A new anuran genus from the fossil sites of Langebaanweg and Cooper’s Cave, South Africa

ABSTRACT Enigmatic frog ilia were recovered from two geographically and temporally disparate fossil sites in South Africa, namely the Early Pliocene (5.1 Ma) fossil site of Langebaanweg (south-western Cape), and Cooper’s Cave D (Northern province), which dates to around 1.38 Ma. The fossil ilia appear to represent an extinct anuran genus that subsisted in southern Africa over several million years, had a previously undocumented mode of locomotion, and possibly exceptional jumping ability. Relative to extant anurans, the fossil ilia show a unique suite of characteristics pertaining to the acetabulum, dorsal protuberance, ventral ridge of the shaft, and dorsal crest; features which would have facilitated and stabilised jumping.


Introduction
Fossil frogs are important paleoenvironmental indicators as their biology and breeding habits are integrally linked to rainfall and temperature.The frog taxa from fossil sites are thus uniquely placed to provide information on rainfall, and seasonality of rainfall (Matthews et al. 2015(Matthews et al. , 2016)).They show many skeletal differences across both families and genera (Buttimer et al. 2022).The ilium is particularly useful for identifying frog fossils as this bone is relatively robust and frequently recovered from frog-bearing deposits (Gardener and Rage 2016).In addition, ilia are generally identifiable even if some damage is sustained, and do not show sex-related morphological differences.Frogs and toads are unique among tetrapods in the skeletal arrangement of their pelvis and hindlimbs, which are specialised according to functionality and lifestyle.These bones are noted as the most important anuran apomorphies and are highly modified and specialized compared to most vertebrates (Holman 2003;Prikryl et al. 2009).They play an integral part in locomotion, reflect the behaviour (ecology) of the animal, and may be used to identify to family, genus, and sometimes species level (Matthews et al. 2019).The ilium, therefore, remains one of the most consistently useful bones for identifying fossil frogs (Holman 2003;Gardener et al. 2010;Matthews et al. 2015Matthews et al. , 2016)).
The fossil site of Langebaanweg (LBW) is situated on the west coast of South Africa, approximately 136 km north of Cape Town.This phenomenally rich fossil site was discovered during phosphate mining from the 1940s through to the 1980s (Hendey 1983).The then South African Museum (now Iziko Museums of South Africa) collected and curated fossils uncovered during mining operations.The site is now the seat of the West Coast Fossil Park Museum and has contributed substantially to understanding the evolution of ecosystems in the south-western Cape.It is the site of the first appearance in the fossil record for numerous vertebrate taxa and includes a rich and diverse frog fauna (Van Dijk 2003;Matthews et al. 2015Matthews et al. , 2016;;Roberts et al. 2011;Stynder 2011;Manegold et al. 2013).
The frog taxa from the Early Pliocene site of Langebaanweg (LBW) have been the focus of research in recent years.The fossil anuran community indicates that aridification of the currently arid west coast had not yet begun, and that the current winter-rainfall pattern may not yet have been established in the region (Matthews et al. 2015(Matthews et al. , 2016)).This research led to the discovery of unidentifiable, enigmatic frog ilia, which exhibited unusual features relating to the size and shape of the acetabulum, dorsal crest, and dorsal protuberance.
Cooper's Cave is situated in the Cradle of Humankind in the Gauteng Province of South Africa, approximately 50 km north of Johannesburg.Cooper's deposits are hosted in dolomites of the Monte Cristo Formation (Malmani Subgroup, Transvaal Supergroup), and at present, three distinct fossiliferous localities, A, B and D, have been identified.The fossils reported in this paper come from Cooper's D, which preserves the richest fossil-bearing deposit of the three localities and is dated 1.37 Ma ± 0.113 Ma by uranium-series (Pickering et al. 2019).The combination of decalcified sediments and an efficient sieving regime yielded an abundant and diverse fossil assemblage (de Ruiter et al. 2009).Several faunal studies have been undertaken of the Cooper's D assemblage, including hominins (de Ruiter et al. 2009), Cercopithecidae (DeSilva et al. 2013;Folinsbee and Reisz 2013), Carnivora (Hartstone-Rose et al. 2007, 2010;O'Regan et al. 2013;Kuhn et al. 2017;O'Regan and Steininger 2017;Cohen et al. 2019;Hanon et al. 2022), Equidae (Badenhorst and Steininger 2019), Aves (Pavia et al. 2022), and micromammals (Linchamps et al. 2019(Linchamps et al. , 2022)).The palaeoecological reconstruction based on bovid taxa suggests that Cooper's D was dominated by grassland, was sparsely covered, and had a nearby water source (Hanon et al. 2022).Around 2018, two anuran ilia (which showed similarities to the unidentified ilia from LBW) were recovered from Cooper's D (CD) while sorting of sieved material by TM.
This paper presents the results of a comparison between the fossil ilia from CD and LBW with extant southern African taxa.This comparison was facilitated by Matthews et al. (2019), who reviewed the ilia of all 39 modern anuran genera currently found in southern Africa, including Angola, Botswana, Malawi, Mozambique, Namibia, Lesotho, South Africa, Swaziland, Zambia, and Zimbabwe.CT scans were used to provide measurements and to describe features and characteristics of the 13 families (50 species).

