On the Occurrence Of HolocHilus cHacarius ( cricetidae : SigmOdOntinae ) in Brazil , with taxOnOmic nOteS On HolocHilus SpecieS

Currently four species of Holochilus (Cricetidae: Sigmodontinae) are recognized. According to the literature, three species are recorded for Brazil: H. sciureus, H. brasiliensis and H. chacarius. Samples from western Brazil (Mato Grosso and Mato Grosso do Sul states) are usually reported as H. sciureus and, when referring to specimens from the Pantanal, as H. chacarius. However, the taxonomic status of specimens from this region has not been properly evaluated through detailed morphological analyses or voucher specimens. About 110 specimens of Holochilus deposited in Brazilian collections were qualitatively and quantitatively analyzed (univariate and multivariate tests). Our results suggested the existence of three species (H. sciureus, H. brasiliensis and H. chacarius) in Brazil, which present significant morphological and morphometric differences, thus confirming that the Mato Grosso and Mato Grosso do Sul samples, formerly misidentified as H. sciureus, belong to H. chacarius. This species differs from H. sciureus and H. brasiliensis by a series of pelage and skull characters, such as: the coloration of pelage, light orange-brown in the dorsum, light orangish in the flanks and white in the venter; ridges of the masseteric crest confluent at the level (or slightly above) of the mental foramen and at the anterior part of m1, alternated main molar cusps, lophids compressed and with acute outer margins (rarely prismatic), anteromedian fossetid labially displaced and subcircular (small to medium), metaflexid less developed (frequently not reaching the midline of the tooth), proto and hipoconid subrectangular in outline and transversely orientated, mesoflexid transversely orientated and mesoloph absent. The updated geographic distribution of Holochilus in Brazil shows that H. sciureus is present in the northern portion of the country, being the only species present at the Amazon and at the Caatinga and with limit in the Cerrado biome at the center of Goiás state. Holochilus brasiliensis is restricted to eastern Brazil, with its northern record at the Atlantic Forest of south Bahia. Holochilus chacarius occurs in the Pantanal areas of the Mato Grosso and Mato Grosso do Sul states. A new record extends the distribution of H. chacarius to the Cerrado biome at eastern Mato Grosso, confirming its presence in this biome. Key-Words: Holochilus chacarius; Holochilus sciureus; Holochilus brasiliensis; Mato Grosso; Brazil. http://dx.doi.org/10.1590/0031-1049.2015.55.03 Volume 55(3):47‐67, 2015 1. Universidade Federal de São Carlos, Campus Sorocaba, Departamento de Biologia, Laboratório de Diversidade Animal. Rodovia João Leme dos Santos, km 110 (SP-264), Itinga, CEP 18052-780, Sorocaba, SP, Brasil. E-mail to correspondence: puerabio@gmail.com 2. Museu de Zoologia, Universidade de São Paulo, Mastozoologia. Avenida Nazaré, 481, CEP 04263-000, São Paulo, SP, Brasil. E-mail: fabnasc@gmail.com 3. Museu Paraense Emílio Goeldi, Mastozoologia. Avenida Perimetral, 1.901, CEP 66017-970, Belém, PA, Brasil.

Holochilus brasiliensis is the largest species of the genus and is found in southeastern Brazil, eastern Paraguay, Uruguay, and east/central Argentina (Musser & Carleton, 2005, Pardiñas et al., 2013).In Brazil, this species is mostly associated to the Atlantic Forest, with poorly documented limits from the states of Rio Grande do Sul to Bahia (Massoia, 1981;Marques, 1988).Furthermore, there are scarce records of H. brasiliensis along riverine forests in the Cerrado biome (Carmignotto, 2005).Regarding the smallest species, H. sciureus is widespread in the northern limits of the genus, inhabiting remarkably different biomes that ranges from humid environments in Amazonia, through central Brazil and up to the Caatingavegetation on northeastern limits of Brazil (Moojen, 1952;Patton et al., 2000;Carmignotto et al., 2012).Holochilus chacarius, in turn, is widespread in the southwestern limits of the genus, where is commonly associated to the Chaco of Argentina, Paraguay and south Bolivia, whereas H. lagigliai is known only from the type locality at Mendoza Province, Argentina (Pardiñas et al., 2013: fig. 2).As a result, field identification of Holochilus spp. is usually associated to their size and geographic distribution.
Except for H. lagigliai, all species of the genus were already mentioned to occur in Brazil (e.g., Bonvicino et al., 2008;Cáceres et al., 2008Cáceres et al., , 2011;;Oliveira & Bonvicino, 2006, 2011;Paglia et al., 2012), stating that H. chacarius occurs in the Pantanal region of Mato Grosso do Sul state.However, none of these reports listed any voucher specimens or explained the decision of reporting the presence of this taxon in Brazil.Furthermore, there is no reference in literature that justifies with details the occurrence of H. chacarius in Brazil.Therefore, the northernmost vouchered record previously known for H. chacarius was made for Bolivia by Anderson (1997) -reported as H. sciureus but latter some specimens were reidentified by Pardiñas & Galliari (1998) -and almost all reports of a small-sized species of Brazilian Holochilus are restricted to H. sciureus.
We provide here a short review on morphological diagnostics characters in order to distinguish the species of Holochilus from Brazil currently recognized, and to discuss the presence of H. chacarius in this country, reporting a distribution extension for this taxon.Finally we provide updated records of H. brasiliensis and H. sciureus, reevaluating both distributional limits in Brazil.

