UNIVERSITY OF ILLINOIS LIBRARY AT URBAiMA-CHAMPAIGN BIOLOGY MP DDDD > CO o > O — 1 O CO co o c: n: cz 3 co m co ^ ZJ CO -1 2 m co 2 o o < 2 F* DDDD h r c A H 3 o tr qd rj U • ^ H Q CO 0 C* 0 a. o DD □□DDDD U^ Ifl H Hd Oh- ■4 3 H 0 *] be o o o •1 l\3 CO »r. til IT. a 3 T M, u ■ _ (O < U • ■ CD 4 I U.' Ul ui tn CTi O in u> CD CD CD I I to li.i ID cn "ST 8 f J M r» n H 0 0. H ■fl- ID U CO 0 TU ridfiici H CO 3- H- 1 II 3 (t ■£" I- QD Zoology 590 .5 FI N.S. navi J*1?"1"'1 -J-SlQ BIX graphic Variation and wy Ktionships Among Broad-Clawed ws of the Cryptoti dmani-G roup immalia: Insectivora: Soricidat & 999 197 T : i > LISHED BY F . I i^WACserr:* m FIELDIANA Zoology JEW SERIES, NO. 91 eographic Variation and Evolutionary Relationships Among Broad-Clawed Shrews of the Cryptotis goldmani-Group Mammalia: Insectivora: Soricidae) s ea I Woodman lepartment of Biological Sciences ast Stroudsburg University ast Stroudsburg, Pennsylvania 18301 Robert M. Timm Natural History Museum and Department of Systematics and Ecology University of Kansas awrence, Kansas 66045-2454 accepted August 7, 1997 Published January 13, 1999 Publication 1497 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY © 1999 Field Museum of Natural History ISSN 0015-0754 PRINTED IN THE UNITED STATES OF AMERICA Table of Contents Abstract 1 Resumen 1 Introduction 2 Methods 3 Systematic Biology 5 "Cryptotis mexicana-group" Choate, 1970 .... 5 "Cryptotis goldmani-group" 5 Cryptotis goodwini Jackson, 1933 8 Cryptotis goodwini goodwini Jackson, 1933 8 Cryptotis goodwini magnimana, new subspecies 11 Cryptotis griseoventris Jackson, 1933 ... 16 Cryptotis goldmani (Merriam, 1895) 19 Cryptotis alticola (Merriam, 1895) 21 Phylogeny of the Cryptotis goldmani- Group 23 Evolution of the Forelimb 25 Zoogeography of the Cryptotis goldmani- Group 30 Key to the Cryptotis coldm an /-Group 31 Acknowledgments 31 Literature Cited 32 Appendix I: Transition Series Used in Phylogenetic Analysis 33 Appendix II: Additional Specimens Exam- ined 34 List of Illustrations 1. Left humerus of Cryptotis goldmani, indicating anatomical features men- tioned in the text 3 2. Measurements of the skull used in this study 4 3. Lateral view of the left zygomatic plates of Cryptotis parva orophila, C. nigrescens, and C. mexicana 7 4. Map of southern Mexico and north- western Central America showing geo- graphic distribution of the Cryptotis goldmani- group 9 5. Three-dimensional plot showing simi- larities and differences among speci- mens of Cryptotis goodwini from Gua- temala and El Salvador 12 6. Dorsal and ventral views of the crani- um and lateral view of the skull of the holotype of Cryptotis goodwini magni- mana, new subspecies 13 7. Plot of the first two factors from prin- cipal components analysis of speci- mens of Cryptotis griseoventris, C. goodwini goodwini, and C g. magni- mana, new subspecies 14 8. Plot of size against geographic locality for specimens of Cryptotis griseoven- tris, C goodwini goodwini, and C. goodwini magnimana, new subspecies ... 16 9. Plot of length of unicuspid toothrow against condylobasal length for Crypto- tis goodwini goodwini, C. goodwini magnimana, new subspecies, and C. griseoventris 17 10. Plot of breadth of zygomatic plate against length of palate for Cryptotis goodwini goodwini, C. goodwini mag- nimana, new subspecies, and C. gri- seoventris 17 1 1 . Plot of width of palate (M2B) against condylobasal length for Cryptotis altico- la, C goldmani, and C. griseoventris 18 12. Lateral views of the orbital areas of the skulls of Cryptotis griseoventris and C. goldmani 20 13. Topologies of the nine shortest trees from phylogenetic analysis of the Cryptotis mexicana-group using 29 morphological transition series 24 14. Topology of the strict consensus tree of the Cryptotis mexicana-group con- structed from the nine shortest-length trees 26 1 5 . Left humeri of selected Cryptotis 27 16. Ventral view of the right forefoot of Cryptotis parva orophila, C. peregrina, C. mexicana, and C. goldmani 28 17. Dorsal views of bones of the left ma- nus of Cryptotis parva orophila, C. peregrina, C mexicana, and C. gold- mani 28 18. Dorsal views of bones of the left pes of Cryptotis parva orophila, C. nigres- cens, C. peregrina, C mexicana, C. al- ticola, and C. goldmani 29 List of Tables 1 . Measurements of Cryptotis used in this study 6 2. Comparisons of characters among taxa in the Cryptotis goldmani-group 8 3. Factor loadings for the first two factor axes from principal components analysis of Cryptotis griseoventris, C. goodwini goodwini, and C. goodwini magnimana, new subspecies 14 4. Correlation matrix for the eight variables used in principal components analysis of Cryptotis griseoventris, C. goodwini goodwini, and C. goodwini magnimana, new subspecies 15 5. Latitude and longitude for collecting sites of Cryptotis griseoventris, C goodwini goodwini, and C. goodwini magnimana, new subspecies Factor loadings for the first three factor axes from principal components analysis of Cryptotis goodwini from Guatemala and El Salvador Complete character matrix for phyloge- netic analysis of members of the Crypto- tis mexicana-group and two outgroups, Cryptotis parva parva and Cryptotis ni- grescens 2 Geographic Variation and Evolutionary Relationships 4mong Broad-Clawed Shrews of the Cryptotis roldmani-Group (Mammalia: Insectivora: Soricidae) Veal Woodman and Robert M. Timm Abstract The Cryptotis goldmani-group of small-eared shrews consists of species that occupy high- elevation (>1000 m) habitats in Mexico and northern Central America. Previously, this group was viewed as consisting of only two species (C. goldmani and Cryptotis goodwini) that were characterized by extreme enlargement of the forefeet and foreclaws. Phylogenetically, C. gold- mani and C goodwini were placed in Choate's (1970) Cryptotis mexicana- group, which oth- erwise consisted of the four subspecies of C. mexicana. Our reevaluation of these shrews indicates that the subspecies of C. mexicana are well-differentiated taxa that we recognize as separate species (C. mexicana, Cryptotis nelsoni, Cryptotis obscura, and Cryptotis peregrina). The C. goldmani-group consists of at least four distinct species: Cryptotis alticola, in Colima, Jalisco, Mexico, Michoacan, Puebla, and the Distrito Federal of Mexico; C. goldmani, in the Sierra Madre del Sur of Guerrero and Oaxaca; C. goodwini, in Chiapas, Guatemala, El Salvador, and Honduras; and Cryptotis griseoventris, in the northern highlands of Chiapas and Guatemala. A single new specimen of C. goodwini indicates the presence of this species in Honduras, which is outside of the previously recognized geographic range of the C. mexicana-group. This new specimen possesses characters suggesting that the Honduran population is phylogenetically distinct from other C. goodwini, and we herein describe a new subspecies for C. goodwini. Our morphological analysis indicates that enlargement of the forefeet and foreclaws is a trend found throughout the C. mexicana-group, reaching its xenith in members of the C. goldmani-group. This trend is accompanied by functional modifications of the forelimb skel- eton. Our phylogenetic analysis, based on 29 transition series, indicates that the C. mexicana- group forms a well-supported clade and justifies nesting of the C. goldmani-group within the C. mexicana-group. Topology of each of the shortest trees shows that the species previously recognized as subspecies of C. mexicana are paraphyletic with respect to the C. goldmani- group. Resumen Las musarafias de orejas pequenas del grupo de especies Cryptotis goldmani ocurren en habitats de altas elevaciones (> 1000 msnm) en Mexico y el norte de Centroamerica. Anter- iormente, se consideraba que este grupo se constaba de solamente dos especies (C. goldmani y Cryptotis goodwini) que se caracterizaban por la agrandacion excesiva de las patas y las unas delanteras. Filogeneticamente C. goldmani y C. goodwini se encontraban en el grupo de es- pecies Cryptotis mexicana sensu Choate (1970), un grupo que ademas incluia solomente C. mexicana con cuatro subespecies. Nuestra revision de estas musaranas indica que las subes- pecies de C. mexicana son taxones bien diferenciados que reconocemos como especies distintas (C. mexicana, Cryptotis nelsoni, Cryptotis obscura, y Cryptotis peregrina). El grupo de especies IELDIANA: ZOOLOGY, N.S., NO. 91, DECEMBER 31, 1998, PP. 1-35 C. goldmani consta de por lo menos cuatro especies distintas: Cryptotis alticola, en Colima, Jalisco, Mexico, Michoacan, Puebla y el Distrito Federal de Mexico; C. goldmani, en la Sierra Madre del Sur en Guerrero y Oaxaca; C. goodwini, en Chiapas, Guatemala, El Salvador, y Honduras; y Cryptotis griseoventris, en las montanas del norte de Chiapas y Guatemala. Un ejemplar unico de C. goodwini muestra la presencia de tal en Honduras, fuera de la distribution anteriormente conocida para el grupo de especies C. mexicana. Este especimen nuevo tiene caracteristicas que sugieren que la poblacion hondurena es filogeneticamente distinta a los demas C goodwini; por lo tanto, se lo describe aqui como una nueva subespecie de C. good- wini. ., Nuestro analisis morfologico muestra que la agrandacion de las patas y las unas delanteras es una tendencia presente en todo el grupo C. mexicana, y que alcanza su desarrollo maximo en las especies del grupo C. goldmani. Tal tendencia se relaciona con modificaciones funcion- ales del esqueleto del miembro delantero. Nuestro analisis filogenetico, basado en 29 caracteres, indica que el grupo C. mexicana es un clado bien definido y justifica la position del grupo C. goldmani dentro del grupo anterior. La topologia de cada uno de los arboles filogeneticos mas cortos demuestra que las especies anteriormente tratadas como subespecies de C. mexicana son parafileticas con respecto al grupo de especies C. goldmani. Introduction Small-eared shrews of the New World genus Cryptotis range geographically from southernmost Canada, through the eastern half of the United States, Mexico, and Central America, to the An- dean highlands of northwestern South America. A large array of names has been proposed for the Mexican and Central American members of the genus; however, the paucity of specimens avail- able has hindered assessment of geographic and nongeographic variation. Choate (1970), in his monograph on the Middle American members of this group, recognized eight species that he ar- ranged into three informal groupings: the "Cryp- totis mexicana- group" (containing three species: Cryptotis goldmani, Cryptotis goodwini, and C. mexicana), the "Cryptotis parva-group" (two species: C. parva and Cryptotis nigrescens), and "relict species" (three species: Cryptotis endersi, Cryptotis gracilis, and Cryptotis magna). Recent- ly, we (Woodman & Timm, 1992) described a ninth species for the region, Cryptotis honduren- sis, from high-elevation pine forest and pine-oak forest of south-central Honduras. In addition, we (Woodman & Timm, 1993) recognized C. nigres- cens, a member of Choate's C. /?arva-group, to be a complex of at least five species distributed from southern Mexico to Colombia. Thus, the genus Cryptotis is considerably more diverse in Central America than previously thought. The three species recognized by Choate (1970) as comprising his C. mexicana- group (C. gold- mani, C. goodwini, and C. mexicana) occur in middle- to high-elevation habitats distributed from; the Mexican state of Tamaulipas to El Salvador. Among these species, C. goldmani and C. good- wini can be distinguished most readily from C mexicana by their greatly enlarged front feet and foreclaws, characters unique among members of the genus. Under Choate's (1970) taxonomy, C. goldmani included two subspecies, C. goldmani^ alticola and C. goldmani goldmani, that werd known to occur from the Mexican state of Jalisco southeast into Guatemala. The monotypic C\ goodwini was known from Guatemala, El Salva- dor, and Chiapas, Mexico. We recently obtained a new specimen of the C. goldmani-group from Honduras, which is outside of the previously known geographic range for these broad-clawed shrews. Our attempts to iden- tify this specimen and adequately document its phylogenetic and biogeographic relationships to other members of the C. goldmani-group led us to reevaluate the taxonomic status of these poorly known shrews. It is the purpose of this paper to review pub- lished knowledge of the broad-clawed shrews of the C. goldmani-group and analyze their phylo- genetic relationships. We redescribe four species described previously (Cryptotis alticola, C. gold- mani, C. goodwini, and Cryptotis griseoventris) and describe a new subspecies for C. goodwini., Our accounts for these species include a review of all available information on reproduction, ele- vational and geographic distributions, and habitat, and we provide a key for their identification. We present diagnostic characteristics of the postcra- nial skeleton of Cryptotis for the first time. FIELDIANA: ZOOLOGY Fig. 1 . Left humerus of Cryptotis goldinani, indicating natomical features mentioned in the text. Abbrevia- pns: CA, capitulum: DP. deltoid process; GT. greater (jberosity; HD, head; LE, lateral epicondyle; ME, me- lial epicondyle; PP, pectoral process; TR, trochlea; and JT. teres tubercle. Anatomical terminology follows Reed 11951). vlethods 1 For the purposes of this paper, we use Cryptotis itexicana-group to refer informally to the set of axa that includes C. mexicana, Cryptotis nelsoni, tryptotis obscura, Cryptotis peregrina, Cryptotis llticola, Cryptotis goldmani, Cryptotis goodwini, nd Cryptotis griseoventris. The Cryptotis gold- jjiam'-group, or broad-clawed shrews, refers to the jjubset of the C. mexicana-group that includes C. llticola, C. goldmani, C. goodwini, and C. gri- seoventris. The Cryptotis nigrescens-group in- cludes Cryptotis colombiana, Cryptotis honduren- tis, Cryptotis mayensis, Cryptotis mera, Cryptotis \ierriami, and Cryptotis nigrescens (Woodman & j'imm, 1993). In the Cryptotis parva-group, we liclude all taxa currently included as subspecies f C. parva (C. parva berlandieri, C p. elasson, l. p. floridana, C. p. orophila, C. p. parva, C. p. aieblensis, soricina, and tropicalis). The C. mex- :ana-group and the C. goldmani-group appear to e natural groupings of species. The C. nigres- ens-group and the C. parva-group may represent lonophyletic groupings of species, but they re- lain informal, awaiting a comprehensive phylo- enetic study of the genus. Terminology of dentition and dental character- ises follows Choate (1970). Anatomical termi- ology of the humerus (Fig. 1) and other parts of le postcranial skeleton follows Reed (1951). leasurements used in our analyses follow Wood- lan and Timm (1993). Measurements of the skull ig. 2) were taken to the nearest 0.1 mm using either an ocular micrometer