Historic, Archive Document Do not assume content reflects current scientific knowledge, policies, or practices. ■1 sou United States ^ ; Department of Agriculture Forest Service Rocky Mountain Forest and Range Experiment Station Fort Collins, Colorado 80526 General Technical Report RM-166 Management of Amphibians, Reptiles, and Small Mammals ir North America Proceedings of the Symposium July 19-21,1988 Flagstaff, Arizona ■■■■■■ "-T^^j^-^ .■■^'■i^^<'r~^ ACKNOWLEDGEMENTS This meering owes its success to sev- eral organizations and individuals. First, we thank the sponsoring or- ganizations (listed on the title page) whose financial support and encour- agement helped make the conference a reality. The local committee on ar- rangements, J. Kevin Aitkin, Marga- ret Bailey, Tom Britt, Roxanne Britt, Charles BuUington, Glen Dickens, and Katherine Holly did a superb job of handling room setup, registration, providing rides, and running the slide projector. We are especially grateful to the session chairman, K. Bruce Jones, George Dalrymple, Robert M,Closkey, David Germano, Winifred Sidle, Constantine Slobod- chikoff, Michael Morrison, Gregory Adler, Martin Raphael, and Ray- mond Dueser, for their help and for keeping the meeting on schedule. Our thanks to those who attended for their enthusiastic participation. We thank Randall Babb for the line drawings in the proceedings and De- borah Johnson and J. Kevin Aitkin for their help in organizing manu- script files and standardizing word processing formats. We would like to extend our sin- cere thanks to the following peer re- viewers who generously gave their time to improve the quality of this proceedings: Gregory H. Adler, Stanley H. Anderson, Michael J. Armbruster, David M. Armstrong, Walter Auffenberg, Keith B. Aubry, Gary C. Bateman, Ronald E. Beiswenger, Kristin H. Berry, Wil- liam M. Block, Michael A. Bowers, Richard C. Bruce, James H. Brown, K. A. Buhlmann, Russell Burke, R. Bruce Bury, Ronald K. Chesser, Ste- ven P. Christman, Tim W. Clark, Tames P. Collins, Stephen Corn, Stephen P. Cross, George Dalrymple, Joan E. Diemer, James G. Dickson, C. Kenneth Dodd, Jr., Raymond D. Dueser, Gary M. Fellers , Henry S. Fitch, Jerran Flinders, Vagn F. Flyger, Kenneth Feluso, Richard Fitzner, David J. Germano, Lowell L. Getz, William E. Grant, Patrick T. Gregory, Marc P. Hayes, Clyde Jones, K. Bruce Jones, Donald W. Kaufman, Brian J. Klatt, Thomas Kunz, J. Larry Lan- ders, James N. Layne, Harvey B. Lil- lywhite, Raymond Linder, William Mannan, S. Clark Martin, Robert T. M'Closkey, David A. McCullough, Gary K. Meefe, Joseph C. Mitchell, Paul E. Moler, Henry R. Mushinsky, Thomas J. O'Shea, William S. Parker, Kenneth H. Pollock, Mary V. Price, Martin G. Raphael, O. J. Reichman, Fred B. Samson, D. J. Schmidly, Nor- man Scott, Steven W. Seagle, Ray- mond D. Semlitsch, Henry L. Short, Lee H. Simons, Graham W. Smith, Hobart M. Smith, Dan Speake, James R. Spotila, Judy A. Stamps, Thomas P. Sullivan, Daniel W. Uresk, Laurie J. Vitt, Peter D. Weigl, Gary C. White, Daniel F. Williams, Richard G. Zweifel. Finally, we thank the speakers for following our schedule for submit- ting the various stages of their manu- scripts and providing us with excel- lent manuscripts in computer format to expedite and enhance the publica- tion of the proceedings. The opinions expressed in these papers are the au- thors' and do not necessarily reflect those of the U.S. Department of Agri- culture. 30301 Baltimore SVd Beltsv/lle. MO 20705.2351 ral Lfbrary USDA Forest Service General Technical Report RM-166 November 1988 Management of Amphibians, Reptiles, and Small Mammals in North America Proceedings of the Symposium July 19-21, 1988 Flagstaff, Arizona Robert C. Szaro, Kieth E. Severson, and David R. Patton technical coordinators^ Sponsored by: Arizona Chapter of the Wildlife Society Arizona Game and Fish Department Northern Arizona University, School of Forestry USDA Forest Service, Rocky Mountain Forest and Range Experiment Station USDA Forest Service, National Wildlife and Fish Ecology Program USDA Forest Service, Southwestern Region 'Szaro and Severson are with the USDA Forest Service, Rocky Mountain Forest and Range Experiment Station, at the Station's Research VJork Unit in fempe, in cooperation with Arizona State University. Patton is with the School of Forestry. Northern Arizona University, Flagstaff. The Management of Amphibians, Reptiles and Small Mammals in North America: Historical Perspective and Objectives Robert C. Szaro 1 The Management of Amphibians, Reptiles and Small Mammals in North America: The Need for an Environmental Attitude J. Whitfield Gibbons 4 Douglas-Fir Forests in the Oregon and Washington Cascades: Relation of the Herpetofauna to Stand Age and Moisture R. Bruce Bury and Paul Stephen Corn 11 Long-Term Trends in Abundance of Amphibians, Reptiles, and Mammals in Douglas-Fir Forests of Northwestern California Martin G. Raphael 23 Use of Woody Debris by Plethodontid Salamanders in Douglas- Fir in Washington Keith B. Aubry, Lawrence L C. Jones, and Patricia A. Hall 32 Forestry Operations and Terrestrial Salamanders: Techniques in a Study of the Cow Knob Salamander, Plethodon punctafus Kurt A. Buhlmann, Christopher A. Pague, Joseph C. r\/litchelL and Robert B. Glasgow 38 Conserving Genetically Distinctive Populations: The Case of the Huachuca Tiger Salamander (Ambystoma tighnum sfebbinsi Lowe) James P. Collins, Thomas R. Jones, and Howard J. Berna 45 Habitat Requirements of New Mexico's Endangered Salamanders Cynthia A. Ramotnik and Norman J. Scott, Jr. 54 Utilization of Abandoned Mine Drifts and Fracture Caves By Bats and Salamanders: Unique Subterranean Habitat in the Ouachita Mountains David A. Saugey, Gary A. Heidt, and Darrell R. Heath 64 The Herpetofauna of Long Pine Key, Everglades National Park, in Relation to Vegetation and Hydrology George H. Dalrymple 72 The Herpetofaunal Community of Temporary Ponds in North Florida Sandhills: Species Composition, Temporal Use, and Management Implications C. Kenneth Dodd, Jr. and Bert G. Charest 87 (Continued) Management of Amphibians, Reptiles, and Small Mammals in Xeric Pinelands of Peninsular Florida /. Jock Stout, Donold R. Richordson, ond Richord E. Roberts 98 Distribution and Habitat Associations of Herpetofauna in Arizona: Comparisons by Habitat Type K. Bruce Jones 109 Multivariate Analysis of thie Summer Habitat Structure of Rana pipiens Sctireber, in Lac Saint Pierre (Quebec, Canada) N. Beouregord ond R. Lecloir Jr. 1 29 Habitat Correlates of Distribution of the California Red-Legged Frog (Rona aurora draytonii) and the Foothill Yellow- Legged Frog (Rana boylii): Implications for Management More P. Hoyes ond Mork R. Jennings 144 Integrating Anuran Amphibian Species into Environmental Assessment Programs Ronold E. Beiswenger 159 PrelimirKjry Report on Effect of Bullfrogs on Wetland Herpetofaunas in Southeastern Arizona Cecil R. Schwolbe ond Philip C. Rosen 166 Developing Management Guidelines for Snapping Turtles Ronold J. Brooks, Dovid A. Golbroith, E. Grohom Noncekiveli, ond Christine A. Bishop 1 74 Spatial Distribution of Desert Tortoises (Gopherus agassizii) at Twentynine Palms, California: Implications for Relocations Ronold J. Boxter 180 Changes in a Desert Tortoise (Gopherus agassizii) Population After a Period of High Mortality Dovid J. Germono ond Michele A. Joyner 190 A Survey Method for Measuring Gopher Tortoise Density and Habitat Distribution Doniel M. Spillers ond Don W. Speoke 199 Evaluation and Review of Field Techniques Used to Study and Manage Gopher Tortoises Russell L Burke ond Jomes Cox 205 Talus Use by Amphibians and Reptiles in the Pacific Northwest Robert E. Herrington 216 Comparison of Herpetofounas of a Natural and Altered Riparian Ecosystem K. Bruce Jones 222 Critical Habitat, Predator Pressures, and the Management of Epicrates monoensis (Serpentes: Boidae) on the Puerto Rico Bank: A Multivariate Analysis Peter J. Tolson 228 The Use of Timed Fixed-Area Plots and a Mark-Recapture Technique in Assessing Riparian Garter Snake Populations Robert C. Szaro, Scott C. Belfit, J. Kevin Aitkin, and Randall D. Babb 239 Design Considerations for the Study of Amphibians, Reptiles and Small Mammals in California's Oak Woodlands: Temporal and Spatial Patterns William M. Block, Michael L Morrison, John C. Slaymaker, and Gwen Jongejan 247 The Importance of Biological Surveys in Managing Public Lands in the Western United States Michael A. Began, Robert B. Finley, Jr, and Stephen J. Petersburg 254 Sampling Problems in Estimating Small Mammal Population Size George E. Menkens, Jr. and Stanley H. Anderson 262 The Design and Importance of Long-Term Ecological Studies: Analysis of Vertebrates in the Inyo-White Mountains, California Michael L. Morrison 267 An Ecological Problem-Solving Process for Managing Special- Interest Species Henry L. Short and Samuel C. Williamson 276 Comparative Effectiveness of Pitfalls and Live-Traps in Measuring Small Mammal Community Structure Robert C. Szaro, Lee H. Simons, and Scott C. Belfit 282 The Role of Habitat Structure in Organizing Small Mammal Populations and Communities Gregory H. Adier 289 Microhabitat as a Template for the Organization of a Desert Rodent Community Michael A. Bowers and Christine A. Flanagan 300 (Continued) Response of Small Mammal Communities to Silvicultural Treatments in Eastem Hardwood Forests of West Virginia and Massachusetts Robert T. Brooks and William M. Healy 313 Habitat Structure and ttie Distribution of Small Mammals in a Northiern Hardwoods Forest JefferyA. Gore 319 Thie Value of Rocky Mountain Juniper (Juniperus scopulorum) Woodlands in Soutti Dakota as Small Mammal Habitat Carolyn Hull Sieg 328 Postfire Rodent Succession Following Prescribed Fire in Southern California Chaparral William O. Wirfz, II, David Hoekman, John R. Muhm, and Sherrie L Sauza 333 Douglas-Fir Forests in the Cascade Mountains of Oregon and Washington: Is the Abundance of Small Mammals Related to Stand Age and Moisture? Paul Stephen Corn, R. Bruce Bury, and Thomas A. Spies 340 Evaluation of Small Mammals as Ecological Indicators of Old- Growth Conditions Kirk A. Nordyke and Steven W. Buskirk 353 Habitat Associations of Small Mammals in a Subalpine Forest, Southeastern Wyoming Martin G. Raphael 359 Differences in the Ability of Vegetation Models to Predict Small Mammal Abundance in Different Aged Douglas- Fir Forests Cathy A. Taylor, C. John Ralph, andArlene T. Doyle 368 Small Mammals in Streamside Management Zones in Pine Plantations James G. Dickson and J. Howard Williamson 375 Patterns of Relative Diversity Within Riparian Small Mammal Communities, Platte River Watershed, Colorado Thomas E. Olson and Fritz L. Knopf 379 Estimated Carrying Capacity for Cattle Competing with Prairie Dogs and Forage Utilization in Western South Dakota Daniel W. Uresk and Deborah D. Paulson 387 (Continued) Cattle Grazing and Small Mannmals on the Sheldon National Wildlife Refuge, Nevada John L Oldemeyer and Lydia R. Allen-Johnson 391 Effect of Seed Size on Removal by Rodents William G. Standley 399 Habitat Use by Gunnison's Prairie Dogs C, N. Slobodchikoff, Anthony Robinson, and Clark Schaack 403 Environmental Contaminants and the Management of Bat Populations in the United States Donald R. Clark, Jr. 409 Habitat Structure, Forest Composition and Landscape Dimensions as Components of Habitat Suitability for the Delmarva Fox Squirrel Raymond D. Dueser, James L. Dooley, Jr., and Gary J. Taylor ....414 Effects of Treating Creosotebush with Tebuthiuron on Rodents William G. Standley and Norman S. Smith 422 Foraging Patterns of Tassel-Eared Squirrels in Selected Ponderosa Pine Stands Jacks States, William S. Gaud, W. Sylvester Allred, and William J. Austin 425 Small Mammal Response to the Introduction of Cattle into a Cottonwood Floodplain Fred B. Samson, Fritz L. Knopf, and Lisa B. Mass 432 Old Growth Forests and the Distribution of the Terrestrial Herpetofauna HartwellH. Welsh, Jr. and Amy L. Lind 439 The Management of Amphibians, Reptiles and Small Mammals in North America: Historical Perspective and Objectives^ Robert C. Szaro^ Historically the management of pub- lic lands from a multiple use perspec- tive has led to a system that empha- sizes those habitat components or faunal elements that primarily re- sulted in some sort of definable eco- nomic value. While this often benefit- ted other species that were not even considered in the original prescrip- tions, it also negatively impacted oth- ers. We no longer can afford to take this simplistic view of ecosystem management. We need to use a more holistic approach where ecological landscapes are considered as units, and land management practices in- corporate all elements into an inte- grated policy. This includes examin- ing the impacts of proposed land uses on amphibian, reptile, and small mammal populations. With the passage of the National Forest Management Act of 1976, the monitoring of all renewable natural resources became law. Even with this legislation, most emphasis by Na- tional Forests in the United States has been placed on big game, other game species, or threatened and endan- gered species. Yet, the act lists five 'Paper presented at symposium. Man- agement of Amphibians, Reptiles, and Small Mammals in Nortt) America. (Flag- staff, AZ, July 19-21 1988). ^Robert C. Szaro is Researct) Wildlife Bi- ologist, USDA Forest Service, Rocky Moun- tain Forest and Range Experiment Station, at thie Station's Research) Work Unit in Tempo, in cooperation with Arizona State University. Station Headquarters is in Fort Collins, in cooperation with Colorado State University. categories of management indicator species: (1) endangered and threat- ened plants and animals; (2) species with special habitat needs; (3) species commonly hunted, fished, or trapped; (4) nongame species of spe- cial interest; and (5) plant and animal species selected because their popu- lation changes are believed to indi- cate the effects of management activi- ties on other species of selected ma- jor biological communities or on wa- ter quality. Nongame birds have been the first group to benefit from changing man- agement practices and public con- cern. The management of nongame birds within the National Forest Sys- tem received a big boost from the "Symposium on Management of For- est and Range Habitats for Nongame Birds" held in Tucson in May 1975 (Smith 1975). Since that initial sym- posium, four regional workshops were held emphasizing the manage- ment of nongame birds in forest and range habitats (Degraaf 1978a, 1978b; Degraaf and Evans 1979; Degraaf and Tilghman 1980). There have also been Forest Service sponsored sym- posia targeting specific bird groups such as owls (Nero et al. 1987) and birds using specific habitat features such as snags (Davis et al. 1983). Only recently has the management of other nongame species gained in- creased recognition. The landmark symposium on "Herpetological Communities" held in Lawrence, Kansas, August 1977, as part of the joint meeting of the Herpetologists' League and the Society for the Study of Amphibians and Reptiles, was the first attempt to organize a vehicle for the incorporation of papers dealing with herpetological communities (Scott 1982). Yet, as Gibbons (this volume) clearly shows, little progress has been made in the recognition of amphibians, reptiles, and small mammals as being important focal points for research and management efforts. It is encouraging that recent comprehensive symposia have incor- porated papers dealing with these groups. There was an entire session on Amphibians and Reptiles in the symposium "Riparian Ecosystems and Their Management" (Johnson et al. 1985), and almost 30% of the Southern Evaluation Project Work- shop reports work on amphibians, reptiles, and small mammals (Pear- son et al. 1987). The intent of this symposium was to bring scientists and managers to- gether to exchange knowledge and ideas on habitat requirements, man- agement needs, and other informa- tion on these often overlooked com- ponents of North American fauna. Another purpose was to summarize the state-of-the-science of habitats and habitat requirements of species within these groups. Of particular interest were papers emphasizing habitat models, habitat requirements, sampling techniques and problems, community dynamics, and manage- ment recommendations. 1 The overwhelming response to our announcement for papers was unexpected. More than 60 abstracts were originally submitted for presen- tation. In order to overcome recent criticism concerning so-called "gray" literature (Bart and Anderson 1981, Capen 1982, Finch et al. 1982, Scott and Ralph 1988), we made every ef- fort to improve the quality of the symposium and its subsequent pro- ceedings. All authors were required to submit their first drafts 5 months prior to the meeting in order to en- sure adequate time for peer review and editing. Each manuscript was reviewed by two experts familiar with the topic, and edited for style and content by one of the sympo- sium editors. We found the meeting itself to be a fertile exchange of ideas and tech- niques between managers and re- searchers from all over the country. Those attending found the meeting extremely enlightening both for re- searchers and managers because of their exposure to new viewpoints. It is a testament to those attending and the quality of the presentations that very little discussion occurred out- side the meeting hall when papers were in progress. Virtually all partici- pants were present throughout the symposium, from the first session to the last. We hope this symposium will prove to be the boost that these fau- nal groups need to get increased re- search and management recognition. For only with an adequate data base can models be developed that predict diversity in relation to natural or man-made disturbance of ecosys- tems. These holistic models are of the utmost importance for the mainte- nance of worldwide biodiversity (Wilson and Peters 1988). Ecosystem diversity is a key correlate with bio- logical productivity and has recently attracted considerable interest both from theoreticians and from profes- sionals concerned with management of land and water systems (Suffling et al. 1988). We feel that amphibians. reptiles, and small mammal popula- tions may prove to be the ultimate indicators of habitat quality and health, because of their sedentary characteristics which make them much more susceptible to manage- ment activities than do highly mobile bird species and ubiquitous species such as deer and turkey. Literature Cited Bart,], and D. R. Anderson. 1981. The case against publishing sym- posia proceedings. Wildlife Soci- ety Bulletin 9:201-202. Capen, David E. 1982. Publishing symposia proceedings: another viewpoint. Wildlife Society Bulle- tin 10:183-184. Davis, Jerry W., Gregory A. Good- win, and Richard A. Ockenfeis (Technical Coordinators). 1983. Snag habitat management: Pro- ceedings of the symposium. USDA Forest Service General Technical Report RM-99. Rocky Mountain Forest and Range Ex- periment Station, Ft. Collins, Colo. 226 p. Degraaf, Richard M. (Technical Coor- dinator). 1978a. Proceedings of the workshop on nongame bird habi- tat management in the coniferous forests of the western United States. USDA Forest Service Gen- eral Technical Report PNW-64. Pacific Northwest Forest and Range Experiment Station, Port- land, Oregon. 100 p. Degraaf, Richard M. (Technical Coor- dinator). 1978b. Proceedings of the workshop: Management of south- ern forests for nongame birds. USDA Forest Service General Technical Report SE-14. Southeast- ern Forest Experiment Station, Asheville, North Carolina. 176 p. Degraaf, Richard M. and Keith E. Evans (Proceedings Compilers). 1979. Management of north central and northeastern forests for nongame birds. USDA Forest Service General Technical Report NC-51. North Central Forest Ex- periment Station, St. Paul, Minn. 268 p. Degraaf, Richard M. and Nancy G. Tilghman (Proceedings Compil- ers). 1980. Workshop proceedings: Management of western forests and grasslands for nongame birds. USDA Forest Service General Technical Report INT-86. Inter- mountain Forest and Range Ex- periment Station, Ogden, Utah. 535 p. Finch, Deborah M., A. Lauren Ward, and Robert H. Hamre. 1982. Com- ments in defense of symposium proceedings: response to Bart and Anderson. Wildlife Society Bulle- tin 10:181-183. Johnson, R. Roy, Charles D. Ziebel, David R. Patton, Peter F. Ffolliott, and Robert H. Hamre (Technical Coordinators). 1985. Riparian eco- systems and their management: reconciling conflicting uses. First North American Riparian Confer- ence. USDA Forest Service Gen- eral Technical Report RM-120. Rocky Mountain Forest and Range Experiment Station, Ft. Collins, Colo. 523 p. Nero, Robert W., Richard J. Clark, Richard J. Knapton, and R. H. Hamre (Editors). 1987. Biology and conservation of northern for- est owls. USDA Forest Service General Technical Report RM-142. Rocky Mountain Forest and Range Experiment Station, Ft. Collins, Colo. 309 p. Pearson, Henry A., Fred E. Smeins, and Ronald E. Thill (Proceedings Compilers). 1987. Ecological, physical, and socioeconomic rela- tionships within southern national forests: Proceedings of the south- ern evaluation workshop. USDA Forest Service General Technical Report SO-68. Southern Forest Ex- periment Station, New Orleans, Louisiana. 293 p. Scott, J. Michael and C. John Ralph. 1988. Quality control of symposia and their published proceedings. Wildlife Society Bulletin 16:68-74. 2 Scott, Norman J., Jr. 1982. Herpeto- logical communities. USDI Fish and Wildlife Service, Wildlife Re- search Report 13. 239 p. Smith, Dixie R. (Technical Coordina- tor). 1975. Proceedings of the sym- posium on management of forest and range habitats for nongame birds. USDA Forest Service Gen- eral Technical Report WO-1. Washington, D.C. 343 p. Suffling, Roger, Catherine Lihou, and Yvette Morand. 1988. Control of landscape diversity by cata- strophic disturbance: a theory and a case study in a Canadian Boreal Forest. Environmental Manage- ment 12:73-78. Wilson, E. O. (Editor) and Frances M. Peter (Associate Editor). 1988. Bio- diversity. National Academy Press, Washington, D.C. 521 p. ■-. ;.. 3 The Management of Amphibians, Reptiles and Small Mammals in North America: The Need for an Environmental Attitude Adjustment^ Abstract.— Amphibians, reptiles, and smail mammals need special consideration in environmental management and conservation because (1) they are significant biotic components in terrestrial and freshwater habitats; (2) research and management efforts have lagged behind those on other vertebrates; (3) a stronger understanding of their ecology and life history is needed to guide management decisions; and (4) their importance has not been promoted satisfactorily to develop the proper public attitude. J. Whitfield Gibbons^ My objective is to provide an over- view and perspective of the amphibi- ans, reptiles, and small mammals of North America as a group that de- serves more careful consideration from an environmental management and conservation standpoint. The justification of the need for and time- liness of a careful examination of am- phibian, reptile, and small mammal assemblages is based on the premises stated below. One intent is to bring the problem into focus so that both scientists and managers can identify problem areas and conjoin in an ef- fort that will result in the manage- ment of these animals in North America in a prudent and far-sighted manner. I offer four premises to support the contention that amphibians, rep- tiles, and small mammals deserve special attention with regard to man- agement considerations: 1. Amphibians, rephles, and small mammals are a signifi- cant and important wildlife component of the fauna in most terrestrial and freshwa- ter habitats in North Amer- ica. 'Paper presented at symposium. Man- agement of Amphibians, Reptiles, and Small Mammals in North America. (Flag- staff, AZ, July 19-21,1988). 'J. Whitfield Gibbons, Head, Division of Stress and Wildlife Ecology, Savannah River Ecology Laboratory, Drawer E, Aiken, SC 29801. 2. Research and management publication efforts as well as funding have lagged behind those of many of the more obvious faunal components (e.g., game species of large mammals, birds and fishes, and many insects, because of their importance as pests). 3. The direct empirical meas- urements of habitat require- ments, species interactions, and life history patterns needed for proper manage- ment are often lacking for amphibians, reptiles, and small mammals. 4. An attitude that amphibians, reptiles, and small mammals should be of concern in envi- ronmental management deci- sions has not been satisfacto- rily instilled among some managers, the general public, and political officials. Support for Premises Premise 1 —Amphibians, reptiles, and small mammals are a significant and important wildlife component in North American ecosystems. One way for a taxonomic group or species assemblage to qualify as im- portant to an environmental manager is to be identified as making a major contribution to biological complexity in terms of species diversity, trophic dynamics, and interactions within communities. Some groups clearly have the potential for overall com- munity influence by virtue of abun- dance. Salamanders at Hubbard Brook were demonstrated to have a higher biomass than other vertebrate groups (Burton and Likens 1975). The capture of as many as 88,000 amphibians in one year (SREL Report 1980) and large numbers in most years (Pechmann et al. 1988) at a 1 ha temporary pond in South Caro- lina suggest that they dominate the higher trophic level in some habitats. Other studies support the postula- tion that amphibians are often the top predators in some aquatic sys- tems (Taylor et al. in press). Freshwa- ter turtles represent the majority of vertebrate biomass in many aquatic habitats (Congdon, Greene, and Gib- bons 1986), and their potential sig- nificance as vectors for seeds and parasites among temporary aquatic habitats has been suggested (Cong- don and Gibbons 1988). Box turtles (Terrapene Carolina) have also been implicated as seed vectors (Braun and Brooks 1987). Small rodents are noted for their impact on plant com- munities under certain environ- mental conditions (Hayward and Phillipson 1979); desert granivores affect the density, biomass, and com- position of annual plants (Brown et 4 Table 1 .—Publications on different taxonomic groups in major North American journals in general ecology and wildlife ecology. Issues from 1983-1988 were selected at random until 200 titles were chosen. Assignment to taxonomic categories was based on the appearance of study organism names in the titles. Not all papers used in tabulation were based on North American fauna. The definition of small mammals is that used in this Symposium. JOURNAL (total) A R S L F B 1 (218) AMN 9 12 43 14 43 18 61 (201) ECOL 10 22 28 10 24 50 58 (213) CJZ 8 9 36 39 34 53 34 (614) Total 27 42 107 63 101 121 153 % 4 7 17 10 16 20 25 (139) Hsr 1 4 4 13 50 67 (204) JWM 0 2 6 103 93 (343) Total 1 6 110 126 50 160 % <1 2 3 37 15 47 A= Amphibians R = Reptiles S = Small mammals L = Large mammals F = Fishes B = Birds I = Insects General Ecology AMN = American Midland Naturalist ECOL = Ecology CJZ = Canadian Journal of Zoology Wildlife Ecology HSI = U.S. Fish and Wildlife Service Habitat Suitability Index Models JWM = Journal of Wildlife Management *Only 139 fifles were available. al. 1986). These represent only a few of the available examples for am- phibians, reptiles, and small mam- mals; however, many more studies are needed that document the role and importance of species in these groups in enhancing biological com- plexity. Another way for a group to as- sume importance is for it to have a direct, measurable economic value or impact. Several examples can be given of the importance of amphibi- ans, reptiles, and small mammals from the economic perspective, but their impact has been trivial in com- parison to large game mammals or insect pests, and controls and regula- tions have been comparatively loose. The limited economic importance of most small terrestrial or semi-aquatic vertebrates is presumably one expla- nation for their being given minimal attention in many management schemes. A few species such as American alligators (Joanen and McNease 1987), bullfrogs (Shifter 1987), and snapping turtles (Bushey, no date) are commercially important as human food items. Other species assume an economic value in the le- gal pet trade (Conant 1975) or as re- search animals sold by biological supply houses (Carolina Biological Supply 1987). Some venomous snakes, especially eastern (Crotalus adamanteus) and western (C. atrox) diamondback rattlesnakes, are an economic irony in that the venom is necessary to make antivenin (Parrish 1980). Of course, such species achieve some level of importance simply by being potentially injurious. Small mammals have been indicted in a variety of situations for negative economic impacts, such as prairie dog damage (Walker 1983), rabies in bats (Constantine 1970), and grain- eating by rodents (Rowe 1981). Another measure of importance of some species or groups is the intan- gible aesthetic value that some people place on them. Many species assume an undeniable importance to many people and may ultimately ac- quire protected status. Legal protec- tion of "the species" often provides protection to certain habitats. This circle of protection is a factor that can work to great advantage for those persons interested in preservation — the species is protected because it is important (aesthetic) and becomes even more important (legal) because it is protected and results in preser- vation of the habitat. For example, the legal status offered the desert tor- toise {Xerobates agassizi; Luckenbach 1982) and the Morro Bay kangaroo rat (Dipodomys heermanni morroensis; USDI 1980) in California or the American crocodile (Crocodylus acutus; Kushlan and Mazzotti 1986) in Rorida serves to provide some level of environmental protection for the entire community where they oc- cur. The protection given the black footed ferret has resulted in protec- tion of its prey. The World Wildlife Fund recognizes this effect in its con- servation programs by designating "flagship" species such as great apes or monkeys, for which funds are more easily raised, in order to pro- tect entire communities or ecosys- tems. Premise 2— Ecological research on herpetofouna and small mammals has lagged behind that of other animal groups. Support for the contention that the level of ecological research on am- phibians, reptiles, and small mam- mals is lower than that of certain other animal groups can be given in several ways. These include annual publications on particular groups (table 1) and the proportion of funded grants that fall into each cate- gory (table 2). 5 The reasons for the lower levels of publication and funding in research on amphibians, reptiles, and small mammals are varied and in part con- jectural. One seemingly obvious rea- son is that most species in these groups have low profile in health, hunting, agricultural, or other eco- nomic issues and therefore receive minimal attention from some quar- ters. The comparatively low level of attention given to small, non-game terrestrial and semi-aquatic verte- brates by certain sectors of society is reflected in lower overall funding and subsequently in fewer general publications. Research funding is inequitable because of the emphasis on species that have important economic status; thus, the life history and ecology of even moderately abundant herpe- tofaunal or small mammal species are seldom understood at a level that would permit prudent management. Even those with potential economic importance receive less emphasis than many birds, large mammals, and fish. As an example, the Ameri- can alligator represents a reptile spe- cies of vital concern from a manage- ment standpoint, yet the number of publications that focus on the life his- tory, ecology, behavior, and genetics of the sp)ecies is limited (see Brisbin et al. 1985) compared to the hun- dreds on large mammal game species such as white-tailed deer (Halls 1984; Johns and Smith 1985). Premise 3— The basic ecological and life history information necessary to make thoughtful environmental management decisions is often absent for many of the amphibians, reptiles, and small mammals in a community. As indicated above, the research ef- fort directed toward amphibians, reptiles, and small mammals by ecologists appears to be below that for other vertebrate groups. Al- though difficult to measure, it would also be expected that the fundamen- tal data bases necessary for thought- ful management decisions would ex- ist in lower proportions for herpe- tofaunal and small mammal species. One reason is that, compared to many large mammals, birds, and Table 2.— Number of grant proposals funded by selected U.S. granting agen- cies on particular groups of animals. A R S L F B 1 NSF(1987) 1 3 4 2 11 7 24 Sigma Xi 4 8 9 7 9 24 10 (March 1987) National Geograpl-iic 0 3 2 28 8 15 14 (1988) World Wildlife Fund 0 15 0 49 0 14 1 (1987-1988) Total 5 29 15 86 28 60 49 % 2 11 6 32 10 22 18 A=Amphibians R=Reptiles S=Small mammals L=Large mammals F=Fishes B=Birds l=lnsects fishes, certain aspects of field studies on many of the amphibians, reptiles, and small mammals are sometimes perceived as being more difficult be- cause of factors such as small body or population sizes, fossorial or cryp- tic habits, patchy distribution, and unpredictable seasonality. Conse- quently, fewer papers are likely to be published in general ecology journals that expect quantitative ecological and life history research results rather than ones that are descriptive and qualitative. An exception to this may be manipulative field experi- ments in which small rodents have been used in almost half of the stud- ies involving vertebrates. The actual or apparent rarity or unpredictability of occurrence of many amphibian, reptile, and small mammal species makes it difficult or impossible for the research ecologist to gather useful data without a fund- ing base that is accepting of the un- certainty of whether data will actu- ally be forthcoming in a particular year. The environmental manager in turn cannot incorporate such species into a management plan, and thus their perceived imp>ortance is dimin- ished. The unpredictability of occur- rence of some species can be demon- strated with amphibians and reptiles on the Savannah River Plant (SRP) in South Carolina. In spite of more than a quarter of a century of field studies and the capture of more than half a million reptiles and amphibians across all available habitats, species previously unreported from the SRP continue to be discovered (Gibbons and Semlitsch 1988; Young 1988). Or, some species have gone for intervals as long as one decade (e.g., pickerel frog. Ram palustris) or two (e.g., glossy water snake, Regina rigida) be- tween sightings (Gibbons and Sem- litsch 1988). Clearly, developing a basic ecological field study on such species in a region is not feasible un- der typical funding situations. Resolutions to the problem of gar- nering information about rare sp>ecies include intensifying survey efforts in 6 geographic regions of interest by supporting long-term research pro- grams that can ultimately reveal the presence of rare or fossorial species. Once a species is identified to be present in a habitat, the decision should be made on whether an eco- logical research effort is warranted. Long-term studies may be neces- sary to reveal certain life history traits, even about common species, because of the inherent variability in some life history features that can result from natural environmental variation (Semhtsch et al. 1988). Such studies may be essential to identify the extent of variability due to an- nual weather patterns and climatic variation (Semlitsch 1985; Pechmann et al. 1988). Long-term research pro- grams may be needed because some species are long- lived, or in the case of many, because the potential lon- gevity is great but unknown (Gib- bons 1987). For many species that have eco- nomic value (e.g., snapping turtle, Chelydra serpentina; Congdon et al. 1987), the impact of harvesting has not been properly assessed. Because of the limited baseline ecological and life history data for most species, a priority goal should be the establish- ment of a moratorium on the whole- sale removal of all native species of amphibians, reptiles, and small mammals until it can be verified that regional populations can sustain the removal rate. State permits should be required of, and possession limits should be set for, all commercial col- lectors for all species of amphibians, reptiles, and small mammals. Today's emphasis should be on protection of each species until con- vincing evidence is supplied that har- vesting has no long-term impact, rather than placing the burden on herpetologists and mammalogists to demonstrate population irrecovera- bility before harvesting is discontin- ued. The negative consequences of the latter, and current, approach (i.e., demonstrating the impact of removal while harvesting is in progress) is that some populations will be re- duced to the point of no recovery be- fore the necessary evidence can be collected. Each species should be protected until proven harvestable. The appropriate basic research should be conducted by scientists with no economic or emotional in- vestment in the outcome. Research support should be provided by state or federal agencies and by special interest groups that have no influ- ence over the final management deci- sions. The ideal approach is that sci- entists would gather the facts and that environmental managers would interpret them in the context of har- vesting quotas. The development and use of predation (Holling 1966) or harvest (Ricker 1975) models may be effective approaches for address- ing the issue of human predation (i.e., harvestability by man). One area that deserves attention in strengthening the study of small ter- restrial or semi-aquatic vertebrates is the use of innovative techniques to address physiological, ecological, and behavioral questions under natu- ral conditions. Non-destructive field sampling techniques are critical in the study of both rare and endan- gered species but are also important for preserving the integrity of any study population. These include techniques for capture, field identifi- cation of individuals, non-disruptive handling or observation, recapture, and the acquisition of non- destruc- tive physiological, genetic, behav- ioral, and life history data. Some ex- amples include radiography (Gib- bons and Greene 1978) or sonogra- phy for determination of clutch sizes, blood sampling for genetic and hormonal analyses (Scribner et al. 1986), and cyclopropane for measur- ing lipid levels (Peterson 1988). A broader use of such techniques in field studies could strengthen the foundation of ecological and life his- tory understanding that is necessary for environmental management. A direct contribution to environ- mental managers could be achieved by attempts to verify the several am- phibian, reptile, and small mammal Habitat Suitability Index models of the U.S. Fish and Wildlife Depart- ment. The concept has the potential value of providing an initial quanti- tative approach that gives a tangible product. However, to be of greatest value, the HSI models must be evalu- ated and modified as appropriate. It is perhaps noteworthy that the HSI models prepared for amphibians (1), reptiles (4), and small mammals (4) collectively represent only 6% of the 139 that have been completed on ver- tebrates (table 1). For these to be- come an effective tool in manage- ment of herpetofauna and small mammals, more herpetologists and mammalogists need to volunteer to develop HSI models for these groups. A distinction must be made be- tween (1) problem oriented applied research on specific systems that re- lies on qualitative assessments or in- direct measurements of variables with minimal inference power and (2) basic research that is founded on quantitative or direct measurements of variables, has a conceptual or theoretical base or orientation, and can be strongly inferential through general field or laboratory experi- ments. The latter approach will be necessary if environmental managers are to have a reliable data base that is founded on broad applicability, lev- els of predictability, and clear direc- tions for future research. Premise 4— The attitude of most people in North America toward most amphibians, reptiles, and small mammals is either negative or neutral, in part because efforts to develop an attitude change have been insufficient or ineffective. Although documentation is difficult, it would appear that in North Amer- ica we are far from a suitable accep- tance level toward these groups of 7 organisms. People still try to run over snakes on highways, have little awareness that many conspicuous predators rely on small mammals for their basic diets, and give no thought to how many small vertebrates will be eliminated by the draining of a swamp or damming of a stream. I think the situation is an embarrassing one for the scientists and general public of a nation that espouses edu- cation and knowledge. Evidence that a more positive atti- tude and less environmental leniency has developed over the last several years is the recent federal listings of snakes (e.g., indigo snake, Dry- marchon corias; San Francisco garter snake, Thamnophis sirtalis tetrataenia) and small rodents (e.g., Utah prairie dog, Cynomys parvidens; salt marsh harvest mouse, Reithrodontomys raviventris; Key Largo cotton mouse, Peromyscus gossypinus allapaticola) as protected species. However, many of the listings involving amphibians, reptiles, and small mammals have been hard fought ones against public and political opinions that such spe- cies hardly deserve such concessions. The failed efforts at protection far outnumber the successful ones. The attitude that these animals are unim- portant is pervasive throughout the general public, politicians, and even some environmental managers. The basic responsibility for eliminating ignorance and effecting the proper environmental attitude adjustment must start with the scientist. It is my firm opinion that many scientists have lost sight of who their patrons are (for most of us, the U.S. taxpayers) and of their responsibility to communicate findings to all levels of society. This communication proc- ess entails a level of cooperation and an educational spirit that allows each individual to contribute in the most effective manner. However, we must all accept and work toward the com- mon goals of establishing a thorough and general foundation of ecological information for amphibians, reptiles, and small mammals and of being generous in the distribution of the findings in a form palatable to and usable by the intended audience. Conclusions An environmental attitude adjust- ment model must be developed and promoted that considers where we want to end up, who we must edu- cate and influence, and what we must know and do to achieve the goal of education in a convincing manner. The desired end point is a nationwide attitude among scientists, managers, politicians, and the public that amphibians, reptiles, and small mammals are critical wildlife compo- nents. Each species population and community must be identified as having an intrinsic value in maintain- ing the integrity of the natural eco- systems of North America. Scientists have a responsibility for collecting extensive and intensive in- formation on the life history patterns and habitat requirements of native amphibians, reptiles, and small mammals. The required data must be collected in a rigorous experimental manner that promotes an under- standing of these species and com- munities through strong inferences and syntheses. Politicians have a responsibility to assure that the approval of a govern- ment project is as contingent on envi- ronmental consequences as on budg- etary considerations. Our attitude must graduate to become one of ac- ceptance of a proposed project only after environmental impact determi- nations have led to an objective deci- sion that the gain from the project warrants the loss to the environment. Managers have a responsibility for promoting basic research, for apply- ing the findings to habitat manage- ment, and for having the patience to wait for the completion of long-term studies as required. In situations where removal of animals or elimina- tion of habitat is an issue, the burden of proof should be borne by the har- vester or developer, and not by the scientist or manager. The status of a species should be determined before the decision to proceed is made, cer- tainly not after harvesting begins or during the physical development of a project. This assessment should be made and evaluated before the proj- ect is approved. Each species should be protected until proven harves- table. Both scientists and managers have a responsibility to inform the public and political arena that the protec- tion and ecological understanding of inconspicuous and non-game species are vital to proper ecosystem man- agement and to the preservation and maintenance of North America's natural heritage. Acknowledgments I thank Justin D. Congdon, Nat B. Frazer, Trip Lamb, William D. McCort, Joseph H. K. Pechmann, David E. Scott and Raymond D. Semlitsch for commenting on the original manuscript. I appreciate the efforts of Marianne Reneau, Marie Fulmer, Jeff Lovich, Tony Mills, and Tim Owens in manuscript prepara- tion. Manuscript preparation was aided by Contract DE- AC09- 76SR00819 between the U.S. Depart- ment of Energy and the University of Georgia's Savannah River Ecology Laboratory. Literature Cited Brisbin, I. Lehr, Charles A. Ross, M.C. Downes, Mark A. Staton and Brad R. Gammon. 1986. A bibliog- raphy of the American alligator (Alligator mississippiensis). Savan- nah River National Environmental Research Park Publ. Aiken, SC. Brown, J. H., D. W. Davidson, J. C. Munger, and R. S. Inouye. 1986. Experimental community ecology: The desert granivore system, pp. 41-61. In: Community Ecol- 8 ogy. J. Diamond and T. J. Case, editors. Harper and Row. NY. Brown, J. and G. R. Brooks, Jr. 1987. Box turtles (Terrapene Carolina) as potential agents for seed dispersal. Am. Midi. Nat. 117:312-318. Burton, T.M. and G.E. Likens. 1975. Salamander populations and bio- mass in the Hubbard Brook Ex- perimental Forest, New Hamp- shire. Copeia 1975:541-546. Bushley, S. Date of publication un- known. How to profit from snap- ping turtles. Reptiles Unlimited, York, Pennsylvania. Carolina Biological Supply Com- pany. 1987. Biology /Science mate- rials. Catalog 58. Burlington, NC. Conant, Roger. 1975. A field guide to reptiles and amphibians of east- ern/central North America. Houghton-Mifflin: NY. Congdon, Justin D. and J. Whitfield Gibbons. 1988. Biomass productiv- ity of turtles in freshwater wet- lands: A geographic comparison. In Freshwater Wetlands and Wild- life. Rebecca R. Sharitz and J. Whitfield Gibbons, editors. Office of Scientific and Technical Infor- mation, U.S. Department of En- ergy. Oak Ridge, TN. Congdon, Justin D., Gary L. Breiten- bach and Richard C. van Loben Sels. 1987. Reproduction and nest- ing ecology of snapping turtles (Chelydra serpentina) in southeast- ern Michigan. Herpetologica 43:39-54. Congdon, Justin D., Judith L. Greene, and J. Whitfield Gibbons. 1986. Biomass of freshwater turtles: A geographic comparison. Am. Midi. Nat. 115:165-173. Constantine, D. G. 1970. Bats in rela- tion to the health, welfare, and economy of man. pp. 320-449. In: W. A. Wimsatt, editor. Biology of bats. Vol. II. Academic Press. New York. Gibbons, J. Whitfield. 1987. Why do turtles live so long? Bioscience 37:262-269. Gibbons, J. Whitfield and Judith L. Greene. 1979. X-ray photography: A technique to determine repro- ductive patterns of freshwater turtles. Herpetologica 35:86-89. Gibbons, J. Whitfield and Raymond D. Semlitsch. 1988. Guide to the reptiles and amphibians of the Savannah River Plant. SRO-NERP 18. U.S. Department of Energy's Savannah River National Environ- mental Research Park. Aiken, SC. Halls, L. K. 1984. White-tailed deer: Ecology and management. Stack- pole Books. Harrisburg, PA. Hayward, G. F. and J. Phillipson. 1979. Community structure and functional role of small mammals in ecosystems, pp. 135-211. In: Ecology of small mammals. D. M. Stoddart, editor. John Wiley and Sons. New York. Holling, C.S. 1966. The functional re- sponse of invertebrate predators to prey density. Mem. Entomol. Soc. Can. 48:1-86. Joanen, Ted and Larry McNease. 1987. The management of alliga- tors in Louisiana, USA. p. 33-42. In: Wildlife Management. Croco- diles and Alligators. Grahame J.W. Webb, S. Charlie Manolis and Pe- ter J. Whitehead, editors. Surrey Beatty & Sons Pty. Limited. Chip- ping Norton, NSW, Australia. Johns, Paul E. and Michael H. Smith. 1985. Bibliography for the white- tailed deer on the Savannah River Plant. University of Georgia, Insti- tute of Ecology and Savannah River Ecology Laboratory. Aiken, SC. Kushlan, James A. and Frank J. Maz- zotti. 1986. Population biology and status of the American crocodile in South Florida, pp. 188-194. In: Proceedings of the 7th working meeting of the crocodile specialist group. lUCN. Caracas, Venezuela. Luckenbach, R. A. 1982. Ecology and management of the desert tortoise (Gopherus agassizii) in California, pp. 1-37. In: R. B. Bury (ed.). North American tortoises: Conservation and Ecology. USDI, Fish and Wildlife Service. Wildlife Resource Department. 12. Parrish, Henry M. 1980. Poisonous snakebites in the United States. Vantage Press, Inc. New York. Pechmann, Joseph H.K., David E. Scott, J. Whitfield Gibbons and Raymond D. Semlitsch. 1988. In- fluence of wetland hydroperiod on diversity and abundance of meta- morphosing juvenile amphibians. Wetlands Ecology and Manage- ment 1:1-9. Peterson, C. 1988. Presentation at Desert Tortoise Council Meeting, Laughlin, NV. Ricker, N.E. 1975. Computation and interpretation of biological statis- tics of fish populations. Bull. Fish. Res. Board Can. 191:1-382. Rowe, P.P. 1981. Wild house mouse biology and control. Symposium of the Zoological Society of Lon- don. 47:575-589. Scribner, Kim T., Joseph E. Evans, Stephen J. Morreale, Michael H. Smith and J. Whitfield Gibbons. 1986. Genetic divergence among populations of the yellow-bellied slider turtle (Pseudemys scripta) separated by aquatic and terres- trial habitats. Copeia 1986:691-700. Semlitsch, Raymond D. 1985. Analy- sis of climatic factors influencing migrations of the salamander Am- bystoma talpoideum. Copeia 1985:477-489. Semlitsch, Raymond D., David E. Scott and Joseph H.K. Pechmann. 1988. Time and size at metamor- phosis related to adult fitness in Ambystoma talpoideum. Ecology 69:184-192. Shiffer, Clark N. 1987. The bullfrog. Pennsylvania Angler. August 1987. SREL Report. 1980. Savannah River Ecology Laboratory' Annual Re- port, FY-1980. A biological inven- tory of the proposed site of the Defense Waste Processing Facility on the Savannah River Plant in Aiken, South Carolina. NTIS Publ. SREL-7UC-66e. Aiken, SC. Taylor, Barbara E., Ruth A. Estes, Jo- seph H. K. Pechmann and Ray- mond D. Semlitsch. Trophic rela- 9 tions in a temporary pond: Larval salamanders and their microin- vertebrate prey. Can. J. Zool. In Press. USDI (United States Department of the Interior). 1980. Republication of the lists of endangered and threatened species and correction of technical errors in final rules. Federal Register. 45:33768-33781. Walker, E. P. 1983. Mammals of the world. Vol. I. John Hopkins Uni- versity Press, Baltimore. 568 p. Young, David P. 1988. Rhadinaea flav- ilata (Pine woods snake). Herp. Rev. 19:20. Douglas-fir Forests in tlie Oregon and Washington Cascades: Relation of the Herpetofauna to Stand Age and Moisture^ R. Bruce Bury^ and Paul Stephen Corn^ Abstract.— Pitfall traps effectively sampled amphibians but not reptiles in Douglas-fir (Pseudofsugo menz/es/O forests. The abundance of only one amphibian species varied across an age gradient or a moisture gradient. Salamanders and frogs that breed in ponds or streams vjere captured in large numbers in some stands, likely due to the presence of nearby breeding habitat rather than forest conditions. Lizards occurred mostly in dry stands and clearcuts. Time-constrained searches showed different use of dov^ned v\/oody debris among terrestrial salamanders. The occurrence and abundance of species in naturally regenerated forests markedly differed from clearcut stands. The value of old-growth forests for wildlife is highly debated (Fosburg 1986, Harmon et al. 1986, Harris 1984, Kerrick et al. 1984, Ruggiero and Carey 1984, Salwasser 1987, Wilcove 1987). Most attention has been directed toward the spotted owl (Strix occidentalis), which is one of several hundred vertebrate species occurring in the Pacific Northwest (Bruce et al. 1985). Franklin and Spies (1984) distinguished old-growth for- ests of Douglas-fir (Pseudotsuga men- ziesii) as having a wide range of tree sizes and ages, a deep mulhlayered crown canopy, large individual trees, and accumulations of coarse woody debris (CWD), including snags and downed logs of large dimension. They reported that these forests are productive, diverse ecosystems, and highly specialized habitats. We need to evaluate sampling techniques continually to better de- scribe, understand and predict the species richness, abundance and bio- mass of herpetological assemblages. However, few herpetological com- munities or their habitats have been ^ Roper presented at symposium, Mon- agement of AmphibioDS. Reptiles, and Smoll Mammals in North America (Flagstaff, AZ.July 17-21, 1988). Bruce Bury is Zoologist (Research)), USDA Fishi and Wildlife Sen/ice, National Ecology Research Center, 1300 Blue Spruce Drive, Fort Collins, CO 80524. ^Paul Stephen Corn is Zoologist, USDA Fish and Wildlife Service, National Ecology Research Center, 1300 Blue Spruce Drive, Fort Collins. CO 80524. sampled using more than one quanti- tative technique. Recently, field techniques for the study of herpetological communities have improved (Scott 1982). Some of the most promising methods employ pitfall traps and drift fences to cap- ture amphibians and reptiles. Several promising pitfall designs have been developed for varied habitats in Aus- tralia (Friend 1984, Webb 1985) and in North America (Bennett et al. 1980, Bury and Corn 1987, Bury and Raphael 1983, Campbell and Christ- man 1982, Enge and Marion 1986, Gibbons and Semlitsch 1981, Jones 1981, 1986, Raphael 1984, Raphael and Rosenberg 1983, Rosenberg and Raphael 1986, Vogt and Hine 1982). Pitfall traps are effective for capture of commmon terrestrial species and they are particularly valuable in sam- pling secretive or rare forms. Searches by hand (either based on specific areas or time of collecting) or observation are used to sample her- petofaunas (see reviews by Bury and Raphael 1983, Jones 1986, Rough et al. 1987). Campbell and Christman (1982) suggested that time-con- strained collecting (searching within a specific period of time by trained collectors) can sample terrestrial spe- cies that are under-sampled or not taken in pitfall traps. The first year of our old-growth study (1983) was partly devoted to refining field techniques. A compari- son of different pitfall designs is re- ported elsewhere (Bury and Corn 1987). Here, we employ a standard- ized pitfall array and time-con- strained searches to determine the occurrence and abundance of the ter- restrial (upland) herpetofauna in the Cascade Mountains of the Pacific Northwest. The current work on small mam- mals (Anthony et al. 1987, Com et al. 1988, West 1985), birds (Carey 1988, Manuwal and Huff 1987), and bats (Thomas in press) are part of an inter- disciplinary effort to better under- stand the relationship of nongame wildlife in old-growth forest stands (Ruggiero and Carey 1984). Our study is the first to attempt to iden- tify which species of the herpe- tofauna, if any, are associated with age and moisture gradients in forests of the Cascade Mountains. Our specific objectives were (1) to compare effectiveness and relative merits of time-constrained collecting versus pitfall trapping, (2) to com- pare the species richness and relative abundance of amphibians and rep- tiles between different forest stands, and (3) to examine the association of the herpetofauna with old-growth forest conditions. DESCRIPTION AND CLASSIFICATION OF STUDY SITES We sampled 30 sites: 18 in or near the H. J. Andrews Experimental For- est in eastern Linn and Lane coun- ties, Oregon, and 12 stands in the 11 Figure 1 .—Conducting time-constrained searches in an old-growtti stand, Oregon. Note large amounts of downed woody debris. Wind River Experimental Forest, Skamania County, Washington. All sites are on the western slopes of the Cascade Mountains. Specific loca- tions, stand classification, elevations and other details are provided in Corn et al. (this volume). Study sites represent a range of forest development across a chronosequence (principally age) and, for old-growth, a moisture gra- dient. These stands were independ- ently selected and assessed by Spies et al. {in press). They were all in natu- rally regenerated forest caused by wildfire. There were three develop- ment stages in moderate moisture conditions: young (30-76 years old), mature (105-150 years) and old- growth (195-450 years). Clearcut sites represent recent timber harvest (<10 years old). For old-growth stands only, there were representative mois- ture conditions: wet, moderate and dry sites. Stand classification was based on age determined by incre- ment boring of trees or other meth- ods, characteristic plant species in the understory, physiography, and soils. These methods and other para- meters are described by Corn et al. (this volume). Franklin et al. (1981) and Spies et al. {in press). Following the initial stand selec- tion, there were minor adjustments in assignment of stand classification (Corn et al., this volume). We re- jected a few sites that were either not continually accessible for our weekly checking of pitfall traps or were being actively logged. MATERIAL AND METHODS Time-Constrained Searches (TCS) Details of this technique are pro- vided elsewhere (Campbell and Christman 1982, Bury and Raphael 1983, Raphael and Rosenberg 1983). A team of 3-8 people intensively searched each stand for 8 person-hrs in the spring (8-25 April 1983 in Ore- gon and 3-12 May 1983 in Washing- ton). We turned over moveable sur- face objects (twigs to logs <1 m dia- mater), dug into decayed wood, and removed bark from downed wood or the bases of standing snags by hand or with potato rakes (fig. 1). Collectors remained within boundaries of habitat typical of the stand, avoiding conspicuous special- ized habitats such as ponds, creeks or rock outcrops. Further, we searched 4 sites in each state again during warm weather (July- Aug 1983). These surveys were performed for 4 hrs per plot. We recorded infor- mation on exact position of capture for each animal, including vertical position (e.g., on or under litter; on, under or in log; etc.), identification of cover object, length and diameter of object, time of capture, total length, and mass of animal. We determined the decay class of coarse woody debris occupied by animals on the forest floor. Large woody debris or felled trees (logs) occur in five progressive broad decay classes (Bartels et al. 1985, Franklin et al. 1981, Harmon et al. 1986, Maser et al. 1979, Maser and Trappe 1984): (1) intact, recently downed trees; (2) logs with loose bark; (3) loss of bark and stem partly rotted; (4) invasion of roots and deep decomposition of stem; and (5) hummocks of wood chunks and organic material. Once fallen, a large tree might require 200 or more years to progress from class 1 to 5 (Spies et al. in press), providing habitat for many generations of resi- dent wildlife. Pitfall Arrays We installed a pitfall array at each site in Oregon and Washington (de- tails in Bury and Corn 1987). Each array had two triads with their cen- ters 25 m apart. Each triad was com- posed of three drift fences 5 m long and 0.5 m tall; about 0.3 m of fence was above ground. Fences radiated at 120° angles, beginning 3 m from the center point. The compass direc- tions of the arms depended on open- ings between trees or large logs on the forest floor. Pitfall traps were constructed from two stacked #10 tin cans (3.2 1 volume) connected with 12 Table 1 .—Numbers of amphibians and reptiles captured during time-constrained searches (TCS) corKiucted 8-25 April 1983 at the H. J. Andrews Experimental Forest In Oregon. Old-growth stands are arranged in order of increasing dryness. Old growth Wet Moderate Dry Mature Young Clearcut Species Stand No. 15 03 24 «02 17 33 25 29 11 35 42 39 47 48 75 55 291 391 Clouded Salamander 3 8 6 9 3 11 17 4 2 1 2 12 2 Oregon Slender Salamander 2 6 4 12 9 n 5 1 9 1 1 Oregon Ensatina 4 3 1 9 5 7 22 2 10 6 4 5 3 9 8 9 4 1 Dunn's Salamander 2 1 Rough-skinned Newt 2 1 1 Pacific Tree Frog ] 4 1 1 1 Western Skink Norhtern Alligator Lizard Western Fence Lizard °Two surveys were conducted in this stand and the results are combined here. duct tape. A pit trap was placed fl'jsh with the ground surface at each end of the fence. Funnel traps were constructed of aluminum screening, rolled into a tube 1 m long by 0.1 m diameter, with inward funnels stapled at each end of the trap. A funnel trap was placed midway on either side of the fence. No water or preservatives were added to the traps. A wooden shingle was propped over each pitfall and funnnel trap, but water entered pit- falls during heavy rains. We rou- tinely removed water from traps with scoops or a hand-operated aq- uarium siphon. We operated pitfall traps conti- nously for 180 days, from the last week of May to late November 1983. Traps were checked 1-2 times each week. Captures were usually taken to a field laboratory for identification and measurements. All retained specimens are deposited at the Na- tional Museum of Natural History. RESULTS Tinne-Constrained Searches (TCS) Yield During spring TCS, we collected 258 amphibians and 4 reptiles (table 1) at the 18 Oregon sites (1.8 animals per person-hr) and we took 78 amphibi- ans and 4 reptiles (table 2) at 12 Washington sites (0.85 per person- hr). For summer TCS, all Washington captures included only 4 lizards from one clearcut, one mature (drier as- pect) and an old-growth dry stand (0.25 animals p>er hr) whereas in Ore- gon we captured 13 salamanders (no new species) and 2 lizards from 4 sites (0.9 animals per hr). Although we report the abun- dance of herpetofauna collected by TCS (tables 1 and 2), we did not ana- lyze these results based on the age and moisture gradients because such abundance data can be biased. Habitat Use TCS provided useful information on the exact position where individuals were found (table 3). Oregon ensati- Table 2.— Numbers of amphibians and reptiles captured during TCS 3-12 May at the Wind River Experimental Forest In Washington. Old-growth stands are arranged in order of increasing dryness. Old growth Wet Moderate Dry Mature Young Clearcut Species Stand No. 14 12 21 20 31 41 42 50 60 61 70 71 Olympic Salamander 2 Oregon Ensatina 3 7 13 5 5 4 1 1 1 1 Larch Mountain Salamander 14 Western Red-backed Salamander 6 Rough-skinned Nev4 3 2 1 Red-legged Frog 1 Pacific Tree Frog 1 Rubber Boa 2 1 Common Garter Snake 1 13 Table 3.— Number of salamanders (Oregon data only) captured in different microhabitats. Percentages are In parentheses. Oregon Oregon Clouded Slender Position Ensotina Salamander Salamander On/Under Litter 3 (2.4) 0 (0) 1 (1.6) On/Under Rock 3 (2.4) 0 (0) 1 (1.6) On/Under Log 14(11.5) 8 (10.2) 6 (6.8) Inside Log 52 (42.6) 27 (34.2) 38 (62.3) Under Bark on Log 12 (9.8) 37 (46.8) 7 (11.5) Under Bark on Ground 38 (31.1) 7 (8.9) 8 (13.1) nas {Ensatim eschscholtzi; fig. 2) oc- curred more evenly and in more mi- crohabitats than did the other two species. Clouded salamanders (Aneides ferreus) were mostly under bark on logs and, secondarily, often were in logs (81% of the sites occu- pied were related to logs). The Ore- gon slender salamander (Batrachoseps wrighti) predominately occurred in logs (62%) and then under bark on ground or on logs (87% in or near logs). Most bark on the ground oc- curred in piles sloughed from fallen trees or snags and is essentially an extension of the log environment. Terrestrial salamanders that were captured in or near downed wood markedly differed in their use of dif- ferent decay classes of CWD (fig. 3). We did not include decay class 1 logs, because few of these were searched and none had salamanders. These logs are intact material and offer little cover for salamanders. We calculated Chi-square statistics for three species in Oregon. The clouded salamander was most abun- dant in younger (class 2) logs (P <0.001), while Oregon slender sala- manders were found more often than expected in the more decayed class 4 and 5 logs (P < 0.05). Numbers of Oregon ensatina generally followed the pattern of log abundance (fig. 3), except that they were found less of- ten than expected in class 3 logs (P <0.05). These results are consistent with microhabitats where the sala- manders were captured (table 3). Pitfall Trapping Total Nunnbers Pitfall arrays at 18 Oregon sites pro- vided 1,028 captures (table 4): 685 salamanders, 252 frogs, 64 lizards and 27 snakes. Pitfalls at 12 Washing- ton sites yielded 1,152 animals (table 5): 460 salamanders, 663 frogs and 29 snakes. Two Washington sites had exceptional catches: 253 tailed frogs (Ascaphus truei) at #21 Old-growth Moderate and 119 red-legged frogs (Ram aurora) at #42 Mature. HtL Alive FREQUCNCV Figure 3.— Frequency of occurrence of clouded salannanders, Oregon slender salannanders, and Oregon ensatinos occu- pying downed wood in decay classes 2-5. Density of logs in each decay class are provided. Data are from 18 sites at the H. J. Andrews Experimental Forest, Oregon. Yield Summer operation of the pitfall ar- rays added a few reptiles but the bulk of the catch was amphibians in the fall months during and after heavy seasonal rains (Bury and Corn 1987). There was a low catch of rep- tiles (Oregon, mean = 5 per site; Washington, mean = 2.4). Species richness did not differ across the chronosequence gradient (table 6, fig. 4). Moderate and dry old-growth stands had the highest mean abundance across the moisture gradient, which was caused by the capture of large numbers of several migratory species. Figure 2.— Adult ensatina (Ensatina eschscholtzi) from Douglas Co., Oregon. 14 r Table 4.— Abundance of amphibians and reptiles captured by pitfall arrays at ttie H. J. Andrews Experimental Forest in Oregon. Arrays of pitfall traps withi drift fences were operated continuously for 180 days in 1983. Old-growth stands are arrariged in order of increasing dryness. Old growth Wet Moderate Dry Mature Young Clearcut Species Stand r4o. 15 03 24 02 17 33 25 29 11 35 42 39 47 48 75 55 291 391 Northwestern Salamander Pacific Giant Salamander Clouded Salamander Oregon Slender Salamander 1 Oregon Ensatina 8 Dunn's Salamander Rough-skinned Newt 21 Tailed Frog Red-legged Frog Pacific Tree Frog 2 Western Skink Norhtern Alligator Lizard Western Fence Lizard Rubber Boa Northwestern Garter Snake 1 Common Garter Snake 1 1 3 2 28 2 1 10 3 5 18 26 3 1 22 5 4 13 1 17 26 119 3 1 27 7 4 21 1 62 46 23 3 4 1 4 28 3 11 14 5 1 2 10 16 14 20 30 12 10 1 15 13 36 5 14 16 2 6 28 30 3 2 4 1 3 3 2 5 9 8 3 1 11 Table 5.— Abundance of amphibians and reptiles captured by pitfall arrays at the Wind River Experimental Forest in Washington. Arrays of pitfall traps with drift fences were operated continuously for 180 days in 1983. Old- growth stands are arranged in order of increasing dryness. Old growth Wet Moderate Dry Mature Young Clearcut Species Stand No. 14 12 21 20 31 41 42 50 60 61 70 71 Northwestern Salamander 2 5 15 4 1 1 1 9 10 2 Pacific Giant Salamander 1 Olympic Salamander 3 1 1 Oregon Ensatina 7 35 29 18 39 14 13 3 24 25 0 1 Larch Mountain Salamander 10 Western Red-backed Salamander 19 Rough-skinned Newt 10 4 5 40 1 10 4 7 38 37 7 4 Tailed Frog 44 22 253 4 27 50 4 2 1 4 Red-legged Frog 8 1 3 15 1 19 119 40 5 23 6 Pacific Tree Frog 3 9 Northern Alligator Lizard 1 1 12 1 Northwestern Garter Snake 2 1 4 Common Garter Snake Differences in Closed-Canopy Stands For Oregon and Washington data combined, mean abundance of com- mon species (3 salamanders, 2 frogs) appeared to differ across either forest development (age) or moisture gradi- ent (fig. 5). However, except for the Oregon ensatina, none of the differ- ences were statistically significant (table 6). High numbers of individu- als at a few stands resulted in large variances in catch at stand types. Large numbers of both the rough- skinned newt (Taricha granulosa) and Northwestern salamander (Ambystoma gracile) were captured in a few stands (tables 4-5). Most of the tailed frogs taken were juveniles at one old-growth site in Washington (table 5), and these were apparently dispersing away from a nearby stream. Similarly, most (78%) of the red-legged frogs were taken at 5 sites (tables 4-5); the largest number (n = 15 119) were juveniles captured at one mature stand in Washington. The only species showing a signifi- cant difference (table 6) across the chronosequence of stands was the Oregon ensatina. Its numbers were lower in mature stands (fig. 5), per- haps related to amounts of CWD in different age classes (fig. 6). Abun- dance of Oregon ensatinas was most highly correlated with the number of decay class 4 and 5 logs per hectare (Pearson r = 0.48, n = 29, P < 0.01) and the mean diameter (d.b.h.) of large-sized canopy trees (r = 0.51, n = 29, P < 0.01). A discussion of the habitat variables used here is pro- vided in Corn et al. (1988). Mean abundance of Oregon ensatina also differed across the moisture gradient in old-growth stands with fewer present in wetter sites than drier. Paradoxically, most OGW stands have large amounts of CWD (fig. 6). Oregon ensatina may be associated with the amount of CWD, but there are other components of the habitat that may be under represented in OGW stands. Clearcut Stands We also trapped 5 clearcut sites (all <10 years old) to describe herpe- tofauna occurrence in managed stands. The relative abundance of the herpetofauna in these clearcuts markedly differed from 6 compa- rable young stands (fig. 7). Reptiles predominate in clearcuts, most likely responding to increased ambient temperature in such areas. The Pa- cific treefrog (Hyla regilla) also was most abundant in clearcuts. DISCUSSION Comparison and Improvements in Tectiniques Time-constrained searches (TCS) provided insufficent animals for quantitative analyses in most stands. The technique might be more worth- while under optimal environmental conditions (e.g., after heavy rains for amphibians) and with increased ef- fort (16+ person-hr per site). Summer searches added the occurrence of liz- ards to some stands, but in general the effort was not worth the time in- vestment in forested stands of the Cascade Mountains. However, TCS can be effective to sample terrestrial species of salaman- ders. Our pitfall trapping (180 days) caught 257 ensatina, 44 clouded sala- manders, and 13 Oregon slender salamanders, whereas TCS yielded 113 ensatina (0.44 times that of pit- r '. " ^ Table 6.— Analysis of variance of species richness and abundance (log transformed) categorized by age (old growth, mature, and young) and moisture (wet, moderate, and dry). Wet and dry old growth stands were not used in the analysis of stand age, and mature and young stands were not used in the analysis of stand moisture. Age (n = 1 7) Moisture < _J Ol 7 6 5 4 3 2 1 1 2 3 4 5 6 7 S IMATUREI YOUNG laEAR-l ao GROWTH ^ Figure 6.--Biomass of all (top) and class 4 and 5 (bottom) downed wood at 18 stands at the H. J. Andrews Experimental Forest, Oregon. YOUNG I — I TAl£D ENSATMA ROUGH- NORTHWEST RED- PACFIC FROG 8KJNNED SALA- UE.QGED TREE KEWT MANDER FROG FROG SNAKES UZARDS n CLEARCUT 78 Figure 7.— Relative abundance of herpetofauna in young stands and clearcuts. Above the horizontal: species more abundant in young stands. Below: species more abundant in clearcuts. Values are the greater mean adundance divided by the lesser, e.g., lizards were 78 times more abundant in clearcuts than in young forest stands. 18 salamanders, and this species merits special study. The Olympic salamander (Rhyacot- riton olympicus) occurs in or near small streams, which can be dis- rupted by timber harvest (Bury 1988, Bury and Corn 1988, Welsh, this vol- ume). Our techniques sampled ter- restrial habitats and we found few of this species (pitfall traps took only 4 in old-growth and 1 in mature stands). Many tailed frogs were cap- tured in pitfall traps in closed-can- opy forests, but they were absent or rare in clearcuts (only 1% of the total catch). Both the Olympic slamander and the tailed frog seem to be sensi- tive to timber harvest, and the sur- vival of these species may depend on protection of cool, flowing streams (required for breeding and larval de- velopment) as well as adjacent for- ested habitats (for shade and reten- tion of stream substrate quality, see Bury and Corn 1988). There is a need to assess the effects of logging in streamside and upland forests, which may directly or indirectly affect am- phibians in headwaters and small streams (Cooper et al. 1988, Bury and Corn 1988). Adults of the rough-skinned newt and Northwestern salamander mi- grate to ponds for breeding and, later, the adults and juveniles move back to land, which obfuscates their relation to forest type. The red- legged frog breeds in slow-moving creeks or ponds, and the proximity of such waters may have influenced the abundance of the frog in adjacent stands. Tailed frogs breed in small streams and the location of these wa- ters can greatly influence the occur- rence of the species in nearby forest stands. Also, we captured some juve- nile and adult tailed frogs 100 to >300 m from the nearest stream (Bury 1988). Before our study, tailed frogs were not thought to move far from water (Metter 1964, Nussbaum et al. 1983). Proximity of aquatic breeding sites apparently influenced the capture of several species in up- land habitat. At the same time, aquatic and semi-aquatic species might depend on the forest habitat for part of their life history, e.g., dis- persal. We suggest that future re- search emphasize the life history re- quirements and movement patterns of amphibians, which might help to resolve which factors are most im- portant to their continued local oc- currence and abundance. Fewer Oregon ensatina were cap- tured in mature forests than either young or old-growth stands, and this salmander might be associated with large amounts of CWD in the Oregon Cascades. Mature forests lack input from large trees and snags (see dis- cussions by Franklin et al. 1981, Har- mon et al. 1986, Spies et al. in press). Disturbance (fire or blow-down) cre- ates new young stands with appre- ciable amounts of CWD. Similar to our results, Raphael and Barrett (1984) found that the abun- dance of Oregon ensatina in northern California was correlated to density of large Douglas-fir trees. However, they found few ensatina in the youngest stands (<150 years) they studied, and they included ensatina with species associated with old- growth stands. In the Oregon Cas- cades, ensatina were ubiquitous and there is no apparent correlation with old-growth stands. Clouded salamanders were most abundant under the bark of relatively young logs. They may prefer class 2 and 3 logs, particularly occupying logs with loose bark. Also, clouded salamanders appear to be common in clearcuts (table 1). This species does not appear to be associated with old- growth conditions. In Washington, we only found the Larch Mountain salamander (Pletho- don larselli) at one old-growth stand (table 2). This species may be associ- ated with forested stands (Herring- ton and Larson 1985), but the relation needs further inquiry and verifica- tion. Management Considerations Current evidence suggests that rich, abundant populations of herpe- tofauna occur in naturally regener- ated forests. Within these stands, however, we found few differences in amphibians between wet, moder- ate, and dry old-growth sites and be- tween young, mature, and old- growth stands. These results might be related to '"old-growth" features occurring in many or all of these stands. For example, young and ma- ture sites retained many characteris- tics of old-growth forests: complex structure, snags, and large amounts of downed woody debris, particu- larly in older decay classes (fig. 6). Such material is the result of wildfire that burns and kills larger trees, which later fall to the ground. Wildfire often burns unevenly through stands, resulting in patches of lightly burned or unburned vege- tation surrounded by areas more in- tensively affected by fire. Some large trees might not be killed during fires and these persist into the regenerated stand. Burned trees become snags that later fall to the forest floor, creat- ing huge amounts of CWD. This heterogeneity and large amounts of CWD in naturally regenerated forest likely maintain favorable conditions for many species of the herpetofauna. Managed stands (clearcuts) had little downed CWD in older decay clases (fig. 6) and, generally, no snags nor trees (except for a rare spar pole or small planted trees). Current for- estry practices usually fell all trees and snags at sites, eliminating vari- ability in stand age and structure. Logging is generally followed by pre- scribed burning of slash and cull logs, reducing CWD by 50% or more (Bartels et al. 1985, Maser et al. 1979). The large amount of CWD at one of our Oregon clearcuts reflects light burning (fig. 6). Also, this site was surrounded by dense, old-growth forest, which probably contributed large amounts of CWD before burn- ing. 19 Often, the result of current timber harvest is even-aged stands with little CWD, especially in larger sizes. Present logging differs from that per- formed 30 or more years ago, when more CWD was left on the forest floor and smaller trees were left in- tact or ignored. Also, earlier practices tended to harvest larger, more valu- able trees with little or no site prepa- ration (except tree-planting), particu- larly on private lands. These were economic decisions, but the resultant second-growth stands may differ markedly from current intensive management of forests. In contrast to clearcuts, young stands (naturally regenerated) we studied were closed-canopy and had much downed woody debris. The predominant species were the tailed frog and ensatina, and young stands had more newts. Northwestern sala- manders and red-legged frogs than did clearcuts (fig. 7). Thus, there seem to be major differences in the herpetofaunas of pre-canopy clearcuts and naturally regenerated stands (young to old-growth). There is a critical need to compare differences in wildlife in intensively managed stands and those subjected to other treatments (e.g., prior log- ging practices, select-cut). At this time, there is a lack of information on herpetofaunas or other wildlife in managed second-growth forests. Managed forests soon will be the predominate forest type in the Pacific Northwest and the bulk of our wild- life probably will occur in these stands. Wise management of these forests should be of foremost concern for wildlife managers, and done in concert with protection of isolated habitat patches (old-growth forest). ACKNOWLEDGMENTS We thank our field crew for their untiring efforts: S. Boyle, R. Hayes, L. Hanebury, S. Martin, T. Olson, and S. Woodis. We thank J. Dragavon, L. Jones, P. Morrison, R. Pastor, and D. Smith for checking traps. A. McKee and J. Moreau at the H. J. Andrews Experimental Forest, Oregon, and personnel at Wind River Experimen- tal Forest and the Carson National Fish Hatchery, Washington, supplied logistical support. Data in figure 6 was provided by T. Spies. We also thank R. E. Beiswenger, A. B. Carey, and M. G. Raphael for review com- ments. This is contribution number 67 of the USD A Forest Service Proj- ect, Wildlife Habitat Relationships in Western Washington and Oregon. LITERATURE CITED Anthony, R. C, E. D. Forsman, G. A. Green, G. Witmer, and S. K. Nel- son. 1987. Small mammal popula- tions in riparian zones of different- aged coniferous forests. Murrelet 68:94-102. Bartels, Ronald, John D. Dell, Rich- ard L. Knight, and Gail Schaefer. 1985. Dead and down woody ma- terial, p. 171-186. In E. Reade Brown, tech. ed.. Management of wildlife and fish habitats in forests of western Oregon and Washing- ton. Part 1 - chapter narratives. USDA Forest Service, Pacific Northwest Publ. No. R6-F&WL- 192-1985. Bennett, Stephen H., J. Whitfield Gib- bons, and Jill Glanville. 1980. Ter- restrial activity, abundance and diversity of amphibians in differ- ently managed forest types. American Midland Naturalist 103:412-416. Bruce, Charles, Dan Edwards, Kim Mellen, Anita McMillan, Tom Owens, and Harold Sturgis. 1985. Wildlife relationships to plant communities and stand condi- tions, p. 33-55. In E. Reade Brown, tech. ed.. Management of wildlife and fish habitats in forests of west- ern Oregon and Washington. Part 1 - chapter narratives. USDA For- est Service, Pacific Northwest Publ. No. R6-F&WL-192-1985. Bury, R. Bruce. 1983. Differences in amphibian populations in logged and old growth redwood forest. Northwest Science 57:167-178. Bury, R. Bruce. 1988. Habitat rela- tionships and ecological impor- tance of amphibians and reptiles. [In press] In Kenneth J. Raedeke, ed., Streamside management: ri- parian wildlife and forestry inter- actions. Proceedings of a Sympo- sium [University of Washington, Seattle, February 11-13, 1987] Insti- tute of Forest Resourses, Univer- sity of Washington. Bury, R. Bruce, and Paul Stephen Com. 1987. Evaluation of pitfall trapping in northwestern forests: trap arrays with drift fences. Jour- nal of Wildlife Management 51:112-119. Bury, R. Bruce, and Paul Stephen Corn. 1988. Responses of aquatic and streamside amphibians to tim- ber harvest. [In press] In Kenneth J. Raedeke, ed., Streamside man- agement: riparian wildlife and for- estry interactions. Proceedings of a Symposium [University of Wash- ington, Seattle, February 11-13, 1987] Institute of Forest Resourses, University of Washington. Bury, R. Bruce, and Martin G. Ra- phael. 1983. Inventory methods for amphibians and reptiles, p. 416-419. In John F. Bell and Toby Atterbury, eds.. Renewable re- source inventories for monitoring changes and trends: Proceedings of an international conference. [Corvallis, Oregon, August 15-19, 1983] Society of American Forest- ers 83-14. Campbell, Howard W., and Stephen P. Christman. 1982. Field tech- niques for herpetofaunal commu- nity analysis, p. 193-200. In Nor- man J. Scott, Jr., ed., Herpetologi- cal Communities. U.S. Fish and Wildlife Service, Wildlife Research Report No. 13. Carey, Andrew B. 1988. The influ- ence of small streams on the com- position of upland bird communi- ties. [In press] In Kenneth J. Raedeke, ed., Streamside manage- 20 ment: riparian wildlife and for- estry interactions. Proceedings of a Symposium [University of Wash- ington, Seattle, February 11-13, 1987] Institute of Forest Resourses, University of Washington. Cooper, Edward J., William J. Trush, and Allen W. Knight. 1988. Effects of logging on Pacific giant sala- manders: influence of age-class composition and habitat complex- ity. Bulletin of the Ecological Soci- ety of America 69(suppl.):104-105. Enge, Kevin E., and Wayne R. Mar- ion. 1986. Effects of clearcutting and site preparation on herpe- tofauna of a north Rorida flat- woods. Forest Ecology and Man- agement 14:177-192. Fosburg, Whit. 1986. Wildlife issues in the National Forest system, p. 159-173. In Roger L. Di Silvestro ed., Audubon Wildlife Report 1986. The National Audubon Soci- ety, New York. 1094 p. Franklin, Jerry F., and Thomas A. Spies. 1984. Characteristics of old- growth Douglas-fir forests, p. 328- 334. In Proceedings, Society of American Foresters national con- vention. [Portland, Oregon, Octo- ber 16-20, 1983] Society of Ameri- can Foresters, Washington, D.C. Franklin, Jerry F., Kermit Cromack, Jr., William Denison, Arthur McKee, Chris Maser, James Sedell, Fred Swanson, and Glen Juday. 1981. Ecological characteristics of old-growth Douglas-fir forests. U.S. Forest Service, General Tech- nical Report PNW-118. 48 p. Friend, Gordon R. 1984. Relative effi- ciency of two pitfall-drift fence systems for sampling vertebrates. Austalian Zoologist 21:423-433. Gibbons, J. Whitfield, and Raymond D. Semlitsch. 1981. Terrestrial drift fences with pitfall traps: an effec- tive technique for quantitative sampling of animal populations. Brimleyana 7:1-16. Harmon, M. E., J. F. Franklin, F. J. Swanson, P. Sollins, S. V. Gregory, J. D. Lattin, N. H. Anderson, S. P. Cline, N. G. Aumen, J. R. Sedell, G. W. Lienkaemper, K. Cromack, Jr., and K. W. Cummins. 1986. Ecology of coarse woody debris in temperate ecosystems. Advances in Ecological Research 15:133-302. Harris, Larry D. 1984. The Frag- mented Forest: Island Biogeogra- phy Theory and the Preservation of Biotic Diversity. University of Chicago Press. 211 p. Herrington, Robert E., and John H. Larson. 1985. Current status, habi- tat requirements and management of the Larch Mountain salamander Plethodon larselli Bums. Biological Conservation 34:169-179. Jones, K. Bruce. 1981. Effects of graz- ing on lizard abundance and di- versity in western Arizona. South- west Naturalist 26:107-115. Jones, K. Bruce. 1986. Amphibians and reptiles, p. 267-290. In Allen Y. Cooperrider, Raymond J. Boyd, and Hanson R. Stuart eds.. Inven- tory and monitoring wildlife habi- tat. U.S. Dept. Interior, Bureau of Land Management, Service Cen- ter, Denver, Colorado. 858 p. Kerrick, Michael A., Kathy Johnson, and Richard J. Pedersen. 1984. What information is necessary for planning the management of old- growth forest for wildlife? p. 383- 386. In Proceedings, Society of American Foresters national con- vention. [Portland, Oregon, Octo- ber 16-20, 1983] Society of Ameri- can Foresters, Washington, D.C. Manuwal, David A., and Mark H. Huff. 1987. Spring and winter bird populations in a Douglas-fir forest sere. Journal of Wildlife Manage- ment 51:586-595. Maser, Chris, Ralph G. Anderson, Kermit Cromack, Jr., Jerry T. Wil- liams, and Robert E. Martin. 1979. Dead and down woody material, p. 78-95. In Jack Ward Thomas, tech. ed.. Wildlife habitats in man- aged forests: the Blue Mountains of Oregon and Washington. Agri- cultural Handbook 553. U.S. Dept. Agriculture, Washington, D.C. Maser, Chris, and James M. Trappe (tech. eds.). 1984. The seen and unseen world of the fallen tree. USDA Forest Service General Technical Report PNW-164. 56 p. Metter, Dean E. 1964. A morphologi- cal and ecological comparison of two populations of the tailed frog, Ascaphus truei Stejneger. Copeia 1964:181-195. Nussbaum, Ronald A., Edmund D. Brodie, Jr., and Robert M. Storm. 1983. Amphibians and Reptiles of the Pacific Northwest. University Press of Idaho, Moscow. 332 p. Rough, F. Harvey, Ellen M. Smith, Donald H. Rhodes, and Andres Collazo. 1987. The abundance of salamanders in forest stands with different histories of disturbance. Forest Ecology and Management 20:1-9. Raphael, Martin G. 1984. Wildlife populations in relation to stand age and area in Douglas-fir forests of northwestern California, p. 259- 274 In William R. Meehan, Theo- dore R. Merrell, Jr., and Thomas A. Hanley, eds.. Fish and wildlife relationships in old-growth for- ests. Proceedings of a symposium. [Juneau, Alaska, April 12-15, 1982] 425 p. American Institute of Fish- ery Research Biologists, (available from John W. Reintjes, Rt. 4, Box 85, Morehead City, North Caro- lina 28557). Raphael, Martin G., and Reginald H. Barrett. 1984. Diversity and abun- dance of wildlife in late succes- sional Douglas-fir forests, p. 352- 360. In Proceedings, Society of American Foresters national con- vention. [Portland, Oregon, Octo- ber 16-20, 1983] Society of Ameri- can Foresters, Washington, D.C. Raphael, Martin G., and Kenneth V. Rosenberg. 1983. An integrated approach to wildlife inventories in forested habitats, p. 219-222. In John F. Bell and Toby Atterbury, eds.. Renewable resource invento- ries for monitoring changes and trends: Proceedings of an interna- tional conference. [Corvallis, Ore- gon, August 15-19, 1983] Society of American Foresters 83-14. 21 Rosenberg, Kenneth V., and Martin G. Raphael. 1986. Effects of forest fragmentation on vertebrates in Douglas-fir forests, p. 263-272. In Jared Verner, Michael L. Morri- son, and C. John Ralph, eds.. Wildlife 2000. University of Wis- consin Press. 470 p. Ruggiero, Leonard F., and Andrew B. Carey. 1984. A programmatic approach to the study of old- growth forest — wildlife relation- ships, p. 340-345. In Proceedings, Society of American Foresters na- tional convention. [Portland, Ore- gon, October 16-20, 1983] Society of American Foresters, Washing- ton, D.C. Salwasser, Hal. 1987. Spotted owls: turning a battleground into a blue- print. Ecology 68:776-779. Scott, Norman J., Jr. (Ed.). 1982. Her- petological Communities. U.S. Fish and Wildlife Service, Wildlife Research Report 13. 239 p. Spies, Thomas A., Jerry F. Franklin, and Ted B. Thomas. In press. Coarse woody debris in Douglas- fir forests of western Oregon and Washington. Ecology. Stebbins, Robert C. 1985. A Field Guide to Western Reptiles and Amphibians. Houghton Mifflin Co., Boston. 336 p. Thomas, Don. In press. Forest age as- sociation of bats in western Ore- gon and Washington. Journal of Wildlife Management. Vogt, Richard C, and Ruth L. Hine. 1982. Evaluation of techniques for assessment of amphibian and rep- tile populations in Wisconsin, p. 201-217. In Norman J. Scott, Jr., ed., Herpetological Communities. U.S. Fish and Wildlife Service, Wildlife Research Report 13. Webb, G. A. 1985. Habitat use and activity patterns in some southeast Australian skinks. p. 23-30. In Gor- don Grigg, Richard Shine, and Harry Ehmann, eds.. Biology of Australasian Frogs and Reptiles. Surrey Beatty & Sons Pty Ltd, New South Wales, Australia. West, Stephen D. 1985. Differential capture between old and new models of the Museum Special snap-trap. Journal of Mammalogy 66:798-800. Wilcove, David S. 1987. Public lands management and the fate of the spotted owl. American Birds 41:361-367. 22 Long-Term Trends in Abundance of Annphlbians, Reptiles, and Mammals in Douglas- Fir Forests of Northwestern California^ Martin G. RaphaeP Abstract.— Relative abundance of 55 species of amphibians, reptiles, and mammals was estimated at 166 sites representing early clearcut through old- growth Douglas-fir forest in northwestern California. Nine species were strongly associated with older stands and 1 1 species were strongly associated with younger stands. The remaining species were either too rare to analyze statistically (22 species) or exhibited no clear trends of abundance in relation to stand age (13 species). Estimates of relative abundance of each species in each stage, coupled with data on historical, present, and future acreage of timber in each serai stage, were used to approximate the long-term impacts of timber harvest on the fauna of the Douglas-fir region in northwestern California. Management of old-growth Douglas- fir (Pseudotsuga menziesii) forests is controversial in the Pacific North- west, primarily because of the pos- sible value of old-growth as habitat for certain wildlife species versus the revenues represented by old-growth trees (Meslow et al. 1981, Harris et al. 1982). Management to provide wild- life habitat requires an inventory of associated wildlife species and an assessment of their old-growth de- pendency. An analysis of the size and distribution of habitat patches necessary to support viable popula- tions of those species is also critical (Burgess and Sharp 1981, Rosenberg and Raphael 1986, Scott et al. 1987). This study describes the relative abundance of amphibians, reptiles, and mammals in six serai stages rep- resenting clearcuts, young timber stands, and mature forest in north- western California. These estimates of relative abundance were used to project probable long-term changes in population size of amphibians, reptiles, and mammals as each serai ^ Paper presented at Symposium, Man- agement of Amphibians, Reptiles and Small Mammals in North America (Flagstaff, AZ, July 19-21, 1988). 