MATERIAL AND METHODS
We have studied skins and skulls of H. chacarius, H. sciureus and H. brasiliensis deposited in the mammal collections of Museu de Zoologia da Universidade de São Paulo, São Paulo (MZUSP), and Museu Nacional, Universidade Federal do Rio de Janeiro, Rio de Janeiro (MN) (Appendix I).
The external qualitative characters refers to the pelage color pattern of the head, body, feet and tail.For the external quantitative characters, we used body measurements taken from the specimen's tags as follows: (1) head and body length (HB); (2) tail length (T); (3) hind foot length (HF); and (4) ear length (E).When only total length (TL) was provided, we subtracted the recorded tail length from total length to obtain the values of the head and body length.
Three age classes were recognized based on maxillary tooth eruption and abrasion and fusion of the skull sutures: juveniles, subadults and adults.Specimen is only classified as an adult if covered by full adult pelage -not the juvenile coat or the transitional molt from juvenile to adult fur -with permanent dentition (third molar fully erupted) and completely ossified spheno-occipital suture.In our analysis we used only adults that present all the mentioned characters.

Statistical Analysis
We apply the Kolmogorov-Smirnov test for all variables of the studied species.When the variables were normally distributed, we performed Student's t-test to evaluate the existence of sexual dimorphism.
We performed one-way analysis of variance (ANOVA) and Tukey's pos hoc test to check whether are significant differences (p < 0.05) among the species on craniodental variables.
For the multivariate analysis, specimens with missing values were removed, reducing the sample from 65 to 48 specimens, and all craniodental variables were log 10 transformed.Principal component analysis (PCA) was extracted from the covariance matrix and it applied as an exploratory tool for investigating the patterns of variation among the species of Holochilus, as well as to evaluate the degree of separation among them.Only principal components with eigenvalues greater than 1 were extracted and after they were rotated using the varimax criterion.Discriminant function analysis (DFA) was performed to investigate whether the three species of Holochilus studied could be distinguished based on skull morphology and also to construct a predictive pattern of different group memberships.All statistical analyses were performed with SPSS 17.0 software.

The identity of the Mato Grosso specimens
In the course of a research on Holochilus specimens from Mato Grosso state, Brazil, which has a significant portion of Amazon rainforest in its north limit, we verified that many references found for small mammals assemblages from this biome and adjacent transitional areas (e.g., Bezerra et al., 2009;Patton et al., 2000;Rocha et al., 2011a, b) mentioned a small sized Holochilus species currently identified as H. sciureus, which Patton and collaborators (2000) distinctively described its ventral color as gray washed with orange or buff (Fig. 1B).Our hypothesis of the presence of a different species in Mato Grosso came from the fact that specimens of Holochilus from this state, although had a similar size, clearly present a self-white venter (Fig. 1A).
For that reason, we initially compared specimens from Amazonas, Rondônia and Pará state, both adjacent to Mato Grosso.All specimens from those areas present venter coloration as the one described by Patton and collaborators (2000), been remarkably different from the Mato Grosso specimens.Latter, we also find that all specimens from northeastern Brazil present a similar venter coloration to those from Brazilian Amazon (excluding Mato Grosso), just as the morphology described by Oliveira & Langguth (2004) for Paraíba e Pernambuco specimens.Specimens that present whitish venters that resembles those from Mato Grosso were from large-sized Atlantic Forest specimens, which clearly represent H. brasiliensis (Fig. 1C), the largest species.For this reason, this name could not be assigned to the Mato Grosso specimens.Additionally some of these specimens present self-white ventral fur only restricted to throat and inguinal areas (see details further in the text).The only specimens that matched both morphological and size features were from Mato Grosso do Sul, the Brazilian state adjacent to Mato Grosso at the south, and therefore also could not be assigned to H. brasiliensis and H. sciureus.
Studying records from species of Holochilus known from other countries, H. lagigliai and H. chacarius, we made progress on Mato Grosso specimens identification.The morphology of H. lagigliai is quite different from the Mato Grosso specimens (see details further in the text), refuting the hypothesis that they are different populations of the same species.However, H. chacarius has similar size and pelage characters similar to Mato Grosso specimens (Massoia, 1971(Massoia, , 1976(Massoia, , 1980)).Furthermore, H. lagigliai is known only from the central western Argentina, while H. chacarius was already confirmed to inhabit Bolivia (Anderson, 1997; see Pardiñas & Galliari, 1998), Argentina andParaguay (Massoia, 1971, 1976;Pardiñas & Teta, 2011;Pardiñas et al., 2013: fig.2), which are adjacent regions to Mato Grosso and Mato Grosso do Sul states.
In view of that, the evidences suggest that the specimens from Mato Grosso are in fact neither H. sciureus nor H. brasiliensis.However, the Mato Grosso specimens present external similarity with both the other congeneric species in Brazil: while ventral pelage pattern resembles those of H. brasiliensis, the size is consistently different, whereas the opposite is observed when compared to H. sciureus.Therefore, in order to consistently distinguish Mato Grosso specimens from the mentioned Brazilian species, we investigate in details the craniodental and external characters.