in a binocular micro- scope or a handheld dial caliper. Standard external measurements are those recorded by the collector, except for head-and-body length (HB), which we calculated by subtracting the recorded tail length (TL) from the total length. Other abbreviations of measurements found in the text include distance from articular condyle to posterior edge of M, (AC3); breadth of articular condyle (BAC); cra- nial breadth (CB); condylobasal length, not in- cluding the upper incisors (CBL); height of artic- ular condyle (HAC); height of coronoid process (HCP); height of coronoid valley (HCV); interorbital breadth (IO); length of lower first mo- lar (M,L); breadth of palate across second molars (M2B); length of molariform toothrow, P4 through M3 (MTR); mandibular length, from inferior sig- moid notch to mental foramen (ML); palatal length (PL); upper toothrow length, U1 through M3 (TR); lower toothrow length, P3 through M, (TRD); breadth of palate across first unicuspids (U'B); breadth of palate across third unicuspids (U3B); length of unicuspid toothrow (UTR); and breadth of zygomatic plate (ZP). All measure- ments are in millimeters. Univariate statistics in- clude mean ± standard deviation. Multivariate analyses and analyses of variance (ANOVA) were carried out using BMDP on the University of Kansas Academic Computing Service IBM VM/CMS computer system and Minitab release 8.0 for DOS-based personal computers. F-values and P-values for ANOVAs (Table 1) were cal- culated using a Brown-Forsythe equality of means test that does not assume equality of the variances. Localities and elevations that we report are taken directly from specimen tags and repre- sent the descriptions provided by the original col- lectors. We provide corrections and additions in brackets. Latitudinal and longitudinal coordinates for specimens used in the analysis of geographic variation of C. goodwini and C. griseoventris are from the U.S. Board on Geographic Names (1956a,b, 1965). Species and subspecies synony- mies list only published uses of names. We attempted to accurately characterize varia- tion in pelage coloration in our descriptions of species; all capitalized color names follow Ridg- way (1912). However, foxing is likely to have oc- curred, expecially in older specimens, and minor variations in pelage color should not be miscon- strued as real differences among species without further testing. In general, pelage in living mem- bers of the C. mexicana-group is dark chocolate- brown or darker in coloration, and it is not pos- /OODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS Fig. 2. Measurements of the skull used in this study (see Table 1). The lines represent the orientation of the measurements. Abbreviations: AC3, distance from articular condyle to posterior edge of M3; BAC, breadth of articular condyle; CB, cranial breadth; CBL, condylobasal length, not including the upper incisors; HAC, height of articular condyle; HCP, height of coronoid process; HCV, height of coronoid valley; IO, interorbital breadth; mlL, length of lower first molar; M2B, breadth of palate across second molars; MTR, length of molariform toothrow, P4 through M3; ML, mandibular length, from inferior sigmoid notch to mental foramen; PL, palatal length; TR, upper toothrow length, U1 through M3; TRD, lower toothrow length, P, through M,; U1B, breadth of palate across first unicuspids; U3B, breadth of palate across third unicuspids; UTR, length of unicuspid toothrow; ZP, breadth of zygomatic plate. Measurements follow those of Woodman and Timm (1993). sible to distinguish them on the basis of pelage alone. Previous investigations of sexual dimorphism in Cryptotis (Choate, 1970 — C. mexicana; Wood- man, 1992 — Cryptotis gracilis; Woodman & Timm, 1993 — C nigrescens) revealed no clear pattern of morphological differences between males and females. We had insufficient sample sizes of complete individuals of both sexes to car- ry out adequate analyses of secondary sexual vari- ation for any one species. Based on the lack of obvious sexual variation in these or other mem- bers of the genus, we assumed that there was no confounding sexual variation in our analyses. Therefore, males, females, and specimens of un- known sex were combined. In our investigations of the C. goldmani-group, we operated under the philosophical influence of the evolutionary species concept as redefined by Wiley (1978, 1981). This concept requires that the species category (and taxonomy in general) be logically consistent with reconstructed phyloge- netic history. In application, we considered a spe- cies as the largest monophyletic entity whose con- stituent parts interact and are not on different evo- lutionary trajectories (Alternative 2 of Frost & Hillis, 1990). Operationally, we used unique dis- tributional patterns of morphological characters among populations to distinguish presumably ge- netically cohesive groups from other genetically cohesive groups. This results in a clear, support- able, testable framework for further phylogenetic, biogeographic, and ecological studies. Similarly, we consider the subspecies category to represent a phylogenetically distinct entity; in practice, we use it to separate poorly defined taxa that require further taxonomic study. Phylogeny and character evolution in the C. goldmani-group were analyzed using PAUP 3.1.1 (see Swofford, 1993) and MacClade 3.0 (see Maddison & Maddison, 1992). Phylogenetic anal- ysis was carried out using an exhaustive search of 29 unordered transition series (TS; see Appendix I). In addition to the four species in the C. gold- mani-group, we included the four taxa previously treated as subspecies of Cryptotis mexicana (C. mexicana, C. nelsoni, C obscura, and C pere- grind). One or more of these taxa should represent the first outgroup to the C goldmani-group, and we thought that the C goldmani-group might FIELDIANA: ZOOLOGY >rove to be paraphyletic with respect to one or nore of these taxa. As outgroups, we chose C. yarva parva from Kansas and C. nigrescens from Dosta Rica because they were considered previ- >usly to be phylogenetically distinct from the C. nexicana-group (Choate, 1970), and large series )f specimens of those taxa were readily available o us. Specimens from the following institutions were lsed in this study: American Museum of Natural iistory, New York (amnh); Angelo State Univer- ity Natural History Collection, San Angelo asnhc); California Academy of Sciences, San -rancisco (cas); Escuela Nacional de Ciencias iologicas, Mexico City (encb); Field Museum, Chicago (fmnh); Instituto de Biologia, Universi- lad Nacional Autonoma de Mexico, Mexico City ibunam); Instituto Nacional de Biodiversidad, Janto Domingo de Heredia, Costa Rica (iNBio); Jniversity of Kansas Natural History Museum, ^awrence (ku); Natural History Museum of Los Vngeles County, Los Angeles (lacm); Museum of Comparative Zoology, Cambridge (mcz); James brd Bell Museum of Natural History, St. Paul mmnh); Museo Nacional de Costa Rica, San Jose mncr); Museo de Zoologia, Facultad de Ciencias, Jniversidad Nacional Autonoma de Mexico, Mexico City (mzfc); Forschungsinstitut und Na- urmuseum Senckenberg, Frankfurt am Main smf); Texas Cooperative Wildlife Collection, College Station (tcwc); University of Michigan Museum of Zoology, Ann Arbor (ummz); and Na- ional Museum of Natural History, Washington usnm). Systematic Biology Based on elevational and geographic distribu- ions of specimens and the correlation of character tates to these elevational and geographic distri- •utions, we recognize four distinct species within he C. goldmani-group. These are C. goodwini, C. ".oldmani, and two species previously synony- nized with C. goldmani (C. alticola and C. gri- eoventris). In addition, we recognize C. goodwini -s a polymorphic species consisting of C. g. good- vini and a new subspecies that we describe below. Family Soricidae von Waldheim, 1817 Subfamily Soricinae von Waldheim, 1817 Genus Cryptotis Pomel, 1848 "Cryptotis mexicana -group" Choate, 1970 Description — These shrews are small to me- dium-sized members of the genus with long, dark dorsal pelage. They are characterized by variably broadened forefeet; moderately elongate and moderately broad to greatly elongate and very broad foreclaws; posterior border of the zygo- matic plate usually positioned equal to, or slightly posterior to, the posterior base of the maxillary process (Fig. 3C); upper toothrow uncrowded; dentition not bulbous; anterior border of the cor- onoid process of the mandible joins the horizontal ramus at a relatively low angle; posterior border of lower incisor extends to posterior cingulum of P4; relatively long distance from the coronoid pro- cess to the posterior border of M3; tall, wide ar- ticular face of the articular process; deep lower sigmoid notch; relatively long, low P,; relatively short, broad metacarpals; shortened and broad- ened humerus with enlongated processes and a dorsoventrally elongate head; and deeply pocket- ed posterior edge of the falciform process of the tibia. Included Taxa — Cryptotis alticola, C. gold- mani, C. goodwini, C. griseoventris, C. mexicana, C. nelsoni, C. obscura, and C. peregrina. "Cryptotis goldmani-group" Description — A subset of the C. mexicana- group, the C. goldmani-group consists of medi- um-sized members of the genus with relatively short tails (mean TL < 39% of HB— Tables 1 and 2); greatly broadened forefeet; extremely long, broad foreclaws; fourth upper unicuspid usually aligned with the unicuspid toothrow and partially visible in labial view; protoconal basin of M1 re- duced relative to hypoconal basin; M1 simple, hy- pocone absent or poorly developed and lacking metacone; relatively low coronoid process of the mandible; entoconid of M, vestigial or absent; and extremely broad humerus, with greatly enlongated processes. Included Taxa — Cryptotis alticola, C. gold- mani, C. goodwini, and C. griseoventris. VOODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS Table 1. Measurements of Cryptotis used in this study. Statistics presented are mean ± standard deviation of the mean and observed extremes. Sample sizes (n) are different for skin measurements, skull measurements, and weight; sample sizes for cranial breadth are different than those for other variables of the skull. The F-value and P- value for each variable are from an analysis of variance (ANOVA) of the variables from C. goldmani and the two species previously considered conspecific with it (C. alticola and C. griseoventris). Cg. goodwini Cg. griseoventris C. alticola C. goldmani C. g. goodwini (El Salvador) magnimana (holotype) C. External measurements n = 22 n = 28 n = 34 n = 1 n = 1 n = 27 Head-and-body length 79 ± 5 76 ± 5 84 ± 5 80 77 ± 3 69-87 62-84 k 75-94 73-85 Tail length 26 ± 2 29 ± 3 29 ± 2 25 29 ± 1 23-30 24-36 25-34 27-32 Skull measurements n = 16 n = 16 n = 20 Condylobasal length (F = 6.53, P = 0.003) 20.2 ± 0.5 19.6 ±0.5 21.1 ± 0.7 19.3-21.1 18.9-20.5 20.0-22.8 Cranial breadth 10.4 ± 0.2 10.2 ±0.2 11.1 ± 0.3 9.9-10.8 9.8-10.5 10.6-11.6 (n = 15) (n = 18) (n = 15) Breadth of zygomatic plate (F = 1.91, P = 0.164) 1.9 ± 0.1 1.8 ± 0.2 1.9 ± 0.1 1.7-2.1 1.4-2.2 1.6-2.2 Interorbital breadth (F = 4.92, P = 0.012) 4.9 ± 0.2 5.0 ± 0.2 5.6 ± 0.2 4.7-5.3 4.7-5.3 5.3-5.8 Breadth across first unicuspids (F = 31.53, P = 0.0000) 2.7 ± 0.1 2.6 ± 0.1 2.7 ± 0.1 2.5-2.8 2.4-2.7 2.6-2.9 Breadth across third unicuspids (F = 19.51, P = 0.0000) 3.2 ± 0.1 3.0 ± 0.1 3.3 ± 0.1 3.0-3.2 2.8-3.2 3.0-3.5 Breadth across second molars (F = 95.12, P = 0.0000) 6.2 ± 0.2 5.8 ± 0.1 6.2 ± 0.2 5.8-6.4 5.6-5.8 6.0-6.6 Palatal length (F = 0.20, P = 0.816) 8.7 ± 0.3 8.7 ± 0.3 9.2 ± 0.3 8.1-9.2 8.0-9.2 8.8-10.1 Length of upper toothrow (F = 4. 72, P = 0.014) 7.6 ± 0.2 7.4 ± 0.2 7.9 ± 0.3 7.1-7.9 7.0-7.8 7.5-8.6 Length of unicuspid toothrow (F = 55.19, P = 0.0000) 2.5 ± 0.1 2.3 ± 0.1 2.7 ± 0.1 2.3-2.7 2.0-2.5 2.5-2.9 Length of molariform toothrow (F = 17.71, P = 0.0000) 5.5 ± 0.2 5.4 ± 0.1 5.6 ± 0.2 5.1-5.7 5.3-5.7 5.3-5.9 Length of mandible (F = 14.35, P = 0.0000) 6.6 ± 0.2 6.3 ± 0.3 6.6 ± 0.2 5.9-6.8 5.8-6.8 6.2-6.8 Height of coronoid process (F = 16.35, P = 0.0000) 4.6 ± 0.2 4.6 ± 0.2 4.8 ± 0.1 4.3-4.9 4.3-4.9 4.6-5.2 20.8 10.8 5.6 2.8 3.3 6.2 9.1 7.8 2.5 5.7 4.8 20.3 10.8 2.2 5.3 2.6 3.0 5.9 8.8 7.5 2.4 5.5 6.5 4.7 n = 20 19.9 ± 0.4 18.8- -20.4 10.2 ± 0.2 9.8- -10.7 (n = 19) 1.9 ± 0.1 1.6-2.1 5.1 ± 0.2 4.8- -5.3 2.5 ± 0.1 2.3- -2.6 3.0 ± 0.1 2.8- -3.2 5.6 ± 0.1 5.4- -5.9 8.7 ± 0.2 8.1- -9.0 7.6 ± 0.2 7.3- -7.8 2.7 ± 0.1 2.5- -2.9 5.2 ± 0.1 5.0- -5.4 6.2 ± 0.2 5.7- -6.4 4.4 ± 0.1 4.2- -4.6 FIELDIANA: ZOOLOGY Table 1. Continued. C. alticola C. goldmani C g. goodwini e.g. goodwini (El Salvador) e.g. magnimana (holotype) C. griseoventris Height of coronoid valley {F - 2.9 ± 0.1 2.8-3.1 10.75. P = 0.0001) 2.8 ± 0.1 2.7-3.0 3.0 ± 0.1 2.8-3.4 2.8 2.9 2.8 ± 0.1 2.6-3.0 Height of articular condyle (F = 4.1 ± 0.2 3.7-4.3 = 9.22, P = 0.0005) 3.9 ± 0.2 3.8-4.2 4.2 ± 0.2 3.8-4.6 4.1 4.1 3.9 ± 0.1 3.7-4.1 Breadth of articular condyle (F 3.3 ± 0.1 3.0-3.5 = 31.46. P = 0.0000) 3.1 ± 0.1 2.9-3.2 3.3 ± 0.2 3.0-3.6 3.3 3.1 3.0 ± 0.1 2.8-3. 1 Articular condyle to M, (F = 3.64, P = 0.035) 5.2 ± 0.2 5.1 ± 0.2 4.8-5.6 4.8-5.3 5.6 ± 0.2 5.2-5.9 5.6 5.3 5.1 ± 0.1 4.8-5.3 Length of lower toothrow (F = 6.2 ± 0.2 5.8-6.5 1.11, P = 0.341) 6.1 ± 0.2 5.4-6.4 6.4 ± 0.2 6.1-6.8 6.5 5.9 6.1 ± 0.1 5.8-6.3 Length of M, 1.8 ± 0.1 1.7-2.0 1.8 ± 0.1 1.7-2.0 1.9 ± 0.1 1.8-2.0 2.0 1.8 1.8 ± 0.1 1.7-1.9 Weight (g): 11 ± 3 8-16 (n = 10) 8 ± 1 6-10 (n = 15) 16 ± 2 12-19 (n = 10) — — — I 1 1 mm zygomatic plate Fig. 3. Lateral views of the left zygomatic plates of A, Cryptotis parva orophila; B, C. nigrescens; and C, C. mexicana. The arrangement of the zygomatic plate in C. mexicana, in which the posterior border is even with and confluent with the posterior root of the maxillary process, is representative of the C. mexicana-group. Among members of the C. nigrescens-group, the posterior border of the zygomatic plate and the posterior root of the maxillary process are separated by the posterior edge of the palate. Scale bar = 1 mm. WOODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS Table 2. Comparisons of characters among taxa in the Cryptotis goldmani-group. Relative measurements are percentages. Sample sizes are presented in accounts for each species. Abbreviations as in Methods section of the text. C. alticola C. goldmani C. g. goodwini C. g. magnimana C. griseoventris Foramen of sinus canal vestigial well developed absent absent absent Foramen dorsal to dorsal articular facet present 39% 16% 90% present 81% Two dorsal foramina present 54% 61% 82% present 92% Posteroventral border of unicuspids straight concave concave concave concave to convex Vestigial entoconid of M, present 64% 52% 14% ? 