﻿A new species of Umbilia Jousseaume, 1884 (Mollusca, Cypraeidae) from the Pliocene fauna of the Roe Plains, Western Australia

﻿Abstract A new morphologically distinct species of cowry (family Cypraeidae Rafinesque, 1815) is described from the Late Pliocene Roe Calcarenite of the Roe Plains, Western Australia. Previously assigned to Umbiliahesitata (Iredale, 1916), the new species differs morphometrically from related taxa and is differentiated from U.hesitata by a number of shell features including a prominent, projecting protoconch, less extended posterior and anterior terminals, coarser columellar teeth extending onto the base, and well-developed, thickened anterior flanges, supporting a rounded anterior extremity with blunt anterior tips. Umbiliatomdarraghi sp. nov. is the third Umbilia species to be described from the Pliocene.


Introduction
The Roe Calcarenite is a fossil-rich Late Pliocene deposit covering an area of around 12,000 km 2 of the Roe Plains in south-eastern Western Australia (Kendrick et al. 1991(Kendrick et al. , 1997James et al. 2006). Fossils of a broad range of molluscs are common within the Roe Calcarenite with around two-thirds of these originally considered to represent living species (Ludbrook 1978). Six species of cowries (Gastropoda, Cypraeidae) have been reported from the Roe Calcarenite, of which five are extinct; Austrocypraea amae Fehse & Kendrick, 2000, Notocypraea darraghi Fehse, 2010, N. goudeyi Fehse, 2011, Zoila campestris Darragh, 2011, and Umbilia fodinata (Darragh, 2011). Darragh (2002) assigned the sixth species to U. hesitata (Iredale, 1916), while noting that the Roe Plains is considerably further west than the western-most range of living U. hesitata, represented by the subspecies U. h. suprastrata Govaert et al., 2015. Subsequent authors have expressed uncertainty regarding assignment of these Roe Plains fossils to U. hesitata. Wilson and Clarkson (2004), for example, commented that the U. hesitata-like fossils from Roe Plains are atypical of living U. hesitata and noted their "rather flat base and stronger teeth", while Lorenz (2018) referred to these fossils as an "ancestral hesitata-like species". Only one living member of the genus, Umbilia armeniaca (Verco, 1912), is currently found in offshore habitats adjacent to the Roe Plains. Goudey (2015), however, noted that Roe Plains U. hesitata-like fossils are less inflated than the shells of living U. armeniaca.
Detailed study of the Roe Plains U. hesitata-like fossils has not previously occurred, probably because of the rarity of intact specimens. However, we recently identified a number of specimens in museum and private collections, available for study, allowing detailed morphometric examination of this taxon for the first time. Past research on fossil and extant species of Umbilia (Southgate et al. 2021;Southgate and Militz 2023) generated morphometric data relating to shell form (i.e., shell length, shell height, shell width, columellar and labral tooth counts, and relative mass) for each of the recognised subspecies of U. hesitata and U. armeniaca. In the present study, these data were used to support multivariate morphometric comparisons between the Roe Plains U. hesitata-like fossils and extant specimens of U. hesitata and U. armeniaca, at both species and subspecies levels. Results showed clear morphometric separation of the Roe Plains U. hesitata-like fossils from both U. hesitata and U. armeniaca, and sufficient differentiation from both living and fossil taxa to justify recognition of the Roe Plains U. hesitata-like fossils as a new species, described here as Umbilia tomdarraghi sp. nov.

Examined materials
All examined specimens of the Roe Plains U. hesitata-like fossils were recovered from material excavated from the Roe Calcarenite at various sites within the general area between Madura (31°53'58"S, 127°01'11"E) and the Hampton repeater tower (31°57'52"S, 127°34'50"E), Western Australia, including the Main Roads quarry 16 km south of Madura (32°02'22"S, 127°02'50"E), which was a major source of material used in construction, upgrades, and maintenance of the Eyre Highway, which dissects the Roe Plains. Only specimens supporting accurate assessment of all morphometric characters, outlined below, were included in this study.

Data collection
Primary data were generated for 11 specimens of the Roe Plains U. hesitata-like fossils. Shell length (L), shell width (W), and shell height (H) were measured to the nearest 0.1 mm using vernier callipers. Counts of columellar teeth (CT) included the posterior-most denticle that merges with the anterior edge of the columella callus bordering the posterior canal. All labral teeth (LT) were counted.

