EFFECTS OF POPULATION DENSITY ON PATTERNS OF MOVEMENT AND BEHAVIOR OF GOPHER TORTOISES (GOPHERUS POLYPHEMUS)

Transcription

1 Herpetological Monographs, 26, 2012, E 2012 by The Herpetologists League, Inc. EFFECTS OF POPULATION DENSITY ON PATTERNS OF MOVEMENT AND BEHAVIOR OF GOPHER TORTOISES (GOPHERUS POLYPHEMUS) CRAIG GUYER 1,3,VALERIE M. JOHNSON 1,2, AND SHARON M. HERMANN 1 1 Department of Biological Sciences, Auburn University, Auburn, AL 36849, USA ABSTRACT: Patterns of burrow size, tortoise size, home range size and overlap, movement distances, and mating rates were compared among six sites that differed in density of Gopher Tortoises. Burrow sizes differed among sites because tortoise size distributions differed among sites, but this was due principally to the unusually small size of animals on the Conecuh National Forest. A linear relationship between tortoise density and burrow density was documented from the six sites, suggesting that tortoises, on average, created 2.5 burrows per site or that the burrow-to-tortoise conversion factor for our sites was The average distance from a burrow to its nearest three neighbors was greater for low-density sites than for high-density sites, indicating that animals probably were more isolated from each other on sites with low tortoise densities. Tortoise home ranges were larger in males than females, a feature documented in other studies of tortoise movements. Home range sizes were greatest for densities of approximately 0.4 tortoises/ha and decreased in size above and below this density. This suggests that animals moved to visit close neighbors in areas of high density, expanded movements to maintain contact with neighbors that became more widely dispersed as density decreased, and then restricted movements to a few close neighbors as density reached extremely low levels. Home range overlap increased linearly with increasing density, suggesting that opportunities for social interactions decreased with decreasing density. When tortoises moved between burrows, males moved longer distances than females and tortoises of both sexes moved shorter distances on high-density sites than did tortoises on low-density sites, suggesting greater movement costs for males than females and for tortoises on low-density sites. Males traveled up to 500 m to visit female burrows, but most movements were, 80 m. Median movement distances of males to visit females were negatively correlated with burrow density, suggesting that cost of male movements to find mates increased as population density decreased. Based upon patterns of 95% confidence limits, rates of mountings of female tortoises approached zero when females occupied burrows approximately 200 m from neighboring burrows. If burrows were uniformly distributed 200 m apart, then reproductive failure would be a statistically supportable outcome at a density of 0.3 burrows/ha (0.12 tortoises/ha). These values are similar to the values of 0.4 tortoises/ha (1.0 burrows/ha) that our data suggest is the density at which social structure associated with movements within home ranges are altered. Key words: Density dependence; Home range; Home range overlap; Reproductive costs; Gopher Tortoise; Gopherus polyphemus GOPHER TORTOISES (Gopherus polyphemus) are land-dwelling turtles that are federally protected in the western portion of their geographic range and protected by state legislation throughout the remainder of the distribution (Diemer, 1986). This species is particularly important in conservation planning for the southeastern United States because it is thought to be a keystone taxon within the Longleaf Pine (Pinus palustris) ecosystem of the Coastal Plain. This habitat contains an unusually diverse flora and fauna (Noss et al., 1995), owing in part to the burrows created by Gopher Tortoises (Guyer and Bailey, 1993). These burrows serve two 2 PRESENT ADDRESS: US Geological Survey, Western Ecological Research Center, 6924 Tremont Road, Dixon, CA 95687, USA 3 CORRESPONDENCE: , guyercr@auburn.edu functions in Longleaf Pine forests. First, they create overwintering sites, homes, or shelters for. 300 species of animals, some of which are found nowhere else in nature (Jackson and Milstrey, 1989). Second, they represent a small-scale disturbance to the understory landscape. This disturbance results from a large volume of sand being brought to the surface when a tortoise digs a new burrow. This soil is piled at the entrance of the burrow, creating a bare patch of ground that initially may include increased levels of soil nutrients and that alters demographics of some understory plants (Kazor and Harnett, 1990). Habitat loss, lack of effective prescribed fire, and fragmentation associated with land use practices of humans are primary causes of imperilment of both the Longleaf Pine ecosystem (Noss et al., 1995; Outcalt, 2000) and Gopher Tortoises (Mushinsky and McCoy, 122

2 2012] HERPETOLOGICAL MONOGRAPHS ). Long-term conservation of Gopher Tortoises depends upon distinguishing areas where management efforts can maintain or increase population density from those where populations are likely to decline. Many areas of conservation concern contain tortoises at low population densities. Populations on these sites may not be viable because animals are so widely dispersed that neighboring individuals cannot find each other, causing reproduction to fail. Because tortoises are such long-lived creatures, declining populations can persist in the landscape for decades before becoming extirpated (McCoy et al., 2006). Three key variables needed to assess population viability are age at first reproduction, fertility, and mortality (Cox et al., 1987). However, these variables are not well known for populations of Gopher Tortoises (Smith et al., 1997) because tortoises are long lived (Landers et al., 1982), making these variables difficult to measure directly. Here, we evaluate population viability indirectly by examining the rates of behavioral interactions of tortoises at burrows occupied by females. One such rate, the rate at which males attempt to copulate with females, measures important aspects of reproduction that may affect opportunities for multiple paternity