# Demographic features and habitat preferences of Osgoodomys banderanus (Osgood's deermouse) in Colima, Mexico.

Osgoodomys banderanus (Osgood's deermouse) is a nocturnal,
solitary, medium-sized cricetid that is endemic to western Mexico, in
particular the western portion of the Trans-Mexican Neovolcanic Belt (Fa
and Morales, 1991). The species occurs throughout the state of Colima
(Ceballos and Miranda, 2000; Fig. 1; Appendix lists localities). Its
distribution is limited along the coastal plain from southern Nayarit to
southern Guerrero and into the interior of Michoacain and Guerrero along
the basin of the Rio Balsas, Mexico (Musser and Carleton, 2005). The
range of O. banderanus is extensive (230,000 km2), and the species is
not considered to be of particular conservation concern (Glazier, 1980),
but it occurs primarily in Pacific tropical to subtropical dry forest,
an area that increasingly is being subjected to habitat degradation and
widespread clearing (Sanchez-Azofeifa et al., 2005). In addition, O.
banderanus occurs in hotspots for biodiversity and endemism, as well as
in regions considered a priority for conservation of nonvolant mammals
(Ceballos and Navarro L., 1991; Fa and Morales, 1991; Arita et al.,
1997).

[FIGURE 1 OMITTED]

Relatively little is known about the basic ecology and demography of O. banderanus, a condition fostered in part by its varied taxonomic treatment. There has been debate over several decades concerning placement of the species within the subfamily Neotominae. Originally assigned in the genus Peromyscus, Carleton (1980), in a morphological study, was the first to advocate that the species be classified in its own genus, a recommendation followed by Musser and Carleton (2005).

The ecology and habitat use of O. banderanus has been described only in general terms, mostly in brief field remarks, with only a few studies reporting on specific demographic parameters. The primary work on the species has been by Ceballos Gonzalez (1989) as part of his study of small mammals of deciduous and semi-deciduous forests at the Chamela Biological Station in Jalisco.

Given the relative paucity of information on demographic features and ecology of O. banderanus, mark-recapture studies were conducted with the specific objectives to describe habitat preferences of O. banderanus, as well as to report sex ratio, reproductive status, age structure, and population density. Additional objectives were to acquire information on use of understory, probability of capture, and movements of males and females.

Materials and Methods--Study Area and Trapping--This study was conducted at Rancho La Angostura, Colima (locality 14, Fig. 1) in 2006 and sites in the vicinity of the Miguel de la Madrid Airport (locality 17) in 2006 and 2007. Five trapping grids were established each year (10 total) during the annual dry season in January. Days were warm and nights were cool (January average during 1961-2005 for Manzanillo was 24.6[degree]C; Instituto Nacional de Estadistica, Geografia e Informatica, 2006), with little day-today variation in temperature and no rain. Grids were placed to include the major types of vegetation present. Each grid consisted of 100 trapping stations (10 by 10 square grid), with stations 10 m apart. Two Sherman live traps (7.5 by 9.0 by 23.0 cm) baited with rolled oats were placed at each station, one on the ground and one elevated 1-2 m above ground on a thin plywood platform (12.5 by 34.5 cm) attached to a tree or shrub. Grids were sampled for 8 nights each; thus, overall sampling effort on the grids involved 16,000 trap-nights.

Grids 1-2 in 2006 at Rancho La Angostura (sampled 2-6, 10-12 January) were 700-765 m in elevation, and vegetation on grids was tropical deciduous forest (bordered by a small stream) in an overall landscape that was fragmented due to cultivated agriculture (mainly oranges and limes) and raising of livestock. However, the area containing the grids had native vegetation (some Ficus trees [less than or equal to] 5 m in circumference) with heavily shaded gallery forest (trees 5-40 m in height). Numerous species of trees were present, the most abundant being Alnus, Salix humboldtiana, S. bonplandiana, Cecropia obtusifolia, Ficus, and Inga eriocarpa (Schnell et al., 2008b). Cultivated coffee trees (Coffea arabica) were in the understory on both grids at this location.

The second locality, adjacent to the Miguel de la Madrid Airport, included grids 3-5 in 2006 (sampled 3-6, 10-13 January) and all five grids in 2007 (grids 1-3 sampled 2-6, 1012 January and grids 4-5 sampled 3-6, 10-13 January). None of the grids overlapped within or between years. The study area had been fragmented and used for agriculture in the past but was protected and relatively undisturbed for 30 years prior to our study. Much of the area studied had reverted to tropical moist forest intermixed with grass-dominated habitat (Andropogon, often >2 m tall). Characteristic trees were Brosimum alicastrum, Sideroxylon cartilagineum, Coussapoa purpusii, Hura polyandra, Guarea glabra, Dendropanax arboreus, and Populus guzmanantlensis (Schnell et al., 2008b).

Grids 3-5 in 2006 were predominately thorn forest. Grid 4 had patches of tall grass associated with small stream and riparian habitat of dense vines and large deciduous trees. Grids 3 and 5 had sparse understory vegetation intermixed with patches of tall grass (1.0-1.5 m). In 2007, grids 1 and 2 had tropical deciduous forest with mature trees and thorn-forest elements, with grid 1 featuring a sparse understory and a partially cleared, unimproved road traversing the grid. Grids 3 and 4 had dense thorn forest with vines creating low, thick canopy cover, while some areas had mature deciduous trees and patches of tall grass in open areas. Grid 5 featured tall dense grass (1.0-3.0 m) with isolated trees.

Traps were checked once each day starting at dawn, rebaited if needed, and opened for the full 24-h period. For each O. banderanus captured, we recorded position of trap (i.e., trapping station on the grid and ground versus elevated), gender, reproductive status (based on external condition of reproductive organs), and age (adult, subadult, or juvenile based on size, mass, and pelage coloration). Typically, animals were tagged in both ears using uniquely numbered Monel No. 1 ear tags (National Band and Tag Company, Newport, Kentucky) to ensure that individuals could be identified throughout the course of study. Mice were released at site of capture.

Taxonomy follows Musser and Carleton (2005), except Liomys is treated as being part of Heteromys as proposed by Hafner et al. (2007). Secondary verifications of identity of species were made by comparison of mitochondrial DNA sequences. Sequence patterns from a region 680 nucleotides long of the cytochrome b gene were generated from tissue samples collected at the time of original capture (data not shown). Sequences were compared visually to multiple sequences available in the GENBANK repository (Benson et al., 2009) representing known cytochrome b sequences from Osgoodomys. Because of the close morphological association between Osgoodomys and Peromyscus, comparisons also were made to multiple species of Peromyscus (P. aztecus, P. hylocetes, P. spicilegus) reported to be present in the state of Colima. Similarity between nucleotide sequences of the field-collected specimens and GENBANK sequences of Osgoodomys were all >99%.

Estimation of Movement, Probability of Capture, Abundance, and Density--For each O. banderanus captured more than once, we calculated the centroid of capture sites. Distance from centroid was calculated for each trapping location and the mean 0determined for each individual. This measure of movement has been used previously for Baiomys musculus and Sigmodon mascotensis (Schnell et al., 2008b, 2010b). Movement also was assessed by determining average distance between successive captures (Slade and Swihart, 1983), which was calculated weighting each occurrence equally (e.g., an individual caught five times would contribute four data points while one caught two times would provide one).

The computer program MARK (Version 6.0; G. C. White, http://warnercnr.colostate.edu/~gwhite/mark/mark.htm) was used to estimate abundance (i.e., population size) of O. banderanus and to estimate probability of capture on a given trapping night. Assuming a closed population during the trapping period, the program compared relative proportions of marked and unmarked animals in successive samples to estimate abundance. We used the Huggins closed-captures option (Huggins, 1989, 1991), in which the two parameters estimated were p, the probability of capture, and c, the probability of recapture; these in turn were involved in deriving N, the abundance for a given grid. Parameters p and c were considered equal in our model but could vary with time (t; i.e., the 8 trapping nights were estimated independently of one another); this model is indicated by p(t) = c(t). Probabilities of capture and recapture for the 8 trapping nights were considered to be the same for grids in 2006 and the same for grids in 2007, but not necessarily the same for the 2 years; thus, 16 parameters in total were being estimated. This model was chosen because it produced the most reliable results based on the Akaike Information Criterion as modified to adjust for small-sample bias ([AIC.sub.c]; Version 6.0; G. C. White, http://warnercnr.colostate.edu/~gwhite/mark/mark.htm). Other models (i.e., where all grids were considered separately, grids were grouped by specific locality, or all grids were considered equal) yielded relatively similar estimates of total numbers of individuals present.

Converting abundance values to densities requires an estimate of the effective area of a grid. The immediate area of the grid (8,100 [m.sup.2]; 90 by 90 m) does not include area used outside the outer traps. To estimate the edge area, we employed average distance from centroid, with equal weight given to each individual mouse caught more than once, a technique used by Schnell et al. (2008b, 2010b). The resulting value was added to each side of the central portion of the grid and corners of the grid area were treated as being rounded with a radius equal to average distance from centroid. Using distance from centroid as reported in Results, the effective area of a grid was estimated to be 1.320 ha.

We also analyzed probability of capture relative to moonlight. Information on the fraction of the moon illuminated for each trapping night was obtained from the United States Naval Observatory (http://www.usno.navy.mil/USNO/astronomical-applications/data-services/frac-moon-ill).

Structure of Vegetation--Following Schnell et al. (2008a, 2008b), we quantified environmental variation by evaluating 14 characteristics, most of which reflected structure of vegetation (Table 1). Measurements were made at points 1 m from each trapping station (1,000 points total). Percentage of ground cover was estimated (to nearest 5%) for a 1-m square (first seven variables in Table 1). Number of vegetation stems hitting a 1-m bar at 1-m height was determined four times (once in each cardinal direction from central point) and the average calculated (variable 8). Canopy cover (i.e., percent closed, variable 9) was estimated with a spherical densitometer (model C; Forest Densitometers, Bartlesville, Oklahoma), as was slope (variable 10) with a clinometer. Using a 7.5-m vertical pole marked at each decimeter, we recorded the number of decimeter intervals within which vegetation touched the pole; data were totaled for 0-2.5 m (maximum of25 hits; variable 11) and for 2.5-7.5 m (maximum of50 hits; variable 12). Maximum height of canopy was estimated to the nearest 0.5 m (variable 13). Distance to nearest tree ([greater than or equal to] 10 cm dbh; diameter at breast height) was determined in each of four quadrants (with edges being the cardinal directions) and the average taken (variable 14); distances [greater than or equal to] 10 m were tabulated as 10 m.

Habitat Analyses--For each grid, we obtained the mean and SD of the 14 environmental variables, comparing these 28 measurements (means and SDs) for grids where O. banderanus was present with those for grids where it was not detected. Parallel analyses involved the contrast of means and SDs of environmental variables for trapping stations where the species was captured and not captured.

Stepwise logistic regression (SYSTAT Software, Inc., 2007) was used to examine the relationship between environmental measures at trapping stations where O. banderanus was captured and not captured. The dependent variable was logit (Y), which is the natural logarithm of the odds of the species being present, and 14 environmental measures were used as potential independent variables. A variable was entered into the equation if [alpha] < 0.05; after adjustment of the equation taking into account the additional variable, a variable was removed if [alpha] > 0.045. McFadden's [[rho].sup.2]-statistic (range, 0-1) was used to evaluate resulting models as a whole.

Nonparametric multiplicative regression (program Hyper-Niche; McCune and Mefford, 009) also was used to model habitat preferences based on the 14 environmental variables. Nonparametric multiplicative regression effectively represents response of a species to multiple habitat factors, factor interactions are automatically accommodated for, and overall form of the response surface need not be specified in advance (B. McCune, http://www.pcord.com/NPMRintro.pdf). The technique has been used in analysis of habitat preference of other small mammals, including Nyctomys sumichrasti and Sigmodon mascotensis (Schnell et al., 010a, 010b).

We employed the local-mean model type, where weighting around each target point was based on a Gaussian (humpbacked) function. In local-mean models, tolerance (i.e., SD of a Gaussian weighting function) is related inversely to importance of variables in models (B. McCune, http://www.pcord.com/NPMRintro.pdf). To select the best overall model, we used Log B as a measure of fit.

Other Statistical Analyses and Software--SigmaPlot 11 (Systat Software, Inc., 2008) was used to conduct linear- and multiple-regression analyses. BIOMstat for Windows 3.3s (Rohlf and Slice, 1999) was used in univariate analyses, including RxC contingency, goodness-of-fit, and Fisher's exact tests for analysis of frequencies, and linear regression, and analysis of variance (ANOVA) to assess continuous variables.

Results--Trapping--Osgoodomys banderanus was caught on 6 of 10 grids, with 1-40 individuals on a grid (Table 2), for a total of 123 individuals and 385 captures. On average, we caught males 2.9 [+ or -] 1.7 (SD) times and females 3.5 [+ or -] 2.1 times, with an overall average of 3.1 [+ or -] 1.9 captures/individual for the 8-night trapping sessions. We recaptured 69.9% of individuals, with no difference for sex or year (Fisher's exact tests; P = 1.000 and 0.156, respectively). However, there was a significant difference for age, with 75.7% of adults being recaptured in contrast to 42.1% of juveniles or subadults (Fisher's exact test, P = 0.006).

One-third of all captures were in elevated traps (128 of 385; 33.2%) with a range of 27.3-42.4% on individual grids, but no significant difference was detected among grids (RxC test, [G.sub.4] = 5.62, P = 0.229). Of individuals captured more than once (86 individuals), 68.6% were captured in both elevated and ground traps, 25.6% only in ground traps, and 5.8% only in elevated traps.

[FIGURE 2 OMITTED]

Probability of Capture Relative to Trapping Night and Moonlight--For trapping nights, we tabulated the proportion of the moon illuminated, which during our study was 0.07-1.00. Lunar illumination, when considered alone or in combination with trapping night, statistically did not explain the probability of capturing O. banderanus. However, based on data for both years, a nonlinear regression (Fig. 2) was highly significant for trapping night, with an increase in probability of capture from nights 1-6 with a gradual decline thereafter ([F.sub.3, 12] = 5.81, P = 0.011, [R.sup.2] = 0.592). A similar pattern emerged when analyzing years separately. For both years, the first trapping night had the lowest capture probability (<0.3) and trapping probability tended to increase with successive nights until it peaked and decreased on night 6 or 7 (Fig. 2).

Demographic Characteristics--In both years, adult O. banderanus were captured more frequently than subadults or juveniles: 2006, 81.6% adults (62 adults of 76); 2007, 89.1% adults (41 of 46); and both years, 84.4% adults (103 of 122). No significant difference between years was detected (Fisher's exact test, P = 0.312).

