SPATIAL SCALE DEPENDENCE OF RODENT HABITAT USE

Similar documents
HABITAT HETEROGENEITY, HABITAT ASSOCIATIONS, AND RODENT SPECIES DIVERSITY IN A SAND SHINNERY-OAK LANDSCAPE

Computational Ecology Introduction to Ecological Science. Sonny Bleicher Ph.D.

Long-term insights into the influence of precipitation on community dynamics in desert rodents

Welcome! Text: Community Ecology by Peter J. Morin, Blackwell Science ISBN (required) Topics covered: Date Topic Reading

Gary G. Mittelbach Michigan State University

Oecologia. The effects of owl predation on the foraging behavior of heteromyid rodents

PATH ANALYSIS: A CRITICAL EVALUATION USING LONG-TERM EXPERIMENTAL DATA

Diversity partitioning without statistical independence of alpha and beta

Occasional Papers. Non-geographic Variation of Chaetodipus eremicus and Chaetodipus nelsoni. Museum of Texas Tech University Number July 2018

Zero Sum, the Niche, and Metacommunities: Long-Term Dynamics of Community Assembly

Zoogeographic Regions. Reflective of the general distribution of energy and richness of food chemistry

Modeling Fish Assemblages in Stream Networks Representation of Stream Network Introduction habitat attributes Criteria for Success

Questions. Questions. Biodiversity. Biodiversity. Questions. Questions

Grant Opportunity Monitoring Bi-State Sage-grouse Populations in Nevada

NICHE BREADTH AND RESOURCE PARTIONING

Interactions among Land, Water, and Vegetation in Shoreline Arthropod Communities

Community Structure. Community An assemblage of all the populations interacting in an area

Patterns of morphology and resource use in North American desert rodent communities

Southwest LRT Habitat Analysis. May 2016 Southwest LRT Project Technical Report

Ryan P. Shadbolt * Central Michigan University, Mt. Pleasant, Michigan

Predation risk, unequal competitors and the ideal free distribution

Additional Case Study: Calculating the Size of a Small Mammal Population

of a landscape to support biodiversity and ecosystem processes and provide ecosystem services in face of various disturbances.

Desert Animals Survive Because

5 th Grade Ecosystems Mini Assessment Name # Date. Name # Date

7/29/2011. Lesson Overview. Vegetation Sampling. Considerations. Theory. Considerations. Making the Connections

EARTH SYSTEM: HISTORY AND NATURAL VARIABILITY Vol. III - Global Biodiversity and its Variation in Space and Time - D. Storch

CHAPTER 1 - INTRODUCTION. Habitat fragmentation, or the subdivision of once-continuous tracts of habitat into

Statistical Forecast of the 2001 Western Wildfire Season Using Principal Components Regression. Experimental Long-Lead Forecast Bulletin

SPECIES DIVERSITY CHANGES AND HABITAT ASSOCIATIONS OF SMALL MAMMALS AT ASH MEADOWS NATIONAL WILDLIFE REFUGE, NYE COUNTY, NEVADA THESIS

Sensitivity of FIA Volume Estimates to Changes in Stratum Weights and Number of Strata. Data. Methods. James A. Westfall and Michael Hoppus 1

Microhabitat Selection by Small Mammals

Bryan F.J. Manly and Andrew Merrill Western EcoSystems Technology Inc. Laramie and Cheyenne, Wyoming. Contents. 1. Introduction...

Climate Change and Community Dynamics: A Hierarchical Bayesian Model of Resource-Driven Changes in a Desert Rodent Community

Georgia Performance Standards for Urban Watch Restoration Field Trips

Approach to Field Research Data Generation and Field Logistics Part 1. Road Map 8/26/2016

BEHAVIORAL MECHANISMS OF COEXISTENCE IN SYMPATRIC SPECIES OF DESERT RODENTS, DIPODOMYS ORDII AND D. MERRIAMI

The Effect of Fire on Bird and Small Mammal Communities in the Grasslands of Wind Cave National Park

SPRING GROVE AREA SCHOOL DISTRICT PLANNED INSTRUCTION. Course Title: Wildlife Studies Length of Course: 30 Cycles

How Do Human Impacts and Geomorphological Responses Vary with Spatial Scale in the Streams and Rivers of the Illinois Basin?

ISLAND BIOGEOGRAPHY Lab 7

IUCN Red List Process. Cormack Gates Keith Aune

176 Index. G Gradient, 4, 17, 22, 24, 42, 44, 45, 51, 52, 55, 56

How does the greenhouse effect maintain the biosphere s temperature range? What are Earth s three main climate zones?

MODELING LIGHTNING AS AN IGNITION SOURCE OF RANGELAND WILDFIRE IN SOUTHEASTERN IDAHO

Utilization. Utilization Lecture. Residue Measuring Methods. Residual Measurements. 24 October Read: Utilization Studies and Residual Measurements

AP Environmental Science I. Unit 1-2: Biodiversity & Evolution

Priority areas for grizzly bear conservation in western North America: an analysis of habitat and population viability INTRODUCTION METHODS

SOIL AND VEGETATIVE ASSOCIATIONS OF HETEROMYID RODENTS IN CENTRAL AND SOUTH TEXAS WITH COMMENTS ON TRAPPING TECHNIQUES. Michelle E. Adcock, B.S.