Quantitative analysis
The Kolmogorov-Smirnov test (applied to test the normality of the sample) showed a normal distribution of all variables (results not included herein).According to Student's t-test, there are no significant differences between the sexes (results not included herein).Therefore, both males and females were included together in the descriptive statistics and the following tests.
A total of 48 specimens (14 H. chacarius, 21 H. brasiliensis and 13 H. sciureus) with no missing data were included in the PCA.Table 3 shows the three first components resulting from the analysis, as well as their respective coefficients, eingevalues and variances.The first principal component was responsible for 60.59% of variance, and in combination with the second component, both were responsible for 69.90% of variance.All coefficients of the first component showed positive signal, indicating a positive correlation with each other, and it can be interpreted as a component usually associated to the overall size of the skull (Jolicoeur & Mosimann, 1960).The greatest coefficient is associated to the breadth of the palatal bridge (BPB), indicating it as the dominant craniodental measurement in the first component.Other variables, breadth across the squamosal zygo- matic processes (ZB), breadth of the zygomatic plate (BZP), length of the rostrum (LR), zygomatic length (ZL) and condyle-incisive length (CIL), in decreasing order, also contributed in a similar way to each other for the first component.
The variables that contributed to the second component were, in decreasing order, breadth of the incisive foramina (BIF), length of the incisive foramina (LIF), length of nasals (LN), breadth of the palatal bridge (BPB), length of the diastema (LD) and breadth across the squamosal zygomatic processes (ZB).The two first variables are related to size of incisive foramina, and the remaining variables (except ZB) are related to the anterior region of the skull.Moreover, regarding to the second component, there is a connection between form and size.
In the two first principal components plot (Fig. 2) is possible to observe that H. sciureus specimens are widely mixed to H. chacarius, mainly, and to H. brasiliensis, but these latter two species, in turn, show a certain overlapping between their respective specimens.Concerning to the first component, and  The discriminant function analysis created two canonical functions from the original craniodental variables used in the analysis.The first canonical variable is responsible by 87.51% of total variance and the second by 12.49% (Table 4 and Fig. 3).The canonical correlation also shows a high relation between each discriminant function and the pre-defined groups of the analysis, indicating that both functions are useful to discriminate the groups (Table 4 and Fig. 3).Discriminant analysis shows a clear separation of each Holochilus species (Fig. 3), and H. brasiliensis is well separated from the two other species based on canonical coefficients of the first function, which are mainly related to the following variables, in decreasing order: length of the incisive foramina (LIF), length of the diastema (LD), breadth across the squamosal zygomatic processes (ZB), length of nasals (LN), breadth of the palatal bridge (BPB) and breadth of the incisive foramina (BIF) (Table 4; Fig. 3).Based on the canonical coefficients of the second function (Table 4), there is an overlapping among the individuals of the three species, but H. sciureus shows a slight separation from H. brasiliensis and H. chacarius (Fig. 3).Condyleincisive length (CIL), length of the molar tooth row (MTR), breadth of the braincase (BB), length of the incisive foramina (LIF), length of nasals (LN), and breadth across the squamosal zygomatic processes (ZB) are the variables that contribute more, in decreasing order, to the second function.

