INFLUENCE OF LANDSCAPE ELEMENTS ON POPULATION DENSITIES AND HABITAT USE OF THREE SMALL-MAMMAL SPECIES

Transcription

1 Journal of ammalogy, 84(1):0 5, 003 INLUENCE O LANDSCAPE ELEENTS ON POPULATION DENSITIES AND HABITAT USE O THREE SALL-AAL SPECIES KAREN E. ABRY,* ERIN A. DREELIN, AND GARY W. BARRETT Institute of Ecology, University of Georgia, Athens, GA , USA Present address of KE: Section of Evolution and Ecology, University of California, Davis, CA 95616, USA Corridor effects on population densities and habitat use of 3 small-mammal species were assessed during in an experimentally fragmented landscape. Corridor presence did not have a statistically significant effect on population densities of cotton rats (Sigmodon hispidus) or cotton mice (Peromyscus gossypinus); however, a significant effect was observed for old-field mice (Peromyscus polionotus) during 000. Cotton rats were captured more frequently than expected in corridors, while old-field mice were captured more frequently than expected in habitat-patch interior; and cotton mice exhibited a more uniform distribution across habitat types. These results suggest that landscape fragmentation and habitat structure may have varying effects on population densities of different species. Key words: corridor, fragmentation, habitat use, matrix, patch, Peromyscus gossypinus, Peromyscus polionotus, Sigmodon hispidus Habitat patches with corridors between them should have higher population growth rates than patches without corridors (ahrig and erriam 1985; Rosenberg et al. 1997), with increased population densities. There is limited empirical support, however, for this theoretical prediction (Beier and Noss 1998). Some landscape-manipulation experiments have shown that corridor presence is associated with higher population densities of certain butterflies (Haddad and Baum 1999) and meadow voles (icrotus pennsylvanicus La Polla and Barrett 1993). Other workers, however, have reported that patches with corridors do not support higher population densities of small mammals than do patches without corridors (Coffman et al. 001; Davis-Born and Wolff 000). Species may respond to the same landscape differently (cintyre and Hobbs 1999); for example, corridors are more ef- * Correspondent: kemabry@ucdavis.edu fective for butterfly species that exhibit increased turning behavior at habitat edges (Haddad 1999). While matrix habitat is not necessarily a total barrier to movement (Bowne et al. 1999), a matrix is typically less conducive to movement than patch habitat, and the ability to move through matrix habitat may vary with species (c- Intyre and Hobbs 1999), matrix habitat type (Ricketts 001), and edge hardness (Stamps et al. 1987). Habitat generalists may be less affected by habitat fragmentation than are habitat specialists. or example, corridor presence was associated with smaller genetic differences among populations of red-backed vole (Clethrionomys gapperi), the habitatspecialist, but not the habitat-generalist, the deer mouse (Peromyscus maniculatus ech and Hallett 001). ossman and Waser (001) found that the genetic structure of white-footed mouse (P. leucopus), the habitat-generalist, is not significantly 0 Downloaded from on 01 arch 018

2 ebruary 003 ABRY ET AL. LANDSCAPE EECTS ON SALL AALS 1 different between populations in continuous forest and populations in isolated woodlots, suggesting that movement between patches by this species may be fairly common even in fragmented landscapes without corridors. The degree of habitat specialization of the 3 small-mammal species considered in this study varies from cotton mouse (P. gossypinus), the generalist, to the highly specialized old-field mouse (P. polionotus Golley et al. 1965). Cotton rats (Sigmodon hispidus) are most often found in habitat with a high percentage of herbaceous cover, particularly grasses and vines (Cameron and Spencer 1981; Goertz 1964; Kaufman et al. 000; Lidicker et al. 199). Cotton mice are found in most habitat types, but they are most plentiful in hardwood forests (Golley et al. 1965). Old-field mice are found in relatively open, sandy habitats with herbaceous vegetation (Gentry 1966), such