Aggregations on larger scales. Metapopulation. Definition: A group of interconnected subpopulations Sources and Sinks

Rank-abundance. Geometric series: found in very communities such as the

Assessing state-wide biodiversity in the Florida Gap analysis project

Terrestrial Flora and Fauna

environmental conditions of Jacobsen s Beach, Kigoma, Tanzania.

Vegetation Structure Assessment (VSA):

John Erb, Minnesota Department of Natural Resources, Forest Wildlife Research Group

Minimizing interspecific competition by different foraging strategies in two North African desert rodents

Population Ecology. Study of populations in relation to the environment. Increase population size= endangered species

Fear and loathing on the landscape: What can foraging theory tell us about vigilance and fear? Commentary on Beauchamp on Fear & Vigilance

DISSERTATION DETERMINANTS OF HABITAT USE AND COMMUNITY STRUCTURE OF RODENTS IN NORTHERN SHORTGRASS STEPPE. Submitted by. Paul T.

What Shapes an Ecosystem Section 4-2

BIOS 6150: Ecology Dr. Stephen Malcolm, Department of Biological Sciences

Effect of Species 2 on Species 1 Competition - - Predator-Prey + - Parasite-Host + -

Community Ecology. Classification of types of interspecific interactions: Effect of Species 1 on Species 2

VarCan (version 1): Variation Estimation and Partitioning in Canonical Analysis

A Note on Bayesian Inference After Multiple Imputation

Overview. How many species are there? Major patterns of diversity Causes of these patterns Conserving biodiversity

Overhead cover 54.7 ± 33.6% 21.6 ± 27.5% 45.5 ± 34.7% 41.5 ± 36.8% 16.4 ± 24.5% 46.4 ± 34.4% 16.7 ± 20.0% 64.8 ± 27.6% 25.3 ± 30.

Metacommunities Spatial Ecology of Communities

ANIMAL ECOLOGY (A ECL)

14.1. Every organism has a habitat and a niche. A habitat differs from a niche. Interactions in Ecosystems CHAPTER 14.

Effects of Weather Conditions on the Winter Activity of Mearns Cottontail

Stability Of Specialists Feeding On A Generalist

Trait-based Australian mammal distribution patterns and extinction risks

Multivariate Analysis of Ecological Data using CANOCO

Weather is the day-to-day condition of Earth s atmosphere.

Community phylogenetics review/quiz

Evolution Common Assessment 1

DETECTING BIOLOGICAL AND ENVIRONMENTAL CHANGES: DESIGN AND ANALYSIS OF MONITORING AND EXPERIMENTS (University of Bologna, 3-14 March 2008)

Vegetation and Terrestrial Wildlife

Supplementary material: Methodological annex

Chapter 52 An Introduction to Ecology and the Biosphere

Describing Greater sage-grouse (Centrocercus urophasianus) Nesting Habitat at Multiple Spatial Scales in Southeastern Oregon

How to quantify biological diversity: taxonomical, functional and evolutionary aspects. Hanna Tuomisto, University of Turku

Chapter 6. Field Trip to Sandia Mountains.

Ecoregions Glossary. 7.8B: Changes To Texas Land Earth and Space

Lecture 2: Individual-based Modelling

An Introduction to Day Two. Linking Conservation and Transportation Planning Lakewood, Colorado August 15-16, 16, 2006

HOME RANGE SIZE ESTIMATES BASED ON NUMBER OF RELOCATIONS

TEST SUMMARY AND FRAMEWORK TEST SUMMARY

Sensitivity Analysis of Boundary Detection on Spatial Features of Heterogeneous Landscape

ASSESSING SEXUAL SEGREGATION IN DEER

ANOVA approach. Investigates interaction terms. Disadvantages: Requires careful sampling design with replication

SIF_7.1_v2. Indicator. Measurement. What should the measurement tell us?

14.1. KEY CONCEPT Every organism has a habitat and a niche. 38 Reinforcement Unit 5 Resource Book

Student Name: Teacher: Date: District: London City. Assessment: 07 Science Science Test 4. Description: Life Science Final 1.

СПИСАНИЯ с Импакт фактор за 2016 година в областта на екологията

Lesson 9: California Ecosystem and Geography

2017 Pre-AP Biology Ecology Quiz Study Guide

Maintenance of species diversity

Transcription:

SPATIAL SCALE DEPENDENCE OF RODENT HABITAT USE ERIC E. JORGENSEN AND STEPHEN DEMARAIS 78 South Monte Vista. Ada, OK 7482 (EEl) Department of Wildlife and Fisheries, Mississippi State University, Mississippi State, MS 39762 (SD) Present address of EEl: United States Environmental Protection Agency, P.O. Box 1198, Ada, OK 7482 Many insights into community ecology over the past 3 decades were derived from investigations of associations of rodent species with microhabitats. Nonetheless, studies of microhabitat use of rodents are inconsistent, suggesting spatially dependent interacting factors. We investigated the relative ability of microhabitat and macrohabitat to predict rodent captures in traps placed in 48 trapping grids of 9 traps each during spring and autumn of 1993 and 1994 (17,28 data points). Trapping grids represented eight replications of six discrete macrohabitats. We used discriminant function analysis and random null models to compare the ability of microhabitat and macrohabitat to predict use of individual traps by 13 rodent species. Classification rates for presence at a trap by dummy variables of macrohabitats exceeded those obtained with principle components of microhabitats for nine of 13 species. In seven of those cases, classification rate exceeded that expected from a random distribution of dummy variables. Of the four cases where principle components of microhabitats out-classified dummy variables of macrohabitats, only two exceeded rates expected from a random distribution of dummy variables. Thus, microhabitat partitioning for many species is constrained by local macrohabitat conditions. Key words: small mammals, Chihuahuan Desert, New Mexico, microhabitat, macrohabitat, spatial scale Partitioning of microhabitats among competing species is thought to contribute to coexistence among rodent species (Price, 1978), especially in relatively simple desert ecosystems. The role of partitioning of microhabitats in detennining rodent habitat associations has been well studied (Brown et al., 1979; Harris, 1986; Holbrook, 1978; M'Closkey and Fieldwick, 1975; Price, 1978; Price and Brown, 1983; Smartt, 1978). Further, the cause of this partitioning has been the focus of much research and theory since Rosenzweig and Winakur's (1969) first investigaiion (Brown, 1975; Brown and Bowers, 1984; Brown and Lieberman, 1973; Brown et ai., 1979; Bowers et al., 1987; Harris, 1984, 1986; Hutto, 1978; Kotler and Brown, 1988; M'Closkey, 1978; Rosenzweig, 1973; Wondolleck, 1978), Although differential use of microhabitat between species is almost certainly a real phenomenon, research results have not been consistent across studies. Thompson (1982b:I35) found that direct observation produced results that varied from concurrent live-trapping results and stated that "although microhabitat use patterns are in the same direction as described by earlier studies, the magnitude of resource overlap is such as to preclude partitioning of foraging effort along this spatial axis." Thompson (1987) also found that use of microhabitats determined from seed use did not correspond to patterns revealed by livetrapping. Rodent species trapped in a desert arroyo, including presumed open-microhabitat specialists, were captured more frequently in patch types characterized by thick vegetation (Jorgensen et al., 1995). It has been suggested that failure to consistently detect partitioning of microhabitats Journal of Mall/llll/logy. 8(2):421-429. 1999 421

422 JOURNAL OF MAMMALOGY Vol. 8, No.2 across study sites may be explained if it is specific to individual sites (Bowers, 1988; Thompson, 1982b). Such a pattern may occur when sites are the product of unique historical influences, (Brown, 1973, 1995). We believe the pattern also may arise when there is a relationship between spatial scales of habitat use where results depend upon the size of the microhabitat relative to the macrohabitat. Investigations of these spatial elements are highly dependent upon initial experimental conditions, particularly size of trapping grid. Also, investigations of scale dependency should be conducted in several habitats to ensure that affects and interpretations attributable to microhabitat are not confounded by unknown and experimentally uncontrolled macrohabitat causes. Morris (1984a, 1987a) tested the extent to which habitat scale governs abundance of Peromyscus and Microtus and found that microhabitat measurements were less effective predictors of density of rodents than was macrohabitat (Morris 1987a). Scale profoundly influences habitat use, interpretation of habitat analyses, (Morris, 1984a, 1984b, 1984c, 1987a, 1987b), and impedes a deeper understanding of ecological relationships, (Morris, 1987a). Morris suggested that scale-related research needs to be conducted in arid ecosystems where partitioning of microhabitats in rodent communities has been detected most clearly. We follow Morris' (1984b) suggestion that scale-related research needs to be conducted in ecosystems, of the arid southwestern United States. where partitioning of microhabitats among rodents may be strongly expressed. We compared ability of macrohabitat and microhabitat variables to predict presence or absence of 13 species of rodents at individual trap sites in six habitats associated with arroyos (dry desert drainages) and adjacent uplands in foothills of the Sacramento Mountains in south central New Mexico. MATERIALS AND METHODS Our research was conducted near the northern limit of the Chihuahuan Desert in foothills of the Sacramento Mountains, New Mexico, at an elevation of 1,37-1,68 m. Forty-eight grids (.58 ha each) were placed in eight replicate blocks of six macrohabitats of 126 km 2 Topography was formed by arroyos cutting through uplands. Arroyos and uplands were not isotypic in their vegetative and substrate condition. Because biotic conditions changed with elevation, we recognized three physiographic conditions (upper, middle, lower) within arroyos and uplands. Each sampling site consisted of a 3 by 3 grid, with lo-m spacing between traps. Grids were approximately rectangular following topography. One Shennan live trap (7.7 by 7.7 by 25.6 cm) baited with rolled oats was placed at each trap site for 4 nights. An entire block (arroyo-upland pair) was trapped simultaneously, and traps were subsequently rotated to another block. All sites were trapped during late spring and early autumn during 2 years (2 March-16 May 1993, 12 August.:...12 October 1993, 9 April-7 June 1994, and 2 August-23 September 1994). Analyses were conducted with SPSS for Windows (Norusis, 1994) and SAS (SAS Institute Inc. 1988). Concurrent with each trapping session, we measured the microhabitat associated with each trap site using a modification of Canfield's (1941) line-transect method. Transects consisted of two, 2-m poles oriented perpendicular to each other on a random azimuth and centered on each of the 9 trap sites for a total of 36 m of transect per trapping grid. We measured, or constructed from measurements, 59 microhabitat variables for each trap site from which we selected 1 that represented generalized habitat structure including: 1) distance from trap to nearest shrub (dm); 2) canopy cover of shrubs (%); 3) maximum shrub height (cm); 4) shrub point diversity (number of shrub species at a trap site); 5) distance to nearest grass or forb (dm); 6) cover of grasses and forbs to 4-cm height (cm out of 4 possible); 7) cover of grasses and forbs >4 cm in height (%); 8) cover of detritus (%); 9) cover of sandy material (%); and 1) cover of stones and rocks (%). Macrohabitat characteristics (representation of substrate constituents. grasses, forbs, shrubs) of each trap grid

