently the oldest known dated vertebrate assemblages within the caves of the Broken River region (see also Price et al. 2017).

Materials and methods

Fossil breccias from the Broken River cave sites were collected as part of a larger project investigating the Cenozoic emergence of modern faunas and ecosystems in the region. Lithified breccias collected from Beehive Cave were acid-digested in the Palaeo Lab of The University of Queensland using dilute (<5%) acetic acid. Sediment residues were then wet-sieved using mesh-sizes of 10 and 1 mm, thus recovering numerous fossils of vertebrate taxa (see Price et al. 2020 for additional details). The unlithified sediments of Tripot Cave were excavated using hand tools and bagged on site before sieving as per samples from Beehive Cave.

Identification of Notomys fossil specimens was made through comparison with published descriptions (e.g., Thomas 1921, Tate 1951, Mahoney et al. 2007) and specimens in museum collections (see Supplemental Data). Fossil specimens were measured using Craftright 150 mm digital callipers (China) and were photographed using a Pro 12MP camera system with an ultra-wide f/2.4 aperture, 120° field of view (California, USA), and a Hitachi TM3030 scanning electron microscope (Chiyoda, Tokyo, Japan) at the School of Earth and Environmental Sciences, The University of Queensland. Measurements included the molar width (W), length (L) and total molar row length (TMR). Molar width was measured perpendicular to the long axis of the tooth.

Valid Australian murine species follow the Australasian Mammal Taxonomy Consortium list (AMTC 2021) Upper molar terminology (Fig. 2) and dental numbering follow Musser et al. (2005) except that ‘labial’ has been replaced by ‘buccal’, for ease of comparison with other published descriptions of Australian murines. Upper molar cusps are labelled T1 to T9 (rather than the homologous cusp terminology of Wood & Wilson 1936) for the same reason.

Figure 2. Diagrammatic representation of M^1-3 of Notomys magnus sp. nov., showing Notomys molar terminology used throughout the text. Cusp terminology follows Musser et al. (2005).

Molar row length and body mass are tightly correlated (Freudenthal & Martín-Suárez 2013). Therefore, in order to estimate the approximate mass of our fossil Notomys, a bivariate plot of natural logarithms of mean mass as a function of the mean molar row length was plotted for species for which weight data, or estimated weights, were available in the literature (Notomys amplus, Notomys longicaudatus, Notomys cervinus, Notomys fuscus, Notomys aquilo, Notomys mitchellii) (Supplemental Data). Data were transformed to natural logarithmic values following Legendre (1989) and Freudenthal & Martín-Suárez (2013) to account for comparison of a linear with a three-dimensional parameter (body mass). The approximate mass of fossil Notomys was calculated on the basis of the resulting regression. Conversion of the natural logarithmic values to grams may be skewed because the logarithmic values are normally distributed around the regression line and the unlogged values therefore have a positively skewed distribution (Myers 2001, Freudenthal & Martín-Suárez 2013). A correction factor (CF) was therefore applied using the equation: (1) $$\text{CF}={\text{e}}^{\left(\text{SEE2}\right)/\text{2}}$$(1) where e is the antilog and SEE is the standard error of estimate (Smith 1993, Myers 2001, Freudenthal & Martín-Suárez 2013). The calculated mass obtained from the regression was multiplied by CF to correct for the skew.

Institutional abbreviations

AM, Australian Museum (‘M’, mammal specimen), Sydney, Australia; NMV, Melbourne Museum, Museums Victoria, Melbourne, Australia; QM, Queensland Museum (‘F’, fossil specimen; ‘JM’, mammal specimen; ‘L’, locality register), Brisbane, Australia.

Systematic palaeontology

RODENTIA Bowditch, 1821

MYOMORPHA Brandt, 1855

MUROIDEA Illiger, 1811

MURIDAE Illiger, 1811

MURINAE Illiger, 1811

Notomys Lesson, 1842

Type species

Notomys mitchellii (Ogilby, 1838).

