Caribbean Journal of Science, Vol. 36, No. 3-4, 289-299, 2000 Copyright 2000 College of Arts and Sciences University of Puerto Rico, Mayagiiez Cryptotis merriami Choate in Costa Rica: Syntopy with Cryptotis nigrescens (Allen) and Possible Character Displacement (Mammalia: Insectivora) NEAL WOODMAN Department of Biological Sciences, East Stroudsburg University, 200 Prospect Street, East Stroudsburg, Pennsylvania 18301-2999 USA nwoodman@po-box.esu.edu ABSTRACT.?A recent review of small-eared shrews, genus Cryptotis (Mammalia: Insectivora: Soricidae), in the collection of the Museo Nacional de Costa Rica, revealed a previously unidentified specimen of Cryptotis merriami Choate from the Monteverde Cloud Forest Reserve. This specimen is only the second record of C. merriami from Costa Rica, it represents the southernmost record for the species, and it documents the first case of syntopy between C. merriami and Cryptotis nigrescens (Allen). The co-occurence of C. merriami and C. nigrescens provides conclusive support for the hypothesis they are distinct species within the C. nigrescens- group (sensu Woodman and Timm, 1993). The soricid community at Monteverde is comprised of three species in three evenly-spaced size classes: the numerically-dominant C. nigrescens, C. merriami, and an undescribed, endemic species the size of typical C. merriami. Morphometrical analysis indicated that C. merriami exhibits no apparent geographic trends in size throughout most of its range, yet the Monteverde specimen is signifi- cantly larger than all other C. merriami. Its large size at Monteverde apparently represents character dis- placement in response to the presence of a more abundant sister species, C. nigrescens; the presence of a second, unrelated shrew the size of more typical C. merriami; or interactions with both. INTRODUCTION Small-eared shrews of the genus Crypto- tis occur from southernmost Canada and the eastern half of the United States, through Mexico and Central America, to northwestern South America. Choate (1970) partitioned Middle American spe- cies into three informal divisions: the Cryp- totis mexicana-group, the Cryptotis parva- group, and a group of "relict species" of unknown, but supposed primitive affini- ties. The C. parva-growp included C. parva (Say), with six Middle American subspe- cies, and Cryptotis nigrescens (Allen), with three subspecies. Woodman and Timm (1993, 1999) dem- onstrated that C. nigrescens is a complex of six species (C. colombiana, C. hondurensis, C. mayensis, C. mera, C. merriami, and C. nigre- scens), which they informally called the Cryptotis nigrescens-growp. Members of the group share a number of morphological characters, some of which may be plesio- morphic within the genus (Woodman and Timm, 1999). These species are small- to medium-sized (head-and-body length, 53- 79 mm; Table 1) small-eared shrews with short- to medium-length tails (usually av- eraging 39-47% of head-and-body length); unspecialized forefeet; short, narrow fore- claws; simple humeri; broad zygomatic plates; posteriorly-convex unicuspids; and unrecessed upper molariform dentitions. The group is distributed from southern Mexico through Central America to the Eastern Cordillera of Colombia (Woodman and Timm, 1992, 1993; Woodman, 1995, 1996). Although each species is morpho- logically and morphometrically distinctive, all evidence indicated that species were al- lopatric, precluding tests of morphological intergradation. Two species of the Cryptotis nigrescens- group occur in Costa Rica. Cryptotis nigre- scens (sensu stricto) occurs at middle to high elevations (800-2900 m) throughout much of Costa Rica and western Panama (Fig. 1). Cryptotis merriami Choate has a discontinu- ous distribution above 900 m from Chiapas, Mexico, to northern Costa Rica, although 28<:> 290 N. WOODMAN TABLE 1. Measurements of Cryptotis nigrescens and Cryptotis merriami used in this study. Statistics presented are mean ? standard deviation of the mean, and observed extremes. A dash (?) indicates no measurement is available. C-pminUi niorpsrpns Cryptotis merriami jr o (Monteverde) Finca Main geographic range n = 41 (except as noted) Monteverde San Bosco n = 22 (except as noted) head-and-body length (HB) 66 ? 