Data analysis
Data analysis combined qualitative appraisal of key conchological features, such as the aperture, spire, columellar teeth, anterior and posterior terminals, and base, with a quantitative appraisal of overall shell form. For the quantitative component of this analysis, we adapted the multivariate approach of Southgate and Militz (2023) to compare specimens of the Roe Plains U. hesitata-like fossils with extant U. armeniaca and U. hesitata, after aggregating subspecies data, and to compare specimens of the Roe Plains fossils with each of the extant U. hesitata subspecies.
Shell form was represented by the following morphometric characters: L, width to length ratio (W/L), height to length ratio (H/L), height to width ratio (H/W), normalised LT (nLT), and normalised CT (nCT). For each specimen, nLT and nCT were calculated from LT and CT, respectively, for a hypothetical shell length of 25 mm as described by Schilder (1937) and W/L, H/L and H/W were expressed as a percentage (Lorenz 2017). While the multivariate approach of Southgate and Militz (2023) also incorporated relative mass (sensu Bridges and Lorenz 2013), this approach is inappropriate with fossils where the influence of mineralisation and residual matrix may compromise resulting data. Aside from this alteration, the multivariate approach taken in this study followed that of Southgate and Militz (2023). Briefly, values for the morphometrics outlined above were transformed to Z-scores and atypical specimens (i.e., |Z-score| > 3) either validated (primary data) or censored (secondary data) before computing a resemblance matrix based on Euclidean distance between specimens. Non-metric multidimensional scaling (nMDS) was then used for dimensionality reduction to permit visualisation in two dimensions. Visual observations of the nMDS configuration were validated by estimating the probability that a priori assigned groups (i.e., taxa) shared the same central tendency (i.e., centroid) and variation (i.e., dispersion) in shell form. Specifically, a one-factor permutational analysis of variance (PERMANOVA) was used to estimate the probability that groups shared the same central tendency in shell form; pairwise comparisons proceeded detection of a significant group effect, using PERMANOVA for each comparison and controlling for the family-wise error rate with the Holm (1979) procedure. Permutation-based tests for homogeneity of multivariate dispersions were used to compare the distance of specimens from their centroid, controlling for the family-wise error rate with the Holm (1979) procedure.
All statistical computing was performed using R (version 4.2.1) with the stats (R Core Team 2022) and vegan (Oksanen et al. 2022) packages. For statistical tests, significance was accepted at a value of P < 0.01 as recommended by Southgate and Militz (2023) to conservatively establish inter-group differences. Data summaries for a specific morphometric are presented in-text as means (x) ± standard deviation (SD) and for all morphometrics, collectively, means are presented using the "shell formula" [L (W/L-H/L-H/W) nLT: nCT] (Bridges and Lorenz 2013 CG collection (paratype 9) • 1; among spoil material at Hampton microwave repeater tower (31°57'52"S, 127°34'50"E); J. Fell leg.; JF collection (paratype 10).
Other material. Australia; Roe Plains, same location as holotype; dry specimen (fossil); among spoil material; CG (1 repaired specimen). Diagnosis. Shell pyriform to ovately pyriform, humped; dorsal summit towards posterior, W/L = 59%, H/L = 48%; spire impressed; protoconch projecting and prominent, positioned to the left side and visible when the shell is viewed from a dorsal aspect. Coarse columellar teeth extending onto base. Anterior and posterior terminals extended; anterior lateral flanges well-developed, thickened; anterior extremity broad, flattened, rounded; anterior tips blunt. Anterior dorsal tubercules absent; a small, raised callus on left side only; anterior groove absent. Anterior and posterior extremities extended; anterior extremity rounded, supported by broad, thickened, lateral flanges; anterior tips rounded, not pointed; posterior terminal curved to left with right posterior tip extending further; dorsum smooth. Single anterior tubercule evident as small, raised callus on left side only; anterior groove absent; base convex, broad, flattened anteriorly. Aperture widening anteriorly, narrowest at anterior end of posterior canal; apertural teeth coarse, evenly spaced, and well developed along whole length and on both sides of the aperture; columellar teeth extending onto base. Labral teeth (28-35) more numerous than columellar teeth (22-30). Shell margins rounded, smooth; spire umbilicate; protoconch large (4.8 ± 0.4 mm diameter; n = 6), projecting and prominent, positioned to the left so that the penultimate body whorl has greater exposure on the right side of the spire ( Fig. 2A, B). Fossula narrow, smooth, and slightly concave. Differential diagnosis. When compared to extant Umbilia armeniaca and U. hesitata, shell form of U. tomdarraghi sp. nov. is morphometrically more similar to U. hesitata (F = 7.9, R 2 = 0.09, P < 0.001) than to U. armeniaca (F = 23.2, R 2 = 0.14, P < 0.001, Fig. 3A). The explicit distinction between U. tomdarraghi sp. nov. and U. hesitata, when independently assessed using morphometric data for all three U. hesitata subspecies (U. hesitata hesitata, U. h. beddomei and U. h. suprastrata), showed clear separation (F = 6.9, R 2 = 0.08, P < 0.001), with U. hesitata being significantly more variable in shell form (F = 10.1, P < 0.01) than U. tomdarraghi sp. nov. (Fig. 3B). When compared to individual subspecies of U. hesitata, shell form of U. tomdarraghi sp. nov. is most similar to U. h. suprastrata, but U. h. suprastrata is more similar to both U. h. hesitata and U. h. beddomei than it is to U. tomdarraghi sp. nov. (Fig. 3B; Table 2). It is notable that variability in shell form among the type series specimens of U. tomdarraghi sp. nov. is similar to that of Figure 3. A nMDS ordination (stress = 0.14) of the resemblance matrix for Umbilia armeniaca, U. hesitata and U. tomdarraghi sp. nov. B nMDS ordination (stress = 0.19) of the resemblance matrix for the three U. hesitata subspecies and U. tomdarraghi sp. nov. Shaded ellipses indicate the 95% confidence interval of taxa (species or subspecies) centroids. The coefficient of determination (R 2 ) and probability that distances between centroids arose by random chance (P) are presented. (Table 3). Univariate comparisons of L, W/L, H/L, H/W, nLT, and nCT showed U. tomdarraghi sp. nov. to differ from U. h. hesitata by significantly greater W/L, lower H/W and lower nCT, from U. h. beddomei by significantly greater L and lower H/L, and from U. h. suprastrata by significantly lower H/L and lower nCT (Fig. 4). Key conchological features differentiating U. tomdarraghi sp. nov. from U. h. hesitata, U. h. beddomei, and U. h. suprastrata include coarser, extended columellar teeth; the broader, flatter, blunter anterior extremity; lack of both distinct anterior dorsal tubercules and an anterior groove; a flatter base; and a much larger, protruding protoconch (Table 4). Table 2. PERMANOVA results testing the hypotheses that there were no differences in central tendency (i.e., centroid) of shell form among the Umbilia hesitata subspecies and U. tomdarraghi sp. nov. The Euclidean distance (D) between centroids, coefficient of determination (R 2 ), and Holm-adjusted probability that the distance between centroids arose by random chance (P) are presented.   A second species of Umbilia, U. fodinata (Darragh, 2011), occurs with U. tomdarraghi sp. nov. within the Roe Calcarenite. While Darragh (2011) originally assigned this species to the genus Zoila Jousseaume, 1884, in the most recent review of the family, Lorenz (2017Lorenz ( , 2018 placed the species within Umbilia and this position is adopted here. Like U. tomdarraghi sp. nov., the spire of U. fodinata protrudes beyond the last shell whorl, but it is readily distinguished from U. tomdarraghi sp. nov. by its smaller size (72 mm), shell form (W/L 65%; H/L 54%), well-developed fossula, less extended posterior extremity, more tapered
Fine, restricted to aperture.
Fine, restricted to aperture.
Less convex and broader than U. h. hesitata; flattened anteriorly.