Remarks

Watts & Aslin (1981) listed the following skull characters as diagnostic for Notomys: lingual cusps better developed than buccal cusps; anterior half of zygomatic arch broadened and abruptly constricted; anterior edge of zygomatic plate concave, commonly with prominent spine; and moderate to large bullae. The presence of well-developed lingual cusps relative to the buccal cusps is a consistent feature in all species of Notomys, while development of the buccal cusps is variable (see Mahoney et al. 2007). The concave anterior edge of the zygomatic plate is characteristic of many species of Notomys and Pseudomys Gray, 1932 (Watts & Aslin 1981); however, in Notomys, the radius of the circle described by the anterior edge of the zygomatic plate generally decreases towards the spine at the dorsal end, i.e., the spiral tightens towards the top. Additionally, the molar row in Notomys shows slight vertical arching and the dorsal surface of the maxilla above the molars generally shows a flat area above M² and M³. Although some characters possessed by the specimens described here overlap between species of Notomys and Pseudomys, the weight of evidence is in favour of them being assigned to Notomys based on the characters discussed above and the consistency of the preserved characters amongst all described specimens.

Notomys magnus sp. nov.

Figs 3–5

Figure 3. Notomys magnus sp. nov. A, QM F55835. Holotype. Left maxilla in left lateral view. B, The same in dorsal view, highlighting the proximal part of the zygomatic arch. Red arrows point to the broken part of the zygomatic arch, indicating the presence of additional missing bone from which it is inferred that the overall thickness of the anterior zygomatic arch would have been thicker.

Figure 4. Notomys magnus sp. nov. A, QM F55835. Holotype. Left maxilla with M^1-3. B–D, SEMs of M¹, M² and M³ of holotype QM F55835. E, QM F55832. SEM image of left upper first molar (M¹); F, QM F60725. Left M¹. G, H, QM F60723. Associated left maxilla fragments with M¹ (G) and M^2-3 (H). I, J, QM F60724. Associated left maxilla fragments with M^1-2 (I) and M³ (J). Red bars indicate the positions of the incipient anterior lingual cusps on M1s. All scale bars equal 1 mm.

Figure 5. Comparison between the maxillae of two Notomys species. A, N. magnus sp. nov. from the Broken River region. B, Notomys longicaudatus from Mount Etna (see Fig. 1). Note the well-developed buccal cusps of N. magnus and the relatively narrow posterior lophs of M^1-2 of N. magnus, compared with those of N. longicaudatus. Scale bars equal 1 mm.

2020, Notomys sp. 2, Price et al., p. 201, fig. 7E.

Diagnosis

Relatively large-bodied species of Notomys with large, stout upper molars and an average M^1-3 length of 6.6 mm (Tables 1,2). Distinguished from other Notomys species by a combination of the following characters (see Figs 2,4B): (1) a deep cleft between the bases of cusps of T1 and T4 of M¹, almost completely separating T1 from T4; (2) relatively narrow T8–9 complex (posterior loph) on M^1-2; (3) the maximum dimension of T4 is greater than one-third of the overall maximum dimension of the middle loph (T4 + T5–T6 complex) when measured at the bases of the cusps on M¹; (4) a small, incipient accessory cusp on the anterolingual surface of the anterior loph of M¹; (5) relatively well-developed buccal cusps on M^1-2 (Fig. 5A).

Table 1. Molar (M^1-3) and zygomatic plate measurements for the examined specimens of Notomys magnus sp. nov.

		M^1		M^2		M^3		TMR	ZP
Specimen #	Description	L	W	L	W	L	W	L	L
QM F55835	Left maxilla	3.01	2.18	2.11	1.92	1.6	1.13	6.57	4.46
QM F55832	Left M^1	2.98	2.17
QM F60723	Left maxilla	3.05	2.1	2.15	1.88	1.6	1.17	6.71
QM F60724	Left maxilla	2.96	2.1	2.04	1.76	1.61	1.28	6.52	4.18
QM F60725	Left M^1	3.08	2.36

Abbreviations: L, length; W, width; TMR, total molar row; ZP, zygomatic plate. All measurements are in mm.

Table 2. Measurements of range, mean and standard deviation for upper molar dimensions for Notomys magnus sp. nov.

n	Description	Range	Mean	STDV
5	M^1 L	2.96–3.08	3.016	0.049295
5	M^1 W	2.1–2.36	2.182	0.106395
3	M^2 L	2.04–2.15	2.1	0.055678
3	M^2 W	1.76–1.92	1.853333	0.083267
3	M^3 L	1.6–1.61	1.603333	0.005774
3	M^3 W	1.4–1.53	1.45	0.07
2	ZP L	4.18–4.46	4.32	0.19799
3	TMR	6.52–6.71	6.6	0.098489

Abbreviations: n, number of specimens; L, length; W, width; TMR, total molar row; ZP, zygomatic plate. All measurements are in mm.