5 69 ? 69 ?4 56-79 60-77 (n = 23) tail length (TL) 29 ?3 27 21 29 ?3 20-34 22-33 (n = 23) condylobasal length (CBL) 18.1 ? 0.5 20.3 19.2 19.4 ? 0.4 16.9-19.3 18.7-20.0 (n = 37) (n = 16) cranial breadth (CB) 9.0 ? 0.2 10.4 9.4 9.5 ? 0.2 8.6-9.5 9.2-10.0 (n = 31) (n = 17) breadth of zygomatic plate (ZP) 1.9 ? 0.2 2.5 2.1 2.3 ? 0.2 1.7-2.2 2.0-2.6 interorbital breadth (IO) 4.2 ? 0.2 4.8 4.7 4.7 ? 0.1 3.9-4.7 4.4-4.9 breadth across first unicuspids (U1B) 2.3 ? 0.1 3.2 2.6 2.6 ? 0.2 2.1-2.5 2.1-2.8 breadth across third unicuspids (U3B) 2.7 ?0.1 3.5 3.1 3.0 ? 0.1 2.4-3.0 2.7-3.3 breadth across second molars (M2B) 5.3 ? 0.2 6.3 6.0 5.8 ? 0.2 4.9-5.6 5.2-6.2 palatal length (PL) 7.7 ? 0.3 9.3 8.3 8.4 ? 0.3 7.1-8.6 7.9-8.9 length of upper toothrow (TR) 6.9 ? 0.2 8.1 7.7 7.3 ? 0.2 6.3-7.3 6.8-7.7 length of unicuspid toothrow (UTR) 2.3 ? 0.1 2.6 2.4 2.5 ? 0.1 2.0-2.5 2.2-2.7 length of molariform toothrow (MTR) 5.1 ? 0.2 5.9 5.6 5.3 ? 0.1 4.8-5.6 5.1-5.5 (n = 10) posterior width of Ml (WM1) 1.7 ?0.1 2.1 1.8 1.9 ?0.1 1.5-1.8 1.7-2.0 length of mandible (ML) 6.1 ? 0.2 7.5 6.3 6.5 ? 0.3 5.5-6.6 5.8-7.2 CRYPTOTIS MERRIAMI IN COSTA RICA 291 TABLE 1. Continued. Cryptotis nigrescens (Monteverde) n = 41 (except as noted) Cryptotis merriami Monteverde Finca Main geographic range San Bosco n = 22 (except as noted) height of coronoid process (HCP) 4.3 ? 0.2 3.9-4.7 (n = 40) height of coronoid valley (HCV) 2.6 ? 0.2 2.3-2.9 (n = 40) height of articular condyle (HAC) 3.6 ? 0.2 3.3-4.0 (n = 40) breadth of articular condyle (BAC) 2.9 ? 0.1 2.6-3.1 articular condyle to M3 (AC3) 4.4 ? 0.2 4.0-4.8 length of lower toothrow (TRD) 5.4 ? 0.2 5.1-5.8 length of lower molariform toothrow (ml3) 4.1 ? 0.1 3.9-4.4 length of ml (mlL) 1.7 ?0.1 1.6-1.8 weight 5.4 ?1.0 3.0-7.5 (n = 38) 5.4 3.1 4.3 3.5 5.3 4.9 2.1 4.7 2.8 3.7 3.2 4.7 5.9 4.5 1.8 5.0 ? 0.2 4.5-5.3 2.9 ? 0.2 2.4-3.3 3.8 ? 0.2 3.3-4.3 3.0 ? 0.2 2.6-3.2 4.6 ? 0.2 4.3-4.9 5.7 ? 0.2 5.3-5.9 4.3 ? 0.2 4.1-4.6 1.8 ?0.1 1.7-1.9 its presence in Costa Rica was based on a single specimen from Finca San Bosco (Fig. 1), near Tilaran (Woodman and Timm, 1993). The two species are relatively easy to distinguish: C. merriami is larger than typi- cal Costa Rican C. nigrescens, and it has a more robust skull (particularly the man- dibles); bulbous dentition (sensu Choate, 1970); a ventral branch of the sinus canal and associated foramen (see Woodman and Timm, 1999); no foramen dorsal to the dor- sal articular facet; and a tendency toward only a single, large dorsal foramen. Crypto- tis nigrescens and C. merriami were not thought to come into geographic contact. I studied specimens in the Museo Natio- nal de Costa Rica, including recently- collected, unidentified shrews. Among these was a specimen of Cryptotis merriami (MNCR 358) from Monteverde (Fig. 1) that represents the second record of the species from Costa Rica. The Monteverde specimen was found dead along the Chomogo Trail (10?18'05"N, 84?47'27"W) in the Mon- teverde Cloud Forest Reserve on 14 June 1994. This trail extends from above 1550 m to just below 1700 m, running entirely through cloud forest. Collecting in the com- munity of Monteverde, the reserve, and