Thomasomys cinereus is the type species of Thomasomys, type genus of the sigmodontine tribe Thomasomyini. As currently recognized, Thomasomys includes 48 species, all of which are endemic to humid montane or premontane forests in the tropical Andes. Although it has been suggested that T. cinereus is a species complex, this hypothesis has yet to be critically evaluated. Herein we provide a revision of the species based on a qualitative assessment of external, craniodental, and soft morphological traits; morphometric analyses; a phylogenetic analysis based on cytochrome b gene sequences; species delimitation methods; and first-hand examination of type material. Our analyses of genetic data recovered four distinct clades within T. cinereus, one corresponding to T. cinereus sensu stricto (restricted to the montane forests delimited by the Río Marañón, Río Huancabamba, and Río Tablachaca in Cajamarca department, Peru) and three new species: Thomasomys lojapiuranus, sp. nov., from the montane forests of Piura department, Peru, and Loja province, Ecuador; T. shallqukucha, sp. nov., restricted to the Kañaris montane forests in the Peruvian department of Lambayeque; and T. pagaibambensis, sp. nov., restricted to the montane forests of Pagaibamba in Cajamarca department, Peru. These species can be distinguished by several discrete morphological traits of the skull, dentition, mandible, stomach, palatal rugae, and glans penis. Genetic distances among these taxa range from 5.06%–7.65% at the cytochrome b locus, and delimitation analyses based on cytochrome b sequence data support their recognition as distinct species. Our results suggest the existence of previously unsuspected dispersal barriers in the Andes of northern Peru, and they confirm that the Río Marañón is a formidable barrier that limits the distribution of species of Thomasomys as well as other sigmodontine rodents.
INTRODUCTION
Thomasomys Coues, 1884, is the type genus of the tribe Thomasomyini in the cricetid rodent subfamily Sigmodontinae. The genus currently includes 48 species (Pacheco et al., 2015; Brito et al., 2019, 2021; Ruelas and Pacheco, 2021a; Lee et al., 2022), all of which are endemic to the tropical Andes from about 1150 m to 4600 m above sea level, where most of them appear to be restricted to montane forests (Patton, 1987; Pacheco et al., 2015). Although several new species have been described in recent years (e.g., T. onkiro Luna and Pacheco, 2002; T. ucucha Voss, 2003; T. andersoni Salazar-Bravo and Yates, 2007; T. salazari Brito et al., 2019; T. pardignasi Brito et al., 2021; T. antoniobracki Ruelas and Pacheco, 2021; and T. burneoi Lee et al., 2022) the genus as a whole remains unrevised.
The type species of Thomasomys is Hesperomys cinereus Thomas, 1882, which was based on a holotype collected at Cutervo in Cajamarca department, Peru. Thomas (1882) described T. cinereus using external morphology and basic morphometric values, and Voss (1993) subsequently described additional key external, cranial, dental, and soft-anatomical characters. Pacheco (2003) confirmed these diagnostic morphological attributes and added several others. More recently, Pacheco (2015) provided an updated account of Thomasomys cinereus, describing the species afresh, comparing it with other congeners known at that time, and reporting marginal localities from the Peruvian departments of Cajamarca (west of the Río Marañón), La Libertad, Lambayeque, and Piura. Subsequently, T. cinereus has also been reported from the province of Loja in southern Ecuador (Lee et al., 2018; Moreno-Cárdenas and Novillo-Gonzalez, 2020). Although Pacheco (2015) considered T. cinereus monotypic, he suggested that it might be a species complex. To date, however, no taxonomic revision of the material currently referred to T. cinereus has been undertaken.
Based on a phylogeny using external, cranial, dental, postcranial, and soft-anatomical characters, Pacheco (2003, 2015) proposed seven informal species groups for Thomasomys, of which the Cinereus Group was the most diverse with 23 species. This group includes mainly small to medium-size rodents (head and body length: 89–161 mm) distributed in montane forests from Venezuela and Colombia to Bolivia. The species that Pacheco (2003, 2015) included in the Cinereus Group included T. australis Anthony, 1925; T. bombycinus Anthony, 1925; T. caudivarius Anthony, 1923; T. cinereiventer J.A. Allen, 1912; T. cinereus (Thomas, 1882); T. cinnameus Anthony, 1924; T. contradictus Anthony, 1925; T. daphne Thomas, 1917; T. dispar Anthony, 1925; T. emeritus Thomas, 1916; T. erro Anthony, 1926; T. fumeus Anthony, 1924; T. hudsoni Anthony, 1923; T. hylophilus Osgood, 1912; T. laniger (Thomas, 1895); T. monochromos Bangs, 1900; T. niveipes (Thomas, 1896); T. onkiro Luna and Pacheco, 2002; T. paramorum Thomas, 1898; T. silvestris Anthony, 1924; T. ucucha Voss, 2003; T. vestitus (Thomas, 1898); and T. vulcani (Thomas, 1898). Recently, Brito et al. (2019) described T. salazari and included it along with two undescribed taxa in the Cinereus Group, bringing the potential total membership to 26 species.
Phylogenetic analyses of cytochrome b sequence data have provided molecular support for some of Pacheco's (2003, 2015) species groups (Salazar-Bravo and Yates, 2007; Lee et al., 2015, 2018; Brito et al., 2019, 2021). In particular, Lee et al. (2018) and Brito et al. (2019) both recovered a monophyletic Cinereus Group, but T. cinereus was represented in their analyses only by Ecuadorian specimens, the identification of which has yet to be convincingly established. Genetic data obtained from topotypical specimens of T. cinereus and specimens from other Peruvian populations currently referred to T. cinereus have yet to be analyzed. As a result, the phylogenetic position of T. cinereus is still unclear, nor has the distributional range of the species been critically evaluated with either phenotypic or genetic analyses.
Herein, we review the taxonomy of Thomasomys cinereus sensu lato (i.e., as this name was applied by Pacheco, 2015) and provide a hypothesis about its phylogenetic relationships. Based on phylogenetic analyses of cytochrome b sequence data, qualitative morphological comparisons, and morphometric analyses, we recognize four species among specimens previously identified as T. cinereus, three of which are described as new. Our results contribute to a better taxonomic understanding of the genus, and they provide a framework for future biogeographic studies of small mammals in northern Peru.
MATERIAL AND METHODS
Source of Material: For morphological assessment we examined 350 specimens (skulls, skins, and fluid-preserved material) previously referred to Thomasomys cinereus (including the holotype) from throughout the distributional range of the species as it has been identified in the past (appendix 1). This morphological material is deposited in the following institutions (acronyms in parentheses): American Museum of Natural History, New York (AMNH); Natural History Museum, London (formerly the British Museum of Natural History, BMNH); Field Museum of Natural History, Chicago (FMNH); Museum of Zoology, Louisiana State University, Baton Rouge (LSUMZ); Museo de la Escuela Politécnica Nacional, Quito (MEPN); Museo de Historia Natural, Universidad Nacional Mayor de San Marcos, Lima (MUSM); Museum of Vertebrate Zoology, University of California, Berkeley (MVZ); Sección de Mastozoología, Museo de Zoología de la Pontificia Universidad Católica del Ecuador, Quito (QCAZ-M); University of Michigan Museum of Zoology, Ann Arbor (UMMZ); and the National Museum of Natural History, Smithsonian Institution, Washington D.C. (USNM).
Collection localities for examined specimens of Thomasomys cinereus (sensu Pacheco, 2015: including the new specimens described herein) are mapped in figure 1 and gazetteered in appendix 1. The specimens we sequenced for cytochrome b are listed in table 1.
TABLE 1
Ingroup and outgroup taxa sequenced for cytochrome b used in phylogenetic tree reconstruction. Species include new taxa (marked with an asterisk) described in this report. Voucher provides museum catalog number or collector's field number. Locality numbers in parentheses (for members of the Thomasomys cinereus complex) are mapped in figure 1.
continued
continued
continued
Molecular Analyses
DNA Isolation: We isolated genomic DNA from 28 specimens of Thomasomys cinereus (sensu Pacheco, 2015) from Piura, Cajamarca, Lambayeque, and La Libertad departments in Peru and from one specimen of T. cinereus (sensu Moreno-Cárdenas and Novillo-Gonzalez, 2020) from Loja province, Ecuador (table 1). DNA was isolated from portions of liver and muscle preserved in alcohol or from fragments of dried skin using Wizard® SV Genomic DNA Purification System Kit (Promega Corp.). We amplified and sequenced the mitochondrial gene cytochrome b (cytb) by polymerase chain reaction using primers MVZ05, MVZ45, MVZ04, MVZ16, and MVZ14 of Smith and Patton (1993). The thermal profile of PCR reactions consisted of an initial denaturation step at 94° C for 4 min, followed by 33 cycles of PCR (denaturation at 93° C for 1 min, annealing at 45° C for 1 min, polymerization at 72° C for 2 min), and a final extension at 72° C for 10 min. Gene isolation and amplification were performed in the Evolutionary Biology Lab of the Mammalogy Department of MUSM. Amplified products were sequenced by an external sequencing service (Macrogen. Inc. Seoul, South Korea). The sequences were manually checked using CodonCode 6.0.2 (CodonCode Corporation, Centerville, Massachusetts). New sequences were deposited in GenBank (table 1).
Intraspecific Analyses: Genetic relationships among geographic samples of Thomasomys cinereus sensu lato were determined by the median-joining approach (Bandelt et al., 1999) using R 4.1.3 (R Core Team 2022) with the packages: “ape,” “pegas,” and “scales.” Then, we performed an Analysis of Molecular Variance (AMOVA; Excoffier et al., 1992) and computed fixation indices (FST) using Arlequin 3.5 (Excoffier and Lischer, 2010) to assess genetic isolation among the geographic populations currently referred to this species.
Phylogenetic Analyses: For phylogenetic analysis, we used the following taxa as outgroups: 10 representatives of other sigmodontine tribes (Sigmodon hispidus, Wiedomys pyrrhorhinos, Microryzomys minutus, Oecomys bicolor, Abrothrix olivaceus, Akodon boliviensis, Oxymycterus rufus, Andinomys edax, Auliscomys micropus, and Phyllotis magister), 5 representatives of other thomasomyine genera (Chilomys instans, Rhagomys longilingua, Rhagomys rufescens, Rhipidomys albujai, Rhipidomys latimanus), and 14 species of Thomasomys representing other species groups of Thomasomys (Thomasomys sp. 1 [sensu Pacheco, 2003], T. andersoni, T. aureus, T. auricularis, T. baeops, T. gracilis, T. cf. incanus, T. kalinowskii, T. ladewi, T. macrotis, T. notatus, T. oreas, T. pardignasi, and T. taczanowskii). Members of the Cinereus Group (as defined in the Introduction) were considered the ingroup (table 1), among which we included 34 partial sequences of Thomasomys cinereus sensu lato.
Phylogenetic relationships were inferred by maximum likelihood (ML) and Bayesian inference (BI). The model and parameters used in both analysis were previously selected under the Akaike Information criterion using JModelTest 2.1.10 (Darriba et al., 2012). The best model was GTR+I+G with the following parameters: rate matrix (1.0886, 6.9496, 1.0512, 0.3431, 11.6214, 1.0000), base frequencies (0.3101, 0.3319, 0.0820, 0.2760), p-inv = 0.4530, gamma distribution shape parameter (0.8350). ML analyses were performed using IQ-TREE multicore 1.6.3 (Nguyen et al., 2015) with 10,000 nonparametric bootstrap replicates to search for the best tree and to compute bootstrap support (BS) values. Bayesian analysis was performed in MrBayes 3.2.6 (Ronquist et al., 2012) with 10 million generations, four simultaneous Markov chains, and trees sampled every 1000 generations per chain. The first 25% of the trees were discarded as burn-in, and the remaining trees were used to compute a 50% majority rule consensus tree and to obtain the Bayesian posterior probability (BPP) for each node. Chain convergence was evaluated using the ESS (effective sample size) indicator, which was greater than 200 in all priors using Tracer 1.7 (Rambaut et al., 2018). Consensus trees were edited using iTOL v 4 (Letunic and Bork, 2016).
Species Delimitation: We performed four single-locus species-delimitation analyses using the cytb dataset: Automatic Barcode Gap Discovery (ABGD; Puillandre et al., 2011), Assemblage of Species by Automatic Partitioning (ASAP; Puillandre et al., 2021); Poisson Tree Processes (PTP; Zhang et al., 2013), and General Mixed Yule Coalescent (GMYC; Fujisawa and Barraclough, 2013; Pons et al., 2006). For the ABGD method, we uploaded the FASTA alignment in the ABGD online version ( https://bioinfo.mnhn.fr/abi/public/abgd/) with settings: pmin = 0.001, pmax = 0.10, X = 1.00, and using simple distance. For ASAP method, we uploaded the FASTA alignment in the ASAP web ( https://bioinfo.mnhn.fr/abi/public/asap/) using the default settings and simple distance. For PTP method, we uploaded trees of BI in bPTP web server ( https://species.h-its.org/; Zhang et al., 2013) with 5 x 105 MCMC generations. For these three methods, we used only ingroup species (members of the Cinereus Group as defined above). The GMYC method was performed in the GMYC web server ( https://species.h-its.org/gmyc/), using a calibrated tree by the uncorrelated relaxed-clock log-normal model with multiple secondary calibration points performed in Beast (Bouckaert et al., 2019). We followed Parada et al. (2021) for the crown ages of sigmodontine tribes.
Genetic Distances: We calculated the corrected cytb genetic distances using the Kimura 2-parameter model and pairwise deletion option within and among candidate species using MEGA v. 7.0 (Kumar et al., 2016).
Species Concept: We recognize species in accordance with the general lineage species concept (de Queiroz, 2007) considering the morphological (qualitative and quantitative evidence), reciprocal monophyly, large genetic distances (Baker and Bradley, 2006), ecological evidence, and molecular species delimitation.
Morphological and Morphometric Analyses
External and Craniodental Measurements: Skin measurements (total length, TL; length of tail, LT; hind foot length, HFL) and weight were transcribed from specimen tags. However, we sometimes remeasured the length of hind foot (HFL) on fluid or dry specimens to check the accuracy of values recorded by the collector, and we used our value rather than the collector's value whenever large discrepancies were found. Head-and-body length (HBL) was estimated by subtracting tail length from total length. We use the ratios “Tail%” (= LT/HBL × 100) and “HFL%” (= HFL/HBL × 100) to quantify proportions of the tail and hind foot with respect to head-and-body length for descriptive purposes. All external measurements are reported to the nearest millimeter (mm) and all weights to the nearest gram (g). All cranial and mandibular variables were measured to the nearest 0.01 mm using dial calipers following Tribe (1996), Luna and Pacheco (2002), and Voss (2003). The following variables were measured (fig. 2):
Greatest length of skull (GSL): measured from the tip of the nasals to the posterior margin of occiput
Condyloincisive length (CIL): distance from the greater curvature of an upper incisor to the articular surface of the occipital condyle on the same side
Condylomolar length (CML): distance from the anterior edge of the first upper molar to the articular surface of the occipital condyle on the same side
Length of orbital fossa (LOF): greatest distance of the orbital fossa inside the maxillary and squamosal roots of the zygomatic arch.