LSID of new species

LSIDurn:lsid:zoobank.org:pub:8518F81F-2E8D-435E-917F-28ED7C610515

Etymology

The specific epithet is derived from Latin; ‘magna’ meaning ‘large’.

Holotype

QM F55835, a left maxilla (Figs 3, 4A–D, 5). This specimen was excavated from the Tripot Cave Surface Deposit QM L1091S.

Referred material

QM F60722 right maxilla (see Price et al. 2020) from the Beehive Cave deposits; QM F60723 left maxilla, QM F60724 left maxilla (Fig. 4G–J) from Tripot Cave Unit 1; QM F60725 left M¹ (Fig. 4F) from Tripot Cave Unit 2. QM F55832 left M¹ (Fig. 4E) from QM L1091S.

Type locality, unit and age

Tripot Cave and Beehive Cave on the Broken River in northeastern Queensland. QM L1091S and the Tripot Cave Unit 1/Unit 2 deposits are considered Middle Pleistocene in age (Price et al. 2017). Beehive Cave correlates with the Early Holocene Marine Isotope Stage (MIS) 1, which is chronometrically dated to ca 8.5 ka (Price et al. 2020).

Description

Maxilla

Anterior half of zygomatic arch deep, although dorsal edge damaged in all specimens so full extent not assessed; nevertheless, its presence is inferred from the jagged dorsal edge (Figs 3, 4A, I). Anterior edge of zygomatic plate concave, with length (from the centre of the anterior edge of the plate to its posterior edge) of 1.19 mm and 4.3 mm respectively (Table 1). All specimens missing anterodorsal corner of zygomatic plate, so presence of zygomatic spine unknown. Posterior edge of zygomatic plate curves at slight angle into zygomatic arch in holotype specimen QM F55835 but more evenly in QM F60724, so feature possibly variable. Anterior palatal foramen narrow, with relatively straight buccal side. The posterior end of the anterior palatal foramen is lingual and slightly anterior to the T1 of M¹. M¹ occupies approximately 45% of the molar row, followed by M² and M³ at approximately 32% and 24%, respectively. Additionally, the maxillary portion of the palate shows a longitudinal ‘step’ between the molar teeth and the centre line of palate. This feature is variable and is shared by many species of Notomys and Pseudomys. Additionally, the anterior face of the anterior loph is inclined approximately at 60° to molar axis when seen in lateral view; middle loph inclined at approximately 45°; and posterior loph very weakly inclined.

M¹

Crown proportionally broad in occlusal view with well-developed, bulging lingual cusps and posterobuccally tapering T8–T9 complex (posterior loph). T1 subcircular in occlusal view and separated from T2 by a deep groove, particularly evident on anterolingual surface. T1 tapers buccally towards lingual point of T2. T1 posterolingual to T2, fused to T2 at base. T2 semicircular in occlusal view, bulging anteriorly, forming majority of median part of anterior loph. Highest point of T2 along the midline, on molar row axis. T3 well-defined, small, fused to T2. T3 rounded buccally. Small, incipient accessory cusp present on anterolingual end of anterior loph, positioned directly opposite to cleft between T1 and T2. No specimens have evidence of wear on the incipient accessory cusp, so may not be functionally occlusive. T4 positioned directly posterior of T1, tear-shaped to subtriangular in occlusal view. Trough between bases of T1 and T4 very deep, little sign of any connection between bases of T1 and T4. T4 well-defined, attached via very thin buccal tail to lingual edge of T5. Maximum dimension of T4 greater than one-third of the overall maximum dimension of the middle loph (T4 + T5–T6 complex) when measured at the bases of the cusps on M¹. T4 and T5 separated by a cleft on anterior face of middle loph, less developed than cleft separating T1 and T2. T5 broadest cusp in the middle loph. T5 lies posterobuccally relative to T2. T6 small, fused to T5. T6 commonly buccal and slightly posterior of T5, although one specimen (QM F55832) (Fig. 4E) has T6 directly buccal of T5. T6 rounded buccally. T6 slightly larger than T3 and extends further buccally, relative to T3. Middle and posterior lophs separated by a deep groove, which becomes much shallower immediately posterior of T4. T7 absent, although a tiny T7-like lingual extension of the posterior loph (T8–T9) is seen in QMF 55832 and QM F60725; this feature is variable as it is not seen in other specimens. T8–T9 complex small. Posterior loph sub-lenticular in occlusal view, with anterior edge terminating into a peak, approximately along molar axis, while posterior edge is posterobuccal with respect to the molar axis. Posterior edge of posterior loph has posterobuccal end directly appressed against anterobuccal edge of median loph of M² while posterolingual end is directly appressed against anterobuccal edge of T1 of M². Posterior loph narrower than middle loph. T9 well-defined but joined to T8, even in unworn specimens. T9 sub-triangular in occlusal view, situated slightly posterolingually relative to T6. M¹ has three roots; anterior, lingual and posterobuccal. Anterior and lingual roots elongated relative to posterobuccal root.