Qualitative analysis
The morphometric evidence described above suggests the presence of a third species of Holochilus in Brazil.Indubitably, the specimens from western Brazil (Mato Grosso and Mato Grosso do Sul states) are related to H. chacarius as judged from their size, pelage color and geographical distribution pattern.Craniodental characteristics also proved to be informative in the identification of the current recognized Discriminant Function 2 species.Remarkable contributions have already been made by others authors as described next.
After Hershkovitz (1955), who lumped 13 nominal species of Holochilus under H. brasiliensis, the most important contributions were the reports made by Massoia (1971Massoia ( , 1976Massoia ( , 1980Massoia ( , 1981)).His reports were mostly related to comparisons between H. brasiliensis to one or another of the others currently recognized species (except for the recently described H. lagigliai), providing consistent diagnostic characters to distinguish them.Latter, following and expanding Massoia's work, significant contributions to diagnostic characters on Holochilus were made by Voss & Carleton (1993), Carleton & Olson (1999), and Machado et al. (2013), whom respectively compared this genus to Lundomys, Noronhomys, Carletonomys and Reigomys in order to recognize them as valid genus.Recently Pardiñas & Galliari (1998) and Pardiñas & Teta (2011) provided a substantial contribution to the knowledge on the fossil history of Holochilus, summing up all known characters for the species of this genus (except for the recently described H. lagigliai).All these mentioned works were the most significant references to the analysis of the specimens reported herein.
Currently it seems to be a general consensus that Holochilus spp.are easily distinguished based on cranial morphological traits into two groups: "H.brasiliensis group", characterized by skulls with longer dimensions (Table 1, Fig. 4), main molar cusps opposite or slightly alternated, proto/hypoconid with posterior faces oriented 45° inwards, presence or at least vestige of mesoloph on M1 and M2 (Fig. 5E) and ridges of the masseteric crest confluent at the level of the posterior face of m1 and continuing forward to the mental foramen (see Voss & Carleton, 1993: fig. 11;Voglino et al., 2004: fig.2); and "H.sciureus group", characterized by skulls with shorter dimensions (Table 1, Fig. 4), main molar cusps evidently alternated, proto/ hypoconid with posterior faces transversally oriented, mesoloph on M1 and M2 absent (Fig. 5A and C) and ridges of the masseteric crest confluent at the level or slightly anterior to the mental foramen (see Voss & Carleton, 1993: fig. 11;Voglino et al., 2004: fig.2).
The majorities of the differences mentioned above consist on diagnostics characters that regard molar morphology.Most of them were summarized by Pardiñas & Teta (2011) and, since we consistently agree with several of the diagnostics characters cited by them (see Pardiñas & Teta, 2011: table 3), those characters are summarized here (Table 6).However, few of them are discussed here to clarify some issues on the identification of specimens of this genus.
Most of the specimens observed by Pardiñas & Teta (2011) were from Argentina and a few specimens from Brazil (three from Pirapora, Minas Gerais, and one from Barreiras, Bahia).Studying a higher number of specimens from Brazil, we consistently agree with the characters cited by the mentioned authors, but we realized that morphological variation within specimens examined by us were significant and it would be hard to distinguish few specimens of H. sciureus from H. chacarius looking only for molar differences.As can be observed in Pardiñas & Teta (2011: table 3), few differences were mentioned to distinguish H. sciureus from H. chacarius, been those restricted to: form of lophids (FL), form and size of the anteromedian fossetid (AF), and occlusal development of metaflexid (DM).We analyzed specimens that presented the mentioned differences (Fig. 5); but we also seen few specimens that presented characters referent to the other species (see Table 7 and Fig. 6).
For instance, although H. sciureus and H. chacarius present FL quite different to H. brasiliensis, we hardly agree that there are differences between H. sciureus and H. chacarius that can be easily detected for some specimens, although exist differences between these species.Following Pardiñas & Teta (2011) descriptions, most of the specimens of sciureus group analyzed by us presented FL similar to those of H. chacarius but also considerable amount presented the FL most similar to H. sciureus.It is worth mentioning that the specimen found by us that presents FL most remarkably similar to H. sciureus, which is described as distinctly compressed and with strongly acute outer margins (almost prismatic), actually belongs to an Argentinean specimen (Fig. 6A .Finally, we also find that DM is quite variable among those species and connection between metaflexid with the protoflexid, a characteristic anteriorly associated to H. sciureus, could be observed in all species of the genus considered here (Fig. 6B and D), including specimens of H. brasiliensis.Also, some specimens of H. sciureus presented the anterior murid and, therefore, the connection between metaflexid with the protoflexid is absent.
All issues discussed above indicate that differences between the two species from sciureus group, made by previous authors, maybe are related to individual or to geographic variation found within these species, since distinctiveness between lower molar morphology can be sometimes quite difficult to observe.
It is worth mentioning that few specimens present good condition for the dental analyses due to the fact that most of them were much worn.Although Holochilus is a member of the tribe Oryzomyini, which usually present pentalophodont molars (generally associated to forested habitats), its species present simplification of molar oclusional surface through lost/ reduction of dental structures, been now classified as a tetralophodont (generally associated to open/transitional areas) (see details on Voss & Carleton, 1993;Weksler, 2006;Machado et al. 2013).Therefore, the high degree of tooth wear is quite common for Holochilus populations due to its species diets, which usually include abrasives and fibrous items that might cause early tooth wear.This was especially true for H. sciureus specimens from semi-arid Caatinga areas, where xeric vegetation is much abrasive.Therefore, it is very hard to identify specimens from sciureus group to specific level based only on the form of the anteromedian fossetid or lophids.For that reason, although we include these characters at Table 6, they should be used with caution and never alone because we did not observed a distinctive difference between H. sciureus and H. chacarius that could not be frequently attributed to morphological variation for each species or if the phenotype observed is just related to tooth wear.Additional specimens referring to younger individuals with less degree of tooth wear, would probably solve these issues in the future.