May 1999 JORGENSEN AND DEMARAIS-RODENT HABITAT USE 423 TABLE I.-Hypothetical matrix summarizing analyses for one species. Habitat type corresponds to six habitats in eight replicate blocks that were assigned corresponding dummy variables; "" or "1." Discriminant analysis contrasts "Yes" and "No" in the Captured column; 1) on the basis of macrohabitat variables, 2) on the basis of microhabitat variables. A null distribution of 1, dis~ criminant analysis classification rates, contrasting Captured "Y" from "N" was created by randomly assigning trap sites to macrohabit. During application one macrohabitat would be coded with all O's to provide a full rank model. Macrohabitat Habitat Arroyo Upland type Upper Middle Lower Upper Middle Lower Trap site Captured 1 2 1 3 4 1 5 6 were detennined by pooling measurements ob~ tained from all 9 individual trap sites. During each of the 4 trapping periods, each trap site was classified as either capturing an in~ dividual of a particular species or not. Thus, we compared used sites with unused sites and made no effort to further quantify levels of use. Used sites were contrasted with unused sites with crossvalidated discriminant analyses (SAS Insti~ tute Inc., 1988) for each species on the data set that contained 17,28 observations. We first contrasted trap sites on the basis of five binary dummy variables representing ma~ crohabitats 1-6 (Table 1). We then contrasted trap sites using the first five varimax~rotated principal components of the 1 microhabitat variables. Principal components analysis was used to equalize the number of variables and to remove autocorrelation. Five principal compo~ nents were appropriate in this case, irregardless of eigenvalue, because the discriminant analysis used five macrohabitats (one macrohabitat was coded with all zeros to provide a full-rank mod~ ej). Importantly, we produced 1, random null data sets of macrohabitat dummy variables by randomly assigning trap sites to macrohabitat. Crossvalidated discriminant analyses were con~ ducted on each data set for each species. That procedure produced a distribution of crossvali~ dated classification success rates for predicting trap use in the dummy variable data set. By doing so, we compared the classification rate achieved by our observed data set to the distri~ 1 3 No 78 y" 1 31 No II No 1 65 y" 21 No bution of classification rates expected according to the properties of our data set That null model allowed assessment of significance of the observed classification rates. RESULTS In 69,12 trap nights, we captured 5,127 individuals and 18 species. Data for 13 species were used to analyze scale effects of macrohabitat and microhabitat. Modeling and analysis of macrohabitat associations of rodents from those sites was presented by Jorgensen (1996) and Jorgensen et a1. (1998). Briefly, sand tended to be the predominate substrate on uplands (especially at middle and lower physiographies), and detritus was prevalent in arroyos (especially at middle and lower physiographies; Table 2). Regarding vegetation, grass cover was especially great in arroyos at lower physiographies, shrub cover was greatest in arroyos (especially at middle physiographies), and forb cover tended to be highest on uplands (especially at upper and middle physiographies; Table 2). The first five principal components accounted for 83.2% of the variance in the microhabitat data set (Table 3). Because the use of principal components was strictly a numerical consideration, interpretation of the components is not needed for our anal-

