Fig. 9. N in Table 2 in Fig. 2 in Zavreliella shidai Cao & Tang, 2017, sp. n.

Patterns of Differentiation and Disparity in Cranial Morphology in Rodent Species of the genus Megadontomys (Rodentia: Cricetidae) Rachel M. Vallejo1,3, José Antonio Guerrero2,*, and Francisco X. González-Cózatl3 1División de Posgrado, Instituto de Ecología, A. C. Xalapa, Veracruz, México. E-mail: [email protected] 2Facultad de Ciencias Biológicas, Universidad Autónoma del Estado de Morelos. Cuernavaca, Morelos, México 3Centro de Investigación en Biodiversidad y Conservación, Universidad Autónoma del Estado de Morelos. Cuernavaca, Morelos, México. E-mail: [email protected] (Received 12 September 2016; Accepted 9 May 2017; Published 7 June 2017; Communicated by Benny K.K. Chan) Rachel M. Vallejo, José Antonio Guerrero, and Francisco X. González-Cózatl (2017) The genus Megadontomys is a Mexican endemic group of rodents with allopatric populations occurring in fragmented patches of cool-humid forest. In this study we used geometric morphometrics methods to assess patterns of morphological variation and differentiation in skull and mandible among and within species of the genus. ANOVA showed that sexual dimorphism was significant for skulls size (P < 0.01) but not for mandibles, and MANOVA indicated that both structures did not differ in shape between sexes. ANOVA reveled a significant difference among the three species (P < 0.01), M. nelsoni exhibit the largest skull. Canonical variate analyses and Goodall’s test found differences in both skulls and mandibles shape among species, being M. cryophilus and M. thomasi the most divergent. The comparison between phylogroups within M. thomasi also revealed significant differences in shape for both structures. Disparity assessment showed that M. thomasi is the species that contributed the most to the overall shape disparity (51.80% for skull and 38.29% for mandible). The permutation test of phylogenetic signal in morphometric data was significant for the skull but not for the mandible. Morphometric data support the recognition of three morphotypes whitin the genus. The sister species M. nelsoni and M. thomasi displayed a grater shape similarity in the skull and mandible shape between them. In contrast, M. cryophilus exhibited the greatest shape divergence relative to the other species. The morphological evidence supports the existence of the two different phylogroups within M. thomasi, supporting their recognition as Evolutionary Significant Units previously suggested on molecular data. The lack of phylogenetic signal in the mandible corresponds with the environmental plasticity of this structure as compared with the skull. Key words: Disparity, Evolutionary history, Geometric morphometrics, Megadontomys, Morphological differentiation. *Correspondence: E-mail: [email protected] BACKGROUND Morphological structures constitute fundamental features for identification and description of new taxonomic groups (Arnold and Ahearn 1972; Rzhavsky 1993; Martin et al. 1996; Scotland et al. 2003; Solari 2004; GonzálezSponga 2009) and for understanding rates of species diversification (Abramov et al. 2016). Also, morphological characters have been used for reconstructing phylogenetic patterns, under the assumption that the phenotype is the result of evolutionary history (Caumul and Polly 2005). However, it is clear that environmental features may also play an important role in molding phenotypic attributes, which in some cases, may result in homoplastic characteristics (Collard and O’Higgins 2001; Caumul and Polly 2005; Gilbert and Rossie 2007). Therefore, the use of morphological characters in phylogenetic reconstruction should be carefully considered because the establishment of homology in phenotypic attributes may become complicated, and ultimately, may lead to an imprecise genealogical reconstruction (Scotland Zoological Studies 56: 14 (2017) doi:10.6620/ZS.2017.56-14 1 et al. 2003; Wiens 2004; Collard and Wood 2007; Cardini and Elton 2008). With the development of new approaches, the use of morphological characters has been reconsidered, and even, these have been employed under a different context. A relatively novel technique with the potential of formally dissociating size and shape is geometric morphometrics. This approach is based on the analysis of landmark coordinates or Cartesian geometric coordinates of morphological structures under rigorous statistical methods (Adams et al. 2004; Mitteroecker and Gunz 2009; Lawing and Polly 2011). Geometric morphometrics allows the description of patterns of shape variation within and among groups, and graphically display changes or differences among morphological characters (Adams et al. 2004). Even though geometric morphometrics is a promising approach for phylogenetic reconstruction (Catalano et al. 2010, 2014), it has been pointed