According to Voss & Carleton (1993) another useful character is the point of confluence between the upper and lower ridges of the masseteric crest at the mandible, which is different among Holochilus species.However, no comparisons between the three species studied here are yet available.We observed that while in H. chacarius this point is horizontally at the level (or slightly above) of the mental foramen and vertically aligned to approximately the anterior part of m1, in H. sciureus and H. brasiliensis this point is below the level of the mental foramen level and aligned to the middle and posterior part of m1 respectively (Fig. 7).
Regarding pelage morphology, we realized that comparisons have been little explored to distinguish Holochilus species.In the present study, pelage morphology proved to posses important characters and, for most cases, species distinctiveness could be done  Voss & Carleton (1993), Voglino et al. (2004) and Pardiñas & Teta (2011).
Brandão, M.V. & Nascimento, F.O.: On the presence of HolocHilus cHacarius in Brazil even in the field (Table 6).We observed that Holochilus species present orangish-brown dorsal pelage that differs in tone coloration between species and seems to be somewhat gradually darker as follows: H. chacarius, H. brasiliensis and H. sciureus (Fig. 1A, C and E).In H. chacarius, the dorsal pelage color tends to be more homogenous, while H. sciureus seems to have a slightly darker stripe along the midline of the dorsum and H. brasiliensis is somewhat streaked with darker hairs, except for the head area.Another useful difference is the presence of a lateral stripe, which shows hairs with gray basis and orangish-brow tip that also seems to present gradual intensity as follows: H. chacarius, H. brasiliensis and H. sciureus (Fig. 1B, D and F).In H. chacarius, this stripe is only slightly perceptible, while H. sciureus present a more notable orange stripe that is somewhat continuous with the ventral pelage and H. brasiliensis also presents a distinct lateral stripe but that gradually seems to became paler towards the venter.The most distinctive pelage character is the ventral coloration, which differs conspicuously among these species.While H. sciureus present grayish ventral pelage, H. chacarius and H. brasiliensis present whitish venters (Fig. 1B, D and F).Yet, in H. chacarius the orangish tones of ventral pelage are none existent and distinction between dorsum and venter is unambiguously perceptible, whereas in H. brasiliensis and H. sciureus the orange tones are constantly present and somewhat continuous toward the venter.However, in H. sciureus, this is notably much more distinct and the orange wash is continuous from lateral to venter, making transition between these areas almost unnoticeable.Regarding to H. brasiliensis, the orangish wash is much more discrete (light-orange), been frequently more evident at the shoulder region or faintly spread at the thorax and abdominal region (see Fig. 1D), toward a whitish venter.All mentioned pelage characters, as well as the cranial characters, are summarized in Table 6.Some variation in ventral pelage coloration has been noticed among each species.In H. sciureus the orange wash was sometimes absent, but seems to be related only to age since most of these specimens were young, except for a single adult specimen (MN 3359).It is worth mentioning that specimens of H. sciureus from Bahia presented general pelage much brighter than typical H. sciureus, specially at the ventral pelage 4202,8255,67401).However, smaller size and dental characters unambiguously classify them as H. sciureus.As related to H. chacarius, a few specimens from Mato Grosso present a light graybased with white tipped hair associated to region between members (MZUSP 15336, 21216-17, 35145).However, the observed variation seems to have no taxonomic meaning since these specimens do not differ in any other aspect from typical H. chacarius from the same localities (Fig. 6E and F).Finally, the specimen from Cocalinho present a much vivid orange tone at the lateral area, but venter coloration and skull characters evidently classify it as H. chacarius.
The most remarkable variation in ventral morphology is related to H. brasiliensis from southeast areas (MZUSP 2788, 3179-90, 6366, 6542, 11213-14, 11226, 21219), whom presented whitish venter (with gray-based white-orangish tip fur) but the self-white fur was restricted to throat and inguinal areas.We suggest that this could prove to have taxonomic meaning in future analyses -see comparisons between H. brasiliensis subspecies (Massoia, 1976) -as they differ from specimens from southern localities (MZUSP 1027, 3179-80, 10083, 10087) which all present self-white ventral fur or, if graybased, only ¼ of total hair length and only perceptible if pressed backwards.However, we only had few specimens available from areas that are crucial to resolve this issue and our major concern was to report H. chacarius for Brazil and compare it to species of Holochilus currently recognized by Musser & Carleton (2005).
It is worth mentioning that the reports of H. sciureus made by Bezerra et al. (2009) and by Rocha et al. (2011a) for Ilha do Bananal, Tocantins, should be reviewed for the reasons explained next.Both mentioned studies provided a detailed morphological description of the Holochilus specimens collected at Ilha do Bananal, mentioning a small body sized species that present whitish venter.Therefore, we hypothesized that these records could represent H. chacarius.We examined pictures of three specimens reported by Rocha et al. (2011a) and stated a high degree of morphological variation, which found no rematch with any other single locality analyzed by us or even in all the distribution range of sciureus species group.As we did not examine personally the skull of these specimens, we were not able to provide a more precise diagnostic; therefore, we could not attribute those specimens to any Holochilus species.However, these specimens clearly are from sciureus species group judging from body size (Rocha et al., 2011a: 14) and could represent a contact area between H. chacarius and H. sciureus, representing a sympatric locality or Finally, it is essential to mention that Holochilus lagigliai was recently described for the west-central Argentina (Pardiñas et al., 2013).This species was described based on a single specimen with skin and skull, and three additional right mandibles.In our present view, characters of the skull mentioned in the original description of this species are quite different from all specimens analyzed by us.Some characters of H. lagigliai draw attention to us, such as the broad braincase, short and broad rostrum, and dental morphology, which present mesoloph on upper molars (similar to H. brasiliensis), anteromedial fossetid laterally displaced (similar to H. sciureus group) and absence of hypoflexus and hypocone on third lower molar (exclusive of H. lagigliai).Yet, H. lagigliai locality is quite far away from the distribution range of H. chacarius and the other congeners.Therefore, although we consider a valid species, we did not compare H. lagigliai to the Holochilus specimens from Brazil because it seems very unlike that the former species might be present in this country and its morphological variation are yet poorly known.