Length of nasals (LN): the greatest length of either nasal bone
Rostral length (RL): the greatest diagonal distance from the zygomatic notch to the tip of the nasal bone on the same side.
Diastema length (LD): measured from the crown of the first upper molar to the exposed lesser curvature of the upper incisor on the same side
Length of incisive foramina (LIF): distance from the anterior to the posterior edge of one incisive foramen
Length of maxillary toothrow (LM): occlusal length of the upper molar row
Breadth of incisive foramina (BIF): the greatest distance across the incisive foramina
Breadth of rostrum (BR): the least breadth between the anteroventral edges of the zygomatic plates
Breadth of palatal bridge (BPB): measured between the protocones of first maxillary molars
Breadth of first upper molar (BM1): measured across the protocone-paracone cusp pair of the first upper molar
Breadth of nasals (BN): the greatest distance across both nasal bones
Least interorbital breadth (LIB): the least interorbital breadth
Zygomatic breadth (ZB): the greatest distance across the zygomatic arches
Braincase breadth (BB): measured immediately posterodorsal to the squamosal roots of the zygomatic arches
Breadth of zygomatic plate (BZP): the distance between the anterior and posterior edges of the zygomatic plate
Depth of incisor (DI): distance between greater and lesser curvature of the upper incisor
Height of braincase (HBC): distance from the top of the braincase to the ventral surface of the basisphenoid and basioccipital bones
Mesopterygoid fossa breadth (MFB): greatest internal width across the mesopterygoid fossa
Age Criteria: An estimate of relative age was based on molar toothwear. Five categories are here defined using a modification of the scheme presented by Voss (1991) for Zygodontomys and by Carleton and Musser (1989) for Microryzomys:
Age class 0: M3 incompletely erupted
Age class 1: M3 erupted but essentially unworn
Age class 2: M3 little worn (some dentine exposed), paracone tuberculate and prominent, paraflexus distinct. M1–2 primary cusps prominent
Age class 3: M3 worn, occlusal surface flat or concave but hypoflexus still distinct; M1–2 moderately worn, primary cusps well-defined, distinct, but not prominent; flexus distinct
Age class 4: M1–3 worn, occlusal surface essentially flat or concave, hypoflexus obliterated
Hereafter, we consider specimens in age classes 0–1 as “juveniles,” age class 2 as “subadult,” and age classes 3 and 4 as “adults.”
Morphology: We followed Smithe (1975) for the characterization of fur coloration. Anatomical terminology follows Brown (1971) and Brown and Yalden (1973) for external characters; Reig (1977, 1980) and Carleton and Musser (1989) for dental nomenclature; Voss (1988) for cephalic arteries; Wahlert (1985) for cranial foramina; Howell (1926) for postcranial characters; Sprague (1941) and Carleton (1980) for the hyoid apparatus; Carleton and Musser (1984) and Voss (1988) for general features of muroid anatomy; Carleton (1973) for gross stomach morphology; Quay (1954), Carleton (1980), and Pacheco (2003) for transverse palatal rugae; Pacheco (2003, 2015) and Zavodszky and Russo (2020) for chevron bones of the caudal vertebrae; Hooper and Hart (1962), Hooper and Musser (1964), and Myers et al. (1990) for the glans penis; and Pacheco (2003, 2015) for other anatomical features of thomasomyines. Descriptions of right upper and right lower molar rows were used. Glandes penes were dissected carefully from the prepuce of fluid-preserved specimens, and some were lightly stained (for 1–3 hours) with 0.3% alizarin. Dissected glandes were maintained in 70% ethanol. The presence or absence of preputial glands was also observed macroscopically. Stomachs were dissected following Carleton (1973) and preserved, with the gastric contents, in 70% ethanol; some contents were identified.
Morphometric Analyses: Standard univariate statistics (mean, range, standard error, coefficient of variation, skewness, and kurtosis) were calculated for each external and craniodental variable for each taxon in the Thomasomys cinereus complex (i.e., T. cinereus and the three species recognized as new in this report). Multivariate normality was assessed utilizing the results of the Royston's H test. ANOVAs were calculated to test the null hypothesis of no significant differences in each variable between the sexes. In addition, MANOVA with Wilk's Lambda and Pillai trace criteria based on the largest available population sample (Nfemales = 29, Nmales = 24) from Huancabamba, department of Piura, was performed for an overall sex effect, but no significant differences were found (λ = 1.021, p >0.05). Therefore, the sexes were combined for subsequent morphometric analyses. Then, ANOVA followed by the statistical post hoc analysis Tukey-Kramer test for unbalanced groups was performed for each variable among taxa. Only normally distributed variables were used in ANOVAs.
Multivariate analyses were performed on measurements of adult specimens (age classes 3 and 4) to reduce intraspecific variation due to age (Rengifo and Pacheco, 2015; Pacheco, 2021). The variables RL and MFB (which were measured only on MUSM specimens) and GSL and CML (which were highly correlated with CBL) were not used in the multivariate analyses. We extracted principal components (PCs) from the variance-covariance matrix using log10-transformed data for the four species in the Thomasomys cinereus complex and in pairwise analyses. Specimens scores of the first three components were used to project two-dimensional patterns of variation. Additionally, we performed a discriminant function analysis (DFA) with Jackknife (Lance et al., 2000) to estimate the percentage of accurate classification. Statistical significance for all analyses was assessed as p <0.05. All analyses were executed in SPSS v 23 and R v 4.1.3 (R Core Team, 2022) with the complementary packages “MVN” (Korkmaz et al., 2014) for the multivariate normality test and “MASS” (Venables and Ripley, 2002) for DFA.
RESULTS
In order to simplify the presentation of our results in the following text, tables, and figures, we anticipate our taxonomic conclusions and use Latin binomina for the allopatric lineages we recognize as full species within Thomasomys cinereus sensu lato. The geographic sense in which these names (T. cinereus sensu stricto, T. lojapiuranus, T. pagaibambensis, and T. shallqukucha) are applied is illustrated in figure 1. The phrase “intraspecific” as applied below, however, refers to T. cinereus sensu lato (i.e., as previously employed by Pacheco, 2015).
Intraspecific Analyses
More than 30 mutational steps separate Thomasomys cinereus sensu stricto. from the candidate taxa recognized in this study (fig. 3). Additionally, AMOVA recovered a high fixation index (FST = 0.89) among them and between species pairs (T. cinereus versus T. lojapiuranus, FST = 0.88; T. cinereus versus T. pagaibambensis, FST = 0.87; T. cinereus versus T. shallqukucha, FST = 0.88; T. lojapiuranus versus T. pagaibambensis, FST = 0.87, T. lojapiuranus versus T. shallqukucha, FST = 0.94; and T. pagaibambensis versus T. shallqukucha, FST = 0.96). Both lines of evidence suggest the absence of gene flow among these taxa.
Phylogenetic Analyses
We analyzed fragments (403–801 bp) of the cytb gene from 116 specimens, among which 399 sites were invariant, 74 were singletons, and 328 were parsimony informative. Both ML and BI analyses of cytb recovered similar topologies (fig. 4). The monophyly of Thomasomys was recovered with strong support in both analysis (BS = 99%, BPP = 1.00) (appendix 2). The monophyly of the Cinereus Group was also recovered, but with low or moderate support (BS = 49%, BPP = 0.74). Within the Cinereus Group, two strongly supported clades (labelled A and B in fig. 4) were recovered, but two members of the Cinereus Group, T. erro and T. onkiro, were not members of either clade; their relationships remain effectively unresolved by our analyses.
Clade A includes Thomasomys australis, T. daphne, T. fumeus, T. bombycinus, and T. vulcani. Within this clade, a lineage consisting of T. australis and T. daphne, and another consisting of T. fumeus, T. bombycinus, and T. vulcani were each recovered with strong support. None of the members of this clade have ever been identified as or associated with T. cinereus.
Clade B includes Thomasomys ucucha, T. silvestris, T. cinnameus, T. caudivarius, T. paramorum, T. hudsoni, T. salazari, two unnamed forms (Thomasomys sp. 1 and Thomasomys sp. 2), and four species belonging to a complex that correspond to T. cinereus sensu lato (i.e., sensu Pacheco [2015] and Moreno-Cárdenas and Novillo-Gonzalez [2020]): T. cinereus sensu stricto, T. lojapiuranus, T. pagaibambensis, and T. shallqukucha. Within clade B, T. ucucha is sister to a lineage that includes T. silvestris and T. sp. 1, whereas T. cinereus sensu stricto is sister to a lineage that includes T. shallqukucha, T. lojapiuranus, and T. pagaibambensis (of which the last two are sister species). All other relationships within this clade have only weak to moderate support, but it is noteworthy that the Ecuadorean taxon identified as T. cinereus by Lee et al. (2018)—denoted as T. “sp. 2” in these results—is not closely associated with the species complex that includes T. cinereus sensu stricto.
Observed genetic distances among species of the Cinereus Group of Thomasomys (table 2) range from 6.67 ± 0.92% (between T. cinereus and T. lojapiuranus) to 16.39 ± 1.65% (between T. onkiro and T. fumeus). Within the T. cinereus complex, the genetic distances among T. cinereus sensu stricto, T. lojapiuranus, T. pagaibambensis, and T. shallqukucha are all in the range of 5.06%–7.65%. By comparison, intragroup divergences in this complex range from 0.22 ± 0.07% in T. lojapiuranus (11 sequences from five localities) to 1.39 ± 0.2% in T. cinereus sensu stricto (14 sequences from eight localities).
The results of the ABGD, ASAP, and PTP species-delimitation analyses all suggest that Thomasomys cinereus sensu lato is comprised of four putative species as recognized above, whereas the GMYC analysis inferred eight putative species, of which five correspond to T. cinereus sensu stricto (fig. 4). The divergence threshold estimated by the GMYC analysis was 0.210 Mya.
Morphometric Analyses
Descriptive statistics for adult specimens of Thomasomys cinereus, T. lojapiuranus, T. shallqukucha, and T. pagaibambensis are provided in table 3. The variables CML, LN, LD, LIF, LM, BPB, BN, LIB, BZP, DI, and HBC had distributions that did not depart significantly from normality. The results of ANOVAs applied to these variables (appendix 3) showed that T. cinereus sensu stricto and T. lojapiuranus differ significantly in LD, LIF, LM, and DI; that T. cinereus and T. shallqukucha differ significantly in CML, LD, LIF, and LIB; that T. cinereus and T. pagaibambensis differ significantly in all variables except LN, LM, BN, and BZP; that T. lojapiuranus and T. pagaibambensis differ significantly in all variables except LD and BZP; that T. lojapiuranus and T. shallqukucha differ significantly only in CML and LIB; and that T. pagaibambensis and T. shallqukucha differ significantly only in BPB, BN, and HBC.
Factor loadings and percentages of variation of the first three principal components (PC1–PC3) for the analysis with the Thomasomys cinereus complex are presented in appendix 4, where the first principal component has uniformly positive coefficient values, indicating that this axis is predominantly related to variation in skull size (and size-correlated proportions). Plots of PC1 versus PC2 (fig. 5A) exhibit broad overlap among Thomasomys cinereus sensu stricto, T. lojapiuranus, T. shallqukucha, and T. pagaibambensis; but plots of PC2 versus PC3 (fig. 5B) are less overlapped. These three axes account for 56.31% of the total variance (PC1: 31.75%, PC2: 12.79%, PC3: 11.77%). By contrast, discriminant function analysis provided an almost complete separation by group (fig. 5C), with each group identified by a candidate taxon. DF1 explained 52.48% of the variance, and the most important variables were CML and LOF; DF2 explained 41.53% with LM and BB as the most important variables, and DF3 explained 5.98% with ZB and BM1 as the most important variables (appendix 4). Overall, 86.4% of the specimens were accurately classified by Jackknife. By species, 90.91% of T. cinereus sensu stricto, 87.18% of T. lojapiuranus, 95% of T. pagaibambensis, and 60% of T. shallqukucha were accurately classified. Additionally, plots of PC1 versus PC2 of Thomasomys cinereus sensu stricto versus T. lojapiuranus (appendix 5A), T. cinereus versus T. pagaibambensis (appendix 5B), and T. cinereus versus T. shallqukucha (appendix 5C) showed these taxa partially overlapped. Plots of PC1 and PC2 for T. lojapiuranus versus T. pagaibambensis (appendix 5D) and T. lojapiuranus versus T. shallqukucha (appendix 5E) showed these taxa also partially overlapped. Finally, the plot of PC1 and PC2 for T. pagaibambensis versus T. shallqukucha showed these taxa overlapped (appendix 5F).
TAXONOMIC ACCOUNTS
Our molecular analyses revealed that Thomasomys cinereus sensu lato comprises four genetically differentiated populations that represent reciprocally monophyletic lineages, and our morphometric analyses suggest modest phenotypic divergence in craniodental size and proportions. Additionally, side-by-side visual comparisons of specimens indicate that these lineages can be diagnosed by differences in multiple external, skeletal, and soft-anatomical characters (table 4). We hypothesize that they are distinct species that should be recognized as follows.
TABLE 2
Corrected average pairwise genetic distances using the Kimura 2-parameter model (below diagonal) and standard errors (above diagonal) among species of the Cinereus Group of Thomasomys
Diagonal values (in parentheses) are average intraspecific distances.
TABLE 3
External and craniodental measurements (mm) and weights (g) for members of the Thomasomys cinereus complex
Table entries include the sample mean, standard deviation, observed range (in parentheses), and sample size. GSL and CML (marked with an asterisk) were not used in multivariate analyses to reduce the number of highly correlated variables.
Thomasomys cinereus (Thomas, 1882)
Olive-gray Thomasomys
Figures 6A, 7, 8, 9, 10A, 11A, 11E, 12A, 13A, 14, 15A, 16A, 17A, 18A, 19A, 19E, 20A, 21, 22A
H[esperomys]. (Rhipidomys) cinereus Thomas, 1882: 108; type locality “Cutervo, 9,200 feet [2,804 m]” Cajamarca, Peru.
Hesperomys (Vesperimus) cinereus: Thomas, 1884: 449; name combination.
Hesperomys (Thomasomys) cinereus: Coues, 1884: 1275; name combination.