out that several practical and theoretical issues should be considered in the implementation of any method that seeks to recover historical patterns (Scotland et al. 2003; Adams et al. 2011). For now, geometric morphometrics analysis has been employed to assess patterns of morphological variation (Hernández-Romero et al. 2015) or to study morphological evolution and to detect phylogenetic signal on previously recognized monophyletic groups as defined for phylogenetic analysis based on alternative kind of characters (i.e. molecular data; Cardini 2003; Adams et al. 2004). The application of this strategy has allowed the quantification of the diversity of forms within a group, using the disparity as a measure of morphological variation (Foote 1997; Collar et al. 2005). Particularly, disparity estimation may play an important role, from a conservation viewpoint, in the identification of the diversity of forms within a species that may eventually lead to the recognition of Evolutionary Significant Units (ESUs). According to Crandall et al. (2000) the categorization of population distinctiveness as ESUs, should include genetic and ecological evidence. Thus, diagnosis of distinct populations must emphasize variation in phenotypes, allowing preservation of important adaptative characters and their associated underlying genetic variation. Genetically, this variation can be shaped by gene flow, and ecologically, genetic drift and natural selection are mainly responsible for variation in phenotypes. Certainly, although morphological variation may be the result of only one or several forces, it is clear that assessment of disparity may contribute to identification of potential ESUs within species. The genus Megadontomys is a Mexican endemic group of rodents with allopatric populations occurring in fragments of cool-humid forest in the highlands of the states of Guerrero, Hidalgo, Oaxaca, Puebla and Veracruz (Fig. 1; Musser, 1964; Heaney and Birney 1977; Werbitsky and Kilpatrick 1987; Ceballos and Oliva 2005; Vallejo and González-Cózatl 2012). Recently, Vallejo and González-Cózatl (2012) reevaluated the systematics of the genus Megadontomys based on mitochondrial cytochrome b sequence data, and found support for the recognition of three species within the genus: M. cryophilus (Sierra de Juárez, Oaxaca), M. nelsoni (Sierra Madre Oriental/Sierra Mazateca) and M. thomasi (Sierra Madre del Sur/Sierra Mixteca). Their data also support a closer evolutionary relationship between M. nelsoni and M. thomasi, relative to M. cryophilus, a view compatible with Musser (1964) and Werbitsky and Kilpatrick (1987; Fig. 2). At the intraspecific level, M. thomasi is formed by two genetically differentiated lineages that are proposed as distinct Evolutionary Significant Units (Vallejo and González-Cózatl 2012; Fig. 2). Although previous taxonomic decisions and hypotheses on the evolutionary relationship within the genus Megadontomys have been addressed using morphological characters (Merriam 1898; Musser 1964; Carleton 1980, 1989), treatment of data has not included the use of formal phylogenetic or morphometric methods. Additionally, overall sample size and geographic representation have been limited. Therefore the amplification of geographical sampling and the use of morphometric techniques are justified and, even more provide finer resolution on the degree of relatedness and degree of divergence within this group. In this context, the goal of this study was to examine patterns of variation and differentiation in skull and mandible morphology among and within species of Megadontomys to assess whether these patterns are consistent with previous views on the evolutionary history of the genus, particularly those depicted by phylogenetic analyses of molecular evidence. Also, we were interested in assessing if the degree of morphological disparity is congruent with levels of molecular differentiation among species and between phylogroups within M. thomasi as pointed out by Vallejo and Gonzálezpage 2 of 15Zoological Studies 56: 14 (2017) Fig. 1. Map of México showing sampling localities of the 3 species of Megadontomys (modified from Vallejo and González-Cózatl 2012). M. cryophilus = circles; M. nelsoni = triangles; M. thomasi = diamonds. The locality code corresponds to those listed in Appendix I, where the letter stands for the respective State in México (H = Hidalgo; V = Veracruz; O = Oaxaca; G = Guerrero). 