within clutches (Moon et al., 2006), fertility of females (Boglioli et al., 2003), or mating system (Tuberville et al., 2011). Because tortoises in low-density populations are widely dispersed, their mating opportunities may depend upon the ability of male tortoises to recall the location of burrows occupied by potential mates. This recall probably becomes more problematic in an environment in which animals are required to move their burrows frequently to avoid canopy closure created by growth of overstory trees or understory shrubs (Aresco and Guyer, 1999). Here, we summarize the findings of 7 yr of research designed to examine the effect of Gopher Tortoise density on patterns of movement and socialization. This information is vital because it allows land managers to determine the tortoise density at which conservation strategies should switch from management of tortoises in situ, presumably because a viable population is present, to a strategy of moving animals to reserve areas, presumably because this will allow animals from nonviable areas to participate in a viable population. Because tortoises affect so many other taxa within the Longleaf Pine ecosystem, conservation of these turtles should provide an umbrella for the preservation of other sensitive indicator taxa (Guyer and Bailey, 1993; Means, 2006). In addition, implementation of conservation measures throughout the rest of the geographic range of tortoises could allow recovery without requiring federal protection range wide. Three specific objectives are addressed here. First, we document variation among sites in tortoise and burrow sizes and densities. Second, we examine whether tortoise density correlates with movement patterns of adult males and females. Third, we determine whether the degree of burrow isolation correlates with mating rate of females that occupy those burrows. We use these objectives to determine whether conservation plans for Gopher Tortoises that are based on average home range size should be modified because movement patterns are affected by density. In addition, we use these data to determine whether the degree of burrow isolation can be used to infer patterns of burrow dispersion for which population viability is likely to be lost because individuals cannot find each other for mating. METHODS Study Sites Six sites that differ in tortoise density were used in this study (Fig. 1; Table 1). The site with the greatest burrow density was the Wade Tract (WT), a privately owned oldgrowth Longleaf Pine forest (Platt et al., 1988) located in Thomas County, Georgia, USA, and currently managed by Tall Timbers Research Station, Tallahassee, Florida, USA. There are old-growth trees on the site, the understory plant community has never been in agriculture, and fire has been an important feature of the landscape. For. 70 yr, the site has been burned every 1 2 yr for quail management, generally in late February to early April. For the past two decades, growing season (May August) burns have been applied in some years. This fire regime has maintained high

3 124 HERPETOLOGICAL MONOGRAPHS [No. 26 FIG. 1. Map of southeastern United States showing geographic range of Gopher Tortoises (shaded) and location of study sites. Study sites are Wade Tract (WT), Green Grove (GG), Conecuh National Forest (CNF), Dixon Forestry Education Center (DC), Mobile County (MC), and Camp Shelby (CS). habitat quality. Soils on this site are complex, with tortoises occupying Orangeburg, Faceville, Fuquay, Leefield, Lucy, and Norfolk series. Burrows on this site were marked and mapped in previous studies (Guyer and Hermann, 1997; Hermann et al., 2002) and were resurveyed during during the current study. The second high-density site, Green Grove (GG) on the Jones Ecological Research Center located in Baker County, Georgia, also was selected to take advantage of previous studies of Gopher Tortoises (Ott et al., 2002; Boglioli et al., 2003). This site has been managed as a quail-hunting plantation since the 1920s and contains widely spaced Longleaf Pine with a generally undisturbed understory plant community. Fire has been an important component of the landscape, and in recent years, frequent growing-season fires have been used to maintain habitat quality on this site. Green Grove is surrounded by dirt roads, and the topography slopes gradually toward a large wetland. Soils of Norfolk, Wagram, Orangeburg, and Lucy series are present and occupied by Gopher Tortoises. Data summarized in this report were collected during The third site, located in the Conecuh National Forest (CNF) of Covington County, Alabama, USA, was intermediate in burrow density. This site contained mixed Longleaf and Slash Pine (Pinus elliottii) growing on a ridge of sandy soils bounded on two sides by drainages of the Blackwater River and on two sides by dirt roads. The forest stand was thinned in 1995 and has been managed with growing-season fires on a 3-yr cycle since then. Soils of the Florala and Troup series were occupied by Gopher Tortoises on this site. Because the site had greater shrub and tree densities it was intermediate in habitat quality. Data summarized in this report were collected during The fourth site was the Dixon Forestry Education Center (DC), a forested property located in Covington and Escambia counties of southern Alabama and managed by the Auburn University School of Forestry and Wildlife Science. This site contained tracts of Slash and Longleaf Pines that have been managed for timber production and wildlife management for the past 30 yr. The particular area used in the study was comprised of a series of three ridges separated by small drainages. The three ridges contained soils of the Florala and Troup series with widely spaced trees creating open areas surrounded by areas of hardwoods. Dormant-season fire TABLE 1. Descriptive information for the six study sites (WT 5 Wade Tract; GG 5 Green Grove; CNF 5 Conecuh National Forest; CS 5 Camp Shelby; MC 5 Mobile County; DC 5 Dixon Center). Number of males and females includes number with radiotransmitters and number (in parentheses) captured and measured throughout study period. Dispersion index based on nearest neighbor analysis and ranges from 0 (aggregated) through 1.0 (random) to 2.15 (uniform). WT GG CNF CS MC DC Area (ha) No. of males 28 (67) 76 (107) 13 (19) 17 (20) 16 (21) 2 (13) No. of females 32 (56) 56 (70) 12 (23) 14 (17) 15 (23) 4 (11) Females with cameras No. of burrows Burrow density (ha 21 ) Tortoise density (ha 21 ) Dispersion index