Materials
The fossil material reported in this article consists of 24 frog ilia from LBW and two from CD. Fossil frog ilia from LBW came from two of the main fossil-bearing members; the Muishondfontein Pelletal Phosphate Member (MPPM) and the Langeberg Quartzose Sand Member (LQSM), which form part of the Varswater Formation.See Roberts et al. (2011) for further details on the stratigraphy and palaeontology of the site.The MPPM and LQSM sediments are inter-fingered in parts (suggesting they are of a similar age) and are considered to be around 5.1 Ma (Roberts et al. 2011;Cohen 2017).The MPPM is a river channel lag deposit which is thought to contain reworked deposits from an estuarine environment, and also from the LQSM which contains a mix of marine and terrestrial fauna (Hendy 1981;Roberts et al. 2011).Since 1998, the MPPM deposits have been the focus of several excavation seasons, and the frog material recovered from two 1 m 2 squares (A18 and B10) have been analysed and published in Matthews et al. (2015).The fossil specimens reported in this paper were recovered from these two excavation units, except for one ilium from the LQSM, recovered from a mining dump area (East Stream Dump 2) during phosphate mining at the site.The Iziko South African Museum, Cape Town, currently curates the LBW fossil material.The Cooper's D frog material consists of two accessioned specimens (CD18280A and CD18283B), and is curated at the Evolutionary Studies Institute, University of the Witwatersrand.

Methods
Gómez and Turazzini (2016) usefully reviewed, and standardized, the diverse terminology used in the literature to describe ilial features and morphology.This paper follows the terminology suggested by these authors.Taxonomy of anurans follows Frost (2017) (http:// research.amnh.org/vz/herpetology/amphibia).The pelvic and thigh musculature of frog ilia is presented and illustrated in detail in Přikryl et al. (2009: table 1, p. 102).Likewise, a detailed description and summary of the main features of the ilium are available in Gómez and Turazzini (2016).Figure 1 illustrates various features of the ilium discussed in this paper.For comparison, ilial features and characteristics (detailed below) were assessed and scored according to the same methodology and categories used on the extant southern African frog taxa (Matthews et al. 2019).Additional visual comparisons were made with frog ilia (available online on Morphosource (www.morphosource)) from elsewhere in Africa (23 genera and species, plus three additional species from sub-Saharan Africa), and from other continents and islands (35 genera and species) to increase the geographical research area (see Appendix A for list of specimens).

Computed Tomography (CT) Scanning and measurements
The LBW fossil frog material was CT scanned by the Central Analytical Facilities (CAF) Department at Stellenbosch University (South Africa).Appendix B contains scan information from the .pcafile.Volume Graphics (Versions 3.1.and 3.2) was used to render the scans and carry out linear and angle measurements.Measurements were made on CT scans of the ilia in lateral view (as per Figure 2), using the dimensioning functions ('distance' and 'angle') available in Volume Graphics.
The linear and angle measurements carried out on the ilia are illustrated in Figure 2. The VSA (the angle between the anterior margin of the ventral acetabular expansion and the iliac shaft) was measured as described in the literature (Holman 2003;Lynch 1971;Gómez and Turazzini 2016).Other measurements follow Sanchiz et al. (1993) and Matthews et al. (2019).
Measurement 1 (height of dorsal crest at the base of dorsal protuberance) was found to be broadly related to ilium size in the analysis of extant taxa (Matthews et al. 2019).The six longest ilia of the sample had the highest crests; generally speaking, the smaller the ilia in length, the smaller the height of the dorsal crest.Measurement 1 assessed the height of the dorsal crest at the base of the dorsal protuberance (DPT) but did not consider specimens where the crest reached maximum height, not at the base of the DPT, but more proximally, on the actual DPT.Measurement 4 was added to accommodate these specimens and to compare maximum crest height across fossil and extant specimens.The extant southern African frog taxa had an additional measurement (Measurement 7), namely, the length of the ilium (Matthews et al. 2019).This measurement was not feasible on any of the fossils due to the breakage of ilial shafts.