0% Humerus derived most derived derived derived ? (Fig. 15E) (Fig. 15H) (Fig. 15G) (Fig. 15F) Tail length as % of head-and-body length (TL/HB X 100) 33 ± 4 38 ± 5 35 ± 3 31 38 ± 2 27-41 31-48 30-41 34-42 Relative length of rostrum (PL/CBL X 100) 43.2 ± 1.2 44.2 ± 1.0 43.6 ± 0.7 43.3 43.8 ± 0.8 40.1-45.5 41.9-45.6 42.5-44.8 42.1-45.0 Relative breadth of zygomatic plate (ZP/PL X 100) 21.8 ± 1.2 21.1 ± 2.4 21.1 ± 1.9 25.0 22.3 ± 1.7 19.6-24.1 15.7-23.4 17.2-24.4 18.2-24.7 Relative length of unicuspid toothrow (UTR/CBL X 100) 12.3 ± 0.4 11.9 ± 0.6 13.0 ± 0.5 11.8 13.7 ± 0.5 11.6-13.0 10.5-13.0 12.0-14.0 12.9-14.5 Relative palatal breadth (M2B/PL X 100) 70.9 ± 3.0 66.4 ± 2.3 67.6 ± 2.4 67.1 64.3 ± 2.1 64.1-77.8 63.0-71.2 62.4-71.6 60.7-68.2 Relative height of coronoid process (HCP/ML X 100) 70.5 ± 3.3 73.6 ± 4.4 72.7 ± 2.2 72.3 70.5 ± 1.8 64.2-80.0 66.2-81.0 68.7-76.5 67.2-73.7 Relative length of posterior portion of the mandible (AC3/ML X 100) 79.3 ± 2.0 80.5 ± 2.6 84.2 ± 2.8 81.5 82.9 ± 2.3 76.2-83.6 76.1-84.5 79.1-90.5 79.7-86.9 Cryptotis goodwini Jackson, 1933 Cryptotis goodwini goodwini Jackson, 1933 Cryptotis goodwini Jackson, 1933:81; Goodwin, 1934:6; Felten, 1958:218; Hall and Kelson, 1959:61; Genoways and Choate, 1967:204; Choate, 1970:249; Hall, 1981:60; Medellin, 1988:84. Cryptotis nigrescens: Burt and Stirton, 1961:21 (in part). C[ryptotis]. goodwini: Musser, 1964:7. Holotype — Skin and skull of adult male, usnm 77074; taken 13 January 1896 by E. W. Nelson (collector number 9073); from "Calel" (Jackson 1933:81), Quezaltenango, Guatemala, 10,200 ft. Distribution — Contiguous highland regions above 1100 m in the Mexican state of Chiapas and Guatemala, and in a separate highland area in northern El Salvador. The population in El Sal- vador probably extends northward along the Mer- endon Cordillera into western Honduras (Fig. 4). Description — Size large for the genus (Table 1); tail short, averaging 29 mm, or ca. 35% (Table, 2) of HB; the dorsal guard hairs 6-8 mm long; dorsal pelage Saccardo's Umber/Prout's Brown/ Mummy Brown; ventral pelage somewhat paler than dorsum, Light Drab/Drab to almost Wood Brown; dorsal fur three-banded: basal five-sixths FIELDIANA: ZOOLOGY WOODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS of hairs silvery gray, followed by a thin, very pale brown band grading into dark brown tip. Rostrum of moderate length (PL/CBL = 43.6% ± 0.7, n = 20); usually two well-developed dorsal foram- ina (82%, n = 28); no ventral extension of the sinus canal or associated foramen posterior to dor- sal articular facet (see account for C. goldmani, below); a foramen dorsal to the dorsal articular facet usually present on one or both sides of the skull (90%, n = 20); zygomatic plate of moderate breadth (ZP/PL = 21.1% ± 1.9, n = 20), anterior border usually aligned with mesostyle-metastyle valley or metastyle of M1; P4, M1, and M2 slightly to moderately recessed on posterior border; M3 usually with paracrista, paracone, precentrocrista, mesostyle, very short postcentrocrista (all pig- mented), and a well-developed, pigmented proto- cone. Mandible relatively long and of moderate breadth for the genus; articular process generally moderately tall and wide, with a moderately broad lower articular facet; entoconid usually absent (86%, n = 29), but when present, poorly devel- oped. Comparisons — Cryptotis goodwini goodwini is the largest of the broad-clawed shrews in overall body size. Cryptotis alticola — Cryptotis g. goodwini has somewhat shorter, narrower foreclaws; lacks ven- tral extension of sinus canal and associated fora- men posterior to dorsal articular facet; has a fo- ramen dorsal to dorsal articular facet; has a rela- tively narrower palate; and has a shorter mandi- ble. Cryptotis goldmani — Cryptotis g. goodwini has a relatively shorter tail; lacks ventral extension of sinus canal and associated foramen posterior to dorsal articular facet; has a foramen developed dorsal to the dorsal articular facet; usually lacks an entoconid on M3; and has a longer, narrower humerus with less enlarged bony processes. Cryptotis griseoventris — Cryptotis g. goodwini is larger overall. Remarks — The known biology of Cryptotis goodwini goodwini was summarized previously by Choate (1970) and Choate and Fleharty (1974). Cryptotis goodwini goodwini has been found in high-elevation pine forests, sometimes mixed with firs, oaks, and other trees. Severe frosts are as- sociated with at least some of their habitats (Gold- man, 1951). Most specimens are from localities of 1200 m or more in elevation. Although Choate (1970:251) reported a specimen from Finca Xic- acao, Guatemala, as from ca. 3000 ft, no elevation is given on the specimen tag for this specimen, and Xicacao is located in a region higher than 1000 m. James W. Bee (ku unpublished field cat- alog, 1954, 1955) recorded taking Sorex verae- pacis, Microtus guatemalensis, Peromyscus gua- temalensis, Reithrodontomys microdon, Reithro- dontomys sumichrasti, and Reithrodontomys ten- uirostris along with C. g. goodwini 3.5 mi SW of San Juan Ixcoy, Guatemala, at 10,120 ft, on 27 December 1954. At 6000 ft, 5 mi N and 1 mi W of El Choi, Guatemala, on 30 January 1955, he captured C. g. goodwini with Oligoryzomys ful- vescens, Peromyscus aztecus, Peromyscus levipes, Reithrodontomys fulvescens, Reithrodontomys mexicanus, R. sumichrasti, and Scotinomy tegui- na. Reproductive data for Cryptotis goodwini goodwini are lacking. Our inspection of study skins of males taken from 4 to 15 January (n = 6), 5-22 May (n = 2), 20 August (n = 3), and 27 December (n = 1) yielded none with well-devel- oped lateral glands. On study skins, lateral glands typically appear as paired, oval regions approxi- mately 6-9 mm in length and 5-6 mm in width. These glandular areas lack long guard hairs and normal underfur, but they have a sparse covering of short, fine, pale hairs. Among most genera of Soricidae, both males and females possess lateral glands, but those of females are smaller and more difficult to see (Murariu, 1976; Bee et al., 1980). Eadie (1938) found that the lateral glands of Blar- ina showed increased activity with enlargement of the testes, and they may serve a function in sexual and social communication. The occurrence of en- larged lateral glands on study skins has been used to infer timing of sexual activity in adult male Cryptotis (Woodman & Timm, 1993). The occurrence of Cryptotis goodwini goodwini in El Salvador is documented by a single speci- men from Hacienda Montecristo on Cerro Mon- tecristo (Felten, 1958; Choate, 1970). Cerro Mon- tecristo is an isolated mountain, separated from the Guatemalan Highlands by a deep, wide valley occupied by tributaries of the Rio Motagua. This lowland valley supports xeric vegetation (Stuart, 1954), unlike that found in cool, moist highlands generally inhabited by C. g. goodwini, and it probably provides an effective barrier to gene flow. Because of this, we hypothesized that the specimen from El Salvador might prove to rep- resent a distinct species. This individual has less emarginate upper dentition than C. g. goodwini, but otherwise it cannot be separated from them on qualitative characteristics. To test the overall similarity of the specimen from El Salvador to 10 FIELDIANA: ZOOLOGY those from Guatemala, we carried out a principal components analysis using seven uncorrelated variables (CBL, ZP, 10, M2B, UTR, MTR, and HCP) from 20 Guatemalan C g. goodwini, plus the specimen from El Salvador. A plot of the first three factor axes from this analysis is shown in Figure 5. The specimen from El Salvador shows up in the centers of factor axis 1 (size; see Table 6) and factor axis 2 (UTR and HCP), but it rep- resents the lower extreme of variation for factor 3 (ZP and UTR). This is due to the individual's relatively narrow zygomatic plate and its relative- ly short unicuspid toothrow (Table 1). The spec- imen from El Salvador falls within the overall range of variation for C. g. goodwini, and, lacking any definitive characters that would set it off from that species, we consider it most appropriate to continue to refer it to C. g. goodwini. Specimens Examined (36)— EL SALVADOR: SANTA ANA: Hacienda [Finca] Montecristo [13°40'N, 89°29'W] (smf 14837). GUATEMA- LA: ALTA VERAPAZ: Finca Xicacao (ummz 87869). BAJA VERAPAZ: 5 mi N, 1 mi W El Choi, 6000 ft (KU 64611). CHIMALTENANGO: Santa Elena [14°48'N, 91°01'W], 9900-10,000 ft (fmnh 41791-41794). HUEHUETENANGO: 3.5 mi SW San Juan Ixcoy [15°36'N, 91°27'W], 10,120 ft (ku 64610); San Mateo Ixtatan, ca. 4 km NW Santa Eulalia [15°45'N, 91°29'W], Yayquich, 2950 m (ummz 117843); Hacienda Chancol, 9500-1 1,000 ft (usnm 77069). JALAPA: Mataquescuintla [14°32'N, 90°11'W], 8400 ft (usnm 275681). QUEZALTENANGO: Calel [15°04'N, 91°34'W], 10,200 ft (usnm 77070, 77072-77073, 77075-77084, includes holotype); Volcan Santa Maria [14°45'N, 91°33'W], 9000- 11,000 ft (usnm 77086-77087). SAN MARCOS: S slope Volcan Tajumulco, 10,000 ft (ummz 99541). TOTONICAPAN: Cumbre Maria Teciin [14°52'N, 91°13'W], 3000 m (ummz 112004- 1 1201 1). MEXICO: CHIAPAS: 17 km SE Finca Prusia, Reserva Ecologica El Triunfo, 2000 m (ibunam 22784). Additional Records— GUATEMALA: CHI- MALTENANGO: Tecpam [= Tecpan], 9700 ft (Goodwin, 1934); SAN MARCOS: Finca La Paz, 1200 m (Choate, 1970:251). Cryptotis goodwini magnimana, new subspecies Honduran Broad-clawed Shrew (Fig. 6) Holotype — Fluid-preserved pregnant adult fe- male with skull and humerus removed and cleaned, ku 144611; collected 21 November 1991 by Peter Holm and Gustavo Cruz (no field num- ber). Fluid-preserved body intact; skull nearly complete, but with a hole in braincase dorsally and lacking left tympanic. Type Locality — 2.5 km N, 1.6 km E Cerro San Juanillo [14°30'N, 87°53'W], Reserva Biologica Cordillera de Montecillos, Comayagua Depart- ment, Honduras, 1730 m (see Fig. 4). The only known specimen was found dead in mixed pine and broadleaf forest. Distribution — Known only from the type lo- cality; probably occurs throughout the Cordillera de Montecillos and adjoining highland areas in high-elevation (> 1000 m) pine forest and mixed pine and broadleaf forest (Fig. 4). Measurements of Holotype — HB, 80; TL, 25; hindfoot, 14; ear, 6; CBL, 20.3; PL, 8.8; M2B, 5.9; IO, 5.3; TR, 7.5; CB, 10.8; TRD, 5.9 (see Table 1). Because no external measurements were re- corded by the collectors, we measured the fluid- preserved body prior to removal of the skull. Diagnosis — A taxon possessing all of the com- mon characteristics of the Cryptotis goldmani- group (see above). Within this group, the holotype is characterized by its relatively short tail; rela- tively broad zygomatic plate; lack of a ventral ex- tension of the sinus canal; a well-developed fo- ramen dorsal to the dorsal articular facet; two well-developed dorsal foramina (= foramina or- bitalia of Biihler, 1964); relatively short upper unicuspid toothrow; unicuspid teeth with concave posteroventral border; P4, M1, and M2 very slight- ly recessed on posterior border; and a short, broad humerus similar in grade to that of Cryptotis g. goodwini. Description — Size medium for the genus (Ta- ble 1); tail short, 25 mm (31% of HB— see Table 2); dorsal guard hairs about 5-7 mm long; dorsal pelage of fluid-preserved holotype Mummy Brown to Fuscous when air-dried; Brownish Olive venter appears to be somewhat paler than dorsum; dorsal fur two-banded, basal three-fourths silvery gray, distal one-fourth grading from pale brown to brown at the tip. Rostrum of moderate length (PL/CBL = 43.3%); two well-developed dorsal foramina; no ventral extension of the sinus canal or associated foramen posterior to the dorsal ar- ticular facet (see account for C. goldmani, below); well-developed foramina dorsal to the dorsal ar- ticular facet on both sides of the skull; zygomatic plate broad (ZP/PL = 25%), anterior border aligned with metastyle of M1; posterior border of P, M', and M2 only very slightly recessed; M-1 WOODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS 11 Guatemala ) El Salvador W Fig. 5. Three-dimensional plot showing similarities and differences among specimens of Cryptotis goodwini from Guatemala (solid squares) and El Salvador (open circle) based on the first three factors from principal components analysis of seven skull variables. Factor 1 represents size, factor 2 contrasts length of unicuspid toothrow with height of coronoid process, and factor 3 represents length of zygomatic plate and length of unicuspid toothrow (Table 6). has paracrista, paracone, precentrocrista, meso- style, very short postcentrocrista, and well-devel- oped protocone (because of wear, it cannot be de- termined whether protocone was pigmented). Mandible relatively long and of moderate breadth; articular process generally tall and wide, with a broad lower articular facet and a relatively shal- low lingual notch between articular facets; be- cause of wear, it cannot be determined whether an entoconid was present in talonid of M3. Humerus similar to C. g. goodwini (see below). Comparisons — Cryptotis goodwini magnimana has a relatively broader zygomatic plate and a rel- atively shorter unicuspid toothrow compared to those of the other four taxa of broad-clawed shrews. Cryptotis goodwini goodwini — Cryptotis g. magnimana is smaller overall (see Remarks, be- low); has a relatively and absolutely shorter uni- cuspid toothrow; and has less deeply recessed posterior borders of P4, M1, and M2. Cryptotis alticola — Cryptotis g. magnimana has somewhat shorter, narrower foreclaws; lacks ventral extension of sinus canal and associated fo- ramen posterior to dorsal articular facet; possesses a foramen dorsal to posterior dorsal articular facet and two well-developed dorsal foramina; has a relatively and absolutely narrower palate; has con- cave posteroventral borders of unicuspids; and has less deeply recessed posterior edges of P, M1, and M2. Cryptotis goldmani — Cryptotis g. magnimana is larger overall and has a relatively shorter tail; lacks ventral extension of sinus canal and asso- ciated foramen posterior to dorsal articular facet; possesses a foramen dorsal to dorsal articular fac- et and two well-developed dorsal foramina; and has a relatively longer, narrower humerus with less enlarged bony processes (see below). Cryptotis griseoventris — Cryptotis g. magni- mana is larger overall (see Remarks, below); has a relatively and absolutely shorter tail; has a rel- atively and absolutely broader zygomatic plate; and has a relatively and absolutely shorter uni- cuspid toothrow. Remarks — The only known specimen of this subspecies is an adult female that was pregnant when collected on 21 November 1991. She car- ried four embryos, three in the right horn of the 12 FIELDIANA: ZOOLOGY 5 mm Fig. 6. Dorsal and ventral views of the cranium and lateral view of the skull of the holotype of Cryptotis