as the clear-cuts used in this study. We examined the effect of corridors on patch-population densities of 3 small-mammal species in an experimentally fragmented, 110-ha landscape (Haddad 1999). We also assessed use of various landscape elements by each species. We made predictions: patch-population densities will be higher in patches with corridors than in patches without corridors, and habitat generalists (cotton mouse) will use all habitat types (patch edge, patch interior, corridor, and matrix) equally, while habitat specialists (cotton rat and old-field mouse) will use the interior of habitat patches preferentially. ATERIALS AND ETHODS This study was conducted in an experimentally fragmented landscape on the Savannah River Site, a national environmental research park in Aiken County, South Carolina ( N, W). The fragmented landscape consisted of clear-cut patches and corridors embedded within a matrix of managed loblolly pine (Pinus taeda; ig. 1). Nine 1.6-ha habitat patches were used in this study, 6 patches with corridors and 3 patches without corridors. Corridors 3 m wide and 18, 56, or 384 m long linked IG. 1. Aerial photograph (ebruary 001) of experimentally fragmented landscape on the Savannah River Site, Aiken County, South Carolina. Each large patch is 14 by 14 m and each small patch is 64 by 64 m. atrix trapping grids are outlined in white. pairs of habitat patches (ig. 1). Patches were established in by clear-cut tree removal and burning of remaining vegetation (Haddad 1999). Patches were burned by the United States orest Service and were seeded with Lespedeza cuneata during winter spring Three trapping grids (1.6 ha) were established within the pine forest matrix, approximately 64 m from the edge of the nearest habitat patch, during arch 000 (ig. 1). The magnitude of a drought during was assessed using data from the National Oceanic and Atmospheric Administration (001, Palmer drought-severity indices relate rainfall and temperature in a region to quantify the severity of drought. An index of 0 indicates normal conditions, while increasingly negative indices indicate severe drought conditions. Vegetation sampling was conducted in patches and corridors during August in 1998, 1999, and 000. Two quadrats, each 4 by 4 m, were randomly selected within 16 m of each trapping station. Total percent vegetative cover in each quadrat was estimated visually. Canopy cover within the forest matrix was measured during September 000. Densiometer readings were taken at randomly selected points per trapping station; species and diameter at breast height of the nearest tree (with diameter at breast height 10 cm) were recorded. Census procedures. Livetrapping was conducted weekly during June November 1998, ay December 1999, and arch November 000. Trap stations were placed in a 4-by-4 grid pattern at 3-m intervals within each habitat patch and forest matrix grid. orest matrix grids were not established until 000; trapping in Downloaded from on 01 arch 018

3 JOURNAL O AALOGY Vol. 84, No. 1 these grids followed the same schedule as trapping in habitat patches during 000. Trap stations were spaced at 3-m intervals for the length of each corridor. Two Sherman live traps (5 by 7.5 by 7.5 cm) baited with sunflower seeds were placed within 5 m of each station and typically set for consecutive nights each week. Cotton was provided as bedding on nights when the minimum temperature was expected to fall below 10 C. Captured animals were identified to species, weighed, individually marked by toe-clipping, and sexed, and reproductive condition was determined before release at site of capture. Animal care and procedures used were consistent with recommendations by the American Society of ammalogists (1998) and were approved by the University of Georgia Animal Care and Use Committee (permit #A343701). Each trap location was categorized as patch interior, patch edge, corridor, or forest matrix. Captures at the 4 trap stations located in the center of each patch were considered interior captures, while those at the 1 stations within 16 m of the boundary between the clear-cut and the forest were considered edge captures. Captures