FIELDIANA Zoology NEW SERIES, NO. 74 Intraspecific and Interspecific Variation in the Cryptotis nigrescens Species Complex of Small-Eared Shrews (Insectivora: Soricidae), with the Description of a New Species from Colombia Neal Woodman Robert M. Timm Museum of Natural History and Department of Systematics and Ecology University of Kansas Lawrence, Kansas 66045-2454 Accepted April 16, 1993 Published September 30, 1993 Publication 1452 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY Table of Contents ABSTRACT 1 RESUMEN 1 INTRODUCTION 1 METHODS 2 SEXUAL VARIATION 3 SYSTEMATIC DESCRIPTIONS 5 Cryptotis mayensis (Merriam, 1901) 7 Cryptotis merriami Choate, 1970 14 Cryptotis nigrescens (J. A. Allen, 1895) ... 17 Cryptotis mera Goldman, 1912 22 Cryptotis colombiana, new species 24 ACKNOWLEDGMENTS 27 LITERATURE CITED 28 SPECIMENS EXAMINED 29 10. Skull and mandible of C. merriami .... 16 11. Map showing the distribution of C. merriami 17 12. Plots of PCA factor scores of C. merria- mi and C. nigrescens 18 13. Skull and mandible of C. nigrescens .... 19 14. Map showing the distribution of C. ni- grescens 20 15. Skull and mandible of C. mera 22 16. Skull and mandible of C. colombiana .. 24 17. Map showing the distribution of C mera and the type locality of C. colom- biana 25 18. Tympanic region of the skulls of C. merriami, C. nigrescens, and C colom- biana 26 19. Plot of PCA factor scores of C. colom- biana, C. mera, and C. nigrescens 26 List of Illustrations 1. Cranial measurements used in this study 4 2. Plot of PCA factor scores for male and female C. nigrescens from Monteverde, Costa Rica 8 3. Plot of PCA factor scores for male and female C. nigrescens from San Felix, Panama 8 4. Skull and mandible of C. mayensis .... 9 5. Map showing the distribution of C. mayensis 10 6. Plot of C. mayensis and C. merriami on factor axes 1 and 3 from PCA of cranial variables 11 7. Plot of C mayensis and C. merriami on factor axes 1 and 2 from PCA of man- dibular variables 14 8. Plot of C mayensis and C. merriami on canonical axes 1 and 2 from discrimi- nant function analysis of cranial vari- ables 14 9. Plot of C. mayensis and C. merriami on canonical axes 1 and 2 from discrimi- nant function analysis of mandibular variables 15 List of Tables 1. Selected measurements of male and fe- male C nigrescens 6 2. Factor loadings from PCA of male and female C nigrescens from Monteverde, Costa Rica 7 3. Factor loadings from PCA of male and female C. nigrescens from San Felix, Panama 7 4. Selected measurements of members of the C. nigrescens group 12 5. Factor loadings from PCA of cranial variables of C. mayensis and C. merriami 13 6. Factor loadings from PCA of mandibular variables of C. mayensis and C merriami 14 7. Factor loadings from PCA of cranial variables of C merriami and C. nigres- cens 17 8. Factor loadings from PCA of cranial variables of C. colombiana, C. mera, and C. nigrescens 27 Intraspecific and Interspecific Variation in the Cryptotis nigrescens Species Complex of Small-Eared Shrews (Insectivora: Soricidae), with the Description of a New Species from Colombia Neal Woodman and Robert M. Timm Abstract The Cryptotis nigrescens species complex of small-eared shrews previously was considered to consist of a single, wide-ranging species with three subspecies distributed from southern Mexico to Panama. Our study of the patterns of morphological variation within this complex indicates that it is much more diverse and speciose than earlier believed. The group includes populations that occur only at high elevations (> 500 m) as well as the only taxon in the genus restricted to elevations below 100 m. Based on our investigations, we recognize five species: C. mayensis occurs on the Yucatan Peninsula and is known from an isolated locality in Guerrero; C. merriami has a patchy distribution from Chiapas, Mexico, to northern Costa Rica; C. nigrescens inhabits conterminous highlands in Costa Rica and Panama; C. mera is known from two isolated mountaintops along the Panama/Colombia border; and we describe a new species from the Central Cordillera of Colombia, which extends the known distribution of the complex to the southeast. In addition, recent studies indicate that C. hondurensis and C. magna may be imbedded phylogenetically within the C. nigrescens complex. Resumen Cryptotis nigrescens es un complejo de especies que solia considerarse como una sola especie con tres subespecies distribuidas a lo largo de America Latina desde el sur de Mexico hasta Panama. Nuestro estudio de patrones de variation morfologica en este complejo indica que es diverso y con mas especies de las que se habia pensado antes. El grupo incluye poblaciones que habitan solamente en elevaciones por encima de los 500 m, y tambien el unico taxon en el genero que esta restringido a elevaciones bajo 100 m. Basados en nuestras investigaciones, reconocemos cinco especies. La primera, C. mayensis, se encuentra en la Peninsula de Yucatan y en una localidad aislada en Guerrero, Mexico. La segunda, C. merriami, tiene una distribution discontinua desde Chiapas en Mexico hasta el norte de Costa Rica. La especie C. nigrescens habita en las cordilleras de Costa Rica y Panama. Se sabe que C. mera vive en las cimas de dos montafias aisladas en la frontera entre Panama y Colombia. La ultima es una especie nueva que describimos en esta publication y que mora en la Cordillera Central de Colombia y que extiende la distribution del complejo hacia el sureste. Ademas, estudios recientes indican que las especies C. hondurensis y C. magna pueden estar relacionadas filogeneticamente con el complejo C. nigrescens. Introduction America. The genus reaches its greatest diversity in Mexico and Central America. Hall and Kelson Small-eared shrews of the genus Cryptotis have (1959) recorded 25 species of Cryptotis in this re- a wide distribution, occurring in parts of North gion. In his revision of the Middle American Cryp- America, Central America, and northwestern South totis, Choate (1970) recognized eight species of FIELDIANA: ZOOLOGY, N.S., NO. 74, SEPTEMBER 30, 1993, PP. 1-30 1 small-eared shrews in Mexico and Central Amer- ica, and his taxonomy has been followed by most subsequent authors, including Hall (1981), Hona- cki et al. (1982), Corbet and Hill (1991), and Wil- son and Reeder (1993). We (Woodman & Timm, 1992, submitted) recognized four additional species from Guatemala, Honduras, and Mexico, bringing the total to 12 species. One poorly understood Central American group of Cryptotis is the C. nigrescens complex, which was known to occur from Guerrero, Mexico, through Central America to the Panama/Colom- bia border. Most species of small-eared shrews that occur from central Mexico through northern South America are high-elevation forms, inhab- iting regions over 500 m. The C. nigrescens com- plex, which includes both high-elevation and low- elevation taxa, is an exception. Choate (1970) provided the only recent comprehensive treat- ment of the taxonomy and distribution of C. ni- grescens. He recognized one widely distributed species, C. nigrescens, which included three sub- species: C. n. mayensis, occurring on the Yucatan Peninsula and known from one locality in Guer- rero, Mexico; C. n. merriami, which Choate de- scribed and is distributed from Chiapas, Mexico, to Honduras and El Salvador; and C. n. nigrescens, inhabiting Costa Rica and Panama to the Panama/ Colombia border. Of the previously recognized species, Choate (1970) synonymized C. mera, C. micrura, C. tersus, and C. zeteki with C. n. ni- grescens. Cryptotis mayensis was reduced to a sub- species of C. nigrescens. In his review of the Cryptotis of Middle Amer- ica, Choate (1970) had few complete specimens of C. nigrescens. Of the 237 C. n. mayensis available at that time, all but 10 were mandibles and partial crania from owl pellets, cave deposits, or Mayan ruins. Similarly, 15 of the 25 specimens available for C. n. merriami were from owl pellets, and only 24 skins with accompanying skulls of C. n. ni- grescens were available for study. Recent collect- ing in Mexico has added important new specimens of C. n. mayensis and C. n. merriami, and work in Panama and Costa Rica has tremendously in- creased the number of C. n. nigrescens in museum collections. This wealth of new material led us to review this complex more thoroughly than was possible previously. Cryptotis nigrescens was considered to be dis- tinguished by its short rostrum; bulbous dentition; equal development of anterior and posterior por- tions of M1; unrecessed posterior margins of P4, Ml, and M2; simple M3; and lack of an entoconid in m3 (Choate, 1970). However, there is much variation in these characters both within this taxon and within the genus, and their polarizations are difficult to define. Systematic relationships within the genus Cryp- totis remain mostly unresolved. Woodman (1992) showed that the C. nigrescens group may be para- phyletic with respect to C hondurensis and C magna. Cryptotis hondurensis recently was de- scribed from montane pine forests in Honduras (Woodman & Timm, 1992). Cryptotis magna was described by Merriam (1895) from mountains of southern Oaxaca, Mexico. A more comprehensive redescription of this species was provided by Choate (1970), and it was illustrated by Robertson and Rickart (1975), who summarized its biology. In this paper, we use the "C. nigrescens group" to refer informally to only those shrews previously called C. nigrescens. The purposes of this paper are to (1) revise the taxonomy and delineate the distributions of taxa previously included in C nigrescens; (2) describe and illustrate a species new to science; and (3) review the available information on reproduction, elevational distribution, and habitat for the mem- bers of this complex. Methods Cranial and mandibular measurements were re- corded to the nearest 0.1mm using either an ocular micrometer in a binocular microscope or a hand- held dial caliper. Skin measurements are those re- corded by the collector, except for head and body length (HB), which we calculated by subtracting the recorded tail length (TL) from the total length. All measurements are in millimeters. Dental ter- minology follows Choate (1970). The following craniomandibular measurements (fig. 1) were used in our analyses: condylobasal length, not including the upper incisors (CBL); cranial breadth (CB); breadth of zygomatic plate (ZP); interorbital breadth (IO); breadth of palate across first uni- cuspids (U1B); breadth of palate across third uni- cuspids (U3B); breadth of palate across second molars (M2B); palatal length (PL); upper tooth row length, Ul to M3, parallel to the long axis of the skull (TR); unicuspid tooth row length, parallel to the unicuspid tooth row (UTR); molariform tooth row length, P4 to M3, parallel to the long axis of the skull (MTR); posterior width of M1, across hypocone and metastyle (WM1); mandibular FIELDIANA: ZOOLOGY length, from inferior sigmoid notch to posterior edge of mental foramen (ML); height of coronoid process (HCP); height of coronoid valley (HCV); height of articular condyle (HAC); breadth of ar- ticular condyle (BAC); articular condyle to pos- terior edge of m3 (AC3); lower tooth row length, p3 to m3 (TRD); and length of lower first molar (mlL). Unless otherwise stated, univariate statis- tics are mean ? standard deviation. All capitalized color names follow Ridgway (1912). Localities and elevations were taken directly from specimen tags and represent the descriptions of the original col- lectors. Corrections and additions are provided in brackets. We did not convert distances measured in miles or elevations measured in feet to the met- ric system to avoid inferring a level of accuracy greater than that originally recorded by the col- lector. Specimens from archeological or paleon- tological contexts were not used in statistical anal- yses or plotted on maps because of the possibility of temporal variation in size, shape, or geographic