goodwini magnimana. Scale bar = 1 mm. uterus and one in the left (crown to rump length = 8 mm). Cryptotis goodwini magnimana, C. goodwini goodwini, and C. griseoventris are similar in many respects, and the three taxa share several synapomorphies (see below) that distinguish them from C. alticola and C. goldmani. Many of the differences among C. g. magnimana, C. g. good- wini, and C. griseoventris are related to variation in body size or proportions of certain features (Ta- ble 2). To help evaluate the distinctiveness of these three populations, we examined their overall similarity using principal components analysis on eight variables (CBL, ZP, 10, UTR, MTR, HCP, BAC, and M,L) measured on 20 C. griseoventris, 22 C. g. goodwini, and the holotype of C. g. mag- nimana. The specimens of C. griseoventris in- cluded seven from Chiapas and 13 from Guate- mala; those of C. g. goodwini included 21 from Guatemala and one from El Salvador. Each of these subsets was plotted separately to facilitate recognition of geographic variation within and among taxa. A plot of the first two factor axes from the principal components analysis (Fig. 7) shows all three named taxa to be distinct. Six of the eight variables are weighted relatively evenly on the first factor axis (Table 3), suggesting that this axis represents overall size of the individuals. Breadth of zygomatic plate (ZP) and length of un- icuspid toothrow (UTR) are weakly weighted on the first axis, indicating that they do not correlate strongly with size. Correlation coefficients of ZP and UTR with each of the other variables in the analysis fall below 0.24, and many are negative (Table 4). The first factor axis clearly separates the larger C. g. goodwini from the smaller C. gri- seoventris, with the holotype of C. g. magnimana between them. The difference in size between C. g. goodwini and C. griseoventris indicated by the first factor axis is supported by Mests between means of each of the individual variables; of 21 variables tested, only three — length of tail (TL), breadth of zygomatic plate (ZP), and length of unicuspid toothrow (UTR) — did not show statis- tical differences between means for the two taxa. Means of each of the other 1 8 variables were sig- nificantly different (P < 0.001). On the second factor axis, ZP and UTR weigh out heavily, in- WOODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS 13 2 - 1 - CM O -2 -3 - O O o o o o <9 o V ▼ v T V — I — -4 — I — -3 -2 -1 factor 1 C.g. goodwini: O Guatemala • El Salvador C. g. magniTnaTia: a Honduras C. griseoventris: V Chiapas T Guatemala Fig. 7. Plot of the first two factors from principal components analysis of specimens of Cryptotis griseoventris, C. goodwini goodwini, and Cryptotis g. magnimana. Factor 1 represents size; factor 2 contrasts length of zygomatic plate and length of unicuspid toothrow (Table 3). dicating that the axis represents a contrast be- tween these two variables (Table 3). The second factor axis separates the holotype of C. g. mag- nimana from C. g. goodwini and C. griseoventris, reflecting the relatively shorter unicuspid tooth- row and relatively broader zygomatic plate of C. g. magnimana. We subsequently investigated size variation Table 3. Factor loadings for the first two factor axes from principal components analysis of Cryptotis gri- seoventris, C. goodwini goodwini, and C. goodwini mag- nimana. These two axes (PCI, PC2) together accounted for 74% of the variation. Variables are listed in descend- ing order by their loadings on the first factor axis. Ab- breviations as in the Methods section. Variable PCI PC2 ZP 0.052 -0.666 UTR -0.048 0.638 mlL -0.349 -0.263 IO -0.408 -0.075 MTR -0.410 0.145 BAC -0.419 0.056 HCP -0.424 -0.185 CBL -0.427 0.123 among Cryptotis goodwini magnimana, C. g. goodwini, and C. griseoventris in relation to their geographic distributions to determine whether dif- ferences among them might be attributable to clinal variation. Cryptotis g. goodwini and C. gri- seoventris appear to be parapatric; their distribu- tions may overlap on a regional scale, but the two taxa have never been reported from the same site (Fig. 4). The distance between the closest known occurrences of C. g. goodwini and C. griseoven- tris is much less than the dimensions of either of their geographic ranges. If the two populations were members of a single species, we would ex- pect them to grade into each other where their distributions are in closest proximity. To examine this, we determined the approximate latitude and longitude of the collecting locality for each of the 43 specimens used in the principal components analysis of morphology (above) and carried out a second principal components analysis on these pairs of geographic coordinates. Latitude and lon- gitude were evenly weighted (0.707) on the first factor axis; hence each score on that first factor axis represented the coordinates of one collecting locality (Table 5). Plotting the factor scores from 14 FIELDIANA: ZOOLOGY Table 4. Correlation matrix for the eight variables used in principal components analysis of Cryptotis griseo- ventris, C. goodwini goodwini, and C. goodwini magnimana. Abbreviations as in Methods section. CBL ZP IO UTR MTR HCP BAC ZP -0.128 IO 0.715 -0.109 UTR 0.234 -0.173 -0.004 MTR 0.824 -0.187 0.641 0.184 HCP 0.798 0.053 0.830 -0.054 0.738 BAC 0.813 -0.169 0.784 0.055 0.772 0.774 mlL 0.612 0.132 0.623 -0.025 0.602 0.652 0.536 the first factor axis ("size") of the initial analysis of morphological variation against the factor scores from the first factor axis of the analysis of latitude and longitude ("geographic locality") provides a graphical representation of geographic variation in size among the three taxa (Fig. 8). We calculated regressions of size on geographic lo- cality for C. g. goodwini and C. griseoventris to aid in the identification of geographic trends in size. The slope of the regression line was not sig- nificantly different from zero (horizontal) for ei- ther C. g. goodwini (y = -1.83 - 0.024*; F = 0.00, P = 0.956) or C griseoventris (y = 1.72 + 0.208jc; F = 0.80, P = 0.383), indicating that no clear trends in size variation occur within either taxon. Therefore, the difference in body size be- tween these two parapatric populations is not a gradual transition but represents a sharp change. One possible explanation might be that some local Table 5. Latitude and longitude for collecting sites of Cryptotis griseoventris, C. goodwini goodwini, and C. goodwini magnimana. "Geographic locality" is a sin- gle factor score derived from the first factor axis of a principal components analysis of sets of coordinates for collecting sites of the three species. Because latitude and longitude loaded equally (0.707) on this first factor axis, each factor score represents latitude and longitude even- ly at a given site. Geographic locality Latitude Longitude (factor score) 16°45' 92°38' 2.29730 15°36' 91°37' 0.80463 14°48' 9i°or -0.15638 15°36' 91°27' 0.75103 14°52' 91°13' -0.05716 15°45' 91°29' 0.84030 15°04' 91°34' 0.50925 14°45' 91°33' -0.01106 14°32' 9o°ir -0.77840 13°40' 89°29' -2.02087 14°30' 87°53' -2.17865 selective factor maintains this size difference be- tween the two populations. We believe that it is much more likely that the difference in body size has a genetic basis, and C. g. goodwini and C. griseoventris represent two closely related taxa that differentiated in isolation and subsequently expanded their ranges so that they are now in con- tact or near contact. Competition may prevent them from coexisting in the same sites. Variation in Cryptotis goodwini magnimana is impossible to ascertain given a single specimen. The holotype of C. g. magnimana is intermediate in size between C. g. goodwini and C. griseo- ventris, and the population of C. g. magnimana probably partially overlaps one or both of the lat- ter two taxa in this factor. Cryptotis g. magnimana is distinguished morphologically from both C. g. goodwini and C. griseoventris by the combination of its broader zygomatic plate and its shorter un- icuspid toothrow. It is additionally separated from C g. goodwini by the less deeply recessed P4, M1, and M2 (but see below), and from C. griseoventris by its absolutely and relatively shorter tail. We believe that the morphological distinctiveness of the holotype of C. g. magnimana from C. g. good- wini probably represents a phylogenetic distinc- tiveness as well. Unfortunately, the extremely small sample size by which C g. magnimana is currently represented prevents us from adequately testing this hypothesis. Future study of this prob- lem, based on larger sample sizes of the new tax- on, will show that it is (1) a species distinct from C. goodwini; (2) an individual of C. goodwini that shows greater than normal variation for the spe- cies and does not warrant a separate name; or (3) one extreme of a morphological (not size) cline that does not merit a separate name. In support of the last hypothesis is the less deeply recessed up- per dentition and the short unicuspid toothrow (Fig. 9, Tables 1 and 2), two characters that C g. magnimana shares with the specimen of C g. WOODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS 15 geographic locality C.g. goodwini: O Guatemala • El Salvador C . g .Tn.ctg7vima.TiCL'. □ Honduras C. griseoventris: V Chiapas ▼ Guatemala Fig. 8. Plot of size against geographic locality for specimens of Cryptotis griseoventris, C. goodwini goodwini, and Cryptotis g. magnimana. Size is represented by the scores from the first factor axis of a principal components analysis of morphological variation among the three species. Because variables were negatively weighted on the first factor axis (Table 3), larger animals are more negative on the y axis, smaller animals more positive. Geographic locality is represented by scores from the first factor axis of a principal components analysis of latitude and longitude of collecting localities. Localities more to the northwest are more positive on the x axis, those to the southeast are more negative. The regression lines for C. g. goodwini (y = -1.83 - 0.024x; F = 0.00, P = 0.956) and C. griseoventris (y = 1.72 + 0.208x; F = 0.80, P = 0.383) are not significantly different from zero. goodwini from El Salvador. In contrast to C. g. magnimana, however, the individual from El Sal- vador has an extremely narrow zygomatic plate (Fig. 10, Tables 1 and 2). Etymology — The subspecific epithet magni- mana is an adjectival construction (Latin magnus "large" + -i- a connective vowel + manus Table 6. Factor loadings for the first three factor axes from principal components analysis of Cryptotis goodwini from Guatemala and El Salvador. These three axes (PCI, PC2, and PC3) together accounted for 77% of the variation. Abbreviations as in Methods section. Variable PCI PC2 PC3 ZP 0.272 0.388 0.754 UTR -0.204 -0.493 0.613 IO -0.324 0.327 -0.131 HCP -0.401 0.463 0.168 MTR -0.434 -0.350 0.014 M2B -0.460 0.349 0.003 CBL -0.467 -0.202 0.101 "hand") meaning "large-handed." This is in ref- erence to the enlarged forefeet and foreclaws that distinguish this taxon from other shrews in Hon- duras. The subspecific name takes a feminine ter- mination to agree in gender with the feminine ge- neric name, Cryptotis (Woodman, 1993). Specimen Examined (1)— HONDURAS: CO- MAYAGUA: Reserva Biologica Cordillera de Montecillos; 2.5 km N, 1.6 km E Cerro San Juan- illo [14°30'N, 87°53'W], 1730 m (ku 144611— holotype). Cryptotis griseoventris Jackson, 1933 Cryptotis griseoventris Jackson, 1933:80; Hall and Kelson, 1959:60. Cryptotis goldmani goldmani: Choate, 1970:247 (in part); Hall, 1981:59 (in part). Holotype — Skin and skull of adult male, usnm 16 FIELDIANA: ZOOLOGY 5 o .c • V V ■*-J o o -t— ' 2.8 - • • •• • V V V V w V ■a a. CO •• • • — V V V D O c 2.6 - • • • V V V V D O r: • • C. griseoventris en 2.4 - a C. g. magnimana c a C.g.goodwini: O a 6.0 - 5.5 - # C.alticola v C.goldmani T C.griseoventris • • • • • vv V • V, V vv V V V V jyT VTV 19 20 21 condylobasal length Fig. 11. Plot of width of palate (M2B) against condylobasal length (CBL) for Cryptotis alticola, C. goldmani, ant C. griseoventris, all of which were considered previously as conspecific. 75894; taken 4 October 1895 by E. W. Nelson and E. A. Goldman (collector number 8545); from "San Cristobal" (Jackson, 1933:80) [San Cristo- bal de las Casas], Chiapas, Mexico, 9500 ft. Distribution — Above at least 2000 m in con- terminous highland areas of Guatemala and Chia- pas (Fig. 4). Description — Size medium for the genus (Ta- ble 1); tail moderately long, averaging 29 mm, or ca. 38% of HB (Table 2); the dorsal guard hairs about 6-8 mm long; dorsal pelage Mummy Brown to Bister; ventral fur Light Drab to Drab, somewhat paler than dorsum; dorsal fur three- banded: basal five-sixths of hairs silvery gray, fol- lowed by a thin, lightly colored band grading into dark brown at the tip. Rostrum of moderate length (PL/CBL = 43.8% ± 0.8, n = 20); usually two well-developed dorsal foramina (92%, n = 25); no ventral extension of the sinus canal or associ- ated foramen (see account for C. goldmani, be- low); a foramen dorsal to the dorsal articular facet usually present on one or both sides of the skull (81%, n = 27); zygomatic plate of moderate breadth (ZP/PL = 22.3% ± 1.7, n = 20), anterior border usually aligned with mesostyle-metastyle valley, metastyle of M1, or parastyle of M2; upper toothrow uncrowded, U4 aligned with toothrow and partially visible in labial view; P4, M1, and M2 only very slightly recessed on posterior bor- der; M3 usually with paracrista, paracone, precen trocrista, mesostyle, very short postcentrocrist; (all pigmented), and a well-developed, lightly pig mented protocone. Mandible relatively long am of moderate breadth for the genus; articular pro cess tall and wide, with a broad lower articula facet and a relatively shallow lingual notch be tween articular facets; entoconid absent (100%, i = 23). Comparisons — Cryptotis griseoventris has . relatively longer unicuspid toothrow than any o the other species except C. g. goodwini. Cryptotis alticola — Cryptotis griseoventris ha somewhat shorter, narrower foreclaws; a relativel; longer tail; lacks any vestige of the ventral exten sion of the sinus canal or the associated foramen possesses a foramen dorsal to posterior dorsal ai ticular facet and two well-developed dorsal fo ramina; has a relatively and absolutely narrowe palate (Fig. 11); has a relatively and absolutel longer unicuspid toothrow; and relatively shorte mandible. Cryptotis goldmani — Cryptotis griseoventri lacks the ventral extension of the sinus canal an i associated foramen; possesses a foramen dorsal t posterior dorsal articular facet and two well-de veloped dorsal foramina; has a relatively and ab solutely narrower palate (Fig. 11), relatively an absolutely longer unicuspid