'Research Ecologist, Forestry Sciences Laboratory, USDA Forest Service, Rocky Mountain Forest and Range Experiment Station, 222 South 22nd Street, Laramie, Wyoming 82070. Stage responds to forest management practices. METHODS Stand Selection Study stands were on the Six Rivers, Klamath, and Shasta-Trinity National Forests within a 50-km radius of Wil- low Creek, Calif. Forest cover was dominated by Douglas-fir, usually in association with an understory of tanoak (Lithocarpus ensiflorus) and Pa- cific madrone (Arbutus menziesii). Ele- vations varied from 400 to 1300 m. Stage 1 2 3 4 5 6 Raphael and Barrett (1984) describe methods for aging these stands. Ground surveys were used to verify stand conditions. Forest Service stand designations were used to guide stand selection, but the final classification of each stand into serai stages was based on measured vege- tation characteristics. The study region is characterized by warm, dry summers and cool, wet winters; total precipitation averages 60-170 cm per year. After selecting potential study stands using timber maps and aerial photographs, I then located all stands that were accessible by road, were relatively homogeneous with respect to tree cover, included no large clear- ings or other anomalous features, and were free from scheduled timber harvest for at least the next 3 years. From this restricted subset of stands, I randomly chose 10 to 15 stands representing each of six serai stages: Vegetation Sampling The structure and composition of vegetation on each stand in the three older serai stages was measured in three, randomly selected, 0.04-ha cir- cular subplots within a 90-m radius of each plot center. Within each sub- plot, observers recorded species. Serai state Age (yrs) Early <10 Late 10-20 Pole 20-50 Sawtimber 50-150 Mature 150-250 Old-growth >250 Classification Clearcut (brush/sapling) Young forest (pole/sawtimber) Mature forest 23 height, diameter at breast height (d.b.h.) and crown dimensions of each tree or shrub >2.0 m tall. In ad- dition, all trees >90-cm d.b.h. were counted on one 0.50-ha circular sub- plot centered on the plot. This sample permitted a better estimate of the density of large-diameter trees. Numbers of larger (>8-cm diameter) logs and volume of other downed woody debris were estimated along a 30-m transect crossing the center of each 0.04-ha subplot (Brown 1974). Marcot (1984) sampled vegetation in a similar manner on stands in the three early-seral stages. Vertebrate Sampling All field data were collected by a team of three to six biologists. We used a variety of techniques to sample various taxonomic groups. Pitfall Arrays We used pitfall arrays to capture small mammals (especially insecti- vores), reptiles, and salamanders. An array was composed of ten 2-gallon plastic buckets buried flush with the ground and covered with plywood lids, arranged in a 2 x 5 grid with 20- m spacing. We placed one array within each stand center and checked traps at weekly to monthly intervals from December 1981 (sawtimber, mature, old-growth; n = 27, 56, and 52 sites in each stage, respectively) or August 1982 (early shrub-sapling, late shrub-sapling, pole; n = 10 sites each) until October 1983. All live ani- mals were marked and released; re- captures were excluded from analy- ses. Dead animals were collected and prepared for permanent deposit in museum collections. Results for each species were expressed as captures per 1000 trapnights on each stand. Raphael and Rosenberg (1983) dem- onstrated that abundance estimates (capture rates) had stabilized after 15 months of continuous trapping. Drift Fence Arrays To better sample snakes, we installed a drift fence array (Campbell and Christman 1982, Vogt and Hine 1982) on each of 60 randomly selected stands (10 of each of the three early stages and sawtimber, 8 mature, and 12 old-growth). An array consisted of two 5-gallon buckets placed 7.6 m apart and connected by an aluminum fence 7.6 m long and 50 cm tall with two 20 X 76 cm cylindrical funnel traps, one on each side of the center of the fence. These fences were oper- ated from May through September 1983. All captures were combined with those from the pitfall arrays along with the associated trapnights from each stand. Track Stations Tracks of squirrels and other larger mammals were recorded on each site on a smoked aluminum plate baited with tuna pet food (Barrett 1983, Ra- phael and Barrett 1981, Raphael et al. 1986, Taylor and Raphael 1988). Based on results of a pilot study (Ra- phael and Barrett 1981), observers monitored each station for 8 days in August or September in 1981-1983, sampling 20 stations in each of the three early stages and 81, 168, and 157 stations in the sawtimber, ma- ture, and old-growth stages, respec- tively. The proportion of stations in each serai stage on which a species occurred was as an index of that spe- cies' abundance. Livetrap Grids To better estimate abundance of small mammals that were liable to escape from pitfalls, we established 27 livetrap grids (3 in each of the three earliest stages and 5, 7, and 6 in the three later stages), each of which usually consisted of 100 25-cm Sher- man livetraps arranged in a 10 x 10 grid with 20-m spacing. Other grid sizes or shapes were used when the plot configuration would not contain the standard grid. Traps were checked each day for 5 days (based on pilot studies, Raphael and Barrett 1981) during July in 1981 (late stages only), 1982, and 1983 (all stages). Re- sults for each species were expressed as mean number of captures per 100 trapnights. Surface Searcti To better sample certain amphibian species, we conducted time- and area-constrained searches (Bury and Raphael 1983, Raphael 1984) on a subset of sites in 1981 (late stages), 1982, and 1983 (all stages). A two- person team searched under all mov- able objects and within logs on three randomly located 0.04-ha circular subplots (fall 1981, 1982) or within a 1-ha area for 4 working hours (spring 1983). We conducted 20 surveys in each of the three early stages and 29, 39, and 48 surveys in the three late stages. Opportunistic Observations Observers recorded the presence of vertebrates or identifiable vertebrate sign incidental to the above proce- dures. We tallied observations to cal- culate frequency of occurrence of rarer species within each stage. Forest Area Trends Estimates of historical, current, and future acreage in each serai stage were taken from Raphael et al. (in press). For these analyses, I com- bined similar pairs of serai stages into three generalized stages repre- senting brush/sapling, pole/ sawtim- ber, and mature timber. I then com- puted relative abundance of each vertebrate species in these three stages using a weighted average (weights based on sampling effort) of 24 estimates from each of the two stages forming the pair. Population esti- mates for historical, present, and fu- ture time periods were computed using the formula: 3 where P.^ was the relative population size of the zth vertebrate species at time t,D.. was the relative abundance of the zth vertebrate in the /th serai stage, and A .^ was the total area of each of the tkree serai stages at time RESULTS Vegetation Structure Comparisons of vegetation structure among the serai stages (table 1) showed that older stands had greater canopy volume, basal area, litter depth, and density of Douglas-fir stems >90 cm d.b.h. Downed wood mass differed among stages, but the greatest volume occurred in the youngest stands, probably in the form of logging slash, and the lowest volume occurred in pole and sawtim- ber stages. Early-seral stands were higher in elevation than older stands, probably because of the logistics of timber harvest in the area (most clearcuts were located along ridg- etops). Stands in the two earliest serai stages, also because of logging, were smaller in area than stands in the four older stages. Vertebrate Abundance and Diversity Among all plots and years of study, we recorded 9,928 captures of all Table 1 .—Comparisons of vegetation characteristics among serai stages of Douglas-fir forest, northwestern California, 1981-1983. Characteristic Early Late brush/ brush/ Old- sapiing sapling Pole SawtimberMature growth Canopy volume (jTT'lrv?) 10.77 n.26 ^3.64 7.15 7,52 7.47 Live stem basal area (m^hc) ^2.6 ^52.8 50.5 60.2 65.6 Snog basal area (m^/ha) 2_ 4.7 6.1 5.3 Downed wood moss (metric tons/ha) <8 cm diameter ^9.7 V.9 m.9 12.9 12.3 11.5 >8 cm diameter ^81.4 V4.7 ^52.4 32.3 43.6 67.3 Utter depth (cm) ^2.2 M.8 ^6.0 6.2 5.1 7.1 Douglas-fir >90 cm d.b.h. (n/ha) — 3.6 19.3 25.7 Elevation (m) 1128 1016 972 660 832 904 Stand area (ha) 12,3 21.9 41.2 47.1 62.0 84.2 Solar radiation index^ 0.34 0.41 0.51 0.49 0.49 0.43 Slope (%) 48 30 31 36 41 52 Age (years since clearcut. or index) 9 14 123 206 294 'Dofo are from Marcot(1984), with permission, and represent a larger number of sites than v/ere sampled in tt)e present study. 'Dasties indicate no values were available. ^ Index of total yearly solar energy flux (Frank and lee 1966). Larger values indicate warmer, drier sites. J Species during 898,431 trapnights from pitfalls and drift fences; 1,636 captures of amphibians during sur- face searches; 3,066 small mammal captures during 35,070 trapnights from Hvetrap grids; and 510 detec- tions of larger mammals from track stations. Relative abundances of 55 species, based on the most appropri- ate sampling method for each spe- cies, are summarized in table 2. Val- ues are comparable across stages but not among taxa if different sampling methods were used. Amphibians were much more abundant in for- ested than in clearcut stands, whereas reptiles were more abun- dant in clearcuts. None of the am- phibians and reptiles [except rarer species such as northwestern sala- mander (see appendix for scientific names of vertebrates)] was absent from any stage. Mammals exhibited a greater vari- ety of responses to serai stage. Some (e.g., Douglas' squirrel, western red- backed vole) increased in abundance from earliest to latest serai stages; others (e.g., deer mouse) decreased along this gradient. A number of spe- cies (e.g., Allen's chipmunk, dusky- footed woodrat, pinyon mouse, Cali- fornia vole) were most abundant both in late shrub-sapling and ma- ture or old-growth stands. Mean numbers of mammal and reptile species recorded per stand differed among serai stages, but mean numbers of amphibian species did not differ significantly (fig. 1). Among mammals, mean numbers of species were greatest in mature and old-growth stages. In contrast, mean numbers of reptile species were greatest in the two earliest stages. Long-Term Trends Estimates of land area in each serai stage through time (table 3) indicate more area is occupied by early serai stages currently than during historic or future times. Mature and old- growth stages currently occupy 25 Table 2.~Mean relative abundance of amphibians, reptiles, and mammals among serai stages of Douglas-fir forest, nortliwestern California, 1981 -1983. Species' Sampling . metliod(s)2 Total captures Relative abundance among serai stages^ 6 Significance* Salamanders Northwestern salamander^ Pacific giant salamander Olympic salamander^ Rough-skinned newt Del Norte salamander Ensatina Black salamander Clouded salamander Frogs and toads Tailed frog^ Western toad Pacific treefrog Foothill yellow-legged frog^ Bullfrog^ Turtles Western pond turtle^ PD. TC^ PD TC TC^ PD,TC^ 6 0 0 0 1 3 4 PD 28 0 0.05 0 0.01 0.02 0.04 PD,TC6 5 0 0 0 1 1 3 PD 68 0.02 0 0 0.05 0.09 0.04 TC 196 0.70 0.60 0.05 0.07 1.92 1.92 TC 1116 2.40 1.85 8.10 6.28 8.15 7.69 TC 32 0.05 0.05 0.05 0.03 0.21 0.42 TC 103 0.35 1.55 0.50 0.10 0.31 0.83 0,114 0.403 0.035 0.001 o.on 0.009 3 0 0 0 0 2 0 54 0.18 0.03 0.02 0.08 0,06 0.01 0.035 51 0.60 0.05 0,10 0.55 0,03 0,06 0.000 6 1 0 0 1 0 0 3 0 0 0 1 0 0 5 0 0 0 4 0 0 Lizards Western fence lizard PD Sagebrush lizard PD Western skink PD Southern alligator lizard PD Northern alligator lizard PD Snakes Rubber boa^ Rtngneck snake^ Sharp-tailed snake^ Racer^ Gopher snake-^ Common kingsnake^ Common gartersnake5 Western terrestrial gartersnakeS Western ratt-lesnake5 Mammals Pacific shrew PD Trowbridge's shrew PD Shrev/-mole PD Coast mole^ PD Allen's chipmunk LT Weste rn g ray squi rrel TP^ Douglas' squirrel TP* Northern flying squirrel TP* 523 1.77 2.38 0.30 0.94 0.54 0.11 0.000 196 2.66 0.76 0.25 0.09 0.11 0.01 0.000 584 3.05 3.47 0.78 0.73 0,42 0.13 0,000 41 0.03 0 0 0.11 0,05 0.03 0.085 586 0.81 1.03 0.90 0.97 0.60 0.44 0,029 CO, PD^ 7 0 20 10 0 4 0 OCPD* 6 0 0 0 0 4 0 PD^ 22 10 20 30 0 5 4 OCPD^ 8 0 20 0 4 5 4 GO, PD* 0 0 10 0 2 0 00,PD* 1 0 0 0 4 0 0 GO, PD6 19 20 10 20 0 5 11 400,PD 11 0 20 10 7 4 6 oo. pd;tc6 5 0 10 0 0 2 6 89 0.02 0.08 0 0.07 0.07 0.17 2384 2.70 4.01 2.83 3.04 3.16 3.80 479 0.04 0.16 0.25 0.76 0.55 0.55 15 0 0.03 0 0.02 0.05 0.06 254 16.7 29.5 0.8 2.8 5.2 5.0 48 0 0 10 12 12 9 104 0 0 20 16 22 30 43 0 0 15 9 18 13 0.004 0.215 0.002 0.003 0.378 0.001 0.046 (continued) 26 Table 2. —(continued). Species' Sampling method(s)2 Total captures Relative abundance among serai stages^ 6 Significance* Deer mouse PD 1 Ml 5.09 0 C\~l o.U/ U.OV u.bo U.Vo 1.28 0.000 Brush mouse LT 33 0 0.33 3.67 0.25 0.25 0 0.216 Pinyon mouse LT 222 1.35 10.34 4.67 10.63 3.86 2.76 0.086 Dusky-footed woodrat LT 115 1.9 3.5 0.2 1.2 4.4 3.4 0.000 Western red-backed vole PD 669 0,35 0.36 0.46 0.45 0.82 0.97 0.015 Red tree vole PD 19 0 0.10 0 0.07 0.11 0.15 0.586 California vole PD 106 0.89 1.70 0.03 0.02 0.01 0.01 0.000 Creeping vole PD 22 0.09 0.03 0.05 0.04 0.01 0.01 0,038 Western jumping mouse^ PD 2 0 0 0 0 0.04 0.02 Coyote^ ALL^ 7 10 30 0 15 9 15 Gray fox TP^ 63 20 15 10 30 11 8 0.001 Black bear TP 196 20 25 5 42 45 48 0.028 Ringtail TP 25 0 0 0 10 6 4 0.249 Raccoon^ TP 3 0 0 0 0 1 1 Fisher TP 58 0 5 25 6 13 15 0.060 Ermine^ PD 2 0 0 0 0 0,02 0.02 Western spotted skunk TP 70 10 15 5 10 18 15 0.426 Striped skunk^ TP 17 0 0 0 7 6 1 Bobcat^ TP 3 5 5 0 1 2 0 'A// names follow Laudenslayer and Grenf ell (1983). ^PD = Pitfall plus drift fence, TC = Time- and area-constrained searcti. OO = Opportunistic observations, TP = Track plots. LT= Live traps, ALL = all observation methiods combined. ^Seral stages (and numbers of stands sampled) are: 1 — early brustt/sapling (n= 10): 2— late brusti/sapling (n= 10): 3— pole (n= 10): 4— sawtimber (n=27): mature (r]=56); 5—old-growtht (n=63). "Significance from analysis of variance (means) or cN-square analysis (frequencies) comparing abundances among stages. A dashi indicates th\at no test was performed. ^Too rare for subsequent analyses. '^Abundance values based on percent frequencies. about half of historic acreage, and these stages will probably occupy only about 30% of current acreage under the most likely harvest pat- terns of the future (table 3). The implications of these changing distributions of serai stages for am- phibians, reptiles, and mammals are summarized in figure 2. Nearly equal numbers of species are likely to have increased or decreased by more than 25% relative to historic abundance at present and in the future. Three of the five reptile species are presently more abundant than in historic times and all five species will likely be more abundant in the future. Am- phibians showed an opposite pattern. Four of the eight species are pres- ently less abundant and five of the eight may be less abundant in the fu- ture. Among the 20 mammal species, seven are presently less abundant than in historic times whereas five are more abundant. Eight species will probably be less abundant in the future and six more abundant. DISCUSSION Abundance in Serai Stages Results suggest late brush/ sapling and mature/old-growth serai stages provided more productive wildlife habitat than early brush/ sapling, pole, and sawtimber stages. Among amphibians, only ensatinas were cap- tured frequently in pole sites. Clouded salamanders were generally under bark or inside downed logs and persisted in clearcut stands as long as adequate numbers of logs were retained, especially in late sites (Raphael 1987, Welsh, this volume). Lizards were more abundant in earlier serai stages than in pole and mature stages. Among snakes, only sharp-tailed snakes were observed on early sites; other species occurred on later sites. However, sampling was not sufficient for definitive con- clusions. 27 With the exception of the deer mouse, small mammals were more abundant on late brush/ sapling sites. Dusky-footed woodrats were of spe- cial interest in this regard as we ob- served many woodrat nests built among the stems of tanoak and Pa- cific madrone in late brush/ sapling sites. The combination of abundant mast, good nesting substrate, and protection from predation (spotted owls rarely forage in old, brush- dominated clearcuts) provided by the dense, brushy cover were proba- bly the reasons that woodrats and other small mammals were more numerous in late clearcut sites (Ra- phael 1987). Tree squirrels were most abun- dant in mature forest sites and ground squirrels were more abun- dant in early clearcut sites. Chip- munks were the only squirrel that reached peak abundance in early serai sites. Their abundance was cor- related with the cover of tanoak in the understory (Raphael 1987). Man- agement actions, such as herbicide treatments, that shorten or delete the late brush/ sapling stage are probably detrimental to chipmunks, woodrats, and certain other rodents. Several carnivorous mammals were abundant in the late brush/sap- ling stage. Greater prey density in late compared to early and pole sites may explain this higher frequency of carnivores although more data will be necessary to confirm this observa- tion. Of the 55 species observed, 20 were strongly associated with either older (9 species) or younger (11 spe- cies) stands (table 4). Three salaman- ders and six mammals were associ- ated with older stands. One toad, one frog, five lizards, and four mam- mals were associated with younger stands. Five species associated with old-growth were also abundant in late (brushy) clearcut stages (table 2). These species peak in abundance in old stands and late clearcuts, with low abundance in intermediate age classes. Table 3 — Approximate area (millions of ha) of serai stages in Douglas-fir forests of northwestern California In historic, present, and future time peri- ods (after Raphael et al., in press). Serai stage Historical Present Likely future Worst case future' Brush/sapling 0.14 0.49 0.20 0.24 Young forest 0.14 0.20 0.77 0.85 Mature forest 0.81 0.40 0.12 0.00 'Assumes that all mature and old-growth stands ore harvested and all lands man- aged under short rotatiorts. CO PRESOff 12 r 3 4 SERAL STAGE Figure 1 .—Mean numbers of amphibian, reptile, and mammal species observed in serai stages of Douglas-fir forest, northwest- ern California, 1981-1983. Serai stages (and numbers of stands sampled) are: 1 - early brush/sapling (n = 10); 2 - late brush/sap- ling (n = 1 0); 3 - pole (n = 1 0); 4 - sawtimber (n = 27); 5 - mature (n = 56); 6 - old-growth (n = S3). Vertical lines indicate 95% confi- dence intervals. UttLYFUIUt iMMi jlBTlB <-75 -75 -50 -TS • CHANGE h ASUNQANCE (1 Figure 2.— Percent change in population size of amphibian, reptile, and mammal species at present and in the future relative to estimated historical populations. Histo- grams represent the numbers of species increasing or decreasing by specified per- centages. 28 I examined habitat associations among each of the above 9 species by computing correlations of their abun- dance with specific habitat compo- nents (table 5). Density of large trees and hardwood volume were corre- lated with the abundance of most species. Moisture, as measured by the presence of surface water, mois- ture-loving tree species, or north-fac- ing slopes, was important for most mammals and one salamander spe- cies. Four mammal S}:>ecies were sig- nificantly more abundant on higher elevation stands. Downed wood vol- ume also was significantly and posi- tively correlated with abundance of four amphibian and mammal spe- cies. The abundance of hardwoods in the understory was important for many species in each group. In con- trast, snag density was not positively correlated with the abundance of any species. Long-Term Trends The list of sensitive species (table 4) is tentative pending results of addi- r Table 5.— Habttat components that were correlated with relative abun- dance of amphibians and mammals associated with late-seral Douglas-fir forests of northwestem California. Density of Hardwood Downed conifers under- wood Standing Species >90-cm d.b.h. story mass snags Moisture Elevation Del Norte salamander X Black salamander X X Clouded salamander X X Pacific shrew X X X X Douglas' squirrel X ^(X) X X X Northern flying squirrel X (X) Dusky footed woodrat X X X X Western red-backed vole X X X X Fisher X X 'Parentheses indicate negative correlations. tional surveys and more intensive, species-specific research. The projec- tions, although based on an intensive sampling effort, are highly specula- tive. Three assumptions must be rec- ognized to interpret these results. First, I assumed that greater relative abundance in a serai stage indicates a species' preference for that stage and that preferences remain constant with shifting distribution of acreage r Table 4.— Amphibian, reptile, and mammal species most strongly affected by future harvest of old-growth Douglas-fir forest, northwestern California.' Decreasers— associated with late-seral forest Increasers— associated with early-seral forest Species decline^ Species % increase^ Del Norte salamander 75 Western toad 45 Black salamander 71 Pacific treefrog 160 Clouded salamander 29 Western fence lizard 60 Pacific shrew 39 Sagebrush lizard 44 Douglas' squirrel 31 Western skink 59 Northern flying squirrel 31 Southern alligator lizard 60 Dusky-footed woodrat 55 Northern alligator lizard 43 Western red-backed vole 37 Pinyon mouse 70 Fisher 26 California vole 44 Creeping vole 102 Gray fox 78 'Species were listed if thieir estimated future abundance differed by more ftian 25% from estimated historical abundance and if mean abundance differed significantly (P <0.10) among serai stages (table 2). 'Percent increase or decrease in estimated hjture abundance compared with estimated historic abundance. in each stage. Some species have (or could) adapt to new stages over time. Second, I assumed total acreage of each serai stage can be used to esti- mate responses of vertebrates with- out regard to size and juxtaposition of stands comprising each stage. However, continued fragmentation of forest habitats may result in dis- junct patches so small they cannot support a species that would other- wise find the habitat suitable. Rosen- berg and Raphael (1986) found that at least eight species of amphibians (2), reptiles (2), and mammals (4) were significantly less abundant in stands <10 ha in size than in larger stands. Some of these (e.g., western gray squirrel) were not listed in this study among the sensitive species (table 4), but the effects of habitat fragmentation may nonetheless be cause for concern. A third assumption is that young forested stands (pole, sawtimber) in this study represent young stands of the future. Naturally occurring pole and sawtimber stands contain some large Douglas-fir stems and a sub- stantial amount of standing and downed wood (table 1). If future management activities result in fewer large live trees, snags, and downed logs, the abundance of vertebrates associated with these habitat compo- nents may also decline. In this case. 29 responses of vertebrates to forest management may be more extreme than those projected. The overall trend is for increased abundance among species of south- ern affinity that are associated with open, drier habitats in other parts of their ranges, and decreased abun- dance among species of boreal affin- ity that are primarily associated with moist coniferous forest throughout their ranges. Furthermore, most of the increasers are widespread species with large distributions that include many nonthreatened habitats. In con- trast, the decreasers are almost all species with rather restricted total ranges, most of which are in threat- ened habitats. Therefore, even though total numbers of increasers and decreasers are nearly equal, the effects of old-growth reduction should not be viewed as neutral. Because many of the decreasers are affected by soil moisture and other microclimatic conditions, man- agement to protect stream edges, moist ravines, and other moist sites may provide refuges for species that can later recolonize maturing stands. Management efforts to retain (or rec- reate) natural components of regen- erating stands, such as hardwood understory, snags, and logs, may help mitigate against wildlife losses in future forests. It is not stand age, per se, but the structural characteris- tics of forests of various ages that are important to survival of most spe- cies. Finally, results of this study ad- dress another important forest man- agement issue in the northwest; What should managers use as a baseline for evaluation of impacts: historic or present conditions? It is apparent that many species are pres- ently much less abundant compared with historic numbers (fig. 2). Addi- tional reductions because of contin- ued timber harvest will cause further declines in some species but most major declines have already oc- curred. Therefore, I believe that esti- mates of historic populations should be used as baselines for monitoring biological diversity, rather than pre- sent populations. ACKNOWLEDGMENTS Field studies were funded by the Pa- cific Southwest Region and the Pa- cific Southwest Forest and Range Experiment Station of the USDA For- est Service and by the University of California, Agricultural Experiment Station 3501 MS. I especially thank my field assistants (Paul Barrett, John Brack, Cathy Brown, Christopher Canaday, Lawrence Jones, Ronald LaValley, Kenneth Rosenberg, and Cathy Taylor) for their dedication and blisters; R. H. Barrett, C. J. Ralph, and J. Verner for their sup- port; Bruce G. Marcot for freely shar- ing information from his studies and for valuable discussions; and Ken- neth V. Rosenberg, Fred B. Samson, and Hobart M. Smith for their com- ments on an earlier draft of this manuscript. LITERATURE CITED Barrett, Reginald H. 1983. Smoked aluminum track plots for deter- mining furbearer distribution and abundance. California Fish and Game 69:188-190. Burgess, Robert L., and David M. Sharpe. 1981. Forest island dy- namics in man-dominated land- scpates. Springer- Verlag, New York. 310 p. Brown, James K. 1974. Handbook for inventorying downed woody ma- terial. USDA Forest Service Gen- eral Technical Report INT-16. 24 p. Bury, R. Bruce, and Martin G. Ra- phael. 1983. Inventory methods for amphibians and reptiles, p. 426-419. In J. F. Bell and T. Atter- bury (eds.). Renewable Resource Inventories for Monitoring Changes and Trends. College of Forestry, Oregon State University, Corvallis, Oregon. Campbell, H. W., and S. P. Christ- man. 1982. Field techniques for herptofaunal community analysis, p. 193-200. In N. J. Scott (ed.). Her- petological Communities. USDI Fish and Wildlife Service Wildlife Research Paper 13. 239 p. Frank Ernest C, and Richard Lee. 1966. Potential solar beam irradia- tion on slopes. USDA Forest Serv- ice Research Paper RM-18. Harris, Larry D., Chris Maser, and Arthur McKee. 1982. Patterns of old growth harvest and implica- tions for Cascades wildlife. Trans- actions of North American Wild- life and Natural Resource Confer- ence 47:374-392. Laudenslayer, William F., Jr., and William E. Grenfell, Jr. 1983. A list of amphibians, reptiles, birds and mammals of California. Outdoor California 44:5-14. Marcot, Bruce G. 1984. Habitat rela- tionships of birds and young- growth Douglas-fir in northwest- ern California. Corvallis, OR: Ore- gon State University; 282 p. Ph.D. dissertation. Meslow, E. Charles, Chris Maser, and Jared Verner. 1981. Old- growth forests as wildlife habitat. Transactions of North American Wildlife and Natural Resource Conference 46:329-344. Raphael, Martin G. 1984. Wildlife di- versity and abundance in relation to stand age and area in Douglas- fir forests of northwestern Califor- nia, p. 259-274. In Meehan, W. R., T. T. Merrell, Jr., and T. A. Hanley (tech. eds.). Fish and Wildlife Rela- tionships in Old-growth Forests: proceedings of a symposium (Jun- eau, Alaska, 12-17 April 1982). Bookmasters, Ashland, Ohio. Raphael, Martin G. 1987. Wildlife tanoak associations in Douglas-fir forests of northwestern California, p. 183-189. In Plumb, T. R., N. H. Pillsbury, (tech. coord.). Proceed- ings of the Symposium on Mul- tiple-Use Management in Califor- nia's Hardwood Resources; No- vember 12-14, 1986, San Luis 30 Obispo, CA. General Technical Report PSW-100. Berkeley, CA: Pacific Southwest Forest and Range Experiment Station, Forest Service, U.S. Department of Agri- culture, 462 p. Raphael, Martin G., and Reginald H. Barrett. 1981. Methodologies for a comprehensive wildlife survey and habitat analysis in old-growth Douglas-fir forests. Gal-Neva Wildlife 1981:106-121. Raphael, Martin G., and Reginald H. Barrett. 1984. Diversity and abun- dance of wildlife in late succes- sional Douglas-fir forests, p. 352- 360. In New Forests for a Chang- ing World. Proceedings 1983 Con- vention of the Society of American Foresters. 650 p. Raphael, Martin G., and Kenneth V. Rosenberg. 1983. An integrated approach to inventories of wildlife in forested habitats, p. 219-222. In J. F. Bell and T. Atterbury (eds.). Proceedings, conference on renew- able resource inventories for monitoring changes in trends. Corvallis, Oregon, 1983. Raphael, Martin G., Kenneth V. Rosenberg, and Bruce G. Marcot. In press. Large-scale changes in bird populations of Douglas-fir forests, northwestern California. Bird Conservation 3. Raphael, Martin G., Cathy A. Taylor, and Reginald H. Barrett. 1986. Sooted aluminum track stations record flying squirrel occurrence. Pacific Southwest Forest and Range Experiment Station Re- search Note PSW-384. Rosenberg, Kenneth V., and Martin G. Raphael. 1986. Effects of forest fragmentation on wildlife commu- nities of Douglas-fir. p. 263-272. In Appendix Verner, J., M. L. Morrison, and C. J. Ralph (eds.). Modeling habitat relationships of terrestrial verte- brates. University of Wisconsin Press, Madison, WI. Scott, J. Michael, Blair Csuti, James D. Jacobi, and John E. Estes. 1987. Species richness — a geographic ap- proach to protecting future bio- logical diversity. Bioscience 37:782-788. Taylor, Cathy A., and Martin G. Ra- phael. 1988. Identification of mam- mal tracks from sooted track sta- tions in the Pacific Northwest. California Fish and Game 74:4-11. Vogt, R. C, and R. L. Hine. 1982. Evaluation of techniques for as- sessment of amphibian and reptile populations in Wisconsin, p. 201- 217. In N. J. Scott, (ed.). Herpeto- logical Communities. USDI Fish and Wildlife Service Research Re- port 13, 239 p. Common and scientific names of vertebrates mentioned in text (nomenclature follows Laudenslayer and Grenfell (1983)). Salamanders Northwestern salamander Amhystoma gracile Pacific giant salamander Dicamptodon ensatus Olympic salamander Rhyacotriton olympicus Rough-skinned newt Taricha granulosa Del Norte salamander Plethodon elongatus Ensatina Ensatina eschscholtzi Black salamander Aneides flavipunctatus Qouded salamander Aneides ferreus Frogs and toads TaUed frog Ascaphus truei Western toad Bufo boreas Pacific treefrog Hyla regilla Foothill yellow-legged frog Rana boylei Bullfrog Rana catesbeiana Turtles Western pond turtle Clemmys marmorala Lizards Western fence lizard Sceloporus occidentalis Sagebrush lizard Sceloporus graciosus Western skink Eumeces skiltonianus Southern alligator lizard Gerrhonotus multicarinatus Northern alligator lizard Gerrhonotus coeruleus Snakes Rubber boa Charina botlae Ringneck snake Diadophis punctatus Sharp-tailed snake Phyllorhynchus decurtatus Racer Coluber constrictor Gopher snake Pituophis melanoleucus Common kingsnake Lampropeltis zonula Common gartersnake Thamnophis sirtalis Western terrestrial gartersnake Thamnophis elegans Western rattlesnake Crotalis viridis Mammals Pacific shrew Sorex pacificus Trowbridge's shrew Sorex trowbridgii Shrew-mole Neurotrichus gibbsii Coast mole Scapanus orarius AUen's chipmunk Tamias senex Western gray squirrel Sciurus griseus Douglas' squirrel Tamiasciurus douglasii Northern flying squirrel Glaucomys sabrinus Deer mouse Peromyscus maniculatus Brush mouse Peromyscus boylii Pin yon mouse Peromyscus truei Dusky-footed woodrat Neotoma fuscipes Western red-backed vole Clethrionomys californicus Red tree vole Arborimus longicaudus California vole Microtus californicus Creeping vole Microtus oregoni Western jumping mouse Zapus princeps Coyote Canis latrans Gray fox Urocyon cinereoargenteus Black bear Ursus americanus Ringtail Bassanscus astutus Raccoon Procyon lotor Fisher Martes pennanti Ermine Mustela erminea Western spotted skunk Spologale gracilis Striped skunk Mephitis mephitis Bobcat Lynx rufus 31 Use of Woody Debris by Plethodontid Salamanders In Douglas-Fir Forests In Washington Keith B. Aubry,^ Lawrence L C. Jones,^ and Patricia A. Hali^ Abstract.— Ensaf/no eschscholfzii \f^os found most often under pieces of bark, whereas Plefhodon vehiculum occurred primarily under logs. Captures of both species were highest in young stands, but occurred in all age classes. Our results suggest that the retention of coarse woody debris in managed forests would provide for the habitat needs of these species. The harvesting of old-growth Douglas-fir (Pseudotsuga menziesii) forests in the Pacific Northwest, and its potential effects on wildlife spe- cies has been the focus of much con- cern in recent years (e.g.. Lumen and Nietro 1980, Franklin et al. 1981, Meslow et al. 1981, Meehan et al. 1984, Gutierrez and Carey 1985). Most of this attention has been di- rected towards birds and mammals such as the spotted owl (Strix occiden- talis), Vaux's swift (Chaetura vauxi), northern flying squirrel (Glaucomys sahrinus), and red ti'ee vole (Ar- borimus longicaudus); little concern has been expressed about amphibi- ans and reptiles. These groups have not been studied extensively in the Pacific Northwest. Only recently has research been conducted on habitat associations among different forest age classes (Raphael 1984, Raphael and Barrett 1984, Ruggiero and Carey 1984). From 1983 to 1986, the USDA For- est Service and USDI Bureau of Land 'Paper presented at symposium. Man- agement of Amphibians. Reptiles, and Small Mammals in Nortti America. (Flag- staff, AZ, July 19-21, 1988). ^Researcti Wildlife Biologist, USDA Forest Service, Pacific Nortt)west Research) Station, 3625 93rd Ave. SW, Olympia, WA 98502. ^Biological Technician, USDA Forest Serv- ice, Pacific Northwest Research Station, 3625 93rd Ave. SW, Olympia, WA 98502. "Wildlife Biologist, USDA Forest Service, Pacific Northwest Research Station, 3625 93rd Ave. SW, Olympia, WA 98502. Management funded a major re- search effort aimed at identifying wildlife species that occur in highest abundances in old-growth Douglas- fir forests and investigating the eco- logical basis of observed patterns of association Amphibian communities were sampled using pitfall traps, stream surveys, and time-constrained searches (Standard Sampling Proto- cols on file at the Forestry Sciences Laboratory, Olympia, WA). Some of the results of these studies are re- ported elsewhere in this volume (Bury and Corn 1988, Welsh 1988). Here, we report the results of time- constrained searches conducted in southern Washington in 1984. Our objectives are to (1) identify potential habitat associations, (2) examine pat- terns of cover object use, and (3) evaluate the efficacy of this technique for studying amphibians in this re- gion. Study Area Forty-five forest stands were sampled in the southern portion of the Cascade Range in Washington (fig. 1). Stands ranged in age from 55 to 730 yr and were at least 20 ha in size. All stands were located within the western hemlock (Tsuga heter- phylla) zone and lower elevations of the Pacific silver fir (Abies amabilis) zone (Franklin and Dyrness 1973), which are characterized by a wet and Figure 1 .—Location of study stands by age class in the southern Washington Cascade Range. mild maritime climate. Snow rarely accumulates at our sites. Old-grovv^th stands (210-730 yr) typically contained high proportions of Douglas-fir and western hemlock and, in wet sites, western redcedar (Thuja plicata). Mature (95-190 yr) and young (55-80 yr) stands were 32 dominated by Douglas-fir. In all age classes, other species such as red alder (Alnus rubra), vine maple (Acer circinatum), bigleaf maple (A. macro- phyllum), Pacific silver fir, and west- ern hemlock occurred in lesser amounts. Average age of each stand was determined through growth ring counts, either by increment coring or examination of cut stumps in nearby stands. Old-growth stands were clas- sified into wet, moderate, and dry moisture classes on the basis of flo- ristic and physiographic characteris- tics; all young and mature stands were in the moderate moisture class (T. A. Spies, unpubl. data). All stands had resulted from natural regenera- tion following fires; none had under- gone silvicultural treatments. Methods Surveys for terrestrial amphibians were conducted from 16 April to 12 June 1984; all but four high-elevation stands were sampled by 4 May. A time- constrained search method was used (Campbell and Christman 1982). A crew of two to four persons actively searched each stand for am- phibians for a total of 4 person- hours. An initial search area was se- lected at least 50 m within the stand to avoid edge effects. In general, woody debris such as logs, snags, and pieces of bark was abundant in each stand and consti- tuted virtually all potential cover ob- jects. An area was searched for 0.5 person-hours, after which we moved a minimum of 25 m to search another suitable area; sampling areas were not spatially constrained. This was repeated until the sampling period was over. All potential cover objects were searched by hand or with po- tato rakes, but no single object was searched for more than 20 min. Logs of all sizes in advanced stages of de- composition were pulled apart with potato rakes. Areas beneath large undecomposed logs could not be Table 1.— Amphibian species captured during time-constrained searches in the southern Washington Cascade Range by stand type.' Species Mean Captures ± Standard Error (N=9) 30 cm), but our inability to ade- quately search this cover type may account for these results. Virtually all logs where ensatinas and redback 34 salamanders were found were in in- termediate stages of decay (fig. 5) (see Maser et al. 1979, p. 80). Only a few captures of either species oc- curred in association with intact or extensively decomposed logs. Nei- ther species was commonly found under rocks, but this cover type is relatively rare in Douglas-fir forests. No correlations between slope or as- pect and amphibian capture sites could be detected. Discussion Old-growth forests do not appear to provide unique habitat for either en- satinas or western redback salaman- ders; both species were well-repre- sented in all age classes. Our results suggest that abundance levels of these salamanders are more likely a function of the availability of woody debris for cover than age of the over- story. Wet old-growth stands in southern Washington, however, ap- parently provide low quality habitat for these plethodontids, especially ensatinas (table 1, fig. 2). Soils in these stands were often saturated INTACT MODERATELY DECOMPOSED DECAY CLASS DECOI^OSED Figure 5.— Use of logs by Ensatina esc/)scho//z//(E NES) and Plethodon vehiculum (PLVE) by decay class In the southern Washington Cascade Range. with water, and such conditions may reduce the availability of microenvi- ronments suitable for cover, mainte- nance of water balance, and success- ful reproduction. In addition, these < 10 CM FINE WOODY DEBRIS 10 -30 CM > 30 CM COARSE WOODY DEBRIS Figure 4.— Use of logs by Ensatina eschscholtzii i£N£S) and Plethodon vehiculum (PLVE) by dianneter class in the southern Washington Cascade Range. Stands were located in topographi- cally low sites where cold air accu- mulates, which may create unfavor- able microclimatic conditions for ple- thodonhd salamanders. Our results also suggest that plethodontid sala- manders may prefer certain types of woody debris as cover, especially those associated with large, moder- ately to well-decomposed snags and logs. Captures of ensatinas were most common under pieces of bark, especially in bark piles at the base of well- decayed snags (fig. 3). Snags in the early stages of decomposition with shallow or no bark piles at their bases provide few suitable mi- crohabitats for salamanders. Depth of these bark piles increases as sloughing continues until all bark has fallen off. Later stages of snag de- composition provide no additional bark to the pile and habitable spaces become compressed as the lower lay- ers of bark decay and mix with the underlying substrate. Bark microhabitats formed by the deterioration of snags differ in struc- ture from those formed by the de- 35 composition of logs. As logs decay, a single layer of bark is deposited on the forest floor, whereas bark slough- ing from snags forms multilayered, structurally complex cover. Such bark piles could provide microcli- matic conditions more resistant to fluctuations in temperature and moisture than those found under bark on the ground. Additional for- aging habitat may also be available. Redback salamanders, on the other hand, were most often found under moderately decayed logs 10-30 cm in diameter (figs. 3-5). In the early stages of decay, bark has not begun to slough and branches suspend the log above the ground. As the bark begins to slough and branches dete- riorate, increased cover and moisture are provided along the length of the bole where it comes in contact with the forest floor (Maser and Trappe 1984). The quality of this environ- ment for salamanders continues to improve with further decay until the organic matter becomes incorporated into the underlying substrate and habitable interstices become com- pressed in the advanced stages of decomposition. All known nest sites of ensatinas in the Pacific Northwest have been found in association with large, mod- erately decayed logs (Norman and Norman 1980, Maser and Trappe 1984, Jones and Aubry 1985, Norman 1986, L. L. C. Jones unpubl. data). This habitat feature may be impor- tant for the persistence of ensatinas in these forests. We do not know to what extent coarse woody debris may be important for reproduction of redback salamanders in Douglas- fir forests; only one nest site has been found, and this was in moist talus in the Oregon Coast Range (Hanlin et al. 1978). In Douglas-fir stands of the Cas- cade Range that have regenerated after catastrophic fires, levels of coarse woody debris (CWD) (logs and snags > 10 cm in diameter) are moderate in young stands, lowest in mature stands, and highest in old- growth stands (Spies et al. in press). In general, this is due to the inheri- tance of high levels of CWD in young stands from the preceding old- growth stands, a low accumulation of CWD in mature stands as CWD decays but inputs are low, and high inputs of CWD in older stands as the large Douglas-firs die and accumu- late as snags and logs. Intensive for- est management results in levels of CWD substantially lower than that encountered in unmanaged forests (Spies and Cline in press). This is be- cause plantations inherit little CWD from the preceding stand when it is clearcut and existing CWD is re- moved and fragmented. In addition, thinning operations reduce the input of CWD from suppression mortality and short rotations prevent the accu- mulation of CWD. Maintaining even moderate amounts of CWD in man- aged forests will require modifica- tions of current harvesting and silvicultural practices (Harmon et al. 1986, Spies et al. in press). Virtually all available cover ob- jects we encountered were woody debris, and both species were found most often in association with large, moderately decayed logs and snags. Our results suggest that the availabil- ity of coarse woody debris may be important for maintaining salaman- der populations in Douglas-fir for- ests. Additional studies of terrestrial salamanders in managed vs. unman- aged forests are necessary to deter- mine the extent to which they may be affected by intensive forest manage- ment. In general, our study yielded a relatively low number of captures. Only two common species (Nuss- baum et al. 1983) were captured in high enough numbers to permit analyses of the data; captures of all other species were incidental. The total number of species detected was also low in relation to known species richness: pitfall trapping for approxi- mately 1000 trap nights in each of the same study sites in the fall of 1984 yielded 916 captures of 13 species (K. B. Aubry unpubl. data). Research us- ing time-constrained searches to study all but the most common spe- cies in this region would require sub- stantially more search time. Sam- pling should also be conducted dur- ing all seasons of the year to detect seasonal shifts in habitat selection or cover object use, and to sample spe- cies that are active at other times of the year. Acknowledgements We thank R. W. Lundquist, J. B. Buchanan, B. A. Schrader, A. B. Humphrey, M. Q. Affolter, M. J. Reed, B. F. Aubry, and M. J. Crites for assistance. T. A. Spies at the For- estry Sciences Laboratory, Corvallis, OR provided data on stand charac- teristics. R. W. Lundquist provided the map used in figure 1 . This study was funded under USDA Forest Service Cooperative Agreement PNW-83- 219 to S. D. West and D. A. Manuwal at the Univ. of Washing- ton. We thank personnel of the Gif- ford Pinchot National Forest and Mount Rainier National Park for their cooperation and support. M. G. Raphael, K. E. Severson, T. A. Spies, and A. B. Carey provided construc- tive comments on a previous draft of the manuscript. This paper is Contri- bution No. 65 of the Old-growth For- est Wildlife Habitat Project, USDA Forest Service, Pacific Northwest Re- search Station, Olympia, WA. Literature Cited Bury, R. B. and P. S. Corn. 1988. Douglas-fir forests in the Oregon and Washington Cascades: rela- tionship of herpetofauna to stand age and moisture. This volume. Bury, R. B. and M. G. Raphael. 1983. Inventory methods for amphibians and reptiles, p. 416-419. In]. F. Bell and T. Atterbury, eds. Renewable resource inventories for monitor- ing changes and trends: Proceed- 36 ings of an international conference (Corvallis, Oregon, August 15-19, 1983) Society of American Forest- ers 83-14). 737 p. Campbell, H. W. and S. P. Christ- man. 1982. Field techniques for herpetofaunal community analy- sis, p. 193-200. In N. J. Scott, Jr., ed. Herpetological communities. USDI Fish and Wildlife Service Wildlife Research Report 13. 239 p. Franklin, J. F., K. Cromack, Jr., W. Denison, A. McKee, C. Maser, J. Sedell, F. Swanson, and G. Juday. 1981. Ecological characteristics of old-growth Douglas-fir forests. USDA Forest Service General Technical Report PNW-118, 48 p. Franklin, J. F. and C. T. Dyrness. 1973. Natural vegetation of Ore- gon and Washington. USDA For- est Service General Technical Re- port PNW-8, 417 p. Gutierrez, R. J and A. B. Carey, eds. 1985. Ecology and management of the spotted owl in the Pacific Northwest. USDA Forest Service General Technical Report PNW- 184. 119 p. Hanlin, H. G., J. J. Beatty, and S. W. Hanlin. 1978. A nest site of the western red-backed salamander Plethodon vehiculum. Journal of Herpetology 13(2):212-214. Harmon, M. E., J. F. Franklin, F. J. Swanson, P. Sollins, S. V. Gregory, J. D. Lattin, N. H. Anderson, S. P. Cline, N. G. Aumen, J. R. Sedell, G. W. Lienkaemper, K. Cromack, Jr., and K. W. Cummins. 1986. Ecology of coarse woody debris in temperate ecosystems. In A. Mac- Fadyen and E. D. Ford, eds. Ad- vances in Ecological Research 15:133-302. Jones, L. L. C. and K. B. Aubry. 1985. Ensatina eschscholtzii (Oregon en- satina). Reproduction. Herpeto- logical Review 16(1):26. Lumen, I. D. and W. A. Neitro. 1980. Preservation of mature forest serai stages to provide wildlife habitat diversity. Transactions of the North American Wildlife and Natural Resources Conference 43:78-88. Maser, C, R. G. Anderson, K. Cromack, Jr., J. T. Williams, and R. E. Martin. 1979. Dead and down woody material, p. 78-95. In J. W. Thomas, ed. Wildlife habitats in managed forests- the Blue Moun- tains of Oregon and Washington. USDA Forest Service Agricultural Handbook 553, 512 p. Maser, C. and J. M. Trappe. 1984. The seen and unseen world of the fallen tree. USDA Forest Service General Technical Report PNW- 164, 56 p. Meehan, W. R., G. R. Merrell, Jr., and T. A. Hanley, eds. 1984. Fish and wildlife relationships in old- growth forests: Proceedings of a symposium. (American Institute of Fishery Research Biologists, Juneau, Alaska, April 1982). 425 p. Meslow, E. C, C. Maser, and J. Vemer. 1981. Old-growth forests as wildlife habitat. Transactions of the North American Wildlife and Natural Resources Conference 46:329-335. Norman, B. R. 1986. Ensatina es- chscholtzii oregonensis (Oregon en- satina). Reproduction. Herpeto- logical Review 17(4):89. Norman, C. E. and B. R. Norman. 1980. Notes on the egg clusters and hatchlings of Ensatina es- chscholtzi oregonensis. Bulletin of the Chicago Herpetological Soci- ety 15(4):99-100. Nussbaum, R. A., E. D. Brodie, Jr., and R. M. Storm. 1983. Amphibi- ans and reptiles of the Pacific Northwest. 332 p. University Press of Idaho, Moscow. Raphael, M. G. 1984. Wildlife diver- sity and abundance in relation to stand age and area in Douglas-fir forests of northwestern California, p. 259-274. In W. R. Meehan, T. R. Merrell, Jr., and T. A. Hanley, eds. Fish and wildlife relationships in old-growth forests: Proceedings of a symposium. (American Institute of Fishery Research Biologists, Juneau, Alaska, April 1982). 419 p. Raphael, M. G. and R. H. Barrett. 1984. Diversity and abundance of wildlife in late successional Douglas-fir forests, p. 352-360. In New Forests for a changing world. (Society of American Foresters Convention, Portland, Oregon, 1983). Ruggiero, L. F. and A. B. Carey. 1984. A programmatic approach to the study of old-growth forest - wild- life relationships, p. 340-345. In New forests for a changing world. (Society of American Foresters Convention, Portland, Oregon, 1983). Spies, T. A. and S. P. Cline. In Press. Coarse woody debris in manipu- lated and unmanipulated coastal Oregon forests. In C. Maser, R. F. Tarrant, J. M. Trappe, and J. F. Franklin, eds. From the forest to the ocean — a story of fallen trees. USDA Forest Service General Technical Rep>ort. Spies, T. A., J. F. Franklin, and T. B. Thomas. In Press. Coarse woody debris in Douglas-fir forests of western Oregon and Washington. Ecology. Thomas, J. W., R. G. An- derson, C. Maser, and E. L. Bull. 1979. Snags. In J. W. Thomas, ed. Wildlife habitats in managed for- ests - the Blue Mountains of Ore- gon and Washington. USDA For- est Service Agricultural Handbook 553. 512 p. Welsh, H. H., Jr. 1988. Old growth forests and the distribution of the terrestrial herpetofauna. This vol- ume. 37 Forestry Operations and Terrestrial Salamanders: Techniques in a Study of the Cow Knob Salamander, Plethodon punctatus^ Kurt A. Buhlmann,2 Christopher A. Pague,^ Joseph C. Mitchell,^ and Robert B. Glasgow^ Abstract.— The status and ecology of Plethodon punctaf us \^/as investigated in George Washington National Forest, Virginia to determine potential effects of logging. Pitfall traps and mark-recapture supplemented searching by hand. Elevation, aspect, soil characteristics, and number of cover objects (rocks) ore the most important features that identify P. panc^ofus habitat. Intensive logging operations appear to be detrimental to this species. Increasing emphasis is being placed on conservation and preservation of biological diversity worldwide (Norse et al., 1986; Wilson, 1988). U.S. federal and state agencies have become concerned about the bio- diversity of their managed lands and are directing efforts towards preserv- ing natural biota. From a manage- ment perspective, research on am- phibians and reptiles lags behind that devoted to game animals, such as some mammals, birds, and fish (Bury et al., 1980). This is partly due to a previous lack of interest in these groups, but also because some spe- cies can be more difficult to observe or investigate. The Cow Knob salamander, Ple- thodon punctatus, is a dark, moder- ately large (to 74 mm snout-vent length), woodland, fossorial amphib- ian (Martof et al., 1982) found only 'Paper presented at symposium. Man- agement of Amphibians, Reptiles, and Small Mammals in Nortti America. (Flag- staff, AZ, July 19-21, 1988). ^ Kurt A. Buhlmann is a consulting biolo- gist for tlie U.S. Forest Service. Buhlmann's current address is 2001 N. Main St., Blacksburg, VA 24060. ^Christophier A. Pague is a doctoral can- didate in ttie Ecological Sciences Program, Department of Biological Sciences, Old Dominion University, Norfolk, VA 23508. ^Josept) C. Mitchiell is a Research) Biolo- gist, Department of Biology, University of Richmond, Richmond, VA 23173. ^Robert B. Glasgow is the Wildlife Biolo- gist for the George Washington National Forest, Harrison Plaza, P.O. Box 233, Harri- sonburg, VA 22801. on Shenandoah and North Mountain of western Virginia and eastern West Virginia (Highton, 1972, Tobey, 1985). Most of the known range of this recently described species (Highton, 1972) is in the George Washington National Forest. Fraser (1976) compared some aspects of the ecology of this species with a sympa- tric congener, Plethodon hoffmani. Little else is known of the ecology of this salamander. Because of its rela- tively small range and unknown status, P. punctatus was added to the U.S. Fish and Wildhfe Service's Cate- gory 2 list (U.S. Fish and Wildlife Service, 1985). Potential timber har- vesting within the range of this spe- cies (USD A Forest Service, 1986) prompted us to examine its status in forest stands of various ages. In this paper we report the following as- pects of this study: techniques of cap- ture and data collection, salamander habitat characteristics, and potential effects of logging operations. Our ob- jective in this paper is to make other researchers aware of the techniques we used and the problems we en- countered in developing useful man- agement recommendations for the protection of an apparently rare ter- restrial salamander. Materials and Methiods We conducted this study on Shenan- doah Mountain, Augusta and Rock- ingham Counties, George Washing- ton National Forest, Virginia. Before its purchase, between 1911 and 1940, by the U.S. government, this area was repeatedly logged and burned (Leichter, 1987; original land deed documents). Few virgin stands of forest remain, and regrowth and log- ging operations has resulted in a mo- saic of mixed hardwoods of various ages. We selected five sites of different aged forest to determine the relative abundance of Plethodon punctatus (fig. 1) to see if its presence was affected by logging. All sites selected had similar aspects (S-SE) and elevation (914- 1127 m) (table 1). We used USDA Forest Service compartment descriptions and maps to aid in se- lection of sites and to obtain informa- tion on the history, physical and bio- logical descriptions, and future man- agement goals for each site. A com- partment is divided into a series of stands, each of which defines a for- ested area of similar tree species by composition, age, and stand condi- tion. Stand age is defined by the age of dominant canopy trees. Final choices of sites were made only after each was checked in the field and tree age was verified by tree ring analysis. In each site we erected drift fence arrays (Campbell and Christman, 1982) consisting of four 60 cm x 7.5 m secHons of aluminum flashing ar- ranged in a cross pattern. Opposite arms of the cross were separated by 15 m and all sections were sunk in 38 the ground approximately 10 cm. A 5-gal plastic bucket was placed in the center of each arm and #10 cans were placed in the ground on either side of the ends of each arm so that the tops of the pitfalls were flush with the ground surface. Sites A and C con- tained two drift fence arrays, and the remaining three sites had one array each. In each pitfall we put 4-10 cm of 10% formal-dehyde solution to in- sure adequate preservation of the salamanders. We selected this method to obtain samples of all the terrestrial fauna for a range of stud- ies on reproductive cycles and ecol- ogy. Pitfalls were checked and all captures (including other vertebrates and all invertebrates) were collected weekly May 5 - June 18, 1987, bi- weekly July 7 - November 22, 1987 and monthly December 1987 and January 1988. Samples were sorted in the laboratory and the vertebrates stored in 10% neutral buffered for- maldehyde. Invertebrates were stored in 70% isopropanol. Hand-collecting supplemented drift fence collection and was used to determine the elevational range of P. Figure 1 .—Plethodon punctatus, the Cow Knob salamander, from Shenandoah Mountain, Augusta County, Virginia. Photograph by Kurt A. Buhlmann. punctatus and to obtain information on range and habitat characteristics. Results from timed collecting periods allowed comparison among sites and dates of collection. Between April 20 and June 2, 1987, we collected data on eleven microhabitat variables at 67 sites to evaluate those most im- portant in predicting the presence of this salamander. These variables were elevation, aspect, slope, soil temperature under cover object, soil moisture, soil pH, soil description, canopy cover, number of cover ob- jects available within a 2 m circle, type of cover object (e.g., rock, log), and forest type. One site >1 km away from any of the collection sites was selected for estimation of population size and data on individual movements. We searched for salamanders in daylight by turning and replacing all surface objects and at night while they were active on the surface (i.e., during conditions of near 100% relative hu- midity [sensu Heatwole, 1960; Jaeger 1978]). Each individual was meas- ured (snout-vent length, tail length to nearest mm), weighed (nearest 0.1 g), the sex determined, assigned to adult or juvenile age-classes, marked by toe-clipping, and released at its cap- ture site. We marked each capture site with survey flags on which the salamander's number and capture Table 1.— Descriptions of drift fence study sites for Plethodon punctatus on ShenarKloah Mountain, Virginia. Slope angle is in degrees and site age is in years since last logging activity. Site Timber descrip. Slope Manag. type Site age Stand condition Past logging history A 1 yr old white pine several hardwood seed trees 30 white pine 2 seedling/ sapling 90% clearcut few hardwood trees B white oak/ red oak/hickory . 45 oak/hickory 8 sparse saw timber thinned due to ice damage, 1979 C white pine/mixed 25 white pine 30 immature cut in 1956, planted in hardwoods pole timber white pine, some hardwood seed trees D white oak/hickory 30 oak/hickory 60-100 mature saw timber no recent management E white oak/ red oak/ hickory 5 none virgin? low quality saw timber none known 39 date were written. We noted all re- captures and measured movements in linear fashion (0.1 m) between cap- ture points. Results and Discussion Capture Tectiniques Nineteen P. punctatus were caught in the pitfall traps, 2.0% of the total number of salamanders. Of the 17 recorded, 12 were caught in 5-gal. buckets and 5 in #10 cans. Eleven P. punctatus were caught in Site E, six in Site B, and two Site D. None were caught in Site A or the Site C. In con- trast, by hand collecting in areas ad- jacent to Site E, we found 38 P. punc- tatus in 7.7 man hours of searching. The drift fence method appears only moderately effective in sampling this salamander. It is feasible that P. punc- tatus is less likely to fall into the pit- falls than other salamander species. We observed several individuals climbing rocks and tree trunks dur- ing nocturnal surface activity. This suggests that this salamander is able to detect precipices and avoid falling into pitfalls. Also, this species may be active on the horizontal surface only for limited periods of time and under specific environmental conditions. Thus, the drift fence technique, which depends on horizontal activ- ity, may not be an effective sampling method for this salamander (R.D. Semlitsch, pers. comm.). Data from pitfall traps, combined with data from hand collecting, can provide information for management decisions. For instance, seasonal trends in surface activity were simi- larly indicated by both drift fence re- sults (fig. 2) and captures based on hand collection (fig. 3). Comparison of P. punctatus with that of its sympa- tric congener P. cinereus (fig. 3) re- veals concordance in seasonal activ- ity and suggests similar responses to surface environmental conditions. This information could be used to determine the times logging opera- MAY JUN JUL AUG SEP OCT NOV DEC JAN Figure 2. —Seasonality of drift fence captures of Plethodon cinereus and P. punctatus at Site E (Tomat>awl< Mountain), George Washington National Forest. Adults and juveniles are in- cluded, but not tiatctilings. Sampling period is 5 May 1987 to 24 January 1988. 20 APR 22 MAY 5 MAY 24 JUL 6 JUL 21 AUG 31 SEP 28 OCT 12 NOV 22 Figure 3.— Seasonality of captures per man hiour of Plett)odon cinereus and P. punctatus on Tomatiawk Mountain, George Wastiington National Forest. Black bars represent P. punctatus and bars witli diagonal lines represent P. cinereus. Sampling dates are 22 April to 22 Novem- ber 1987. 40 tions would cause the least impact on salamanders at or near the surface. The benefits of the drift fence tech- nique outweighed the low numbers of captures of P. punctatus. We probably would not have otherwise found this species in Site D because there were few surface rocks to turn over. Although the individuals caught may have been transients, this species does occasionally occur at this site. This result would not have been obtained by hand-collecting alone. The drift fence method also pro- vided estimates of the relative abun- dance of the salamander fauna and other species in the community. The relative numbers of these species and species groups can generate addi- tional information on the structure of the community in which the focal species lives. Drift fence techniques have been used for a variety of eco- logical studies (e.g.. Gibbons, 1970; Gill, 1978; Pechmann and Semlitsch, 1986) but only recently to answer questions about vertebrate communi- ties in relation to forest management (e.g., Bennett et al., 1980; Gibbons and Semlitsch, 1981; Enge and Mar- Table 2.— Seasonal differences in surface abundance of Plethodon punctatus at Flagpole Knob and Skidmore Tract, Shenandoah f^ountain, George Washington For- est. These sites are <1 km apart. Flagpole Knob is a rocky, grassy ridge habitat containing young oak (Quercus sp.) and maple (Acer sp.) pole timber, and Skid- more is a virgin hemlock (Tsuga canadensis)/ye\\ow birch (Betula lutea) forest. Numbers of salaman- ders are followed by number of man hours In parentheses. All data are based on hand-collecting re- sults. Date Flagpole Skidmore June 2 11 (0,5) 10 (1.7) June 8 0 (0.5) 2 (2.0) Sept. 28 0 (1.0) 3 (3.0) V J ion, 1986; Bury and Com, 1987). Our results indicate this technique can be effective in mountainous terrain and can be used to gain information on apparently rare terrestrial salaman- ders. If an endangered or otherwise protected species is the focus of study and cannot be collected, then slight modifications of the drift fence-pitfall design must be made. Traps would need to be checked on a daily basis, or nearly so, in order to release the animals unharmed (Gib- bons and Semlitsch, 1981). Water or wet leaves can be placed in the pit- falls for cover and moisture. Poten- tial problems include killing of the salamanders in the pitfalls by small mammals, especially shrews, and desiccation. The loss of animals by shrew or raccoon predation in pitfall containers affects the samples and may prevent quantitative compari- sons among sites. Data obtained from visitation frequencies of every three days (Bury and Corn, 1987) to once a week (Enge and Marion, 1986) probably underestimate actual cap- tures. The detection of P. punctatus at a particular site depends on the time of year, substrate type, soil depth, soil moisture, soil temperature, and weather conditions (see Habitat Re- quirements). A simple survey of sites by hand searching and rock turning in daylight hours without attention to weather and seasonality will un- derestimate actual abundance and fail to detect presence of a species. Table 2 contains comparative data for two sites searched the same day at different times of the year and demonstrates a strong seasonal ef- fect. In order to construct effective management plans, the range and abundance of a terrestrial salaman- der must be known. Therefore, re- searchers conducting distributional surveys must take seasonal and diet changes in surface activity into con- sideration. Results of our 1987 mark-recap- ture efforts are preliminary; only four recaptures were made. One P. punctatus captured 28 May was re- captured on 15 October. It had moved 17.4 m. Three salamanders were recaptured within ten days of original capture and had moved < 2 m. Knowledge of movement capabili- ties by P. punctatus is an important part of evaluating the consequences of population fragmentation through logging operations. Are salamanders able to move out of a logged area or repopulate it when suitable habitat conditions return? We believe mark- recapture studies can provide useful information on rare terrestrial sala- manders, but realize that data may need to be collected over several years and under standardized condi- tions in order to provide direct an- swers. Habitat Characteristics Preliminary evaluation of microhabi- tat data indicate that four site charac- teristics are most important in deter- mining the presence of P. punctatus. We found P. punctatus at elevations between 732 m and 1317 m (fig. 4). Most sites (87%) with this species oc- curred above 960 m. Plethodon punc- tatus occurred on all slopes but were more common on north-facing as- pects (87% of 11 sites) than east (38% of 13), south (36% of 8), or west as- pects (40% of 7). Most of the captures (67% of 21) were on slopes of 20-45°. Seven sites were on slopes less than 20° and between 46° and 60°. Sites without this salamander were on a similar range of slopes (< 20°, 28.6%; 20-45°, 57.1%; > 45°, 14.3%). Soil temperatures under cover ob- jects at sites with P. punctatus (x = 12.3 C, 9.4-16.1, n = 36) were nearly identical to temperatures at sites without this species (x = 12.8 C, 9.4- 15.8, n = 15). Soil pH under cover ob- jects were also similar (with P. punc- tatus: X = 6.3, 5.4-6.8; without P. punc- tatus: X = 6.4, 5.8-6.8). Average soil moisture at sites with P. punctatus was 37.1% (12- 70%) and 42.8% (24- 41 80%) at sites without this species. Soils in which P. punctatus were found are characterized by shallow black humus intermixed with rocks (72% of 39 sites). One site where eleven salamanders were captured consisted of brown humus and ex- tensive log cover, but few rocks. Cover objects under which this sala- mander was found were rocks < 645 cm2 (13.6%), rocks 645-1290 cm^ (40.0%), rocks > 1290 cm^ (34.8%), and logs (10.6%). Over 89% of the captures were found under rock cover. Number of cover objects within a 2 m circle of the captured salamander averaged 15.1 (1-45). Sites without P. punctatus ranged from 100% rock cover to 0% rock cover. Sites with canopy cover equal to or greater than 50% accounted for 88.2% of the captures (n = 52). We found P. punctatus in the fol- lowing forest types: mature oak/ hickory (38.5% of 13), oak/maple/ birch (62.5% of 8), oak/pine (33.3% of 3), young oak/ maple/ hemlock (50% of 8), virgin hemlock /yellow birch (100% of 2), hemlock/maple/bass- wood (62.5% of 8), white pine (0% of 2), and grassy balds (20% of 5). Of the site characteristics we examined, the following appear to be most im- portant in identifying P. punctatus habitat: elevation, aspect, soil charac- teristics, and number of cover objects (rocks). Habitats of terrestrial salamanders differ among species and, in some cases, among geographic areas within species (e.g., Semlitsch, 1980; Tilley, 1973). Data derived from the literature for management studies and plans must be used with caution. Baseline habitat and life history stud- ies should be conducted on the focal species at the location in question be- fore developing management plans. Effects of Logging Tree removal effects the terrestrial salamander community in several ways. Removal of canopy cover eliminates the moisture-retaining po- tenUal of the soil and leaf litter, al- lows an increase in insolation (with a concomitant increase in soil tempera- tures), and increases soil erosion (Bury, 1983). The use of heavy machinery com- pacts soil and destroys leaf litter. Enge and Marion (1986) found that machine site preparation and clearcutting had little effect on am- phibian species richness in a Florida slash pine forest. However, of the 15 amphibian species they recorded, none was a terrestrial salamander. On Shenandoah Mountain, where most of the terrestrial amphibian community is comprised of terres- trial salamanders, logging and clearcutting are likely to have detri- mental effects. Salamander abun- dance in a 60-100 yr-old deciduous forest in another Virginia site was more than four times that in 2 yr-old and 6-7 yr-old clearcuts (Blymer and McGinnes, 1977). Bury (1983) found that terrestrial salamanders were more abundant in old growth com- pared to logged redwood forest habitats. Plethodon cinereus was sig- nificantly less abundant in a clearcut site compared to an old-growth site in a deciduous forest in New York (Pough et al., 1987). Populations of Plethodon punctatus inhabiting rocky substrates with a thin soil cover may be able to with- stand some logging operations. Our Site B was logged in a salvage opera- tion after ice storm damage. Not all trees were removed and the sub- strate was not as damaged as that in Site A, which was clearcut. These fac- tors, combined with the presence of a seep near the drift fence array, probably explain the high numbers of P. punctatus found at Site B com- pared to other logged sites. We found no P. punctatus on Sites A and C for apparently different rea- sons. Site A was clearcut, the sub- strate was greatly disturbed, and the lack of canopy cover prevented mois- ture retention. The fact that P. punc- tatus occurred on the same ridge in a nearby hardwood stand suggests this salamander may have occurred on Site A prior to logging. Site C was 4(37-610 611-762 763-914 915-1067 1068-1219 1220 +■ Elevation (m) Figure 4.— Elevational distribution of Pleftiodon punctatus on Stienandoati Mountain, George Washiington National Forest. Solid bars represent sites wtiere P. punctatus was not found and bars witti diagonal lines represent sites whiere thiis species was found. 42 logged 30 years ago but was re- planted with white pine (table 1). The logging operation and change in vegetation type may have affected the salamander populations previ- ously present. However, because of the lack of rocky substrate, we can- not disprove the hypothesis that P. punctatus may not have occurred there historically. Plethodon punctatus appears to oc- cur in greatest abundance on rocky sites that contain virgin hardwoods (Site E) and sites that are not heavily disturbed by logging operations (Site B). Clearcutting and associated dis- turbance does appear to eliminate populations of this salamander. Sala- mander mortality can be minimized if logging operations are conducted outside the seasonal activity period. If size of the area logged is small, or if the area is logged in a mosaic, or if corridors are allowed to remain, rein- vasion may eventually be possible from peripheral populations when suitable conditions return. Fragmen- tation of the limited range of P. punc- tatus by a patchwork of clearcuts could seriously affect its long-term survival. Conclusions Because of budget and time con- straints, our study attempted to ob- tain baseline data and evaluate the effects of logging simultaneously. We offer the following conclusions to re- searchers and managers who must study a salamander whose ecology is little known. 1 . Multiple capture techniques should be used when study- ing an apparently rare terres- trial salamander. 2. The life history and basic ecology of the study species needs to be understood be- fore the project^ s experimen- tal design can be erected to evaluate logging effects. 3. Seasonal and daily activity patterns of salamander activ- ity must be taken into con- sideration when surveys are conducted to determine range and population abun- dance. 4. Project proposals to federal and state agencies should contain a two step process, a field survey phase to obtain baseline data on ecology and life history and an experi- mental phase in which log- ging or other concerns are evaluated. The design of the experimental phase should be based on the results of the field survey. Acknowledgments We are grateful to the following people for field assistance: Christian A. Buhlmann, Kara S. Buhlmann, Lana C. Buhlmann, Susan J. Fortuna, Joshua C. Mitchell, and Scott M. Smith. David A. Young deserves spe- cial thanks for helping install the drift fence arrays. The rangers and foresters at the Dry River District provided logistical support, equip- ment, and help with site selection. This study was supported by a Chal- lenge Cost Share from the U.S. Forest Service. Additional support was pro- vided by the Nongame Wildlife and Endangered Species Program of the Virginia Department of Game and Inland Fisheries. Literature Cited Bennett, Stephen H., J. Whitfield Gib- bons, and Jill Glanville. 1980. Ter- restrial activity, abundance and diversity of amphibians in differ- ently managed forest types. American Midland Naturalist 103:412-416. Blymer, Michael J., and Burd S. McGinnes. 1977. Observations on possible detrimental effects of clearcutting on terrestrial amphibi- ans. Bulletin of the Maryland Her- petological Society 13:79-83. Bury, R. Bruce. 1983. Differences in amphibian populahons in logged and old growth redwood forest. Northwest Science 57:167-178. Bury, R. Bruce, Howard W. Campbell, and Norman J. Scott, Jr. 1980. Role and importance of nongame wildlife. Transactions of the 45th North American Wildlife and Natural Resources Confer- ence, 45:197- 207. Bury, R. Bruce, and Paul Stephen Corn. 1987. Evaluation of pitfall trapping in northwestern forests: trap arrays with drift fences. Jour- nal of Wildlife Management 51:112-119. Campbell, Howard W., and Stephen P. Christman. 1982. Field tech- niques for herpetofaunal commu- nity analysis, p.193-200. In Herpe- tological Communities. N. J. Scott, Jr., editor. U.S. Fish and Wildlife Service, Wildlife Research Report 13. Enge, Kevin M., and Wayne R. Mar- ion. 1986. Effects of clearcutting and site preparation on herpe- tofauna of a north Rorida flat- woods. Forest Ecology and Man- agement 14:177- 192. Eraser, Douglas F. 1976. Coexistence of salamanders in the genus Ple- thodon: A variation on the Santa Rosalia theme. Ecology 57:238-251. Gibbons, J. Whitfield. 1970. Terres- trial activity and the population dynamics of aquatic turtles. American Midland Naturalist 83:404-414. Gibbons, J. Whitfield, and Raymond D. Semlitsch. 1981. Terrestrial drift fences with pitfall traps: an effec- tive technique for quantitative sampling of animal populations. Brimleyana 7:1-16. Gill, Douglas E. 1978. The metapopu- lation ecology of the red-spotted newt, Notophthalmus viridescens (Rafinesque). Ecological Mono- graphs 48:145-166. 43 Heatwole, Harold. 1960. Burrowing ability and behavioral responses to desiccation of the salamander, Ple- thodon cinereus. Ecology 41:661- 668. Highton, Richard. 1972. Distribu- tional interactions among eastern North American salamanders of the genus Plethodon. p. 139- 188. In The Distributional History of the Biota of the Southern Appalachian Mountains. Part III: Vertebrates. Virginia Polytechnic Institute and State University, Research Divi- sion Monographs 4. Jaeger, Robert G. 1978. Plant climb- ing ability by salamanders: peri- odic availability of plant- dwelling prey. Copeia 1978:686- 691. Leichter, Bill. 1987. The land nobody wanted. Highlands of the Vir- ginia's Magazine, Autumn 1987/ Winter 1988, pp. 3-7. Martof, Bernard S., William M. Palmer, Joseph R. Bailey, and Jul- ian R. Harrison, III. 1980. Am- phibians and Reptiles of the Caro- linas and Virginia. University of North Carolina Press, Chapel Hill, 264 p. Norse, Elliott A., Kenneth L. Rosen- baum, David S. Wilcove, Bruce A. Wilcox, William H. Romme, David W. Johnston, and Martha L. Stout. 1986. Conserving Biological Diversity in our National Forests. The Wilderness Society, Washing- ton, D.C., 116 pp. Pechmann, Joseph H. K., and Ray- mond D. Semlitsch. 1986. Diel ac- tivity patterns in the breeding mi- grations of winter-breeding anu- rans. Canadian Journal of Zoology 64:1116-1120. Pough, P. Harvey, Ellen M. Smith, Donald H. Rhodes, and Andres Collazo. 1987. The abundance of salamanders in forest stands with different histories of disturbance. Forest Ecology and Management 20:1-9. Semlitsch, Raymond D. 1980. Geo- graphic and local variation in population parameters of the slimy salamander Plethodon gluti- nosus. Herpetologica 36:6-16. Tilley, Stephen G. 1973. Life histories and natural selection in popula- tions of the salamander Desmognathus ochrophaeus. Ecology 54:3-17. Tobey, Franklin J. 1985. Virginia's amphibians and reptiles, a dis- tributional survey. Privately pub- lished, Virginia Herpetological So- ciety, 114 p. USDA Forest Service. 1986. Final land and resource management plan for the George Washington National Forest. U.S. Department of Agriculture. U.S. Fish and Wildlife Service. 1985. Endangered and threatened wild- life and plants: review of verte- brate wildlife. Federal Register 50:37958-37967. Wilson, Edward 0. 1988. Biodiver- sity. National Academy Press. Washington, D.C. 44 Conserving Genetically Distinctive Populations: The Case of the Huachuca Tiger Salamander (Ambystoma tigrinum stebbinsi Lowey James P. Collins,^ Thomas R. Jones,^ and Howard J. Berna^ Abstract.— Huachuca tiger salamanders are a genetically distinctive race of Ambystoma tigrinum found only in 1 7 localities in the San Rafael Valley (SRV) in southeastern AZ. Populations of SRV salamanders are threatened by introduction of exotic fishes and disease. Salamanders were largely eliminated from four habitats after introduction of sunfish and/or catfish. An unknown fatal disease killed all aquatic morphs in two other habitats. An additional threat includes possible hybridization and introgression of SRV populations resulting from introduction of exotic salamanders. Introduced bullfrogs may also prey on salamanders, or act as vectors for disease. Technological advances in genetics now enable characterization of vari- ation within a species at increasingly finer levels of description. These de- velopments are allowing us to begin the difficult task of identifying which gene pools should be protected to preserve genetic attributes significant for conserving present and future generations of a species (Echelle 1988, Meffe and Vrijenhoek 1988, Ryman and Utter 1987). Rather than considering simply which species to conserve, we can now ask whether a conservation effort should be di- rected at the species, subspecies, or population levels (Allendorf and Leary 1988, Behnke 1972, Ryder 1986). Tiger salamanders, Ambystoma ti- grinum Green, range throughout much of North America from south- ern Canada to the central Mexican Plateau, and from the east coast of the United States to California (Gehlbach 1967). This complex spe- cies is divided into eight subspecies (Collins et al. 1980, Gehlbach 1967, ' Paper presented of symposium. Man- agement of Amphibians, Reptiles, and Small Mammals in Nortt) America. (Flag- staff, AZ, July 19-21, 1988). 'James P. Collins is Associate Professor of Zoology. Department of Zoology, Arizona State University, Tempe, AZ 85287-1501 . ^Ihomas R. Jones and Howard J. Berna are Graduate Students, Department of Zo- ology, Arizona State University, Tempe, AZ 85287-1501. Jones et al. 1988), several of which (tigrinum, mavortium, nebulosum, and melanostictum) are widespread geo- graphically, locally abundant, and apparently not in need of protection at this time. More information is needed on the Mexican subspecies, velasci, and a north-central USA race, diaboli, before conservation needs can be confidently assessed. Two races need consideration now. A. t. californiense occurs only in the Central Valley and adjacent oak woodlands of California, placing it among the more geographically re- stricted tiger salamanders. Further, A. t. californiense appears to have been isolated from the other races of A. tigrinum for several million years, and has a level of genetic divergence equalling species-level differences among many ambystomatid taxa (Jones 1988). Two factors suggest this taxon warrants special conservation efforts. First, California populations are as distinct genetically from other races of tiger salamanders, as other species of Ambystoma are from each other. Second, the geographic isola- tion and apparent spatial subdivision of A. t. californiense populations (Gehlbach 1967) likely increases their probability of extinction (Soule 1987). A. t. stebbinsi has properties like A. t. californiense, suggesting it too needs special conservation efforts despite being classified as only part of a very wide-ranging species. Populations of A. t. stebbinsi occur only in the San Rafael Valley (SRV) in the border- lands between Arizona and Sonora, Mexico. In addition to being geo- graphically restricted, the race is also genetically distinctive. Average heterozygosity among SRV popula- tions is the lowest in Ambystoma (Jones et al. 1988). Electrophoretic analysis, as well as variation in exter- nal morphology, indicates A. t. steb- binsi is phylogenetically most closely related to A. t. mavortium. In contrast, analysis of the mitochondrial DNA (mtDNA) in these populations indi- cates there is a single mitochondrial clone in the San Rafael Valley. This clone is derived from A. t. nebulosum, not A. t. mavortium, suggesting A. t. stebbinsi actually arose from hybridi- zation between A. t. nebulosum and A. t. mavortium (Collins 1988). A recent paper describes patterns of variation in external morphology, allozymes, and geographic isolation that suggest A. t. stebbinsi is a distinc- tive race within the A. tigrinum com- plex (Jones et al. 1988). In a future paper we will describe mitochondrial DNA variation in these populations (Collins et al., in prep.). Our present goal is to summarize several aspects of the population biology of A. t. steb- binsi. In addition to being restricted geographically, our research indi- cates salamander populations in SRV are threatened by several factors in- cluding disease, and factors sur- rounding the introduction of exotic fishes and salamanders. 45 Materials and Methods SRV is a Plains grassland-Madrean evergreen woodland habitat extend- ing from southeastern AZ into north- eastern Sonora (Brown 1982). A sur- vey of aquatic habitats in southern AZ and northern Sonora and Chi- huahua indicated salamanders refer- able to A. i. stebbinsi occurred only in SRV (Jones et al. 1988). From June 1979 to February 1988, we sampled seven natural and 23 man-made or man-altered aquatic habitats in SRV and adjacent slopes of the Patagonia and Huachuca mountains. Altered habitats were primarily livestock watering tanks constructed where natural water for- merly existed. Bog Hole tank is a large, impounded cienega (sensu Hendrickson and Minckley 1985), and another may be an impounded spring. Salamanders occurred in only 17 of the 30 habitats sampled in SRV (appendix 1, fig. 1). We report life history variation in A. t. stebbinsi, and the influence of disease and intro- duced exotic animals on this taxon. For describing life history variation we emphasize four tanks (Parker Canyon #1, Huachuca, Upper 13, and Bodie Canyon) sampled routinely. We also present additional informa- tion from irregular collections at all other SRV tanks with salamanders. We usually collected specimens using seines and dipnets, but occa- sionally used gill nets. Depending on our plans for using a particular col- lection, we either marked and re- leased salamanders, returned them to the laboratory alive, or preserved them in the field for later analysis. All preserved specimens are in the Lower Vertebrate Collections at Ari- zona State University. To summarize life history vari- ation in A. t. stebbinsi, we classified salamanders by life history stage and morphology using internal and exter- nal characters (table 1). Stages 1 or 2 were immature, and 3-5 were ma- ture. Metamorphosed salamanders lack gills and a caudal fin, while lar- vae and mature branchiate salaman- ders have those structures. All meas- urements are in mm; snout-vent length (SVL) is the distance from snout to posterior margin of the vent. Results Life History Variation Ambystoma tigrinum has the most complicated pattern of morphologi- cal and life history variation known in salamanders. After an egg hatches a larva begins growing in an aquatic habitat. At about 30 mm SVL, larvae of A. t. nebulosum, A. t. mavortium, or A. t. tigrinum can continue develop- ment as a typical larva, or develop as a cannibalistic larval morph. This dimorphism is unknown in the other subspecies (Collins et al. 1980). At about 70 mm SVL, larvae of all sub- species except A. t. californiense con- tinue developing in one of two ways. They may metamorphose, often leave the aquatic habitat, and must eventu- ally return to freshwater to breed. Alternatively, a larva continues growing beyond 70 mm SVL, ma- tures, and breeds as a larval-like form, or paedomorph (Gould 1977). Thus, depending on the subspecies, a single population might have two juvenile morphs, typical or cannibal, and four adult morphs, typical and cannibal, mature, branchiate morphs or metamorphosed morphs of either type. Relative frequency of each morph varies among populations in a subspecies (Collins 1981, Rose and Armentrout 1976). In SRV, most populations have mature, typical, branchiate morphs as well as mature, typical, metamor- phosed morphs. Judy Tank is one population in which we have col- Figure 1 .—Map of the San Rafael Valley, Arizona. Symbols indicate sampling sites (see ap- pendix 1). Electrophoretic samples were from sites 1-8; mtDNA samples were from sites 1,2,5,6,9; F=sites with fish; D=sites with diseased salamanders; arrow=J.F. Jones Ranch, type locality for stebbinsi. 46 lected no mature, branchiate morphs thus far. Mature, typical, branchiate morphs dominated the SRV popula- tions. From July 1979 to August 1985, we collected more than 1200 mature, branchiate salamanders and only 64 mature, metamorphosed animals. We conservatively estimated popula- tion sizes of mature, branchiate morphs as varying from 50 (Upper 13 Reservoir, 1984) to several hundred (Huachuca Tank, 1983, 1984). No population in the SRV had cannibalistic morphs. Absence of the cannibal morph is a distinctive fea- ture of these populations, since the morph can be common in A. t. nebu- losum and A. t. mavortium, the nearest relatives of A. t. stehbinsi. Salamanders in SRV bred as early as mid-February and as late as early May. Most egg laying occurred from mid-March to late April. Animals hatched within several weeks and grew rapidly, so that larvae <40 mm SVL were often abundant by late spring (tables 2-5). By mid-July, lar- vae were usually about 60 mm SVL, and those that metamorphosed gen- erally did so from late July to early September. A relatively small per- centage of larvae metamorphose an- nually—about 17% to 40% based on estimates from Bodie Canyon Tank. r Table 1 .—Criteria used to classify salamanders into stages of breeding readiness. Numerals In parentheses refer to diameter in mm (after Collins 1981). Oviduct, ovary, peritoneum, and cloacal characters Wolffian duct, testes, peritoneum, and cloacal characters 1 . Gonadal tissue primarily white and flacid; Wolffian ducts or oviducts narrow, with few folds; cloacal margins not swollen; peritoneum largely unpigmented. 2. Oviducts enlarged (0.5-1), white, weall), convoluted, cream colored with localized black pigment; testes turgid; cloaca! margins swollen , grey to grey-black, rugose borders, especially posterior; peritoneum black, especially densely pig- mented dorsally. 4. Duct large, convoluted, cream colored with scattered black pig- ment spots, distended in coils; tes- tes turgid, enlarged; cloaca and peritoneum as in 3. By early autumn, first year ani- mals that did not metamorphose be- gan to mature (tables 2-5). From late autumn through winter most SRV branchiate salamanders were >100 mm SVL (tables 2-5), and ready to breed (figs. 2-3). These data indicate branchiate salamanders in SRV breed for the first time when one year old. Disease During July and August 1985, all branchiate salamanders in Inez, Huachuca, and Parker Canyon #1 Tanks were killed by an undiagnosed disease (fig 1). Salamanders in the field and laboratory showed little re- sistance to the disease which was 100% fatal within a few days of the appearance of symptoms. Attempts to culture the pathogen(s) were in- conclusive, but many symptoms re- sembled those characteristic of Aero- monas infection [red legl (Fowler 1978), including lethargy, loss of ap- petite, and the epidermis can become red from infusion of blood. This type of epidemic disease in the aquatic environment is particularly devastat- ing in A. t. stebbinsi, because popula- tion structure in SRV is strongly skewed toward larvae and mature branchiate animals. In addition to death of larvae, therefore, most adults may have been killed in highly infected populations. Parker C!anyon #1 and Inez were recolonized by metamorphosed sala- manders that presumably escaped the disease while in terrestrial sites. We collected two metamorphosed adults (male and female) and one larva in Inez Tank in April 1986 and collected eggs in April 1987. We also collected eggs in Parker Canyon #1 in April 1987, and five mature branchi- ate morphs (3 males, 2 females) in January 1988. Since all branchiate morphs in Parker Canyon were killed in 1985, and none was collected in 1986, these five animals also sup- ported our conclusion that in SRV branchiate salamanders can reach 47 sexual maturity when a year old. We collected no salamanders in Huachuca Tank as late as spring 1988 (see below). Introduction of Exotic Animals Fishes. — A few exceptional species of salamanders can coexist with fishes, but most cannot. In SRV exotic fishes, especially centrarchids and ictalurids, invariably eliminate sala- manders. We do not know the effect of native fish on A. t. stebbinsi, but no salamanders occur in four natural SRV habitats (Heron Spring, Sheehy Spring, Sharp Spring, Santa Cruz River and tributaries) that have na- tive fishes (Gila topminnow, Poecili- opsis 0. occidentalis and Gila chub, Gila intermedia). We base our general conclusions concerning exotic fishes and salamanders in SRV on the fol- lowing observations (fig.l). J.F. Jones Ranch Tank. — This is the type locality for A. t. stebbinsi. Largemouth bass (Micropterus salmoi- des) and bluegill (Lepomis macrochirus) were introduced in the 1950s, and salamanders no longer occur here (see photograph of this site in Lowe 1964:106). It is apparently a popular local fishing spot. FS 58 Tank.— We first collected mature, branchiate and larval sala- manders here in July 1979. There were only yellow bullheads (Ameiurus natalis) in June 1980. In August 1984, we collected 19 mature, branchiate salamanders, no catfish, and hundreds of sunfish (Lepomis sp.). Huachuca Tank. — First sampled in May 1982, this tank was a reliable source of salamanders and natural- history information for the next two years. On 22 August 1984 we found one yellow bullhead, plus many lar- val and mature branchiate salaman- ders. On 5 July 1985 we netted >100 salamanders in each of several seine hauls. Routine sampling on 24 Au- gust 1985 yielded several thousand fingerling catfish and no salaman- > Table 2.— Seasonal variation in number and size (SVL) of salamanders in eacti breeding stage collected from Parker Canyon Tank. Snout/Vent Length (mm) Date Stage 0-19 20-39 40-59 60-79 80-89 100-119 120-140 8 Jan 4 5 28-29 Mar 1 13 4 7 2 22-28 Apr 1 4 6 4 39 8 5 4 13-25 Jun 1 3 36 2 1 3 6 9 1 4 1 19 10 5 4 4 14 5-10 Jul 1 1 18 4 7 1 5 1 22 Aug 1 7 2 2 3 7 3 4 9 1 2 Dec 4 10 18 Table 3.— Seasonal variation in number and size (SVl^ of salamanders In each breeding stage collected from Upper 13 Resen/oir. Snout/Vent Length (mm) Date Stage 0-19 20-39 40-59 60-79 80-89 100-119 8 Jan 4 2 17Mar 4 5 7 May 1 n 43 24 Jun- 1 20 79 24 8 9 Jul 2 13 4 1 7 5 2 5 23-28 Aug 1 1 1 2 3 4 n 4 3 9 Oct 1 2 2 4 2 3 1 1 2 4 4 10 Nov 1 1 2 3 1 2 Dec 4 5 48 ders. We resampled this site several times through February 1988. Each time we caught only catfish, although salamanders were abundant in nearby tanks. In this instance disease as well as predation may have contributed to decline of the salamander popula- tion. On 24 August 1985 we found three dead mature, branchiate morphs in the tank. We also ob- served a significant decline in sala- mander p)opulations on this date at two other tanks with diseased sala- manders. Yellow bullheads are highly carnivorous (Minckley 1973), and we do not expect salamanders to successfully recruit at Huachuca Table 4.— Seasonal variation In number and size (SVL) of salamanders In each breeding stage collected from Huachuca Tank. Snout/Vent Length (mm) Date Stage 29 Mar 1 4/5 21 Apr 1 4/5 25 Jun 1 4/5 5 Jul 1 3 4 5 22-28 Aug 1 2 3 4 5 2 Dec 1 4 Stage 0-19 20-39 40-59 60-79 80-89 100-119 120-140 13 46 78 21 24 18 1 9 9 1 2 5 13 1 2 14 2 16 20 1 12 17 8 7 11 1 1 4 Table 5.