Taxonomic review of genus Sooretamys Weksler, Percequillo & Voss (Rodentia: Cricetidae: Sigmodontinae): an integrative approach

Transcription

1 bs_bs_banner Zoological Journal of the Linnean Society, 2014, 171, With 12 figures Taxonomic review of genus Sooretamys Weksler, Percequillo & Voss (Rodentia: Cricetidae: Sigmodontinae): an integrative approach ELISANDRA DE ALMEIDA CHIQUITO 1, GUILLERMO D ELÍA 2 and ALEXANDRE REIS PERCEQUILLO 1 * 1 Departamento de Ciências Biológicas, Escola Superior de Agricultura Luiz de Queiroz, Universidade de São Paulo, Av. Pádua Dias, 11, Caixa Postal 9, Piracicaba, São Paulo, Brazil 2 Instituto de Ciencias Ambientales y Evolutivas, Universidad Austral de Chile, campus Isla Teja s/n, Valdivia, Chile Received 26 July 2013; revised 5 February 2014; accepted for publication 7 February 2014 Sooretamys is a monotypic genus of the family Cricetidae, subfamily Sigmodontinae, that is distributed throughout eastern South America in the Atlantic Forest Biome, including Argentina, Brazil, and Paraguay. The taxonomic history of the forms associated with this genus is long and relatively complex, and few studies have evaluated the taxonomic problems of this genus. To this end, our goal was to describe the degree and geographical pattern of morphological and molecular variation in this genus to test the current hypothesis that the genus is monotypic, and, as a consequence, to determine the status of the nominal forms associated with Sooretamys. Accordingly, we employed morphometric, morphological, and molecular tools, according to an integrative taxonomy approach. The results show that some level of morphometric discontinuity is present between the individuals from Paraguay and those from adjacent localities in Brazil and Argentina; sharp discontinuities were not observed in qualitative traits. Molecular analyses of the mitochondrial cytochrome b gene showed that the Paraguayan populations have some degree of genetic differentiation, but the haplotypic variants do not form a monophyletic group. Thus, the evidence so far suggests a difference in the genes and morphology of the Paraguayan population, but there is no consistent resolution (e.g. lack of monophyly) to show that specimens from Paraguay represent a distinct population that would merit taxonomic recognition. Thus, we recognize a single species within the genus Sooretamys, named Sooretamys angouya. The pattern of morphological and genetic differentiation of Sooretamys could be the result of divergence with gene flow. However, our data also correspond in some aspects with the model advanced by Carnaval & Moritz, which claims the existence of stable Atlantic Forest areas where the forest biota persisted during the Quaternary climatic fluctuations. Whatever process has occurred, S. angouya represents one species with a complex evolutionary history, and the analysis of additional samples would be welcome to further elucidate the process of diversification of this taxon.. doi: /zoj ADDITIONAL KEYWORDS: biogeography cytochrome b geographical variation morphology morphometry Neotropical region Oryzomyini phylogeography South America taxonomy. INTRODUCTION The tribe Oryzomyini consists of 30 living genera and c. 125 species (Weksler, Percequillo & Voss, 2006; *Corresponding author. percequillo@usp.br Percequillo, Weksler & Costa, 2011; Weksler & Percequillo, 2011) and is the most diverse and widely distributed of the subfamily Sigmodontinae (Cricetidae). Its area of distribution ranges from the south-eastern USA to Tierra del Fuego and neighbouring islands in the southernmost portion of South America. Species 842

2 INTEGRATIVE TAXONOMY OF GENUS SOORETAMYS 843 of Oryzomyini occupy all biomes existing in this large area (Prado & Percequillo, 2013), and most species are terrestrial, although scansorial, arboreal, and semiaquatic forms also exist (Musser et al., 1998; Weksler, 2006). The oryzomyine genus Sooretamys (Weksler et al., 2006) inhabits the Atlantic Forest, one of the most diverse tropical forests in the world (Fonseca, 1985; Myers et al., 2000; Ribeiro et al., 2009; Percequillo et al., 2011). Known collecting localities for this genus extend from southern Espírito Santo to northern Rio Grande do Sul in eastern Brazil, reaching Misiones in northeastern Argentina and several departments in western and eastern Paraguay (Myers, 1982; Musser & Carleton, 2005; D Elía et al., 2008; Prado & Percequillo, 2013). The species, Sooretamys angouya (sensu Musser & Carleton, 2005; Weksler et al., 2006), and its associated name Mus angouya Fischer, 1814, has a long and relatively complex taxonomic history. Other names currently associated with Sooretamys, namely Hesperomys leucogaster Wagner, 1845, Hesperomys ratticeps Hensel, 1872, Calomys rex Winge, 1888, Oryzomys ratticeps tropicius Thomas, 1924, and Oryzomys ratticeps paraganus Thomas, 1924, have simpler histories, and their link to Sooretamys resulted from type specimenbased studies (Musser et al., 1998; Musser & Carleton, 2005). However, detailed morphometric and morphological comparisons have not yet been performed. Currently, all aforementioned names are considered to be synonyms of S. angouya, and no subspecies are recognized (Musser & Carleton, 2005; Weksler & Percequillo, 2011). The unique hypothesis regarding the evolutionary history of this species and molecular variation amongst populations (Miranda et al., 2007) considers, mainly, the southern portion of the species distribution in Brazil and does not discuss taxonomic and nomenclatural issues. Instead, only phylogeographical issues have been evaluated, focusing on geographical structure and genetic divergence. Consequently, in the absence of a formal revision of the genus, the status of the species-group names associated with this generic taxon is pending assessment. Recently, a new taxonomic approach, so-called integrative taxonomy (Dayrat, 2005; Padial et al., 2009), suggests the combination of several sources of evidence and evolutionary theories to clarify taxonomic questions and evolutionary history (see also Patton et al., 1997). However, it must be noted that this approach is by no means new to sigmodontine taxonomy: for more than two decades, researchers have integrated morphological, karyotypic, and molecular data (e.g. Patton & Hafner, 1983; Patton, da Silva & Malcolm, 2000; Pardiñas, D Elía & Cirignoli, 2003; D Elía & Pardiñas, 2004; Pardiñas et al., 2005; Percequillo, Hingst-Zaher & Bonvicino, 2008; Percequillo et al., 2011). Our goal in this study was to describe the degree and geographical pattern of morphological and molecular variation in the genus Sooretamys to test the current hypothesis that the genus is monotypic and, as such, to determine the status of the nominal forms associated with Sooretamys. Similarly, we aim to provide information on the evolutionary history of this endemic Atlantic Forest genus. MATERIAL AND METHODS SPECIMENS AND SAMPLING We studied 478 specimens (Appendix 1) of the genus Sooretamys housed in the following museums and institutional collections: American Museum of Natural History, New York (AMNH); Departamento de Genética da Universidade Federal do Rio Grande do Sul, Porto Alegre (UFRGS); Departamento de Sistemática e Ecologia da Universidade Federal da Paraíba, João Pessoa (UFPB); Facultad de Ciencias Naturales y Museo, Universidad Nacional La Plata, La Plata (MLP); Fundação Universidade Regional de Blumenau, Blumenau (FURB); Fundação Zoobotânica, Porto Alegre (FZB/RS); Instituto Nacional de Pesquisas da Amazônia, Manaus (INPA); Laboratório de Mamíferos Aquáticos, Universidade Federal de Santa Catarina, Florianópolis (UFSC); Museo Argentino de Ciencias Naturales Bernardino Rivadavia, Buenos Aires (MACN); Museu de Ciências Naturais da Universidade Luterana do Brasil, Canoas (MCNU); Museu de Historia Natural Capão da Imbuia, Curitiba (MHNCI); Museu de Zoologia da Universidade de São Paulo, São Paulo (MZUSP); Museu Nacional da Universidade Federal do Rio de Janeiro, Rio de Janeiro (MN); Museu Paraense Emilio Goeldi, Belém (MPEG); Museum of Comparative Zoology, Harvard University, Cambridge (MCZ); Museum of Vertebrate Zoology, University of California, Berkeley (MVZ); Museum of Zoology of University of Michigan, Ann Arbor (UMMZ); National Museum of Natural History, Smithsonian Institution, Washington, D.C. (NMNH); Naturistorisches Museum Wien, Wien (NMW); The Natural History Museum, London (NHM). We also analysed un-catalogued specimens provided by Meika Alessandra Mustrangi (MAM), Alexandre Uarth Christoff (AUC), Renata Pardini (RP), Ana Cristina Monteiro Leonel (ACL), and Estação Ecológica do Bananal (EEB), housed at MZUSP and Guillermo D Elía (GD), housed at Colección de Mamíferos, Universidad Austral de Chile. In addition, tissue samples were loaned by institutions, such as Texas Tech Museum (TK), and collectors, such as Marcelo Passamani (MP), Ricardo Siqueira Bovendorp (RSB), and Cibele Rodrigues Bonvicino (CRB). Two acronyms of sequences downloaded from GenBank could not be identified: AFV (Miranda et al., 2007) and EM (Bonvicino & Moreira, 2001).