toothrow, and a lor FIELDIANA: ZOOLCKT ger, narrower humerus with less enlarged bony processes. Remarks — Cryptotis griseoventris was origi- nally described by Jackson (1933) based on a specimen from Chiapas. This was one of many lames that Choate (1970) synonymized under the subspecies C. goldmani goldmani — an under- standable move, given the confused taxonomic state of the genus and lack of adequate sample sizes at that time. However, a major discontinuity ;xisted in the range of Choate's C. goldmani gold- mani at the Isthmus of Tehuantepec, and popula- ions on either side of the isthmus are completely isolated from each other. These two populations ire easily distinguishable based on a number of :ranial and postcranial characters (see Compari- sons, above), and it is now clear that they repre- sent distinct species. Specimens of Cryptotis griseoventris have been :aptured above at least 2100 m in forests domi- nated by pines and firs and in oak-dominated :loud forest. Some areas inhabited by this small- jared shrew undergo nightly winter frosts and oc- :asional snow (Goldman, 1951). No reproductive data are available for Cryptotis griseoventris. None of the skins we inspected of idult males collected 25-30 September (n = 4), 2-6 October (n = 2), and 25-31 December (n = 10) had obvious lateral glands. Specimens Examined (28) — GUATEMALA: HUEHUETENANGO: Todos Santos Cuchumatan [15°36'N, 91°37'W], 10,000 ft (usnm 77051- 77068). MEXICO: CHIAPAS: San Cristobal de las Casas [16°45'N, 92°38'W], 8000-9500 ft (usnm 75886-75894, includes holotype); 6 mi SE San Cristobal de las Casas (mcz 48061). Cryptotis goldmani (Merriam, 1895) Blarina mexicana goldmani: Merriam, 1895:25; Elliot, 1904:560. Blarina mexicana machetes: Merriam, 1895:26; Elliot, 1904:561. Blarina fossor: Merriam, 1895:28; Elliot, 1904: 562. Cryptotis frontalis: Miller, 1911:222; Hall and Kelson, 1959:60; Goodwin, 1969:40. Cryptotis mexicana goldmani: Miller, 1912:27; Davis and Lukens, 1958:350; Hall and Kelson, 1959:59; Genoways and Choate, 1967:204. Cryptotis mexicana machetes: Miller, 1912:27; I Hall and Kelson, 1959:60; Schaldach, 1966: 288; Goodwin, 1969:40 (in part). Cryptotis fossor: Miller, 1912:28; Hall and Kel- son, 1959:62; Goodwin, 1969:41. Cryptotis guerrerensis: Jackson, 1933:80; Hall and Kelson, 1959:60. Notiosorex phillipsii: Goodwin, 1969:43 (in part). Cryptotis mexicana mexicana: Goodwin, 1969:39 (in part). Cryptotis goldmani goldmani: Choate, 1970:247 (in part); Hall, 1981:59 (in part). Holotype — Skin and skull of adult male, usnm 70244; taken 23 December 1894 by E. W. Nelson and E. A. Goldman (collector number 7231); "from mountains near Chilpancingo, Guerrero, Mexico" (Merriam, 1895:25), 10,000 ft. Distribution — Conterminous high mountain areas in the Sierra Madre del Sur and Sierra Ma- dre de Oaxaca (Fig. 4); known from above 1500 m in the Mexican state of Guerrero and above 2000 m in Oaxaca. Description — Size medium for the genus (Ta- ble 1); tail moderately long, averaging 29 mm, or ca. 38% of HB (Table 2); dorsal guard hairs about 6-8 mm long; dorsal pelage Mummy Brown; ven- tral pelage somewhat paler than that on dorsum, between Light Drab and Drab; dorsal fur three- banded: basal five-sixths of hairs silvery gray, fol- lowed by a thin, lightly colored band grading into brown tip. Rostrum of moderate length (PL/CBL = 44.2% ± 1.0, n = 16); usually two well-de- veloped dorsal foramina (61%, n = 41); an ob- vious, well-developed foramen (that opens into a ventral extension of the sinus canal) typically is present posterior to the dorsal articular facet on one or both sides of the skull (97%, n = 38; see Remarks, below); a foramen dorsal to the dorsal articular facet only occasionally present on one or both sides of the skull (16%, n = 38); zygomatic plate of moderate breadth (ZP/PL = 21.2% ± 2.4, n = 16), anterior border usually aligned with me- sostyle-metastyle valley of M', but may extend as far posteriorly as the parastyle of M:; upper tooth- row uncrowded, U4 aligned with toothrow and usually partly visible in labial view; P4, M1, and M2 only very slightly recessed on posterior bor- der; M-1 with paracrista, paracone, precentrocrista, and a small, lightly pigmented or unpigmented protocone; mesostyle of M3 is usually small and uncolored, and occasionally has a short, uncolored postcentrocrista. Mandible of moderate length and breadth; articular process tends to be tall and wide, with a narrow ventral articular facet and a shallow lingual notch between articular facets (Oaxaca), or tall and moderately wide, with a WOODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS 19 Fig. 12. Lateral views of the orbital areas of the skulls of A, Cryptotis griseoventris (usnm 75887); and B, Cryptotis goldmani (ku 98727). Arrows indicate ap- proximate locations of foramen dorsal to (A) and fora- men posterior to (B) the dorsal articular facet and as- sociated with the ventral extension of the sinus canal. Dashed line on B indicates the path of the ventral branch of the sinus canal. broad ventral articular facet and a deep notch be- tween articular facets (Guerrero); entoconid of M3 often present, but usually poorly developed (48%, n = 23), occasionally well developed (4%). Comparisons — Cryptotis goldmani is the only member of the C. mexicana-group with an obvi- ous, well-developed foramen posterior to the dor- sal articular facet. The humerus in this species is the shortest and broadest in the genus, and it has the most elongate and enlarged bony processes. Cryptotis alticola — Cryptotis goldmani aver- ages smaller in HB, CBL, and weight (Table 1); has a relatively longer tail; has smaller forefeet and foreclaws; averages narrower in relative pal- atal breadth; and has less recessed posterior bor- ders of P4, M1, and M2. Remarks — Cryptotis goldmani is unique among the broad-clawed Cryptotis in its posses- sion of a well-developed foramen associated with a ventral branch of the sinus canal (Fig. 12B). This ventral branch canal exits through a foramen that usually is located posterior to the dorsal ar- ticular facet and is approximately the size of the anterior orifice of the sinus canal. The foramei associated with the ventral branch of the sinus ca nal is a convenient character for distinguishin; crania of C. goldmani from those of other mem bers of the C. mexicana-group. This foramen i absent from the holotype of C. g. magnimana anc all specimens of C. g. goodwini, C griseoventrh and C. nelsoni that we inspected. A minute fora men associated with an equally diminutive ventre branch canal is present in many C. alticola (549J n = 22), C. mexicana (22%, n = 92), C. obscur (21%, n = 34), and C. peregrina (85%, n = 20; however, the foramina in these species never ap proach the size of the well-developed foramen c the ventral branch canal present in C. goldman Another foramen, dorsal to the dorsal articuk facet, often is present among members of the C mexicana-group, but this foramen is much les prevalent in species (and individuals) that posses a ventral branch of the sinus canal, regardless c size. Cryptotis goldmani, C mexicana, and C. pen grina all occur in Oaxaca, and they can be diff cult to tell apart, as attested by the number c misidentified specimens we have encountere( Cryptotis goldmani tends to be larger (HB = 7 ± 5; CBL = 19.6 ± 0.5; see Table 1) than Oa> acan C. mexicana (HB = 67 ± 3, n = 16; CB = 18.6 ± 0.4, n = 13; C. mexicana from otht Mexican states are larger) and C. peregrina (H = 66 ± 4, n = 17; CBL = 19.7 ± 0.4, n = 17 but there is sufficient overlap in size to make th an undesirable character for identifying individi als with any certainty. Cryptotis goldmani can b identified most easily by its much longer ar broader foreclaws and, cranially, by its very wel developed ventral sinus canal. From C. mexicanc C. goldmani usually can be distinguished by i vestigial or absent entoconid on M3. Cryptot mexicana can be distinguished most conclusive from C. peregrina by its wider foreclaws and for feet and its relatively shorter tail (34% ± 3 of H vs. 48% ± 5). Cryptotis mexicana is also muc more likely (72%) to have a distinct entoconid < M3 than is C. peregrina (6%). Specimens of Cryptotis goldmani have bet taken in high-elevation pine forest in Oaxaca ai Guerrero. The climates of some localities whe the species has been found include severe wint frosts and heavy summer rains (Goldman, 1951 Percy L. Clifton (ku field notes and catalog, 196 took C. goldmani in snap traps set under rod and logs in pine forest with scattered oak trees 7300 ft at Omiltemi, Guerrero, on 15 July 196 20 FIELDIANA: ZOOLOG )ther species taken in the same habitat included Veotoma mexicana, Peromyscus aztecus, Pero- nyscus megalops, and Reithrodontomys sp. Paul J. Robertson (ku field notes and catalog, 1970) :aptured two C. goldmani while trapping along a mall rivulet and a larger stream lined with hard- vood trees at 2300 m at Campemento Rio Moli- lo, Oaxaca, from 14 to 16 April 1970. The nearby lopes were covered with pines. Musser (1964) eported that the canyon bottom at Rio Molino :ontained relatively moist, open forest with bro- neliad-covered oaks and a dense understory of hrubs, ferns, and herbaceous vegetation. The dgher slopes of the canyon were drier and cov- :red with a mixed secondary growth of oaks and >ines. In the same trapline, Robertson (ku field lotes and catalog, 1970) took Cryptotis peregrina, lorex saussurei, Microtus mexicanus, Oryzomys 'hapmani, Peromyscus megalops, Reithrodonto- nys mexicanus, and Rattus sp. In a nearby over- grown field, he captured Cryptotis peregrina, Rei- hrodontomys megalotis, R. sumichrasti, and Sig- nodon alleni. Reproductive data for Cryptotis goldmani are acking. Lateral glands were well developed on hree males collected 21-25 March, three of four nales from 18-28 April, three males from 11-15 une, six males from 25-31 July, and one male rom 5 August. A male taken on 26 December lad lateral glands that were visible, but not par- icularly well developed. Lateral glands were not leveloped on single specimens taken during the nonths of January, February, and October. The iming of gland development that we ascertained "or C. goldmani suggests that adult males may be eproductively active from March through June, iowever, data are lacking entirely for five months )f the year, and there are no large series of males rom any single month. Specimens Examined (58)— MEXICO: GUER- RERO: S slope of Cerro Teotepec, 3150 m (ummz 14710); mountains near [N and NW of] Chil- lancingo, 9500-9800 ft (usnm 70243-70245 [in- :luding holotype], 70247); Los Retrocesos, 1550 n (mzfc 3485-3486); El Iris (mzfc 3481-3482); )miltemi, 5700-8700 ft, 2450 m (asnhc 3493- 494; ibunam 29471, 32006; ku 98725-98727; acm 74156-74161; mzfc 3480, 3483-3484; rsNM 126895-126897 [including holotype of :ryptotis guerrerensis], 126947, 127500, 127506); mi NW Omiltemi, 2300 m (usnm 329427); 2 mi V Omiltemi, 7800-7900 ft (tcwc 5573-5575, 665); 3 mi W Omiltemi, 8200 ft (mvz 113491); ; mi SW Filo de Caballo, 8200 ft (tcwc 41948- 41949). OAXACA: Campamento Rio Molino, 2300 m (ku 124278, 124301); Vista Hermosa (ku 143749); Cerro Zempoaltepec, 8000-10,500 ft (usnm 68531, 68542, 68545 [holotype of Cryp- totis fossor], 68547); Cerro Zempoaltepec, 4.5 km N of Santa Maria Yacochi, Mpio. Tlahuitoltepec, 2450 m (ibunam 33601-33602); Puerto Angel Road, lumber camp, km 158, 8375 ft (cas 15477); 2 km NE San Andres Chichuaxtla, 2300 m (ummz 113888); Santa Maria Ozolotepec [mountains near La Cieneguilla], 10,000 ft (usnm 71454- 71460, includes holotype of Cryptotis mexicana machetes); "near the City of Tehuantepec" (usnm 123429, holotype of Cryptotis frontalis). Additional Records— MEXICO: OAXACA: Lachao; San Miguel Suchixtepec; San Juan Ozo- lotepec; Mixteguilla (Choate, 1970:249). Cryptotis alticola (Merriam, 1895) Blarina alticola Merriam 1895:27, Elliot, 1904: 561. Cryptotis alticola Miller: 1912:27; Davis, 1944: 376; Hooper, 1957:3; Hall and Kelson, 1959: 60; Genoways and Choate, 1967:204. Cryptotis euryrhynchis: Genoways and Choate, 1967:203. Cryptotis goldmani alticola: Choate, 1970:245; Hall, 1981:59. Holotype — Skin and skull of adult male, usnm 52047; collected 25 February 1893 by E. W. Nel- son (collector number 4396); "from Mount Po- pocatepetl" (Merriam, 1895:27) [Volcan Popoca- tepetl], Mexico, Mexico, 11,500 ft. Distribution — Highland areas above 2000 m in the Mexican states of Colima, Jalisco, Michoa- can, Mexico, and Puebla, and the Distrito Federal (Fig. 4). Description — Size medium for the genus (Table 1); tail short, averaging 26 mm, or ca. 33% of HB (Table 2); dorsal guard hairs about 6-8 mm long; dorsal pelage Olive Brown/Mum- my Brown/Bister/Clove Brown; venter some- what paler, Light Grayish Olive/Drab/Buffy Brown to Olive Brown; dorsal fur distinctly three-banded: basal five-sixths of hairs silvery gray, followed by a thin, lightly colored band grading abruptly to the brownish tip. Rostrum of moderate length (PL/CBL = 43.2% ± 1.2, n = 16); more than half of specimens examined (54%, n = 22) have two well-developed dorsal foramina; a minute foramen opening on an JVOODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS 21 equally minute, ventral extension of the sinus canal may be present posterior to the dorsal ar- ticular facet on one or both sides of the skull (52%, n = 23; see account for C. goldmani, above); a foramen dorsal to the dorsal articular facet may be present on one or both sides of the skull (39%, n = 23); zygomatic plate of mod- erate breadth (ZP/PL = 21.8% ± 1.2, n = 16), anterior border aligned with mesostyle-meta- style valley of M1; palate wide; P4, M\ and M2 slightly to moderately recessed on posterior border, emargination reaching about halfway to hypocone; M3 with paracrista, paracone, and precentrocrista, as well as small, lightly colored or uncolored protocone; mesostyle of M3 usu- ally small and uncolored, and sometimes having a short, uncolored postcentrocrista. Mandible relatively long and of moderate breadth; artic- ular process generally tall and moderately wide, with a moderately broad lower articular facet; entoconid of M3 often present but very reduced (64%, n = 14), or absent. Comparisons — Comparisons of Cryptotis alti- cola with other broad-clawed shrews are present- ed in the above accounts for those species. Se- lected measurements are presented in Table 1. Remarks — Choate (1970) noted many of the differences between Cryptotis alticola and C. goldmani (sensu stricto). Although he considered the two taxa to be "incipient species," he treated C. alticola formally as a subspecies of C. gold- mani. We concur with Choate (1970) that it is unlikely that there is intergradation between C. alticola and C. goldmani. The large number of differences in external, cranial, and postcranial characters clearly indicates