Rodent community structure and microhabitat associations at the Bernard Field Station, Claremont, California Lindsay Bledsoe, Lyn Davis, Allyson Degrassi, Jennifer Edwards, Patricia Gonzalez, Sean Hauser, Leslie Herington, Brian Kronenfeld, Ji Lee, Christina Mirzaian, Tiffany Oberg, Erick Ortiz, Eric Peralta, Thomas Sterner, Tracie Treybig and Dr. Paul Stapp Final Report prepared for BIOL478 Mammalogy California State University, Fullerton March 2010
INTRODUCTION Natural habitats in the valley and foothill areas of coastal southern California have been diminished over the past century as a result of intensive human development and sprawl. The loss, conversion and fragmentation of native vegetation have altered animal communities throughout the region as well, leaving populations isolated from one another and the remaining open space in the surrounding mountains and parks. The Bernard Field Station (BFS), located in Claremont, California, is a 38-ha tract of scrub, woodland and grassland habitat embedded in a matrix of residential and commercial development. Expansion of Interstate 210 in 2002 approximately 0.5 km north of the BFS further isolated native plant and animal communities from the nearby San Gabriel Mountains. Rodent populations at the BFS haved been studied occasionally by students taking courses at the Claremont Colleges, e.g., Dr. Nina Karnovsky's studies of woodrats (http://pages.pomona.edu/~njk04747/woodrat/), but information on population densities and microhabitat associations of other rodents is not widely available. As part of our Mammalogy course at Cal-State Fullerton, we live-trapped rodents over three nights in March 2010 to determine what rodent species were abundant in representative habitats at the BFS and to determine if species tended to be associated with particular microhabitat features. METHODS Study Area The Bernard Field Station (BFS) is located immediately north of the Claremont Colleges, on an alluvial fan at an elevation of approximately 400 m. Vegetation is a mixture of coastal sage scrub, Riversidian alluvial fan scrub, live oak woodland, and non-native grassland. The climate is Mediterranean, with cool wet winters and hot dry summers. The BFS and adjacent Rancho Santa Ana Botanical Gardens are surrounded on all sides by residential and commercial development. Field Sampling One live-trapping grid (0.49 ha) was established in grassland (GR), coastal sage scrub (SW) and Riversidian alluvial fan scrub (RS) communities (Fig. 1). Each grid consisted of 49 traps in a 7-x-7 array with 10 m between stations. A single extra-large folding Sherman live-trap was placed at each station. Traps were baited with a ball of peanut-butter and oats and raw cotton was provided for insulation. Traps were set for three consecutive nights from 3-5 March 2010 and checked at dawn each morning. All individuals were identified, weighed and given a uniquely numbered aluminum ear tag (numbers between 1026 and 1398). Individuals were then released at their capture location. We used the Schnabel estimator (Krebs 1999) to estimate population size and density. Density was estimated by dividing N-hat by the effective trapping area, which was calculated by adding a boundary strip equal to ½ the mean maximum distance moved to the area bounded by the traps (60 x 60 m). Individuals that escaped before tagging (n = 3) were included in the tally of the total number of captures, but not counted for the unique number of individuals captured. We measured microhabitat at trap stations where individuals were first captured. At each station, we recorded percent cover of substrate, shrub density, cactus density, and woody plant species richness. Substrate cover was recorded in two, randomly placed quadrats (0.04-m 2 ; 20 x 20 cm) within 2 m of each trap station. Percent cover of shrub, grass, litter, rock (grain size >5 cm), and bare ground were visually estimated to the nearest 10% inside each quadrat, and quadrat values were averaged to obtain one value for each trap station. We measured the number of individual shrubs, individual cacti stems, and woody species rooted inside a 2-m radius around
each trap station. Substrate cover, shrub density, cactus density, and species richness were averaged for each species at each site and compared using 95% confidence intervals. RESULTS No rodents were captured at the GR trapping grid, although California ground squirrels (Spermophilus beecheyi) were seen nearby and runways of California voles (Microtus californicus) and mounds of pocket gophers (Thomomys bottae) were observed on the grid. We caught four rodent species at the RS grid (Pacific kangaroo rat, Dipodomys agilis; deer mouse, Peromyscus maniculatus; desert woodrat, Neotoma lepida; big-eared woodrat, Neotoma macrotis) and three species (all except N. lepida) at the SW grid (Fig. 2). At both grids, only a single individual of N. macrotis was captured, both of which were at trap stations near large trees. Peromyscus maniculatus was the most abundant rodent captured at both grids, with densities of 26.2 ha -1 at the RS grid and 22.1 ha -1 at the SW grid (Table 1). Conversely, D. agilis was twice as abundant at the SW grid (15.5 ha -1 ) than at the RS grid (7.8 ha -1 ). Neotoma lepida was only captured at the RS grid, where it reached a relatively high density (26.7 ha -1 ). No Neotoma previously marked by Dr. Karnovsky were captured. Peromyscus maniculatus tended to be the most generalized in its habitat associations and was caught in most areas of both grids (Fig. 3). At the RS grid, N. lepida was caught along the border and in the heavily wooded areas, especially near the taller shrubs and trees. There were many instances of overlap between N. lepida and P. maniculatus. Dipodomys agilis was caught only a few times at the RS site, but was common on the western and southern edges of the SW grid (Fig. 3). At both sites, the most common substrate type at capture locations was litter (Fig. 4a, 5a). Grass was also common, especially at the SW grid and at locations were D. agilis was captured. Shrub and cactus densities tended to be higher at the RS grid than at the SW grid and wood plant species richness was also higher at the RS grid (Fig. 4b, 5b). However, except for the higher cover of bare ground at traps capturing P. maniculatus at the RS grid, there were no discernible differences in microhabitat variables between species at either site, based on the overlap between confidence intervals. Because we did not sample microhabitat variables at random or noncapture locations, we could not determine if species preferred certain microhabitats over others, or compare microhabitat associations between grids that had very different vegetation (Fig. 6). DISCUSSION Rodents reached their highest abundance at the RS grid, where shrub and cactus density and plant species richness was highest. Collectively, density at the RS site was 1.6 times higher than at the SW site and four species were captured there, compared to three at the SW site. Neotoma lepida and P. maniculatus were much more numerous than D. agilis at the RS grid. In contrast, N. lepida was not captured in the grassy, coastal sage scrub site (SW), but was replaced by high numbers of D. agilis. Neotoma macrotis was rare at both sites but we intentionally avoided densely wooded areas, where this species tends to be more abundant. We found little evidence of microhabitat preferences for the common species, suggesting that they are fairly generalized in their habitat affinities or that we did not have sufficient numbers of captures to detect significant differences. Unfortunately, we did not capture any of the rarer rodents such as P. californicus, P. eremicus or R. megalotis, which have not been seen on the BFS for >20 years. Additional, more intensive trapping in other locations of the BFS may reveal if these species are still present.