3 844 E. A. CHIQUITO ET AL. Not all specimens were included in both approaches: the morphological (either quantitative or qualitative) and molecular ones. For instance, 22 specimens were analysed by both approaches, 29 only for the molecular one because of local small sample sizes or age criteria not being useful in the morphological approach, and 427 only morphologically, with no tissues available, represented by old specimens from museums. Appendix 1 lists the samples used in each analysis performed throughout this study. Type specimens of most nominal forms associated with S. angouya were assessed in the study. Regrettably, no tissue samples from these specimens were available; however, specimens from localities close to the type localities of most nominal forms were included in the analyses. For all collecting localities, we obtained information on geographical coordinates and altitude (when available) from collector labels, online databases (National Geospatial-Intelligence Agency website: geonames.nga.mil/namesgaz), or gazetteers (Paynter Jr., 1989, 1995; Paynter Jr. & Traylor Jr., 1991); data were organized in a Gazetteer (Appendix 2). To obtain larger sample sizes with at least four adult individuals for the statistical analyses, specimens from different collecting localities were pooled according to the criteria established by Vanzolini & Williams (1970): close geographical proximity, absence of major geographical barriers amongst localities, such as altitudinal levels or major rivers, and lack of obvious discrepancy in size and shape amongst contiguous samples. Considering these criteria, we were able to group specimens into 15 samples that cover most of the geographical distribution of the genus (Table 1, Fig. 1). Comparative analyses amongst geographical samples were performed using the method of transects (Vanzolini, 1970; Vanzolini & Williams, 1970), which are a series of localities more or less linearly arranged between, and including, major samples. This method is used to recognize sharp discontinuities throughout the geographical range of collection samples. In addition, we mapped the frequency of qualitative and quantitative characters throughout the geographical samples available, as performed previously by Musser (1968). We established two transects: the first, which connects samples along the Atlantic Forest, is associated with the hills and highlands of Serra do Mar across a latitudinal gradient, ranging from warmer evergreen forest (in the northern localities) to colder evergreen forests, some mixed with Brazilian Pine, Araucaria angustifolia, in the south; the second transect is orientated along a vegetation and moisture gradient, from the coastal region, with evergreen forests, to the interior forests, with drier and semideciduous forests, as well as riparian grasslands of the Paraná and Paraguay river basins. MORPHOLOGICAL-BASED ANALYSES Several qualitative traits were surveyed, but after discarding those that were nonvariable and those that showed no informative variation, we analysed the following characteristics: colour of dorsal and ventral body surfaces, length of tufts of ungual hair relative to the length of claws, dorsoventral countershading of the tail, presence of anteromedian flexus in the anterocone of molar 1 (M1), and length of incisive foramina relative to M1 alveoli. To describe the qualitative character variation, we used the character states proposed by Hooper & Musser (1964), Carleton (1973, 1980), Reig (1977), Voss & Linzey (1981), Voss (1988, 1993), Carleton & Musser (1989), Voss & Carleton (1993), Steppan (1995), and Weksler (2006). Qualitatively, the samples exhibit a marked variation in relation to specimen age regarding pelage colour and texture, molar wear, suture fusion, and other cranial features. To standardize this variation, we used four age classes based on molar wear, modified from Voss (1991) as follows. Age class 1: first and second molars with no apparent wear, third molar usually non-erupted or newly erupted with main cusps still closed, labial lophs well developed and isolated, labial and lingual flexus deep and distinct. Age class 2: first and second molars with minor wear and small exposure of dentine, third molar already showing minimal to moderate wear, anteroloph and mesoloph may be connected to paracone through marginal lophules, posteroloph nearly fused to metacone, marginally. Age class 3: first and second molars with moderate wear; third molar with marked wear and a nearly flat surface; anteroloph and mesoloph fused marginally to paracone, forming long anterofosset and mesofosset, respectively; posteroloph completely fused to metacone, forming a distinct mesofosset. Age class 4: first and second molars with heavy wear, indistinct cusps, and massive exposure of dentine; third molar quite flat, with major exposure of dentine; anteroloph, mesoloph, and posteroloph indistinct, fused to major cusps. However, no noticeable variation in shape is present regarding sex, with males and females being quite similar. For morphometric analyses, 15 cranial measurements (Fig. 2) were obtained with digital callipers to the nearest 0.01 mm: occipitonasal length (ONL); condylo-incisive length (CIL), measured from the greater curvature of one upper incisor to the articular surface of the occipital condyle on the same side; length of diastema (LD), from the crown of the first upper molar to the lesser curvature of the upper incisor on the same side; length of molars (CLM1 3), crown length from molar 1 (M1) to molar 3 (M3); breadth of M1 (BM1), greatest crown breadth of the first maxillary molar across the paracone protocone; length of incisive foramen (LIF), greatest anterior posterior dimension

4 INTEGRATIVE TAXONOMY OF GENUS SOORETAMYS 845 Table 1. Available samples pooled to allow the definition of 15 large-size samples and the localities included. The numbers in parentheses are the sample size of adults in each pooled sample, and the numbers before the locality name correspond to the gazetteer Pooled samples (N) Localities (N) South of Minas Gerais (15) 43 Alto da Consulta, Poços de Caldas (3) 46 Morro do Ferro, Poços de Caldas (3) 48 Poços de Caldas (7) 49 Posses, 13 km south-east of Itanhandú (2) Boracéia-Casa Grande (16) 135 Boracéia, Estação Biológica de Boracéia (7) 139 Casa Grande, Biritiba-Mirim (9) Riacho Grande (8) 156 Furnas, Riacho Grande (5) 169 Paranapiacaba (2) 173 Ribeirão Pires (1) Cotia-Piedade (18) 142 Caucaia do Alto (7) 147 Cotia (8) 155 Furnas, Piedade (1) 170 Piedade, São Paulo (2) Upper Rio Paranapanema (22) 131 Apiaí (7) 136 Canaleta AB, Companhia de Cimento Nassau, Ribeirão Grande (1) 137 Canaleta T3,4, Ribeirão Grande (1) 144 Corrego Água Limpa, C C Nassau, Ribeirão Grande (1) 145 Corrego Barracão, Ribeirão Grande (1) 146 Corrego Fernandes, C C Nassau, Ribeirão Grande (1) 152 Fazenda Intervales Sede (2) 153 Fazenda Intervales, Base da Bocaina (2) 154 Fazenda Intervales, Base do Carmo (3) 162 Itararé (3) Upper Rio Iguaçú (11) 55 Guaricana, São José dos Pinhais (1) 65 Usina de Guaricana, São José dos Pinhais (1) 104 Bugre, Três Barras (1) 108 Estação Ecológica Bracinho/Piraí, Joinville (4) 111 Fazenda Sta Alice, Rio Negrinho (2) 112 Floresta Nacional Três Barras (1) 120 Reserva Biológica Sassafrás, Dr. Pedrinho (1) Rio Itajaí-Açú Basin (16) 102 Barragem do Garcia, Angelina (1) 103 Barragem do Rio São Bento, Siderópolis (1) 114 Hotel Plaza Caldas da Imperatriz, Caldas da Imperatriz (1) 117 Mono, Parque das Nascentes, Indaial (4) 119 Pinheiro Alto, Anitápolis (= Pinheiros, Alto-Anitápolis) (1) 124 Terceira Vargem, Parque das Nascentes, Blumenau (2) 128 Vale do Espingarda, Parque das Nascentes, Indaial (4) 129 Vale da Indústria de Fosfatados Catarinense, Anitápolis (2) Rio Tramandaí Basin (8) 82 Faxinal, Norte da Lagoa Itapeva, Torres (2) 85 Itapeva, Parque Estadual de Itapeva, Torres (1) 95 Pontal do Norte, Lagoa Palmital, Osório (2) 96 Pró-Mata, São Francisco de Paula (1) 99 Tramandaí, Lenha Seca, Lagoa de Tramandaí (1) 100 Vale da Encantada/Barra do Ouro, Maquiné (1) RioTaquari-Antas Basin (9) 79 Cruzeiro do Sul (6) 84 Guaporé (1) 92 Nova Roma do Sul (2) North-west Rio Grande do Sul (4) 78 Cruz Alta (1) 83 Fazenda Aldo Pinto, São Nicolau (1) 87 Lajeado Grande, Alpestre (1) 88 Lajeado Grande, Rio dos Índios (1) Ponta Grossa (5) 59 Parque Estadual Vila Velha, Ponta Grossa (5)

5 846 E. A. CHIQUITO ET AL. Table 1. Continued Pooled samples (N) Localities (N) Lower Rio Iguaçú (5) 54 Foz do Rio Capoteiro, Pinhão (2) 61 Reserva, Pinhão (1) 64 Usina Hidroelétrica Salto Caxias, Foz do Chopim, Cruzeiro do Iguaçú (1) 66 Vila UHE Segredo, Copel, Pinhão (1) West Santa Catarina (17) 106 Canteiro de Obras, Pequena Central Hidroelétrica Alto Irani, Arvoredo (2) 107 Canteiro de Obras, Pequena Central Hidroelétrica Plano Alto, Xavantina (1) 116 Linha Voltão, Xaxim (1) 126 Usina Hidroelétrica Itá (2) 127 Usina Hidroelétrica Quebra Queixo, Rio Xapecó, São Domingos/Ipuaçú (11) Misiones (9) 7 30 km from Puerto Bemberg, Rio Uruguay (Rio Urugua-í) (7) 8 60 km from Puerto Iguazú, Rio Iguazú (1) 12 Arroyo Uruguay, Departamento General Belgrano (1) 26 Junção dos rios Iguazú e Alto Paraná, Puerto Aguirre (1) 39 Tobuna, Depto. San Pedro (5) Río Tebicuary Basin (27) 184 Orillas del Río Tebicuary (1) km by road north of San Antonio (9) 186 Ayolas, 5 km by road east-north-east of Ayolas (4) 188 Costa del Río Tebicuary (1) 189 Orillas del Río Tebicuary (1) 190 San Ignacio (3) 192 Costa del Río Tebicuary (4) 195 Sapucay (4) BRAZIL S Minas Gerais Boracéia-Casa Grande PARAGUAY Ponta Grossa Rio Tebricuary basin Misiones Lower Rio Iguaçú Riacho Grande Cotia-Piedade Upper Rio Paranapanema Upper Rio Iguaçú W Santa Catarina ARGENTINA NW Rio Grande do Sul Rio Taquari-Antas basin Rio Itajaí-Açú basin ATLANTIC OCEAN Rio Tramandaí basin Figure 1. Samples of genus Sooretamys used in the morphometric and morphological analysis of geographical variation. Ovals indicate the samples that were pooled to allow the definition of 15 large-size samples (as explained in the text and Table 1). of one incisive foramen; breadth of incisive foramen (BIF), greatest dimension measured across the internal surface of both incisive foramen; breadth of rostrum (BR), greatest dimension measured across the external border of the nasolacrimal capsules; length of nasals (LN), greatest anterior posterior dimension of one nasal bone; length of palatal bridge (LPB), measured from the posterior border of the incisive foramen to