in corridors were categorized as corridor captures, and those in the forest matrix were considered matrix captures. We used t-tests to compare the percent vegetative cover between patches with and without corridors each year. Analysis of variance (AN- OVA) was used to determine if differences in vegetative cover were significant among patch interior, patch edge, and corridor habitat types. Scheffé pairwise comparisons were then used to determine which habitat types were different from each other. Weekly patch-population densities were estimated using the minimum number of individuals known to be alive (Krebs 1966). Patch-population densities of each species were compared between patches with and without corridors using repeated-measures AN- OVA in SAS ANALYST v. 8. (SAS Institute 00). Chi-square tests were used to determine whether the number of captures in each habitat type differed from number expected if captures were proportional to number of traps located in each habitat type. RESULTS IG.. Vegetative cover in each of 3 habitat types (patch interior, patch edge, and corridor) during , shown as mean SE. Dominant species in clear-cut habitat patches were oak seedlings (Quercus), grasses (Heterotheca, Panicum, Andropogon), smooth sumac (Rhus glabra), greenbrier (Smilax rotundifolia), and dog fennel (Eupatorium capillifolium). Vegetative cover was greater in patches with corridors (65%) than in those without corridors (58%) during 1998; but there were no significant differences in percent cover between patches with and without corridors during 1999 (with corridors, 70%; without corridors, 71%) or 000 (with corridors, 8%; without corridors, 80%). Dominant tree species in the forest matrix were loblolly pine, slash pine (P. elliotii), and willow oak (Q. phellos). Percent canopy cover differed among the 3 forest matrix grids (ANOVA, 15.3, d.f., 95, P ); however, percent canopy cover was high in all grids, ranging from (mean SE) to Percent vegetative cover was significantly higher in corridors than in patch interior or patch edge habitat types during 1999 (ANOVA, 7.55, d.f., 333, P , ig. ); differences in vegetative cover were not significant during 1998 and 000. This study took place during an extremely dry period. A 3-year drought began during the summer of 1998 and persisted throughout the remainder of the study (National Oceanic and Atmospheric Administration 001, Palmer drought-severity indices were below 0 Downloaded from on 01 arch 018

4 ebruary 003 ABRY ET AL. LANDSCAPE EECTS ON SALL AALS 3 IG. 3. Population density in patches with and without corridors during 000 for a) cotton rats (Sigmodon hispidus), b) cotton mice (Peromyscus gossypinus), and c) old-field mice (P. polionotus). Symbols indicate means. Note that scales of y axes are not equal. throughout the study, indicating drought conditions. We recorded 506 captures of cotton mice, 438 of old-field mice, and 340 of cotton rats during 9,088 trap nights between 1998 and 000. We also recorded 1 golden mouse (Ochrotomys nuttalli), 10 southern shorttailed shrews (Blarina carolinensis), and 1 rice rat (Oryzomys palustris). Corridor presence did not affect cotton rat (S. hispidus) patch-population densities (repeated measures ANOVA: 1998, 0.89, d.f. 1, 7, P ; 1999, 0.99, d.f. 1, 7, P ; 000, 0.38, d.f. 1, 7, P ; ig. 3a). While statistical analyses were conducted on data from all 3 years of the study, graphs of average population densities in patches with and without corridors are presented for 000 only. The effect of corridor presence on patch-population densities was not statistically significant for cotton mice (P. gossypinus; repeated measures ANOVA: 1998, 0.1, d.f. 1, 166, P 0.745; 1999, 1.11, d.f. 1, 40, P 0.97; 000, 0.89, d.f. 1, 41, P ; ig. 3b). Corridor presence did have a statistically significant effect on patch-population density of old-field mice during 000 (P. polionotus; repeated measures ANOVA: 1998, 0.64, d.f. 1, 7, P 0.451; 1999, 1.05, d.f. 1, 7, P 0.340; 000, 8.18, d.f. 1, 7, P 0.043; ig. 3c); however, patch-population densities were higher in patches without corridors. Habitat use by cotton mice and old-field mice was nonrandom during 1999 