4 2012] HERPETOLOGICAL MONOGRAPHS 125 was the primary management tool used on this site with a fire return interval of 5 7 yr. Because of this, shrub density throughout the site was high, resulting in poor habitat quality. Data were collected on this site during A fifth site was on lands previously managed by the International Paper Company and located north of Chunchula in Mobile County (MC), Alabama. This site was an active Slash Pine plantation managed on a short rotation for production of pulp wood. In 1998, this site was clearcut, treated with herbicides, and replanted with Slash Pine. Tortoise burrows were found on soils of the Troup-Heidel and Shubuta-Troup series. Fire was not used as a management tool for at least 10 yr previous to our evaluation of the site. Natural regeneration of Longleaf Pine was allowed, but no further management activities occurred, causing the site to be of poor quality. Data were collected during The sixth site was located at Camp Shelby (CS) in Forrest County, Mississippi, USA, near the western limit of the geographic range of Gopher Tortoises. The site was along a north south ridge, bounded to the east by a drainage and to the west by a dense forest bordering a highway. Tortoise burrows were found on soils of the McLaurin, Benndale, and Heidel series. Forested areas were dominated by mature Longleaf Pines that were thinned and received a prescribed fire 3 yr before use in this study. The site served as an orienteering course for military training and also was used for hunting, management activities that caused it to be intermediate in habitat quality. Data on tortoises were collected during Field Methods Each site was surveyed for burrows of Gopher Tortoises. On CNF, CS, and MC, site boundaries were generated by expanding out from a central core of burrows until no further burrows were located within 100 m of any peripheral burrow. In this way, an entire cluster of burrows was delimited. At WT, all burrows were marked and mapped on a conservation easement established for the study of old-growth forest (Platt et al., 1988). A contiguous subset of these was selected for study because they represented the core of the conservation area and generated a study area of comparable characteristics to the other sites. At GG, roads were used to delimit an artificial boundary to the study site. On DC, an area containing five distinct clusters of isolated burrows was selected to represent sites with extremely low burrow densities. These clusters were separated by habitat with dense canopy cover created by hardwood encroachment into the uplands. This landscape required tortoises to travel great distances (up to 2000 m) through inhospitable habitat to find a neighboring cluster of burrows. Burrows on all sites were marked (numbered aluminum tag), recorded as to status (active, inactive, abandoned; as in Hermann et al., 2002), mapped (nearest 1.0 m based on global positioning system), and measured (width to nearest 1.0 mm at 50 cm depth from entrance). Mapped locations were used to determine minimum distances between burrows as well as the degree of burrow isolation (distance to nearest three neighboring burrows) for each burrow. Tortoises were captured at active burrows via wire live traps placed at a burrow entrance. We covered each trap with burlap to provide adequate shading of captured animals and checked each trap twice daily (generally 1000 and 1400 h). Each captured tortoise was weighed (nearest 1.0 g) and measured (nearest 1.0 mm; following McRae et al., 1981a), categorized as to sex (based on McRae et al. 1981a) and age (following Aresco and Guyer, 1998), marked (permanent mark via holes drilled in marginal scutes; temporary mark via numbers applied with paint sticks on top of shell), affixed with a radiotransmitter (American Wildlife Enterprises; 2-yr battery life), and released to the burrow of capture. At all sites except GG, we had enough traps to place one at each burrow likely to have a tortoise during an early spring (April May) trapping period. At GG tortoises in sections of approximately 50 contiguous burrows were trapped for up to 1 mo starting in May and ending in August, by which time all active burrows had been sampled. At all sites except MC, tortoises were trapped during two spring trapping periods before data used in this study were collected,

5 126 HERPETOLOGICAL MONOGRAPHS [No. 26 followed by one complete season (May October) of sampling used in this study. Because of the presample effort, we assumed that all resident tortoises were known and included in the study. For MC, tortoises were trapped and monitored over two consecutive years without prior sampling of tortoises, and we assumed all residents were included. Burrow Surveys and Tortoise Sizes Three sets of statistical tools were used to examine data describing characteristics of burrows and tortoises. First, distributions of burrow size (counts of burrows in bins of 50-cm width based on measures of burrow width), tortoise size (counts of tortoises in bins of 50-cm width based on measures of carapace length), and burrow isolation (counts of burrows in bins of 100-m width based on mean distance to nearest three burrows) were compared among sites by two-way (site, bin class) contingency tables (G-test of association). Second, linear regression was used to examine the effect of tortoise density (total number of animals captured divided by study area) on burrow density. Finally, burrow dispersion on each site was calculated based on the method of Clarke and Evans (1954) and compared against an expectation of 1.0 for random dispersion (. 