Description of ilial features
For comparison, ilial features and characteristics were assessed and scored according to the same methodology and categories used on the extant southern African frog taxa (Matthews et al. 2019), as per Table 1 and Table 2.An additional category was added to accommodate the large-angled VSA observed in the fossil material, and category three (VSA = >140 ̊) was added to the existing two categories (VSA = 140 ̊-90 ̊, and VSA < 90 ̊).
Obtaining a clear view of the bone surface of the two CD specimens was complicated as the bones were coated in a matrix that could not be separated from the bone cortex during rendering (of the CT scans) without eroding the surface.Assessing small features, such as the presence of a supra-acetabular or tubercular fossa was thus not possible on the CD material.
Phrynobatrachus africanus Hallowell, 1858 was added to the list of extant southern African frog species, and the measurements of this taxon are thus included with those from Matthews et al. (2019).In addition, taxa with a curved VAE were remeasured for the VSA to ensure that the angle between the shaft and the entire anterior portion of the VAE was considered, rather than the angle between the shaft and the most proximal portion of the VAE.
Some genera reported on in Matthews et al. (2019) as having ilial crests, were not considered a relevant comparison to the fossil material as ilial morphology and measurements were so different, clearly indicating different adaptations and locomotion.These included the burrower Hemisus (Hemisotidae), Trichobatrachus (Arthroleptidae), as well as several taxa that exhibited small crests and/or crests confined to the proximal portion of the ilia, including Kassina (Hyperoliidae), and the pyxicephalids, Poyntonia, Nothophryne, and Microbatrachella.

Linear and angle measurements
The measurements made on the fossil ilia (see Figure 2) are presented in Table 3.The same is presented in Table 4 for extant southern African taxa with dorsal crests (excepting the taxa noted above).Figures 3 and 4 present southern African pyxicephalids, and arthroleptids, phrynobatrachids, ptychadenids and ranids with dorsal crests, respectively.

Features and characteristics of the fossil ilia
Tables 5 and 6 present the characteristics and features of the fossil ilia.Images of the fossil specimens tabulated in Tables 5 and 6 are shown in Figure 5.  Figure 6 compares fossil taxa with assorted extant taxa with shared similarities, such as high crests, large acetabula, and an extended DT.
In order to provide a clear picture of crest morphology, Figure 7 presents the inner surface of the fossil ilia of specimens with relatively well-preserved dorsal crests, and includes SAM-PQL-B10A (Figure 7, specimen E), the largest fossil specimen recovered.Specimens A) and B), which are relatively well preserved along the ilioischiatic junction, exhibit an interiliac tubercle.

Dorsal and ventral acetabular expansions
Assessing the relative size of the dorsal and ventral acetabular expansions was complicated because these areas of the ilia were not preserved in any specimen.However,

Tubercular fossa
This fossa is a shallow depressed area dorsal to the iliac shaft and ventral to the DPT, and has no obvious functional correlate (Gómez and Turazzini 2016).Preacetabular fossa This feature is variable and if present, varies in the degree of development.The preacetabular fossa is the place of origin of the m.iliacus internus (Gómez and Turazzini 2016).Supraacetabular fossa This fossa is not always present and is absent from basal salientians and many anurans.When present, it varies in definition and depth and is generally delimited by a ridge that merges with the base of the DPT (Gómez and Turazzini, 2016).Interiliac tubercle This is a tubercle that varies quite considerably in size across taxa and is situated in the middle or ventral region of the ilial body (Gómez and Turazzini 2016).

Features on the iliac shaft
There are a number of grooves and ridges on the iliac shaft: . The 'calamita ridge' -a small ridge on the ventrolateral surface of the top third of the iliac shaft is the area for attachment of the m.tensor fasciae latae (Gómez and Turazzini 2016).In most anurans the m.tensor fasciae latae attaches further down the iliac shaft (Přikryl et al. 2009).

○
The 'spiral groove' is a groove on the dorsal surface of the head of the ilia which wraps around the shaft from dorsolateral to ventromedial (Gómez and Turazzini 2016).

○
The 'medial oblique groove' is a shallow groove which runs from the medial surface on the proximal iliac shaft, and then extends dorsoproximally to ventrodistally from near the dorsal edge, almost as far as the ventral margin (Gómez and Turazzini 2016).

○
The 'lateral oblique ridge' is a short ridge which is situated distal to the acetabulum, and extends from the midpoint of the acetabular rim to the ventral surface of the iliac shaft (Gómez and Turazzini 2016).the most complete specimens (Figure 5, specimens C, J, K, T, and U) suggest that they are of a similar size, as seen today in Amnirana (Figure 4, specimens G and H), Anhydrophryne (Figure 3, specimen F) and Arthroleptella (Figure 3, specimens G and H).Some of the largest, more massive fossil specimens (Figure 5, specimens F, N, and S) hint that the VAE may be slightly larger than the DAE, but this impression cannot be confirmed.The VAE is of medium size and well-developed in the relatively complete specimens.The DAE is triangular in shape and projects backwards.