69-89-0011820 16 18 20 16 69-89-001RIO SANTO DOMINGO H1 H2 V1 V2 V4 O3 V3 O4 O5 O1 O2 G5 G6 O6 O7 VALLE TEHUACAN - CUICATLAN G4 G2 G1 G3 SIERRA MADRE ORIENTAL Cózatl (2012). To this end, we substantially increased the number of sampling localities and the total number of specimens for each species, comparing with previous studies, and analyzed morphological attributes using geometric morphometric methods. MATERIALS AND METHODS Specimens The skulls and mandibles were obtained from Colección de Mamíferos del Centro de Investigación en Biodiversidad y Conservación, Universidad Autónoma del Estado de Morelos N page 3 of 15Zoological Studies 56: 14 (2017) Fig. 2. Phylogenetic relationships of the genus Megadontomys based on cytochrome b sequence data (modified from Vallejo and González-Cózatl 2012). (CMC), and Colección Nacional de Mamíferos, Universidad Nacional Autónoma de México (CNMA). A total of 153 adults from 19 localities, representing the three species of the genus, were examined (M. cryophilus = 61 individuals, M. nelsoni = 51 individuals, M. thomasi = 41 individuals; Fig. 1; Appendix I). Also, 24 specimens of Peromyscus mexicanus were exclusively included for comparison purposes in the phylogenetic signal test (see below; Appendix I). Adults were categorized based on the wear of cusps of teeth, the presence of the third molar, and hair color (Hoffmeister 1951). Digital images, landmarks, and geometric morphometrics The ventral view of 153 skulls and lateral view of 127 mandibles (26 had broken structures and were not included in the analysis) were photographed with a digital camera Canon Power Shot A300, with a resolution of 300 pixels per inch. The coordinates X and Y of 37 landmarks for the ventral view of the skull (Fig. 3a, Table 1) and 18 landmarks for the lateral view of the mandible (Fig. 3b, Table 1), which are assumed to be homologous from one individual to another, were digitized using the tpsDig 2.15 software (Rohlf 2010). For the ventral view of the skull, only landkmarks on the left side were registered to avoid redundant information due to the bilateral symmetry condition of this structure. For each structures (skull and mandible), a Generalized Procrustes Superimposition procedure was implemented in CordGen module of IMP software (Sheets 2002) to eliminate all variation that is not related to shape, additionally, it produced an overall average shape that was used as a reference in subsequent analyses (Zelditch et al. 2004). Moreover, the coordinates of the superimposed configurations Outgroup Populations from western Sierra Madre del Sur Populations from eastern Sierra Madre del Sur / Sierra Mixteca Populations from Sierra de Juárez M. nelsoni M. thomasi M. cryophilus Populations from Sierra Madre Oriental page 4 of 15Zoological Studies 56: 14 (2017) were transformed into shape variables (partial warps) through a Thin-plate spline analysis (Bookstein 1991). This analysis produces a geometric description of shape using the partial warps to detect deformations relative to a general consensus to explain the shape change within and between species (Singh et al. 1997; Rosas and Bastir 2002). Statistical analysis Two-way univariate analysis of variance (ANOVA) for centroid size and multivariate for shape variables (MANOVA) were performed to test for the differences between sexes and among species. Statistical analyses were executed using Statistica 6.0 software (Statsoft 2001). Canonical Variate Analysis (CVA) was employed to analyze among and within species shape variation using CVA Gen6 (IMP series; Sheets 2002). This method extracts a number of axes (canonical variables) from a multidimensional space, which explain the higher proportion of the variance between the groups (Klingenberg et al. 2003). The choice of the canonical variate axes was based on the Wilks’ λ value, which is the sum of squares within groups divided by the total sum of squares within and between groups (IMP series; Sheets 2002). To graphically visualize the shape changes associated with the canonical variables, we generated deformation grids with the Thin-plate spline interpolation function (Bookstein 1991), considering only the extreme points of each axis on the CVA plot and magnifying the changes three times. Moreover, we evaluated differences in mean centroid size among the three species of Megadontomys implementing a one-way ANOVA with a Bonferroni correction as implemented in Statistica 6.0 software (Statsoft 2001). Additionally, although this not a traditional morphometric study, we performed an ANOVA on standard measures (Total length, tail length, hind foot, ear from notch, and weight) to test if there is an unequal pattern of differentiation in size, among these species. Data were obtained from specimens deposited at the CMC (Appendix I). A permutation test was performed on Procrustes distances to examine the shape differences among species and between phylogroups of M. thomasi, as recognized by Vallejo and González-Cózatl (2012). Goodall’s F statistical test is specifically designed to the coordinates produced by Procrustes superimposition (Goodall 1991). The Procrustes Fig. 3. Locations of landmarks for (a) ventral view of the skull and (b) lateral view of the mandible. Landmarks numbers