distribution. Number of specimens from archeo- logical or paleontological sites is calculated as the minimum number of individuals. Multivariate analyses were used to look at pat- terns of variation in overall similarity among spe- cific members of the C. nigrescens complex. Al- though these analyses often are useful for distinguishing among known groups, and occa- sionally point out variables useful for distinguish- ing taxa, they were not relied upon to determine species or provide taxonomic diagnoses. In gen- eral, members of the C. nigrescens complex were found to be very conservative in overall form, particularly as determined by principal compo- nents analyses (PCA), which is a descriptive rather than a discriminatory tool. Most variation de- tected by PCA was in size rather than shape. De- spite this similarity in shape, species differ dis- tinctly in their possession of specific characters, and we used these characters to separate them. Analyses of variance (ANOVA), correlation ma- trices, and multivariate analyses were carried out using BMDP on the University of Kansas Aca- demic Computing Services' IBM VM/CMS com- puter system. All data were log-transformed prior to carrying out these analyses. Elevational and geographic distributions and the correlation of character states to these distribu- tions were examined for all members of the C. nigrescens group. Specimens used in our analyses are listed in the Specimens Examined section. These specimens are deposited in the following institutions: American Museum of Natural History, New York (AMNH); Angelo State Natural History Collections, San An- gelo, Texas (ASNHC); Escuela Nacional de Ciencias Biologicas, Mexico (ENCB); Field Museum of Nat- ural History, Chicago (FMNH); Instituto de Biolo- gia, Universidad Nacional Autonoma de Mexico, Mexico City (IBUNAM); Instituto Nacional de Bio- diversidad, Santo Domingo de Heredia, Costa Rica (iNBio); University of Kansas Museum of Natural History, Lawrence (KU); Los Angeles County Mu- seum, Los Angeles (LACM); Museum of Compar- ative Zoology, Harvard University, Cambridge (MCZ); Museum of the High Plains, Fort Hays State University, Hays, Kansas (MHP); James Ford Bell Museum of Natural History, University of Min- nesota, Minneapolis (MMNH); Museo Nacional de Costa Rica, San Jose (MNCR); Museum of Verte- brate Zoology, University of California, Berkeley (MVZ); Royal Ontario Museum, Toronto (ROM); University of Iowa Museum of Natural History, Iowa City (sui); Departamento de Biologia, Uni- versidad Autonoma Metropolitana Iztapalapa, Mexico (UAMI); University of Michigan Museum of Zoology, Ann Arbor (UMMZ); Universidad Na- cional Autonoma de Honduras, Tegucigalpa (UNAH); and U.S. National Museum, Washington, D.C. (USNM). Sexual Variation Male and female Cryptotis are difficult to tell apart, and few tests for sexual dimorphism within the genus have been carried out, at least in part because there are few good series of individuals of known sex. The two studies that have investigated sexual dimorphism in Cryptotis indicate that sex- ual differences in mensural variables are minor at most. Choate (1970) found only one of seven vari- ables (length of maxillary tooth row) that he ex- amined to be significantly different between the sexes in his morphometric analysis of C. mexicana from near Jalapa, Veracruz, Mexico. Only 1 of 19 variables differed significantly between males and females of C. gracilis in Costa Rica and Panama (Woodman, 1992); zygomatic plate length was sig- nificantly longer in females. We tested for secondary sexual variation in C. nigrescens using our two largest series of sexed individuals. One included 17 males (14 crania, 16 skins) and 16 females (12 crania, 16 skins) from Monteverde, Costa Rica, and the other consisted of 13 males (13 crania, 9 skins) and 10 females WOODMAN & TIMM: SMALL-EARED SHREWS FIG. 1. Cranial and mandibular measurements used in this study. Abbreviations of variables are explained in the Methods section of text. (10 crania, 8 skins) from near San Felix (including Cerro Bollo), Panama. A correlation matrix was calculated for 19 craniomandibular variables (CBL, CB, ZP, IO, U1B, M2B, PL, TR, UTR, MTR, WM1, ML, HCP, HAC, BAC, TRD, mlL, AC3, HAV). Because sample sizes were different for skin variables and for craniomandibular variables, skin measurements (HB, TL) were not included in the correlation matrix. Variables then were culled so that no two variables in the remaining data set had a correlation coefficient of 0.75 or greater. This yielded eight variables (CBL, ZP, UTR, MTR, WM1, HAC, BAC, HCV); all other variables cor- related strongly with CBL, except HCV, which was strongly correlated with HAC. A two-way ANO- VA was calculated for each of these eight cranio- mandibular variables and the two skin variables. Both sex and locality (Monteverde vs. San Felix) were tested as sources of variance to separate ef- fects of sexual dimorphism and geographic vari- ation. Statistical significance was determined using a Brown-Forsythe test for equality of means, which does not assume equality of the variances. Pro- tected alpha values were not used despite a lack of independence among the variables tested, be- cause protected values would have increased the likelihood of making Type II errors (i.e., incor- rectly accepting no difference between males and females or between localities). Only one variable, UTR (P = 0.0042), exhibited a statistically significant difference between sexes, with females larger than males. In addition, CBL (P = 0.0566) was close to being significantly dif- ferent; again, females were larger than males. Five of the 10 variables (ZP, MTR, WM1, BAC, HCV) had P > 0.30, indicating no distinction between the sexes. The remaining three variables (HB, TL, HAC) had probabilities falling between 0.10 and 0.30, providing no clear evidence of whether a difference existed between males and females. Fe- males averaged slightly larger than males for five