We captured more adult males than adult females: 2006, 56.5% males (35 males and 27 females); 2007, 65.9% males (27 and 14); and both years, 60.2% males (62 and 41). For 2006, deviation from a 1:1 male: female ratio could not be demonstrated statistically (goodness-of-fit test, [G.sub.1] = 1.04, P = 0.309). However, for 2007, significantly more males were caught than females ([G.sub.1] = 4.19, P = 0.041). Likewise, for both years pooled, there were significantly more males than females ([G.sub.1] = 4.31, P = 0.038). No statistical heterogeneity was detected in sex ratio among years ([G.sub.1] = 0.92, P = 0.338), and overall sex ratio of adults (males: females) of captured individuals was 1.51:1.

For adult females, percentages of pregnant or lactating individuals were: 2006, 48.1% (13 of 27); 2007, 28.6% (4 of 14); and both years, 41.5% (17 of 41). No significant difference between years was demonstrable (Fisher's exact test, P = 0.321).

Mean mass of eight pregnant females was 40.17 [+ or -] 1.66 g, while 33 non-pregnant females averaged 35.63 [+ or -] 6.57 g; these two groups were not significantly different (one-way ANOVA, [F.sub.1, 39] = 3.71, P = 0.061). Mean mass for adult males was 37.48 [+ or -] 6.19 g (n = 62, range 22.7-51.0 g). For 95 adult animals (males and non-pregnant females), a two-way ANOVA detected no significant difference in mass for sex or year, nor was there significant interaction (respectively, [F.sub.1, 91] = 1.30, 0.03, and 0.86; P = 0.257, 0.859, and 0.356).

Movement, Abundance, and Density--Overall mean of average distances from centroid for adult males was 14.01 [+ or -] 10.35 m (n = 47, range 0-40.6 m), while that for adult females was 11.18 [+ or -] 5.69 m (n = 31, range 0-28.9 m). A two-way ANOVA detected no significant difference for sex, year, or interaction ([F.sub.1, 74] = 0.02-1.75, all P > 0.050). Average distances from centroid were not significantly different among grids (one-way ANOVA, [F.sub.4, 73] = 0.19, P = 0.942). Pooling O. banderanus, irrespective of sex or age, average distance from centroid was 12.74 [+ or -] 9.48 m (n = 86, range 0-40.6 m). Using this value, the effective area of a grid was estimated to be 1.320 ha.

Average distance moved between successive captures did not differ statistically between sexes (one-way ANOVA, [F.sub.1, 239] = 2.64, P = 0.105). Adult males moved 21.40 [+ or -] 21.95 m (n = 132, range 0-94.9 m), females 17.47 [+ or -] 13.83 m (n = 109, range 0-70.7 m), and both combined 19.62 [+ or -] 18.79 m (n = 241, range 0-94.9 m). We also did calculations weighing each individual equally with respect to movement between successive captures, but results were not appreciably different from those reported above.

Based on our model derived in program MARK, abundances for grids were 1.00-40.31 individuals/grid (Table 2). These values translate to density estimates of 0.76-30.55 individuals/ha with an average of 15.85 individuals/ha on grids where the species was present.

Co-occurring Species and Relative Abundance--There was a significant difference between years in the percent of small mammals captured that were O. banderanus (Fisher's exact test, P < 0.001), with the species comprising 16.5% (77 of 468) in 2006 and 5.5% (46 of 836) in 2007. The two localities (localities 14 and 17, Fig. 1) also significantly differed, with O. banderanus constituting 40.1% (61 of 149) of all small mammals at locality 14, but 5.4% (62 of 1,155) at locality 17 (Fisher's exact test, P < 0.001). Relative abundance on grids varied as well, with O. banderanus comprising 0.6-43.4% of the small mammals.

Other species most commonly encountered in 2006-2007 were Sigmodon mascotensis (West Mexican cotton rat; 352 individuals; 27.0%), Heteromys pictus (painted spiny pocket mouse; 313; 24.0%), Baiomys musculus (southern pygmy mouse; 301; 23.1%), and Reithrodontomys fulvescens (fulvous harvest mouse; 131; 10.1%). Numbers of individuals captured of Oryzomys couesi (Coues' oryzomys; 34; 2.6%), Tlacuatzin canescens (gray mouse opossum; 29; 2.2%), Nyctomys sumichrasti (Sumichrast's vesper rat; 14; 1.1%), and Hodomys alleni (Allen's woodrat; 6; 0.5%) were lower than for O. banderanus (123; 9.4%). All of the abovementioned species occurred on grids with O. banderanus.

Habitat--Contrasting means and SDs of the 14 environmental variables for the six grids where O. banderanus was present versus the four grids where it was not found, four significant differences were identified: mean percent grass, 12.0% for grids where present versus 47.2% for grids where absent ([F.sub.1, 8] = 7.98, P = 0.022); mean total hits high on a vertical pole, 5.14 versus 1.28 ([F.sub.1, 8] = 9.86, P = 0.014); mean average distance from nearest tree, 3.4 versus 6.9 m ([F.sub.1, 8] = 11.63, P = 0.009); and SD of percent forbs, 5.93 versus 15.10% ([F.sub.1, 8] = 21.37, P = 0.002). Thus, grids with O. banderanus present had less grass, thicker vegetation high above the ground, shorter distances to nearest tree, and less variation in percentage of forbs.

Evaluating mean values and SDs of environmental variables for trapping stations where O. banderanus was caught and not caught for 2006, 2007, and the combined years (Table 1), 10 variables were statistically different in all three comparisons. Ground cover at stations where O. banderanus was captured had lower percentages of grasses and higher percentages of leaf litter and rocks. In addition, capture sites relative to noncapture sites had fewer hits at 1 m, a more closed canopy, greater slope, fewer hits low and more hits high on a vertical pole, a higher canopy, and shorter average distance to nearest trees.

Stepwise logistic regression for 2006-2007 combined (10 grids, 1,000 trapping stations), contrasting stations where O. banderanus was caught and not caught, resulted in the following equation containing four of the 14 environmental variables:

logit(Y) = -0.883 - 0.027[X.sub.3] - 0.375[X.sub.14] + 0.025[X.sub.6] + 0.041[X.sub.13];

where logit (Y) is the natural logarithm of the odds of the species being present, [X.sub.3] is percent grass, [X.sub.14] is average distance to nearest tree, [X.sub.6] is percent rocks, and [X.sub.13] is maximum canopy height (Table 3). McFadden's p2 was 0.263, indicating a robust model. Projection values for sites where O. banderanus was trapped tended to be greater than for those where the species was not caught (Fig. 3a). Coefficients for the first two variables were negative, indicating that stations where O. banderanus was trapped tended to have less grass (Fig. 3b) and closer trees (Fig. 3c). Coefficients for the last two environmental variables were positive, with stations where the species was caught typically having more rocks (Fig. 3d) and a higher canopy (Fig. 3e).

Using nonparametric multiplicative regression, the best model (log B = 51.58) for predicting likelihood of occurrence of O. banderanus incorporated three variables: average distance to nearest tree (tolerance 0.65), percent rocks (tolerance 7.20), and maximum canopy height (tolerance 8.75). Average neighborhood size of the resulting model was 73.5. Likelihood of occurrence of O. banderanus was highest (ca. 0.3-0.4) when on average trees were 1-3 m away, with likelihood falling sharply to <0.1 when trees were >4 m distant (Fig. 4a). For percent rocks (Fig. 4b), likelihood of occurrence tended to increase as percentage of rocks increased, with likelihood low (<0.3) in areas with <20% rocks and highest (0.7) at 80% rocks; however, areas with 90% rocks also had a low likelihood for occurrence. For maximum canopy height (Fig. 4c), likelihood of occurrence increased as canopy height rose with likelihood peaking above 0.5 when canopies were 32 m in height.

DISCUSSION--Trapping Response--Little information exists on arboreal activity of tropical rodents (Delany, 1971), which Dominguez-Castellanos et al. (2007) suggested could lead to a misrepresentation of population densities and activity patterns. We had equal numbers of ground and elevated traps, resulting in one-third of our captures of O. banderanus coming in elevated traps. Using traps placed 1-3 m above ground, Ceballos Gonzalez (1989) had a nearly identical proportion of captures from elevated traps (33.5%; n = 95) during monthly, 3-night studies (15 months total) on grids in deciduous and semideciduous forests at Chamela Biological Station in Jalisco. However, only ca. 20% of his traps were placed above ground during his 20,412 trap-nights. Dominguez-Castellanos et al. (2007) reported a lower aboveground use for O. banderanus of 10% at Chamela Biological Station but did not indicate relative numbers of elevated and terrestrial traps. In Para, Brazil, Lambert et al. (2005) compared species richness and rates of species accumulation using traps placed at three heights: ground, understory (similar to our elevated traps), and canopy. Many species of small mammals at their site were captured in the understory and rarely on the ground; however, no species was significantly more abundant in the canopy than in the understory traps, rendering understory traps as effective as canopy traps. Yet, in Manaus, Brazil, Malcolm (1991) reported that canopy trapping revealed a rich and abundant fauna not evident when traps were set on the ground or in the understory. However, there was little difference in numbers of species or individuals caught in ground and understory traps.

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

Over 68% of O. banderanus we captured more than once frequented both elevated and ground levels. For another semiarboreal species, Reithrodontomys fulvescens, we found a significantly lower proportion of animals (<49% of individuals) used both levels (Fisher's exact test, P = 0.037; unpublished data). It is clear that O. banderanus readily and regularly made use of both ground and elevated sites.

Probability of Capture Relative to Trapping Night and Moonlight--Some studies suggest that bright moonlit nights negatively affect trapping success for certain rodents (Blair, 1951; Price et al., 1984). Subsequently, fieldwork often has been conducted intentionally during the new moon when rates of capture are presumed to be highest (Prugh and Brashares, 2010), as done by Ceballos Gonzalez (1989) and Dominguez-Castellanos et al. (2007). We detected no relationship between lunar phase and probabilities of capturing O. banderanus. In an analysis of similar data from Colima (Januarys of 2003-2007), Schnell et al. (2008b) also detected no association for Baiomys musculus. It appears that the influence of moonlight on activity of small mammals is greater in more open habitats, such as deserts and grasslands.

In our 8-night study, night of trapping was a significant factor in the probability of catching O. banderanus, with probability of capture increasing nightly until night 6 or 7 and decreasing thereafter. Only a few comparable investigations have reported nightly temporal affects on trapping success. In an 8-night study during February in South Africa, van Hensbergen and Martin (1993) determined that probability of capturing the murid rodent Rhabdomys pumilio (xeric four-striped grass rat) increased steadily during nights 1-4 and decreased thereafter. Other studies conducted for 8-nights in Colima during January also found trapping night to influence significantly the probability of capturing Baiomys musculus (Schnell et al., 2008b) and Sigmodon mascotensis (Schnell et al., 2010b). Two basic patterns were documented in both species; probabilities of capture increased from nights 1-4 then decreased thereafter, or continued increasing until night 6 or 7 before decreasing.

Demographics and Movement--Mass of O. banderanus seldom has been documented. Davis (1944) examined six specimens and reported non-pregnant females in Guerrero averaged 43 g, while a pregnant female (in August) containing three nearly full-term embryos weighed 57.9 g (suggesting young weigh about 4 g each at birth). Ceballos Gonzalez (1989) reported average mass of 14 individuals in Jalisco to be 38.3 [+ or -] 5.6 g, a value similar to our average mass of adults of 37.1 g.

Reproduction in O. banderanus likely occurs year-round with the reproductive pattern reported to be continual polyestrous (Mendoza Duran and Miranda, 2002; Arroyo-Cabrales, 2005). Of females we captured in January, > 40% were reproductively active. This is similar to another study conducted during January by Sanchez-Hernandez et al. (2009) at Ejido Ranchitos, Colima. They reported that about one-third of 44 female O. banderanus were pregnant, lactating, or both.

Overall, males significantly out-numbered females 1.51:1 in our traps, and nearly 2:1 in 2007. Sanchez-Hernandez et al. (2009) reported a 1.41:1 sex ratio (62 males, 44 females), although this was not statistically different from 1:1 (goodness-of-fit test, [G.sub.1] = 3.07, P = 0.080). Ceballos Gonzalez (1989) reported the overall sex ratio of O. banderanus in semideciduous forest to be 1.36:1 (19 males, 14 females), while Dominguez-Castellanos et al. (2007) obtained 1:1.28 (25 males, 32 females); neither differ statistically from 1:1. The explanation often given for a male-skewed sex ratio in natural populations is that males travel over larger areas than females and, thus, have greater exposure to traps (Terman, 1968). Ceballos Gonzalez (1989) noted that female O. banderanus were caught more frequently; we also had a higher average number of captures per individual for females. Thus, exposure to traps per se does not explain the deviation in sex ratio. However, in our study, male O. banderanus moved farther on average than females, albeit the difference was not statistically significant.

Ceballos Gonzalez (1989) concluded that male O. banderanus moved farther than females, but no statistical test was reported. He calculated an average distance moved between successive captures of 24.0 [+ or -] 15.3 m for males (greater than our 21.40 m), 15.4 [+ or -] 13.5 m for females (less than our 17.47 m), and an overall mean distance between successive captures of 17.8 m (n = 107; slightly lower than our 19.62 m). Dominguez-Castellanos et al. (2007) marked animals with fluorescent powder, released them, and measured distances traveled. They too reported longer average distances for males (14.74 m) than females (9.95 m), but these were not statistically different. In addition, they determined that O. banderanus moved significantly farther during the dry season than the wet (15.2 [+ or -] 11.9 versus 7.0 [+ or -] 5.3 m, respectively) and farther in deciduous forest than in semideciduous forest (19.7 [+ or -] 13.7 versus 8.6 [+ or -] 6.5 m). For trapping grids in Jalisco, Collett et al. (1975) reported that one individual traveled 27.3 m between capture sites and another 68.3 m, both values within the range of our data.

Abundance and Density--Ceballos Gonzalez (1989) provided density estimates for O. banderanus in semi-deciduous forest using minimum number of individuals known to be alive. He reported a January density of 3 individuals/ha and a yearly average of 2.1 individuals/ha, estimates considerably lower than our overall average of 15.9 individuals/ha, but within the range of densities among our grids (0.8-30.6 individuals/ha). Ceballos Gonzalez (1989) also recorded changes in density over a 14-month period, noting an increase from low values in December beginning in January to a peak of 8 individuals/ha in March.

Dominguez-Castellanos et al. (2007) caught 57 O. banderanus on four 1-ha plots, which extrapolates to 14.3 individuals/ha and is a value similar to our overall estimate of density. Collett et al. (1975) caught 11 individuals on three trapping grids with effective areas of 3.11 ha (1.18 individuals/ha).