— Seasonal variation In number and size (SVL) of salamanders in each breeding stage collected from Bodle Canyon Tank. Snout/Vent Length (mm) Date Stage 0-19 20-39 40-59 60-79 80-89 100-119 120-140 28 Mar 1 3 25 Aug 1 2 3 4 26-27 Sep 3 4 11 Nov 3 4 10 74 21 18 3 21 18 8 2 2 10 2 3 2 6 3 1 2 3 1 4 1 Tank as long as the catfish popula- tion remains high. Catfish will pre- sumably eat eggs, larvae, and all but the largest salamanders. We know of no experiments demonstrating the minimum number of catfish that will prohibit salamander reproduction. Bog Hole Tank. — We collected salamanders here in 1979, 1980, and one larva in 1982. Native fishes com- prised longfin dace (Agosia chrysogas- ter) and Gila topminnow. Since the 1970s, several exotic fishes including Gambusia affinis, Cyprinodon macular- ius eremus, Lepomis spp., and Microp- terus salmoides (W.L. Jvlinckley, pers. comm.; Minckley and Brooks 1985) have become established. Disappear- ance of A. t. stebbinsi, and the two na- tive fish species, correlates with es- tablishment of non-native fish popu- lations. Frogs. — During the last decade bullfrogs (Ram catesbeiana) were in- troduced in SRV. Their introduction correlates with reduction in native frog populations in the valley, but the impact of bullfrogs on A. t. steb- binsi is unknown. Bullfrog larvae may eat salamander eggs, while adults may prey on larval salaman- ders. Bullfrogs may also act as vec- tors for disease, since in the three tanks where salamanders were heav- ily affected by disease, bullfrog populations were large and appar- ently unaffected. Frogs may be a natural reservoir for disease, and suf- fer few negative effects from the pathogen(s). Since they disperse readily to colonize surrounding habi- tats, they may also help spread dis- ease among amphibian populations. S alam anders. — Commercial baitdealers (waterdoggers), fisher- men, and private landowners intro- duce native and exotic salamanders into aquatic habitats in Arizona (Collins 1981). Salamanders are used commonly as bait by fishermen in the American Southwest (table 6), and Lowe (1955) first noted that salaman- ders were being introduced into Ari- zona for this purpose. SRV is closed to "waterdog" collecting under Ari- 49 zona Game and Fish Commission order #R1 2-4-311. Enforcement is dif- ficult, because SRV is large and sparsely settled. It would be easy to introduce exotic A. Hgrinum into this valley. Pre-mating and post-mating isolating mechanisms in the A. H- grinum species group within Amby- stoma are weak (Brandon 1972, Nel- son and Humphrey 1972). Introduced A. Hgrinum would be ex- pected, therefore, to easily interbreed with native tiger salamanders. Discussion In theory, average heterozygosity or gene diversity of organisms in an area can be decomposed into gene diversities within and between any subpopulations comprising the total number of organisms in the popula- tion (Nei 1987). If all organisms in a population are a panmictic aggre- gate, then the component describing variation between subpopulations is zero. We have no information on dis- persal between tanks in SRV, so for this discussion we arbitrarily con- sider each tank a subpopulation and together all tanks comprise the total population of SRV salamanders. Within this context our results high- light several factors to consider in trying to understand the evolution- ary genetics of SRV tiger salaman- ders. Mean heterozygosity (.0015) for A. t. stebbinsi is the lowest reported for any salamander (Jones et al. 1988). Salamanders in SRV went through one or more bottlenecks at some point in their history, but cause(s) and time of reduction in numbers and associated genetic diversity are unknown. The effect of a one-time bottleneck is a drastic decrease in ex- pected heterozygosity of a popula- tion, and in theory, repeated bottle- necks could reduce gene diversity even more (Motro and Thomson 1982). Current factors affecting changes in SRV salamander numbers may provide some insight into the origin and /or perhaps maintenance of low gene diversity in SRV. Increased heterozygosity generally correlates positively with traits associated with high individual vigor and fitness, plus population stability (Mitton and Grant 1984). Susceptibility to disease or apparent reduced ability to over- come infection may thus be conse- quences of reduced genetic variation in SRV salamanders. A historical bot- tleneck in population size with asso- ciated loss of gene diversity in SRV salamanders, therefore, could have resulted in populations more suscep- tible to disease. This susceptibility, as seen in contemporary stocktanks, could easily cause severe reductions in numbers of salamanders and re- tard any expected increase in gene diversity. O'Brien et al. (1985) pro- vide a related example. They de- scribe how extremely low genetic variation in the South African chee- tah may derive from a population bottleneck. Low genetic variation seen in structural loci also extends to the major histocompatibility com- plex. This extreme monomorphism correlates with a hypersensitivity in cheetahs to some viral pathogens, and they feel the sensitivity of this genetically uniform species to patho- gens provides an example of the pro- tection against disease genetic vari- ation provides to species. The mecha- nism connecting low genetic vari- ation revealed by electrophoresis and susceptibility to disease is unclear. Hence, for both cheetahs and SRV salamanders it is uncertain if reduc- tion of population size and loss of genetic variation increased suscepti- bility to disease, or alternatively, sus- ceptibility increased for some other reason, and this lead to reductions in population numbers. Two additional factors, again found in present stocktanks, would reinforce this pattern of change in numbers of salamanders and reduc- > 120" 100" 80" 60" 40" 20 2 3 4 Breeding Stage Figure 2.— Variation in SVL with breeding stage for aninnals from four SRV populations: soiid llne=Upper 13 Reservoir, dotted line=Parl50 cm dbh) + LFIR small spruce (<20 cm dbh) medium spruce (20-50 cm dbh) large spruce (>50 cm dbh) MSPRUCE + LSPRUCE aspen (all sizes) non-oak deciduous (all sizes) oak (all sizes) pine (all sizes) snags (all sizes) 56 without salamanders. The Statistical Analysis System computer package (SAS, Version 5) was used for all analyses (SAS Institute Inc 1982). Sig- nificance levels were set at P < 0.05 unless otherwise indicated. Results Jemez Mountains Salanaander Salamanders (N = 28) were present on 10 of 43 transects (23%) with a mean density of 3/100 m^ in occu- pied areas. One hundred twenty salamanders were found in areas off the transects. Transects with sala- manders occurred on significantly steeper slopes and at lower eleva- tions than transects without salaman- ders (table 2). Analysis of size classes of fir and spruce showed no signifi- cant differences between transects with and without salamanders. Pro- portions of decay classes of CWD also did not differ significantly be- tween the two groups of transects (X^ = 0.28, df = 2, P > 0.90). The amount of CWDl was similar between groups but amounts of CWD3 and CWD5 were higher on transects with salamanders. Although no south-fac- ing slopes were searched, propor- tions of other aspects occupied by salamanders were not different from the proportions of total aspects searched (X^ = 1.3, df = 2, P > 0.50). Three of the original 20 variables were selected by the stepwise vari- able entry procedure for inclusion in the descriptive discriminant model: SLOPE, TPINE, and LSPRUCE (table 3). Subsequent analysis by DDA re- tained these variables. The resultant discriminant function explained 38% of the between-group variance; how- ever, it did not have significant power in discriminating between groups (F = 2.34, P = 0.09). This func- tion describes a multivariate gradient that ranges from steep slopes with Table 2,— Comparison of habitat variables measured on transects with and without Jemez Mountains salamanders, Santa Fe National Forest, 1986- 1987. Significance is based on one-way analysis of variance. Mnemonic codes for habitat variables are explained in table 1 . Transects (N = 10) with salamanders Transects (N = 33) without salamanders Mnemonic X ± se (range) X ± se (range) Significance ELEV 2526 +35.8 (2359-2621) 2635 +22.0 (2332-2886) • SLOPE 66 + 2.5 (55-84) 44+ 2.8 (0-82) CANOPY 62 + 1,8 (56-65)^ 64+ 2.1 (21-82)2 NS TFIR 72 ±10,4 (29-156) 95 + 10.3 (22-292) NS TSPRUCE 17 + 6,6 (0-59) 20+ 5,9 (0-163) NS TPINE 25 ± 7,8 (0-63) 9 + 2.1 (0-56) NS TASPEN 20+ 8.8 (1-96) 17 + 2.5 (0-60) NS TOAK 10+ 6,6 (0-59) 7± 2.4 (0-50) NS TSNAGS 33 + 6.1 (5-64) 27 + 3.3 (3-82) NS TNOD 29 + 10.4 (0-103) 8+ 2.0 (0-51) NS ROCK 11 + 2,6 (3-26) 7 + 1.6 (0-37) NS FWD 4+ 1.1 (2-12) 4+ 0,5 (0-15) NS BARK 1+1.0 (0-10) 1 ± 0.1 (0-3) NS CWD 10+ 1.9 (1-20) 9± 0.8 (1-26) NS 'P<0.05 "P< 0.005 'Data are available for 5 frar\secfs. 'Data are available for 29 transects. many pine and large spruce trees containing salamanders, to shallow slopes with few pine or large spruce trees without salamanders. SLOPE had the highest discriminating power (r^ = 0.73). PDA correctly classified 91% of the 33 transects without sala- manders and 80% of the 10 transects with salamanders. The 10 transects and additional searches produced 148 Jemez Moun- tains salamanders; the type of cover item was known for all but one sala- mander. Ninety-six percent (141/ 147) of salamanders were distributed among the four major cover classes as follows: CWD, 100 (68%); ROCK, 40 (27%); FWD, 1 (1%). No salaman- ders were found under BARK. Three salamanders (2%) were found on transects under surface litter and three salamanders (2%) were found under aspen logs. The frequency of salamanders associated with CWD by decay class was CWDl — 4%; CWD3— 66%; CWD5— 30%. Of 28 salamanders found on transects, 24 salamanders were associated with one of the four classes of cover items. Because of the small sample size, we were unable to determine a correla- tion between cover item availability and use. Sacramento Mountain Salanaander Salamanders (N = 233) were present on 26 of 80 transects (33%) with a mean density of 6/100 m^ in occu- pied areas. We located 387 salaman- ders in areas off the transects. Transects with and without salaman- ders differed in several respects: transects with salamanders occurred at significantly higher elevations, on shallower slopes, and had higher numbers of spruce and lower num- bers of pine than transects without salamanders (table 4). Analysis of size classes of fir and spruce revealed that densities of large fir and all size classes of spruce were significantly higher on transects with salamanders 57 (LFIR: t = 3.38, P = 0.001; SSPRUCE: t = 2.85, P = 0.008; MSPRUCE: t = 2.56, P = 0.016; LSPRUCE: t = 3.04, P = 0.003) (fig. 4). Although the total amount of CWD on transects with and without salamanders was not significantly different, there was sig- nificantly more CWD5 on transects with salamanders (X^ = 6.93, df = 2, P > 0.05). The proportions of transects by aspect did not differ between the two groups (X2 = 3.83, df = 3, P > 0.10). Because numbers of the three size classes of spruce were significantly higher on transects with salaman- ders, we substituted TSPRUCE for SSPRUCE, MSPRUCE, and LSPRUCE in subsequent analyses. A stepwise variable entry procedure se- lected eight of the original 20 vari- ables for inclusion in the descriptive discriminant model (table 5). Subse- quent DDA kept all but three (SLOPE, CWDl, and TAPSEN) in the model. The resultant discriminant function explained 49% of the be- tween-group variance and had sig- nificant power in discriminating be- tween groups (F = 6.87, P < 0.0001). This function can be interpreted ecol- ogically to describe a gradient that ranges from low elevations with many pine, few spruce and large fir, and infrequent CWD5 without sala- manders, to higher elevations, few pine, many spruce and large fir, and abundant CWD5 that contain sala- manders. ELEV had the highest dis- criminating power (r^ = 0.64). PDA correctly classified 96% of the 54 transects without salamanders and 58% of the 26 transects with salaman- ders. The 26 occupied transects and ad- ditional searches produced 620 Sac- ramento Mountain salamanders. Ninety-five percent (589) were dis- tributed among the four major cover classes as follows: CWD, 377 (64%); ROCK, 127 (22%); BARK, 58 (10%); and FWD, 27 (4%). Fourteen sala- manders (2%) were found under as- pen logs and 17 salamanders (3%) were above or below surface litter. The frequency of salamanders associ- ated with CWD in the three decay classes was CWDl— 13%; CWD3— 62%; CWD5— 25%. Of 233 salaman- ders found on transects, 209 sala- manders were associated with one of the four classes of cover items. Ex- amination of cover item availability and use for these salamanders re- vealed that salamanders are associ- ated with some cover items dispro- portionate to their availability (X^ = 59.9, df = 3, P < 0.001). In particular, Aneides was found in association with FWD proportionately less fre- quent than expected, and used well- decayed and moderately decayed logs to a greater extent than expected (X2 = 62.1, df =2, P< 0.001). Discussion Jemez Mountains Salamander While canonical analysis did not dis- criminate between transects with and without salamanders, it did identify steep slopes as the most useful vari- able in determining the occurrence of Jemez Mountains salamanders. It is possible that steep slopes contain more interstitial spaces in the soil than do shallower slopes. The soils of steep slopes may be less compacted than those of more gentle slopes due to the combined effects of gravity, and movement of water and soil. As a consequence of steep slope and the presence of underlying volcanic rock characteristic of the Jemez Mountains (Burton 1982), spaces within this ma- Tobie 3.— Correlations of habitat variables with discriminant scores for transects with and without Jemez Mountains salamanders. Mnemonic DFl SLOPE TPINE LSPRUCE 0.73 0.52 0.35 Table 4.--Comparlson of habitat variables measured on transects with and without Sacramento Mountain salamanders, Lincoln National Forest, 1986- 1987. Significance Is based on one-way analysis of variance. Mnemonic codes for habitat variables are explained In Table 1 . Transects (N = 26) with salamanders Transects (N = 54) without salamanders Mnemonic x ± se (Range) X ± se (Range) Significa ELEV 2779 + 17.6 (2618-2890) 2682 + 8.7 (2450-2792) " SLOPE 39 + 2.7 (21-65) 41 + 1.6 (17-70) ♦* CANOPY 72 + 1.3 (59-88) 71 + 1.3 (53-90) NS TFIR 67 + 6.3 (8-122) 64 + 4.0 (14-144) NS TSPRUCE 17 + 7.6 (0-186) 1 + 0.6 (0-30) *• TPINE 7 + 2.1 (0-50) 22 + 2.3 (0-71) ♦ TASPEN 14 + 4.1 (0-74) 17 + 3.3 (0-107) NS TOAK 5 + 2.4 (0-59) 18 3.8 (0-104) NS TSNAGS 24 + 2.8 (6-56) 25 + 2.5 (1-106) NS TNOD 33 + 7.7 (4-180) 34 + 5.6 (0-222) NS ROCK 7 + 1.7 (0-33) 7 + 0.9 (0-29) NS FWD 6 + 0.6 (2-13) 5 + 0.5 (0-14) NS BARK 1 + 0.3 (0-6) 1 + 0.2 (0-10) NS CWD 12 + 1.2 (4-24) 8 + 0.8 (0-26) NS "P< 0.005 'P < 0.05 58 130 SPRUCE 50 -I 40 30 20 - 10 - 0 130 WITH SALAMANDERS WITHOUT SALAMANDERS 120 A 80 FIR 50 4-/-^ 40 30 20 - 10 - <20CM 20-50 CM > 50 CM D.B.H. Figure 4.— Comparisons of average size classes (d.b.h.) of spruce and fir on transects witti and without Sacramento Mountain salamanders. Boxes indicate 95% confidence Inten/als for ttie mean. Levels of significance indicated by asterisks are 0.05 (*) and 0.005 (**). trix of rocky soil may provide refugia for salamanders during inhospitable times and, thus, may provide a clue to the survival of this salamander in the harsh environment of the Rocky Mountains. The largest concentra- tions of P. neomexicanus have been found in association with talus slopes (Whitford and Ludwig 1975, Clyde Jones pers. comm.), which are also important to many other western Ple- thodon (Brodie 1970). Other pletho- dontids are virtually restricted to ar- eas with a loose rocky soil (Aubry et al. 1987, French and Mount 1978, Herrington and Larsen 1985, Jaeger 1971). The variables selected by canoni- cal analysis showed some predictive value. Although three transects with- out salamanders were misclassified by PDA as transects with salamanders, Plethodon was found in areas adjacent to the transects. The two transects misclassified as transects without salamanders had values for TPINE and LSPRUCE closer to values usually associated with transects without salamanders. Because a larger percentage of transects without salamaders were correctly classified by PDA, these three variables may better describe the conditions under which salaman- ders are absent from an area, rather than describing favorable conditions under which they would occur. The limited discriminatory and predictive power of the variables se- Table 5.— Correlations of habitat variables v/ith discriminant scores for transects with and v^lthout Sac- ramento Mountain salamanders. Mnemonic DFl ELEV TSPRUCE TPINE CWD5 LFIR CWDl SLOPE TASPEN 0.55 0.42 -0.47 0.44 0.34 -0.05 -0.06 -0.02 59 lected by multivariate techniques may reflect our inability to reliably and consistently detect the presence of Plethodon at a site. We believe that our ability to detect salamanders is fairly good and repeatable, but we realize that environmental factors can influence the relative numbers of salamanders. During repeated visits to the same sites, Plethodon was more abundant when we searched under wet conditions, and other studies have reported a significant correla- tion between movement and activity of salamanders, and precipitation (Barbour et al. 1969, Kleeberger and Werner 1982, MacCullough and Bider 1975). Low densities and patchiness of P. neomexicanus popula- tions also can hinder detection of the animal. In comparison with densities of red-backed salamanders, P. cin- ereus, (0.9-2.2 individuals/ m^; Heat- wole 1962, Jaeger 1980), our density estimates for Jemez Mountains sala- manders are extremely low (0.03 in- dividuals/m^). Although Williams (1972) reported estimates of Jemez Mountains salamanders ten times greater than ours, he noted that their distribution was spotty. A better fit to a discriminant model might be obtained by includ- ing variables that we did not meas- ure, e.g., fire and logging history and soil characteristics (moisture, pH, and compaction). Williams (1976) suggested that logging may have eliminated Jemez Mountains sala- manders from part of Peralta Canyon due to dry conditions resulting from removal of most of the canopy. How- ever, there was no documentation that salamanders occurred at the site prior to logging. Soil characteristics, which can be affected by fire and log- ging practices (Childs and Hint 1987, DeByle 1981, Krag et al. 1986), also can influence the distribution of ple- thodontid salamanders, that occupy the soil-litter interface. Plethodon cin- ereus was excluded from 27% of for- est habitat in eastern deciduous for- ests because of low soil pH (Wyman and Hawksley-Lescault 1987), while the distributions of up to 10 amphibi- ans in southeastern New York were significantly influenced by soil pH and moisture (Wyman 1988). Salamanders also may be absent from a given site for reasons other than unsuitability of habitat. For ex- ample, access to a particular area by salamanders may be impossible due to the unsuitability of the area that surrounds it, e.g., dry, open field. Or, a climatic event may have eliminated salamanders from a given area with- out sufficient time occurring for them to recolonize the site. Sacramento Mountain Salamander The variables selected by canonical analysis were able to discriminate be- tween transects with and without salamanders. However, these vari- ables had limited predictive value. Although a larger percentage of transects without salamanders were correctly classified by PDA, there is still a one-in-five chance of being wrong in predicting that salaman- ders are absent from a site. For most management decisions, this level of uncertainty will not be acceptable, and ground-truth searches will have to be made. High elevation was the best pre- dictor of the presence of Sacramento Mountain salamanders (table 5). Weigmann et al. (1980) also found significantly more Sacramento Mountain salamanders on transects at higher elevations. The higher ele- vations of the Sacramento Mountains experience greater rainfall, cooler temperatures, and lower evapotranspiration rates than the lower elevations and therefore may be more hospitable to plethodontid salamanders. The low critical ther- mal maximum of Aneides probably reflects adaptations to the low tem- peratures characteristic of their mi- crohabitat (Whitford 1968) and may restrict salamanders to high eleva- tions. Aneides is often present where the best habitat predictors indicate they should not occur. While high-eleva- tion, wet, north-facing slopes with a mature mixed-conifer forest do har- bor Aneides, salamanders are also found less predictably in areas that may be drier and more exposed than the model would indicate. With the exception of elevation, the ranges of habitat variables on transects occu- pied by salamanders are not strik- ingly different from those on plots without salamanders (table 4). This overlap may be due to factors not measured, e.g., fire and logging his- tory, and it may show an ability of salamanders to persist after habitats have been altered. Management Guidelines Our data show that, despite some predictive power of the habitat vari- ables, the level of uncertainty in pre- dicting salamander occurrence may preclude their use by the USFS. At this time, we feel the best survey technique for salamanders is ground- truth surveys in wet weather during the activity season of each species. Under proper conditions, both spe- cies are easy to find and relatively unskilled persons can be quickly trained to survey habitats. Our im- pression was that Plethodon was more difficult to survey, because it tended to retreat underground dur- ing dry periods. Aneides, however, can usually be found even during ex- tended dry periods. Our attempts to explain the ab- sence of salamanders from a given area, i.e., potential difficulty of de- tecting all salamanders present, and low density or patchy distribution of populations, may overlook the possi- bility that absence is not solely due to unsuitable habitat. Absence does not necessarily mean avoidance, but may be due to insufficient time for the animal to recolonize an area, or inac- cessibility of a suitable area due to unsuitable habitat surrounding it. 60 In lieu of specific recommenda- tions, the USPS needs interim man- agement guidelines to protect the salamanders from population de- clines. We suggest the following steps: 1. Salamander surveys should be made on specific sale ar- eas as early in the planning process as possible. The USPS could maintain a team of seasonal employees for such surveys and for other activities related to endan- gered species. 2. To the extent possible, inten- sive logging operations (i.e., clearcuts, seed-tree cuts, trac- tor logging) should not be conducted in areas occupied by salamanders. Cable log- ging in winter, when the ground is frozen and the salamanders are under- ground, is probably the least damaging activity. In com- parison, tractor logging on wet soils can compact the soil to such a degree that salamanders cannot use it. 3. Modifications of current practices, such as leaving slash where it falls or leaving as much canopy as possible, help prevent the soil surface from drying out and will probably benefit salaman- ders. 4. Because current timber har- vest schedules will inevitably lead to younger-aged stands with few or only small downed logs, a mix of young and old logs should be main- tained to ensure short-term and long-term habitat com- ponents. Old logs provide cover to Aneides and Pletho- don, while younger logs are potential sources of cover in future years. Other studies provide some evi- dence for negative effects of logging on amphibian populations (Bennet et al. 1980, Blymer and McGinnes 1977, Bury 1983, Gordon et al. 1962, Her- rington and Larsen 1985, Pough et al. 1987, Ramotnik 1988, Staub 1986, and Williams 1976) and we suspect that intensive logging, slash removal, and burning will reduce or eliminate populations of Plethodon neomexica- nus and Aneides hardii. Only intensive observations of salamander popula- tions throughout the logging cycle will provide the information needed to make management recommenda- tions. These studies are in progress, but may require years before defini- tive results are available to assess the effects of logging on Plethodon and Aneides. Acknowledgments We thank the following U.S. Porest Service personnel: Santa Pe National Porest — R. Alvarado, D. Delorenzo, and M. Morrison; Lincoln National Porest — R. Dancker, D. Edwards, S. Lucas, J. Peterson, and D. Zaborske; and L. Pisher, Regional Office. Much of the funding was provided by the U.S. Porest Service (Southwestern Region). Pield personnel included M. J. Al- tenbach, R. R. Beatson, A. Bridegam, R. B. Bury, C. Campbell, S. Com, T. H. Pritts, B. E. Smith, and M. C. Tremble. S. Stefferud (Endangered Species, U.S. Pish & Wildlife Service) and C. Painter (Endangered Species Program, New Mexico Department of Game & Pish) were welcome field companions. S. Corn provided pho- tographs. K. Aubry, K. Buhlmann, S. Corn, C. K. Dodd, and C. Painter provided helpful criticism of earlier drafts. Literature Cited Aubry, Keith B., Clyde M. Senger, and Rodney L. Crawford. 1987. Discovery of Larch Mountain sala- manders Plethodon larselli in the central Cascade Range of Wash- ington. Biological Conservation 42:147-152. Barbour, Roger W., James W. Har- din, James P. Schafer, and Michael J. Harvey. 1969. Home range, movements, and activity of the dusky salamander, Desmognathus fuscus. Copeia 1969:293-297. Bennet, Stephen H., J. Whitfield Gib- bons, and Jill Glanville. 1980. Ter- restrial activity, abundance and diversity of amphibians in differ- ently managed forest types. American Midland Naturalist 103:412-416. Blymer, Michael J. and Burd S. McGinnes. 1977. Observations on possible detrimental effects of clearcutting on terrestrial amphibi- ans. Bulletin of the Maryland Her- petological Society 13:79-83. Brodie, Edmund D., Jr. 1970. West- ern salamanders of the genus Ple- thodon: systematics and geo- graphic variation. Herpetologica 26:468-516. Burton, Barry W. 1982. Geologic evo- lution of the Jemez Mountains and their potential for future volcanic activity. Los Alamos National Laboratory, LA-8795-GEOL. Los Alamos, New Mexico. 31 p. Bury, R. Bruce. 1983. Differences in amphibian populations in logged and old growth redwood forest. Northwest Science 57:167-178. Bury, R. Bruce and Martin G. Ra- phael. 1983. Inventory methods for amphibians and reptiles, p. 416-419. In Renewable resource inventories for monitoring changes and trends. J. P. Bell and T. Atterbury, eds. Oregon State University, Corvallis. Campbell, Howard W. and Stephen P. Christman. 1982. Pield tech- niques for herpetofaunal commu- nity analysis, p. 193-200. In Herpe- tological communities. Norman J. Scott, ed. U.S Pish & Wildlife Service, Wildlife Research Rep. 13. Castetter, Edward, P. 1956. The vege- 61 tation of New Mexico. New Mex- ico Quarterly 26:256-285. Childs, S. W. and L. E. Hint. 1987. Ef- fect of shadecards, shelterwoods, and clearcuts on temperature and moisture environments. Forest Ecol. and Manage. 18:205-217. Corn, P. Steven and R. Bruce Bury. In press. Sampling terrestrial am- phibians and reptiles. In Popula- tion Monitoring Techniques for Wildlife in Pacific Northwest For- ests. A. B. Carey and L. F. Ruggi- ero, eds. USDA Forest Service General Technical Report. Pacific Northwest Research Station, Port- land, Oregon. DeByle, Norbert V. 1981. Clearcut- ting and fire in the larch/ Douglas- fir forests of western Montana-a multifaceted research summary. General Technical Report INT-99. USDA Forest Service, Intermoun- tain Forest and Range Experiment Station. 73 p. Franklin, J. F., K. Cromack, Jr., W. Denison, A. McKee, C. Maser, J. Sedell, F. Swanson, and G. Juday. 1981. Ecological characteristics of old-growth Douglas-fir forests. General Technical Report PNW- 118. USDA Forest Service, Port- land, Oregon. 48 p. French, Thomas W. and Robert H. Mount. 1978. Current status of the Red Hills salamander, Phaeognathus huhrichti Highton, and factors affecting its distribu- tion. Journal of the Alabama Academy of Science 49:172-179. Gordon, Robert E., James A. MacMa- hon, and David B. Wake. 1962. Relative abundance, microhabitat and behavior of some Southern Appalachian salamanders. Zool- ogica 47:9-14. Heatwole, Harold. 1962. Environ- mental factors influencing local distribution and activity of the salamander, Plethodon cinereus. Ecology 43:460-472. Herrington, Robert E. and John H. Larsen. 1985. Current status, habi- tat requirements and management of the Larch Mountain salamander Plethodon larselli Bums. Biological Conservation 34:169-179. Hubbard, John P., Marshall C. Con- way, Howard Campbell, Gregory Schmitt, and Michael D. Hatch. 1979. Handbook of species endan- gered in New Mexico. New Mex- ico Department of Game & Fish. Jaeger, Robert G. 1971. Moisture as a factor influencing the distributions of two species of terrestrial sala- manders. Oecologia 6:191-207. Jaeger, Robert G. 1980. Microhabitats of a terrestrial forest salamander. Copeia 1980:265-268. Kleeberger, Steven R. and J. Kirwin Werner. 1982. Home range and homing behavior of Plethodon cin- ereus in northern Michigan. Copeia 1982:409-415. Krag, R., K. Higginbotham, and R. Rothwell. 1986. Logging and soil disturbance in southeast British Columbia. Canadian Journal of Forest Research 16:1345-1354. Kunkel, Kenneth E. 1984. Tempera- ture and precipitation summaries for selected New Mexico locations. Climate report. Office of State Cli- matologist. New Mexico Depart- ment of Agriculture, Las Cruces, New Mexico. Lemmon, Paul E. 1956. A spherical densiometer for estimating forest overstory density. Forest Science 2:314-320. Lowe, Charles H., Jr. 1950. The sys- tematic status of the salamander Plethodon hardii, with a discussion of biogeographic problems in Aneides. Copeia 1950:92-99. MacCullough, Ross D. and J. R. Bider. 1975. Phenology, migra- tions, circadian rhythm and the ef- fect of precipitation on the activity of Eurycea b. bislineata in Quebec. Herpetologica 31:433-439. Rough, F. Harvey, Ellen M. Smith, Donald H. Rhodes, and Andres Collazo. 1987. The abundance of salamanders in forest stands with different histories of disturbance. Forest Ecol. and Manage. 20:1-9. Ramotnik, Cynthia A. 1986. Status report Plethodon neomexicanus Stebbins and Riemer, Jemez Mountains salamander. Submitted to the U.S. Fish & Wildlife Service, Region 2, Office of Endangered Species, Albuquerque, NM. 55 p. Ramotnik, Cynthia A. 1988. Habitat requirements and movements of Jemez Mountains salamanders, Plethodon neomexicanus. M.S. the- sis, Colorado State University, Fort Collins. 84 p. Raphael, Martin G. and K. V. Rosen- berg. 1983. An integrated ap- proach to wildlife inventories in forested habitats, p. 219-222. In Renewable resource inventories for monitoring changes and trends. J. F. Bell and T. Atterbury, eds. Oregon State University, Corvallis. Reagan, Douglas P. 1972. Ecology and distribution of the Jemez Mountains salamander, Plethodon neomexicanus. Copeia 1972:486-492. SAS Institute Inc. 1982. SAS user's guide: statistics, 1982 ed. Gary, North Carolina: SAS Institute Inc. 584 p. Scott, Norman J. Jr., Cynthia A. Ra- motnik, Marilyn J. Altenbach, and Brian E. Smith. 1987. Distribution and ecological requirements of endemic salamanders in relation to forestry management. Summary of 1987 activities. Part 2: Santa Fe National Forest. Final report sub- mitted to the USDA Forest Serv- ice, Albuquerque, New Mexico. 33 p. Shelford, Victor E. 1963. The ecology of North America. University of Il- linois Press, Urbana. 610 p. Snedecor, George W. and William G. Cochran. 1967. Statistical methods. 6th ed. The Iowa State University Press, Ames, Iowa. 593 p. Staub, Nancy. 1986. A status survey of the Sacramento Mountain salamander, Aneides hardii, with an assessment of the impact of log- ging on salamander abundance. Final report submitted to the U.S. Fish and Wildlife Service, Region 2, Endangered Species Program, Albuquerque, New Mexico. 42 p. 62 U.S. Fish & Wildlife Service. 1986. Distribution and ecological re- quirements of endemic salaman- ders in relation to forestry man- agement. Part 2: Santa Fe National Forest. Report submitted to the U.S. Forest Service, Albuquerque, New Mexico. 40 p. Weigmann, Diana L., Mark Hakkila, Karl Whitmore, and Richard A. Cole. 1980. Survey of Sacramento Mountain salamander habitat of the Cloudcroft and Mayhill dis- tricts of the Lincoln National For- est. Final report submitted to the U.S. Forest Service, Lincoln Na- tional Forest. 45 p. Whitford, Walter G. 1968. Physiologi- cal responses to temperature and desiccation in the endemic New Mexico plethodontids, Plethodon neomexicanus and Aneides hardii. Copeia 1968:247-251. Whitford, Walter G. and John Ludwig. 1975. The biota of the Baca geothermal site. Report for Union Oil Co. -Geothermal Divi- sion. 140 p. Williams, Byron K. 1983. Some obser- vations on the use of discriminant analysis in ecology. Ecology 64:1283-1291. Williams, Stephen R. 1972. Repro- duction and ecology of the Jemez Mountains salamander, Plethodon neomexicanus. M. S. thesis, Univ. of New Mexico, Albuquerque. 98 p. Williams, Stephen R. 1976. Compara- tive ecology and reproduction of the endemic New Mexico pletho- dontid salamanders, Plethodon ne- omexicanus and Aneides hardii. Ph. D. thesis. University of New Mex- ico, Albuquerque. 152 p. Wyman, Richard L. 1988. Soil acidity and moisture and the distribution of amphibians in five forests of southcentral New York. Copeia 1988:394-399. Wyman, Richard L. and Dianne S. Hawksley-Lescault. 1987. Soil acidity affects distribution, behav- ior, and physiology of the salamander Plethodon cinereus. Ecology 68:1819-1827. utilization Of Abandoned Mine Drifts and Fracture Caves By Bats and Salamanders: Unique Subterranean Habitat In The Ouachita Mountains^ Abstract.— Twenty-seven abandoned mine drifts and four fracture caves constitute one of the most unique habitats in and adjacent to the Ouachita National Forest, an area devoid of solutional caves. Six species of salamanders and nine species of bats were found to utilize these areas. David A. Saugey,^ Gary A. Heidt,^ Darrell R. Heathi^ Caves and mines play an important role in the ecology of many species, serving as permanent or temporary habitats. Culver (1986) stated, "the variety of species that depends on caves during some critical time in their life cycle, such as hibernation in bats, is impressive and usually underestimated." To this statement, we add mines. Bear Den Caves are located in Winding Stair Mountain, LeFlore County, in southeastern Oklahoma. These four caves occur in an outcrop belt of a massive sandstone unit and were formed by a number of factors, the most important being gravita- tional sliding and slumpage of sand- stone. These four caves have more than 365 meters of mapped passage- way and represent the only known caves in the Ouachita National Forest (Puckette 1974-75). Additional subterranean habitat was formed from 1870 to 1890, when the area extending west from Hot Springs to Mena, Arkansas was the scene of a gold, lead, silver and zinc ' Paper presented at symposium. Man- agement of Amphibians, Reptiles, and Small Mammals in North America. (Flag- staff , AZ, July 19-2h 1988). ^David A. Saugey is a Wildlife Biologist, U.S. Forest Sen/ice, Ouachita National For- est, Hot Springs, AR. 71902. ^GaryA. Heidt is Professor of Biology, University of Arkansas at Little Rock, AR. 72204. "Darrell R. Heath is an Undergraduate Student, University of Arkansas at Little Rock, AR. 72204. rush. During the period of greatest activity, 1885 to 1888, over a dozen gold mines were in operation, rang- ing from shallow test holes to exten- sive linear and L-shaped drifts ex- tending up to 150 meters into the surrounding mountains (Harrington 1986, Hudgins 1971, U.S. Army Corps of Engineers 1980). The "gold and silver boom" effectively ended with the issuance of a report which in effect stated there were no pre- cious metals in paying quantities to be found in the area (Branner 1888). Soon thereafter, many mines were abandoned as prospectors moved West (Harrington 1986, Hudgins 1971). Through the years, other min- erals, such as manganese and mer- cury, have been mined from the Ou- achitas resulting in the excavation of numerous additional drifts; but for a variety of reasons, most have been abandoned (Clardy and Bush 1976, Stone and Bush 1984). The legacy of these mining activities has not been riches and new-found wealth, but the creation of unusual and unique wild- life habitat. The objectives of this study were to review, compile, and consolidate existing literature concerning utiliza- tion of caves and mine drifts by bats and salamanders in the Ouachita Mountains. In addition, we provide new data and propose recommenda- tions concerning management of caves and mines in the Ouachita Na- tional Forest and on other public and private lands. METHODS During the past six years, 27 aban- doned mines in Garland (8), Montgomery (3), Pike (4) and Polk (12) counties, Arkansas (fig. 1) were located and visited a minimum of eight times (at least once each sea- son). In several cases, where endemic or Category II (U.S. Federal Register 1985) species occurred or breeding populations were found, mines were visited much more often. Mist net- ting of entrances for bats was con- ducted in spring, summer, and fall. Bear Den Caves came to our atten- tion during 1987 and were visited several times. Collections were mini- mal (mines only) and voucher speci- OKLA Figure 1 .—Location of Ouachita National Forest (backslastied area) and study area (crosshiatchied area). 64 mens are located in the Vertebrate Collections at the University of Ar- kansas at Little Rock and Arkansas State University. Following McDaniel and Smith (1976), we include the probable eco- logical position of the species in the cave and mine environments. This is followed by comments concerning the status or life history of each spe- cies. Following Barr (1963) and McDaniel and Smith (1976) the terms "troglophile" (commonly found in caves), "trogloxene" (may be com- mon in caves but must leave to com- plete their life history), and "acciden- tal" (unable to survive long in the cave environment) have been em- ployed in the species accounts. RESULTS Nine species of bats and six species of salamanders were found to utilize caves and abandoned mine drifts during some portion of their annual cycles. Annotated List of Bats and Salamanders Utilizing Caves and Abandoned Mine Drifts CLASS AMPHIBIA Order Urodela Family Plethodontidae Desmognafhus brimleyorum (Stefneger). Troglophile. Means (1974) stated the Ouachita dusky salamander was confined to rocky, gravelly, streams in the Ou- achita Mountains. Rock falls along the upper portions of streams repre- sented particularly good adult habi- tat. This species was most abundant where water percolated through rocky substrate in streambeds and along stream sides. Description of egg clutch characteristics and stream/streamside deposition were given by Means (1974) and Trauth (1988) provided descriptions of deposition sites in seepage areas dur- ing the severe summer drought in 1980. Heath et al. (1986) reported the occurrence of this endemic salaman- der in four drifts, with egg clutches deposited on the underside of rocks in one mine and the presence of lar- vae in two others. In those mines with larvae, pools contained abun- dant leaf litter and isopods. On one occasion, larvae were observed feed- ing on isopods. Since these observa- tions were made, numerous addi- tional visits to these four mines re- vealed the presence of D^smognathus when epigean conditions would be considered ideal. The pools within these and other drifts are the result of seepage through walls which, in some instances, provided sufficient volumes of water to have small streams flowing from their entrances. However, unlike the preferred, gravel-bottomed stream habitat, pools typically exhibited silted sub- strates with very little rubble and few rocks large enough for egg at- tachment. Eurycea multiplicata (Cope). Trogloptiile. The many-ribbed salamander is pri- marily an aquatic species endemic to the Interior Highland region and ad- jacent areas that contain suitable habitat. It may be found under stones, logs, and other debris in clear, rock or gravel-bottomed streams (Bishop 1943, Ireland 1971, Reagan 1974). It inhabits essentially the same habitat as Desmognathus brimleyorum (Strecker 1908). Hurter and Strecker (1909) noted Desmognathus eating Eurycea indi- viduals with which they were con- fined. Heath et al. (1986) reported both larvae and adults in two mines and in one, larvae shared the same pools with Desmognathus larvae. Both mines contained shallow streams with a gravel substrate. One addi- tional mine contained larvae of this species. A seepage stream in this mine was approximately five centi- meters wide, one centimeter deep, and extended a distance of sixty centimeters before dropping into a large pool at the entrance. The pool connected directly to an epigean stream. Plethodon caddoensis Pope and Pope. Trogloptiile. Large aggregations of the endemic Caddo Mountain salamander using drifts as refugia to escape heat and dryness during summer and fall were first reported by Saugey et al. (1985). Over 100 individuals were discovered in each of two drifts, from June through September 1983. Subsequent visits to these and other drifts revealed limited use of three additional drifts and use of one of the original aggregation sites for egg deposition and breeding (Heath et al. 1986). Since these observations were made, summer aggregations of this salamander have numbered as high as 383 individuals and additional egg clutches have been observed and monitored. Known only from the Novaculite Uplift area of the Ou- achita Mountains in Howard, Montgomery, and Polk counties in Arkansas (Blair and Lindsey 1965, Robison and Smith 1982), this sala- mander and its habitat are of special concern to the Arkansas Natural Heritage Commission (ANHC) (Smith 1984). In 1985, the U.S. Fish and Wildlife Service (USFWS) desig- nated it a Category II species. In 1986, the U.S. Forest Service (Ou- achita National Forest) began infor- mal consultation with the USFWS (Jackson, Mississippi, Endangered Species Field Station) and requested field assistance from the ANHC con- cerning preservation of critical mine aggregation sites and protection of their vulnerable populations. Place- ment of a gate at one sensitive site is planned in 1988 (fig. 2). 65 Plethodon glutinosus glutinosus (Green). Troglophile. The slimy salamander, a woodland species, is widely distributed, ex- ploiting virtually every available ter- restrial habitat. This species is com- monly found under rocks, in and under well rotted logs and stumps, and buried deep in moist layers of leaf litter. During hotter and drier portions of the year, they usually re- treat deeper into the substrate. Al- though primarily epigean, this sala- mander has been reported to use caves for aggregation sites, egg depo- sition and brooding, and escape from inhospitable surface environmental conditions (Barnett 1970, Noble and Marshall 1929). Heath et al. (1986) reported this salamander from five mines; two contained breeding popu- lations and brooding behavior has been observed several times. Subse- quent observations have confirmed another of the five mines as an egg deposition and brooding site. One of the mines reported with a breeding population (Heath et al. 1986) is the site of an annual aggregation of slimy salamanders exceeding 600 individu- als. A gate (fig. 2) has been con- structed by the U.S. Army Corps of Engineers to protect this population. Continuing studies to determine the effect of gating will allow compari- son of pre- and post-gating data. Plethodon ouachitae Dunn and Heinze. Troglophile. Endemic to the Ouachita Mountains of Arkansas and Oklahoma, the Rich Mountain salamander may be found living beneath rotting logs and stumps. However, it lives primarily under pieces of sandstone on heavily overgrown talus north slopes (Black 1974, Dunn and Heinze 1933, Pope and Pope 1951, Sievert 1986). Reagan (1974) listed this species as "endan- gered and vulnerable" in Arkansas. Ashton (1976) and Black (1980) both considered this salamander "threat- ened" in Oklahoma. Sievert (1986) proposed it as a species of "special concern," conditional on his recom- mendations concerning silvicultural practices on National Forest lands. Black (1974) reported this salaman- der in Bear Den Caves where they were found throughout, but most commonly within the first 19 meters or twilight zone. A small juvenile with a snout-vent length (SVL) of < 7 mm was found in an entrance and the presence of numerous juveniles with SVLs of > 30mm may indicate egg deposition and brooding activi- ties. One of the authors (DAS) visited these caves in December, 1987 and observed one adult Rich Mountain salamander near the entrance of one cave. An additional visit in June 1988 resulted in the observation of 30+ salamanders of various size classes. Considerable human refuse and a well worn path indicated substantial numbers of visitors. Considering the uniqueness of this area and the Cate- gory II status of this salamander, steps are being taken to exclude ex- cessive visitation and protect this population from vandalism and overcollection. These caves are util- ized by the small-footed bat, Myotis leibii, (Caire 1985) also a Category II species. Plethodon serratus Grobman. Troglophile. The endemic Ouachita Red-backed salamander is commonly found be- neath rocks, logs, and in leaf litter at all elevations throughout the Ou- achita Mountains. This species has been observed in one mine on two separate occasions. In both cases, it has been in association with large aggregations of the Caddo Mountain salamander during extremely dry epigean conditions. Reagan (1974) frequently found this species in asso- ciation with the Caddo Mountain and Rich Mountain salamanders. CLASS MAMMALIA Order Chiroptera Family Vespertilionidae Myotis austroriparius (Rhoads). Trogloxene. The first Arkansas specimens of the southeastern bat were collected from one of several drifts located 12 miles northwest of Hot Springs, Garland County, Arkansas (Davis et al. 1955). 66 At the time of collection (November 1952) and during a subsequent visit, this species was found in association with the little brown bat, Myotis luci- fugus, and Keen's bat, Myotis keenii. This particular drift was inundated by the filling of Lake Ouachita in 1955 and, since that time, no addi- tional specimens have been observed in nearby drifts. The second occur- rence of this species in the Ouachita Mountain area was from abandoned Cinnabar mines located on an penin- sula in Lake Greeson, Pike County, Arkansas (Heath et al. 1986). During a winter visit (January 1984) over 150 individuals of both red and gray color phases were observed in deep torpor. A subsequent early spring visit (March 1986), revealed 15 indi- viduals. During December, 1986, only a few scattered individuals were found. According to personnel famil- iar with the drift, considerable hu- man visitation and disturbance may have been the cause of sharp decline in use of this excavation. Mumford and Whitaker (1982) suggested the southeastern bat does not tolerate disturbance and is likely to change its roosting and hibernation sites quite readily. Caire (1985) did not report this species, but records exist for the Little River drainage in southeastern Oklahoma (Glass and Ward 1959). The southeastern bat is listed as a Category II species in the U.S. Fed- eral Register (1985). Myotis keenii (Merriam). Trogloxene. Utilization of caves and mines by Keen's bat has been well documented (Barbour and Davis 1969, Heath et al. 1986, McDaniel and Gardner 1977). Sealander and Young (1955) first re- ported the occurrence of Keen's bat from the Ouachita Mountain area when three specimens were collected from the drift located 12 miles north- west of Hot Springs. Caire (1985) mist-netted a number of specimens at Bear Den Caves; the majority were males with a few postlactating fe- males. Heath et al. (1986) found this bat in 12 drifts. The largest hibernat- ing aggregation consisted of 12 bats, including both males and females. Normally, from one to three indi- viduals (usually males) were found hibernating in small cracks and crev- ices near entrances. On occasion, two have been found together in drill holes in ceilings and walls and, less frequently, individuals were ob- served hanging in the open. The larg- est non-hibernating cluster was 57 females found in the spring of 1985. Three were collected and found to be pregnant (drifts were not used as maternity roosts). Although utilized more frequently during winter months, these drifts contained from one to several Keen's bats through- out most of the year. Myotis ieibii (Audubon and Bachman). Trogloxene. The small-footed bat is very common and widespread in the western United States where it readily uses caves and mines for hibernation. In the eastern United States it is consid- ered to be rare (Barbour and Davis 1969, Smith 1984). Caire (1985) re- ported mist-netting four males, three adults and one subadult, at Bear Den Caves. Specimens collected in Sep- tember had descended testes. Heath et al. (1986) did not record this bat from drifts in Arkansas. According to Barbour and Davis (1969), the only known winter habitats for this spe- cies are caves and mines. Preferred hibernation sites are near entrances where temperatures drop below freezing and humidity is relatively low. Abandoned drifts in the Ou- achitas generally have one, small, partially collapsed entrance which ensures relatively warm interiors (18 C) with high humidities, which is un- suitable hibernating habitat. Mist- netting of creeks and drift entrances and subsequent winter visits to drifts have been unsuccessful in locating this bat. Caire (1985) indicated this species is probably restricted to cave areas. Thus, the few caves in south- eastern Oklahoma are critical to the species survival and are in need of protection. The small-footed bat is a Category II species (U.S. Federal Register 1985). Myotis lucifugus (LeConte). Trogloxene. The little brown bat appears to be extremely rare in the Ouachita Mountains. It had been reported from one drift by Sealander and Young (1955), but an additional specimen was reported by Heath et al. (1986) from a drift in Arkansas. In Oklahoma, the little brown bat has been collected only from Beavers Bend State Park in the southeastern part of the state (Glass and Ward 1959). Myotis sodalis Miller and Allen. Trogloxene. Sealander and Young (1955) reported a misidentified Indiana bat from a now inundated drift northwest of Hot Springs. There is a confirmed record of the species from a south- eastern Oklahoma cave (Glass and Ward 1959). Neither Caire (1985) nor Heath et al. (1986) found this species inhabiting mines or caves in the Ou- achitas. Pipistrellus subflavus (F. Cuvier). Trogloxene. The eastern pipistrelle was described as fairly abundant in southeastern Oklahoma (Caire 1985) and as wide- spread and abundant in the Arkansas portion of the Ouachitas (Heath et al. 1986). Barbour and Davis (1969) de- scribed it as the most abundant bat over much of the eastern United States. Caves and mines appear to be important habitats for winter hiber- 67 nation sites and for summer night roosts (Barbour and Davis 1969, McDaniel and Gardner 1977). Caire (1985) reported capturing many indi- viduals at Bear Den Caves during summer months. Heath et al. (1986) reported this species had been ob- served in every drift at all times of the year and that, over a three year period, one drift had an annual population of between 600-800 hiber- nating individuals. Visits to this hibernaculum over the past three years have revealed the number of individuals to be fairly constant. Pre- liminary observations of a drift that has had a gate in its entrance for two years have indicated an increase in numbers of hibernating pipistrelles. Epfesicus fuscus (Palisot de Beauvois). Trogloxene. Heath et al. (1986) reported that, al- though common in the Ouachita Mountain area, the big brown bat was rarely found hibernating in drifts. The four drifts used during hibernation had larger, less re- stricted, openings that created a vari- able temperature zone. Rarely were more than two or three observed in any drift. This species characteristi- cally chose hibernating sites near the entrance where temperature and humidity levels were lower. Similar hibernating behavior has been docu- mented in other caves and mines (Barbour and Davis 1969, Lacki and Bookhout 1983). Caire (1985) re- ported this species from Bear Den Caves. Lasionycteris noctivagans (LeConte). Trogloxene. Typically considered a tree bat, the silver-haired bat has been found in numerous caves and mines (Barbour and Davis 1969, Saugey et al. 1978, Whitaker and Winter 1977). Heath et, al. (1986) discovered a single speci- men hibernating in a breeze way of a drift near Lake Greeson; the ambient temperature was 2 C. The three following species of La- siurus, normally considered tree bats, have been captured during swarm- ing activities at the entrances of, but not inside drifts (Heath et al. 1983, 1986). Similar behavior in tree bats has been observed at caves (Barbour and Davis 1969, Harvey et al. 1981). Lasiurus borealis (Muller). Accidental. The red bat was captured at the en- trances of three drifts. Caire (1985) reported capturing this species at Bear Den Caves. Red bats were re- ported from inside two Ozark caves by McDaniel and Gardner (1977). Saugey et al. (1978) discovered the remains of 140 red bats in one Ozark cave. Lasiurus seminolus (Rhoads). Accidental. Heath et al. (1983) reported the cap- ture of a female Seminole bat at the entrance to a drift in Polk County, Arkansas, during September. Lasiurus cinereus (Palisot de Beauvois). Accidental. Previously unreported, a male hoary bat was captured simultaneously with the above mentioned Seminole bat. The occurrence of this species in mines and caves has been well docu- mented (Barbour and Davis 1969, Saugey et al. 1978). DISCUSSION Caves are common and widely dis- tributed in the United States. Caves are known in every state and, in some, are very common. It has been found that most caves contain a bio- logically interesting fauna (Culver 1986). Where caves are scarse, aban- doned mineshaf ts occasionally pro- vide the same specialized habitat as do natural caves (Barbour and Davis 1969). Abandoned mine drifts and frac- ture caves represent important habi- tat features in the Ouachita Moun- tains. Six species of salamanders and nine species of bats utilize these structures for some purpose. In addi- tion, four of the six salamanders are endemic to the Ouachita Mountains, and a fifth is endemic to the Interior Highlands. Two of these salamanders, Plethodon caddoensis and P. ouachitae, are Category II spe- cies. For all of these salamanders, caves and mines may only represent larger versions of existing subterra- nean microhabitats, complimenting existing situations and not replacing them. However, caves and mines do provide ''natural laboratories'' where insights into life histories and species interactions, otherwise unobservable, may be studied with the knowledge gained applied to management of surface populations. Six of the nine species of bats regularly frequent caves or mines during some portion of their annual cycles and two of these are listed as Category II species {Myotis austrori- parius and M. leibii). Mines provide a key habitat component for bats where natural subterranean hiber- nacula are scarce. Hibernacula can be viewed as islands of different sizes and complexities in an ocean of habi- tat inhospitable for hibernation (Gates et al. 1984). Most caves and mines in the Ouachitas are small and marginal as hibernacula when com- pared with extensive and complex cave systems of other regions. How- ever, minor hibernacula may become major ones (depending on their size, configuration, and microclimate), if the latter are destroyed. Further, they may function to promote range ex- pansions (Gates et al. 1984). In addi- tion, small populations become in- creasingly important in species man- agement when large populations are 68 continually threatened (Humphrey 1978). Fifty- three vertebrate taxa use Ozark caves (McDaniel and Gardner 1977). Heath et al. (1986) reported the occurrence of 27 vertebrate taxa util- izing abandoned mine drifts in the Ouachita Mountains. Caire (1985) and Black (1974) reported two spe- cies from Bear Den Caves. We report two additional species from aban- doned mines (Lasiurus cinereus and Plethodon serratus). Of the 31 re- corded species that use caves and mines in the Ouachita Mountains, 22 are common to both the Ouachitas and Ozarks. These data further support Maser et al (1979) when they stated, "Unique habitats occupy a very small percent of the total forest land base, yet they are disproportionately important as wildlife habitats." From our measurement, the total area of all known and inventoried caves and drifts in the Ouachita Mountains is approximately one acre in a forest with nearly 1 .6 million surface acres. For these reasons, resource managers should not overlook opportunities to protect and conserve what may ap- pear to be marginal sites, especially in areas where these unique habitats may be a limiting factor. MANAGEMENT RECOMMENDATIONS While the National Forest Manage- ment Act (1976) and Endangered Species Act (1973) specify objectives and set policy, the Forest Service Manual provides guidance and di- rection to realize these objectives re- lating to species of special concern and their habitats. These documents mandate consideration of these unique and valuable resources in all phases of planning and project im- plementation. Nieland and Thornton (1985), Nie- land (1985), Hathom and Thornton (1986), and Chaney (1984) provide additional information, guidance and considerations concerning manage- ment, inventory and evaluation of caves. Caire (1985) made recommen- dations about habitat management for bats, including Bear Den Caves in southeastern Oklahoma, and Sievert (1986) proposed guidelines for pres- ervation of habitat for the endemic Rich Mountain salamander. Because management of cave re- sources are adequately addressed in these references, the following rec- ommendations address issues con- cerning needed management of aban- doned mine drifts whose importance to bats and other vertebrates has been demonstrated by Heath et al. (1986), Lacki and Bookhout (1983), Saugey et al. (1985), Whitaker and Winter (1977) and this study. In line with these studies, we rec- ommend the following actions be taken on National Forests, other pub- lic lands, and private lands: 1. Address abandoned mine drifts and shafts as "unique subterranean habitat" in the Cave Management section of the Forest Service Manual. Most of the language in this chapter is directly applicable to these excavations. 2. Incorporate management prescriptions for abandoned mine drifts into Forest Land Management Plans and other resource management plan- ning documents, where ap- plicable. 3. Develop specific supple- ments, for individual Na- tional Forests, to the Forest Service Manual concerning the inventory, evaluation, and management of these excavations. 4. Prepare a chapter in the Ou- achita National Forest Wild- life Handbook providing di- rection and guidance con- cerning management of abandoned mine drifts and coordination with other re- sources. 5. Use full seasonal or partial closures to protect species of special concern during criti- cal periods of the year. 6. Acquire lands within agency administrative authority that contain caves and aban- doned mine drifts. 7. Prohibit extraction of miner- als and other materials from abandoned mine drifts. 8. Identify and designate aban- doned mine drifts, caves, and associated above ground habitat as "key areas" for wildlife during the silvicultu- ral prescription process. 9. Set aside and preserve travel corridors to prevent isolation and loss of use by terrestrial vertebrates. 10. Establish monitoring activi- ties to assess changes in the drift environment and asso- ciated wildlife utilization. 11. Continue inventory of spe- cies utilizing drifts and de- termine how and what they are using them for. 12. Cooperate, consult, and coor- dinate with state and federal resource management agen- cies, universities and col- leges, public and private con- servation organizations, and other interested publics to promote conservation, edu- cation, and research. "Ultimately, the survival of most animal species depends more on habitat protection than on direct shielding of the creatures them- selves" (Smith 1984). 69 ACKNOWLEDGMENTS We thank District Rangers John M. Archer and Rex B. Mann, Resource Assistant Clifford F. Hunt, and Wild- life Staff Officer Dr. David F. Urb- ston, all of the Ouachita National Forest, for their support and encour- agement during this study. Special appreciation is extended to Clark Efaw, Belinda Jonak, Stan Neal, Di- anne Saugey, and Derrick Sugg for valuable assistance in the field. Le- onard Aleshire and David Heath were most helpful in locating aban- doned mines in the Polk County area. The Arkansas Geological Com- mission provided useful information concerning the location of mines. This study was supported, in part, by the U.S. Forest Service (Ouachita Na- tional Forest), a University of Arkan- sas Faculty Research Grant and the University of Arkansas at Little Rock College of Science's Office of Re- search, Science and Technology. LITERATURE CITED Ashton, Ray E. 1976. Endangered and threatened amphibians and reptiles in the United States. Soci- ety Study Amphibians Reptiles, Herpetological Circular 5, 65 pp. Barbour, Roger W., and Wayne H. Davis. 1969. Bats of America. Uni- versity Press of Kentucky, Lexing- ton. 286 pp. Barnett, Douglas Eldon. 1970. An ecological investigation of cavemi- cole populations in Mansell Cave, Randolph County, Arkansas. M.S. Thesis, Northwestern State Uni- versity, Natchitoches, LA. 51 pp. Barr, Thomas C. 1963. Ecological classification of cavernicoles. Cave Notes 5:9-12. Bishop, Sherman C. 1943. Handbook of Salamanders. Comstock Pub- lishing Company, Ithaca. 555 pp. Black, Jeffery H. 1974. Notes on Ple- thodon ouachitae in Oklahoma. Pro- ceedings Oklahoma Academy Sci- ence 54:88-89. Black, Jeffery H. 1980. Amphibians of Oklahoma — a checklist. Bulletin Oklahoma Herpetological Society 4:78-80. Blair, Albert P., and Hague L. Lind- say, Jr. 1965. Color pattern vari- ation and distribution of two large Plethodon salamanders endemic to the Ouachita Mountains of Okla- homa and Arkansas. Copeia 3:331- 335. Branner, John C. 1888. Annual report of the Geological Survey of Arkan- sas. Volume 1. Little Rock. 103pp. Caire, William. 1985. Summer ecol- ogy of the bats in southeastern Oklahoma. Final Report. Okla- homa Department Wildlife Con- servation, Oklahoma City. 22 pp. Chaney, Steve W. 1984. Cave Man- agement Plan. USDI, National Park Service, Buffalo National River, Arkansas. 24 pp. Clardy, Benjamin F., and William V. Bush. 1976. Mercury District of southwest Arkansas. Information Circular 23. Arkansas Geological Commission, 57 pp. Culver, David C. 1986. Cave Faunas. Pp. 427-443. In Conservation Biol- ogy: The science of scarcity and diversity. Michael E. Soule, editor. Sinaeur Assoc. Inc., Publishers, Sunderland, MA. 584 pp. Davis, Wayne H., William Z. Lid- icker, Jr., and John A. Sealander. 1955. Myotis austroriparius in Ar- kansas. Journal of Mammalogy 36:288. Dunn, Emmett R., and Albert A. Heinze. 1933. A new salamander from the Ouachita Mountains. Copeia 3:121-122. Gates, J. Edward, George A. Feld- hamer, Lizabeth A. Griffith, and Richard L. Raesly. 1984. Status of cave-dwelling bats in Maryland: Importance of marginal habitats. Wildlife Society Bulletin 12:162- 169. Glass, Byron P., and Claud M. Ward. 1959. Bats of the genus Myotis from Oklahoma. Journal of Mam- malogy 40:194-201. Harrington, Donald. 1986. Bear City, Arkansas, p. 381-427. In: Let us build us a city. Harcourt Brace Jovanovich, Publishers, New York. Harvey, Michael J., John J. Cassidy, and Gary G. O'Hagan. 1981. En- dangered bats of Arkansas: Distri- bution, status, ecology and man- agement. Final Report to Arkansas Game and Fish Commission, U.S. Forest Service (Ozark National Forest) and National Park Service (Buffalo National River). Memphis State University, 80 pp. Hathom, Jim, and Jer Thornton. 1986. The common sense guide to cave gates. American Cave Con- servation Association 1:23-44. Heath, Darrell R., Gary A. Heidt, David A. Saugey and V. Rick McDaniel. 1983. Arkansas range extensions of the seminole bat (La- siurus seminolus) and eastern big- eared bat (Plecotus rafinesquii) and additional county records for the hoary bat (Lasiurus cinereus), sil- ver-haired bat (Lasionycteris nocti- vagans) and evening bat (Nycticeius humeralis). Proceedings Arkansas Academy Science 37:90-91 . Heath, Darrell R., David A. Saugey, and Gary A. Heidt. 1986. Aban- doned mine fauna of the Ouachita Mountains, Arkansas: Vertebrate taxa. Proceedings Arkansas Acad- emy Science 40(in press). Hudgins, Mary D. 1971. Gold and silver boom town of Bear thrived for a decade. The Record: Hot Springs-Garland County Histori- cal Society 12:98-102. Humphrey, Stephen R. 1978. Status, winter habitat, and management of the endangered Indiana bat, Myotis sodalis. Florida Science 41:65-76. Hurter, J. and J.K. Strecker, Jr. 1909. Amphibians and reptiles of Ar- kansas. Transactions Academy Science St. Louis 18:11-27. Ireland, Patrick H. 1971. Systematics, reproduction and demography of the salamander Eurycea multipli- cata. Ph.D. Dissertation. Univer- sity of Arkansas, Fayetteville. 127 pp. 70 Lacki, Michael ]., and Theodore A. Bookhout. 1983. A survey of bats in Wayne National Forest, Ohio. Ohio Journal Science 83:45-50. Maser, Chris, Jon E. Rodiek, and Jack Ward Thomas. 1979. Cliffs, talus and caves. Pp. 96-103. In Wildlife habitats in managed forests: The Blue Mountains of Oregon and Washington. Jack Ward Thomas, editor. Agriculture Handbook No.553, USDA Forest Service, Washington D.C. 512 pp. McDaniel, V. Rick, and Kenneth L. Smith. 1976. Cave fauna of Arkan- sas: selected invertebrate taxa. Proceedings Arkansas Academy Science 30:57-60. McDaniel, V. Rick, and James E. Gardner. 1977. Cave fauna of Ar- kansas: vertebrate taxa. Proceed- ings Arkansas Academy Science 31:68-71. Means, D. Bruce. 1974. The Status of Desmogmthus brimleyorum Stejneger and an analysis of the genus Desmognathus (Amphibia:Urodela) in Florida. Bulletin Rorida State Museum, Biological Sciences 18. 100 pp. Mumford, Russell E. and John O. Whitaker. 1982. Mammals of Indiana. Indi- ana University Press, Blooming- ton. 537 pp. Nieland, Jim. 1985. Cave manage- ment prescriptions. American Cave Conservation Association 1:36-41. Nieland, Jim, and Jer Thornton. 1985. Spelean inventory and evaluation guide. American Cave Conserva- tion Association 1:1-35. Noble, G.K., and Byron C. Marshall. 1929. The breeding habits of two salamanders. American Museum Novitates 347:2-12. Pope, Clifford H., and Sarah H. Pope. 1951. A study of the salamander, Plethodon ouachitae, and the description of an allied form. Bulletin Chicago Academy Science 9:129-152. Puckette, Bill. 1974-75. Bear Caves. Oklahoma Underground 7:15-17. Reagan, Douglas P. 1974. Threatened native amphibians of Arkansas. Pp. 93-99. In Arkansas natural area plan. Arkansas Department of Planning. Little Rock, AR. 248 pp. Robison, Henry W., and Kenneth L. Smith. 1982. The endemic flora and fauna of Arkansas. Proceed- ings Arkansas Academy Science 36:52-57. Saugey, David A., Robert H. Baber, and V. Rick McDaniel. 1978. An unusual accumulation of bat re- mains from an Ozark cave. Pro- ceedings Arkansas Academy Sci- ence 32:92-93. Saugey, David A., Darrell R. Heath, and Gary A. Heidt. 1985. Summer use of abandoned mines by the Caddo Mountain salamander, Ple- thodon caddoensis (Plethodontidae) in Arkansas. Southwestern Natu- ralist 30:318-319. Sealander, John A., and Howard Young. 1955. Preliminary observa- tions on the cave bats of Arkansas. Proceedings Arkansas Academy Science 7:21-31. Sievert, Gregory. 1986. An investiga- tion of the distribution and popu- lation status of the Rich Mountain Salamander (Plethodon ouachitae) in Oklahoma. Final Report. Okla- homa Department Wildlife Con- servation, Oklahoma City. 37 pp. Smith, Kenneth L. 1984. Animal spe- cies. Pp. 48-70. In Arkansas's natu- ral heritage. Bill Shepherd, editor. August House Publishers, Little Rock. 116 pp. Stone, Charles G., and William V. Bush. 1984. General geology and mineral resources of the Caddo River Watershed. Information Cir- cular 29. Arkansas Geological Commission, Little Rock. 32 pp. Strecker, John K. Jr. 1908. Notes on the habits of two Arkansas sala- manders and a list of batrachians and reptiles collected at Hot Springs. Proceedings Biological Society Washington 21:85-89. Trauth, Stanley E. 1988. Egg clutches of the Ouachita dusky salamander, Desmognathus brimley- orum (Caudata: Plethodontidae), collected in Arkansas during a summer drought. Southwestern Naturalist (in press). U.S. Army Corps of Engineers. 1980. Lake Ouachita design memoran- dum No. 2A. Re-evaluated and Updated Master Plan for Develop- ment and Management of Lake Ouachita and Blakely Mountain Dam. Vicksburg, MS. 254 pp. U.S. Federal Register. 1985. Endan- gered and threatened wildlife and plants; review of vertebrate wild- life; notice of review. Volume 50, No. 181. September 18. 10 pp. Wells, Kentwood D., and Roger A. Wells. 1976. Patterns of movement in a population of the slimy salamander, Plethodon glutinosus, with observations on aggrega- tions. Herpetologica 32:165-162. Whitaker, John O., and Francis A. Winter. 1977. Bats of the caves and mines of the Shawnee National Forest, southern Illinois. Transac- tions Illinois State Academy Sci- ence. 70:301-313. 71 The Herpetofauna of Long Pine Key, Everglades National Park, in Relation to Vegetation and Hydrology' George H. Dalrymple^ Abstract.— The amphibians and reptiles of the Long Pine Key region. Everglades National Park, were surveyed between 1984 and 1986. This herpetofauna, with 51 species, is well represented by habitat generalists and Prairie species, but the compliment of Upland species, primarily Pineland species, is low due to the lack of natural soil development and the isolation of the area. Many authors have noted a general reduction in species diversity among animal groups as latitude decreases in peninsular Florida (Dinnen 1984, Loftus and Kushlan 1987, for fishes; Duellman and Schwartz 1958, Kiester 1971, for amphibians and reptiles; Cook 1969, Robertson and Kushlan 1984, for birds; Simpson 1964, Layne 1984, for manmials). Simpson (1964) considered such a "peninsular ef- fect to be due to a greater rate of extinction and, or a lower rate of immigration along peninsulas in comparison to the mainland. Species area curves (Preston 1962, Mac Arthur and Wilson 1967) for liz- ards and snakes evaluated by Busack and Hedges (1984) showed that there was no significant peninsular effect in Florida. There was, however, a general trend for reduced species numbers as one proceeds down the peninsula of Rorida, most likely caused by a reduction in habitat quality. Moreover, Robertson's (1955) study of breeding land birds of the Long Pine Key region of Ever- glades National Park, the southern most Upland region on the mainland, revealed both lower species richness and lower densities within species 'Paper presented at symposium. Man- agement of Amphibians, Reptiles, and Small Mammals in Nortti America. (Flag- staff, AZ, July 19-21, 1988.) 'George H. Dalrymple is Associate Pro- fessor, Department of Biological Sciences, Florida International University, Miami, FL 33199. than in other areas. This reduced abundance of animals agrees with the general belief that productivity is low in southern Florida Pinelands (oligotrophic, Snyder 1986). When Duellman and Schwartz (1958) de- scribed the southern Florida herpe- tofauna as "depauperate... for a warm lowland area" they were refer- ring to the lower number of species (table 1). It has remained unclear whether characterization of the her- petofauna as depauperate applies to all habitat types in the region, in- cludes both low species and popula- tion numbers and applies to all taxa. The main objectives of this study are to: 1. develop a species list of am- phibians and reptiles in Long Pine Key-Paradise Key area (abbreviated LPK herein), 2. describe species associations with vegetation characteris- tics. Table 1.— The number of species of amphibians and reptiles found In Flor- ida, southern Florida and In Long Pine Key' Taxa Florida Southern Florida # (%) Long Pine K # (%) # Salamanders 24 4 (17) 3 (13) Frogs and toads 29 16 (55) 12 (41) Amphibian Subtotal 53 20 (38) 15 (28) Turtles 20 11 (55) 8 (40) Crocodilians 2 2(100) 1 (50) Lizards 16 11 (69) 6 (38) Snakes 41 28 (68) 21 (52) Reptile Subtotal 79 52 (66) 36 (46) Totals 132 72 (55) 51 (39) 'The data for Florida and souttiern Florida are based upon current species lists (Wilson arud Porras, 1983; Auffenberg, 1982). The numt>ers for Long Pine Key are for the current study (see text). Since Long Pine Key column includes the exotic species Eleutherodactylus planirostris. Osteopilusseptentrionalis onaf Anolissagrei they have been included in the counts for the first two columns also. (Salamander list includes Stereochilus marginatus; frog list includes the new species Rana okaloosae (Moler, 1986). 72 3. evaluate correlations be- tween species' phenologies and rainfall patterns in the area, 4. estimate abundances of spe- cies and compare them to other areas in North Amer- ica. Study Area The Long Pine Key (LPK) region was chosen for study because this 8000 ha area is the principal remaining natu- ral upland region of the original Mi- I ami (or Atlantic) Rock Ridge physi- ographic province (Davis 1943) and jj as part of Everglades National Park it ' has been protected from human interference for nearly 40 years. The region includes about 4650 ha of Pi- nelands (Snyder 1986) with a series of ''transverse or finger glades," or seasonally flooded Prairies, inter- spersed throughout the Pinelands (fig. 1). Within the Pinelands there is a series of at least 120 tropical hard- wood Hammocks (Olmsted et al. 1983, fig. 2) varying in size from .1 ha to 91 ha (Olmsted, Loope and Hilsen- beck 1980). Most Hammocks are completely surrounded by Pineland and are kept rather small due to the frequent fires (prescribed burns and natural fires from lightning) in the region. The largest Hammock, Royal Palm, is surrounded by seasonally flooded Prairies and has almost com- pletely overgrown the limestone ele- vation known as Paradise Key (these names are sometimes used inter- changeably). Because Paradise Key figured importantly in the study of Duellman and Schwartz (1958), I have included it in the present study as part of the general area described herein as LPK. On the southern border of LPK about 3600 ha of land were farmed until 1975 (abandonment was an at- tenuated process from the 1960's to 1975), when this agricultural area, known as the "hole-in-the donut," was purchased by the Park Service. Early farming was limited to areas with deeper soil, and involved little alteration of the underlying bedrock. Starting in 1954 (W.B. Robertson, Jr. pers. comm.) rock-plowing of the upper 20 cm of the ground surface created an artificial soil: "deeper, better drained, better aerated, and possibly more nutrient-rich than the pre-farming soil" on 1600 of the 3600 ha (Ewel et al. 1982:1-2). The sub- strate alteration proved conducive to the establishment of exotic vegeta- tion, especially Brazilian Pepper (Sch- inus terebinthifolius) after the farm- land was abandoned (Ewel et al. 1982). Existing detailed surveys of the region's vegetation in relation to ele- vation, fire and hydrology (e.g. Olm- sted et al 1980; Olmsted et al. 1983; Olmsted and Loope 1984; Taylor and Herndon 1981) as well as an ex- tremely detailed vegetation map of the area (Johnson et al. 1983) have made it much easier to plan the cur- rent project. Historical surveys of the literature in the above cited refer- ences, among many others, make it clear that the LPK region has not been completely free from distur- bances: logging of the Pinelands dur- ing the 1930's and 1940's; farming, as described above; invasion by exotic vegetation; development of elevated roadways with marl dug from local pits and their resulting small canals, culverts and ponds bordering the former farmlands (all of which dis- tort the original associations of eleva- tion, soil, vegetation and surface wa- ter); fire roads, to help control pre- scribed bums; and the inevitable presence of humans and their build- ings (both those for visitors and the complex of staff facilities). All of these factors play a role in determin- ing the present herpetofauna. Cur- rent park management fosters a de- Figure 1.— Aerial photograph of Pineland and Prairie of Long Pine Key. 73 large enough to ensure lasting pres- ervation of this unique ecosystem type. Materials and Methods General Collecting and Road Cruising For the 3 years of the study reported on herein many hours were spent surveying and trapping in areas for evidence of amphibians and reptiles. Each time the traps were checked, a 50 km section of unimproved dirt roads was driven over by van, and an additional 15 km paved road was systematically covered by van for a total of 8 to 16 hours per week, dur- ing which all animals were captured and identified. Searches on foot, by teams of two to four people, were conducted in all of the major habitats each week, during which animals were searched for at the surface and under rocks and logs. The time spent collecting and road cruising was di- vided between day and night to en- sure that all species in LPK might be found. Trapping I used a system of funnel traps at- tached to drift fences and transects (referred to throughout as '"arrays"). Many researchers have used arrays to study amphibians and reptiles (Campbell and Christman 1982b, Clawson and Baskett 1982, Vogt and Hine 1982, Gibbons and Semlitsch 1981, Clark 1970), however they all employed arrays that included both funnel traps and pit traps. Usually the pit traps are placed at regular in- tervals by digging holes in the ground. However, the lack of well developed soils coupled with an ir- regular limestone surface made the use of pit traps impractical to use in the everglades. Each array was constructed of four fifteen meter long sheets of shade cloth (one meter tall) that intersected in the middle to form an "x." The shade cloth was kept up- right by tieing it to iron rebars that were hammered into the limestone. Traps were made of cylinders of one- eighth inch hardware cloth approxi- mately 1 m in length and 30 cm in diameter. Each trap was fitted with two funnels (one funnel on each side of the shade cloth fencing) made of the same material. Funnels were at- tached to the free ends of the four arms of the array. Shade cloth had 12-cm flaps sewn onto the bottom edge to conform to the irregular sur- faces of the everglades terrain. Flaps were covered with natural soils and or leaf litter so that animals would not crawl under them (figs. 3 and 4). The square area encompassed by each array was .10 ha. Arrays were placed in each of four main habitat types: seasonally flooded Prairies, Pinelands, tropical hardwood Hammocks, and in the area of secondary succession from former farming, the "hole-in-the-do- nut." The latter area is referred to throughout as "Disturbed." Thirteen arrays were maintained starting in May, 1984, and the arrays are still checked to the current date. Three arrays were placed in each habitat type within Long Pine Key and one extra hammock array was main- tained in Royal Palm Hammock on Paradise Key (fig. 5). Arrays were temporarily taken down during park service prescribed burns and re- placed after the bums. Because ar- rays were in place for different dura- tions, I assessed yield in terms of rate of capture, rather than absolute cap- ture yield, and capture rate was as- sessed separately for wet and dry seasons. At each array we main- tained two 1-m^ pieces of tar-paper, under which we commonly collected seasons. At each array we main- tained two pieces of tar-paper, under which we commonly collected animals. All animals caught along the fences or under the tar paper at an array were counted as part of the capture rate at the array in question. Figure 3.— Aerial photograph of locality known as New Wave Prairie in Long Pine Key with "x"-shaped trapping array visible at left (each of the four arnr>s of the array is 15 m long). Symbolic Star Plot Analysis Symbolic Star Plot Analysis (Cham- bers at al. 1983) was chosen as a use- ful multivariate method for graphi- cally depicting the rates of capture of species in the major habitats. Only species for which there were at least ten captures were chosen, and the analyses were based on the number of animals trapped per 1000 array days because the raw data does not reflect the fact that arrays were op- erational for varying time periods. The data values are used as the lengths of the rays of the stars for each habitat. All data values were rescaled to range from 1 to c, where c is the length of the smallest ray (set to 0.1 for these analyses). According to Chambers et al. (1983:158): "If x.. is the measurement of the i"" variable then the scaled variable [x*. ] is ') x*j = (1 - c)(X|j - min,X|P / (maX|X|j - min|X,j) + c." The scaled variables are arranged around a circle at equal angles, the number of angles determined by the number of variables, and the actual rays are drawn by connecting points trigonometrically calculated for an arbitrarily chosen maximum radius for the circle. The lengths of the rays (not the area adjoining the rays) in the four habitat stars for a given species rep- resent the proportion of all captures for that species in each habitat. The result is intended to form a simple yet "dramatic and memorable" im- pression of the relationships within species and between habitat types, for further details see Chambers et al. (1983:158-163). Figure 4.— Ground level view of trapping array fencing in Pineland, 75 Population Abundance Estimates For most species the actual numbers presented are actual numbers of indi- viduals captured. All snakes and turtles were individually marked. The anurans and lizards were marked only during 1984, but due to the lack of recaptures I stopped marking in 1985. The marking method used for snakes was that of Brown and Parker (1976), and even though snakes were marked for four consecutive years (1984-1987) the re- capture rate remained very low (<0.05, Dalrymple, in prep.). Concentrations of amphibians and reptiles around one or more re- sources, such as water (ponds or lakes. Carpenter 1952, Reichenbach and Dalrymple 1986), hibemacula (caves, pits and dens, Woodbury 1951, Brown and Parker 1982a, Aleksiuk and Gregory 1974) breeding sites (Crump 1982, Brown and Parker 1982b, Wiest 1982) and or food (Ha- milton 1951) lead to recaptures that allow for density estimates with con- fidence limits (cf. Turner 1971). These estimates are dependent on seasonal fluctuations, and may differ greatly from estimates of crude density. However, few concentrations were found on LPK particularly because water was readily available in nu- merous solution holes in every habi- tat. Moreover, mild winters allowed most species to be active throughout the year, and the ability of animals to readily go underground through the porous limestone and plentiful solu- tion holes found in all habitats re- sulted in the absence of group hiber- nacula. Further complicating density estimation were widespread move- ments in search of mates, and the fact that major food sources were not clumped. All these factors lead to a wide spread distribution of most species in the region and most were not habitat specialists, at least at the major vege- tation type level. The lack of concen- trations and the limited number of recaptures permit only the presenta- tion of total numbers of captures and not accurate density estimates at this time. Results Species List Starting in January, 1984, 51 species of amphibians and reptiles were ob- served or collected in LPK (table 2). Some species were rare because they are most commonly associated with more permanently aquatic habitats, such as the Sloughs (e.g. Acris gryllus, Ram grylio, Trionyx ferox, Farancia abacura, Nerodia cyclopion, Nerodia tax- ispilota, Regim alleni). A few species that have been recorded in the larger geographic region were not found in LPK during this study (Scaphiopus holbrooki, Pseudobranchus striatus, Semimtrix pygaea, Masticophis flagel- lum, Heterodon platyrhinos, Ophisaurus ventralis, Sternotherus odoratus). Trapping Results Between May, 1984 and December, 1986, 1709 amphibians and reptiles were collected either in the traps, under associated tar paper, or along array fences (table 3). These animals represent 37 of the 51 species (73%) known from our overall surveys. I compared the four habitats by re- cording the number of animals per Figure 5.— Map of the Long Pine Key-Paradise Key region of Everglades National Park. Array locations are nunnbered and referred to in \he text as follows: 1 . Pine Block B, 2. New Wave Prairie, 3. Pine Block E, 4. Junk Hammock, 5. Serenoa Prairie, 6. Wrigfit Hammock, 7. Mud Prairie, 8. Pine Block H, 9. Palma Vista I Hammock, 10. Royal Palm Hammock, 11. Burnout Dis- turbed, 12. Sctiinus Disturbed, 13. Grass Disturbed. 76 Table 2.— List of species of amphibians and reptiles observed In the Long Pine Key - Paradise Key region of Ever- glades National Park during present study, between January, 1984 and December, 1986. The reglonwide natural habitat associations of Duellman and Schwartz (1958), as they apply In the study area, are given after the scientific name for each species. Pr = Prairie, Pi = Pine, H = Hammock, A= Permanently Aquatic, I.e. Slough, Canals. Scientific name Common name Scientific name Common name Urodela Amphiumo means -Pr A Siren lacerfina -Pr A Nofophfhalmus viridescens -Pr A Anura Acris gryllus -Pr Bufo quercicus -Pr ?\ H Bufo ferresfris -Pr Pi H Eleufherodacfylus planirosfris -Pi Gasfrophryne carolinensis -Pr Pi H Hyla cinerea -Pr Pi H Hyla squirella -Pr Pi H Limnaoedus ocularis -Pr Osfeopilus sepfenfrionalis -H * Pseudacris nigrifa -Pr Pi Rana grylio -Pr A Rana splienocephala -Pr A Testudines Chelydra serpentina -Pr A Clirysemys floridana -Pr A Chrysemys nelsoni -Pr A Deirocheiys reficularia -A Gopherus polyphemus -Pi Kinosfernon bauri -Pr A Terrapene Carolina -Pr Pi H Trionyx ferox -A Crocodylia Alligator mississippiensis -Pr A two-toed ampl-ii- uma greater siren peninsula newt Florida cricket frog oak toad southern toad greenhouse frog eastern narrow- mouthed toad green treefrog squirrel treefrog little grass frog Cuban treefrog Florida chorus frog pig frog southern leopard frog snapping turtle peninsula cooter red-bellied turtle chicken turtle gopher tortoise striped mud turtle box turtle Florida soft-shelled turtle American alligator Squamatalacertilia Anolis carolinensis -Pr Pi H Anolis sagrei -P\ Eumeces inexpectatus -Pr Pi H Ophisaurus compressus -Pr Pi Scincella laterale -Pi Sptiaerodactylus notatus -Pi Squamata ,Serpentes Agkistrodon piscivorus -Pr A Cemopt)ora coccinea -Pi Coluber cor^strlctor -Pr Pi H Crotalus adamanteus -Pi Diadophis punctatus -Pr Pi H Drymarct)on corais -Pr Pi H Eiaphe guttata -Pr Pi H Elaphe obsoleta -Pr Pi H Farancia abacura -Pr A Lampropeltis getulus -Pr Pi H Lampropeltis triangulum -Pi Micrurus fulvius -Pi Nerodia fasciata -Pr A Nerodia cyclopion -Pr A Nerodia taxispilota- A Opheodrys aestivus -Pr Pi H Regina alleni -Pr Sistrurus miliarius -Pr Pi Storeria dekayi -Pr Pi H Thamnophis sauritus -Pr Pi H Ttiamnophis sirtalis -Pr Pi H green anole brown anole southeastern five- lined skink island glass lizard ground skink reef gecko cottonmouth scarlet snake black racer eastern diamond- back ringnecked snake indigo snake corn snake yellow rat snake mud snake kingsnake scarlet kingsnake coral snake banded water snake green water snake brown water snake rough green snake striped crayfish snake pigmy rattlesnake brown snake ribbon snake garter snake array day. The highest capture rates were in seasonally flooded Prairie, which had both the most individuals and the most species collected, fol- lowed by Disturbed areas. Hammock and Pineland (table 3). Monthly total rainfall for LPK and maximum water level from well sta- tion NP-72 in the same area for data from 1984-1986 were provided from hydrological stations maintained by the South Florida Research Center, Everglades National Park. These data were correlated with the monthly values for animals trapped per check day. There were significant correla- tions between number of animals caught per check day and both monthly rainfall (r = 0.55, p = .001), and monthly maximum water levels (r = 0.50, p = .004) for the three year period (fig. 6). Rates of capture were significantly greater during the wet season than the dry season (table 4; Wilcoxin matched pairs test, T = 3.0, p < .005). Differences in overall cap- ture rates between the dry and wet seasons is greater in Hammock and Disturbed areas than in the Pinelands and Prairie. Relative Abundance Although 37 species were found at arrays they were not all equally com- 77 Table 3.— Total numbers of amphibians and reptiles trapped, May 1984-Dec 1986. "Ctieck days" are number of days on which traps were checked. "Array days" are number of total days arrays were standing. Numbers in parentheses are animals per 1000 array days. Acronyms at right of table are for species used in figures 7-9. Taxa Prairie Pineland Hammock Disturbed Total A. means 9 (3.5) 0 (0) 0 (0) 0 (0) 9 A. gryllus 1 (0.4) 0 (0) 0 CO) 0 (0) \^y 1 B. quercicus 95 (37.2) 7 (2,8) 3 (0 9) 9 (6.2) 114 Bq B. ferresfris 45 (17.6) 24 (9.4) 50 (15.5) 31 (21.3) 150 Bt E. plonirosfris 15 (5.9) 17 (6.7) 50 (15.5) 6 (4.1) 88 EP G. carol inensis 10 (3.9) 1 (0.4) 21 (6 5) 33 (22.6) 65 Go H. cinerea on 1 1 7 u-2; 31 He H. squkella 32 (12.5) 3 (1.2) 6 d 9) 4 (2.7) 45 Hs OiSepfenfrionalis 2 (0.8) 1 (0.4) 3 CO 9) 6 (4.1) 12 Os P.nigrifa 5 (2.0) 8 (3.1) 0 CO) 0 (0) \^y 13 Pn R. grylio 5 (2,0) V,*- >^ J 0 (0) 0 CO) 0 (0) \^y 5 R. sphenocephala 135 (52.8) 10 (3.9) 106 (32 8) 20 (13.7) 271 Rs A. carolinensis 170 (66,5) 136 (52.3) 19 C5 9) 19 (13.0) 344 Ac A. sagrei 0 (0) 0 (0) 50 (15.5) 103 (70.7) 153 As E. inexpecfatus 23 (9.0) 21 (8.2) 42 CI 3.0) 3 (2.1) 89 El O. compressus 1 (0.4) 1 (0.4) 0 (0) \^y 1 (0,4) 3 S. late rale 30 (11.7) 9 (3.5) 3 CO 9) 0 (0) \^y 42 SI S. not a f us 0 (0) 0 (0) 29 (9.0) 0 (0) 29 Sn K. bauri 12 (4.7) 2 (0.8) 1 (0.3) \ V • ^y 1 (0,7) 16 Kb T. Carolina n (4.3) 1 (0.4) 2 (0.6) 3 (2.1) 17 To A. piscivorus 1 (0.4) 0 (0) 0 (0) 2 (1.4) 3 C. coccinea 2 (0.8) 0 (0) 0 (0) 0 (0) \^y 2 C. constrictor 8 (3.1) 30 (1 1.8) 14 C4 3) 14 (9.6) v » • ^y 66 Cc C. adamanteus 0 (0) 0 (0) 0 (0) \^y 1 (0,7) 1 D. punctatus 3 (1.0) 3 (1.0) 13 (4 0) 0 (0) \^y 19 Dp D. corals 1 (0.4) 2 (0.8) 2 (0,6) 0 (0) 5 E. guttata 0 (0) 1 (0.4) 0 (0) 0 (0) 1 E. obsoleta 0 (0) 0 (0) 4 (1.2) 1 (0.7) 5 L.getulus 0 (0) 0 (0) 0 (0) 1 (0.7) 1 L triangulum 1 (0.4) 0 (0) 0 (0) 0 (0) 1 M. fulvius 0 (0) 0 (0) 4 (1.2) 0 (0) 4 N. fasciata 3 (1.2) 0 (0) 0 (0) 0 (0) 3 R. alleni 1 (0.4) 0 (0) 0 (0) 0 (0) 1 S. millarius 14 (5.5) 8 (3.1) 3 (0.9) 6 (4.1) 31 Sm S, dekayi 2 (0.8) 0 (0) 4 (1.2) 0 (0) 6 T. sauritus 8 (3.1) 1 (0.4) 10 (3.1) 0 (0) 19 Tsa T. sirtalis 30 (11.7) 5 (2.0) 2 (0.6) 7 (4.8) 44 Tsi Totals 695 292 448 274 1709 No. Check days 669 663 789 361 2482 Anis/Check day 1.04 0.44 0.57 0.76 0.70 No. Species 30 22 24 21 37 No. Array days 2555 2550 3229 1458 9792 Anis/ Array day 0.27 0.12 0.14 0.19 0.18 men. The most common species were anurans and lizards (table 3): Ram sphenocephala, Bufo terrestris, and Ano- lis carolinensis. Of the 20 species of snakes collected during the study, 17 were trapped but only five were cap- tured in high enough frequency to allow for more detailed study (Col- uber constrictor, Thamnophis sirtalis, Sistrurus miliarius, Diadophis punc- tatus, and Thamnophis sauritus). As a preliminary method, abundance can be minimally estimated as the actual counts from the 'Total" column of table 3 as the number per hectare (12 arrays, each one covering approxi- mately one-tenth of a hectare makes this a conservative estimate). 78 Habitat Use And Preference A species' likelihood of being trapped is more a funcrion of the number of individuals in the vicinity of an array than a result of any dif- Figure 6.— Comparison of number of ani- mals trapped per check day per month with monthly rainfall and water table values from study area between May 1984 and December 1986. ference in trap functioning between habitats. For species with high cap- ture rates, there were significant dif- ferences in habitat use for: Coluber constrictor, more common in Pine- lands (chi square = 14.59, p = ,0007); Thamnophis sirtalis, Sistrurus miliarius, Scincella laterale and Bufo quercicus all more common in Prairie (chi squares of 42.9, 9.6, 26.4, 71.8 respectively, all with p's < .01); while Bufo terrestris is equally common in all habitats (chi square = 2.36, p = .51). In most cases, species were found in more than one and usually three habitats (cf. Duell- man and Schwartz 1958). Among trapped species, 41% were found in all four habitat types, 27% in two or three, and 32% in only one habitat type. Seven of the 13 species from only 1 habitat type were from Prairie. Table 4.— Results of 1985 trapping of all individuals of amphibians and rep- tiles at 13 array sites organized by vegetation type, and season (dry = No- vember-April; wet = May-October). "Check-days" are the number of days on which an array was checked for animals. Note that there is no data for the wet season for "Grass" array (see Materials and Methods). Variation within habitat types is as great as between habitat types. Habitat/array No. No. No. Animals Individuals species check-days per check day Season: Dry Wet Dry Wet Dry Wet Dry Wet Prairie New Wave 65 118 12 20 54 54 1,2 2.2 Mud 38 64 10 18 56 54 0.7 1.2 Serer^oa 12 20 2 10 50 51 0,2 0.4 Pineiand Pine Block B 26 23 8 8 50 53 0,5 0.4 Pine Block H 25 39 8 11 56 51 0,5 0,8 Pine Block E 16 17 4 8 51 52 0,3 0,3 Hammocks Royal Palm 18 110 7 17 56 28 0.3 3.9 Pclma Vista 1 11 50 6 12 56 33 0,2 1.5 Wright 15 21 6 8 52 53 0.3 0.4 Junk 17 23 7 7 53 52 0,3 0.4 Disturbed Schinus 11 76 6 12 55 33 0.2 2.3 Burnout 11 16 6 7 45 17 0.2 0.9 Grass 14 4 18 0.7 Symbolic star plot analyses (Chambers et al. 1983) were applied to the 1984-1986 trap data for the number of animals per 1000 array days as the data set (table 3), for the anurans (fig. 7), lizards and turtles (fig. 8), and snakes (fig. 9). Since the qualitative general habitat associa- tions of Duellman and Schwartz (1958) were corroborated in this study, I restricted this quantitative analysis to those species for which there were at least 10 captures. It is obvious from the anuran plot that the majority of individuals and species are most prevalent in Prairie. Pseudacris nigrita is strongly repre- sented in Pineiand, as was noted by Duellman and Schwartz 1958). In Hammocks, Eleutherodactylus planirostris, Bufo terrestris, Gastro- phryne carolinensis, and Hyla cinerea were dominant. Rana sphenocephala was most common in Prairie but was very abundant in two Hammocks that are adjacent to wet Prairie and that retained water in solution holes throughout most of the year (Royal Palm and Palma Vista I). Bufo terrestris, G. carolinensis and the exotic Cuban tree frog, Osteopilus septentri- onalis, were dominant in Disturbed habitat (fig. 7). For the trap data for turtles, Kinos- ternon bauri and Terrapene Carolina, and the lizards. Prairie again had the greatest abundance; but T. Carolina was commonly found in the Dis- turbed habitat. Anolis carolinensis was well represented in Pineiand and Prairie, as were the skinks, Eumeces inexpectatus and Scincella laterale. Ano- lis sagrei was restricted to Disturbed sites and Hammocks, especially those close to roads and parking lots. Sphaerodactylus notatus is most often found in leaf litter of Hammocks, and E. inexpectatus is also well repre- sented in Hammocks (fig. 8). For snakes, the star diagram analysis was restricted to the five most common species; again the greatest diversity and abundance is found in Prairie. Coluber constrictor was clearly the dominant snake in 79 prairie pineland Pn Rg Figure 7.