A Survey Method for Measuring Gopher Tortoise Density and Habitat istributionl Daniel M. Spillers and Dan W. Speake2 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. The only tortoise to occur in the southeast, the gopher tortoise (Gopherus polyphemus) (fig. I), is limited to six states. Of these six states, legal protection is offered by South Carolina, Mississippi, Georgia, Florida and Alabama; Louisiana does not restrict 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 populations has been noted. Bozeman (1971) and Wharton (1978) noted the rapid loss and alteration of sand ridge habitat, the habitat in which most gopher tortoise populations occur, and argued for the preservation of these habitats not only for gopher tortoises but also for other aspects of their ecological signifi- 'A contribution of the Alabama Cooperative F5h and Wildlife Research Unit: Auburn Universify Agricultural Experiment Sfation and Department of Zoology and Wildlife Science. Game and Fish Division of the Alabama Department of Conservation and Natural Resources, the US. Fish and Wildlife Service and the Wildlife Management Institute cooperating. Presented at the Symposium on Management of Amphibians, Reptiles, and Small Mammals in North America, July 19, 1988. 2Spiilers is a research technician and Speake is assistant unit /eader/wildlife with the Alabama Cooperative Fish and Wildlife Research Unit, Auburn University, Alabama 36849-54 14. cance. Auffenberg and Franz (1982) documented a decline of gopher tortoise 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 response to habitat manipulation or neglect. Other conceivable reasons for this apparent decline were noted by Diemer (1986). Sand ridge habitat is not only important for gopher tortoises, but also for many other animals that use gopher tortoise burrows for nesting, feeding, or escape cover. Three subspecies of the crawfish-gopher frog complex that are closely associated with gopher tortoise burrows are the dusky gopher frog (Ram areolata sevosa), the Florida gopher frog, (R. a. aesypus), and the Carolina gopher frog (R. a. capito). The threatened eastern indigo snake (Drymarchon corais couperi) is dependent on tortoise 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 tortoise 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 important to be able to accurately measure tortoise density in an area and to determine quantity and distribution of suitable tortoise habitat. Tortoise density has been previously estimated by means of a correction factor applied to counts of burrows (Auffenberg and Franz, 19821, digging of burrows, and use of listening devices. Previous methods do not ensure accurate determination of tortoise density without burrow destruction and prohibitive labor. Determination of quantity and location of tortoise habitat is becoming neces-
sary due to rapid changes in land use and increasing relocation and restocking efforts (Diemer, 1984; Landers, 1981). The objectives of this study were to develop and employ a method to: (1) accurately measure gopher tortoise density and (2) locate and quantify 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 cooperation in giving us access to the Landsat remote sensing system. Appreciation is extended to Joe Exum, Raymond Metzler, and Nick Wiley for their assistance in experimental design and data analysis. Special appreciation is extended to Dr. Charles Williams of the Research and Data Analysis Department, Auburn University, 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, conserva tion 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 supported or had supported tortoise populations, and names of landowners or other persons who might have additional knowledge of tortoise populations. A map was included with each questionnaire so that locations could be marked. A total of 132 questionnaires was mailed out and 58% were returned. Soil conservation offices were visited in each surveyed county and further 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 preferably contained a variety of habitat types were delineated on maps. These areas were considered potential tortoise habitat (Garner and Landers, 1981; Landers, 1981; Linders and Garner, 1981) and were used to sample tortoise densities. After evaluation of the information from the questionnaires, personal 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 populations and habitat. Class I1 counties (n=4) contained relict or disjunct populations and scattered, spotty habitat. Class 111 counties (n=6) were those in which no tortoise populations could be found. Sampling Scheme In Class I counties, regions delineated by the soil conservation agents (sandy soil > 1 m) were located on 1 :24,OOO 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 visiting the site. Where necessary, permission 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, oldfields, 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 located within the habitat types available; edge h-ansects 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 device enabled us to insert a closed- U not in study Figure 2.-Distribution of the gopher tortoise in 24 counties of Alabama.