the surrounding area by biologists and local residents has yielded more than 70 Crypto- tis in various museum collections?the largest collection of shrews from any local- ity in Central America. These specimens in- clude 70 C. nigrescens, the new specimen of C. merriami, and one specimen of an unde- 2}:>2 N. WOODMAN 800 m contour shown FIG. 1. Map of Costa Rica and western Panama showing the distribution of Cryptotis nigrescens (inferred from Woodman and Timm, 1993) and the locations of Finca San Bosco and Monteverde. scribed, endemic form (Cryptotis sp. "a") probably related to Cryptotis gracilis Miller and Cryptotis endersi Setzer in Choate's (1970) group of "relict species." The nu- merically dominant C. nigrescens has been found in a variety of habitats from 870-1760 m, including cloud forest along the Cho- mogo Trail. The new specimen of C. merriami con- firms the presence of the species in the Tila- ran Cordillera, extends the known distribu- tion of C. merriami ca. 25 km to the south, and provides the first certain record of syn- topy between C. merriami and C. nigrescens. Detailed examination of the Monteverde specimen indicates that it is within the range of variation for C. merriami in exter- nal measurements and in most non- mensural cranial and dental characters. The skull of this individual, however, is very large, measuring beyond the upper ex- tremes of most cranial variables docu- mented for other members of the species (Table 1) and for the C. nigrescens-group in general (see Woodman and Timm, 1993: Table 4). The purpose of this paper is to review morphometrical variation within C. mer- riami and C. nigrescens and examine the possibility that the large C. merriami at Monteverde may represent a case of char- acter displacement in the presence of a closely-related species. METHODS Terminology of dentition and dental characteristics follows Choate (1970). All measurements are in mm and follow Woodman and Timm (1993, 1999) and Woodman (1995); abbreviations are ex- plained in Table 1. Skull measurements were taken to the nearest 0.1 mm using ei- CRYPTOTIS MERRIAMI IN COSTA RICA 293 TABLE 2. Factor loadings for the first two factor axes from principal components analysis of nine cranial variables from C. merriami and Monteverde C. nigre- scens. Abbreviations as in Table 1. Variable Factor 1 Factor 2 UTR -0.281 0.761 ZP -0.304 -0.440 IO -0.329 -0.281 WM1 -0.332 0.072 TR -0.337 0.320 PL -0.348 -0.031 U3B -0.351 -0.118 U1B -0.355 -0.049 M2B -0.355 -0.159 eigenvalue: 6.9845 0.5983 proportion of dispersion explained: 0.776 0.066 ther an ocular micrometer in a binocular microscope or a hand-held dial caliper. Standard external measurements are those recorded by the collector. Principal components analyses (PCAs) and regressions were carried out using Minitab Release 8. Orthogonal principal components were calculated from the cor- relation matrix of nine untransformed cra- nial variables (ZP, IO, U1B, U3B, M2B, PL, TR, UTR, WM1). An initial PCA to evaluate the overall similarity of the Monteverde specimen with C. merriami and C. nigrescens included 24 C. merriami from throughout the species' geographic range (including Monteverde) and 41 C. nigrescens from Monteverde (Table 2, Fig. 2). To test for cli- nal variation in size, I regressed "size" on -2 0 o O C. nigrescens ? C. merriami ? San Bosco ?k Monteverde ? O n oo o o ?? 0? ? ?0 ? o? ? 0 6>o o o ? CT o o ?? -8 factor 1 FIG. 2. Plot of factor scores from principal components analysis of nine cranial variables from Cryptotis merriami and C. nigrescens. The first factor axis represents overall size; because factor loadings are negative, the largest specimens have the lowest scores on this axis (Table 2). The second axis represents length-of-unicuspid- toothrow and length-of-upper-toothrow contrasted with breadth-of-zygomatic-plate and interorbital-breadth. The two axes account for 85% of the variation. 