— Star plot diagrams of anuran data from table 3, comparing th»e frequen- cies of trapping (anurans per 1000 array days) of ttie species in \he four tiabitat types. Genus and species names abbrevi- ated on key at bottom of figure correspond to acronyms given in table 3. Pineland. Sistrurus miliarius was well represented in all habitats, but is least common in Hammocks. Tham- nophis sirtalis was most abundant in Prairie, while T. sauritus was most common in Prairie and Hammocks. Diadophis punctatus is the snake spe- cies most difficult to keep in traps (because of their small size they could more readily escape) but cur- rent data indicate that they are most common in the leaf litter environ- ment of Hammocks (fig. 9). The most similar habitats with re- gard to trap data were Prairie and Pineland, the least similar were Pine- land and Hammock (table 5). Table 5 includes the only data from the ar- rays and therefore some species are excluded from the similarity index (because the index used, Morisita's index (Horn 1966; Brower and Zar 1984) requires data on both the num- ber of species and the number of in- dividuals per species in the estima- tion of degree of similarity). hammock disturbed prairie pineland Figure 8.— Star plot diagrams of lizard and turtle data from table 3, comparing fre- quencies of trapping (lizards or turtles per 1000 array days) in the four tiabitat types. Genus and species names abbreviated on key at bottom of figure correspond to acro- nyms given in table 3. Discussion Species List Duellman and Schwartz (1958) gave a complete list of the localities from which they examined specimens but, unfortunately this list does not serve as an effective species list for this study. Since the intention of their study was a survey of all of southern Florida, they did not collect as exten- Pralrie Prairie 30 Pine .736 Hammocks .608 Disturbed .314 V hammock disturbed prairie pineland Figure 9.— Star plot diagranr>s of snake data from table 3, comparing frequencies of trapping (snakes per 1000 array days) in ttie four tiabitat types. Genus and species names abbreviated on key at bottom of figure correspond to acronynns given in table 3. sively in one area as we have been able to. Nevertheless, the descrip- tions of habitat preferences they gave make it clear that a few more species might be found in the Long Pine Key region if I continue the study. There are some noticeable absences from their list for the Long Pine Key and Paradise Key areas however: Storerk dekayi and Diadophis punctatus. It is possible that these species were merely overlooked in their surveys Pine Hammocks Disturbed 21 20 17 22 19 16 ,303 24 17 .253 .589 21 Table 5.— Measures of similarity among arrays grouped by vegetation type based on data from table 3 (1984-1986, above). Numbers above the di- agonal are the numbers of species shared between habitats; numbers along the diagonal, boldfaced, are numbers of species occurring in each habitat. Numbers below the diagonal, underlined, are Morisita's indices. 80 and it is extremely unlikely that these species were not present in the local area thirty years ago (Duellman and Schwartz, personal communications). Salamanders were the taxon most poorly represented in LPK, only four of the state's 24 salamanders were found in southern Florida (table 1), and only three of these were found in LPK. The reason for the low count is obviously the low elevation and poor soil development of the region. The majority of Florida's salaman- ders are members of the family Ple- thodontidae, and this family is pri- marily distributed in the Appala- chian mountains and foothills of the eastern U.S. Many species are stream dwellers, others are forest litter in- habitants that require a moist thick leaf litter and soil development. The mole salamanders, family Ambysto- matidae, also require soils for bur- rowing. Moreover, salamander lar- vae are frequently absent from aquatic settings in which fish are common. One notable exception is the newts (family Salamandridae), but even the one member of this family from the region, Notophthalmus viridescens, is rare. The only successful salaman- ders in the region are fully aquatic, neotenic, eel-like animals: Amphiuma means, Siren lacertina and Pseudobran- chus striatus. Their cryptic life styles and easy access to the underground aquifer through the porous limestone bedrock may be important reasons for their success. The number of anuran, lizard and turtle species are all rather low in southern Florida (tables 1 and 2). Several species of lizards extend southward past the mainland into the Rorida Keys, but appear to have completely by-passed the western extension of the Miami Rock Ridge (in particular LPK) e.g. Eumeces egregius and Cnemidophorous sexlinea- tus. Two species are endemic to the sandhills and scrub habitats of Ror- ida (Sceloporous woodi and Neoseps reynoldsi) and their absence in the area is again probably due to the lack of suitable soils and substrates. The reason for the absence of the other two species of Ophisaurus (O. attenu- atus and O. ventralis) listed by Duell- man and Schwartz (1958) is not clear, although they did note that Ophisau- rus compressus was the "most abun- dant" of the three species in southern Florida. The only notable introduced lizard was Anolis sagrei. This species is so common in southern Florida now that it is no surprise that large popu- lations are found in some parts of the current study area (Wilson and Por- ras 1983). In LPK it was generally limited to areas where there was a greater rate of contact with visitors, and in Disturbed settings. In remote Hammocks anoles were rarely ob- served, but Palma Vista I and Royal Palm Hammocks (both sites that are popular with visitors and adjoin roads) Anolis sagrei is extremely com- mon, as well as throughout the hole- in-the-donut. At the current time the park appears to have a limited "load" of exotic lizards. Hemidactylus garnoti was observed at the parking lot at Pahayokee visitors site, and there are occasional reports of this species and of Anolis equestris in the LPK campground area and the "Pine Island" residential area for park staff. Of the few specimens of Gopherus polyphemus seen during the study, the only one from the study area was crossing the road into the hole-in-the- donut (several others were seen in the Pine Island residential area and one shell was near a pond, but no one is certain of the source of these animals, and some visitors have been known to release gopher tortoises near the entrance to the park). Whether the sighting within the study area (the turtle was measured, and marked) is indicative of a small population or is a captive released by a visitor is not at all clear. The presence of a population of gopher tortoises on Cape Sable (Kushlan and Mazzotti 1985) does not help in explaining the single specimen, and Duellman and Schwartz (1958) list only one speci- men for Dade County. Duellman and Schwartz (1958:260) described Ster- notherus odoratus as "the least abun- dant of the three southern Florida kinosternids," and I have found it in the Shark River Slough region but not LPK. Kinosternon subrubrum is de- scribed by Duellman and Schwartz (1958:265) as avoiding "the main part of the Everglades, an area where K. bauri reaches its greatest abundance. When the above three rare species are noted the turtle list for Long Pine Key is typical of the southern Rorida region. Some of the species listed by Du- ellman and Schwartz were not com- mon in the southern everglades, but were found in other areas of south- ern Rorida. There were no species of anurans that I expected to find and did not. The burrowing nature of Scaphiopus holbrooki probably pre- vents it from being common in LPK, and it was never seen or heard dur- ing this study. The crocodilian fauna of LPK is composed of only one species, the American alligator (although there have been rare occurrences of the American crocodile, Crocodylus acutus, in the freshwater reaches of the Taylor Slough drainage in the vi- cinity of the study area, W.B. Robertson, Jr. pers. comm). The alli- gator is found in almost every place in the everglades where there is wa- ter. We commonly found evidence of alligators in the seasonally flooded Prairie (alligator trails) and in the willow heads and Hammocks ("ga- tor holes," a few nests seen, juvenile and adult alligators observed). The LPK region is certainly peripheral to the main distribution of the species in the park. The snake fauna is clearly the best represented fauna in LPK. Of the 26 species listed for southern Florida, 21 were collected during the study. Of the five not found during this study only one was expected, Seminatrix pygaea, and the technique for trap- 81 ping this species described by Lo- raine (1985) will be tried in the study- area in the future. Heterodon platy rhi- nos was described by Duellman and Schwartz (1958) as not being abun- dant in southern Florida, and there is only one report of it from the LPK area (Roger L. Hammer pers. comm.). Masticophis flagellum is still re- ported from the pineland remnants of southwest Dade County. Duell- man and Schwartz (1958) had no rec- ords of this species from the park, but since then there has been one rec- ord from the park. Pituophis melanoleucus was repre- sented in the work of Duellman and Schwartz by a single specimen from Miami, and a single specimen of this species was collected in 1984 in North Miami Beach. The snake was probably a captive pet released in the area, since its feces contained white mouse remains (Robert J. Nodell, pers. comm.). Tantilla oolitica (T. coromta wagneri of Duellman and Schwartz) has never been recorded from the park, and its range is lim- ited to isolated Atlantic Coastal Ridge remnants on the eastern coast and the Florida Keys (Wilson and Porras 1983). Habitat Use and Preferences Within the LPK region. Prairie habi- tat has the most diverse and abun- dant herpetofauna. The Prairie is a broad transition zone or ecotone be- tween the longer hydroperiod Slough habitat and the drier Uplands, and they are seasonally inhabited by most species from those two habitats as well as a semi-aquatic fauna of their own. Duellman and Schwartz (1958:206- 213) characterized the habitats of southern Florida, as they pertain to Long Pine Key, as: Xeric (including the rocky Pineland of Long Pine Key), Mesic (including the tropical hardwood Hammocks of Long Pine Key), and Altemohygric (including Prairie), and their characterization for each species is given in table 2. All of the 18 species that Duellman and Schwartz (1958:211) character- ized as generalists i.e. "common to all three" (i.e. Prairie, Pineland, and Hammock) were found in Long Pine Key. Seventeen of the 21 species (81%) they characterized as inhabi- tants of the Prairie (or Altemohygric habitat) were found in the study area. Only 9 of the 22 species (40%) that Duellman and Schwartz (1958:210) characterized as Xeric or Pineland species are found in the region. Four of these 9 species were actually more common in Hammocks (Eleutherodac- tylus planirostris, Sphaerodactylus notatus, Anolis sagrei, and Micrurus fulvius), one (Scincella laterale) was common in Prairie, three were rare (Gopherus polyphemus, Lampropeltis triangulum, and Cetnophora coccinea) and only one (Crotalus adamanteus) was actually most common in Pine- land (see table 2). Using the species associations of Duellman and Schwartz (1958), of the 51 species from Long Pine Key, 35% (18) are generalists, 33% (17) are Prairie species, 18% (9) are Pineland or Xeric in habitat association, 6% (3, Limmoedus ocularis, Pseudacris nigrita and Ophisaurus compressus) are com- mon to Prairie and Pineland, 6% (3, Alligator mississipiensis, Trionyxferox and Deirochelys reticularia) are pri- marily Slough or Hygric (Duellman and Schwartz 1958:212), and 2% (1, Osteopilus septentrionalis) from Edifi- carian-Ruderal and Hammock (Me- sic) habitats. The limit to the preservation of overall diversity of the Long Pine Key region is the extent of rocky Pi- neland habitat, because it is the ma- jor habitat type of the area with the smallest percentage (40%) of its her- petofauna (as defined by Duellman and Schwartz 1958) represented. It is important to note that the common use of interdigitating finger glades, i.e. the local Prairie, and Hammocks by some of the Pineland species makes it clear that overall diversity depends upon continued manage- ment to preserve the current patch- iness of the area. Sixty two percent of the species trapped in the Disturbed habitat are characterized as generalists by Duell- man and Schwartz (1958), 14% are from Pineland and Prairie, 14% are from Pineland and 10% are from Prairie. While the vast majority of am- phibians and reptiles were either trapped and, or seen in the Disturbed habitat, a few were rarely or never seen in the Disturbed habitat: Limnaoedus ocularis, Pseudacris nigrita, Scincella laterale and Sphaerodactylus notatus. In contrast to these native species, which were not common to the Disturbed habitat, the two exotic species, Osteopilus septentrionalis and Anolis sagrei were most common there. Species composition of the Dis- turbed habitat primarily depends on the historical topography of the area. The vast majority of species there are generalists, but the area is large enough that local variations in hy- droperiod attract a number of species more commonly associated with drier or wetter conditions and future analyses of this very complex area will involve a more specific separa- tion of habitat types within the area. Clearly, most of the species of am- phibians and reptiles are responding to basic microhabitat requirements that have little to do with the actual species composition of the vegetation (Campbell and Christman 1 982a: 170- 171). Abundance It is impossible to accurately com- pare the trapping results of this study to other studies. The methods, objectives and local circumstances of each study vary widely. Perhaps most confounding is the variability in the number of months per year dur- ing which species are active, and this 82 makes comparisons based on animals per check day difficult. There are also differences in types of arrays used, the purposes of the trapping effort, substrate characteristics and ability to use pit traps, all of which preclude valid comparisons. Campbell and Christman (1982b) summarized their results from north- ern Florida, in which they operated 30 arrays for 7432 array-days. They collected 1644 animals of 43 species from 11 habitats for an average of 0.22 animals per array-day. In LPK, 13 arrays operated a total of 9792 ar- ray-days and collected 1709 animals of 37 species in 4 habitats for an aver- age of 0.18 animals per array day, a similar catch rate per array day. Campbell and Christman (1982b) used both funnel traps and pit traps, and they estimated that only 36% of their collection came from funnel traps. They also state that 69% of the animals trapped were Eleutherodacty- lus planirostris, and that 90% of their trappings were of E. planirostris and Gastrophryne carolinensis. Both of these species were readily trapped in their pit traps. If their pit trap ex- cluded, and look at the percent from funnel traps, there was a much trap yield. There are so many differences in the two studies that the only conclu- sion to be drawn is that the results compare favorably with that the LPK region has a moderate diversity and comparable abundance of animals, based upon similar trapping effort. Comparisons to other studies are even more difficult, since studies in more temperate climates are done only during the warmer months of the year. For example, Clawson and Baskett (1982), in Missouri, used 13 arrays a total of 3159 array days in the spring, summer, and fall, and captured 2545 animals, for an aver- age of 0.81 animals per array day. This much higher figure may well be representative of the greater concen- tration of both animals and resources typically found in more temperate climes. Species Diversity Species richness for southern Florida was described by Duellman and Schwartz (1958:205) as "depauper- ate" and "impoverished." They state that "an impoverished herpetofauna is what might be expected at the end of a long peninsula, through the length of which certain habitats and their inhabitants disappear." The difficulty in evaluating this statement arises from the fact that there is much more involved in the biogeography of the peninsula of Florida than a simple "peninsula ef- fect" due to reduced area and dis- tance from centers of distribution (Robertson and Kushlan 1984). There is also the recent geological origin of the land area, the poor development of soils in the area during the time since emergence, the lack of variation in relief of the area (Olmsted and Loope 1984), and the severe human disturbance. All of these factors need to be considered in evaluating the possible reasons for an "impover- ished" fauna. Finally there is the is- sue of deciding whether the fauna deserves the label of "impoverished" in the first place. A reduced species list does not by itself determine whether the biomass of the existing species is high or low, e.g. while the species list for fresh water fish is considered low for the area (Loftus and Kushlan 1987) they are the principal food of an enor- mous biomass of wading birds. Robertson and Kushlan (1984:234) have addressed this point: "...the nearly unique ability of the South Florida ecosystem to support such large numbers of 14 species of super- ficially similar secondary and tertiary consumers on a resource base that is reduced in species diversity by bio- geographic factors is generally unap- preciated." and the nesting efforts (1972 or 1974 numbers) of the White Ibis and Wood Storks alone are esti- mated to have required "in excess of 3 billion kilocalories or approxi- mately 2500 metric tons of food..." As the impact of the remaining 12 species of wading birds is not known and the secondary productivity of South Florida habitats has not yet been studied, the meaning of this en- ergy requirement to the total system is undeterminable." During this study we have col- lected data on 51 species of amphibi- ans and reptiles (table 2). This is not a low figure for an area the size of LPK (8000 ha). Vogt and Hine (1982) list 34 spe- cies of amphibians and reptiles from their study area in southern Wiscon- sin. Clawson and Baskett (1982) list 35 species from their Missouri study area. Clarke (1958) lists 39 species from Osage County, Kansas. In trap- ping studies in the Florida sandhills of Tampa, Mushinsky (1985) lists 27 species. Campbell and Christman (1982b) list 60 species from their ex- tensive study in northern Florida, and this number comes from a vari- ety of sampling techniques in, at least, 11 different habitat types. Gibbons and Harrison (1981) list 68 species from coastal mainland South Carolina and Gibbons and Pat- terson (1978) list 94 species from the Savannah River Plant in South Caro- lina. Myers and Rand (1969) list 100 species for Barro Colorado Island, Panama. Crump (1971) lists 116 spe- cies for the Belem area of Brazil. From the temperate to tropic lati- tudes there is an obvious increase in overall diversity, but the species rich- ness for the LPK is not very low for its latitude. The presence of 51 spe- cies and the fact that many are abun- dant makes it clear that the applica- tion of terms such as impoverished or depauperate must be used in con- text. Rather than pondering the ab- sence of some species (especially when for the group with the least representation in the area, the sala- manders, it is quite clear why they are not common, see above) I find myself, like Robertson and Kushlan (1984, above), more impressed with the actual abundance of animal life in this unique area. 83 Conclusions 1. The species list for the LPK includes at least 51 species, 15 species of amphibians and 36 species of reptiles. The most poorly represented group is the salamanders, the best represented group is the snakes. The survey of current species composition is basi- cally the same as reported 30 years ago for the area by Du- ellman and Schwartz (1958). The fact that there has been no reduction in species rich- ness of the local area should be considered a major benefit of the preservation of the re- gion inside the national park. 2. Amphibians and reptiles of LPK are primarily habitat generalists, usually being found in three of the four major habitat types in the area. The principal separa- tion by habitat is related to the characteristics of the sub- strate, there being a subset of herptiles most commonly found in areas with greater soil development (Ham- mocks and the Disturbed ar- eas) and another subset of herptiles that are more com- mon in seasonally flooded Prairie. The most poorly rep- resented group is that de- scribed as primarily from Xeric, Pineland habitat, and the absence of sandy soils in the rocky Pineland makes this the most fragile compo- nent of the Everglades herpe- tofauna. The findings of this study do not differ signifi- cantly from those of Duell- man and Schwartz (1958) from thirty years ago. The results point out that there is a significant portion of the local herpe to fauna that relies upon the preservation of large contiguous areas of na- tive Pineland interspersed with Hammocks and season- ally flooded Prairie for its continued success. 3. Phenologies of amphibians and reptiles of the LPK can be described as modified temperate zone patterns. While the subtropical charac- ter of the southern coastal portion of peninsular Rorida results in a year long grow- ing season, with only occa- sional frosts, the seasonality of rainfall and the temperate zone origin of the herpe- tofauna results in a tradi- tional spring emergence of the herptiles, tied to increas- ing day length, warmer tem- peratures and the onset of heavy rainfall. 4. Estimates of density and relative abundance remain difficult to give at the current time. Comparison of current trapping results with those of Campbell and Christman (1982a, 1982b) from 11 habi- tats in northern Florida indi- cate a similar level of abun- dance for the two areas, but differences in the actual spe- cies lists, habitat types and methodologies make such conclusions tenuous. Com- parisons of the fauna of the area with those of a wide va- riety of other regions indicate that the herpetofauna of LPK, with the exception of the salamanders, has a mod- erate level of diversity. Acknowledgnrjents I wish to thank all the dedicated stu- dents of ecology and herpetology at F.I.U. who gave their time so will- ingly during the study. Doug Barker, Peter Beck, Laura Brandt, Teresa De- Francesco, Bob Dunne, Ernesto Her- nandez, Liz Lewis, Nancy O'Hare, and Arlene Sackman helped with trap checking and collecting. To Frank S. Bernardino, Jr., Bob Nodell, Todd Steiner, and Joe Wasilewski I owe a great debt for their dedication to the field work. I thank David But- ler and SARLON Industries for the donation of the shade cloth used to make the fencing, and David W. Lee, of F.I.U., for suggesting the use of shade cloth to us. I thank the staffs of the South Florida Research Center and the Division of Resources Man- agement of the park for their pa- tience, generosity, perspectives and spontaneous collection of specimens for our studies. Most of all I wish to thank Gary Hendrix and William B. Robertson, Jr. for their interest and support. This research was sponsored by the U.S. National Park Service-Flor- ida International University Coop- erative Agreement (CA-5000-3-8005, Supplemental Agreement No.2, 1984) and the Horida International Univer- sity Foundation. Literature Cited Aleksiuk, Michael, and Patrick T. Gregory. 1974. Regulation of sea- sonal mating behavior in Thamno- phis sirtalis parietalis. Copeia 1974:681-689. Auffenberg, Walter. 1982. Horida environments and their herpe- tofaunas. Part III. Herpe togeogra- phy. Florida Herpetologist.4:l-36. Brower, James E., and Jerrold H. Zar. 1984. Field and ecological methods for general ecology. Wm C. Brown. Dubuque. 226p. Brown, William S, and William S. Parker. 1976. A ventral scale clip- ping system for permanently marking snakes (Reptilia, Serpen- tes). Journal of Herpetology 10:247-249. Brown, William S., and William S. Parker. 1982a. Niche dimensions and resource partitioning in a Great Basin Desert snake commu- 84 nity. p.59-81. In Herpetological Communities. Norman J. Scott, Jr., editor. U.S. Fish and Wildlife Serv- ice, Wildlife Research Report 13. Washington D.C. Brown, William S., and William S. Parker. 1982b. Growth, reproduc- tion and demography of the racer. Coluber constrictor mormon, in northern Utah. In Contributions to vertebrate ecology and sys- tematics: a tribute to Henry S. Fitch. Richard A. Seigel and oth- ers, editors. Museum of Natural History University of Kansas, Spe- cial Publication No. 10. Lawrence. Busack, Stephen D. and S.Blair Hedges. 1984. Is the peninsular effect a red herring? American Naturalist 123:266-275. Campbell, Howard W., and Steven P. Christman. 1982a. The herpeto- logical components of Florida sandhill and sand pine scrub asso- ciations, p. 163-171. In Herpetologi- cal Communities. Norman J. Scott, Jr., editor. U.S. Fish Wildlife Serv- ice Wildlife Research Report 13. Washington D.C. Campbell, Howard W., and Steven P. Christman. 1982b. Field tech- niques for herpetofaunal commu- nity analysis, p.193-200. In Herpe- tological Communities. Norman J. Scott, Jr., editor U.S. Fish and Wildlife Service Wildlife Research Report 13. Washington D.C. Carpenter, Charles C. 1952. Com- parative ecology of the common garter snake (Thamnophis s. sir- talis), the ribbon snake (Thamno- phis s. sauritus) and Butler's garter snake (Thamnophis butleri) in mixed populations. Ecological Monographs 22:235-258. Chambers, John M., William S. Cleveland, Beat Kleiner and Paul A. Tukey. Graphical Methods for Data Analysis, p. 395. Duxbury. Boston, Mass. Clark, Donald R. Jr. 1970. Ecological study of the worm snake, Car- phophis vermis (Kenicott). Museum of Natural History Publication, University of Kansas 19:85-194. Clark, Robert F. 1958. An ecological study of reptiles and amphibians in Osage County, Kansas. The Emporia State Research Stud- ies.7:l-52. Clawson, Mary E., and Thomas S. Baskett. 1982. Herpetofauna of the Ashland Wildlife Area, Boone County, Missouri. Transactions, Missouri Acad, of Science 16:5-16. Cook, R.E. 1969. Variation in species density in North American birds. Systematic Zoology 18:63-84. Crump, Martha L. 1971. Quantitative analysis of the ecological distribu- tion of a tropical herpetofauna. Museum of Natural History, Uni- versity of Kansas, Occasional Pa- per 3:1-62. Crump, Martha L. 1982. Amphibian reproductive ecology on the com- munity level, p. 21 -36. In Herpeto- logical Communities. Norman J. Scott, Jr., editor. U.S.Fish and Wildlife Service Wildlife Research Report 13. Washington D.C. Davis, J. H. 1943. The natural features of southern Rorida. Geological Bulletin 25, Rorida Geological Survey. Dinnen, J. Walter. 1984. The fishes of the everglades, p. 258-268. In Envi- ronments of South Rorida Present and Past II. Patrick J. Gleason, edi- tor. Miami Geol. Soc, Miami. Duellman, William E., and A. Schwartz. 1958. Amphibians and reptiles of southern Rorida. Bulle- tin of the Rorida State Museum 3:181-324. Ewel, John J., Dennis S. Ojima, Dori A. Karl, and William F. DeBusk. 1982. Schinus in successional eco- systems of Everglades National Park. Report T-676. Homestead, Rorida: Everglades National Park. South Florida Research Center. Gibbons, J. Whitfield, and Karen K. Patterson. 1978. The reptiles and amphibians of the Savannah River Plant. Savannah River Ecology Plant. DuPont Press, SRO-NERP-2: 1-24. Gibbons, J. Whitfield, and Julian R. Harrison III. 1981. Reptiles and amphibians of Kiawah and Caper Islands, South Carolina. Brimley- ana 5:145-162. Gibbons, J. Whitfield, and Raymond D. Semlitsch. 1981. Terrestrial drift fences with pitfall traps: an effec- tive technique for quantitative sampling of animal populations. Brimleyana. 7:1-16. Hamilton, W.J. 1951. The food and feeding behavior of the garter snake in New York State. Ameri- can Midland Naturalist 46:385- 390. Heatwole, Harold. 1982. A review of structuring in herpetofaunal as- semblages. p.1-19. In Herpetofau- nal Communities. Norman J. Scott, Jr., editor. U.S. Fish and Wildlife Service Wildlife Research Report 13. Washington D.C. Horn, Henry S. 1966. Measurement of "overlap" in comparative eco- logical studies. American Natural- ist 100:419-424. Johnson, J.M., I.C. Olmsted, and O.L. Bass Jr. 1983. Vegetation map of Long Pine Key, Everglades Na- tional Park. South Rorida Re- search Center. U.S. National Park Service. Homestead. Kiester, A. Ross 1971. Species diver- sity of North American amphibi- ans and reptiles. Systematic Zool- ogy 20:127-137. Kushlan, James A. and Frank J. Maz- zotti. 1985. Environmental effects on a coastal population of gopher tortoises. Journal of Hepetology 18:231-239. Layne, James N. 1984. The land mammals of south Florida, p.269- 296. In Environments of South Rorida Present and Past II. Patrick J. Gleason, editor. Miami Geologi- cal Society, Miami. Loftus, William F. and James A. Kushlan. 1987. Freshwater fishes of southern Rorida. Bulletin of the Rorida State Museum, Biological Sciences. 31:147-344. Loraine, Raymond K. 1985. Seasonal changes in foraging success and diet composition of Seminatrix pygaea. Abstract from 1985 com- 85 bined meeting of the Society for the Study of Amphibians and Rep- tiles and Herpetologisf s League, Tampa, Florida. August 4-9, 1985. MacArthur, R.H., and Wilson. E.O. 1967. The theory of island biogeo- graphy. Princeton University Press. Princeton N.J. Moler, Paul E. 1985. A new species of frog (Ranidae: Ram) from north- western Florida. Copeia. 1985:379- 383. Mushinsky, Henry R. 1985. Fire and the Rorida sandhill herpetofaunal community: with special attention to responses of Cnemidophorus sexlineatus. Herpetologica 41:333- 342. Myers, Charles W., and A. Stanley Rand. 1969. Checklist of amphibi- ans and reptiles of Barro Colorado Island, Panama, with comments on faunal change and sampling. Smithsonian Contributions to Zo- ology. 10:1-11. Smithsonian Insti- tution Press. Washington D.C. Olmsted, Ingrid C, Lloyd L. Loope, and Charles E. Hilsenbeck. 1980. Tropical hardwood Hammocks of the interior of Everglades National Park and Big Cypress National Preserve. National Park Service, South Florida Research Center Report T-604. Everglades National Park, Homestead, FL. 58 p. Olmsted, Ingrid C, William B. Robertson Jr., Jill Johnson, and Oron L. Bass Jr. 1983. The vegeta- tion of Long Pine Key, Everglades National Park. National Park Serv- ice, South Horida Research Center Rep. Sn 10 mm. Movements in 1985 occurred despite bitter cold and prolonged drought. In contrast, striped swamp snakes did not leave the pond during the cold weather of 1985, but waited un- til temperatures moderated in early March (fig. 5). Unlike newts, how- ever, they did not return in appre- ciable numbers later in 1986 or 1987 despite favorable habitat and climatic conditions. Orientation and Movement Patterns: Gastrophryne carolinensis and Notophthalmus perstriatus <0 - M • 108B N-34 f!' !•!« ;! A -1016 N-364 ■•10B7 N-449 III in!' i rrrpr HP ■ I 101 •» i i TTT *l 1 I 1. JAN FEB MAR APR —I Figure 4.— Comparison of thie numbers of striped newts (Notophthalmus persfriafus) cap- tured from January 16 ttiroughi April 16, 1985-1987. The stars indicate days of > 10 mm rain- fall. The frequency of bucket capture, both inside and outside the drift fence, varied significantly for both adult G. carolinensis and N. perstria- tus in 1986 and 1987 (table 2). These data indicate non-random movement into and out of the pond. There was no significant correlation between inside and outside bucket capture frequency for G. carolinensis in 1986 (r^ = -0.20, 22 df) or 1987 (r^ = -0.25, 22 df). There was significant correla- tion between inside bucket captures between 1986 and 1987 (r = 0.35, 22 s ' df) but not between outside bucket captures between years (r^ = 0.06, 22 df). These results indicate that nar- row-mouthed toads left the pond in similar directions but entered it from different directions. Juvenile G. carolinensis entering and exiting Breezeway Pond showed distinct differences between capture frequency at different stations (X^ = 535.73, df = 22, P < 0.001). However, 91 they showed no correlation with adult capture frequency per station (r^ = 0.09, 22 df). There also was no correlation in bucket capture fre- quencies for juveniles caught inside and outside the drift fence (r^ = 0.26, 22 df). These data apply only to 1986 because no juveniles were observed in 1987. For N, perstriatus, there was like- wise no significant correlation in in- side versus outside bucket capture frequency in 1986 (r^ = 0.23, 22 df) or 1987 (r^ = 0.03, 22 df). Capture fre- quencies were compared outside the fence in 1986 versus 1987 (r^ = 0.07, 22 df, P > 0.05) and inside the fence in 1986 versus 1987 (r^ = 0.55, 22 df, P < 0.01). As with Gastrophryne, these results suggest that newts were leav- ing the pond in similar directions be- tween years, but that they were en- tering it from different directions. Habitat Relationships Adult Gastrophryne did not m.ove to- ward specific habitats in either 1986 (X2 = 2.62, 2 df, P = 0.27) or 1987 (X^ = 0.32, 2 df, P = 0.85). On the other hand, juvenile narrow-mouthed toads moved toward the sandhills at a higher frequency than would be expected if movements were random (X2 = 13.31, 2 df, P = 0.001), but not toward the pond from any particular direction (X^ = 2.26, 2 df, P = 0.32). Striped newts showed non-random movement in 1986 (X^ = 7.79, 2 df, P = 0.02) toward the sandhills but in 1987 moved toward the Panicum meadow more often than would be expected by chance alone (X^ = 9.42, 2 df, P = 0.009). Movement in rela- tion to nearby habitat is illustrated in figure 6. Discussion Was Sampling Effective? Although we caught 39 species in > 7,000 captures, it is likely that more 14^ II • > < o c U I n < > • o z 4- SEMINATRIX PYQAEA - BREEZEWAY POND t A I t 1 1 I ff 1 1 O' — - " 1985 N-73 1986 N-2 * 1987 N-1 II ■ It JAN FEB a ill I ill I III I . I i i I Mill I IIII.I MAR APR te Figure 5— Comparison of the numbers of swamp snal 10 mm rainfall. species of amphibians and reptiles occasionally visit Breezeway Pond. Some species, such as the eastern coach whip snake (Masticophis flagel- lum), Florida pine snake (Pituophis melanoleucus), and gopher tortoise (Gopherus polyphemus), are common in adjacent sandhills but have not been observed in or near the pond. Large snakes (e.g., Pituophis, Mastico- phis) could easily go over the fence and thus avoid capture. The barking treefrog (Hyla gratiosa) bred in the pond before the initiation of our r Table 2.— Is the frequency of bucket capture random inside and outside the drift fence? For ail analyses, there were 23 stations and 22 df. A signifi- cant value indicates non-random movement. Species Year Orientation X2 P Gosfrophryne 1986 Inside 55.68 < 0.001 carolinensis 1986 Outside 81.25 < 0.001 1987 Inside 84.00 < 0.001 1987 Outside 100.69 < 0.001 Nofophfhalmus 1986 Inside 243.56 < 0.001 perstriatus 1986 Outside 93.44 < 0.001 1987 Inside 88.45 < 0.001 1987 Outside 145.48 < 0.001 92 study (R. Franz, pers. comm.), but we have never captured it or heard it calling from the pond. Some species, particularly treefrogs such as Hyla femoralis, might be able to climb over the fence and thus go undetected (Gibbons and Semlitsch 1982). Newts (N. viridescens) are known to scale drift fences (Semlitsch and Pechmann 1985) although we have not observed N. perstriatus doing so. We have ob- served a substantial number of un- marked newts inside the drift fence even after two years of study, but we do not know if they were residents that were moving after remaining in the pond area for several years, or if they entered by crawling over or un- der the drift fence. Harris et al. (1988) noted that many adult N. viridescens burrowed into mud at the edge of North Carolina sandhills ponds as the ponds dried. For these reasons, our data proba- bly underrepresent both the number of species and individuals using the pond during the two years of obser- vation. On the other hand, it is un- likely that some species (e.g., Bufo, Scaphiopus) are able to climb the fence. As such, capture results of these species may provide a reasona- bly accurate estimate of pond use. Activity Patterns It is difficult to interpret data on ac- tivity patterns of species with only two years of data because there are many variables that influence activity cycles and the timing of reproduc- tion. These variables, such as rainfall amount and distribution, maximum and minimum temperatures, and hydroperiod (Wiest 1982, Semlitsch 1985, Pechmann et al. 1988), vary daily, seasonally and yearly, and may affect different species in differ- ent ways. The subtle interaction of these parameters probably accounts for the variation in activity patterns observed between years (Semlitsch 1985, Semlitsch and Pechmann 1985). HABITAT DISTRIBUTION GC 1986 JUV : LEAVING POND GO 1986 ADULT GC 1987 NP 1986 NP 1987 Figure 6.— Diagram illustrating the relationship between buckets, emigration from the pond, and nearby habitat for Gastrophryne corolinensis (GC) and Notophthalmus perstriatus (NP). 93 Amphibians breeding in sandhills ponds are faced with substantial un- certainty as to whether or not suit- able conditions will prevail for repro- duction. Breezeway Pond was cho- sen as the site for our study because it had consistently held water from the spring of 1983 through January 1985 (R. Franz, pers. comm.). Begin- ning in January, climatic conditions changed resulting in two years of drought with only sporadic free wa- ter. Temporary ponds may allow re- production free of certain predators, but their use comes at the cost of re- productive uncertainty. Amphibians are active during or immediately after periods of rainfall or high humidities. However, the interaction of moisture and tempera- ture and how they affect condensa- tion probably affects diel activity (Semlitsch and Pechmann 1985, Du- ellman and Trueb 1986, Pechmann and Semlitsch 1986) but also seasonal activity. The extremely dry conditions at Breezeway Pond during the study makes it difficult to predict whether patterns observed in early 1985 and from late 1985 through late 1987 are "typical" for the amphibian commu- nity using the pond. Observations from other long-term studies of her- petofaunal communities suggest that there is wide variation in numbers of individuals at a site and in reproduc- tive success from year to year (Gill 1978, Semlitsch 1983, 1985, 1987, Pechmann et al. 1988). Because of their lack of depend- ence on standing water, temperature is probably more important than hy- droperiod in governing reptile daily and seasonal activity, at least for spe- cies in direct spatial proximity to the pond. However, reptile predators that opportunistically visit tempo- rary ponds, such as garter snakes (Thamnophis sp.), might increase the number of visits and duration of stay if a sufficiently long hydroperiod al- lows amphibian reproduction to take place. Our data are insufficient as yet to answer this question. Some individuals are active even during unfavorable environmental conditions of drought and unseason- ally cold temperatures. Amphibians and reptiles are generally, but not always inactive during cold or dry weather. For instance, Semlitsch (1983, 1985) noted that mole sala- manders {Amhystoma sp.) in South Carolina bred during the coldest but not necessarily the wettest months. He felt that most animals moved to breeding ponds at this time to allow sufficient time for larval develop- ment prior to pond drying (Semlitsch 1987). Such may not explain winter/ early spring breeding in N. perstriatus because the breeding period is ex- tended (Bishop 1947) and larvae have been found from April through De- cember (Christman and Means 1978). The larval period is unknown, but its duration is critical to successful re- production in temporary sandhills ponds. Individuals moving at times of unusually cold and dry weather may be searching for more favorable re- treats or escaping adverse condi- tions. If the onset of migration (sensu Semlitsch 1985) commenced during unusually adverse conditions, and the unfavorable conditions extended for a long period of time, the popula- tion could be vulnerable to local ex- tinction via mortality or emigration. Prolonged drought brought about the local extinction, via emigration, of the resident Semimtrix population. Movement Patterns and Orientation Because of the small size of Breeze- way Pond, it is difficult to ascribe directed movements of individuals as migrating to, or originating from, a specific habitat type. Because the pond was located in an ecotone, an animal captured at buckets facing the interface between sandhills and xeric hammock could move in either direc- tion once beyond the fence. Likewise, an animal originating from one habi- tat type could be misclassified if it moved a relatively short distance and fell into a bucket facing a differ- ent habitat type. The open field was also rather small and, although we did not feel comfortable assigning buckets 4-6 to sandhills or xeric ham- mock, it is likely that animals exiting or entering the pond through these buckets came from or went to one or the other habitat. Given these qualifications, adult Gastrophryne did not exhibit habitat preferences, although juveniles left the pond primarily toward sandhills. Gastrophryne are commonly recorded in sandhills (Carr 1940, Campbell and Christman 1982, Mushinsky 1985) and have been found in sandhills > 100 m from the nearest water source (Franz 1986, Dodd pers. obs.). Xeric hammock or sandhills apparently provide narrow-mouthed toads suitable cover and resources away from the breeding pond, but why juvenile Gastrophryne would move toward sandhills is unknown. Striped newts are most commonly found in flatwoods ponds in pine- palmetto habitats (Christman and Means 1978) as well as ponds in sandhills and scrub areas (Campbell and Christman 1982). To what extent they use sandhills habitats away from ponds is unknown. Carr (1940, reported as N. v. symmetrica) re- corded efts in high and mesophytic hammocks in light, porous soil. However, striped newts at Breeze- way Pond moved toward sandhills or meadow rather than hammock. Migration distances of striped newts are unknown although displaced N. viridescens can move 400 m through deciduous forest to return to a resi- dent pond (Gill 1979). N. perstriatus probably can travel similar distances in its migrations. Management implications The Florida sandhills are undergoing extensive habitat alteration because of rapid human population growth 94 and associated development. In the late 1970's, Auffenberg and Franz (1982) estimated that 70.6% of the sand pine-scrub oak, 57% of the long- leaf pine, and 37.7% of the xeric ham- mock communities had been de- stroyed by forest plantation agricul- ture and urbanization. In Putnam County, the site of our study, > 50% of the land area originally supporting such communities no longer does so. With projected human population increases of more than 300% between 1972 and 2000 (Auffenberg and Franz 1982), there has been increasing con- cern for the loss of sandhills habitats in northern and central Florida. Ex- tensive loss of habitat is occurring in other portions of the state and South- east, such that only 14% of the long- leaf pine (Pinus palustris) forests re- main from estimates of over 70 mil- lion acres that once comprised this community (Means and Grow 1985). Because of habitat loss, amphibian and reptile populations dependent upon sandhills probably are declin- ing. Many of the amphibians, such as the Rorida gopher frog. Ram areolata aesopus, and the striped newt, N. per- striatus, are considered endangered, threatened, or rare (Fogarty 1978, Christman and Means 1978), yet there are few data on their life histo- ries or population dynamics. The paucity of information on spe- cies composition and population dy- namics of amphibians and reptiles that use temporary ponds in xeric habitat masks the probable impor- tance of such habitats. Variation in annual habitat use, both intraspecifi- cally and inter-specifically, appears to be considerable. Long-term eco- logical studies of the herpetofaunal community are needed to under- stand the magnitude of such vari- ation and its potential significance. Information on the biology of the species comprising the sandhills her- petofaunal community could be im- portant in planning for the manage- ment of sandhills ecosystems by State and Federal agencies. For in- stance, Florida Statutes Section 373.414 required Water Management Districts to adopt rules to establish specific permitting-criteria for small isolated wetlands, including size thresholds below which impacts on fish and wildlife habitats would not be considered. When these rules were adopted, almost no data were available on herpetofaunal communi- ties on which to make recommenda- tions for size threshold considera- tions. Lack of information led, in part, to variation among regulations adopted by the different Water Man- agement Districts. There is considerable interest among Rorida biologists, conserva- tionists, and land use planners in the concept of wildlife corridors to main- tain biotic diversity (Harris 1985). Unfortunately, most discussions have centered on riparian habitats. The lack of data on sandhills habitat use, especially by candidate endan- gered or threatened species, could hamper the long-term survival of such species. Many sandhills species are likely dependent on small iso- lated wetlands for at least a portion of their life cycle. By focusing on ri- parian habitats, planners may be overlooking the importance of up- land habitats and their associated small wetlands to the maintenance of biotic diversity. The following are the most impor- tant implications of our study for the conservation and management of small isolated wetlands and their as- sociated herpetofaunal communities in "high pine" xeric habitats in northern and central Rorida. These should be kept in mind when evalu- ating impacts of habitat loss and planning assessment studies. 1. Many species use these habi- tats: some are permanent residents, some are migrants, and some wander through the area on an irregular ba- sis. All pond-breeding spe- cies live in surrounding ter- restrial habitats during the non-breeding season. Thus, the pond and a portion of the terrestrial habitat are both critical to species persistence. 2. Such habitats are used year- round despite seemingly un- favorable periods of drought and cold weather. 3. Species composition varies within a year: some species are found only in one season, some predominate at one time but are found com- monly at other times, some are very rarely observed. 4. Reproductive output among species varies considerably: in one year spring breeders may be successful, in other years summer breeders may be successful, in some years both probably produce young, in other years neither may successfully reproduce. The longer that studies are conducted, the greater is the likelihood that multiple pat- terns will emerge. 5. Activity patterns change sea- sonally and annually proba- bly in response to environ- mental cues, particularly rainfall, temperature, and hydroperiod. 6. To determine the total num- ber of species using such wetlands, spring and early summer sampling produces the best results, but single season or even yearly sam- pling will not catch all spe- cies. 7. Quick surveys underestimate both numbers of species and individuals, as well as an- nual variation, and thus un- derestimate the importance of temporary isolated wet- lands in sandhills. 95 8. To adequately understand complex communities, long- term studies are absolutely- essential for management and conservation. Acknowledgments We thank H. 1. Kochman for advice on statistical analyses, and R. Ash- ton, R. Franz, J. Oldemeyer, J. H. K. Pechmann, R. Seigel and R. D. Sem- litsch for their comments on the manuscript. R. L. Burke, K. M. Enge, and J. N. Stuart assisted with various phases of fieldwork. Literature Cited Anderson-Bell. 1984. ABSTAT. Re- lease 4. Anderson-Bell, Canon City, Colorado. Auffenberg, Walter, and Richard Franz. 1982. The status and distri- bution of the gopher tortoise (Go- pherus polyphemus). p. 95-126. In North American Tortoises: Con- servation and ecology. R. Bruce Bury, editor. Fish and Wildlife Service, Wildlife Research Report 12. Bishop, Sherman C. 1947. Handbook of Salamanders. Cornell Univer- sity Press, Ithaca, New York. Campbell, Howard W., and Steven P.Christman. 1982. The herpeto- logical components of Horida sandhill and sand pine scrub asso- ciations, p. 163-181. In Herpeto- logical Communities. Norman J. Scott, editor. Fish and Wildlife Service, Wildlife Research Report 13. Carr, Archie Fairly, Jr. 1940. A contri- bution to the herpetology of Ror- ida. University of Florida Publica- tion, Biological Science Series 3(1):1-118. Christman, Steven P., and D.Bruce Means. 1978. Striped newt Notoph- thalmus perstriatus. p. 14-1 5. 7n Rare and Endangered Biota of Horida, Vol. 3, Amphibians & Reptiles. Roy W. McDiarmid, editor. Uni- versity Presses of Horida, Gainesville. Duellman, William E., and Linda Trueb. 1986. Biology of Amphibi- ans. McGraw-Hill Book Co., New York. Fogarty, Michael J. 1978. Florida go- pher frog Rana areolata aesopus. p.5-6. In Rare and Endangered Bi- ota of Florida, Vol. 3, Amphibians & Reptiles. Roy W. McDiarmid, editor. University Presses of Flor- ida, Gainesville. Franz, Richard. 1986. Gopherus pol- yphemus (Gopher Tortoise). Bur- row commensals. Herpetological Review 17:64. Gibbons, J. Whitfield, and David H. Bennett. 1974. Determination of anuran terrestrial activity patterns by a drift fence method. Copeia 1974:236-243. , and Raymond D. Semlitsch. 1982. Terrestrial drift fences with pitfall traps: an effective technique for quantitative sampling of ani- mal populations. Brimleyana 7:1- 16. Gill, Douglas E. 1978. The metapopu- lation ecology of the red-spotted newt; Notophthalmus viridescens (Rafinesque). Ecological Mono- graphs 48:145-166. . 1979. Density dependence and homing behavior in adult red- spotted newts Notophthalmus viridescens (Rafinesque). Ecology 60:800-813. Harris, Larry D. 1985. Conservation corridors, A highway system for wildlife. ENFO (Florida Conserva- tion Foundation), November, 10 p. Harris, Reid N., Ross A. Alford, and Henry M. Wilbur. 1988. Density and phenology of Notophthalmus viridescens dorsalis in a natural pond. Herpetologica 44:234-242. Jones, K. Bruce. 1986a. Amphibians . and reptiles, p.267-290. In Inven- tory and Monitoring of Wildlife Habitat. A. Cooperrider, R.J.Boyd and H.R.Stuart, editors. Bureau of Land Management Service Center, Denver, Colorado. . 1986b. Data types, p. 11-28. In Inventory and Monitoring of Wildlife Habitat. A. Cooperrider, R. J. Boyd and H. R. Stuart, edi- tors. Bureau of Land Management Service Center, Denver, Colorado. Macan, T. T. 1966. The influence of predation on the fauna of a moorland fishpond. Archiv fur Hydrobiologie 61:432-452. Means, D. Bruce, and Gerald Grow. 1985. The endangered longleaf pine community. ENFO (Horida Conservation Foundation), Sep- tember, 12 p. Moler, Paul E., and Richard Franz. 1988. Wildlife values of small, iso- lated wetlands in the southeastern coastal plain, p. 234-241, In Pro- ceedings of the Third Southeastern Nongame and Endangered Wild- life Symposium. R. R. Odom, K. A. Riddleberger, and J. C. Ozier, edi- tors. Georgia Department of Natu- ral Resources, Athens, Georgia. Mushinsky, Henry R. 1985. Fire and the Horida sandhill herpetofaunal community: with special attention to responses of Cnemidophorus sexlineatus. Herpetologica 41:333- 342. Pechmann, Joseph H. K. and Ray- mond D. Semlitsch. 1986. Diel ac- tivity patterns in the breeding mi- grations of winter-breeding anu- rans. Canadian Journal of Zoology 64:1116-1120. , David E. Scott, J. Whitfield Gib- bons, and Raymond D. Semlitsch. 1988. Influence of wetiand hy- droperiod on diversity and abun- dance of metamorphosing juvenile amphibians. Wetlands Ecology and Management 1:1-9. SAS Instihite Inc. 1985. SAS Intro- ductory Guide for Personal Com- puters, Version 6 Edition. SAS In- stitute Inc., Gary, North Carolina. Semlitsch, Raymond D. 1983. Struc- ture and dynamics of two breed- ing populations of the eastern ti- ger salamander, Amhystoma ti- grinum. Copeia 1983:608-616. . 1 985. Analysis of climatic factors influencing migrations of the 96 salamander Amhy stoma talpoideum. Copeia 1985: 477-489. . 1987. Relationship of pond drying to the reproductive success of the salamander Amby stoma talpoideum. Copeia 1987:61-69. , and Joseph H. K. Pechmann. 1985. Diel pattern of migratory activity for several species of pond-breeding salamanders. Copeia 1985:86-91. Sexton, Owen J., and Christopher Phillips. 1986. A qualitative study of fish-amphibian interactions in 3 Missouri ponds. Transactions, Missouri Academy of Science 20:25-35. Wiest, John A., Jr. 1982. Anuran suc- cession at temporary ponds in a post oak-savanna region of Texas, p. 39-47. In Herpetological Com- munities. Norman J. Scott, Jr., edi- tor. Fish and Wildlife Service, Wildlife Research Report 13. 97 Management of Annphlbians, Reptiles, and Snnall Mammals In Xeric PInelands of Peninsular Florida^ I. Jack $tout,2 Donald R. Richardson,^ and Richard E. Roberts^ Abstract.— The primary xeric pinelands of peninsu- lar Florida are longleaf pine/turi 18 spe- cies). This diversity includes large forms, e.g., Drymarchon corais couperi and Cro talus adamanteus, and small, specialized species like Stilosoma ex- tenuatum. This latter ophiophagous species feeds largely on Tantilla relicta; Tantilla, is in turn specialized on Tenebrionidae larvae (Mushinsky 1984). Small Mammals At least 19 species of small mammals with body masses less than 6.0 kg may be anticipated in LLP/TO sand- hills (table 2). Two are fossorial, Scalopus aquations and Geomys pinetia, 1 semi-fossorial, P. polionotus, and 2 occur in the surface litter, Blarina carolinensis and Cryptotis parva. Arboreal species include Sciurus carolinensis, S. niger, Glaucomys volans, P. gossypinus, and Ochrotomys nuttalli. Podomys floridanus nests in the burrows of the gopher tortoise and the pocket gopher (Layne 1969); it may enlarge other openings in the soil to establish burrows independ- ently of the gopher tortoise (R. E. Roberts, personal observation). Dasypus novemcinctus is the only ex- otic species of mammal that is clearly established in the sandhill commu- nity. Sand Pine Scrub Amphibians and Reptiles Campbell and Christman (1982) listed 64 species of reptiles and am- phibians that may be found in LLP/ TO sandhills and SPS. Pitfall trap- ping in six different even-aged stands of SPS on the Ocala National Forest by Christman et al. (unpub- lished manuscript and personal com- munication) revealed 27 species (table 1). Of 1,624 individuals Table 1.— Herpetofauno trapped or observed within the xeric pinelonds of peninsula Florida. Standard herp arrays were used in each study to sample for a period of at least one year. Species Long leaf pine/ Sand pine Species Long leaf pine/ Sand pine turkey oak scrub turkey oak scrub Campbell Mushinsky Stout Christman Campbell Mushinsky Stout Christman & Christman 1965 etaL etai. & Christman 1985 etal. etal. 1982 unpubl. unpubl. 1982 unpubl. unpubl. Nofophfhalmus viridescens N. persfriafus ScopNopus holbrookii Bufo ferresfris Bufo quercicus Eleufh ero dac fylus planirosfris HyJa fern or alls Hyla grafiosa Hyla squirella Hyla cinerea Acris gryllus Rana gryllo Rana areolata Rana ufricularia Gasfrophryne carolinensis Kinosfernon bauri Terrapene Carolina bauri Gopherus polyphemus Anolis carolinensis Anolis sagrei Sceloporus undulafus Sceloporus wood! Ophisaurus compressus Cnemldophorus sexlineatus X X X X X X X X X X X X X X X X X X X X X X X X X X Scincella lateralis x Eumeces inexpectatus x E, egregius lividus — E. egregius onocrepis x Neoseps reynoldsi x Rhineura floridgna x Nerodia fasciafa — Th amn ophis saurifus — Riiadinaea flavilafa — Diadophis punc tafus x Paranoia abacura — Coluber constrictor x Masticophis flagellum x Opheodrys aestivus x Drymarclion corais — Baphe guttata — Pituophls melanoleucus x Lampropeltis triangulum x Stilosoma extenuatum x Cemophiora coccinea x Tantilla relicta x Heterodon platyrhinos x Heterodon simus — Micrurus fulvius fulvius x Sistrurus millarius x Crotalus adamanteus — Totals 29 X X X X 27 X X X X X X X X X X X 33 27 100 trapped, the common species were Bufo terrestris (n=332), Cnemidophorus sexlineatus (n=329), and Sceloporus woodi (n=216); five species were rep- resented by single captures. Christ- man et al. concluded that the herpe- tofaunal diversity declined with in- creasing age of SPS stands. Gopherus polyphemus is the key- stone species in SPS but is less com- mon there than in LLP /TO (Auf fen- berg and Franz 1982). Many, if not most, of the burrow commensals are common in SPS (Cox et al. 1987). Podomys floridanus is an example. [ Small Mammals Fourteen species of small mammals commonly inhabit SPS (table 2). Pod- omys floridanus is a predictable mem- ber of the assemblage throughout the range of scrubs in peninsular Florida (Layne 1978). Three subspecies of Peromyscus polionotus occur in scrubs of the interior and east coast portions of the peninsula. Common small mammals in central peninsular Flor- ida scrubs include Podomys floridanus, Peromyscus gossypinus, Ochrotomys nuttalli, and Glaucomys volans (Swin- dell 1987). Podomys floridanus is the predominate small mammal in scrubs of southeast Florida (Richardson etal. 1986). Limited data suggest Spilogale putorius is a major predator on small mammals in scrubs with lesser roles played by Mephitis mephitis and Mustela frenata (Stout and Roberts, personal observations). Table 2.— Small mammal community structure In sandhill and sand pine scrub plant associations of peninsular Florida. The upper limit of body mass of small mammals was arbitrarily set at 6.0 kg. Mammal Species Longleaf pine/turkey oak' -^-^ Sand pine scrub* Didelphis virginiana X X Crypfofis parva X Blarina carolinensis X X Scalopus aquaficus X Dasypus novemcinctus X X SyMlagus floridanus X X Sciurus carolinensis X X Sciurus niger X Glaucomys volans X X Geomys pinefis X Peromyscus polionotus X X Peromyscus gossypinus X X Podomys floridanus X X Ochrotomys nuttalli X X Sigmodon hispidus X X Urocyon cinereoargenfeus X Procyon lotor X X Mustela frenata ? X Spilogale putorius X X Mephitis mephitis X No. Species 19 14 ^ St out et al., unpublished ^Arata 1959 ^Humphrey et al. 1985 "Stout 1982 ENDANGERED AND THREATENED SPECIES Ten species of amphibians, reptiles, and small mammals associated with xeric pineland are currently listed as having some level of threatened, en- dangered, or sensitive status by ei- ther the state of Florida or the De- partment of Interior (table 3). The extensive overlap in species composi- tion between the two pineland com- munities results from the high num- ber of species common to both types. The Endangered Species Act charges federal agencies with the responsibil- ity to manage federally listed species on federally owned lands. At the state level, preservation of these listed species is of major concern when they occur on parcels of land scheduled for large-scale develop- ment. Preserve design and manage- ment practices for these species have largely evolved on an ad hoc basis without adequate time for an evalu- ation of the management or the long- term implications for the species. MANAGEMENT OF XERIC PINELANDS ON PUBLIC LANDS Of three national forests in Rorida, only the Ocala National Forest is lo- cated in the peninsula. It totals 153,846 ha of which 85,020 ha are SPS and 18,219 ha LLP/TO. The National Forest Management Act (1976) and pursuant regulations (36 CFR 219) require that each forest be managed to maintain well-distributed and vi- able populations of wildlife species, including species that are endan- gered or threatened (Norse et al. 1986). Silvicultural systems differ be- tween the two pineland communi- ties. On the Ocala National Forest sand pine scrub is routinely har- vested in patchy clearcuts that range from 16-24 ha in area. Scrub under- story vegetation is allowed to regen- erate naturally; however, sand pine is seeded following site preparation 101 by a single roller chopping. The har- vest rotation length is about 50 years. In contrast, LLP /TO is ostensibly managed on a 80-100 year rotation and shelterwood cutting favors natu- ral regeneration of the longleaf pine (Don Bethancourt, personal commu- nication). In practice, harvesting of longleaf pine may occur in 60 years. Effectiveness of ecosystem man- agement in the SPS community v;^ill be judged by the response of desig- nated indicator species, such as go- pher tortoises and scrub jays (Aphelo- coma coerulescens) (table 3). The go- pher tortoise is also a designated in- dicator species for the LLP/TO com- munity. The significance of the go- pher tortoise as a keystone species was emphasized in 1986 when har- vesting of the species on national for- ests in Florida was made illegal through an agreement between the U.S. Forest Service and the Florida Game and Fresh Water Fish Commis- sion. Other species-specific manage- ment practices involving amphibians, reptiles, or small mammals have not been deemed necessary to carry out on the Ocala National Forest (Don Bethancourt, personal communica- tion). In fact, the impact of timber harvesting on small vertebrates of LLP/TO and SPS communities is simply not known. Public lands in Florida supporting xeric pinelands include, but are not limited to, state forests and state parks. State forests with large acre- ages of LLP/TO, e.g., the Withla- coochee State Forest, are managed at the ecosystem level. Prescribed burn- ing is done every 3-8 years and fu- ture timber sales will follow a rota- tion length of 80-120 years; currently rotation lengths are about 60 years and are not regarded as favorably for endemic wildlife. Wildlife manage- ment areas overlap the state forest holdings and are managed for sus- tained yields of wildlife by the Flor- ida Game and Fresh Water Fish Commission based on a memoran- dum of understanding between agencies (Cathy Ryan, personal com- munication). Table 3.— Endangered and potentially endangered amphibians, reptiles, and small mammals (Wood 1 987) inhabiting xeric pinelands of peninsular Florida. Species group Xeric pineland Designated status' LLP/TO SPS FGFWFC2 USFWS^ Amphibians and Reptiles Drymarchon cords couperi Eumeces egregius lividus Gopherus polyphemus Neosepsreynoldsi Pifuophis melanoieucus mugifus Rona areolafa Sfilosoma exfenuafum Mammals Geomys pinefis goffi Podomys floridanus Sciurus niger shermani X X X X X X X X X X X X X X X X X X T T I ssc T SSC SSC T I E SSC SSC T T UR2 T UR2 UR2 UR2 UR3 UR2 UR2 'f= Endongeredj T=Jhreatened; SSC= Species of Special Concern: UR2= Under re- view for listing, but substantial evidence of biological vulnerability and/or threat is lacking: UR3 = Still formally under review for listing, but no longer being considered for listing due to existing pervasive evidence of extinction. ^Florida Game and Fresh Water Fish Commission ^United States Fish and Wildlife Service State parks are managed by the Division of Recreation and Parks of the Rorida Department of Natural Resources (FDNR). An ecosystem approach is taken in the restoration and management of xeric pinelands on state park lands (Jim Stevenson, personal communication). Prescribed burning has been used since 1969 to control hardwood invasion of LLP/ TO stands and to stimulate growth and flowering of grasses and herbs. Burning in spring and early summer appears to best duplicate the historic timing of lightning initiated fires in xeric pinelands. The impact of these management practices on the plant community has been documented (Davis 1984); the response of reptiles, amphibians, and small mammals is currently under study (Stout et al. unpublished). Generally, mature stands of SPS have not been burned until recently, due to the unpredict- able behavior of fire in the commu- nity; however, a prescription for burning this fuel type has been writ- ten and tested on private land and state parks (Doran et al. 1987). Early recovery stages of SPS appear to sup- port the greatest diversity of reptiles and amphibians. However, as can- opy closure occurs in SPS, ground cover diminishes and habitat quality for gopher tortoises declines (Cox et al. 1987). In contrast, similar numbers of Podomys have been observed in early (R. E. Roberts, unpublished data, J. Dickinson State Park); inter- mediate (Stout 1982); and old growth SPS (James N. Layne, unpublished data, Archbold Biological Station). State parks, reserves, and pre- serves appear to be ideal lands to ex- plore species-specific management measures for herp tiles and small mammals. For example, sand swim- ming herptiles (Smith 1982) require openings that are relatively root free in LLP/TO and SPS habitats. The natural occurrence of such openings may have been due to "hot" spots associated with the combustion of high fuel loads, e.g., fallen trees (Ron Myers, personal communication). 102 Concentrarion of natural fuels prior to prescription bums in SPS would offer a means to create microhabitat conditions favorable for the sand swimmers. MANAGEMENT OF XERIC PINELAND ON PRIVATE LAND Development of Regional Impact Concern with management of am- phibians, reptiles, and small mam- mals on private lands in Rorida de- rives from state and federal protec- tion of endangered species and the development guidelines promul- gated during the Development of Regional Impact (DRI) process. "The Rorida Environmental Land and Water Management Act of 1972" (Chapter 380, Florida Statutes) de- fines developments of regional im- pact in Section 380.06(1), Rorida Stat- utes, as "...any development which, because of its character, magnitude, or location, would have a substantial effect upon the health, safety, or wel- fare of citizens of more than one county (Anonymous 1976)." Large scale development projects in penin- sular Florida commonly involve hun- dreds to several thousand acres of relatively natural landscape. The DRI process requires bona fide studies of wildlife populations and their associ- ated habitats; emphasis is placed on listed species. Developers must pre- pare viable management strategies to accommodate wildlife resources de- pendent upon their lands (Cox et al. 1987; Richardson et al. 1986). Management strategies of devel- opers with xeric pinelands generally follow one of two somewhat overlap- ping approaches to preserve habitat and/or species values: (1) conserva- tion set asides or (2) mitigation. Con- servation set asides are, in principle, the preferred solution. In practice some habitat is dedicated in perpetu- ity as a nature preserve; preserve de- sign currently is a somewhat ad hoc process and will be discussed more completely in a subsequent section of this paper. Very high land values may dictate mitigation rather than on site preservation of habitat. Mitigation may take many forms to compensate for development of xeric pinelands. Restoration of de- graded land (Humphrey et al. 1985), not necessarily xeric pinelands, is one method. Another tactic is to purchase comparable land or some other type of land of equivalent natural value elsewhere and dedicate it to preser- vation. A formal process for accom- plishing this option is presently un- der study by the Florida Game and Fresh Water Fish Commission. Preservation of habitat is the basic purpose of conservation set asides and mitigations. The value of these efforts depends on the proximity to larger, undeveloped tracts of land, travel corridors, area of preserves, and future management options. Another form of mitigation is the relocation of sensitive species from tracts of land to be developed to land dedicated to purposes that are con- sistent with the long-term survival of the relocated species. In Rorida, the gopher tortoise has been the focus of numerous relocation efforts. Diemer (1984) discussed the advantages and disadvantages of relocation of go- pher tortoises as a species manage- ment strategy. Formal research on gopher tortoise relocation was re- cently reported (Proced. Gopher Tor- toise Relocation Symp., 27 June 1987, Gainesville, FL, in press). The Rorida Game and Fresh Water Fish Commis- sion regulates relocations by a permit system based on a standardized relo- cation protocol. Preserve Design Preserve design is an evolving and controversial area of conservation biology (Diamond 1975, 1978; Gilbert 1980; Higgs 1981; Margules 1982; Pickett and Thompson 1978; Pyle 1980; Soule and Simberioff 1986). Large preserves encompassing a mo- saic of xeric pinelands, mesic forests, and seasonal and f>ermanent wet- lands would perhaps offer the ideal landscape unit for long-term preser- vation of amphibians, reptiles, and small mammals in peninsular Ror- ida. Because preserves on private lands must be justified and dedicated through the DRI process, economics dictates preserve units of minimal size. Rarely do we have the opportu- nity to cluster or juxtapose these small units to take advantage of the so called "rescue effect" (Brown and Kodric-Brown 1977). In practice, conservation set asides tend not only to be small in acreage but also only of one habitat type. The latter presents a dilemma for species whose requirements often include two or more contrasting habitats. For example, the gopher frog lives in tor- toise burrows in LLP/TO sandhills during late spring, summer and early fall and migrates to temporary wet season depressions to breed in win- ter and early spring (Moler and Franz 1987). Thus a mosaic of up- land-wetland habitats in close prox- imity are essential to maintain viable populations of this species. Other species such as the indigo snake have home range requirements that in- clude 122-202 ha of several upland- wetland habitat types (Moler 1985; Moler unpublished data). It is obvi- ous that large landscape units are necessary to preserve viable popula- tions of these animals. We have prepared a detailed pre- serve design for a SPS community within the city of Boca Raton, Rorida (Richardson et al. 1986; Stout et al. 1987; manuscript in preparation). The approach taken anticipated Soule and Simberioff (1986) and rec- ommended the area of the preserve be sufficient to support a minimum viable population (Franklin 1980) of gopher tortoises because of their status as the keystone species. Al- though biologically reasonable, this basis for determining preserve size is often economically unrealistic from the view point of the private land- 103 owner. A consortium of public land- owners would, however, permit the purchase and long-term management of the preserve as recommended. Cox et al. (1987) offer guidelines for the design of preserves on private lands to maintain gopher tortoise populations. They employed the computer simulation model POPDYN (Perez-Trejo and Samson manuscript) to determine population viability based on different initial sizes. Populations of 40-50 individu- als were found to be likely (>90%) to persist 200 years. Based on existing literature on home range require- ments. Cox et al. (1987) recom- mended a minimum preserve of 10- 20 ha, depending on habitat quality, to support 40-50 tortoises. Another approach to determining the area of a preserve employs ''inci- dence functions" (Diamond 1978). Incidence functions are species spe- cific and derived from data sets which reveal the fraction of plots (discrete habitats) of different areas that actually support the species. It is a matter of judgement as to the probability of occurrence, e.g. 0.5 as opposed to 0.7, that would set a lower limit to area for an acceptable preserve. Data sets useful for evalu- ating this approach with respect to amphibians, reptiles, and small mammals in xeric pinelands are pres- Table 4.— Incidence of Gopherus polyphemus a keystone species, and Podomys fioridanus in xeric pinelands of peninsular Florida. Presence (+) or absence <-) Is indicated. Study sites are ranked according to area within the xeric pinelands. Quantitative sampling of the 12 LLP/TO study sites con- sisted of 5 days of live-trapping and observation at invervals of 3 months over a period of 18 months (1986-1988). Study sites In SPS were sampled by live -trapping and observation a minimum of 3 consecutive days, often in the same season of consecutive years (Stout et al. unpublished). Incidence of species in xeric pineland Study sites Area LLP/TO SPS (ha) Gopherus Podomys Gopherus Podor Lake Mary 1.2 + — Morningside Nature Center 2.0 + — San Felasco 4.1 + — Spruce Creek 4.1 + — Orange City 5.6 + — Bok Tower 9.3 + — Wekiwa Springs 9.7 + + Suwannee River 10.1 + + O'Lena 10.5 + + J. Butterfield Brooks 15.8 + + Starkey Well Field 16.2 + — Sandhill Boy Scout Camp 16.2 + — Interlachen 21.8 + — Yamato Plaza 2.8 + — Yamato Scrub, B 3.2 + + Quantum Park, A 4.4 + + Quantum Park, B 4.4 + + Quantum Park, C 4.8 + + Yamato Scrub, A 8.5 + + Summit Place 10.5 + — Potomac Road 17.8 + + Cedar Grove 21.5 + + J. Dickinson 256.2 + + ently lacking. Table 4 provides data we have gathered on area of discrete habitats and the presence or absence of gopher tortoises and Florida mice. It is apparent that tortoises are less area sensitive than Florida mice and that Florida mice are patchy in occur- rence in LLP /TO, perhaps only sec- ondarily related to area. Incidence functions do not neces- sarily reveal the minimum area re- quired to support minimum viable populations (Franklin 1980). We be- lieve preserve area should be based on providing this requirement, par- ticularly when preserves are isolated relative to average dispersal dis- tances of keystone species. However, clusters of preserves within dispersal distances of keystone species may be of less area per preserve due to a high likelihood of reinvasion from nearby populations following local population extirpations (Noss and Harris 1986). Monagennent of Preserves in Xeric Pinelands The future viability of preserves de- pends largely on their ownership af- ter development of the surrounding landscape. It is unlikely that home- owners associations will assume the cost of management if preserves re- main as a part of the overall develop- ment's "commons." Public owner- ship is an alternative and might rest with a city, county, or state. Local governments seem more appropriate; however, funds and expertise to manage may be lacking. One pre- serve in south Rorida is designed to border a city park, thus allowing its maintenance and /or management costs to be assumed over time as part of the existing park system (Richardson, personal observation). Regardless of the ownership, a com- mitment to long-term management must be achieved if a preserve is to retain natural values. Management options for nature preserves range from a decision 1) to 104 do nothing and let nature take its course; 2) to manage for maintenance of a viable ecosystem, which implies the natural biota, including amphibi- ans, reptiles, and small mammals, will be present in proportion to their normal abundance; or 3) to focus management on the needs of one or more species. White and Bratton (1980) have exposed the folly of the first management option. The deci- sion to emphasize ecosystem or spe- cies management depends on the en- tity responsible for management, type of preserve, management objec- tives, area of the preserve, nature of the surrounding lands, relative over- all or regional rarity of particular species, and the resources available for management. Management objectives of any preserve should focus on: 1) mainte- nance of normal ecosystem proc- esses; 2) conservation of soil; 3) maintenance or restoration of normal hydrologic conditions; 4) prevention of establishment of exotic species.; 5) and prevention of human encroach- ment (e.g., dumping, ATVs, etc.) Be- yond these generalities, management of preserves is an idiosyncratic proc- ess that may concern endemic spe- cies, genetics of inbred populations, or restoration of periodic wild fires. Xeric pinelands of peninsular Flor- ida depend on periodic fires to main- tain their structure and function (Laessle 1958a; Abrahamson 1984). Thus a burning program is essential in the management of LLP /TO or SPS preserves. Spring or early sum- mer prescribed bums are routinely used to maintain LLP/ TO communi- ties on state parks. Doran et al. (1987) have documented prescribed burns of SPS preserves in an urban setting based on rather esoteric fire models developed by the U.S. Forest Service. Gopher tortoises respond favorably to the bums (Stout et al. 1988). A mo- saic of recovery stages in SPS may favor beta diversity of herptiles and small mammals. Mushinsky (1985) has carefully documented the re- sponse of the herpetofauna to a vari- ety of buming schedules in LLP/TO. Diversity and abundance of amphibi- ans and reptiles was increased on experimental plots relative to un- bumed controls. Re-establishment of the pine overstory may be necessary to produce needle cast for carrying fire (Landers and Speake 1980). Management of conservation set asides and /or easements may focus on particular species or combinations of species. The smaller the preserve the more likely that a reduced suite of species will be present (Richardson et al. 1986). Given that a fixed area is available for manage- ment, major efforts to enhance or maintain habitat should target those species that can maintain viable populations within the preserve (Shaffer 1986). A species whose mini- mum area requirements for a mini- mum viable population exceeds the preserve area should not be of major concern (Shaffer and Samson 1985); nonetheless, such species can benefit from the preserves if travel corridors exist (Harris 1984). DISCUSSION Xeric pinelands of peninsular Florida support a species-rich assemblage of reptiles, amphibians, and small mammals. (Growth and development continues to diminish LLP/ TO and SPS habitats to the detriment of the associated biota. Land in public own- ership, e.g. state parks and forests, national forests, and private hold- ings, e.g., the Archbold Biological Station, and institutional lands such as the Ordway and Swisher Pre- serves, jointly owned and managed by the University of Florida and The Nature Conservancy, will be increas- ingly valuable as other xeric pine- lands are converted to land uses not favorable to the biota. Thus, manage- ment of these xeric pinelands will become more important in the fu- ture. At present management is largely limited to prescribed bums to maintain what were historically fire climax communities. Thus, fire man- agement is tantamount to small ver- tebrate management. In the future as air quality stan- dards are modified, prescribed bum- ing, particularly in or near urbanized areas, will be restricted or eliminated as a management option. Alternative means of habitat manipulation need to be developed, particularly for SPS. Basic information on the life his- tory of many amphibians, reptiles, and small mammals of xeric pine- lands is lacking. The Nongame Wild- life Program of the Rorida Game and Fresh Water Fish Commission has initiated and funded rather large scale studies of SPS and LLP/TO communities. These studies are at the community level and largely obser- vational. Management needs of indi- vidual species may be derived only secondarily from this research. Stud- ies that focus on particular species will ultimately lead to more refined habitat management guidelines. The report by Cox et al. (1987) will likely serve as a model for the preparation of habitat protection guidelines; man- agement follows protection (White and Bratton 1980). Management alternatives at the ecosystem and species level are needed now for xeric pinelands on private lands undergoing develop- ment. Regulation of development in these habitats as currently practiced will result in a patchwork of small, isolated nature preserves. Preserva- tion of natural habitat in a developed landscape is, of course, desirable. However, several problems remain: (1) who will own the preserves, (2) how will a management plan be pre- pared, and (3) who will be respon- sible for management? Even another decade of rapid growth in peninsular Florida may result in a few hundred nature preserves, which will not nec- essarily be restricted to xeric pine- land habitat. Ignoring the question of ownership, no public land manage- ment agency is currently capable of assuming the charge of managing these preserves. Lack of manage- 105 ment, e.g., failure to conduct pre- scribed burning, will allow succes- sional changes to occur to the detri- ment of many small vertebrates nar- rowly adapted to xeric pinelands. Loss of habitat and species values originally used by jurisdictional agencies to secure preserve set asides provides a potential basis for private land owners to request development rights on the land. This action would defeat the entire purpose of having conservation set asides. An alternative to on site habitat protection is offered by Cox et al. (1987) in regard to preserving habitat for the gopher tortoise. The alterna- tive, a Wildlife Resource Mitigation Fund (WRMF), allows a developer to contribute money to the fund to miti- gate losses of valuable wildlife habi- tat on lands being developed. The collective monies of several develop- ment projects would allow an inde- pendent group such as the Trust For Public Lands to assist in the purchase of commensurate lands to expand an existing public park, preserve or for- est. Management is more likely to be applied to these lands and ultimately the resources are better served by the public agencies. ACKNOWLEDGMENTS We thank the authors of the papers cited herein for their efforts and dedication to science. Biologists who contributed to our knowledge of xeric pineland include but are not limited to the following individuals: Dan Austin, Don Bethancourt, Russ Burke, Steve Christman, David Cook, David Corey, Jim Cox, Joan Diemer, Dick Franz, Larry Harris, Randy Kautz, Jim Layne, Wayne Marion, Paul Moler, Ron Myers, Reed Noss, Cathy Ryan, Jim Stevenson, and Don Wood. Support for research on the ecology of sandhill communities was provided by the Nongame Wildlife Program, RFP86-003, of the Rorida Game and Fresh Water Fish Commis- sion. Development groups that funded work by the authors on xeric pinelands include Hardy-Lieb Devel- opment Corporation, The Adler Group, and Deutsch-Ireland Proper- ties. The Division of Recreation and Parks (FDNR), Department of Biol- ogy, University of South Florida, and the Department of Biological Sci- ences, University of Central Florida assisted in our studies in a variety of ways. We thank Beverly Bamekow, Rita Greenwell, Barbara Erwin, and Nancy Small for typing the manu- script. Lastly, we thank Paul E. Moler, James N. Layne and Robert C. Szaro for providing excellent sugges- tions to improve the paper. LITERATURE CITED Abrahamson, Warren G. 1984. Post- fire recovery of Florida Lake Wales ridge vegetation. American Journal of Botany 71:9-21. Abrahamson, Warren G., Ann F. Johnson, James N. Layne, and Pa- tricia A. Peroni. 1984. Vegetation of the Archbold Biological Station, Florida: an example of the South- ern Lake Wales Ridge. Florida Sci- entist 47:209-250. Anonymous. 1976. Developments of regional impact: guidebook for preparation of the application for development approval. 65 p. Divi- sion of State Planning, Department of Administration, Tallahassee, Florida. Arata, Andrew A. 1959. Effects of burning on vegetation and rodent populations in a longleaf pine tur- key oak association in north Cen- tral Florida. Quarterly Journal Florida Academy Sciences 22:94- 104. Auffenberg, Walter, and Richard Franz. 1982. The status and distri- bution of the gopher tortoise (Go- pherus -polyphemus). p. 95-126. In North American tortoises: conser- vation and ecology. R. Bruce Bury, editor. Wildlife Research Report 12. 126 p. U.S. Fish and Wildlife Service, Washington, D.C. Beckenbach, J. R., and J. W. Ham- mett. 1962. General soil map of Florida. 1 p. Florida Agricultural Experiment Stations and the Soil Conservation Service. Brown, James H., and Astrid Kodric- Brown. 1977. Turnover rates in insular biogeography: effects of immigration on extinction. Ecol- ogy 58:445-449. Campbell, Howard W., and Steven P. Christman. 1982. The herpetologi- cal components of Horida sandhill and sand pine scrub associations, p. 163-171. In Herpetological com- munities: a symposium of the So- ciety for the Study of Amphibians and Reptiles and the Herpetolo- gists' League, August 1977. Nor- man J. Scott, Jr., editor. 239 p. U.S. Fish and Wildlife Service, Wildlife Research Report 13. Christman, Steven P., Howard I. Kochman, Howard W. Campbell, Charles R. Smith, and Wallace S. Lippincott, Jr. Successional changes in community structures: amphibians and reptiles in Horida sand pine scrub. Unpublished manuscript. Cox, James, Douglas Inkley, and Randy Kautz, 1987. Ecology and habitat protection needs of gopher tortoise (Gopherus polyphemus) populations found on lands slated for large-scale development in Horida. Horida Game and Fresh Water Fish Commission Nongame Wildlife Program Technical Report no. 4. 75 p. Tallahassee, FL. Davis, John H. 1980. General map of natural vegetation of Florida. Agric. Exp. Sta. Institute of Food and Agricultural Sciences Circular S-178. 1 p. University of Horida, Gainesville. Davis, Mary M. 1984. Phenology of flowering plants on experimental fire plots, St. Marks National Wildlife Refuge. M. S. Thesis. 54 p. Horida State University. Tallahas- see, FL. Diamond, Jared M. 1975. The island dilemma: lessons of modem bio- geographic studies for the design 106 of natural reserves. Biological Conservation 7:129-146. Diamond, J. M. 1978. Critical areas for maintaining viable populations of species, p. 27-40. In The break- down and restoration of ecosys- tems. M. W. Holdgate and M. J. Woodman, editors. Plenum Press, New York, N. Y. Diemer, Joan E. 1984. Tortoise reloca- tion in Florida: solution or prob- lem? Proc. 1984 Symposium Des- ert Tortoise Council: 131-135. Doran, R. F., Donald R. Richardson, and Richard E. Roberts. 1987. Pre- scribed burning of the sand pine scrub community: Yamato Scrub, a test case. Florida Scientist 50:184- 192. Eisenberg, J. F. 1983. The gopher tor- toise as a keystone species, p. 1-4. In The gopher tortoise: a keystone species. R. J. Bryant and R. Franz, editors. Proceedings of the 4th Annual Meeting of the Gopher Tortoise Council. Rorida State Museum, Gainesville, FL. Franklin, Ian Robert. 1980. Evolution- ary change in small populations, p. 135-149. In Conservation biol- ogy: an evolutionary-ecological perspective. Michael E. Soule and Bruce A. Wilcox, editors. 395 p. Sinauer Assoc., Sunderland, MA. Gilbert, F. S. 1980. The equilibrium theory of island biogeography: fact or fiction? Journal of Biogeo- graphy 7:209-235. Givens, Kerry T., James N. Layne, Warren G. Abrahamson, and Susan C. White-Schuler. 1984. Structural changes and succes- sional relationships of five Florida Lake Wales Ridge plant communi- ties. Bulletin Torrey Botanical Club 111:8-18. Harris, Larry D. 1984. The frag- mented forest. 211 p. The Univer- sity of Chicago Press, Chicago, IL. Higgs, A. J. 1981. Island biogeogra- phy theory and nature reserve de- sign. Journal of Biogeography 8:117-124. Humphrey, Stephen R., John F. Eis- enberg, and Richard Franz. 1985. Possibilities for restoring wildlife of a longleaf pine savanna in an abandoned citrus grove. Wildlife Society Bulletin 13:487-496. Kalisz, Paul J., and Earl L. Stone. 1984. The longleaf pine islands of the Ocala National Forest, Florida: a soil study. Ecology 65:1743-1754. Kurz, Herman. 1942. Florida dunes and scrub vegetation and geology. Florida Geological Survey Bulletin 23:15-154. Laessle, Albert M. 1942. The plant communities of the Welaka Area. University of Florida Publication, Biological Science Series 4:1-143. Laessle, Albert M. 1958a. The origin and successional relationship of sandhill vegetation and sand pine scrub. Ecological Monographs 28:361-387. Laessle, Albert M. 1958b. A report on succession studies of selected plant communities on the Univ. of Fla. conservation reserve, Welaka, Florida. Quarterly Journal Florida Academy Science 21:101-112. Laessle, A. M. 1967. Relationship of sand pine scrub to former shore lines. Quarterly Journal Florida Academy Science 30:269-286. Landers, J. Larry, and Dan W. Speake. 1980. Management needs of sandhill reptiles in Southern Georgia. Annual conference Southeastern Association Fish and Wildlife Agencies. 34:515-529. Latham, Pamela J. 1985. Structural comparisons of sand pine scrubs of east-central Florida. M. S. The- sis. 106 p. University of Central Florida, Orlando. Layne, James N. 1969. Nest-building behavior in three species of deer mice, Peromyscus. Behaviour 35:288-303. Layne, James N. 1978. Peromyscus flo- ridanus (Chapman), p. 21-22. In Rare and endangered biota of FLorida. Vol. 1: mammals. James N. Layne, editor. 52 p. University Presses of Florida, Gainesville, Fl. Margules, C, A. J. Higgs, and R. W. Rafe. 1982. Modern biogeographic theory: are there any lessons for nature reserve design? Biological Conservation 24:115-128. Moler, Paul E. 1985. Distribution of the eastern indigo snake, Dry- marchon corais couperi, in Rorida. Herpetological Review 116:37-38. Moler, Paul E., and Richard Franz. 1987. Wildlife values of small, iso- lated wetlands in the southeastern coastal plain, p. 234-241. In Proc. 3rd S. E. Nongame and Endan- gered Wildlife Symposium. R. R. Odom, K. A. Riddleberger and J. C. Ozier, editors. Georgia Depart- ment Natural Resources, Game and Fish Division. Monk, Carl D. 1968. Successional and environmental relationships of the forest vegetation of North Central Rorida. The American Midland Naturalist 79:441-457. Mushinsky, Henry R. 1984. Fire and the Rorida sandhill herpetofaunal community: with special attention to responses of Cnemidophorus sexlineatus. Herpetologica 41:333- 342. Myers, Ronald L. 1985. Fire and the dynamic relationship between Rorida sandhill and sand pine scrub vegetation. Bulletin Torrey Botanical Club 112:241-252. Norse, Elliott A., Kenneth L. Rosen- baum, David S. Wilcove, Bruce A. Wilcox, William H. Romme, David W. Johnston, and Martha L. Stout. 1986. Conserving biological diversity in our national forests. 116 p. The Wilderness Society. Noss, Reed F., and Larry D. Harris. 1986. Nodes, networks, and MUMS: preserving diversity at all scales. Environmental Manage- ment 10:299-309. Pickett, S. T. A., and John N. Th- ompson. 1978. Patch dynamics and the design of nature reserves. Biological Conservation 13:27-37. Pyle, Robert Michael. 1980. Manage- ment of nature reserves, p. 319- 327. In Conservation Biology: An evolutionary-ecological perspec- tive. Michael E. Soule and Bruce A. Wilcox, editors. 395 p. Sinauer Associates, Sunderland, Mass. 107 Richardson, Donald Robert. 1977. Vegetation of the Atlantic coastal ridge of Palm Beach County, Flor- ida. Florida Scientist 40:281-330. Richardson, Donald R., I. Jack Stout, Richard E. Roberts, Daniel F. Austin, and Taylor R. Alexander. 1986. Design and management recommendations for a sand pine scrub preserve: the Yamato scrub. 124 p. Ecological Consultants, Tampa, Florida. Shaffer, Mark L. 1986. The met- apopulation and species conserva- tion: the special case of the north- ern spotted owl. p. 86-99. In Ecol- ogy and management of the spot- ted owl in the Pacific Northwest, Ralph J. Gutierrez, and Andrew B. Carey, tech editors. USDA Forest Service. General Technical Report PNW-185, 119 p. Pacific North- west Forest and Range Experiment Station, Portland, Oregon. Shaffer, Mark L., and F. B. Samson. 1985. Population size and extinc- tion: a note on determining critical population sizes. American Natu- ralist 125:144-152. Smith, Charles R. 1982. Food re- source partitioning of fossorial Rorida reptiles, p. 173-178. In Her- petological communities: a sympo- sium of the Society for the Study of Amphibians and Reptiles and the Herpetologists' League, Au- gust 1977. Norman J. Scott, Jr., editor. 239 p. U.S. Fish and Wild- life Service, Wildlife Research Re- port 13. Soule, Michael E., and Daniel Sim- berloff. 1986. What do genetics and ecology tell us about the de- sign of nature reserves? Biological Conservation 35:19-40. Stout, I. Jack. 1981. Conservation bi- ology of Peromycus floridanus. Flor- ida Scientist 44 (Supplement 1):37. Stout, I. Jack. 1982. Small mammal community structure in sand pine scrub. Bulletin Ecological Society America 63:68-69. Stout, I. Jack, Donald R. Richardson, and Richard E. Roberts. 1988. Re- sponse of resident and relocated gopher tortoises to a prescribed bum in a sand pine scrub commu- nity. Proced. Gopher Tortoise Re- location Symp., 27 June 1987, Gainesville, FL. Stout, I. Jack, Donald R. Richardson, Richard E. Roberts, and Daniel F. Austin. 1987. Design of a nature preserve in a subtropical urbaniz- ing landscape: application of ecol- ogic and genetic principles. Bulle- tin Ecological Society of America 68:423-424. Swindell, C. E. 1987. Multivariate analysis of competition and habi- tat use by small mammals in Flor- ida Sand Pine Scrub. M. S. Thesis. 59 p. University of Central Florida, Orlando. Veno, Patricia A. 1976. Successional relationships of five Florida plant communities. Ecology 57:498-508. White, Peter S., and Susan P. Bratton. 1980. After preservation: philo- sophical and practical problems of change. Biological Conservation 18:241-255. White, W. A. 1970. The geomorphol- ogy of the Florida peninsula. Bu- reau of Geology, Tallahassee. 164 p. Geological Bulletin No. 51. Wood, Don A. 1987. Official lists of endangered and potentially en- dangered fauna and flora in Flor- ida. 19 p. Florida Game and Fresh Water Fish Commission. 