424 JOURNAL OF MAMMALOGY Vol, 8, No.2 TABLE 2.-Average percent cover of structural habitat components, including characteristics of substrate, short «4 em) and tall (>4 em) grasses and forbs, and shrub cover from six macrohabitats associated with arroyos and uplands from three physiographies (Upper, Middle, and Lower) in the foothills of the Sacramento Mountains during spring and autumn of 1993 and 1994. Arroyo Upland Component Upper Middle Substrate Sand 3.6 34.6 Stone/rock 35.2 17.8 Detritus 26. 4,7 Vegetative Grass <4 em 1.4 8.6 Forbs <4 em 5.8 5. Grass >4 em.8 1.6 Forbs >4 em. 1.3 Shrub cover 35.6 62.8 yses nor would it be an issue in our discussion. However. the first principal component tended to represent shrub cover and associated detritus, the second principal component represented grass and forb cover, the third principal component represented a contrast between sandy substrata and stone, and the fourth principal component represented distance to nearest vegetation. The biological representation of the fifth principal component was difficult to assertain (Table 3). Significance levels derived from our null model of random classification rates (Table 4) were calculated as P = (number of ran- Lower Upper Middle Lower 31.5 43.8 61.2 68.5 4.2 31.6 13.2 5.7 52.2 16.5 2.2 2.2 29.7 5.5 2.8 9.4 5.1 12.5 13.7 5.6 15..2..7 2.6..1. 41.8 13.6 21.9 27.2 dom classification rates which exceeded observed classification ratell,ooo). Thus, a significance level of P =.176 (macrohabitat rate for Dipodomys merriami) means that 176 random rates exceeded the observed rate of 58.8%. Dummy variables of macrohabitats classified with greater success than principal components of microhabitats for nine species. Of those, rates for seven species exceeded those expected from a random distribution by large amounts (P <.1; Table 4; i.e., <1 random data sets exceed the observed classification rate). Conversely, principal components of mi- TABLE 3.-Principal component (PC) varimax-rotated loadings and eigenvalues on microhabitat structural measures for trap sites dispersed in six macrohabitats in the foothills of the Sacramento Mountains during spring and autumn of 1993-1994. Microhabitat PCl PC2 PC3 PC4 PC5 Shrub cover.88 -.2.12 -. -.2 Shrub height.75 -.6.28.2 -.14 Shrub diversity.68 -.34 -.5 -.19.31 Distance to shrub -.5.38.31.42 -.28 Distance to grass or forb.14 -.4.35.72.33 Grasses and forbs «4 em).7.83 -.4 -.2.27 Grasses and forbs (>4 cm).13.8.7.18.39 Detritus.75.43.6.13 -.33 Sand -.55 -.32.68 -.29.15 Stone and rock -.17 -.25 -.87.26.8 Eigenvalue 2.98 2.14 1.55.98.67

May 1999 JORGENSEN AND DEMARAIS-RODENT HABITAT USE 425 TABLE 4.-bserved discriminant-analysis classification rate contrasting trap sites that captured an animal from trap sites that did not for n individuals of each species. Macrohabitat rate is rate observed from five dummy variables. Microhabitat rate is rate observed from five principal components. Significance level (P) is rate at which classification by a null model data set of 1, random observations of macrohabitat exceeded observed rates for macrohabitat and microhabitat. The range of classification rates obtained from the 1, random observations is presented. Species Dipodomys merriami Dipodomys ordii Chaetodipus intennedius Chaetodipus penicillatus Perognathus jiavus Perognathus jiavescens Peromyscus leucopus Peromyscus eremicus Peromyscus maniculatus Reithrodontomys megalotis Neotoma albigula Neotoma micropus Sigmodon hispidus If) c:: o :;::: CO ~i If) ~~ 16 14 12 1 (3 ~ 8 b ~g 6 U;;' c:: " Q):i 4 :::J - ~ 2 U. a Qulnmodal DlstribuUan of Rarnlom Varlables n 1,366.588 177.866 42.769 55.795 225.84 263.651 48.596 319.87 7S.586 4S.654 415.728 97.275 189.935 Macro Dummy o ~ ~ g ~ ~ ~ R g g g - Percentage of correct classifications FIG. l.-classification success rates; with dummy variables of macrohabitats exceed rates for microhabitat principal components and are Significantly higher than those expected from a random distribution, Example is for captures of Chaetodipus intermedius. Range of Macrohabitat Microhabitat classification Rate P Rate P rates.176.73.5 84.7-17.3..718.44 75.7-17.6.7.676.143 87.6-17.6.4.679.12 86.5-17.2..673.71 76.4-17.9.82.77.65 76.-17.1.134.695.91 86.7-17.5..546.285 73.9-17.5.184.517.38 92.4-..14.67.14 87.-17.1.5.651.118 87.1-18..886.627.45 81.-17.6..694.47 79.-. crohabitat classified more successfully than did random macrohabitat dummy variables for only four species; Dipodomys merriami, Perognathus fiavescens, Peromyscus leucopus, and Neotoma micropus (Table 4). Of those however, rates differed from random only for Dipodomys merriami (P = :5) and Neotoma micropus (P =.45; Table 4). Finally, observed classification rates of macrohabitats for Dipodomys ordii and Sigmodon hispidus exceeded all random sets while classification rates of microhabitats for these species were exceeded by about 5% of random data sets (Table 4). The random distribution of classification rates produced by the null-model discriminant analysis was not normal but was quinmodal (e.g., Chaetodipus intermedius; Fig. 1). This is because five macrohabitats were coded with a «1" in a full rank analysis. Thus, small differences in classification rates between macrohabitat and microhabitat (i.e., C. intermedius, 77.9% macrohabitat versus 67.6% microhabitat; Fig. 1 and Table 4) can produce large differences in