corresponded to those listed in Table 1. (a) (b) page 5 of 15Zoological Studies 56: 14 (2017) Table 1. Morphological definition of landmarks for the occlusal view of the skull and the lateral view of the mandible Cranial View Number Landmarks Definition Skull 1 Anterior tip of the nasal 2 Anterior tip of suture between nasal and premaxilla 3Base of the incisivor 4Meeting point between the incisivors 5Anterior tip of incisive forame 6Meeting point between incisive forame and palatine process 7Medium point of incisive forame 8Medium point of palatine process 9 Posterior point of incisive forame 10 Posterior point of palatine 11 Posterior point lateral of palatine 12 Meeting point between external and internal pterigoid processes 13 Posterior point of internal pterigoid process 14 Lateral point of tympanic bulla 15 Anterior point of occipital 16 Meeting point between occipital condyle 17 Posterior point of foramen magnum 18 Meeting point between foramen magnum and occipital condyle 19 Posterior meeting point between occipital condyle and occipital 20 Posterior lateral point of external auditory meatus 21 Anterior lateral point of external auditory meatus 22 Posterior point of external pterigoid process 23 Anterior point of tympanic bulla close to external pterigoid process 24 Posterior point of squamosal 25 Lateral medium point of squamosal 26 Anterior point of squamosal 27 Meeting point between squamosal and jugal 28 Meeting point between jugal and zygomatic 29 Posterior point of third molar 30 Lateral point of third molar 31 Meeting point between 1st and 2nd molar 32 Anterior point of 1st molar 33 Posterior point of malar process 34 Anterior point of malar process 35 Meeting point between malar process and maxilla 36 Lateral point of maxilla 37 Anterior point of maxilla Mandible 1 Base of the incisor 2 Anterior point of diastema 3 Inferior point of maxillary toothrow 4 Posterior point of diastema 5Meeting point between 1st and 2nd molar 6Meeting point between 2nd and 3rd molar 7 Base of the coronoid process 8Tip of the coronoid process 9Medium point of incisura mandibulae 10 Anterior tip of the condyle 11 Medium tip of the condyle 12 Posterior tip of the condyle 13 Medium point of the condyle 14 Medium point of the angular process 15 Posterior tip of the angular process 16 Anterior point of the angular process 17 Inferior medium point of mandible 18 Anterior point of the masseteric ridge page 6 of 15Zoological Studies 56: 14 (2017) distances express the differences between shapes from each group especially when sample size is unequal (Cardini 2003; Zelditch et al. 2004). Goodall’s F test was applied employing the software TwoGroup6 (IMP series; Sheets 2002) with 2500 permutations. Morphological diversity in the shape of both cranial structures among and within species was assessed using the Procrustes distances following the method proposed by Foote (1993). This approach considers that morphological disparity depends on a measure that reflects distances among points in morphological space. In this case, the variances are the measures used for explain the disparity because they have the property of being additive allowing the calculation of the partial disparity to obtain the overall disparity as the result of the partial contributions for each group (Foote 1993; Zelditch et al. 2004). Partial disparity estimations were obtained with the software PairDisparity6 (IMP series; Sheets 2002). Phylogenetic signal test The phenotype could be the resulted either from of phylogenetic history or adaptations to local environments (Caumul and Polly 2005). If the first case, the morphology of two groups that share a common ancestor will be more similar in comparison with that of most distant groups. Several studies have used a permutation test to evaluate the presence or absence of phylogenetic signal in the morphometric data (Figueirido et al. 2010; Meloro et al. 2011; Klingenberg et al. 2012). This test assumes that closely related forms tend to occupy the same portion of the morphometric space, because they share a common ancestor in comparison with distantly related species, which are found at different segments of the morphometric space. This approach is implemented by mapping morphometric traits onto know phylogenies by the method of squared-change parsimony to reconstruct ancestral shape of morphometric data. The test simulates the null hypothesis of the complete absence of phylogenetic structure by permutation of the shape data among the terminal taxa (Klingenberg and Gidaszewski 2010). In order to perform this test we selected Peromyscus mexicanus as outgroup. Although there is uncertainty about the sister group of the genus Megadontomys, several studies have pointed out a close phylogenetic affinity between these taxa (Rogers 1983; Rogers et. al. 1984; Vallejo and González-Cózatl 2012). The