108 Distribution and Habitat Associations of Herpetofauna in Arizona: Comparisons by Habitat Type^ Abstract. —Between 1977 and 1981 , the Bureau of Land Management conducted extensive surveys of Arizona's herpetofauna in 16 different habitat types on approximately 8.5 million acres of public lands. This paper describes results of one of the most exten- sive surveys ever conducted on amphibian and rep- tile communities in North America. K. Bruce Jones^ With the passage of the Federal Land PoUcy and Management Act in 1976, the Bureau of Land Management (ELM) was mandated to keep an in- ventory of resources on public lands. Information collected during inven- tories or surveys was then to be used to identify issues for land use plan- ning and opportunities for land man- agement. The BLM made a decision to collect data on all major wildlife groups and their habitats Early in the development of its in- ventory program, the BLM recog- nized a need to devise a strategy that would compare animal distributions and abundance to habitats. This strategy was important since the BLM manages wildlife habitats and not wildlife populations. In 1977 the BLM initiated invento- ries of wildlife resources on public lands. At that time, considerable in- formation was already available on game species. However, data on nongame species were mostly lack- ing. As a result, priority was given to collecting data on nongame species and their habitats. Amphibians and reptiles are im- portant members of the nongame fauna. They use a wide range of habi- ' Paper presented at symposium. Man- agement of Amphibians. Reptiles, and Small Mammals in North America. (Flag- staff , Arizona, July 19-21. 1988). ^K. Bruce Jones is a Research Ecologist with the Environmental Protection Agency. Environmental Monitoring Systems Labora- tory. Las Vegas. Nevada 89193. tats and are often good indicators of habitat conditions (Jones 1981a). Therefore, in order to obtain infor- mation on these animals, principally for land-use planning, the BLM con- ducted extensive inventories of am- phibians and reptiles by habitat type. This inventory included a scheme whereby associations between am- phibians and reptiles and certain n\i- crohabitats could be determined. The inventory, conducted between 1977 and 1981, was one of the most com- prehensive surveys of herpetological communities ever conducted in North America (27,885 array-nights in 16 habitat types over a five-year period). It also represents the first large-scale effort to quantitatively compare herpetofaunas associated with ecosystems. This paper reports the results of these surveys, includ- ing species distributions and associa- tions with microhabitats and habitat types (plant communities). STUDY AREA The study area consisted of approxi- mately 3,441,296 ha (8.5 million acres) of public lands located in cen- tral, west-central, southwestern, and northwestern Arizona (fig. 1). Sixteen different habitat types were deline- ated within this area, primarily from an existing map of vegetation asso- ciations (Brown et al. 1979). Field re- connaissance allowed more local as- sociations to be recognized within Figure 1.— The study area. those presented by Brown et al. (1979). For example, because of the scale of their map. Brown et al. (1979) failed to recognize several small, rel- ict stands of chaparral woodland, although Brown (1978) had noted the presence of chaparral woodland vegetation at several small sites (see Jones et al. 1985 for the importance of small woodland stands to certain herpetofauna). Therefore, the habitat type map used to allocate samples in this study drew upon the Brown (1978) and Brown et al. (1979) maps. 109 and results of field reconnaissance. For detailed descriptions of these habitat types see Jones (1981b) and Buse (1981). SAMPLING METHODS Amphibian and reptile distribution and abundance by habitat type were determined by on-the-ground sam- pling efforts between October, 1977, and July, 1981. Samples were ob- tained by three methods. The most extensive sampling was accom- plished with a pit-fall trapping method (array) consisting of a series of 18.3 1 (5 gal) plastic containers bur- ied in the ground and connected by 0.41 m (8 inches) high aluminum drift fence; one trap was located in the center with three evenly dis- persed (120°) peripheral traps 7.14 m (25 ft) from the center (Jones 1981a, Jones 1986). This modified array method was designed specifically for sampling amphibians and reptiles in desert habitats (see Jones 1986 for a comparison of this procedure with the original array trapping scheme designed by Christman and Campbell 1982). A total of 183 arrays were used to sample 16 different habitat types (see table 1 for sum- Table 1 .—Sampling effort in each habitat type. #of arrays # of trap nights # of road riding road transects # of field searches Elevation range (m (ff.)) Ponderosa Pine Woodland (PP) 5 745 10 15 1677-2531 (5500-8300) Pinyon-Juniper Woodland (PJ) 9 945 14 20 1311-1921 (4300-6300) Sagebrush (Great Basin Desert) (SB) 3 270 12 12 1311-1830 (4300-6000) Closed Chaparral (CC) 18 2168 18 20 1250-2287 (4100-7500) Open Chaparral (OC) 13 1950 22 25 762-1311 (2500-4300) Desert Grassland (DO) 11 1155 15 14 1006-1525 (3300-5000) Disclimax Desert Grassland (DD) 3 300 11 10 884-1311 (2900-4300) Mixed Broadleaf Riparian (MB) 6 784 8 18 884-2287 (2900-7500) Cottonwood-Wiilow Riparian (CW) 13 3145 23 28 549-1372 (1800-4500) Juniper Woodland (mixed shrub) (JM) 9 1080 19 22 793-1342 (2600-4400) Canotia Mixed Shrub (CA) 3 265 11 16 884-1189 (2900-3900) Mesquite Bosque (floodplain woodland) (ME) 15 3025 18 22 213-915 (700-3000) Mixed Riparian Scrub (Xeroriparian) (MR) 16 2640 23 18 229-1220 (750-4000) Mojave Desertscrub (MD) 15 1803 25 24 610-1220 (2000-4000) Sonoran Desertscrub (Arizona Upland) (SD) 22 3970 33 27 335-1189 (1100-3900) Creosotebush (Lower Colorado) (CB) 22 3640 32 18 213-915 (700-3000) mary of sampling effort in each habi- tat typ>e). Arrays were placed so that microhabitat variability within each habitat type was sampled. The num- ber of arrays used to sample habitat types was partially influenced by the size of habitats; generally, more ex- tensive habitats received prop)ortion- ally larger samples. However, certain habitats (e.g., riparian) were known to be great sources of diversity within desert regions; therefore, pri- ority was given to obtaining larger samples within these habitats. Once placed into the ground, arrays were continuously open for a minimum of 60 days. Some arrays (60) were open for 9 months. Generally, samples were taken during the spring, sum- mer, and fall. However, some arrays (17) were open only during spring months and others only in the fall (12). The opening of new arrays at different locations, and the closing of other arrays, were often dictated by BLM's predetermined resource plan- ning schedule. Since some amphibians and many snakes could not be effectively sampled by pit-fall traps, it was nec- essary to use two other field tech- niques. Road riding, consisting of traveling roads from dusk to ap- proximately 2300 h throughout de- lineated habitat types, was used to determine the occurrence of amphibi- ans and medium and large snakes (see table 1 for sampling effort within each habitat type). Time-constraint searches (Bury and Raphael 1983), consisting of walking along permanent and tem- porary water sources (natural and man-made) at night, were used to verify the presence of frogs and toads at waters within habitat types (see table 1 for sampling effort within each habitat type). Finally, to get an idea of the known distribution of amphibians and reptiles within the study area, I obtained records from 7 museums known for their outstanding collec- tions of amphibians and reptiles from the Southwest: the University of 110 Michigan, Arizona State University, the University of New Mexico, Northern Arizona University, the University of Arizona, the Los Ange- les County Museum, and the Univer- sity of California at Berkeley. In addi- tion, these data were used to com- pare the past distribution of amphibi- ans and reptiles within the study area with that obtained during the BLM's inventories. Microhabitat data were collected on each array site and along roads by a modified point-intercept method consisting of 100 sample points sepa- rated by 8 m (26 ft) along a randomly determined compass line; on array sites, the center of the line crossed over the array. At each point, the fol- lowing measurements were taken: (1) vertical distribution of vegetation be- tween 0-0.6 m (0-2 ft), 0.6-1.7 m (2-6 ft), 1.7-6.0 m (6-20 ft), and > 6 m (20 ft) (each time vegetation occurred in a height class above the point, a con- tact or "hit" was recorded); (2) pene- tration to the nearest cm into the soil by a pointed metal rod (1 cm in di- ameter); (3) depth of leaf litter (if present); (4) depth of other litter such as debris heaps (piles of logs, leaves and other dead vegetative material) and rotting logs; (5) characterization of surface rock into size classes of sand, gravel (< 1 cm or 0.4 inches in diameter), cobble (1 to 5 cm or 0.4 to 2 inches in diameter), stone (> 5 cm or 2 inches in diameter), and bed- rock. Vegetation cover and percent- age of the surface occupied by each rock and litter size class was deter- mined by comparing the number of "hits" in each category (e.g., litter) with the total number of sample points (100). Plant species were also recorded along each 100 point transect (see table 1 for the number of microhabitat samples taken in each habitat type). DATA ANALYSIS I calculated relative abundance of each amphibian and reptile species as the total number of any species caught during a 24-hour period (ar- ray-night). Relative abundance was determined for each species on array sites by taking the greatest number of individuals of a species trapped dur- ing a 30-day period and dividing by the number of days. This calculation was used because of monthly differ- ences in species' activity patterns. The number of arrays in which a spe- cies was trapped in each habitat type also was compiled to determine how widespread a species was within in- dividual habitat types. A principal components analysis (Pimental 1979) was performed to compress microhabitat data into a smaller, depictable subset. Mean fac- tor scores of compressed microhabi- tat data were computed for each habitat type and plotted on a 3 vector (axis) graph. Similarly, mean factor scores of compressed microhabitat data were computed for each am- phibian and reptile species (turtles were excluded because aquatic mi- crohabitats were not measured). These scores were calculated for each species by averaging mean factor scores for microhabitats on which a species occurred. Species richness (total number of species) and species diversity were calculated for each habitat type. Two calculations of species richness for habitats were used; one that used only array data and one that used all data (array, road-riding, and field- search data). In addition, the average number of species collected per array (30-day period) was calculated and compared to overall, array-deter- mined, species richness. Species di- versity of each habitat was deter- mined from a Shannon-Weaver di- versity index (Hair 1980): H' = E p, logjQ Pi/ where s = the number of spe- cies and Pj is the proportion of the total number of individuals consist- ing of the i**^ species. Average species diversity per array was calculated for each habitat type. Because road-rid- ing and field searches did not yield estimates of relative abundance simi- lar to arrays, only array data were used to calculate species diversity. Two types of cluster analysis were used to determine similarities among habitat types. The first cluster analy- sis was performed only on array data, and it was based on euclidean distances (Pimental 1979). Calcula- tion of euclidean distances between habitats were based on a combina- tion of species' presence or absence on a site and similarity in species' dominance (relative abundance) be- tween habitats. Since medium and large snakes (> 0.5 m or 1.5 ft) are not readily caught in pit-fall traps, their relative abundances could not be cal- culated accurately. To compare the overall herpetofaunas of habitat types, a second cluster analysis was performed. This procedure involved calculation of Simpson similarity co- efficients (Pimental 1979). These coef- ficients were then submitted to a cluster analysis. Unlike the analysis of array data via euclidean distances, the use of Simpson similarity coeffi- cients in a cluster analysis did not consider relative dominance in calcu- lating distances between habitats. Several thousand site specific dis- tributional records were obtained for amphibians and reptiles within the study (to 16.2 ha or 40 acre accu- racy). These individual records were too numerous to report here; detailed locality records for each species are kept at the Bureau of Land Manage- ment's Phoenix District Office. RESULTS Microhabitats A principal components analysis (PCA) of microhabitats yielded 3 compressed habitat components (axes), and the cumulative propor- tion of eigenvalues was < 1 .0 with 83% of the variability accounted for by the matrix (p < .05). This analysis revealed large differences in the mi- crohabitat among habitat types (fig. 2). Desert grassland, disclimax desert 111 grassland, and creosotebush habitats had open canopies and low-height vegetative structure, whereas pinyon-juniper, mixed riparian scrub, cottonwood-willow riparian, mixed broadleaf riparian, and pon- derosa pine had tree canopies and large amounts of vegetative debris, such as leaf litter and logs, on their surfaces (fig. 2). Closed and open chaparral habitats consisted of shrubs with rocky surfaces, and Sonoran Desert had a combination of trees and shrubs and rocky surfaces (fig. 2). Species Distributions and Abundances A total of 28 species of lizards, 30 snakes, 4 turtles, 9 toads, 3 frogs, and 1 salamander were observed or trapped during the study. Sceloporus magister, Urosaurus ormtus, Uta stansburiam, and Cnemidophorus tigris were the most widely distributed and abundant lizards throughout the study area's habitat types (table 2). These lizards also consistently oc- curred on a large number of sites within each habitat type (table 2). Certain lizards, such as Gambelia wis- lizeni, Phrynosoma solare, and Dip- sosaurus dorsalis occurred only on lower elevation (< 915 m or 3000 ft), desert habitats, and other lizards, such as Sceloporus undulatus, Gerrhon- otus kingi, and Phrynosoma douglassi occurred only on higher elevation (> 1220 m or 4000 ft) habitats (table 2). Some species, such as Eumeces gilberti and Cophosaurus texam, were princi- pally found on higher elevation habi- tats, but also inhabited cottonwood- willow riparian habitats at lower ele- vations (549-915 m or 1800-3000 ft) (table 2). Certain lizards, such as 1.5 Trees Component II Figure 2.— Mean factor scores of microhabitats for habitat types. (Abbreviations correspond to thiose listed for hiabitats in table 1 .) Cnemidophorus burti and Eumeces ob- soletus, had limited distributions within the study area (table 2); C. burti is principally distributed in the Sonoran Desert and Desert Grass- land habitats in extreme southern Arizona and Mexico, and E. obsoletus only occurs in the chaparral habitat type in the extreme eastern portion of the study area. Although re- stricted to higher elevation and ripar- ian habitats throughout most of the study area, C. texana was found in Sonoran Desert in the extreme east- ern portion of the study area. Most lizards occurred throughout the study area where suitable habitat was present and were not restricted by geographic range. A PCA revealed that lizards dif- fered in their associations with cer- tain microhabitats (fig. 3). Some of the widely distributed species, such as Cnemidophorus tigris and Uta stansburiana, showed little association with any of the principal components (fig. 3), although the distribution of other common species, such as Sce- loporus magister and Urosaurus ornatus was highly correlated with the pres- ence of vegetation debris (fig. 3). More than half of the lizards oc- curred on sites with relatively open canopies and shrubs or grasses, and many also preferred rocky substrates (fig. 3). Dipsosaurus dorsalis, Callisau- rus draconoides, and Gambelia wislizeni occurred on sites with sand substrate. Gerrhonotus kingi and Eumeces gilberti occurred on sites with large amounts of vegetative de- bris, medium to high canopies, and rocky substrates, and Xantusia vigilis on sites with similar substrate but with a more open canopy (fig. 3). Crotaphytus collaris and Sauromalus obesus occurred on sites that were open, rocky, and shrubby or grassy (fig. 3). Snakes showed similar distribu- tional patterns to lizards. Some snakes, such as Lampropeltis getulus, Pituophis melanoleucus, Rhinocheilus leconti, Crotalus atrox, and Crotalus molossus, occurred in many habitat 112 Table 2.— Relative abundance of lizards by habitat type. Relative abundance the number of an individual species caught in an array per 24 h period. * Indicates species verified in a habitat type via road-riding and searches. The number below the Habitat Type in ( > = the total number of arrays. The number In ( ) to the right of the species' relative abundance = the number of arrays in which the species was trapped. PP PJ SB CC OC DG DD MB cw JM CA ME MR MD SD CB (5) (9) (3) (18) (13) (11) (3) (6) (13) (9) (3) (15) (16) (15) (22) (22) Gerrhonotus kingi - - - .03(1) .03(1) Coleonyx vahegafus 03(1) - - .03(1) .03(1) .01(1) .05(2) .01(5) .03(8) .02(6) .04(11) .06(11) Heloderma suspectum ♦ » .03(1) ♦ ♦ .03(1) .03(1) .03(1) .03(2) Callisaurus draconoides - - - .06(1) - - - - .10(3) .01(6) 05(2) .03(2) .05(7) .08(4) .06(10) .04(6) Cophosaurus fexona .07(1).09(5) - .10(5) .03(1) - - .08(2) .10(4) - - .01(1) .03(2) - .02(2) - Crofaphyfus collaris - * * .03(2) • .10(5) .03(2) .03(1) .04(2) - Dipsosaurus dorsalis Gambelia wislizenii ------- .01(1) - .03(1) .08(9) - .07(1) - ♦ _ _ .01(1) .03(2) .01(2) .02(3) .02(3) Holbrookia maculafa - - - .08(1) - ,03(1) Phrynosoma douglassi .06(3).04(3) .13(1) .04(6) _ _ _ Rirynosoma plafyrhinos Phrynosoma sol are Sauromalus obesus - ' - .07(1) - - .01(1) .11(3) - .03(1) .02(3) .02(3) .05(7) ---- ---- - .03(1) - .03(2) .02(1) - - - - .03(1) _ _ _ .06(1) .01(7) .02(1) - .03(1) * .02(1) - Sceloporus clarki - .03(1) _ _ _ _ . .03(2) .03(2) - - - .03(1) .03(1) .03(1) - Sceloporus magister - .05(5) - .05(7) .03(3) .03(2) - .11(4) .23(7) .03(8) .19(3) .13(10) .1 1(16) .10(15) .07(14) .03(6) Sceloporus undulafus .13(3). 13(4) .17(3) .07(13) - .10(3) - .02(1) .04(2) - - - - - - Urosaurus graciosus ---------- - .07(7) .07(11) .01(3) .04(2) .07(13) Urosaurus omafus .03(1).04(4) - .04(6) .03(7) .05(3) - .15(4) .20(5) .03(1) - .08(5) .04(5) .03(3) .06(7) ,02(3) Ufa sfansburiana - .03(2) - .04(4) ,04(7) .05(1) .10(1) - .11(7) .05(8) .05(2) .08(5) .1 1(13) .05(12) .13(17) .09(15) Eumeces gilberfi .03(1).06(3) - .05(9) .11(10) .03(1) - .02(2) .04(4) .03(1) ______ Eumeces obsolefus Cnemidophorus burfi Cnemidophorus flagellicaudus - .05(3) - .04(5) - .07(2) - .08(1) .02(1) Cnemidophorus inornatus - .03(2) - * - .03(1) (continued) 113 c Table2.— (continued). PP PJ SB CC OC DG (5) (9) (3) (18) (13) (11) DD (3) MB CW JM CA (6) (13) (9) (3) ME (15) MR (16) MD (15) SD (22) CB (22) Cnemidophorus figris nom 10 f/^^ — 07(6) 05(3) 09(4) 23(2) 10(3) 07(7) .14(9) .25(3) . 14(9) .25(16) .13(15) .17(21) .15(21) {^noniioopnoius ui ]ii~juit3i lo _ _ _ .04(1) - .03(1) — — — — — — Cnemidophorus velox - * .49(3) .14(5) - .01(1) .05(2) .02(2) - - — — — — Xanfusia vigilis - .01(1) - .07(1) - .02(1) - .05(1) .08(7) - — Total Number of Species (includes species verified by road-riding and searclnes) 7 14 4 20 12 14 4 10 16 11 9 10 16 14 17 12 Mean Relative Abundance .37 .69 .79 .96 .43 .74 .47 .67 1.06 .54 .72 ,59 .91 .58 .78 .63 Species Diversity (H') .56 1.00 .40 1.18 .89 1.07 .54 .91 1.00 .76 .72 .86 1.05 1.00 1.09 .95 J 15 .75 Component I -.75 Sknitn/ Grasses Component III 1 - EwMcd obsoWM 2 - Ctleonyi foritfohis ] - Mcdemn ml^mian 4 - CdtMM *ocanoid«l 5 - Cophosounn kenun 6 - OuttphyfaiS cdbnt 7 - D^jMnwva towh I - GonMio (WiOii $ - Hotmcldo mooioto 10 - Rnynosoni dougkns I I - Rfjbosoto plotjrtinos 12 - Rf)iwswno vim 13 - Sann«lut cbsw U - Scelopann darti 15 - Scikpana najim IS - Scilcpmf uiduMif 17 - Umwvn fTOCWSus 18 - UoNwvs omotn 19 - Uto ilmlMioM 20 - EiiffltcM ffbvti 21 - DinMcts obsolekis 22 - OMnUai^mi barb' 23 - Ca«nidg|]hona 9aj/kKin 24 - Omidofiham •wmtba 25 - CfwrMoftow ttf* 26 - Cneiitfoftow u^mn 27 - Cn«n4«|iharui lim 21 - bnhso liglil Opsn -.75 -.50 -J5 J5 JO Component II Figure 3.— Mean factor scores of microhabitats for lizards. types. Others, such as Chilomeniscus cinctus, Chiomctis occipitalis, Phyl- lorhynchus browni, Phyllorhynchus de- curtatus, and Crotalus cerastes, oc- curred primarily on lower elevation (< 915 m or 3000 ft), desert habitats, and some, such as Lampropeltis py- romelam and Crotalus viridis cerberus, occurred only on higher elevation (>1525 m or 5000 ft) habitats (table 3). Lichanura trivirgata and P. browni oc- cur primarily outside the study ar- eas, and their distributions only overlap the extreme southern and southwestern portions of the study area. Therefore, they were limited to the small number of sites with suit- able habitat. Thamnophis cyrtopsis and Thamnophis marcianus were restricted to sites with water, with the former occurring on a large number of habi- tats and the latter only in a mesquite bosque habitat along the Gila River south of Phoenix. Similar to Copho- saurus texana, Tantilla hobartsmithii was found on higher elevation (>1220 m or 4000 ft) and riparian habitats throughout most of the study area, but also in Sonoran Des- ert in the eastern portion of the study area. A PC A of microhabitats on which snakes occurred revealed that, simi- 114 Table 3.— Relative abundance of snakes by habitat type. Relative abundance = the number of an Individual species caught In an array per 24 h period.* Indicates species verified In a habitat type via road-riding and searches. The number below the Habitat Type in ( ) = the total number of arrays. The number in () to the right of the species' relative abundance = the number of arrays in which the species was trapped. Arizona elegans * ♦ Chilomeniscus cine f us Chionacfis occipitalis Diadophis puncfafus Hypsiglena for quota Lampropeltis getulus « » ♦ Lampropeltis pyromelana ♦ Lichanura trivirgata - - - - • • - * .02(1) * .03(2) - - - - .05(3) - - .07(7) .08(7) - .02(2) .03(1) ------- .03(3) .12(2) - .05(4) .06(5) .02(2) .02(1) - - - .03(2) * .03(2) .02(1) .02(1) * ,02(1) * - Masficophis bilineatus - - - - .03(2) .03(1) - • * * _ _ _ _ _ _ Masticophis flagellum - ,03(1) - * * - - - .02(1) * .02(1) .02(1) .02(1) ,02(1) ,02(1) .03(2) Masficophis taeniatus ,03(2) .03(2) - ,02(1) * _ _ _ _ _ _ Pituophis melanoleucus .02(1) * .02(1) * .02(1) * * • .01(1) Pliyllorliynclius browni Phyilorhynclius decurfafus Rhinoclieilus leconfi - — - - - - ,02(2) - * .03(2) - .02(1) .02(1) * * * * * .02(1) .02(1) Salvadora liexalepis - - - .02(1) .02(1) * - - .03(2) • .03(1) .03(1) .02(1) .04(2) .02(1) Sonora semiannulafa - - - * * .10(2) - - * .03(1) * ♦ _ * .05(3) Tanfilla hobarf smith ii - - - .05(5) .08(8) ,07(2) - .02(1) .05(4) .03(1) - .03(2) - - Thamnophis cyrtopsis *->»»• »•♦♦♦♦»»♦ Thamnophis marcianus « Trimorphodon biscutafus lambda _ » Crofalus afrox Crofalus cerastes Crofalus mifchelli .02(1) .02(2) (continued) 115 > Table 3.— (continued). PP PJ SB CC OC DG DD MB CW JM CA ME MR MD SD CB (5) (9) (3) (18) (13) (11) (3) (6) (13) (9) (3) (15) (16) (15) (22) (22) Crofalus molossus * « • • • • ♦ ♦ Crofalus scututatus .02(1) « » ■■ ♦ Crofalus figris Crofalus viridis cerberus • * « Micruroides euryxanfhus Leptotyphlops humilis - .04(3) - - .05(2) .05(3) .05(3) .03(2) - - * - .09(6) .03(2) .02(1) .03(2) .06(6) Total Number of Species (includes species verified by road-riding and searches) 2 4 11 6 17 22 12 Mean Relative Abundance - .07 - .07 .18 .23 Species Diversity (H') - .30 - .26 .63 .54 12 20 18 13 17 18 16 25 16 .18 .24. 16 .07 .36 .28 .12 .22. .29 .75 .81 .68 .47 .93 .68 ,68 .86 .90 15 TrMt 75 _ Component I Crane* 29 - Arizona ejegans . 4-6 • 30 - Chilomeniscus cinctus 47 • 31 - Chionactis occipitalis 4-8 • 32 - Diodophis punclatus 4-9 • 33 - Hypsiglena torquata 50 ■ 34 - Lompropeltis qetuius 51 ■ 35 - Lompropeltis pyromelaiM • 36 - Lichanura trivirqata 53 ■ 37 - Mosticophis bilineatus 54 • 38 - Mosticophis flagellum 55 • 39 - Masticophis taeniatus 56 • 40 - Pituophis melanoleucu57 ■ 41 - Phyllorhynchus browni 58 • 42 - Phyllorhynchus decurtotus 43 - Rhinocheilus leconti 44 - Salvadoro hexolepis 45 - Sonora semiannulqja Tontilla hobortsmithii Thomnophis cyrtopsis Thomnobhis mcrcianus Trimorpnodon biscutctus lambdc Crotolus otrox Crotolus cerastes Crotalus mitchelli Crotolus molossus Crotalus scutulatus Crotalus tigris Crotolu? viridis cerberus Micruroides euryxonthus Leptotyphlops humilis " 51 54 Component 50 A2 M 5:i 51) 57 49 58 48 46 56 Open Canopy Component Figure 4.— Mean factor scores of microhabitats for snakes. lar to those of lizards, microhabitat associations differed among snakes (fig. 4). Many of the widely distrib- uted snakes, such as Hypsiglena torquata, Lampropeltis getulus, Mastico- phis flagellum, and Pituophis melano- leucus, showed no strong relationship with any of the compressed habitat components (fig. 4). Conversely, most species with limited distribu- tions showed a strong relationship with certain components (fig. 4). Chionactis occipitalis, Crotalus cerastes, Crotalus scutulatus, and Phyllorhyn- chus browni consistently occurred on open, sandy sites, and Chilomeniscus cinctus occurred on sites with sandy substrate but taller canopy (fig. 4). Other species, such as Crotalus mitch- elli and Sonora semiannulata, were found on sites with open canopies but rocky substrates (fig. 4). Thamno- phis marcianus and Tantilla hobartsmithii occurred on sites with sandy substrates but closed canopies and large amounts of vegetative de- bris, and Lampropeltis pyromelana oc- curred only on sites with high amounts of vegetative debris (fig. 4). Other species, such as Diadophis 116 pundatus, Thamnophis a/rtopsis, and Crotalus viridis cerherus, occurred on rocky sites with high amounts of vegetative debris (fig. 4). Except for a single Gopherus agas- sizii captured in an array, all turtle records came from road-riding and field searches. Four species of turtles were recorded within the study area, three aquatic and one terrestrial (table 4). Of these, G. agassizii was the most widely distributed (verified in 9 habitat types, table 4). A more thor- ough account of this turtle's distribu- tion is described by Burge (1979, 1980). Pseudemys scripta, an intro- duced species, was limited to a stretch of the Gila River from the 99th Street bridge in southwest Phoe- nix to Gillespie Dam, located ap- proximately 24 km (15 miles) south of Buckeye. Trionyx spiniferus oc- curred at Alamo Lake (confluence of the Big Sandy and Santa Maria rivers in western Arizona) and along peren- nial stretches of the Gila River from Phoenix to Yuma. Kinosternon sonori- ense occurred on several permanent streams and rivers throughout the study area. In contrast to the observed distri- bution patterns among lizards and snakes, the distribution of amphibi- ans did not shown an elevational pat- tern. Although certain species such as Bufo punctatus and Scaphiopus couchi occurred in a large number of habitat types, most species were found in at least one lower (< 915 m or 3000 ft) and one higher (> 1220 m or 4000 ft) elevation site (table 5). Similar to lizards and snakes, there are some amphibians whose ranges are principally outside the study area and are, therefore, found only on a few sites (table 5). The ranges of Bufo debilis, Bufo retiformes, and Gastro- phyrne olivacea are primarily in north- ern Mexico, or east and south of the study area in the Chihuahuan Desert; within the study areas, their ranges are limited to desert grassland habi- tats in the extreme southern portion (Vekol Valley, 48 km or 30 mi west- southwest of Casa Grande). All populations of Ambystoma tigrinum were located at earthen stock tanks (dirt tanks). Presumably, all of these populations were introduced. A PCA demonstrated correlations between occurrence of amphibian species and particular microhabitats (fig. 5). Bufo debilis, B. retiformes, and Gastrophyrne olivacea occurred on sandy, grassy sites, and Bufo cognatus on sandy, shrubby sites (dg. 5). Bufo microscaphus and B. punctatus oc- curred on rocky sites, and Hyla areni- color on rocky sites generally occu- pied by trees and large amounts of vegetation debris (fig. 5). Certain species, such as Scaphiopus couchi, Bufo alvarius, and Bufo woodhousei oc- curred on sites with a wide variety of substrates (fig. 5). The occurrence and frequency of water was not quantitatively meas- ured at each site; therefore, the influ- ence of water was not considered in the development of figure 5. How- ever, all sites with amphibians had surface water during some part of the year, especially during summer months. AH sites with Bufo mi- croscaphus, Rana pipiens, R. catesbe- iana, and Hyla arenicolor had perma- nent water (e.g., springs, creeks, and rivers). At the start of the survey in 1977, populations of Bufo microscaphus and B. woodhousei sympatric on major drainages, such as the Hassayampa, Santa Maria, Agua Fria, and New rivers, could be easily distinguished from one another. By 1981, popula- tions on all of these drainages were indistinguishable. Range Extensions Thirty-five range extensions were recorded for amphibians and reptiles within the study area. Except for the following discussion, range exten- Table 4.— Distribution of turtles by habitat type C). Records are entirely from road-riding and searches (except where otherwise Indicated. All turtles except Gopherus agassizii occurred only at sites with permanent water within habitat types listed below. PP PJ SB CC OC DG DD MB CW JM CA ME MR MD SD CB Gopherus agassizii « Pseudemys scripta Trionyx spiniferus Kinosternon sonoriense " Trapped in an array Number of species 117 — - .. , ,, Table 5 -Relative abundance of amphibians by habitat type.Relative abundance = the number of an individual spe- cies caught in an array per 24 h period. ' indicates species verified in a habitat type via road-riding and searches. The number below the Habitat Type in ( ) = the total number of arrays. The number in ( ) to the right of the species relative abundance = the number of arrays in which the species was trapped. PP (5) PJ (9) SB (3) CC (18) OC (13) DG (11) DD (3) (6) CW (13) JM (9) CA (3) ME (15) MR (16) MD (15) SD (22) CB (22) Bufo olvorius - .07(1) Bufo cognatus * — — Bufo debilis Bufo mlcroscaphus' - .05(2) -• Bufo puncfafus .03(1).03(1) * .15(8) .11(8) Bufo refiformis Bufo woodhouseP Hyla orenicolor - .07(1) .03(1) Gastrophyrne olivacea .06(2) .14(3) .03(1) .03(2) .03(4) .09(3) - .06(4) 18(2) .13(3) .06(3) V - -07(4) _ - - - .20(1) .16(3) .12(2) .23(1) .28(6) .05(2) .06(2) .10(7) - _ • _ - - .03(1) - Scaphiopus couchi Rana pipiens • ♦ ♦ Rona cafesbeiana Ambysfoma figrinum .12(2) .05(2) .10(3) • .20(1) .15(6) .06(2) .07(6) .06(5) .11(6) .03(1) '95% of these were a cross between the two species (B, microscaphus x B. woodhousei) Total Number of Species (includes species verified by road-riding and searches) 4 4 3 6 6 8 2 4 6 5 5 Mean Relative Abundance .03 .03 - .34 .26 .56 - .33 .45 .12 .23 Species Diversity (H') _ - - .56 .42 .71 - .29 .51 - 8 6 5 6 3 .65 .14 .13 .16 .17 .65 .46 .30 .29 .28 sions discovered during this study have been described elsewhere (Jones et al. 1981, Jones et al. 1982, Buse 1983, Jones et al. 1983, Jones et al. 1985). The southernmost distribution of Tantilla hobartsmithii was extended from the Salt River east of Phoenix, southwest in the mesquite bosque habitat along the Gila River to 56 km (35 miles) east-northeast of Yuma (fig. 6). A population of T. hobart- smithii was also discovered in a 10 ha (25 acres) open chaparral habitat in the Eagletail Mountains (fig. 6). The westernmost distribution of Cnemido- phorus burti was extended from the Tucson area northwest by discovery of isolated populations in desert grassland habitats on summits of the Tabletop and Estrella mountains (fig. 6). An isolated population of Mastico- phis bilineatus lineolatus was discov- ered on the summit of Tabletop Mountain in a relict desert grassland habitat (fig. 6). This population ex- tends the known distribution of this subspecies approximately 100 km (62 mi) to the north of the only other known population (Ajo Mountains). Finally, an isolated population of Diadophis punctatus was discovered in a relict desert grassland commu- nity on the summit of the Estrella Mountains southwest of Phoenix (fig. 6). 118 Comparison of Habitat Types Based on data compiled from pit-fall trapping, road-riding, and searches, the Sonoran Desert habitat had the greatest species richness (49 species, fig. 7). Closed chaparral and cotton- wood-willow riparian habitats were the second richest habitats (44 spe- cies), and open chaparral and mixed riparian scrub were third (41 species, fig. 7). Disclimax desert grassland had the fewest species (8), and sagebrush and ponderosa pine had the second and third fewest species (13 and 15 species, respectively, fig. 7). All other habitats had at least 27 species but not more than 39 (fig. 7). Although Sonoran Desert had the richest lizard and snake faunas, mesquite bosque and desert grassland habitats had the richest amphibian fauna (fig. 7). The mesquite bosque habitat type had the greatest number of turtle species (four species, fig. 7). When only array data are com- piled, disclimax desert grassland, sagebrush, and ponderosa pine habi- tats still had by far the lowest num- ber of species, but Sonoran Desert and mesquite bosque had the great- est number of species (fig. 8). As when all data were taken into ac- count, mixed riparian scrub, cotton- wood-willow riparian, closed chap- arral, and open chaparral had high species richness (fig. 8). However, desert grassland was relatively more diverse using only array data (fig. 8). The difference between array vs. all data appears to result from the inability of arrays to consistently ver- ify (trap) turtles and medium and large snakes, although many larger snake species were verified because young-of-the-year were easily trapped. A more revealing statistic is the average number of species verified by an array (fig. 8). This analysis re- veals which habitats consistently had the largest number of species at sample sites. Certain habitats, such as desert grassland, although high in overall species richness, had rela- tively few species verified at each array site (fig. 8). Other habitats, such as ponderosa pine, sagebrush, and disclimax desert grassland, had the lowest number of total species and the lowest average number of species per array site (fig. 8). Many of the habitats that had high overall species richness also had high overall richness at each array site; however, cotton wood-willow had a higher av- erage number of species per array site than did Sonoran Desert (fig. 8). Species diversity indices (H') cal- culated from array data reveal pat- terns similar to those described above (fig. 9). Disclimax desert grass- land, sagebrush, and ponderosa pine continue to exhibit low diversity, and Sonoran Desert, closed chaparral, cotton wood-willow riparian, mixed riparian scrub, and desert grassland continue to be diverse (fig. 9). How- ever, as in the previous analysis, the average diversity per array site is low when compared to total diver- sity for individual habitats (fig. 9). Of the habitats with high overall diver- sity, mixed broadleaf riparian and cottonwood-willow riparian had relatively high average diversity per array site (fig. 9). A comparison of herpetofaunas of each habitat type by cluster analyses revealed that all desert habitats, such as creosotebush, Sonoran Desert, Mohave Desert, and mixed riparian scrub had very similar herpetofaunas (figs. 10 and 11). In both cluster analyses, open and closed chaparral had similar herpetofaunas, and sage- brush and disclimax desert grassland had a herpetofauna different from any other habitat. However, there were differences in results of the two cluster analyses for other habitats. Whereas the cluster analysis of array 1.5 Tro« 75 Component I -.75 Grauei Component 59 - aufo olvoriui 60 - Bufo cognotm 61 - Biifo deUD] 62 - SjTo mlcroicaphui 63 - auto punctatui - Buto rotltormS 66 - Buf 0 woodhomel 66 - Hyla aronlcolof 67 - Gaitrophyme ollvocea 68 - ScopmopLa coucni 69- nana plplem 70- nono coteiDelana 71 - AmbyitDtno ttgrtnum iJ 71 Open Canopy 0 .25 60 VeQerotlva Component I Figure 5.— Mean factor scores of microhabitats for amphibians. 119 data revealed large differences be- tween the herpetofaunas of cotton- wood-willow and desert habitats, such as Sonoran and Mohave Des- erts, these habitats had a relatively moderate degree of overlap when all data were analyzed (figs. 10 and 11). Additionally, ponderosa pine and pinyon-juniper habitats were similar when array data were analyzed and relatively dissimilar when all data were submitted to cluster analysis (figs. 10 and 11). DISCUSSION Overall, western Arizona has an ex- tremely diverse herpetofauna, pri- marily because of its large variety of habitats zoogeographic location. The Hualapai Mountains, located in northwestern Arizona, are adjacent to three major deserts: the Mohave Desert to the northwest, the Great Basin Desert to the northeast, and the Sonoran Desert to the south. No- where else on the North American continent does such a phenomenon exist. The diversity of habitat in this area is also enhanced by the occur- rence of several woodland islands. Number of Species 50 40 30 _ 20 - 10 _ Tuttes Amphibians Snakes Uzofds PP PJ SB CC OC DG DO MB CW JM CA rvE fv« MD SO CB Habitat Type Figure 7.— Number of species by taxonomic group by habitat type. (Abbrev. correspond to ttiose listed for liabitats In table 1 .) Number of Species 30 25 — 20 15 - 10 — 6 — #Of Species Ave # of species/ array PP PJ SB CC OC DG DD MB CW JM CA ME MR MD SD CB Figure 6.— Map of range extensions. Habitat Type Figure 8.— Total number of species caughit in arrays by habitat type vs. the average number of species caught per array by habitat type. (Abbrev. correspond to those listed for habitats In table 1 .) 120 Species Diversity (H') Patterns of Species Distributions Habitat Type Figure 9.— Total species diversity (H') by hiabitat type vs. average species per array by tiabl- tat type. (Abbrev. correspond to thiose listed for htabitats in table 1 .) Similarity MR SD MD CA OC JM CW CC ME CB DG PJ MB PP SB DD 1.0 Figure 10.— Cluster analysis (dendrogram) of array data Illustrating similarities in tiabitat type tierpetofaunas. (Abbrev. correspond to thiose listed for tiabitats in table 1 .) This survey reveals that certain spe- cies are widespread, occurring in several habitats, but many species are limited to specific habitat types. Also, some species occur on most sample sites within a habitat type and others on only a few. There ap- pear to be at least 3 major factors contributing to distributional pat- terns of amphibians and reptiles in the study area. Geographic Limitations The ranges of certain species only peripherally occur in western Arizona. Cnemidophorus burti, Phyl- lorhynchus browni, Masticophis bilinea- tus lineolatus, and Bufo retiformis oc- cur principally in northern Mexico whereas others such as Holbrookia maculata, Eumeces obsoletus, Gastro- phyrne olivacea, and Bufo debilis are mostly east and north of the study area (Stebbins 1985). Bufo retiformis, Gastrophyrne olivacea, and Bufo debilis are associated with low elevation (457-915 m or 1500-3000 ft) desert grassland (Jones et al. 1983), and these habitats are mostly absent in the central and northern portions of the study area. However, habitat suitable for other species listed above appears to be available throughout most of the study area. Physical barriers, such as topogra- phy, elevation, and climate may have presented these species from coloniz- ing or immigrating into suitable habi- tats to the north and west (see Con- nor and Simberloff 1979, Case 1983, Jones et al. 1985 for discussion of the influence of physical barriers on colo- nization/immigration). In addition, competition between species may have limited individual species' ranges during initial and subsequent colonization of suitable habitats (e.g., during periods of large climatic changes). Perhaps the best example of this is the distributional relation- ship between Eumeces gilberti and E. 121 ohsoletus. E. gilberti belongs to the skiltonianus group of skinks, whose evolutionary center is the western United States (Taylor 1935, Rogers and Fitch 1947). Conversely, E. obsoletus evolved in the Great Plains region (Fitch 1955). Both of these lizards occupy seem- ingly identical, but separate, habitats in central Arizona, and their distribu- tions come together in chaparral and desert grassland habitat types near Cordes Junction; the westernmost range of E. obsoletus is just east of Interstate Highway 17 and the east- ernmost range of E. gilberti is just west of the highway. These lizards are similar in appearance, with E. ob- soletus averaging slightly larger in size. Although subtle differences in mi- crohabitat cannot be ruled out as fac- tors influencing their ranges, it ap- Similarity pears that these lizards are mutual exclusive (competitive exclusion). Several remnant stands of chapar- ral and desert grassland occur in western and northwestern Arizona at or near the summits of mountain ranges. These relict stands or habitat islands are isolated within creo- sotebush and Sonoran Desert habi- tats as a result of the retreat of the last Ice Age (see Van Devender and Spaulding 1977). Data collected in my study show that several reptiles typically found in "upland" habitats (e.g., large continuous stands of des- ert grassland and woodlands associ- ated with the Colorado Plateau of central and northern Arizona) inhabit these isolated mountain stands, al- though the number and composition of these upland species vary among mountains. Habitat island size ap- pears to be of primary importance in determining the number of upland present species (see Jones et al. 1985). The turtles Pseudemys scripta and Trionyx spiniferus are present along the Gila River as a result of introductions. P. scripta is a popular pet, and specimens have been re- leased along the Gila River in south- west Phoenix. T. spiniferus was intro- duced along the Colorado River in the early 1900's (Stebbins 1985); pre- sumably, these populations ex- panded into the Gila River at the confluence of the Gila and Colorado rivers near Yuma. Microhabitats and Physical Characteristics of Habitat Many studies have shown a strong relationship between the distribution and abundance of amphibians and reptiles and the presence and amount of certain microhabitats (Norris 1953, Pianka 1966, Zweifel and Lowe 1966, Fleharty 1967, Pianka and Parker 1972). The distribution of a number of species within western Arizona area appears to be influenced by the presence of microhabitats on sites, although most of the widespread species, such as Cnemidophorus tigris, Pituophis melanoleucus, and Lam- propeltis getulus show no strong rela- tionship with any specific habitat components, others (e.g., Urosaurus ornatus and Sceloporus magister) occur on sites with trees and downed litter. Many sites in the study area, includ- ing desert and upland habitat types, have trees and downed logs, and this probably accounts for these species' wide distributions. The habitat analy- sis revealed that several species are associated with specific substrate types (e.g., rock), density or height of the vegetation canopy, type of vege- tation (shrubs or grasses vs. trees), or presence of downed litter. Species' associations with certain microhabitats may reflect their physi- cal or behavioral limitations. For example, Eumeces gilberti may be re- stricted to sites with large amounts MR SD MD CA OC JM CW CC ME CB DG PJ MB PP SB DD 1.0 Figure 1 1 .—Cluster analysis (dendrogram) of all data illustrating similarities in tiabitat type hterpetofaunos. (Abbrev. correspond to thiose listed for hiabitats in table 1 .) 122 of downed litter (primarily leaves and logs) because of its low preferred body temperature and feeding habits (Jones 1981b, Jones and Glinski 1985). Large amounts of surface litter on certain riparian sites may explain the occurrence of this lizard in cotton- wood-willow riparian sites within desert regions (down to 549 m or 1800 ft) (see Jones and Glinski 1985). Several other species typically found on upland habitats (e.g., chaparral), such as Tantilla hobartsmithii, Copho- saurus texana, Masticophis bilineatus, and Diadophis punctatus, also may persist on riparian habitats within deserts because of the high moisture regime associated with surface litter, higher humidity, and surface water (Jones and Glinski 1985). A similar relationship appears to exist in desert habitats occupied by Xantusia vigilis. This lizard also has a low preferred body temperature, and it only occurs on Mojave Desert sites occupied by agaves (Agave spp.) and yuccas (Yucca spp. and Nolim spp.); these plants create cool, moist mi- crohabitats within desert habitats. In the southern part of its range, X. vig- ilis only occupies Sonoran Desert on steep slopes in mountain canyons, or on top of mountains (> 1220 m or 4000 ft) in chaparral habitats. This shift in habitat association may re- flect increased average temperature and aridity associated with decreas- ing latitude; canyons and mountain summits may be the only sites mod- erate enough to support this lizard. A similar moisture or temperature relationship may also account for dif- ferences observed in habitat type as- sociations of Tantilla hobartsmithii, Cophosaurus texana, and Diadophis punctatus in the eastern and western portions of their ranges. In the west- ern portion of the study area, these reptiles occur only in chaparral or riparian habitat types (excluding mixed riparian scrub habitats). In the eastern and southeastern portions of the study area, these species also oc- cur in the Sonoran Desert habitat type. Eastern and southeastern Sono- ran Desert habitats within the study area are more extensive than those to the west and northwest, and they are not interrupted by large creo- sotebush habitats; western and northwestern sites are restricted mostly to mountain slopes, separated by extensive creosotebush flats. In addition, eastern and southeastern sites appear to have more springs and perennial creeks than western and northwestern sites, and this ad- ditional moisture might contribute to the presence of these species on these sites. The presence of surface water also has a profound affect on the distribu- tion and abundance of certain species within the study area. Kinosternon sonoriense, Trionyx spiniferus, Thamno- phis cyrtopsis, Bufo alvarius, Bufo mi- croscaphus, Bufo woodhousei, Rana pipi- ens, Rana catesbeiana, Hyla arenicolor, and Ambystoma tigrinum occur only on sites with permanent water (springs, creeks, rivers, dirt tanks). All of these species are restricted to permanently watered sites because of a combination of physiological (Walker and Whitford 1970), mor- phological (Mayhew 1968), reproduc- tive (Justus et al. 1977), or behavioral (Hulse 1974) limitations. In addition to occurring near permanent water, Bufo punctatus also occurs in rock- bound canyons with intermittent wa- ter, and Bufo cognatus, B. debilis, B. retiformis, and Gastrophyrne olivacea occur on sites with clay and clay- loam soils that accumulate surface water during summer convectional rainstorms. All of these species pos- sess adaptations, such as a rapidly developing embryo, that are condu- cive to survival in areas with inter- mittent surface water (Creusere and Whitford 1976). A number of species were verified on fewer than half of the array sites within habitat types. These low per- centages may reflect species' associa- tion with specific microhabitats and the abundance and distribution of microhabitats within habitat types. For example, Chilomeniscus cinctus occurred on less than half of the cot- tonwood-willow and mixed riparian scrub array sites. The habitat analysis shows that this species is associated with sandy and fine gravel soils, but many of the cotton wood-willow ri- parian and mixed riparian scrub sample sites have rocky substrates. Therefore, the substrate type limits this species' range within these habi- tat types. However, there were other spe- cies, especially snakes in excess of 0.5 m (1.5 ft), that were not readily caught in pit-fall traps, although a small percentage of arrays captured a few large snakes; these snakes were feeding on small rodents at the bot- tom of traps. Therefore, the paucity of large snakes on samples sites within habitats probably reflects the ability of larger snakes to escape from pit-fall traps rather than the dis- tribution and abundance of mi- crohabitats within habitat types. Ad- ditionally, amphibians and reptiles with restricted activity patterns (e.g., toads) or home ranges (Xantusia vig- ilis) also were rarely trapped and, therefore, verified on few sites within a habitat. The limited number of mixed broadleaf and chaparral array sites with Gerrhonotus kingi probably reflect a low sampling effort in these habitats during the fall; this lizard's peak activity is during its breeding season in the fall (Robert Bowker personal comm.). Habitat Conditions The condition of habitats may play an important role in determining the distribution and abundance of am- phibians and reptiles. In Arizona, the large variety of land uses within the area may affects the distribution and abundance of certain microhabitats and may account for variation in spe- cies composition within habitats. A number of studies have shown the effects of land uses on amphibians and reptiles and their habitats. These include grazing (Bury and Busack 123 1974, Jones 1981a, Szaro et al 1985), off-road vehicle use (Bury et al. 1977, Bury 1980), forest management (Ben- nett et al. 1980), and stream modifi- cation resulting from water im- poundments (Jones, this volume). Generally, these affect habitat struc- ture. For example, excessive, long- term livestock grazing reduces the abundance and diversity of forbs and perennial grasses. Many former des- ert grassland habitats are now domi- nated by shrubs such as creosotebush (Larrea tridentata) and mesquite (Prosopis glandulosa) (York and Dick- Peddie 1969). Jones (1981a) showed large differences in the presence and abundance of certain lizards on heav- ily vs. lightly grazed sites, especially on riparian, desert grassland, and woodland habitats, attributable to differences in lizard ecology and dif- ferences in habitat structure between heavily vs. lightly grazed areas. Cer- tain lizards, such as Cnemidophorus tigris, prefer open, shrubby sites; these lizards are more abundant on heavily grazed sites where shrubs have replaced grasses and forbs (Jones 1981a). Conversely, certain lizards, such as Eumeces gilberti, pre- fer grassy, moist sites, and are, there- fore, less abundant on or absent from sites where grazing has reduced tree reproduction (e.g., cottonwoods, Populus fremontii on riparian sites) or suppressed grasses (e.g., on desert grassland sites) (Jones 1981a). The reduction of naturally-occur- ring water and the modification of river and stream habitats has been shown to affect the composition of amphibians and reptiles within habi- tats, especially riparian sites (Jones 1988). Platz (1984) attributes the ex- tinction of Ram onca to modification of stream habitats along the Virgin River. Species that prefer lentic or pool habitats should increase on sites with water impoundments, whereas species that prefer lotic or running water should decrease. Natural phenomena, such as fire, also affect species composition within habitats (Kahn 1960, Simovich 1979). Simovich (1979) showed that fire set back succession within chap- arral habitats (grass/forb succes- sional stage), and that these changes resulted in increases in certain spe- cies and decreases in others. As suc- cession proceeded to shrubs and trees, reptiles that were abundant in the grass/forb successional stage (e.g., Phrynosoma coromtum) became less abundant, and others that pre- ferred wooded sites (e.g., Sceloporus occidentalis) became more abundant. Historical vs. Present Distributions Prior to this study, records of am- phibians and reptiles on the study area were limited; one of the primary reasons for which this study was conducted was to assemble basic dis- tribution information. Therefore, range expansions or reductions were hard to document. This study re- sulted in range extensions of ap- proximately 35 species, and clarified the relationship of Arizona habitats to habitats in adjacent geographic regions. Many species, such as Helod- erma suspectum, Eumeces gilberti, Sce- loporus clarki, Tantilla hobartsmithii, and parthenogenic whiptail lizards (Cnemidophorus flagellicaudus, C. uni- parens, and C. velox) proved to be considerably more widespread than previous records indicated — not sur- prising since many areas had never been intensively sampled. The expan- sion of E. gilberti' s range results from the discovery of the California subspecies, E. g. rubricaudatus, in chaparral and pinyon-juniper habi- tats; the distribution of E. g. ari- zonenis is limited to a cottonwood- willow riparian habitat along an 18 km (11 mi) stretch of the Has- sayampa River immediately south of Wickenburg (see Jones et al. 1985, Jones and Glinski 1985). Only one species demonstrated a range reduction. Pure populations of Bufo microscaphus have apparently been reduced due to hybridization with Bufo woodhousei, especially on major drainages. Water impound- ment and diversion-associated changes in aquatic habitats from per- manent riffles and runs to pools may have caused the immigration of B. woodhousei into areas formerly occu- pied by only B. microscaphus (Brian Sullivan personal comm.). There is considerable taxonomic confusion about a population of Kinosternon sonoriense on the Big Sandy River near Wikieup. Because specimens with raised 9th marginal scales had been taken from this area, Stebbins (1966) considered this popu- lation to be Kinosternon flavescens, but Iverson (1978) considered it to be K. sonoriense, based on specimens with- out 9th marginals. Of the 12 indi- viduals observed during this study, 6 had raised 9th marginals and 6 did not. Based on its large separation from the nearest population of K. flavescens, Iverson (personal comm.) considers this population to be an aberrant form of K. sonoriense. Similarity of Habitats Types It is possible to discern definite pat- terns in the diversity of and similari- ties between the herpetofaunas of different habitat types within the study area. There is an apparent ele- vational gradient affecting species diversity. Desert habitats between 610 and 1067 m (2000-3500 ft), ripar- ian habitats between 549 and 1220 m (1800-4000 ft), and chaparral habitats between 1067 and 1525 m (3500-5000 ft) had greater species richness than higher elevation woodland (> 1677 m or 5500 ft, e.g., Ponderosa pine) and desert habitats (> 1220 m or 4000 ft, e.g., sagebrush). Additionally, low elevation desert habitats (> 610 m or 2000 ft, e.g., creosotebush), had rela- tively low species diversity. Higher species diversity on middle elevation habitat types may reflect these habi- tats' moderate environmental and climatic conditions, whereas higher and lower elevation habitats possess 124 extreme environmental and climatic conditions (e.g., temperature). For example, low elevation creosotebush habitats have sparse canopies, and temperatures often exceed 60 C near the surface in summer (Costing 1956). High elevation sites are cold and are often snowcovered until late April so that the growing season is short. Although possessing relatively low species richness, low elevation creosotebush habitats are more di- verse than high elevation sites. These differences in diversity may reflect thermal conditions at these eleva- tional extremes. Many of the species that occur within creosotebush are nocturnal, and, therefore, these ani- mals avoid exposure to extreme sur- face heat. On higher elevation habi- tats, the problem is not avoiding heat but, rather, gaining heat for activity. Other than along rock outcrops, rapid heating is difficult for reptiles at higher elevations. Differences be- tween diversity and species composi- tion on medium elevation habitat types probably reflect differences in microhabitat abundance and diver- sity on habitat types (see earlier dis- cussion on microhabitats). Lack of diversity on disclimax desert grass- land sites probably reflects the lack of vegetation structure on these sites. There was similarity in the herpe- tofaunas of certain habitat types. All desert habitats, except sagebrush, had very similar herpetofaunas, as did most moderate elevation habitats (e.g., chaparral, pinyon-juniper, and mixed riparian scrub). This is pre- dictable because all of these habitats occur in close proximity and are structurally similar. There was a moderate degree of similarity be- tween cottonwood-willow riparian and desert habitats, chaparral and cottonwood-willow riparian, and chaparral and desert habitats. Be- cause cottonwood-willow riparian habitats traverse through both desert habitats and upland habitats, many of the species associated with the surrounding habitats also frequent riparian sites; riparian sites are im- portant sources of food and cover (Ohmart and Anderson 1986). Simi- larities between chaparral and desert habitat types, such as Mohave Des- ert, Sonoran Desert, and mixed ripar- ian scrub, result from occurrence of typical desert species (e.g., Callisau- rus draconoides) on upland sites rather than the occurrence of upland spe- cies (e.g., E. gilberti) on desert sites. The diversity of and similarities among amphibian and reptile com- munities of habitat types also may have been affected by the proximity of habitat types to evolutionary cen- ters. Because of the many new rec- ords for herpetofauna generated by this study, we now have a better pic- ture of the sources of diversity for this area. Many of the amphibians and reptiles occurring in the Sonoran and Mohave Deserts evolved in Baja California and along the western sec- tion of mainland Mexico; these areas were linked until their separation 13 million years ago (Murphy 1983). With the retreat of pleistocene glacia- tion and spread of xerophyllous and desert habitats, amphibians and rep- tiles moved northward into southern California and southwestern Ari- zona; hence, Sonoran and Mohave Desert habitat types have similar her- petofaunas. Although many species immigrated into what is today the Sonoran and Mohave Deserts, only a few species immigrated as far north as the Great Basin Desert. Higher ele- vations may have precluded many of these species from colonizing the Great Basin desert habitat types and, hence, it's herpetofauna is different from and less rich than those of the other two deserts. The discovery of the subspecies Eumeces gilberti rubricaudatus, for- merly unknown in Arizona, suggests that Arizona chaparral was closely associated with (Zalifomia chaparral during Pleistocene glaciation; E. g. rubricaudatus evolved in California sclerophyll woodland (Taylor 1935). That parthenogenic whiptail lizards, such as Cnemidophorus flagellicaudis, C. uniparens, and C. velox, are absent from California chaparral suggest that these species evolved after Pleis- tocene glaciation. There were a few inconsistencies in the results of the two analyses used to determine similarity between habitats (the cluster analysis of all data vs. the cluster analysis of only array data). These inconsistences par- tially result from the inconsistency of arrays to capture turtles and medium and large-sized snakes, and partially from the analyses themselves (see the Methods Section for a more detailed explanation). Conclusions and Recommendations This survey indicates that most spe- cies present within western Arizona are widespread, and that few war- rant special management considera- tion. However, it is evident that cer- tain species are more vulnerable to range or population reduction than others. Generally, these species are those that require microhabitats that are easily affected by land uses. It appears that habitat moisture and moderated surface temperatures are of primary importance to many species in western Arizona. Downed and dead surface litter (debris), such as logs and leaves, play a major role in moderating surface temperature and enhancing moisture (Dauben- mire 1974). Horizontal and vertical vegetation structure also help moder- ate temperatures and increase mois- ture. In developing management schemes, priority should be given to maintaining or enhancing surface lit- ter and vegetation structure. It is im- portant to maintain tree reproduc- tion, and to leave litter on the surface rather than piling and burning it. The latter practice is especially important on cottonwood-willow riparian sites within deserts, since many species in riparian sites are totally dependent on surface litter for their survival (Jones and Glinski 1985). Many ripar- ian sites within the study area have 125 reduced amounts of trees and sur- face litter, principally because live- stock have greatly reduced the repro- duction of Cottonwood trees by re- ducing the survival of seedlings (Jones 1981a). Management prescrip- tions are needed on these sites to in- crease the survivorship of seedling and young cottonwood trees. Populations of ''upland" species (e.g., Eumeces gilberti) on habitat is- lands are more vulnerable to impacts associated with certain land uses than populations occurring on major, continuous stands. Jones et al. (1985) described these habitat islands, some only 10 ha (25 acres) in size. Loss or fragmentation of any portion of these islands could result in the local extir- pation of one or several upland spe- cies (see Bury and Luckenbach 1983 and Harris 1984 for the effects of habitat fragmentation and habitat loss on species occurring on habitat islands). Because even small modifi- cations to island habitats can result in the extirpation of upland species, proposed projects should be moved to alternative sites whenever pos- sible; mitigation strategies should be used only as a last resort. Top prior- ity should be given to protecting these sites in land-use and on-the- ground activity plans (see Jones et al. 1985 for specific locations of these sites). Although all amphibians in the study area (excluding Bufo mi- croscaphus) appear to be stable, water in many habitats continues to be de- veloped. In addition, new informa- tion (Bruce Bury personal comm. Com and Fogleman 1984) suggest that several populations of ranid frogs have been extirpated from western North America, although there is no apparent cause for their extirpation. Considering the heavy use of spring and creek water, and the reported loss of many ranid populations in the West, high prior- ity should be given to monitoring amphibian populations at springs and creeks in Arizona. Additionally, high priority should be given to de- termining the extent of hybridization between the toads B. microscaphus and Bufo woodhousei. Pure popula- tions of B. microscaphus should be lo- cated and protected against hybridi- zation with B. woodhousei. If only a few pure populations are found, the Arizona Game and Fish Department and/or the U.S. Fish and Wildlife Service should set up a captive breeding program to reduce this toad's risk of extinction. Although I obtained distributional records of Gopherus agassizii, Burge (1979, 1980) and Schneider (1980) provide considerably more detail on the needs of this species. However, many biologists consider G. agassizii to be declining throughout most of its range. The U.S. Fish and Wildlife Service (1987) continues to list G. agassizii as a species that needs fur- ther study to determine its status, although it has determined that the Federal listing of the tortoise throughout its range is warranted but precluded by species needing more immediate listing (e.g., species in more eminent danger of extinc- tions). The BLM should continue to give high priority to the study and management of this species in Ari- zona. If the few measures suggested in this paper are implemented, western Arizona should continue to support one of North America's most diverse herpetofaunas. ACKNOWLEDGMENTS I am indebted to several people for the completion of this project. Don Seibert, Bob Furlow, and Ted Cor- dery were instrumental in obtaining funding, equipment, and personnel for this study. Lauren Kepner, Tim Buse, Dan Abbas, Terry Bergstedt, Kelly Bothwell, William Kepner, Dave Shaffer, Bob Hall, Ted Cordery, Scott Belfit, Ted Allen, Ken Relyea, Becky Peck, Brian Millsap, Jim Zook, Jim Harrison, and Greg Watts helped collect both animal and habitat data. Special thanks to W.L. Minckley and M.J. Fouquette for technical contribu- tions to this project's study design, and to the Bureau of Land Manage- menf s line managers and supervi- sors. Bill Barker, Roger Taylor, Barry Stallings, Dean Durfee, Gary McVicker, and Malcolm Schnitkner, for their continuous support of re- source inventories on public lands. I thank John Fay, Scott Belfit, R. Bruce Bury, and Robert Szaro for review of this manuscript. Finally, all of us who strive for the conservation of nongame wildlife on public lands are indebted to Gary McVicker, Bill McMahan, and Don Seibert for their tireless efforts in getting top-level management to support nongame programs. LITERATURE CITED Bennett, Stephen H., J. Whitfield Gib- bons, and Jill Glanville. 1980. Ter- restrial activity, abundance, and diversity of amphibians in differ- ently managed forest types. American Midland Naturalist 103:412-416. Brown, David E. 1978. The vegeta- tion and occurrence of chaparral and woodland flora on isolated mountains within the Sonoran and Mohave Deserts in Arizona. Jour- nal of Arizona Academy Sciences 13:1-12. Brown, David E., Charles H. Lowe, and Charles P. Pase. 1979. A digit- ized classification system for the biotic communities of North America, with community (series) and association examples from the Southwest. Journal of Arizona Academy of Science 14, Suppl. 1, p. 16. Burge, Betty L. 1979. A survey of the present distribution of the desert tortoise (Gopherus agassizii) in Ari- zona. U.S. Bureau of Land Man- agement, Denver, Colorado. Con- tract No. YA-512-CTB-108. Burge, Betty L. 1980. Survey of the present distribution of the desert 126 tortoise, Gopherus agassizii, in Ari- zona. U.S. Bureau of Land Man- agement, Denver, Colorado. Con- tract No. YA-512-CT8-108. Bury, R. Bruce. 1980. What we know and do not know about off-road vehicle impact on wildlife. In R.N. Andrews and P.F. Nowah (eds.). Off-road vehicle use: a manage- ment challenge. U.S.D.A. Office of Environmental Quality, 748 p. Bury, R. Bruce and Stephen D. Busack. 1974. Some effects of off- road vehicles and sheep grazing on lizard populations in the Mojave Desert. Biological Conser- vation 6:179-183. Bury, R. Bruce, Roger A. Luckenbach, and Stephen D. Busack. 1977. Effects of off-road vehicles on vertebrates in the California Desert. U.S. Fish and Wildlife Service Wildlife Re- search Report No. 8. Bury, R. Bruce and M.G. Raphael. 1983. Inventory methods for am- phibians and reptiles. Proceedings of the International Conference on Renewable Resources, Inventories for monitoring changes and trends. Oregon State University, Corvallis. Bury, R. Bruce and Roger A. Lucken- bach. 1983. Vehicular recreation in arid lands drives: biotic responses and management alternatives, p. 217-221. In R.H. Webb and H.G. Wilshire (eds.). Environmental ef- fects of off-road vehicles: impacts and management in arid regions. Springer- Verlag, New York, New York. Buse, Timothy C. 1981. Distribution, ecology, and habitat management of the reptiles and amphibians of the Cerbat planning unit, Mohave County, Arizona. U.S. Bureau of Land Management, Kingman, Ari- zona. Unpubl. Man. Buse, Timothy C. 1983. Herpetologi- cal records from northwestern Arizona. Herpetol. Rev. 14:53-54. Campbell, Howard W. and Stephen P. Christman. 1982. Field tech- niques for herpetofaunal commu- nity analysis, p. 193-209. In Norm J. Scott (ed.), Herpetological com- munities, U.S. Fish and Wildlife Service, Wildlife Research Report Number 13. Case, Thomas J. 1983. The reptiles: ecology. In T.J. Case and M.L. Cody (eds.). Island biogeography in the Sea of Cortez. University of California Press, Berkeley. Connor, Edward F. and Daniel Sim- berloff. 1979. The assembly of spe- cies communities: chance or com- petition? Ecology 60:1132-1140. Com, Paul Stephan and James C. Fogleman. 1984. Extinction of montane populations of the north- ern leopard frog (Rana pipiens) in Colorado. Journal of Herpetology 18:147-152. Creusere, F. Michael and Walter G. Whitford. 1976. Ecological rela- tionships in a desert anuran com- munity. Herpetologica 32:7-18. Daubenmire, Rexford F. 1974. Plants and environment: a textbook of autecology. John Wiley and Sons, New York, New York. 3rd Ed. Fitch, Henry S. 1955. Habits and ad- aptations of the Great Plains skink (Eumeces obsoletus). Ecological Monographs 25(3):59-83. Reharty, Eugene D. 1967. Compara- tive ecology of Thamnophis elegans, Thamnophis cyrtopsis, and Thamno- phis rufipunctatus in New Mexico. Southwestern Naturalist 12(3):207- 230. Hair, Jay D. 1980. Measurements of ecological diversity, p. 269-275. In S.D. Schemnitz (ed.). Wildlife Management Techniques Manual. The Wildlife Society, Washington, D.C. Harris, Larry D. 1984. Island biogeo- graphy applied: old growth is- lands and wildlife conservation in the western Cascades. University of Chicago Press, Chicago, Illinois. Hulse, Arthur C. 1974. An autecol- ogy study of Kinosternon sonoriense leconte (Chelonia: Kinosternidae). Phd. Dissertation, Arizona State University, Tempe. Iverson, John. 1978. Distributional problems of the genus Kinosternon in the American Southwest. Copeia 1978:476-479. Jones, K. Bruce. 1981a. Effects of grazing on lizard abundance and diversity in western Arizona. Southwestern Naturalist 26(2): 107- 115. Jones, Kenneth Bruce. 1981b. Distri- bution, ecology, and habitat man- agement of the reptiles and am- phibians of the Hualapai- Aquarius planning areas, Mohave and Yavapai Counties, Arizona. U.S. Bureau of Land Manage. Technical Note No. 353, Denver, Colo. Jones, K. Bruce. 1987. Amphibians and reptiles, p. 267-290. In A.Y. Cooperrider, R.J. Boyd, and H.R. Stuart (eds.). Inventory and moni- toring of wildlife habitat. U.S. Bu- reau of Land Management, Den- ver, Colorado xviii, 858 p. Jones, K. Bruce., Dan R. Abbas, and Terry A. Bergstedt. 1981. Herpeto- logical records from central and northwestern Arizona. Herpeto- logical Review 12(1):16. Jones, K. Bruce and Patricia C. Glinski. 1985. Microhabitats of liz- ards in a southwestern riparian community, p. 355-358. In R. Roy Johnson et. al.. Riparian ecosys- tems and their management: rec- onciling conflicting uses. First North American riparian confer- ence. Rocky Mountain Forest and Range Experimental Station, Gen- eral Technical Report Number RM-120., Fort Collins, Colo. Jones, K. Bruce, Lauren P. Kepner, and William G. Kepner. 1983. Anurans of Vekol Valley, central Arizona. Southwestern Naturalist 28(4):469-470. Jones, K. Bruce, Lauren P. Kepner, and Thomas E. Martin. 1985. Spe- cies of reptiles occupying habitat islands in western Arizona: a de- terministic assemblage. Oecologia 66:595-601. Jones, K. Bruce, Lauren M. Porzer, and Kelly J. Bothwell. 1982. Her- petological records from westcen- tral Arizona. Herpetological Re- view. 13(2):54. 127 Justus, J. T., Mark Sandomir, Tom Urquhart, and Barbara Orgel Evan. 1977. Developmental rates of two species of toads from the desert Southwest. Copeia 1977:592-594. Kahn, Walter C. 1960. Observations on the effect of a bum on a popu- lation of Sceloporus occidentalis. Ecology 41(2):358-359. Mayhew, William W. 1968. Biology of desert amphibians and reptiles, p. 195-356. In G.W. Brown, Jr. (ed.). Desert biology. Academic Press, New York, New York. Murphy, Robert W. 1983. Paleobio- geography and genetic differentia- tion of the Baja California herpe- tofauna. Occasional Papers Cali- fornia Academic Sciences 137:1-48. Norris, Kenneth S. 1953. The ecology of the desert iguana, Dipsosaurus dorsalis. Ecology 34:265-287. Ohmart, Robert D. and Bertin W. Anderson. 1987. Riparian habitats, p. 169-200. In A.Y. Cooperrider, R.J. Boyd, and H.R. Stuart (eds.). Inventory and monitoring of wild- life habitat. U.S. Bureau of Land Management, Denver, Colorado xviii, 858 p. Oosting, Henry J. 1956. The study of plant communities. W.H. Freeman and Company, San Franciso, Cali- fornia. Pianka, Eric R. 1966. Convexity, des- ert lizards, and spatial heterogene- ity. Ecology 47:1055-1059. Pianka, Eric R. and William S. Parker. 1972. Ecology of the igua- nid lizard, Callisaurus draconoides. Copeia 1972:493-508. Pimental, Richard A. 1979. Mor- phometries: the multivariate analysis of biological data. Ken- dall/Hunt Publishing Company, Dubuque, Iowa. Platz, John E. 1984. Status report for Rana onca Cope. Status report pre- pared for the U.S. Fish and Wild- life Service, Albuquerque, New Mexico. 28 p. Rogers, Thomas L. and Henry S. Fitch. 1947. Variation in the skinks (Reptilia: Lacertilia) of the skil- tonianus group. University of California Publications in Zoology 48(4):169-220. Schneider, Paul B. 1980. A popula- tion analysis of the desert tortoise in Arizona during 1980. U.S. Bu- reau of Land Management, Phoe- nix, Arizona. Contract No. AZ- 950-279-0014. Simovich, Marie A. 1979. Post fire reptile succession. Cal-Neva 1979:104-113. Stebbins, Robert C. 1966. A field guide to western reptiles and am- phibians. Houghton Mifflin Co., Boston, 1st Edition. Stebbins, Robert C. 1985. A field guide to western reptiles and am- phibians. Houghton Mifflin Co., Boston, 2nd Edition. Szaro, Robert C, Scott C. Belfit, and J. Kevin Aitkin. 1985. Impact of grazing on a riparian garter snake, p. 359-363. In R. Roy Johnson et. al.. Riparian ecosystems and their management: reconciling conflict- ing uses. First North American riparian conference. Rocky Moun- tain Forest and Range Experimen- tal Station, General Technical Re- port Number RM-120., Fort Collins, Colorado. Taylor, Edward H. 1935. A taxo- nomic study of the cosmopolitan scincoid lizards of the genus Eumeces with an account of the distributions and relationships of its species. University of Kansas Science Bulletin 23:1-643. U.S. Fish and Wildlife Service. 1987. Notice of findings on petitions and initiation of status review. Federal Register 52(126):24485-24488. Van Devender, Thomas R. and Wil- liam G. Spaulding. 1977. Develop- ment of vegetation and climate in the southwestern United States. Science 204:701-710. Walker, Robert F. and Walter G. Whitford. 1970. Soil water absorp- tion capabilities in selected species of anurans. Herpetologica 26(4):411-416. York, John C. and William A. Dick- Peddie. 1969. Vegetation changes in southern New Mexico during the past hundred years, p. 157-166. In W.G. McGinnes and B.J. Goldman (eds.). Arid lands in per- spective. University of Arizona Press, Tucson. Zweifel, Richard G. and Charles H. Lowe. 1966. The ecology of a population of Xantusia vigilis, the desert night lizard. American Mu- seum Novitates 2247:1-57. 128 Multivariate Analysis of the Summer Habitat Structure of Rana pipiens Schreber, in Lac Saint Pierre (Quebec, Canada)^ N. Beauregard^ and R. Leclair Jr^ Abstract.— Thirty stations representing various ripar- ian habitats typical of the Lac Saint Pierre area were sampled with a system of drift fences and funnel traps to characterize the summer habitat structure of a leopard frog population. A discriminant analysis indicates that habitats with high frog density (1) ore close to the marsh line, (2) have a tall herbaceous stratum with high richness and (3) have a low moss cover. A stepwise multiple regression model used 5 of the vegetation structure variables, and explains CO. 70% of the variability associated with frog density among stations. The leopard frog, Rana pipiens, is the most abundant frog species in the Lac Saint Pierre area (Leclair 1985, Leclair and Baribeau 1982, Paquin 1982), and also one of the most com- mon vertebrates in aquatic communi- ties in North America (Dole 1965a). Despite this apparent abundance, many herpetological surveys made in the last fifteen years have shown dra- matic reductions in leopard frog den- sities. Gibbs et al. (1971) estimated a 50% drop in the global population of leopard frogs in the USA, during the 1960's. Many other workers have re- ported population reductions and local extinctions in Canada and the USA (Collins and Wilbur 1979, Cook 1984, Degraaf and Rudis 1983, Froom 1982, Hayes and Jennings 1986, Hine et al. 1981). In area where hypothesis of preda- tion or competition by introduced species (Bullfrogs or predatory fishes) (Hayes and Jennings 1986) does not apply, two major causes have been invoked as responsible for ^ Paper presented at symposium. Man- agement of Amphibians, Reptiles, and Small t^ammals in North America. (Flag- staff . AZ, July 19-21, 1988). ^Norman Beauregard is a graduate stu- dent in Environmental Sciences, Universite du Quebec a Trois-Rivieres, Department of chimie-biologie, CP. 500 Trois-Rivieres, Que- bec, Canada, G9A 51-17. ^Raymond Leclair Jr. is Professor of Her- pefology and Ecology. Universite du Quebec a Trois-Rivieres, Department of chimie-biologie, CP. 500 Trois-Rivieres, Que- bec, Canada, G9A 5H7. this situation (1) overexploitation of natural stocks, and (2) loss or altera- tion of habitat rendering it unsuitable for R. pipiens (Cook 1984, Frier and Zappalorti 1984, Leclair 1985, Mar- cotte 1981, Rittschof 1975). Riparian habitats have been especially affected by human activities (Sarrazin et al. 1983, MLCP 1985). In Canada, 50% of the wetlands that once supported wildlife have now been reclaimed for agricultural, industrial or urban de- velopment, or have been altered by pollution (SCF 1980). Even greater riparian habitat has occurred along the St.Lawrence river, where 70% of the riparian habitats have been elimi- nated. According to Ministere des Loisirs, de la Chasse et de la Peche (MLCP) (1985), essential habitats are those vital to population or species survival, whether these habitats are used temporarily or permanently. This definition emphasizes several crucial aspects of amphibian habitat use, i.e. the use of aquatic as well as terrestrial habitats, and of migratory routes between the two. Up to now, quantitative studies of habitat re- quirements for anuran species have focused mostly on the aquatic habi- tats (Beebee 1977, Clark and Euler 1982, Dale et al. 1984, Gascon and Planas 1986, Hine et al. 1981). This situation largely results from the lack of appropriate quantitative sampling method for amphibian populations in terrestrial habitats (Bury and Raphael 1983, Clawson et al. 1984). Recently, Campbell and Christman (1982), and Vogt and Hine (1982) have devel- oped adequate techniques that help overcome this situation. The aims of the present study were (1) to characterize the structural aspects (biotic and abiotic) of the ter- restrial habitats of Rana pipiens and (2) to develop a model relating frog abundance to habitat descriptors. Study Area The study area is a 30 X 0.9 km strip extending from Trois-Rivieres to Ber- thierville (Quebec, Canada), on the north shore of Lac Saint Pierre (73*30' W X 46°05' N). The Lac Saint Pierre covers about 300 km^ and is formed by a widening of the St.-Lawrence river (fig. 1). The lake flood plain is extensive (Tessier et al. 1984) and consequently, spawning sites for amphibians are abundant in spring. The habitats most frequently used by Rana pipiens (based on mating call frequencies) are flooded fields of reed phalaris (Phalaris arundinacea) and of purple loosestrife (Lythrum salicaria), mixed with willow {Salix sp.) (Leclair 1983). From these fields, numerous bays, small rivers, drain- ing canals and natural or man-made pools facilitate movement of frog to- wards adjacent terrestrial habitats. According to the maps produced by Denis Jacques (1986) and by Tessier and Caron (1980) on the ri- parian vegetation of Lac Saint Pierre, 129 at least ten plant communities may be recognized on the criterion of dominant species. These plant com- munities can be grouped in six differ- ent physionomic types (table 1). Thirty stations were selected in order to sample the diversity of habitats. From the maps, sampling sites were located in habitat patches not having less than 2500 m^ of homogeneous vegetation. The final choice of sites was determined by physical and le- gal accessibility. Materials and Methods Sampling Technique At each station, frogs were sampled with 12 funnel traps placed on each side of two 15 m drift fences made of polyethylene and forming a right angle (fig. 2). Dirt and/or litter was brushed into the mouth of each fun- nel to simulate a natural entrance (Clawson and Baskett 1982). This de- sign has been shown to allow for sampling in various kinds of terres- trial habitats, and to provide data for the estimation of demographic para- meters and for comparison between various habitats (Campbell and Christman 1982, Clawson et al. 1984). Funnel traps were opened for at least 10 consecutive days in each pe- riod (10 days in May, 10 in June, 12 in July, 10 in august, 22 in Fall) and were checked every other day. Data recorded for each capture were: date, station number, direction of capture (N, S, E or W), species, sex and snout-urostyle length. Captured frog were marked by clipping the fourth digit of the hindfoot. Clipped phalanges were kept for age determi- nation trough skeletochronological examination (Leclair and Castanet 1987). Because of the way the arrays were used for sampling, captures re- flected the relative abundance of frogs among stations, not their abso- lute density. Environmental Variables^ Each station was characterized by 6 spatial variables: distance to the marsh line (DMARSH), to the nearest permanent pool (DWATERP), to the nearest temporary pool (DWATERP), ^See appendix 1 for all abbreviations used in fhe text. ^Variable measured monthly. ^ 7m h 15 m -I Figure 2.— Schematic representation of the trapping arrays. Figure 1 .—General location of the study area (upper map) and the study area's relationship to lake St. Pierre (Quebec, Canada) (lower map). 130 to the nearest human aUeration (road, path, residence, crop) (DHU- MAN), and to the nearest open habi- tat without shrub or tree canopy (DOPEN) or closed habitat with can- opy (DCLOSE). All distances were measured in the field with a topofil marker (lost thread measure appara- tus), except for some measures of DMARSH taken from a 1:10 000 to- pographical map. Elevation from the marsh ground (ALTREL) was taken with a Keuffel and Essel altimeter. Water table level (WTABLE^) was measured with a piezometer, placed 1 m deep. Edaphic variables measured were: soil moisture (MOIST^), from oven- dried soil samples (80 °C, 24 hrs); soil fractions (SAND, SILT, CLAY), as determined by the Bouyoucos method (Bouyoucos 1936); soil water Table 1 .—Characteristics of the sampiing stations according to physiog- nomic type and to major plant species. Sta. Physionomic type Code 1 Open dry field O 2 Brushy dry field B 3 Wooded swamp F 4 Riparian marsh M 5 Shrub swamp S 6 Wet prairie P 7 Open dry field O 8 Wooded swamp F 9 Wet prairie P 10 Wooded swamp F 11 Wet prairie P 12 Shrub swamp S 13 Wet prairie P 14 Shrub swamp S 15 Riparian marsh M 16 Wet prairie P 17 Brushy dry field B 18 Wet prairie P 19 Brushy dry field B 20 Open dry field O 21 Open dry field O 22 Wet prairie P 23 Shrub swamp S 24 Riparian marsh M 25 Wooded swamp F 26 Shrub swamp S 27 Wooded swamp F 28 Riparian marsh M 29 Wet prairie P 30 Wet prairie P Code Major plant species Solidago canadensis, Aster umbellafus Spirea lafifolia, Populus tremuloides Acer saccharinum, Laporfea canad- ensis Sparganium eurycarpum, Scirpus flu- viafilis Spirea lafifolia. On ode a sensibilis Calamagrosfis canadensis, Phalaris arundinacea Solidago rugosa. Aster umbellafus Acersaccharium, Laporfea canaden- sis Carex lacustris, Lyft-)rum salicaria Salix nigra, Laporfea canadensis Typha lafifolia, Onoclea sensibilis Salix spp . . Myrica gale Calamagrosfis canadensis Salix cordafa, Phalaris arundinacea Sparganium eurycarpum, Equisetum fluviafile Phalaris arundinacea Spirea lafifolia, Populus tremuloides Carex lacustris, Lythrum salicaria Spirea lafifolia, Salix ssp. Solidago canadensis. Aster umbellafus Phleum pratense, Agrostis alba Calamagrosfis canadensis, Phalaris arundinacea Salix ssp . , Rorippa amphibia Sparganium eurycarpum, Sagittaria lafifolia Acer saccharinum . Populus deltoides Salix ssp . , Spirea lafifolia Acer saccharinum, Onoclea sensibilis Sparganium eurycarpum, Rorippa am- phibia Carex lacustris, Lythrum salicaria Calamagrosfis canadensis, Lythrum salicaria pH (PH), as determined with a Fisher pH-meter, and soil temperature (TEMP2). The soil temperature vari- able used in the statistics is ex- pressed as the sum (5 reading per month) of the deviations from the daily mean taken over all stations. Percent of ground covered by lit- ter (LITTER5), dead wood (DEAD- WOOD), mosses (MOSSCOV^), her- baceous plants (HERBCOV^), and percent bare ground (BAREGRND^) was estimated by two independent observers in 5 X 5 meters quadrat, and the mean was recorded. Litter thickness (LITTHICK^) and height of the herbaceous stratum (HERBHGHT^) represented the mean of 5 measurements taken with a me- terstick. Quantitative assessment of vegeta- tion structure was represented by Fox's photometric index (Fox 1979) as: = In (la/lb) H (b-a) where represents the photometric index for the amount of vegetation present in a layer between two levels, when la and lb are the light intensi- ties immediately above and below the layer and H(b-a), the layer thick- ness. Readings of light intensity were taken with a Sekonic light meter, at 0, 20, 50, and 100 cm above ground, above the herbaceous canopy and in the open field adjacent to stahon having closed canopy. At each site, measurements were taken at five points which were then averaged to provide one value. Five photometric index were computed: vegetation in- dex in the 0-20 cm layer (PHOT20^); from 20 to 50 cm (PHOT505); from 50 to 100 cm (PHOTIOO^); herb layer above 100 cm (PHOT+s); and shrub and tree strata (PHOTCAN^). Vegetation structure was also de- scribed in 8 growth-form categories: (TREE) woody plants > 10 cm diame- ter; (SHRUBHI) woody plants > 2.5 m tall); (SHRUBLO) woody plants < 2.5 m tall; (HGH') high graminoid herbs > 100 cm tall; (MGH^) medium 131 size graminoid herbs from 20 to 100 cm tall; (HBLH^) high broad-leaf herbs > 100 cm tall; (MBLH^) me- dium size broad-leaf herbs from 20 to 100 cm tall; (SMALL^) herbs layer below 20 cm tall. Basal area (BA- SARE A) was calculated by measur- ing tree diameter at breast height with a caliper. Richness in herbs spe- cies (NSPHERB), shrubs (NSPSHRUB), and trees (NSPTREE) was determined in a 400 m^ quadrat. Minimal area of homogeneous vege- tation patch (MINAREA) was esti- mated according to the graphical method of Braun-Blanquet (1964). Statistics Spearman rank correlations and chi- square tests were used to test for non-random distribution of captured frogs among age class and among periods of sampling. Chi-square tests were also used to detect a significant movement of frogs. Because some variables were not normally distrib- uted (Kolmogorov-Smirnov test), they were square-root transformed before analysis (indicated on appen- dix 1). For final analyses, the number of variables was reduced by screening an initial principal component analy- sis (PCA), and by using Pearson rank correlations (Green 1979). Because of heterogeneity in the variables meas- ured, the correlation matrix was used to extract the principal components that explained the greatest propor- tion of variability. A second PCA with the 22 extracted variables served to define the structural differ- ences among stations, and to reduce the data set to a few important di- mensions that could identify most of the structural variability among measured habitats. To construct a classification model for potential habitats, a discriminant analysis (DFA) was done on three groupings of stations based on frog abundance. The model was validated through a simulation. A stepwise multiple regression was used to identify which habitat characteristics account for most of the variability in the analyzed data (Clawson et al. 1984). An independ- ent variable was included in the model when its partial F-value was significant (a = 0.05). Partial correla- tion coefficients were used to verify the statistical relation between the dependent variables and the inde- pendent one. This analysis has been identified as the most appropriate to study the combined effects of various habitat variables on wildlife density (Legendre and Legendre 1984). Inter- pretation of the models obtained from such analyses takes into ac- count combinations of variables, but not variables taken individually (Sch- errer 1984). Statistical analyses were performed with SPSS (Nie et al. 1975). Results A total of 798 individuals represent- ing 4 species of anurans (Ram pipiens, r Table 2.— Capture data by sampling period, and by age class. Number of captures By month By age class Sum Sum^ WV4I 1 1 Station M J J A S-O Adult Juv. NMY' adjusted 1 0 1 0 2 2 4 0 1 5 5 2 0 0 3 2 6 5 3 3 11 11 3 1 0 0 1 3 2 2 1 5 5 4 8 n 5 3 7 11 17 6 34 34 5 6 3 4 2 2 15 1 1 17 17 6 7 13 3 3 9 20 11 4 35 35 7 0 1 1 3 1 5 0 1 6 6 8 0 0 1 0 2 0 2 1 3 3 9 8 8 4 2 5 14 7 6 27 27 10 3 9 3 2 2 12 3 4 19 19 11 5 3 1 2 8 7 4 8 19 19 12 11 9 4 4 3 21 3 7 31 31 13 15 8 7 12 5 28 13 4 47 47 14 22 7 10 39 47 16 12 78 107 15 12 15 8 12 39 51 21 11 86 86 16 3 0 2 0 3 1 1 5 7 17 0 0 0 0 0 0 0 0 0 0 18 0 0 0 0 0 0 0 0 0 19 0 0 3 0 0 3 0 0 3 3 20 0 0 0 0 0 0 0 0 0 0 21 0 0 3 1 1 0 0 5 5 5 22 5 2 6 2 0 8 4 0 15 15 23 10 10 2 1 0 9 13 0 23 23 24 26 21 3 2 7 28 23 1 59 59 25 8 3 1 4 3 17 2 0 19 19 26 5 1 1 3 7 12 1 3 17 17 27 0 1 4 9 6 3 5 14 19 28 2 10 12 10 20 3 n 34 47 29 5 2 0 1 1 3 5 0 9 9 30 3 5 6 7 11 3 7 21 29 Total 135 150 86 98 178 362 161 103 647 704 ^ Newly metamorphosed young. 'Sum adjusted for stations not inventoried in May. 132 R. catesheiana, R. sylvatica, Bufo ameri- canus) were captured during the study. Many small rodents (n = 188) and a few weasels (Mustela ermim) were also captured. The results pre- sented here relate only to R. pipiens. Table 2 presents the capture data for the various stations and sampling periods, along with data on popula- tion age structure. The mean capture rate is 0.35 capture/day/ station; sta- tions range from 0 to 1.77 captures/ day/ station. Preliminary trials on three stations in fall 1986 had given 4.8 captures/day/station. Spearman's correlation coefficients (table 3) from among all possible age groups and sampling period pairs were all significant except those be- tween captures at period 1 and newly metamorphosed young (R = 0.3471, P = 0.097). We also compute a contingency table (table 4) to check for independence of the two vari- Table 3.— Spearman rank correlations and signifiance level between cap- tures for all possible age groups and sampling periods pairs. Periods Age Period/Age M J J A s-o Adult Juv NMY' May «*« •♦» NS June 0,88 » July 0.59 0.57 «« *** August 0.63 0.56 0.64 ♦ ♦• **• Sep.-Oct, 0.52 0.47 0.52 0.75 ••• Adult 0.88 0.83 0,74 0.81 0.67 Juvenile 0.82 0.79 0.59 0.49 0.60 0.70 ♦ NMYl 0.35 0.44 0.62 0,70 0,81 0.56 0.46 'Newly metamorphosed youngs. 'P<0.05. "P<0.01 '"P< 0.001. NS = non significant. Table 4.--Contingency table for non random distribution of age group cap- tures among the physlonomic types of habitat. Physionomic types of habitat Age Dry Shrub Wooded Wet Riparian groups habitat' swamp swamp prairie marsh NMY 10 23 11 30 29 Count 4.9 26.5 9.9 28.3 33.4 Exp. vol. Juvenile 3 34 12 48 64 Count 7.7 41.4 15.4 44.2 52.2 Exp. vol. Adult 17 104 37 94 no Count 17.3 93.1 34.7 99,5 117.4 Exp. vol. For all habitats: D.F. = 8. X' = 16.64, 0.025 25. seasonal patterns of movement (i.e. movement away from aquatic over- wintering sites in Spring, movement within a summer foraging range, and movement towards aquatic overwin- tering sites in Fall). Chi-square values (table 7a) showed significant movement for period 1,2 and 5. Individuals cap- tured in the Fall seem to move back towards the lake where they pre- sumably overwinter. A stepwise re- gression model associated with this period (model 3) would thus charac- terize habitat used during Fall migra- tion. Although we got significant chi- square in early season (sampling pe- riods 1 and 2), interpretation is doubtful whether or not there was a migration movement from the over- wintering site (i.e. from south and east). To test for an actual movement, we associated the two compass di- rections in the general direction to- wards the overwintering site and we tested them against the two compass directions in the general direction away from the overwintering site (i.e. north and west). No significant movement was then noted (table 7b). Consequently, we referred to the phenology of the leopard frog de- scribed by Dole (1967) and Rittschof 135 (1975) to decide for grouping of sam- pling periods. In May, as leopard frogs remained at proximity of their reproductive site and because we had only 24 sam- pling stations at that time, data from period 1 were analyzed separately (model 1). Data from June, July and August (periods 2, 3 and 4) were grouped together to construct a single model (model 2) because in June individuals normally tend to disperse in their summer foraging habitats (Rittschof, 1975), and in July and August no definite movement direction was observed (that is typi- cal when foraging habitat is occu- pied). We also analyzed the data for all periods in two general models (models 4 and 5). Model 1 (table 8) explains ca. 82% of the variation in frog density for the month of May using 6 variables. The first one is distance to marsh F U N C T Table 7a.— Capture data by sampling period and by direction and chi- square values for tests of goodness of fit. P values „ 0.05 are considered significant. Month North West South East Exp.vaiue P May 45 24 24 41 33.50 9.42 <0.025 June 42 28 27 52 37.25 11.56 <0.010 July 26 T9 15 25 21.25 3.80 >0.25 August 19 24 31 23 24.25 3.08 >0.25 Sep.-Oct. 106 13 26 33 44.50 117.96 <0.001 Table 7b.