426 JOURNAL OF MAMMALOGY Vol. 8, No.2 significance (P =.7 versus P =.143; Table 4). Following our procedure, small differences in classification rate (1.3% for C. intermedius) could not be assumed to represent equivalent or similar results. The quinmodal distribution of random classification rates could result in intervening peaks, as was the case for C. intennedius where one of the random distribution peaks occurred between the macrohabitat and microhabitat classification rates (Fig. 1) and strongly influenced significance of the ou11- model analysis (Table 4). DISCUSSION It is clear that durruny variables of macrohabitats are better at predicting if any particular trap catches an animal than are microhabitat variables (represented by principal components) specific to the trap site. Morris (1987a) identified the contrary nature of this result for two species of northern rodents. We confirm this result and note its relatively clear expression at our Chihuahuan Desert sites. Presuming the primacy of microhabitat partitioning, we expected that classification rates with a multivariate suite of five principal components created from 1 microhabitat variables specific to the trap site would exceed classification rates attained with an equal number of (very crude) "" or "1," binary dummy variables of macrohabitat. Yet, this typically did not occur. Microhabitat and macrohabitat exist along a gradient of changing patch types and scales (Rosenzweig, 1973). While our results do not deny the potential reality of microhabitat partitioning for some species (i.e., microhabitat classification rates for D. merriami were significantly greater than random; Table 4), they do suggest that microhabitat partitioning is constrained by macrohabitat. That is, partitioning of microhabitats exists only within limits set by as yet poorly understood macrohabitat factors (poorly understood in the sense that interaction of these factors with microhabitat is not known). A direct interpretation of our data is that in unused macrohabitats (macrohabitats that support or presently include only a few representatives of a species), there are many unused microhabitat sites that are _ equivalent to microhabitat sites in used macrohabitats. Thus, in contrast to prevailing knowledge, density (as measured by trapuse) is not determined by local availability of microhabitats (sensu Reichman and Price, 1993). Suitable microhabitat is found in all macrohabitats. Density is determined by macrohabitat. To make an illustrative analogy, brush specialists are not found in every bush; they are found in brushy macrohabitats. Predation-related mechanisms represent some of the most persuasive evidence supporting microhabitat partitioning (Kotler, 1984, 1989; Kotler et aj.. 1988, 1991; Longland. 1994; Langland and Price. 1991; Pierce et aj.. 1992; Price et aj.. 1984). If predation-related mechanisms are causing microhabitat partitioning. it might be expected that macrohabitat specificity of potential predators would play an important role in expression of the microhabitat partitioning by rodent species. Dipodomys merriami was the only species for which classification by microhabitat variables substantially (P <.1) exceeded classification by crude binary macrohabitat variables, which contrasts to the seven species for which macrohabitat variables performed substantially better (Table 4). Therefore, D. merriami may not be an appropriate model organism for extrapolation of microhabitat partitioning to other species. Although our method differs slightly from Morris' (l987a), our results are highly concordant with his and others (Bowers, 1988; Bowers and Flanagan, 1988; Thompson, 1982a, 1982b). Our data are not in agreement with the prevailing expectation that microhabitat partitioning allows sympatry between coexisting rodent species. There is a need to interpret these results in light of the many cited observations of mi-

May 1999 JORGENSEN AND DEMARAIS-RODENT HABITAT USE 427 crohabitat partitioning. If use of microhabitat is constrained by macrohabitat. many environmental factors have the potential to confound observations. We draw attention to two. First, Monis' (1987a) and our data suggest that microhabitat partitioning is constrained by macrohabitat. Therefore. previ- us microhabitat studies, conducted in single habitats. are not widely generalizable because of potentially confounding macrohabitat factors. Second, Thompson's (1982a, 1982b) data suggest that the conventional openspecialist, shrub-specialist dichotomy may not even be real but may be an artifact of sampling methodology. Thompson (1982b) made the connection between observer-induced disturbance and incre~sed trap success. Physical disturbance associated with maintenance of trap lines may have a relatively greater, or qualitatively different, impact in open (sandy) areas than areas covered with detritus. Rodents may simply be attracted to potentially large or qualitatively different disturbances to substrata caused by researchers as traps are set. Root (1997) found evidence to support this hypothesis. The uncertainty principal (Heisenberg. 1927, 1977, 1983) holds that the effect of observing an experiment on its outcome cannot be known. Observer-induced bias may be a factor confounding previous microhabitat investigations. Morris (1987a) suggested that microhabitat data do little to explain density relative to macrohabitat data and our data supports this conclusion for desert rodents. Thompson's (1982b) data suggested that lack of concordance between studies of microhabitat may have a meth9,dological bias. In either case, generalization to other ecosystems is inadvisable (Bowers, 1988; Bowers and Flanagan, 1988). Given the central place that investigations of rodent communities have played in ecology over the past 3 decades. development of clear understandings of microhabitat phenomena is important. Although questions differ slightly, we prefer the discriminant-analysis approach described herein to Morris' (1987a) use of multiple regression of factor scores for the following reason; the biological meaning of Morris' (1987a) densities within subplots and the ability of these subplots to mimic microhabitat are sources of uncertainty. Morris' (1987a) subplots seem to be more like patches (Jorgensen et ai., 1995). We believe our approach, using each trap site as a 'subplot' and capture or no-capture as 'density' estimators effectively eliminates this uncertainty. Our method is at least equivalent to Morris' (1987a) in this regard and represents an alternative, but one which sacrifices less interpretability than Morris' factor-score approach regarding applications to microhabitat investigations. hnportantly, the approach we present here allows generation of random distributions (null models) that allows direct assessment of significance. Our findings show that macrohabitat based models (Jorgensen et al., 1998) are probably best for conservation and management of rodent species in the southwestern United States. Microhabitat data, although potentially of local applicability, will probably not be of use ~o modelers except under very specific conditions. ACKNOWLEDGMENTS We thank the United States Army Construction Engineering Research Laboratories and The Directorate of Environment, Fort Bliss for financial and logistical support, particularly K. von Finger, B. Russell, D. Price, and W. Whitworth. We thank our technicians: S. Sell, S. Lerich. M. Vogel, S. Rockafellow, H. Wilson, K. WeiSt E. Bartz. P. Blanton, T. Monasrnith, S. Neff, K. Hallman, V. Vicenti, and M. Hensley. J. Ryan helped write the randomization procedure. Consultations with P. Westfall were beneficial. Reviews by D. Morris, M. Willig. C. Jones. N. Mathews, and L. Densmore have improved the manuscript. During the conduct of this research the authors were employed by the Department of Range, Wildlife, and Fisheries Management, Texas Tech University.