2i:)4 N. WOODMAN "geographic locality" individually for each species (Figs. 3,4). For C. merriami, I used 23 specimens from throughout its geographic range (excluding the Monteverde speci- men) to calculate the regression. For C. ni- grescens, I used 49 specimens from Costa Rica and 33 from Panama. Each individual was treated as an single point in the regres- sions, although there may be a lack of in- dependence among individuals from the same locality. "Size" in both regressions is represented by scores on the first factor axes of two PCAs carried out individually on C. nigrescens and on C. merriami (Table 3). "Geographic locality" is represented by scores on the first factor axis of a PCA of the untransformed decimal equivalents of the latitude and longitude for all collecting lo- calities (Table 4). Latitude and longitude loaded equally (latitude = 0.707; longitude = 0.707) on the first factor axis, which ac- counted for 98.6% of variation. Factor 1 scores therefore describe the relative geo- graphic position of each locality. Specimens used in this study are depos- ited in the following institutions: American Museum of Natural History, New York (AMNH); Escuela Nacional de Ciencias Biologicas, Mexico City (ENCB); Field Mu- seum, Chicago (FMNH); University of Kan- sas Natural History Museum, Lawrence (KU); Natural History Museum of Los An- geles County, Los Angeles (LACM); Mu- seum of Comparative Zoology, Harvard University, Cambridge (MCZ); James Ford Bell Museum of Natural History, St. Paul a 4- _ o ?"----^ n 2 ? ? 0- t ? ^^-^ 1 -2- i ? ? -4- S Guatemala ? Honduras ? El Salvador -6- 0 Nicaragua * X San Bosco -8- i ' i i i i ?k Monteverde i l -0.5 0.5 1 geographic locality 1.5 2.5 FIG. 3. Regression of factor 1 scores ("size") from principal components analysis of nine cranial variables against "geographic locality" for C. merriami. "Size" factor scores are negative, so that the smallest individuals have the largest positive scores (Table 3). "Geographic locality" is a factor score derived from a PCA of latitude and longitude for collecting localities (Table 4). The regression (y = 1.32 - 0.742x) is not significantly different from zero (F = 1.92, P = 0.180, adjusted R2 = 4.0%), indicating no recognizable geographic trends in size. CRYPTOTIS MERRIAMI IN COSTA RICA 295 -1.2 -1 geographic locality 0.4 FIG. 4. Regression of factor 1 scores ("size") from a PC A of nine cranial variables against "geographic locality" for C. nigrescens in Costa Rica and Panama. The regression (y = 3.39 + 3.55x) indicates a significant trend of increasing size from northern Costa Rica through western Panama (F = 103.21, P < 0.001, adjusted R2 = 55.8%). Specimens listed as "s. Costa Rica" are from Costa Rican localities south of Monteverde; those from "w. Panama" are from west of San Felix. (MMNH); Museo Nacional de Costa Rica, San Jose (MNCR); University of California Museum of Vertebrate Zoology, Berkeley (MVZ); University of Michigan Museum of Zoology, Ann Arbor (UMMZ); Departa- mento de Biologia, Universidad Nacional Autonoma de Honduras, Tegucigalpa (UNAH); U.S. National Museum of Natural History, Washington (USNM). RESULTS A plot (Fig. 2) of factor scores on the first two factor axes from the PCA of C. merriami and C. nigrescens indicates that the two spe- cies overlap broadly on the second factor axis (length-of-unicuspid-toothrow and length-of-upper-toothrow contrasted with breadth-of-zygomatic-plate and interor- bital-breadth?Table 2). However, both