measurements in the groups from both Monte- verde (TL, CBL, ZP, UTR, HAC) and San Felix (CBL, UTR, MTR, HAC, mlL), although not for all the same variables. Males from Monteverde were larger than females for HB, and males from FIELDIANA: ZOOLOGY San Felix were larger for HB and TL. However, actual differences between male and female means in both groups were small (table 1). If we use the differences, taken to four decimal places, between the male and female means for the eight cranio- mandibular variables, we find that they average 0.0470 (range = 0.0169-0.0892) for the specimens from San Felix and 0.0926 (range = 0.0107-0.4630) for specimens from Monteverde. Most are below our ability to measure accurately. In contrast to the sex term, 9 of the 10 variables tested were significantly different between locali- ties. Seven of these variables (HB, TL, CBL, UTR, WM1, HAC, BAC) had P < 0.001, and two (MTR, mlL)hadP < 0.05. Only ZP showed no difference between the populations at Monteverde and San Felix (P = 0.3531). For all variables, specimens from San Felix averaged larger than those from Monteverde. In fact, males from San Felix aver- aged larger than females from Monteverde for all variables except ZP, UTR, and m 1L. The interaction between sex and locality was not significant for any variables, and P < 0.30 for only TL (P = 0.0734), CBL (P = 0.0947), and ZP (P = 0.2593). Body mass has been shown to be a good cor- relate of other estimators of size in some other small mammals (Iskjaer et al., 1989). Average weights of males and females from Monteverde were the same (table 1), giving no indication of sexual dimorphism. However, our use of body mass was hampered by the imprecision with which very small animals can be weighed in the field and small sample sizes available. Geographic variation could not be evaluated, because weights were not avail- able for specimens from San Felix. To test whether all craniomandibular variables together would show size differentiation between males and females, we ran PCAs separately for specimens from Monteverde and from San Felix using the original 19 craniomandibular variables and the same individuals as for the ANOVAs. Because all variables generally load heavily on fac- tor axis 1 in a PCA, this axis provides a good estimator of overall size. Therefore, factor 1 scores should prove to be useful in determining any ob- vious patterns of size differentiation between the sexes. In the PCA of specimens from Monteverde, factor 1 estimates size and factor 2 is a shape score representing contrast between a combination of mlL and WM1 vs. the negatively weighted AC3 (table 2). A plot of factor 1 and factor 2 scores for these specimens (fig. 2) indicates nearly complete overlap of males and females on both the size and shape axes. In the PCA of specimens from San Felix, factor 1 represents size and factor 2 is a contrast between shape of the articular condyle of the mandible (HCV and HAC) vs. a negatively weighted mlL (table 3). In the plot of factor 1 and factor 2 scores for these specimens (fig. 3), males and females exhibit complete overlap along the shape axis. Along the size axis, however, there is separation of the two sexes. Males range along the entire length of factor axis 1 but are more con- centrated at the lower end of the scale. Females all cluster toward the upper end of the scale. Our analysis of sexual variation in C. nigrescens shows no clear pattern of sexual dimorphism in size within this species. Although females gener- ally averaged larger in most craniomandibular measurements in which there were differences be- tween the sexes, males averaged larger in HB, es- pecially among specimens from San Felix. The overlap between males and females for all mea- surements was great. Only two variables were sig- nificant or close to being significant when tested for differences between the sexes, and the statis- tical significance of those variables may have been a result of multiple comparison of dependent means. No other variables tested were clearly dif- ferent. PCA indicated a tendency for male and female C. nigrescens from San Felix to separate out on size, but this was not the case for those from Monteverde, and the size range for males from San Felix overlapped completely that of fe- males. These results indicate that any sexual variation that exists is beyond our ability to measure ac- curately and precisely in these diminutive mam- mals. Differences between sexes were considerably less than differences between geographic localities, and they did not influence statistical tests among populations or contribute significantly to overall patterns of geographic variation in which we were interested. Possessing few sexed individuals from each locality, we could not partition our sample into tooth-wear classes (Rudd, 1955) and thus pro- portion out possible effects of age-related differ- ences. However, because sexual variation did not inhibit our ability to detect geographic variation within C. nigrescens, we used males, females, and unsexed individuals in all subsequent analyses. Systematic Descriptions In the following descriptions, species are ar- ranged geographically, roughly from northwest to WOODMAN & TIMM: SMALL-EARED SHREWS TABLE 1. Selected measurements of male and female C. nigrescens from Monteverde, Costa Rica, and San Felix, Panama. Abbreviations of measurements are explained in the Methods section of text. The statistics presented are mean ? standard deviation of the mean, and observed extremes. Number of individuals for each species is in parentheses. HB TL CBL CB ZP IO U1B M2B PL TR UTR MTR WM1 ML HCP HCV HAC Monteverde San Felix Males Females Males Females 68 ? 5 67 ? 7 77 ? 3 74 ?4 60-76 56-79 72-83 68-79 (16) (16) (9) (8) 28 + 4 30 ? 2 34 ? 2 33 ? 2 20-34 27-33 29-37 30-36 (16) (16) (9) (8) 17.9 ?0.5 18.4 ?0.5 19.2 ? 0.4 19.3 ? 0.4 16.9-18.7 17.7-19.3 18.7-20.2 18.3-19.8 (14) (12) (13) (10) 9.0 ? 0.2 9.0 ? 0.3 9.7 ? 0.2 9.7 ? 