428 JOURNAL OF MAMMALOGY Vol. 8, No.2 LITERATURE CITED BOWERS, M. A. 1988. Seed removal experiments on desert rodents: the microhabitat by moonlight effect. Journal of Mammalogy, 69:21-24. BOWERS, M. A., AND C. A. FLANAGAN. 1988. Microhabitat as a template for the organization of a desert rodent community. Pp. 3-312, in Management of amphibians, reptiles, and small mammals in North America, Proceedings of the symposium (R. C. Szaro, K. E. Severson, and D. R. Patton, eds.). United States Department of Agriculture. Forest Service, General Technical Report, RM-166:1-458. BOWERS, M. A., D. B. THOMPSON, AND 1. H. BROWN. 1987. Spatial organization of a desert rodent community: food addition and species removal. Oecologia, 72:77---S2. BROWN, 1. H. 1973. Species diversity of seed-eating desert rodents in sand dune habitats. Ecology, 54: 775-787. ---. 1975. Geographical ecology of desert rodents. pp. 315-341, in Ecology and evolution of communitics em. L. Cody and J. M. Diamond, eds.). The Belnap Press of Harvard University Press, Cambridge, Masschusetts. ---. 1995. Macroecology. The University of Chicago Press, Chicago, Illinois. BROWN, J. H., AND M. A. BOWERS. 1984. Patterns and processes in three guilds of terrestrial vertebrates. Pp. 282-296, in Ecological communities conceptual issues and the evidence (D. R. Strong, Jr., D. Simberloff, L. G. Abele, and A. B. Thistle, eds.). Princeton University Press, Princeton, New Jersey. BROWN, J. H., AND G. A. LIEBERMAN. 1973. Resource utilizationand coexistence of seed-eating rodents in sand-dune habitats. Ecology, 54:788-797. BROWN, J. H., O. J. REICHMAN, AND D. W. DAVIDSON. 1979. Granivory in desert ecosystems. Annual Review of Ecology and Systematics, 1:21-227. CANFIELD, R. H. 1941. Application of the line interception method in sampling range vegetation. Journal of Forestry, 39:388-394. HARRIS, J. H. 1984. An experimental analysis of desert rodent foraging ecology. Ecology, 65:1579-1584. ---. 1986. Microhabitat segregation in two desert rodent species: the relation of prey availability to diet. Oecologia, 68:417-421. HEISENBERG, W. 1927. Uber den anschaulichen inhalt der quantentheoretischen kinematik und mechanik. Zeitschrift for Physik, 43: 172-198. ---. 1977. Remarks on the origin of the relations of uncertainty. Pp. 3-6, in The uncertainty principal and foundations of quantum mechanics, a fifty years' survey (W C. Price and S. S. Chissick, eds.). John Wiley & Sons, London, United Kingdom. ---. 1983. The physical content of quantum kinematics and mechanics. Pp. 62-84, ill Quantum theory and measurement (J. A. Wheeler and W. H. Zurek, eds.). Princeton University Press, Princeton, New Jersey. HOLBROOK, S. J. 1978. Habitat relationships and coexistence of four sympatric species of Pemmyscu.~ in northwestern New Mexico. Journal of Mammalogy, 59:18-26. HUTIo, R. L. 1978. A mechanism for resource allocation among sympatric heteromyid rodent species. Oecoiogia, 33:115-126. JORGENSEN, E. E. 1996. Small mammal and herpetofauna communities and habitat associations in foothills of the Chihuahuan Desert. Ph.D. dissertation, Texas Tech University, Lubbock. JORGENSEN, E. E., S. DEMARAIS, AND S. NEFF. 1995. Rodent use of microhabitat patches in desert arroyos. The American Midland Naturalist, 134:193-199. JORGENSEN, E. E., S. DEMARAIS, S. M. SELL, AND S. P. LERICH. 1998. Modeling habitat suitability for small mammal in Chihuahuan Desert foothills of New Mexico. The Journal of Wildlife Management, 62: 989-996. KOTLER, B. P. 1984. Risk of predation and the structure of desert rodent communities. Ecology, 65:689-71. ---. 1989. Temporal variation in the structure of a desert rodent community. Pp. 127-139, in Patterns in the structure of mammalian communities (D. W. Morris, Z. Abramsky, B. J. Fox, and M. R. Willig, eds.). Texas Tech University Press, Lubbock. KOTLER, B. P., AND J. S. BROWN. 1988. Environmental heterogeneity and the coexistence of desert rodents. Annual Review of Ecology and Systematics, 19: 281-37. KOTLER, B. P., J. S. BROWN, AND O. HASSAON. 1991. Factors affecting gerbil foraging behavior and rates of owl predation. Ecology, 72:2249-226. KOTLER, B. P., J. S. BROWN, R. J. SMITH, AND W O. WIRTZ, II. 1988. The effects of morphology and body size on rates of owl predation on desert rodents. Oikos, 53:145-152. LaNGLAND, W. S. 1994. Effects of artificial bush canopies and illumination on seed patch selection by heteromyid rodents. The American Midland Naturalist, 132:82-9. LaNGLAND, W. S., AND M. V. PRICE. 1991. Direct observations of owls and heteromyid rodents: can predation risk explain microhabitat use? Ecology, 72: 2261-2273. M'CLOSKEY, R. T. 1978. Niche separation and assembly in four species of Sonoran Desert rodents. The American Naturalist, 112:683-694. M'CLOSKEY, R. 1:, AND B. FIELDWICK. 1975. Ecological separation of sympatric rodents (Peromyscus and Microtus). Journal of Mammalogy, 56:119-129. MORRIS, D. W 1984a. Microhabitat separation and coexistence of two temperate-zone rodents. The Canadian Field-Naturalist, 98:215-218. ---. 1984b. Sexual differences in habitat use by small mammals: evolutionary strategy or reproductive constraint? Oecologia, 65:51-57. ---. 1984c. patterns and scale of habitat use in two temperate-zone small mammal faunas. Canadian Journal of Zoology, 62:154-1547. ---. 1987a. Ecological scale and habitat use. Ecology, 68:362-369. ---. 1987 h. Tests of density-dependent habitat selection in a patchy environment. Ecological Monographs, 57:269-281. NORUSIS, M. J. 1994. SPSS, SPSS for Windows. Advanced Statistics 6.1. SPSS, Inc., Chicago, Illinois. PIERCE, B. M., W S. LaNGLAND, AND S. H. JENKINS. 1992. Raltlesnake predation on desert rodents: mi-