Peromyscus (Thomasomys) cinereus: Trouessart, 1898: 512; name combination.
Thomasomys cinereus: Thomas, 1906: 443; first use of current name combination.
Type: A subadult (age 2) female specimen (BMNH 81.9.7.29) collected at Cutervo, Chota [Cutervo] Province, department of Cajamarca, Peru; 9200 ft (2804 m), in February or March 1879 by the well-known Polish collector Jan Stanislaw Stolzmann. The specimen is preserved in alcohol with skull removed. Thomas's description was based on a single uncataloged specimen that Gyldenstolpe (1932) later identified as BMNH 81.9.7.29 (the holotype by monotypy).
Emended Diagnosis: A medium-sized species (table 3) of Thomasomys distinguished from congeneric taxa by the following combination of characters: pelage long and soft, above ashy gray with hair tips brownish; mystacial vibrissae short, extending barely beyond posterior margin of pinnae; metatarsals whitish; tail equal to or less than head-and-body length, moderately bicolored (paler ventrally than dorsally), without a white tip and without a terminal pencil; thenar and hypothenar pads closely approximated; nasals long, expanded anteriorly, tapering posteriorly to lacrimals, and extending slightly behind premaxillae; zygomatic notch relatively deep; interorbital region narrow; zygomatic plate moderately broad, anterior edge vertical; incisive foramina long, extending posteriorly to or between first molars; posterior margin of palatal bridge usually with a median process; mesopterygoid fossa usually narrow with subparallel sides; Eustachian tube short and narrow; stapedial process of bulla conspicuous; capsular process of lower incisor alveolus absent; upper incisors orthodont; M1 anteromedian flexus distinct; M3 hypoflexus conspicuous; and m1 anteromedian flexid distinct.
Morphological Description: Medium sized Thomasomys (HBL = 114–148 mm) with long, soft, and dense dorsal fur (individual hairs about 13–14 mm). Dorsal coloration is grizzled ashy gray to dark gray with hairs being slate colored at the base (Blackish Neutral Gray, color 82) and white at tips, sprinkled with longer and brownish hairs (Vandyke Brown, color 221) or Raw Umber (color 123). Ventral pelage is pale yellowish (Pale Horn Color, color 92) or pale gray (Pale Neutral Gray, color 86), hairs are also slate colored at the base, and moderately countershaded with dorsal pelage (figs. 8, 9). The ears are short (EL = 17–22 mm) and pale. Postauricular patches of cream-colored fur are conspicuous. The mystacial and supraorbital vibrissae are short, extending barely behind the posterior margin of the pinnae when laid back alongside the head. The genal vibrissae are absent. Females have three pairs of mammae in thoracic, abdominal, and inguinal position (sensu Pacheco, 2003). The tail is comparatively thick, moderately bicolored, and without a conspicuous white tip. The tail is, on average, about as long as the combined length of the head and body (Tail% = 103). The manus is whitish with digit II larger than digit V. The hind feet are moderately long (27–33 mm) and narrow, with metatarsals covered by pure white shining hairs that extend to the digits. The pedal digits (dII–dV) exhibit long and dense ungual tufts that cover the claws. The plantar surface of the hind foot is pale (unpigmented), and the thenar and hypothenar pads are not widely separated (fig. 10A). Digit I of the hind foot (hallux) is moderately long, its claw extending close to or to the first interphalangeal joint of dII. Digit V of hind foot is long, with the claw extending about half the length of phalanx 2 of dIV.
Cranium: The skull is moderately large (CIL = 28.55–34.45 mm) with a flat or barely convex profile, without a bump on the frontal region. The rostrum is relatively long and moderately broad (fig. 11A). A rostral tube is absent, and the gnathic process is small (figs. 12A, 15A). The nasals are long and tapering; the posterior margins are narrow, extending posteriorly to the lacrimals and beyond the premaxillae. The zygomatic notches are usually comparatively deep (55%) or shallow (fig. 13A). The lacrimals are small without a posterior process. The interorbital region is narrow and hourglass shaped with rounded margins, and the frontal sinuses are not inflated. The braincase is broad and squarish. The lateral parietal processes are moderately large and subtriangular. The interparietal is wide and moderately long anteroposteriorly, straplike in some specimens, but rhomboidal in others (fig. 14). The zygomatic arches converge anteriorly to a moderate degree. The zygomatic plates are moderately broad, subequal in breadth to the length of M1, and vertically oriented (fig. 15A). The premaxillae are slightly produced anteriorly beyond the incisors (fig. 15A). The posterior margin of the inferior zygomatic root is dorsal to the anterior margin of M1. The zygomatic processes of the maxilla and the squamosal are closely approximated; the jugal is slender (fig. 16). The postglenoid foramen is small and anterior to a relatively larger subsquamosal fenestra. The tegmen tympani is robust and overlaps a distinct suspensory process of the squamosal. The carotid circulatory pattern is pattern 1 (sensu Voss, 1988): a stapedial foramen, a groove on the inner surface of the squamosal and the alisphenoid, and a sphenofrontal foramen are all present. In the medial wall of the orbit, the ethmoid foramen is placed dorsal to M2; the ethmoturbinals are moderate in size; the sphenopalatine foramen is bordered by the maxillary, ethmoid, and palatine bones; and the optic foramen is moderate in size and slightly posterior to M3. The paraoccipital process is small, slightly curved, and not bifurcated. The incisive foramina are long, and extend posteriorly to, or usually (in 73% of examined specimens) slightly between, the anterior margins of the first molars; the maxillary septum is long but less than half the incisive foramina length (fig. 17A). The palate is short and wide (sensu Hershkovitz, 1962). The mesopterygoid fossa is narrow (in 76% of examined specimens) with subparallel sides; a median posterior process of the palate is present in 75% of examined specimens. The sphenopalatine vacuities are usually represented by narrow slits along the basisphenoid and presphenoid bones. The parapterygoid fossa is triangular, shallow, and usually (in 80% of examined specimens) without vacuities. The foramen ovale is moderate in size, the alisphenoid strut is present, and the middle lacerate foramen is narrow but conspicuous. The auditory bullae are small, flask shaped, and not inflated; the Eustachian tube is short and narrow, with two spines that usually do not reach the alisphenoid; the stapedial process of the bulla is conspicuous. The internal carotid canal is bounded by the basioccipital and the ectotympanic (auditory bulla), but not by the periotic. The basioccipital is moderately broad, and the mastoid (the occipital exposure of the periotic) is either imperforate or has only a small fenestra. The lamina of the malleus is squarish and not deep, and the orbicular apophysis is short and knob shaped due to a conspicuous basal constriction (fig. 18A). The processus brevis of the incus is long, narrow, and delicate.
Teeth: The upper incisors are orthodont with orange enamel on their anterior surfaces. The upper molar rows are moderately long (5.00–5.74 mm) and parallel. The upper molars are brachyodont and pentalophodont (sensu Hershkovitz, 1962) without interpenetration of the labial and lingual flexi; labial and lingual cingula are not developed; and accessory labial roots are absent. The procingulum of the first upper molar (M1) is slightly narrower than the protocone-paracone cusp pair, and the anteromedian flexus conspicuously divides a slightly smaller anterolingual conule from a larger anterolabial conule. The anteroloph is conspicuous and the paraflexus is recurved to deeply recurved. The paraloph is oriented transversally to the median mure and, in some cases, to the base of the mesoloph. The mesoloph is well developed, reaching the labial margin of the tooth, and it is often fused to the metacone producing a closed metaflexus that is curved or sometimes angular. The posteroloph is usually coalesced with the metacone or persists as a minute structure in a few specimens. M2 exhibits a strong anteroloph with a recurved paraflexus; the mesoloph is narrow and usually complete, the metaflexus is “comma-shaped,” and the posteroloph is coalesced with the metacone. M3 is much smaller than M2; its paraflexus, mesoflexus, and hypoflexus are conspicuous; and its metacone is obsolete (fig. 19A).
On the lower molars, the main cuspids are conspicuous and slightly alternating. The first lower molar (m1) has a distinct anteromedian flexid; the protolophid is short; the anterolophid is short or obsolete; the anterolabial and anterolingual cingula are poorly developed; the mesoflexid is broad and recurved; the entoflexid is small or coalesced; the posteroflexid is broad and almost straight; the hypoflexid is broad and oriented perpendicularly; and the mesolophid is narrow but complete. On m2, the mesolophid is short or coalesced with the entoconid, and the anterolabial cingulum is small. The m3 has a subtriangular occlusal shape produced by the less-developed entoconid; the mesoflexid and hypoflexid are conspicuous; and the posteroflexid is reduced or coalesced (fig. 19E).
Mandible: The coronoid process of the mandible is narrow, long, falciform, and taller than the condylar process, and produces a deep sigmoid notch. The angular process is short and anterior to the condylar process. The capsular process of the lower incisor alveolus is absent or indistinct. Superior and inferior masseteric ridges are posterior to the anterior margin of m1. The mental foramen is below the diastemal border on the lateral surface of the mandible (fig. 7).
Hyoid apparatus: The basihyal has a slightly convex anterior border, and the posterior border is concave; an entoglossal process is absent. The thyrohyal is shorter than the basihyal, and it is arched laterally. The ceratohyal is much smaller and more slender than the thyrohyal (observation made in five specimens).
Penis: Four fluid-preserved male specimens (MUSM 38005, 38006, 38493, 38497) were examined. The glans is medium sized and subcylindrical, with the dorsum convex and the venter slightly flat in profile. On average, the midshaft diameter (2.55 mm) is less than the overall length of the glans (4.54 mm) (fig. 20A). A dorsal groove is present and confluent with the crater lip; a ventral groove is also present but is shallow and less developed. Laterally, other shallow depressions divide the glans into 5 or 7 sections on each side. The epidermal spines are rather sparse overall except along the rim, which is spineless. Each spine is in a shallow pit, and the spines are larger at the base of the glans than they are near the apex. The crater lip circumscribes the entire crater opening, and is separated from the spinous epithelium by a narrow fold. The medial bacular mound projects conspicuously from the crater lip and is thus visible externally; the much smaller lateral bacular mounds project only a little and are barely visible externally. The tip of the medial mound is usually oriented ventrally, except in one specimen where it is slightly dorsally oriented. The urethral flaps are well developed, triangular, undivided, and relatively long, but deeply buried within the crater; their tips are well separated, and a few minute spines are present on each. The dorsal papilla is well developed, conical, and bears several small spines on the tip or dorsal surface; it is surrounded by tissue folds connecting the bacular mound with the inner surface of the crater.
Palatal rugae: Two complete (diastemal) and five incomplete (interdental) transverse palatal ridges (terminology following Quay, 1954, and Carleton, 1980) characterize most known species of Thomasomys (Pacheco 2003), including T. cinereus and T. lojapiuranus. In T. cinereus, the first diastemal ridge is slightly triangular and the second ridge is slightly convex. The interdental rugae i1, i2, i3, and i5 are slightly convex and each extends to the palate midline, whereas the interdental ruga i4 is very short and does not reach the palate midline (fig. 21).
Stomach: The stomach corresponds to the unilocular-hemiglandular morphotype (Carleton, 1973; Pacheco, 2003). The incisura angularis is shallow, and the bordering fold is thin and crosses the lesser curvature of the stomach slightly anterior to the incisura angularis; thus, some cornified epithelium extends into the antrum; the bordering fold also recurves moderately to the left but not beyond the esophageal orifice (fig. 22A).
Gall bladder: Present.
Reproductive glands: As described by Pacheco (2003), adult males have one pair each of dorsal prostate, anterior prostate, ampullary, vesicular, and bulbourethral glands, and two pairs of ventral prostate glands. The medial ventral prostate is much smaller than the lateral ventral prostate. Macroscopic preputial glands are absent.
Skeleton: Two partial skeletons each have 7 cervical, 13 thoracic, 6 lumbar, 4 sacral, and more than 36 caudal vertebrae; and the same skeletons each have 13 pairs of ribs. The first two caudal vertebrae in one specimen, and the first three caudal vertebrae in the second specimen, have chevron bones fused forming a closed arch. In both specimens, the closed arch of the first caudal vertebra has a small spinous process.
Karyotype: Unknown.
Measurements of Holotype: GSL, 33.05; CIL, 30.95; CML, 20.15; LOF, 11; LN, 12.95; LD, 9.25; LIF, 7.4; LM, 5.7; BIF, 2.4; BR, 6.2; BPB, 3.3; BM1, 1.8; BN, 4.3; LIB, 4.9; ZB, 18; BB, 15.05; BZP, 2.9; DI, 1.7. Measurements of additional specimens are provided in table 3.
Distribution: Thomasomys cinereus is distributed in northern Peru, west of the Río Marañón and south of the Huancabamba Depression and the Río Chamaya. On the western side of the Andes, the species is distributed south of the Río Chancay from Monteseco in the department of Cajamarca southward to the Río Tablachaca (a main tributary of the Río Santa) in the department of La Libertad. On the eastern side of the species range, all records are on the west (left) side the Río Marañón, from Cutervo (fig. 1: locality 25) in the north to the Río Crisnejas in the south, all within the department of Cajamarca (fig. 1). The altitudinal range is from 1524 m to 3818 m.
Natural History: This species inhabits the humid montane forests or Yungas of northern Peru, which has also been called the Peruvian Yungas ecoregion of the Tropical and Subtropical Moist Broadleaf Forest biome (Dinerstein et al., 2017). Specimens were collected on the ground among dense shrubs or inside forests with small trees (fig. 23A). Arboreal or semiarboreal habits are suggested because we observed that some specimens easily climb shrubs and small trees when released.
Among specimens with reproductive data, 21 males were collected from July to October (usually considered the dry season); nine of them had scrotal testes and 12 had abdominal testes. Of eight male specimens collected from November to March (usually considered the wet season), four had scrotal testes and four had abdominal testes. Of 16 females collected from July to October, 14 had closed vaginas, one had an open vagina, and one was lactating. For the wet season, there is only one record of one female with an open vagina in April.
The stomach contents of one specimen was full of insect remains and a second specimen had some insect remains together with plant material.
Johnson (1972) and Durden and Musser (1994) reported Thomasomys cinereus as the host for the anopluran louse Hoplopleura angulata, but Pacheco (2015) corrected the host identification to T. ischyrus. Several species of nematodes of the genus Trichuris, Vianella, Malvinema, Aspidodera, and Pterygodermatites have been recently reported from specimens of T. cinereus collected at Bosque Cachil, Cajamarca (Polo-Gonzales, 2020).
At several localities, Thomasomys cinereus is sympatric with T. taczanowskii (Thomas, 1882) and T. pyrrhonotus Thomas, 1886 (MUSM database). Other natural history information about this species is still unknown.