Dorsal Tubercle
The dorsal prominence and the dorsal protuberance were clearly distinguishable in the fossils and did not merge, as they do in many extant taxa.The dorsal prominence (DPM) was smooth and flat, and the dorsal protuberance (DPT) was large, broad, and exceptionally well-defined in all the fossils.All fell into the 'tubular, elongated, large' category, and just under half were additionally described as 'asymmetrical'.This asymmetry manifested in the larger specimens (e.g., Figure 5, specimens N, F, A and S) in the distal portion of the DPT, which was bulbous and tear-shaped.All the fossils fell into the 'massive' category regarding dorsal expansion of the DPT, which was positioned posterior to the acetabulum in all cases.This is similar to the positioning observed in many extant taxa and was one of the two most common categories (the other being where the proximal portion of the dorsal prominence was at the same level as the acetabulum) (Matthews et al. 2019).Regarding extant taxa with dorsal crests, Hildebrantia and Amietia had the largest DPTs.Arthroloptella is noted as having a relatively large DPT (Matthews et al. 2019), but it is thin and elongated, very different in shape and robustness compared to the fossils.Despite being a relatively small taxon, the DPT of the fossil species differentiates from the three extant taxa in terms of its large size and robustness, in shape, and the extent of the dorsal expansion (see Figure 6, which compares the fossils with extant taxa with large DPTs).The exact shape of the DPT of the CD specimens is hard to assess due to coating on the fossils, but it is also clearly large and of a similar shape to the LBW specimens.Matthews et al. (2019) note that in some taxa with dorsal crests (such as Hildebrandtia, Ptychadena, Arthroleptis, and Leptopelissee Figure 4), the DT extends into the proximal portion of the crest and merges with it.This merging of the dorsal crest with the dorsal prominence is also noted in Gómez and Turazzini (2016).In the case of the fossils, however, the DT remains clearly differentiated from the crest (to an extent not observed in any extant taxa) and rests upon it.The gluteus maximus always originates from the dorsolateral border of the dorsal tubercle, or, if a dorsal protuberance is lacking, has its origins      in a corresponding part of the iliac shaft, just dorsolateral to the origin of the iliofibularis and iliofemoralis (Přikryl et al. 2009).This muscle does not seem to have changed position compared to early-diverging anurans (Přikryl et al. 2009).

The acetabulum
The acetabulum was expanded and covered a large portion of the acetabular expansion.The general acetabular shape was that of a rounded 'V', similar to that of Natalobatrachus and Ptychadena, which have similarly expanded acetabula.The acetabula of the fossils have proximally divergent margins.In many fossils, a rugose, V-shaped area may be observed on the margin of the acetabulum, proximal to the VAE, in the general area where the cruralis muscle attaches in several frog taxa (Přikryl et al. 2009).

Iliac shaft and dorsal crest
The dorsal crest was observed to taper in width proximally to distally in the majority of extant taxa with dorsal crests e.g., Arthroleptis, Leptopelis, Amnirana, and Ptychadena.
No fossils preserved the entire length of the shaft.However, the most complete specimen (see Figure 5, F) indicates that the shaft does not start tapering in what is estimated to be the top third to half of the proximal portion, as observed in extant taxa such as Ptychadena, Phrynobatrachus, and Hildebrantia.Gómez and Turazzini (2016) note that the iliac shaft length is approximately four times that of ilial body length in the majority of anurans, with exceptions noted as being both shorter (less than three times the length of the ilial body) in some extinct salientians (Rage and Roček 1989), and extant Ascaphus (Přikryl et al. 2009), and longer (more than five times the length of the ilial body) in pipids, bombinatorids, and some hylids.Unfortunately, ilial length could not be assessed in the fossils due to damage.All the fossil specimens (Figure 5) show a particularly broad ventral ridge which is thicker and better defined than that of relatively larger extant taxa (e.g. Figure 6, Hildebrantia, Amnirana and Leptopelis).The ventral ridge of the fossils projects outwards from the lateral surface of the dorsal crest, giving the impression it is resting on it, rather than merging with it, as is commonly seen in extant taxa (a similar differentiation is noted above regarding the DT).
The lateral oblique ridge, which merges to become one with the ventral ridge, is present in all the fossils, and is much better defined than in any of the extant southern African taxa with this feature, which include Trichobatrachus (Arthroleptidae), Vandijkophrynus (Bufonidae), Natalobatrachus and Tomopterna (Pxyicephalidae), and Heleophryne (Heleophrynidae) (Matthews et al. 2019).