- Results of test for nonrandom distribution of captures among the two general directions of movement from and away overwintering sites. Month North + West South + East P May 69 65 0.119 0.067 >0.75 June 70 79 0.272 0.215 >0.50 July 45 40 0.294 0.188 >0.50 August 43 54 1.247 1.031 >0.25 Sep.-Oct. 119 59 20.224 19.556 <0.001 X'. = Chi-square wifh Yates correction for cor^tinuity. 0 N S c 0 R E 2.0-- 1 .0-- 0 -- -1,0- NSPHERB DM ARS ,'22 3.0 - 2.0 - 1,0 0 1.0 2.0 3.0 FUNCTION SCORE 1 Figure 5.— Localization of the sampling stations (represented by tfieir abundance class) in \he discriminant space according to their function score. The relative contribution of each variable (NSPHERB, PHOT+, DMARSH AND MOSSCOV) involved in the two discriminant func- tions is indicated by the length of each vector. Class centroids are represented by *. Mis- classified stations are circled. line. Four of the five other variables are related to soil characteristics: temperature, moisture, silt fraction and bare ground. In model 2 (sum- mer feeding habitats), about 70% of variation in frog density is explained by only three variables: distance to marsh line, number of herb species and clay fraction. The third model, for the month of September and Oc- tober, explains only 34.6% of vari- ation in Fall captures with two vari- ables: DMARSH and NSPHERB. It should be noted that the same two variables explain 61.5% of the vari- ation in model 2. In the next two models (table 8) the seasonal captures were corrected to account for the lower number of stations sampled in May. Model 4 includes five variables: DMARSH and NSPHERB again, and three vari- ables related to vegetation structure (PHOT+, PHOT20, PHOT50). These last three variables explain an addi- tional 21.6% of the variation in frog density in the model. Hooding of St. Lawrence river over our study sites is a major mani- festation in the Lac Saint Pierre area 136 having a strong impact on frog distri- bution as indicated by the presence of the variable DMARSH in all previ- ous models. However, when water recesses, we get a mosaic of habitats that can be found elsewhere in North America but independently of the presence of such marsh line. That is the reason why we ran another mul- tiple regression (model 5) after hav- ing removed DMARSH. This last model emphasizes the significance of vegetation, all 5 variables included being related to vegetation structure. This model explains 69.2% of the variation in total captured frogs. To facilitate the understanding of our interpretation, we present in ap- pendix 2 the significant level of the Pearson rank correlations between Table 8.— Multiple regression models for frog captures. Variable Coefficient Probability Adjusted (p ±SE) (a value for F) Model 1 Capture in May (24 stations) (Intercept) -2.51 ± 3.60 0.4950 0,4950 DMARSH -0.0116 ± 0.0027 0.0005 0,484 TEMP 0.176 ± 0.050 0.0027 0,574 MOIST 0.230 ± 0.053 0.0004 0.636 PHOT50 -1.430 ± 0.429 0.0039 0.691 SILT 0.284 ± 0.074 0.0013 0.770 BAREGRND -0.999 ± 0.419 0.0290 0,818 Model 2 Captures in June, July and August (Intercept) 6.05 ± 2.99 0.0532 DMARSH -0.0355 ± 0.0044 0.0000 0.298 NSPHERB 0.649 ± 0.153 0.0002 0.615 CLAY 0.151 ± 0.051 0.0068 0.700 Model 3 Captures in September and October (Intercept) 1.21 ± 3.54 0.7358 DMARSH -0.0196 ± 0.0056 0.0016 0.121 NSPHERB 0.663 ± 0.204 0.0030 0.346 Model 4 Adjusted total captures (Intercept) -6.80 ± 8.03 0.4053 DMARSH -0.0600 ± 0.0099 0.0000 0.351 NSPHERB 1.745 ± 0.362 0.0001 0.553 PHOT+ -14.859 ± 3.001 0.0000 0.607 PHOT20 4.100 ± 1.274 0.0037 0.693 PHOT50 4.804 ± 1.576 0.0055 0.769 Model 5 Adjusted total captures* (Intercept) 17.64 ± 11.37 0,1339 HGH 20.307 ± 2.732 0.0000 0.196 LITTHICK -6.275 ± 1 .264 0.0000 0,362 PHOT+ -14.060 ± 3.081 0.0001 0.450 PHOTCAN -5.234 ± 1.374 0,0009 0.631 MBLH 5.195 ± 2.135 0,0228 0.692 "DMARSH removed from the model 4. the variables used in the models (1 to 5 and DFA) and all other variables measured in the field. Discussion Model-Related Assumptions In order to use density (estimated by captures) as the dependent variable in multivariate analysis to model sea- sonal habitat structure selected by leopard frogs, certain assumptions must be made. Moreover, we cannot recommend the use of the models presented in table 8 to predict den- sity for leopard frog populations for which the pattern of seasonal fluctua- tion and causes of those fluctuations are unknown (Clawson et al. 1984, Hine et al. 1981). 1. Density as estimated by cap- ture reflects density in the sampled habitats as regards to immigration or emigration to or from neighboring habi- tats (Collins and Wilbur 1979). Ram pipiens is known to be very mobile (Merrell 1977, Rittschof 1975), and is capable of nocturnal excur- sions of 100 m or more (Dole 1965a). Nevertheless, leopard frogs rarely move more than 10 m away from their home range, estimated by Dole (1965b) to vary between 68 and 503 m^. 2. Favorable habitats are char- acterized by frog densities that are higher than those in unfavorable habitats (Par- tridge 1978). However, if density is low (as observed on our study site in 1987 when compared to 1986), all favorable habitats may not be occupied (Partridge 1978). 3. Multivariate analyses are based on matrices of linear correlation between environ- 137 mental variables and an in- dex of abundance (Legendre and Legendre 1984), which neglects saturation and nega- tive feedback effects, as well as non-linear patterns in the species response to environ- mental factors. 4. Competition and predation or the presence of sites for reproduction may control frog distribution patterns but active habitat selection with respect to vegetation struc- ture also plays an important role. Dole (1971) has ob- served that newly metamor- phosed young do not neces- sarily select the first suitable site during dispersal. Finally, in models, it is apparently essential to assume that factors vital for species survival, i.e. those vari- ables actively selected by individu- Figure 6.— Ordination of the stations in relation to marsti distance (DMARSH) and number of hierb species (NSPHERB). Stations withi thie same abundance class are circled by an ellipsoid als, and those identified by the analy- sis do not necessarily coincide. In fact, apparent cause-and-effect rela- tionships are not often testable and require specific study on the func- tional responses of species to the se- lected variables. Weller (1978) indi- cates that the study of habitat stimuli as attractants for wildlife remains to be done. The approach used in this study is valuable when variables de- scribing favorable habitat are re- quired (Clark and Euler 1982, Green 1971,Grier 1984). Classification of Habitats The PCA analysis facilitated under- standing of the multidimensional models, and so allowed for system- atic description of the various habi- tats found in the Lac Saint Pierre floodplain. We found that our pre- established groupings were not an analytical artefact but rather con- firms that there is a structure that can be defined by environmental vari- ables not related to species specific local vegetation. Our results have shown that dif- ferent age groups of R. pipiens are not differently distributed among habi- tats (tables 3 and 4). This conclusion have been drawn with recently meta- morphosed young representing only 16% of total captures but is sup- ported by others studies describing the habitats used by young (Dole 1971, Hine et al. 1981, Rittschof 1975, Whitaker 1961). Our proposed mod- els are those independent of age or size groups. This might not be the same however, for other species as Clark and Euler (1982) and Roberts and Lewin (1979) have noted for Ram clamitans and for R. sylvatica, respectively. The models presented in this pa- per reveal the importance of distance to marsh line in habitat classification. This variable has a high degree of predictive power as to the extent habitat will be utilized by leopard frogs, in the Lac Saint Pierre 138 floodplain. However, systematic sampling in habitats of unknown value indicates the presence of a sig- nificant number of leopard frogs in some wooded and shrub swamps stations far from the marsh (fig. 6). The DFA model is then relevant to show the importance of variables re- lated to structural components of habitat such as herbaceous vegeta- tion (PHOT+, NSPHERB) and moss cover (MOSSCOV). In a similar analysis on Missouri herpetofauna, Clawson et al. (1984) concluded that proximity to water appeared to over- ride other variables in determining the abundance of amphibians. Other multivariate studies (Beebee 1977, Clark and Euler 1982, Dale et al. 1985 and Gascon and Planas 1986) on anuran species habitat have shown that bio-physico-chemical variables related only to the body of water cannot give predictive infor- mation about the absence or presence of a respective amphibian species. Frog Abundance Models In spring, before the growing season, frog distribution is related to soil characteristics, such as temperature. This variable is not significantly cor- related with any other variable meas- ured. It results from the interaction of many variables and may be a key element in habitat selection during that period. The activity of ectoth- erms is known to be related to ambi- ent temperatures (Putnam and Ben- nett 1981), by selecting warmer habi- tat, ectotherms might improve their mobility, thus escaping more easily to predators. Soil moisture is the third most important variable in the first model and appears only in this model. In spring, soil moisture re- flects the speed of water recess after snowmelt and obviously is a variable linked with the proximity of over- wintering and spawning sites. The model proposed for the sum- mer period is the simplest of the models presented in this paper with only 3 descriptors (DMARSH, NSPHERB, CLAY). Soil moisture is not included into this model al- though it has been shown to be the major factor limiting the distribution of anuran species in terrestrial habi- tats (Clark and Euler 1982, Dole 1965a, 1971, Rittschof 1975, Roberts and Lewin 1979). It may be that this variable contains an information al- ready carried in DMARSH variable; its presence in the summer model would then be a redundancy. Clay, on the other hand, is a variable known to play an important role in soil water retention (Ramade 1984). Sampling during Fall migration have shown a significant movement towards aquatic overwintering sites. Model 3 however, with two variables explaining only 34.6 % of frog abun- dance, did not allow identification of preferred migratory corridors. It seems that leopard frogs en route to overwintering sites do not select any particular pathway. The last two models use data from all sampling periods. Model 4, which improves on model 2 (summer model), is interesting because its photometric variables are signifi- cantly correlated (appendix 2) with many of other variables describing the habitat structure. This suggests the value of such indices (Fox 1979) in habitat modeling to quantify vege- tation structure since they can be measured with an instrument (light meter) easy to use. The last model, with 69.2% vari- ability explained, is of more general interest because the local variable DMARSH has been removed. In model 5, the importance of vegeta- tion structure in habitat selection is obvious, and the model can be ap- plied to the entire distributional range of R. pipiens. HGH indicates the importance of graminoids (grasses, sedges, etc.) usually abun- dant in open wetlands. This vegeta- tion cover provides a refuge from many predators and may thus con- tribute to maintaining an abundant frog population (Whi taker 1961). Lit- ter thickness has a negative coeffi- cient in the model, but is positively correlated with HGH, which sug- gests the existence of an optimum foliage density. Dole (1965b, 1967, 1971) mentions that litter may pre- clude direct contact betv-^een the in- dividual and the moist substrate and thus cause higher cutaneous evapo- ration. The three other descriptors summarize the information on vege- tation structure. PHOT+ corresponds to the presence of broad-leaf herbs > 100 cm tall (Rp = .4066, P = 0.026), and graminoids (Rp = .3765, P = 0.040); PHOTCAN represents tree and shrub cover; MBLH indicates broad-leaf plant obstruction between 20 and 100 cm from the ground. These results seem to indicate that vegetation structure, more than spe- cific species composition, is an im- portant factor in habitat selection for Rana pipiens. This finding is similar to that of MacArthur and MacArthur (1961) who have demonstrated that bird species occupying forests and prairies choose their habitat on the basis of foliage density at different levels from the ground, irrespective of plant species composition. Conclusion In summary, we present three types of complementary analysis dealing with wet habitats used by the leop- ard frog during Summer. First, a PCA gives a qualitative description of five kinds of habitats typical to the St. Lawrence river floodplain and offering potential supports to leop- ard frog populations. Second, a DFA model with four easily measured variables allows classification of habitats into three groups of frog abundance. This is a very helpful way to map potential frog species habitats for protective purpose. Fi- nally, five regression models (accord- ing to each phenological periods or whole active season) explain frog abundance variations with only a few important structural variables. 139 Although the models described in this paper cannot fully demonstrate functional relationships between model variables and frog density, suitable modifications of some of these variables (litter thickness, for instance) may increase frog popula- tion. Refinement of these models will require experimental studies on func- tional responses of leopard frogs to specific habitat features. Acknowledgments Thanks are due to Benoit Levesque, Sylvain Cote and Jean-Louis Benoit for field assistance, to Bernard Robert for graphical art, to Gille Houle for the English version of the text and to Marc P. Hayes and Gary K. Meefe for their very constructive comments on the first draft of this paper. Financial support came through grants to N.B. from National Research Council of Canada, Cana- dian Wildlife Federation, and Centre d'Etude Universitaire (Quebec) and from direct funding from Ministere Quebecois du Loisir, de la Chasse et de la Peche, and Universite du Quebec a Trois-Rivieres. Literature Cited Beebee, Trevor J.C. 1977. Habitats of the British amphibians (l):Chalk uplands. Biological Conservation 12:179-193. Bouyoucos, G.J. 1936. Directions for making Mechanical Analysis of Soils by the Hydrometer Method. Soil Science 32:225-228. Braun-Blanquet, Josias. 1964. Plant sociology. McGraw-Hill, New York. 865 p. Bury, R. Bruce and Martin G. Ra- phael. 1983. Inventory methods for amphibians and reptiles, p. 416-419 in J. F. Bell and T. Atter- bury, eds. Renewable resource in- ventories for monitoring changes and trends. Oregon state Univer- sity, Corvallis. Campbell, Howard W., and Steven P. Christman. 1982. Field techniques for herpetofaunal community analysis, p. 193-200; in Scott N.J.Jr.(ed.), Herpetological com- munities. United States Depart- ment of the interior. Fish Wildlife Service, Wildlife Research Report 13. 239p. Clark, Karen and David Euler. 1982. The importance of pH and habitat disturbance in amphibian distribu- tions in Central Ontario. Technical Report Wildlife Habitat compo- nent, Lakeshore Capacity Study, Ontario Ministry of Natural Re- sources, Toronto, Ontario. Clawson, Mary E. and Thomas S. Baskett. 1982. Herpetofauna of the Ashland Wildlife Area, Boone County, Missouri. Transactions of the Missouri Academy Science 16:5-16. Clawson, Mary E., Thomas S. Baskett and Michael J. Armbruster. 1984. An approach to habitat modeling for herpetofauna. Wildlife Society Bulletin 12:61-69. Collins, James P. and Henry M. Wilbur. 1979. Breeding habits and habitats of the amphibians of the Edmin S. George Reserve, Michi- gan with notes on the local distri- bution of fishes. Occasional Papers of the Museum of Zoology Uni- versity of Michigan no 686. 34 p. Cook, Francis R. 1984. Introduction aux amphibiens et reptiles du Canada. Musees Nationaux du Canada, Ottawa, 211 p. Denis Jacques, Enr. 1986. Vegetation des terres humides du lac St-Pi- erre. Cartographic 1:10 000. Min- istere du Loisir, de la Chasse et de la Peche du Quebec, Corporaton pour la mise en valeur du lac St- Pierre. Feuillet 7 a 12. Dale, J.M., B. Freedman and J. Kere- kes. 1984. Acidity and associated water chemistry of amphibian habitats in Nova Scotia. Canadian Journal of Zoology 63:97-105. Degraaf, R.M. and D.D. Rudis. 1983. Amphibians and reptiles of New England Habitats and natural his- tory. The University of Massachu- setts. 85 p. Dole, Jim W. 1965a. Spatial relations in natural populations of leopard frog, Rana pipiens Schreber, in northern Michigan. American Midland Naturalist 74:464-478. Dole, Jim W. 1965b. Sommer move- ments of adult leopard frogs, Rana pipiens Schreber, in northern Michigan. Ecology 46(3):236-255. Dole, Jim W. 1967. Spring move- ments of leopard frogs, Rana pipi- ens Schreber, in northern Michi- gan. American Midland Naturalist 78(1):167-181. Dole, Jim W. 1971. Dispersal of re- cently metamorphosed leopard frogs, Rana pipiens. Copeia 1971(2):221-228. Fox, Barry J. 1979. An objective method of measuring the vegeta- tion structure of animal habitats. Australian Wildlife Research 6:297-303. Frier, Jo Ann and Robert T. Zap- palorti. 1984. Reptile and amphib- ian management techniques. Transactions of the N.E. Section Wildlife Society 40:142-148. Frontier, Serge. 1976. Etude de la decroissance des valeurs propres dans une analyse en composantes principales: comparaison avec le modele du baton brise. Journal of experimental Marine Biology and Ecology 25:67-75. Froom, Barbara 1982. Amphibians of Canada. McClelland and Stewart, Toronto. 120 p. Gascon, Claude and Dolors Planas. 1986. Spring pond water chemis- try and the reproduction of the wood frog, Rana sylvatica. Cana- dian Journal of Zoology 64:543- 550. Gibbs, E.L., G.W. Nace, and M.B. Emmons. 1971. The life frog is al- most dead. Bioscience 21:1027- 1034. Green, Roger H. 1971. A multivariate statistical approach to the Hutch- insonian niche: Bivalve molluscs of central Canada. Ecology, 52 (4):543-556. 140 Green, Roger H. 1979. Sampling de- sign and statistical methods for environmental biologists. Wiley Inter-science Pub., Wiley and Sons, Inc. 257 p. Grier, James.W. 1984. Biology of ani- mal behavior. Times Mirror/ Mosby College Publishing, St. Louis. Hayes, Marc P. and Mark R. Jen- nings. 1986. Decline of Ranid Frog Species in Western North Amer- ica: Are Bullfrog (Ram catesbeiam) responsible? Journal of Herpetol- ogy 20(4):490-509. Hine, Ruth L., Betty L. Les and Bruce F. Hellmich. 1981. Leopard frog populations and mortality in Wis- consin 1974-76. Technical Bulletin No. 122, Department of Natural Resources, Wisconsin. 39 p. Leclair, Raymond Jr. 1983. Utilisation de differents types d'habitats par la grenouille leopard pour fin de reproduction dans la region de Bainville au lac St-Pierre. Rapport technique, Ministere du Loisirs, de la Chasse et de la Peche du Quebec. 24 p. Leclair, Raymond Jr. 1985. Les am- phibiens du Quebec: Biologie des especes et problematique de con- servation des habitats. Ministere du Loisirs, de la Chasse et de la Peche du Quebec. 119 p. Leclair, Raymond Jr. and L. Bar- ibeau. 1982. Problematique des anoures au Quebec. Ministere du Loisirs, de la Chasse et de la Peche du Quebec. 69 p. Leclair, R. Jr., and J. Castanet. 1987. A skeletochronological assessment of age and growth in the Frog Ram pipiens Schreber (Amphibia, Anura) form Southwestern Que- bec. Copeia 1987(2):361-369. Legendre, Pierre and Louis Legen- dre. 1984. Ecologie numerique (2^ ed.). Tome 1: Le traitement mul- tiple des donnees ecologiques; Tome 2: La structure des donnees ecologiques. Presses de I'Univer- site du Quebec et Masson Cie. 260; 335 p. MacArthur, Robert H., and John W. MacArthur. 1961. On bird species diversity. Ecology. 42:594-598. Marcotte, Alice. 1981. L'exploitation des grenouilles au Quebec. Rap- port technique, Ministere du Loisirs, de la Chasse et de la Peche du Quebec, Service de I'Amenage- ment de la Faune, Memoire Uni- versite du Quebec a Montreal. 75 p. Merrell, David J. 1977. Life history of the leopard frog. Ram pipiens, in Minnesota. University of Minne- sota Bell Museum Natural History Occasionnal Paper 15: 23 p. MLCP. 1985. Des habitats fauniques a conserver. CoUoque national sur la faune. Ministere du Loisirs, de la Chasse et de la Peche du Quebec. 81 p. Nie, Norman H., C. Hadlai Hull, Jean G. Jenkins, Karin Steinbrenner and Dale H. Bent. 1975. Statistical package for the social sciences. McGraw-Hill, New-York. 675 p. Paquin, Diane. 1982. L'exploitation de trois expeces d'anoures dans la region du lac St-Pierre. Rapport technique, Ministere du Loisirs, de la Chasse et de la Peche du Quebec. 64 p. Partridge, Linda. 1978. Habitat selec- tion. Pages 351-376 in J.R. Krebs and N.B. Davies. (ed.). Behavioral ecology: an evolutionary ap- proach. Blackwell Scientific Publi- cations, Oxford. Putnam, Robert W. and Albert F. Bennett. 1981. Thermal depend- ence of behavioral performance of anuran amphibians. Animal be- haviour. 29(2):502-509. Ramade, Francois. 1984. Elements d'ecologie; ecologie fondamentale. McGraw-Hill, Paris. 397 p. Rittschof, Daniel 1975. Some aspects of the natural history and ecology of the leopard frog. Ram pipiens. University of Michigan, Ann Ar- bor. Ph. D. Thesis. 212 p. Roberts, Wayne and Victor Lewin. 1979. Habitat utilization and pop- ulation densities of the amphibi- ans of northeastern Alberta. Cana- dian Field-Natur. 93(2):144-154. Sarrazin, R., M. Cantin, A. Gagnon, C. Gauthier and G. Lefebvre. 1983. La protection des habitats fauniques au quebec. Gouv. du Quebec. Dir. Gen. Faune. Min. Loisir Chasse et Peche. 256 p. Scherrer, B. 1984. Biostatistique. Gaetan Morin (ed.). Chicoutimi, Quebec. 850 p. Service Canadien De La Faune. 1980. Plan de gestion des oiseaux aqua- tiques au Canada, Environnement Canada, 26 p. Tessier, Claude and Paul Caron. 1980. Cartographie ecologique de la vegetation de la rive nord du lac Saint-Pierre. Rapport technique Service de I'amenagement et de l'exploitation de la faune, Min- istere du Loisirs, de la Chasse et de la Peche du Quebec. 19 p. + carte. Tessier, Claude, Antoine Aubin and Daniele Chenevier. 1984. Les facteurs elevation et distance dans la structure et la productivite d'une communaute riveraine de macrophytes. Canadian Journal of Botany 62:2260-2266. Vogt, Richard C. and Ruth K. Hine. 1982. Evaluation of techniques for assessment of amphibian and rep- tile populations in Wisconsin, p. 201-217; in Scott, N.J.Jr. (ed.). Her- petological communities. United States Department of the interior. Fish Wildlife Service, Wildlife Re- search Report 13. 239p. Weller, M.W. 1978. Management of freshwater marshes for wildlife, p. 267-284 in Freswater wetlands; ecological processes and manage- ment potential. New York, 365 p. Whitaker, John 0.,Jr. 1961. Habitat and food of mousetrapped young Ram pipiens and Ram clamitans. Herpetologica 17(3):173-179. 141 Appendix 1. Abbreviations for variables used in \he text, figures and tables. Abbreviations Variables Abbreviations Variables Abbreviations Variables DOPEN Distance to nearest ^ BASAREA Basal area PHonoo* Photometric index, open habitat MINAREA,, PH Minimal area between 50 et 100 DCLOSE Distance to nearest pH of soil solution cm closed habitat SAND Sand fraction in soil PHOT+* Photometric index. DMARSH Distance to marsh SILT Silt fraction in soil for herbs > 100 cm line CLAY Clay fraction in soil PHOTCAN* Photometric index DWATERP Distance to nearest permanent pool DWATER* Distance to nearest temporary pool under shrub and tree strata DHUMAIN Distance to nearest MOIST* % soil moisture HERBHGHT* Height of herb stra- human artefact TEMP* Temperature at the tum ALTREL Altitude relative to soil surface HERBCOV* % herb cover shore line WTABLE* % bare ground HGH* Cover class for high NSPHERB Number of herba- BAREGRNDj^* Water table level graminoid herbs ( > ceous species LITTER* % ground covered 100 cm tall) NSPSHRUB Number of shrub with litter HBLH* Cover class for high species MOSSCOVj^* % moss cover broad-leaf herbs NSPTREE Number of tree spe- DEADWOOD* % ground covered MGH/ Cover class for me- cies by dead wood dium graminoid TREE Cover class for tree LITTHICK* Litter thickness herbs (20 to 100 cm) stratum PHOT20* Photometric index. MBLH* Cover class for me- SHRUBHI Cover class for High shrub stratum between 0 and 20 cm dium broad-leaf herbs SHRUBLO Cover class for lov^ shrub stratum PHOT50* Photometric index, between 20 and 50 SMALLH* Cover class for herbs < 20 cm tall cm N: Variable normalized by square-root transformation. ': Variable measured monthly. 142 Appendix 2. Significance levels of Pearson rank correlations between ttie variables included in ttie nnodels and all vari- ables measured. (Significance levels 1 : P < 0.05; 2: P < 0.01 ; 3: P < 0.001 ; 4: P < 0.0001 ; +: positive; negative.) 4 Models 1 2 4 5 D D T M P s B N C P P P H L P M M M E o H I A s L H H H G I H B o A M I o L R P A o o o H T o L s R P s T T E H Y T T T T T H s s T 5 G E + 2 5 H c c H 0 R R 0 0 I A o N B c N V D K DOPEN -1 -1 -2 -1 +1 +1 DCLOSE +1 DMARSH -1 -1 DHUMAN ALTREL +3 -1 +1 NSPHERB NSPSHRUB +3 -1 -1 -1 -1 +1 +3 NSPTREE +2 -1 -1 -2 -1 -3 -1 +1 +2 +2 BASAREA +2 -2 -2 -3 -2 +1 +2 +1 MINAREA -1 -3 +2 +2 PH SAND -2 -4 -1 SILT -2 CLAY BAREGRND -2 +1 +1 LITTER -2 +1 +1 +2 -2 MOSSCOV -1 -2 -1 DEADWOOD -1 -1 +1 LITTHICK -4 +2 +3 -2 -2 -3 TREE +1 -1 -1 -1 -3 -2 +2 +2 SHRUBHI -2 -1 -1 -2 +3 +1 SHRUBLO -1 _i +2 +2 MOIST -1 +1 -1 -1 TEMP PHOT20 +1 +1 +2 +2 -1 PHOT50 +2 +1 +2 -1 PHOTIOO -1 -1 +1 -2 PHOT+ +1 +2 +1 -2 PHOTCAN -1 -2 -1 -1 -2 +1 HERBHGHT +1 +3 +1 -2 WTABLE +3 D.WATER HERBCOV +1 -2 -2 HGH -1 +1 +2 +2 +3 -1 -2 -2 HBLH -1 +1 +1 MGH +2 MBLH -1 -2 SMALLH -2 -2 D: DM mode/. 143 Habitat Correlates of Distribution of the California Red -Legged Frog (Rana aurora draytoriii) and the Foothill Yellow-Legged Frog (Rana boylii): Implications for Management^ Marc P. Hayes and Mark R. Jennings^ Abstract.— We examined features of the habitat for the California red-legged frog and foothill yellow- legged frog from the Central Valley of California. Limited overlap exists in habitat use between each frog species and introduced aquatic macrofaunal predators. Temporal data implicate aquatic preda- tors that restrict red-legged frogs to intermittent stream habitats as explaining limited overlap. Identi- fication of responsible predators is currently pre- vented because the alternative of limited overlap simply due to differential habitat use between frogs and any one putative predator cannot be rejected. Until the predators causing the negative effects are identified, efforts should be made to isolate these frogs from likely predators and minimize alteration of key features in frog habitat. Wright 1920). Despite this history of exploitation, few attempts have been made to link species-specific habitat requirements of ranid frogs to their management (but see McAuliffe 1978; Treanor 1975a, b; Treanor and Nicola 1972). Most ''management" literature has either simply reviewed the biolog}'' of selected ranid frog species or indicated vulnerable life history stages needing study (Baker 1942, Bury and Whelan 1984, Storer 1933, Willis et al. 1956, Wright 1920). In this report, we examine the habitat features of two "non-game" species, the California red-legged frog (Rana aurora draytonii) and the foothill yellow-legged frog (Rana boylii), two ranid frogs found in low- land California. Each species has dis- appeared from sizable areas of its historic range (Hayes and Jennings 1986, Sweet 1983). Although histori- cal disappearance of red-legged frogs has been linked to its exploitation as food (Jennings and Hayes 1985), causal factors in the continuing de- cline of both species remain poorly understood. Insufficient documenta- tion of the habitat requirements of each species has especially impeded identification of the causes of decline (Hayes and Jennings 1986). In this report, we reduce this gap by identi- fying the habitat requirements that characterize each frog. We then use these data to suggest the direction for management of these two species The application of habitat analysis to management has a long, complex history. The Greek philosopher Aristotle inferred that seasonal variation in the distribu- tion of certain commercially ex- ploited fishes was related to changes in their food resources and habitat temperatures (Cresswell 1862). In the 13th century, the Mongol emperor Kublai Khan encouraged the gather- ing of data on foraging patterns of sport-hunted birds to facilitate ma- nipulating their populations (Leo- pold 1931). Since these efforts, many individuals have used diverse habitat data to help understand factors that influence the distribution and success of various species. Most often, such data have been used to address com- mercially important or game species, usually to identify management al- ternatives intended to enhance exist- ing populations or avert population declines (Bailey 1984, Leopold 1933). This emphasis has resulted in most studies addressing selected birds, 'Paper presented at symposium. Man- agement of Amphibians. Reptiles and Small Mammals in North America. (Flagstaff. AZ. July 19-21. 1988.) ^Environmental Scientist, Gaby & Gaby, Inc.. 6832 SW 68th Street, Miami, FL 33 143- 3 1 15 and Department of Biology, P.O. Box 249118, University of Miami, Coral Gables. FL 33 124-91 18; Research Associate, Depart- ment of Herpetology, California Academy of Sciences, Golden Gate Park, San Fran- cisco. CA 94118-9961. fishes, and large mammals. In con- trast, species historically having lim- ited economic importance (i.e., "non- game" species) have been largely ne- glected (Bury 1975; Bury et al. 1980a, b; Pister 1976). Only over the last 15 years has an appreciation been broadly realized that non-game spe- cies are also in need of management. Non-game species are often linked to economically important ones, and as such, provide significant direct and indirect benefits to humans (Kellert 1985, Neill 1974). Although this ap- preciation has led to greater empha- sis in their study (Bury et al. 1980a, Pister 1976), a broader understand- ing of the biology of non-game spe- cies is increasingly urgent because of widespread habitat modification in- fluencing declines among ever-great- er numbers of such species (Dodd 1978, Hayes and Jennings 1986, Hine et al. 1981, Honegger 1981). Amphibians are prominent among groups of organisms given a non- game label (Bury et al. 1980a). For ranid frogs, among the most familiar of amphibian groups, non-game is really a misnomer (Brocke 1979) be- cause they have a history of human exploitation which has its roots in European and aboriginal cultural tra- ditions (Honegger 1981, Zahl 1967) and has included significant com- mercial enterprises (Abdulali 1985, Chamberlain 1898, Husain and Rah- man 1978, Jennings and Hayes 1985, 144 until experiments can identify the causes of decline. METHODS Our analysis draws upon two data sets, one addressing R. a. draytonii and the other, R. hoylii. The former is based on all known occurrences of R. a. draytonii (n = 143) from the Central Valley of California, which we define as the collective drainage area of the Kaweah, Kern, Sacramento-San Joaquin (to Carquinez Strait), and Tule River systems. We assembled these data from museum records and field notes or direct observations of the many investigators listed in the acknowledgments or whose data are cited in Childs and Howard (1955), Cowan (1979), Fitch (1949), Grinnell and Storer (1924), Grinnell et al. (1930), Hallowell (1854, 1859), Ingles (1932a, b; 1933; 1936), Storer (1925), Walker (1946), Williamson (1855), and Wright and Wright (1949). We used records not authenticated by museum specimens if they were cor- roborated by at least two sources. We then determined the subset (n = 131) of records that could be both mapped (i.e., where we could iden- tify the aquatic system likely to be Table 1 .—Habitat variables recorded for the California red-legged frog (Rana aurora draytor)ll) data set. Subset scored refers to the subset of lo- calities for which we were able to score each variable. Percent scored re- fers to the percentage of the entire data set (n = 143) for which we were able to score each variable. See text regarding further details concerning the method of data collection for each variable. Variable Subset scored % scored (n=) DefirtJtion 1. Habitat type 140 98 As (1) stream or (2) pond 2. Temporal status 137 96 As (1) perennial or (2) inter- mittent 3. Drainage area 129 90 In km^ 4. Local gradient 139 97 In angular degrees C) from horizontal 5. Water depth 74 52 As (1) presence or (2) absence of water >0.7 m deep 6. Vegetation matrix (emergent or shoreline) 44 31 As (1) dense (area >25% thickly vegetated) (2) limited (some, but <25% of area) (3) absent 7. Native fishes 56 39 As (1) present or (2) absent 8. Introduced fishes 32 22 As (1) present or (2) absent 9. Introduced bullfrogs 115 80 As (1) present or (2) absent 10. Substrate alteration 113 79 As (1) present or (2) absent n. Vegetation reduction 106 74 As (1) present or (2) absent 12. Stream order 127 89 As defined by Strahler (1957) the site of origin of the source popu- lation upon which the record was based), and identified as being from different "point" localities i>OA km apart). Although our data set was developed primarily from this sub- set, we used a few data from the re- maining 12 localities for the habitat variables described below. We used this additional data because they were either available with the origi- nal records or could be determined independent of accurate mapping. For each locality, we recorded as many of 12 habitat variables as pos- sible (table 1). For aquatic habitat type, we used the term "stream" for localities with both a well-defined drainage inflow and outflow, whereas we used "pond" for locali- ties lacking a well-defined inflow and little or no outflow. Temporal . .atus of the aquatic habitat was scored as perennial or intermittent based on 7.5'and 15' United States Geological Survey (USGS) topographic maps, but the status of some localities was modified based on field reconnais- sance or data provided by other in- vestigators. For many localities, lack of change in the temporal status of the aquatic habitat during the time R. a. draytonii was recorded was veri- fied by examining USGS topographic maps bracketing the frog record date(s). We used the designation intermittent to describe the interrup- tion of surface flow in streams or complete dry-down in ponds, either occurring at least once seasonally. Drainage area indicates the size of the hydrographic basin influencing the recorded locality. The drainage area, local gradient, and stream or- der were largely estimated from 7.5' USGS topographic maps. We esti- mated large drainage areas (>130 km^) by extrapolation to the recorded locality on topographic maps from either the drainage area for the near- est upstream gauging station (United States Geological Survey 1970a, b) or section counts on United States For- est Service and county maps. Local gradient was estimated from map 145 distances of 0.5-1.0 km across the re- corded locality except in the few cases where pronounced local relief required reduction of this distance for an accurate estimate. Data for the remaining variables (water depth, vegetation matrix, na- tive and introduced fishes, intro- duced bullfrogs [Ram catesbeiam], substrate alteration, and vegetation reduction) were obtained for subsets of the larger data set from the sources indicated earlier supple- mented by Leidy (1984), Moyle and Nichols (1973), Moyle et al. (1982), and Rutter (1908). The exact values used to partition water depth and vegetation matrix variables are arbi- trary. However, we chose their gen- eral dimensions with the intent of identifying whether the habitat re- quirements of red-legged frogs sug- gested by anecdotal data (moderately deep water associated with dense vegetation; see Hayes and Jennings 1986) were supported by this data set. Variation in the collective data set required scoring the fish and in- troduced bullfrog data as presence/ absence, but we also used available data on which fish species were pres- ent to interpret the habitat require- ments of red-legged frogs. Substrate alteration and vegetation reduction variables indicate alteration of aquatic habitats that was, directly or indirectly, human-effected. We scored substrate alteration as present if evidence existed that the shoreline or substrate topography of the aquatic habitat had been markedly altered (e.g., dams, rip-rap, bank- trampling by cattle). Marked altera- tion meant that at least 25% of the area of substrate of a locahty ap- peared altered. We scored vegetation as being reduced when data indi- cated that at least 25% of pre-existing shoreline or emergent vegetation had been removed. We also gathered current data on a subset of the described localities through field reconnaissance and some information provided by others (data gathered during the interval 1980-1987 represented ''current'' data). We used these data to help identify temporal changes that may have occurred at sites or in drainage systems for which we had historical data. For this analysis, we used "drainage system" to mean only the primary and highest-order (fide Strahler 1957) secondary tributaries of the Sacramento-San Joaquin drain- age system. These data were particu- larly important for indicating where red-legged frogs were probably ex- tinct. The data set addressing R. hoylii consists of data published by Moyle (1973) and Moyle and Nichols (1973) from which we re-examined selected elements. Collection methods for these data are thoroughly described therein. Our reanalysis used most of the variables described by Moyle (1973) with some modifications. We used the original estimates of the numbers of each fish species rather than the coded values; the numbers of yellow-legged frogs and bullfrogs remained coded because the original data were recorded as coded. Moyle' s stream type variable was reduced to two categories by com- bining his three intermittent and three perennial stream categories. We also added two variables, one which combines Moyle's cobble and boulder/bedrock substrate catego- ries. The other describes the stream morphology category designated in Moyle's original data as smooth wa- ter and fits the definition of a run (Armour et al. 1983). For correlations between yellow-legged frogs and other species, we used only the sub- set of localities where either or both of yellow-legged frogs and the spe- cies being compared was present. We re-examined these data for four reasons. First, Moyle (1973) summarized data from only some of the sites where yellow-legged frogs were not found. We were equally interested in habitat variation among all sites sampled where yellow- legged frogs had not been found as well as sites where they were found. Second, Moyle (1973) found that the collective abundance of all fish spe- cies was inversely correlated with that of yellow-legged frogs, but also commented that yellow-legged frogs were most abundant where native fishes were present. Because original estimates of the numbers of each fish species were available and an inverse relationship between the abundance of native frogs and introduced fishes had already been identified (Hayes and Jennings 1986), we were espe- cially interested in relationships be- tween the abundance of specific na- tive and introduced fishes and that of yellow-legged frogs. Third, Moyle (1973) coded fish abundance when the data, as originally recorded, per- mit at least ranking, so, where pos- sible, we analyzed the original data directly to minimize bias that can re- sult from coding (Sokal and Rohlf 1981). Lastly, the fish abundance data displayed skewed distributions for several species, so we used non-par- ametric analyses to avoid having to make any assumptions about sample distributions. Statistical treatments used are de- scribed in Sokal and Rohlf (1981) and Zar (1974). All contingency table comparisons performed had one de- gree of freedom (df), so all Chi- square values were calculated with the correction for continuity (X^^). For those analyses that required more than one comparison using some of the data, alpha (a) was evaluated based on the number of comparisons to a level equivalent to 0.05 using Si- dak's multiplicative inequality (Sokal and Rohlf 1981). RESULTS California Red-Legged Frog Ram aurora draytonii was recorded primarily from aquatic habitats that were intermittent streams which in- cluded some area with water at least 0.7 meters deep, had a largely intact emergent or shoreline vegetation. 146 Table S.—Frequency of fish species co-occurrence with Rana aurora dray- tonii. Percentage Is the number of sites respective fish species were re- corded as a function of all sites where fishes were recorded as co-occur- ring with R. a. drayfonii. An asterisk (*) indicates introduced species. Co-occurrence Percentage Species (n => <%) California roach (Lavinia symmefricus) 19 47 Mosquitofish (Gambusia affinis)* 10 25 Hitch (Lavinia exilicauda) 6 15 Green sunfish (Lepomis cyanellus)* 6 15 Threespine stickleback (Gasferosfeus aculeafus) 3 8 Sacramento squowfish (Piychocheiius grandis) 2 5 Sacramento sucker (Cafosfomus occidenfalis) 2 5 Prickly sculpin (Coffus asper) 1 3 Hardhead (Mylopliarodon conoceplialus) 1 3 Rainbow trout (Salmo gairdnerii) 1 3 Brown trout (Salmo fruffa)* 1 3 and lacked introduced bullfrogs (table 2). We found descriptions ade- quate to characterize vegetation for 77% (33) of sites where the emergent or shoreline vegetation variable could be scored. With three excep- tions, descriptions indicated that ei- ther, or both of, an emergent vegeta- tion of cattails {Typha spp.) or tules {Scirpus spp.), or a shoreline vegeta- tion of willows (Salix spp.) were present. Shrubby willows were re- corded at 67% (22) of the sites with vegetative descriptions, and were identified as arroyo willow (Salix la- siolepis) in the eight instances where a species name was provided. Only juvenile frogs were recorded at five of the six sites where a limited emer- gent vegetation was present and at the only site that lacked a water depth greater than 0.7 m. We found no significant difference in the num- bers of intermittent versus perennial sites with red-legged frogs that had a dense vegetation and a water depth of >0.7 m (X2^ = 0.338, p = 0.561, for vegetation; X^^ = 0.017, p = 0.897, for water depth; X^^^^ ^^^^s = 5-024 for both). Rana aurora draytonii was also more frequently recorded at sites with native fishes and with substrate alteration, but less frequently re- corded at sites with introduced fishes. Fishes were present at 69% (40 of 58) of sites where data as to their occurrence were recorded; 26 sites had only native fishes, seven had only introduced fishes, and seven had both. Only four fish species, California roach (Lavinia symmet- ricus), hitch (Lavinia exilicauda), green sunfish (Lepomis cyanellus), and mosquitofish (Gambusia affinis), were recorded as co-occurring with R. a. draytonii at more than three sites (table 3), and only California roach was recorded at more than 25% (10) of sites. Sixty of the 70 sites described as being substrate-altered at the time R. a. draytonii was recorded were small impoundments. California red-legged frogs were also most frequently recorded at sites influenced by a small drainage area, having a low local gradient, and in streams having a low stream order. Drainage areas of sites from which R. a. draytonii was recorded vary from 0.02 km2 to over 9000 km^, but two- Table 2.— Variation among habitat variables for California red-legged frogs (Rana aurora draytonii). Number of localities (percentages of localities) in each category are indicated. See table 1 and text for explanation of vari- able categories. Variable Variable categories 1. Aquatic habitat type (a) stream 129 (92%) (b) pond 10 (8%) 2. Temporal status of (a) perennial 49 (36%) aquatic site (b) intermittent 88 (64%) 3. Water depth (a) > 0.7 meters 73 (99%) (b) < 0.7 meters 1 (1%) 4. Emergent and (a) absent 0 (0%) shoreline vegetation (b) limited 9 (20%) (c) dense 35 (80%) 5. Native fishes (a) present 33 (65%) (b) absent 18 (35%) 6. Introduced fishes (a) present 14 (44%) (b) absent 18 (56%) 7. Introduced bullfrogs (a) present 13 (1 1%) (b) absent 102 (89%) 8. Significant substrate (a) present 70 (62%) alteration (b) absent 43 (38%) 9. Significant removal (a) present 1 (2%) vegetation (see #4) (b) absent 44 (98%) 10. Current status (a) probably extant 86 (72%) (among localities) (b) probably extinct 34 (28%) 11. Current status (a) probably extant 18 (42%) (among drainages) (b) probably extinct 25 (58%) 147 thirds (n = 83) are from localities with drainage areas <40 km^ (fig. 1). Local gradient (slope) at California red-legged frog localities varies from 0.04° to 12.8° from horizontal, al- though 87% (n = 100) occur at sites with slopes <2°. California red- legged frogs have been recorded in 1st to 6th order streams, but 94% (n = 119) of these localities are 4th- or lesser-order streams and 42% are 1st- order streams (fig. 2). Based on the subset for which cur- rent data were available (n = 120), California red-legged frogs are probably extinct at >25% of the lo- calities where they were historically recorded. When clustered into a sample representing drainage sys- tems (n = 43; see methods), this sub- set indicates that California red- legged frogs are probably extinct in over 50% of the drainage systems in the Central Valley area. Three habitat variables (temporal status of aquatic habitat, drainage area, and intro- duced bullfrogs) showed a signifi- cant relationship to the probability of survival of local populations of C!ali- fornia red-legged frogs (table 4). We found that R. a. draytonii is likely ex- tant at 82% (n = 70) of localities with an intermittent aquatic habitat, whereas it is probably extinct at 71% (n = 22) of the sites with a perennial aquatic habitat. Grouping localities based on drainage area, R. a. dray- tonii is probably extant at 83% (n = >«01 1 3 3001-4000 1 1 MOt-3000 |2 ATM ISOI-MOO 2 (H km) lOOl-ISOO t SOI-tOOO 1 iKillUMbc) Figure 1 .—Frequency distribution of locali- ties wtiere Rana aurora draytonii \nas been recorded in thie Central Valley, California based on drainage area. Thie inset details \he frequency distribution of localities witti drainage areas < 280 kpn*. 7A\-2tO 201-240 12 161-200 121-160 81-120 13 41-60 0-40 I I I I I 0 20 40 60 60 100 ■ \ Table 4.— Contingency analysis relating selected habitat variables to an estimate of the lilceiihood that historically recorded California red-legged frog populations are extant. Status of frog populations at recorded locali- ties are indicated as extant (= probably extant) and extinct <= probably extinct). A double asterisk (") denotes significant contingency tables, based a critical ^^^^y^^^^^Q^^j = 7.3, a adjusted for seven comparisons (see methods). Locality Status Variable Condition extcnt extinct x% Probability 1. Temporal status Perennial 9 22 27.326 o.ooor* Intermittent 70 15 2. Drainage area >300 km2 0 11 31.466 o.ooor* <300 km^ 85 18 3. Native fishes ■f 13 6 0.276 0.5991 14 11 4 Introduced bullfrogs + 0 10 27.140 0.0001" 70 16 5. Substrate alteration^ 25 14 0.983 0.3215 47 14 6. Introduced fishes + 5 9 0.003 0.9524 7 10 7. Substrate alteration'' -1- 21 3 <0.001 0.9944 26 5 "Analysis with all localities. ^Analysis with subset of localities having a drainage area <25 km'. \ affected by the largest drainage areas (n = 10). Similarly, R. a. draytonii is probably extant at 81% (n = 70) of localities lacking introduced bull- frogs and is probably extinct at all localities (n = 10) where it has been recorded with bullfrogs. Remaining variables either failed to show a sig- nificant relationship to the probabil- ity of California red-legged frog sur- vival (table 4), or one of the variable categories was so rare that this analy- sis was not applicable (see table 2). Foothill Yellow- Legged Frog Rana boylii was recorded primarily from shallow, partly shaded stream sites with riffles and at least a cobble- sized substrate. All 29 stream sites at which either post-metamorphic or larval R. boylii were recorded were <0.6 m in average water depth (fig. 3) and had at least some shading (fig. 4). Rana boylii was recorded more 85) of sites influenced by a small (<300 km^) drainage area, whereas it is probably extinct at all recorded localities (n = 11) influenced by a large (>300 km^) drainage area. Moreover, available data indicate that R. a. draytonii is extinct at all re- corded localities on the Central Val- ley floor, which includes all localities 0 10 » M 4Q SC M Figure 2.— Frequency distribution of locali- ties wt^ere Rana aurora draytonii has been recorded in the Central Valley, California based on stream order. 148 frequently at sites with a stream area that was >20% shaded than at sites with >20% shading. Only one of 29 R. hoylii sites lacked riffle habitat and R. boylii was recorded significantly more frequently at sites with >40% riffle area than at sites with a riffle area of <40% [X^ = 8.680, p = 0.003, X'df=,a(2)=o.o25 = 5.024; fig. 5]. Only four of 29 R. boylii sites lacked at least a cobble-sized substrate and R. boylii was recorded most frequently (20 of 29) at sites with >40% of the sub- strate that was at least cobble-sized (fig. 6). Few other patterns could be identified from among the environ- mental variables that we re-analyzed. Rana boylii was recorded more fre- quently from perennial streams (n = 19) than from intermittent ones (n = 10), but the difference was not sig- nificant when compared to the total number of perennial (n = 71) and intermittent (n = 59) stream sites sampled [X^'^ = 1.268, p = 0.260, X'df=i,a(2)=o.o25 = 5-024]. Of 13 environ- mental variables that we re-exam- ined, only the percentage of stream area in riffles was significantly corre- lated with the abundance of R. boylii (table 5). Rana boylii occurred with 1-5 Cx = 2.5) of the vertebrate members of the aquatic macrofauna at 26 of the 29 localities where it was recorded. Figure 3.— Histogram of the proportion of sites in stream depthi categories whiere Rana boylii hias been recorded in thie Sierra Nevada foothiiils, California. Sample sizes as a function of thie total sample in eachi stream deptti category are: <0.20 (n=8/24), 0.21=0.40 (n=9/43), 0.41-0.60 (n=12/57), and >0.60 (n=0/18). 0 I-» Jl-« . Foothill yellow-legged frogs were recorded as occurring with 12 differ- ent species, but co-occurrence, ex- pressed as the percentage of total sites at which either R. boylii or the co-occurring species were recorded, did not exceed 31% (table 6). Intro- duced species (n = 6) occurred with R. boylii less frequently Tx = 2, 1-3) than native species \x = 9.3, 1-17) and native species had a significantly higher percentage of co-occurrence (3-31%, X = 16.5%) than introduced species [n = 6; 2-9%, x = 3.7%; Mann- Whitney test, U' = 32.5, p = 0.0275, U criHcala(2)=0.05 = 31]. Only four native 0 1-20 21-« 4l-«0 6l-« SI-100 PrcirUgi if SkiM Slmm Vm Figure 4.— Histogram of the proportion of sites in stream shading categories where Rana boylii has been recorded in the Sierra Nevada foothills, California. Sample sizes as a function of the total sample In each sh'eam shading category are: 0% (n=0/5), 1 -207o (n=3/37>, 2 1 -40% (n=7/38), 41 -607* (n=8/30), 61 -807. (n=9/23), and 81 -1007, (n=2/8). 0 1-20 2l-« 41-M i\-m 81-100 Figure 6.— Histogram of the proportion of sites in substrate categories where Rana boylii has been recorded In the Sierra Ne- vada foothills, California. Sample sizes as a function of the total sample in each sub- strate category are: 07, (n=4/19), 1-207, (n=3/32), 21 -407, (n=2/23), 41 -607, (n=7/29), 61-807, (n=9/26). and 81-1007, (n=4/12). fishes, California roach, Sacramento sucker (Catostomus occidentalis), Sac- ramento squawfish (Ptychocheilus grandis), and rainbow trout (Salmo gairdnerii), occurred with R. boylii at more than three of the 29 sites where the latter was recorded, and of these, only California roach occurred with R. boylii at more than 50% of the sites where R. boylii was recorded. Only one species assemblage, that consist- ing of California roach, Sacramento squawfish, and Sacramento sucker, occurred with R. boylii more often than expected by chance alone (table 7). Correlation analysis indicated that the abundance of 10 of the 12 co-oc- curring species was significantly in- versely correlated with the abun- dance of R. boylii (table 8). DISCUSSION Habitat Variation California Red-Legged Frog A dense vegetation close to water level and shading water of moderate depth are habitat features that ap- pear especially important to Califor- nia red-legged frogs. Previous au- thors have suggested or implied the occurrence of at least one of these habitat features. Storer (1925) noted 149 that R. a. draytonii in streams was re- stricted to large pools, which implies a moderate water depth. Stebbins (1966, 1985) emphasized vegetative cover as important to red-legged frogs, but his comments confound habitat characteristics that may be attributable to northern versus Cali- fornia (southern) red-legged frogs; data on these two forms should re- main partitioned until it is well-es- tablished that they are not different species (Hayes and Miyamoto 1984, Hayes and Krempels 1986). Zweifel (1955) coupled the water depth and vegetation features of California red- legged frog habitat, but he empha- sizes a herbaceous shoreline vegeta- tion. Chir data indicate that a more complex vegetation is a feature of sites where R. a. draytonii occurs. Cattails, bulrushes, and shrubby wil- /fdble 5. —Spearman rankcorrelatlorr between selected environmental variables and the coded abun- dance of R. boylil as measured by Moyie (1973). Sample size for each variable is n = 1 30. A double asterisk (**) Indicates significant correlations; based on a critical r, = 0,267 at an aCtwo -tailed) = 0.002, adjusted for 24 comparisons (13 below and 1 1 In | table 8; see methods). Variable Correlation coefficient ecles (n=) (r,=; (p=) Cafosfomus occidenfalis 71 -0.404" <0.001 -0.363 Gambusia affinis* 62 -0.835" <0.001 -0.388 Icfalurus cafus* 41 -0.798** <0.001 -0.473 Lovinia eydlicauda 40 -0.760" <0.001 -0.479 Lavinla symm e trie us 55 -0.316 0.020 -0.411 Lepomis cyonellus* 88 -0.742" <0.001 -0.327 Lepomis macrochirus' 59 -0.827" <0,001 -0.397 Micropferus dolomfeur 35 -0.538" 0.001 -0.510 Mylopharodon conocephalus 38 -0.607" <0.001 -0.491 PlychocheHus grandis 66 -0.54r* <0.001 -0.376 Rana cafesbeiana* 90 -0.800" <0.001 -0.323 Salmo gairdneriJ 44 -0.425 0.005 -0.458 1976). Even the only two exceptions to R. a. draytonii not occurring in ver- nal pools support this hypothesis. A large vernal pool in San (Dbispo County, California is known to have a population of California red-legged frogs (D. C. Holland, pers. comm.). However, this vernal pool is atypical because it possesses significant mac- rovegetation and water depth. These features appear to be present because this large (ca. 20 ha) pool does not dry down each year. The second ex- ception is a vernal pool in coastal southern California in which two frogs with abnormal numbers of legs were found (Cunningham 1955). Cunningham thought that the defects were induced by exposure to high temperatures during early develop- ment, a condition facilitated by the limited vegetative cover that was present. His speculation may be valid if California red-legged frog embryos have a low critical thermal maximum (Hayes and Jennings 1986). Storer (1925) thought that R. a. draytonii was excluded from tempo- rary (vernal) pools because its larval period is relatively long, but the more likely mechanism is that frogs immigrating to such pools were un- able to establish because suitable habitat was lacking. The latter hy- pothesis is supported because Cali- fornia red-legged frogs are not re- corded from the many vernal pools that hold water for intervals longer than the minimum time required by R. a. draytonii to complete metamor- phosis (10 weeks; Hayes, unpubl. data; see also Jain 1976, Zedler 1987). Rana a. draytonii also appears to have responded to the creation of habitat with the appropriate vegeta- tion and water depth characteristics. A significant aspect of the changes in aquatic habitats that have occurred in the Central Valley below 1500 m is an increase in the number of perma- nent ponds (Moyle 1973). Storer (1925) reported that R. a. draytonii occurred in a number of water stor- age reservoirs and artificial ponds, but the habitat features of those sites 151 were not described. Thus, it was of special interest to find that no signifi- cant difference could be identified between the probability of extinction of R. a. draytonii at substrate-altered sites (mostly small impoundments) and at sites lacking such alteration. Moyle (1973) concluded that the de- cline of R. a. draytonii was related in part to human-induced alteration, including creation of impoundments. Our data suggest that human-in- duced alteration creating small im- poundments cannot be related di- rectly to the disappearance of Cali- fornia red-legged frogs. We empha- size that these data do not exclude the alternative, discussed later, which indicates that the creation of small impoundments is likely to have an indirect negative effect on R. a. draytonii by facilitating the dispersal of introduced aquatic predators. Besides features of habitat struc- ture associated with R. a. draytonii, its isolation from one or more aquatic macrofaunal predators is the other key element suggested by these data. No significant variation was found in the features of habitat structure im- portant to R. a. draytonii between intermittent and perennial aquatic sites, so differences in habitat struc- ture cannot explain why R. a. dray- tonii is recorded most frequently from intermittent aquatic sites. We believe that California red-legged frogs were recorded most frequently from intermittent sites because the likelihood of extinction at perennial sites is now higher than at intermit- tent sites (see table 4) and few his- torical data are available from when frogs were often found at perennial sites. California red-legged frogs are now extinct from all sites on the Cen- tral Valley floor, all of which were perennial and, except for one, were recorded prior to 1950. We believe that the disadvantage associated with perennial sites and the advan- tage associated with intermittent sites is the degree to which the for- mer allow, and the latter restrict, the access of aquatic macrofaunal preda- tors. The remaining variation in fea- tures of R. a. draytonii habitat we have identified can be directly, or indirectly, linked to a hypothesis in- voking the influence of one or more aquatic macrofaunal predators. The significantly lower likelihood of ex- tinction at sites with small drainage areas (table 4) and R. a. draytonii being recorded from a greater num- ber of localities with smaller drain- age areas (fig. 1) and lower stream orders (fig. 2), are probably unrelated to either drainage area or stream or- der effects per se. Rather, they are a function of both the bias against re- cording historical data and the fact that sites with smaller drainages or lower stream orders have a higher probability of being intermittent aquatic habitats, which have a higher probability of excluding aquatic predators. Limited co-occurrence with aquatic predators, namely bull- frogs and predatory fishes, and a sig- nificantly higher likelihood of extinc- tion at sites where bullfrogs were re- corded (table 4) may indicate a nega- tive interaction with one or more of these species. Rana a. draytonii did not co-occur with any fish species frequently. It co-occurred most often with California roach, a small, om- nivorous native fish that is thought to have declined, in part, due to pre- dation by introduced fishes (Moyle and Nichols 1974, Moyle 1976). We did not detect a significantly higher likelihood of extinction at sites with introduced fishes. However, the sample was too small to partition to permit testing individual fish species, the level at which we believe such an effect is most likely. While we are reasonably con- vinced that the greater restriction of R. a. draytonii to intermittent aquatic habitats is an effect due to novel aquatic predators, we emphasize that these data cannot identify which are the aquatic predators producing such an effect. The inability to identify the responsible predators is complicated by the condition of limited overlap between each potential predator and R. a. draytonii. That condition pre- vents excluding the alternative that different habitat requirements rather than any predatory interaction may explain the limited overlap in habitat use between each putative predator and California red-legged frogs (compare Moyle 1973 for bullfrogs and Moyle and Nichols (1973) for various fishes, but especially mosqui- tofish and green sunfish; see also Hayes and Jennings 1986 for a dis- cussion). It is this fact and the appar- ent intolerance of R. a. draytonii to unshaded habitat that leads us to suggest that some alteration of ripar- ian vegetation may be necessary to create the conditions for a negative interaction. Foothill Yellow-Legged Frog Partly shaded, shallow streams and riffles with a rocky substrate that is at least cobble-sized are the habitat features that appear to be important to foothill yellow-legged frogs. Previ- ous authors agree that R. hoylii oc- curs in streams (Moyle 1973; Stebbins 1966, 1985; Storer 1925; Zweifel 1955), but variation exists in the fea- tures of streams associated with these frogs. Of environmental vari- ables that appear important to R. boy Hi, the percentage of stream area in riffles is the only one we were able to correlate significantly, albeit weakly, with its abundance. Moyle (1973) obtained a similar positive correlation in his original analysis of the same data, and Stebbins (1966, 1985) also emphasized riffles as one of the key aspects of R. hoylii habitat. The reason for the weak correlation we found is uncertain, but one or more of three factors probably pro- duced that result. First, as intermit- tent streams lose surface flow during late summer, riffles disappear, and R. hoylii can then be found associated with stream pools (Fitch 1938, Slevin 1928, Storer 1925, Zweifel 1955). 152 Moyle's data were collected in late summer and 10 of the 29 stream sites at which R. boylii was recorded were intermittent, so data from these sites may have diluted the correlation. Second, riffle area may be correlated with the abundance of R. boylii only above or below certain values (see fig. 5). Lastly, R. boylii has been re- ported from sites with little or no riffle habitat unrelated to seasonal patterns (Fitch 1938, Zweifel 1955). Apart from riffles, our reanalysis of environmental variables differs from that of Moyle (1973), who found that five of the other variables that we re-examined were either positively (i.e., shading and boulder/ bedrock; compare table 1 in Moyle [1973] and our table 5) or negatively (i.e., rooted vegetation [= our aquatic vegetation], pools, man modified [= our human alteration]) significantly correlated with the abundance of jR. boylii. We attribute this difference, in part, to our analysis being more con- servative because we adjusted a for the experimentwise error rate, our analysis was not restricted to locali- ties where only frogs were found, and we used non-parametric tests. Some of the correlations that Moyle (1973) observed with R. boylii abun- dance may have been significant due to one or more of these differences. We must emphasize, however, that several of the variables that Moyle found correlated with R. boylii abun- dance vary differentially in their oc- currence between riffles and pools (e.g., boulder /bedrock; see Moyle [1973] and Moyle and Nichols [1973]). Those variables are also sus- ceptible to the seasonal correlation- altering effects discussed for the riffle variable. Thus, a conservative analy- sis, like ours, is less likely to detect variables related to frog abundance within such a data set. Nevertheless, variables identified as important to R. boylii need not be correlated to its abundance. Stream depth, shading, and substrate type may represent such variables. Out reanalysis of Moyle's data suggests that sites with a shallow average stream depth are somehow advanta- geous (see fig. 3). Moyle (1973) found no significant correlation between the abundance of R. boylii and stream depth, and he did not discuss stream depth with respect to foothill yellow- legged frogs in any other context. Zweifel (1955) noted that streams in which R. boylii occurred were seldom more than 0.3 m deep, and Fitch (1936), Storer (1925), and Wright and Wright (1949) found that R. boylii usually lays eggs in shallow water. Still, overall importance of stream depth to R. boylii remains unclear. Our reanalysis also suggests that some advantage is linked to in- creased shade up to some intermedi- ate level (see fig. 4). Zweifel (1955) described shading in typical R. boylii habitat as interrupted, whereas Moyle (1973) reported a positive cor- relation between frog abundance and the degree of shading. Some workers have emphasized the degree of openness or insolation in R. boylii habitat, rather than ad- dressing shading (Fitch 1938; Steb- bins 1966, 1985). Nevertheless, even the latter imply that some shading is present. Fitch's (1938) suggestion that yellow-legged frogs are excluded by dense canopy may be supported by Moyle's data because he recorded no R. boylii at sites with >90% shading (see also fig. 4). Our reanalysis also suggests that some advantage is as- sociated with sites possessing at least a cobble-sized substrate (see fig. 6). Although workers have most fre- quently emphasized the rocky aspect of R. boylii habitat (Fitch 1936, 1938; Moyle 1973; Stebbins 1966, 1985; Storer 1925), substrate descriptions of that habitat are probably as varied as any other single variable. Moyle (1973) identified a positive correla- tion between the percentage of stream area with bedrock and boul- ders and the abundance of R. boylii, yet sites with gravely (Gordon 1939), sandy (Zweifel 1955), or muddy sub- strates have also been recorded (Fitch 1938, Storer 1925). Because Moyle's data do not provide frog age, we could not determine whether sites having a substrate that was less than cobble-sized were simply mar- ginal habitat with juvenile R. boylii (see Zweifel 1955), or whether they represented real variation in habitat used by established populations. Fitch (1938) and Zweifel (1955) re- ported on a few sites with adult frogs that lacked a substrate that was cobble-sized or larger and appeared to have few predators. They sug- gested that yellow-legged frogs are rarely recorded from such sites be- cause their predators may access the "atypical" habitat more easily. Nev- ertheless, data on the aforementioned variables reinforce the conclusion al- ready arrived at with R. a. draytonii: Existing data cannot distinguish hy- potheses explaining the differential occurrence of R. boylii among habitat categories due to mechanistic or physiological restriction (i.e., "habi- tat preference") from hypotheses in- voking habitat restriction because of some novel predator (Hayes and Jen- nings 1986). The data for R. boylii dif- fer from that of R. a. draytonii in that we cannot confidently reject the al- ternative that no restriction is occur- ring. For example, it remains unclear whether earlier reports of "atypical" habitat use by R. boylii were simply rare occurrences, or whether those instances actually reflect a general pattern of broader habitat use in years prior to when Moyle (1973) ob- tained his data, indicating that habi- tat restriction had occurred. Management Implications Both R. a. draytonii and R. boylii need immediate management considera- tion if many remaining populations are to survive into the next century. Rana a. draytonii is extinct on the floor of the Central Valley, and is probably extinct from over half of the drainage systems in the Central Val- ley from where it was historically re- corded. We consider many of the 153 remaining populations at risk since over half of the localities are within areas projected to be flooded by res- ervoirs proposed for the Coast Range slope of the Central Valley (Wemette et al. 1980; C. J. Brown, Jr., pers. comm.). Populations at an additional 10 localities are at an unknown, but probably high level of risk. Although these additional localities will not be flooded by the proposed reservoirs, flooding will isolate the frogs present in small (<10 km^) drainage basins upstream of the reservoirs. We lack data on how isolation in very small drainage basins may increase the probability of extinction (see Fritz 1979), but the only four localities iso- lated by reservoirs for which data exist now lack red-legged frogs (Hayes, unpubl. data). California red-legged frogs were recorded at each of the latter sites up to 20 years ago, between one and five years after flooding of the adjacent reservoir had taken place. Comparable data on the decline of R. boylii in the Central Valley are lacking, but observations by experienced workers indicate that R. boylii no longer occurs at many localities in the Central Valley drain- age basin where it was historically recorded (Moyle 1973; R. Hansen, D. Holland, S. Sweet, D. Wake, pers. comm.; Jennings, unpubl. data). Modal habitat requirements for both frog species suggested by exist- ing data should be given special at- tention in any management attempt. Since our comments here are based on data for both species in the Cen- tral Valley of California, attempts to apply the management recommenda- tions we make to other areas within the geographic range of each species should be done cautiously. We can- not overemphasize that preservation of what appears to be the preferred (modal) habitat condition for either species should be stressed where it is ambiguous whether restriction is due either to the negative impact of the introduced aquatic macrofauna, or to intrinsic mechanical or physiological limitations. Preservation of non-mo- dal habitat is not only likely to incur a greater cost to ensure frog survival, but more importantly, it may still not allow survival if the worst-case sce- nario (restriction of habitat by the introduced aquatic macrofauna) is true. The modal habitat features of R. a. draytonii and R. boylii are similar in two ways. First, the aquatic habitat of each has some shading. Yet, shad- ing associated with California red- legged frogs differs because of the apparently crucial aspect of having dense vegetation at or near water level. We lack details on just how the streams Moyle (1973) sampled were shaded, but knowledge of some of the species providing shade suggests that a higher overstory was typical. Rana a. draytonii will always be at greater risk than R. boylii where al- teration of riparian vegetation is a problem simply because of its shade requirement; even altered stream en- vironments may retain some shad- ing, but a lesser probability will al- ways exist that the shading that re- mains will have the structure needed by R. a. draytonii. Second, each spe- cies occurs most frequently in the ab- sence of any aquatic macrofauna, and both species have probably expe- rienced some habitat restriction due to introduced aquatic predators. Only one small native minnow co- occurs at over one-third the sites where each frog species was re- corded, and even that species was not positively correlated with frog abundance. For R. a. draytonii, the data are reasonably convincing that restriction has occurred away from perennial aquatic sites. For R. boylii, data do not clearly indicate habitat restriction. Still, the fact that R. boylii was found at fewer intermittent sites leads us to believe that if habitat re- striction has taken place, it has oc- curred away from intermittent aquatic sites. We reason that since riffles disappear seasonally in inter- mittent streams, such streams lack the condition found in perennial streams that may be an advantage if riffle habitat is a refuge, i.e., that per- ennial streams have riffle habitat year-round. CXir analysis indicates that at- tempts at management of these two frogs should address at least three other habitat variables: water depth, stream morphology, and substrate type. Rana boylii appears to require a shallow water depth of <0.6 m, whereas R. a. draytonii seems to re- quire some water _0.7 m deep. Data on stream morphology and substrate type, which were recorded only for R. boylii, suggest that both of a per- centage of riffle area and at least cobble-sized substrate of greater than 40% best suit this species. Parallel data for R. a. draytonii are lacking, but since data on other habitat para- meters measured for R. a. draytonii are largely "reciprocals" of the corre- lates of riffle habitat associated with R. boylii, we anticipate that some re- lationship to the more lentic water stream morphology categories (i.e., pools and runs) and their associated finer substrate categories (i.e., silt and sand) will be demonstrated for R. a. draytonii. Experiments may ultimately iden- tify the introduced aquatic predators likely responsible for the declines of these frogs, but management based on current knowledge should ad- dress no less than the worst-case sce- nario; i.e., that any member of the introduced aquatic macrofauna pres- ents a risk to the survival of popula- tions of R. a. draytonii and R. boylii. Thus, the sound management deci- sion is to implement measures that will maximize the degree of isolation between existing populations of each frog species and any members of the introduced aquatic macrofauna. Just how isolation should be maintained will vary depending on the site con- sidered, but some general sugges- tions can be made. First, passive measures promoting isolation are preferable because they are less costly and are less likely to affect non-target species. Simply avoiding habitat modification where the mo- 154 dal habitat features for each frog spe- cies already exist is a passive meas- ure that will provide some degree of within-habitat isolation since mem- bers of the introduced aquatic macrofauna show little overlap in their habitat requirements with each frog. Yet, populations of either frog species currently coexisting in a habi- tat mosaic with members of the in- troduced aquatic macrofauna may still be doomed. This possibility leads us to suggest that most efforts at management should be spent on frog populations at sites that cur- rently lack introduced aquatic preda- tors. We consider protection of the entire hydrographic basins of drain- age systems tributaries (see methods for definition) an important part of such management attempts because intrusion by introduced aquatic predators is probably most easily controlled if the only natural access route is via upstream movement. To our knowledge, no locality within the Central Valley drainage area having an extant California red-legged frog population has its entire hydro- graphic basin protected. Moreover, only two California red-legged frog populations within this area occur at sites where the habitat is currently offered some protection. Second, iso- lation strategies may differ depend- ing on whether proximate popula- tions of introduced aquatic predators are bullfrogs or fishes or both. Apart from being physically transported, fishes are effectively prevented from moving upstream by a barrier (see Hayes and Jennings 1986), whereas bullfrogs, capable of overland move- ment under wet conditions (Hayes and Warner 1985), are less likely to be barrier-limited. We indicated ear- lier that creation of small impound- ments may enhance the ability of R. a. draytonii to establish at certain sites through the creation of features found in its habitat, but attention to the positioning of such impound- ments is an equally important con- siderahon. If impoundments are close enough that bullfrogs reach them from an adjacent source popu- lation, such sites can also act as local refuges at which new bullfrog popu- lations can become established, and can serve as new focal points from which to disp)erse. Moreover, new impoundments probably favor the establishment of bullfrogs simply be- cause their unvegetated condition more closely matches the habitat re- corded for bullfrogs (Moyle 1973). These arguments simply indicate that particular attention should be given to avoiding the creation of "step- ping-stone" pathways, i.e., provision of access into currently isolated drainages by the positioning of im- poundments that permit introduced predators, like bullfrogs, to encroach progressively by dispersal. The limits of our analysis indicate that significant aspects of habitat variation for both frog species re- main to be understood. In particular, an understanding is needed as to how key variables influence repro- duction and refuge sites. Although available data on oviposition pat- terns suggest a link between R. a. draytonii and the presence of emer- gent vegetation (Hayes and Miyamoto 1984), and R. boylii and a rocky substrate (Fitch 1936, 1938; Storer 1925; Zweifel 1955), it is un- clear for either species to what de- gree the substrate can vary before oviposition may be prevented and also how aspects of reproduction be- sides oviposition may be linked to habitat variation. Perhaps the most crucial gap is a lack of understanding of what aspects of habitat variation are related to frog refuge sites, in- cluding the often temix)rary refuges used as an escape from predators as well as those refuges used during the season of inactivity. The former type of refuge site may be related to the deep-water and dense vegetation habitat associated with R. a. draytonii, and the riffle habitat associated with R. boylii, but what aspects of those habitat features really comprise the refuge and to what degree they may vary before they are no longer a ref- uge is unknown. A understanding of the latter is pivotal to the identifica- tion of predator-induced habitat re- striction. Most importantly, an understanding of how reproduction and refuge sites are related to habitat variation for these two frogs is essen- tial if management is to ever be re- fined to a level where habitat vari- ables, either individually or in con- cert, may be manipulated. Finally, if habitat manipulations are attempted, they will have to be implemented with caution in aquatic systems where both R. a. draytonii and R. boylii co-occur; differences in habitat characteristics between each species suggest that whatever way one or more of several habitat variables are manipulated, they will probably re- sult in a tradeoff between habitat losses and habitat gains for R. a. dray- tonii versus R. boylii. In summary, habitat analysis for the two ranid frogs, R. a. draytonii and R. boylii, indicates that each spe- cies is most frequently associated with discemibly different aquatic habitats, the former with densely vegetated, deep water and the latter with rocky, shallow-water riffles in streams. The species are similar in that they infrequently co-occur with any aquatic vertebrates, especially the introduced aquatic macrofauna. Low levels of co-occurrence between frogs and the introduced aquatic macrofauna have two confounded explanations: 1) preferential use of different habitats between the intro- duced aquatic macrofauna and frogs, and 2) habitat restriction because frogs and their life stages are preyed upon by the introduced aquatic macrofauna. However, even though it is presently impossible to idenrify the responsible predator, temporal data strongly suggest that R. a. dray- tonii has been restricted by some in- troduced aquatic predator and the same possibility cannot be excluded for R. boylii. For t>oth species, a man- agement scheme is necessary to avert existing trends of decline, and ulti- mately, extinction. A management 155 scheme that minimizes the risk of ex- tinction based on current data must address the worst-case scenario among the ahernatives imphcated in limiting frog distributions. To ad- dress anything less increases the risk of extinction if that alternative is true. Since that alternative is habitat restriction by an introduced aquatic macrofauna, management should strive to isolate both frog species from the introduced aquatic macro- fauna. Moreover, available data indi- cate that preservation of modal con- ditions for habitat variables identi- fied as associated with each species is a suitable interim strategy, since it is more likely to promote isolation. Sig- nificant refinements of this manage- ment scheme will require a thorough understanding of how habitat vari- ables associated with each frog spe- cies are linked to their refuge re- quirements and their reproductive patterns. ACKNOWLEDGMENTS Special thanks go to Charles J. Brown, Jr., Peter B. Moyle, and David B. Wake for allowing us to use data in their care. Sean J. Barry, John M. Brode, Charles W. Brown, Mark L. Cay wood, Henry E. Childs, Jr., Arthur L. Cohen, Nathan W. Cohen, Lawrence R. Cory, John B. Cowan, Robert G. Crippen, Henry S. Fitch, William J. Hamilton, Jr., George H. Hanley, George E. Hansen, Robert W. Hansen, John Hendrickson (Woo- dleaf, Calif.), John R. Hendrickson (University of Arizona), Daniel C. Holland, Samuel B. Horowitz, Alex- ander K. Johnson, William F. Johnson, Donald R. Kirk, J. Ralph Li- chtenfels. Amy R. McCune, Roy W. McDiarmid, Milton D. Miller, Rich- ard R. Montanucci, Garth I. Murphy, Robert T. Orr, Thomas L. Rodgers, Stephen B. Ruth, Robert C. Stebbins, the late Ruth R. Storer, Samuel S. Sweet, Richard Terry, Walter Tor- doff, Jens V. Vindum, Conrad Yamamoto, and Richard G. Zweifel all contributed ancillary data. Addi- tionally, important data were ex- tracted from the unpublished field notes or voucher specimens collected by the following workers no longer living: Adrey E. Borell, Harold C. Bryant, Charles L. Camp, Joseph S. Dixon, Adolphus L. Heermann, Henry W. Henshaw, Carl L. Hubbs, Lloyd G. Ingles, Henry C. Kellers, William N. Lockington, Donald R. McLean, Joseph R. Slevin, Tracy I. Storer, John Van Denburgh, and Al- bert H. Wright. Phyllis A. Buck, Peter B. Moyle, C. Mindy Nelson, and Richard G. Zweifel kindly reviewed the manuscript. LITERATURE CITED Abdulali, Humayun. 1985. On the export of frog legs from India. Journal of the Bombay Natural History Society 82:347-375. Armour, C. L., K. P. Bumham, and W. S. Platts. 1983. Field methods and statistical analyses for moni- toring small salmonid streams. 200 p. United States Fish and Wildlife Service, FWS/OBS-83/33. Bailey, James A. 1984. Principles of wildlife management. 373 p. John Wiley and Sons, Inc., New York. Baker, RoUin H. 1942. The bullfrog: A Texas wildlife resource. Texas Game, Fish and Oyster Commis- sion, Bulletin (23):3-7. Brocke, R. H. 1979. The name of nongame. Wildlife Society Bulletin 7:279-282. Bury, R. Bruce. 1975. Conservation of non-game wildlife in California: A model programme. Biological Conservation 7:199-210. Bury, R. Bruce, Howard W. Campbell, and Norman J. Scott, Jr. 1980a. Role and importance of nongame wildlife. 197-207 p. In Transactions of the 45th North American Wildlife and Natural Resources Conference, Wildlife Management Institute, Washing- ton, D.C. Bury, R. Bruce, C. Kenneth Dodd, Jr., and Gary M. Fellers. 1980b. Con- servation of the Amphibia of the United States: A review. Depart- ment of the Interior, United States Fish and Wildlife Service, Re- source Publication (134):l-34. Bury, R. Bruce, and Jill A. Whelan. 1984. Ecology and management of the bullfrog. Department of the Interior, United States Fish and Wildlife Service, Resource Publica- tion (155):l-23. Chamberlain, F. M. 1898. Notes on the edible frogs of the United States and their artificial propaga- tion. 249-261 p. In Report of the United States Commissioner of Fish and Fisheries for the Year Ending June 30, 1897. United States Government Printing Of- fice, Washington, D.C. Childs, Henry E., Jr., and Walter E. Howard. 1955. The vertebrate fauna of the San Joaquin Experi- mental Range. United States De- partment of Agriculture, Califor- nia Range and Experiment Station, Miscellaneous Paper (19):l-20. Cowan, J. B. 1979. Mammals, reptiles and amphibians of Gray Lodge Wildlife Area. 12 p. Publication of the Department of Fish and Game, Resources Agency, State of Cali- fornia, Sacramento. Cresswell, Richard. 1862. Aristotle's history of animals in ten books. 326 p. Henry G. Bohn, York Street, Covent Garden, London. Cunningham, John D. 1955. Notes on abnormal Rjina aurora draytonii. Herpetologica 11:149. Dodd, C. Kenneth, Jr. 1978. Amphibi- ans and reptiles: the declining spe- cies. Water Spectrum 10:24-32. Fitch, Henry S. 1936. Amphibians and reptiles of the Rogue River Basin, Oregon. American Midland Naturalist 17:634-652. Fitch, Henry S. 1938. Rana hoylii in Oregon. Copeia 1938:148. Fitch, Henry S. 1949. A study of snake populations in central Cali- fornia. American Midland Natu- ralist 41:513-579. 156 Fritz, R. S. 1979. Consequences of in- sular population structure: distri- bution and extinction of spruce grouse (Camchites canadensis) populations. Oecologia 42:57-66. Gordon, Kenneth. 1939. The amphib- ian and reptiles of Oregon. Oregon State College Monographs (l):l-82. Grinnell, Joseph, Joseph Dixon, and Jean M. Linsdale. 1930. Vertebrate natural history of a section of northern California through the Lassen Peak region. University of California Publications in Zoology 35:1-594. Grinnell, Joseph, and Tracy I. Storer. 1924. Animal life in the Yosemite; an account of the mammals, birds, reptiles, and amphibians in a cross-section of the Sierra Nevada. 752 p. University of California Press, Berkeley. Hallowell, Edward. 1854. Descrip- tions of new reptiles from Califor- nia. Proceedings of the Academy of Natural Sciences of Philadel- phia 7:91-97. Hallowell, Edward. 1859. Report on the reptiles collected on the Sur- vey. 27 p. Pacific Railroad Reports 10 (part 4, Williamson's Report), zoology report no. 1. Hayes, Marc P., and Mark R. Jen- nings. 1986. DecHne of ranid frog species in western North America: Are bullfrogs (Rana catesbeiana) responsible? Journal of Herpetol- ogy 20:490-509. Hayes, Marc P., and Dana M. Krem- pels. 1986. Vocal sac variation among frogs of the genus Rana from western North America. Copeia 1986:927-936. Hayes, Marc P., and Michael M. Miyamoto. 1984. Biochemical, be- havioral and body size differences between Rana aurora aurora and R. a. draytoni. Copeia 1984:1018-1022. Hayes, Marc P., and Joan Warner. 1985. Life history notes: Rana catesbeiana, food. SSAR Herpeto- logical Review 16:109. Hine, Ruth L., Betty L. Les, and Bruce F. Hellmich. 1981. Leopard frog populations and mortality in Wisconsin. Department of Natural Resources, Madison, Wisconsin, Technical Bulletin (122):l-39. Holland, Robert F. 1978. The geo- graphic and edaphic distribution of vernal pools in the Great Valley, California. California Native Plant Society (4):1-12. Honegger, Rene E. 1981. Threatened Amphibians and Reptiles in Eu- rope. 158 p. Akademie Verlagsges, Wiesbaden, Germany. Husain, K. Z., and M. M. Rahman. 1978. The amphibian fauna of Bangladesh. Bangladesh Journal of Zoology 6:157-158. Ingles, Lloyd G. 1932a. Four new species of Haematoloechus (Trema- toda) from Rana aurora draytonii from California. Univ. of Calif. Publications in Zool. 37(7):189-201. Ingles, Lloyd G. 1932b. Cephalogoni- mus brevicirrus, a new species of Trematode from the intestines of Rarm aurora from California. Uni- versity of California Publications in Zoology 37(8):203-210. Ingles, Lloyd G. 1933. The specificity of frog flukes. Science 78:168. Ingles, Lloyd G. 1936. Worm para- sites of California amphibia. Transactions of the American Microscopical Society 55:73-92. Jain, Subodh. (editor) 1976. Vernal pools: their ecology and conserva- tion. A symposium sponsored by the Institute of Ecology, Univer- sity of California, Davis, May 1 and 2, 1976. Institute of Ecology Publication (9):l-93. Jennings, Mark R., and Marc P. Hayes. 1985. Pre-1900 overharvest of California red-legged frogs (Rana aurora draytonii): The in- ducement for bullfrog (Rana catesbeiana) introduction. Herpetol- ogica 41:94-103. Kellert, Stephen R. 1985. Social and perceptual factors in endangered sp)ecies management. Journal of Wildlife Management 49:528-536. Leidy, R. A. 1984. Distribution and ecology of stream fishes of the San Francisco Bay drainage. Hilgardia 52(8):1-175. Leopold, Aldo S. 1931 . A history of ideas in game management. Out- door America (June):22-24,38- 39,47. Leopold, Aldo S. 1933. Game man- agement. 481 p. Charles Scribner's Sons, New York. McAuliffe, J. R. 1978. Biological sur- vey and management of sport- hunted bullfrog populations in Nebraska. 78 p. Nebraska Game and Parks Commission, Lincoln, Nebraska. Moyle, Peter B. 1973. Effects of intro- duced bullfrogs, Rana catesbeiana, on the native frogs of the San Joaquin Valley, California. Copeia 1973:18-22. Moyle, Peter B. 1976. Inland fishes of California. 405 p. University of California Press, Berkeley. Moyle, Peter B., and Robert D. Nichols. 1973. Ecology of some native and introduced fishes of the Sierra Nevada Foothills in Central California. Copeia 1973:478-490. Moyle, Peter B., and Robert D. Nichols. 1974. Decline of the fish fauna of the Sierra Nevada foot- hills. Central California. American Midland Naturalist 92:72-83. Moyle, Peter B., Jerry J. Smith, Roberts A. Daniels, Thomas L. Taylor, Donald G. Price, and Donald M. Baltz. 1982. Distribu- tion and ecology of stream fishes of the Sacramento-San Joaquin drainage system, California. Uni- versity of California Publications in Zoology 115:1-256. Neill, Wilfred T. 1974. Reptiles and amphibians in the service of man. 248 p. Biological Sciences Curricu- lum Study Book, Pegasus Press, The Bobbs-Merrill Company, Inc., Indianapolis, Indiana. Pister, E. P. 1976. A rationale for the management of nongame fish and wildlife. Fisheries 1:11-14. Rutter, Cloudsley. 1908. The fishes of the Sacramento-San Joaquin Basin, with a study of their distribution and variation. Bulletin of the United States Bureau of Fisheries 27:103-152. 157 Slevin, Joseph R. 1928. The amphibi- ans of western North America. Occasional Papers of the Califor- nia Academy of Sciences (16):1- 144. Sokal, Robert R., and R. James Rohlf. 1981. Biometry. 687 p. W. H. Free- man and Co., San Francisco. Stebbins, Robert C. 1966. A field guide to western reptiles and am- phibians. 279 p. Houghton Mifflin Company, Boston. Stebbins, Robert C. 1985. A field guide to western reptiles and am- phibians. Second edition, revised. 336 p. Houghton Mifflin Com- pany, Boston. Storer, Tracy 1. 1925. A synopsis of the Amphibia of California. Uni- versity of California Publications in Zoology 27:1-342. Storer, Tracy 1. 1933. Frogs and their commercial use. California Fish and Game 19:203-213. Strahler, A. N. 1957. Quantitative analysis of watershed geomor- phology. Transactions of the American Geophysicists Union 38:913-920. Sweet, Samuel S. 1983. Mechanics of a natural extinction event: Ram boylii in southern California. Ab- stract of a paper presented at the 26th annual meeting of the Society for the Study of Amphibians and Reptiles, and the 31st annual meet- ing of the Herpetologists' League, University of Utah, August 7-12, 1983. Treanor, Robert R. 1975a. Manage- ment of the bullfrog (Ram catesbe- iam) resource in California. Cali- fornia Department of Fish and Game, Inland Fisheries Adminis- trative Report (75-l):l-30. Treanor, Robert R. 1975b. Manage- ment of the bullfrog resource in California. Cal-Neva Wildlife Transactions 1975:81-92. Treanor, Robert R., and Stephen J. Nicola. 1972. A preliminary study of the commercial and sporting utilization of the bullfrog. Ram catesbeiam Shaw, in California. California Department of Fish and Game, Inland Fisheries Adminis- trative Report (72-4):l-23. United States Geological Survey. 1970a. Surface water supply of the United States 1961-1965. Part II. Pacific slope basin in California. Volume 3. Southern Central Valley basins. U. S. Geological Survey, Water-Supply Paper (1930):1-155. United States Geological Survey. 1970b. Surface water supply of the United States 1961-1965. Part II. Pacific slope basin in California. Volume 4. Northern Central Val- ley basin. Uruted States Geological Survey, Water-Supply Paper (1931):1-111. Walker, M. V. 1946. Reptiles and amphibians of Yosemite National Park. Yosemite Nature Notes 25:1- 48. Wemette, F. G., F. A. Hall, Jr., C. J. Brown, Jr., C. L. Mayer, and N. A. Villa. 1982. Los Vaqueros project — fish and wildlife impacts. 263 p. California Department of Fish and Game, Status Report. Williamson, R. S. 1855. Report of Lieutenant R. S. Williamson. 43 p. In Report of Exploration in Cali- fornia for Railroad Routes, to con- nect with the Routes Near the 35th and 32nd Parallels of North Latitude; In Report of Explora- tions and Surveys to Ascertain the Most Practicable and Economical Route for a Railroad from the Mis- sissippi River to the Pacific Ocean, Made Under the Direction of the Secretary of War, in 1853-54. Vol- ume 5, part I. Beverly Tucker, Washington, D.C. (1856) [Senate Executive Document No. 78, 33rd Congress, 2nd Session]. Willis, Yuell L., Don L. Moyle, and Thomas S. Baskett. 1956. Emer- gence, breeding, hibernation, movements and transformation of the bullfrog. Ram catesbeiam, in Missouri. Copeia 1956:30-41. Wright, A. H. 1920. Frogs: their natu- ral history and utilization. Depart- ment of Commerce, United States Bureau of Fisheries Document (888):l-44. Wright, Albert Hazen, and Anna Al- len Wright. 1949. Handbook of frogs and toads of the United States and Canada. Third edition. 640 p. Comstock Publishing Com- pany, Inc., Ithaca, New York. Zahl, Paul A. 1967. In quest of the world's largest frog. National Geographic Magazine 132:146-152. Zar, Jerrold H. 1974. Biostatistical analysis. Prentice-Hall, Inc. 620 p. Englewood Cliffs, New Jersey. Zedler, Paul H. 1987. The ecology of southern California vernal pools: a community profile. Department of the Interior, United States Fish and Wildlife Service, Biological Report 85(7.11):1-136. Zweifel, Richard G. 1955. Ecology, distribution, and systematics of frogs of the Ram boylei group. University of California Publica- tions in Zoology 54(4):207-292. 