Remarks: Pacheco (2015) corrected the identification of some specimens previously identified as Thomasomys cinereus, including one collected from Maraynioc (near Chanchamayo), Junín department (Thomas, 1884), another reported near Uchco, Amazonas (Osgood, 1914), and other records from Antioquia (Hooper and Musser, 1964) and Huila (Carleton, 1973) in Colombia. Therefore, the descriptions of penis morphology (Hooper and Musser, 1964) and stomach morphology (Carleton, 1973) previously attributed to T. cinereus do not correspond to this species. However, Pacheco (2003) reported on these morphological attributes based on correctly identified specimens, and his observations are expanded here based on examination of additional specimens.
One specimen of Thomasomys cinereus LSUMZ 20313 (Esselstyn, 2017) has locality information and geographic coordinates that we believe to be wrong, and this problem merits comment. The specimen was collected from “33 road km SW Chiriaco” in Piura Department by K.R. Thomas. However, the recorded coordinates for this locality (5.372°S, 77.929°W) correspond to a point in the lowlands of Amazonas department far from all other known records of T. cinereus. We suggest that either the locality datum is a typographical error for “33 road km SW Huancabamba” in Piura Department (where Thomas also collected), or the specimen (which we have not examined) is misidentified.
Two specimens (AMNH 73127, 73129) were said to have been collected at Seques in Lambayeque department by Watkins on his way to Taulis in Cajamarca, although Lambayeque currently has no locality named Seques, but Cajamarca does. In the province of San Miguel (district of La Florida) there are localities known as “Pampa de Seques” and “Mountain of Seques” ( https://www.deperu.com/centrospoblados/pampa-de-seques-37687), with approximate coordinates: 6.890384°S, 79.088967°W. These places are on the route that Watkins travelled to Taulis, so it seems likely Watkins collected these specimens in Cajamarca rather than Lambayeque.
TABLE 4
Selected morphological comparisons among Thomasomys cinereus sensu stricto, T. lojapiuranus, T. shallqukucha, and T. pagaibambensis
Mystacial vibrissae are “short” when they barely extend behind the posterior margins of pinnae or “long” if they extend conspicuously behind the pinnae. Incisive foramina are “long” when they extend to or between the first molar crowns or “short” if they do not extend to or between the first molars. Capsular process is of the lower incisor alveolus.
Thomasomys lojapiuranus, new species
Loja and Piura Thomasomys
Figures 6B, 8, 9, 10B, 11B, 11F, 12B, 13B, 15B, 16B, 17B, 18B, 19B, 19F, 20B, 21, 22B, 24
Thomasomys cinereus Voss, 1993: part,
Thomasomys sp. 2 Brito et al., 2019: 9.
Holotype: An adult female specimen housed in the Museo de Historia Natural de la Universidad Nacional Mayor de San Marcos (MUSM 23758), collected by V. Pacheco on April 28, 2006 (original field number VPT 3149). The holotype is preserved as skin and skull in good condition.
Paratypes: Four paratypes from the same locality, two females (MUSM 23759, 23762) and two males (MUSM 23760, 23761).
Type locality: Peru, Department of Piura, Province of Huancabamba, Pariamarca Alto; 5.15867°S, 79.54901°W, 2990 m above sea level.
Diagnosis: A medium-sized Thomasomys (table 3) that can be distinguished from its congeners by the following combination of characters: dorsal pelage long and soft, brownish with hair tips yellowish; mystacial vibrissae long, extending beyond posterior margin of pinnae; tail longer than head-and-body length, unicolored and without a white tip; thenar and hypothenar pads separated; plantar surface of hind foot pale; nasal relatively short, expanded anteriorly, and tapering posteriorly to the maxilla-frontal-lacrimal intersection or shorter; zygomatic notch shallow; incisive foramina long, extending posteriorly to the anterior margins of first molars, not between; palatal median process usually present (78%); Eustachian tube short and broad; stapedial process of bulla reduced; capsular process of lower incisors alveolus absent; M1 anteromedian flexus weak; M3 hypoflexus weak; and m1 anteromedian flexid weakly produced.
Morphological Description: Thomasomys lojapiuranus is a medium-size Thomasomys (HBL = 107–146 mm) with long, soft, and dense dorsal fur (individual hairs about 14–16 mm). Dorsal pelage is brownish (Dark Drab, color 119B) with hairs slate colored at the base. Ventral pelage is gray whitish (Smoke Gray, color 45) or washed with pale brown, individual hairs slate colored basally. Dorsal and ventral pelage are moderately countershaded (figs. 8, 9). The ears are moderately long (EL = 18–26 mm) and paler. Postauricular patches of cream-colored fur are conspicuous. The mystacial vibrissae are long, extending behind the posterior margin of the pinnae by about half the pinnae length when laid back alongside the head. The supraorbital vibrissae are also relatively long, extending behind the posterior margins of pinnae when laid back alongside the head. The genal 1 vibrissae are absent. The tail is on average longer than the combined length of head and body (Tail% = 117), comparatively thick, mostly unicolored, and without a white tip. The metacarpals are pale brownish. Digit II of manus is larger than digit V. The hindfeet are moderately long (HFL = 28–32 mm) with metatarsal patch pale brownish, and dense ungual tufts. The plantar surface of the hind foot is pale. The thenar and hypothenar pads are separated by a distinct gap (fig. 10B). Digit I of the hind foot (hallux) is moderately long, its claw extending close or to the first interphalangeal joint of dII. Digit V of the hind foot is long, with the claw extending about half the length of phalanx 2 of dIV.
Cranium: The skull is moderately large (CIL = 29–32 mm) with a relatively long and moderately broad rostrum. The skull dorsal profile is slightly convex, never straight, with a small bump at the interorbital region level. A rostral tube is absent, and the gnathic process is small (figs. 12B, 15B). The nasals are short and tapering; the anterior half is expanded; the posterior margins are narrow, extending posteriorly to the maxilla-frontal-lacrimal intersection or shorter and barely beyond the premaxillae. The zygomatic notches are shallow (fig. 13B). The interorbital region is moderately broad and hourglass shaped with rounded margins, and the frontal sinus are relatively inflated. The interparietal is wide and moderately long anteroposteriorly, straplike in some specimens, but rhomboidal in others. The premaxillae are slightly produced anteriorly beyond the incisors (fig. 15B). The zygomatic arches converge anteriorly to a moderate degree. The zygomatic plates are moderately broad, subequal in breadth to the length of M1, and moderately sloped backward (fig. 15B). The zygomatic process of the maxilla and the zygomatic process of the squamosal are closely approximated (80%) or separated by a distinct gap (20%) (fig. 16B). The postglenoid foramen is small and placed anterior to a relatively larger subsquamosal fenestra. The tegmen tympani is robust and overlaps a distinct suspensory process of the squamosal. The carotid circulatory pattern is pattern 1 (sensu Voss, 1988): a stapedial foramen, a groove on the inner surface of the squamosal and the alisphenoid, and a sphenofrontal foramen are all present. In the medial wall of the orbit, the ethmoid foramen is placed dorsal to M2; the ethmoturbinals are moderate in size; the sphenopalatine foramen is bordered by the maxillary, ethmoid, and palatine bones; and the optic foramen is moderate in size and slightly posterior to M3. The paraoccipital process is small. The incisive foramina are long and extend posteriorly to the anterior margins of the first molars, not between, and are widest behind the premaxillary-maxillary suture; the posterior margins are rounded (fig. 17B). The mesopterygoid fossa is broad and expanded anteriorly, the roof is closed, and the sides are subparallel; the palatal median process is present (in 78% of examined specimens). The parapterygoid fossa is triangular, shallow, and without vacuities. The foramen ovale is moderate in size, the alisphenoid strut is present, and the middle lacerate foramen is narrow but conspicuous. The auditory bullae are small, flask shaped, and moderately inflated with short and broad Eustachian tubes; the stapedial process of bulla is reduced. The internal carotid canal is bounded by the basioccipital and the ectotympanic (auditory bulla), but not by the periotic. The basioccipital is moderately broad, the mastoid (the occipital exposure of the periotic) is either imperforate or has only a small fenestra. The lamina of the malleus is squarish and not deep, and the orbicular apophysis is elongated without a basal constriction (fig. 18B). The processus brevis of the incus is long, narrow, and delicate.
Teeth: The upper incisors are orthodont with orange enamel on their anterior surfaces. The upper molar rows are moderately long (4.8–5.5 mm). The upper molars are brachyodont and pentalophodont (sensu Hershkovitz, 1962) without interpenetration of the labial and lingual flexi; labial and lingual cingula are not developed; and accessory labial roots are absent. The procingulum of M1 is slightly narrower than the protocone-paracone cusp pair, and the anteromedian flexus is weak or coalesced. The anteroloph is conspicuous and the paraflexus is recurved. The paraloph is oriented transversally to the median mure, to the base of the mesoloph, or the mesoloph. The mesoloph is narrow and well developed and reaches the labial margin of the tooth. The metaflexus is slightly curved. The posteroloph is coalesced with the metacone. M2 exhibits a strong anteroloph with a recurved paraflexus; the mesoloph is narrow and complete, the metaflexus is comma shaped, and the posteroloph is coalesced with the metacone. M3 is smaller compared to M2; its paraflexus is conspicuous but the hypoflexus is reduced (fig. 19B).
On the lower molars, the main cuspids are conspicuous and slightly alternating. The first lower molar (m1) has a weak anteromedian flexid; the protolophid and anterolophid are short and quickly coalesced; the anterolabial and anterolingual cingula are poorly developed; the mesoflexid is narrow and recurved; the entoflexid is small or coalesced; the posteroflexid is narrow and almost straight; the hypoflexid is narrow and oriented perpendicularly; and the mesolophid is narrow but complete. On m2, the mesolophid is short and the anterolabial cingulum is small. The m3 has a subtriangular occlusal shape produced by the less developed entoconid; the mesoflexid and hypoflexid are conspicuous; and the posteroflexid is coalesced (fig. 19F).
Mandible: The coronoid process is long, narrow, falciform, and taller than the condylar process, producing a deep sigmoid notch. The angular process is well developed and on the same plane as or slightly anterior to the condylar process. The ventral border of the mandible is deeply concave. The capsular process of the lower incisor alveolus is absent (fig. 24).
Hyoid apparatus: Hyoid morphology like that described for Thomasomys cinereus (observation made only in one specimen).
Penis: Two fluid-preserved male specimens (MUSM 23760, 23761) were examined. The glans is medium size, subcylindrical, relatively long (4.48 mm), and narrow (2.28 mm) (fig. 20B). A dorsal groove is absent, and a ventral groove is exhibited as a shallow sulcus. The external surface of the glans is covered by small epidermal spines overall except along the crater rim. The spineless crater lip circumscribes the entire crater opening and is separated from the spinous epithelium by a thick fold. Distally, the glans is not divided on the lateral sides, but a shallow notch is present (fig. 20B). The medial bacular mound projects prominently from the crater lip and is thus visible externally; the lateral bacular mounds are much smaller than in Thomasomys cinereus, lack ornamentation, and are not visible externally. The tip of the medial bacular mound is broad and oriented vertically or slightly ventrally oriented. The urethral flaps are small, conical, deeply buried within the crater, and lack ornamentation. The dorsal papilla is small, conical, and lack spines or other ornamentation.
Palatal rugae: Two complete (diastemal) and five incomplete (interdental) transverse palatal ridges like in Thomasomys cinereus, except that the interdental ruga i4 extends to the palate midline and the interdental ruga i5 is nearly straight (fig. 21).
Stomach: The stomach corresponds to the unilocular-hemiglandular morphotype (Carleton, 1973; Pacheco, 2003). The incisura angularis is shallow, and the bordering fold is thicker and crosses the lesser curvature of the stomach slightly anterior to the incisura angularis; the bordering fold also gently recurves to the left but not beyond the esophageal orifice (fig. 22B).
Gall bladder: Present.
Reproductive glands: Macroscopic preputial glands are absent.
Skeleton: The vertebral count is 7 cervical, 13 thoracic, 6 lumbar, 4 sacral, and 37 to 40 caudal vertebrae; and the same skeletons each have 13 pairs of ribs. The supratrochlear fenestra of humerus is usually absent (6 of 7 samples). The first caudal vertebra has chevron bones fused forming a closed arch, mostly without a spinous process (in 5 of 7 examined specimens). As described by Pacheco (2003), on the calcaneus, the trochlear process levels its posterior articular facet; and the peroneal process of the fifth metatarsal is moderately long but not extending to the proximal edge of the cuboid.
Karyotype: Unknown.
Measurements of Holotype: GSL, 33.1; CIL, 30.15; CML, 19.52; LOF, 10.5; LN, 12.03; RL, 11.28; LD, 8.94; LIF, 6.31; LM, 5.15; BIF, 2.67; BR, 5.39; BPB, 3.36; BM1, 1.7; BN, 4.27; LIB, 5.2; ZB, 17.56; BB, 14.15; BZP, 2.73; DI, 1.7; HBC, 9.35; MFB, 2.45. Measurements of additional specimens are provided in table 3.
Distribution: Thomasomys lojapiuranus is distributed in the montane forests of the western slope of the department of Piura, in northern Peru, and the province of Loja in southern Ecuador. All examined specimens are from north of the Huancabamba depression, most of them in the headwaters of the Río Huancabamba, a main tributary of the Río Chamaya (fig. 1). The altitudinal range is from 1198 m (but see Remarks) to 3481 m.
Taxonomic Comparisons: Thomasomys lojapiuranus differs from T. cinereus by its brownish dorsal pelage (vs. grizzled ashy gray to dark gray in T. cinereus); whitish gray or pale brown ventral pelage (vs. pale yellowish or pale gray); mystacial vibrissae long, extending behind the posterior margin of pinnae by about half the pinnae length when laid back alongside the head (vs. shorter in T. cinereus, table 4); tail longer than head-and-body length (Tail% = 117±11% vs. 103±8%) and mostly unicolored (vs. moderately bicolored); thenar and hypothenar pads separated by a distinct gap (vs. thenar and hypothenar pads closely approximated); zygomatic notches shallow (vs. deep [55%] or shallow in T. cinereus); incisive foramina long, but never extending posteriorly between the first upper molars (vs. usually extending between the first upper molars in T. cinereus); stapedial process of bulla reduced (vs. conspicuous); orbicular apophysis elongated without basal constriction (vs. knob shaped in T. cinereus); M1 anteromedian flexus, M3 hypoflexus, and m1 anteromedian flexid weak (vs. all these traits distinct; table 4). T. lojapiuranus also differs from T. cinereus in the morphology of glans penis by exhibiting a lateral notch, lack of dorsal groove, and spineless urethral flaps and dorsal papilla. Additionally, interdental ruga i4 extends to the palatal midline (vs. interdental ruga i4 shorter in T. cinereus).