M²

Slightly narrower than M¹ (Tables 1, 2). T1 well-developed, nearly circular in occlusal view. T1 separated from T5 by deep cleft, although not as isolated in profile as T1 on M¹. Lingual cusps on M² do not bulge at bases to same extent as those on M¹. T2 and T3 absent. T4 similar in size to T1, subtriangular in outline and situated posterobuccally of T1, separated by deep cleft. T4 rounded/sub-rounded lingually, tapering buccally to fuse directly with lingual side of T5. T4 posterolingual of T5. T5 is triangular in occlusal view with anterior edge forming apex along molar axis while posterior edge is slightly concave but relatively straight. T6 well-defined, subtriangular in occlusal view, situated directly buccally of T5. T6 rounded/sub-rounded buccally and tapers lingually to fuse directly with buccal side of T5. Apex of T6 at same level as that of T4 in occlusal view. Posterior base of T6 anterior relative to posterior base of T4. Posterior loph (T8–T9 complex) lenticular in occlusal view, subequal in size to T4 and oriented posterobuccally relative to T4. Posterior loph situated directly posterior to T5, along molar axis. M² has three roots: anterolingual, anterobuccal and posterobuccal. Anterior and lingual roots elongated relative to posterobuccal root.

M³

Crown broad, sub-triangular in occlusal view. Smaller than M² (Tables 1, 2). T1 well-developed, subtriangular in occlusal view. T1 completely separated from middle loph, isolated from anterolingual bases of T4 and T5 by deep groove. T1 rounded lingually, tapers buccally, terminating just before posterolingual tip of posterior loph of M². Apex of T1 elevated above plane passing through T1 and apex of T6 in occlusal view. T1 subequal or slightly smaller than posterior loph (T8–T9). T2 and T3 absent. T4 slightly smaller than T1. T4 posterolingual of T5–6 complex. T4 well-rounded lingually and joined to T5 buccally without tapering. T5 and T6 virtually indistinguishable as distinct cusps, T5–T6 complex joins buccal end of T4 lingually. T5–T6 complex subtriangular in occlusal view. T5–T6 complex rounded buccally, slightly tapers lingually to continue into T4. Apex of T5–T6 complex raised relative to apex of T4 in occlusal view, lies in same plane as posterobuccal end of T1. Posterior loph (T8–T9) isolated from middle loph by prominent groove which is slightly constricted along posterolingual end of T4 and widens along posterolingual end of T6. Posterior edge of posterior loph perpendicular to molar axis. M³ has three roots; anterolingual, anterobuccal and posterobuccal.

Remarks

Notomys magnus is placed within the genus Notomys on the basis of the combination of the following characters. (1) Anterior half of zygomatic arch is inferred to have been broadened. The anterior half of the zygomatic arch is partially preserved in two specimens, QM F55835 and QM F60722 (UQPL27 of Price et al. 2020). The latter specimen has been further fragmented since the publication of Price et al. (2020); nevertheless, in both specimens, the anterior half of the zygomatic arch is broken along the dorsal margin (Fig. 3). Although the total depth of the anterior half of the zygomatic arch is thus unknown, it is inferred that it is deeper than that of larger-bodied species of Pseudomys such as the long-tailed mouse (Pseudomys higginsi Trouessart, 1897) (see Driessen & Rose 1999). (2) Lingual cusps of upper molars better developed than buccal cusps. (3) Concave anterior edge of the zygomatic plate.