Tubercles, fossae, and grooves
The interiliac tubercle proved to be a useful feature for the identification of some of the extant taxa as it was not common, and appeared only in the three arthroleptid genera, in Phrynomantis and four out of the seven hyperoliid genera, namely Cryptothylax, Afrixalus, Phlyctimantis and Semnodactylus.An interiliac tubercle (see Figure 7, specimens A and B) was present in some seven fossil specimens, but could not be properly assessed in the remainder due to damage along the ilioischiatic border.
A tubercular fossa, preacetabular fossa, and supraacetabular fossa were discernible in the majority of the fossils.A preacetabular fossa was present in eight extant genera: Trichobatrachus, Sclerophrys, Schismaderma, Vandijkophrynus, Poyntonophrynus, Nothophryne, Pyxicephalus, and Tomopterna, while a supraacetabular fossa was found in Arthroleptis, Leptopelis, Trichobatrachus, Tomopterna, Poyntonia, Pyxicephalus, Cryptothylax, Kassina and Amnirana (Matthews et al. 2019).A tubercular fossa was present only in Arthroleptella.This fossa was well-developed in the LBW fossils.None of the extant southern African taxa were found to exhibit all three fossae (although Matthews et al. (2019) note that the tubercular fossa may have been obscured by an advanced rendering of the bone's surface in some cases).The presence of a spiral groove was noted in only one extant southern African taxon (Trichobatrachus) (Matthews et al. 2019) and was not present in the fossils.Linear measurements (Measurements 1, 2, 3 and 4) As indicated by measurements 1 and 2, SAM-PQL-71161G (Figure 5, specimen L) had the narrowest crest of all the fossils.The results for this taxon are reported on separately from the other fossils in this section as it differed in several other features (discussed further on) and is thus considered to represent a different taxon.Of all the extant southern African taxa, Amietia vertebralis Hewitt, 1927 and Hildebrandtia ornata Nieden, 1907 (also Trichobatrachus, excluded from this discussion) exhibited the highest crests (measurement 1) (Matthews et al. 2019).The six longest ilia of the extant sample had the highest crests (as measured at the base of the dorsal protuberance), indicating that the size of the individual (and presumably age) generally has a positive relationship with crest height (Matthews et al. 2019).Measurement 1 ranged from 1.71 mm to 3.38 mm for the fossils, and from 0.54 mm to 3.33 mm for all the extant southern African taxa.Measurement 4, which also assessed crest height, ranged from 0.59 mm to 4.40 mm for the modern species, and 1.28 mm to 3.43 mm for the fossils.Both measurements indicate that, relative to the extant taxa, the fossils have exceptionally high crests.This is illustrated by the fact that only the fossil specimens with the smallest crest heightthat is, SAM-PQL-71161F (Figure 5, specimen K) for measurement 1, and SAM-PQL-B10F (Figure 5, specimen Q) for measurement 4have narrower crests than that of the largest extant taxa, which also had the largest (longest) ilia, i.e.Leptopelis, Hildebrandtia, Ptychadena, Amietia and Amnirana, and Strongylopus springbokensis Channing 1986 (the latter in the case of measurement 4, not measurement 1).Arthroleptella has a narrow crest, as reflected in the values obtained for measurement 4 (and 6).
Measurement 2 (neck depth) reflects the morphology of the ilia at the point where the shaft joins the proximal ilia.No clear pattern emerged from the analysis of extant taxa, and it was found to be a very variable feature, even among genera within the same family; however, this measurement was found to be size-dependent in that the seven longest ilia had the deepest necks (Matthews et al. 2019).Amietia vertebralis, which was the fourth longest ilia measured, has the deepest neck of all extant species investigated, as well as a particularly broad acetabulum.As no complete fossil ilia were recovered, it is not possible to quantify length vs neck depth; however, measurement 2 scores high considering the fact that the fossil specimens are small compared with extant taxa such as Amietia (Figure 3), and Leptopelis, Hildebrantia, and Amnirana (Figure 6).Excluding the anomalous SAM-PQL-71161G (Figure 5, specimen L), neck depth ranged from 1.46 to 2.71 mm for the fossils, and from 0.48 to 3.56 mm in extant taxa (Table 4).
In the extant taxa, the relationship between the breadth of acetabulum below the ilioischiatic junction (measurement 3) with ilial size, is similar to that noted above, with longer ilia generally having larger, and broader acetabula, although exceptions exist (Matthews et al. 2019).However, rather more homogeneity was noted among species from the same genus, reflecting a common general shape and morphology (Matthews et al. 2019).Results are complicated in that breakage disqualified eight fossil ilia from this measurement, while another four were estimates.Measurement 3 ranged from 0.69 to 3.71 mm for the extant taxa, with both the median and average measurement for all southern African frog taxa being 1.9 mm (Matthews et al. 2019), while the fossils ranged from 1.46 to 3.35 mm (average 2.53 mm, median 2.22 mm).Arthroleptis, both Amnirana species, Ptychadena, Hildebrantia, Leptopelis and A. vertebralis (but not A. hymenopus Channing 2015) fall within the range of the fossils for this measurement, once again indicating that the acetabula of the fossils are exceptionally wide relative to those of the larger, extant taxa.Both CD specimens fell within the range of LBW specimens with the widest acetabula.SAM 71161G (Figure 5, specimen L), is once again anomalously small and was 0.34 mm smaller than the next smallest fossil.
Angle measurements (Measurements 5 and 6) Matthews et al. (2019) noted quite marked inter-generic and inter-specific variation in the VSA, indicating that it is of variable usefulness for identification.The VSA is, however, the measurement which distinguishes the LBW and CD specimens from all the extant southern African taxa (for which the VSA ranged from 61°to 139°, including taxa with and without dorsal crests).The fossils had a VSA which ranged from 140.5°to 165.9°, except SAM 71161G, which has the smallest VSA at 135.27°(this is, however, still greater than the VSA of extant taxa with dorsal crests by 4.9°).The two CD specimens fit comfortably in the lower half of VSA fossil measurements.
The angle between the dorsal protuberance and the ventral ridge (measurement 6) showed a wide degree of variation in the extant taxa between genera and species, (ranging from 19°to 66°) and was, like the VSA, of variable usefulness for identification.The fossils (excluding the anomalous SAM-PQL-71161G) were also variable, with a range of 28°to 61°.SAM-PQL-71161G exhibited marginally, the largest angle at 61.55°( the next closest fossil values being 61.45°and 61.24°).The two Amietia species (A.vertebralis, A hymenopus) and Strongylopus grayii Smith 1849, fall within the upper range shown by the fossils, and Amnirana, Arthroleptis, Natalobatrachus, Hildebrantia, and Phrynobatrachus overlap with the lower range of values.Leptopelis, Ptychadena, Phrynobatrachus, and Anhydrophryne fall below, and show no overlap with the fossil taxa in terms of measurement 6, while Arthroleptella and Strongylopus overlap in the case of one of the two species representing these genera.