Thomasomys lojapiuranus differs from T. shallqukucha by its long mystacial vibrissae that extend conspicuously behind the posterior margin of pinnae when laid back alongside the head (vs. mystacial vibrissae shorter, barely extending behind the posterior margin of the pinnae in T. shallqukucha); plantar surface of hind foot pale (vs. dark); zygomatic notch shallow (vs. very shallow); interparietal bone moderately long anteroposteriorly (vs. conspicuously long anteroposteriorly); incisive foramina long, extending posteriorly to the anterior margins of the first molars (vs. short, not approaching the anterior margins of the first molars); palatal median process usually present (vs. absent); Eustachian tube short and broad (vs. short and narrow); orbicular apophysis elongated without basal constriction (vs. elongated with basal constriction on anterior margin); M1 anteromedian flexus, M3 hypoflexus, and m1 anteromedian flexid weak (vs. all flexi distinct; table 4). Thomasomys lojapiuranus also differs from T. shallqukucha in the morphology of glans penis by presenting a lateral notch.
Thomasomys lojapiuranus differs from T. pagaibambensis by having a tail without a white tip (vs. tip usually white in T. pagaibambensis); plantar surface of hind foot pale (vs. slightly darker); zygomatic plates moderately sloped backward (vs. vertically oriented); zygomatic process of maxilla and zygomatic process of squamosal closely approximated (vs. processes distinctly separated by a narrow jugal); incisive foramina long, extending posteriorly to the anterior margins of the first molars (vs. short, not approaching the anterior margins of the first molars); palatal median process usually present (vs. usually absent); Eustachian tube short and broad (vs. short and narrow); orbicular apophysis elongated without basal constriction (vs. elongated with basal constriction and recurved); capsular process of lower incisor alveolus absent (vs. distinctly swollen; table 4). Thomasomys lojapiuranus also differs from T. pagaibambensis in penis morphology by exhibiting a lateral notch and lack of dorsal groove, and in stomach morphology by exhibiting a shallower incisura angularis and a bordering fold little recurved to the left.
Thomasomys lojapiuranus could also be confused with some species of similar size of the Cinereus Group from Ecuador, such as T. silvestris, T. caudivarius, and the recently described T. salazari. However, T. lojapiuranus has the tail proportionally shorter (Tail% = 117%), whereas T. silvestris, T. caudivarius, and T. salazari have the tail longer (Tail% >130%; Brito et al., 2019: table 2). Thomasomys lojapiuranus has a broad rostrum with anterior half of nasals expanded that clearly differs from the narrower rostrum and nasals of T. caudivarius, T. salazari, and T. silvestris (fig. 24; Brito et al., 2019: fig. 6, S2). Also, the cranium of T. lojapiuranus has a more rectangular dorsal profile versus a more triangular profile in T. caudivarius, T. salazari, and T. silvestris (fig. 12B; Brito et al., 2019: fig. 6, S2). Additional comparisons are presented in table 4 (this work) and Brito et al. (2019: table 2).
Natural History: This species inhabits the humid montane forests of northern Peru and southern Ecuador, which has also been called the Peruvian Yungas ecoregion of the Tropical and Subtropical Moist Broadleaf Forest biome (Dinerstein et al., 2017). At several collecting localities, the vegetation consisted of high shrubs and small trees, 10 to 15 m high.
Among specimens with reproductive data, 13 males were collected from July to August (usually considered the dry season); eight of them had scrotal testes (length >10 mm) and five had abdominal testes (length ≤9 mm). No data is available for the wet season. Among 24 female specimens collected in July and August, 18 specimens had closed vagina (75%); three females had embryos (one with a fetus on the left ovary; the second with two embryos, one on each side; and the third with three embryos, two on the left, one on the right side); and three females were lactating; suggesting that during the dry season most specimens of T. lojapiuranus are in nonreproductive condition. Six specimens collected in November and December had no embryos.
Stomach contents of four specimens contained insect remains and seeds. Some nematodes were also found in the antrum.
In Yacurí National Park, Ecuador, Thomasomys lojapiuranus is sympatric with T. taczanowskii and T. caudivarius (Lee et al., 2018).
Etymology: It is a pleasure to dedicate this species to the people of Loja and Piura departments, from Ecuador and Peru, respectively, who share similar costumes and traditions. We recognize in the distributional range of this mouse the fact that nature does not observe political boundaries.
Remarks: Brito et al.'s (2019: fig. 3) phylogenetic analysis of the Cinereus Group contains several problematic identifications. Among others, they identified several specimens that appear as terminals in a phylogenetic tree (QCAZ 15612–15630) as Thomasomys cinereus. The authors did not report the collection localities of those specimens, but according to Lee et al. (2015: fig. 1), these were collected in 2015 near Laguna Negra (4.71308°S, 79.43025°W; 3407 m) in Loja province, Ecuador. We found that the sequences of four of these specimens are not T. cinereus sensu stricto nor any of the new species reported here (table 1, fig. 4). Additionally, Brito et al. (2019: apéndice 1) identified several specimens (QCAZ 16230, 16231, 16328–16330) from Jimbura (4.71167°S, 79.4403°W; 3226 m) as T. cinereus or “Thomasomys sp. 2” (Brito et al., 2019: fig 3); based on our sequence analyses (fig. 4), we would identify these specimens as T. lojapiuranus. Lastly, we sequenced one specimen (MEPN 12549) from Espínola, Parque Nacional Yacuri, Loja (4.719803°S, 79.439031°W; 3274 m) that Brito et al. (2019) identified morphologically as T. cinereus and determined that it too is T. lojapiuranus (fig. 4).
Several specimens from Piura, Huancabamba, Canchaque, Tambo, at 1198 m (FMNH 83444–83454) were collected at an unusually low elevation for Thomasomys (Pacheco 2015). Unusually too, most of the habitat around Canchaque corresponds to dry forest (VP, personal obs.), which suggests that the elevation provided for those specimens is inaccurate. On the other hand, two specimens were collected nearby (15 road km E Canchaque; LSUMZ 19315, 19316) but at 1737 m, which suggests that the locality of the FMNH specimens is likely correct, but perhaps not the elevation. The latter locality (at 1737 m) would be the lowest elevation record for T. lojapiuranus if the elevation data of the FMNH specimens were not considered.
Thomasomys shallqukucha, new species
Shallqa Thomasomys
Figures 6C, 8, 9, 10C, 11C, 11G, 12C, 13C, 14, 15C, 16C, 17C, 18C, 19C, 19G, 20C, 22C, 25
Thomasomys cinereus, Pacheco 2015; part, not Thomas, 1882
Holotype: An adult male specimen in the Museo de Historia Natural de la Universidad Nacional Mayor de San Marcos (MUSM 46875). The holotype is a fluid-preserved specimen with skull removed, collected by Víctor Pacheco (field number VPT 4685) on 14 October 2016.
Paratypes: Nineteen paratypes from the same locality (MUSM 46870–46874, 46876–46889).
Type Locality: Peru, Department of Lambayeque, Province of Ferreñafe, District of Kañaris, Geomarca; 6.098573°S, 79.239928°W. 3266 m above sea level.
Diagnosis: A medium-sized Thomasomys (table 3) that can be distinguished from its congeners by the following combination of characters: pelage dark brownish; metatarsal patch pale brownish; tail long, unicolored, without a white tip; thenar and hypothenar pads separated; plantar surface of hind foot dark; nasals long, expanded anteriorly, and tapering posteriorly to lacrimal; zygomatic notches very shallow; interparietal bone conspicuously long anteroposteriorly; zygomatic plate moderately sloped backward; incisive foramina short, extending posteriorly close to the anterior margin of the first molar; palatal median process absent; Eustachian tube short and narrow; stapedial process of bulla reduced; bullae flask shaped and moderately inflated; capsular process of lower incisors alveolus absent; M1 anteromedian flexus distinct; M3 hypoflexus distinct; and m1 anteromedian flexid distinct.
Morphological Description: Thomasomys shallqukucha is a medium-size species (HBL = 116–142 mm) with long, soft, and dense dorsal fur (about 14 mm). Dorsal pelage is dark brown (Vandyke Brown, color 221; Sepia, color 219) with slate colored basally (Blackish Neutral Gray, color 82). Ventral pelage is Pearl Gray (color 81) or Pale Neutral Gray (color 86) with individual hairs slate colored basally, and moderately countershaded with dorsal pelage (figs. 8, 9). The ears are moderately long (EL = 19–26 mm) and dark brown. Postauricular patches are not conspicuous. The mystacial vibrissae are short, extending slightly behind the posterior margin of pinnae when laid back alongside the head. The supraorbital vibrissae are short, not extending to the posterior margins of pinnae when laid back alongside the head. The genal 1 vibrissae are absent. The tail is on average longer than the combined length of head and body (Tail% = 119), comparatively thick, unicolored without a white tip. The manus is pale brown with digit II larger than digit V. The hindfeet are moderately long (HFL = 30–33 mm) with metatarsals pale brownish. The plantar surface of the hind foot is dark. The thenar and hypothenar pads are separated by a conspicuous gap (fig. 10C). Digit I of the hind foot (hallux) is moderately long, its claw extending close to or to the first interphalangeal joint of dII. Digit V of hind foot is long, with claw extending about half the length of phalanx 2 of dIV.
Cranium: The skull is moderately large (CIL = 30.4–32.5 mm) with flat or barely convex profile without a bump on the frontal region. The rostrum is relatively long and moderately broad. A rostral tube is absent, and the gnathic process is small (figs. 12C, 15C). The nasals are long and tapering; the anterior half are expanded; the posterior margins are narrow, extending posteriorly to the lacrimals and beyond the premaxillae. The zygomatic notches are extremely shallow (fig. 13C). The interorbital region is moderately broad and hourglass shaped with rounded margins, and the frontal sinuses are relatively inflated. The braincase is wide and squarish. The lateral parietal processes are moderately large and subtriangular. The interparietal is wide and long anteroposteriorly, straplike in some specimens, but rhomboidal in others (fig. 14). The zygomatic arches converge anteriorly to a moderate degree or subparallel. The zygomatic plates are moderately broad, subequal in breadth to the length of M1, with the anterior margins slightly sloped backward (fig. 15C). The premaxillae are conspicuously produced anteriorly beyond the incisors (fig. 15C). The zygomatic processes of the maxilla and of the squamosal are closely approximated (fig. 16). The postglenoid foramen is small and placed anterior to a relatively larger subsquamosal fenestra. The tegmen tympani is robust and overlaps a distinct suspensory process of squamosal. The carotid circulatory pattern is pattern 1 (sensu Voss, 1988): a stapedial foramen, a groove on the inner surface of the squamosal and the alisphenoid, and a sphenofrontal foramen are all present. In the medial wall of the orbit, the ethmoid foramen is placed dorsal to M2; the ethmoturbinals are moderate in size; the sphenopalatine foramen is bordered by the maxillary, ethmoid, and palatine bones; and the optic foramen is moderate in size and slightly posterior to M3. The paraoccipital process is small. The incisive foramina are short and extend posteriorly to just in front of the anterior margins of the first molars, and are widest at the premaxillary-maxillary suture (fig. 17C). The palate is short and wide (sensu Hershkovitz, 1962). The mesopterygoid fossa is broad (in 68% of examined specimens), expanded anteriorly, with subparallel sides; a median posterior process of the palate is absent. The sphenopalatine vacuities are absent or produced as small slits along the basisphenoid and the presphenoid. The parapterygoid fossa is triangular, shallow, and perforated by a vacuity. The foramen ovale is small, the alisphenoid strut is present, and the middle lacerate foramen is relatively wide. The auditory bullae are flask shaped, moderately inflated, with short and narrow Eustachian tubes; the stapedial process of bulla is reduced. The internal carotid canal is bounded by the basioccipital and the ectotympanic (auditory bulla), but not the periotic. The basioccipital is moderately broad, and the mastoid (the occipital exposure of the periotic) is either imperforate or has only a small fenestra. The lamina of the malleus is squarish and not deep, and the orbicular apophysis is elongated, with a basal constriction on the anterior margin (fig. 18C). The processus brevis of the incus is narrow and delicate.
Teeth: The upper incisors are orthodont with orange enamel on their anterior surfaces. The upper molar rows are moderately long (5.1–5.4 mm). The upper molars are brachyodont and pentalophodont (sensu Hershkovitz, 1962) without interpenetration of the labial and lingual flexi; labial and lingual cingula are not developed; the accessory labial roots are absent. The procingulum of M1 is slightly narrower than the protocone-paracone cusp pair, and the anteromedian flexus conspicuously divides a slightly smaller anterolingual conule from a larger anterolabial conule. The anteroloph is conspicuous and the paraflexus is recurved. The paraloph is oriented transversally to the median mure or the base of the mesoloph. The mesoloph is well developed and reaches perpendicularly the labial margin of the tooth; the metaflexus is nearly straight. The posteroloph is absent. M2 exhibits a strong anteroloph with a recurved paraflexus; the mesoloph is mostly absent or very reduced, the metaflexus is comma shaped, and the posteroloph is coalesced with the metacone. M3 is smaller than M2; its paraflexus, mesoflexus, and hypoflexus are conspicuous; and its metacone is obsolete (fig. 19C).
On the lower molars, the main cuspids are conspicuous and slightly alternating. The first lower molar (m1) has a narrow anteroconid and a distinct anteromedian flexid; the protolophid and anterolophid are short or coalesced; the anterolabial and anterolingual cingula are poorly developed. The mesoflexid is narrow and curved; the entoflexid is small or coalesced; the posteroflexid is narrow and almost straight; the hypoflexid is broad and oriented perpendicularly; and the mesolophid is very short or coalesced with the entoconid. On m2, the mesolophid is usually absent or short; and the anterolabial cingulum is conspicuous. The m3 has a subtriangular occlusal shape produced by the less developed entoconid; the mesoflexid and hypoflexid are conspicuous; and the posteroflexid is enclosed in an island (fig. 19G).
Mandible: The coronoid process is long, slender, and falciform. The angular process is short and anterior than the condylar process. The capsular process of the lower incisor alveolus is absent (fig. 25).
Hyoid apparatus: Hyoid morphology like that described for Thomasomys cinereus (observation made in six specimens).