Co-occurring Species and Relative Abundance--Osgoodomys banderanus is considered common in humid areas like tropical semideciduous forests while scarce with patchy distributions in deciduous forests (Ceballos Gonzalez, 1989; Ceballos and Miranda, 2000). In north-central Colima, we found the relative abundance of O. banderanus to vary significantly between nearby localities and between years, with the species ranking fifth overall in abundance of small mammals. Such variation in abundance is likely in large part due to habitat heterogeneity. At Ejido Ranchitos, Colima, O. banderanus was the most abundant species (Sanchez-Hernandez et al., 2009), comprising 26.1% of 406 individuals of all species. Dominguez-Castellanos et al. (2007) reported O. banderanus and Peromyscus perfulvus (tawny deermouse) to be the most abundant species in both the deciduous and semi-deciduous forests of Jalisco. Also in Jalisco, Ceballos Gonzalez (1989) noted that O. banderanus ranked third in terms of biomass (g/ha) and comprised 9.2% of individuals captured. He reported O. banderanus to be most similar to Xenomys nelsoni (Magdalena woodrat), Nyctomys sumichrasti, and Reithrodontomys fulvescens in use of habitat. Hooper (1968) considered O. banderanus to be associated ecologically with P. perfulvus, which in Colima is restricted to coastal areas (Schnell et al., 2008a) and where O. banderanus occurs together with it. In Michoacan, Winkelmann (1962) found O. banderanus along with Heteromys pictus and Hodomys alleni.

Habitat--Average distance to nearest tree was a significant variable in habitat preference, with capture sites having closer trees. Percent rocks and maximum canopy height were of importance, with O. banderanus occupying areas with more rocks and higher canopies. Osgoodomys banderanus also chose microhabitats with less grass, more leaf litter, and more dead wood. The species has been reported to build burrows between rocks or construct nests (lined with leaves and other plant material) under accumulations of litter on the forest floor and in hollows of trees (Ceballos Gonzalez, 1989; Ceballos and Miranda, 2000).

Ceballos Gonzalez (1989) frequently found O. banderanus in rocky habitats, concluding its distribution was restricted to areas with rocks or large accumulations of brush on the ground. Collett et al. (1975) reported more often capturing the species in parts of grids with higher elevations. Winkelmann (1962) caught O. banderanus at Canon de Zopilote, Guerrero, near a stream that was surrounded by scrub forest and cacti-covered rocky hillsides. He also found O. banderanus at lower elevations in the mountains of Michoacan along a boulder-strewn stream with humid tropical vegetation. Davis (1944:397) reported sampling 44 locations along both coasts and the interior of Mexico, finding six O. banderanus at Rio Aguacatillo in Guerrero. He remarked, "The nearly impenetrable brushy hillsides at this station harbored numbers of these long-tailed mice. Apparently this species is restricted to such a habitat as none were taken elsewhere."

Thus, O. banderanus has a strong microhabitat preference for rocky areas at various localities in Mexico, including Colima. We found areas frequented by O. banderanus to possess other important characteristics as well, such as a more open understory (with less dense ground cover) that is well enclosed by nearby trees that form a tall canopy.

We thank E. Meneses for help in facilitating our studies of mammals in Colima, and the numerous students and others from the United States and Mexico for their participation in fieldwork. In addition, we appreciate helpful comments and suggestions on the manuscript from M. A. Mares, I. Schlupp, and two anonymous reviewers. Permits for conducting this investigation were provided by the Instituto Nacional de Ecologia, Direccion General de Vida Silvestre FAUT.0103 to C. Sanchez-Hernandez. The following individuals made specimens available for study from their institutions: R. C. Randall, American Museum of Natural History; L. K. Gordon and C. Ludwig, National Museum of Natural History; R. M. Timm and T. Holmes, Natural History Museum, University of Kansas; F. Cervantes, Instituto de Biologia, Universidad Nacional Autonoma de Mexico; and M. A. Mares and J. K. Braun, Sam Noble Oklahoma Museum of Natural History, University of Oklahoma.

APPENDIX--Gazetteer with localities in Colima, Mexico, where Osgoodomys banderanus (Osgood's deermouse) has been captured. Locality numbers correspond to those in Fig. 1 (locality numbers with appended lowercase letters were geographically close and are represented by a single circle identified by locality number on the map). Localities are ordered by municipality. Latitude and longitude in decimal degrees are included in brackets.

Comala: (1) Hacienda San Antonio [19.44820428N, 103.71665158W].

Minatitlan: (2) El Sauz, 8 km NE Minatitlan, 1,200 m, 19826.330N, 103860.000W [19.43883338N, 104.00000008W]; (2a) 8.5 km NE El Sauz, 1,200 m, 19826.330N, 103860.000W [19.43883338N, 104.00000008W]; (3) Rastrojitos, 3 km NW Ranchitos, 1,650 m [19.42616658N, 103.98984278W]; (4) Rastrojitos, 2 km NW Ranchitos, 1,650 m [19.42021138N, 103.97630578W]; (5) 1 km NW Ranchitos, 1,550 m [19.41193968N, 103.96907708W]; (6) El Salto, 4 km WSW Minatitlan, 630 m, 19822.160N, 10485.060W [sic; 19.36580998N, 104.08291178W]; (7) 9 km SW Minatitlan, 810 m, 19819.710N, 10486.810W [19.32850008N, 104.11350008W]; (7a) 7 km SW Minatitlan, 780 m, 19820.760N, 10486.150W [19.34600008N, 104.10250008W].

Villa de Alvarez: (8) 11 miles [17.7 km] W Comala [19.31563368N, 103.93511038W].

Cuauhtemoc: (9) 2 km E Palmillas, 950 m 19821.250N, 103831.580W [19.35416678N, 103.52633338W]; (10) 6 km S Cuauhtemoc, 710 m, 19816.083 0N, 103836.000 0W [19.26805008N, 103.60000008W]; (11) 2.5 km SE Palmillas, 950 m, 19820.050N, 103832.220W [19.33416678N, 103.53700008W]; (12) 4.5 km SE Palmillas, 860 m, 19819.670N, 103831.360W [19.32783338N, 103.52266678W]; (12a) 4 km SE Palmillas, 860 m, 19819.670N, 103831.360W [19.32783338N, 103.52266678W]; (12b) 4.5 km SE, 1 km E Palmillas, 850 m, 19819.650N, 103831.190W [19.32750008N, 103.51983338W]; (12c) 4.5 km SE, 0.5 km E Palmillas, 850 m, 19819.670N, 103831.360W [19.32783338N, 103.52266678W]; (12d) 4.5 km SE, 0.5 km S Palmillas, 850 m, 19819.670N, 103831.360W [19.32783338N, 103.52266678W]; (12e) 5 km SE Palmillas, 860 m, 19819.590N, 103831.400W [19.32650008N, 103.52333338W]; (12f) 5.5 km SE, 1 km E, Palmillas, 810 m, 19819.380N, 103831.100W [19.32300008N, 103.51833338W]; (12g) 5.5 km SE, 1.5 km E Palmillas, 810 m, 19819.380N, 103831.100W [19.32300008N, 103.51833338W]; (13) 5.5 km SE Cerro Colorado, 763 m, 19817.7140N, 103831.4700W [19.29523338N, 103.52450008W]; (13a) 2.5 km SE Cerro Colorado, 873 m, 19818.8640N, 103832.5480W [19.31440008N, 103.54246678W]; (13b) 2.2 km SE Cerro Colorado, 887 m, 19819.1550N, 103832.6330W [19.31925008N, 103.54388338W]; (14) Rancho La Angostura, 761 m, 19818.837 0N, 103839,997 0W [19.31395008N, 103.66661678W]; (15) El C obano [19.30438718N, 103.64125518W]; (16) El Cobano, 2.25 km E Hacienda El Cobano [19.30511278N, 103.63015348W]; (17) Colima airport, Colima, Mexico, 1981603.500N, 103835054.500W [19.26763898N, 103.59847228W]; (17a) Colima airport, Grid 1 in 2007, Colima, Mexico, 19816056.3 00N, 10383403.9 00W [19.28230568N, 103.56775008W]; (17b) Colima airport, Grid 4 in 2007, Colima Mexico, 1981702.8 00N, 103834028.4 00W [19.28411118N, 103.57455568W].

Manzanillo: (18) 4.5 km N Los Parajes [19.27570428N, 104.42093728W]; (19) El Charco, 50 feet [15 m] [19.24390858N, 104.49455878W]; (20) 2 km NW Los Parajes [19.24401418N, 104.44006828W]; (20a) 4 km NW Los Parajes, 1981503800N, 10482504000W [19.26055558N, 104.42777778W]; (21) Don Tomas, 270 m, 1981304300N, 10482302800W [19.22861118N, 104.39111118W]; (21a) 1 km WNW Don Tomas, 265 m, 1981305800N, 10482305600W [19.23277788N, 104.39888898W]; (22) La Tigrera, 2 km SW Don Tomas, 19813.260N, 104824.350W [19.22100008N, 104.40583338W]; (23) La Huiscolotila [19.22119678N, 104.37838898W]; (24) 4 km E Don Toma s, 1981205100N, 10482200300W [19.21416678N, 104.36750008W]; (24a) 1.6 km SE La Huiscolotila, 250 m, 19812.730N, 104821.930W [19.21216678N, 104.36550008W]; (25) 2 miles [3.2 km] NE Camotlan [19.23059868N, 104.20521848W]; (26) 3.4 km SE Las Canoas, 550 m, 19813.880N, 10487.60W [19.23133338N, 104.12666678W]; (26a) 2.6 km SE Las Canoas, 630 m, 19813.890N, 10488.40W [19.23150008N, 104.14000008W]; (27) 3 km WNW San Jose de Lumbert, 480 m [19.18971838N, 104.14188818W]; (28) 0.6 km NE La Rosa, 700 m, 19811.5230N, 10486.2250W [19.19205008N, 104.10375008W]; (28a) 0.5 km NE La Rosa, 893 m, 19811.4830N, 10486.1460W [19.19138338N, 104.10243338W]; (28b) Rio San Jose, 0.3 km NE La Rosa, 893 m, 19811.4830N, 10486.1460W [19.19138338N, 104.10243338W]; (29) 4 km NW (by road) San Jose de Lumbert, 19810.540N, 10489.230W, 466 m [sic; 19.18216348N, 104.15344298W]; (30) Casa del Senor Ra ul Ochoa Duarte, San Jose de Lumbert, 555 m, 19810.9220N, 10487.6770W [19.18203338N, 104.12795008W]; (31) Playa de Oro, 1987.9350N, 104830.7860W [19.13225008N, 104.51310008W]; (31a) Playa de Oro, 1988.028 0N, 104830.4840W [19.13380008N, 104.50806678W]; (31b) Playa de Oro, 3 m, 1987.9470N, 104830.8020W [19.13245008N, 104.51336678W]; (31c) Playa de Oro, 0 m [19.13913338N, 104.52881678W]; (31d) 3 km NE Playa de Oro, 1988.6980N, 104830.5580W [19.14496678N, 104.50930008W]; (31e) Playa de Oro, 1987. 987 0N, 104831.130 0W [19. 1 3 311678N, 104.51883338W]; (31f) 0.5 km E Playa de Oro [19.13311678N, 104.51883338W]; (32) 1.3 km E Playa de Oro, 80 m, 1988.4240N, 104830.2600W [19.14040008N, 104.50433338W]; (32a) Playa de Oro, Torrecillas [19.14040008N, 104.50433338W]; (33) 4 km SW La Central, 40 m, 1987.5140N, 104827.8110W [19.12523338N, 104.46351678W]; (34) Rancho Majagua, 1986.3350N, 104826.8140W [19.10558338N, 104.44690008W]; (34a) 0.5 mile [0.8 km] S, 7 miles [11.2 km] W Santiago, 0 m [19.10559008N 104.44690008W]; (35) 1 mile [1.6 km] NE Santiago, 3 m [19.12750008N, 104.33781628W]; (36) 3 km E Santiago [19.12099638N, 104.33205298W]; (37) 5 miles [8 km] N Manzanillo [19.11265228N, 104.31619338W]; (38) 1 km W Paso Ancho, 80 m [19.07106208N, 104.17877808W]; (39) Agua Blanca, 141 m, 1983.9610N, 10483.1520W [19.06601678N, 104.05253338W]; (40) Manzanillo [19.05361308N, 104.31671208W].

Coquimatlan: (41) 9 km W Pueblo Juarez [19.16725228N, 104.01712708W]; (42) La Fundicion, 1988.4980N, 10481.1790W [19.14163338N, 104.01965008W]; (43) Pueblo Jua rez [19.16742968N, 103.93144688W]; (43a) Hacienda La Magdalena [19.16742968N, 103.93144688W]; (44) 2 km NW La Esperanza, 260 m [19.17683108N, 103.88157428W]; (45) 2.8 km WSW Coquimatlan, 300 m, 19811.50N, 103850.2 0W [19.19166678N, 103.83666678W]; (46) 3 km NE Jala, 300 m, 1989.1410N, 103849.920W [19.15235008N, 103.83200008W]; (46a) 3.5 km NE Jala, 460 m, 1989.8090N, 103849.7440W [19.16348338N, 103.82906678W].

Colima: (47) 3.5 miles [5.6 km] NNW Colima [19.27518248N, 103.71670148W]; (48) Colima [19.25593408N, 103.71400668W]; (49) near Colima [19.23799638N, 103.68815418W]; (50) 6.5 km N Trapichillas, 627 m, 19812.844 0N, 103831.329 0W [19.21406678N, 103.52215008W]; (51) 1.5 km S, 3.0 km E Asmoles, 359 m [19.10447008N, 103.74720008W]; (52) 3 km NE Tepames, 480 m, 1986.5220N, 103835.6770W [19.10870008N, 103.59461678W]; (52a) 3.5 km NE Tepames, 460 m, 1987.0200N, 103835.7540W [19.11700008N, 103.59590008W]; (53) 1.4 km S Puerta de Anzar, 244 m, 198405400N, 10383201800W [sic; 19.10653108N, 103.55266248W]; (53a) 2 km SE Puerta de Anzar, 460 m, 1986.0210N, 103832.3010W [19.10035008N, 103.53835008W]; (54) 1 km E El Hervidero, 5 km S Puerta de Anzar, 1984.8540N, 103832.1530W [19.08090008N, 103.53588338W]; (54a) El Hervidero, 5 km S Puerta de Anzar, 244 m, 1984 054 00N, 103832 018 00W [19. 0 8 166678N, 103.53833338W]; (54b) 1 km NNW El Hervidero, 400 m, 1985.1300N, 103832.3750W [19.08550008N, 103.53958338W].

Armeria: (55) 5 miles [8.0 km] NE La Cofradia (Juarez) [19.03582908N, 103.91566238W]; (56) La Taberna, 8 km N Nuevo Cuyutlan, 250 m, 1984.0820N, 10487.0770W [sic; 19.02082038N, 104.03469018W].

Tecoman: (57) 4.5 km N, 5 km E Tecolapa, 300 m [19.05466678N, 103.78333338W]; (58) 0.5 km E Tecoman [18.97479188N, 103.86736738W]; (59) 4 km E Tecoman, 10 m [18.92394928N, 103.84891018W]; (59a) 5 km E Tecoman, 10 m, 18856.130N, 103849.840W [18.93550008N, 103.83066678W] (60) 3 km NE Tecuanillo, 10 m, 18850.100N, 103853.490W [18.83500008N, 103.89150008W].