Geographical distribution
The most recent contribution to the knowledge of the geographical distribution of this genus is provided by Prado & Percequillo (2013).Fortunately, these authors were very cautious while reporting records of Holochilus since no recent published revision of the genus exists.Their records were based only in museum databases and Hershkovitz (1955), without assigning specimens to any species in order to -at least to our view of Prado & Percequillo (2013) article -not cause further error of taxa assignments.However, despite Prado & Percequillo (2013) contribution, records of Holochilus in Brazil remain poorly discussed.Herein we provided a new map based on the analyses of the Brazilian Holochilus specimens studied by us (Fig. 8).
As previously noted, H. chacarius had already been mentioned for Brazil but no museum catalogued number has been provided.The only excep-tion is Cáceres et al. (2008) which reported H. chacarius and H. sciureus for Mato Grosso do Sul state.However, the vouchers are only referent to H. sciureus while H. chacarius is only mentioned as a reference to Oliveira & Bonvicino (2006).These last authors report H. chacarius to Mato Grosso do Sul state but no detailed information on specimens is presented.Also, we analyzed specimens MN 1987, MZUSP 3780 and MZUSP 27430 reported as H. sciureus by Cáceres et al. (2008) and re-identified them as H. chacarius; therefore, the presence of H. sciureus was mistakenly attributed to Mato Grosso do Sul.It is worth to mention that Allen (1916) had already mentioned two species of Holochilus collected in the Roosevelt Expedition in Brazil: H. amazonicus (= H. sciureus) cited for Rio Solimões, Amazonas, and H. vulpinus from Urucúm, Mato Grosso do Sul, which probably refers to H. chacarius judgment by external measurements (see Allen, 1916: 571) and locality (MN 1987 and1988 are H. chacarius -Present study), even though vulpinus is currently junior synonym of H. brasiliensis (Musser & Carleton, 2005).Recent list of mammals species from Brazil that include rodents, report the presence of H. chacarius exclusively for Mato do Grosso Sul state (e.g., Bonvicino et al., 2008;Oliveira & Bonvicino, 2006, 2011).Here we report that this species is also present at Mato Grosso state.
Sympatry records in Brazil are nonexistent for Holochilus spp.However, this should only be a matter of time before new inventories efforts are made in localities where these species nearly contact each other (Fig. 8).
Although sympatry between H. sciureus and H. chacarius have not yet been reported, it is very likely that these species distribution contact each other in central Bolivia (see Pardiñas et al., 2013) and in center-west of Brazil (Fig. 8) -Mato Grosso/Pará limits (northeast and southwest limits of H. chacarius and H. sciureus distribution in Brazil respectively) and Mato Grosso/Mato Grosso do Sul/Goiás limits (east and southeast limits of H. chacarius and H. sciureus distribution in Brazil respectively).As mentioned earlier, the northernmost previously known record with museum catalogued specimens of H. chacarius was made for the center/south of Bolivia by Anderson (1997) -reported as H. sciureus but latter some specimens was reidentified by Pardiñas & Galliari (1998) where also H. sciureus might be present.For that historical confusion, all records from Anderson (1997) should be reviewed.
Sympatric records between H. chacarius and H. brasiliensis in Argentina have been mentioned (Voglino et al., 2004).However, since the distribu-tion areas are somewhat far from each other in Brazil (Fig. 8), this is very unlikely to occur in this country or it is only restricted to a small area.On the other hand, H. sciureus and H. brasiliensis are very likely to occur simpatricaly since these species were reported here in very near localities in Minas Gerais.Another possible area of occurrence might be Bahia, where it is present the northern limit of H. brasiliensis distribution at Caravelas (17°45'S, 39°15'W), as previously noted by Massoia (1981), and the southeast limit of H. sciureus distribution at Jaguaquara (13°32'S, 39°58'W).These observations highlight the fact that only H. sciureus is known from Caatinga biome in northeastern Brazil, since we found no specimen of H. brasiliensis and morphological descriptions in literature (see qualitative analyses above) that could be assigned to this species.
Considering the previous Bolivian record from Beni (see Anderson, 1997), the present record not only provides a remarkable extension of 1,500 km toward the east of the distribution range of H. chacarius, but also represent an important record on a biogeographical standpoint.As expected, considering the distribution map observed in Bonvicino et al. (2008: 34), we report here the presence of this species to the Pantanal area of Brazil, but also the first record for the Cerrado.However, some consideration must be done regarding the presence of this species in the mentioned vegetation formations.
The record of H. chacarius in Cerrado is not totally unexpected, since the diversity of this biome have grown in the last few years, raising from 100 species (Redford & Fonseca, 1986) to 227 (Carmignotto et al., 2012).In addition, H. chacarius is a semi-aquatic species that feed mainly on riverine vegetation (mostly grass); consequently its presence might be mostly related to the presence of water bodies and phytophysiognomies suitable to the species preferences, than to a determined biome itself.It is important to mention that Carmignotto et al. (2012) already considered this species as component of Cerrado fauna, though this decision was made based on their criteria that Pantanal and Cerrado represent a single zoogeographical unit.
Although it is evident that the new record of H. chacarius provided here much further into the Cerrado of central Brazil is quite relevant, it is also important to mention that all the records we include here to H. chacarius in Brazil are areas seasonally susceptive to floods, including the new record from Cocalinho (Carmignotto pers. com.).This locality resembles swampy areas from Pantanal, which is known as the biggest flooded area in South America and where H. chacarius has already been reported.Taking this fact into perspective, this data is consistent to Carmignotto et al. (2012) criteria, which treated the Pantanal and Cerrado as a single zoogeographical unit and mention a high number of species (25) shared between Cerrado and Chaco, where H. chacarius were already known to occur.
After all we discussed above, although the IUCN (2013) category of threatened species should not change with the present work, we propose some corrections on the distribution map of Holochilus spp.provided by the mentioned institution, because the current map reflect the historical errors of identity of the specimens of this genus (Fig. 8).Furthermore, the question mark indicates those Holochilus specimens from Ilha do Bananal, Tocantins, which presents uncertain identities since we could not evaluate properly with only pelage pictures of the specimens reported by Rocha et al. (2011a).This locality might represent a contact area between H. chacarius and H. sciureus, which could bring valuable genetic and morphological information to this genus.
According to our results, all literature records for Holochilus from Pantanal region should be recognized as H. chacarius.We found no evidence for the presence of another species of Holochilus in this biome and no other species are yet known for Mato Grosso state besides H. chacarius, although we considered that the presence of H. sciureus should be only a matter of time.Finally, since Bonvicino et al. (2008) and Paglia et al. (2012) report H. vulpinus for the Pampas area in Brazil, we believe that the validity of vulpinus is current under evaluation and this species should prove to be considered taxonomically valid in a near future, adding another species to Pampas biodiversity.