0.2 8.7-9.4 9.4-10.0 9.4-10.3 9.2-10.0 (14) (12) (13) (10) 1.9 ? 0.2 2.0 ?0.1 2.0 ? 0.2 2.0 ? 0.2 1.6-2.2 1.8-2.2 1.5-2.5 1.6-2.2 (14) (12) (13) (10) 4.2 ?0.1 4.2 ? 0.2 4.6 ? 0.1 4.6 ? 0.2 4.0-4.3 4.0-4.7 4.4-4.9 4.3^1.8 (14) (12) (13) (10) 2.3 ? 0.1 2.3 ?0.1 2.6 ?0.1 2.6 ?0.1 2.2-2.5 2.2-2.5 2.5-2.8 2.5-2.7 (14) (12) (13) (10) 5.3 ? 0.1 5.3 ? 0.2 5.7 ?0.1 5.8 ?0.1 5.1-5.6 4.9-5.6 5.5-5.9 5.5-6.0 (14) (12) (13) (10) 7.6 ? 0.2 7.9 ? 0.3 8.1 ? 0.2 8.3 ? 0.3 7.2-7.9 7.4-8.6 7.8-8.3 7.8-8.8 (14) (12) (13) (10) 6.8 ? 0.2 7.1 ? 0.2 7.4 ? 0.2 7.5 ? 0.3 6.4-7.2 6.7-7.3 7.2-7.6 6.9-7.8 (14) (12) (13) (10) 2.3 ? 0.2 2.4 ? 0.1 2.4 ? 0.1 2.5 ?0.1 2.0-2.5 2.3-2.5 2.2-2.6 2.4-2.6 (14) (12) (13) (10) 5.2 ? 0.2 5.2 ? 0.2 5.3 ?0.1 5.4 ? 0.2 4.8-5.5 4.8-5.6 5.2-5.5 4.9-5.7 (14) (12) (13) (10) 1.7 ? 0.1 1.7 ? 0.1 1.8 ?0.05 1.8 ? 0.1 1.5-1.8 1.6-1.8 1.7-1.9 1.7-1.9 (14) (12) (13) (10) 6.1 ?0.2 6.3 ? 0.2 6.4 ? 0.3 6.5 ? 0.2 5.7-6.5 5.9-6.6 6.0-6.9 6.1-6.9 (14) (12) (13) (10) 4.3 ? 0.2 4.4 ? 0.2 4.7 ? 0.1 4.8 ?0.1 4.0-4.6 4.1^t.7 4.6-5.0 4.6-5.0 (14) (12) (13) (10) 2.6 ?0.1 2.7 ? 0.1 2.8 ? 0.1 2.9 ?0.1 2.4-2.8 2.5-2.9 2.6-3.0 2.8-3.0 (14) (12) (13) (10) 3.6 ? 0.2 3.7 ? 0.2 3.8 ? 0.2 3.9 ?0.1 3.3-3.9 3.4-4.0 3.5-4.1 3.7-4.0 (14) (12) (13) (10) FIELDIANA: ZOOLOGY TABLE 1. Continued. Monteverde San Felix Males Females Males Females BAC AC3 TRD mlL Weight (g) 2.9 ?0.1 2.7-3.1 (14) 4.4 ? 0.2 4.1-4.7 (14) 5.4 ? 0.2 5.1-5.6 (14) 1.7 ? 0.1 1.6-1.8 (14) 6 ? 1 4-8 (13) 2.9 ?0.1 2.7-3.1 (12) 4.5 ? 0.2 4.2-4.8 (12) 5.5 ? 0.2 5.2-5.8 (12) 1.7 ?0.1 1.6-1.8 (12) 6 ? 1 5-7 (9) 3.0 ? 0.1 2.9-3.2 (13) 4.6 ? 0.2 4.4-5.1 (13) 5.8 ? 0.1 5.7-6.2 (13) 1.7 ? 0.1 1.6-1.9 (13) 3.0 ? 0.1 2.9-3.2 (10) 4.7 ? 0.05 4.7^1.8 (10) 5.9 ? 0.2 5.3-6.2 (10) 1.8 ? 0.1 1.6-1.9 (10) southeast. Accounts are not provided for C. hon- durensis or C. magna, although there is evidence that these two species may be included within the C. nigrescens group (Woodman, 1992). However, comparisons of these two species are made with the species described below. Cryptotis mayensis (Merriam, 1901) (fig. 4) Blarina mayensis C. H. Merriam, 29 November 1901, Proc. Washington Acad. Sci. 3:559. Cryptotis mayensis: G. S. Miller, Jr., 31 December 1912, Bull. U.S. Natl. Mus. 79:26; R. T. Hatt, 18 August 1938, J. Mamm. 19:334; P. Hershkovitz, 10 July 1951,FieldianaZool. 31:522; R. T. Hatt et al., March 1953, Cranbrook Inst. Sci. Bull. 33:59; E. R. Hall and K. R. Kelson, 31 March 1959, The Mammals of North America 1:61; T. Alvarez and A. Martinez G., 4 August 1967, Southwest. Nat. 12: 205; R. L. Peterson, 26 November 1968, J. Mamm. 49:796. Blarina mexicana: G. F. Gaumer, 1917, Monografia de los mamiferos de Yucatan, p. 249 (part). Cryptotis micrura: A. Murie, 15 July 1935, Misc. Publ. Mus. Zool., Univ. Michigan 26:17 (part); E. R. Hall TABLE 2. Factor loadings from PCA of male and fe- male C. nigrescens from Monteverde, Costa Rica. Ab- breviations of measurements are explained in the Meth- ods section of the text. TABLE 3. Factor loadings from PCA of male and fe- male C. nigrescens from the vicinity of San Felix, Pan- ama. Variable Factor 1 Factor 2 Variable Factor 1 Factor 2 CBL 0.875 -0.179 TRD 0.901 -0.291 TR 0.860 0.221 MTR 0.879 -0.167 PL 0.857 -0.090 CBL 0.872 -0.278 HCP 0.824 -0.260 TR 0.836 -0.051 ML 0.786 -0.175 PL 0.817 0.117 TRD 0.755 0.485 HCP 0.738 0.474 U1B 0.728 0.007 IO 0.716 0.088 HAC 0.708 -0.284 BAC 0.713 0.379 IO 0.692 0.102 CB 0.708 -0.248 BAC 0.667 -0.243 ML 0.700 -0.417 HCV 0.663 -0.390 M2B 0.682 0.176 M2B 0.658 0.440 AC3 0.570 -0.032 UTR 0.652 0.160 HCV 0.566 0.634 MTR 0.642 0.392 U1B 0.527 0.324 AC3 0.606 -0.589 ZP 0.496 -0.219 CB 0.596 -0.218 UTR 0.459 -0.123 ZP 0.539 -0.283 HAC 0.449 0.631 mlL 0.535 0.584 WM1 0.443 -0.290 WM1 0.445 0.582 mlL 0.340 -0.703 WOODMAN & TIMM: SMALL-EARED SHREWS o o 2 - 0 ? 1 - o o o ? o 3 ? o O 0 - o ? ? ? ? 1 - ? o 0 * 2 - 0 ? males females o i i factor 1 FIG. 2. Plot of PC A factor 1 and factor 2 scores for males and females of C. nigrescens from Monteverde, Costa Rica. o J - 2 - 0 * O 0 1 - o 8 ? o 0 - o 0 0 o o o ? ? 1 -I o ma es ?o ? ? ? females ? i 1 -2-10 1 2 factor 1 FIG. 3. Plot of PCA factor 1 and factor 2 scores for males and females of C. nigrescens from the vicinity of San Felix, Panama. and K. R. Kelson, 31 March 1959, The Mammals of North America 1:62 (part). Cryptotis nigrescens mayensis: J. R. Choate, 30 De- cember 1970, Univ. Kansas Publ. Mus. Nat. Hist. 19:275; E. R. Hall, 3 April 1981, The Mammals of North America 1:63; R. C. Dowlerand M. D. Eng- strom, 15 December 1988, Ann. Carnegie Mus. 57: 160. HOLOTYPE?Skin and skull of subadult female, U.S. National Museum of Natural History no. 108087, obtained by E. W. Nelson and E. A. Gold- man (collector number 14495) 5 February 1901 from Mexico, Yucatan, Chichen Itza. DISTRIBUTION?Known primarily from below 100 m on the Yucatan Peninsula of Mexico and adjacent Belize and Guatemala (fig. 5). Also known from remains isolated from owl pellets collected in a cave at 650 m, in the Canon del Zopilote, Guerrero, Mexico. DESCRIPTION?A medium-sized Cryptotis, HB averaging 69 (table 4); tail short, averaging 41% (? 5%, n = 11) of HB; dorsal hairs about 3 mm long, ranging from about 2 to 4 mm, three-banded; dorsum often has mottled, salt and pepper ap- pearance; dorsal pelage variable: Pale Ecru-Drab and Drab, Light Drab, Light Mouse Gray to Deep Mouse Gray, Pale Mouse Gray or Mouse Gray with touches of Olive Brown; lateral pelage from Smoke Gray to Mouse Gray; venter slightly paler than dorsum, may include one or more of the following: Pale Olive Gray, Light Olive Gray, Light Grayish Olive, Pale Mouse Gray to Light Mouse Gray. Rostrum narrow and of normal length (PL/CBL = 43.4% ? 1.2, n = 10); interorbital area mod- erately wide; usually only one dorsal foramen (81%, n = 65) located on either the right or left frontal, small to medium in size; a well-developed fora- men leading to a ventral extension of the sinus canal typically present posterior to the dorsal ar- ticular facet on one (8%, n = 62) or both (86%) sides of the skull; normally no foramen dorsal to dorsal articular facet (93%, n = 61); anterior pro- cess of the petromastoid low and very narrow (fig. 18A); paroccipital process prominent; zygomatic plate broad in proportion to CBL (11.8% ? 