1.0 indicating uniform dispersion;, 1.0 indicating aggregated dispersion). Telemetry Data Tortoises were relocated with a radioreceiver three to five times each week during one active season. The burrow occupied on each date was recorded. Based on these locations, minimum movement distances were measured as the straight line connecting burrows occupied consecutively by individual tortoises. Similarly, home range size for each animal was calculated as the 100% minimum convex polygon (Kie et al., 1994) of burrows occupied by individual animals. The polygon method was used instead of the kernel method because the polygon method allowed all sites to be compared with information from GG published by Ott et al. (2002). Home range overlap was measured by recording the total number of tortoises that used each burrow and then comparing the frequency distribution of these data among sites. Data were examined in two ways. First, variation in burrow isolation, movement distance, and home range area were evaluated with nonparametric analysis of variance (ANOVA). For burrow isolation a one-way (site) design was used and for movement distances and home range area a two-way (site and sex) design was used. The second analytical tool implemented was linear regression, which was used to assess the effect of tortoise density on movement distance, home range area, and home range overlap. Camera Data A camera method developed by Guyer et al. (1997) was used to assess the effect of burrow isolation on rates at which males mounted females. With this method, 4 13 females were targeted on each study site and cameras recorded behavioral interactions at all burrows used by these animals. The cameras were triggered by a pressure-sensitive switch and recorded any individual entering or exiting a burrow occupied by a target female. Because target females could be followed via telemetry, the camera was moved to follow each target female for an entire field season. Because males were followed via telemetry, an estimate of minimum distance traveled was recorded for each male that appeared in a photo. In addition, because mating typically occurs on the aprons of burrows occupied by females (Boglioli et al., 2003), the rate at which females were mounted by males (total number of mount events divided by total time that a camera was functional) was recorded for each burrow occupied by a target female. For any burrow occupied by more than one female, an average of the rate of mounting for each female was used to describe a single overall rate for that burrow. A nonparametric one-way ANOVA was used to test for differences among sites in the distances that males traveled to visit females. Linear regression was used to examine the effect of tortoise density on median distance moved by males and nonparametric local polynomial regression (LOESS) was used to examine the relationship between mounting rates and degree of burrow isolation. The 95% confidence limits (CL) around the LOESS regression were used to estimate

6 2012] HERPETOLOGICAL MONOGRAPHS 127 FIG. 2. Distribution of tortoise (carapace length, closed bars) and burrow (width, open bars) sizes. Study sites are Conecuh National Forest (CNF), Camp Shelby (CS), Dixon Forestry Education Center (DC), Green Grove (GG), Mobile County (MC), and Wade Tract (WT). the interburrow distance at which reproductive failure might occur and this was assumed to be the distance at which confidence limits included a mounting rate of zero. Statistical Analysis System (SAS Institute, 1996) was used for all tests, and a was set at All regressions were considered to be one-tailed because a particular direction to the slope was expected in each case. RESULTS Burrow Surveys and Tortoise Sizes The width of tortoise burrows spanned mm and burrows differed in size among the six sites (x , df 5 5, P, ). The difference was associated with the CNF that had a higher proportion of smaller burrows than the other five sites that, in turn, were remarkably similar to each other (Fig. 2). The carapace lengths of tortoises were mm. Tortoise sizes also differed among the six sites (x , df 5 5, P, ) because CNF had a higher proportion of small tortoises and WT had a higher proportion of large animals compared with the other sites (Fig. 2). The abundance of burrows was marginally significantly correlated with tortoise abundance (Fig. 3), with 60% of variation in burrow abundance being explained by tortoise abundance. The regression equation indicated that there are 2.5 burrows/tortoise, or alternatively, that approximately 40% of burrows were occupied by a tortoise. Degree of burrow isolation, as measured by distance to the nearest three neighbor burrows, differed among sites (x , df 5 5, P, ) because the maximum mean isolation distance was negatively related to tortoise density (Fig. 4). Overall, burrows at all sites were aggregated, but more strongly so for WT, MC, and DC than the other three sites (Table 1).

7 128 HERPETOLOGICAL MONOGRAPHS [No. 26 FIG. 3. Regression of burrow abundance on tortoise abundance. Telemetry Data When tortoises moved, the minimum distances traveled between consecutive telemetry locations was m (N for all sites combined; modal and median elapsed time 5 1 and 2 d, respectively). Movement distances differed between sexes (F , df 5 1, P ) and among sites (F , df 5 5, P, ). The sex difference resulted from longer movement distances by males compared with females; site differences resulted from a negative association between tortoise density and median movement distance (Fig. 5A). Home range size was ha for females and ha for males. Home ranges differed significantly between sexes (F 5 4.5, df 5 1, P ) and among sites (F , df 5 5, P, ) because males at all sites maintained larger home ranges than females and tortoises FIG. 5. Regression of interburrow movements (A, solid circles, males; solid triangles, females) and median male movement distance to visit females (B) on tortoise density. on sites of extremely low and high densities moved over smaller areas than animals at sites of intermediate densities (Fig. 6). Between 1 and 12 tortoises were observed to use each active burrow, and this measure of overlap differed significantly among sites (x , df 5 16, P, ) because the maximal FIG. 4. Regression of maximal burrow isolation on tortoise density. FIG. 6. Regression of median home range area (adult male, solid circles and dashed line; adult females, solid triangles and solid line) on density.