and 000; habitat use by cotton rats was nonrandom during all years of the study (Table 1). During periods of nonrandom habitat use by cotton mice (both sexes during 000 and females during 1999), there were more captures than expected in interior traps. During 1999 and 000, old-field mice were captured more frequently than expected in interior traps. Cotton rats were captured more frequently than expected in corridors during each year of the study. DISCUSSION Corridor effects. Our prediction that population densities of all 3 small-mammal species would be significantly higher in patches with corridors than in patches without corridors was not supported. Corridor presence had a statistically significant effect only on population densities of old-field mice (P. polionotus) during 000, and patch-population densities were lower, not higher, in patches with corridors in this case (ig. 3c). While we did not observe increased population densities in patches with corridors in this study, other effects of corridors, such as increased interpatch movement, may be more important for habitat specialists than for habitat generalists. We have found that cotton mice were more likely than cotton Downloaded from on 01 arch 018

5 4 JOURNAL O AALOGY Vol. 84, No. 1 TABLE 1. Tests of null hypothesis that Sigmodon hispidus, Peromyscus gossypinus, and Peromyscus polionotus are captured in interior, edge, and corridor habitat types with frequencies proportional to the number of traps in each habitat type. Expected (Exp.) and observed (Obs.) frequencies for male () and female () are shown (chi-square test). Habitat type Interior Edge Corridor Species Year Sex Exp. Obs. Exp. Obs. Exp. Obs. d.f. P S. hispidus P. gossypinus P. polionotus rats or old-field mice to move between patches without corridors linking them, indicating that more-specialized species may have greater difficulty moving through matrix habitat (abry and Barrett, in press). Greater-than-expected cotton rat captures in corridors (Table 1) suggest that old-field mice and cotton rats may be more likely to benefit from corridor presence than are cotton mice. Habitat use. We predicted that cotton mice would use all habitat types equally; this prediction was not supported as cotton mice were captured more frequently in interior sites during 1999 and 000 and were captured very infrequently in the forest matrix. As expected, old-field mice were captured more frequently in habitat-patch interiors than would be expected if captures were distributed proportionally among trap locations. We expected that cotton rats would be captured more frequently than expected in habitat-patch interiors; however, this species was instead captured more frequently than expected in corridors. While all 3 small-mammal species considered in this study moved through matrix habitat (abry and Barrett 00), few, if any, animals were resident in the pine forest matrix. Another potential explanation for low numbers of captures in the forest matrix is that animals moving between patches of suitable habitat may be less susceptible to trapping while in less-suitable habitat (Briese and Smith 1974). Placement of forest trapping grids (ig. 1) may also have influenced our ability to detect animals moving within the matrix, as animals moving between patches would not have been detected by trap grids on the perimeter of the site. The results of this study suggest that landscape elements may have greater effects on population dynamics of habitat specialists than on those of habitat generalists. urther studies of the effects of hab- Downloaded from on 01 arch 018