0.7, n = 10) and PL (27.8% ? 1.6, n = 72); anterior border of zygomatic plate from parastyle/meso- style valley to mesostyle of M1, posterior border from metastyle of M2 to middle of M3, and from posterior one-third of base of maxillary process to posterior to maxillary process; palate long and nar- row; upper tooth row crowded; unicuspids mas- sive; U4 normally displaced medially, so that U3 and P4 in contact or nearly so; U4 not visible in lateral view of skull; lateral view of U3 also some- times partially obstructed by P4; posterior borders of P4, M1, and M2 unrecessed or only very slightly recessed; M3 with well-developed paracrista and paracone, reduced precentrocrista, poorly devel- oped and normally uncolored mesostyle and pro- tocone, and hypocone absent or vestigial and in- corporated into the posterior cingulum; dentition bulbous. Mandible large; horizontal ramus deep; coro- noid process high (HCP/ML = 84.3% ? 3.3, n = FIELDIANA: ZOOLOGY FIG. 4. Dorsal, ventral, and lateral views of the skull and lateral view of the skull and mandible of C. mayensis (ASNHC 6071). Scale bar = 5 mm. 21) and broad, joins mandible at steep angle; viewed posteriorly, horizontal and vertical branches of articular condyle short and broad; lower sigmoid notch very shallow, not extending beneath ventral border of articular process; pos- terior border of lower incisor extends nearly to posterior border of cingulum of p4; lower denti- tion wide; only hypoconid in talonid of m3. COMPARISONS? Cryptotis mayensis has the pal- est pelage of any member of the C. nigrescens group, appearing medium gray (often speckled with patches of white) rather than dark brown to black. Cryptotis magna?C. mayensis is much smaller (HB averaging 69 vs. 86; table 4) and has much paler pelage; much shorter tail (averaging 41% of HB vs. 53%); skull much smaller in all dimensions, but zygomatic plate relatively broader (averaging 11.8% of CBL vs. 10.1% ? 0.6); usually only one dorsal foramen; well-developed foramen posterior to dorsal articular facet (foramen present but al- ways vestigial in C. magnet); rarely possesses fo- ramina dorsal to dorsal articular facet; no ento- conid in talonid of m3 (vestigial entoconid present in C. magnd). Cryptotis hondurensis?C. mayensis is much larger (HB averaging 69 vs. 61 ? 5; table 4) and has much paler pelage; shorter tail (averaging 41 % vs. 46%); M3 less complex, lacking metacone; den- tition bulbous; coronoid process of mandible much higher and broader. REMARKS?The pelage of Cryptotis mayensis is paler than in any other member of the genus, in- cluding C. parva, from which the coloration differs in being more gray and less olive. The overall appearance of C. mayensis is a medium brownish gray. Close inspection of the dorsal pelage of C. mayensis reveals that the individual hairs are three- banded. The proximal one-half to three-quarters of the hairs is a medium gray, followed by a narrow band of light gray. The distal tip (about one-eighth of the total length) of the hair is brownish gray. The transition between bands is not sharp, with the colors instead tending to grade into each other. The great extent of lighter, basal coloration in the hairs of this taxon probably accounts for its overall paler appearance and greater reflectance. Goldman (1951) described the northern Yuca- tan habitat around Chichen Itza, from which the WOODMAN & TIMM: SMALL-EARED SHREWS ? C. mayensis ? Guerrero FIG. 5. Map of southern Mexico, Belize, and Guatemala showing the distribution of C. mayensis. holotype of C. mayensis was collected, as stunted tropical forest about 25-40 ft high with dense, brushy undergrowth. He noted that the weather is highly seasonal, with a long, hot, dry season from about December to June and a rainy season marked by torrential afternoon showers. Dowler and Eng- strom (1988) reported a specimen from 7.5 km W Escarcega, Campeche, as having been trapped in mature, transitional deciduous-evergreen forest. In the same area they also collected Heteromys gaumeri, Ototylomys phyllotis, Peromyscus yuca- tanicus, and Oryzomys melanotis. Two C. may- ensis from 60 km SE of Dzibalchen, Campeche, were taken in pitfalls set in transitional deciduous- evergreen forest dominated by escobo palms (Cry- sophila) and Sabal palms (M. D. Engstrom, in litt.). Alvarez and Martinez (1967) captured a C. mayensis along a road in tropical rain forest op- posite a cornfield, 2 km SE Laguna Chickanka- naab, Quintana Roo. Few reproductive data are available for C, may- ensis, and no pregnant or lactating females have been recorded. Males captured on 4 June, 24 June, and 19 August showed no indications of having lateral glands. When present, these paired glan- dular areas lack long guard hairs and underfur but have a sparse covering of short, fine hairs (Wood- man & Timm, submitted). Both males and females possess lateral glands, but in females they are much smaller and more difficult to see (Murariu, 1976; Bee et al., 1980). It is likely that these glands serve a function in sexual communication. Eadie (1938) found that the lateral glands of male Blarina showed increased activity with enlargement of the testes. Cryptotis mayensis is rarely collected, and there are few complete specimens in museum collec- tions. However, it is unlikely that this shrew is truly rare, because remains from owl pellets are abundant. Of 122 specimens examined in our study, 76 came from owl pellets, 33 are from ar- cheological contexts, and 13 are standard museum specimens. Cryptotis mayensis is distinctive in being the only member of the genus restricted to lowland areas. On the Yucatan Peninsula the species is not known to occur above 100 m. The only specimens of C. mayensis from outside