Ixtlahuacan: (61) 4.3 km NE Las Conchas, 133 m, 1 8 854037.9800N, 1 0 3 83 6 00 8 . 32 00W [18.91055008[degree]N, 103.60231118W]; (61a) 3.9 km NE La[s] Concha[s], 73 m, 18854040.3600N, 103836026.9900W [18.9112111[degree]N , 103.60749728W]; (62) 1.6 km NE Las Conchas, 30 m [18.89746448N, 103.62050088W]; (63) 3.3 km NW La[s] Concha[s], 60 m, 18854020.300N, 103839019.300W [18.90563898N, 103.65536118W].

Submitted 18 October 2011. Accepted 9 November 2012. Associate Editor was Jennifer K. Frey.

LITERATURE CITED

ARITA, H. T., F. FIGUEROA, A. FRISCH, P. RODRIGUEZ, AND K. SANTOS-DELPRADO. 1997. Geographical range size and the conservation of Mexican mammals. Conservation Biology 11:92-100.

ARROYO-CABRALES, J. 2005. Osgoodomys banderanus (J. A. Allen, 1897). Pages 717-718 in Los mamiferos silvestres de Mexico (G. Ceballos and G. Oliva, editors). Comision Nacional para el Conocimiento y Uso de la Biodiversidad, y Fondo de Cultura Economica, Mexico, Distrito Federal, Mexico.

BENSON, D. A., I. KARSCH-MIZRACHI, D. J. LIPMAN, J. OSTELL, AND E.W. SAYERS. 2009. GenBank. Nucleic Acids Research 37:D26-31.

BLAIR, W. F. 1951. Population structure, social behavior, and environmental relations in a natural population of the beach mouse (Peromyscus polionotus leucocephalus). Contributions from the Laboratory of Vertebrate Biology, University of Michigan 48:1-47.

CARLETON, M. D. 1980. Phylogenetic relationships in neotomine-peromyscine rodents (Muroidea) and a reappraisal of the dichotomy within New World Cricetinae. Miscellaneous Publications of the Museum of Zoology, University of Michigan 157:1-146.

CEBALLOS, G., AND A. MIRANDA, EDITORS. 2000. Guia de campo de los mamiferos de la costa de Jalisco, Mexico. Fundacion Ecol ogica de Cuixmala, A.C., Universidad Nacional Autonoma de Mexico, Mexico, Distrito Federal, Mexico.

CEBALLOS, G., AND D. NAVARRO L. 1991. Diversity and conservation of Mexican mammals. Pages 167-198 in Latin American mammalogy: history, diversity and conservation (M. A. Mares and D. J. Schmidly, editors). University of Oklahoma Press, Norman.

CEBALLOS GONZALEZ, G. J. 1989. Population and community ecology of small mammals from tropical deciduous and arroyo forests in western Mexico. Ph.D. dissertation, University of Arizona, Tucson.

COLLETT, S. F., C. SA NCHEZ HERNANDEZ, K. A. SHUM, JR., W. R. TESKA, AND R. H. BAKER. 1975. Algunas caracteristicas poblacionales demograficas de pequenos mamiferos en dos habitats mexicanos. Anales del Instituto de Biologia, Universidad Nacional Autonoma de Mexico, Serie Zoologia 46:101-123.

DAVIS, W. B. 1944. Notes on Mexican mammals. Journal of Mammalogy 25:370-403.

DELANY, M. J. 1971. The biology of small rodents in Mayanja Forest, Uganda. Journal of Zoology (London) 165:85-129. DOMINGUEZ-CASTELLANOS, Y., F. PIMENTEL L., AND G. CEBALLOS. 2007. Uso de habitat de roedores arboricolas en la selva seca de la Reserva de la Biosfera Chamela-Cuixmala, Jalisco. Revista Mexicana de Mastozoologia 11:21-33.

FA, J. E., AND L. M. MORALES. 1991. Mammals and protected areas in the Trans-Mexican Neovolcanic Belt. Pages 167-198 in Latin American mammalogy: history, diversity and conservation (M. A. Mares and D. J. Schmidly, editors). University of Oklahoma Press, Norman.

GLAZIER, D. S. 1980. Ecological shifts and the evolution of geographically restricted species of North American Peromyscus (mice). Journal of Biogeography 7:63-83.

HAFNER, J. C., J. E. LIGHT, D. J. HAFNER, M. S. HAFNER, E. REDDINGTON, D. S. ROGERS, AND B. R. RIDDLE. 2007. Basal clades and molecular systematics of heteromyid rodents. Journal of Mammalogy 88:1129-1145.

HAYNE, D. W. 1949. Calculation of size of home range. Journal of Mammalogy 30:1-18.

HOOPER, E. T. 1968. Classification. Pages 27-74 in Biology of Peromyscus (Rodentia) (J. A. King, editor). American Society of Mammalogists, Special Publication 2:1-593.

HUGGINS, R. M. 1989. On the statistical analysis of capture experiments. Biometrika 76:133-140.

HUGGINS, R. M. 1991. Some practical aspects of a conditional likelihood approach to capture experiments. Biometrics 47:725-732.

INSTITUTO NACIONAL DE ESTADISTICA, GEOGRAFIA E INFORMA TICA. 2006. Anuario estadistico del estado de Colima. 2006 edicion. Instituto Nacional de Estadistica, Geografia e Informatica, Aguascalientes, Aguascalientes, Mexico.

LAMBERT, T. D., J. R. MALCOLM, AND B. L. ZIMMERMAN. 2005. Variation in small mammal species richness by trap height and trap type in southeastern Amazonia. Journal of Mammalogy 86:982-990.

MALCOLM, J. R. 1991. Comparative abundances of Neotropical small mammals by trap height. Journal of Mammalogy 72:188-192.

MCCUNE, B., AND M. J. MEFFORD. 2009. HyperNiche. Nonparametric multiplicative habitat modeling. Version 2.08. MjM Software, Gleneden Beach, Oregon.

MENDOZA DURAN, A., AND A. MIRANDA. 2002. Osgoodomys banderanus J. A. Allen 1897. Pages 431-432 in Historia natural de Chamela (F. A. Noguera, J. H. Vega Rivera, A. N. Garcia Aldrete, and M. Quesada Avendano, editors). Instituto de Biologia, Universidad Nacional Autonoma de Mexico, Mexico, Distrito Federal, Mexico.

MUSSER, G. G., AND M. D. CARLETON. 2005. Superfamily Muroidea. Pages 894-1531 in Mammal species of the world: a taxonomic and geographic reference (D. E. Wilson and D. M. Reeder, editors). Third edition. Johns Hopkins University Press, Baltimore, Maryland.

PRICE, M. V., N. M. WASER, AND T. A. BASS. 1984. Effects of moonlight on microhabitat use by desert rodents. Journal of Mammalogy 65:353-356.

PRUGH, L., AND J. BRASHARES. 2010. Basking in the moonlight? Effect of illumination on capture success of the endangered giant kangaroo rat. Journal of Mammalogy 91:1205-1212.

ROHLF, F. J., AND D. E. SLICE. 1999. BIOMstat for Windows. Exeter Software, Setauket, New York.

SANCHEZ-AZOFEIFA, G. A., M. QUESADA, J. P. RODRIGUEZ, J. M. NASSAR, K. E. STONER, A. CASTILLO, T. GARVIN, E. L. ZENT, J. C. CALVOALVARADO, M. E. R. KALACSKA, L. FAJARDO, J. A. GAMON, AND P. CUEVAS-REYES. 2005. Research priorities for Neotropical dry forests. Biotropica 37:477-485.

SANCHEZ-HERNANDEZ, C., S. B. GONZALEZ-PE REZ, G. D. SCHNELL, M. L. ROMERO-ALMARAZ, M. L. KENNEDY, AND T. L. BEST. 2009. Mammals from Ejido Ranchitos, Colima, Mexico. Southwestern Naturalist 54:409-420.

SCHNELL, G. D., M. L. KENNEDY, C. SA NCHEZ-HERNANDEZ, M. L. ROMERO-ALMARAZ, B. D. N. ESTEVEZ, J. GUERRERO, T. L. BEST, M. C. WOOTEN, AND R. D. OWEN. 2008a. Habitat preference of the endemic tawny deermouse (Peromyscus perfulvus), a species of conservation concern. Southwestern Naturalist 53:9-20.

SCHNELL, G. D., C. J. POINDEXTER, C. SA NCHEZ-HERNANDEZ, M. L. ROMERO-ALMARAZ, M. L. KENNEDY, T. L. BEST, M. C. WOOTEN, AND A. P. JIMENEZ. 2008b. Demographic features and habitat preferences of southern pygmy mice (Baiomys musculus) in Colima, Mexico. Canadian Journal of Zoology 86:507-524.

SCHNELL, G. D., E. A. GILLAND, C. SA NCHEZ-HERNANDEZ, M. L. ROMERO-ALMARAZ, M. L. KENNEDY, T. L. BEST, AND M. C. WOOTEN. 2010a. Microhabitat preferences and spatial distribution of the vesper rat (Nyctomys sumichrasti) in Colima, Mexico. Mammalian Biology 75:482-495.

SCHNELL, G. D., M. L. ROMERO-ALMARAZ, S. T.MARTINEZ-CHAPITAL, C. SA NCHEZ-HERNANDEZ, M. L. KENNEDY, T. L. BEST, M. C. WOOTEN, AND R. D. OWEN. 2010b. Habitat use and demographic characteristics of the west Mexican cotton rat (Sigmodon mascotensis). Mammalia 74:379-393.

SLADE, N. A., AND R. K. SWIHART. 1983. Home range indices for the hispid cotton rat (Sigmodon hispidus) in northeastern Kansas. Journal of Mammalogy 64:580-590.

SYSTAT SOFTWARE, INC. 2007. SYSTAT for Windows. Version 12. SYSTAT Software, Inc., San Jose, California.

SYSTAT SOFTWARE, INC. 2008. SigmaPlot for Windows. Version 11.0. Systat Software, Inc., San Jose, California.

TERMAN, C. R. 1968. Population dynamics. Pages 412-450 in Biology of Peromyscus (Rodentia) (J. A. King, editor). American Society of Mammalogists, Special Publication 2:1-593.

VAN HENSBERGEN, H. J., AND S. C. MARTIN. 1993. Climatic factors affecting trapping success of some South African small mammals. South African Journal of Wildlife Research 23:87-94.

WINKELMANN, J. R. 1962. Mammal records from Guerrero and Michoacan, Mexico. Journal of Mammalogy 43:108-109.

Cassie J. Poindexter, Gary D. Schnell,* Cornelio Sanchez-Hernandez, Maria de Lourdes Romero-Almaraz, Michael L. Kennedy, Troy L. Best, Michael C. Wooten, and Juliann L. Waits

Sam Noble Oklahoma Museum of Natural History and Department of Biology, University of Oklahoma, 2401 Chautauqua Avenue, Norman, OK 73072 (CJP, GDS) Departamento de Zoologia, Instituto de Biologia, Universidad Nacional Autonoma de Mexico, Apartado Postal 70-153, Ccoyoaccan, Mexico, Distrito Federal 04510, Mexico (CS-H) Escuinapa No. 92 bis. Colonia Pedregal de Santo Domingo, C.P. 04360, Mexico, Distrito Federal, Mexico (MLR-A) Ecological Research Center, Department of Biological Sciences, University of Memphis, Memphis, TN 38152 (MLK, JLW) Department of Biological Sciences, 331 Funchess Hall, Auburn University, AL 36849 (TLB, MCW)

Correspondent: gschnell@ou.edu

[FIGURE 1 OMITTED]

Relatively little is known about the basic ecology and demography of O. banderanus, a condition fostered in part by its varied taxonomic treatment. There has been debate over several decades concerning placement of the species within the subfamily Neotominae. Originally assigned in the genus Peromyscus, Carleton (1980), in a morphological study, was the first to advocate that the species be classified in its own genus, a recommendation followed by Musser and Carleton (2005).

The ecology and habitat use of O. banderanus has been described only in general terms, mostly in brief field remarks, with only a few studies reporting on specific demographic parameters. The primary work on the species has been by Ceballos Gonzalez (1989) as part of his study of small mammals of deciduous and semi-deciduous forests at the Chamela Biological Station in Jalisco.

Given the relative paucity of information on demographic features and ecology of O. banderanus, mark-recapture studies were conducted with the specific objectives to describe habitat preferences of O. banderanus, as well as to report sex ratio, reproductive status, age structure, and population density. Additional objectives were to acquire information on use of understory, probability of capture, and movements of males and females.

Materials and Methods--Study Area and Trapping--This study was conducted at Rancho La Angostura, Colima (locality 14, Fig. 1) in 2006 and sites in the vicinity of the Miguel de la Madrid Airport (locality 17) in 2006 and 2007. Five trapping grids were established each year (10 total) during the annual dry season in January. Days were warm and nights were cool (January average during 1961-2005 for Manzanillo was 24.6[degree]C; Instituto Nacional de Estadistica, Geografia e Informatica, 2006), with little day-today variation in temperature and no rain. Grids were placed to include the major types of vegetation present. Each grid consisted of 100 trapping stations (10 by 10 square grid), with stations 10 m apart. Two Sherman live traps (7.5 by 9.0 by 23.0 cm) baited with rolled oats were placed at each station, one on the ground and one elevated 1-2 m above ground on a thin plywood platform (12.5 by 34.5 cm) attached to a tree or shrub. Grids were sampled for 8 nights each; thus, overall sampling effort on the grids involved 16,000 trap-nights.

Grids 1-2 in 2006 at Rancho La Angostura (sampled 2-6, 10-12 January) were 700-765 m in elevation, and vegetation on grids was tropical deciduous forest (bordered by a small stream) in an overall landscape that was fragmented due to cultivated agriculture (mainly oranges and limes) and raising of livestock. However, the area containing the grids had native vegetation (some Ficus trees [less than or equal to] 5 m in circumference) with heavily shaded gallery forest (trees 5-40 m in height). Numerous species of trees were present, the most abundant being Alnus, Salix humboldtiana, S. bonplandiana, Cecropia obtusifolia, Ficus, and Inga eriocarpa (Schnell et al., 2008b). Cultivated coffee trees (Coffea arabica) were in the understory on both grids at this location.

The second locality, adjacent to the Miguel de la Madrid Airport, included grids 3-5 in 2006 (sampled 3-6, 10-13 January) and all five grids in 2007 (grids 1-3 sampled 2-6, 1012 January and grids 4-5 sampled 3-6, 10-13 January). None of the grids overlapped within or between years. The study area had been fragmented and used for agriculture in the past but was protected and relatively undisturbed for 30 years prior to our study. Much of the area studied had reverted to tropical moist forest intermixed with grass-dominated habitat (Andropogon, often >2 m tall). Characteristic trees were Brosimum alicastrum, Sideroxylon cartilagineum, Coussapoa purpusii, Hura polyandra, Guarea glabra, Dendropanax arboreus, and Populus guzmanantlensis (Schnell et al., 2008b).