8 2012] HERPETOLOGICAL MONOGRAPHS 129 FIG. 7. Regression of maximal number of tortoises per burrow on density. number of tortoises sharing a burrow on each site was positively related to tortoise density (Fig. 7). Camera Data Photographic data were used for two purposes. One was to determine distances that male tortoises traveled to visit females. These moves ranged were m and differed significantly among sites (x , df 5 5, P, 0.007). This resulted because male movements were negatively associated with tortoise density (Fig. 5B). The second use of the camera data was to determine rates of mountings at burrows occupied by female tortoises. These rates decreased relatively rapidly for isolation distances of 0 25 m and then decreased slowly for increasingly isolated burrows (Fig. 8); FIG. 8. Local polynomial regression of rate of mounting on burrow isolation (mean distance to nearest three neighbors). Open circles are observed values, solid square (solid line) represents best-fit regression, and solid triangles (dashed lines) are upper and lower 95% confidence limits. N 5 154; two observations (extreme values for ordinate and abscissa) not plotted for clarity. FIG. 9. Regression of reproductive opportunity (mounts/d) on burrow dispersion. females rarely occupied burrows isolated by. 80 m from neighboring burrows. The lower CLs for a LOESS regression included 0 mounts/d when burrows were isolated by 200 m (Fig. 9). There was a significant negative linear relationship between burrow dispersion and reproductive rates at those burrows (Fig. 9). DISCUSSION Our data document widespread density dependence in Gopher Tortoise biology. Patterns of burrow and tortoise abundance found in this study indicate that the number of burrows present on a site is correlated with the number of tortoises that are present. This relationship is not particularly surprising given that tortoises dig burrows and given that other studies have found similar relationships (Burke, 1989; Breininger et al., 1991; McCoy and Mushinsky, 1992). The slope of the relationship in the current study suggests that 2.5 burrows (active and inactive categories of Hermann et al., 2002) are added for every additional tortoise. Because surveys of burrows continue to be the most efficient way to assess density of Gopher Tortoises on plots of land, it is tempting to recommend a burrowto-tortoise conversion factor of 0.4 as being representative of the northern portion of the range of Gopher Tortoises. However, our data also indicate that 40% of the variance in burrow abundance is probably explained by some other factor than the number of tortoises that are present. Others (Burke, 1989; Breininger et al., 1991; McCoy and

9 130 HERPETOLOGICAL MONOGRAPHS [No. 26 Mushinsky, 1992) have provided similar cautions regarding burrow-to-tortoise conversion factors. Nevertheless, we found that patterns of burrow sizes were generally similar to patterns of tortoise sizes. Therefore, some additional aspects of tortoise demography might be inferred from surveys of tortoise burrows (see, e.g., Alford, 1980; Wilson et al., 1991; Doonan and Stout, 1994). However, our data indicate that interpretation of patterns from burrows must account for differences in the size distributions between burrows and tortoises. Large burrows were overrepresented on all six sites, relative to the tortoises that were present. This finding was expected given that large abandoned burrows remain for decades, whereas smaller burrows fill rapidly (Guyer and Hermann, 1997). Modeling will be required to determine whether this factor alone can account for the disparity in size distributions that we observed between tortoises and burrows. Most aspects of movement of Gopher Tortoises also were density dependent. When tortoises moved from one burrow to another, males moved farther than did females, a feature documented in other studies of Gopher Tortoises (McRae et al., 1981b; Diemer, 1986, 1992; Eubanks et al., 2003). Thus, males consistently had greater movement costs than females. Such movements increased linearly as density decreased. This pattern occurred for both males and females. Previous interpretations of this pattern led to suggestions that increased movements of males should occur for them to maintain contact with females as all individuals become more widely dispersed (McRae et al., 1981b; Diemer, 1992; Eubanks et al., 2003). Our data provide additional support for this interpretation of male movements because we were able to document that males move longer distances to visit burrows occupied by females in areas where tortoise density is low. Our observation that females increased movement distances as density decreased is unexpected because mate searching is thought to be largely a feature of males (McRae et al., 1981b; Diemer, 1992; Eubanks et al., 2003). Females might move more widely to find food resources on low-density sites because habitat quality generally is reduced on such sites. Reduction in habitat quality occurs largely because past exclusion or ongoing ineffective use of fire increases overstory and midstory woody growth that creates a shaded environment with limited ground cover (Hermann et al., 2002; McCoy et al., 2006). However, previous correlative studies failed to document a strong relationship between understory forage and burrow abandonment (Aresco and Guyer, 1999) or burrow density (McCoy and Mushinsky, 1995). Therefore, factors other than forage seem to regulate movements of females. One likely alternative explanation is that females play an active role in social behaviors, including reproduction. Support for this idea includes the observations that females on four of the six sites examined in this study frequently moved to share burrows with males and replaced individuals from burrows occupied previously by males much more frequently than burrows occupied