158 Integrating Anuran Amphibian Species into Environmental Assessment Programs^ Ronald E. Beiswenger^ Abstract.— Anurans are often given minimal attention in environmental assessments despite their ecological importance and potential value as indicator species. Habitat and guild-based models must be adopted to include all life cycle stages of anurans. A preliminary habitat suitability model for the American toad shows how this con be accomplished. As a result of our increased under- standing of the roles of wildlife spe- cies in ecosystem structure and func- tion, and legal requirements to de- velop holistic approaches to environ- mental management, it has become increasingly common to include all species of wildlife in resource inven- tories and monitoring programs (Chalk et al. 1984). However, am- phibians are often ignored or given minimal attention in such programs, even though they are important wildlife resources and should be given serious consideration in man- agement evaluations (Bury and Ra- phael 1983, Bury et al. 1980, Jones 1986). If included in resource evalu- ations at all, amphibians are usually lumped with reptiles in a category called herpetofauna and even then are often only represented as items in a species list. This is unfortunate because, in addition to their ecological impor- tance, anurans are potentially valu- able as a unique form of indicator species capable of integrating envi- ronmental changes occurring in both the terrestrial and aquatic phases of their habitats. Furthermore, because they occupy small ponds and the shallow margins of lakes, anurans ' Paper presented at symposium. Man- agement of Amphibians, Reptiles and Small Mammals in North America. (Flagstaff. AZ. July 19-21. 1988.) '^Ronald E. Beiswenger is Professor. De- partment of Geography and Recreation. The University of Wyoming. Laramie, WY 82071. are likely to be the first vertebrates to come in contact with contaminated run-off or acidified snowmelt. This could make them useful as elements of an early warning system for the detection of environmental contami- nation. Campbell (1976) found that the boreal toad, Bufo boreas, would be an especially effective indicator spe- cies for monitoring the impact of cloud seeding in the mountains of Colorado. It is also significant that many anurans require specialized habitats in wetland areas and ripar- ian zones, and could serve as indica- tor species for the overall health of these areas of special ecological im- portance. Despite their potential usefulness, there are several reasons why am- phibians are not given adequate at- tention in environmental assess- ments. The importance of amphibi- ans in ecosystems is generally unrec- ognized, particularly by the general public and the resource managers who must respond to the desires of this public as they set management priorities. Also, the secretive habits during the non-breeding season, and complex life cycles of amphibians make them relatively difficult to study. Consequently, the natural his- tory of many amphibian species is not well known. Another factor is that current models for monitoring and assessment have been developed for either terrestrial or aquatic spe- cies and have not been adapted to species with divergent life cycle stages which depend on both aquatic and terrestrial habitats (table 1). Table 1 .— Habitat components and life cycle stages of anurans. Habitat component Eggs/Pre- feeding tadpoles Feeding IVIetamorphosing tadpoles tadpoles Juveniles Adults Aquatic Phase Spawning sites X Tadpole habitat X Aquatic/Terrestrial Interface Phase Tadpole habitat Juvenile habitat Terrestrial Phase Summer habitat Hibernation sites Movement corridors Interspersion Factors Distribution of habitat components Density of habitat components X X X X X X X X X X X X X X X 159 Approaches for incorporating wildlife into resource evaluations in- clude inventories of relative abun- dance and species richness, develop- ment of databases, the use of indica- tor species, and the development of species diversity indices and models using guild concepts. However, the application of these approaches to species of Amphibia has not kept pace with applications to other spe- cies of vertebrates. The primary purpose of this paper is to suggest ways to use single spe- cies models, and models which use guilds and habitat structure, to more effectively integrate anuran amphibi- ans into resource assessments. A single species model for the Ameri- can toad, stressing the importance of tadpole habitat, is presented in some detail. Models for Anurons Guilds and Habitat Structure Guild-based environmental assess- ments are especially useful from an ecological perspective, although they are most effective when used in com- bination with other methods (Karr 1987). Unfortunately, when amphibi- ans are included in guild-based pro- grams they are usually considered too simplistically. A common proce- dure is to categorize them according to their general spawning and feed- ing habitat, but to include no further detail (e.g. see Thomas et al. 1979). The habitat models developed for Arizona (Short 1984) represent a good starting point for producing effective models for anurans. In these models wildlife guilds are used to correlate habitat use with habitat structure (layers) by associating a species with a particular plant com- munity (habitat or cover type), and then with a habitat layer. Layers of both terrestrial and aquatic habitat are included. This system is as appropriate for terrestrial adult anurans as it is for any small, terrestrial vertebrate. However, the aquatic phases of the model require further development if it is to be used with the aquatic larval stages of amphibians. The adaptive significance of the tadpole stage has been established by Wassersug (1975) and Wilbur (1980), and it is clear that the habitat requirements of larval anurans should be an important component of habitat models. The selection of a spawning site that will provide high quality habitat for the tadpole stage is likely to be critical to the evolutionary success of an anu- ran species. Single Species Models Habitat models for indicator species have been developed by the U.S. Fish and Wildlife Service (1981), the U.S. Forest Service (Berry 1986) and oth- ers (e.g. Clawson et al. 1984) for use in assessing environmental impacts and in making management deci- sions. A comprehensive habitat model for an anuran species must encompass spawning sites, tadpole habitat, metamorphic sites, juvenile and adult feeding habitat, movement corridors and hibernation sites. For example, a model developed for the bullfrog (Rana catesbeiana) illustrates how the approach can be applied to Table 2.— Components of habitat for Bufo americanus (measurable attrib- ute in parentheses). Spawning Habitat Shallow, emphemeral ponds (depth range) m^m:: Emergent or submergent vegetation (% cover) ■ Exposure to direct sunlight (% of area shaded) Tadpole Habitat Ponds with access to shallow shoreline areas (< 10 cm) and to deeper areas (10-100 cm) Substrates with food W-:W-r periphyton (% cover) bottom areas with detritus or microorganisms (% cover) Microorganisms suspended in water column (density) Exposure to direct sunlight (% of area shaded) Metamorphic Habitat Shallow depth gradient at shoreline (< 1 0 cm) Exposure to direct sunlight (% of area shaded) Moist substrate on shore (moisture content) Vegetative cover on shore (% cover) Juvenile and Adult Habitat Availability of insect and other invertebrate prey (prey density) Access to moist substrates and refugia (moisture content and refu- gia density) Access to vegetative cover (distance to cover) Hibernation Site Unoccupied animal burrows (burrow density) Friable soils (soil texture) Root zones of large trees (large tree density) Interspersion Movement corridors between hibernation and spawning sites (distri- bution of continuous open areas with adequate cover) Distribution and density of potential spawning sites within the home range of the population (density of spawning sites) 160 an anuran species that is primarily aquatic (Graves and Anderson 1987). While this model is well constructed, a different modeling approach would be needed for anurans with terres- trial adult stages. A limitation of the bullfrog model is that the habitat re- quirements of the tadpole stage are not given in sufficient detail. This is important because the larval stage (up to three years in duration) repre- sents a significant proportion of a bullfrog's total lifespan. A different array of habitat com- ponents for a species that is predomi- HABITAT VARIABLES nantly terrestrial is an adult, the American toad (Bufo americanus) is outlined in table 2. This outline is based on extensive field studies in Michigan (Beiswenger 1975, 1977), field observations of related toad species in Oregon and Wyoming (Beiswenger 1978, 1981, 1986), and information found in the literature. Including the terrestrial features of toad habitat in assessments does not represent a particularly difficult chal- lenge because these features can be described using well-established ap- proaches developed for other small COMPONENTS Percent of water area 1 m or less in depth (VI) Percent cover of rooted aquatic vegetation (V2) Percent of shoreline v/ith shading riparian vegetation (V3) Percent of shoreline v/ith strip of invegetated shallow water (V4) Percent of shoreline with terrestrial vegetative cover or ground debris within 1 m of water (V5) Percent tree canopy closure (V5) Percent of trees that are deciduous species (V7) Percent herbaceous canopy cover (V8) Number of burrows, decaying logs, and debris objects larger than 20 cm in diameter on the ground (V9) Distance along a protected dispersal corridor to potential spawning sites (V10) _ Aquatic cover/ reproduction Jerrestrial cover/ hibernation Interspersion Figure 1 .—Relationships of habitat variables to components of an HSI model for the Ameri- can toad. vertebrates that live on and below the surface of the ground. However, tadpole habitat is also important and must be incorporated into habitat as- sessment procedures. This is some- what more challenging because less is known about tadpole ecology and techniques for describing tadpole habitat are not well developed. A Habitat Model for the American Toad A preliminary version of a habitat suitability model for the American toad is described here to show how the requirements of all life cycle stages could be incorporated into such a model (figs. 1 and 2). The model includes 10 variables and is based primarily on the author's expe- rience and a partial literature review. Consequently, the model should be refined through a more extensive analysis of the literature and a peer review process before it is field tested. The habitat requirements of spawning adults and tadpoles are included in the aquatic cover/repro- ductive component of the model. The quality of spawning sites selected by American toads is influenced by structural features such as depth gra- dients and vegetation. Adult toads typically lay their eggs in shallow, unshaded, vegetated areas (variables 2 and 3), distributing them in strands on the vegetation. At first the newly hatched tadpoles do not feed, but remain at the site where the eggs were laid. Older tadpoles are active swim- mers and display a variety of feeding modes that arc influenced to a large measure by structural features of the habitat (e.g. aquatic vegetation and depth gradients) (variables 1, 2, and 4). Wassersug (1975) has shown that tadpoles are essentially non-discrimi- nant suspension feeders, although they use a variety of means for ob- taining food. Tadpoles of the Ameri- can toad most commonly graze 161 t 0.5- 100!? PERCEN"0- WATER AREA I n OR LE55 IN DEPTH (Variable 1) m% PERCENT SHORELINE WITH SHADING RIPARIAN VEGETATION (variaole 3) PERCFHT f.OVFR OF ROOTED ACUATIC VEGETATION (Variable 2) TO-i 05- l(50« PERCENT SHORELINE WITH 30-50 cn WIDE STRIP OF UNVFGFTATFD SHALLOW WATER ;iOcn OR LESS DEEP) (Variable 4) PERCENT SHORELINE WITH VEGETATIVE COVER OB GROUND DEBRIS WITHIN 1 m 0- WATER (Variable 5) PERCENT OF TREES THAT ARE DEC DUOUS SPECIES (Variable 7) 100^ PERCENT TREE CANOPY CLOSURE (Variable 6) 0.5- PERCENT HERBACEOUS CANOPY COVER (V.¥lable a) 150 m 300 n NUMBER OF BURROWS, DECAYING LOGS, AND DEORISODjECTS LARGER THAN 20 cm INDiAMETEROM THE GROUND (Va-.able DISTANCE ALONG A PROTECTED D 3PER5AL CORRIDOR TO POTENTIAL SPAWNING SITES (VariablelO) Figure 2.— The assumed relationships among habitat variables and suitability index values for the American toad. periphyton from emergent or sub- mergent vegetation, or scrape micro- organisms and detritus from the pond bottom and other substrates. However, when blooms of sus- pended algae are present, the tad- poles become midwater filter feed- ers. They also feed on organic mate- rial supported by the surface film of the pond. At other times, the tad- poles are facultatively cannibalistic or coprophagic. The particular feed- ing mode employed is usually influ- enced by a combination of factors including the type of food available, depth and temperature gradients, vegetation structure and the degree of social behavior exhibited by the tadpoles (Beiswenger 1975). Most of the time toad tadpoles feed from substrates provided by the structural features of their environment. Diaz- Paniagua (1987) also found structural features of aquatic vegetation to be important in the distribution of the tadpoles of five anuran species in Spain. Habitat use by tadpoles is strongly influenced by temperature, which in the shallow ponds they occupy is highly correlated with depth and so- lar radiation (variables 1, 3, and 4). For example, in northern Michigan ponds were early summer tempera- tures varied greatly over the diel pe- riod, toad tadpoles consistently se- lected the warmest available water in thermally stratified ponds (Beiswenger 1977). Thus, they occu- pied the deepest areas of the pond (greater than 50 cm in depth) at night, avoiding the shallow pond margin where temperatures were 5.5 C cooler. During the day tadpoles moved to shallow areas near shore which were 9 C warmer than the deeper areas of the pond. During those times when there was no ther- mal stratification (e.g. cloudy days), or later in the summer when pond temperatures were uniformly high, the tadpoles used all parts of the pond (Beiswenger 1977). These ob- servations indicate that tadpole habi- tat quality is partly determined by thermal stratification associated with depth gradients and exposure to di- rect sunlight. Habitat quality for mctamorphic tadpoles is strongly influenced by their vulnerability to predation (vari- ables 4 and 5). As Arnold and Was- sersug (1978, p. 1019) expressed it, "the transforming anuran is neither a good larva nor a good frog." The lar- vae develop forelimbs which impede swimming, the tail remnant on the newly emergent juvenile interferes with its jumping ability. Conse- quently, the availability of structural features such as hiding cover and moist substrates is important for the successful emergence and dispersal of metamorphosing tadpoles. Habitat quality for juvenile and adult toads is determined by factors generally associated with deciduous or mixed coniferous/deciduous for- ests. These factors include moderate temperature regimes, invertebrate prey density, protected microhabitats with moist substrates, vegetative cover, and access to hibernation sites. Some of the variables used as surro- gate measures of substrate moisture and other forest floor conditions in the HSI model for the red-spotted newt (Sousa 1985) were adapted for the American toad model (variables 6, 7, and 8). Juvenile and adult toads 162 also need moist cover during hot dry- periods and for winter hibernacula. These can be provided by soils which are suitable for burrowing, existing small mammal burrow systems, or decaying logs and other debris ob- jects on the ground (variable 9). The American toad model in- cludes interspersion as a habitat-re- lated factor. Movement corridors interconnecting spawning areas, summer habitat and hibernation sites are an im|X)rtant component of juve- nile and adult habitat (variable 10). Brode and Bury (1984) have pointed out (cited in Ohmart and Anderson 1986), that such corridors are impor- tant for dispersal and genetic conti- nuity, and anurans use riparian zones as travel lanes. Habitat frag- mentation by road construction (Rittschof 1975), or other forms of habitat destruction can disrupt these travel lanes and prevent anurans from reaching spawning ponds or hibernation sites. Attention must also be paid to other aspects of interspersion. For example, the reproductive success of toads depends on the continuing availability of shallow water habitats. Ponds with optimum spawning con- ditions in a given year may be dry in years with low precipitation, or too deep in years when flooding pre- vails. At the same time, changing wa- ter levels may result in the availabil- ity of new spawning sites, apparently in response to this kind of variation, some species of toads do not use the same spawning site every year (Kelleher and Tester 1969) and in some years may not breed at all. Be- cause of variation like this, it is im- portant to describe the distribution of habitat components, such as spawn- ing sites and movement corridors, in a broad geographic area and over a range of environmental conditions. Relationships among the habitat variables and habitat components are expressed by equations in HSI mod- els. A value for the aquatic cover/ reproduction (SIA) component is ob- tained by combining the suitability index values for variables 1 through 4, as shown in the following equa- tion. SIA = SIVl X SIV2 x(SI\/3+SIV4) 2 This assumes that the suitability of aquatic habitats is primarily deter- mined by the presence of water depths ranging from less than 10 cm to 1 m, rooted aquatic vegetation to provide cover and substrates for food, and shallow, unshaded shore- line areas. It is assumed that terrestrial habi- tat suitability (SIT) is determined by the availability of cover with moist substrates, invertebrate prey and hi- bernation sites. The following equa- tion shows how these habitat values could be evaluated using variable 5 to assess cover for metamorphic stages, 6, 7, and 8 as surrogate meas- ures of substrate moisture, and vari- able 9 for the availability of hibernac- ula. SIT= (SIV5+SIV6+SIV74-SIV9) 4 Overall habitat suitability (HSI) is determined by combining the suita- bility values for the aquatic (SIA) and terrestrial (SIT) habitat components with the suitability value for inter- spersion (SII) as shown in the follow- ing equation. HSI = (SIA x SIT x 511)^^3 This form is used because a value of zero for the suitability index for any one of the three components indi- cates a lack of habitat to maintain vi- able populations of American toads. Once it has been fully developed, a habitat model for the American toad could be used to assess the ef- fects of such activities as road build- ing, housing construction, environ- mental pollution, landfill operations, clearing of deciduous forests, drain- ing or dredging of ponds and wet- lands, intensive recreational use of wetlands, floodplains and the shore- line areas of lakes, and large changes in water level by removing or intro- ducing water. Habitat Models and Endangered Species Protection The Wyoming toad (Bufo hemiophrys baxteri) has recently been listed as endangered by the U.S. Fish and Wildlife Service (Baxter et al. 1982). As of June 1988, there was only one small breeding population known to exist. There are no habitat models available for this subspecies and there have been few studies of its natural history. This is unfortunate because there is an urgent need to begin a recovery program. Informa- tion about the related Manitoba toad (Bufo hemiophrys) which has been more extensively studied could be used to infer habitat relationships, but this is obviously not as valid as studying the Wyoming toad directly. This situation illustrates why it is important to intensify our efforts to develop databases and habitat mod- els for all species before they reach the point of becoming endangered. It also exemplifies the role a habitat model can play in identifying infor- mation gaps and focusing research efforts. Discussion Resource assessments require the development of models for the quan- titative assessment of habitat suitabil- ity. It is essential that such models be developed in combination with com- prehensive databases. A long range goal should be to develop databases with efficient retrieval systems so that it is possible to access all of the site-specific natural history informa- tion available in the literature, and in the files of researchers and resource managers. The databases should also be constructed so that information gaps and priority areas for research can be identified. This paper has emphasized pro- ducing habitat models for individual species as if these species exist in iso- lation. Hutto et al. (1987) have criti- cized the overemphasis on species 163 approaches in conservation pro- grams as too narrow and they point out that we must not lose sight of the higher order patterns and processes which occur among interacting spe- cies. They suggest supplementing the species approach with approaches that consider such things as land- scape patterns that maintain ecosys- tem level processes, the use of geo- graphic information systems, and other land-based approaches. Studies emphasizing the role of anurans in ecosystems should result in a better understanding of ecologi- cal process occurring at the terres- trial-aquatic interface, and could also contribute to more effective manage- ment of species which depend on these edge habitats and ecotones. Literature Cited Arnold, Stevan J. and Richard J. Was- sersug. 1978. Differential preda- tion on metamorphic anurans by garter snakes (Thamnophis): social behavior as a possible defense. Ecology 59: 1014-1022. Baxter, George T., Mark K. Stromberg and C. Kenneth Dodd. 1982. The status of the Wyoming toad (Bufo hemiophrys baxteri). En- vironmental Conservation 9: 348, 338. Beiswenger, Ronald E. 1975. Struc- ture and function in aggregations of tadpoles of the American toad, Bufo americanus. Herpetologica 31: 222-233. Beiswenger, Ronald E. 1977. Diel pat- terns of aggregative behavior in tadpoles of Bufo americanus, in re- lation to light and temperature. Ecology 58: 98-108. Beiswenger, Ronald E. 1978. Re- sponses of Bufo tadpoles to labora- tory gradients of temperature. Journal of Herpetology 12: 499- 504. Beiswenger, Ronald E. 1981. Preda- tion by gray jays on aggregating tadpoles of the boreal toad (Bufo boreas). Copeia 1981: 459-460. Beiswenger, Ronald E. 1986. An en- dangered species, the Wyoming toad, Bufo hemiophrys baxteri — the importance of an early warning system. Biological Conservation 37: 59-71. Berry, K.H. 1986. Introduction: de- velopment, testing, and applica- tion of wildlife-habitat models. Pages 3-4. In J. Verner, M.L. Mor- rison, and C.J. Ralph, eds. Wildlife 2000: modeling habitat relation- ships of terrestrial vertebrates. Univ. Wise. Press, Madison. 470 p. Brode, J.M. and R.B. Bury. 1984. The importance of riparian systems to amphibians and reptiles. Pages 30- 36. In Warner, R.E. and K.M. Hen- drix, eds. California riparian sys- tems: ecology, conservation, and productive management. Univ. California Press, Berkeley. 1035 p. Bury, R. Bruce, Howard W. Camp- bell and Norman J. Scott, Jr. 1980. Role and importance of nongame wildlife. Transactions North American Wildlife and Natural Resources Conference 45: 197-207. Bury, R. Bruce and Martin G. Ra- phael. 1983. Inventory methods for amphibians and reptiles. Pages 416-419. In John F. Bell and Toby Atterbury, eds. Proceedings of the international conference. Renew- able resource inventories for monitoring changes and trends. Oregon State University, Corval- lis. 737 p. Campbell, James B. 1976. Environ- mental controls on boreal toad populations in the San Juan Moun- tains. Pages 289-295. In Harold W. Steinhoff and Jack D. Ives (eds). Ecological impacts of snowpack augmentation in the San Juan Mountains, Colorado. Final report San Juan ecology project. Colo- rado State University Publ., Fort Collins. 480 p. Chalk, David E., Stephen A. Miller and Thomas W. Hoekstra. 1984. Multispecies inventories: integrat- ing information on wildlife re- sources. Wildlife Society Bulletin 12: 357-364. Clawson, Mary E., Thomas S. Bas- kett, and Michael J. Armbruster. 1984. An approach to habitat mod- eling for herpetofauna. Wildlife Society Bulletin 12: 61-69. Diaz-Paniagua, Carmen. 1987. Tad- pole distribution in relation to vegetal heterogeneity in tempo- rary ponds. Herpetological Journal 1: 167-169. Graves, Brent M. and Stanley H. An- derson. 1987. Habitat suitability index models: bullfrog. U.S. Fish Wildl. Serv. Biol. Rep. 82 (10: 138). 22 p. Hutto, Richard L., Susan Reel and Peter B. Landres. 1987. A critical evaluation of the species approach to biological conservation. Endan- gered Species UPDATE 4(12): 1-4. Jones, K. Bruce. 1986. Amphibians and reptiles. Pages 267-290. In Co- operrider, A.Y., R.J. Boyd and H.R. Stuart eds. Inventory and monitoring of wildlife habitat. U.S. Department of the Interior, Bureau of Land Management, Service Center. Denver, Co. xviii, 858 p. Karr, James R. 1987. Biological moni- toring and environmental assess- ment: a conceptual framework. Environ. Manage. 11: 249-256. Kelleher, Kelvin E. and John R. Tester. 1969. Homing and survival in the Manitoba toad, Bufo hemio- phrys, in Minnesota. Ecology 50: 1040-1048. Ohmart, Robert D. and Bertin W. Anderson. 1986. Riparian habitat Pages 169-199. In Cooperrider, A.Y., R.J. Boyd and H.R. Stuart, eds. Inventory and monitoring of wildlife habitat. U.S. Dept. Inter., Bur. Land Manage. Service Center. Denver, Co. xviii, 858 p. Rittschof, Daniel. 1975. Some aspects of the natural history and ecology of the leopard frog, Rana pipiens. Ph.D. Dissertation, University of Michigan, Ann Arbor. Short, H.L. 1984. Habitat suitability index models: The Arizona guild and layers of habitat models. U.S. Fish Wildl. Serv. FWS/OBS-82/ 10.70. 37 p. 164 Sousa, P.J. 1985. Habitat suitability index models: Red-spotted newt. U.S. Fish Wildl. Serv. Biol. Rep. 82(10.111). 18 p. Thomas, Jack Ward (ed). 1979. Wild- life habitats in managed forests. The Blue Mountains of Oregon and Washington. Agricultural Handbook No. 553. USDA Forest Service, Washington, D.C. 512 p. U.S. Fish and Wildlife Service. 1981. Standards for the development of habitat suitability index models. ESM103. U.S. Depart. Int. Fish Wildl. Serv., Division of Ecological Services. Washington, D.C. 68 p. and appendices. Wassersug, Richard J. 1975. The adaptive significance of the tad- pole stage with comments on the maintenance of complex life cycles in anurans. American Zoologist 15: 405-417. Wilbur, Henry M. 1980. Complex life cycles. Annual Review of Ecology and Systematics 11: 67-93. 165 Preliminary Report on Effect of Bullfrogs on Wetland Herpetofaunas in Southeastern Arizona^ Cecil R. Schwalbe and Philip C. Rosen^ Abstract.— Ranid frogs (Rona cofesbeiano, R. chiricohuensis, and R. yavopaiensis), garter sr^akes (Thamnophis eques, T. marcianus) and Sonoran mud turtles (Kinosfernon sonoriense) were surveyed in soutt^eastern Arizona. Distribution of the introduced bullfrog (Rano cofesbeiano) was negatively correlated witt^ distributions of the two leopard frogs and garter snakes, The hypothesis that bullfrog predotion caused decline of a native wetland herpetofouno is supported by data on bullfrog diet, on garter snake, leopard frog and mud turtle population structure, and natural history observations on the snakes. An experimental removal of bullfrogs has been initiated at the San Bernardino National Wildlife Refuge. The bullfrog (Ram catesbeiam) is North America's largest frog and one of the most widely distributed anu- rans on the continent. Occurring naturally from Florida to Nova Scotia and west into central Texas, Okla- homa, and Kansas, the bullfrog has been introduced widely into perma- nent waters throughout the West (Bury and Whelan 1984, Stebbins 1985, Wright and Wright 1949) Known to be voracious, opportunis- tic predators, they have been impli- cated in declines of native anuran populations (Bury and Luckenbach 1976, Bury et al. 1980, Conant 1975, 1977, Jameson 1956, Moyle 1973, Nussbaum et al. 1983, Vitt and Ohmart 1978 and others). Much less is known about their impacts on other vertebrate classes. A recent investigation of factors producing decline of Mexican garter snakes (Thamnophis eques) in Arizona (Rosen and Schwalbe 1988) sug- gested that predation by introduced bullfrogs (see fig. 1) is a present and 'Paper presented of the Symposium, The Monagemenf of Amphibians, Repfiles, and Small Mammals in North America. July 18- 22. 1988. Flagstaff. Arizona. ^Cecil R. Schwalbe is Nongame Herpe- tologist and Philip C. Rosen is Contract Bi- ologist. Arizona Game and Fish Depart- ment. 2222 West Greenway Road. Phoenix. Arizona 65023-4399. Rosen's present ad- dress is Department of Ecology and Evolu- tionary Biology. University of Arizona. Tucson. Arizona 85721. serious impact on some of the few remaining snake populations. Obser- vations during the garter snake sur- vey suggested a similar effect on leopard frogs (Rana yavapaiensis, R. chiricahuensis). Recently, Hayes and Jennings (1986) questioned the importance of bullfrog predation in declines of western North American ranid frogs. They include predation by bullfrogs as one of three major hypotheses to explain decline of ranid frogs in Cali- fornia, but suggest that predation by introduced fish has had greater im- pact on native frogs. Hayes and Jen- nings (1986) indicate further that their hypotheses need to be tested to determine actual causal factors in population declines. In this paper we present distributional and natural historical data implicating bullfrogs in population declines of native wet- land reptiles and amphibians in southeastern Arizona. We then de- scribe an experimental program of bullfrog removal we have initiated to test the direct and indirect effects of this introduced predator on wetland herpetofaunas. 166 Methods and Materials We report on two phases of our work. The first phase involves exten- sive surveys, principally for garter snakes. The second focuses on inten- sive surveying and experimental manipulation at one locality that is heavily infested with bullfrogs. Extensive Phase We sampled over 80 localities throughout much of central and southern Arizona during 1985-1987, searching appropriate aquatic and semi-aquatic habitats (Rosen and Schwalbe 1988). Methods and results are briefly summarized here. Lotic habitats were surveyed for 2-6 mile reaches on foot. Lentic habitats were also examined on foot, in their en- tirety in most cases. During these surveys, attempts were made to cap- ture, measure, mark and release all garter snakes seen. Detailed observa- tions were made on distribution and abundance of other biota on the sites sampled, with special attention to anurans, turtles and other snakes. Intensive mark-recapture studies were conducted at four sites using trapping methods described below. Intensive Phase San Bernardino National Wildlife Refuge (SBNWR), one of four sites where mark-recapture procedures were initiated during the extensive phase of our work, was selected for ongoing observation and experimen- tation. Beginning in September 1986, we visited the refuge in September and May of each year, marking snakes, observing herpetofaunal dis- SAN BEffmom NWR Mexico Figure 2.— Diagrammatic map of San Bernardino National Wildlife Refuge. Stippled line indi- cates boundary between upland Chihuahuan desertscrub and riparian scrub and wood- land vegetation types. tributions and abundances, and ex- perimentally removing bullfrogs. Intensive Site Description SBNWR (fig. 2) consists of 984 ha in the San Bernardino Valley on the Mexican border in Cochise County, Arizona. Elevations range from 1134 to 1183 m. Higher, rocky slopes and mesas supporting Chihuahuan de- sertscrub and lower terraces grading into desert grassland comprise al- most two-thirds of the refuge. The heart of the refuge is a low- land supporting dense mesquite (Prosopis velutina) bosques and sacaton (Sporobolus) grasslands inter- spersed with four spring-fed ponds and seven additional springs. In the center of this low ground is deeply incised Black Draw, headwater of the Rio Yaqui, which normally arises at a natural spring about halfway be- tween the Mexican border and north boundary of the refuge. Large, iso- lated, living and dead cottonwood trees (Populus fremontii) occur near almost all aquatic habitats. Broad swamplikc ciencgas with little open water occur at the artesian wells that do not supply ponds. Vegetation in Black Draw varies from rank herbaceous plants and tall grass in the northern one-half, through open riparian thicket and cat-tail (Typha domingensis) stands, into almost impenetrable thickets of sapling cottonwood and willow (Salix gooddingi) throughout the lower 1.2 km to the border. Cienega pools are cold and reach a depth of about 2 meters. North Pond, focal point for the experimental removal of bullfrogs, contains 0.1 ha of open water sur- rounded by earthen levees. Artesian well flow is piped into the pond and into a small marshy area north of the pond. North and west banks are lined with mesquite. South and west banks are open or overgrown with herbaceous vegetation. Cat-tail is spreading rapidly around the pond 167 margin from foci in northeast and southwest corners. Open water is largely choked with submergent macrophytes. The wetland herpetofauna of the refuge includes bullfrogs (Rana catesbeiana), lowland leopard frogs (R. yavapaiensis), Mexican garter snakes (Thamnophis eques), checkered garter snakes (T. marcianus) , and Sonoran mud turtles (Kinosternon sonoriense). Intensive Field Procedures Garter snakes were collected by hand at all times of day and night, and with minnow traps connected by aquatic drift fences (see Rosen and Schwalbe 1988 for details). Four drift fences, each with a trap at each end, were set in North Pond during each visit to the refuge. Two drift fences with traps were set in Twin Pond in August 1985 and August-September 1986. Twin Pond was drained during summer 1987 and remains dry. The following data were recorded for each snake captured: date, loca- tion, sex, snout-vent length (SVL), tail length, total weight, presence/ absence and number of food items, and injuries. Females were palped to determine presence /absence and number of developing young. For hand-caught snakes we recorded ac- tivity at time of first sighting, mi- crohabitat, time, and cloacal and am- bient temperatures. Each individual was uniquely marked by clipping subcaudal scales. Bullfrogs were collected mostly with four-pronged spears at night by using head lamps to find and blind them. Additionally, many were col- lected in turtle hoop nets, which were set along seine nets rigged as aquatic drift fences. Some hoop nets were baited to capture turtles, and these captured bullfrogs, as well. A few were collected by hand and with air guns and light arms. Initial col- lecting efforts were focused on larger (>100 mm SVL) bullfrogs. Every aquatic habitat on the refuge was checked for frogs by listening for their calls and searching visually at night. Captured bullfrogs were kept on ice overnight and the following data were recorded the next day: capture location and date, sex, snout- vent length, total weight. Most were dissected to determine stomach con- tents and reproductive condition. Results Distribution and Natural History Leopard frogs are significantly less common where bullfrogs abound (table 1: Spearman rank correlation r^=-0.434, p<0.025, Rosner 1982, Sokal and Rohlf 1981). SBNVVR is the only site where we found both bullfrogs and leopard frogs. Among the sites shown in table 1, introduced, non- native predatory fish were found in abundance only at Bog Hole and Babocomari Cienega, where ranid frogs were absent. Historical records indicate that leopard frogs once were abundant in two areas now support- ing dense bullfrog populations, Ari- vaca Creek (Wright and Wright 1949) and SBNWR (Lanning 1981, Lowe personal communication). Mexican garter snakes also are sig- nificantly less abundant in the pres- ence of bullfrogs (table 1: Spearman rank correlation r^=-0.420, p<0.03). At the Potrero Canyon locality, Mexican garter snakes were known as late as 1970 (Rosen and Schwalbe 1988), but we found only checkered garter snakes (N = 24) during 19854987. At SBNWR, all museum records of Thamnophis prior to 1970 (N=7) were Table 1. —Distribution and abundance of ranid frogs and garter snalclified fronn Rosen and Schwalbe 1988). Upper histogrann represents snakes fronn populations where bullfrogs were scarce or absent. Lower histogrann represents San Bernardino National Wildlife Refuge sample. Figure 4.— Bullfrog damage to tail of large Mexican garter snake, San Bernardino Na- tional Wildlife Refuge, Cochise County, Ari- zona, 1986. known to have been markedly re- duced by human activities (Rosen 1986). Including captures obtained by all methods, only six Sonoran mud turtles have been found by us on the refuge. All were large adults, and, according to growth ring analysis (see Rosen 1987), all were born prior to 1981. In all other populations, ju- veniles comprised over 207o of the sample (Rosen, unpublished data). Bullfrog Diet Stomach contents confirmed the op- portunistic feeding behavior of bull- frogs (table 3). Invertebrates consti- tuted the majority of food items, with the snail, Planorbella tenuis, and in- sects of the orders Coleoptera, Diptera, Hemiptera, Hymenoptera, Odonata and Orthoptera commonly eaten. Arthropods consumed in- cluded adults and larvae of terres- trial, aquatic and flying forms. Vertebrates were found in 14.6 percent of the stomachs that con- tained some food. The most com- monly consumed vertebrates were other frogs, including bullfrogs. At least two species of native fishes, both endangered, were eaten, the Yaqui chub (Gila purpurea) and the Yaqui topminnow (Poeciliopsis oc- cidentalis sonoriensis). Mammal prey included Peromyscus, a Sigmodon and other as yet unidentified small ro- dents. The two reptile food items were a neonate checkered garter snake in a frog from House Pond and a spiny lizard (genus Sceloporus). Not shown in table 3 was a nestling bird, thought to be a red-winged blackbird, Agelaius tricolor, found in the stomach of a subadult bullfrog (100 mm SVL). Bullfrog Density Using the numbers of bullfrogs re- moved from North Pond (table 4), we can estimate density and bio- 169 mass. After removing 74 adult bull- frogs in spring 1987, we estimated 5 adults remained. Including the small area of marsh north of the levee, there was 0.11 ha of habitat for this population, giving a minimum den- sity estimate of 718 adults/ha. Mean weight for all frogs removed in the spring 1987 census at North Pond was 217.1 g, yielding a total biomass of 23.7 kg, or 215.5 kg/ha. Excluded from this biomass estimate were re- maining adults, and numerous juve- niles that were not hunted. These es- timates are conservative since we had already removed 51 adults and 23 juveniles during fall 1986, before we had determined the most effec- tive means of removing the frogs. The fall 1987 census at North Pond reflects thorough removal the previ- ous spring, with only about 10 frogs either maturing into adults or immi- grating between May 24 and Septem- ber 5, 1987. We estimated that 4-6 adults remained in North Pond at the end of our 1987 collecting. Because of extremely cool, windy weather dur- ing the spring 1988 trip, we were un- able to collect bullfrogs effectively during the last night and left an esti- mated 15-20 adults. A total of 552 bullfrogs has been removed from SBNWR as of June 1988 (tables 4-5), including 358 of adult size, from a total area of 2.4 ha of open water. We estimate that take to represent 55-80% of the adult bull- frogs on the refuge at that time. Preliminary Experimer^tal Results Leopard frogs bred successfully at the spring source in central Black Draw in early 1987, a time of unusu- ally good rainfall. This area was vir- tually devoid of bullfrogs because it is open enough for predators and re- source managers to kill all or almost all adults. In May 1987, leopard frog tadpoles and juveniles were moder- ately abundant from the spring to the northernmost reach of cienega- stream and dense sapling thicket. where they were replaced by bull- frogs. The first confirmation of leop- ard frogs in North Pond was five found in bullfrog stomachs in May 1987. No noticeable further increase in leopard frog numbers or distribu- tion was observed in May 1988. The first juvenile Mexican garter snake on the refuge during this study was recorded in fall 1987. The cap- ture rate of garter snakes on the ref- uge doubled between May and Sep- tember 1987 following bullfrog re- moval (table 2). Extremely cold. windy weather on the May 1988 trip greatly depressed reptile activity. Thus, the 2.0 garter snakes captured per day (table 2) may reflect a de- crease in activity rather than a de- crease in the numbers of garter snakes on the refuge. Discussion Distributional and natural historical data from southeastern Arizona pro- vide prima facie evidence that bull- Table 3 -Stomach contents of aduit (>120 mm snout-vent length) bullfrogs. San Bernardino National Wildlife Refuge. Arizona. Sampling date Prey type 30 Aug- 1 Sep 86 5-6 Sep .87 22-24 May 87 29-30 May 88 Total Amphibians Bullfrogs Tadpoles Juveniles Leopard frogs Juveniles Unknown anurans Fishes Yaqui chub Yaqui topminnovv Unidentified I Mammals Reptiles Invertebrates Detritus Empty stomachs Total food items No. frogs dissected r sidered to be adults. Sampling period Adult males Fall 1986 Spring 1987 Fall 1987 Spring 1988 ^ Totals 33 43 14 17 107 Adult females Total juveniles Total removed 18 31 1 15 23 35 13 48 74 109 28 80 65 119 291 170 frogs play a causative role4n popula- tion decline and disappearance of native wetland amphibians and rep- tiles (table 1; Results). For Mexican garter snakes, this evidence is bol- stered by data on population struc- ture (fig. 3) and by observations of injuries caused by bullfrogs (fig. 4; Rosen and Schwalbe 1988). That bullfrogs are predatory gen- eralists has been thoroughly docu- mented (see extensive review of bull- frog foods in Bury and Whelan 1984). In Arizona alone, bullfrogs have con- sumed such vertebrate prey as a nestling bird, young muskrat (On- datra zibethicus), cotton rat (Sigmo- don), softshell turtle (Trionyx spinif- erus), spiny lizard (Sceloporus), kingsnake (Lampropeltis getulus), sev- eral species of fish and frogs, garter snakes, even a rattlesnake (Crotalus atrox) (fig. 1, table 3; Clarkson and deVos 1986). To our knowledge, in southeastern Arizona, the only place where bull- frogs abound and where leopard frogs and Mexican garter snakes also still occur, albeit rarely, is SBNVVR. We beheve the native species persist there because the extent and diver- sity of aquatic habitats is greater than elsewhere in the region. Specifically, the relatively sparse vegetation and absence of deep pools at the spring source area in central Black Draw has remained largely free of adult bull- frogs. This is where leopard frogs have bred and where the smallest Mexican garter snakes have been found. We believe the reason only five leopard frogs and one garter snake were found in bullfrog stomachs is due to already severe reduction of leopard frog and garter snake popu- lations. The same reasoning may ap- ply to the absence of hatchling Sono- ran mud turtles in bullfrog stomachs. The bullfrog density at North Pond (SBNWR) was quite high for Arizona populations, although not necessarily high for other parts of its range (Currie and Bellis 1969). Such a density is equalled and possibly ex- ceeded at Arivaca, Pima County, Ari- zona, where both leopard frogs and Mexican garter snakes have been ex- tirpated or become extremely rare (Rosen and Schwalbe 1988). Concen- trations of bullfrogs similar to that in lower Black Draw have only been seen in comparable habitat in por- tions of one cienega in the San Ra- phael grasslands of Santa Cruz County. Abundances comparable to those in House Pond occur at a gravel mine south of Arizona High- way 90 on the San Pedro River, Co- chise County; at Page Springs, Yavapai Count}'; and possibly at Parker Canyon Lake, Cochise and Santa Cruz counties and Potrero Canyon marsh, eight kilometers north of Nogales, Santa Cruz County. r Table 5.— Bullfrog removals from aquatic habitats other than North Pond, San Bernardino National Wildlife Refuge, Arizona. Individuals > 120 mm snout-vent length are considered adults. Adult Adult Juveniles Total Locality Date males females removed Twin Pond Fall 86 2 2 1 5 Tula Pond Spring 87 2 3 3 8 House Pond Spring 87 35 42 34 111 Black Draw Spring 87 32 25 15 72 Tuie Pond Spring 88 0 0 3 3 House Pond Spring 88 9 10 11 30 Black Draw Spring 88 6 18 8 32 Totals 86 100 75 261 At Potrero Canyon marsh, Mexi- can garter snakes have disappeared and checkered garter snakes are abundant. In the preceding three lo- calities, checkered garter snakes are absent, and Mexican garter snakes persist in low numbers. Both garter snakes occur along the San Pedro River but neither utilize the gravel pit pond (Rosen and Schwalbe 1988, Rosen personal observations). Natural cienega-streams, includ- ing Turkey and O'Donnell Creeks, where bullfrogs are absent, and Cienega Creek, where they are rare, have high densities of Mexican garter snakes and include many juveniles and young adults. One spring fed pond north of Canelo Hills, which is structurally and vegetatively similar to North Pond, contained about 95 Mexican garter snakes at a density near 1055 individuals/ha, and yielded an average of 5.4 snakes per trapping day (Rosen and Schwalbe 1988). In contrast, only seven garter snakes have been trapped on SB- NWR in fifteen days of similar trap- ping. Central Black Draw would ordi- narily be regarded as relatively poor habitat for Mexican garter snakes, because the vegetative cover is too thin, particularly at the water's edge. The abundance of Mexican garter snakes there and the regular occur- rence of checkered garter snakes at North Pond display an inversion of the usual habitat preferences of the two species in Arizona. In competi- tion, in a broad sense, with Mexican garter snakes, checkered garter snakes may be favored by the pres- ence of bullfrogs because they are less aquatic and hence less affected by the increased predation pressure. Hayes and Jennings (1986) argued that predation by introduced bull- frogs was not a compelling hypothe- sis to explain population declines of native ranid frogs in western North America. They suggest that preda- tion by introduced fish, mainly cen- trarchids, is a more promising hy- pothesis. In southeastern Arizona we 171 found that bullfrogs have invaded a greater variety of wetland environ- ments than exotic predatory fish, and, in some instances, have achieved population densities suffi- cient to impact the native herpe- tofauna. While we do suspect that introduced fish impact native wet- land herpetofaunas in Arizona (see Rosen and Schwalbe 1988), our data for the southeastern portion of the state compellingly incriminate the bullfrog. Our approach is to attempt to manage or eliminate bullfrogs from selected areas. It is principally in- tended to develop practical manage- ment techniques for controlling bull- frogs, but should also provide an ex- perimental test of the bullfrog preda- tion hypothesis. Effective January 1, 1988, the Ari- zona Game and Fish Commission opened the season year round and set an unlimited bag and possession limit on dead bullfrogs statewide ex- cept for La Paz, Mohave, and Yuma counties (Arizona Game and Fish Commission 1988). The stipulation of unlimited possession of dead frogs was to decrease the likelihood of ac- cidental or intentional release of bull- frogs into new habitats. The new regulations will make it easier for agencies, organizations and individu- als to put pressure on bullfrog popu- lations in specific areas in favor of native species. No data exist to show impacts of bullfrogs on native species in the three western counties, so they have retained a July 1 to November 30 sea- son with a bag and possession limit of 12 per day or in possession live or dead. Because Arizona's amphibian and reptile regulations are reviewed annually, new data can be incorpo- rated into management decisions. Conclusions There is evidence that bullfrogs have negatively impacted populations of native amphibians and reptiles in Arizona. Although some of the trends are encouraging, preliminary data from bullfrog removal exp>eri- ments are inconclusive as to whether or not bullfrog control measures may augment recruitment in lowland leopard frogs, Mexican garter snakes or Sonoran mud turtles. More inten- sive efforts will be required to elimi- nate bullfrogs from even local habi- tats when such habitats are structur- ally complex. Acknowledgments We are thankful to the U.S. Fish and Wildlife Service Office of Endan- gered Species in Albuquerque for funding parts of this study. We thank the U.S. Fish and Wildlife Service and Refuge Manager Ben Robertson in particular for permission to con- duct this research on the refuge, and C.H. Lowe for his information on the history of the herpetofauna on the refuge and elsewhere in the South- west. For enthusiastic assistance in the field we gratefully acknowledge the following: Ron Armstrong, Randy Babb, Howard Berna, Andrew, Cindy and Ted Cordery, Mary Gilbert, Rich Glinski, Dean and Gar- rett Hendrickson, Julia Hoffman, Terry Johnson, Charlie Painter, Bruce Palmer, David Parizek, David Propst, Cathy Schmidt, Adam and Ethan Schwalbe, Berney Swinburne, Ross Timmons and Sabra Tonn. Literature Cited Arizona Game and Fish Commission. 1988. Commission Order 41: Am- phibians. Arizona Game and Fish Department Publication. 1 p. Bury, R. Bruce, C. Kenneth Dodd, Jr., and Gary M. Fellers. 1980. Conser- vation of the amphibia of the United States: a review. U.S. De- partment of the Interior, Fish and Wildlife Service, Resource Publica- tion 134, 34 p. Washington, D.C. Bury, R. Bruce, and Roger A. Luck- enbach. 1976. Introduced amphibi- ans and reptiles in California. Bio- logical Conservation 10:1-14. Bury, R. Bruce, and Jill A. Whelan. 1984. Ecology and management of the bullfrog. U.S. Department of the Interior, Fish and Wildlife Service, Resource Publication 155, 23 p. Washington, D.C. Clarkson, Robert W., and James C. deVos, Jr. 1986. The bullfrog. Ram catesbeiam Shaw, in the Lower Colorado River, Arizona-Califor- nia. Journal of Herpetology 20:42- 49. Conant, Roger. 1975. A field guide to reptiles and amphibians of eastern and central North America. 429 p. Houghton Mifflin Company, Bos- ton. Conant, Roger. 1977. Semiaquatic reptiles and amphibians of the Chihuahuan Desert and their rela- tionships to drainage patterns of the region, p. 455-491. In Transac- tions of the symposium on the bio- logical resources of the Chihua- huan Desert region. United States and Mexico. R.H. Wauer and D.H. Riskind, editors. U.S. National Park Service, Transaction Proceed- ings Series 3, 658 p. Currie, W., and E. D. Bellis. 1969. Home range and movements of the bullfrog. Ram catesbeiam Shaw, in an Ontario pond. Copeia 1969:688-692. Dowe, Brian John. 1979. The effect of time of oviposition and microenvi- ronment on growth of larval bullfrogs (Ram catesbeiam) in Ari- zona. Unpubl. masters thesis, 55 p. Arizona State University, Tempe. Fritts, Thomas H., Randy D. Jen- nings, and Norman J. Scott, Jr. 1984. A review of the leopard frogs of New Mexico. Unpubl. re- port to New Mexico Department of Game and Fish (Santa Fe, New Mexico). 157 p. Hayes, Marc P., and Mark R. Jen- nings. 1986. Decline of ranid frog species in western North America: Are bullfrogs (Ram catesbeiam) 172 ogy 20:490-509. Hendrickson, Dean A., and W.L. Minckley. 1984. Cienegas — vanish- ing climax communities of the American Southwest. Desert Plants 6:130-175. Hulse, Arthur C. 1974. An autecol- ogical study of Kinosternon sonori- ense Lcconte (Chelonia: Kinoster- nidae). Unpubl. doctoral disserta- tion, 105 p. Arizona State Univer- sity, Tempe. Jameson, David L. 1956. Growth, dis- persal and survival of the Pacific treefrog. Copeia 1956:25-29. Lanning, Dirk V. 1981. The verte- brates of San Bernardino Ranch, Cochise County, Arizona. Unpubl. report from the Arizona Natural Heritage Program (Tucson, Ari- zona) to The Nature Conservancy and the U.S. Fish and Wildlife Service, Albuquerque, New Mex- ico. 88 p. Moyle, Peter B. 1973. Effects of intro- duced bullfrogs, Rana catesbeiana, on the native frogs of the San Joaquin Valley, California. Copeia 1973:18-22. Nussbaum, Ronald A., Edmund D. Brodie, Jr., and Robert M. Storm. 1983. Amphibians and reptiles of the Pacific Northwest. 332 p. Northwest Naturalist Books, The University of Idaho Press, Moscow. Rosen, Philip C. 1986. Population de- cline of Sonoran mud turtles at Quitobaquito Springs. Final Re- port to National Park Service, Co- operative Park Service Research Unit, Tucson, Arizona. 24 p. Rosen, Philip C. 1987. Variation of female reproduction among popu- lations of Sonoran mud turtles (Kinosternon sonoriense) in Arizona. M.S. thesis, Arizona State Univer- sity, Tempe. Rosen, Philip C, and Cecil R. Sch- walbe. 1988. Status of the Mexican and narrow-headed garter snakes (Thamnophis eques megalops and Thamnophis rufipunctatus) in Ari- zona. Unpubl. report from Ari- zona Game and Fish Dept. (Phoe- nix, Arizona) to U.S. Fish and Wildlife Service, Albuquerque, New Mexico. 69 p. Rosner, Bernard. 1982. Fundamentals of Biostatistics. 547 p. Duxbury Press, Boston. Sokal, Robert R. and F. James Rohlf. 1981. Biometry, the Principles and Practice of Statistics in Biological Research. 859 p. W.H. Freeman Co., San Francisco. Stebbins, Robert C. 1985. A field guide to western reptiles and am- phibians. 336 p. Houghton Mifflin Company, Boston. Vitt, Laurie J., and Robert D. Ohmart. 1978. Herpetofauna of the lower Colorado River: Davis Dam to the Mexican border. Proceed- ings of the Western Foundation of Vertebrate Zoology 2:35-72. Wright, Albert Hazen, and Anna Al- len Wright. 1949. Handbook of frogs and toads of the United States and Canada. 640 p. Com- stock Publishing Company, New York. 173 Developing Management Guidelines for Snapping Turtles^ Ronald J. Brooks,^ David A. Galbraith,^ E. Graham Nanceklvell/ and Christine A. Bishop^ Abstract.— We examined demographic features of 2 Ontario populations of snapping turtles (Chelydra serpentina) io provide an empirical basis for developing management guidelines. The northern population matured later (18-20 yr) than did the southern populations (<10 yr), and displayed an older age distribution. Long-lived, "bet-hedging" species have low annual reproductive success and are unusually susceptible to exploitation. A preliminary life table is presented for the northern population. Our results indicate that the northern population cannot sustain even minimal levels of exploitation by humans without undergoing a decline in numbers. In general, turtles have not been a major concern of wildlife managers in North America, and in many juris- dictions they are given little or no protection. They are perceived to have limited ecological, commercial, aesthetic or recreational value, and because they are usually cryptic and slow moving they are uninteresting to most people. Partly for these rea- sons, there have been remarkably few studies of their life history and ecology. In addition, their great lon- gevity makes them difficult to study, except on a long-term basis. Never- theless, turtles are, or should be, of interest to wildlife managers for at least three major reasons. First, they are major components of a variety of both terrestrial and aquatic ecosystems and therefore play significant, though often unrec- ognized roles as carnivores, herbi- vores and scavengers. In both aquatic and terrestrial habitats, the 'Paper presented at symposium. Man- agement of Amphibians, Reptiles, and Small Mammals in North America. (Flag- staff , AZ, July 19-21, 1988.) ^Professor, Department of Zoology, Uni- versity of Guelph, Guelph, Ontario, Can- ada. N1G2W1. ^Graduate Student, Biology Depart- ment, Queen's University, Kingston, Ontario, Canada. K7L 3N6. "Wildlife Technician, Department of Zo- ology. University of Guelph, Guelph, On- tario, Canada. N1G2W1. ^Graduate Student, Biology Depart- ment, York University, Toronto, Ontario, Canada. Standing-crop biomass of turtles is generally much higher than that of any other reptile (Iverson 1982). In aquatic systems, turtle biomass often exceeds that of sympatric endoth- erms by an order of magnitude and is similar to levels reported for fishes (Iverson 1982). Similarly, annual pro- duction of turtles is comparable to that reported in most other verte- brates, although well below levels found in some fishes (Iverson 1982). Many turtles that are especially long- lived may have low annual produc- tivity. This low productivity may be overestimated because of the high standing-crop biomass of turtles. Their life history is markedly differ- ent from those of the birds and mam- mals that typically occupy the atten- tion of wildlife managers. As such, these species represent special prob- lems in conservation and manage- ment. Therefore, turtles should be of interest to managers, because they are important components of a vari- ety of ecological communities and because in many cases their longevity and low annual production relative to standing crop, characteristic of a "bet-hedger" (Obbard 1983) is a life- history strategy that may be highly susceptible to exploitation or to other sources of mortality of adult animals such as unsuitable overwintering conditions or heavily polluted wa- ters. Secondly, managers should have an interest in turtles because many species are harvested for commercial profit, usually as food or for the pet trade (Bergmann 1983, Congdon et al. 1987, Lovisek 1982). There is evi- dence of marked, recent declines in harvests of most turtle species, but this evidence is difficult to quantify because estimates of total stocks do not exist for any turtle species. For snapping turtles, the annual commer- cial catch in Minnesota was esti- mated at 36000-40800 kg or approxi- mately 6000-6800 average-sized adults (Helwig and Hora 1983). In southern Ontario, Lovisek (1982) esti- mated the annual catch of C. ser- pentina to be 30000-50000 kg or 5000- 8300 adults. There is evidence from trappers (J. Bullard pers. comm.) that numbers of this species are a fraction of former numbers over much of their southern range in Ontario, but again no quantitative estimates exist. At present, therefore, it is necessary to measure the impact of harvesting turtles on a local basis (Hogg 1975). Thirdly, snapping turtles may be of interest to managers because they are often regarded as pests or as a danger to human swimmers, or as destructive predators of waterfowl and game fish (Hammer 1969, Kiviat 1980, Pell 1941). In this paper, we review the biol- ogy of snapping turtles in relation to these three areas of potential impor- tance for wildlife managers. We pres- ent demographic characteristics of 2 populations in Ontario, and in addi- tion, we develop a life table for the more northern population of snap- 174 ping turtles which will allow us to predict the impact of different levels of harvesting pressure on this popu- lation. Snapping Turtles in Aquatic Ecosystems Regulation of Population Density There is at present little understand- ing of what factors regulate popula- tions of any turtle species, but it is known that turtles may reach very high densities and high biomass den- sities (Galbraith et al., in press; Iver- son 1982). It seems likely that pri- mary productivity would be the best predictor of variation in numbers of turtles in a habitat. In snapping turtles, population density ranges from 1-75 adult turtles per ha (Gal- braith et al., in press). Density among populations correlates positively with latitude and primary produc- tion levels and negatively with the size of the body of water (Galbraith et al., in press), although data are too sparse to rely heavily on these corre- lations. Other possible factors influ- encing density are predation pres- sure, especially on nests and hatch- lings, climatic influences on egg sur- vival and embryo development, and availability of suitable nesting sites. Again, the role of these factors has not been studied. Annual Energy Budgets No complete energy budget has been determined for any turtle population, although some efforts have been made to estimate critical components of the energy budget (Congdon et al. 1982). Almost all efforts in this area have concentrated on the energy con- tent and cost of the eggs (Congdon and Gibbons 1985, Congdon and Tinkle 1982, Shine 1980) and on the rates of digestion, especially in rela- tion to temperature (Parmenter 1981). Food-Web Connections Snapping turtles are widely regarded as voracious predators, but most studies of their diet indicate that plant material is a major com|X)nent of their food (Alexander 1943, Ham- mer 1972, Pell 1941). Hammer (1972) found that plants made up the major- ity of the diet of snapping turtles in a North Dakota marsh. In Connecticut, fish (mostly nongame species) and aquatic plants were of equal impor- tance and birds made up only a small fraction of the diet (Alexander 1943). In Maine, snapping turtles ate signifi- cant numbers of ducklings in local areas where both turtles and water- fowl were common, but widespread control of turtles was not recom- mended (Coulter 1957). Lagler (1943), working in Michigan, con- cluded that snapping turtles had minimal impact on waterfowl and pan fish and subsisted primarily on plant material and invertebrates. In general then, snapping turtle preda- tion on waterfowl or game and sport fish poses no serious problem to these valuable species except perhaps in local situations where numbers of turtles may be very high and the turtles have easy access to young wa- terfowl. Adult snapping turtles are largely immune to predation other than by humans over most of their range. A wide diversity of predators prey on snapping turtle eggs (foxes, skunks, raccoons) and hatchlings (herons, large fish), and mortality is very high during these stages. Rationale for \he Development of Life Tables The demography of populations of freshwater turtles under exploitation has not been extensively studied. Some reports have cited large catches being removed from specific loca- tions with apparently little impact on remaining numbers in the short term (e.g. Hogg 1975) but no study has followed an exploited population in detail for any length of time. It is nec- essary, therefore, to infer the effect of harvesting on populations using demographic parameters of unex- ploited populations under long-term study. This paper describes 2 snap- ping turtle populations in Ontario, Canada and presents a life table for one of these populations. Study Areas Lake Sasajewun, Algonquin Provincial Park, Ontario The Ontario Ministry of Natural Re- sources Wildlife Research Area (W.R.A. 45'35' N, 78'30'W, mean an- nual temperature 4.4 'C), is located in the central area of Algonquin Provin- cial Park, in a region of mixed forest last logged in the 1930s. The snap- ping turtles inhabiting the lakes and streams running through the W.R.A. have been studied since 1972. Each year, adult female turtles are cap- tured after nesting and both males and females are captured using baited hoop traps. Of the approxi- mately 185 tagged snapping turtles in the watershed of the North Mada- waska River, about 100 are recap- tured each year. Approximately 70 nests of known females are located each year. Snapping turtles are the largest aquatic vertebrate in the W.R.A., with the exception of beavers (Castor canadensis) and occasional river otters (Lutra canadensis). The only other spe- cies of turtle in this watershed is the midland painted turtle ( Chrysemys picta marginata), present in very small numbers (< 10). The density of the W.R.A. snapping turtle population is approximately 1.5 adults/ha in lakes (Galbraith et al., in press). The study area and the snapping turtle popula- tion have been described extensively elsewhere (Galbraith and Brooks 1987; Galbraith et al. 1987, in press; Obbard 1983). 175 Royal Botanical Gardens, Hamilton, Ontario The Royal Botanical Gardens (R.B.G.) consist of approximately 700 ha of woodlands and waterways within the metropolitan Hamilton area (43'17'N, 79'53'W; mean annual tem- perature 9.8'C). This study area and the snapping turtle population in the R.B.G. have been described previ- ously (Galbraith et al., in press). We have captured, tagged, and released adult and juvenile snapping turtles in this watershed since 1984. In addi- tion to snapping turtles, map turtles (Malaclemys geographica) and painted turtles are common aquatic cheloni- ans in this system. The painted turtle is at least as common as the snapping turtle. The turtles inhabit a highly pro- ductive, eutrophic waterway which is artificially enriched by effluent from a sewage treatment plant. West Pond (9.8 ha), where our trapping has taken place, also connects with heavily-polluted Hamilton Harbour. Despite the contaminants, this popu- lation exhibits one of the highest den- sities yet reported for this species, approximately 60-70 adults/ ha (Gal- braith et al., in press). Methods and Results Life Tables Two approaches are commonly taken in preparing life tables. Static or ver- tical life tables are prepared by deriv- ing mortality rates from the observed population age structure. Cohort- specific, or horizontal life tables are prepared by following a specific co- hort and observing age-specific mor- tality rates throughout life (Deevy 1947). At present, only static life tables can be prepared for snapping turtle populations, because individ- ual cohorts cannot be followed effec- tively in these animals which may have a maximum longevity of over a century (Galbraith and Brooks 1987). Therefore, we will only consider static life tables. Life-Table Parameters for Algonquin Park (W.R.A.) Snapping turtles experience large fluctuations in annual reproductive success (Obbard 1983). In the W.R.A. population, for example, most years do not produce any emergent hatch- lings (R.J. Brooks, unpubl. data) whereas occasional years may pro- duce large numbers of hatchlings. This highly stochastic survivorship throws some doubt on the utility of static life tables, because age curves could be highly biased by errors due to irregular recruitment. Therefore, we will use an average mark-recap- ture survivorship rate (Galbraith and Brooks 1987) for all adult females for the construction of the life table. Several critical pieces of informa- tion have never been obtained for any snapping turtle population. For example, no estimate of survivorship of hatchlings or juveniles has ever been published. A crude estimate of this rate can be obtained by assuming that the number of turtles recruited per year into the population is fairly represented by the average recruit- ment rate, and that the number of eggs being produced per year has not varied greatly between the years when recruits were initially pro- duced (i.e. as eggs) and the present time. In the W.R.A. population, on average, one new nesting female is captured per year on nesting sites used by approximately 85 other fe- males. The mean clutch size of 34 eggs once per year gives an annual egg production of 2890 eggs. Assum- ing half these eggs produce females, the net survivorship across all age classes (including eggs) until age at first nesting (approximately 19 yr, (Galbraith 1986)) is therefore 1/1445 (0.000692). In the W.R.A. population, Obbard (1983) observed a mean rate of emer- gence of hatchlings from eggs of 0.0635, averaged over 142 nests in 5 yr. Taking this into account, in addi- tion to the adult recruitment rate of one mature female per year, the probability of mortality between hatching and maturity for females in this population is 99.17%. Average annual juvenile survivorship from this estimate is therefore 0.7541 from hatching to 19 yr (table 1). High rates of statistical errors within age estimates of individual turtles (Galbraith 1986) make docu- mentation of horizontal rates of age- specific changes in fecundity unreli- able, and therefore we have con- structed our life table using mean clutch size for all age classes. Net fe- cundity, however, is a function of both clutch size and clutch fre- quency. Obbard (1983) estimated that 72.1% of adult females, on average, lay a clutch each year in this popula- tion. Mean annual egg production is therefore 24.514 eggs per female (mean clutch size is 34 eggs). For the purposes of a life table, the female turtles are considered as producing only female offspring. It is also neces- sary, therefore, to consider the effects of biases in hatchling sex ratios. Snapping turtles experience environ- mental sex determination, whereby incubation temperature during the middle third of the incubation period determines offspring sex (Yntema 1976). Between 1981 and 1985, the mean hatchling sex ratio of naturally incubated nests in the W.R.A. was 66% female (R.J. Brooks, unpubl. data). Therefore, each female turUe, on average, produces 16.18 female- destined embryos per nesting season. Although snapping turtles are long- lived, the life table for female snap- ping turtles in the W.R.A. suggests that they do not reproduce enough to sustain the population (table 1). Life-Table Parameters for the Royal Botanical Gardens (R.B.G.) Although data are inadequate to con- struct a meaningful life table for 176 snapping turtles from the R.B.G., some population parameters are known. For example, females in the very large snapping turtle population in the R.B.G. appear to nest for the first time at 10 yr of age (RJ. Brooks, unpubl. data), and the mean clutch size in the R.B.G. population between 1985 and 1987 was 45 eggs. The rate of mortality in this population is likely higher than in the Algonquin population, because numerous dead turtles are found each year (C.A. Bishop, unpubl. data). Essential but currently unavailable information from the R.B.G. population includes Table l.—Ufe table for female snapping turtles In Algonquin Park (W.R.A.), Ontario, Canada. Year class a ' X 1 = X ^x X ml, X X Zm.l * X X n u 1007 A 1 nnn 1 1 .uooo 9 01 '^zL'^ 0470 O oo.ooo .UOO 1 *4 CI o^n 0979 C. '^0 17ft 090^^ . / 0*4 1 A O 90 '^i4A OT^"^ .u I 7^41 7 99 9ft9 01 17 o o iA on A .uuoo 7c; /ll 0 y 19 A7'^ OOAA i^Ay , / 0*4 1 in 0 ^^^7 y . oo / 00 '^O 7'^dl . / 0*4 1 1 1 1 9nft / .zuo 00 "^A l^A^ . 1 1 009 A 7^41 1 o A noo 0091 . /C^ t 15 3.091 .0316 .7541 16 2.331 ,0012 .7541 17 1.758 .0009 .7541 18 1.326 .0007 .7541 19 1.000 .000524 .9660 16.18 .00848 0.00848 20 .000506 .9660 16.18 .00819 0.0167 21 .000489 .9660 16.18 .00791 0.0246 22 .000472 .9660 16.18 .00764 0.0322 23 .000456 .9660 16.18 .00738 0.0396 24 .000441 .9660 16.18 .00714 0.0468 25 .000426 .9660 16.18 .00689 0.0536 30 .000358 .9660 16.18 .03107 0.0847 35 .000301 .9660 16.18 .02615 0.1109 40 .000253 .9660 16.18 .02199 0.1329 50 .000179 .9660 16.18 .03404 0.1633 60 .000127 .9660 16.18 .02409 0.1873 70 .000090 .9660 16.18 .01705 0.1990 80 .000064 .9660 16,18 .01209 0.2111 90 .000045 .9660 16,18 .00853 0.2196 100 .000032 .9660 16.18 .00730 0.2269 ' = numbers of individuals. ax - probability of survival from year class 0 to year class x. ^n, = probability of survival from year class x to year class x+ 1 . = net fecundity at year class x (female-destined embryos produced). *Z//n^ = sum of all reproduction from year class 0 to year class x. equals Ro. total lifetime reproduction, when xisat its maximum. long-term estimates of emergence rates of hatchlings or of adult survi- vorship, annual nesting frequency, and primary sex ratio. Life-Table Implications for Management Guidelines Clearly, exploitation of a population similar to that in Algonquin Park would quickly reduce numbers be- low any chance of recovery by repro- duction within that population. In formulating our life table for the W.R.A., we have had to make several assumptions. The most important concerns our estimate of the rate of survival of hatchlings and juveniles. A comparison between the 2 populations indicates that the advan- tages in the R.B.G. population of hav- ing a larger clutch size than the more northern population and being able to initiate nesting almost 10 yr before the W.R.A. population may be tem- p>ered by overestimating adult survi- vorship in the R.B.G. population. Consequently, lifetime reproduction may not be as high as one might pre- dict. These comparisons must be im- proved by direct observation of sur- vival in the critical juvenile years, and by following individuals of known age throughout life, in a vari- ety of populations. Considerable variation in popula- tion characteristics exists between these 2 populations located about 280 km apart. Trapping guidelines appli- cable to the R.B.G. population may not be suitable to the population in the W.R.A. Regardless, neither could likely tolerate harvests of more than 10% of the adult population. Management Practices to Increase Yields of Snapping Turtles It is evident that unregulated har- vesting of adult snapping turtles will rapidly decrease population sizes, because adult turtles are normally 177 subject to very low rates of mortality (Galbraith and Brooks 1987). Two strategies are possible to increase harvestable numbers of turtles. First, practical experience with sea turtle farming has shown that large numbers of eggs can be incubated under artificial or protected condi- tions (Mrosovsky and Yntema 1980), although care must be taken to incu- bate the eggs at a selection of tem- peratures which will produce a bal- anced sex ratio. Similar propagation of snapping turtles should result in increased numbers of juveniles in populations where adult numbers are not density-dependent. Secondly, enrichment of the envi- ronment could provide faster growth rates for these poikilotherms. In- creases in available protein will probably result in an increase in growth rates of individuals and in- creases in adult carrying capacities (MacCulloch and Secoy 1983). Organochloride Contaminants and Hunnan Consumption Long-lived bottom-dwellers can ac- cumulate high levels of environ- mental toxins, and snapping turtles have been found to carry very high loads of PCBs of various forms (Bryan et al. 1987a). Several studies have considered the way in which PCBs accumulate and in which tis- sues, and snapping turtles are now being employed as biomonitors for organochlorides in some studies (C.A. Bishop et al., unpubl. data). Bryan et al. (1987) demonstrated that local levels of pollutants mark- edly affected the levels of organo- chloride toxins in snapping turtle tis- sues. Tissue-specific accumulation of PCBs is not random in snapping turtles, but is a function of lipopro- tein content of the tissue and the high lipoprotein solubility of the toxins. Especially high concentrations (as high as 1600 ppm PCB in turtles from polluted locations) are found in fat bodies, brain, and testes. However, Bryan et al. (1987) indicated that toxic PCB congeners did not remain in the large fat reserves of female turtles, as some had suggested, but were passed on in bulk to the egg yolks. It is necessary, therefore, to test tissue or egg samples to ensure that turtles being harvested for human consumption are not loaded to a dan- gerous degree with organochloride contaminants. Management of Snapping Turtles as Predators Several studies have considered the impact of snapping turtles on water- fowl populations (Alexander 1943, Hammer 1972, Lagler 1943). Highly- productive bodies of water present ideal habitat for waterfowl and for turtles. Destroying turtle nesting locations may not reduce local populations of snapping turtles, because females may migrate several kilometers be- tween their usual home range and their nesting sites (Obbard 1977). In addition, such habitat interference will remove nesting opportunities for other turtle species. Reduction in numbers of adult snapping turtles through trapping will rapidly deplete isolated popula- tions and should reduce risks to prey species. However, if turtles can emi- grate into the management area, then the expected long-term effect of cull- ing adults will not be realized be- cause the population can increase from these new adult immigrants. Acknowledgments We are grateful to the Ontario Minis- try of Natural Resources and to D. Strickland for their support and for granting permission to conduct re- search in Algonquin Provincial Park, and to the Royal Botanical Gardens, Hamilton, Ontario, for permission to work in Cootes Paradise. We thank C. Bell, M.L. Bobyn, M. Fruetel, J. Hughes, K. Lampman, J. Lay field, and S. Plourde for field and technical assistance, and K. Kovacs and S. In- nes for computational help. This study has been supported by Natural Sciences and Engineering Research Council Canada Grant A5990, and an Ontario Ministry of Natural Re- sources Renewable Resource Pro- gram Grant to Ronald J. Brooks. Literature Cited Alexander, Maurice M. 1943. Food habits of the snapping turtle in Connecticut. Journal of Wildlife Management 7:278-282. Bergmann, Brian. 1983. Carapace ca- pers: uncontrolled trapping threat- ens Ontario's snapping turtles. Equinox 2(5):127. Bryan, A. M., P. G. Olafsson, and W. B. Stone. 1987. The disposition of low and high environmental con- centrations of PCB's in snapping turtle tissues. Bulletin of Environ- mental Contamination and Toxi- cology 38:1000-1005. Congdon, Justin D., Gary L. Breiten- bach, Richard C. van Loben Sels, and Donald W. Tinkle. 1987. Re- production and nesting ecology of snapping turtles (Chelydra ser- pentina) in southeastern Michigan. Herpetologica 43:39-54. Congdon, Justin D., A. E. Dunham, and Donald W. Tinkle. 1982. En- ergy budgets and life histories of reptiles, p. 233-271. In C. Gans and F. H. Rough, editors. Biology of the Reptilia. 13. Physiology D- Physiological Ecology. Academic Press. New York. Congdon, Justin D., and J. Whitfield Gibbons. 1985. Egg components and reproductive characteristics of turtles: relationships to body size. Herpetologica 41: 194-204. Congdon, Justin D., and Donald W. Tinkle. 1982. Reproductive ener- getics of the painted turtle (Chrysemys picta). Herpetologica 38:228-237. 178 Coulter, Malcolm W. 1957. Predation by snapping turtles upon aquatic birds in marine marshes. Journal of Wildlife Manage. 21(1):17-21. Deevy, Edward S., Jr. 1947. Life tables for natural populations of animals. Quarterly Review of Biol- ogy 22:283-314. Galbraith, David A. 1986. Age esti- mates, survivorship, growth, and maturity of female Chelydra ser- pentina Linnaeus in Algonquin Provincial Park, Ontario. M.Sc. Thesis, University of Guelph, Guelph, Ontario. Galbraith, David A., and Ronald J. Brooks. 1987. Survivorship of adult females in a northern popu- lation of common snapping turtles, Chelydra serpentina. Cana- dian Journal of Zoology 65:1581- 1586. Galbraith, David A., Ronald J. Brooks, and Martyn E. Obbard. In Press. Female age and body size at the onset of reproduction in a northern population of snapping turtles (Chelydra serpentina). Cana- dian Journal of Zoology. Galbraith, David A., Mark W. Chan- dler, and Ronald J. Brooks. 1987. The fine structure of home ranges of male Chelydra serpentina: are male snapping turtles territorial? Canadian Journal of Zoology 65:2623-2629. Hammer, Donald A. 1969. Parame- ters of a marsh snapping turtle population: Lacreek Refuge, South Dakota. Journal of Wildlife Man- agement 33: 995-1005. Hammer, Donald A. 1972. Ecological relations of waterfowl and snap- ping turtle populations. Ph.D. Dis- sertation, Utah State University, Salt Lake City, Utah. Helwig, D. D., and M. E. Hora. 1983. Polychlorinated biphenyl, mercury and cadmium concentrations in Minnesota snapping turtles. Bulle- tin of Environmental Contamina- tion and Toxicology 30:186-190. Hogg, D. M. 1975. The snapping turtles of Wye marsh. Ontario Fish and Wildlife Review 14(2):16-20. Iverson, John B. 1977. Reproduction in freshwater and terrestrial turtles of north Rorida. Herpetol- ogica 33:205-212. Iverson, John B. 1982. Biomass in turtle populations: a neglected subject. Oecologia 55:69-76. Kiviat, Erik. 1980. A Hudson River tidemarsh snapping turtle popula- tion. Transactions of the Northeast SecHon, Wildlife Society 37:158- 168. Lagler, Karl F. 1943. Food habits and economic relations of the turtles of Michigan with special reference to fish management. 29:257-312. Lovisek, Jim. 1982. An investigation of the harvesting of turtles in On- tario. Nongame Program Wildlife Branch, Ontario Ministry of Natu- ral Resources. MacCulloch, Ross D., and D. M. Secoy. 1983. Demography, growth, and food of western painted turtles, Chrysemys picta bellii (Gray), from southern Sas- katchewan. Canadian Journal of Zoology 61:1449-1509. Mrosovsky, N., and C. L. Yntema. 1980. Temperature dependence of sexual differentiation in sea turtles: implications for conserva- tion practices. Biological Conser- vation 18:271-280. Obbard, Martyn E. 1977. A radio- telemetry and tagging study of activity in the common snapping turtle, Chelydra serpentina. M.Sc. Thesis, University of Guelph, Guelph, Ontario. Obbard, Martyn E. 1983. Population ecology of the common snapping turtle, Chelydra serpentina, in north- central Ontario. Ph.D. dissertation. University of Guelph, Guelph, On- tario. Parmenter, Robert R. 1981. Digestive turnover rates in freshwater turtles: the influence of tempera- ture and body size. Comparative Biochemistry and Physiology 70(A):235-238. Pell, Stuyvesant Morris. 1941. Notes on the habits of the common snap- ping turtle, Chelydra serpentina (Linn.), in central New York. M.Sc. Thesis, Cornell University, Ithaca, New York. Shine, Richard. 1980. "Costs" of re- production in reptiles. Oecologia 46:92-100. Yntema, C. L. 1976. Effects of incuba- tion temperatures on sexual differ- entiation in the turtle, Chelydra ser- pentina. Journal of Morphology 150:453-462. 179 Spatial Distribution of Desert Tortoises (Gopherus agassizii) at Twentynine Palms, California: Implications for Relocations^ Ronald J. Baxter^ Abstract.— The spatial distribution of desert tortoises in relation to plant communities was compared against randomness. Tortoise captures (n = 120) and tortoise burrows (n = 160) exhibited non- random distributions across a 1.29 square kilometer study plot at Twentynine Palms, California. Results imply high diversity plant ecotones and communities, and possibly soil characteristics are important in determining tortoise densities. Non- randomness in tortoise populations dictates that relocation sites must include specific vegetational, topographic and edaphic habitats used by the parental populations. The desert tortoise (Gopherus agas- sizii) is a species whose future is un- certain. Increased use of the deserts by man (Luckenbach 1982) has led to the point where the tortoise was offi- cially listed as "threatened" in the state of Utah (Dodd 1980). The U.S. Fish and Wildlife Service stated in 1985 that ''...listing [of the desert tor- toise as a threatened or endangered species] is warranted but precluded by other pending proposals of higher priority" (Federal Register. 50(234): 49868-49870, 1985). In California, the desert tortoise is the official state reptile, and is fully protected under law. The tortoise is also protected in Arizona and Ne- vada. As part of a larger population study (Stewart and Baxter 1987) at the Twentynine Palnns Marine Corps Air Ground Combat Center (MCAGCC), the spatial distributions of tortoise captures and burrows were analyzed and compared against randomly generated distributions. Questions asked were: (1) Are tor- toise captures and burrows ran- domly located across the landscape 'Paper presented at symposium. Man- agement of Amphibians, Reptiles, and Small Mammals in Nortt) America. (Flag- staff, AZ, July 19-21, 1988). 'Ronald J. Baxter received his master's degree in biology for working on the desert tortoise while at California State Polytech- nic University, Pomona. He is currently com- pleting his doctorate at the Department of Biological Sciences. Northern Arizona Uni- versity. Flagstaff. AZ, 8601 1-5640. and /or are they associated with cer- tain habitat types or site characteris- tics, and if so, (2) what implications do these distributions have for future management decisions? Mettiods Twentynine Palms MCAGCC is lo- cated approximately 5 kilometers north of Twentynine Palms, San Ber- nardino County, California, in the southwestern extreme of the Mojave Desert. All fieldwork was performed in the Sand Hill Training Area which is in the southwest corner of the MCAGCC. Elevations ranged from 865 meters atop Sand Hill to about 730 meters in the bottom of Surprise Springs wash. Data were collected Monday through Friday, 14 April through 18 July, 1986. Systematic searching methods for tortoises and tortoise burrows were a derivation of procedures described by Berry (1984). A 1.29 square kilo- meter permanent study plot was es- tablished, with its approximate cen- ter being the NE 1 /4, SW 1 /4, NE 1 / 4, of S7, T2N, R7E (San Bernardino Base Meridian) This site offered a wide variety of habitats including washes, sandy basins, rolling hills and alluvial bajadas. The plot was di- vided into 64 equal sized "grids" of 142 meters on a side, with grid cor- ners marked by posts. Grids were searched in parallel belts until the entire plot had been searched twice; once with the belts running north- south, and once with the belts run- ning east-west. The plot was also randomly searched. When an active tortoise was en- countered, it was marked, weighed, sexed, measured and photographed. Each tortoise was assigned a unique number, and marginal scutes were notched with a small triangular file for relatively permanent identifica- tion. The precise location of the cap- ture was noted by its distance (meas- ured by rangefinder) and compass aspect to the nearest grid post. Data collected at each capture site in- cluded plant community, tempera- tures at the ground, 1 centimeter, and 1 meter, cloud cover, wind speed and direction, closest burrow, closest plant, and any unusual be- havior. Precise location of tortoise bur- rows were similarly determined by rangefinder and compass. Data col- lected at each burrow included plant community, distance and identifica- tion of nearest ecotone, distance to nearest wash, distance to nearest Hi- laria rigida, slope aspect and steep- ness, opening compass aspect and position, length, depth, and tunnel characteristics. In this study area, it was difficult to determine if a bur- row high on a slope above a wash was part of the wash "system." Therefore, it was arbitrarily decided to include burrows in the wash plant community only if they were actually found the wash bed. 180 Six visually identified plant com- munities (Latr/ Amdu, Hiri/ Amdu, Mixed, Wash, Sparse Wash and Meadow) were mapped within the study plot, and seven 15-meter line transects (total of 105 meters) were measured which included bare ground as a species. Transects were i:latr/amdu 5: meadow 2:mixed •'.burrow 3:SPARSE WASH ff-mSH NORTH 4:HIRI/AMDU Soo' METERS Figure 1 .—Approximate distribution of piant communities and tortoise burrows across the study piot. See text for expianation of piant community names. randomly located in each of these six communities. Standard transect sta- tistics (density, coverage, frequency, relative density, relative coverage, relative frequency and importance values; Brower and Zar 1984) were computed for each community. Simpson's diversity indices (Simpson 1949) were computed and compared with Student's t-tests (Keefe and Bergerson 1977). Available annuals as well as perennials were used to give the best estimate possible for diversity. In addition, seven soil samples were taken in each commu- nity, and analyzed for soil separates (Brower and Zar 1984) and soil cal- cium (Hach 1983). Finally, nine ''sand scats" were collected during the field work and tested for calcium. A random model for capture and burrow locations was formed by combining a number of statistical tests. First a master map of the plot was constructed from actual field data at a scale of 1:2000. All capture positions, burrows and plant com- munity boundaries were plotted on this map and checked against aerial photographs. The area covered by each plant community was then de- termined by the use of a planimeter. An X-Y scale ranging from 0 to 8 was plotted on the sides of the map, and a list of 328 random numbers was generated by computer. These num- bers were paired, and the pairs be- came the X-Y coordinates of random positions against which observed capture and burrow locations were compared. Distances to the nearest wash and ecotone were determined for these random locations by meas- uring them on the map, and com- pared against observed by Student's t-tests (Zar 1974). Observed capture distances were sometimes combined with previous data recorded in this area (Baxter and Stewart 1986). A lack of habitat preference may be suggested if burrows and captures were found in the same relative abundance as the plant communities. In addition, if the expected plant abundance distribution differed sig- nificantly from random an extrapola- tion of observed distributional char- acteristics could be accomplished. An assumption of this test was that a distribution of randomly generated locations (with randomness con- firmed) produced a random fre- quency distribution. Expected fre- quencies for burrows and captures were generated by multiplying the total number of actual burrows or captures by the percent of the plot encompassed by each plant commu- nity. These values were compared by a goodness-of-fit chi-square test (Zar 1974). In addition, the number of burrows or captures per grid were compared against expected values as derived from the Poisson distribu- tion by a goodness-of-fit test. Results Plant Communities and Soils Vegetation analyses revealed six dis- tinct plant assemblages (table 1; fig. 1). Plant community distributions generally reflected the relief of the plot. The higher, more well-drained hills were dominated by an associa- tion of Larrea tridentata and Ambrosia dumosa, which encompassed plot area the most ("Latr/ Amdu"; table 1) and exhibited relatively high plant diversity. r Table 1 .—Summary of plant community data from plant transects (total transect length = 105 meters). Simpson's Percent Plant No. of No. of diversity of plot community^ species individuals index area Sparse Wash 16 733 0.6069 5.6 Hiri/Amdu 8 501 0.6841 4.0 Mixed 12 349 0,7247 37.2 Latr/ Amdu 11 306 0.7688 50.6 Wash 25 292 0.7914 0.2 Meadow 15 662 0.7497 1.7 Bore Areas 0.7 °See text for explanation of community names. 181 Found on 37.2% of the plot area was the "n\ixed" community that generally occupied intermediate ar- eas between the Latr/ Amdu and ei- ther washes or areas of high Hilaria rigida density. It was characterized by the association of L. tridentata, A. dumosa and H. rigida, and was found most often on the slopes above, and narrow linings next to washes. The edge, or ecotone, of this community with the Latr/ Amdu community is extensively discussed below. A highly diverse plant community was found in the washes (table 1; ap- pendix 1). Such areas not only con- tained these perennial species, but also a significant number of other species found only in this commu- nity, giving it the highest species richness of any community. Small uplifts within wash channels seemed to support a more open type of wash vegetation, "sparse wash." Such areas had many species com- mon to the washes (appendix 1 ), yet much of this community was essen- tially pure stands of the opportunis- tic grass, Schismus barbatus. A community ("Hiri/Amdu") consisting primarily of H. rigida and A. dumosa was located in upland ba- sins where L. tridentata was not found. Such areas were low in habit and diversity, and very sandy. Finally, near the south boundary of the plot, a small "meadow" of mostly Bailey a multiradiata was found. Since no tortoises or tortoise burrows were found there, it was eliminated from further analyses. Bare ground, when treated as a species in transect analyses, had overriding importance values and dominance in all communities (ap- pendix 1). This is often the case in desert environments. Likewise im- portance values of S. barbatus were extremely high in all communities, pointing to the generally disturbed nature of the site. Comparisons of Simpson indices for the communities revealed significant differences (p < 0.05) in diversity for all communities except two. The Latr/ Amdu and wash communities were not signifi- cantly different (p > 0.50) in their di- versity. Soils were found to be somewhat similar in constituency (table 2), each being composed to a large degree of sand. Soil calcium levels (table 3) were shown to differ significantly. No detectable calcium was found in any of the sand scats tested. Tabie 2.— Summary of percent soli separcstes for plant communities. Plant Silt community" Sand (%) Clay Classification Sparse mixed 87 2 11 loamy sand Hiri/Amdu 90 8 2 sand Mixed 85 12 3 loamy sand Latr/ Amdu 70 20 10 sandy loam Wash 81 3 16 sandy loam Meadow 63 3 33 sandy day loam °See text for explanation of community names. r Table 3.— Summary of soil calcium levels and their significance. Plant Mean soil calcium Significantly community*' Cmeq/100 mg soli) different from Latr/Amdu 6.43 Hiri/Amdu <0.006 Wash <0.05 Mixed NS Hiri/Amdu 3.00 Wash N$ Mixed <0.05 Mixed 1.48 Wash NS Wash 0.57 °Se0 text for explanation of community names. ^2'Sampte t-test; corrected for type t errors: N$ = not significant. Table 4.— Distributions and significance of tortoise burrows per grid (Pols- son, n = 64). °P>0.9d ^P>Q.7S J Number of Number of grids Expected values burrows/grid random observed random" observed^ 0 6 6 4.95 5.52 1 11 17 12.67 13.53 2 15 11 16.21 16.58 3 15 14 13.83 13.54 4 10 8 8.85 8.29 5 or more 7 8 4.67 6.30 182 Table 5.