species form distinct clusters with little overlap on the first factor axis, illustrating their clear distinction in overall size (Table 2). This analysis supports the impression of a strong size difference between the Mon- teverde C. merriami and all other known specimens of the species, including the other Costa Rican specimen from Finca San Bosco. The Monteverde specimen is the largest C. merriami documented. The slope of the regression line from the test for clinal variation in C. merriami (Fig. 3) suggests a trend of decreasing size from north to south in Central America, the op- 296 N. WOODMAN TABLE 3. Factor loadings for the first factor axis from two separate principal components analyses of nine cranial variables from C. merriami and C. nigre- scens, respectively. Abbreviations as in Table 1. Variable C. merriami C. nigrescens ZP -0.120 -0.170 IO -0.174 -0.347 UTR -0.259 -0.264 WM1 -0.345 -0.319 PL -0.372 -0.352 M2B -0.378 -0.381 TR -0.385 -0.370 U3B -0.404 -0.363 U1B -0.420 -0.377 eigenvalue: 4.8973 6.0330 proportion of dispersion explained: 0.544 0.670 posite of what a clinal model based on the large size of the Monteverde specimen would predict. However, the slope is not statistically different from zero, and it ex- plains only a small percentage of the varia- tion (F = 1.92, p = 0.180, d.f. = 22, adjusted R2 = 4.5%), indicating little geographic variation in size. The second smallest indi- vidual in the analysis is from Nicaragua, which neighbors Costa Rica to the north. The Finca San Bosco specimen?the only other specimen of C. merriami from Costa Rica?is average size for the species (first factor axis coordinate = 0.05364). In con- trast, the individual from Monteverde, the only locality where the species occurs in sympatry with C. nigrescens, is considerably TABLE 4. Approximate latitude and longitude for collection localities of C. merriami and C. nigrescens. The factor score for each locality is from the first factor axis of a principal component analysis of all pairs of coordinates (see text). Geographic coordinates derive from gazeteers published by the United States Board on Geographic Names, from maps of the regions, and from original specimen tags. Locality Latitude Longitude Factor score 15?40' 91?44' 2.47670 15?48' 91?18' 2.41113 15?29' 90?29' 2.14180 15?18' 90?04' 2.00206 14?59' 88?18' 1.51491 14?40' 88?38' 1.51182 14?13' 87?04' 1.03979 13?56' 86?57' 0.93859 Guatemala: Jacaltenango Barillas San Pedro Carcha Tucuru Honduras: San Jose de Colinas Las Flores San Juancito Montana de Monserrat El Salvador: Cerro Cacaguatique Nicaragua: Santa Maria de Ostuma Costa Rica: Tilaran Monteverde San Luis Cinchona Volcan Irazu San Isidro del General San Vito Las Tablas Panama: Volcan de Chiriqui Fish Camp 19 km NNW El Volcan Finca Santa Clara Cerro Punta Hato del Volcan 23 to 25 km NNE San Felix 13?46' 12?57' 10?28' 10?18' 10?17' 10?13' 9?59' 9?23' 8?50' 8?57' 8?48' 8?58' 8?56' 8?51' 8?52' 8?46' 8?27' 88?13' 85?58' 84?58' 84?47' 84?47' 84?11' 83?51' 83?42' 82?58' 82?46' 83?33' 82?40' 82=42' 82?45' 82?35' 82?38' 81?48' 1.18915 0.46491 -0.39446 -0.47204 -0.47712 -0.63912 -0.77576 -0.96259 -1.26971 -1.28508 -1.14434 -1.30293 -1.30621 -1.31507 -1.34897 -1.36291 -1.63453 CRYPTOTIS MERRIAMI IN COSTA RICA 297 larger than the range of size variation pre- sent throughout the geographic distribu- tion of the species. Woodman and Timm (1993) demon- strated that C. nigrescens at Monteverde is distinctly smaller than conspecifics from near San Felix, Panama, and suggested that C. nigrescens could be a complex of species. However, the plot of size (Fig. 4, Table 3) against geographic locality (Table 4) shows a gradual transition from populations of