permutation test was implemented in MorphJ software with 50,000 random permutations (Klingenberg 2008). RESULTS Sexual dimorphism For skulls, sexual dimorphism in size was significant. Males were, on average, larger than females in the three species. Interspecific differences in size were also significant, while sex by species interaction was not significant (Table 2). Differences in skull shape were, however, significant only between species (Wilks’ λ = 0.012, P < 0.05, Table 3). For mandibles, neither size nor shape was significantly different between sexes (P > 0.05, Table 2 and Table 3) only significant differences were detected in shape among species (Wilks’ λ = 0.066, P > 0.05, Table 3). For both skull and mandible, interaction between sex and species was not significant (P > 0.05, Tables 2 and 3); therefore, sexes were pooled in all subsequent analyses. Table 2. Two-way ANOVA for sex, species, and sex × species interaction effects for the skull centroid size in Megadontomys. Significant P values (P < 0.05) are indicated in bold and with an asterisk Cranial View Effect Sum of squares F d.f. P-value Skull Sex 97.1 5.67 10.0185* Species 177.2 5.18 20.007* Sex × Species 34.5 1.01 2 0.367 Error 2515.25 147 Mandible Sex 13.6 3.49 10.064 Species 62.8 8.03 20.000* Sex × Species 0.1 0.01 22 0.99 Error 473 121 page 7 of 15Zoological Studies 56: 14 (2017) Fig. 4. Scatterplot of CV1 and CV2 scores for oclussal view of skull. TPS deformation grids for the extreme points of each axis are shown. Deformation grids were 3x exaggerated. Table 3. Two-way ANOVA for sex, species, and sex × species interaction effects for the skull centroid size in Megadontomys. Significant P values (P < 0.05) are indicated in bold and with an asterisk Cranial View Effect Wilks’ λF d.f. P-value Skull Sex 0.526 170 0.49 Species 0.012 8.745 140 0.000* Sex × Species 0.294 0.938 140 0.648 Mandible Sex 0.692 1.249 32 0.206 Species 0.066 8.063 64 0.000* Sex × Species 0.58 0.88 64 0.718 CVA and interspecific morphometric variation Canonical variates analyses, with species as the grouping variable, found differences in both skulls (Fig. 4) and mandibles (Fig. 5). In both cases, the first two canonical variates showed significant differentiation in the shape among three species: M. cryophilus, M. nelsoni and M. thomasi (Fig. 4 and Fig. 5; Skull: CV1: Wilks’ λ = 0.0127, P < 0.0001, CV2: Wilks’ λ = 0.1673, P < 0.0001; mandible: CV1 Wilk’s λ = 0.0674, P < 0.0001, CV2 Wilks’ λ = 0.3479, P < 0.0001). Shape differences detected by canonical variates are illustrated on TPS grids (Figs. 4 and 5). When comparing the skull between M. cryophilus and M. thomasi, CV1 shows that the former has a shorter malar process page 8 of 15Zoological Studies 56: 14 (2017) and maxilla (landmarks 3537), short and narrow incisive foramen (landmarks 59), the region of tympanic bulla and occipital is short and narrow (landmarks 1315, 2223), closed zygomatic arch (landmarks 2627), narrow squamosal (landmark 24) and the region of the maxillary molars is long (landmarks 2932) (Fig. 4). Compared to M. thomasi, M. nelsoni has a large malar process and maxilla (landmarks 3537), long and wide incisive foramen (landmarks 59), a long and narrow tympanic bulla and occipital region (landmarks 1315, 2223), opened zygomatic arch (landmarks 2627), wide squamosal (landmark 24) and the region of the maxillary molars is short (landmarks 2932) (Fig. 4). Canonical variate 1 showed that the mandibles of M. thomasi and M. cryophilus are different (Fig. 5). M. cryophilus has a shorter mandible, a narrower condyle (landmarks 10, 12), a longer coronoid process (landmark 8), a deeper incisura mandibulae (landmark 9), a more opened posterior point of diastema (landmark 4), and a wider and longer angular process (landmarks 1416). Canonical variate 2 revealed that M. nelsoni has a large mandible, but it is narrow in the zone of processes and in the middle portion. Also, it possesses a wider condyle (landmarks 10, 12), shorter coronoid process (landmark 8), wider incisura mandibulae and longer molar region (landmarks 9, 4), and it is long and wide in the angular process (landmarks 1416) (Fig. 5), compared with M. thomasi. Morphometric differentiation among and within species ANOVA showed a significant difference in size among groups/species (skull: F = 5.05, P = 0.007; mandible: F = 7.24, P = 0.001). For both skull and mandible, M. nelsoni was the largest (skull centroid size (SCS) = 71.049, mandible centroid size (MCS) = 30.145), although there were not differences between M. cryophilus (CS = 68.531) - M. thomasi (CS = 69.462) and M. nelsoni (CS = 71.049) - M. Fig. 5. Scatterplot of CV1 and CV2 scores for lateral view of the mandible. TPS deformation grids for the extreme points of each axis are shown. Deformation grids were 3x exaggerated. page 9 of 15Zoological Studies 56: 14 (2017)