6 LN LIF INTEGRATIVE TAXONOMY OF GENUS SOORETAMYS 847 BR ZB LD LOF BZP ONL CZL CIL BIF BM1 Figure 2. Skull of an adult specimen of genus Sooretamys depicting the craniodental measurements employed in the morphometric analysis [specimen MZUSP 21919, from Casa Grande, Salesópolis, São Paulo, Brazil (ONL: mm)]. Abbreviations: BIF, breadth of incisive foramen (greatest dimension measured across the internal surface of both incisive foramen); BM1, breadth of molar 1 (greatest crown breadth of the first maxillary molar across the paracone protocone); BR, breadth of rostrum (greatest dimension measured across the external border of the nasolacrimal capsules); BZP, breadth of zygomatic plate (across central area of zygomatic plate); CIL, condyloincisive length (measured from the greater curvature of one upper incisor to the articular surface of the occipital condyle on the same side); CLM1 3, length of molars (crown length from molar 1 to molar 3); CZL, condylozygomatic length; HBC, height of braincase; LD, length of diastema (from the crown of the first upper molar to the lesser curvature of the upper incisor on the same side); LIF, length of incisive foramen (greatest anterior posterior dimension of one incisive foramen); LN, length of nasals (greatest anterior posterior dimension of one nasal bone); LOF, length of orbital fossa (greatest length of the orbital fossa between the squamosal and maxillary roots of the zygomatic arch); LPB, length of palatal bridge (measured from the posterior border of the incisive foramen to the anterior border of the mesopterygoid fossa); ONL, occipitonasal length; ZB, zygomatic breadth (greatest dimension across the squamosal root of zygomatic arches). LPB HBC CLM1-3 the anterior border of the mesopterygoid fossa; height of braincase (HBC); zygomatic breadth (ZB), greatest dimension across the squamosal root of zygomatic arches; breadth of zygomatic plate (BZP), across central area of zygomatic plate; condylozygomatic length (CZL); length of orbital fossa (LOF), greatest length of the orbital fossa between the squamosal and maxillary roots of the zygomatic arch. Initially, quantitative variables were tested for uniand multivariate normality using Kolmogorov Smirnov and Mardia s Kurtosis tests, respectively. Statistical differences amongst age and sexual classes were diagnosed by ANOVA and Tukey s test. In addition, sexual differences were also tested in adults using the Student s t test. We used a level of significance of 0.05 for all statistical analyses performed. We used distinct sample sizes depending on the analysis performed: to evaluate normality, we used samples with more than 25 specimens; to test dimorphism, our samples included at least eight specimens and contained equal proportions of males and females; and to study age variation, we used samples with more than one individual for each age class. Musser (1968) employed the frequency of morphological character states, both quantitative and qualitative, to assess the geographical variation of Sciurus aureogaster, establishing an important means of coupling geographical variation with nomenclatural attribution. Therefore, variation for each qualitative morphological character was codified in character states. Polymorphic traits were summarized by the frequency at which the state is observed in each geographical sample studied (Musser, 1968). We employed inferential error-bar diagrams (mean ± 95% of confidence interval; Simpson, Roe & Lewontin, 2003; Cumming, Fidler & Vaux, 2007) for the largest available samples of skull measurements (external dimensions were not included in statistical analyses as they were recorded by distinct collectors; descriptive statistics of external dimensions were used only in general comparisons). Multivariate analyses were performed with geographically pooled samples (Table 1, Fig. 1). Discriminant analyses (DA) using natural base logtransformed cranial measurements were performed comparing the 15 geographical samples simultaneously (Simpson et al., 2003; Manly, 2008). We also generated error bars with the scores of the first canonical function to evaluate the multivariate structure along the distribution of geographical samples. Specimens with missing values were discarded from both the univariate and multivariate analyses. To test the hypothesis that the differences amongst geographical samples could be explained by the isolationby-distance model (Wright, 1943), we compared geographical and morphometric distance matrices with a Mantel test (Mantel, 1967; Manly, 2008; Moreira &

7 848 E. A. CHIQUITO ET AL. Oliveira, 2011) with random permutations. Both matrices used in this test were calculated based on sample centroids, with geographical distances in kilometres and Mahalanobis distance for the morphometric matrix. The statistical analyses were performed in R (R Development Core Team, 2005), SPSS 13.0 for Windows (SPSS, Inc., 2004), and STATISTICA v. 9 (StatSoft, Inc. 2009). PHYLOGEOGRAPHICAL ANALYSIS The genetic analyses are based on 53 DNA sequences of Sooretamys (Table 2), 15 from GenBank, and from specimens gathered from 31 localities, covering almost entirely the geographical distribution of the genus (Fig. 3, Appendices 1, 2). The tree was rooted with the outgroup criterion, which was composed of sequences of one specimen of Nectomys squamipes (GenBank accession number AF181283), one of Aegialomys xanthaeolus (EU340015), and one of Cerradomys subflavus (AF181274), all of which are oryzomyines that are closely related to Sooretamys (Percequillo et al., 2011). DNA was extracted from liver samples using the ChargeSwitch gdna Mini Tissue Kit (Invitrogen). A 675-bp region of the cytochrome b gene was amplified via PCR using primers MVZ05 and MVZ16 (Smith & Patton, 1993). The annealing temperature was 52.5 C, the PCR programme was the same as D Elía (2003), and the PCR reaction mixture was as follows for a 25-μL total reaction volume: 2.5 μl primers (10 mol μl 1 ), 2.5 μl buffer (Invitrogen), 1.5 μl MgCl 2 (50 mm), 0.25 μl deoxynucleotide triphosphates (10 mm), 0.20 μl Platinum Taq DNA Polymerase, 1 μl DNA, and μl ultrapure water. Amplicons were purified with the UltraClean PCR Clean-Up Kit (Mo Bio Laboratories, Inc.) and sequenced using the DYEnamic ET Dye Terminator Kit (MegaBACE) in an ABI Prism 3100 Genetic Analyzer sequencer (Applied Biosystems) following the manufacturer s protocols. Chromatograms were checked and manually edited in CHROMAS LITE 2.01 software (Technelysium Pty Ltd, 2007). DNA sequences were aligned with ClustalW (Thompson, Higgins & Gibson, 1994) using the default alignment parameters and were corrected manually. Pairwise genetic distances were calculated in MEGA 5 (Tamura et al., 2011) as p-distances and Kimura two parameters (K2p) (Supporting Information Table S1). Gene trees were constructed using maximum parsimony (MP; Farris, 1982) and Bayesian analysis (BA; Yang & Rannala, 1997). MP analysis was performed in PAUP* (Swofford, 2000) with characters treated as unordered and equally weighted, 500 replicates of heuristic searches with random addition of sequences, and tree bisection reconnection branch swapping. The relative support of the recovered clades was calculated by performing 1000 bootstrap (BS) replications with five replicates of random sequence addition each. BA analysis was performed using MrBayes 3.1 (Ronquist & Huelsenbeck, 2003) by implementing a model of sequence evolution that includes six categories of base substitutions, a gamma-distributed rate parameter and a proportion of invariant sites. Uniform-interval priors were assumed for all parameters except base composition and generalized time-reversible parameters, which assumed a Dirichlet prior process. Two independent runs with four chains were allowed to proceed for generations with trees sampled every 100 generations. The first 25% of the trees were discarded as burn-in, and the remaining trees, sampled well after stationarity was reached, were used to compute posterior probability (PP) estimates for each clade. In addition, as another method of visualizing relationships amongst haplotypes of Sooretamys, a haplotype network was created via statistical parsimony (Templeton, Crandall & Sing, 1992) using the program TCS (Clement, Posada & Crandall, 2000). Finally, to further explore how the observed genetic variation is geographically structured, hierarchical analyses were conducted in the form of analysis of molecular variance (AMOVA; Excoffier, Smouse & Quattro, 1992) using ARLEQUIN v. 3.1 (Excoffier, Laval & Schneider, 2005). Distinct hierarchical haplotype arrangements were defined based on sampling localities (Fig. 3), and these were grouped by geographical regions, by groups found in the morphological analyses (see below), or to match the traditional taxonomic scheme of Thomas (1924). SPECIES AND SUBSPECIES DEFINITION We employed all of the aforementioned phenotypic traits to recover unique combinations of diagnostic characters and monophyletic lineages in order to recognize species in the genus Sooretamys, which is a procedure akin to the phylogenetic species concept advocated by Cracraft (1983). This concept favours diagnosability, as species are the smallest diagnosable cluster of organisms that exhibit a parental pattern of ancestry and descent. Although not in common usage under the phylogenetic species concept (e.g. Frost, Kluge & Hillis, 1992), the subspecies category could be employed eventually, upon the recognition of geographical lineages. According to Mayr (1963), subspecies are evolutionary unities only when they coincide with geographical isolates; otherwise, they represent a convenient system of taxonomic classification. Similarly, Frost et al. (1992) affirmed that subspecies can be either invented (mere artifacts of idealizing diagnosis ) or discoverable items ( temporarily isolated lineages ); if they represent discoverable entities, they are elements of