smaller individuals in northern Costa Rica to populations of larger individuals in western Panama. Regression analysis sup- ports this trend (y = 3.39 + 3.55x; F = 103.21, p < 0.001, d.f. = 81, adjusted R2 = 55.8%). Rather than representing different taxa, C. nigrescens from Monteverde and San Felix represent two extremes of a gradual size cline. DISCUSSION Throughout its distribution, C. merriami exhibits no regular pattern of geographical variation in size, indicating that clinal variation does not account for the ex- tremely large size of the Monteverde speci- men. One possibility is that the Monteverde specimen represents an undescribed spe- cies closely related to C. merriami. How- ever, in that case, the same suite of charac- ters that supports its conspecificity with C. merriami would indicate that the two "spe- cies" are sister taxa, and the original ques- tion remains?why is the Monteverde specimen so big? Environmental conditions at Monteverde and near Finca San Bosco, ca. 25 km to the north, are similar, and they certainly differ less than between northern Costa Rica and anywhere else in the geographic range of C. merriami. The one environmental variable that is most obviously different at Mon- teverde is the presence of C. nigrescens. It seems likely that the large size of C. mer- riami at Monteverde represents character displacement (Brown and Wilson, 1956) to facilitate coexistence by minimizing com- petition with C. nigrescens. Character displacement has been docu- mented in a number of mammals, includ- ing insectivores. Choate (1970:238) ob- served that Cryptotis peregrina and Cryptotis goldmani from Oaxaca, Mexico, were easier to distinguish where the two species lived in sympatry than where they occurred alone. In northern Europe, Malmquist (1985) showed that the European pygmy shrew, Sorex minutus, averaged signifi- cantly smaller in regions where it was sym- patric with the European common shrew, Sorex araneus, than in regions where it ex- isted alone. In contrast, the common shrew exhibited no differences between regions where it was sympatric with S. minutus and regions where the pygmy shrew was ab- sent. The lack of character displacement in S. araneus is similar to the lack of size dis- placement in C. nigrescens at Monteverde. The smaller size of C. nigrescens at Mon- teverde is part of a gradual trend of de- creasing size from western Panama through northern Costa Rica (Fig. 4). But, if larger size is an adaptation that evolved in C. merriami to reduce competition with C. nigrescens at Monteverde, then why does C. nigrescens show no corresponding response (further diminution in size) to the presence of C. merriamil In Europe, the lack of mor- phological response in S. araneus was at- tributed to the low abundance of S. minutus in areas of sympatry and hence, the expo- sure of S. araneus "to only a weak interspe- cific competitive pressure" (Malmquist, 1985:375). A corresponding situation may exist at Monteverde, where the numerically dominant C. nigrescens has remained un- changed, and the less abundant C. merriami exhibits a distinct increase in size. Larger size allows predators to utilize prey sizes unavailable to smaller predators, whereas smaller size less typically bestows a trophic advantage (Wilson, 1975). However, Dick- man (1988) argued that size variation in in- sectivorous mammals does not permit spe- cialization on different sizes of prey. He hypothesized that larger insectivorous spe- cies can partly exclude smaller insectivores from microhabitats where large prey are more abundant. If the larger size of C. mer- riami permits assymetrical competition fa- voring its occurrence at Monteverde, it would be maladaptive