that they are distinct species. Cryptotis alticola is known to occur in high- elevation forests dominated by pines and often mixed with firs, oaks, and other tree species. At least some of the areas inhabited by C. alticola are subjected to winter frosts and occasional snow (Goldman, 1951). Percy L. Clifton (ku field notes and catalog, 1967) noted that he cap- tured three C. alticola along Microtus runways on the relatively dry east side of a hill in an area of bunch grass under scattered pine, oak, and fir at 9000 ft, 20 mi SE of Autlan, Jalisco, during 27-28 April 1967. The moister western slope of the hill supported denser forest. Other small mammals caught in this trapline included Sorex saussurei, Microtus mexicanus, Peromys- cus aztecus, and Reithrodontomys sumichrasti. On 11 June 1967, at 10,000 ft, 12 mi SW of Ciudad Guzman, Jalisco, Clifton caught thre C. alticola in Museum Special snap traps st along a rotten log in a grassy area along th edge of a fir, pine, and hemlock forest. Walte W. Dalquest (ku field notes, 1946) captured C alticola, along with Microtus mexicanus, Nee tomodon alstoni, Peromyscus melanotis, an Reithrodontomys chrysopsis, in a clearing don inated by 3-foot-high clumps of sacaton gras in coniferous forest at 11,500 ft, 12 km ESE c Amecameca, in the state of Mexico, from 14 t 15 June 1946. At Monte Rio Frio, Mexico, D; vis (1944) captured a C. alticola in an unbaite snap trap set along a Microtus mexicanus rui way in short grass meadow bordered by pint at 10,500 ft. In a thick forest of pines, firs, an oaks at 9150 ft, ca. 12 mi W of Ciudad Hidalg. Michoacan, Robert W. Dickerman (ku fie] notes and catalog, 1955) took C. alticola wit Neotomodon alstoni and Reithrodontom) megalotis. Few reproductive data are available for Cry\ totis alticola. A female (ku 17691) from 12 k ESE of Amecameca, Mexico, was lactating c 15 June, and a pregnant female (ku 62311) w; found dead along a trail on Cerro Patamba Michoacan, on 1 July. Our examination of la eral glands on preserved study skins reveals pattern similar to that of C. goldmani: later glands were well developed on one of tv males taken in January, three males captun 21-25 March, three of four males captured 2' 28 April, and three males taken in June. Later glands were not visible on four males captun in February, July, October, and November, r spectively. Specimens Examined (36)— MEXICO: C( LIMA: Volcan de Fuego [= Volcan de Colim; 8800 ft (lacm 29058). DISTRITO FEDERA Cerro de Santa Rosa, 3200 m (ibunam 951 Santa Rosa, 3000 m (ummz 93367); Canyr [Canon] Contreras, 10,200 ft (ummz 9459: JALISCO: 20 mi SE Autlan [de Navarro], 9(X' ft (ku 1 1 1385-1 1 1387); 12 mi SW Ciudad Gu man, 10,000 ft (ku 112039-112041); Volcan J Fuego [= Volcan de Colima], 9800 ft (i 107143, holotype of Cryptotis euryrhynchi.i MEXICO: 12 km ESE Amecameca, 11,500 I (ku 17691); N edge Refugio San Cayetano, i mi S Bosenchere, 8200 ft (ummz 102713); L gunas de Zempoala [10 mi NNW Cuernavaca Choate, 1970] 9100 ft (usnm 329424-32942* j 45 km ESE Mexico City, Monte Rio Frio (to 1927); Estacion Experimental Forestal y de { 22 FIELDIANA: ZOOLOC Table 7. Complete character matrix for phylogenetic analysis of members of the Cryptotis mexicana-gcoup and vo outgroups, Cryptotis parva parva and Cryptotis nigrescens. Transition series are explained in Appendix II. Transitior series Taxon 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 jrx'a 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 2 0 0 0 0 grescens 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 1 0 1 exicana 1 1 1 2 1 0 1 1 1 1 1 2 2 1 1 1 1 0 0 0 1 0 0 2 jlsoni 1 1 1 9 2 1 0 1 9 9 9 9 9 9 9 9 9 0 0 0 0 1 0 0 2 bscura 0 0 1 9 2 0 0 1 1 1 1 1 2 2 1 1 1 1 0 0 0 0 1 0 2 iregrina 0 0 1 2 1 0 1 1 1 1 0 1 1 1 1 1 1 0 0 1 0 1 0 2 ticola 2 2 2 2 0 1 0 2 1 1 1 3 2 2 2 1 1 0 0 1 1 0 1 2 jldmani 2 2 2 2 1 1 0 2 1 1 3 3 2 2 3 1 2 0 0 1 1 0 1 2 wdwini 2 2 2 0 2 1 1 0 2 1 1 2 3 2 2 2 1 0 1 1 2 1 0 1 2 nseoventris 2 2 2 0 2 1 1 0 9 9 9 9 9 9 9 9 9 0 1 1 2 1 0 1 2 agnimana 2 2 2 9 2 1 1 0 2 1 1 2 3 2 2 2 9 0 1 1 9 1 0 1 2 lora "L. M. Arellano," 2500 m (ibunam 4779); Mount Popocatepetl [Volcan Popoca- ipetl], 11,500-13,500 ft (usnm 52043, 52045- 2047, includes holotype); Salazar, 8800- 0,000 ft (usnm 50757-50759); N slope Volcan oluca [= Nevado de Toluca; above San Juan e las Huertas], 1 1,500 ft (usnm 55896); 12 km San Juan de las Huertas (encb 22675); 15.5 m S, 7 km E Zinacantepec [= San Miguel Zin- :antepec], 3470 m (encb 19357). MICHOA- AN: ca. 12 mi W Ciudad Hidalgo, 9150 ft (ku 6280); 17.5 km NW Ciudad Hidalgo, 2980 m •ncb 26210); Sierra Patamba [Cerro Patam- an], 9000 ft (ku 66311). PUEBLA: 10 km W an Juan Tetla, 3300-3400 m (ibunam 26544- 6548). Additional Records — MEXICO: JALIS- CO: N slope Nevado de Colima [= Volcan de blima], 8000-10,000 ft. MEXICO: Cerro jusco, 11,000 ft. MICHOACAN: Cerro de ancitaro [= Pico de Tancitaro], 9000-12,000 (Choate, 1970:246). liylogeny of the Cryptotis oldmani-Group Our phylogenetic analysis of the Cryptotis oldmani- gxonp and the subspecies of Cryptotis lexicana, based upon 29 morphological transition sries (Table 7), provided nine shortest trees ength = 51 steps; Fig. 13). Strict consensus and Ldams consensus trees constructed from the nine lortest trees are identical (Fig. 14) and represent le topology of one of the shortest trees. Areas of isagreement among the nine shortest-length trees inter on two regions. The first contains the three branches leading to C. nelsoni, to C. mexicana, and to the Cryptotis goldmani-group. These are presented alternatively (1) as an unresolved tri- chotomy; (2) with a clade consisting of C nelsoni and C. mexicana representing the first outgroup to the C. goldmani-group; or (3) with C. mexicana as the second outgroup and C. nelsoni as the first outgroup to the C. goldmani-group (Fig. 13). The second region of disagreement consists of the branches leading to C. g. goodwini, C. g. magni- mana, and C griseoventris. This region is repre- sented (1) as a trichotomy among the three taxa; (2) with C. g. goodwini and C. griseoventris as sister species; or (3) with C griseoventris and C. g. magnimana as sister species. Further resolution of these two regions of the trees is hampered by the lack of complete skeletal material for C. ob- scura and C. g. magnimana, and the lack of any skeletons of C. nelsoni or C. griseoventris. De- spite this, we have sufficient information to con- clusively resolve the most basic relationships among these shrews. Choate 's (1970) Cryptotis mexicana-group is supported by as many as 16 synapomorphies. These species all have a similarly shaped skull, and they are most easily identified cranially by a suite of characters that includes the position of the zygomatic plate, the posterior border of which is even with (or slightly posterior to), and confluent with, the posterior root of the maxillary process (Fig. 3). In some North American species of Cryptotis, as well as in Blarina and Sorex, which commonly are considered to be sister groups of Cryptotis (George, 1986), the posterior border of the zygomatic plate is anterior to the posterior root of the maxillary process. Among members of the C. nigrescens complex, the posterior border of GOODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS 23 S-8SS 5 S-S o'g a Bi = ~ £ c C 0) 3 ilia | Q.CfcC(0O)O)O)5oQ. _, & 5 2 JS g J s §, s * & 5 fi Q.C t C «j 0>0)0>t is 5 is en Ml dC t C UOlDlO) 9 -3 5 o u \v u U 03 Sill*8* i: O O C r^ 03 CD O) CJ> C O Q. SS'6 a I S E 2 9 2 0> *: 5 o c ac t c u oioioi o f to •a 'Is tj CO C C 0) 3 2 5 £ I clgi E8| to ? to 05 C c c a> 03 ?llPi11ilf Q.CCCC0O>O>O>COQ. g 8i tt !|l|llilllM 03-9>20>*:OOC2-O0> § § - 1 2 1 i » S (5 P U 5 S i •£ S CtC«lC3iO)0)tO \^f & 24 FIELDIANA: ZOOLOC he zygomatic plate is even with the posterior root jf the maxillary process, as in the C. mexicana- *roup, but in the C. nigrescens-group the two jtructures are separated by the posterior edge of he palate (Woodman & Timm, 1993). Other cran- omandibular characters that link members of the E mexicana-group are the shape of the coronoid jrocess as it joins the horizontal ramus; the rela- ively high and narrow articular condyle; and the ong, low-cusped P,. These three characters are similar to those of Cryptotis gracilis (see Wood- nan & Timm, 1992) and may link C. gracilis Dhylogenetically with the C. goldmani-group. We ire in the process of studying C. gracilis and other southern Central American species, and we will iddress their relationship to the C. mexicana- proup in the future. The members of the Cryptotis nexicana- group also show a number of synapo- norphies associated with trends toward elonga- ion of the foreclaws, broadening of the forefeet, ind modification of the humerus. These are con- idered further below. The four taxa previously considered to be sub- pecies of Cryptotis mexicana (C. mexicana, C. xelsoni, C obscura, and C. peregrina) do not by hemselves form a natural group. The Cryptotis loldmani-gxoup is embedded within these taxa, ndicating that C. mexicana (sensit lato) is para- >hyletic. Among members of the C. mexicana- jroup, C. peregrina and C. obscura appear to be ■he most plesiomorphic species, based primarily ipon the development of their foreclaws, their mailer body sizes, and their shorter tails relative o other members of the C. mexicana-group. Monophyly of the Cryptotis goldmani-group is upported by at least seven synapomorphies (in- luding one reversal). Most of the transition series upporting this clade concern the modification of tructures of the forelimb. Among these species, he two most northern taxa, C. alticola and C. oldmani, appear to be the most plesiomorphic aembers (but see below). The three taxa from outh of the Isthmus of Tehuantepec, C. g. good- rini, C. g. magnimana, and C. griseoventris, form clade supported by three transition series, all ased on cranial foramina. Evolution of the Forelimb The most distinctive and intriguing character- istics of the Cryptotis mexicana-group occur in the forelimb. Our studies of these shrews indicate that modifications of the forelimb are not restrict- ed to the Cryptotis goldmani-group, but represent common trends seen throughout the C. mexicana- group. Based on our phylogeny, the humerus shows a trend of becoming shorter relative to overall body size, and the shaft of the humerus broadens (Fig. 15). The articular regions become enlarged relative to the length of the humerus, and various bony processes (notably the teres tubercle, medial epicondyle, and pectoral process) elongate and enlarge. Due to widening of the base of the lateral epicondyle and a lateral shift in the posi- tion of the head of the humerus, the humerus ap- pears more curved in the mediolateral plane. Nearly parallel trends are seen in the enlargement of the forefeet (Fig. 16), the shortening and broad- ening of the metacarpals and phalanges (Fig. 17), and the elongation and broadening of the fore- claws. The hindfeet show similar shortening and broadening of metatarsals and phalanges (Fig. 18), but this is not reflected to the same degree externally as it is in the forefeet. It is likely that some or all of these characters are linked, but we did not predict this a priori, and the lack of skel- etal material for some taxa makes it impossible to confirm the extent of any linkages among these traits. For these reasons, we treated forelimb char- acters as separate transition series. In general, five grades of evolutionary modifi- cation of the humerus and the rest of the forelimb are exhibited within the Cryptotis mexicana-group as compared to the plesiomorphic condition rep- resented by the two operational outgroup species, Cryptotis parva and Cryptotis nigrescens (Fig. 15 A). These grades are represented in order by (1) C. peregrina (Fig. 15B); (2) C obscura (Fig. 15C); (3) C. mexicana (Fig. 15D); (4) C. alticola and C. goodwini (including C g. goodwini and C. g. magnimana; Fig. 15E-G); and (5) C. goldmani (Fig. 15H). Within the C. mexicana-group, the hu- merus is least modified in C peregrina; the bony Fig. 13. Topologies of the nine shortest trees (length = 51 steps; consistency index = 0.82; resolution index = 86; rescaled consistency index = 0.71) from phylogenetic analysis of the Cryptotis mexicana-group using 29 lorphological transition series. Cryptotis parva par\'a and Cryptotis nigrescens were used as outgroups. Differences mong trees center on two regions: (1) branches leading to C. mexicana, C. nelsoni, and the Cryptotis goldmani- roup; and (2) branches leading to C. griseoventris, C. goodwini goodwini, and Cryptotis g. magnimana. (VOODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS 25 parva nigrescens mexicana nelsoni alticola goldmani goodwini griseoventris magnimana obscura peregrina Fig. 14. Topology of strict consensus tree of the Cryptotis mexicana-group of species constructed from nine shortej length trees. This also is the topology of one of the nine shortest trees and the Adams consensus tree. See Figure for additional explanation. processes of the humerus are relatively small. Rel- ative to the outgroups, however, the humerus of C. peregrina is shortened, broadened, and some- what curved; the head of the humerus is dorso- ventrally elongate; and the distal articular surfaces are enlarged (Fig. 15B). The metacarpals of C. peregrina (Fig. 17B) are relatively shorter and thicker than in C. parva and C. nigrescens (Fig. 17 A). In the preceding grade, represented by Cryptotis obscura (Fig. 15C), the pectoral pro- cess, teres tubercle, and medial epicondyle of the humerus are enlarged relative to C. peregrina. The metacarpals of C. obscura are unknown. Both C. peregrina and C. obscura have elongate, but narrow, foreclaws and enlarged forefeet (Fig. 16B). The humerus of Cryptotis mexicana (Fig. 15D) shows a grade of development similar to that of C. obscura. No postcranial material is known for Cryptotis nelsoni; however, both C mexicana and C. nelsoni have larger forefeet and broader foreclaws (Fig. 16C) than C obscura and C. peregrina, and the metacarpals of C. mexicana (Fig. 17C) are broader than those of C. peregrina. Members of the Cryptotis goldmani-group have the largest forefeet, the foreclaws are the most elongate and broadened (Fig. 16D), and the metacarpals and phalanges are the shortest and thickest (Fig. 17D). Cryptotis alticola and Cryp- totis g. goodwini have the absolutely longest and broadest foreclaws, but this may be attributable in part to their larger overall body sizes (Table 1). Postcranial material is unknown for Cryptot griseoventris, but C. alticola, C g. goodwih and Cryptotis g. magnimana have very simil humeri that are relatively shorter and much mo curved than in C. mexicana (Fig. 15E-G). Tl! bony processes are more elongate, and the ten tubercle and medial epicondyle are