8 INTEGRATIVE TAXONOMY OF GENUS SOORETAMYS 849 Table 2. List of mitochondrial DNA sequences downloaded from GenBank and sequenced in the present study, with the respective accession number. Localities are mapped in Figure 3 Voucher/tissue no. GenBank accession no. BP Locality (locality number) Reference Ingroup RSB6595 KF Cotia, São Paulo, Brazil (147) Present study AFV21 EF Caxias do Sul, Rio Grande do Sul, Brazil (76) Miranda et al., 2007 AFV22 EF Caxias do Sul, Rio Grande do Sul, Brazil (76) Miranda et al., 2007 CNP1998 KF Refúgio Moconá, Misiones, Argentina (30) Present study CNP2524 KF Refúgio Moconá, Misiones, Argentina (30) Present study CRB1271 AF Teresópolis, Rio de Janeiro, Brazil (69) Bonvicino & Moreira, 2001 EM1207 AF Fazenda Intervales, São Paulo, Brazil (152) Bonvicino & Moreira, 2001 FURB12041 KF Terceira Vargem, Parque das Nascentes, Blumenau, Santa Catarina, Present study Brazil (124) FURB12151 KF Reserva Biológica Sassafrás, Dr. Pedrinho, Santa Catarina, Brazil (120) Present study FURB477 KF Vale do IFC, Anitápolis, Santa Catarina, Brazil (129) Present study FURB5070 KF UHE Itá, Itá, Santa Catarina, Brazil (126) Present study FURB5900 KF Vale do Espingarda, Parque das Nascentes, Indaial, Santa Catarina, Present study Brazil (128) FURB9230 KF UHE Quebra Queixo, São Domingos, Santa Catarina, Brazil (127) Present study FURB9238 KF UHE Quebra Queixo, São Domingos, Santa Catarina, Brazil (127) Present study FURB9252 KF UHE Quebra Queixo, São Domingos, Santa Catarina, Brazil (127) Present study FURB9696 KF Mono, Parque das Nascentes, Indaial, Santa Catarina, Brazil (117) Present study FURB9749 KF Reserva Particular do Patrimônio Natural Figueira Branca, Gaspar, Present study Santa Catarina, Brazil (121) FURB9790 KF Mono, Parque das Nascentes, Indaial, Santa Catarina, Brazil (117) Present study FURB9836 KF Mono, Parque das Nascentes, Indaial, Santa Catarina, Brazil (117) Present study FURB9867 KF Mono, Parque das Nascentes, Indaial, Santa Catarina, Brazil (117) Present study GD257 KF Paraguari, Paraguari, Paraguay (194) Present study GD265 KF Paraguari, Paraguari, Paraguay (194) Present study GD273 KF Centu-Cue, Misiones, Paraguay (187) Present study GD274 KF Centu-Cue, Misiones, Paraguay (187) Present study GD52 KF Paraguari, Paraguari, Paraguay (194) Present study GD543 KF Costa Norte, Paraguari, Paraguay (193) Present study MCNU1229 KF Nova Roma do Sul, Rio Grande do Sul, Brazil, (92) Present study MCNU1230 KF Nova Roma do Sul, Rio Grande do Sul, Brazil (92) Present study MCNU1291 KF Encruzilhada do Sul, Rio Grande do Sul, Brazil (81) Present study MCNU1292 KF Encruzilhada do Sul, Rio Grande do Sul, Brazil (81) Present study MCNU1625 KF Cachoeirinha, Rio Grande do Sul, Brazil (74) Present study MCNU622 KF Maquiné, Rio Grande do Sul, Brazil (89) Present study MHNCI4781 KF Córrego Barracão, Ribeirão Grande, São Paulo, Brazil (145) Present study MN37777 EF Florianópolis, Santa Catarina, Brazil (113) Miranda et al., 2007 MN37778 EF Torres, Rio Grande do Sul, Brazil (98) Miranda et al., 2007 MN37780 EF Torres, Rio Grande do Sul, Brazil (98) Miranda et al., 2007 MN37783 EF Torres, Rio Grande do Sul, Brazil (98) Miranda et al., 2007 MN37785 EF Tainhas, Rio Grande do Sul, Brazil (97) Miranda et al., 2007 MN37786 EF Mostardas, Rio Grande do Sul, Brazil (90) Miranda et al., 2007 MN37789 EF Mostardas, Rio Grande do Sul, Brazil (90) Miranda et al., 2007 MN37790 EF Tramandaí, Rio Grande do Sul, Brazil (99) Miranda et al., 2007 MN37794 EF Osório, Rio Grande do Sul, Brazil (93) Miranda et al., 2007 MN50234 EU Teresópolis, Rio de Janeiro, Brazil (69) Hanson, 2008 MP301 KF Itamonte, Minas Gerais, Brazil (44) Present study TK61763 EU Yacare, Ñeembucu, Paraguay (191) Hanson, 2008 UFPB335 EF Hotel Fazenda Monte Verde, Venda Nova, Espírito Santo, Brazil (41) Miranda et al., 2007 UFPB338 KF Hotel Fazenda Monte Verde, Venda Nova, Espírito Santo, Brazil (41) Present study UMMZ KF Costa Norte, Paraguari, Paraguay (193) Present study UMMZ KF Costa del Rio Tebicuary, Paraguari, Paraguay (192) Present study UMMZ KF Paraguari, Paraguari, Paraguay (194) Present study UMMZ KF Paraguari, Paraguari, Paraguay (194) Present study UMMZ KF Centu-Cue, Misiones, Paraguay (187) Present study UMMZ KF Costa Norte, Paraguari, Paraguay (193) Present study Outgroup CRB540 TK CEG42 AF (Nectomys rattus) EU (Aegialomys xanthaeolus) AF (Cerradomys subflavus) 801 Fazenda Da Mata, Maracajú, Mato Grosso do Sul, Brazil Bonvicino & Moreira, Bosque Protector Cerro Blanco, Guayas, Ecuador Hanson & Bradley, Parque Nacional do Rio Doce, Minas Gerais, Brazil Bonvicino & Moreira, 2001

9 850 E. A. CHIQUITO ET AL. ) BRAZIL PARAGUAY ATLANTIC OCEAN ARGENTINA Figure 3. Samples of genus Sooretamys used in phylogeographical analyses. The numbers correspond to the gazetteer, and those in parentheses are the sample size in each locality. evolutionary biology. Therefore, we aimed to identify diagnosable monophyletic lineages, whether species or subspecies (as geographical isolated lineages), through the morphological and molecular integrative approach here employed. RESULTS AGE AND SEX MORPHOLOGICAL VARIATION The nongeographical tests (age and sex variation) were applied to the largest available samples, depending on the analysis. Normality was verified in all of the samples tested: Boracéia-Casa Grande (N = 39), Cotia-Piedade (N = 59), Upper Rio Paranapanema (N = 38), Rio Itajaí- Açú Basin (N = 27), and Río Tebicuary Basin (N = 30). Only one variable (BIF) from Río Tebicuary Basin did not exhibit normal distribution (Z = 2.552, P < 0.05), but we chose not to exclude it from analysis because in all other samples BIF was normally distributed. Mardia s Kurtosis test was applied for all adult individuals and all variables; the coefficient obtained was (P > 0.05), which also indicated the multivariate normality of the data. ANOVA together with post hoc Tukey tests showed that the samples of age classes 3 and 4 from Boracéia- Casa Grande (N = 39) and Cotia-Piedade (N = 59) are not significantly different (P > 0.05). Therefore, individuals assigned to both age classes were pooled for subsequent analyses of variation for all of the samples; individuals of these classes are, presumably, mature adults. Sexual dimorphism was investigated in adult individuals in the samples from Boracéia-Casa Grande (N = 16), Cotia-Piedade (N = 18), Upper Rio Paranapanema (N = 22), Upper Rio Iguaçú (N = 11), Rio Itajaí-Açú Basin (N = 16), Rio Tramandaí Basin (N = 8), Rio Taquari-Antas Basin (N = 9), and Río Tebicuary Basin (N = 27). Six variables of seven samples showed significant differences: CLM1-3, HBC, and BZP for Rio Itajaí-Açú Basin; BR for Rio Taquari-Antas Basin and Río Tebicuary Basin; LPB for Upper Rio Paranapanema; and LOF for Rio Tramandaí Basin. These results show a lack of a consistent pattern of sex variation across the species geographical range. The absence of sexual dimorphism was verified in several other oryzomyine species, including Transandinomys talamancae (Musser & Williams, 1985), Microryzomys minutus and Microryzomys altissimus (Carleton & Musser, 1989), species of Cerradomys (Percequillo et al., 2008), Aegialomys xanthaeolus (Prado & Percequillo, 2011), and other oryzomyine genera (see Musser et al., 1998; Abreu-Junior et al., 2012). Therefore, as our analysis did not reveal a pattern of sexual dimorphism, geographical analyses were conducted with adult individuals from both sexes pooled together. MORPHOGEOGRAPHICAL VARIATION Qualitative characteristics The few traits that showed consistent variation within and amongst samples in Sooretamys are described and summarized below, and represented graphically on Figure 4:

10 G G TRB RGR URI IAB TEB MIS WSC LRI TEB PTG WSC MIS CPI RGR PTG CPI BCG URP LRI RGR URP BCG Figure 4. Morphological variation in Sooretamys throughout South America, expressed in frequency histograms, for 14 of the 15 pooled samples studied here (except Southern Minas Gerais) divided in two transects: left, north south transect; right, east west. Each sample is represented by three letters: BoracéiaCasa Grande (BCG); Riacho Grande (RGR); Cotia-Piedade (CPI); Upper Rio Paranapanema (URP); Ponta Grossa (PTG); Upper Rio Iguaçú (URI); Rio Itajaí-Açú Basin (IAB); Rio Tramandaí-Basin (TRB); Rio Taquari-Antas Basin (TAB); north-west of Rio Grande do Sul (NRS); Lower Rio Iguaçú (LRI); W Santa Catarina (WSC); Misiones; Argentina (MIS); and Río Tebicuary Basin (TEB). The bars represent the frequency of each characteristic (0 to 100%). A: colour of upper parts (1, yellow; 2, reddish; 3, dark; 4, greyish). B: colour pattern of under parts (1, grizzled; 2, pure). C: colour of hind feet (1, brown; 2, white). D: length of ungual tufts (1, do not reach the top of the claws; 2, reach the top of the claws). E: countershading of tail (1, unicoloured; 2, slightly bicoloured). F: incisive foramina (1, anterior to the alveolus of molar 1 (M1); 2, in the same plane of the alveolus of M1; 3, exceeds the plane of the alveolus of M1). G: posterolateral palatal pits (1, unique; 2 multiple). The empty spaces in the figure indicate that there is no information for certain qualitative characteristic in a given sample. NRS TAB URP F F CPI BCG E E BCG D RGR D CPI C URP C URI B IAB B TRB A TAB A NRS INTEGRATIVE TAXONOMY OF GENUS SOORETAMYS 851

11 852 E. A. CHIQUITO ET AL. Colour of upper parts (N = 111): Four patterns of dorsal colours were recognized: yellowish-brown, reddishbrown, dark-brown, and greyish-brown. In most of the samples, the predominant colour is yellowish-brown, but samples from eastern Santa Catarina (Rio Itajaí and Upper Rio Iguaçu) have reddish-brown upper parts, whereas samples from São Paulo state (Boracéia, Riacho Grande, and Cotia-Piedade) are predominantly greyish-brown. Colour of under-parts (N = 112): This characteristic is associated with the presence of hairs with a grey basal portion and a white or buffy apical portion, or hairs entirely white or buffy, resulting in grizzled and pure white or pure buffy appearance, respectively. The predominant pattern of the colour of the under-parts is grizzled, whereas a pure colour was observed only in Río Tebicuary sample. Colour of hind feet (N = 76): Two states were observed: hind feet predominantly white or brown. Hind feet that are predominantly brown occurred in Boracéia and Río Tebicuary only; in all other samples, the hind feet colour of most of the specimens is white. Ungual tufts (N = 67): Ungual tufts are dense and variable in length on digits II to V, usually concealing the claws, but very short or even absent on digit I. Two states were recognized: short tufts not reaching the tip of the claws and long tufts reaching or surpassing the tip of the claws. The ungual tufts are predominantly long on populations to the south and north of the genus distribution. Colour of tail (N = 98): Two states were observed: not countershaded dorsoventrally, resulting in a unicoloured tail, or countershaded dorsoventrally, with the ventral surface slightly paler than dorsal surface, resulting in a weakly bicoloured tail. We recorded the countershaded tail as a predominant state only in specimens from Misiones. Length of incisive foramen (N = 62): Three states were observed: posterior margins of incisive foramen that do not reach the plane of first molars, posterior margins that reach but do not surpass the plane of molars, and posterior margins that surpass the plane of first upper molars and reach the anterocone. In most of the specimens of the upper Rio Paranapanema in São Paulo and Tramandaí, in the north of Rio Grande do Sul, the posterior margin of the incisive foramen lies anteriorly to the M1 alveolus. In all other samples, the posterior margin of the foramen is aligned to the anterior margin of the M1 alveolus. The long incisive foramen surpassing the molar planes occurs only in the northern portion of the distribution, in samples from São Paulo state. Posterolateral palatal pits (N = 115): These perforations on the posterior portion on the palatine are unique or multiple. When single, the pits are usually positioned on the palate level; when multiple, pits are recessed in deep and well-delimited palatine fossa. In all samples, multiple posterolateral palatal pits recessed in deep palatine fossa are the most common state, indeed larger specimens from São Paulo state and Paraguay present even deeper and wider fossa. In general, some trends of morphological variation can be recognized throughout the geographical samples studied. Regarding the colour of upper parts, there is a predominance of greyish-brown specimens amongst the northern samples and a predominance of yellowishbrown samples towards the south. Long ungual tufts are present in high frequency in the northern samples, whereas short tufts are usually found in the southern localities. White hind feet are the most common characteristic throughout the entire geographical distribution, but predominantly brown hind feet occur in high frequency in some groups from Paraguay and eastern Brazil. An incisive foramen with a posterior margin that does not penetrate between molar series is more frequently observed in samples from São Paulo, southern Paraná, and northern Santa Catarina States, in the central portion of the geographical distribution. These patterns of characteristic variation indicate that Sooretamys lacks sharp morphological discontinuities across its geographical distribution. Quantitative characters ONL, LD, CIL, LIF, LN, LPB, CZL, and LOF (Fig. 5): A strong clinal variation is observed in all these variables (although not shown in Fig. 5, all variables share the same clinal pattern) related to the anteroposterior axis of the skull, with mean values decreasing from the north (south of Minas Gerais) to the south (Rio Grande do Sul State) and south-west (Argentina and Paraguay), with the smallest values observed in Santa Catarina. A comparison between the northern and southern samples, southern Minas Gerais and Río Tebicuary Basin, respectively, showed that there are no significant differences (P = 0.05) in these variables, despite the clinal variation [ONL (t = 1.204, P = 0.243), LD (t = 0.663, P = 0.515), CIL (t = 1.386, P = 0.180), LIF (t = 1.288, P = 0.212), LN (t = 0.486, P = 0.632), LPB (t = 0.136, P = 0.893), CZL (t = 1.656, P = 0.113), and LOF (t = 1.352, P = 0.191)]. The only sharp discontinuity in this general trend is located at the western part of the geographical range, between the samples of Misiones and Río Tebicuary Basin, with the latter being noticeably larger. In fact, specimens from the south-western Río Tebicuary Basin are as large as those from the north-eastern region; for instance, specimens of Boracéia-Casa Grande and Río Tebicuary Basin are not significantly different (P = 0.05) in most cranial

12 INTEGRATIVE TAXONOMY OF GENUS SOORETAMYS 853 S Minas Gerais (SMG) S Minas Gerais (SMG) Boracéia-Casa Grande (BCG) Riacho Grande (RGR) Cotia-Piedade (CPI) Upper Rio Paranapanema (URP) Misiones (MIS) Ponta Grossa (PTG) Boracéia-Casa Grande (BCG) Riacho Grande (RGR) Cotia-Piedade (CPI) Upper Rio Paranapanema (URP) Upper Rio Iguaçú (URI) Lower Rio Iguaçú (LRI) NW Rio Grande do Sul (NRS) Rio Itajaí-Açú basin (IAB) Rio Tebicuary basin (TEB) W Santa Catarina (WSC) Rio Taquari-Antas basin (TAB) RioTramandaí basin (TRB) 95% CI CIL 95% CI BR 95% CI CIL 95% CI BR n= NRS TAB TRB IAB URI URP CPI RGR BCG SMG n= NRS TABTRB IAB URI URP CPI RGR BCG SMG n= TEB MIS WSC LRI PTG URP CPI RGR BCG SMG n= TEB MIS WSC LRI PTG URP CPI RGR BCG SMG 95% CI CLM1-3 95% CI BM1 95% CI CLM1-3 95% CI BM1 n= NRS TAB TRB 9 IAB URI URP CPI RGR BCG SMG n= NRS TAB TRB IAB URI URP CPI RGR BCG SMG n= TEB MIS WSC LRI PTG URP CPI RGR BCG SMG n= TEB MIS WSC LRI PTG URP CPI RGR BCG Figure 5. Graphs, including error bars [mean ± 95% confidence interval (CI)], of the 15 pooled samples; the error bars represent four patterns of geographical variation observed in univariate geographical analysis. The figure on the top represents the north south transect; the figure on the bottom, the east west transect, as described in detail in the text. Abbreviations: BM1, breadth of molar 1 (greatest crown breadth of the first maxillary molar across the paracone protocone); BR, breadth of rostrum (greatest dimension measured across the external border of the nasolacrimal capsules); CIL, condyloincisive length (measured from the greater curvature of one upper incisor to the articular surface of the occipital condyle on the same side); CLM1 3, length of molars (crown length from molar 1 to molar 3). SMG measurements [ONL (t = 0.005, P = 0.996), LD (t = 0.192, P = 0.849), CIL (t = 0.949, P = 0.348), LIF (t = 1.375, P = 0.176), LN (t = 0.069, P = 0.946), LPB (t = 1.775, P = 0.083), CZL (t = 1.188, P = 0.241), and LOF (t = 0.091, P = 0.928)]. BIF, BR, HBC, ZB, and BZP (Fig. 5): These measurements also exhibit clinal variation with an abrupt discontinuity between the Argentinean and Paraguayan samples, at the western range of the species. However, for these variables, the specimens from Río Tebicuary Basin are significantly larger (BIF: t = 3881, P = 0.00; BR: t = 2639, P = 0.01; and BZP: t = 3625, P = 0.00) than the north-eastern specimens from Boracéia- Casa Grande. For the variable ZB, the P-value (P = 0.063) is not significant; for HBC, there is no significant difference between the two samples (P = 0.167). In general, these two geographically separate samples are very similar in length, but Paraguayan specimens exhibit wider and more robust rostra, with wider incisive foramina (BR and BIF) and more robust zygomasseteric apparatus (ZB and BZP).