Grids 3-5 in 2006 were predominately thorn forest. Grid 4 had patches of tall grass associated with small stream and riparian habitat of dense vines and large deciduous trees. Grids 3 and 5 had sparse understory vegetation intermixed with patches of tall grass (1.0-1.5 m). In 2007, grids 1 and 2 had tropical deciduous forest with mature trees and thorn-forest elements, with grid 1 featuring a sparse understory and a partially cleared, unimproved road traversing the grid. Grids 3 and 4 had dense thorn forest with vines creating low, thick canopy cover, while some areas had mature deciduous trees and patches of tall grass in open areas. Grid 5 featured tall dense grass (1.0-3.0 m) with isolated trees.

Traps were checked once each day starting at dawn, rebaited if needed, and opened for the full 24-h period. For each O. banderanus captured, we recorded position of trap (i.e., trapping station on the grid and ground versus elevated), gender, reproductive status (based on external condition of reproductive organs), and age (adult, subadult, or juvenile based on size, mass, and pelage coloration). Typically, animals were tagged in both ears using uniquely numbered Monel No. 1 ear tags (National Band and Tag Company, Newport, Kentucky) to ensure that individuals could be identified throughout the course of study. Mice were released at site of capture.

Taxonomy follows Musser and Carleton (2005), except Liomys is treated as being part of Heteromys as proposed by Hafner et al. (2007). Secondary verifications of identity of species were made by comparison of mitochondrial DNA sequences. Sequence patterns from a region 680 nucleotides long of the cytochrome b gene were generated from tissue samples collected at the time of original capture (data not shown). Sequences were compared visually to multiple sequences available in the GENBANK repository (Benson et al., 2009) representing known cytochrome b sequences from Osgoodomys. Because of the close morphological association between Osgoodomys and Peromyscus, comparisons also were made to multiple species of Peromyscus (P. aztecus, P. hylocetes, P. spicilegus) reported to be present in the state of Colima. Similarity between nucleotide sequences of the field-collected specimens and GENBANK sequences of Osgoodomys were all >99%.

Estimation of Movement, Probability of Capture, Abundance, and Density--For each O. banderanus captured more than once, we calculated the centroid of capture sites. Distance from centroid was calculated for each trapping location and the mean 0determined for each individual. This measure of movement has been used previously for Baiomys musculus and Sigmodon mascotensis (Schnell et al., 2008b, 2010b). Movement also was assessed by determining average distance between successive captures (Slade and Swihart, 1983), which was calculated weighting each occurrence equally (e.g., an individual caught five times would contribute four data points while one caught two times would provide one).

The computer program MARK (Version 6.0; G. C. White, http://warnercnr.colostate.edu/~gwhite/mark/mark.htm) was used to estimate abundance (i.e., population size) of O. banderanus and to estimate probability of capture on a given trapping night. Assuming a closed population during the trapping period, the program compared relative proportions of marked and unmarked animals in successive samples to estimate abundance. We used the Huggins closed-captures option (Huggins, 1989, 1991), in which the two parameters estimated were p, the probability of capture, and c, the probability of recapture; these in turn were involved in deriving N, the abundance for a given grid. Parameters p and c were considered equal in our model but could vary with time (t; i.e., the 8 trapping nights were estimated independently of one another); this model is indicated by p(t) = c(t). Probabilities of capture and recapture for the 8 trapping nights were considered to be the same for grids in 2006 and the same for grids in 2007, but not necessarily the same for the 2 years; thus, 16 parameters in total were being estimated. This model was chosen because it produced the most reliable results based on the Akaike Information Criterion as modified to adjust for small-sample bias ([AIC.sub.c]; Version 6.0; G. C. White, http://warnercnr.colostate.edu/~gwhite/mark/mark.htm). Other models (i.e., where all grids were considered separately, grids were grouped by specific locality, or all grids were considered equal) yielded relatively similar estimates of total numbers of individuals present.

Converting abundance values to densities requires an estimate of the effective area of a grid. The immediate area of the grid (8,100 [m.sup.2]; 90 by 90 m) does not include area used outside the outer traps. To estimate the edge area, we employed average distance from centroid, with equal weight given to each individual mouse caught more than once, a technique used by Schnell et al. (2008b, 2010b). The resulting value was added to each side of the central portion of the grid and corners of the grid area were treated as being rounded with a radius equal to average distance from centroid. Using distance from centroid as reported in Results, the effective area of a grid was estimated to be 1.320 ha.

We also analyzed probability of capture relative to moonlight. Information on the fraction of the moon illuminated for each trapping night was obtained from the United States Naval Observatory (http://www.usno.navy.mil/USNO/astronomical-applications/data-services/frac-moon-ill).

Structure of Vegetation--Following Schnell et al. (2008a, 2008b), we quantified environmental variation by evaluating 14 characteristics, most of which reflected structure of vegetation (Table 1). Measurements were made at points 1 m from each trapping station (1,000 points total). Percentage of ground cover was estimated (to nearest 5%) for a 1-m square (first seven variables in Table 1). Number of vegetation stems hitting a 1-m bar at 1-m height was determined four times (once in each cardinal direction from central point) and the average calculated (variable 8). Canopy cover (i.e., percent closed, variable 9) was estimated with a spherical densitometer (model C; Forest Densitometers, Bartlesville, Oklahoma), as was slope (variable 10) with a clinometer. Using a 7.5-m vertical pole marked at each decimeter, we recorded the number of decimeter intervals within which vegetation touched the pole; data were totaled for 0-2.5 m (maximum of25 hits; variable 11) and for 2.5-7.5 m (maximum of50 hits; variable 12). Maximum height of canopy was estimated to the nearest 0.5 m (variable 13). Distance to nearest tree ([greater than or equal to] 10 cm dbh; diameter at breast height) was determined in each of four quadrants (with edges being the cardinal directions) and the average taken (variable 14); distances [greater than or equal to] 10 m were tabulated as 10 m.

Habitat Analyses--For each grid, we obtained the mean and SD of the 14 environmental variables, comparing these 28 measurements (means and SDs) for grids where O. banderanus was present with those for grids where it was not detected. Parallel analyses involved the contrast of means and SDs of environmental variables for trapping stations where the species was captured and not captured.

Stepwise logistic regression (SYSTAT Software, Inc., 2007) was used to examine the relationship between environmental measures at trapping stations where O. banderanus was captured and not captured. The dependent variable was logit (Y), which is the natural logarithm of the odds of the species being present, and 14 environmental measures were used as potential independent variables. A variable was entered into the equation if [alpha] < 0.05; after adjustment of the equation taking into account the additional variable, a variable was removed if [alpha] > 0.045. McFadden's [[rho].sup.2]-statistic (range, 0-1) was used to evaluate resulting models as a whole.

Nonparametric multiplicative regression (program Hyper-Niche; McCune and Mefford, 009) also was used to model habitat preferences based on the 14 environmental variables. Nonparametric multiplicative regression effectively represents response of a species to multiple habitat factors, factor interactions are automatically accommodated for, and overall form of the response surface need not be specified in advance (B. McCune, http://www.pcord.com/NPMRintro.pdf). The technique has been used in analysis of habitat preference of other small mammals, including Nyctomys sumichrasti and Sigmodon mascotensis (Schnell et al., 010a, 010b).

We employed the local-mean model type, where weighting around each target point was based on a Gaussian (humpbacked) function. In local-mean models, tolerance (i.e., SD of a Gaussian weighting function) is related inversely to importance of variables in models (B. McCune, http://www.pcord.com/NPMRintro.pdf). To select the best overall model, we used Log B as a measure of fit.

Other Statistical Analyses and Software--SigmaPlot 11 (Systat Software, Inc., 2008) was used to conduct linear- and multiple-regression analyses. BIOMstat for Windows 3.3s (Rohlf and Slice, 1999) was used in univariate analyses, including RxC contingency, goodness-of-fit, and Fisher's exact tests for analysis of frequencies, and linear regression, and analysis of variance (ANOVA) to assess continuous variables.

Results--Trapping--Osgoodomys banderanus was caught on 6 of 10 grids, with 1-40 individuals on a grid (Table 2), for a total of 123 individuals and 385 captures. On average, we caught males 2.9 [+ or -] 1.7 (SD) times and females 3.5 [+ or -] 2.1 times, with an overall average of 3.1 [+ or -] 1.9 captures/individual for the 8-night trapping sessions. We recaptured 69.9% of individuals, with no difference for sex or year (Fisher's exact tests; P = 1.000 and 0.156, respectively). However, there was a significant difference for age, with 75.7% of adults being recaptured in contrast to 42.1% of juveniles or subadults (Fisher's exact test, P = 0.006).

One-third of all captures were in elevated traps (128 of 385; 33.2%) with a range of 27.3-42.4% on individual grids, but no significant difference was detected among grids (RxC test, [G.sub.4] = 5.62, P = 0.229). Of individuals captured more than once (86 individuals), 68.6% were captured in both elevated and ground traps, 25.6% only in ground traps, and 5.8% only in elevated traps.

[FIGURE 2 OMITTED]

Probability of Capture Relative to Trapping Night and Moonlight--For trapping nights, we tabulated the proportion of the moon illuminated, which during our study was 0.07-1.00. Lunar illumination, when considered alone or in combination with trapping night, statistically did not explain the probability of capturing O. banderanus. However, based on data for both years, a nonlinear regression (Fig. 2) was highly significant for trapping night, with an increase in probability of capture from nights 1-6 with a gradual decline thereafter ([F.sub.3, 12] = 5.81, P = 0.011, [R.sup.2] = 0.592). A similar pattern emerged when analyzing years separately. For both years, the first trapping night had the lowest capture probability (<0.3) and trapping probability tended to increase with successive nights until it peaked and decreased on night 6 or 7 (Fig. 2).

Demographic Characteristics--In both years, adult O. banderanus were captured more frequently than subadults or juveniles: 2006, 81.6% adults (62 adults of 76); 2007, 89.1% adults (41 of 46); and both years, 84.4% adults (103 of 122). No significant difference between years was detected (Fisher's exact test, P = 0.312).

We captured more adult males than adult females: 2006, 56.5% males (35 males and 27 females); 2007, 65.9% males (27 and 14); and both years, 60.2% males (62 and 41). For 2006, deviation from a 1:1 male: female ratio could not be demonstrated statistically (goodness-of-fit test, [G.sub.1] = 1.04, P = 0.309). However, for 2007, significantly more males were caught than females ([G.sub.1] = 4.19, P = 0.041). Likewise, for both years pooled, there were significantly more males than females ([G.sub.1] = 4.31, P = 0.038). No statistical heterogeneity was detected in sex ratio among years ([G.sub.1] = 0.92, P = 0.338), and overall sex ratio of adults (males: females) of captured individuals was 1.51:1.

For adult females, percentages of pregnant or lactating individuals were: 2006, 48.1% (13 of 27); 2007, 28.6% (4 of 14); and both years, 41.5% (17 of 41). No significant difference between years was demonstrable (Fisher's exact test, P = 0.321).

Mean mass of eight pregnant females was 40.17 [+ or -] 1.66 g, while 33 non-pregnant females averaged 35.63 [+ or -] 6.57 g; these two groups were not significantly different (one-way ANOVA, [F.sub.1, 39] = 3.71, P = 0.061). Mean mass for adult males was 37.48 [+ or -] 6.19 g (n = 62, range 22.7-51.0 g). For 95 adult animals (males and non-pregnant females), a two-way ANOVA detected no significant difference in mass for sex or year, nor was there significant interaction (respectively, [F.sub.1, 91] = 1.30, 0.03, and 0.86; P = 0.257, 0.859, and 0.356).

Movement, Abundance, and Density--Overall mean of average distances from centroid for adult males was 14.01 [+ or -] 10.35 m (n = 47, range 0-40.6 m), while that for adult females was 11.18 [+ or -] 5.69 m (n = 31, range 0-28.9 m). A two-way ANOVA detected no significant difference for sex, year, or interaction ([F.sub.1, 74] = 0.02-1.75, all P > 0.050). Average distances from centroid were not significantly different among grids (one-way ANOVA, [F.sub.4, 73] = 0.19, P = 0.942). Pooling O. banderanus, irrespective of sex or age, average distance from centroid was 12.74 [+ or -] 9.48 m (n = 86, range 0-40.6 m). Using this value, the effective area of a grid was estimated to be 1.320 ha.

Average distance moved between successive captures did not differ statistically between sexes (one-way ANOVA, [F.sub.1, 239] = 2.64, P = 0.105). Adult males moved 21.40 [+ or -] 21.95 m (n = 132, range 0-94.9 m), females 17.47 [+ or -] 13.83 m (n = 109, range 0-70.7 m), and both combined 19.62 [+ or -] 18.79 m (n = 241, range 0-94.9 m). We also did calculations weighing each individual equally with respect to movement between successive captures, but results were not appreciably different from those reported above.

Based on our model derived in program MARK, abundances for grids were 1.00-40.31 individuals/grid (Table 2). These values translate to density estimates of 0.76-30.55 individuals/ha with an average of 15.85 individuals/ha on grids where the species was present.

Co-occurring Species and Relative Abundance--There was a significant difference between years in the percent of small mammals captured that were O. banderanus (Fisher's exact test, P < 0.001), with the species comprising 16.5% (77 of 468) in 2006 and 5.5% (46 of 836) in 2007. The two localities (localities 14 and 17, Fig. 1) also significantly differed, with O. banderanus constituting 40.1% (61 of 149) of all small mammals at locality 14, but 5.4% (62 of 1,155) at locality 17 (Fisher's exact test, P < 0.001). Relative abundance on grids varied as well, with O. banderanus comprising 0.6-43.4% of the small mammals.

Other species most commonly encountered in 2006-2007 were Sigmodon mascotensis (West Mexican cotton rat; 352 individuals; 27.0%), Heteromys pictus (painted spiny pocket mouse; 313; 24.0%), Baiomys musculus (southern pygmy mouse; 301; 23.1%), and Reithrodontomys fulvescens (fulvous harvest mouse; 131; 10.1%). Numbers of individuals captured of Oryzomys couesi (Coues' oryzomys; 34; 2.6%), Tlacuatzin canescens (gray mouse opossum; 29; 2.2%), Nyctomys sumichrasti (Sumichrast's vesper rat; 14; 1.1%), and Hodomys alleni (Allen's woodrat; 6; 0.5%) were lower than for O. banderanus (123; 9.4%). All of the abovementioned species occurred on grids with O. banderanus.