for C. nigrescens to evolve smaller size. 298 N. WOODMAN Several studies have attempted to de- termine how much of a size difference is required for two members of the same tro- phic guild to coexist. In general, magni- tudes of difference show great variability, depending upon what taxa and what mea- surements are considered. Hutchinson (1959) suggested that sympatry would re- quire a two-fold difference in mass, which he equated to a size difference ratio of ap- proximately 1.3 (range = 1.1-1.4), with size represented for mammals by length of skull. Similar ratios have been reported among Israeli canids (Dayan et al., 1992) and Miocene hyaenas (Werdelin, 1996), based on carnassial lengths. Sympatric ger- billid rodents exhibit size ratios ranging from 1.0 to 1.3 (Yom-Tov, 1991), whereas, a size ratio of 1.6 may be the norm among folivorous mammals (Maiorana, 1990). Malmquist (1985) used the minimum dis- tance between the upper second molars to investigate character displacement in Sorex minutus and S. araneus. The size ratio for this character in sympatry (1.4) was larger than in allopatry (1.3). Using a similar tro- phic measurement, breadth-across-upper- second-molars (Table 1), I determined the size ratio between C. nigrescens and typical C. merriami is 1.1, whereas the difference between C. nigrescens and the Monteverde specimen of C. merriami is 1.2?an increase in magnitude equivalent to that of the Eu- ropean shrews. Unlike the European soricid community (Malmquist, 1985), the Monteverde com- munity includes a third shrew, Cryptotis sp. "a"?a rare species more closely related to C. gracilis and C. endersi than to C. nigrescens or C. merriami. Using breadth-across-upper- second-molars (5.8 mm), Cryptotis sp. "a" is equal in size to typical C. merriami (Table 1) and is intermediate between C. nigrescens and C. merriami from Monteverde. The size ratios between Cryptotis sp. "a" and C. ni- grescens, and between Cryptotis sp. "a" and C. merriami, are both 1.1. Hence, the larger size of C. merriami at Monteverde may be a response to the presence of an abundant sister species (C. nigrescens), the presence of an unrelated shrew the size of more typical C. merriami {Cryptotis sp. "a"), or, most likely, the presence of both species. Fox and Kirkland (1992; see also Kirkland, 1991) suggested that an assem- blage rule, initially formulated for Austra- lian small mammals, may apply to commu- nities of forest-dwelling shrews in eastern North America. This rule predicts that community assembly proceeds by func- tional groups (i.e., size groups) and that each such group should be represented by a single member before a second member of a functional group can be added. The functioning of this assemblage rule for in- sectivores presumably is enforced by com- petition for distinct classes of prey (e.g., Churchfield, 1991), although it may be brought about through interference compe- tition (Dickman, 1988) or another form of coarse-grained habitat partitioning (Kirk- land, 1991; Churchfield, 1991). If a similar assemblage rule functions in Central American forests, it would preclude the co- occurrence of similarly-sized species of shrews and should promote character dis- placement in size, as seen in C. merriami at Monteverde. The functioning of such an as- semblage rule could be tested by closely examining where and on what different species in the community are feeding throughout the year. Although the Costa Rican fauna has been intensely studied, large gaps remain in our knowledge of the region's mammals. Par- ticularly little is known regarding the dis- tribution and ecology of the region's small- est terrestrial mammals, the shrews. As predators of