Penis: Three fluid-preserved male specimens (MUSM 46875, 46883, 46887) were examined. The glans is medium sized, subcylindrical, relatively long (4.4 mm), and narrow (2.2 mm) (fig. 20C). The external surface is covered by epidermal spines overall except along the crater rim, which is spineless; dorsal and ventral grooves are absent, and the ventral notch is shallow. The crater lip circumscribes the entire crater opening and is separated from the spinous epithelium by a thick fold. The medial bacular mound is large, conical, and projects prominently from the crater lip; the lateral bacular mounds are small, deep, and barely visible externally. The tip of the medial bacular mound is oriented vertically. The urethral flaps are small, conical, and deeply buried within the crater; their tips are well separated and lack spines or other ornamentation. The dorsal papilla is very small, deep, rounded, and lack spines or other ornamentation.
Palatal rugae: Unknown.
Stomach: The stomach corresponds to the unilocular-hemiglandular morphotype (Carleton, 1973; Pacheco, 2003). The incisura angularis is shallow, and the bordering fold is thick and crosses the lesser curvature of the stomach slightly anterior to the incisura angularis; the bordering fold also recurves moderately to the left but not beyond the esophageal orifice (fig. 22C).
Gall bladder: Present.
Reproductive glands: Unknown.
Skeleton: Unknown.
Karyotype: Unknown.
Measurements of Holotype: GSL, 34.07; CIL, 31.46; CML, 20.3; LOF, 11.09; LN, 13.04; LD, 9.62; RL, 12.65; LIF, 6.64; LM, 5.25; BIF, 2.43; BR, 5.15; BPB, 3.35; BM1, 1.77; BN, 4.69; LIB, 5.05; ZB, 17.26; BB, 13.82; BZP, 2.63; DI, 1.78; HBC, 8.96; MFB, 2.40. Measurements of additional specimens are provided in table 3.
Distribution: Thomasomys shallqukucha has a restricted distribution near the northeastern border of the department of Lambayeque, Peru. The Río Kañaryaku flows through this region to the Río Huancabamba, which ultimately joins the Río Marañón on the eastern versant of the Andes. All records are also west of the Río Chotano and south of the Huancabamba Depression (fig. 1). The altitudinal range is from 2550 m to 3330 m.
Taxonomic Comparisons: Thomasomys shallqukucha differs from T. cinereus by its dark brownish dorsal pelage (vs. grizzled ashy gray to dark gray in T. cinereus); metatarsals pale brownish (vs. whitish); tail longer than head and body length (Tail% = [119 ± 6], vs. [103 ± 8] %) and unicolored (vs. moderately bicolor); plantar surface of hind foot dark (vs. pale); thenar and hypothenar pads separated by a distinct gap (vs. thenar and hypothenar pads closely approximated); zygomatic notches very shallow (vs. deep [55%] or shallow); interparietal conspicuously long anteroposteriorly (vs. moderately long); zygomatic plates slightly sloped backward (vs. vertically oriented); incisive foramina short, not approaching the anterior margins of the first molars (vs. long, and usually extended posteriorly between the first molars); median palatal process absent (vs. usually present); bullae moderately inflated (vs. not inflated); stapedial process of bulla reduced (vs. conspicuous); orbicular apophysis elongated with basal constriction on anterior margin (vs. knob shaped; table 4). Also, the glans penis of T. shallqukucha lacks dorsal and ventral grooves, and the urethral flaps and dorsal papilla lack spines, whereas the glans penis of T. cinereus exhibits a conspicuous dorsal groove and a ventral shallow groove, and the urethral flaps and dorsal papilla exhibit some spines.
Thomasomys shallqukucha differs from T. pagaibambensis by its short mystacial vibrissae that barely extend behind the posterior margin of pinnae when laid back alongside the head (vs. longer vibrissae that extend conspicuously behind the posterior margin of pinnae); tail without a white tip (vs. tail usually white tipped); zygomatic notches very shallow (vs. shallow; fig. 13); zygomatic plates gently sloping backward (vs. vertically oriented); zygomatic process of maxilla and zygomatic process of squamosal closely approximated (vs. processes separated by a narrow jugal); orbicular apophysis elongated with basal constriction on anterior margin (vs. elongated with basal constriction and recurved); capsular process of lower incisor alveolus indistinct (vs. distinctly swollen); M3 hypoflexus distinct (vs. weak; table 4). Also, the stomach of T. shallqukucha has a shallower incisura angularis and the bordering fold is not markedly recurved to the left, whereas the stomach of T. pagaibambensis has a deeper incisura angularis and the bordering fold is distinctly recurved to the left (fig. 22); and the glans penis of T. shallqukucha lacks a dorsal groove and a distinct notch on the ventral wall, whereas the glans penis in T. pagaibambensis has a shallow dorsal groove and a conspicuous ventral notch.
Thomasomys shallqukucha is unlikely to be sympatric with T. silvestris, T. caudivarius, or T. salazari; nonetheless, T. shallqukucha could be readily differentiated from them by its broader rostrum and nasals, its extremely reduced zygomatic notch, and its short incisive foramina.
Natural History: This species inhabits the montane forests of northern Peru, which has also been called the Peruvian Yungas ecoregion of the Tropical and Subtropical Moist Broadleaf Forests biome (Dinerstein et al., 2017) (fig. 23B). The main tree components of the montane forests of Kañaris are members of Lauraceae (Persea, Ocotea, Nectandra), Cunoniaceae (Weinmannia), Podocarpaceae (Podocarpus), Cecropiaceae (Cecropia), Myrtaceae (Myrcianthes), Moraceae (Ficus), Rubiaceae (Cinchona), Bignoniaceae (Tabebuia), Arecaceae (Ceroxylon), and tree ferns of the genus Nephelea (Llatas-Quiroz and López-Mesones, 2005). In these forests, Thomasomys shallqukucha is sympatric with T. taczanowskii and T. cf. aureus.
Among specimens of Thomasomys shallqukucha with reproductive data, 10 males were collected in October (usually considered dry season); four had abdominal testes and six had scrotal testes; and one male specimen collected in November had scrotal testes. Of 10 females collected in October (dry season), three had closed vaginas, two had open vaginas, and five were lactating. Four female specimens collected in November had closed vaginas.
The stomach contents of two specimens contained insect remains and seeds. Some nematodes were also found in the antrum.
Etymology: The name shallqukucha is a compound of two words in the Lambayeque dialect of Quichua: shallqua means “Jalca”: a Paramo-like ecoregion, typical of northwestern Peru (Weigend 2002, 2004)—and ukucha means “rodent.” The community of Kañaris speaks the Lambayeque dialect of Quichua, and it is probably one of the few Quichuan-speaking communities in northern Peru. This species is dedicated to the region of Kañaris, its people, and their interest in maintaining their montane forests.
Thomasomys pagaibambensis, new species
Pagaibamba Thomasomys
Figures 6D, 8, 9, 10D, 11D, 11H, 12D, 13D, 15D, 16D, 17D, 18D, 19D, 19H, 20D, 22D, 26
Thomasomys cinereus, Pacheco 2015; part, not Thomas, 1882
Holotype: An adult female specimen in the Museo de Historia Natural de la Universidad Nacional Mayor de San Marcos (MUSM 40986). The holotype is a skin, skull, and carcass, all in good condition, collected by Edith Salas Perez (field number ESP 384) on 29 September 2007.
Paratypes: Twenty-three specimens from the same locality MUSM 40968–40985; 40987–40990.
Type locality: Peru, Department of Cajamarca, Province of Chota, District of Querocoto, Pagaibamba. 6.429585°S, 79.05792°W; 2902 m above sea level.
Diagnosis: A medium-sized Thomasomys (table 3) that differs from its congeners by the following combination of characters: dorsal pelage brownish; metatarsal patch pale brownish; tail long, unicolored, usually with a white tip; thenar and hypothenar pads separated; plantar surface of hind foot slightly dark; nasals long with anterior half expanded; zygomatic notches shallow; interparietal bone wide and long anteroposteriorly; zygomatic plate with anterior margin vertical; incisive foramina short, approaching posteriorly in front of the anterior margins of the first molars; palate median process absent or indistinct; zygomatic process of the maxilla distinctly separated from the zygomatic process of the squamosal by a narrow jugal; capsular process of lower incisors alveolus distinctly swollen; M1 anteromedian flexus distinct; M3 hypoflexus reduced, and m1 anteromedian flexid distinct.
Morphological Description: Thomasomys pagaibambensis is a medium-sized species (HBL = 121–158 mm) with long, soft, and dense dorsal fur (about 14 mm). Dorsal pelage is brownish (Clay Color, color 26 or Dark Drab, color 119B) with individual hairs slate colored basally. Ventral pelage is Pale Horn Color (color 92) or Pale Neutral Gray (color 86) with individual hairs slate-colored basally, and moderately countershaded with dorsal pelage (figs. 8, 9). The ears are moderately long (EL = 20–24 mm) and dark brown. Postauricular patches are not conspicuous. The mystacial vibrissae are long, extending behind the posterior margin of the pinnae when laid back alongside the head. The supraorbital vibrissae are short, extending barely to the posterior margin of the pinnae when laid back alongside the head. The genal 1 vibrissae are absent. The tail is longer than combined length of head and body (Tail% = 122), comparatively thick, unicolored, and usually with a white tip. The hindfeet are moderately long (HFL = 28–34 mm) with metatarsals pale brownish. The manus are also pale brown. The plantar surface of hind foot is slightly dark. The thenar and hypothenar pads are widely separated (fig. 10D). Digit I of the hind foot (hallux) is moderately long, its claw extending close or to the first interphalangeal joint of dII. Digit V of hind foot is long, with claw extending about half the length of phalanx 2 of dIV.
Cranium: The skull is moderately long (CIL = 29.8–33.3 mm) with a flat or barely convex profile and a bump on the frontal region. The rostrum is relatively long and moderately broad (fig. 11D). A rostral tube is absent, and the gnathic process is small (figs. 12D, 15D). The nasals are long and tapering; the anterior half are expanded; the posterior margins are narrow, extending to the lacrimals and beyond the premaxillae. The zygomatic notches are shallow (fig. 13D). The interorbital region is moderately broad and hourglass shaped with rounded margins, and the frontal sinus are relatively inflated. The braincase is wide and oval shape. The lateral parietal processes are moderately large and subtriangular. The interparietal is wide and long anteroposteriorly, straplike, or rhomboidal. The zygomatic arches converge anteriorly to a moderate degree. The zygomatic plates are moderately broad, in breadth longer than the length of M1, and vertically oriented (fig. 15D). The premaxillae are conspicuously produced anteriorly beyond the incisors (fig. 15D). The zygomatic process of the maxilla is distinctly separated from the zygomatic process of the squamosal by a long and narrow jugal (fig. 16). The postglenoid foramen is small and placed anterior to a relatively larger subsquamosal fenestra. The tegmen tympani is robust and overlaps a distinct suspensory process of squamosal. The carotid circulatory pattern is pattern 1 (sensu Voss, 1988): a stapedial foramen, a groove on the inner surface of the squamosal and the alisphenoid, and a sphenofrontal foramen all present. In the medial wall of the orbit, the ethmoid foramen is placed dorsal to M2; the ethmoturbinals are relatively small; the sphenopalatine foramen is bordered by the maxillary, ethmoid, and palatine bones; the optic foramen is moderate in size and is slightly posterior to M3. The paraoccipital process is larger and curved. The incisive foramina are usually short (in 84% of examined specimens) and extend posteriorly in front of the anterior margins of the first molars, and slightly broad with the anterior and posterior margins narrow (fig. 17D). The palate is short and wide (sensu Hershkovitz, 1962). The mesopterygoid fossa is broad, lyre shaped; a median posterior process of the palate is absent (in 77% of examined specimens) or poorly developed (23%). The sphenopalatine vacuities are absent or are represented by narrow slits along the basisphenoid and presphenoid bones. The parapterygoid fossa is triangular and shallow. The foramen ovale is of moderate size, the alisphenoid strut is present, and the middle lacerate foramen is narrow but conspicuous. The auditory bullae are flask shaped, moderately inflated, with short and narrow Eustachian tubes; the stapedial process of bulla is reduced. The internal carotid canal is bounded by the basioccipital and the ectotympanic (auditory bulla), but not by the periotic. The basioccipital is moderately broad, and the mastoid (the occipital exposure of the periotic) is either imperforate or has only a small fenestra. The lamina of the malleus is squarish and not deep; the orbicular apophysis is elongated and recurved, with a basal constriction (fig. 18D). The processus brevis of incus is narrow and delicate.
Teeth: The upper incisors are orthodont with orange enamel on their anterior surfaces. The upper molar rows are moderately long (5–5.8 mm). The upper molars are brachyodont and pentalophodont (sensu Hershkovitz, 1962) without interpenetration of the labial and lingual flexi; labial and lingual cingula are not developed; and the accessory labial roots are absent. The procingulum of M1 is slightly narrower than the protocone-paracone cusp pair, and the anteromedian flexus is distinct. The anteroloph is conspicuous but usually coalesced with the anterocone and the paraflexus is recurved. The paraloph is oriented transversally to the median mure, and sometimes to the base of the mesoloph. The mesoloph is well developed and reaches perpendicularly the labial margin of the tooth; the metaflexus is slightly curved; and the posteroloph is absent. M2 exhibits a strong anteroloph with a recurved paraflexus; the mesoloph is narrow, complete, and reaches the labial margin of the tooth; the metaflexus is comma-shaped, and the posteroloph is coalesced with the metacone. M3 is much smaller than M2, only its paraflexus is distinct, and its hypoflexus is mostly absent or reduced (fig. 19D).
On the lower molars, the main cuspids are conspicuous and slightly alternating. The first lower molar (m1) has a distinct anteromedian flexid; the protolophid and the anterolophid are short or coalesced; the anterolabial and anterolingual cingula are poorly developed; the mesoflexid is narrow and curved; the entoflexid is coalesced; the posteroflexid is narrow and slightly curved; the hypoflexid is broad, deep, and oriented slightly transversally; and the mesolophid is narrow and complete or coalesced with the entoconid in adult or older specimens. On m2, the mesolophid is short and the anterolabial cingulum is small. On m3, the mesoflexid and hypoflexid are conspicuous; but its posteroflexid is reduced or absent (fig. 19H).
Mandible: The coronoid process is long, slender, and falciform. The angular process is short and anterior than the condylar process. The capsular process of the lower incisor alveolus is distinctly swollen (fig. 26).
Hyoid apparatus: Hyoid morphology like that described for Thomasomys cinereus (observation made in six specimens).
Penis: Three fluid-preserved male specimens (MUSM 40965, 40998, 41004) were examined. The glans is subcylindrical, comparatively short (4 mm), and wide (2.34 mm) (fig. 20D). The external surface is covered by epidermal spines overall except along the crater rim, which is spineless. A dorsal shallow groove and a distinct ventral notch are present. Laterally, the glans is undivided. The crater lip circumscribes the entire crater opening and is separated from the spinous epithelium by a thick fold. The medial bacular mound is large and conical, projecting beyond the crater lip; the lateral bacular mounds are small and barely project beyond the crater lip. The tip of the medial bacular mound is oriented vertically. The urethral flaps are small, conical, and deeply buried within the crater; their tips are well separated and lack ornamentation. The dorsal papilla is large, conical, and lack spines or other ornamentation.