13 854 E. A. CHIQUITO ET AL. Table 3. Descriptive statistics of the external dimensions for the 15 pooled samples of the genus Sooretamys. First line, mean ± standard error; second line, sample size (minimum and maximum) Variable Sample* Length of head and body (in mm) Length of tail (in mm) Length of hind foot, including claw (in mm) Length of ear (in mm) Weight (in grams) South of Minas Gerais ± ± ± ± ± ( ) ( ) 6 (35 40) 6 (21 26) 6 (80 135) Boracéia-Casa Grande ± ± ± ± ( ) 13 ( ) 13 (33 38) 13 (13 25) Cotia-Piedade ± ± ± ± ( ) 8 ( ) 8 (35 41) 9 (18 34) 1 Upper Rio Paranapanema ± ± ± ± ± ( ) 13 ( ) 13 (25 42) 13 (18 26) 10 (79 220) Upper Rio Iguaçú ± ± ± ± ± ( ) 7 ( ) 7 ( ) 7 ( ) 7 (84 158) Rio Itajaí-Açú Basin ± ± ± ± ± ( ) 16 ( ) 16 ( ) 16 ( ) 12 (74 164) Rio Tramandaí Basin ± ± ± ± ± ( ) 6 ( ) 6 ( ) 6 ( ) 5 (31 172) RioTaquari-Antas Basin ± ± ± ± ( ) 4 ( ) 8 (35 39) 8 (13 23) 1 North-west Rio Grande do Sul ± ± ± ± ± ( ) 3 ( ) 4 (34 37) 4 (21 24) 3 (80 122) Ponta Grossa ± ± ± ± ± ( ) 3 ( ) 3 (32 38) 2 (21 23) 2 (90 110) West Santa Catarina ± ± ± ± ± ( ) 14 ( ) 15 ( ) 15 ( ) 15 (56 138) Misiones ± ± ± ± ( ) 8 ( ) 8 (32 38) 7 (18 25) 1 Río Tebicuary Basin ± ± ± ± ± ( ) 26 ( ) 26 (35 43) 25 ( ) 14 (91 157) *There are no available external measurements for specimens of the samples from Riacho Grande and Lower Rio Iguaçu. CLM1 3 (Fig. 5): The length of the upper molar series has a mosaic variation, with the greatest mean values observed in the Upper Rio Iguaçú and Lower Rio Iguaçú samples and the smallest in Cotia-Piedade and northwest Rio Grande do Sul. Thus, there is no clear pattern of variation across the geographical range, although the sample from Río Tebicuary Basin also exhibits a clear discontinuity from the samples from Misiones. BM1 (Fig. 5): The breadth of M1 also shows a mosaic variation, with larger mean sizes observed in Riacho Grande and smaller in north-west Rio Grande do Sul and west Santa Catarina, although the samples from Upper Rio Paranapanema also have narrower molars. For this characteristic, the samples from Río Tebicuary Basin have one of the lowest averages (contrasting with previous traits), but it is still larger than the closer geographical samples of Misiones and west Santa Catarina. Considering the external dimensions (Table 3), the specimens from Río Tebicuary Basin are larger in body and tail length, and have larger ears and hind feet. In contrast, smaller specimens were recorded in the samples from the central-western region of southern Brazil (Taquari-Antas Basin, north-west Rio Grande do Sul, Ponta Grossa, west Santa Catarina, and Misiones). The DA performed with all 15 samples showed that 78.65% of the variation is distributed throughout the four first discriminant functions, and CZL is the variable that predominately explains the variation amongst groups, with strong discriminant function coefficients in these functions (Table 4). In the first discriminant function (DF1), which accounts for 36.71% of the variation, the most influential variables are the condyloincisive and the condylo-zygomatic length, which are associated with the anteroposterior axis of the skull and directly correlated to the rostral and neurocranium

14 INTEGRATIVE TAXONOMY OF GENUS SOORETAMYS 855 Table 4. Standardized discriminant function coefficients for 15 log-transformed cranial variables for the 15 pooled samples Standardized canonical discriminant function coefficients Variable First Second Third Fourth ONL CIL LD CLM BM LIF BIF BR LN LPB HBC ZB BZP CZL LOF Canonical correlation Function Wilk s lambda 0.029** 0.081** 0.169** ns Eigenvalue % Variance *Wilk s lambda significant at P < 0.05; **Wilk s lambda significant at P 0.001; ns Wilk s lambda not significant. BIF, breadth of incisive foramen (greatest dimension measured across the internal surface of both incisive foramen); BM1, breadth of molar 1 (greatest crown breadth of the first maxillary molar across the paracone protocone); BR, breadth of rostrum (greatest dimension measured across the external border of the nasolacrimal capsules); BZP, breadth of zygomatic plate (across central area of zygomatic plate); CIL, condylo-incisive length (measured from the greater curvature of one upper incisor to the articular surface of the occipital condyle on the same side); CLM1 3, length of molars (crown length from molar 1 to molar 3); CZL, condylozygomatic length; HBC, height of braincase; LD, length of diastema (from the crown of the first upper molar to the lesser curvature of the upper incisor on the same side); LIF, length of incisive foramen (greatest anterior posterior dimension of one incisive foramen); LN, length of nasals (greatest anterior posterior dimension of one nasal bone); LOF, length of orbital fossa (greatest length of the orbital fossa between the squamosal and maxillary roots of the zygomatic arch); LPB, length of palatal bridge (measured from the posterior border of the incisive foramen to the anterior border of the mesopterygoid fossa); ONL, occipitonasal length; ZB, zygomatic breadth (greatest dimension across the squamosal root of zygomatic arches). regions. The length of the diastema is also highly associated with DF1 and correlated to the rostral region. Moreover, as coefficients exhibit different directions across all variables, we assume that in addition to size variation, there is also shape variation amongst the samples for the most influential variables discussed above. In the second function (DF2), responsible for 21.67% of the variation, the variables ZB and CZL explain most of the variation and are associated with the zygomasseteric apparatus. The dispersion of the individual scores amongst the three discriminant functions (Fig. 6) does not allow the recognition of distinct groups, except for samples from Boracéia-Casa Grande, Rio Taquari-Antas Basin, west Santa Catarina, and Río Tebicuary Basin, which are enclosed by an ellipsis that represents 67.5% of the distribution of points for these four samples. Figure 6 shows that the samples from Boracéia-Casa Grande and Río Tebicuary Basin are similar (DF1), but there is a difference in the cranial shape (DF2 mainly, but also DF1). The scatterplot between the first and the third discriminant function revealed no significant pattern of variation, with wide superimposition of all samples. Figure 7 shows that there is a clear decrease in the mean values of individual scores of the first discriminant function from north to south and from north to west. Samples from southern Minas Gerais and Boracéia-Casa Grande have average scores between 0.5 and 1. In contrast, samples from Riacho Grande, Cotia-Piedade, Upper Rio Paranapanema, Upper Rio Iguaçú, Rio Itajai-Açú Basin, Rio Tramandaí Basin, Rio Taquari-Antas Basin, north-west Rio Grande do Sul, Ponta Grossa, Lower Rio Iguaçú, and Misiones exhibit

15 95% CI DF1 95% CI DF1 856 E. A. CHIQUITO ET AL Western Santa Catarina Rio Tebicuary Basin 3 2 Western Santa Catarina Rio Tebicuary Basin DF Rio Taquari-Antas Basin Boracéia-Casa Grande DF1 DF Rio Taquari-Antas Basin Boracéia-Casa Grande DF1 Samples S Minas Gerais Boracéia-Casa Grande Riacho Grande Cotia-Piedade Upper Rio Paranapanema Upper Rio Iguaçú Rio Itajaí-Açú Basin Rio Tramandaí Basin Rio Taquari-Antas Basin NW Rio Grande do Sul Ponta Grossa Lower Rio Iguaçú W Santa Catarina Misiones Rio Tebicuary Basin Figure 6. Scatterplot of the individual discriminant scores of the three first discriminant functions (DFs), obtained through discriminant analysis, conducted with log-transformed data of 15 craniodental variables from 15 pooled samples. These three functions (DF1, DF2, and DF3) are responsible for 36.71, 21.67, and 13% of the variation, respectively. The ellipses cover 67.5% of the distribution of points for four samples. S Minas Gerais (SMG) S Minas Gerais (SMG) Cotia-Piedade (CPI) Upper Rio Paranapanema (URP) Upper Rio Iguaçú (URI) 2 Boracéia-Casa Grande (BCG) Riacho Grande (RGR) Misiones (MIS) Cotia-Piedade (CPI) Uper Rio Paranapanema (URP) Lower Rio Iguaçú (LRI) 2 Boracéia-Casa Grande (BCG) Riacho Grande (RGR) NW Rio Grande do Sul (NRS) Rio Taquari-Antas Basin (TAB) Rio Itajaí-Açú Basin (IAB) 0-2 Rio Tramandaí Basin (TRB) NRS TRB URI CPI BCG TAB IAB URP RGR SMG Rio Tebicuary Basin (TEB) W Santa Catarina (WSC) 0-2 TEB WSC URP RGR SMG MIS LRI CPI BCG Figure 7. Graphs, including error bars [mean ± 95% confidence interval (CI)], of the scores of the first discriminant function (DF1) for 14 of the 15 pooled samples studied here. The figure on the left represents the north south transect, and the figure on the right represents the east west transect. certain similarity, with mean scores between 0 and 1. The lowest average, approximately 2.5, is from the west Santa Catarina sample. However, moving west from west Santa Catarina, mean scores are higher again (c. 2) in the sample from Río Tebicuary Basin, with an average higher than the eastern samples. A Mantel test performed for the pairwise comparison between the geographical and Mahalanobis distance matrices resulted in a marginally significant P-value (r = 0.206, P = 0.053). This result suggests that there is a correlation, although nonsignificant, between size variation and geographical distances. For craniometrical traits, some noticeable discontinuities occur between southern (Rio Tramandaí Basin, Rio Taquari-Antas Basin, and north-west Rio Grande do Sul) and western samples (west Santa Catarina, Misiones and Río Tebicuary Basin). Additionally, there is strong clinal variation, with a decrease in the overall size of the skull amongst samples in the north south and east west directions. In the westernmost samples (west Santa Catarina, Misiones and Río Tebicuary Basin), an opposite clinal trend is observed, with an increase in the overall size of the skull from east to west. Multivariate analyses provided similar results to those obtained in the univariate analysis, pointing to a reduction in the skull size southwards, with a major and consistent discontinuity between most samples and the sample from Paraguay.