Fortunately, however, we also did not capture any of the non-native commensal rats (Rattus sp.) and mice (Mus musculus) that are probably common in the adjacent urban and suburban development, indicating that at least the most common elements of the native rodent community still persist in some areas and have not been invaded. Lastly, we captured relatively large numbers of D. agilis, especially in the southwest area of the BFS. The southwest part of the BFS seems to be particularly important for these kangaroo rats, perhaps because of the unique combination of grassland and shrub habitats or because of the scarcity of woodrats here. In addition, limited measurements of individual D. agilis suggest that these individuals might actually be D. simulans (Dulzura kangaroo rat), a species that was split from D. agilis based on morphological traits and chromosome number (Sullivan and Best 1997; J. Mammal. 78:775). Dipodomys simulans is believed to be common from Orange County southward and in the Los Angeles Basin to the west, but has not been identified from the foothills near Claremont. Future plans call for additional sampling of kangaroo rats in southern areas of BFS to collect more morphological data as well as tissue samples to resolve, using molecular methods, the taxonomic status of these populations. ACKNOWLEDGEMENTS We thank the Bernard Field Station, especially Stephen Dreher, for permission to work at the BFS, C. Degrassi for the assistance and direction on the GIS analysis and California State University at Fullerton for providing vehicles to drive to and from the site for a week.
Table 1. Population characteristics of rodents caught during 3 nights of live-trapping at the Bernard Field Station, Claremont, California, from 3-5 March 2010. No rodents were captured at a third grassland (GR) grid. Population size was estimated using the Schnabel method and 95% confidence limits were calculated based on the Poisson distribution. Density was calculated by dividing population size by the effective trap area, which was the area bounded by the traps (60-x-60 m) plus a boundary strip equal to ½ the maximum distance moved between captures. Because so few N. macrotis were captured, we used the effective grid area for N. lepida to estimate density. Site/ species Unique individuals (M:F) Total captures Population size (95% CI) Mean max. dist. moved (m) Density (ha -1 ) Riversidian Scrub (RS) Dipodomys agilis 4 (2:2) 8 3.8 (1.6, 11.0) 10.0 7.8 Peromyscus maniculatus 18 (10:8) 45 18.6 (12.9, 27.7) 24.3 26.2 Neotoma lepida 15 (10:5) 28 15.6 (9.2, 28.0) 16.4 26.7 Neotoma macrotis 1 (1:0) 1 1-1.7 Coastal Sage Scrub (SW) Dipodomys agilis 10 (5:5) 23 10.2 (7.5, 24.8) 21.0 15.5 Peromyscus maniculatus 13 (9:4) 27 12.9 (8.5, 27.1) 16.5 22.1 Neotoma macrotis 1 (0:1) 1 1-1.7
Fig. 1. Location of trapping areas (0.49 ha) at the Bernard Field Station, Claremont, California, depicted by the yellow grids. RS = Riversidian alluvial fan scrub, SW = coastal sage scrub in the southwest of the BFS, GR = invasive grassland.
Fig. 2. Images of rodents captured at BFS. Clockwise from left: Peromyscus maniculatus (PEMA), Neotoma lepida (NELE), Neotoma macrotis (NEMA, top), Dipodomys agilis (DIAG, bottom).
Fig. 3. Locations of captures of rodents at trap stations on RS (left) and SW (right) trapping grids from 3-5 March 2010. At each trap station, pie-charts are divided into 3 slices, with each slice representing 1 night of trapping. Colored slices indicate a capture of one of the following species: DIAG = Dipodomys agilis (orange), NELE = Neotoma lepida (green), NEMA = Neotoma macrotis (pink), PEMA = Peromyscus maniculatus (blue). Sprung traps (closed but empty) are filled gray.
Fig. 4. Microhabitat associations of DIAG (n = 4), PEMA (n = 20), and NELE (n = 16) at trap stations where individuals were captured at the RS grid. Measurements were a) mean percent cover of five substrate types and b) mean shrub density, cactus density, and woody plant species richness. Error bars represent 95% CI. Species abbreviations as in Fig. 2.
Fig. 5. Microhabitat associations of DIAG (n = 10) and PEMA (n = 12) at trap stations where individuals were captured at the SW grid. Measurements were a) mean percent cover of five substrate types and b) mean shrub density, cactus density, and woody plant species richness. Error bars represent 95% CI. Species abbreviations as in Fig. 2.
Fig. 6. Representative vegetation at the Riversidian alluvial fan scrub grid (RS; top left), coastal sage scrub grid at the southwest part of the BFS (SW; top right) and the grassland grid (GR).