Palatal rugae: Unknown.
Stomach: The stomach corresponds to the unilocular-hemiglandular morphotype (Carleton, 1973; Pacheco, 2003). The incisura angularis is comparatively deeper, and the bordering fold is also comparatively thicker and crosses the lesser curvature of the stomach slightly anterior to the incisura angularis; the bordering fold recurves to the left slightly beyond the esophageal orifice (fig. 22D).
Gall bladder: Present.
Reproductive glands: Macroscopic preputial glands are absent.
Skeleton: The vertebral count is 7 cervical, 13 thoracic, 6 lumbar, 4 sacral, and 41 caudal vertebrae; and the same skeletons each have 13 pairs of ribs. The supratrochlear fenestra of humerus is usually absent (5 of 6 examined specimens). The first caudal vertebrae have chevron bones unfused (in 5 of 6 examined specimens). On the calcaneus, the trochlear process levels its posterior articular facet. The peroneal process of the fifth metatarsal is moderately long but does not extend to the proximal edge of the cuboid.
Karyotype: Unknown.
Measurements of Holotype: GSL, 34.52; CIL, 31.78; CML, 20.88; LOF, 10.89; LN, 13.56; RL, 12.44; LD, 9.2; LIF, 6.31; LM, 5.7; BIF, 2.56; BR, 5.71; BPB, 3.57; BM1, 1.71; BN, 4.08; LIB, 5.45; ZB, 18.23; BB, 14.1; BZP, 2.63; DI, 1.76; HBC, 9.44; MFB, 2.77. Measurements of additional specimens are provided in table 3.
Distribution: Thomasomys pagaibambensis is distributed in the mountains of the Pagaibamba Protection Forests, province of Chota, department of Cajamarca, Peru. All records are also south of the Huancabamba Depression, west of Río Chotano, and north of Río Chancay (fig. 1). The elevation range is from 2530 m to 3370 m.
Taxonomic Comparisons: Thomasomys pagaibambensis differs from T. cinereus by its brownish dorsal pelage (vs. grizzled ashy gray to dark gray in T. cinereus); mystacial vibrissae long, extending posteriorly behind margin of pinnae when laid back alongside the head (vs. shorter in T. cinereus, table 4); metatarsal pale brown (vs. whitish); tail longer than head-and-body length (Tail% = [122 ± 8], vs. [103 ± 8] %), unicolored (vs. moderately bicolor), and tip usually white (vs. a tail without a white tip); plantar surface of hind foot slightly dark (vs. pale); thenar and hypothenar pads separated by a distinct gap (vs. thenar and hypothenar pads closely approximated); zygomatic notch shallow (vs. deep [55%] or shallow in T. cinereus); interorbital region moderately broad (vs. narrow); frontal sinus relatively inflated (vs. not inflated); interparietal conspicuously long anteroposteriorly (vs. moderately long, fig. 14); the zygomatic process of the maxilla and the zygomatic process of the squamosal are separated by a narrow jugal bone (vs. both processes closely approximated); incisive foramina short, not approaching the anterior margins of the first molars (vs. usually extending between the first upper molars in T. cinereus); palatal median process usually absent (vs. present); bullae moderately inflated (vs. small, not inflated); stapedial process of bulla reduced (vs. distinct); orbicular apophysis elongated with basal constriction and recurved (vs. knob shaped); capsular process distinctly swollen (vs. absent); M3 hypoflexus reduced (vs. distinct in T. cinereus). Also, the stomach of T. pagaibambensis has a deeper incisura angularis and a thicker bordering fold more recurved to the left, whereas the stomach of T. cinereus has a shallower incisura angularis and a thinner bordering fold not recurved to the left (fig. 22); and the glans penis in T. pagaibambensis is comparatively smaller with a conspicuous ventral notch and a shallow dorsal groove, whereas the glans penis in T. cinereus exhibits a shallow ventral notch and a conspicuous dorsal groove.
Thomasomys pagaibambensis is unlikely to be sympatric with T. silvestris, T. caudivarius, or T. salazari; nonetheless, T. pagaibambensis could be readily differentiated from them by its broader rostrum and nasals, its larger interparietal, and its short incisive foramina.
Natural History: This species inhabits the montane forests of the Pagaibamba Cordillera, which is part of the Peruvian Yungas ecoregion of the Tropical and Subtropical Moist Broadleaf Forests biome (Dinerstein et al., 2017) and the Jalca ecoregion, above 3200 m, which has also been called the Cordillera Central Páramo (Dinerstein et al., 2017). In the Jalca, there is a predominance of grasses of the family Poaceae (Stipa, Festuca, and Calamagrostis), and trees of the families Asteraceae (Werneria) and Rosaceae (Polylepis) occur on rocky slopes (Roncal-Rabanal, 2018). Extensive data on the flora and fauna of Bosque de Protección Pagaibamba was recently summarized by Roncal-Rabanal (2018).
In the montane forests of Pagaibamba Protected forests, Thomasomys pagaibambensis is sympatric with T. taczanowskii, T. pyrrhonotus, and T. cf. aureus; and the thomasomyine Chilomys instans (Medina et al., 2016). Medium and large mammals (i.e., Cuniculus taczanowskii, Dinomys branickii, Sylvilagus andinus, Leopardus pardalis, Puma concolor, Lycalopex culpaeus, Tremarctos ornatus, Conepatus chinga, Eira barbara, Mustela frenala, Tapirus pinchaque, Mazama sp., Odocoileus virginianus) have also been reported (Jiménez et al., 2010; Diaz and Pacheco, 2022).
Among specimens with reproductive data, seven males were collected in September and October (usually considered dry season); all had abdominal testes (length ≤10 mm). Among specimens collected in May and November (wet season), nine had scrotal testes and two had abdominal testes. Of 17 females collected in September and October (dry season), 13 had closed vaginas, and four had open vaginas. Of 17 specimens collected in April, May, and November (wet season), 10 had closed vaginas, six had open vaginas, and one was pregnant.
The stomach contents of four specimens contained some insect remains and seeds. One of them was full of seeds of Greigia sp. (Bromeliaceae) and large unidentified nematodes in the antrum.
Etymology: The species epithet refers to the type locality, which is located in the Pagaibamba Protected Forest in the Cordillera Pagaibamba, one of the few relict forests in the department of Cajamarca.
DISCUSSION
This study is based on the largest series of morphological specimens closely related to Thomasomys cinereus and the largest mitochondrial DNA dataset ever assembled for the species in its previously understood sense. Our phylogenetic analyses of cytochrome b sequence data together with our morphological comparisons of voucher material suggest that T. cinereus sensu lato (Pacheco, 2015) is a species complex composed of at least four distinct lineages, herein referred to T. cinereus sensu stricto and three new species: T. lojapiuranus from the montane forests of the Peruvian department of Piura and the Ecuadorean province of Loja; T. shallqukucha from the mountains of the department of Lambayeque (Peru); and T. pagaibambensis from the Cordillera Pagaibamba, department of Cajamarca (Peru). Of these taxa, T. lojapiuranus is distributed north of the Huancabamba Depression—long recognized as a biogeographically important feature of the tropical Andes (e.g., by Parker et al., 1985)—whereas T. cinereus, T. shallqukucha, and T. pagaibambensis are distributed south of this landscape feature.
In addition, our phylogenetic analyses confirm the monophyly of the Cinereus Group as defined by Pacheco (2015) and as emended in the Introduction to this report. Although weakly supported, the Cinereus Group now consists of at least 18 currently recognized species (T. australis, T. bombycinus, T. caudivarius, T. cinereus, T. cinnameus, T. daphne, T. erro, T. fumeus, T. hudsoni, T. lojapiuranus, T. onkiro, T. pagaibambensis, T. paramorum, T. salazari, T. shallqukucha, T. silvestris, T. ucucha, T. vulcani) and two candidate species from Ecuador (Thomasomys sp. 1 and Thomasomys sp. 2). Future revisionary research will doubtless increase this tally. In particular, our species delimitation results suggest that T. daphne includes two candidate taxa as previously suggested by Pacheco (2015: 642). As discussed above (in Remarks in the account for T. lojapiuranus), our results corrected identifications for several terminal taxa previously sequenced by Lee et al. (2018) and Brito et al. (2019) and support the hypothesis that their “Thomasomys sp. 1” (herein also Thomasomys sp. 1) is another candidate species. However, their “T. cinereus” (here Thomasomys sp. 2) appears to be more closely related to T. salazari and T. caudivarius than to T. cinereus sensu stricto and likely also represents an undescribed taxon. Therefore, the Cinereus Group potentially comprises at least 21 species based on these molecular results.
Brito et al. (2019) recognized four clades (labelled A–D) within the Cinereus Group, of which clade B consisted only of T. erro. Of their three multispecies lineages, clades A and C were strongly supported in our analyses, whereas clade D was only weakly supported, probably due to denser taxon sampling in this report. To be more conservative, we opt to recognize only two strongly supported multispecies clades within the Cinereus Group (fig. 4). Our clade A (which corresponds to the clade A of Brito et al. [2019]) includes Thomasomys fumeus, T. bombycinus, T. vulcani, T. australis, T. daphne, and Thomasomys sp. 3 (sensu Pacheco, 2003), whereas our clade B includes T. caudivarius, T. cinereus, T. cinnameus, T. hudsoni, T. lojapiuranus, T. pagaibambensis, T. paramorum, T. salazari, T. shallqukucha, T. silvestris, T. ucucha, and two candidate species from Ecuador (Thomasomys sp. 1 and Thomasomys sp. 2). Only two species are orphaned by this scheme—T. erro and T. onkiro—because their relationships remain inadequately supported in our analyses.
In spite of our improved understanding of the phylogenetic relationships within the Cinereus Group, these results remain preliminary and incomplete. Sequence data from additional genes are clearly needed, and several species of the Cinereus Group from Colombia and Venezuela are still unrepresented by genetic data (i.e., T. cinereiventer, T. contradictus, T. dispar, T. emeritus, T. hylophilus, T. laniger, T. monochromos, T. niveipes, and T. vestitus). Pacheco (2015) included 23 species in his Cinereus group, which now additionally includes T. salazari, T. lojapiuranus, T. pagaibambensis, T. shallqukucha, and the three candidate unnamed species listed in table 1. With a potential total of 30 taxa, the Cinereus Group is by far the largest of the seven species groups proposed by Pacheco (2015) within Thomasomys.
Among other remaining phylogenetic issues, evidence for the monophyly of Thomasomys and for the interrelationships among currently recognized species groups remains unconvincing. Several phylogenies that have recovered a monophyletic Thomasomys with different genetic markers (cytb, IRBP, BRCA1, GRH, RAG1, Acp5; Schenk et al., 2013; Steppan and Schenk, 2017; Parada et al., 2013, 2015; Lee et al., 2011, 2015, 2018; Brito et al., 2019, 2021; Ruelas and Pacheco, 2021a) have sampled only a limited number of species, and none has included representatives of all species groups. About half the currently recognized species have yet to be included in any phylogenetic analysis, so Pacheco (2003)'s hypothesis of Thomasomys as a polyphyletic genus diagnosed by morphological characters still awaits a formal molecular-phylogenetic evaluation. Similarly, the monophyly of Pacheco's (2003, 2015) seven species groups remains untested with molecular data. Although the results of several molecular analyses partially support Pacheco's species-group classification (see Salazar-Bravo and Yates, 2007; Lee et al., 2018; Brito et al., 2019), other results are incongruent.1 Because this is not the place to enumerate such inconsistent analytic results nor to explain the real or assumed misidentifications on which some of them are based, we reserve such topics for future reports.
Biogeographic Implications
The Huancabamba Depression (also called the Huancabamba Deflection) transects the departments of Piura, Lambayeque, and Cajamarca in northern Peru. In this region, the Andes reach their lowest elevation point at the Abra de Porculla (2145 m), and they make a rather abrupt change in direction from NW/SE to SW/NE (Gansser, 1973; Duellman, 1979; Jaillard et al., 1990). The Huancabamba Depression has been proposed as a barrier between northern and central Andean faunas because it is thought to limit the dispersion of species living at high elevation (Parker et al., 1985; Patterson et al., 1992; Pacheco and Patterson, 1992; Albuja and Patterson, 1996; Vivar et al., 1997; Duellman and Pramuk, 1999; Pacheco, 2002; Lunde and Pacheco, 2003). The habitat in the Huancabamba Depression is mostly open shrubby vegetation or dry forests (Linares-Palomino, 2004) that have a mammalian fauna resembling those of adjacent lowland forests (Ruelas and Pacheco, 2021b), Such habitats are not inhabited by species of Thomasomys (Pacheco, 2015) and—consistent with its hypothesized role as a dispersal barrier—the Huancabamba Depression appears to separate the distribution of T. lojapiuranus in the north from the distributions of the other three species in this complex (T. cinereus, T. shallqukucha, and T. pagaibambensis), which are all found to the south. Of the latter, T. shallqukucha and T. pagaibambensis are both distributed in the Cordillera Occidental, and no major river has previously been suggested as a potential barrier between the small ranges that they occupy. However, the Río La Leche (which originates in the Cordillera Occidental and flows westward to the Pacific Ocean; fig. 1) might be such a barrier. Although the montane forests of the Kañaris and Pagaibamba ranges are apparently continuous, the data compiled and analyzed in this report suggest that T. shallqukucha and T. pagaibambensis are independently evolving lineages evolving despite their geographic proximity (about 35 km separates their known distributional ranges). Similarly, the distributions of T. pagaibambensis and T. cinereus are separated by a map distance of only about 45 km, with no known intervening barrier. However, the Río Maichil in Cajamarca department (a tributary of the Río Chancay that runs westward to the Pacific Ocean) might be acting as a barrier, as might the dry-forested valley of the Río Chotano, the floor of which is mostly below 2000 m. To our knowledge, these minor rivers (La Leche, Maichil, Chotano) and the valleys they occupy have never previously been suggested as zoogeographic barriers, and additional collecting is needed to test their local importance as drivers of biodiversity south of the Huancabamba Depression.