Habitat--Contrasting means and SDs of the 14 environmental variables for the six grids where O. banderanus was present versus the four grids where it was not found, four significant differences were identified: mean percent grass, 12.0% for grids where present versus 47.2% for grids where absent ([F.sub.1, 8] = 7.98, P = 0.022); mean total hits high on a vertical pole, 5.14 versus 1.28 ([F.sub.1, 8] = 9.86, P = 0.014); mean average distance from nearest tree, 3.4 versus 6.9 m ([F.sub.1, 8] = 11.63, P = 0.009); and SD of percent forbs, 5.93 versus 15.10% ([F.sub.1, 8] = 21.37, P = 0.002). Thus, grids with O. banderanus present had less grass, thicker vegetation high above the ground, shorter distances to nearest tree, and less variation in percentage of forbs.

Evaluating mean values and SDs of environmental variables for trapping stations where O. banderanus was caught and not caught for 2006, 2007, and the combined years (Table 1), 10 variables were statistically different in all three comparisons. Ground cover at stations where O. banderanus was captured had lower percentages of grasses and higher percentages of leaf litter and rocks. In addition, capture sites relative to noncapture sites had fewer hits at 1 m, a more closed canopy, greater slope, fewer hits low and more hits high on a vertical pole, a higher canopy, and shorter average distance to nearest trees.

Stepwise logistic regression for 2006-2007 combined (10 grids, 1,000 trapping stations), contrasting stations where O. banderanus was caught and not caught, resulted in the following equation containing four of the 14 environmental variables:

logit(Y) = -0.883 - 0.027[X.sub.3] - 0.375[X.sub.14] + 0.025[X.sub.6] + 0.041[X.sub.13];

where logit (Y) is the natural logarithm of the odds of the species being present, [X.sub.3] is percent grass, [X.sub.14] is average distance to nearest tree, [X.sub.6] is percent rocks, and [X.sub.13] is maximum canopy height (Table 3). McFadden's p2 was 0.263, indicating a robust model. Projection values for sites where O. banderanus was trapped tended to be greater than for those where the species was not caught (Fig. 3a). Coefficients for the first two variables were negative, indicating that stations where O. banderanus was trapped tended to have less grass (Fig. 3b) and closer trees (Fig. 3c). Coefficients for the last two environmental variables were positive, with stations where the species was caught typically having more rocks (Fig. 3d) and a higher canopy (Fig. 3e).

Using nonparametric multiplicative regression, the best model (log B = 51.58) for predicting likelihood of occurrence of O. banderanus incorporated three variables: average distance to nearest tree (tolerance 0.65), percent rocks (tolerance 7.20), and maximum canopy height (tolerance 8.75). Average neighborhood size of the resulting model was 73.5. Likelihood of occurrence of O. banderanus was highest (ca. 0.3-0.4) when on average trees were 1-3 m away, with likelihood falling sharply to <0.1 when trees were >4 m distant (Fig. 4a). For percent rocks (Fig. 4b), likelihood of occurrence tended to increase as percentage of rocks increased, with likelihood low (<0.3) in areas with <20% rocks and highest (0.7) at 80% rocks; however, areas with 90% rocks also had a low likelihood for occurrence. For maximum canopy height (Fig. 4c), likelihood of occurrence increased as canopy height rose with likelihood peaking above 0.5 when canopies were 32 m in height.

DISCUSSION--Trapping Response--Little information exists on arboreal activity of tropical rodents (Delany, 1971), which Dominguez-Castellanos et al. (2007) suggested could lead to a misrepresentation of population densities and activity patterns. We had equal numbers of ground and elevated traps, resulting in one-third of our captures of O. banderanus coming in elevated traps. Using traps placed 1-3 m above ground, Ceballos Gonzalez (1989) had a nearly identical proportion of captures from elevated traps (33.5%; n = 95) during monthly, 3-night studies (15 months total) on grids in deciduous and semideciduous forests at Chamela Biological Station in Jalisco. However, only ca. 20% of his traps were placed above ground during his 20,412 trap-nights. Dominguez-Castellanos et al. (2007) reported a lower aboveground use for O. banderanus of 10% at Chamela Biological Station but did not indicate relative numbers of elevated and terrestrial traps. In Para, Brazil, Lambert et al. (2005) compared species richness and rates of species accumulation using traps placed at three heights: ground, understory (similar to our elevated traps), and canopy. Many species of small mammals at their site were captured in the understory and rarely on the ground; however, no species was significantly more abundant in the canopy than in the understory traps, rendering understory traps as effective as canopy traps. Yet, in Manaus, Brazil, Malcolm (1991) reported that canopy trapping revealed a rich and abundant fauna not evident when traps were set on the ground or in the understory. However, there was little difference in numbers of species or individuals caught in ground and understory traps.

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

Over 68% of O. banderanus we captured more than once frequented both elevated and ground levels. For another semiarboreal species, Reithrodontomys fulvescens, we found a significantly lower proportion of animals (<49% of individuals) used both levels (Fisher's exact test, P = 0.037; unpublished data). It is clear that O. banderanus readily and regularly made use of both ground and elevated sites.

Probability of Capture Relative to Trapping Night and Moonlight--Some studies suggest that bright moonlit nights negatively affect trapping success for certain rodents (Blair, 1951; Price et al., 1984). Subsequently, fieldwork often has been conducted intentionally during the new moon when rates of capture are presumed to be highest (Prugh and Brashares, 2010), as done by Ceballos Gonzalez (1989) and Dominguez-Castellanos et al. (2007). We detected no relationship between lunar phase and probabilities of capturing O. banderanus. In an analysis of similar data from Colima (Januarys of 2003-2007), Schnell et al. (2008b) also detected no association for Baiomys musculus. It appears that the influence of moonlight on activity of small mammals is greater in more open habitats, such as deserts and grasslands.

In our 8-night study, night of trapping was a significant factor in the probability of catching O. banderanus, with probability of capture increasing nightly until night 6 or 7 and decreasing thereafter. Only a few comparable investigations have reported nightly temporal affects on trapping success. In an 8-night study during February in South Africa, van Hensbergen and Martin (1993) determined that probability of capturing the murid rodent Rhabdomys pumilio (xeric four-striped grass rat) increased steadily during nights 1-4 and decreased thereafter. Other studies conducted for 8-nights in Colima during January also found trapping night to influence significantly the probability of capturing Baiomys musculus (Schnell et al., 2008b) and Sigmodon mascotensis (Schnell et al., 2010b). Two basic patterns were documented in both species; probabilities of capture increased from nights 1-4 then decreased thereafter, or continued increasing until night 6 or 7 before decreasing.

Demographics and Movement--Mass of O. banderanus seldom has been documented. Davis (1944) examined six specimens and reported non-pregnant females in Guerrero averaged 43 g, while a pregnant female (in August) containing three nearly full-term embryos weighed 57.9 g (suggesting young weigh about 4 g each at birth). Ceballos Gonzalez (1989) reported average mass of 14 individuals in Jalisco to be 38.3 [+ or -] 5.6 g, a value similar to our average mass of adults of 37.1 g.

Reproduction in O. banderanus likely occurs year-round with the reproductive pattern reported to be continual polyestrous (Mendoza Duran and Miranda, 2002; Arroyo-Cabrales, 2005). Of females we captured in January, > 40% were reproductively active. This is similar to another study conducted during January by Sanchez-Hernandez et al. (2009) at Ejido Ranchitos, Colima. They reported that about one-third of 44 female O. banderanus were pregnant, lactating, or both.

Overall, males significantly out-numbered females 1.51:1 in our traps, and nearly 2:1 in 2007. Sanchez-Hernandez et al. (2009) reported a 1.41:1 sex ratio (62 males, 44 females), although this was not statistically different from 1:1 (goodness-of-fit test, [G.sub.1] = 3.07, P = 0.080). Ceballos Gonzalez (1989) reported the overall sex ratio of O. banderanus in semideciduous forest to be 1.36:1 (19 males, 14 females), while Dominguez-Castellanos et al. (2007) obtained 1:1.28 (25 males, 32 females); neither differ statistically from 1:1. The explanation often given for a male-skewed sex ratio in natural populations is that males travel over larger areas than females and, thus, have greater exposure to traps (Terman, 1968). Ceballos Gonzalez (1989) noted that female O. banderanus were caught more frequently; we also had a higher average number of captures per individual for females. Thus, exposure to traps per se does not explain the deviation in sex ratio. However, in our study, male O. banderanus moved farther on average than females, albeit the difference was not statistically significant.

Ceballos Gonzalez (1989) concluded that male O. banderanus moved farther than females, but no statistical test was reported. He calculated an average distance moved between successive captures of 24.0 [+ or -] 15.3 m for males (greater than our 21.40 m), 15.4 [+ or -] 13.5 m for females (less than our 17.47 m), and an overall mean distance between successive captures of 17.8 m (n = 107; slightly lower than our 19.62 m). Dominguez-Castellanos et al. (2007) marked animals with fluorescent powder, released them, and measured distances traveled. They too reported longer average distances for males (14.74 m) than females (9.95 m), but these were not statistically different. In addition, they determined that O. banderanus moved significantly farther during the dry season than the wet (15.2 [+ or -] 11.9 versus 7.0 [+ or -] 5.3 m, respectively) and farther in deciduous forest than in semideciduous forest (19.7 [+ or -] 13.7 versus 8.6 [+ or -] 6.5 m). For trapping grids in Jalisco, Collett et al. (1975) reported that one individual traveled 27.3 m between capture sites and another 68.3 m, both values within the range of our data.

Abundance and Density--Ceballos Gonzalez (1989) provided density estimates for O. banderanus in semi-deciduous forest using minimum number of individuals known to be alive. He reported a January density of 3 individuals/ha and a yearly average of 2.1 individuals/ha, estimates considerably lower than our overall average of 15.9 individuals/ha, but within the range of densities among our grids (0.8-30.6 individuals/ha). Ceballos Gonzalez (1989) also recorded changes in density over a 14-month period, noting an increase from low values in December beginning in January to a peak of 8 individuals/ha in March.

Dominguez-Castellanos et al. (2007) caught 57 O. banderanus on four 1-ha plots, which extrapolates to 14.3 individuals/ha and is a value similar to our overall estimate of density. Collett et al. (1975) caught 11 individuals on three trapping grids with effective areas of 3.11 ha (1.18 individuals/ha).

Co-occurring Species and Relative Abundance--Osgoodomys banderanus is considered common in humid areas like tropical semideciduous forests while scarce with patchy distributions in deciduous forests (Ceballos Gonzalez, 1989; Ceballos and Miranda, 2000). In north-central Colima, we found the relative abundance of O. banderanus to vary significantly between nearby localities and between years, with the species ranking fifth overall in abundance of small mammals. Such variation in abundance is likely in large part due to habitat heterogeneity. At Ejido Ranchitos, Colima, O. banderanus was the most abundant species (Sanchez-Hernandez et al., 2009), comprising 26.1% of 406 individuals of all species. Dominguez-Castellanos et al. (2007) reported O. banderanus and Peromyscus perfulvus (tawny deermouse) to be the most abundant species in both the deciduous and semi-deciduous forests of Jalisco. Also in Jalisco, Ceballos Gonzalez (1989) noted that O. banderanus ranked third in terms of biomass (g/ha) and comprised 9.2% of individuals captured. He reported O. banderanus to be most similar to Xenomys nelsoni (Magdalena woodrat), Nyctomys sumichrasti, and Reithrodontomys fulvescens in use of habitat. Hooper (1968) considered O. banderanus to be associated ecologically with P. perfulvus, which in Colima is restricted to coastal areas (Schnell et al., 2008a) and where O. banderanus occurs together with it. In Michoacan, Winkelmann (1962) found O. banderanus along with Heteromys pictus and Hodomys alleni.

Habitat--Average distance to nearest tree was a significant variable in habitat preference, with capture sites having closer trees. Percent rocks and maximum canopy height were of importance, with O. banderanus occupying areas with more rocks and higher canopies. Osgoodomys banderanus also chose microhabitats with less grass, more leaf litter, and more dead wood. The species has been reported to build burrows between rocks or construct nests (lined with leaves and other plant material) under accumulations of litter on the forest floor and in hollows of trees (Ceballos Gonzalez, 1989; Ceballos and Miranda, 2000).

Ceballos Gonzalez (1989) frequently found O. banderanus in rocky habitats, concluding its distribution was restricted to areas with rocks or large accumulations of brush on the ground. Collett et al. (1975) reported more often capturing the species in parts of grids with higher elevations. Winkelmann (1962) caught O. banderanus at Canon de Zopilote, Guerrero, near a stream that was surrounded by scrub forest and cacti-covered rocky hillsides. He also found O. banderanus at lower elevations in the mountains of Michoacan along a boulder-strewn stream with humid tropical vegetation. Davis (1944:397) reported sampling 44 locations along both coasts and the interior of Mexico, finding six O. banderanus at Rio Aguacatillo in Guerrero. He remarked, "The nearly impenetrable brushy hillsides at this station harbored numbers of these long-tailed mice. Apparently this species is restricted to such a habitat as none were taken elsewhere."

Thus, O. banderanus has a strong microhabitat preference for rocky areas at various localities in Mexico, including Colima. We found areas frequented by O. banderanus to possess other important characteristics as well, such as a more open understory (with less dense ground cover) that is well enclosed by nearby trees that form a tall canopy.

We thank E. Meneses for help in facilitating our studies of mammals in Colima, and the numerous students and others from the United States and Mexico for their participation in fieldwork. In addition, we appreciate helpful comments and suggestions on the manuscript from M. A. Mares, I. Schlupp, and two anonymous reviewers. Permits for conducting this investigation were provided by the Instituto Nacional de Ecologia, Direccion General de Vida Silvestre FAUT.0103 to C. Sanchez-Hernandez. The following individuals made specimens available for study from their institutions: R. C. Randall, American Museum of Natural History; L. K. Gordon and C. Ludwig, National Museum of Natural History; R. M. Timm and T. Holmes, Natural History Museum, University of Kansas; F. Cervantes, Instituto de Biologia, Universidad Nacional Autonoma de Mexico; and M. A. Mares and J. K. Braun, Sam Noble Oklahoma Museum of Natural History, University of Oklahoma.

APPENDIX--Gazetteer with localities in Colima, Mexico, where Osgoodomys banderanus (Osgood's deermouse) has been captured. Locality numbers correspond to those in Fig. 1 (locality numbers with appended lowercase letters were geographically close and are represented by a single circle identified by locality number on the map). Localities are ordered by municipality. Latitude and longitude in decimal degrees are included in brackets.

Comala: (1) Hacienda San Antonio [19.44820428N, 103.71665158W].