previously by females (Johnson et al., 2009). Regardless of the reason for tortoise movements among available burrows, our data indicate that reduction of 1 tortoise/ha results in an increase of m/move for remaining individuals. Although annual energy budgets have been calculated for tortoises (e.g., Peterson, 1996), movement costs are not part of these budgets. However, some have speculated that movement costs differ between sexes and among seasons (Jodice et al., 2006). We now speculate that these costs might become prohibitive as tortoise density decreases. Gopher Tortoises also altered home range size and overlap in a density-dependent manner. Home range size was larger for males than females, a feature documented in other studies (Diemer, 1992; Eubanks et al., 2003), and this pattern occurred regardless of density. However, home range sizes were reduced in high density areas, increased on sites as density diminished to approximately 0.4 individuals/ha, and then were reduced on sites of extremely low density. In conjunction with these changes in movement patterns, the degree of overlap of tortoises was strongly and positively associated with tortoise density. These results suggest that individuals first increase movement distances to maintain contact with neighbors as density decreases,

10 2012] HERPETOLOGICAL MONOGRAPHS 131 even though the number of neighbors is reduced. When density reaches a critically low level, home range movements cease expanding and become reduced as small clusters of individuals restrict interactions to nearest neighbors. Such an explanation is consistent with our observation that rates of mounting of females by males are increased on sites where individuals are most strongly aggregated. At first glance, reduction of home range size in areas of extremely low density is contradicted by movement distances that are linearly related to burrow density. Expanding movement distances on sites with extremely low density would seem to require a linear increase in home range area as density declines. However, this would only be true if movements within home ranges were unaffected by tortoise density. We suspect this is not true. Instead, we argue that tortoise home ranges are characterized by single, central, core areas when animals live in high densities. Within such home ranges, movements are dominated by short distances between neighboring burrows within the core area. As density decreases, this pattern of movements is maintained among increasingly more widely dispersed burrows until a critical low density is reached. Below this critical low density, animals reduce total home range area but continue to increase movement distance by moving principally among widely spaced burrows around the periphery of the home range, presumably to maintain contact with neighboring individuals. Daily rates of mounting behaviors were altered by burrow isolation and dispersion. The majority of burrows selected by females on all sites were within 80 m of the nearest three neighboring burrows. Mounting rates decreased relatively rapidly as interburrow distances increased from the most aggregated burrows to burrows of intermediate isolation (30 m to nearest three neighbors). As burrow isolation continued to increase, further reduction in mounting rates occurred at a slower pace, presumably because of increased movements of both males and females to maintain contact with each other. Thus, mounting behaviors were observed even at our most isolated burrow (800 m to nearest 3 neighbors). This suggests Gopher Tortoises have strong spatial memory and that female locations are known by males even for those rare events when females occupy extremely isolated burrows. More typically, Gopher Tortoises on sites with extremely low densities aggregate in small groups of individuals (as small as one male and one female) within which mating continues to occur, but with a reduced diversity of mating opportunities for males and females. Our main goal was to determine the density of tortoises at which conservation efforts should switch from one in which animals are managed in situ to one in which animals from low density sites are moved to dedicated conservation areas where populations could be augmented and maintained at high density through aggressive habitat management. Our data on home range movement suggest a density of 0.4 individuals/ha should trigger a change in conservation strategy because this is the point at which movement patterns among sites of equal or greater density differ only in the scale across which movements occur. At densities below this level, tortoises alter movements in ways that might alter population viability because of changes in social structure. In particular, we predict that gene flow will be reduced because females are unable to select from among several potential mates and because both sexes incur increased movement costs as they attempt to maintain contact with potential mates. Our data on rates of attempts to reproduce at burrows of differing degrees of isolation provide an independent estimate of the point at which conservation strategies should switch from management in situ to relocation of tortoises to conservation areas. Mounting rates are negatively associated with degree of burrow isolation. This association is seen most strongly over distances (10 30 m) that are unlikely to affect conservation decisions. We specifically chose the DC to serve as a site where population density might be so low that individual females would be completely isolated from contact with males. Our hope was that the observed negative relationship between reproductive rates and burrow isolation eventually would allow us to project the point at which burrows were so isolated that

