Rhagomys, Thomas, 1917

Pardiñas, Ulyses F. J., Tinoco, Nicolás, Barbière, Franck, Ronez, Christophe, Cañón, Carola, Lessa, Gisele, Koch, Claudia & Brito, Jorge, 2022, Morphological disparity in a hyperdiverse mammal clade: a new morphotype and tribe of Neotropical cricetids, Zoological Journal of the Linnean Society 196, pp. 1013-1038 : 1028-1032

RHAGOMYS : GENERIC CONTENT AND UNIQUENESS

Ample evidence suggests that Rhagomys , as currently understood (i.e. composed of three species; see: Moreno Cárdenas et al., 2021), represents two generic entities. The point is controversial because it relies on epistemological aspects, in particular what degree of morphological/genetical distance can be allocated to a genus before its value is diluted [see a discussion of this issue applied to Oryzomyini rodents in Weksler (2006)]. Differences between R. rufescens and ( R. longilingua + R. septentrionalis ) include trenchant morphological (e.g. absence vs. presence of spiny fur, presence vs. absence of the anterolingual conule of the M1), genetical (> 17% in Cytb distance between species pairs, a value well above the median recorded for the order Rodentia ; see: Schrago & Mello, 2020) and geographical (Atlantic Interior Forest vs. eastern Andean forest and contiguous Amazonian lowlands) contrasts. We suggest that western Rhagomys populations, from northern Bolivia ( Villalpando et al., 2006) to south-eastern Ecuador (this paper), including two different species, deserve a new generic assignment. A similar case is exemplified by the erection of the murine Musseromys Heaney et al., 2009 despite its clear morphological and genetical closeness to Carpomys Thomas, 1895 . Heaney et al. (2009: 215) were adamant, indicating that ‘although our molecular data strongly suggest that Carpomys is the closest relative of the Banahaw mouse [i.e. Musseromys ], on the basis of these 16 characters, we reject the option of considering them to be members of a single genus’. Most of these 16 characters, if not all, can be judged as subtle differences (e.g. lower incisors more slender, coronoid process of the mandible shorter and more sharply pointed) in the context of those separating western vs. eastern species in Rhagomys . It is vox populi that R. longilingua was originally stated as a type species of a new genus, but because of subsequent work was finally placed in Rhagomys . We hope that the morphological scrutiny developed in the present contribution triggers a reappraisal of that pending taxonomic issue.

Uncertainties about Rhagomys relationships have been largely due to the poor quality and scarcity of the available material. Although the type species was described by the end of the 19 th century ( Thomas, 1886), the basis for the recognition of the genus was a second animal acquired decades later ( Thomas, 1917). As has been reviewed in detail by previous authors ( Luna & Patterson, 2003: 13–17; Percequillo et al., 2004: 253), the acute perception of O. Thomas pointed to connect Rhagomys with oryzomyines and thomasomyines. This author was mostly guided by external morphology because not only the cranium of the holotype is badly preserved (e.g. Gyldenstolpe, 1932: pl. III, fig. 4), but also skin features were highlighted in successive diagnoses ( Thomas, 1886, 1917). In any case, Thomas (1917: 193) was unusually expressive, pointing to the uniqueness of the genus as reflected in its molars: ‘The teeth are therefore almost as in certain Phyllostomid bats, with smooth glossy surface and simple conical cusps, which are evenly spaced- slightly slanted backwards, 6, 4, and 2 in number on the three teeth. Below the teeth are similarly modified, the cusps slanting forwards.’ He added, ‘Still younger specimens of Rhagomys will be welcome to show what trace of the normal foldings and ridges is exhibited by the molars when absolutely unworn; but it is evident there cannot be much.’