Additionally, some features of N. magnus are consistent with other species of Notomys. (1) The slight vertical arching of the molar row is seen in the holotype specimen QM F55835 in lateral view (Fig. 3A). (2) The presence of a flat area along the dorsal surface of the maxilla above the molars is seen in QM F55835 (Fig. 3A, B). (3) The tightening of the curvature of the anterior edge of the zygomatic plate towards the maxillary spine is clearly seen in QM F55567 (Fig. 5B) and a similar condition is seen in QM F60724 (Fig. 4I), although the full extent of the spine is not preserved. In specimens where this feature has not been preserved, their assignment to Notomys magnus is based on shared dental synapomorphies as discussed below.

Notomys magnus can be distinguished from all other members of the genus. Notomys magnus differs from Notomys alexis, Notomys aquilo, Notomys cervinus and Notomys fuscus by having a relatively large crown length where M^1-3 >6.0 mm (see Mahoney et al. 2007 for dental measurements of N. aquilo, N. cervinus and N. fuscus). Notomys magnus further differs from these small-bodied species by having a proportionately narrower palate, well-developed buccal cusps on M^1-2, proportionately narrower posterior loph (T8–T9 complex) on M¹ and an incipient anterior accessory cusp on M¹. Notomys magnus differs from Notomys macrotis in having a larger crown length for M^1-3 (the M^1-3 crown length for two skulls of N. macrotis are 5.5 mm and 5.3 mm) (Mahoney 1977), better-developed buccal cusps on M^1-2, a proportionately narrower posterior loph and an incipient anterior accessory cusp on M¹. Notomys magnus can be distinguished from Notomys mitchellii on the basis of its M^1-3 crown length (the latter has an M^1-3 crown length of 5.0–5.4 mm in both sexes, with a mean of 5.2 mm), better-developed buccal cusps on M^1-2, a proportionately narrower posterior loph and an incipient anterior accessory cusp on M¹ (Mahoney et al. 2007). Notomys magnus can be distinguished from Notomys mordax by the presence of a small, incipient accessory cusp seen on M¹ as compared to the very prominent one seen in N. mordax (Mahoney 1977, Mahoney et al. 2007). Additionally, the accessory cusp of the holotype of N. mordax is worn, whereas the accessory cusp of N. magnus lacks any wear on all specimens. This suggests a lack of occlusion between the incipient accessory cusp and lower teeth during chewing (at least prior to heavy molar wear), further differentiating N. magnus from N. mordax. Notomys magnus differs from Notomys robustus by the absence of a close association between cusps T1 and T4 of M¹ (Mahoney et al. 2007; see their fig. 4a). Notomys magnus differs from Notomys longicaudatus and Notomys amplus in possessing a combination of the following characters: a more prominently bulging T1 which is almost isolated from T4 and a relatively narrow posterior loph (T8–T9 complex). This is important in distinguishing N. magnus from N. longicaudatus, since N. longicaudatus is the only other species of Notomys known from fossil sites in eastern Queensland (see Price et al. 2020). The presence of well-developed buccal cusps on M^1-2 and a proportionately narrow posterior loph on M^1-2 in N. magnus therefore serves to readily distinguish it from N. longicaudatus (Fig. 5). The incipient accessory cusp on M¹ in N. magnus is present on all examined material and is therefore one of the diagnostic features of the taxon. Accessory cusps are variably present in other species of Notomys (e.g., N. longicaudatus), but are commonly absent in most species. One possible exception is N. mordax, the holotype (and only published specimen) of which possesses a well-developed accessory cusp (Mahoney 1977, Watts & Aslin 1981). The prevalence of this character across that species is currently unknown, pending discovery of more specimens. Regardless, the accessory cusp of N. mordax differs from that of N. magnus by being larger and positioned anteriorly, rather than adjacent to the cleft between T1 and T2. Another unique feature is the prominent isolation of T1 from T4 on M¹. This is unlike other Notomys species where the indent is prominent but does not lead to the almost complete isolation of the bases of T1 and T4 with a conspicuous gap. The posterior loph (T8–T9 complex) of M¹ in N. magnus has the smallest area amongst all Notomys species. M³ of N. magnus is smaller in size relative to M², although it is not as reduced as in N. longicaudatus (Fig. 5B) and certain larger-bodied Pseudomys species (like Pseudomys australis Gray, 1832, P. gouldii [Waterhouse, 1939], P. higginsi). The M³ of N. magnus is nevertheless comparable to the relative size of M³ in N. amplus. The combination of diagnostic autapomorphic M¹ features such as the presence of the incipient accessory cusp, a prominent and isolated T1 and a small posterior loph (T8–T9) together with the overall large total molar row length (M^1-3) clearly merit recognition of N. magnus as a new species. Recent phylogenies (see Smissen & Rowe 2018, Roycroft et al. 2021) for modern Notomys show a split between the clade that includes N. cervinus, N. macrotis and N. longicaudatus and all the other species. Alhajeri (2021), however, suggested, on the basis of size alone, that a closer phylogenetic association existed amongst the three largest species (N. amplus, N. longicaudatus and N. robustus) than with the smaller species. Craniodental characters further hinted at the possible closer phylogenetic association between N. macrotis, N. mordax and the extant smaller species (Watts & Aslin 1981, Ford 2006, Alhajeri 2021). This contrasts with the results of molecular studies that suggest that body size is not a phylogenetically informative character and that larger-bodied species of Notomys occur in two separate clades within the genus (see Roycroft et al. 2022). Owing to the dichotomy between the results from palaeontological and molecular data, no further comment can be made at this point on the phylogenetic placement of N. magnus.