6 ebruary 003 ABRY ET AL. LANDSCAPE EECTS ON SALL AALS 5 itat fragmentation and habitat structure at multiple scales and on a range of species varying in degree of habitat specialization are needed. ACKNOWLEDGENTS Comments from. Buechner and anonymous reviewers improved the manuscript. We thank S. H. Schweitzer and. H. Smith for comments on earlier drafts of this manuscript and J. D. Peles for advice and assistance throughout the project. K. Biris, C. Brooks, C. Christopher, J. Jamell, A. ajeske, L. Paddock, and A. Pruett assisted in the field. We thank N. Haddad and R. Cheney for allowing us to use the experimental landscape, and J. Blake, J. Kilgo, and E. Olson of the United States Department of Agriculture orest Service, Savannah River, for supporting this project. unding was provided by the Department of Energy Savannah River Operations Office through the United States orest Service, Savannah River under Interagency Agreement DE-IA09-76SR This study was conducted under Cooperative Agreement DE-C09-96SR18546 between the United States Department of Energy and the University of Georgia Research oundation, and we acknowledge the Savannah River Ecology Laboratory for use of facilities. K. E. abry was supported by a National Science oundation Graduate Research ellowship during this study. LITERATURE CITED AERICAN SOCIETY O AALOGISTS Guidelines for the capture, handling, and care of mammals as approved by the American Society of ammalogists. Journal of ammalogy 79: BEIER, P., AND R.. NOSS Do habitat corridors provide connectivity? Conservation Biology 1: BOWNE, D. R., J. D. PELES, AND G. W. BARRETT Effects of landscape spatial structure on movement patterns of the hispid cotton rat (Sigmodon hispidus). Landscape Ecology 14: BRIESE, L. A., AND. H. SITH Seasonal abundance and movement of nine species of small mammals. Journal of ammalogy 55: CAERON, G. N., AND S. R. SPENCER Sigmodon hispidus. ammalian Species 158:1 9. COAN, C. J., J. D. NICHOLS, AND K. H. POLLOCK Population dynamics of icrotus pennsylvanicus in corridor-linked patches. Oikos 93:3 1. DAVIS-BORN, R., AND J. O. WOL Age- and sex-specific responses of the gray-tailed vole, icrotus canicaudus, to connected and unconnected habitat patches. Canadian Journal of Zoology 78: AHRIG, L., AND G. ERRIA Habitat patch connectivity and population survival. Ecology 66: GENTRY, J. B Invasion of a one-year abandoned field by Peromyscus polionotus and us musculus. Journal of ammalogy 47: GOERTZ, J. W The influence of habitat quality upon density of cotton rat populations. Ecological onographs 34: GOLLEY,. B., J. B. GENTRY, L.D.CALDWELL, AND L. B. DAVENPORT, JR Number and variety of small mammals on the AEC Savannah River Plant. Journal of ammalogy 76: HADDAD, N Corridor use predicted from behaviors at habitat boundaries. American Naturalist 153:15 7. HADDAD, N.., AND K. A. BAU An experimental test of corridor effects on butterfly densities. Ecological Applications 9: KAUAN, D. W., G. A. KAUAN, AND B. K. CLARK Small mammals in native and anthropogenic habitats in the Lake Wilson area of north-central Kansas. Southwestern Naturalist 45: KREBS, C. J Demographic changes in fluctuating populations of icrotus californicus. Ecological onographs 36: LA POLLA, V. N., AND G. W. BARRETT Effects of corridor width and presence on the population dynamics of the meadow vole (icrotus pennsylvanicus). Landscape Ecology 8:5 37. LIDICKER, W. Z., JR., J. O. WOL, L. N. LIDICKER, AND. H. SITH Utilization of a habitat mosaic by cotton rats during a population decline. Landscape Ecology 6: ABRY, K. E., AND G. W. BARRETT. 00. Effects of corridors on home range sizes and interpatch movements of three small-mammal species. Landscape Ecology 17: CINTYRE, S., AND R. HOBBS A framework for conceptualizing human effects on landscapes and its relevance to management and research models. Conservation Biology 13: ECH, S. G., AND J. G. HALLETT Evaluating the effectiveness of corridors: a genetic approach. Conservation Biology 15: OSSAN, C. A., AND P.. WASER Effects of habitat fragmentation on population genetic structure in the white-footed mouse (Peromyscus leucopus). Canadian Journal of Zoology 79: RICKETTS, T. H The matrix matters: effective isolation in fragmented landscapes. American Naturalist 158: ROSENBERG, D. K., B. R. NOON, AND E. C. ESLOW Biological corridors: form, function, and efficacy. BioScience 47: SAS INSTITUTE INC. 00. SAS Version 8.. SAS Institute Inc., Cary, North Carolina. STAPS, J. A.,. BUECHNER, AND V. V. KRISHNAN The effects of edge permeability and habitat geometry on emigration from patches of habitat. American Naturalist 19: Submitted 10 August 001. Accepted July 00. Associate Editor was John G. Kie. Downloaded from on 01 arch 018