11 132 HERPETOLOGICAL MONOGRAPHS [No. 26 reproduction was unlikely. Our observations that males found all females regardless of degree of isolation and that the LOESS regression model did not cross the axis of the independent variable indicate that Gopher Tortoises might continue to reproduce on sites of remarkably low density because of their remarkable ability to find distant neighbors. Nevertheless, the 95% CLs around the LOESS regression model indicate that females in burrows isolated by 200 m might be at risk of reproductive failure. If burrows were uniformly dispersed at this interburrow distance, then burrow density would be approximately 0.3 burrows/ha, or approximately 0.12 tortoises/ha (burrow-to-tortoise conversion of 0.4). This pattern of dispersion provides an independent estimate of the density at which population viability of Gopher Tortoises is compromised. McCoy and Mushinsky (2007) provide important progress in evaluating how many individual tortoises comprise a population and how much area is required to maintain such populations. Their evaluation increases substantially the target population sizes ( individuals) and minimum area requirements ( ha) compared with previous estimations offered by Cox et al. (1987) and Ott et al. (2002). We concur with their conclusion that conservation efforts for Gopher Tortoises are futile without substantial effort to improve habitat quality so that population densities are high enough to prevent the kinds of slow declines that may be occurring in conservation areas in Florida, USA (McCoy et al., 2006). Our data on movement patterns suggest mechanisms by which reproduction could be maintained on these Florida sites even though populations continue to decline. Data on patterns of burrow abandonment (Aresco and Guyer, 1999) indicate that hardwood encroachment via removal of frequent, effective application of fire can convert a burrow from one that a tortoise will choose to maintain to one that a tortoise will choose to abandon in 5 7 yr. Data on rates of recolonization of areas restored via thinning and use of frequent prescribed fires suggest that recovery of tortoise populations via immigration from surrounding areas of low tortoise density may take decades (Ashton et al., 2008). In addition, mating patterns of translocated tortoises indicate that legacy effects could influence reproduction and patterns of paternity within translocated populations for many years (Tuberville et al., 2011). All of this information suggests that efforts to conserve Gopher Tortoises either in situ or in translocated populations will require long-term monitoring to determine whether the goal of population viability is being met. The consistent message of our data and these previous studies is that conservation efforts for Gopher Tortoises must include active management that provides the open, park-like savannas that used to characterize much of the native Longleaf Pine community. Tortoise densities on such sites will be high, movement costs will be reduced, and population viability will be maximized. Acknowledgments. This project developed first as a joint venture between Auburn University and the Jones Ecological Research Center, where the data for Green Grove were generated and that served as the template for work done at all other sites. The rest of the project developed as a joint venture between Auburn University; US Fish and Wildlife Service; National Council for Air and Stream Improvement, Inc.; and the International Paper Company. We thank Will McDearman (US Fish and Wildlife Service) for conceiving of the overall idea and T. Bentley Wigley (National Council for Air and Stream Improvement) and Jimmy Bullock (International Paper) for overseeing financial, logistical, and analytical support that greatly expanded the scope of the project. Access to each site and other logistical support were provided by the following people: CNF Dagmar Thurmond, Gary Taylor, Rick Lint; CS Col. I. Pylant, Deborah Epperson; DC Rhett Johnson, Dale Pancake; GG Lindsey Boring; MC Jimmy Bullock; and WT Lane Green, Lenny Brennan, Jeptha Wade and family. Arvind Bhutta, Roger Birkhead, Melissa Boglioli, Jeannine Eubanks, Derek Fussel, Valerie Johnson, Paula Kahn, Erica Lee, Abbie Sorenson, Jimmy Stiles, Sierra Stiles, and Wes Wilkerson performed the fieldwork summarized in this report. LITERATURE CITED Alford, R.A Population structure of Gopherus polyphemus in northern Florida. Journal of Herpetology 14: Aresco, M.J., and C. Guyer Efficacy of using scute annuli to determine growth histories and age of Gopherus polyphemus in southern Alabama. Copeia 1998: Aresco,M.J.,andC.Guyer.1999.Burrowabandonmentby gopher tortoises in slash pine plantations of the Conecuh National Forest. Journal of Wildlife Management 63: Ashton, K.G., G.M. Engelhardt, and B.S. Branciforte Gopher tortoise (Gopherus polyphemus) abundance and

12 2012] HERPETOLOGICAL MONOGRAPHS 133 distribution after prescribed fire reintroduction to Florida scrub and sandhill at Archbold Biological Station. Journal of Herpetology 42: Boglioli, M.D., C. Guyer, and W.K. Michener Mating opportunities of female Gopher Tortoises, Gopherus polyphemus, in relation to spatial isolation of females and their burrows. Copeia 2003: Breininger, D.R., P.A. Schmalzer, and C.R. Hinkle Estimating occupancy of gopher tortoise (Gopherus polyphemus) burrows in coastal scrub and slash pine flatwoods. Journal of Herpetology 25: Burke, R.L Burrow-to-tortoise conversion factors: Comparison of three gopher tortoise survey techniques. Herpetological Review 20: Clarke, P.J., and F.C. Evans Distance to nearest neighbor as a measure of spatial relationships in populations. Ecology 35: Cox, J., D. Inkley, and R. Kautz Ecology and habitat protection needs of gopher tortoise (Gopherus polyphemus) populations found on land slated for largescale development in Florida. Florida Game and Fresh Water Fish Commission, Nongame Wildlife Program Technical Report no. 4. Florida