CONCLUSIONS
Several names as amazonicus, balnearum, berbicensis, darwini, guianae, incarum, nanus, venezuelae and vulpinus (see Hershkovitz, 1955: fig. 140;Musser & Carleton, 2005) are yet to be consistently evaluated and, as discussed above, species limits in Holochilus are poorly understood.It is expected that the geographic ranges of the currently recognized species to be somewhat smaller, since its seems likely that all current species of Holochilus are complex of two or more cryptic species, as already indicated by a highly variable karyotypes (see Gardner & Patton, 1976).However, specimens are still scarce for several areas of the broad distribution of Holochilus species, causing uncertainty of correspondence of karyotypical variants to definable morphologies, therefore it is difficult to determine from morphological characters with only the current material available.For that reason, the recognition of the current Holochilus species is only indicative rather than a precise diagnostic for the genus diversity since morphological cryptic species could be present.
As evidenced by the data provided here, Holochilus species inhabits more than one biome, consequently it puts in perspective that our knowledge on its species distribution is limited and the identification of specimens should be based mainly on diagnostic characters and not on geographic ranges, as seemed to be the case for the small forms of Holochilus in Brazil.As a result, many mistakes concerning the identification of Holochilus specimens were made in literature; therefore, a taxonomic revision for this genus is urgently needed.