— Summary of frequency of tortoise burrows compared to plant community abundance. Plant community*' Sparse Wasti HIrl/ Amdu Mixed Latr/ AmHij Wast! Other % of Diot 5 6 A n 37.2 50.6 0.2 2.4 Random'^ observed expected Observed^ 14 9.2 6 6.6 55 61.0 76 83.0 8 0.3 5 3.9 observed expected 11 8.8 2 6.3 68 58.8 75 80.0 1 0.3 1 3.8 °See text for explanation of community names. ''P<0.00hn=164. ^P>0.25,n= 158. _J Table 6.— Comparison of distance to washes between observed and ran- dom tortoise burrows. Plant Mean distance (m (SEM)) community*" random observed t freedom P All 96.83 101.21 0.424 318 >0.50 communities (6.80) (7.83) Mixed 79.54 68.66 0.821 120 >0.50 (9.75) (8.90) Latr/ Amdu 132.05 145.40 0.845 148 >0.50 (10.07) (12.17) °See text for explanation of community names. J r Table 7.— Comparison of distances to ecotone between observed and ran- dom tortoise burrows. Plant community*' Mean distance (m fSFMD random observed t Degrees of freedom P All 96.83 101.21 0.424 318 >0.50 Latr/ Amdu 33.63 15.21 5.360 137 < 0.0005 (2.65) (1.80) Mixed 38.33 12.18 3.650 65 < 0.0005 (9.09) (1.60) Combined 34.05 13.99 6.493 203 < 0.0005 (3.00) (1.26) ""See text for explarxjtion of community names. / Tortoise Burrov^s A total of 164 tortoise burrows was found on the study plot (fig. 1). Sev- enty-five percent were found under bushes, 14% with the opening under a bush but the tunnel proceeding into an open area, 8% with entrances in the open but the tunnels proceeding under a bush, and 3% entirely in an open area. Thus, almost all of bur- rows (97%) were associated with shrubs. Of these, 71% were associ- ated with L. tridentata, 13% each with H. rigida and A. dumosa, and another 3% with other species. Neither the distribution of ob- served or random burrows differed significantly from the Poisson ex- pected frequencies (table 4). Like- wise, when the distribution of ob- served burrows was compared against the distribution of random burrows, no significant difference was found (chi-square = 2.224; DF = 5; p > 0.50). Thus, when the entire plot area is considered, tortoise bur- rows exhibited a random pattern across the landscape. However, this was a relatively large scale test of burrows per arbitrary unit area, and says nothing about the pattern of tor- toise burrows in relation to plant communities. The abundances of tortoise bur- rows (both observed and random) in each plant community were com- pared against expected frequencies generated by the abundances of the plant communities (table 5). Burrows were sparse in the Hiri/Amdu and wash communities. Observed bur- row frequency distribution did not differ significantly (p > 0.25) from the expected frequency distribution. The observed frequency distribution dif- fered significantly from the random distribution (chi-square = 11.74; DF = 5; p < 0.05), as did the expected dis- tribution (chi square = 158.9; DF = 5; p< 0.001). Mean observed burrow distance to the closest wash was compared to the mean distance from the ran- domly located burrows (table 6). 183 Comparisons for the sparse wash and Hiri/Amdu communities were not done because they would be bio- logically meaningless or had too low a sample size, respectively. For all burrows, and for burrows found in either the Latr/ Amdu or mixed com- munities, no significant differences between random and observed wash distances were detected. Thus, ob- served tortoise burrows were not lo- cated closer to washes than a set of random points predicted. However, examination of the spatial pattern (fig. 1) reveals a lack of burrows deep within Latr/Amdu and Hiri/Amdu areas which were furthest away from any possible wash influence. Past observations seemed to indi- cate a correlation between burrow location and the presence of the edge of the H. rigida distribution (Baxter and Stewart 1986). The approximate distribution of observed burrows to this edge may be seen in figure 1 . Mean edge (ecotone) distance of ob- served burrows was compared to that of random sites (table 7). Highly significant differences in ecotone dis- tances were found in both communi- ties, and also when combined. Thus, burrows were found closer to the ecotone than a set of random points. Tortoise Captures Similar analyses were performed for tortoise capture sites. There were a total of 120 tortoise captures and re- captures of 41 individual tortoises. The observed captures per grid, along with the randomly located cap- ture frequencies (same points used for random burrow sites) were com- pared against expected values de- rived from the Poisson distribution (table 8). Observed capture sites showed a statistically significant de- parture from Poisson expected fre- quencies by the goodness-of-fit test (p < 0.05). Frequencies of capture sites in each plant community were com- pared against expected values gener- ated by community abundance (table 9). Observed distributions for both all captures, and for captures of ac- tive tortoises (those found outside of burrows) differed significantly from expected. These two observed distri- butions did not differ from each other (chi-square = 0.5385; DF = 5; p > 0.99), yet differed significantly from the randomly generated distri- bution (chi-square = 18.957 and 19.556, respectively; DF = 5; p < 0.005). Thus, tortoise captures were not found across the plot in a ran- dom fashion as would be predicted by a set of randomly generated points. Habitat preference for washes was seemingly indicated, as was a lack of preference for Hiri/Amdu areas. These results also gave further support to the non-randomness ex- hibited in the Poisson analyses. To further examine this apparent non-random distribution of capture locations, the mean observed capture distance to washes was compared to that of the randomly located sites (table 10). When all capture sites, or captures within the mixed commu- nity were considered, a significant Table 8.—Distrlbutlons and significance of tortoise captures per grid (Pois- son, n= 64). Number of captures/grid Number of grids random observed Expected values random" observed^ 0 1 2 3 4 or more 6 VI 15 15 17 19 17 15 4 9 4.95 12.67 16.21 13.83 13.59 9.97 18.54 17.23 10.68 5.54 ''P>0.90 '>P< 0,026 Table 9.— Summary of frequt^ncy of captures compared to plant commu- nity abundance. Plant community*" Sparse HIrl/ IViixed Latr/ Wash Other Wash Amdu Amdu % of plot 5.6 4.0 37.2 50.6 0.2 2.4 Random'^ observed (n=164) 14 6 55 76 8 5 expected 9.2 6.6 61.0 83.0 0.3 3.9 Observed (oll)^ observed (n=120) 14 1 33 48 23 1 expected 6.7 4.8 44.6 60.7 0.3 2.9 Observed (active)^ observed (n=81) 9 1 20 32 18 1 expected 4.5 3.3 30.1 41.0 0.2 1.9 °See text for explanation of community names. ""Pk 0.001 184 difference between random and ob- served locations was demonstrated. However, mean distance to washes within Latr/ Amdu sites was not sig- nificantly different from the random set of points, possibly because the Latr/ Amdu communities were gen- erally located further away from washes, as well as the high variation in observed Latr/Amdu distances. These results, along with the results of the community analysis above, seemed to indicate a high degree of tortoise activity near the washes. Distances to the edge of the H. rigida were compared between ran- domly generated and observed cap- ture locations (table 11). Highly sig- nificant differences in mean distances were demonstrated for both the Latr/Amdu community, and for cap- tures found in the mixed and Latr/ Amdu communities combined. Cap- tures within the mixed community alone were not significantly different from randomly generated locations. It seems then that captures, like bur- rows, were generally not found far within Latr/Amdu areas, but tended to be near its edge with the H. rigida distribution (i.e. the mixed commu- nity). Because there was no differ- ence within the nnixed community alone, differences from random for captures within the mixed and Latr/ Amdu communities combined were probably significant due to the higher number of observations within the Latr/Amdu community biasing the sample. Thus, it seems that tortoises tended to stay either near the washes, the mixed commu- nity, or its ecotone with the Latr/ Amdu community, and generally were not going far within the Latr/ Amdu community. Table 10.— Comparison of distance to washes between observed and ran- dom capture locations. Plant community*' Mean distance rSfM)) t Degrees of freedom random observed P All 96,83 71.86 2.189 258 <0.05 communities (6.80) (9.39) Mixed 79.54 44.14 2.081 73 <0.05 (9.75) (10.61) Latr/Amdu 132.05 133.66 0.917 117 > 0.50 (10.07) (15.12) ''See text for explanation of community names. r N Table 1 1 .—Comparison of distances to ecotone between observed and random capture locations. Plant community*' Mean distance (m (SEM)) random observed Degrees of t freedom P Latr/Amdu 32.33 18.59 3.389 114 < 0.001 (9.09) (3.05) Mixed 38.63 21.05 1.595 42 >0.10 (2.65) (4.65) Combined 34.05 13.99 3.485 157 < 0.001 (3.00) (2.53) =See text for explanation of community names. Discussion Since the establishment in 1975 of the Desert Tortoise Council, the amount of literature published on the desert tortoise has been considerable. Oddly enough, only a few papers may be found that attempt to say what exactly makes good tortoise habitat. A paper by Schwartzmann and Ohmart (1978) quantified the fre- quency of use by tortoises in a num- ber of "habitat types." Their study took place in the Picacho Mountains of Arizona's Sonoran Desert, where tortoises are known to frequent rocky hillsides and are absent from valley bottoms (Fritts 1985). Habitat preferences are just the opposite in the Mojave Desert, and thus their re- sults may not be applicable. Like- wise, Walchuck and Devos (1982) studied tortoise habitat, but this was also in the Sonoran Desert of Ari- zona. In a draft report, Weinstein et al. (1986) performed several multivari- ate analyses on the large Bureau of Land Management tortoise database. Several attempts were made to corre- late abundance with habitat charac- teristics. Not only were many of these characteristics derived from the extrapolation of large scale map data, but the best fit analysis was found by designating "corrected sign" of the transects (the dependent variable; not actual population numbers) into arbitrary categories. Indeed, one of the authors (Berry and Nicholson 1984) has shown that roughly one- third of population estimates (7 out of 20 and 4 out of 6) based on sign transects did not agree with intensive plot censuses. Also, Turner et al. (1982) stated that sign transects "...cannot provide the accuracy and precision needed..." In addition, Fritts (1985) stated that such transects are "...subject to error." Thus the accuracy of sign transects are open to serious debate, and al- though the discriminant analysis showed some promise as a method 185 for accessing regional abundances, the nature of the analysis and the underlying assumptions of both the data acquisition and techniques leave much to be desired. When viewed from the larger scale of regional or even plot area, these data seem to indicate that bur- rows were found in a random fash- ion when predicted by burrows per unit area. However, different results may have been obtained by changing the size and shape of the grids. For example, 32 larger rectangular grids may very well have produced differ- ent results than the 64 smaller square grids used in this study. In addition, such an analysis said nothing about distributions in relation to habitat characteristics. Therefore, such a test should be used as a starting point and /or support for other tests, and locally is of limited use by itself for describing ecologically meaningful patterns which may exist. With closer examination, these data also indicate that burrow loca- tions were assembled in a pattern similar to the non-random distribu- tion of plant communities. Within- community examinations revealed patterns of burrow site utilization, and such patterns were strongly non- random. At Sand Hill then, while a majority of burrows were not found in washes, they were often found within easy walking distance to a wash. Very often, burrows were on slopes high above washes, and possi- bly within its area of influence. They were not found far within either the Latr/ Amdu or Hiri/ Amdu commu- nities, but were tied strongly to the edge of these communities with the mixed community. Washes are sometimes cited as being of great importance to tortoise populations (Burge 1978, Hohman 1977, Lowe 1964). However, results of this study indicated that tortoise burrows were not significantly closer to washes than a set of randomly se- lected sites. Burge (1978) found 207 (26%) of 783 burrows and pallets were associated with washes. Of these, 56 (27%) were actually within a wash bed. However, Burge appar- ently eliminated some burrows from the analysis due to their physical characteristics. The discrepancy may be due to the definition used. In this study, wash burrows were defined as such, only if they were actually within the sandy wash bottoms. In this way, burrows which were on wash banks, were counted as being in the plant community of the bank. Burrows located on wash banks, and even further away, may have been associated with the wash, and a re- classification of these burrows may show washes to have a more impor- tant influence in burrow analyses. Examinations of the actual burrow distribution (fig. 1) seemed to indi- cate that they were mostly absent from areas highly isolated from wash influence. The significance of capture loca- tions in relation to the washes also seemed to refute the burrow /wash results. Washes clearly supported a disproportionate amount of activity in relation to their abundance on the plot. Preliminary investigations of tortoise communities near Kramer Junction, San Bernardino County, have also shown tortoises are proba- bly localizing their activities in the vicinity of washes (Baxter, unpub. data). Several things may explain the disproportionate amount of captures in the washes. Greater visibility of tortoises in the washes may be a fac- tor. Utilization of highly diverse plant resources there may also con- tribute to the localization of activity. Finally, washes may simply serve a natural highways for tortoise move- ments. For instance, several relocated tortoises at Kramer Junction abruptly turned and followed trails and washes upon their release (Baxter, unpub. data). Regardless, these data seem to support washes as an impor- tant habitat characteristic for tor- toises at Sand Hill. If this population is representative of other Mojave populations, the importance of washes in p>otential relocation sites will be highly significant in assuring the best chance of survival for the relocatees. Further, impacts to washes may have highly significant impacts on a population if it is local- izing its activities there. These data support the impor- tance of large woody shrubs (i.e., L. tridentata) for successful burrow con- struction at this site. Similar results have been reported by Burge (1978) who found 72% of "cover sites" as- sociated with shrubs. Berry and Turner (1984) found 75% of juvenile burrows associated with bushes. Support for the burrow roofs and added protection from predators are likely reasons for this association. Regardless, the absence of L. triden- tata from the Hiri/Amdu community is probably a major reason for the tortoises not utilizing those areas. Unsuccessful burrow construction by virtue of the sandier soils is another possibility. This latter assumption is supported by the Weinstein et al. (1986) analysis which showed "soil diggibility" as a highly significant regression variable. However, the lack of burrows deep within Latr/ Amdu communi- ties is not explained by the spatial abundance of L. tridentata. The high frequency of burrows and captures point out that something is being sought there by the tortoises. Yet, deep ventures within these areas ap- parently do not provide resources that are unavailable at their edges. Perhaps the higher levels of soil cal- cium found there are being utilized. Tortoises must support a massive, ossified shell, as well as lay eggs, and calcium may be a very important nu- trient. Tortoises have been observed eating dirt (geophagy) and then pro- ducing "sand scats," and calcium levels have been hypothesized as an explanation for this behavior (Sokol 1971). The lack of calcium in the sand scats tested seems to support this hypothesis. In contrast, such deep ventures would take the tortoises away from 186 the distribution of H. rigida, and the frequented and diverse washes. Al- though detailed scat analyses were not performed, field examination of hundreds of scats seemed to suggest that H. rigida is a significant dietary component. Turner and Berry (1986) found H. rigida as a part of the diet of tortoises near Goffs, California. It would seem then that tortoises in this area are exhibiting some char- acteristics similar to "edge" species. That is, tortoise activity is centered on the two communities with the highest vegetational diversity that border extensive areas of H. rigida. Since burrows are closely associated with L. tridentata, they in turn are found primarily along the only highly diverse ecotone of the H. rigida distribution where L. tridentata importance is the highest. This im- portance of H. rigida and L. tridentata is further shown in appendix 1 . The two communities where tortoises were not found (i.e., deep Latr/ Amdu and Hiri/Amdu) each lack one of these species. The assumption that they are focusing on high diver- sity areas is further supported by Weinstein et al. (1968) which shows "food availability" as the single most significant regression variable. Fi- nally, Speake (1986) reports that for the gopher tortoise (G. polyphemus), "Edge habitats or ecotonal areas ap- pear important to tortoises. In each habitat type except oldfields tortoises tended to cluster near the edges. In general, the more edge availability in a given habitat, the higher the tor- toise density." In summary, tortoises utilized the environment at Sand Hill in a mostly non-random fashion. Tortoise cap- tures were spread out between two communities of highly diverse re- sources, with clustering occurring at either edge. Tortoises frequented washes and the ecotonal edge of the Latr/Amdu community, with many found in the intermediate mixed community. Tortoises were not found deep within Latr/Amdu or Hiri/Amdu areas. Burrows were found close to the ecotone of the mixed and Latr/Amdu connmunities. Burrows were not found closer to washes than randomly located bur- rows, although this point is far from clear. Burrows were located close to the one highly diverse edge of tor- toise activity area where the impor- tance of L. tridentata and soil calcium were the greatest, and were not found in Hiri/Amdu areas where L. tridentata was absent, and soils were the most unconsolidated. Non-randomness in tortoise popu- lations is especially important for the management considerations of relo- cation. Clearly, despite the best ef- forts of concerned managers, the use of the deserts will continue to in- crease and the frequency of tortoise relocations will also undoubtedly in- crease. If tortoise distributions are random, relocation management es- sentially becomes a search for safe relocation sites roughly similar to the "parental" area. No special consid- erations of unique habitat types are required. If on the other hand they are not, then the relocation site(s) must include such high-use habitats as those found in the parental site. In addition, severe disturbance of such favored habitats will in turn have se- vere impacts on the populations, par- ticularly if small. This study indicates that the non- randomness exhibited by the Sand Hill tortoises is probably a function of the non-randomness of highly di- verse plant assemblages and edaphic characteristics. Thus, the presence of diverse land forms and their associ- ated plant communities and diverse edges within future relocation sites should be of significant importance to the manager. Areas which "look good" to the relocation manager may not supply the needed resources for the relocatees. These data are in need of further support however. If such patterns are exhibited in other popu- lations, biologists and managers may use such techniques to successfully determine possible habitat require- ments, and help insure the survival 187 of one of the Mojave's most enig- matic species. Acknowledgments The author wishes to express sincere thanks to Dr. Glenn R. Stewart of Cal Poly, Pomona for physical help and moral support during the fieldwork, and for his abiding friendship. Many thanks also to the entire staff at the MCAGCC for logistical support. Fi- nally, thanks to K. Berry, D. Speake and R. Szaro for their constructive reviews of this manuscript. This work was supported by United State Navy contract N6247484RPOOV48, which was ad- ministered by the Cal Poly Kellogg Unit Foundation. Additional equip- ment support was supplied by graduate research funds of Cal Poly, and monies received from the Chuck Bayless and Tim Brown memorial scholarship funds. Travel funds were supplied by Sigma Xi, The Scientific Research Society. Literature Cited Baxter, Ronald J., and Glenn R. Ste- wart. 1986. Report of the continu- ing fieldwork on the desert tortoise (Gopherus agassizii) at the Twentynine Palms marine corps base. Proceedings of the sympo- sium, [Palmdale, Calif., March, 1986]. The Desert Tortoise Coun- cil, Long Beach, Calif [in press]. Berry, Kristin H. 1984. A description and comparison of field methods used in studying and censusing desert tortoises. Appendix II. In The status of the desert tortoise (Gopherus agassizii) in the United States. Kristin Berry, editor. Re- port to the U.S. Fish and Wildlife Service, Sacramento, Calif. Order No. 11310-0083-81. Berry, KrisHn H., and Lori L. Nicholson. 1984. The distribution and density of desert tortoise populations in California in the 1970's. In Kristin Berry, editor. The status of the desert tortoise (Gopherus agassizii) in the United States. Report to the U.S. Fish and Wildlife Service, Sacramento, CA. Order No. 11310-0083-81. Berry, Kristin H., and Frederick B. Turner. 1984. Notes on the behav- ior and habitat preferences of ju- venile desert tortoises (Gopherus agassizii) in California. Proceed- ings of the symposium, [Lake Ha- vasu City, Ariz., March, 1984]. The Desert Tortoise Council, Long Beach, CA. Brower, James E. and Jerrold Zar. 1984. Field and laboratory meth- ods for general ecology. William E. Brown, publishers. Dubuque, Iowa. Burge, Betty. 1978. Physical charac- teristics and patterns of utilization of cover sites used by Gopherus agassizii in southern Nevada. Pro- ceedings of the symposium, [Las Vegas, Nevada, 1978]. The Desert Tortoise Council, Long Beach, CA. Dodd, C. Kenneth. 1980. Endangered and threatened wildlife and plants: listing as threatened with critical habitat for the Beaver Dam slope population of desert tortoise in Utah. Federal Register 45(163): 55654-55666. Fritts, Thomas H. 1985. Ecology and conservation of North American Tortoises (genus Gopherus). II. Evaluation of tortoise abundance based on tortoise sign detected in field surveys. Prepared for: U.S. Fish and Wildlife Service, Denver Wildlife Research Center, Univer- sity of New Mexico, Albuquerque. NM. Hach Company. 1983. Soil calcium and magnesium test kit (model 14855) instruction manual. Hach Company, Inc., Loveland, CO. Hohman, Judy P. 1977. Preliminary investigations of the desert tor- toise on the Beaver Dam slope in Arizona. Proceedings of the sym- posium, [Las Vegas, Nev., March 1977]. The Desert Tortoise Coun- cil, Long Beach, CA. Keefe, T. J. and E. Bergerson. 1977. A simple diversity index based on the theory of runs. Water Re- sources 11:689-691. Lowe, Charles. 1964. The vertebrates of Arizona. The University of Ari- zona press, Tucson, Ariz. Luckenbach, Roger A. 1982. Ecology and management of the desert tortoise (Gopherus agassizii) in California. In R. Bruce Bury, edi- tor. North American tortoises: conservation and ecology. U.S. Fish and Wildlife Service, Wildlife research report No. 12, Washing- ton, D.C. Schwartzmann, James L., and Robert D. Ohmart. 1976. Quantitative vegetational data of desert tortoise (Gopherus agassizii) habitat in the lower Sonoran Desert. Proceed- ings of the symposium, [Las Ve- gas, Nev. March, 1978] The Desert Tortoise Council, Long Beach, CA. Appendix 1 Summary of Importance Values^ From Plant Transect Data. Plant Community^ Species 1 2 3 4 5 Dare vjrouna 1 Oft 0 1 IJ.O 1 1 n 1 lU.O 114 7 1 14./ yD.D Dcnismus oaroatus Q Rl ft AQ 1 ^y. 1 OA ft Zo.o 40 A Larrea tridentata 16.5 21.6 36.7 6.7 Ambrosia dumosa lie lie: 1 1.0 07 Q Z/.y 0'2 ft ZO.O A 1 riuaria rtgiaa 0 Q l^Q Q Oy.y 07 L/ .o A 7 O./ Erodtum texanum 11.9 6.8 10.2 26.4 19.3 Malacothrix spp. 7 A OQ 1 1C o oo.z 1 n c Eriogonum spp. /.U o c Z.o 1 "3 1 lO.l o.o Hymenoclea salsola 111 11.1 O 7 Z./ OA A Zo.4 /iTnsincicui spp. Q ft y.o 0 0 z.z Oenothera deltoides 6.4 4.8 13.6 Baileya multiradiata 2.0 6.8 6.5 Abronia villosa 5.6 2.2 2.1 Bromus rubens 2.7 2.2 Langloisia Matthervsii 4.9 2.6 Langloisia Palmeri 2.3 Oryzopsis hymenoides 2.5 Eriophyllum Wallacei 2.0 Menodora spinescens 2.9 5.7 Lesquerella Palmeri 2.3 3.1 Salazaria mexicana 3.0 Dalea Fremontii 10.6 Cucurbita foetidissima 2.3 Euphorbia polycarpa 2.0 Isomeris arborea 3.4 Prunus fasiculata 6.7 Spheralcea ambigua 2.6 Salvia columbariae 4.0 Phacelia spp. 2.2 Petalonyx Thurberi 2.6 Unknown composite #1 2.9 2.3 7.3 Unknown composite #2 2.3 'Importance value = relative density + rel. domin. + rel. freq. 'Plant community: See text for description of community names: Meadow and bare areas not listed: I = Sparse Wash): 2 = Hiri/Amdu: 3 = Mixed: 4 = Latr/Amdu: 5 = Washi. 188 Simpson, E. H. 1949. Measurement of diversity. Nature 163:466-467. Sokol, O. M. 1971. Lithophagy and geophagy in reptiles. Journal of Herpetology 5:69-71. Speake, Daniel W. 1986. Gopher tor- toise density in various south Ala- bama habitats. Alabama Coopera- tive Fish and Wildlife Research Unit, Research Information Bulle- tin 86-105, 1 p. Auburn, Alabama. Stewart, Glenn R., and Ronald J. Baxter. 1987. Final report and habitat management plan for the desert tortoise (Gopherus agassizii) in the West and Sand Hill training areas of the Twentynine Palms MCAGCC. Report prepared for the U.S. Department of the Navy, San Bruno, Calif. Contract number N6247484RPOOV48. Turner, Frederick B., and Kristin H. Berry. 1986. Population ecology of the desert tortoise at Goffs, CaH- fornia, in 1985. University of Cali- fornia, Los Angeles publication number 12-1544. Turner, Frederick B., and Carl Thelander, Daniel Pearson, and Betty Burge. 1982. An evaluation of the transect technique for esti- mating desert tortoise density at a prospective power plant site in Ivanpah Valley, California. Pro- ceedings of the symposium, [Las Vegas, Nev., March, 1982]. The Desert Tortoise Council, Long Beach, CA. Walchuck, Sandra L., and James C. devos, Jr. 1982. An inventory of desert tortoise populations near Tucson, Arizona. Proceedings of the symposium, [Las Vegas, Nev., March, 1982]. The Desert Tortoise Council, Long Beach, CA. Weinstein, Michael and Frederick B. Turner and Kristin H. Berry. 1986. An analysis of habitat relation- ships of the desert tortoise in Cali- fornia. Draft report prepared for Southern California Edison Com- pany, Los Angeles, Calif. Zar, Jerrold H. Biostatistical analysis. 1974. Prentice-Hall, Inc., Engle- wood Cliffs, NJ. Changes in a Desert Tortoise (Gopherus agassizii) Population After a Period of High IVIortality^ David J. Germano^ and Michele A. Joyner^ Abstract.— An apparent high rate of mortality for desert tortoises at the Piute Valley in southern Nevada between 1979 and 1983 significantly decreased mean carapace length and average age of the population by 1983. but not density. By 1987, average size and age of the population had increased and density remained stable. Chelonians, as a group, are charac- terized by high rates of adult sur- vival, delayed maturity, and low rates of juvenile survival (Wilbur and Morin 1988). Many chelonians live a long time after reaching adulthood (Gibbons 1987), potentially leading to a long period of reproduction offset- ting low juvenile survival (Wilbur and Morin 1988). The desert tortoise (Gopherus agassizii) (fig. 1) is an her- bivorous chelonian of the desert Southwest that exhibits these popula- tion traits (Berry 1986, Luckenbach 1982, Osorio and Bury 1982, Turner et al. 1984, 1986). In 1983, a large number of desert tortoise skeletons were collected from a study plot lo- cated in southern Nevada and deaths were believed to have occurred since the initial census in 1979 (unpub- lished report, C. Mortimore and P. Schneider, Nevada Department of Wildlife, Las Vegas, NV). It was re- ported that since 1979, mean cara- pace length of the population de- creased, sex ratio had become male biased, and that population density 'Paper presented at symposium. Man- agement of Amphibians, Reptiles, and Small Mammals in Nortti America. (Flag- staff. AZ, July 19-21. 1988). 'David J. Germane is a doctoral candi- date. Museum of Southwestern Biology. Department of Biology. University of New Mexjco, Albuquerque 87131 ^Michele A. Joyner is an undergraduate. Museum of Southwestern Biology. Depart- ment of Biology. University of New Mexico. Albuquerque 87131 decreased, and that these changes oc- curred because long-term grazing of this plot by cattle weakened tortoises to such a degree that decreased for- age production resulting from below- average rainfall in 1981 killed many individuals (unpublished report, C. Mortimore and P. Schneider, Nevada Department of Wildlife, Las Vegas). We recensused this population in 1987 in order to determine changes that might have taken place since 1983 in age distribution, size distri- bution, sex ratios, and population density in order to address the fol- lowing questions: Of what signifi- cance are such periods of high mor- tality to the p)opulations' probability of survival? How do desert tortoise populations respond to high rates of mortality? Are changes in population demographics long-lasting? Can we predict future changes in desert tor- toise populations? We also reassess possible causes of the high rate of mortality between 1979 and 1983. METHODS Study Area The 2.59 km^ plot is located in the Piute Valley of southern Nevada in 190 Figure 3.— Creosote bush and white bursage are the most conspicuous plants of much of the study plot (top) with Mojave yucca abundant In the northwestern portion (bottom). Other abundant plants at this site are California buckwheat (Eriogonum fasiculatum), rayless goldenhead (Acamtopappus sphaerocephalus), Opuntia spp,, bush muhly (Muhlenbergia porteri), gig galleta (Hilaria riglda), six-week fescue (Festuca octoflora), filaree (Erodium cicutarium), desert dandelion (Malacothrix glabrata), and Chaenactis spp. the eastern Mojave desert (fig. 2). Vegetation is Mojave desert scrub dominated by creosote bush (Larrea tridentata) and white bursage (Ambro- sia dumosa) over the southeastern 2/3 of the plot grading into an area with an overstory of Mojave desert yucca (Yucca schidigera) in the northwestern third (fig. 3). Field Methods The population was censused be- tween April and June 1979 by the Bu- reau of Land Management (unpub- lished report, A. Karl, BLM, Las Ve- gas, NV) and again between April and June 1983 by the Nevada Depart- ment of Wildlife (unpublished re- port, C. Mortimore and P. Schneider, Nevada Department of Wildlife, Las Vegas, NV). Each tortoise encoun- tered was measured, weighed, marked, its sexed determined, and its location, behavior and general Figure 2.— The location of the desert tortoise permanent study plot (PSP) in the Piute Val- ley of southern Nevada. The dashed and dotted lines show major washes. condition noted. Shells were col- lected and are catalogued in the Mu- seum of Southwestern Biology, Uni- versity of New Mexico, Albuquer- que. We recensused the plot 13-27 May and 18-25 August 1987. We collected similar data on tortoises, but in- cluded making casts of the second costal scute using dental casting ma- 191 terial (Galbraith and Brooks 1987). Measurements of growth rings from the impressions on the casts were taken. Growth rings of desert tortoises have been found to be valuable for determining age and growth histo- ries of many individuals (Germano 1988). Shells were collected and de- posited in the Museum of Southwest- ern Biology. Data Analysis Density Densities in 1979 and 1983 were de- termined by the investigators who conducted the censuses using the Schnabel estimator. This method in- volves making periodic estimates of density during the census based on the number of marked and un- marked animals found (Tanner 1978). Because of immigration into the plot, we reestimated density for 1983 us- ing the Jolly-Seber estimator (Tanner 1978), which does not assume a closed population. As a first approximation of den- sity for 1987, we used a simple mark- recapture estimator with May as the period of marking animals and Au- gust as the recapture period. Only 1 / 2 the plot was recensused in August because of time constraints. Density was computed for this half of the plot. Carapace Lengtti Distributions Carapace lengths (CD of individuals were plotted and mean CLs com- puted for live tortoises and remains for each census year. Mean CLs of the total population, tortoises >180 mm CL, and tortoises <180 mm CL were compared among years using anova with comparisons among means using Scheffe's multiple com- parisons test. Age Distributions Ages of individuals were plotted for live tortoises and remains and mean ages compared in a manner similar to CLs. Ages of skeletons and 1987 live tortoises were determined for most individuals using scute annuli, a technique that is accurate up to 20- 25 years (Germano 1988). Several in- dividuals were considered to be older than the number of easily seen annuli based on non-growth since last capture, or scute edge beveling, which indicates continued slow growth. These individuals were cate- gorized as >25 years old. Ages were estimated for live tor- toises found in 1979 and 1983 using an age-CL regression (Age = 0.106 CL - 3.82). The number of scute an- nuli is well correlated with CL (r^ = 0.908, n = 150), although the relation- ship is less accurate in larger indi- viduals. We corrected for the pres- ence of older individuals in our esti- mates by assigning a portion of adults of various sizes to the >25 age category based on the percentage of adults that were into this category from the 1987 live and 1983 and 1987 shell groups. Mortality Rates Age-specific mortality rates were de- termined for 1979-1983 and 1983- 1987 using the equation q^ = (k [f J)/ g^, where q^ is the mortality rate per year for age x, k is the per capita mortality rate of the population, f^ is the proportion of animals age x that are known to have died in the past year, and g^ is the proportion of ani- mals of age X in the preceding live population (Fryxell 1986). In order to compare mortality rates to age distri- butions, we determined mortality rates for age groups 0-14 years, 15-27 years, and >25 years. The per capita mortality rate was divided by 4 to obtain the yearly mortality rate for each time period. Sex Ratios Sex ratios were compared among live tortoises and shells. Sex was assigned to tortoises >180 mm CL based on secondary sex characteristics or, in some instances, for males >170 mm CL when plastron concavity was ob- vious. Sex can be determined reliably in desert tortoises based on shell characters after 180 mm CL (unpub- lished report, F. Turner and K. Berry, Southern California Edison Co., CA) and female tortoises in this part of the Mojave desert reproduce at 189 mm CL (Turner et al. 1986), indicat- ing that sexual maturity probably oc- curs between 180-190 mm CL. Ratios were tested for deviation from a 1:1 sex ratio with Chi-square analysis (p < 0.05). CL/Weigtit Regressions Carapace length to weight regres- sions were constructed for 1979 and 1987 tortoises based on the logarith- mic transformation of both variables. Data for 1983 were not available. Slopes were tested against 0 and against each other using f-tests (Sokal and Rohlf 1981). Growtti Rate Comparisons Individual growth was compared among 1987 live tortoises and shell groups in two ways. Growth rings were compared among groups using mean annual widths (AW) and mean percent growth for rings 1-24 (See Germano 1988 for a description of growth ring measurements). Percent growth for a ring is AW/ estimated CL for the preceding year. CLs were estimated using the length of growth rings from the second costal scute, which are highly correlated to CL (r^ = 0.96, n = 174). Growth estimates based on annuli have been found to accurately reflect carapace growth in gopher tortoises (Landers et al. 1982) and desert tortoises (Germano In 192 Press). Means of these variables for each ring were compared among groups using the nonparametric Wilcoxon sign test. We also com- pared the mean AW and mean per- cent growth of the last two growth rings for the shells found in 1983 to the mean AW and mean percent growth of the 1980 and 1981 growth rings from live tortoises found in 1987 using f-tests. 'rill i'ri"i"l"rT"("i'T'i"r"i"ri"i'TiTi i i i i i — I t s 7 I II II B 17 II n u tt n n >za AGE (yeors) 1 I 7 * II n II IT If ai M tt IT II >ii AGE (y e a r t ) Figure 4.— Population size distributions for iive desert tortoises from \he Piute Valley permanent study plot. Mean carapace lengths and somple sizes are given in table 1. Figure 5.— Population size distrit>utions for desert tortoises found dead in 1983 and 1987 from the Piute Valley permanent study plot. Mean carapace lengths and sample sizes are given in table 1 . Table 1 .—Mean carapace lengths (mm) of tortoises from the Piute Valley permanent study plot. Standard deviation and sample size are given be- low \he mean. All % of %of Group tortoises >180 mm CL total <180 mm CL total 1979 live 186.8 217.1 58 144.5 42 (44.0, 84) (21.0,49) (30.8, 35) 1983 live 148.2 211.8 37 110.8 63 (59.6,81) (24.9, 30) (38.3,51) 1987 live 181.1 213,8 60 125.8 40 (46.6, 48) (20.0, 29) (37.8, 19) 1983 shells 197.6 212.9 78 106.4 22 (93.3, 108) (22.6, 84) (39.0, 24) 1987 shells 165.4 216.3 49 117.2 51 (58.1,37) (19.4, 18) (36.9, 19) Climate Analysis Climate was analyzed using weather information from Searchlight, Ne- vada. Data were compared for 3 time periods; 1970-June 1979, July 1979- 1982, and July 1979-July 1987. Means and variances of rainfall, both annual and winter, were compared among time periods. Mean monthly tem- peratures were compared among time periods and temperatures below freezing were analyzed for duration and relation to unusually warm win- ter daily highs. RESULTS Density Tortoise density was estimated to be 50/km2 in 1979 and 72/km2 in 1983 by the authors of these censuses. Eighty-four and 81 tortoises were found in 1979 and 1983, respectively. We reestimated the 1983 density to be 44 tortoises/km^. We estimated the density in 1987 to be 59 tortoises/ km^ (95% confidence intervals, 19- 173). We found 48 tortoises in 1987, 33 in May and 19 on the southern half of the plot in August, of which 4 had been marked in May. Carapace Length Distributions Distributions of CLs of live tortoise populations varied significantly for each census (fig. 4). Mean CL was significantly smaller in 1983 than in either 1979 (p<.05) or 1987 (p<.05). Mean CLs in 1979 and 1987 were not significantly different, however (p>.05, table 1). No significant differ- ences were found among mean CLs for adults (>180 mm CL). Adults comprised 58% of the 1979 popula- tion, 37% of the 1983 population, and 60% of the 1987 population. The ' mean CL of non-adults (<180 mm CL) was significantly smaller in 1983 than 1979 (p<.05), but was not sig- nificantly different than 1987 (p>.05. 193 table 1). The mean CL of non-adults was not significantly different be- tween 1979 and 1987 (p>.05). Remains of 37 tortoises were found in 1987 compared to 109 found in 1983 (fig. 5). Ten shells were found in 1979. CLs of remains were not sig- nificantly different (p>.05), although mean CL in 1983 was considerably larger than for 1987 (table 1). Mean CLs of adult remains in 1983 and 1987 were similar, as were non-adult CLs, but adults comprised 78% of the 1983 collection and only 49% of the 1987 collection. The mean CL of re- mains from 1983 was not signifi- cantly different from the mean CL of live tortoises in 1979 or 1987, but was significantly larger than live tortoises in 1983 (p<.05). Mean CL of remains from 1987 was not significantly dif- ferent than any live tortoise means. Age Distributions Ages of tortoises varied significantly among years (table 2). Changes in age distributions of live tortoises were similar to the changes seen for CLs (fig. 6). The estimated mean age for 1979 was significantly older than 1983 (p<.05) but not 1987 (p>.05). Mean age for 1987 was not signifi- cantly different than 1983 (p>.05), but non-adults were significantly older (p<.05). Mean age of 1983 remains was significantly older than 1983 live tortoises (p<.05), but was not signifi- cantly different than 1987 live tor- toises or remains (p>.05, fig. 7). Mortality Rotes Death rates for 1983-1987 were lower than for 1979-1983. Per capita mortal- ity rate (k) for 1979-1983 was 0.21/ year (N = 130) and was 0.08/year for 1983-1987 (N = 115). Mortality rates dropped for all age classes after 1983. For 1979-1983 mortality rates were 0.145/year for 0-14 year olds, 0.247/ year for 15-25 year olds, and 0.195/ year for tortoises >25 years. For 1983- 1987 mortality rates were 0.061 /year for 0-14 year olds, 0.093/year for 15- 25 year olds, and 0.103 for tortoises >25 years. Mortality rates for all adults (15-25 years and >25 years) for 1979-1983 was 0.240/year and for 1983-1987 was 0.103/year. Sex Ratios Sex ratios of live tortoises show an increasing proportion of males (table 3), although only 1987 showed a sig- nificantly biased sex ratio. When the 1987 sex ratio was analyzed by size, 92% of tortoises >220 mm CL were males, whereas only 53% of tortoises 180-219 mm CL were males (table 3). When analyzed by age, 63% of tortoises >20 years were males, but 71% of tortoises of known sex be- tween 13-19 years were males, a sig- nificantly higher proportion than fe- males. The sex ratios of dead tor- toises were not significantly different than 1:1 (table 3). CL/Weigtit Regressions The regressions of weight against CL had significant slopes for 1979 and CARAPACE LENGTH (mm) Figure 6.— Population age distributions for live desert tortoises from the Piute Valley perrrKinent study plot. The 1979 and 1983 age distributions are estimates based on a carapace length to annulus number re- gression. A proportion of adults were placed in the >25 age category based on the proportion of adults In this category from the age distributions for which ages were assigned by annuli counts. The 1987 age distribution Is based on annuli counts. r Table 2.--Mean ages of tortoises from the Piute Valley permanent study plot in southern Nevada. Standard deviation and sample size are given below the mean. Ages for 1 979 and 1 983 are estimates based on cara- pace length (see Methods). Ages (years) Group 0-27 0-14 15-27 ^>25 1979 live 16.6 10.9 19.5 (5.1,72) (3.4, 24) (2.9, 48) (12) 1983 live 12.1 7.5 18.8 (6.6.74) (3.7,41) (3.0, 30) (7) 1987 live 14.1 11.3 17.0 (3.8, 43) (3.2, 22) (2.2,21) (5) 1983 shells 17.0 7.8 19.9 (6.2, 94) (3.6, 22) (3.3, 72) (14) 1987 sl-»ells 14.0 8.4 19.3 (62., 31) (2.6, 15) (3.5, 16) (6) 'Mean age cannot be determined. 194 1987 (fig. 8). The regression equation for 1979 is gram weight = 0.000317 CL2.924 ^ 0.952, n = 73) and for 1987 is gram weight = 0.000505 (2L2.826 (jj ^ 0.969, n = 53). Regression slopes were not significantly differ- ent from each other (p>.10). Growth Rate Comparisons No significant differences were found in a ring by ring comparison of growth between 1987 live tortoises and 1983 remains for either annual widths (AW) or percent growth. When 1980 and 1981 rings were com- pared, no significant difference ex- isted between the mean AW for the last two rings of 1983 mortalities (X = 1.98mm, n = 72) and the 1980 and 1981 rings for 1987 live tortoises (X = 1.92nmi, n = 79; p>.10). CARAPAC E LENGTH (mm) Figure 7.— Population age distributions for desert tortoises found dead in 1 983 and 1987 from the Piute Valley pernnanent study plot. Both the 1983 and the 1987 age distri- butions are based on counts of annuli. 9 4.0- l.t- 1.0- IJ t.o- 1,1 1.0 I • T 1 I I \ — r— I — l—l — I — I — r— I — I I I — r— I — I — I — I — r M 1.0 u l.« I.I- 1.0- •.I HIT I T I I I I I I I I I I I I I I I I I — I — r to M no HO NO NO WO tot tW 140 tM CARAPACE LENGTH (mn) Figure 8.— Regressions of carapace length to weight for desert tortoises found in 1979 and 1987. Slopes of both regressions are significantly different from 0 but not from each other. Table 3.— Numbers of males to females for desert tortoises from the Piute Valley permanent study plot. Significant departures from a 1 :1 sex ratio were determined by Chl-square analysis. The 1987 live totals were sub- categorized by size and age. Year Males Females Ratio 1979 live 24 30 0.88 1 0.667 shells 4 3 1.33 1 0.001 1983 live 22 11 2 1 3.667 shells 35 41 0.85 1 0.474 1987 live (total) 20 9 2.22 1 M.172 size: 180-219 mm CL 9 8 1.13 1 0.059 >220mm CL 11 1 11 1 ^8.330 age: 13-19 years 15 6 2.5 1 ^3.857 >20 years 5 3 1.67 1 0.500 shells 11 6 1.83:1 1.471 'Significanf departure from 1:1 ratio (p<.05). Climate Analysis Average precipitation were higher between July 1979 and July 1987 than the previous 10 years (table 4). The highest average precipitation was recorded between July 1979 and De- cember 1982. Winter rainfall (Octo- ber-March) followed the same pat- tern, with both 1979-1987 and 1979- 1982 averages higher than 1970-1979 (table 4). The period 1970-1979 was a drought period with average rainfall 7% below the long-term average of 183.8 mm and 7 of the 10 years were well below average (table 4). When 1978 and 1979 are excluded, average precipitation drops to 129.3 mm, 30% below the long-term average. July 1979-December 1982 averaged 40% higher rainfall than the long-term av- erage with only 1981 experiencing below-average rainfall. Mean monthly high and low temperatures were similar among time periods. No extended periods of freezing tem- peratures were found for daily read- ings between 1979 and 1983. DISCUSSION Population Parameters The desert tortoise p>opulation in the Piute Valley study plot experienced a high rate of mortality, particularly of adults, between July 1979 and 1983. Related to this event was a signifi- cant decrease in the size and age dis- tributions of the population in 1983, although both were returning to 1979 dimensions by 1987. The lower mean age in 1983 is probably a result of in- creased survival of hatchlings and increased immigration. The increased 195 survival of hatchlings, as shown by the significant increase of tortoises in the 1-4 age group in 1983, may be due to more favorable conditions be- cause of lower densities just after the high rate of mortality, or to optimal climatic and habitat conditions. It is possible that the greater num- bers of smaller tortoises found in 1983 could have resulted from better search effort for these sizes (Berry and Turner 1984), but we censused the plot carefully in 1987, specifically looking for small tortoises, yet we found relatively few. While we do not doubt that young are missed be- cause of their inconspicuousness, we believe that the changes in size and age distributions between 1979 and 1987 reflect actual population changes. The size and estimated age distri- butions for 1983 indicate that a sig- nificant number of smaller and younger tortoises came into the plot between 1979 and 1983. Judging by the male-dominated sex ratio after 1979, immigration largely has been by young males. The biased sex ra- tios are not due to higher adult male survival since equal proportions of males and females died. Most of the males in the present population are fairly young, although they are large. Male turtles are known to disperse greater distances than females (Gib- bons 1986). Although many turtle populations have biased sex ratios, evolutionary theory indicates that these ratios should be under selective pressure to be relatively even, in most instances (Fisher 1930, Trivers 1972). However, desert tortoise age to maturity is ca. 15 years (Germano In Press, Woo- dbury and Hardy 1948), therefore a reproductive solution mediated by selection would require hundreds of years. Censuses in other parts of this val- ley in 1983 indicate that this high rate of mortality was confined to this plot and areas close by (unpublished re- port, C. Mortimore and P. Schneider, Nevada Department of Wildlife, Las Vegas, NV). Differences in sex ratios at this plot may be more a reflection of higher male movement rates com- pared to females and not to a real difference in numbers of males and females in the population as a whole. Over time the sex ratios may change by movement of females into the plot from outside. Density may have decreased slightly since 1979, but it does not ap- pear to have changed significantly over the 8 year period, although we recognize the imprecision of these density estimates. The number of tor- toises found has decreased in each census, but investigators and time periods in the field have varied, ren- Table 4.— Annual and winter precipitation (mm) for 1970-1987 and for 3 time periods from the Searchlight, Nevada NOAA Station. Winter precipita- tion is defined by the months October-March. Means and standard devia- tions are given for the 3 time periods. Precipitation for 1987 only Includes the months of January-July. Year 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 Time period Annual Winter total total Jan. 1970- July 1979- July 1979 June 1 979 Dec. 1 982 July 1 987 127.76 68.83 136,65 114.81 184.40 132.08 161.80 107.70 473.71 256.54 313.44 162.81 366.10 376.68 300.48 149.35 166.88 73.66 30,73 17.02 179.02 54,36 100,08 82.79 52.58 183.90 249.43 260.10 67.06 101.09 216.15 61.47 191.52 91.69 126.24 Annual precipitation 170.9 (113.0) 281.2 (86.4) 265.5 (148.8) Winter precipitation 104.4 (33.0) 139.4 (73.7) 161.0 (71.7) 196 dering this comparison unreliable. We believe that the lower number of live tortoises found in 1987 is due to inexperienced field personnel and the shorter duration of time in the field. The most valid of these density estimates is the Jolly-Seber estimate of 44 tortoises /km^, because more assumptions are met with this tech- nique. Unfortunately, estimates can- not be made for the first or last cen- sus with this technique. Density esti- mates, though, are similar in magni- tude and we believe this indicates that density has remained relatively stable since 1979. The population must have experienced a decline af- ter 1979 but we believe that increased survival of young and immigration from adjacent non-affected areas has quickly returned the density to 1979 levels. Mortality Factors Causes of the high rate of mortality have not been demonstrated. The hypothesis that long-term grazing confounded by a drought in 1981 was the cause of the high number of tortoise deaths is not supported by growth analysis of annuli, CL/ weight data, or climate data. Growth did not differ significantly between those that died before 1983 and those that survived to 1987. In addition, the weight to size regressions for 1979 and 1987 were the same and both were almost identical to the re- gression for tortoises from an un- grazed plot in Nevada (Medica et al. 1975). As for a drought in 1981, aver- age rainfall was only 9% below the long-term average (up to 1987) and was actually at the average, up to 1981, given the drought in the 1970s. Preceding 1981 were 3 years of ex- ceptionally high rainfall. In contrast, rainfall in 1977 was 41% below aver- age and followed many drought years (table 4). Desert tortoises are known to store water (Nagy and Medica 1986) and may be able to store fat. It seems doubtful that one average year of rainfall after 3 very good years could cause starvation or lethal dehydra- tion. The 2 years preceding our cen- sus in 1987 were below average in precipitation, yet mortality rates dropped. The period 1970-1977 was a drought, yet only 10 shells were found in 1979. If these low rainfall years didn't produce a high rate of mortality that could be detected in 1979, it is hard to imagine that one average year after 3 good years would result in excess mortality. Es- timates of yearly adult death rates from 1972-1982 for a population only 42 km south of this site was 1.2%, in an area that has been grazed by live- stock for 100 years (Berry and Nicholson 1984a). Other possible causes for this mor- tality could have been disease, pre- dation, or flooding. Diseases are known to affect other turtle species in the wild (Jacobson 1980a,b), but no evidence exists for disease as a fac- tor. Many of the shells show signs of chewing by carnivores, although whether this indicates predation or scavenging cannot be determined. Flooding occurred in or near the plot in 1980 and 1982 (unpublished re- port, J. Jamrog and R. Stager, BLM, Las Vegas, NV). The plot is dissected by numerous washes that are most prevalent in this part of the valley (fig. 2). The exact cause of the high rate of mortality may never be known. Star- vation, disease, flooding, and preda- tion may have all had an effect. No singular explanation is supported by the data. Whatever the causative agent, the population appears to be returning to a density and popula- tion structure as occurred before the f)eriod of high mortality. Management Implications As a long-lived reptile, the desert tor- toise is more vulnerable to fluctua- tions in adult mortality than to simi- lar fluctuations in younger age groups. Many desert tortoise popula- tions consist of adult segments that usually have yearly survivorship rates of 95-98% (Berry and Nicholson 1984b). High adult survivorship is often coupled with low juvenile sur- vivorship (Wilbur and Morin 1988) and part of the concern for tortoise populations is that they may not have the ability to withstand distur- bance because of low juvenile survi- vorship. Female desert tortoises in the eastern Mojave desert have the ability to lay 2-3 clutches in a season (Turner at al. 1986). The significant increase in 1983 of tortoises 1-4 yr of age suggests more hatchlings have survived between 1979-1983 than previously. As with any other popu- lation parameter, juvenile survivor- ship can vary, and this may lead to periodic additions of greater num- bers of young surviving to adult age. It appears that desert tortoises have the ability to recover from dis- turbance in some instances. This ap- pears to be what is happening at the Piute plot. Increased juvenile survi- vorship and immigration are holding the population density stable and the age and size distributions are return- ing to 1979 dimensions. This kind of recovery may not occur if a distur- bance is prolonged or is widespread. Those managing desert tortoises must be aware of the dynamics of each population, but it is apparent that tortoise populations can recover from short-term high mortality. ACKNOWLEDGMENTS We thank T. Fritts and the National Ecology Research Center of the U.S. Fish and Wildlife Service for provid- ing support during data collection and analyses. We also thank R. Wil- ingham, J. Talbert, and C. Isbell for assistance with the May census. R. Haley and B. Turner of the Nevada Department of Wildlife provided re- |X)rts and shells for this site. T. Fritts, M. Molles, N. Scott, H. Snell, K. Sev- erson, and 2 anonymous reviewers 197 read drafts of this manuscript and greatly improved its content, but any errors or omissions are our own. LITERATURE CITED Berry, Kristin H. 1986. Desert tortoise (Gopherus agassizii) research in Cahfornia, 1976-1985. Herpetol- ogica 42:62-67. Berry, Kristin H., and Lori L. Nicholson. 1984a. Attributes of populations at twenty-seven sites in California, p. 154-241. In The Status of the Desert Tortoise (Go- pherus agassizii) in the United States, Kristin H. Berry, editor. Desert Tortoise Council Report to U.S. Fish and Wildlife Service, Sacramento, California. Order No. 11310-0083-81. Berry, Kristin H., and Lori L. Nicholson. 1984b. A summary of human activities and their impacts on desert tortoise populations and habitat in California, p. 61-117. In The Status of the Desert Tortoise (Gopherus agassizii) in the United States, Kristin H. Berry, editor. Desert Tortoise Council Report to U.S. Fish and Wildlife Service, Sacramento, California. Order No. 11310-0083-81. Berry, Kristin H., and Frederick B. Turner. 1984. Notes on the behav- ior and habitat preference of juve- nile desert tortoises (Gopherus agassizii) in California, p. 111-130. In Proceedings of the 1984 Sympo- sium of the Desert Tortoise Coun- cil, Mary W. Trotter, editor. Long Beach, California. Fisher, Ronald A. 1930. The genetical theory of Natural Selection. Ox- ford Univ. Press, London. 272 p. Fryxell, J. M. 1986. Age-specific mor- tality: an alternative approach. Ecology 67:1687-1672. Galbraith, David A., and Ronald J. Brooks. 1987. Photographs and dental casts as permanent records for age estimates and growth studies of turtles. Herpetological Review 18:69-71. Germano, David J. 1988. Age and growth histories of desert tortoises using scute annuli. Copeia 1988: [In Press]. Germano, David J. In Press. Prelimi- nary comparison of growth in des- ert tortoises and gopher tortoises. In Proceedings of the 1987 Sympo- sium of the Gopher Tortoise Council, Richard Franz and C. Kenneth Dodd, Jr., editors. Gibbons, J. Whitfield. 1986. Move- ment patterns among turtle popu- lations: applicability to manage- ment of the desert tortoise. Herpe- tologica 42:104-113. Gibbons, J. Whitfield. 1987. Why do turtles live so long? Bioscience 37:262-269. Jacobson, Elliott R. 1980a. Viral agents and viral diseases of rep- tiles, p. 197-202. In Reproductive Biology and Diseases of Captive Reptiles, James B. Murphy and Joseph T. Collins, editors. Society for the Study of Amphibians and Reptiles, Contribution to Herpe- tology Number 1. Jacobson, Elliott R. 1980b. Mycotic diseases of reptiles, p. 235-241. In Reproductive Biology and Dis- eases of Captive Reptiles, James B. Murphy and Joseph T. Collins, editors. Society for the Study of Amphibians and Reptiles, Contri- butions to Herpetology Number 1. Landers, J. Larry, W. Alan McRae, and James A. Garner. 1982. Growth and maturity of the go- pher tortoise in southwestern Georgia. Bulletin of the Rorida State Museum, Biological Sciences, 27:81-110. Luckenbach, Roger A. 1982. Ecology and management of the desert tor- toise (Gopherus agassizii) in Califor- nia, p. 1-37. In North American Tortoises: Conservation and Ecol- ogy, R. Bruce Bury editor. U.S. Fish and Wildlife Service, Wildlife Research Report 12. Washington, D.C Medica, Philip A., R. Bruce Bury, and Frederick B. Turner. 1975. Growth of the desert tortoise (Gopherus agassizi) in Nevada. Copeia 1975:639-643. Nagy, Kenneth A., and Philip A. Medica. 1986. Physiological ecol- ogy of desert tortoises in southern Nevada. Herpetologica 42:73-92. Osorio, Sandalio Reyes, and R. Bruce Bury. 1982. Ecology and status of the desert tortoise (Gopherus agas- sizii) on Tiburon Island, Sonora. p. 39-55. In North American Tor- toises: Conservation and Ecology, R. Bruce Bury, editor. U.S. Fish and Wildlife Service, Wildlife Re- search Ref)ort 12. Washington, D.C. Sokal, Robert F., and F. James Rohlf. 1981. Biometry. Second edition. W. H. Freeman and Company, New York. 859 p. Tanner, James T. 1978. Guide to the study of animal populations. Uni- versity of Tennessee Press, Kn- oxville. 186 p. Trivers, Robert L. 1972. Parental in- vestment and sexual selection, p. 136-179. In Sexual Selection and the Descentof Man— 1871-1971, Bernard Cambell editor. Aldine Publishing Co., Chicago. Turner, Frederick B., Page Hayden, Betty L. Burge, and Jan B. Rober- son. 1986. Egg production by the desert tortoise (Gopherus agassizii) in California. Herpetologica 42:93- 104. Turner, Frederick B., Philip A. Medica, and Craig L. Lyons. 1984. Reproduction and survival of the desert tortoise (Gopherus agassizii) in Ivanpah Valley, California. Copeia 1984:811-820. Wilbur, Henry M., and Peter J. Morin. 1988. Life history evolution in turtles, p. 387-439. In Biology of the Reptilia, Volume 16, Carl Gans and Raymond B. Huey, editors. Alan R. Liss, Inc., New York. Woodbury, Angus M., and Ross Hardy. 1948. Studies of the desert tortoise, Gopherus agassizii. Eco- logical Monographs 18:145-200. 198 A Survey Method for Measuring Gopher Tortoise Density and Habitat Distribution^ Daniel M. Spillers and Dan W. Speake^ The only tortoise to occur in the southeast, the gopher tortoise (Go- pherus polyphemus) (fig. 1), is limited to six states. Of these six states, legal protection is offered by South Caro- lina, Mississippi, Georgia, Florida and Alabama; Louisiana does not re- strict the harvest on gopher tortoises at present. The gopher tortoise is now federally listed as threatened in the portion of its range west of the Tombigbee river in Alabama. During the past several years, an apparent decline of gopher tortoise p)opulations has been noted. Boze- man (1971) and Wharton (1978) noted the rapid loss and alteration of sand ridge habitat, the habitat in which most gopher tortoise popula- tions occur, and argued for the pres- ervation of these habitats not only for gopher tortoises but also for other aspects of their ecological signifi- 'A contribution of the Alabama Coop- erative Rshi and Wildlife Research Unit: Au- burn University Agricultural Experiment Sta- tion and Department of Zoology and Wild- life Science, Game and Rsh Division of the Alabama Department of Conservation and Natural Resources, the U.S. Fish and Wildlife Service and the Wildlife Management Insti- tute cooperating. Presented at the Sympo- sium on Management of Amphibians, Rep- tiles, and Small Mammals in North America, July 19, 1988. 'Spillers is a research technician and Speake is assistant unit leader/wildlife with the Alabama Cooperative Fish and Wildlife Research Unit. Auburn University, Alabama 36849-5414. Abstract.— An underground closed-circuit television camera and Landsat satellite imagery were utilized in a 2-year study to examine status of the gopher tortoise in southern Alabama. Use of this camera resulted in a complete count of gopher tortoises in the sample transects. The transects were located precisely on standard topographic maps and on Landsat images. An estimation was then made of the amount of each habitat type in southern Alabama based on light reflectance of the vegetation and soil type of the sample transects. Density measurements were then expanded to estimate tortoise numbers for the entire area. This method is effective for estimating gopher tortoise numbers and for determining quantity and location of gopher tortoise habitat. cance. Auffenberg and Franz (1982) documented a decline of gopher tor- toise populations on specific sites in the Southeast. Landers et al. (1980) found that gopher tortoises have such a low reproductive rate that human exploitation of tortoises can drastically reduce local populations. Landers and Speake (1980) showed that population densities of gopher tortoises can fluctuate widely in re- sponse to habitat manipulation or neglect. Other conceivable reasons for this apparent decline were noted by Diemer (1986). Sand ridge habitat is not only im- portant for gopher tortoises, but also for many other animals that use go- pher tortoise burrows for nesting, feeding, or escape cover. Three sub- species of the crawfish-gopher frog complex that are closely associated with gopher tortoise burrows are the dusky gopher frog (Ram areolata sevosa), the Rorida gopher frog, (R. a. aesypus), and the Carolina gopher frog (R.a. capito). The threatened eastern indigo snake (Drymarchon corais couperi) is dependent on tor- toise burrows for winter cover in the northern part of its range (Speake et al., 1978; Landers and Speake, 1980; Diemer and Speake, 1981). Several species of mammals and birds use gopher tortoise burrows, most often as escape cover. Several authors have noted the diversity of animal life (both vertebrate and invertebrate) Figure 1 .—A gopher tortoise from southern Alabama. inhabiting tortoise burrows and the dependence of some species on tor- toise burrows for survival (Allen and Neill, 1951; Hubbard, 1894; Hutt, 1967; Landers and Speake, 1980; Speake et al., 1978; Woodruff, 1982). In view of the apparent decline of gopher tortoise populations, it is im- p)ortant to be able to accurately meas- ure tortoise density in an area and to determine quantity and distribution of suitable tortoise habitat. Tortoise density has been previously esti- mated by means of a correction fac- tor applied to counts of burrows (Auffenberg and Franz, 1982), dig- ging of burrows, and use of listening devices. Previous methods do not ensure accurate determination of tor- toise density without burrow de- struction and prohibitive labor. De- termination of quantity and location of tortoise habitat is becoming neces- 199 sary due to rapid changes in land use and increasing relocation and re- stocking efforts (Diemer, 1984; Lan- ders, 1981). The objectives of this study were to develop and employ a method to: (1) accurately measure gopher tor- toise density and (2) locate and quan- tify tortoise habitat in a 24-county area of southern Alabama. We are indebted to James Altiere, Eugene Carver, Kevin Dodd, Lane Knight, Sonny Mitchell, Claud Searcy, and William Sermons, who assisted in collecting field data. We are especially indebted to Walter Stephenson, Chief of the Resource Development Section of the State Planning Division, Department of Economic and Community Affairs, State of Alabama for his help and co- operation in giving us access to the Landsat remote sensing system. Ap- preciation is extended to Joe Exum, Raymond Metzler, and Nick Wiley for their assistance in experimental design and data analysis. Special ap- preciation is extended to Dr. Charles Williams of the Research and Data Analysis Department, Auburn Uni- versity, for his advice and aid with statistical design and analysis. The project was funded by a grant from the U.S. Fish and Wildlife Service and by the Alabama Cooperative Fish and Wildlife Research Unit. Methods Study Area Determination and Questionnaires Our study area was determined by the reported historical range of the gopher tortoise in Alabama (Mount, 1978; Auffenberg and Franz, 1982). This included 24 counties in the coastal plain of Alabama (excluding the counties west of the Tombigbee river which were surveyed by other researchers). Questionnaires were sent out to wildlife biologists, conser- vation officers, herpetologists, county agents, soil conservation agents and other people who were likely to have knowledge of gopher tortoise populations in our 24-county study area. These questionnaires asked for locations of areas that sup- ported or had supported tortoise populations, and names of landown- ers or other persons who might have additional knowledge of tortoise populations. A map was included with each questionnaire so that loca- tions could be marked. A total of 132 questionnaires was mailed out and 58% were returned. Soil conservation offices were vis- ited in each surveyed county and fur- ther inquiries were made concerning tortoise population occurrence and habitat availability. Areas in each county that had soils with sand to a depth of at least 1 m and that pref- erably contained a variety of habitat types were delineated on maps. These areas were considered poten- tial tortoise habitat (Garner and Lan- ders, 1981; Landers, 1981; Landers and Garner, 1981) and were used to sample tortoise densities. After evaluation of the informa- tion from the questionnaires, per- sonal interviews, and discussion with soil conservation agents, the 24- county study area was divided into three classes (fig. 2). Class I counties (n=14) contained widely distributed gopher tortoise {X)pulations and habitat. Class II counties (n=4) con- tained relict or disjunct populations and scattered, spotty habitat. Class III counties (n=6) were those in which no tortoise populations could be found. Sampling Sctieme In Class I counties, regions deline- ated by the soil conservation agents (sandy soil > 1 m) were located on 1:24,000 scale topographic maps. Within these areas, a reference point for initiation of sampling was chosen from the map which had a variety of habitat types (at least 2) within a 1 km radius of the reference point. These points were chosen before vis- iting the site. Where necessary, per- mission was obtained for sampling on private property. Upon arrival at the location as many of the following habitat types were located as possible: unburned pine/ scrub oak, burned pine/ scrub oak, planted pines, clearcuts, old- fields, agricultural fields, pasture, and corresponding edges for each type. The example of each habitat type nearest to the reference point was then sampled. Belt transects measuring 265 x 15 m (0.4 ha) were systematically lo- cated within the habitat types avail- able; edge transects were centered on and followed the edge. If there were open burrows in the transect, the burrows were examined using the MUTVIC (Miniature Underground Television Inspection Camera) (Speake and Altiere, 1983). This de- vice enabled us to insert a closed- Figure 2.— Distribution of the gophier tortoise in 24 counties of Alabama. 200 circuit television camera to the bot- toms of the burrows and determine if they were occupied (figs. 3-5). Bur- row width measurements were made with calipers inserted approximately 70 cm into the burrow. Data gathered for each transect included habitat type, number of open burrows, num- ber of active burrows (burrows with sign of recent tortoise use), number of tortoises, and width of burrows. In Class II counties we searched each area where tortoise populations had been reported or where gopher Figure 3.— Closed-circuit television camera with protective glass globe. Figure 4.— Crew inserting closed-circuit television camera into gopher tortoise burrow. tortoise habitat (sandy soil > 1 m) ex- isted. Observations were made of the total number of burrows, and total number of active burrows. Since these counties lay along the northern border of the gopher tortoise's range in Alabama, tortoise populations were scattered and did not occur as uniformly in specific habitat types as those populations in Class I counties. Therefore we did not sample here but instead used a correction factor similar to the one described by Auffenberg and Franz (1982). The correction factor (0.67 tortoises/ac- tive burrow) was obtained from our sampling of Class I counties by di- viding the total number of tortoises by the total number of active bur- rows. The estimated total number of tortoises for Class II counties was very low (56), and did not signifi- cantly affect our population estimate. Landsat Satellite Imagery Having measured tortoise density on sample areas of the habitat types, Landsat digital satellite imagery was used to obtain an estimate of the area of each habitat type in Class I coun- ties. Characteristics and usage of this remote sensing technique are de- scribed by Anderson, Wentz and Treadwell (1980), Brabander and Barclay (1977), Diemer and Speake (1983), Graham et al. (1981), Taranik (1978a), and Taranik (1978b). The system we used makes a scan of the earth every eighteen days from a geosynchronous orbit. The multis- pectral scanner operates in seven dif- ferent wavelengths of light — four vis- ible and three infrared. We used near infrared because it showed vegeta- tion characteristics more clearly. By making several passes, the scanner senses light reflectance based on 0.1 ha pixels. Each 0.1 ha of the earth's surface is assigned 1 of 256 gray val- ues based on its reflectance. Using these gray values we separated the following habitat types based on their spectral signature: unburned 201 pine/scrub oak, burned pine/ scrub oak, planted pine, old-field, agricul- tural fields, pasture and composite edge. Before sampling, we used ground- truthing to determine if it was fea- sible to attempt to classify each habi- tat type using Landsat imagery. On 70-0.4 ha sample plots in Baldwin County (10 plots in each habitat type), each plot was correctly classi- fied. Clearcuts were not included be- cause they were a rapidly changing transient stage (1-2 years) leading to planted pine habitat, and as such could not be identified on Landsat images accurately due to their rapid vegetational change. Habitat was considered planted pine if pine was a prominent understory or midstory component (at least 0.3 m tall). Indi- vidual edge types were combined because edge transects had similar vegetation characteristics and thus a similar spectral signature. Combined edge habitat was identifiable. NASA software used with Land- sat imagery includes a program for referencing Landsat digital data to any scale map. We referenced our data to standard 1:24,000 topo- graphic maps using known control points. This enabled us to use Uni- versal Trans Mercator coordinates to locate each transect on the Landsat image and obtain the correct gray value for each transect. We then as- signed a range of gray values to each habitat type based on the reflectance of the sample transects. The accuracy of the habitat classifications was checked throughout this process. A polygon was then constructed enclosing all the Class I counties, and areas of each gray value within this polygon were measured. From these measurements we determined the total area for each habitat type in Class I counties. Data Analysis We had two concerns relative to data analysis: (1) to derive a population estimate based on mean tortoise den- sity f>er hectare multiplied by the es- timated area of the respective habitat type, and (2) to identify and locate gopher tortoise habitat. In order to obtain a population estimate we multiplied the mean density of gopher tortoises per hec- tare in a specific habitat type by the total area of that habitat type in Class I counties. An allowance was made for standard error of the mean. The habitat totals were then summed to give a final population estimate of the Class I counties. In addition to these concerns we examined age class structure. Lan- ders et al. (1982) noted that gopher tortoises pass through two general life-history stages before they reach sexual maturity. The juvenile stage lasts until the carapace is approxi- mately 100-120 mm. During the juve- nile stage, the shells are very soft and carapacial scutes usually have dis- tinct yellow centers. This stage usu- ally lasts until about 5 years of age. Juvenile coloration fades and the shells begin to harden during the subadult stage which generally lasts from 5 to 21 years of age. Carapace lengths range from about 120-220 mm. At sexual maturity, body vol- ume has drastically increased and sexual dimorphism is apparent. This occurs at approximately 21 years of age and a carapace length of 230 mm. Alford (1980) established a mathe- matical relationship between the widths of gopher tortoise burrows and the carapace lengths of their oc- cupants in northern Florida (this rela- tionship has not been thoroughly tested in other states). Using Alford's equation log^^y = 0.879 log^^x + 0.149, where y is carapace length and x is burrow width, we used our burrow width measurements of occupied burrows to divide tortoise popula- tions into juvenile, subadult, and adult age classes. We considered age class structure to be an important cri- Figure 5.— Closed-circuit television monitor displaying picture of a gopher tortoise inside a burrow. 202 ^ . Table 1 .—Summary of sample variables and derived estimates for Class I counties from 339-0>4 ha transects in south- em Alabama, 1984-1985. Standard Area Population Habitat n Habitat totals Mean densitles/ha error*" (ha) estimate Open Active Open Active burrows burrows Tortoises burrows burrows Tortoises Old-field 21 23 17 13 2.72 2.00 1.53 0.47 35,822 207,808 ± 63,836 Planted Pine 17 7 6 5 1.01 0.87 0.72 0.35 99,855 71,896 ± 34,949 Burned Pine/ Scrub Oak 34 36 13 9 2.62 0.94 0.64 0.27 209,108 133,829 ± 56,459 Edge 129 85 54 34 1.63 1,04 0.64 0.15 102,408 65,541 ± 15,361 Pasture 46 1 1 1 0.05 0.05 0.05 0.05 61,225 3,061 ± 3,061 Agriculture 31 0 0 0 0.00 0.00 0.00 0,00 210,386 0 Unburned Pine/ Scrub Oak 10 0 0 0 0.00 0.00 0.00 0.00 133,004 0 Clearcuts 51 1 1 0 0.05 0.05 0.00 0.00 Totals 339 153 92 62 951,808 482,135 ± 173,666 "Standard error o f the tortoise mean dertsity/ha. teria along with density in evaluating tortoise population viability. Re- search has not yet revealed an opti- mum age class structure. Intuitively, in a long-lived animal such as the gopher tortoise, the age class struc- ture of a healthy population would be skewed toward the adult class. The presence of juvenile and sexually mature adult tortoises does definitely indicate recent reproduction. Results Gopher Tortoise Densities and Habitat Areas Tortoise densities and habitat areas were measured in Class I counties. These results are summarized in table 1, which includes sampling variables by habitat type along with estimates derived from sampling. Age Class Structure Five percent of the sampled popula- tion (n=100 tortoises) were juvenile tortoises, 48% were subadult, and 47% were adults. This structure shows that there has been recent re- production, and that there is a large segment of breeding size adults pres- ent. This suggests that the potential for successful population mainte- nance over the estimated 951,808 ha area of tortoise habitat in Class I counties is good. Discussion Using the referenced Landsat data and knowing the range of gray val- ues for each habitat type, we were able to examine any area in Class I counties and determine the size and quantity of gopher tortoise habitat units. Using a plotter, figures can be made of all the 0.1 ha pixels that cor- respond to a given habitat type and then the figure can be overlaid on a map. For our purposes we only needed the area of each habitat type in Class I counties. This technique has two distinct sources of error. First is the variation of the gopher tortoise densities within habitat types. These variations are inherent in sampling biological populations. In this study the vari- ance was fairly low. Increased sample size would likely lower this J error. The second source of error is in estimating total areas of the habi- tat types over a large region. Al- though in our preliminary ground- truthing, Landsat imagery correctly classified all our habitat types (ex- cluding clearcuts and individual edge types), we suspect that when this technique is applied to a large diverse region some areas will be misclassified. Ground-tru thing should be done after the classifica- tion to determine what percentage has been misclassified, which would allow the researcher to make allow- ances for this error in final computa- tions. Conclusions We found this technique to be useful for measuring tortoise density and for determining quantity and loca- tion of tortoise habitat. The error in this technique seems to b>e less than that for techniques used for census- ing most other animals. Although it is difficult to estimate numbers of animals over a large area, it is helpful to be able to accurately measure den- sity in small areas and then extrapo- late this density on the basis of a 203 quantitative measurement of a desig- nated area. This method should be especially valuable for surveys of animals that are habitat specific. Literature Cited Alford, Ross A. 1980. Population structure of Gopherus polyphemus in northern Florida. Journal of Herpetology. 14:177-182. Allen, Ross, and Wilfred. T. Neill. 1951. The gopher. Florida Wildlife. 5(7):10, 32. Anderson, William H., W. Alan Wentz, B. Dean Tread well. 1980. A guide to remote sensing for wild- life biologists. Pages 291-303. In S. D. Schemnitz (ed.). Wildlife man- agement techniques manual. The Wildlife Society, Washington, D.C. viii + 686 p. Auffenberg, Walter and Richard Franz. 1982. The status and distri- bution of Gopherus polyphemus. p. 95-126. In R.B. Bury (ed) North American tortoises: Conservation and ecology. U.S. Fish and Wild- life Service Research Rep. No. 12. Bozeman, John. R. 1971. A sociologic and geographic study of the sand ridge vegetation in the Coastal Plain of Georgia. Ph.D. disserta- tion. University of North Carolina, Chapel Hill. 244 p. Brabander, Jerry J. and John S. Barclay. 1977. A practical applica- tion of satellite imagery to wildlife habitat evaluation. Proceedings of the Annual Conference of South- eastern Fish and Wildlife Agen- cies. 31:300-306. Diemer, Joan E. 1984. Tortoise reloca- tion in Florida: solution or prob- lem? Proceedings of the 9th An- nual Symposium of the Desert Tortoise Council, p. 131-135. Diemer, Joan E. 1986. The ecology and management of the gopher tortoise in the southeastern United States. Herpetologica 42(1):125- 133. Diemer, Joan E. and Dan W. Speake. 1981. The status of the eastern in- digo snake in Georgia. In Ron R. Odom and J. W. Guthrie (eds.). Proceedings of the Nongame and Endangered Wildlife Symposium. Ga. Dept. Nat. Resources, Game & Fish. Tech. Bull. WL5. p. 52-61. Diemer, Joan E. and Dan W. Speake. 1983. The distribution of the east- ern indigo snake (Drymarchon co- rais couperi) in Georgia. Journal of Herpetology 17:256-264. Garner, James A. and J. Larry Lan- ders. 1981. Foods and habitat of the gopher tortoise in southwest- em Georgia. Proceedings of the Annual Conference of the South- eastern Association of Fish and Wildlife Agencies. 35:120-133. Graham, Marcellus H., Ronnie W. Pearson, Benjamin R. Seyfarth, Bobby G. Junkin, Maria T. Kalcic. 1980. ELAS, earth resources labo- ratory applications software. Re- port No. 183 of the NASA, Na- tional Space Technologies Labora- tories, Earth Resources Labora- tory. Hubbard, Henry. G. 1894. The insect guests of the Florida land tortoise. Insect Life 6:302-315. Hutt, Art. 1967. The gopher tortoise, a versatile vegetarian. Florida Wildlife. 21(7):20-24. Landers, J. Larry. 1981. Techniques for restocking gopher tortoise populations. In The Gopher Tor- toise: Distribution, Ecology, and Effects of Forest Management. Fi- nal Report to Georgia Department of Natural Resources. Landers, J. Larry and James A. Gar- ner. 1981. Status and distribution of the gopher tortoise in Georgia, p. 45-51. In Ron R. Odom and J. W. Guthrie (eds.). Proceedings of the Nongame and Endangered Wildlife Symposium. Ga. Dept. Nat. Resources, Game & Fish Tech. Bull. WL5. Landers, J. Larry, James A. Garner, and W. Alan McRae. 1980. Repro- duction of gopher tortoises (Go- pherus polyphemus) in southwest- em Georgia. Herpetologica 36(4):353-361. Landers, J. Larry and Dan W. Sf)eake. 1980. Management needs of sandhill reptiles in southern Georgia. Proceedings of the An- nual Conference of the Southeast- em Association of Fish and Wild- life Agencies. 34:515-529. Mount, Robert H. 1975. The Reptiles and Amphibians of Alabama. Ala- bama Agricultural Experiment Station, Auburn. 347 p. Speake, Dan W. and James A. Alti- ere. 1983. A device for viewing and filming the contents of tree cavities, p. 185-187. In Davis, J. W., Gregory, A. F. and R. A. Ocken- tield (Tech. Coords.) 1983 Snag Habitat Management: proceedings of the symposium. USDA Forest Service General Technical Report RM-99. 226 pages. Rocky Moun- tain Forest and Range Experiment Station, Ft. Collins, CO. Speake, Dan W., Joe A. McGlincy, Thaggard A. Colvin. 1978. Ecology and management of the eastern indigo snake in Georgia: a prog- ress report. Pages 64-73. In Ron R. Odom and Larry Landers (eds.). Proceedings of the Rare and En- dangered Wildlife Symposium. Ga. Dept. Nat. Resources, Game & Fish Tech. Bull. WL4. Taranik, James V. 1978a. Characteris- tics of the Landsat Multispectral Data System: U.S. Geological Sur- vey Open-file Report 78-187, 76 p., Sioux Falls, South Dakota. Taranik, James V. 1978b. Principles of computer processing of Landsat data for geologic applications: U.S. Geological Survey Open-file Re- port 78-117, 50 p., Sioux Falls, South Dakota. Wharton, Charles H. 1978. The natu- ral environments of Georgia. Georgia Dept. Nat. Res. 277 p. Woodruff, Robert E. 1982. Arthro- pods of gopher burrows, p. 24-48. In R. Franz and R. J. Bryant (Eds.), The Gopher Tortoise and Its Sandhill Habitat. Proceedings of the 3rd Annual Meeting, Gopher Tortoise Council. 204 Evaluation and Review of Field Techniques Used to Study and Manage Gopher Tortoises^ Russell L. Burke^ and James Cox^ Abstract. —This paper reviews methods used to census gopher tortoises as well as techniques for demographic, reproduction, and movement studies. We also evaluate a refinement for line transect estimates of gopher tortoise abundance. In situations where dense vegetation structure may hinder abilities to locate burrows along transects, Fourier series estimators of abundance con be used to overcome the problem, However, our results indicate that many transects may be needed to provide precise estimates of gopher tortoise abundance over large areas. The collection of vegetation data along transects may also be helpfu in evaluating habitat preference in this species. Introduction Of the approximately 107 genera and 267 species of North American rep- tiles, two species of tortoises have received a relatively large amount of scientific attention. Organizations dedicated to the conservation and protection of the gopher tortoise (Go- pherus polyphemus) (The Gopher Tor- toise Council) and the desert tortoise (G. agassizi) (The Desert Tortoise Council) attest to heightened levels of amateur and scientific interest in these species. Past bibliographies (Diemer 1981, Douglass 1975, Douglass 1977, Hohman et al. 1980) together record over 775 different publications concerning the genus, and more have been published since then. Compared to most other reptile species, an exceptional diversity of techniques has been employed, and many field methods have been devel- oped and used to study their status and biology. The gopher tortoise is a large ter- restrial turtle (15-37 cm carapace length, 3.6-5.0 kg) that exhibits low rates of juvenile recruitment, extreme 'Paper presented at syrDposium, Man- agement of Amphibiarts. Reptiles, and Small Mammals in Northi America. (Flag- staff. AZ, July 19-21 1988.) 'Researct) Associate. Tall Timbers Re- search! Station. Route 1. Box 678. Tallahas- see. Florida. 32312. 'Biologist. Nongame Wildlife Program. Florida Game and Fresh Water Fish Com- mission. 620 S. Meridian Street. Tallahassee. Rorida. 32399-1600. adult longevity, and persistent use of a small number of burrows, often in a loose aggregation of 10 to 15 indi- viduals. As a result, tortoises display a social system that involves indi- viduals who may have interacted regularly for decades (Douglass 1976, Landers et al. 1980, McRae et al. 1980). Tortoises were once a common feature of the upland habitats of the southeastern coastal plain (Auffen- berg and Franz 1982), but the species is now less common and appears on several state and federal lists of rare or endangered species (Lohoefener and Lohmeier 1984, Wood 1987). The principal forces driving these popu- lation declines are rapid urbaniza- tion, certain forest management prac- tices, and human predation (Diemer 1986). Gopher tortoise burrows are im- portant to a large wildlife commu- nity, and 332 other species have been documented to use tortoise burrows at least occasionally (Jackson and Milstrey in press). Included among the several rare species that rely heavily on tortoise burrows are the Florida mouse (Podomys floridanus), Florida and dusky crawfish frogs (Ram areolata aesopus and R. areolata sevosa), sand skink (Neoseps reynoldsi), Florida pine snake (Pituophis melano- leucus mugitus), and eastern indigo snake (Drymarchon corais couperi). In this paper we review tech- niques used in field research on the gopher tortoise community. We also discuss future areas of research and analyze the use of Fourier series esti- mators (Burnham et al. 1980) in line transect censusing techniques. In doing so we suggest appropriate methods for future work, standard- ize some techniques, bring some lesser known techniques to the fore, and suggest refinements to com- monly used methods. Estinnating Population Size Burrow Count Transects Burrow-count transects are currently the most widely used method for es- timating the size of local gopher tor- toise populations, though some tor- toise populations do not dig burrows (Auffenberg 1969), while others may use seven or more burrows per indi- vidual (McRae et al. 1980). Burrows are particularly amenable to transect analysis since they are stationary and generally visible in many of the open areas occupied by gopher tortoises. Transects also require little equip- ment, can be used to cover relatively large areas in a short time, and can be used to estimate abundance over a large area using random or stratified- random sampling procedures. A con- version factor (Auffenberg and Franz 1982) is used to relate the number of different tortoise burrows to the number of gopher tortoises in an area. The dimensions of reported transects ranges from 100 to 250 m in 205 length to 7 to 10 m in width (Auffen- berg and Franz 1982, Cox et al. 1987, Lohoefener and Lohmeier unpub. rep.). Lohoefener (in press) points out that strip transect burrow counts assume that all burrows are detected within a strip. Breininger et al. (in press), however, expressed concern that dense vegetation could make strip-transect estimates unreliable unless the transects were narrow. The thick oak scrub (Quercus spp.) vegetation common on many of their study sites, for example, would have prohibited surveyors from seeing burrows more than a few meters from transect lines. A possible method of correcting this problem (Cox et al. 1987, Lohoefener in press) is to take {per- pendicular distance measures from transect lines to observed gopher tor- toise burrows. Perpendicular dis- tances can be used in Fourier series density estimators (or other estima- tors) (Bumham et al. 1981) to account for differences in the detectability of burrows due to vegetation or the size of the burrow. To look at this problem in more depth, we compared strip transects and line transects by establishing 12 transects (250 m by 20 m) in each of three areas containing gopher tor- toise populations. The areas selected had noticeable differences in vegeta- tive structure. The first site was a mixed longleaf pine (Pinus palustris), turkey oak (Quercus laevis) habitat on a private ranch; the second site was an early successional sand pine scrub (P. clausa) forest on private timber lands; and the third site was a ma- ture longleaf pine forest in the Apalachicola National Forest. The starting points and directions of transects within these areas were randomly selected. Perpendicular distances from bur- rows to transect lines were measured to the nearest 0.25 m, and only bur- rows detected from the transect line were recorded (i.e., burrows located while measuring perpendicular dis- tances to burrows seen from the transect line were ignored). Burrow densities for each of the three areas were estimated directly using the number recorded on transects and Fourier series estimators obtained from perpendicular distance data (table 1)". Fourier series estimators were calculated using the TRANSECT program developed by Laake et al. (1979) and are presented in table 1 for the three sites. Vegetation structure appeared to influence the estimate of burrow den- sity on the early successional site (Site 2), but the Fourier series esti- mate of density was no different than the estimate provided by direct com- putations on the other sites. The early successional site had a very thick shrub component that made it difficult to locate burrows several meters from the transect line. Ten meters was probably too wide a transect width in this particular set- ting. The direct computation of bur- row density from transect data on Site 2 is only half the density estimate developed by the Fourier series esti- mate. The level of variation observed among transects (whether they be strip or line transects) within a site can be used to estimate the number of additional transects needed to at- tain a higher level of accuracy for the estimate of density (Bumham et al. 1981). To increase the precision of our estimates by 10%, for example, an additional 24 transects would be needed for Site 1, 40 for Site 2, and 78 for Site 3. Such an analysis can help determine whether additional sur- veys are needed, given the level of accuracy desired. For some ques- tions, levels of accuracy of 20-30% may be acceptable. Detecting small burrows of juve- nile tortoises in transect sampling can be particularly difficult even in fairly open habitats (Douglass 1978). This problem weakens the reliability of transect data in estimating the abundance of juveniles. Fourier se- ries estimators again could be used, in conjunction with an estimate of burrow size, to gauge detectability of small burrows, but extremely large samples are probably needed to ob- tain an accurate detectability func- tion and estimate of abundance for smaller tortoises. Point-Center Burrow Counts Tortoises often form small colonies of aggregated burrows (McRae et al. 1980), and H. Mushinsky and E. McCoy (Pers. comm.. University of South Florida, Tampa, Florida) use a point-center method (Cottam and Curtis 1956) to estimate the size of tortoise colonies. The approximate center of the aggregation of burrows is estimated, and the center point of the census station is placed there. The distance from the center point to several tortoise burrows is deter- mined, and a burrow density esti- mate is derived using standard point-center calculations (Cottam and Curtis 1956). If the abundance of tortoises over a large area is desired, all aggregations should be located. Ottier Indirect Estimates of Density In some situations (e.g., intensive col- ony analysis or preparation for popu- lation relocation), complete burrow counts are needed. We have used teams of 6 to 12 inexf)erienced field assistants, spaced at arm's length, to Table 1 .—Mean burrow density esti- mates (burrows per ha) and stan- dard deviations calculated from transect data using Fourier series estimators (D) and direct computa- tions. Data were collected at three sites in north Florida. Location Fourier series Direct estimator