circuit television camera to the bst- type, number of open burrows, numtoms of the burrows and determine if ber of active burrows (burrows with they were occupied (figs. 3-5). Bur- sign of recent tortoise use), number row width measurements were made of tortoises, and width of burrows. with calipers inserted approximately In Class I1 counties we searched 70 cm into the burrow. Data gathered each area where tortoise populations for each transect included habitat had been reported or where gopher tortoise habitat (sandy soil > 1 m) existed. 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/active burrow) was obtained from our sampling of Class I counties by dividing the total number of tortoises by the total number of active burrows. The estimated total number of tortoises for Class I1 counties was very low (561, and did not significantly affect our population estimate. Landsat Satellite Imagery Figure 3.-Closed-circuit television camera with protective glass globe. Figure 4.-Crew inserting closed-circuit television camera into gopher tortoise burrow. 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 counties. Characteristics and usage of this remote sensing technique are described by Anderson, Wentz and Treadwell (1980), Brabander and Barclay (1977), Diemer and Speake (1983), Graham et al. (1981), Taranik (1978a1, and Taranik (1978b). The system we used makes a scan of the earth every eighteen days from a geosynchronous orbit. The multispectral scanner operates in seven different wavelengths of light-four visible and three infrared. We used near infrared because it showed vegetation 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 values based on its reflectance. Using these gray values we separated the following habitat types based on their spectral signature: unburned
pine/scrub oak, burned pine/scrub oak, planted pine, old-field, agricultural fields, pasture and composite edge. Before sampling we used groundtmthing to determine if it was feasible to attempt to classify each habitat type using Landsat imagery. On 70-0.4 ha sample plots in Baldwin County (10 plots in each habitat type), each plot was correctly classified. Clearcuts were not included because 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 rn tall). Individual 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 Landsat imagery includes a program for referencing Landsat digital data to any scale map. We referenced our data to standard 1:24,000 topographic maps using known control points. This enabled us to use Universal Trans Merca tor coordinates to locate each transect on the Landsat image and obtain the correct gray value for each transect. We then assigned 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. estimate based on mean tortoise density per hectare multiplied by the estimated 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 hectare 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. Landers 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 approximately 100-120 mm. During the juvenile stage, the shells are very soft and carapacial scutes usually have distinct 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 volume 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 mathematical relationship between the widths of gopher tortoise burrows and the carapace lengths of their occupants in northern Florida (this relationship has not been thoroughly tested in other states). Using Alford's equation logl0y = 0.879 loglox + 0.149, where y is carapace length and x is burrow width, we used our burrow width measurements of occupied burrows to divide tortoise populations into juvenile, subadult, and adult age classes. We considered age class structure to be an important cri- Data Analysis We had two concerns relative to data analysis: (1) to derive a population Figure 5.-Closed-circuit television monitor displaying picture of a gopher tortoise inside a burrow,
teria along with density in evaluating tortoise population viability. Research has not yet revealed an optimum age class structure. Intuitively, in a long-lived animal such as the gopher tortoise, the age class structure 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 population (n=100 tortoises) were juvenile tortoises, 48% were subadult, and 47% were adults. This structure shows that there has been recent reproduction, and that there is a large segment of breeding size adults present. This suggests that the potential for successful population maintenance 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 values 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 correspond 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 variance was fairly low. Increased sample size would likely lower this error. The second source of error is in estimating total areas of the habitat types over a large region. Although in our preliminary groundtruthing, Landsat imagery correctly classified all our habitat types (excluding 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 classification to determine what percentage has been misclassified, which would allow the researcher to make allowances for this error in final cornputations. Conclusions We found this technique to be useful for measuring tortoise density and for determining quantity and location of tortoise habitat. The error in this technique seems to be less than that for techniques used for censusing 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 density in small areas and then extrapolate this density on the basis of a
quantitative measurement of a designated 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 ~olyphemus in northern Florida. Journal of Herpetology. 14:177-182. Allen, Ross. and Wilfred. T. Neill. 19%. The gopher. Florida Wildlife. 5(7):10,32. Anderson, William H., W. Alan Wentz, B. Dean Treadwell. 1980. A guide to remote sensing for wildlife biologists. Pages 291-303. In S. D. Schemnitz (ed.), Wildlife management techniques manual. The Wildlife Society, Washington, D.C. viii + 686 p. Auffenberg, Walter and Richard Franz. 1982. The status and distribu tion of Gopherus polyphemus. p. 95-126. In R.B. Bury (ed) North American tortoises: Conservation and ecology. U.S. Fish and Wildlife 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. dissertation. University of North Carolina, Chapel Hill. 244 p. Brabander, Jerry J. and John S. Barclay. 1977. A practical application of satellite imagery to wildlife habitat evaluation. Proceedings of the Annual Conference of Southeastern Fish and Wildlife Agencies. 31:300-306. Diemer, Joan E. 1984. Tortoise relocation in Florida: solution or problem? Proceedings of the 9 th Annual 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 eastern indigo snake (Drymarchon corais couperi) in Georgia. Journal of Herpetology 17:256-264. Garner, James A. and J. Larry Landers. 1981. Foods and habitat of the gopher tortoise in southwestern Georgia. Proceedings of the Annual Conference of the Southeastern 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 laboratory applications software. Report No. 183 of the NASA, National Space Technologies Laboratories, Earth Resources Laboratory. 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 Tortoise: Distribution, Ecology, and Effects of Forest Management. Final Report to Georgia Department of Natural Resources. Landers, J. Larry and James A. Garner. 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. Reproduction of gopher tortoises (Gopherus polyphemus) in southwestern Georgia. Herpetologica 36(4):353-361. Landers, J. Larry and Dan W. Speake. 1980. Management needs of sandhill reptiles in southern Georgia. Proceedings of the Annual Conference of the Southeastern Association of Fish and Wildlife Agencies. 34:515-529. Mount, Robert H. 1975. The Reptiles and Amphibians of Alabama. Alabama Agricultural Experiment Station, Auburn. 347 p. Speake, Dan W. and James A. Altiere. 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. Ockenfield (Tech. Coords.) 1983 Snag Habitat Management: proceedings of the symposium. USDA Forest Service General Technical Report RM-99.226 pages. Rocky Mountain 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 progress report. Pages 64-73. In Ron R. Odom and Larry Landers (eds.), Proceedings of the Rare and Endangered Wildlife Symposium. Ga. Dept. Nat. Resources, Game & Fish Tech. Bull. WL4. Taranik, James V. 1978a. Characteristics of the Landsat Multispectral Data System: U.S. Geological Survey 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 Report 78-117,50 p., Sioux Falls, South Dakota. Wharton, Charles H. 1978. The natural environments of Georgia. Georgia Dept. Nat. Res. 277 p. Woodruff, Robert E. 1982. Arthropods 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.