Game and Fresh Water Fish Commission, USA. Diemer, J.E The ecology and management of the gopher tortoise in the southeastern United States. Herpetologica 42: Diemer, J Home range and movements of the gopher tortoise (Gopherus polyphemus) in northern Florida. Journal of Herpetology 26: Doonan, T.J., and I.J. Stout Effects of gopher tortoise (Gopherus polyphemus) body size on burrow structure. American Midland Naturalist 131: Eubanks, J.O., W.K. Michener, and C. Guyer Patterns of movement and burrow use in a population of gopher tortoises (Gopherus polyphemus). Herpetologica 59: Guyer, C., and M.A. Bailey Amphibians and reptiles of longleaf pine communities. Pp in S.M. Hermann (Ed.), Proceedings of the Tall Timbers Fire Ecology Conference, The Longleaf Pine Ecosystem: Ecology, Restoration, and Management. Tall Timbers Research Station, Florida, USA. Guyer, C., and S.M. Hermann Patterns of longevity of gopher tortoise (Gopherus polyphemus) burrows: Implications for the longleaf pine-wiregrass ecosystem. Chelonian Conservation and Biology 2: Guyer, C., C.T. Meadows, S.C. Townsend, and L.G. Wilson A camera device for recording vertebrate activity. Herpetological Review 28: Hermann, S.M., C. Guyer, J.H. Waddle, and M.G. Nelms Sampling on private property to evaluate population status and effects of land use practices on the gopher tortoise, Gopherus polyphemus. Biological Conservation 108: Jackson, D.R., and E.G. Milstrey The fauna of gopher tortoise burrows. Pp in J.E. Diemer, D.R. Jackson, J.L. Landers, J.M. Layne, and D.A. Woods (Eds.), Proceedings of the Gopher Tortoise relocation symposium. Nongame Wildlife Program Technical Report 5. Florida Game and Fresh Water Fish Commission, USA. Jodice, P.G.R., D.R. Epperson, and G.H. Visser Daily energy expenditure in free-ranging gopher tortoises (Gopherus polyphemus). Copeia 2006: Johnson, V.M., C. Guyer, S.M. Hermann, J. Eubanks, and W.K. Michener Patterns of dispersion and burrow use support scramble competition polygyny in Gopherus polyphemus. Herpetologica 65: Kazor, S.A., and D.C. Harnett Gopher tortoise (Gopherus polyphemus) effects on soils and vegetation in a Florida sandhill community. American Midland Naturalist 123: Kie, J.G., J.A. Baldwin, and C.J. Evans CALHOME: Home range analysis program. Electronic user s manual. US Forest Service, California, USA. Landers, J.L., W.A. McRae, and J.A. Garner Growth and maturity of the gopher tortoise in southwestern Georgia. Bulletin of the Florida State Museum, Biological Sciences 27: McRae, W.A., J.L. Landers, and G.D. Cleveland. 1981a. Sexual dimorphism in the gopher tortoise (Gopherus polyphemus). Herpetologica 37: McRae, W.A., J.L. Landers, and J.A. Garner. 1981b. Movement patterns and home range of the gopher tortoise. American Midland Naturalist 106: McCoy, E.D., and H.R. Mushinsky Studying a species in decline: Gopher tortoises and the dilemma of correction factors. Herpetologica 48: MCCOY, E.D., and H.R. Mushinsky The demography of Gopherus polyphemus (Daudin) in relation to size of available habitat. Project report, Nongame Wildlife Program project GFC Florida Game and Fresh Water Fish Commission, USA. McCoy, E.D., and H.R. Mushinsky Estimates of minimum patch size depend upon the method of estimation and the condition of the habitat. Ecology 88: McCoy, E.D., H.R. Mushinsky, and J. Lindzey Declines of the gopher tortoise on protected lands. Biological Conservation 128: Means, D.B Vertebrate faunal diversity of longleaf pine ecosystems. Pp in J. Shibu, E.J. Jokela, and D.L. Miller (Eds.), The Longleaf Pine Ecosystem: Ecology, Silviculture, and Restoration. Springer Science, New York, USA. Moon, J.C., E.D. McCoy, H.R. Mushinsky, and S.A. Karl Multiple paternity and breeding systems in the gopher tortoise, Gopherus polyphemus. Journal of Heredity 97: Mushinsky, H.R., and E.D. McCoy Comparison of gopher tortoise populations on islands and on the mainland in Florida. Pp in R.B. Bury and D.J. Germano (Eds.), Biology of North American Tortoises. Fish and Wildlife Research Report 13. U.S. Department of the Interior, National Biological Survey, Washington, DC, USA. Noss, R.F., E.T. LaRoe, and J.M. Scott Endangered ecosystems of the United States: A preliminary assessment of loss and degradation. National Biological Service Biological Report 28. US Department of the Interior, Washington, DC, USA. Ott, J.A., J.W. Hollister, C. Guyer, and W.K. Michener Area requirements of gopher tortoises (Gopherus polyphemus): An evaluation of guidelines for estimating reserve size. Chelonian Conservation and Biology 4: Outcalt, K.W Occurrence of fire in longleaf pine stands in the southeastern United States. Pp in W.K. Moser, and C.E. Moser (Eds.), Proceedings of the

13 134 HERPETOLOGICAL MONOGRAPHS [No. 26 Tall Timbers Fire Ecology Conference, Fire and Forest Ecology: Innovative Silviculture and Vegetation Management. Tall Timbers Research Station, Florida, USA. Peterson, C.C Ecological energetics of the Desert Tortoise (Gopherus agassizii): Effects of rainfall and drought. Ecology 77: Platt, W.J., G.W. Evans, and S.L. Rathbun The population dynamics of a long-lived conifer (Pinus palustris). American Naturalist 131: SAS Institute Statistical Analysis System. SAS Institute, North Carolina, USA. Tuberville, T.D., T.M. Norton, B.J. Waffa, C. Hagen, and T.C. Glenn Mating system in a gopher tortoise population established through multiple translocations: Apparent advantage of prior residence. Biological Conservation 144: Smith, K.R., J.A. Hurley, and R.A. Seigel Nesting ecology, female home range and activity patterns, and hatchling survivorship in the gopher tortoise (Gopherus polyphemus). Chelonian Conservation and Biology 2: Wilson, D.S., H.R. Mushinsky, and E.D. McCoy Relationship between gopher tortoise body size and burrow width. Herpetological Review 22: Accepted: 1 February 2012.Associate Editor: Michael Freake