Peromyscus is one of the most representative genera of rodents in North America (^{Bedford & Hoekstra, 2015}) displaying a wide range of morphological, physiological, and behavioral variation (^{Kirkland & Layne, 1989}). Nevertheless, for some species, overlapping of morphological characters, color pattern, and shared distribution areas may make it difficult to accurately identify them in the field and scientific collections (^{Ramírez-Pulido et al., 2001}; ^{González-Ruíz et al., 2005}). This is the case of the Southern rock deer mouse Peromyscus difficilis (J. A. Allen, 1891), widely distributed from Chihuahua to Oaxaca (^{Fernández et al., 2010}), and the Perote mouse Peromyscus bullatus (^{Osgood, 1904}), endemic to the Oriental basin in Veracruz and Puebla (^{González-Ruíz & Álvarez-Castañeda, 2005}). Peromyscus difficilis is an abundant species found mainly in rocky volcanic areas with xerophytes and in forests with pinyon pines, junipers, and oaks. On the other hand, populations of P. bullatus are small, isolated, and restricted to flat sandy arid areas with grasses and scattered trees (^{Galindo-Leal & Krebs, 2004}; ^{González-Ruíz et al., 2005}). According to Mexican legislation, the Perote mouse is classified under special protection and is considered critically endangered by the IUCN (^{Álvarez-Castañeda, 2018}; ^{SEMARNAT, 2018}). To contribute to the correct discrimination of P. bullatus and P. difficilis, our objective was to evaluate the usefulness of using tympanic bulla morphometric measures, recommended by Osgood (1904) and ^{Hoffmeister (1951)}, and corroborated by ^{González-Ruíz et al. (2005)} as key characters for the accurate discrimination between these species. For this purpose, we search for P. bullatus specimens erroneously identified as P. difficilis in the mammal collection of the Instituto de Investigaciones Biológicas of the Universidad Veracruzana (IIB - UV). Since misassignment of P. bullatus specimens as P. difficili IUCN s is frequent (^{González-Ruíz et al., 2005}), we expected to find at least one miscataloged P. bullatus specimen, achieving hence an accurate discernment between studied species with the proposed tympanic bulla measures.

We obtained morphometric data from 109 adult specimens (according to ^{Hoffmeister, 1951}) cataloged as P. difficilis in the IIB-UV mammal collection (SEMARNAT VER-MAM-191-10-06); 52 females and 57 males. All specimens were from the region of San Antonio Limón (Totalco), currently a Protected Natural Area (^{Gobierno del Estado de Veracruz, 2016}), in central Veracruz, Mexico. From each specimen’s tag, we obtained sex and body measurements: total length - TL, tail length - CL, hindfoot length - HL, ear length - EL. We measured seven dimensions from the right side of craniums: occipito-nasal length (OL), braincase breadth (BC), postorbital constriction (PC), nasal length (NL), maxillary tooth row length (LX), bulla length (BL), and bulla breadth (BW) after ^{Santos-Moreno et al. (1998)} and ^{González-Ruíz et al. (2005)}. All measurements were taken using an electronic caliper (Mitutoyo series 500; 0.01 mm accuracy) by the same person (IF). We calculated the hindfoot to ear length ratio (HL-EL) and the occipito-nasal length to bulla length ratio (OL-BL; ^{González-Ruíz et al., 2005}).

To find specimens which were different within the P. difficilis group, possibly P. bullatus whose identity (i.e. catalog number) we did not know, we generated a hierarchical cluster dendrogram (Euclidian distance, average linkage; ^{Kettenring, 2006}) in R (^{R Core Team, 2018}) for the joint sexes, since sexual dimorphism is absent in both species (^{González-Ruíz & Álvarez-Castañeda, 2005}; ^{Fernández et al., 2010}). For the dendrogram, we used the BL, BW, OL-BL and LX measurements; reported by ^{González-Ruíz et al. (2005)} as key characters to differentiate P. bullatus from P. difficilis. We used hierarchical cluster analysis because it is based on groups’ similarities/dissimilarities and is not necessary to know a priori the number of groups nor the identity of their elements (Kettenring, 2006).

According to the cluster dendrogram, four specimens were different from the remaining sample (IIB-UV catalog number: 1052, 1053, 1673, 1695) at a linkage distance > 1.5 (Fig. 1). After a thorough inspection, we identified these four specimens as P. bullatus since they showed, on average, smaller somatic (TL, CL, HL) and cranial (OL, NL, LX, BC, PC) values and larger ear (EL) and bulla measurements (BL, BW, OL-BL ratio) than P. difficilis (Table 1). These specimens were corroborated as P. bullatus by González-Ruíz from the Universidad Autónoma Metropolitana and were re-cataloged in the IIB-UV mammal collection. The four specimens identified as P. bullatus constituted the first ones for the IIB-UV mammal collection and were added up to those already deposited in other collections.

Figure 1 Cluster dendrogram for the joint sexes. Dashed box indicates the specimens different on average from the rest of the group, which was re-cataloged as P. bullatus in the IIB-UV mammal collection. 

For both P. bullatus and P. difficilis our measurements were lower than those recorded by ^{González-Ruíz et al. (2005)}, likely due to variation of the studied populations or to inter-measurer variation (^{Santos-Moreno et al., 1998}). Regardless of these differences, the data and description of species provided by ^{González-Ruíz et al. (2005)} greatly helped us to discriminate specimens’ species. While the BL, BW, and LX measurements were valuable for discriminating specimens through hierarchical cluster analysis, we found that the OL-BL ratio was the most important measurement for species identification. In P. bullatus specimens OL-BL ratio exceeded the average value for P. difficilis, corroborating the findings reported by ^{González-Ruíz et al. (2005)}. Furthermore, we did not find overlapping average ± SD values for the OL-BL ratio between species. While average ± SD values for P. bullatus cranial measurements should be interpreted cautiously due to our n = 4, the showed tendency by morphometric data for this species was clear and agreed with the findings of ^{González-Ruíz et al. (2005)}. However, contrary to descriptions by ^{Osgood (1909)} and conclusions by ^{González-Ruíz et al. (2005)}, the HL-EL ratio was not useful to us since all the studied specimens presented values within the same range, possibly due to HL values were taken by several measurers.

The contribution by ^{González-Ruíz et al. (2005)}, together with cluster analysis, were valuable for discriminating species when craniums were studied, nevertheless, for identification of P. bullatus in the field, besides total length and coloration (slightly smaller and lighter than P. difficilis), the habitat type where the specimens are captured is a valuable criterion due to the specific requirements of the species, (^{Hoffmeister, 1951}; ^{González-Ruíz et al., 2005)}. Since only 3.67% of the studied specimens turned out to be P. bullatus, whose rarity and small populations in the field are corroborated, our findings support the conservation efforts for this species, which has been left with a reported 6.3% of remnant habitat (^{Sánchez-Cordero et al., 2005)}.

Since the scientific collections are a valuable bank of information, an accurate assignment of species to specimens is fundamental. Especially for underrepresented species, which in turn are often endemic or poorly studied ones. Additionally, erroneous identification can be detrimental for systematic, taxonomic, morphometric, and conservation studies of species. Even more, due to information deposited in collections has a greater scope and availability through digital platforms. Because of this, a routine review of the specimens to update or correct the information has been widely recommended (^{Lorenzo et al., 2012}). Given our findings and the importance of having accurately designated specimens, we support a further search for P. bullatus among the P. difficilis specimens in collections, when they have been collected from the Oriental basin region in Mexico. The use of the bulla variables, particularly the OL-BL ratio, is useful for performing such task.