Although we have not explicitly tested the hypothesis of isolation by distance among taxa in the Thomasomys cinereus complex, the FST values we computed among them are similar and appear uncorrelated with geographic proximity. For example, the FST value between T. lojapiuranus and T. pagaibambensis on the one hand and between T. pagaibambensis and T. cinereus on the other are the same (0.87) in spite of the fact that the geographic distributions of T. lojapiuranus and T. pagaibambensis are separated by about 390 km, whereas the distributions of T. pagaibambensis and T. cinereus are separated by about 50 km. Also, the highest fixation index we computed (FST = 0.96) was that between T. shallqukucha and T. pagaibambensis, which are geographically closer to one another than to either of the other species.
Although its geographic range is smaller now than was previously thought, Thomasomys cinereus is still widely distributed: from the Río Marañón westward to the Pacific versant of the Cordillera Occidental, a large extent of montane forest, jalca, and puna. The northern and southern limits of this distribution are still not precisely known, although they might correspond to the Río Chancay on the north, and Río Tablachaca and Río Crisnejas on the south. Morphological and genetic variability within T. cinereus sensu stricto appears to be geographically structured, and the average intraspecific genetic distance we computed for this species (1.39%) is substantially higher than the corresponding values for T. lojapiuranus, T. shallqukucha, and T. pagaibambensis (0.22%–0.39%). At present, montane forests within the geographic range of T. cinereus are highly fragmented, probably due to human interventions that date to pre-Hispanic times (Sylvester et al., 2017), but also by naturally occurring nonforest (puna or jalca) vegetation. Because our collection data suggest that T. cinereus is closely associated with montane forest, its current distribution suggests that either most of this region was once covered by montane forests that have since disappeared by natural or anthropogenic causes or that the species actually survives at low densities in open habitats.
Thomasomys cinereus is not known to occur east of the Río Marañón despite frequent and extensive collecting in the eastern cordillera dating back to the late 19th century (Thomas, 1894, 1926a, 1926b, 1926c, 1927a, 1927b; Osgood, 1914; Thomas and Leger, 1926; Gardner and Romo R., 1993; Leo L. and Gardner, 1993; Ruelas et al., 2021), so its absence in the eastern cordillera is unlikely to be a sampling artifact. Pacheco (2015) suggested that the Río Marañón is a formidable dispersal barrier for species of Thomasomys, as it apparently is also for species of Akodon (Jiménez et al., 2013), Neacomys (Hurtado and Pacheco 2017), Phyllotis (Pacheco et al., 2014; Rengifo and Pacheco, 2015), and Nephelomys (Ruelas et al., 2021).
Taking a larger biogeographic perspective, our results suggest that species richness of the Cinereus Group is higher in the Northern Andes (north of the Huancabamba Depression; i.e., from Venezuela to northern Peru) than it is in the Central Andes (south of the Huancabama Depression), where only a few species are known to occur (i.e., T. daphne, T. australis, and T. onkiro), all to the east of the Río Marañón (Pacheco, 2015). This biogeographic pattern is consistent with Voss's (2003) suggestion that Ecuador is a center of endemism for Thomasomys. Previous studies (Brito et al., 2019, 2021; Ruelas and Pacheco, 2021a) also support this pattern—at least for the Cinereus and Baeops groups—but we should emphasize that other congeneric clades (including the Notatus, Incanus, Macrotis, and Oreas groups) are endemic to the montane forests of the Central Andes east of the Río Marañón (Pacheco, 2015).
Ongoing taxonomic research will doubtless refine our biogeographic knowledge of Thomasomys. Among other currently deficient data, several additional species remain to be described (Pacheco (2003, 2015; Lee et al., 2018; Brito et al., 2019; Brito et al, 2021; Ruelas and Pacheco, 2021a), mostly in the Aureus and Cinereus groups. The discovery of new species is not unexpected because analyses of molecular sequence data have revealed the existence of cryptic species in several other sigmodontine genera, including Neacomys (Sánchez-Vendizú et al., 2018; Semedo et al., 2020), Nephelomys (Ruelas et al., 2021), and Rhipidomys (Brito et al., 2017). This taxonomic research and accumulating new records of Thomasomys (e.g., Moreno-Cárdenas and Novillo-Gonzalez, 2020; Pacheco, 2021) are improving our knowledge of the diversity and distribution of the genus, but we are still far from a comprehensive analysis of diversification and biogeography of these quintessentially montane-forest rodents.
ACKNOWLEDGMENTS
We thank R. Cadenillas, K. Cervantes, L. Huamani, C. Jiménez, E. Rengifo, M. Peralta, E. Salas, A. Solis, C. Tello, L. Luna-Wong, S. Velazco, and E. Vivar, who have collected with us or independently several specimens of Thomasomys cinereus sensu lato for the MUSM collection employed in this work. J.L Patton, M.D. Carleton, D. Lunde, R.S. Voss, B.D. Patterson, P. Myers, M.S. Hafner, L. Albuja, S. Burneo, and P. Jenkins graciously allowed access to specimens under their care and/or facilitated the visit to their institution (to V.P.). Special thanks to Lucìa Luna-Wong, who kindly provided the photographs of the holotype of Thomasomys cinereus, and to Mario López Mesones for sharing his deep knowledge of the Lambayeque region. Klauss Cervantes identified seeds found in stomach contents and shared his expertise on habitat description. Gerardo Lamas helped resolve issues of taxonomic nomenclature. Financial support was provided by INNOVATE (formerly FINCyT) with Contract No. 402-PNICP-PIBA-2014 (to VP), which funded several expeditions to northern Peru and implemented our Laboratory of Evolutionary Biology in the Mammalogy Department of the MUSM. Additional support came from the Vicerrectorado de Investigación y Posgrado de la Universidad Nacional Mayor de San Marcos (UNMSM) with the projects B1710043e, B18104092e, B19100991, and B20100321 (to the Study Group Diversidad de Mamíferos y sus Parásitos [DIMAPA] lead by VP). We thank Guillermo D'Elía for his help with molecular analyses at the beginning of the project, and Pablo Moreno for kindly sharing some samples from Ecuador. Silvia Diaz efficiently helped to handle specimens and the database, and skillfully improved and composed the figures.
Copyright © American Museum of Natural History 2023
REFERENCES
Appendices
APPENDIX 1
Selected Collecting Localities
List of examined specimens of Thomasomys cinereus, T. lojapiuranus, T. pagaibambensis, and T. shallqukucha with their respective localities.
Names of the largest administrative unit (department, state, etc.) within each country are italicized and geographic coordinates are provided.
Thomasomys cinereus
PERU
Cajamarca, Cajamarca, Encañada (locality 40: 6.944976°S, 78.362122°W; MUSM 42453, 42454).
Cajamarca, Cajamarca, Páramo de Cajamarca (not located; BNMH 0.3.15.4, 0.3.15.5).
Cajamarca, Cajamarca, Quengorrio (locality 37: 6.8446100°S, 78.4318300°W; MUSM 38350).
Cajamarca, Cajamarca, San Juan de Yerbabuena (locality 38: 6.9766300°S, 78.3799800°W; MUSM 38351, 38352).
Cajamarca, Celendín, CCPP La Florida (locality 47: 7.041686°S, 78.137400°W; MUSM 38005).
Cajamarca, Celendín, CCPP Lalucma (locality 46: 6.923999°S, 78.244523°W; MUSM 38006).
Cajamarca, Celendín, CCPP Sorochuco (locality 45: 6.893929°S, 78.244760°W; MUSM 38007).
Cajamarca, Celendín, Coñicorgue (locality 41: 6.8330600°S, 78.3149600°W; MUSM 38349, 38493–38495).
Cajamarca, Celendín, El Punre (locality 44: 6.994457°S, 78.279074°W; MUSM 39369–39373).
Cajamarca, Celendín, Hacienda El Punre (locality 43: 7.007557°S, 78.288909°W; MUSM 38008).
Cajamarca, Celendín, Laguna Milpo (locality 42: 6.975250°S, 78.307950°W; MUSM 34949).
Cajamarca, Celendín, Porvenir (locality 39: 6.977116°S, 78.379967°W; MUSM 38496, 38497).
Cajamarca, Contumaza, Bosque Cachil (locality 48: 7.395183°S, 78.780277°W; MUSM 51108–51113, 51115–51117, 52399–52406).
Cajamarca, Cutervo, Cutervo (locality 24: 6.378980°S, 78.817920°W; BMNH 81.9.7.29).
Cajamarca, Cutervo, Parque Nacional Cutervo, 100 m over Tragadero (locality 25: 6.249972°S, 78.766528°W; MUSM 46716).
Cajamarca, San Miguel, Seques (locality 26: 6.900000°S, 79.300000°W; AMNH 73127, 73129).
Cajamarca, San Miguel, Taulis (locality 27: 6.833300°S, 79.166670°W; AMNH 73136, 73147).
Cajamarca, Santa Cruz, 2 Km E Monteseco (locality 28: 6.857110°S, 79.098200°W; MUSM uncatalogued [VPT 1629, VPT 1640, VPT 1650).
Cajamarca, Santa Cruz, 3.81 km NE from Monteseco (locality 29: 6.8437300°S, 79.0812540°W; MUSM 46756–46759; UMMZ 178461, 178463, 178467, 178482, 178488).
Cajamarca, Santa Cruz, 35 mi NW Cajamarca (locality 30: 6.971760°S, 78.987800°W; MVZ 137937–137939).
Cajamarca, Santa Cruz, Alcaparrosa, T 28 (locality 34: 6.786230°S, 78.890020°W; MUSM 25704).
Cajamarca, Santa Cruz, Bancuyoc, T 07 (locality 32: 6.823410°S, 78.903620°W; MUSM 25705).
Cajamarca, Santa Cruz, Bosque San Pedro (locality 33: 6.806293°S, 78.896919°W; MUSM 46909).
Cajamarca, Santa Cruz, La Zanja y La Redonda (locality 35: -6.821666°S, -78.88166°W; MUSM 23346).
Cajamarca, Santa Cruz, Pampa Verde (locality 31: 6.814830°S, 78.912050°W; MUSM 25706–25708).
Cajamarca, Santa Cruz, Pisit T 25 (locality 36: 6.811390°S, 78.870520°W; MUSM 25709).
La Libertad, Daniel A. Carrión, La Arena (locality 51: 7.88955°S, 78.13786°W; MUSM 24136–24142, 24166–24172).
La Libertad, Daniel A. Carrión, South of Huamachuco (locality 50: 7.814830°S, 78.050030°W; BMNH 0.6.6.12).
La Libertad, Gran Chimú, La Clambay (locality 49: 7.622123°S, 78.435071°W; MUSM 52972, 52973).
La Libertad, Santiago de Chuco, Cachicadán (locality 55: 8.0646300°S, 78.1732100°W; MUSM 17285–17299, 17302).
La Libertad, Santiago de Chuco, El Fundo Río Chuyuhual (locality 54: 7.933627°S, 78.220142°W; MUSM 41102, 41103, 42008–42012).
La Libertad, Santiago de Chuco, Lagunas del Norte, río Chuyuhual (locality 53: 7.9304800°S, 78.2180200°W; MUSM 24928).
La Libertad, Santiago de Chuco, Lagunas Norte, río Perejil, San Pedro (locality 52: 7.9262000°S, 78.2790800°W; MUSM 24774).
PERU
Piura, Ayabaca, Aypate (4.70756179°S, 79.57981948°W; (locality 1: MUSM uncatalogued: ASA06, ASA11, ASA12, ASA16, ASA23, ASA27; MUSM 38501).
Piura, Ayabaca, Bosque de Huamba, 44 km E Ayabaca (locality 3: 4.718720°S, 79.530850°W; MUSM 511).
Piura, Huancabamba, Pariamarca Alto (locality 6: 5.158670°S, 79.549010°W; MUSM 23758–23762).
Piura, Huancabamba, 15 road Km E Canchaque (locality 12: 5.400000°S, 79.464660°W; LSUMZ 19315, 19316).
Piura, Huancabamba, 33 road Km SW Huancabamba (locality 13: 5.455450°S, 79.662160°W; LSUMZ 19301–19314, 19707, 20313).
Piura, Huancabamba, Cerro Chinguela, ca. 5 km NE Zapalache (locality 5: 5.116666°S, 79.383333°W; LSUMZ 27147).
Piura, Huancabamba, Cruz Blanca, ca. 33 rd. Km SW Huancabamba (locality 7: 5.333333°S, 79.533333°W; LSUMZ 27045–27081, 27130–27146).
Piura, Huancabamba, Tambo (locality 9: 5.350000°S, 79.550000°W; FMNH 81297–81299, 81301–81306, 81308–81312, 81325–81330, 81334, 81336).
Piura, Huancabamba, Canchaque (locality 11: 5.400000°S, 79.600000°W; FMNH 83444–83454).
Piura, Huancabamba, El Tambo (locality 10: 5.3594000°S, 79.5511000°W; USNM 304535–304537).
Piura, Huancabamba, Sondorillo, Ulpamache (locality 8: 5.338900°S, 79.491400°W; USNM 551643).
Thomasomys pagaibambensis
PERU
Cajamarca, Chota, El Mirador, Agua de Montaña (locality 18: 6.380954°S, 79.126427°W; MUSM 40961–40967).
Cajamarca, Chota, La Granja, Qda. Agua de la Montaña (locality 17: 6.359960°S, 79.142670°W; MUSM 40357, 40358, 41591).
Cajamarca, Chota, Pagaibamba (locality 22: 6.432256°S, 79.070808°W; MUSM 40968–40990).
Cajamarca, Chota, Pagaibamba Alto (locality 23: 6.433360°S, 79.072210°W; MUSM 39524–39527).
Cajamarca, Chota, Pagaibamba Bajo (locality 20: 6.419030°S, 79.058920°W; MUSM 39528–39532).
Cajamarca, Chota, Pagaibamba Medio (locality 21: 6.427160°S, 79.065380°W; MUSM 39533–39536, 40991–40995).
Cajamarca, Chota, Paja Blanca (locality 19: 6.389970°S, 79.126910°W; MUSM 39537–39556, 40996–41011).
APPENDIX 3
ANOVA Test Results ANOVA test results with post-hoc Tukey-Kramer test comparing Thomasomys cinereus sensu stricto, T. lojapiuranus, sp. nov., T. shallqukucha, sp. nov., and T. pagaibambensis, sp. nov., using 13 variables with normal distribution. The meaning of the abbreviations of the variables are explained in Material and Methods. Abbreviations: F, F value; p, p-value; diff, mean differences; lwr, lower confidence interval; upr, upper confidence interval. Statistically significant values are shown in bold (P <0.05).
continued
APPENDIX 4
Factor Loadings for Principal Components and Discriminant Function Analyses
Factor loadings of principal components analyses (PC1, PC2, PC3), discriminant function analysis (DF1, DF2, DF3), and percentage of variance from 17 morphometric variables of Thomasomys cinereus, T. lojapiuranus, T. shallqukucha, and T. pagaibambensis.