16 INTEGRATIVE TAXONOMY OF GENUS SOORETAMYS 857 Minas Gerais and Rio de Janeiro states Espírito Santo state São Paulo state Santa Catarina state W Santa Catarina state Rio Grande do Sul state Paraguay Argentina Figure 8. Unrooted network of 34 haplotypes of 675 bp of the mtdna cytochrome b gene obtained from 53 individuals of Sooretamys. Dots between adjacent haplotypes represent missing haplotypes. The symbol fill type indicates the geographical region in which the haplotypes were sampled; the symbol size indicates the frequency of this haplotype at the collection location. PHYLOGEOGRAPHICAL VARIATION The 53 sequences of Sooretamys have 59 variable sites that define 34 haplotypes (Fig. 8). Eleven haplotypes were retrieved from more than one specimen; from these, six haplotypes were collected at more than one locality. Observed p-distances amongst all pairwise sequence comparisons range from 0 to 3.0% (average: 1.6%). For the 12 localities with more than one specimen sequenced, the observed variation average is 0.5% (range: 0 to 1.2%); whereas the observed divergence between locality pairs averages 1.4% (range: 0 to 2.8%). Genetic distances within localities (p-distance) and between locality pairs (p-distance and K2p) are presented as supporting information (Table S1). Results of the AMOVAs are presented in Table 5. Of the four locality grouping schemes used, the one that maximizes the differences amongst groups (44.8%) and minimizes the differences amongst populations within groups (31.4%) is a scheme with two locality groups: east (Paraguay) and west (Argentina and Brazil) of the Parana River. The recovered genealogies, obtained by maximum parsimony and Bayesian inference (Fig. 9), revealed that all Sooretamys sequences showed several polytomies, with most clades lacking significant support in both analyses. The genus is strongly supported (BS = 100; PP = 1) and shows at its base a polytomy involving four lineages in the Bayesian analysis and 28 (21 formed by a single sequence each) in the parsimony analysis. Of these lineages, the only one that is strongly supported (BS = 91; PP = 0.95) is exclusively constituted by haplotypes recovered from the Paraguayan specimens. However, not all Paraguayan variants are part of this clade; two others are located in a marginally supported clade (PP = 0.65; not found in the MP analysis), with one variant found at nearby Misiones Argentina, and from two geographically distant localities in the Brazilian states of Espírito Santo and Minas Gerais. As such, this is a widely distributed phylogroup. The other two main lineages have low support and are distributed in the central portion of the geographical distribution; both overlap at several localities. The haplotype network (Fig. 8) has only two loops and shows several missing haplotypes. As expected given the MP and BA trees, no clear geographical groups (clans sensu Wilkinson et al., 2007) can be delimited in the network. However, it is of interest to note that haplotypes from the easternmost (i.e. Argentina, Paraguay, west Santa Catarina in Brazil) and northernmost localities (i.e. northern Brazil) occupy external positions in the network; however, the centre of the network is formed by variants collected in the centre and south-eastern parts of the distributional range of Sooretamys (e.g. coastal Santa Catarina, Rio Grande do Sul, São Paulo). DISCUSSION GEOGRAPHICAL DIFFERENTIATION: MORPHOLOGICAL AND MOLECULAR EVIDENCE Nearly a century ago, Thomas (1924) detected differences in this genus throughout its geographical distribution and used the names Oryzomys ratticeps ratticeps, O. r. paraganus (= S. angouya), and O. r. tropicius to account for this variation. According to Thomas, the typical subspecies O. r. ratticeps

17 858 E. A. CHIQUITO ET AL. Table 5. AMOVA results for four arrangements of locality samples (see Fig. 3 and Appendices 1 and 2). Percentage of variation amongst groups (AG), amongst populations within groups (APWG), within populations (WP) F-statistics sensu Wright (1950) Percentage of variation F-statistics AG APWG WP FSC FST FCT Number of groups Group: description Localities in each group Grouping criteria * 0.72* 0.37* G1: 187, G2: 30, 74, 76, 81, 89, 90, 92, 93, 97 99, 117, 120, 121, 124, 126, G1: Paraguay G2: Argentina, Rio Grande do Sul, Santa Catarina G3: rest f the distribution Thomas (1924) classification scheme * 0.73* 0.38* G3: all the others G1: 187, G2: 30, 126, 127 G 3: all the others G1: 187, G2: all the others G1: 30, 126, 127, 187, G2: all the others 3 G1: Paraguay G2: Argentina, west Santa Catarina G3: rest of the distribution Geographical groups * 0.76* 0.45* Paraná River 2 G1: Paraguay G2: Argentina, Brazil * 0.72* 0.30* 2 G1: Paraguay, Argentina, west Santa Catarina Geographical groups 2 G2: rest of the distribution *P-values < inhabits Rio Grande do Sul and Santa Catarina (Brazil) and Misiones (Argentina) and is characterized by a small size and a general greyish-brown colour. Oryzomys r. paraganus is restricted to Paraguay and is defined by a large size and a general buffy-brown colour that is richer and brighter, with lighter buffy sides and a buffy-whitish under surface. Oryzomys r. tropicius is found in the Brazilian states of Paraná and São Paulo and is recognized by a size similar to that of O. r. ratticeps and general buffy-brown colour with dark buffy sides and a buffy under-surface (Thomas, 1924). More recently, Musser et al. (1998), overlooking this variation, synonymized these taxa, plus Mus leucogaster Brandt, 1835 and Calomys rex Winge, 1887, under the senior available name Mus angouya Fischer, Our qualitative results point to a direction different from that of Thomas (1924) that is similar, but not identical, to that of Musser et al. (1998). Individuals with a greyish-brown upper surface were more frequently found in the northern part of the distribution (São Paulo), a region corresponding to O. r. tropicius; however, Thomas (1924) suggested that specimens from the southern part of the range (O. r. ratticeps) are greyish brown. Specimens from Paraguay, corresponding to O. r. paraganus, have two exclusive characteristics according to Thomas (1924): nongrizzled under-parts, a trait that we observed in only 28.6% of specimens, and a bright dorsum, which was not more frequent than in specimens from other localities in our work (see above). We are confident in the adequacy of the methodological approach used in this study (see also Moreira & Oliveira, 2011) for evaluating intra- and interpopulation variation and, therefore, species boundaries. Our results suggest the existence of only one living species of Sooretamys. The geographical variation approach allowed us to recognize subtle and gradual differences amongst populations that would be not evident when a taxonomic study is based on taxa delimited a priori, based on previous knowledge published on the group (e.g. Percequillo et al., 2008). Aiming to enhance the importance of a geographical approach, we tested the subspecific taxa proposed by Thomas (1924) to contrast with our results. Thomas (1924), after recognizing the subspecies, established geographical boundaries for them, mentioning the state, province, or country as the geographical limit: all taxa are parapatric and replace one another from north to south and the south-west. Therefore, the specimens included in this analysis are assembled from the political unities mentioned in the original description of the subspecies: specimens from Rio Grande do Sul and Santa Catarina States in Brazil and Misiones Province in Argentina correspond to O. r. ratticeps; specimens from the Brazilian states of São Paulo and Paraná correspond to O. r. tropicius; and specimens from

18 INTEGRATIVE TAXONOMY OF GENUS SOORETAMYS 859 outgroup GD257 PY 194 GD265 PY 194 UMMZ PY 194 UMMZ PY 194 GD52 PY 194 UMMZ PY 193 GD543 PY 193 GD274 PY 187 UMMZ PY 193 UMMA PY 192 TK61763 PY 191 FURB9230 SC 127 FURB9238 SC SP 147 FURB5900 SC 128 FURB9836 SC 117 MHNCI4781 SP 145 FURB9790 SC 117 FURB477 SC 129 FURB9252 WSC 127 FURB9749 SC 121 MCNU1229 RS 92 MN37786 RS 90 MN37789 RS 90 UFPB338 ES 41 UFPB335 ES 41 CRB1271 RJ 69 MN50234 RJ 69 MCNU1625 RS 74 MN37778 RS 98 AFV21 RS 76 AFV22 RS 76 EM1207 SP 152 FURB12041 SC 124 FURB12151 SC 120 FURB5070 WSC 126 FURB9696 SC 117 FURB9867 SC 117 GD273 PY 187 UMMZ15083 PY 187 CNP1998 AR 30 CNP2524 AR 30 MCNU1230 RS 92 MCNU1291 RS 81 MCNU1292 RS 81 MCNU622 RS 89 MN37777 SC 113 MN37780 RS 98 MN37783 RS 98 MN37785 RS 97 MN37790 RS 99 outgroup FURB9696 SC 117 AFV22 RS 76 CNP2524 AR 30 MCNU622 RS 89 GD257 PY EM1207 SP 152 FURB9867 SC 117 MCNU1230 RS 92 MCNU1291 RS 81 MCNU1292 RS 82 MN37790 RS 99 MN37794 RS 93 MCNU1229 RS 92 MN37786 RS 90 MN37789 RS 90 1 CRB1271 RJ 69 MN50234 RJ 69 MCNU1625 RS 74 MN37778 RS 98 FURB12041 SC 124 FURB5070 SC 126 MN37777 SC 113 MN37780 RS 98 MN37785 RS 97 MN37783 RS 98 FURB9230 SC 127 FURB9238 SC SP 147 FURB5900 SC 128 FURB9836 SC 117 MHNCI4781 SP 145 FURB9790 SC 117 FURB477 SC 129 FURB9252 SC 127 FURB9749 SC 121 AFV21 RS 76 FURB12151 SC GD265 PY 194 UMMZ PY 194 UMMZ PY 194 GD52 PY 194 UMMZ PY 193 GD543 PY 193 GD274 PY 187 UMMZ PY 193 UMMZ PY 192 TK61763 PY 191 GD273 PY 187 UMMZ PY 187 CNP1998 AR 30 UFPB338 ES 41 UFPB335 ES 41 MN37794 RS 93 MP301 MG 44 MP301 MG 44 A B 0.3 Figure 9. Gene genealogies for 675 bp of the mtdna cytochrome b gene obtained from 53 individuals of Sooretamys recovered through maximum parsimony (A) and Bayesian (B) methods. The terminal branches are individuals, each represented by the institution number, followed by the Brazilian state or country acronym (AR, Argentina; PY, Paraguay; ES, Espírito Santo; RJ, Rio de Janeiro; MG, Minas Gerais, SP, São Paulo; SC, Santa Catarina; and RS, Rio Grande do Sul) and the locality number, according to Appendix 2 and Figure 3. In A, the numbers represent the support for each branch; bootstrap values under 50 are not shown. In B, the numbers represent the posterior probability support for each branch; support values under 0.50 are not shown. Outgroups used for both analyses were Cerradomys subflavus, Nectomys squamipes, and Aegialomys xanthaeolus.