Discussion
The large VSA distinguishes the LBW and CD specimens from all extant southern African taxa.The fossils are also characterised by a markedly robust and large DT, and a very expansive dorsal expansion.Relative to extant species, the ventral ridge of the fossils is markedly broad and well-defined, projecting outwards from the lateral surface of the dorsal crest, and (like the DT) remains clearly differentiated, not merging with the crest at any point.The lateral oblique ridge is present in all the fossils and is exceptionally well-defined.Three fossae (the tubercular, preacetabular, and supraacetabular) are present.Relative to extant taxa, the crest is exceptionally high, and the acetabulum wide and deep.
SAM-PQL-71161G is anomalous as compared to the other fossil specimens in four aspects: it has a narrower crest, exhibits the broadest angle obtained for measurement 6 out of all the fossils, has the smallest VSA, and also has a smaller acetabulum.The fact that this specimen was consistently different from the other fossils in the linear and angle measurements, as well as the fact that the differences do not appear to be size-related (another similarly sized fossil (SAM-PQL-71161F, Figure 5 specimen K) shows inverse patterning regarding the VSA and measurement 6), suggests that it represents a different taxon to the other specimens.
The fact that no specific suite of features can be used to consistently identify extant frog families and genera (Matthews et al. 2019) complicates the interpretation of the fossil record.Modern pyxicephalids, for example, are exceptionally diverse and there are no intra-familial features or characteristics which can be used to identify genera within this family.Upon visual inspection, the two CD specimens are morphologically similar to the LBW fossils and fall within the range of the latter for all measurements.It is, however, unclear at what taxonomic level they may be differentiated, an issue complicated by the fact that they are temporally (∼3.6 Ma younger) and geographically disparate from the LBW specimens.Another issue is whether the fossils from the two sites represent an extinct anuran family/families, or genus.Although there are similarities between the modern and fossil taxa in some features and measurements, most notably Amnirana (in terms of a wide VSA, and height of dorsal crest), Amietia (angle between the dorsal protuberance and the ventral ridge, height of dorsal crest and neck depth), and Strongylopus (angle between the dorsal protuberance and the ventral ridge, wide VSA), the fossils exhibit a suite of unique features.As research into these taxa is in the early stages a conservative approach is considered best, and it is therefore suggested that they represent a gen.nov.Although Matthews et al. (2019) focussed on southern African taxa, research for this paper included consideration of the morphology of frog ilia from elsewhere in Africa and the world (see Appendix A), reviews covering geographically disparate anuran ilia (Holman 2003;Jorgenson and Reilly 2013;Gómez and Turazzini 2016), and pelvic and thigh musculature in Anura (Přikryl et al. 2009).Given the advanced degree of parallel and convergent morphological evolution observed among ranoid frogs (Maxson and Wilson 1974;Wells 2007) (even across arboreal, aquatic, and fossorial microhabitats (Feng et al. 2017;Gomes et al. 2009), it is interesting that this research (admittedly not exhaustive) has not yet come up with a comparative ilial morphology to that of the fossils.
The fact that this gen.nov. is found at both LBW and CD suggests it may have been relatively widespread in southern Africa and comprised several species.Supporting fossil evidence for this comes from the anomalous specimen SAM-PQL-71161G, and the fact that a taxon with similar characteristics and features (including those that make the CD and LBW fossils unique) was identified from Member 1 at the hominid-bearing site of Swartkrans (also situated in the Cradle of Humankind) (TM pers.ob.) The fossil micromammal (mice, rats, mole rats and shrews) communities from the Cradle and LBW have been the focus of multiple studies (Avery 2000;Avery 2001;Hopley et al. 2006;Matthews et al. 2007;Denys and Matthews 2007;Matthews andStynder 2011, Linchamps et al. 2022), and the sharing of anuran taxa is not unexpected as the two sites have in common at least seven murid (mice and rat) genera (TM pers.ob.).Other microfaunal similarities exist between LBW and other Cradle of Humankind sites such as Gondolin (TM pers.ob.), Makapansgat, Sterkfontein, Swartkrans, and Kromdraai (Denys 1999;Avery 2001;Matthews 2004;Matthew et al. 2007).LBW's faunal assemblages include many extant genera in terms of birds, and large and small mammals, and is the site of first appearance for many; however, also present are taxa which span the Late Miocene-Early Pliocene, and a number of relict Miocene taxa (Matthews 2004;Matthews et al. 2007;Roberts et al. 2011).The LBW micromammal assemblages include two shrew species diagnosed as representing an extinct lineage (Matthews and Stynder 2011), and the relict Miocene murid genera, Stenodontomys and Euryotomys, which also appear in some Cradle sites (Matthews 2004).The presence of an extinct anuran genus at LBW and CD is thus not surprising.However, the uniqueness and novelty of some of the features of the fossil taxa are unexpected and intriguing.Relating the morphology of the fossils to locomotion or ecology is complicated by lacunae in our understanding of the relationship between skeletal morphology and locomotor mode in frogs (Jorgenson and Reilly 2013).Although the majority of anurans are specialised in swimming or jumping (Ahn et al. 2004), there is considerable variety in the modes of locomotion which include walking, hopping, swimming, jumping, and "flying" (Emerson et al. 1990;Du Preez and Carruthers 2017).There are variations within these modes; for example, Hylambates maculatus (previously Kassina) walks and runs with the hindlimbs alternating (Ahn et al. 2004), and many anurans whose main form of locomotion is walking or hopping, are also capable of burrowing (Wells 2007).Anurans are found