Body mass estimate

A bivariate plot of natural logarithmic values of body mass measurements against total upper molar row length (Fig. 6) shows a strong positive correlation between body mass and molar row length (R² = 0.9112). The regression equation obtained is: (2) $$\text{ln}\left(y\right)=\text{2}.\text{5837 ln}\left(x\right)–0.\text{4544}$$(2) where ln (y) = ln (mean mass in grams), the y-axis coordinate; ln (x) = ln (mean total molar row in mm), the x-axis coordinate.

Figure 6. Plot of log mean mass against log mean total molar row (length) to estimate the mass of Notomys magnus sp. nov. Data for Notomys cervinus and Notomys fuscus from Tate (1951), Notomys aquilo, Notomys mitchellii and Notomys amplus from Mahoney et al. (2007) and Notomys longicaudatus from Thomas (1921). Horizontal dashed line indicates the lower limit of the CWR (see text). Dental measurements for Notomys alexis were not clearly reported in literature and since it is a small-bodied species, it is likely to cluster with other smaller-bodied species and its absence therefore does not have any major effect on the regression line of the graph and the estimation of the mass of N. magnus.

With a mean upper molar row length of 6.6 mm, Notomys magnus yields a ln (y) value of 1.92, which equates to a mass of ca 83.2 g. The calculated SEE is 0.05 and CF is 1.0, so no major correction is required. Notomys magnus is therefore a large-bodied Notomys species, clearly separated from the smaller/intermediate-sized members (Fig. 6). The lower end of the CWR range (horizontal dashed line) at 35 g (ln value of 3.55 on the y-axis) places all examined Notomys species within the CWR group (see Fig. 6; Supplemental Data). The upper end of the CWR range (ca 3500 g) exceeds the maximum limit of the y-axis and is therefore omitted.

Discussion

Notomys magnus is a large-bodied Notomys species (ca 83 g) that occurs in the Broken River fossil deposits alongside remains of arid-habitat adapted taxa like the Northern Pig-footed Bandicoot, Chaeropus yirratji Travouillon Simoes, Miguez, Brace, Brewer, Stemmer, Price, Cramb & Louys, 2019, the long-tailed hopping-mouse, Notomys longicaudatus, and the plains mouse, Pseudomys australis (Price et al. 2020). The presence of arid-zone taxa in the Broken River region suggests open habitats in the area, so the most parsimonious interpretation of the palaeoecology of N. magnus is that it also inhabited open-habitats. This seems reasonable since most extant Notomys species are arid-adapted (Mahoney et al. 2007). As noted by Price et al. (2020), it remains unclear whether open habitats were more common than wooded areas in the region or if the abundance of open-habitat taxa is more indicative of a strong feeding bias on the part of the owls. However, as owls normally hunt within a 4 km radius of their roosting sites (see Comay & Dayan 2018), it is likely that open habitat environments existed proximally to the caves that yielded the fossils.

The deposits yielding N. magnus are late Quaternary in age. The temporal span of N. magnus is at least middle Pleistocene to early Holocene (ca 8 ka). It remains uncertain if its local distribution was continuous through the Late Pleistocene; more excavations coupled with dating will be necessary to establish this.