Minatitlan: (2) El Sauz, 8 km NE Minatitlan, 1,200 m, 19826.330N, 103860.000W [19.43883338N, 104.00000008W]; (2a) 8.5 km NE El Sauz, 1,200 m, 19826.330N, 103860.000W [19.43883338N, 104.00000008W]; (3) Rastrojitos, 3 km NW Ranchitos, 1,650 m [19.42616658N, 103.98984278W]; (4) Rastrojitos, 2 km NW Ranchitos, 1,650 m [19.42021138N, 103.97630578W]; (5) 1 km NW Ranchitos, 1,550 m [19.41193968N, 103.96907708W]; (6) El Salto, 4 km WSW Minatitlan, 630 m, 19822.160N, 10485.060W [sic; 19.36580998N, 104.08291178W]; (7) 9 km SW Minatitlan, 810 m, 19819.710N, 10486.810W [19.32850008N, 104.11350008W]; (7a) 7 km SW Minatitlan, 780 m, 19820.760N, 10486.150W [19.34600008N, 104.10250008W].

Villa de Alvarez: (8) 11 miles [17.7 km] W Comala [19.31563368N, 103.93511038W].

Cuauhtemoc: (9) 2 km E Palmillas, 950 m 19821.250N, 103831.580W [19.35416678N, 103.52633338W]; (10) 6 km S Cuauhtemoc, 710 m, 19816.083 0N, 103836.000 0W [19.26805008N, 103.60000008W]; (11) 2.5 km SE Palmillas, 950 m, 19820.050N, 103832.220W [19.33416678N, 103.53700008W]; (12) 4.5 km SE Palmillas, 860 m, 19819.670N, 103831.360W [19.32783338N, 103.52266678W]; (12a) 4 km SE Palmillas, 860 m, 19819.670N, 103831.360W [19.32783338N, 103.52266678W]; (12b) 4.5 km SE, 1 km E Palmillas, 850 m, 19819.650N, 103831.190W [19.32750008N, 103.51983338W]; (12c) 4.5 km SE, 0.5 km E Palmillas, 850 m, 19819.670N, 103831.360W [19.32783338N, 103.52266678W]; (12d) 4.5 km SE, 0.5 km S Palmillas, 850 m, 19819.670N, 103831.360W [19.32783338N, 103.52266678W]; (12e) 5 km SE Palmillas, 860 m, 19819.590N, 103831.400W [19.32650008N, 103.52333338W]; (12f) 5.5 km SE, 1 km E, Palmillas, 810 m, 19819.380N, 103831.100W [19.32300008N, 103.51833338W]; (12g) 5.5 km SE, 1.5 km E Palmillas, 810 m, 19819.380N, 103831.100W [19.32300008N, 103.51833338W]; (13) 5.5 km SE Cerro Colorado, 763 m, 19817.7140N, 103831.4700W [19.29523338N, 103.52450008W]; (13a) 2.5 km SE Cerro Colorado, 873 m, 19818.8640N, 103832.5480W [19.31440008N, 103.54246678W]; (13b) 2.2 km SE Cerro Colorado, 887 m, 19819.1550N, 103832.6330W [19.31925008N, 103.54388338W]; (14) Rancho La Angostura, 761 m, 19818.837 0N, 103839,997 0W [19.31395008N, 103.66661678W]; (15) El C obano [19.30438718N, 103.64125518W]; (16) El Cobano, 2.25 km E Hacienda El Cobano [19.30511278N, 103.63015348W]; (17) Colima airport, Colima, Mexico, 1981603.500N, 103835054.500W [19.26763898N, 103.59847228W]; (17a) Colima airport, Grid 1 in 2007, Colima, Mexico, 19816056.3 00N, 10383403.9 00W [19.28230568N, 103.56775008W]; (17b) Colima airport, Grid 4 in 2007, Colima Mexico, 1981702.8 00N, 103834028.4 00W [19.28411118N, 103.57455568W].

Manzanillo: (18) 4.5 km N Los Parajes [19.27570428N, 104.42093728W]; (19) El Charco, 50 feet [15 m] [19.24390858N, 104.49455878W]; (20) 2 km NW Los Parajes [19.24401418N, 104.44006828W]; (20a) 4 km NW Los Parajes, 1981503800N, 10482504000W [19.26055558N, 104.42777778W]; (21) Don Tomas, 270 m, 1981304300N, 10482302800W [19.22861118N, 104.39111118W]; (21a) 1 km WNW Don Tomas, 265 m, 1981305800N, 10482305600W [19.23277788N, 104.39888898W]; (22) La Tigrera, 2 km SW Don Tomas, 19813.260N, 104824.350W [19.22100008N, 104.40583338W]; (23) La Huiscolotila [19.22119678N, 104.37838898W]; (24) 4 km E Don Toma s, 1981205100N, 10482200300W [19.21416678N, 104.36750008W]; (24a) 1.6 km SE La Huiscolotila, 250 m, 19812.730N, 104821.930W [19.21216678N, 104.36550008W]; (25) 2 miles [3.2 km] NE Camotlan [19.23059868N, 104.20521848W]; (26) 3.4 km SE Las Canoas, 550 m, 19813.880N, 10487.60W [19.23133338N, 104.12666678W]; (26a) 2.6 km SE Las Canoas, 630 m, 19813.890N, 10488.40W [19.23150008N, 104.14000008W]; (27) 3 km WNW San Jose de Lumbert, 480 m [19.18971838N, 104.14188818W]; (28) 0.6 km NE La Rosa, 700 m, 19811.5230N, 10486.2250W [19.19205008N, 104.10375008W]; (28a) 0.5 km NE La Rosa, 893 m, 19811.4830N, 10486.1460W [19.19138338N, 104.10243338W]; (28b) Rio San Jose, 0.3 km NE La Rosa, 893 m, 19811.4830N, 10486.1460W [19.19138338N, 104.10243338W]; (29) 4 km NW (by road) San Jose de Lumbert, 19810.540N, 10489.230W, 466 m [sic; 19.18216348N, 104.15344298W]; (30) Casa del Senor Ra ul Ochoa Duarte, San Jose de Lumbert, 555 m, 19810.9220N, 10487.6770W [19.18203338N, 104.12795008W]; (31) Playa de Oro, 1987.9350N, 104830.7860W [19.13225008N, 104.51310008W]; (31a) Playa de Oro, 1988.028 0N, 104830.4840W [19.13380008N, 104.50806678W]; (31b) Playa de Oro, 3 m, 1987.9470N, 104830.8020W [19.13245008N, 104.51336678W]; (31c) Playa de Oro, 0 m [19.13913338N, 104.52881678W]; (31d) 3 km NE Playa de Oro, 1988.6980N, 104830.5580W [19.14496678N, 104.50930008W]; (31e) Playa de Oro, 1987. 987 0N, 104831.130 0W [19. 1 3 311678N, 104.51883338W]; (31f) 0.5 km E Playa de Oro [19.13311678N, 104.51883338W]; (32) 1.3 km E Playa de Oro, 80 m, 1988.4240N, 104830.2600W [19.14040008N, 104.50433338W]; (32a) Playa de Oro, Torrecillas [19.14040008N, 104.50433338W]; (33) 4 km SW La Central, 40 m, 1987.5140N, 104827.8110W [19.12523338N, 104.46351678W]; (34) Rancho Majagua, 1986.3350N, 104826.8140W [19.10558338N, 104.44690008W]; (34a) 0.5 mile [0.8 km] S, 7 miles [11.2 km] W Santiago, 0 m [19.10559008N 104.44690008W]; (35) 1 mile [1.6 km] NE Santiago, 3 m [19.12750008N, 104.33781628W]; (36) 3 km E Santiago [19.12099638N, 104.33205298W]; (37) 5 miles [8 km] N Manzanillo [19.11265228N, 104.31619338W]; (38) 1 km W Paso Ancho, 80 m [19.07106208N, 104.17877808W]; (39) Agua Blanca, 141 m, 1983.9610N, 10483.1520W [19.06601678N, 104.05253338W]; (40) Manzanillo [19.05361308N, 104.31671208W].

Coquimatlan: (41) 9 km W Pueblo Juarez [19.16725228N, 104.01712708W]; (42) La Fundicion, 1988.4980N, 10481.1790W [19.14163338N, 104.01965008W]; (43) Pueblo Jua rez [19.16742968N, 103.93144688W]; (43a) Hacienda La Magdalena [19.16742968N, 103.93144688W]; (44) 2 km NW La Esperanza, 260 m [19.17683108N, 103.88157428W]; (45) 2.8 km WSW Coquimatlan, 300 m, 19811.50N, 103850.2 0W [19.19166678N, 103.83666678W]; (46) 3 km NE Jala, 300 m, 1989.1410N, 103849.920W [19.15235008N, 103.83200008W]; (46a) 3.5 km NE Jala, 460 m, 1989.8090N, 103849.7440W [19.16348338N, 103.82906678W].

Colima: (47) 3.5 miles [5.6 km] NNW Colima [19.27518248N, 103.71670148W]; (48) Colima [19.25593408N, 103.71400668W]; (49) near Colima [19.23799638N, 103.68815418W]; (50) 6.5 km N Trapichillas, 627 m, 19812.844 0N, 103831.329 0W [19.21406678N, 103.52215008W]; (51) 1.5 km S, 3.0 km E Asmoles, 359 m [19.10447008N, 103.74720008W]; (52) 3 km NE Tepames, 480 m, 1986.5220N, 103835.6770W [19.10870008N, 103.59461678W]; (52a) 3.5 km NE Tepames, 460 m, 1987.0200N, 103835.7540W [19.11700008N, 103.59590008W]; (53) 1.4 km S Puerta de Anzar, 244 m, 198405400N, 10383201800W [sic; 19.10653108N, 103.55266248W]; (53a) 2 km SE Puerta de Anzar, 460 m, 1986.0210N, 103832.3010W [19.10035008N, 103.53835008W]; (54) 1 km E El Hervidero, 5 km S Puerta de Anzar, 1984.8540N, 103832.1530W [19.08090008N, 103.53588338W]; (54a) El Hervidero, 5 km S Puerta de Anzar, 244 m, 1984 054 00N, 103832 018 00W [19. 0 8 166678N, 103.53833338W]; (54b) 1 km NNW El Hervidero, 400 m, 1985.1300N, 103832.3750W [19.08550008N, 103.53958338W].

Armeria: (55) 5 miles [8.0 km] NE La Cofradia (Juarez) [19.03582908N, 103.91566238W]; (56) La Taberna, 8 km N Nuevo Cuyutlan, 250 m, 1984.0820N, 10487.0770W [sic; 19.02082038N, 104.03469018W].

Tecoman: (57) 4.5 km N, 5 km E Tecolapa, 300 m [19.05466678N, 103.78333338W]; (58) 0.5 km E Tecoman [18.97479188N, 103.86736738W]; (59) 4 km E Tecoman, 10 m [18.92394928N, 103.84891018W]; (59a) 5 km E Tecoman, 10 m, 18856.130N, 103849.840W [18.93550008N, 103.83066678W] (60) 3 km NE Tecuanillo, 10 m, 18850.100N, 103853.490W [18.83500008N, 103.89150008W].

Ixtlahuacan: (61) 4.3 km NE Las Conchas, 133 m, 1 8 854037.9800N, 1 0 3 83 6 00 8 . 32 00W [18.91055008[degree]N, 103.60231118W]; (61a) 3.9 km NE La[s] Concha[s], 73 m, 18854040.3600N, 103836026.9900W [18.9112111[degree]N , 103.60749728W]; (62) 1.6 km NE Las Conchas, 30 m [18.89746448N, 103.62050088W]; (63) 3.3 km NW La[s] Concha[s], 60 m, 18854020.300N, 103839019.300W [18.90563898N, 103.65536118W].

Submitted 18 October 2011. Accepted 9 November 2012. Associate Editor was Jennifer K. Frey.

LITERATURE CITED

ARITA, H. T., F. FIGUEROA, A. FRISCH, P. RODRIGUEZ, AND K. SANTOS-DELPRADO. 1997. Geographical range size and the conservation of Mexican mammals. Conservation Biology 11:92-100.

ARROYO-CABRALES, J. 2005. Osgoodomys banderanus (J. A. Allen, 1897). Pages 717-718 in Los mamiferos silvestres de Mexico (G. Ceballos and G. Oliva, editors). Comision Nacional para el Conocimiento y Uso de la Biodiversidad, y Fondo de Cultura Economica, Mexico, Distrito Federal, Mexico.

BENSON, D. A., I. KARSCH-MIZRACHI, D. J. LIPMAN, J. OSTELL, AND E.W. SAYERS. 2009. GenBank. Nucleic Acids Research 37:D26-31.

BLAIR, W. F. 1951. Population structure, social behavior, and environmental relations in a natural population of the beach mouse (Peromyscus polionotus leucocephalus). Contributions from the Laboratory of Vertebrate Biology, University of Michigan 48:1-47.

CARLETON, M. D. 1980. Phylogenetic relationships in neotomine-peromyscine rodents (Muroidea) and a reappraisal of the dichotomy within New World Cricetinae. Miscellaneous Publications of the Museum of Zoology, University of Michigan 157:1-146.

CEBALLOS, G., AND A. MIRANDA, EDITORS. 2000. Guia de campo de los mamiferos de la costa de Jalisco, Mexico. Fundacion Ecol ogica de Cuixmala, A.C., Universidad Nacional Autonoma de Mexico, Mexico, Distrito Federal, Mexico.

CEBALLOS, G., AND D. NAVARRO L. 1991. Diversity and conservation of Mexican mammals. Pages 167-198 in Latin American mammalogy: history, diversity and conservation (M. A. Mares and D. J. Schmidly, editors). University of Oklahoma Press, Norman.

CEBALLOS GONZALEZ, G. J. 1989. Population and community ecology of small mammals from tropical deciduous and arroyo forests in western Mexico. Ph.D. dissertation, University of Arizona, Tucson.

COLLETT, S. F., C. SA NCHEZ HERNANDEZ, K. A. SHUM, JR., W. R. TESKA, AND R. H. BAKER. 1975. Algunas caracteristicas poblacionales demograficas de pequenos mamiferos en dos habitats mexicanos. Anales del Instituto de Biologia, Universidad Nacional Autonoma de Mexico, Serie Zoologia 46:101-123.

DAVIS, W. B. 1944. Notes on Mexican mammals. Journal of Mammalogy 25:370-403.

DELANY, M. J. 1971. The biology of small rodents in Mayanja Forest, Uganda. Journal of Zoology (London) 165:85-129. DOMINGUEZ-CASTELLANOS, Y., F. PIMENTEL L., AND G. CEBALLOS. 2007. Uso de habitat de roedores arboricolas en la selva seca de la Reserva de la Biosfera Chamela-Cuixmala, Jalisco. Revista Mexicana de Mastozoologia 11:21-33.

FA, J. E., AND L. M. MORALES. 1991. Mammals and protected areas in the Trans-Mexican Neovolcanic Belt. Pages 167-198 in Latin American mammalogy: history, diversity and conservation (M. A. Mares and D. J. Schmidly, editors). University of Oklahoma Press, Norman.

GLAZIER, D. S. 1980. Ecological shifts and the evolution of geographically restricted species of North American Peromyscus (mice). Journal of Biogeography 7:63-83.