of the Yucatan Pen- insula are crania and unassociated mandibles col- lected by William Lopez-Forment C. in 1969 from below a barn owl (Tyto alba) roost in Macuiltzingo 10 FIELDIANA: ZOOLOGY Cave [Cueva del Canon de Zopilote], located at 650 m in the Canon de Zopilote, Guerrero, Mexico (Choate, 1970). The precise locality and habitat where the shrews were captured by owls is un- known, and subsequent work in Macuiltzingo Cave has not produced additional specimens (Lopez- Forment & Urbano, 1977; W. Lopez-Forment C, pers. comm.), suggesting a change in feeding pat- terns by the owls or local extinction of the pop- ulation of shrews upon which they were feeding. When one of us (Woodman) visited the cave with W. Lopez-Forment C. in August 1991, no owls were occupying the cave, and there was no evi- dence of any recent use of the cave by owls. Despite repeated collecting in the region surrounding the cave (Choate, 1970; W. Lopez-Forment C, pers. comm.; fieldwork by Woodman in 1991), no ad- ditional specimens of these shrews have been ob- tained. The presence of shrews identifiable as C. may- ensis in Guerrero presents an interesting biogeo- graphical problem. These shrews are from an area that clearly is isolated geographically from the Yu- catan Peninsula and is climatically and vegeta- tionally distinct. Choate (1970) noted the great distance (> 950 km) and extensive biogeograph- ical barriers between Zopilote Canyon and the nearest known locality of C. mayensis on the Yu- catan Peninsula and suggested that the specimens from Zopilote Canyon eventually might prove to be subspecifically distinct. All known specimens from Guerrero are mandibles and incomplete skulls, making comparisons with other taxa, in- cluding C. mayensis, difficult. Because the specimens from Guerrero were from far outside the range of all other populations on the Yucatan Peninsula, and because of the geo- graphical barriers and habitat differences between the two areas, we considered that specimens from Guerrero might prove to be either misidentified or an undescribed species. The fragmentary nature of the specimens from Guerrero limited the mea- surements that could be taken from them and did not yield characters useful for separating them from C. mayensis on the Yucatan Peninsula. We carried out PCAs and discriminant function analyses in order to test the overall similarity of the two groups of specimens. Crania and mandibles were tested separately, because these elements are disasso- ciated in all known specimens from Guerrero. Analyses of crania included six variables (ZP, IO, M2B, PL, MTR, WM1) measured on 20 C. may- ensis from the Yucatan Peninsula, 16 C. merriami, and 20 specimens from Guerrero. Analyses carried n u a O 2 - 1 - O ? o ?~ ? ? o o * 0? ? 0*8 V V 0 - ?v o S 0 ?? V v V ? ? 9 V 1 - ? ? ? V o w V V 2 - ? V V V 3 - O Guerrero ? C. mayensis V C. merriami 1 0 1 factor 1 FIG. 6. Plot of specimens of C. mayensis from Guer- rero and the Yucatan Peninsula and C. merriami on factor axes 1 and 3 from PCA of cranial variables. out on mandibles utilized eight variables (ML, HCP, HCV, HAC, BAC, TRD, mlL, AC3) from 20 C. mayensis, 24 C. merriami, and six mandi- bles from Guerrero. Specimens of C. merriami were included to determine whether or not the variables used were useful for distinguishing groups. PCAs indicated wide overlap of C, mayensis from the Yucatan with specimens from Guerrero. In the PCA of crania, all variables except ZP load- ed out heavily on factor 1, providing a measure of overall size of the specimens (table 5). Factor 3 was a shape axis, providing a contrast between PL and M2B. Factor 2 from this analysis repre- sented a contrast between a very heavily weighted ZP and a slightly negatively weighted WM1; it did not aid in separating any groups. In a plot of factor scores (fig. 6), C. mayensis and specimens from Guerrero are essentially indistinguishable. Cryp- totis mayensis and C. merriami are separated by a combination of size and shape. Cryptotis mer- riami tends to be larger cranially than C. mayensis, and, at any given size, C. mayensis has a longer but narrower palate than C. merriami. In the PCA carried out using mandibles, factor 1 is a size axis and factor 2 is a contrast between a combination of variables representing length of the mandible (mlL, TRD, ML) and HCP (table 6). A plot of specimens on factor 1 and factor 2 (fig. 7) shows C. mayensis from the Yucatan Pen- insula and specimens from Guerrero again inter- mixed. In contrast to the PCA using cranial vari- ables, C. merriami and C. mayensis do not separate out on the size axis but can be discriminated by WOODMAN & TIMM: SMALL-EARED SHREWS 11 TABLE 4. Selected measurements of members of the C. nigrescens group. Abbreviations of measurements are explained in the Methods section of text. The statistics presented are mean ? standard deviation of the mean, and observed extremes. Number of individuals for each species is in parentheses. HB TL CB ZP IO U1B U3B M2B PL TR UTR MTR WM1 ML C. mayensis C. magna C. merriami C. hondurensis (holotype and 2 paratypes) C. C. colombiana nigrescens C. mera (holotype) TL as % of HB CBL 69 ? 8 61-90 (11) 28 ? 3 24-33 (11) 41 ? 5 31-49 (11) 19.0 ? 0.4 18.2-19.6 (10) 9.2 ? 0.3 8.8-9.8 (10) 2.3 ? 0.1 2.0-2.5 (21) 4.4 ? 0.2 4.1-4.8 (21) 2.6 ? 0.1 2.2-2.7 (21) 3.0 ?0.1 2.8-3.3 (21) 5.5 ? 0.1 5.2-5.7 (21) 8.3 ?0.3 7.8-9.0 (21) 7.3 ? 0.2 6.9-7.8 (21) 2.7 ? 0.2 2.1-3.1 (21) 5.0 ?0.1 4.8-5.3 (21) 1.7 ? 0.0 1.7-1.8 (21) 6.2 ? 0.2 5.8-6.8 (24) 86 ? 6 70-100 (47) 46 ? 4 34-53 (47) 54 ? 6 34-64 (47) 22.8 ? 0.4 21.9-23.4 (17) 11.6 ?0.2 11.2-12.3 (17) 2.3 ?0.1 2.1-2.6 (17) 5.6 ? 0.1 5.4-5.8 (17) 3.2 ? 0.1 3.0-3.4 (17) 3.7 ? 0.2 3.5-3.9 (17) 6.8 ? 0.1 6.5-7.0 (17) 10.1 ?0.3 9.6-10.8 (17) 9.1 ? 0.2 8.8-9.3 (17) 3.2 ? 0.1 3.1-3.4 (17) 5.1 ?0.6 5.1-5.2 (17) 1.8 ?0.1 1.7-1.9 (17) 7.3 ? 0.2 6.8-7.6 (17) 69 + 4 60-77 (23) 29 ?4 22-38 (23) 43 ? 6 32-55 (23) 19.4 ? 0.4 18.7-20.0 (16) 9.6 ? 0.2 9.2-10.0 (18) 2.3 ? 0.2 2.0-2.6 (21) 4.7 ? 