in various microhabitats, including terrestrial, aquatic, arboreal and underground (burrowers).Emerson (1979Emerson ( , 1982) ) identified three distinctive ilio-sacral joint morphologies and noted that the condition of the dorsal crest on the ilium, relative ilial length, and posterior transverse process length, are significantly correlated with the degree of sacral expansion.The presence of a dorsal crest on the fossils places them into class IIB in terms of Emerson's (1982) sacral joint categories, which means that they would have lacked sacral expansion, and would (presumablythe research was carried out on extant taxa) have had relatively shorter ilia and longer transverse processes (Emerson 1982).These features result in no movement in the horizontal plane, and such frogs are noted as being jumpers that do not typically walk (Emerson 1982).Zug (1978) noted in a study on performance, that intra-and inter-specifically, larger frogs jump further than smaller ones; however, in terms of relative ability, smaller frogs are stronger jumpers.The fossil ilia suggest the animals had a relatively small body size, which was accompanied by a number of features which presumably supported and enhanced jumping performance.Ponssa et al. (2018) note that the cross-sectional area of muscle, and bone crest attachment sites, are crucial for locomotory performance.The first serves as a proxy of the force exerted by the muscle, and crests (such as the DT and dorsal crest on the ilium, and urostylar crest on the urostyle) are muscle attachments sites related to muscle force, where bone crest size can be used to indicate the magnitude of muscle force in the fossil record (Zumwalt 2005;Ponssa et al. 2018).Přikryl et al. (2009) note that the iliac shaft and urostyle support muscles which are predominantly protractors of the femur and are necessary in attaining a crouching position to facilitate rapid escape.
A lateral oblique ridge is present in all the fossils and is particularly well-defined relative to extant southern African taxa; however, the functional correlate of this feature is unknown (Přikryl et al. 2009), and it is of limited use in phylogenetic studies (Gómez and Turazzini 2016).Other muscles which originate on the ischium and insert on the femur or knee apoeurosis include: the adductor magnus, obturator internus, obturator externus, quadratus femoris, gemellus, semitendinosus, gracilis major, gracilis minor, and the sartorius.These muscles are involved in the femur's protraction, adduction and retraction, knee flexion, and cranial long axis rotation (Přikryl et al. 2009).The semimembranosus originates on the ilium and ischium, inserts into the knee aponeurosis, and facilitates retraction of the femur and slight knee flexion.It is not possible to comment on the areas of attachment of the above muscles as the ischium and pubis are missing from all the fossils.However, the bone at the ilioischiatic junction is thickened, suggesting that these too were robust and sturdy.James et al. (2007) note that the coordination of the skeleton and muscles of the pelvic girdle is critical to force transmission between the hindlimbs and axial skeleton when jumping.The fossil specimens have a markedly large and robust DPT, which presents an extensive area of attachment for the glutaeus maximus (the glutaeus maximus originates on the ilium at the posterior end of the dorsal crest, inserts on the cruralis, and contributes to knee extension (Přikryl et al. 2009)) and would have facilitated strong, high, jumps.This would have been supported by the iliofibularis and iliofemoralis which both originate on the lateral surface of the dorsal tubercle's posterior part, insert on the tibiofibula, and contribute to retraction of the femur and flexion at the knee (Přikryl et al. 2009).Previously, it was noted that many of the fossils exhibited a rugose, Vshaped area on the margin of the acetabulum, proximal to the VAE, in the area where the cruralis muscle attaches in several frog taxa (Přikryl et al. 2009).This muscle originates on the ilium and inserts into the knee aponeurosis (aponeuroses are sheet-like elastic tendon structures that act as insertion sites for muscle fibres, and absorb energy during the movement of the muscle (https://www.physio-pedia.com/Aponeurosis),and, like the glutaeus maximus, enables knee extension and hip flexion (Nauwelaerts et al. 2007;Přikryl et al. 2009).The fact that the site of this muscle is evident on many of the fossils suggests it was robust and large, and presumably supported efficient jumping.The high dorsal crest would likewise have provided a large surface area for muscle attachment.Such large ilial muscles would control the movement of the pelvis relative to the body, and stabilise the movement of the limbs relative to the pelvis (Emerson and De Jongh 1980;Duellman and Trueb 1994;Přikryl et al. 2009).These muscles would have included the iliacus externus, which originates on the lateral surface of the dorsal crest, inserts on the femur, and is involved in protraction of the femur.The tensor fasciae latae also originates on the iliac shaft, and inserts into the fascia lata upon which it puts pressure (Přikryl et al. 2009).The coccygeoiliacus originates on the urostyle and inserts on the iliac shaft.The high dorsal crest of the fossils would thus have provided ample area for muscles which supported strong, high jumps.
The preacetabular fossa is the place of origin for the iliacus internus (Gómez and Turazzini 2016).The well-developed state of this fossa suggests that the associated muscle (which inserts into the femur and is responsible for its protraction and abduction (Přikryl et al. 2009)) was robust.Muscles involved with the retraction and adduction (movement of a body part toward the body's midline) of the femur are the adductor longus (which originates on the ilium and pubis), and the pectineus (which originates on the ischium) which is involved in fixing the femur in the acetabular cavity (Přikryl et al. 2009).The broad, deep acetabulum of the fossil taxa clearly articulated with a robust femoral trochanter, which would have helped stabilize the pelvis during jumping.