positiont much more closely to each other. The most dl rived condition of the humerus is observed | Cryptotis goldmani (Fig. 15H). The bony pr< cesses are the most pronounced in this specie and the teres tubercle and medial epicondyle a most elongate and most closely situated to eac other, their internal edges forming a symmetric reverse C-shape. Cryptotis goldmani was n identified as one of the most derived species our phylogenetic analysis of the C. mexicam group because the characters that most distil guish its grade of forelimb development are ai tapomorphic. More interesting is the fact th; although it has the most derived humerus, < goldmani does not possess the most enlarged for claws or the broadest forefeet or metacarpals. Th suggests that enlargement of forefeet and for claws and the change in the shape of the humeri are not entirely linked, although the difference degree of development of the forefeet and for claws may be more closely associated with tl body size of the animals involved. Small-eared shrews in general have been cor sidered to be semi-fossorial or semi-cursori 26 FIELDIANA: ZOOLOG Fig. 15. Left humeri of selected Cryptotis: A, C. nigrescens (ku 142054, HB not recorded; however, ku 143384 as a slightly longer humerus and HB = 70 mm); B, C. peregrina (ku 124298, HB = 75 mm); C, C. obscura (mzfc 45, HB = 65 mm); D, C. mexicana (ku 29541, HB = 78 mm); E, C. alticola (ibunam 26544, HB = 85 mm); F, ".. g. magnimana (ku 14461 1, HB = 80 mm); G, C. g. goodwini (usnm 275681, HB = 86 mm); and H, C. goldtnani ibunam 29471, HB = 84 mm). All humeri are drawn to the same scale. The humeri of C. nelsoni and C. griseoventris re unknown. Note the tendency of the humerus to shorten relative to head-and-body length (HB) as the humerus ecomes more modified. Scale bar = 1 mm. Reed, 1951), although most information about lie habits of the genus is derived from studies of jst one species, Cryptotis parva, in the United tates. This shrew has been observed to be an ctive burrower in soft soil, and its nests are often Dcated at or below ground level (see Whitaker, 974, and references therein). However, the small ;et and foreclaws with which C. parva excavates how no obviously specialized adaptations for igging relative to other small mammals. Changes i forelimb structure that we document for mem- ers of the Cryptotis mexicana-group (especially the Cryptotis goldmani-group) are distinctive and suggest selected specialization, probably for ex- cavating. It seems likely that these modifications permit this group of shrews to have a more fos- sorial niche than other members of the genus. Un- fortunately, no behavioral or ecological studies are available that might support this hypothesis or help us understand the functional significance of these forelimb structures. Many of the modifications of the humerus (shortening, dorsoventral elongation of the head, lengthening and enlargement of various processes, VOODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS 27 Fig. 16. Ventral views of the right forefoot of A, Cryptotis parva orophila (HB = 60 mm); B, C. peregrina (I 71 mm); C, C. mexicana (HB = 71 mm); and D, C. goldmani (HB = 68 mm). Scale bar = 1 mm. B Fig. 17. Dorsal views of bones of the left manus of A, Cryptotis parva orophila (ku 142693, HB = 59 mm); C. peregrina (ku 124298, HB = 75 mm); C, C. mexicana (ku 29541, HB = 78 mm); and D, C. goldmani (ibun 33602, HB = 76 mm). The metacarpals of C. p. orophila are nearly the same length as those of C. goldmani, desp the large difference in size between the two species. The metacarpals of C. mexicana are slightly shorter and broa< than those of C. peregrina. Scale bar = 1 mm. 28 FIELDIANA: ZOOLOC 1 mm Fig. 18. Dorsal views of bones of the left pes of A, Cryptotis pana orophila (ku 142693, HB = 59 mm); B, C. qrescens (ku 143385, HB = 67 mm); C, C. peregrina'(KU 124298, HB - 75 mm); D, C. mexicana (ku 29541, 3 = 78 mm); E, C. alticola (ku 62311, no measurements); and F, C. goldmani (ibunam 33602, HB = 76 mm). ements with stippled borders are claws. Scale bar = 1 mm. OODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS 29 and close association of teres tubercle and medial epicondyle), metacarpals (shortening and broad- ening), and foreclaws (lengthening and broaden- ing) are similar to characteristics seen in the fore- limbs of talpids, and they even surpass the degree of modification of the humerus observable in some of the most primitive grades of moles (no- tably Uropsilus). The C. mexicana-group provides a living series of structural grades that may pro- vide clues as to the early evolution of digging adaptations in talpids and possibly other strongly fossorial mammals. Zoogeography of the Cryptotis goldmani-Group Choate (1970:297) previously suggested that southern Mexico was "the center of origin and dispersal" for most modern members of the genus Cryptotis because of the large number of species there, the high degree of differentiation among subspecies, and the presence of what he consid- ered to be the most derived species in the genus. We concur, based on our understanding of the modern biogeography and phylogeny of the Cryp- totis mexicana-group, that a highland region of central or southern Mexico west of the Isthmus of Tehuantepec was the most likely location for the evolution of the common ancestor of this clade. Most of the eight species of shrews that make up the C. mexicana-group are restricted to the west of the isthmus, including the least specialized spe- cies, Cryptotis peregrina and C obscura. In gen- eral, the members of the group inhabit high ele- vations; although C. mexicana has been found as low as ca. 500 m elevation in Veracruz, elsewhere this and all other species in the clade are known only from elevations above 1000 m (Choate, 1970; this paper). The distribution of the more specialized Cryp- totis goldmani-group is discontinuous from west central Mexico to western Honduras. In southern Mexico, the occurrence of these shrews corre- sponds roughly with the distribution of the tree genus Pinus (see Styles, 1993), an observation that seems to be supported by habitat descriptions in collectors' field notes (see comments on spe- cies, above). One of the major gaps in the distri- butions of both the Cryptotis goldmani-group and Pinus is at the Isthmus of Tehuantepec, where there is a break between the highlands of the Si- erra Madre del Sur to the west and the Sierra Ma- dre de Chiapas to the east. The continental divid at the isthmus drops to as low as 250 m, and mon tane plant communities give way to vegetatioi transitional between lowland rainforest and semi arid scrub (Duellman, 1960). Duellman (196C 1966) recognized the break in the highlands at th< Isthmus of Tehuantepec as a major barrier to th« dispersal of montane amphibians. In general, th< isthmus does not appear to have been regarded a a major barrier to most small mammals (Hersh kovitz, 1958; Hall, 1981), although it may be im plicated in the evolution of subspecies or specie of montane Peromyscus (see Carleton, 1989) Many middle- to high-elevation mammal species including four taxa of soricids {Cryptotis parvi pueblensis, C. mexicana, Sorex saussurei, and So rex veraepacis), putatively occur on both sides o the isthmus (Choate, 1970; Hall, 1981). Howevei no species of shrews are known from the isthmu itself, and this dry lowland probably is now ai effective barrier to the dispersal of all shrews Clearly, the isthmus prevents contact betweei eastern and western members of the C. goldmani group; two species {Cryptotis alticola and C goldmani) are found to the west of the isthmus and three taxa (C. g. goodwini, C. g. magnimana and C. griseoventris) occur on the eastern side. Dispersal of the Cryptotis goldmani-grou] across the Isthmus of Tehuantepec probably tool place during one of the Pleistocene glacial epochs as suggested by Duellman (1960, 1966) for mon tane amphibians. Climatic amelioration, combine< with lowered sea level during one or more of tht at least nine Pleistocene glacial maxima, wouh have resulted in a lowering or mixing of climati< and vegetational zones that permitted montane forest plants and animals to extend across the isth mus. The subsequent upward migration of cli matic zones preceding a subsequent interglacia would have isolated populations on either side o the isthmus. The sparse paleofloral evidence available for southern Mexico and Central Amer ica provides support for fluctuation in elevationa climatic zones in the past, although regional evi dence for the timing of specific vegetationa changes remains scant (Graham, 1993). The topology of our phylogeny of the Cryptoti. goldmani-group (Fig. 14) suggests that thest shrews derived from a common ancestor that orig inated to the west of the Isthmus of Tehuantepec The two species to the west of the isthmus, C alticola and C. goldmani, are the two most prim itive shrews in the C. goldmani-group in terms o numbers of shared-derived characters. Althougl 30 FIELDIANA: ZOOLOGY C. goldmani is clearly the most specialized mem- ber of the C. goldmani-group in terms of its fore- limb structure, this degree of specialization is mostly autapomorphic. The three eastern taxa (C g. goodwini, C. g. magnimana, and C. griseo- ventris) form a clade that is supported by up to four transition series (TS 22-25 in Appendix I). A single character of the humerus (TS 16 in Ap- pendix I) links C. goldmani with the three eastern taxa, suggesting that they share a common ances- tor. Based on our phylogeny and the modern dis- tributions of the species in the C. goldmani-group, we speculate that the occurrence of members of this group on both sides of the Isthmus of Te- huantepec is the result of a single dispersal event from west to east; as habitat conditions on the isthmus became more favorable for shrews during a glacial maximum, the common ancestor (of C. goldmani and of the three western taxa) expanded its range from the west across the isthmus to east. As climatic conditions on the isthmus shifted with the onset of the succeeding interglacial, the geo- graphic range of the ancestral population contract- ed into two smaller populations, isolated from each other by the isthmus. From the western pop- ulation, C. goldmani ultimately evolved, whereas the eastern population gave rise to C. goodwini and C. griseoventris. (In this scenario, the lineage leading to modern C. alticola branched off prior to the dispersal across the isthmus.) Two lines of evidence provide clues as to the timing of the dispersal of the Cryptotis goldmani- group across the isthmus. The modern occurrence of four species of shrews on both sides of the Isthmus of Tehuantepec (but absent from the isth- mus itself; see above) suggests that these four spe- cies crossed the isthmus during the ultimate (Wis- consinan) glacial epoch, but that populations of these species on either side of the isthmus, iso- lated since the end of that epoch, have had insuf- ficient time to differentiate (speciate). (Alterna- tively, these "species" may represent unresolved taxonomic problems.) Given similar rates of evo- lution among these species and the C. goldmani- group shrews, we can speculate that the eastern and western members of the C. goldmani-group separated earlier, possibly subsequent to the pen- ultimate (Illinoian) glaciation at the latest. This is supported by the recent rediscovery of fossil re- mains of C. goodwini-C. griseoventris-grade shrews from a site in Honduras believed to be late Pleistocene in age (D. Croft, in lit.; Woodman, unpublished data). Given an Illinoian or earlier dispersal, migrations of the C. goldmani-group across the isthmus either to the east or to the west during subsequent glacial epochs may have been blocked by the presence of sister species (and po- tential ecological competitors) in suitable habitat on both sides of the isthmus. Key to the Cryptotis goldmani-Group 1. West of the Isthmus of Tehuantepec: usually a foramen posterior to the dorsal articular fac- et (Fig. 12B); usually lacking a foramen dor- sal to external capitular facet; vestigial ento- conid may be present onM, 2 1'. East of the Isthmus of Tehuantepec: always lacking a foramen posterior to the dorsal ar- ticular facet (Fig. 12A); usually a foramen dorsal to dorsal articular facet; entoconid lack- ing from M, 3 2. Larger species (averaging 11 ± 3 g) with shorter tail (23-30 mm, averaging 33% of HB); foramen posterior to dorsal articular fac- et is minute, if present C. alticola 2'. Smaller species (averaging 8 ± 1 g) with lon- ger tail (24-36 mm, averaging 38% of HB); large foramen posterior to dorsal articular fac- et present C. goldmani 3. Zygomatic plate broader (ca. 2.2 mm, 25.0% of PL); unicuspid toothrow shorter (ca. 2.4 mm, 1 1.8% of CBL) C. goodwini magnimana 3'. Zygomatic plate narrower (^ 2.2 mm, ^ 24.7% of PL); unicuspid toothrow longer (^ 2.5 mm, ca. > 12.0% of CBL) 4 4. Larger (CBL > 20.0 mm; HB averaging 84 ± 5 mm) C. goodwini goodwini 4'. Smaller (CBL < 20.4 mm; HB averaging 77 ± 3 mm) C. griseoventris Acknowledgments Special thanks to Gustavo Cruz of the Univer- sidad Nacional Autonoma de Honduras and to Pe- ter Holm of the University of Arizona for making important new specimens of Honduran Cryptotis available to us for study. We thank the following curators and collection managers for loans or for permission to examine specimens under their care: Guy G. Musser (amnh); Robert C. Dowler (asnhc); Luis F. Baptista (cas); Ticul Alvarez and Sergio Alvarez (encb); Bruce D. Patterson and WOODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS 31 William T. Stanley (fmnh); Fernando Cervantes R. (ibunam); Lynn J. Barkley, Sarah B. George, and John E. Heyning (lacm); Maria E. Rutzmoser (mcz); Elmer C. Birney (mmnh); Livia Leon P. (mzfc); Gerhard Storch (smf); George D. Baum- gardner (tcwc); Phil Myers (ummz); and Michael D. Carleton, Alfred L. Gardner, and Linda K. Gor- don (usnm). Robert S. Hoffmann, Guy Musser, Ronald H. Pine, Norman A. Slade, and an anon- ymous reviewer provided helpful comments on previous versions of this manuscript. Amy La- throp provided the illustrations used as Figure 6. Robert P. Anderson, Rafael L. Joglar, and Adrian Nieto Montes de Oca aided with the Spanish translation of our abstract. Kate Shaw kindly pro- vided technical assistance for our phylogenetic analyses. Portions of this project were funded by the Panorama Society Fund and the E. Raymond Hall Endowment Fund, both of the University of Kansas Natural History Museum; the KU Depart- ment of Systematics and Ecology, the KU General Research Fund; the KU Fellowship Program for Latin America Studies; and the National Science Foundation (BSR 89-04195); we are indebted to each for helping make this study possible. Literature Cited Bee, J. W., D. Murariu, and R. S. Hoffmann. 