HAFNER, J. C., J. E. LIGHT, D. J. HAFNER, M. S. HAFNER, E. REDDINGTON, D. S. ROGERS, AND B. R. RIDDLE. 2007. Basal clades and molecular systematics of heteromyid rodents. Journal of Mammalogy 88:1129-1145.

HAYNE, D. W. 1949. Calculation of size of home range. Journal of Mammalogy 30:1-18.

HOOPER, E. T. 1968. Classification. Pages 27-74 in Biology of Peromyscus (Rodentia) (J. A. King, editor). American Society of Mammalogists, Special Publication 2:1-593.

HUGGINS, R. M. 1989. On the statistical analysis of capture experiments. Biometrika 76:133-140.

HUGGINS, R. M. 1991. Some practical aspects of a conditional likelihood approach to capture experiments. Biometrics 47:725-732.

INSTITUTO NACIONAL DE ESTADISTICA, GEOGRAFIA E INFORMA TICA. 2006. Anuario estadistico del estado de Colima. 2006 edicion. Instituto Nacional de Estadistica, Geografia e Informatica, Aguascalientes, Aguascalientes, Mexico.

LAMBERT, T. D., J. R. MALCOLM, AND B. L. ZIMMERMAN. 2005. Variation in small mammal species richness by trap height and trap type in southeastern Amazonia. Journal of Mammalogy 86:982-990.

MALCOLM, J. R. 1991. Comparative abundances of Neotropical small mammals by trap height. Journal of Mammalogy 72:188-192.

MCCUNE, B., AND M. J. MEFFORD. 2009. HyperNiche. Nonparametric multiplicative habitat modeling. Version 2.08. MjM Software, Gleneden Beach, Oregon.

MENDOZA DURAN, A., AND A. MIRANDA. 2002. Osgoodomys banderanus J. A. Allen 1897. Pages 431-432 in Historia natural de Chamela (F. A. Noguera, J. H. Vega Rivera, A. N. Garcia Aldrete, and M. Quesada Avendano, editors). Instituto de Biologia, Universidad Nacional Autonoma de Mexico, Mexico, Distrito Federal, Mexico.

MUSSER, G. G., AND M. D. CARLETON. 2005. Superfamily Muroidea. Pages 894-1531 in Mammal species of the world: a taxonomic and geographic reference (D. E. Wilson and D. M. Reeder, editors). Third edition. Johns Hopkins University Press, Baltimore, Maryland.

PRICE, M. V., N. M. WASER, AND T. A. BASS. 1984. Effects of moonlight on microhabitat use by desert rodents. Journal of Mammalogy 65:353-356.

PRUGH, L., AND J. BRASHARES. 2010. Basking in the moonlight? Effect of illumination on capture success of the endangered giant kangaroo rat. Journal of Mammalogy 91:1205-1212.

ROHLF, F. J., AND D. E. SLICE. 1999. BIOMstat for Windows. Exeter Software, Setauket, New York.

SANCHEZ-AZOFEIFA, G. A., M. QUESADA, J. P. RODRIGUEZ, J. M. NASSAR, K. E. STONER, A. CASTILLO, T. GARVIN, E. L. ZENT, J. C. CALVOALVARADO, M. E. R. KALACSKA, L. FAJARDO, J. A. GAMON, AND P. CUEVAS-REYES. 2005. Research priorities for Neotropical dry forests. Biotropica 37:477-485.

SANCHEZ-HERNANDEZ, C., S. B. GONZALEZ-PE REZ, G. D. SCHNELL, M. L. ROMERO-ALMARAZ, M. L. KENNEDY, AND T. L. BEST. 2009. Mammals from Ejido Ranchitos, Colima, Mexico. Southwestern Naturalist 54:409-420.

SCHNELL, G. D., M. L. KENNEDY, C. SA NCHEZ-HERNANDEZ, M. L. ROMERO-ALMARAZ, B. D. N. ESTEVEZ, J. GUERRERO, T. L. BEST, M. C. WOOTEN, AND R. D. OWEN. 2008a. Habitat preference of the endemic tawny deermouse (Peromyscus perfulvus), a species of conservation concern. Southwestern Naturalist 53:9-20.

SCHNELL, G. D., C. J. POINDEXTER, C. SA NCHEZ-HERNANDEZ, M. L. ROMERO-ALMARAZ, M. L. KENNEDY, T. L. BEST, M. C. WOOTEN, AND A. P. JIMENEZ. 2008b. Demographic features and habitat preferences of southern pygmy mice (Baiomys musculus) in Colima, Mexico. Canadian Journal of Zoology 86:507-524.

SCHNELL, G. D., E. A. GILLAND, C. SA NCHEZ-HERNANDEZ, M. L. ROMERO-ALMARAZ, M. L. KENNEDY, T. L. BEST, AND M. C. WOOTEN. 2010a. Microhabitat preferences and spatial distribution of the vesper rat (Nyctomys sumichrasti) in Colima, Mexico. Mammalian Biology 75:482-495.

SCHNELL, G. D., M. L. ROMERO-ALMARAZ, S. T.MARTINEZ-CHAPITAL, C. SA NCHEZ-HERNANDEZ, M. L. KENNEDY, T. L. BEST, M. C. WOOTEN, AND R. D. OWEN. 2010b. Habitat use and demographic characteristics of the west Mexican cotton rat (Sigmodon mascotensis). Mammalia 74:379-393.

SLADE, N. A., AND R. K. SWIHART. 1983. Home range indices for the hispid cotton rat (Sigmodon hispidus) in northeastern Kansas. Journal of Mammalogy 64:580-590.

SYSTAT SOFTWARE, INC. 2007. SYSTAT for Windows. Version 12. SYSTAT Software, Inc., San Jose, California.

SYSTAT SOFTWARE, INC. 2008. SigmaPlot for Windows. Version 11.0. Systat Software, Inc., San Jose, California.

TERMAN, C. R. 1968. Population dynamics. Pages 412-450 in Biology of Peromyscus (Rodentia) (J. A. King, editor). American Society of Mammalogists, Special Publication 2:1-593.

VAN HENSBERGEN, H. J., AND S. C. MARTIN. 1993. Climatic factors affecting trapping success of some South African small mammals. South African Journal of Wildlife Research 23:87-94.

WINKELMANN, J. R. 1962. Mammal records from Guerrero and Michoacan, Mexico. Journal of Mammalogy 43:108-109.

Cassie J. Poindexter, Gary D. Schnell,* Cornelio Sanchez-Hernandez, Maria de Lourdes Romero-Almaraz, Michael L. Kennedy, Troy L. Best, Michael C. Wooten, and Juliann L. Waits

Sam Noble Oklahoma Museum of Natural History and Department of Biology, University of Oklahoma, 2401 Chautauqua Avenue, Norman, OK 73072 (CJP, GDS) Departamento de Zoologia, Instituto de Biologia, Universidad Nacional Autonoma de Mexico, Apartado Postal 70-153, Ccoyoaccan, Mexico, Distrito Federal 04510, Mexico (CS-H) Escuinapa No. 92 bis. Colonia Pedregal de Santo Domingo, C.P. 04360, Mexico, Distrito Federal, Mexico (MLR-A) Ecological Research Center, Department of Biological Sciences, University of Memphis, Memphis, TN 38152 (MLK, JLW) Department of Biological Sciences, 331 Funchess Hall, Auburn University, AL 36849 (TLB, MCW)

Correspondent: gschnell@ou.edu

Table 1-Mean 6 SD of 14 environmental variables on trapping grids, contrasting stations where Osgood's deermouse Osgoodomys banderanus was captured and not captured in Colima, Mexico. (a) 2006 Captured Not captured Variable (n = 97) (n = 403) 1 Percent woody 9.5 [+ or -] 8.39 8.6 [+ or -] 12.73 plants 2 Percent forbs 7.7 [+ or -] 5.30 7.6 [+ or -] 11.65 3 Percent grasses 2.9 [+ or -] 14.17 *** 29.3 [+ or -] 35.57 4 Percent litter 36.0 [+ or -] 15.83 *** 25.8 [+ or -] 21.67 5 Percent dead wood 17.2 [+ or -] 9.13 *** 9.3 [+ or -] 10.41 6 Percent rocks 15.6 [+ or -] 15.73 *** 9.7 [+ or -] 13.44 7 Percent bare 11.0 [+ or -] 10.23 9.7 [+ or -] 11.25 ground 8 Average hits 1.67 [+ or -] 2.981 *** 4.90 [+ or -] 6.754 at 1 m 9 Percent canopy 81.0 [+ or -] 19.44 *** 67.2 [+ or -] 33.88 closed 10 Slope (degrees) 17.0 [+ or -] 9.11 *** 11.0 [+ or -] 10.84 11 Total hits low 4.97 [+ or -] 4.485 *** 7.60 [+ or -] 5.026 12 Total hits high 6.72 [+ or -] 9.273 *** 4.36 [+ or -] 7.746 13 Maximum canopy 20.9 [+ or -] 11.41 *** 10.3 [+ or -] 9.80 height (m) 14 Average distance 2.4 [+ or -] 1.58 *** 4.7 [+ or -] 2.65 to nearest tree (m) 2007 Captured Not captured Variable (n = 65) (n = 435) 1 Percent woody 2.8 [+ or -] 6.19 2.9 [+ or -] 5.48 plants 2 Percent forbs 6.4 [+ or -] 4.55 7.7 [+ or -] 12.50 3 Percent grasses 4.0 [+ or -] 6.51 *** 31.7 [+ or -] 33.65 4 Percent litter 58.5 [+ or -] 17.54 *** 41.0 [+ or -] 28.78 5 Percent dead wood 7.5 [+ or -] 11.32 5.9 [+ or -] 8.45 6 Percent rocks 20.0 [+ or -] 14.60 *** 5.1 [+ or -] 9.52 7 Percent bare 0.8 [+ or -] 2.43 *** 5.8 [+ or -] 8.60 ground 8 Average hits 1.37 [+ or -] 1.340 *** 5.16 [+ or -] 7.287 at 1 m 9 Percent canopy 89.5 [+ or -] 7.74 *** 46.9 [+ or -] 33.29 closed 10 Slope (degrees) 5.4 [+ or -] 3.86 *** 4.2 [+ or -] 2.86 11 Total hits low 5.03 [+ or -] 2.947 *** 6.54 [+ or -] 4.319 12 Total hits high 4.80 [+ or -] 3.256 *** 2.01 [+ or -] 2.781 13 Maximum canopy 10.4 [+ or -] 3.14 *** 4.9 [+ or -] 3.00 height (m) 14 Average distance 2.4 [+ or -] 1.20 *** 5.8 [+ or -] 3.04 to nearest tree (m) 2006-2007 Captured Not captured Variable (n = 162) (n = 838) 1 Percent woody 6.9 [+ or -] 8.24 5.6 [+ or -] 10.07 plants 2 Percent forbs 7.2 [+ or -] 5.04 7.6 [+ or -] 12.09 3 Percent grasses 3.4 [+ or -] 11.70 *** 30.5 [+ or -] 34.58 4 Percent litter 45.0 [+ or -] 19.83 *** 33.7 [+ or -] 26.69 5 Percent dead wood 13.3 [+ or -] 11.11 *** 7.5 [+ or -] 9.60 6 Percent rocks 17.3 [+ or -] 15.40 *** 7.3 [+ or -] 11.79 7 Percent bare 6.9 [+ or -] 9.48 7.7 [+ or -] 10.14 ground 8 Average hits 1.55 [+ or -] 2.457 *** 5.03 [+ or -] 7.033 at 1 m 9 Percent canopy 84.4 [+ or -] 16.33 *** 56.7 [+ or -] 35.06 closed 10 Slope (degrees) 12.3 [+ or -] 9.37 *** 7.5 [+ or -] 8.50 11 Total hits low 4.99 [+ or -] 3.930 *** 7.05 [+ or -] 4.699 12 Total hits high 5.95 [+ or -] 7.509 *** 3.14 [+ or -] 5.849 13 Maximum canopy 16.7 [+ or -] 10.40 *** 7.5 [+ or -] 7.63 height (m) 14 Average distance 2.4 [+ or -] 1.44 *** 5.3 [+ or -] 2.91 to nearest tree (m) (a) Asterisks indicate statistical differences (one-way ANOVA): *, P < 0.05; **, P < 0.01; ***, P < 0.001; no symbol, P > 0.05. Table 2-Total number of individuals captured and estimates of density for Osgood's deermouse Osgoodomys banderanus on individual grids where the species was trapped in Colima, Mexico. Size of population (N) was estimated using MARK based on model p(t) = c(t). Probabilities of capture (p) and recapture (c) were considered equal for a given night (t) and could vary among nights but were the same for all grids in 2006 and the same for all grids in 2007. Year and Number N [+ or -] SE Density grid captured (95% confidence interval) (animals/ha) 2006 1 33 33.90 [+ or -] 0.992 (33.16-38.17) 25.69 2 28 28.76 [+ or -] 0.910 (28.12-32.82) 21.80 4 16 16.44 [+ or -] 0.680 (16.05-19.84) 12.46 2007 1 40 40.31 [+ or -] 0.579 (40.03-43.41) 30.55 3 1 1.00 (a) 0.76 4 5 5.04 [+ or -] 0.200 (5.00-6.38) 3.82 (a) With only one individual caught, program MARK was unable to provide a robust estimate of density. Table 3--Estimates, SEs, Z-scores, and P-values for constant and coefficients of independent variables incorporated sequentially in the logistic-regression equation when evaluating 2006-2007 trapping stations in Colima, Mexico, to predict the dependent variable logit (Y), which is the natural logarithm of the odds of Osgood's deermouse Osgoodomys banderanus being present. Constant or Coefficient independent variable [+ or -] SE Z-score P-value Constant -0.883 [+ or -] 0.296 -2.986 0.003 Percent grasses -0.027 [+ or -] 0.010 -2.671 0.008 ([X.sub.3]) Average distance to -0.375 [+ or -] 0.076 -4.932 <0.001 nearest tree (m) ([X.sub.14]) Percent rocks 0.025 [+ or -] 0.006 3.943 <0.001 ([X.sub.6]) Maximum canopy height 0.041 [+ or -] 0.010 3.990 <0.001 (m) ([X.sub.13])

Printer friendly Cite/link Email Feedback | |

Author: | Poindexter, Cassie J.; Schnell, Gary D.; Sanchez-Hernandez, Cornelio; de Lourdes Romero-Almaraz, Mar |
---|---|

Publication: | Southwestern Naturalist |

Article Type: | Report |

Geographic Code: | 1MEX |

Date: | Mar 1, 2013 |

Words: | 9709 |

Previous Article: | Rapid assessment of sustainable harvesting of leaves from the understory palm, Chamaedorea radicalis. |

Next Article: | Composition and structure of a mixed-hardwood bottomland forest in the West Cross Timbers of north-central Texas. |

Topics: |