Conclusions
The features discussed above, many of which differentiate the fossils from extant taxa, all appear to support a particularly efficient jumping locomotion, possibly far superior to extant frogs.However, although the features noted above may have supported an exceptional jumping ability, the issue is complicated.For example, a lack of correlation between available muscle power and jump power has been observed in extant frogs (Roberts et al. 2011b), andJames andWilson (2008) note further research is needed to understand which adaptations (morphological and physiological) play the most significant part in enhancing jump performance.The marked differences exhibited by the gen.nov.relative to extant taxa appear to indicate a hitherto undocumented mode of locomotion, which would, in turn, have been associated with a number of unknown variables relating to evolution, environment, breeding strategies, and predator avoidance.While these remain uncertain, the reasons for the extinction of the gen.nov., which appears to have subsisted in southern Africa over a period of several million years, remains unknown.Further research into fossil frogs from other southern African and African sites should provide a clearer picture of the taxonomic status of these enigmatic taxa.

Figure 2 .
Figure 2. Linear and angle measurements made on fossil ilia.(SAM-PQL-B10B), right ilium, right lateral view Measurement 1. Height of dorsal crest at base of dorsal protuberance; Measurement 2. Neck depth (distance from the dorsal to the ventral margin of the junction between the ilial body and the iliac shaft); Measurement 3. Breadth of acetabulum below the ilioischiatic junction; Measurement 4. Height of dorsal crest at highest point of dorsal tubercle; Measurement 5. VSA = the angle between the anterior margin of the ventral acetabular expansion and the iliac shaft; Measurement 6.The angle between the dorsal protuberance and the ventral ridge.

Table 3 .
Linear and angle measurements of fossil specimens from Langebaanweg and Cooper's D.

Table 5 .
Features and characteristics of the fossil ilia from Langebaanweg and Cooper's D (VAE, and dorsal and interiliac tubercles).

Table 6 .
Features and characteristics of the fossil ilia from Langebaanweg and Cooper's 'D' (dorsal crest, acetabulum, and fossae).