1980. Histology and histochemistry of specialised integu- mentary glands in eight species of North American shrews (Mammalia: Insectivora). Travaux du Museum d'Histoire Naturelle Grigure Antipa, 22: 547-569. Buhler, P. 1964. Zur Gattungs — und Artbestimmung von Neomys — Schadeln — gleichzeitig eine Einfiihr- ung in die Methodik der optimalen Trennung zweier systematischer Einheiten mit Hilfe mehrerer Merk- male. Zeitschrift fur Saugetierkunde, 29: 65-93. Burt, W. H., and R. A. Stirton. 1961. The mammals of El Salvador. University of Michigan Museum of Zoology, Miscellaneous Publications, 117: 1-69. Carleton, M. D. 1989. Systematics and evolution, pp. 7-141. In Kirkland, G. L., Jr., and J. N. Layne, eds., Advances in the Study of Peromyscus (Rodentia). Texas Tech University Press, Lubbock, 367 pp. Choate, J. R. 1970. Systematics and zoogeography of Middle American shrews of the genus Cryptotis. Uni- versity of Kansas Publications, Museum of Natural History, 19: 195-317. Choate, J. R., and E. D. Fleharty. 1974. Cryptotis goodwini. Mammalian Species, 44: 1-3. Davis, W. B. 1944. Notes on Mexican mammals. Jour- nal of Mammalogy, 25: 370-403. Davis, W. B., and P. W. Lukens, Jr. 1958. Mammals of the Mexican State of Guerrero, exclusive of Chi- roptera and Rodentia. Journal of Mammalogy, 39: 347-367. Duellman, W. E. 1960. A distributional study of th amphibians of the Isthmus of Tehuantepec, Mexicc University of Kansas Publications, Museum of Natr ral History, 13: 21-71. . 1966. The Central American herpetofauna: a ecological perspective. Copeia, 1966: 700-719. Eadie, W. R. 1938. The dermal glands of shrews. Joui nal of Mammalogy, 19: 171-174. Elliot, D. G. 1904. The land and sea mammals of Mic die America and the West Indies. Zoological Serie; Field Columbian Museum, 4(2): v-xiii, 441-850. Felten, H. 1958. Weitere Saugetiere aus El Salvadc (Mammalia: Marsupialia, Insectivora, Primates, Ec entata, Lagomorpha, Carnivora und Artiodactyla Senckenbergiana Biologica, 39: 213-228. Frost, D. R., and D. M. Hillis. 1990. Species in con ; cept and practice: Herpetological applications. Hei petologia, 46: 87-104. Genoways, H. H., and J. R. Choate. 1967. A ne\| species of shrew (genus Cryptotis) from Jalisco, Mex j ico (Mammalia; Insectivora). Proceedings of the Bi ological Society of Washington, 80: 203-206. George, S. B. 1986. Evolution and historical biogeog raphy of soricine shrews. Systematic Zoology, 35 153-162. Goldman, E. A. 1951. Biological investigations i Mexico. Smithsonian Miscellaneous Collections, 115 i-xiii, 1-476. Goodwin, G. G. 1934. Mammals collected by A. V\ Anthony in Guatemala, 1924-1928. Bulletin of th American Museum of Natural History, 68: 1-60. . 1969. Mammals from the State of Oaxac; Mexico, in the American Museum of Natural Histon Bulletin of the American Museum of Natural Histon 141: 1-270. Graham, A. 1993. Historical factors and biological di versity in Mexico, pp. 109-127. In Ramamoorthy, 1 P., R. Bye, A. Lot, and J. Fa, eds., Biological Diversit of Mexico. Oxford University Press, New York, 81 pp. Hall, E. R. 1981. The Mammals of North Amelia 2nd ed. John Wiley & Sons, New York, 600 + 90 pp Hall, E. R., and K. R. Kelson. 1959. The Mammal of North America. The Ronald Press Company, Ne\ York, 546 + 79 pp. Hershkovitz, P. 1958. A geographic classification o Neotropical mammals. Fieldiana: Zoology, 36: 581 620. Hooper, E. T 1957. Records of Mexican mammals University of Michigan Museum of Zoology, Occa sional Papers, 586: 1-9. Jackson, H. H. T. 1933. Five new shrews of the genu Cryptotis from Mexico and Guatemala. Proceeding of the Biological Society of Washington, 46: 79-82 Maddison, W. P., and D. R. Maddison. 1992. Mac Clade: analysis of phylogeny and character evolutior version 3. Sinauer Associates, Sunderland, Massa chussetts, 939 pp. Medellin, R. A. 1988. Prey of Chrotopterus aurituy with notes on feeding behavior. Journal of Mammal ogy, 69: 841-844. Merriam, C. H. 1895. Revision of the shrews of thi 32 FIELDIANA: ZOOLOGY American genera Blarina and Notiosorex. North American Fauna, 10: 5-34. Miller, G. S., Jr. 191 1. Three new shrews of the genus Cryptotis. Proceedings of the Biological Society of Washington, 24: 221-224. . 1912. List of North American land mammals in the United States National Museum, 1911. Bulletin of the U.S. National Museum, 79: 1-455. Murariu, D. 1976. Les glandes tegumentaires de cer- tains insectivores (Mammalia — Insectivora) de Rou- manie. Anatomie, histologic et histochimie. Travaux du Museum d'Histoire Naturelle Grigure Antipa, 17: 387-413. Musser, G. G. 1964. Notes on geographic distribution, habitat, and taxonomy of some Mexican mammals. Occasional Papers of the Museum of Zoology, Uni- versity of Michigan, 636: 1-22. Ieed, C. A. 1951. Locomotion and appendicular anat- omy in three soricoid insectivores. American Midland Naturalist, 45: 513-671. Iidgway, R. 1912. Color standards and color nomen- clature. Published privately by the author, Washington, D.C., 43 pp. + 53 pi. Jchaldach, W J., Jr. 1966. New forms of mammals from Southern Oaxaca, Mexico, with notes on some mammals of the Coastal range. Saugetierkundliche Mitteilungen, 14: 286-297. Jtuart, L. C. 1954. A description of a subhumid cor- ridor across northern Central America, with comments on its herpetofaunal indicators. Contributions from the Laboratory of Vertebrate Biology, University of Mich- igan, 65: 1-26. Ityles, B. T. 1993. Genus Pinus: A Mexican purview, pp. 397-420. In Ramamoorthy, T. P., R. Bye, A. Lot, and J. Fa, eds., Biological Diversity of Mexico. Ox- ford University Press, New York, 812 pp. iwOFFORD, D. L. 1993. PAUP: Phylogenetic analysis using parsimony, version 3.1. Illinois State Natural History Survey, Champaign. J.S. Board on Geographic Names. 1956a. Gazetteer no. 15. Mexico. U.S. Government Printing Office, Washington, D.C. . 1956b. Gazetteer no. 26. El Salvador. U.S. Government Printing Office, Washington, D.C. . 1965. NIS Gazetteer. Guatemala. U.S. Govern- Soricidae), from Honduras. Proceedings of the Bio- logical Society of Washington, 105: 1-12. 1993. Intraspecific and interspecific variation ment Printing Office, Washington, D.C. /hitaker, J. O., Jr. 1974. Cryptotis par\>a. Mamma- lian Species, 43: 1-8. /iley, E. O. 1978. The evolutionary species concept reconsidered. Systematic Zoology, 27: 17-26. 1981. Phylogenetics. John Wiley & Sons, New York, 439 pp. I J/oodman, N. 1992. Biogeographical and evolutionary relationships among Central American small-eared shrews of the genus Cryptotis (Mammalia: Insectivo- ra: Soricidae). Ph.D. diss., University of Kansas, 386 pp. . 1993. The correct gender of mammalian ge- neric names ending in -otis. Journal of Mammalogy, 74: 544-546. Goodman, N., and R. M. Timm. 1992. A new species of small-eared shrew, genus Cryptotis (Insectivora: in the Cryptotis nigrescens species complex of small- eared shrews (Insectivora: Soricidae), with the de- scription of a new species from Colombia. Fieldiana: Zoology, new series, 74: 1-30. Appendix I: Transition Series Used in Phylogenetic Analysis 1. length of foreclaws (Fig. 16): short (0); elon- gate (1); greatly elongate (2). 2. breadth of foreclaws (Fig. 16): narrow (0); broadened (1); greatly broadened (2). 3. forefeet (Fig. 16): small (0); enlarged, broad- ened (1); greatly enlarged and broadened (2). 4. metacarpals (Fig. 17): long, narrow (0); short, broad (1). 5. posterior border of zygomatic plate (Fig. 3): even with or anterior to anterior root of max- illary process (0); even with posterior root of zygomatic process, but separated from it by posterior border of palate (1); even with (or posterior to) and confluent with posterior root of zygomatic process (2). 6. anterior border of coronoid process: steep, forming a narrow angle with horizontal ramus of mandible (0); less steep, forming a wide angle with horizontal ramus of mandible (1). 7. articular condyle: low and broad (0); high and narrow (1). 8. lower sigmoid notch: very shallow (0); deep (1). 9. shape of unicuspids (U'-IP): cone-shaped, posteroventral border straight-edged or con- vex (0); narrow, posteroventral border con- cave (1). 10. protoconal basin of M': about equal in size to hypoconal basin (0); reduced relative to hy- poconal basin (1). 11. M3 morphology: simple, metacone absent (0); complex, metacone present (1). 12. shape of P3: short and high (0); long and low (D- 13. shape of humerus (Fig. 15): long, narrow, rel- atively straight (0); short, robust, curved (1); short, robust, and very curved (2). 14. head of humerus: rounded (0); dorsoventrally elongate (1). 15. ventral edge of proximal face of greater tu- berosity of humerus: rounded (0); with broad, deep pocket (1). /OODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS 33 16. pectoral process of humerus: low (0); high (1); very high (2); extremely high (3). 17. teres tubercle of humerus (Fig. 15): relatively short (0); elongate (1); moderately elongate (2); greatly elongate (3). 18. medial epicondyle of humerus (Fig. 15): short (0); elongate (1); greatly elongate (2). 19. lateral epicondyle of humerus (Fig. 15): small (0); expanded (1); greatly expanded (2). 20. teres tubercle and medial epicondyle of hu- merus (Fig. 15): far apart (0); close together (1); very close (2); extremely close (3). 21. posterior edge of falciform process of tibia: not deeply pocketed (0); deeply pocketed (1). 22. foramen of sinus canal (Fig. 12): absent (0); present, but tiny (1); well developed (2). 23. dorsal foramina: two in < 75% of specimens (0); two in > 75% of specimens (1). 24. foramen dorsal to articular facet (Fig. 12): present in < 75% of specimens (0); present in > 75% of specimens (1). 25. entoconid of M3: well developed, present in > 75% of specimens (0); vestigial, but pres- ent in < 76% of specimens (1); absent (2). 26. body size (head-and-body length): smaller, mean < 69 (0); larger, mean > 72 (1). 27. relative tail length (% of head-and-body length): short, mean < 39% (0); long, mean > 40% (1). 28. upper unicuspid toothrow: crowded, three un- icuspids visible in lateral view (0); uncrowd- ed, four unicuspids visible in lateral view (1). 29. dorsal guard hairs: short, < 4 mm (0); of in- termediate length, 4-5 mm (1); long, > 5 mm (2). Appendix II: Additional Specimens Examined Cryptotis mexicana (59)— MEXICO: CHIA- PAS: 3 mi E Pueblo Nuevo Solistahuacan, 7000 ft (ku 83942); OAXACA: Vista Hermosa, 1500 m (ku 91464); 6.5 mi SSW Vista Hermosa, 7100 ft (ku 99546); 21.8 km S Vista Hermosa, 2100 m (ku 124274); 31.6 km S Vista Hermosa, N slope Cerro Pelon, 2650 m (ku 124275-124276); NE slope Cerro Pelon, 2620 m (ku 120302); Cerro San Felipe, 6 km W La Cumbre, 2670 m (ku 121658, 124293); 2 km W La Cumbre, 2900 m (ku 121414); 7 mi N Ixtlan de Juarez, 10,000 ft (cas 12244-12245); Llano de las Flores, 2800- 3150 m (ku 91465, 120303, 121410-12141 124277); 0.4 mi S Llano de las Flores, 9200 (tcwc 45106); 11 mi NE (Tuxtepec road) Liar de las Flores, 9100 ft (ummz 112571); 27.5 k NNE Llano de las Flores (lacm 74166-74173 Ixtlan, 16 mi WSW La Esperanza (tcwc 2501 1 PUEBLA: 7.5 mi NE Tezuitlan (mmnh 6875); mi NE Tezuitlan (mmnh 4710, 6939). VERV CRUZ: Las Vigas, 8500 ft (ku 29525, 2952* 29540, 29542-29545, 29547-29549); 11 km ^ Las Vigas, 8500 ft (ku 29524); 5 mi E Las Vig, (tcwc 25077); 4 km W Tlapacoyan, 1700 ft (v 23412-23414). Cryptotis nelsoni (9)— MEXICO: VER/ CRUZ: Volcan Tuxtla, 4800 ft (mcz 19747; usn 65429-65433, 65435-65437, includes holotype Cryptotis nigrescens (62) — COSTA RIC/ ALAJUELA: Cinchona, 1600 m (ku 106942 Monteverde Cloud Forest Reserve, 1580-1600 I (ku 143377-143380); Monteverde Cloud Fore] Reserve, Pefias Blancas Valley, 870 m (v j 143371, 143381). CARTAGO: [Volcan] Iraz j 8000-9400 ft (amnh 141200). GU AN AC AST PUNTARENAS border: Monteverde, Cerro Arr gos, 1790 m (ku 143382, 143384). PUNTA1 ENAS: Coto Brus [Canton], Sabalito District, L Tablas, Rio Coton, 1700 m (mncr no numbej Monteverde, 1345-1600 m (fmnh 12410 128415, 135224; iNBio no number; ku 13489 135008, 135083, 142053, 142689, 14278* 142789, 143295-143297, 143383, 14338: 143395, 143636, 143638, 144612; lacm 6484 67443, 67453; mmnh 14095; ummz 11588.' 115884, 117107-117110, 125632); Monteverd Cerro Amigos, 1760 m (ku 142054); Monteven Cloud Forest Reserve, 1530-1660 m (ku 14278( 142787, 143386, 143396, 143496, 143637); S; Luis, 1200 m [ca. 2.5 km S Monteverde] (t 143385); 1 mi SW of Finca Las Cruces, San Vit 4000 ft (lacm 74351-74353). SAN JOSE: S; Isidro (amnh 7952/9691, holotype); 9 mi N of S; j Isidro del General (Pan American Highway), 48( I ft (ummz 111999). Cryptotis obscura (42)— MEXICO: HIDAJ GO: Tlanchinol, 20°59'N, 98°39'W (ibuna 4186); Tulancingo (usnm 55633); Encarnacic (usnm 81125-81127, 81131-81134); 11 km E Acaxochitlan, Lago Tejocotal, 2250 m (]• 81770). QUERETARO: 1 km S Ahuacatlan (mz 637); Pinal de Amoles (usnm 81115-8111 81123-81124, 81129, 81147); 1 km S Pinal i Amoles (mzfc 638-648); 4 km SW Pinal de Am les, 2550 m (ibunam 29106, 29332). TAMA1 LIPAS: 5 mi NW of Gomez Farias, Rancho d 34 FIELDIANA: ZOOLOG :ielo, 3500 ft (mmnh 4301-4305, 4570-4571, ■74). VERACRUZ: Zacualpan, 6000 ft (ku 58241). Cryptotis parva orophila (25) — COSTA RICA: ALAJUELA: Santa Clara (mncr no num- >er). ALAJUELA: Zarcero, 6000 ft (fmnh 43974). ZARTAGO: Cartago (ku 26932; ummz 66465, 37316); Coliblanco (ku 26930-26931); La Estrel- la (amnh 14847); Estrella de Cartago, 4500 ft ummz 64147); Guarco (ku 16563); Navarro (mcz >1656); Irazu Range [Volcan de Irazu] (amnh >64 1/9841, holotype). HEREDIA: Barva Canton, 5an Jose de la Montana, Paso Llano, 1800 m (ku 142692-142694); Barva Canton, San Miguel de a Montana, 1690-1700 m (ku 143372-143374). SAN JOSE: 10 mi S of Cartago, El Muneco, 3800 t (ummz 67315); Cerro Tablazo, 1983 m (usnm 52525); San Jose (usnm 7224, 38477); San Pedro le Montes de Oca (amnh 139282); San Rafael de Pontes de Oca, 4300 ft (ku 147100); Santa Ana ilsu 15753). Cryptotis parva parva (48)— USA: KANSAS: Douglas Co.: Lawrence (ku 22131-22142); 3.5 mi W of Lawrence (ku 125554-125568); 1.5 mi N, 1.7 mi E of Lawrence (courthouse), sandpits (ku 114236-114239, 114241-114247). Cryptotis peregrina (18)— MEXICO: OAXA- CA: mountains 15 mi SW Oaxaca de Juarez, 9500 ft (usnm 68317, holotype); Rio Molino, Puerto Angel road, km 153, 2250-2300 m (cas 14068; ku 121661; ibunam 8447); lumber camp, Puerto Angel road, km 158, 8375 ft (cas 15478); Rio Jalatengo, Puerto Angel road, km 178, 4275 ft (cas 14069, 14071-14072, 15475; ibunam 27518); Puerto Angel road, km 195, 3475 ft (ibunam 26551); La Cima, Puerto Escondido road, km 184.5, 5750 ft (cas 15473); Puerto Escondido road, km 193, 4200 ft (cas 15474); Rio Guajalote, 2000 m (ku 114226); Sinai, 10 km E Nopala, 7200 ft (cas 14940); 20 mi S, 5 mi E Sola de Vega, 4800 ft (ku 98728); 16 km SW Suchixte- pec, 2000 m (encb 3413-3414). OODMAN & TIMM: RELATIONSHIPS AMONG BROAD-CLAWED SHREWS 35