Widening slightly towards anterior
Narrower than U. h. hesitata.
Narrower than U. h. hesitata; slightly constricted at anterior end of posterior canal.
Spire less impressed, broader. Protoconch much broader, protruding; positioned towards left side ( Fig. 2A, B). and shorter anterior extremity, and in the structure of the anterior-most columellar teeth, which are longer and coarser than those elsewhere on the columella and extend further onto the base (Darragh 2011). Pliocene strata of the Cameron Inlet Formation at Flinders Island, off the northeast coast of Tasmania, around 2,000 km from the Roe Plains, contain at least three species of cowries, including two species of Umbilia: U. furneauxensis Southgate et al., 2021 andU. hesitata (Sutherland andKershaw 1970;Darragh 1985;Southgate et al. 2021;Southgate and Roberts 2022). Umbilia furneauxensis differs from U. tomdarraghi sp. nov. by its much smaller size (<60 mm), extension of apertural dentition to at least midway on the base and labrum, and heavily callused margins that may form a dimpled surface extending to the base. Flinders Island fossils assigned to U. hesitata can be distinguished from U. tomdarraghi sp. nov. by their much greater extension of both anterior and posterior extremities, restriction of their finer columellar teeth to the aperture, lengthening of the anterior-most labral teeth, a more convex base, and a more umbilicate spire with less prominent protoconch. However, comparison of U. tomdarraghi sp. nov. with U. hesitata from the Cameron Inlet Formation is done with caution at this stage because available specimens (n = 16, MV collection) vary considerably in shell form and conchological features to such an extent that they may not represent a single taxon.
Umbilia tomdarraghi sp. nov. superficially resembles the Miocene species U. eximia (G.B. Sowerby I, 1845) and U. hallani Hawke, 2020. It differs from the former by its much less prominent extremities, lack of prominent anterior dorsal tubercules, and by differences in the structure of the columellar teeth which, in U. eximia, are generally broad, deeply incised, and rectangular in cross section. Umbilia hallani is readily distinguished from U. tomdarraghi sp. nov. by its much smaller size with a more inflated body whorl and rostrate anterior extremity, and by the moderately formed anterior dorsal tubercules, separated by a diagonal groove.
Etymology. Named to honour Dr T.A. Darragh, invertebrate paleontologist at Museums Victoria, Melbourne, Australia, in recognition of his significant contribution to our understanding of Australian marine molluscs, both fossil and living.
Distribution. Known only from the Roe Calcarenite of the Roe Plains, Western Australia.