The deposit in Beehive Cave is ca 8.5 ka (Price et al. 2020), and the comparatively stable climate of the Holocene, relative to the preceding glacial and deglacial, may suggest that N. magnus persisted locally after that time. It is unknown if it was extant at the time of European colonization. It is plausible that it was extant after 1788 but never documented as a living animal post-contact. If that is the case, such a scenario mirrors that of the Capricorn rabbit-rat, Conilurus capricornensis (Cramb & Hocknull 2010), and Notomys robustus (Mahoney et al. 2007), both described on the basis of remains from owl roosts and younger cave surface deposits. Additionally, C. capricornensis co-occurs with N. magnus in the Broken River cave deposits. Again, more sampling in the region is necessary to determine when it became extinct.

In their study of cave deposits from the southern periphery of the Kimberley region of Western Australia, Start et al. (2012) mentioned two undescribed species of Notomys. While they stated that the specimens are substantially larger than Notomys alexis (which occurs in the sandy deserts far south of their study site), no morphological description of the specimens was given, and it was noted that the species within Notomys are historically unknown from the Kimberley tropics (Start et al. 2012). Importantly, these records indicate the likely extinction of previously undocumented species of hopping-mice in tropical Australia, a proposition reinforced by our description of N. magnus. The modern decline and/or extinction of numerous species of mammals in arid regions of southern Australia is well documented (Woinarski et al. 2015, 2019), but the hopping-mice suggest that similar declines of dry tropical species also may have occurred.

Given the already large number of extinct and endangered species within the genus, the discovery of yet another extinct member raises questions about the susceptibility of species of Notomys to environmental stresses and their consequent fate through time. With an estimated mass of approximately 83 g, N. magnus is one of the larger-bodied species of Notomys, alongside Notomys amplus, N. longicaudatus and N. robustus, all of which are extinct (Fig. 6). Notomys amplus, N. robustus and N. longicaudatus all went extinct post-1788 (Mahoney et al. 2007, Alhajeri 2021). The primary reason for their extinction is suspected to be over-predation from introduced feral predators, principally cats and foxes (Woinarski et al. 2015, 2019, Mahoney et al. 2007). While it is highly likely that the increasing pressure from introduced feral predators was chiefly responsible for several pulses of extinctions across the Australian mainland during the last 200 years (Johnson 2006) (although see Mansergh et al. 2022 for a differing view), it remains unclear whether it was responsible for the extinction of N. magnus.

The stratigraphic record of the various species of Notomys remains unclear owing to the scarcity of well-preserved museum specimens (Alhajeri 2021) and poorly resolved ages of the deposits in which they are found. Most fossil specimens of Notomys are found in surficial deposits that are either not dated or are assumed to be relatively young (<200 years) (see Mahoney et al. 2007 for N. robustus and Start et al. 2012). Additionally, many species of Notomys are absent from the fossil record, although this may be due to lack of adequate sampling. Some studies have used modern dating techniques and demonstrated that species such as N. longicaudatus has a temporal record extending back to the middle Pleistocene (ca 165 ka) (see Hocknull et al. 2007, Price et al. 2020).

Most importantly, a major finding of this study is that our ‘baseline’ understanding of tropical ecosystems in northern Australia is patchy and incomplete. The discovery of N. magnus, along with those mentioned by Start et al. (2012) and C. yirratji (Travouillon et al. 2019), are indicative of the decline (and possibly collapse) of small mammal assemblages in the dry northern tropics of Australia. This is also true of many other pre-European extinct taxa that fall within the CWR range (such as murids, dasyurids and peramelids) as documented in Hocknull (2005), Hocknull et al. (2007) and Price et al. (2020). This raises an important question. If the CWR taxa are so prone to decline today (e.g., Burbidge & McKenzie 1989, Fisher et al. 2014), then could they not have been predisposed to extinctions in the past as well? Despite mounting evidence demonstrating extinctions of small-bodied vertebrates during the Quaternary (e.g., Archer & Baynes 1972, Archer 1976, 1981, Archer & Bartholomai 1978, Godthelp 1997, Price 2002, Martinez 2010, Cramb & Hocknull 2010, Klinkhamer & Godthelp 2015, Cramb et al. 2018, Travouillon et al. 2019), there still appears to be an over-emphasis placed on understanding the fate of ‘megafaunal’ taxa to the exclusion of small-bodied, less ‘charismatic’ taxa (e.g., Flannery 1990, Miller et al. 1999, Roberts et al. 2001, Barnosky et al. 2004, Johnson 2006, van der Kaars et al. 2017). The general under-representation of these CWR mammals in extinction studies ignores the long-term trends that have been responsible for shaping extant populations and distributions (Archer et al. 2019). For instance, the discovery of an extinct taxon like N. magnus adds not only to our knowledge of the pre-European diversity of the hopping-mice but also strengthens the claim that deep-time extinctions have not just impacted upon the ‘megafauna’. Many questions about the fate of Notomys magnus remains: was it widespread in its geographical distribution prior to middle Pleistocene or was it always confined to isolated populations in the Broken River region? If N. magnus survived past ca 8.5 ka into the time of European colonization, could the effects of other factors, such as feral predators, changed land-use patterns, and habitat loss have finally driven it to extinction?