May 1999 JORGENSEN AND DEMARAIS-RODENT HABITAT USE 429 crohabitat and species-specific effects on risk. Journal of Mammalogy, 73:859-865. PRICE, M. V. 1978. The role of microhabitat in structuring desert rodent communities. Ecology, 59:91-921. PRICE, M. V, AND J. H. BROWN. 1983. Patterns of morphology and resource use in North American desert rodent communities. The Great Basin Naturalist Memoirs, 7:117-134. 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. REICHMAN, O. J., AND M. V. PRICE. 1993. Ecological aspects of heteromyid foraging. Pp. 539-574, in Biology of the Heteromyidae (H. H. Genoways and J. H. Brown, eds.). Special Publication, The American Society of Mammalogists, 1: 1-719. ROOT, J. J. 1997. Microsite and habitat boundary influences on small mammal capture, diversity, and movements. M.S. thesis, Texas Tech University, Lubbock. ROSENZWEIG, M. L 1973. Habitat selection experiments with a pair of coexisting heteromyid species. Ecology, 54:111-117. ROSENZWEIG, M. L, ANO J. WINAKUR. 1969. Population ecology of desert rodent communities: habitats and environmental complexity. Ecology, 5:558-572. SAS INSTITUTE INc. 1988. SAS/STAT User's Guide. Release 6.3 Edition. SAS Institute Inc., Cary, North Carolina. SMARTT, R. A 1978. A comparison of ecological and morphological overlap in a Peromyscus community. Ecology, 59:216-22. THOMPSON, S. D. 1982a. Structure and species composition of desert heteromyid rodent species assemblages: effects of a simple habitat manipulation. Ecology, 63:1313-1321. ---. 1982b. Microhabitat utilization and foraging behavior of bipedal and quadrupedal heteromyid rodents. Ecology, 63:133-1312 ---. 1987. Resource availability and microhabitat use by Merriam's kangaroo rats, Dipodomys merriami, in the Mojave Desert. Journal of Mammalogy, 68:256-265. WONDOLLECK, J. T. 1978. Forage-area separation and overlap in heteromyid'rodents. Journal of Mammalogy, 59:51-518. Submitted 3 October 1997. Accepted 2 July 1998. Associate Editor was Janet K. Braun