Key to Pliocene Umbilia species
The four known Pliocene Umbilia species are described in the following key.

Discussion
The genus Umbilia is well represented in the fossil record with at least 11 recognised species. Of the five extant Umbilia species (Lorenz 2017), U. hesitata is the only one with a fossil record (Darragh 2002;Goudey 2015). Darragh (2002) noted that although Late Miocene and Early Pliocene specimens of U. hesitata are uncommon, and often fragmentary or poorly preserved, sufficient well-preserved material is available to confirm their identification. Although similar to U. hesitata and previously assigned to that species (Darragh 2002;Wilson and Clarkson 2004;Goudey 2015;Lorenz 2018), U. tomdarraghi sp. nov. is morphometrically distinct from U. hesitata and differs in shell form and key conchological features, some of which (e.g., coarser teeth and flatter base) have been noted in previous studies (Wilson and Clarkson 2004;Lorenz 2017). As outlined above, there has been speculation in prior studies regarding the relationship between U. tomdarraghi sp. nov., living populations of U. hesitata found to the east, and living U. armeniaca found in adjacent waters in southern Western Australia. Our results clarify that U. tomdarraghi sp. nov. has greater affinity with U. hesitata than with U. armeniaca. Of the U. hesitata subspecies, our results also show closest affinity between U. tomdarraghi sp. nov. and the western most subspecies of U. hesitata, U. h. suprastrata, which has the closest natural range to the Roe Plains. Possible lineages within the Umbilia have been a source of speculation in a number of studies (e.g., Darragh 2002;Wilson and Clarkson 2004;Yates 2008;Hawke 2020). Darragh (2002) noted that U. hesitata probably descended from U. eximia, and Wilson and Clarkson (2004) reasoned that there was progressive change from the ancestral U. eximia towards contemporary U. hesitata, where intermediate stages within this lineage were represented by separate species. In considering the likely position of U. tomdarraghi sp. nov. within this lineage, the possibility of a close ancestral relationship between U. tomdarraghi sp. nov. and living U. hesitata was considered a likely scenario at the start of this study. However, accepting the existence of U. hesitata within both the Miocene and Pliocene (Darragh 2002), and considering the clear morphometric separation of U. tomdarraghi sp. nov. and extant U. hesitata shown in this study, we consider it more likely that U. tomdarraghi sp. nov. is an offshoot from the U. eximia-U. hesitata lineage that became extinct in the Pliocene. Our results suggest that U. hesitata may not be present among the fauna of the Roe Calcarenite, greatly reducing the previously accepted distribution of this species within the fossil record.
Particular thanks go to Mr Chris Goudey, who assisted with specimen sourcing, data collection, and other inputs to this study. We thank Dr Adam Yates and Dr Felix Lorenz for their constructive inputs during review of this manuscript.