Luna & Patterson (2003: 19) hypothesized that the unique morphology represented by the incisors of R. longilingua is related to a singular feeding activity: the collection of invertebrates boring bamboo canes. We elaborate on this proposal here: the entire morphotype of Rhagomys can be understood as a specialization for living and exploiting dense understory habitats consisting mostly of reeds, bamboos and several similar plants (e.g. Gynerium Willd. ex P.Beauv. , Poaceae ). Bamboos cover extensive areas in South America (> 100 000 km 2), including a diversity of about 20 genera and 430 species ( Judziewicz et al., 1999). Even discarding Patagonian records that traced the history of these unique grasses to the Eocene, it is easy to envision the abundance of this kind of vegetation on the continent at least since the Pliocene ( Wilf, 2020). The assumption that a sigmodontine specializes in exploiting bamboo directly or indirectly as a resource is not far-fetched because other rodents have also followed this path (e.g. murids in Asia such as Hapalomys Blyth, 1859 or Kadarsanomys Musser, 1981 ; see: Bartels, 1937; Musser, 1972). Field observations of R. septentrionalis suggest that this animal lives in dense riverine understory, which consists mainly of bamboos and Gynerium , and the latter is used to construct spherical nests about 1 or 2 m high (Supporting Information, Appendix S18). Collection data for R. longilingua also support a close association with bamboos ( Luna & Patterson, 2003), and the same was found for several of the trapping sites in the case of R. rufescens (e.g. Steiner-Souza et al., 2008).

ARBOREAL SIGMODONTINES: TWO STRATEGIES

It is a biological commonplace that different strategies, reflected in differential phenotypes, allow organisms to face similar challenges (e.g. McGhee, 2011). Focusing on sigmodontine rodents, this axiom was masterfully addressed by Hershkovitz (1966: 95) regarding fossoriality, when he wrote: ‘Fossorial cricetines, like fossorial rodents in general, are of two functional types. The first is shrew- or mole-like superficially but with weak teeth and jaws and is primarily insect and worm eating. The second is pocket gopher-like with strong teeth and jaws and is primarily grass, forb and root eating. Kunsia Lichtenstein, 1830 is the outstanding, if not the only living gopher-like cricetine. Soricine or talpine cricetines are represented by neotropical oxymycterines and many akodonts, most notably the monotypic Blarinomys Thomas, 1896 . They are characterized by thick, soft pelage, weak zygomata and mandible. They are generally small and wiry. Gopher-like rodents are comparatively large and stocky and nearly all their specialized characters … contrast with those of the soricine type.’

To date, arboreal-scansorial sigmodontine rodents have been viewed as a single morphotype characterized by a combination of external traits, including ‘… long semi-prehensile tails covered, at least terminally, with long tactile hairs … feet are broad, with long and partially opposable outer digits, recurved claws, and enlarged plantar tubercles adapted for clutching slender branches and twigs. The short muzzle and enlarged eyes of these arboreal mice afford a wide field of frontal vision’ ( Hershkovitz, 1969: 42). Members of several tribes, especially oryzomyines and thomasomyines, but also euneomyines ( Irenomys Thomas, 1919 ), wiedomyines (e.g. Juliomys González, 2000 ), etc., display this suite of features (e.g. Pearson, 1983; Tribe, 1996; Rivas & Linares, 2006).

Here we assume that Rhagomyini represent a second arboreal morphotype that evolved to face the exploitation of the vertical dimension in dense understory, which is mostly dominated by woody plants characterized by moderately thin branches crossing in all directions. To deal with such a habitat, external specializations are likely to be different from those traditionally observed in arboreal sigmodontines ( Hershkovitz, 1969). In this context, manual grasping would be enhanced by a larger system of pads, calluses and grooves, and a shortening of the claws on both manus and pes. Contrary, perhaps, to expectations, the tail has lost its traditional role in balancing ( Cartmill, 1985) through a reduction in length and apical hairiness ( Fig. 12 ).

The Rhagomyini morphotype seems to favour not solely climbing activities but also a particular trophic behaviour associated with an arboreal mode of life: insectivory. The stomach contents of R. longilingua included insects, and the powerful musculature of this mouse has been associated with the ability for cane boring ( Luna & Patterson, 2003: 19). The exclusive incisor features shown by Rhagomys , especially the fine groove and the cutting-edge indentation of the lower incisor, point in the same direction. Functionally, incisor longitudinal striae have been associated with strengthening the tooth while improving retraction when gnawing woody material ( Russell, 1968; Akersten, 1973). Both are desirable qualities for a rodent determined to pierce the hard exterior of cane, presumably to extract boring invertebrates. In this context, manus morphology deserves special attention; they are so impressive when the animal is alive that the suspicion that some other role beyond climbing is involved seems certain (Supporting Information, Appendix S3). Cartmill (1974: 69) emphasized: ‘Forestfloor predators may climb to nose out insects concealed in the dense crowns of palms and similar plants … but their search procedures are poorly adapted to stalking insects amid the slender branches of the canopy and shrub layers. The insects which abound in these strata are attacked by small predators … which locate and track prey with their eyes and seize it with their hands ….’ We believe that the combination of insectivory and arboreality triggered the unique suite of specializations observed in Rhagomyini , which include not only cheiridia or tail features, but also large eyes, a short snout, etc. This morphofunctional pathway is uniquely present in these rodents within the sigmodontine universe; in contrast, all other known arboreal members of the subfamily have focused on the consumption of vegetal items ( Pardiñas et al., 2017).