FIGURE 2 :
FIGURE 2: Distribution of the factorial scores in the first and second principal components of the craniometrical variables (in decimal logarithm) of Holochilus specimens.

FIGURE 3 :
FIGURE 3: Distribution of the factorial scores in the first and second discriminant functions of the craniometrical variables (in decimal logarithm) of Holochilus males specimens.
) identified by us as H. chacarius.As concerned to AF, Pardiñas & Galliari (1998) had anteriorly divided this character into tree types, each belonging to H. sciureus, H. c. balnearum and H. c. chacarius, who seemed to present an intermediate form of AF between the two others (see Pardiñas & Galliari, 1998: Table 2).Differently, Pardiñas & Teta (2011) did not recognized differences among subspecies of H. chacarius and recognized for this species the same AF reported by Pardiñas & Galliari (1998) to H. c. balnearum.Specimens analyzed by us proved to present all forms of AF mentioned by Pardiñas & Galliari (1998), therefore we were not able to distinguish quite easily those species or much less in subspecies level (see Fig. 5B, D and F; Fig. 6B, C and D for comparisons)

FIGURE 8 :
FIGURE 8: Map of the specimens of Holochilus spp.studied.Numbered points correspond to museums records of H. brasiliensis (black circles), H. chacarius (black triangles) and H. sciureus (black squares) listed in the gazetter (see Appendix II).The box A shows in details the museum records of the specimens from southeastern Brazil.The shades of gray are the geographic distribution of the Holochillus spp.suggested by the IUCN Red List (2013).?: Records from literature (Bezerra et al. 2009, Rocha et al., 2011a, b) analyzed in the present work only by photos of skin.

TABLE 1 :
Descriptive statistics for craniodental and external variables of adult specimens (males and females grouped) of H. chacarius, H. brasiliensis

TABLE 2 :
ANOVA and Tukey test to evaluate the existence of differences in the craniodental variables in the three Brazilian species of Holochilus.
Values in bold represent statistical difference at 5% in Tukey test and the * represent statistical differences at 5% in ANOVA.

TABLE 3 :
Factor loadings, eingevalues and percentage of variance of Principal Component Analysis (PCA) for Holochilus spp.using 14 craniodental variables.Variables with high loading are marked with boldface.

TABLE 4 :
Discriminant function analysis (DFA) for Holochilus spp.using two first component scores.Variables with high loading are marked with boldface.Brandão, M.V. & Nascimento, F.O.: On the presence of HolocHilus cHacarius in Brazil also considering it related to the overall size of the skull, it is possible to affirm that there is a trend in H. brasiliensis specimens to have bigger skull dimensions, whereas the H. chacarius and H. sciureus specimens trend to show smaller ones, and furthermore, there are overlaps among the smallest H. brasiliensis specimens and the biggest H. chacarius and H. sciureus ones.With regard to the second component, the specimens of the three species are widely mixed.

TABLE 5 :
Classification matrix for Holochilus spp.obtained by discriminant function analysis concerning the probabilities of classifying each species correctly into one of the three species. .Cross validation is done only for those cases in the analysis.In cross validation, each case is classified by the functions derived from all cases other than that case.b. 93.8% of original grouped cases correctly classified.c. 77.1% of cross-validated grouped cases correctly classified. a

TABLE 6 :
Diagnostic traits of three Brazilian Holochilus species showing qualitative variation among the external and cranial characteristics analyzed*.