19 860 E. A. CHIQUITO ET AL. Paraguay correspond to O. r. angouya/paraganus. We performed univariate comparisons and discriminant analysis amongst these three samples ; we found that specimens assigned the name O. r. ratticeps exhibit the smallest mean values, whereas the species identified as O. r. paraganus have the largest means, and the specimens from São Paulo and Paraná, named O. r. tropicius, are characterized by intermediate mean values. Wilks lambda indicates that the projection of individual scores along the first and second discriminant functions revealed little overlap between specimens assigned to O. r. ratticeps and O. r. paraganus (Table 6). However, individuals identified as O. r. tropicius are completely overlapped to those of O. r. ratticeps and only marginally superimposed to those of O. r. paraganus (Fig. 10). An error bar exhibiting the mean values of the scores of the first discriminant function (Fig. 11) showed three quite distinct groups, as suggested by many of the craniodental variables individually in the univariate analysis. These results partially corroborate those obtained by Thomas (1924): specimens from Paraguay, called O. r. angouya/paraganus, are consistently and significantly larger than the other two taxa; however, between O. r. ratticeps and O. r. tropicius, size variation is subtler, as it does not exist for some variables. These results are supported by analyses of molecular variance: the differences amongst the population groups are higher (44.8%) when localities are grouped into two groups, Paraguay vs. the rest (i.e. O. r. angouya/ paraganus vs. O. r. ratticeps and O. r. tropicius) than when localities are grouped into three groups corresponding to the subspecific arrangement of Thomas (37.4%). These results suggest that an a priori recognition of putative different biological units (either species or subspecies) could be potentially misleading and that careful and detailed analysis of variation, both genetic and morphological, and employing geographical samples is essential for appropriate species recognition. Wright (1943) and Gould & Johnston (1972) postulated that geographically distant samples would accumulate more differences (through selection or genetic drift) than closely located samples. Furthermore, if a taxon represents a morphologically continuous unit with gradual differences accumulated throughout its geographical distribution, it is expected that a positive relationship will exist between morphological and geographical distances (Moreira & Oliveira, 2011). In the case of Sooretamys, the low but still nonsignificant P-value in the Mantel test (P = 0.053) rejects the hypothesis of the isolation-by-distance model and indicates that Sooretamys represents a continuous unit. This trend is apparently broken by the similarity in size between the two most distant samples from Boraceia- Casa Grande and Río Tebicuary Basin, approximately 800 km apart, and by the high discrepancy between two very close samples, west Santa Catarina and Río Table 6. Standardized discriminant function coefficients of the log-transformed data of 15 craniodental variables, using as grouping variables the three subspecies proposed by Thomas (1924): Oryzomys ratticeps tropicius, Oryzomys ratticeps ratticeps, and Oryzomys ratticeps angouya/paraganus Variable Standardized canonical discriminant function coefficients First Second Tests of equality of group means Wilk s lambda ONL ** CIL ** LD ** CLM ns BM ns LIF ** BIF ** BR ** LN ** LPB ns HBC ** ZB ** BZP ** CZL * LOF ** Canonical correlation Function Wilk s lambda 0.40** 0.86 ns Eigenvalue % Variance *Wilk s lambda significant at P < 0.05; **Wilk s lambda significant at P 0.001; ns Wilk s lambda not significant. BIF, breadth of incisive foramen (greatest dimension measured across the internal surface of both incisive foramen); BM1, breadth of molar 1 (greatest crown breadth of the first maxillary molar across the paracone protocone); BR, breadth of rostrum (greatest dimension measured across the external border of the nasolacrimal capsules); BZP, breadth of zygomatic plate (across central area of zygomatic plate); CIL, condylo-incisive length (measured from the greater curvature of one upper incisor to the articular surface of the occipital condyle on the same side); CLM1 3, length of molars (crown length from molar 1 to molar 3); CZL, condylozygomatic length; HBC, height of braincase; LD, length of diastema (from the crown of the first upper molar to the lesser curvature of the upper incisor on the same side); LIF, length of incisive foramen (greatest anterior posterior dimension of one incisive foramen); LN, length of nasals (greatest anterior posterior dimension of one nasal bone); LOF, length of orbital fossa (greatest length of the orbital fossa between the squamosal and maxillary roots of the zygomatic arch); LPB, length of palatal bridge (measured from the posterior border of the incisive foramen to the anterior border of the mesopterygoid fossa); ONL, occipitonasal length; ZB, zygomatic breadth (greatest dimension across the squamosal root of zygomatic arches).

20 DF2 INTEGRATIVE TAXONOMY OF GENUS SOORETAMYS ratticeps tropicius paraganus DF1 Figure 10. Scatterplots of the individual scores from the first two discriminant functions (DF1 and DF2), obtained through discriminant analysis, conducted with logtransformed data of 15 craniodental variables, using as grouping variables the three subspecies proposed by Thomas (1924): Oryzomys ratticeps tropicius (squares), Oryzomys ratticeps ratticeps (triangles), and Oryzomys ratticeps angouya/ paraganus (circles). These two functions (DF1 and DF2) are responsible for 88 and 12% of the variation, respectively. The ellipses cover 67.5% of the distribution of points. Figure 11. Graph, including error bars [mean ± 95% confidence interval (CI)], of the scores from the first discriminant function (DF1) conducted amongst the three subspecies proposed by Thomas (1924), plotted against the alleged geographical distribution of these three taxa. Tebicuary Basin, approximately 400 km apart, as indicated by the nonsignificant P-value of the Mantel test. Qualitatively, there is no consistent discontinuity throughout the geographical range for any of the traits evaluated. The morphological variation that is observed is not concordant with geographical distribution, as aspects of coat colour and cranial morphology are highly variable within samples and also within the subspecies postulated by Thomas (1924). Genetically, Paraguayan populations differentiate from samples from other locations. No haplotype is shared between Paraguayan and non-paraguayan localities. In addition, four of five haplotypes (recovered from 11 of 13 sequenced specimens) from Paraguay form a highly supported clade that constitutes one of the four main lineages of S. angouya recovered in the Bayesian analysis. Meanwhile, the remaining Paraguayan haplotype (found in two of 13 specimens) forms a clade together with haplotypes found in nearby Misiones, Argentina (locality 30), and the Brazilian states of Espirito Santo (41) and Minas Gerais (44). If haplotypes from Paraguay were recovered as a monophyletic group, even without reciprocal monophyly, one scenario would be to recognize Paraguayan populations as a subspecies of the more inclusive clade, as a geographically and temporarily isolated lineage or sublineage (Frost et al., 1992). However, Paraguayan haplotypes do not form a monophyletic group: two haplotypes from Centu Cue, Paraguay (locality 187 on map) are closely related to haplotypes from Itamonte (locality 44), Venda Nova (locality 41), and Refúgio Moconá (locality 30), rather than to other Paraguayan haplotypes. The biological reason behind this pattern is currently unclear: it may be indicative of current gene flow throughout the geographical range of Sooretamys or it just may be the reflect of incomplete lineage sorting. The study of additional specimens and the sequencing of nuclear markers may clarify this issue. The molecular results are similar to the morphological and morphometric results: our integrative approach suggests a certain degree of variation in the genes and morphology of the Paraguayan population, but there is no consistent resolution to suggest that the specimens from Paraguay represent a distinct lineage that would merit taxonomic recognition. As such, we recognize only one unique diagnosable unit and monophyletic cluster of individuals as proposed by Cracraft (1983). Consequently, we conclude that Sooretamys is a monotypic genus (for taxonomic account, see Appendix 3). Considering that the genus Sooretamys exhibits only one evolutionary lineage at the species level, an important point is to designate the appropriate name for this species. Direct morphological and morphometric (principal components analysis, results not shown) comparisons of the nominal specimens of Mus angouya Fisher, 1814 (2.7 km north of San Antonio by road, Paraguay, neotype); Hesperomys ratticeps Hensel, 1872 (Rio Grande do Sul, lectotype); O. r. tropicius Thomas, 1924 (Piquete, São Paulo, lectotype); and O. r. paraganus Thomas, 1924 (Sapucay, Paraguay, holotype) re-