0.1 4.4-4.9 (21) 2.6 ? 0.2 2.1-2.8 (21) 3.0 ? 0.2 2.7-3.1 (6) 5.8 ? 0.2 5.2-6.2 (21) 8.4 ? 0.3 7.9-8.9 (21) 7.4 ? 0.2 6.8-7.7 (21) 2.5 ? 0.1 2.2-2.7 (21) 5.2 ? 0.2 5.0-5.5 (17) 1.9 ?0.1 1.7-2.0 (21) 6.5 ? 0.3 6.0-7.2 (21) 55, 63, 64 29, 27, 27 53, 43, 42 19.2, -, - 9.2, -, - 2.2, 2.0, 2.1 4.3, 4.5, - 2.4, 2.4, - 2.9, 3.0, - 5.4, 5.4, - 8.4, 7.8, - 7.1, 7.1, 7.1 2.6, 2.4, 2.4 5.1, 5.1, 5.2 1.7, 1.7, 1.9 6.1, 5.8,6.3 68 ?7 55-83 (70) 29 ? 4 19-37 (70) 43 ? 5 28-53 (70) 18.6 ? 0.7 16.9-20.2 (70) 9.2 ? 0.4 8.0-10.3 (70) 2.0 ? 0.2 1.5-2.5 (70) 4.4 ? 0.2 3.9^1.9 (70) 2.4 ? 0.2 2.1-2.8 (70) 2.8 ? 0.2 2.4-3.2 (70) 5.5 ?0.3 4.9-6.0 (70) 7.9 ? 0.3 7.2-8.8 (70) 7.1 ? 0.3 6.4-7.8 (70) 2.4 ? 0.1 2.0-2.6 (70) 5.2 ? 0.2 4.7-5.7 (70) 1.8 ?0.1 1.5-1.9 (70) 6.2 ? 0.3 5.5-6.9 (70) 69 ?2 67-73 (6) 27 ? 2 24-31 (6) 39 ?4 34-46 (6) 18.3 ? 0.4 17.7-18.7 (6) 9.4 ? 0.3 9.1-9.8 (6) 2.0 ? 0.1 1.9-2.2 (6) 4.5 ? 0.2 4.3-17 (6) 2.5 ? 0.1 2.4-2.6 (6) 2.9 ? 0.1 2.7-3.0 (6) 5.6 ?0.1 5.5-5.8 (6) 7.8 ?0.1 7.6-7.8 (6) 7.1 ?0.1 6.9-7.2 (6) 2.4 ? 0.1 2.3-2.5 (6) 5.1 ? 0.05 5.1-5.2 (6) 1.8 ? 0.05 1.8-1.9 (6) 5.9 ? 0.4 5.2-6.2 (6) 76 27 36 19.9 10.0 2.0 4.9 2.6 3.2 6.3 7.7 2.5 6.9 12 FIELDIANA: ZOOLOGY TABLE 4. Continued. C. hondurensis C. (holotype and C. colombiana C. mayensis C. magna C. merriami 2 paratypes) nigrescens C. mera (holotype) HCP 5.3 ? 0.3 6.1 ?0.2 5.0 ? 0.2 4.8, 4.4, 4.6 4.5 ? 0.3 4.5 ?0.1 4.8 4.8-5.7 5.7-6.3 4.6-5.3 4.0-5.0 4.3^t.6 (24) (17) (21) (70) (6) HCV 2.9 + 0.1 3.4 ? 0.1 2.9 ? 0.2 2.7, 2.7, 2.7 2.7 ? 0.2 2.7 ? 0.1 2.9 2.7-3.2 3.2-3.7 2.6-3.3 2.3-3.0 2.6-2.8 (24) (17) (21) (70) (6) HAC 3.9 ? 0.2 4.6 ? 0.1 3.8 ? 0.2 3.6, 3.6, 3.8 3.7 ? 0.2 3.7 ? 0.1 4.1 3.4-4.4 4.4-4.8 3.5^1.3 3.3-4.1 3.6-3.8 (24) (17) (21) (70) (6) BAC 3.1 ?0.1 3.6 ? 0.1 3.0 ?0.1 3.0, 2.8, 2.8 2.9 ?0.1 3.0 ?0.1 3.1 2.8-3.3 3.4-3.8 2.6-3.2 2.6-3.2 2.9-3.1 (24) (17) (21) (70) (6) AC3 4.8 ? 0.2 5.8 ? 0.2 4.6 ? 0.2 4.3, 4.3, 4.4 4.5 ? 0.2 4.4 ? 0.1 4.9 4.4-5.3 5.3-6.3 4.3^4.9 3.8-5.1 4.3-4.6 (24) (17) (21) (70) (6) TRD 5.5 ? 0.2 6.9 ?0.1 5.7 ? 0.2 5.5, 5.3, 5.7 5.6 ? 0.3 5.5 ? 0.1 6.0 5.3-5.8 6.6-7.1 5.3-5.9 5.1-6.2 5.4-5.7 (24) (17) (21) (70) (6) mlL 1.7 ? 0.1 2.2 ? 0.1 1.8 ? 0.1 1.7, 1.7, 1.7 1.7 ? 0.1 1.8 ?0.1 1.9 1.6-1.9 2.1-2.4 1.7-1.9 1.6-1.9 1.7-1.8 (24) (17) (21) (70) (6) factor 2 scores. Cryptotis merriami has a relatively longer mandible and tooth row and a shorter cor- onoid process than C. mayensis and the specimens from Guerrero. Discriminant function analysis of crania failed to distinguish effectively among the three groups of shrews. This analysis utilized a three-variable model (WM1, M2B, PL), which correctly identi- fied fewer than 70% of specimens using a jack- knifed classification. A plot of specimens on ca- nonical axes 1 and 2 (fig. 8) shows some distinction between C. merriami and the other two groups, but there is broad overlap between C. mayensis and the specimens from Guerrero. The pattern of linear groupings in this plot is due to one variable, WM1, which loaded out equally on each axis and for which there were only five states. These qual- ities made it act as a discrete variable rather than as a continuous variable. However, this does not affect the validity of the analysis nor the conclu- sion that C. mayensis from the Yucatan Peninsula and from Guerrero are difficult to distinguish. The discriminant function analysis of mandi- bles utilized four variables (ML, HCP, HAC, BAC), which correctly classified 96% of C. merriami, 80% of C. mayensis from the Yucatan, and all of the specimens from Guerrero. Intermixing between the two described taxa was minimal; only one C. merriami and one C. mayensis were incorrectly identified as the other taxon. However, three C. mayensis were misclassified as belonging to the group of specimens from Guerrero. A plot of spec- imens on canonical axes 1 and 2 shows clustering of individuals belonging to each of the three groups but not complete separation among groups (fig. 9). Cryptotis merriami can be distinguished from the other two groups along axis 1, whereas C. mayensis and the specimens from Guerrero cluster sepa- rately along axis 2. Although there is some indication that C. may- TABLE 5. Factor loadings from PCA of cranial vari- ables of C. mayensis from the Yucatan Peninsula and Guerrero and C. merriami. Variable Factor 1 Factor 2 Factor 3 MTR 0.892 -0.099 0.096 M2B 0.869 0.047 -0.311 WM1 0.839 -0.233 -0.162 IO 0.828 0.003 -0.180 PL 0.681 0.071 0.708 ZP 0.198 0.970 -0.065 WOODMAN & TIMM: SMALL-EARED SHREWS 13 2 - 1 - V V V o - V V V o -t-J o o -1 - ? ? ? ? ? -2 - O Guerrero ? ? C. mayensis v C. merriami i i -4-3-2-10 1 2 3 factor 1 FIG. 7. Plot of specimens of C. mayensis from Guer- rero and the Yucatan Peninsula and C. merriami on factor axes 1 and 2 from PCA of mandibular variables. ensis and the shrews from Guerrero exhibit some differentiation, they are very similar, and at this time it seems best to refer the specimens from Guerrero to C. mayensis. We do this tentatively, however, and we believe that once more complete specimens are available the shrews from Guerrero will prove to be specifically distinct from C. may- ensis. Alternatively, the shrews from Guerrero may represent a disjunct population of C. mayensis. This is supported by another species found with the shrews in the owl pellets from Macuiltzingo Cave, Ototylomys phyllotis, which otherwise is un- known north of the Isthmus of Tehuantepec. The specimens from Guerrero are indistinguishable from O. p. phyllotis from the Yucatan Peninsula, which in turn are easy to separate from O. p. con- nectens in Chiapas (M. D. Engstrom, in litt.). Because of possible geographic variation, and because of their fragmentary nature, specimens TABLE 6. Factor loadings from PCA of mandibular variables of C. mayensis from the Yucatan Peninsula and Guerrero and C. merriami. Variable Factor 1 Factor 2 HAC 0.873 -0.265 HCV 0.817 -0.005 BAC 0.814 -0.144 AC3 0.723 -0.290 HCP 0.701 -0.529 ML 0.672 0.560 TRD 0.383 0.813 mlL 0.285 0.858 ?o / W