Our current lack of deep-time knowledge about what pre-European northern Australian mammalian assemblages looked like is a major issue for present-day conservation. Even today, conservation of CWR taxa (e.g., rodents) is facing obstacles. Firstly, indifference prevails because large-bodied and more ‘charismatic’ species receive more attention (not unlike the issues with the fossil record, as highlighted above) (see also Lidicker et al. 2007). Secondly, complacency arises because of the general mindset that owing to their generally high reproductive rates and wide distributions, rodents are adaptable and therefore need no immediate conservation attention (Breed 1979, Amori & Gippoliti 2001, Lidicker et al. 2007, Breed & Leigh 2011). But, on the contrary, the discovery of extinct taxa like N. magnus reinforces the observation that these animals are environmentally sensitive and susceptible to extinction (see also Breed & Ford 2007, Lidicker et al. 2007, Amori et al. 2011). It is possible that N. magnus had a limited geographical distribution as it has not yet been identified in any coeval fossil deposits outside the Broken River region (e.g., Mount Etna region), (see Hocknull 2005, Hocknull et al. 2007) which do host other contemporaneous taxa like C. yirratji and N. longicaudatus (Cramb 2012). This could suggest that N. magnus was more susceptible to extinction than taxa with wider geographical distributions (see Amori et al. 2011). Finally, at least across the mid-late Holocene, it has been clear that the larger-bodied species of Notomys were more susceptible to extinction relative to smaller-bodied species (Fig. 6) (Alhajeri 2021, see also Griffiths 1980). The discovery of yet another larger-bodied Notomys species adds to that list and may have some implications for the identification of the most ‘at-risk’ extant species, providing knowledge that may inform priorities for conservation approaches for remaining populations. Data obtained from palaeontological studies therefore could highlight the deep-time responses of these taxa to long-term environmental changes and could potentially complement modern conservation, ecological and genetic studies (e.g., Roycroft et al. 2021, 2022) in formulating better policies for the conservation of Australia’s declining endemic fauna.

Conclusions

The discovery of a new extinct species of larger-bodied Notomys in the Broken River region has two key implications. Firstly, it underscores the susceptibility to extinction of larger-bodied murids within the threatened CWR taxa, especially within the genus Notomys. Secondly, it highlights the paucity of knowledge of the extent of decline of arid-adapted taxa in northern Australia. These results have conservation implications for protecting the current dwindling populations of these taxa in response to the current global environmental crisis by highlighting the need to incorporate the relatively recent fossil record alongside modern ecological and genetic studies to infer the long-term survival trends of these taxa.

Acknowledgements

The authors would like to thank the late Ken Aplin for initially recognizing the new species of Notomys. Kristen Spring, Andrew Rozefelds and Scott Hocknull and Heather Janetzki (QM), Sandy Ingleby (AM), Graham Medlin (South Australian Museum) and Kenny Travouillon (Western Australian Museum) provided access to comparative material. The authors also thank the volunteers at the UQ Palaeo Lab for assistance with fossil preparation. We are grateful to Douglas Irvin and family, Paul Osborne, Paco Murray and other members of the Chillagoe Caving Club, as well as local landowners for allowing access to caves. Finally, the authors would like to thank Dr Alex Baynes and an anonymous reviewer whose comments and feedback greatly improved our paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was funded by a UQ PhD scholarship to V.V. and Australian Research Council grants DE120101533 and DP120101752.