FACING MORPHOLOGICAL DISPARITY IN SIGMODONTINES

Morphological disparity underpins the entire sigmodontine radiation and clearly constitutes one of the biggest challenges to understanding its evolutionary history, and encapsulating that history in a classificatory scheme. The last two decades have centred on three strategies to deal, at the classificatory level, with this issue: (1) expand already-named tribes to accommodate morphologically divergent taxa (e.g. Smith & Patton, 1999; D’Elía et al., 2006; Gonçalves et al., 2020); (2) erect new tribes to accommodate these taxa (e.g. Pardiñas et al., 2015; Salazar Bravo et al., 2016); or (3) retain these taxa as ‘unique lineages’ ( Sigmodontinae incertae sedis) while awaiting future data (e.g. Voss, 1993; Ventura et al., 2013). Strategies 1 and 2 are permeated by the same basic question: what degree of morphological disparity is acceptable before tribal unity (= diagnosability) is broken? When Smith & Patton (1999) decided to expand Akodontini to include Scapteromyini, the nature of the evidence employed (cytochrome b) meant that the above-mentioned question was not addressed. On the contrary, in the recent proposed expansion of the Wiedomyini to include two sylvan enigmatic genera, Juliomys and Phaenomys Thomas, 1917 (see: Gonçalves et al., 2020), morphological evidence was barely discussed. As such, the resulting new concept of Wiedomyini implies a high degree of variability reflected in the main anatomical systems analysed and the same is true for the current Akodontini . When D’Elía et al. (2006: 562) assigned Rhagomys to Thomasomyini , they did mention morphological differences highlighted by Pacheco (2003), but chose to base their final decision on the shared molecular characters.

Voss (1993: 25) was adamant about how to deal with these ‘refractory’ unique genera: ‘ Delomys is one of many pentalophodont genera that cannot be assigned to any demonstrably monophyletic taxon less inclusive than the Neotropical muroid ingroup identified earlier …. Placing these taxa in a formally named tribe, such as the Thomasomyini … would be a mistake because, inevitably, all coordinate taxa are treated equivalently by nonsystematists. Equally useless is the option of erecting a monotypic tribe for each genus judged to have achieved some arbitrary degree of difference.’ However, knowledge accumulated over the last three decades since Voss’s publication, particularly with the advent of molecular data, has produced a refreshed impetus to face these questions. Contrary to the opinion, popularized by several taxonomists, that people simply name clades because they like to name organisms (vox populi), recognition by naming distinctive tribes has intrinsic value. The case of the sigmodontine Reithrodon Waterhouse, 1837 exemplifies this point. This rat was early assigned to a ‘sigmodont’ group ( Hershkovitz, 1955), but shortly thereafter was recognized as a separate tribe ( Vorontsov, 1959, 1967, 1982). Several morphologically based contributions cemented its assignment to Phyllotini (e.g. Olds & Anderson, 1989; Braun, 1993; Steppan, 1995), but when molecular data retrieved Reithrodon as a separate branch of the sigmodontine radiation (e.g. Smith & Patton, 1999; Jansa & Weksler, 2004), Reithrodontini were rapidly resurrected ( Musser & Carleton, 2005; Cazzaniga et al., 2019). However, if Vorontsov (1959) had not created the tribe, certainly Reithrodon would be a ‘unique lineage’ by now. Nothing impedes a classification composed of tribes and isolated genera, since the latter can (at least potentially) be understood as unnamed monotypic tribes. However, when the history of a clade is added to the equation, the option to maintain numerous morphologically divergent genera seems a poor strategy. Revisionary studies on fossils originally envisioned as Phyllotini , such as Auliscomys osvaldoreigi Quintana, 2002 and Olympicomys vossi Steppan & Pardiñas, 1998 (Barbière, 2019; Barbière et al., 2021b), support placement of both taxa in Reithrodontini (see also: Barbière et al., 2016). In this new enlarged context, and monotypic in its recent expression, the tribe acquires a new dimension to analyse the structure of the entire radiation.