insects and other important soil- and plant-inhabiting invertebrates, shrews may play a crucial role in Neotropi- cal systems. Their role may be better de- fined through additional study of the com- position of soricid communities in tropical habitats and the interactions of community members. Acknowledgments.?Special thanks to Richard K. LaVal of Monteverde, Bernal Rodriguez (MNCR), and Don E. Wilson (USNM) for making important specimens of Costa Rican Cryptotis available for study. Eric Lundquist, John Simmons, and William E. Duellman provided valuable as- sistance regarding the Finca San Bosco lo- cality. Robert M. Timm, Norman A. Slade, CRYPTOTIS MERRIAMI IN COSTA RICA 299 and Thor Holmes permitted use of the KU Mammal Division to conduct this project. Robert P. Anderson, Jerry R. Choate, and Robert M. Timm provided insightful com- ments on previous versions of this manu- script. LITERATURE CITED Brown, W. L., Jr., and E. O. Wilson. 1956. Character displacement. Syst. Zool. 5:49-64. Choate, J. R. 1970. Systematics and zoogeography of Middle American shrews of the genus Cryptotis. Univ. Kansas Publ., Mus. Nat. Hist. 19:195-317. Churchfield, S. 1991. Niche dynamics, food resources, and feeding strategies in multispecies communities of shrews. Mus. Southwest. Biol., Spec. Publ. 1:23- 34. Dayan, T., D. Simberloff, E. Tchernov, and Y. Yom- Tov. 1992. Canine carnassials: character displace- ment in the wolves, jackals, and foxes of Israel. Biol. J. Linn. Soc. 45:315-331. Dickman, C. R. 1988. Body size, prey size, and com- munity structure in insectivorous mammals. Ecol- ogy 69:569-580. Fox, B. J., and G. L. Kirkland, Jr. 1992. An assembly rule for functional groups applied to North Ameri- can soricid communities. J. Mamm. 73:491-503. Hutchinson, G. E. 1959. Homage to Santa Rosalia or why are there so many kinds of animals? Amer. Nat. 93:145-159. Kirkland, G. L., Jr. 1991. Competition and coexistence in shrews (Insectivora: Soricidae). Mus. Southwest. Biol., Spec. Publ. 1:15-22. Maiorana, V. C. 1990. Evolutionary strategies and body size in a guild of mammals. In J. Dalmuth and B. J. MacFadden (eds.), Body size in mammalian paleobiology: estimation and biological implica- tions, pp. 69-102. Cambridge Univ. Press, New York, 104 pp. Malmquist, M. G. 1985. Character displacement and biogeography of the pygmy shrew in northern Eu- rope. Ecology 66:372-377. Werdelin, L. 1996. Community-wide character dis- placement in Miocene hyaenas. Lethaia 29:97-106. Wilson, D. L. 1975. The adequacy of body size as a niche difference. Amer. Nat. 109:769-784. Woodman, N. 1995. Morphological variation between Pleistocene and Recent samples of Cryptotis (Insec- tivora: Soricidae) from the Yucatan Peninsula, Mexico. J. Mamm. 76:223-231. Woodman, N. 1996. Taxonomic status of the enigmatic Cryptotis avia (Mammalia: Insectivora: Soricidae), with comments on the distribution of the Colom- bian small-eared shrew, Cryptotis colombiana. Proc. Biol. Soc. Wash. 109:409-418. Woodman, N, and R. M. Timm. 1992. A new species of small-eared shrew, genus Cryptotis (Insectivora: Soricidae), from Honduras. Proc. Biol. Soc. Wash. 105:1-12. Woodman, N, and R. M. Timm. 1993. Intraspecific and interspecific variation in the Cryptotis nigre- scens species complex of small-eared shrews (Insec- tivora: Soricidae), with the description of a new species from Colombia. Fieldiana: Zool. (n. s.) 74: 1-30. Woodman, N, and R. M. Timm. 1999. Geographic variation and evolutionary relationships among broad-clawed shrews of the Cryptotis goldmani- group (Insectivora: Soricidae). Fieldiana: Zool. (n. s.) 91:1-35. Yom-Tov, Y. 1991. Character displacement in the psammophile Gerbillidae of Israel. Oikos 60:173- 179.