The case we address herein, a disparate morphology ( Rhagomys ) subsumed into a recognized tribe ( Thomasomyini ) by molecular linkage, is not an exception: the taxonomic history of sigmodontines offers numerous other examples. Most of these have been recognized by previous studies (e.g. Gonçalves et al., 2020), including relationships resolved as sisters between Delomys and Phyllotini (a result repeatedly reported since Jansa & Weksler, 2004) or Abrawayaomys and Akodontini (e.g. Ventura et al., 2013). These seem to be iconic examples of pairs where the degree of morphological disparity is so great that no attempt is made to expand the tribes to resolve these relationships. In other words, returning to the case of Delomys commented on above, it was never proposed to be included in Phyllotini . Including Delomys within Phyllotini would be destructive, as it would eliminate an otherwise morphological cohesion. This suggestion is an eloquent indication that molecular-based associations have a limit, and that limit, beyond the fact that we may not like to recognize it, is arbitrary.

There are many other overlooked examples of morphological disparity throughout the sigmodontines, and we are persuaded that these must be carefully reviewed if we are to obtain a complete understanding of the radiation. Examples of neglected cases include, among others, the phylogenetic association of Scolomys Anthony, 1924 and Zygodontomys J.A. Allen, 1897 with the remainder of the oryzomyines (e.g. Weksler, 2003, 2006; Pine et al., 2012; Percequillo et al., 2021). Both genera are so distinct in many aspects of their morphology ( Voss, 1991; Patton & da Silva, 1995), that they probably each deserve a monotypic tribe. This does not imply that one should ignore their basal relationship with the oryzomyines, as the same is valid between Rhagomyini and Thomasomyini .

Nevertheless, the most challenging situation is our current understanding of the main dichotomy embraced by the subfamily: Oryzomyalia vs. Sigmodontalia. This binary scheme, revealed and propagated by seminal genetic studies (e.g. Engel et al., 1998; Casavant et al., 2000; Steppan et al., 2004; Leite et al., 2014), hides a paramount case of morphological disparity: the sister-relationship between the two components of Sigmodontalia. They are the residue of the classical ‘sigmodont’ group of Hershkovitz (1955), limited to Sigmodon Say & Ord, 1825 as the only living element of Sigmodontini and the members of Ichthyomyini . It is difficult to imagine a more trenchant morphological distance than that involved in this tribal dichotomy. It can be argued that Sigmodon represents a conservative, grazing sigmodontine ( Voss, 1992), while ichthyomyines are a hyperspecialized tribe focused on freshwater carnivory ( Voss, 1988), but this does not necessarily explain why they are separated by a morphological gap of such magnitude. To suppose that ‘something’ (one or more genera or tribes) is missing from this evolutionary puzzle (more probably extinct than awaiting discovery) emerges as a crucial necessity. Because Sigmodon is widespread in North America and easy to catch and breed, it has been intensively studied and emerges as the emblematic sigmodontine capable of biologically representing the entire subfamily (e.g. Rinker, 1954; Hershkovitz, 1962; Martin, 1979; Peppers et al., 2002). However, to focus only on the molar pattern of this genus resists any simplistic approach and results in a rara avis ( Barbière et al., 2019).

According to the most recent phylogenetic reconstruction, intendedasthe‘backbone’ofsigmodontine radiation ( Parada et al., 2021), Sigmodontini and Ichthyomyini have a long history as separate branches, with an estimated split prior to the estimated time of diversification of the Oryzomyalia (see also: Steppan et al., 2004; Schenk et al., 2013). As a first step toward improving this molecular ‘backbone’, we propose to change our current understanding of sigmodontine radiation to a ternary scheme. In this new scenario, Sigmodontini , Ichthyomyini and Oryzomyalia are comparable evolutionary units, each encompassing a long and idiosyncratic history of diversification (represented by fossil and living records). While subtle, this proposal is crucial because it enriches our understanding of the full complement of diversity within the clade. That morphological disparity should be treated with the same emphasis that is currently given to molecular evidence is the substantive theme of our contribution.

Figures

Figure 12. Contrasting selected external features between two morphotypes evolved to face arboreal mode of life in sigmodontine rodents.