Home Range Behavior among Box Turtles (Terrapene c. carolina) of a Bottomland Forest in Maryland

Eastern box turtles (Terrapene c. carolina) in a Maryland bottomland forest were studied over a period of years (1944-1981). Home ranges of 51 males averaged 146 ? SD 48 m long and 105 + SD 38 m wide; ranges of 52 females averaged 144 + SD 52 m long and 100 + SD 38 m wide. An approximation of average home range size, based on an ellipse, is 1.20 ha for males and 1.13 ha for females. Sizes of home ranges of individuals did not differ significantly between 1945 and the full term of their captures (x = 14 yr) (AOV; P > 0.05). Mean distance between capture sites, which provides an index to range size, was not significantly different among the years of 1945, 1955, 1965, and 1975 (AOV; P > 0.05). Geographic centers of ranges of 77 males in the bottomlands showed no significant (AOV; P > 0.05) change for 46, and change over relatively short distances (x = 57 ? SD 23 m) for the others. Among 70 females, there was no significant change for 46 and change over short distances (x = 61 + SD 24 m) for the others. Changes in location were more frequent between 1965 and 1975, a period of pronounced population decline, than between previous decades (significant only for females, x2; P < 0.025). Hibernation sites ordinarily (21 of 23 individuals) were within the normal bottomland range; hibernation sites of different years were near each other (all of 4 individuals). In contrast, nesting sites were far distant, extending the home range by 400-700 m, but those of different years were near each other (6 individuals). Mating partners occupied broadly overlapping or contiguous ranges (35 records). Interactions between males (18 records) were identical to courtship behavior, and are believed not to represent territorial aggression. The behavior of animals in their natural environment ordinarily involves utilization of limited areas, defined as home ranges when not actively defended (Burt, 1940, 1943). The concept of home range was shown to be applicable 1 Present Address: 174 Shepherd Creek Road, Franklin, North Carolina 28734, USA. to eastern box turtles, T. c. carolina (Stickel, 1950) and confirmed in subsequent ecological studies (Williams, 1961; Dolbeer, 1969; Yahner, 1974; Madden, 1975) as well as in studies of the threetoed box turtle, T. c. triunguis (Schwartz and Schwartz, 1974) and the ornate box turtle, T. o. ornata (Legler, 1960). Studies over relatively short periods, however, could not answer the questions of whether 40 This content downloaded from 157.55.39.242 on Thu, 21 Apr 2016 07:37:28 UTC All use subject to http://about.jstor.org/terms BOX TURTLE HOME RANGE individuals retained the same home ranges through the years or whether home-range size was related to population density. Neither was it feasible to evaluate the home-range relationships of mating, nesting, hibernation, and male/ male interactions. Results of a long-term study (1944-1981) that are applicable to these questions are summarized here. MATERIALS AND METHODS The primary study area of 11.8 ha was in the bottomland forest of the Patuxent Wildlife Research Center, in the midst of similar habitat along the Patuxent River (Hotchkiss and Stewart, 1947; Stickel, 1950). Farther from the river, the land joins lower and wetter swamp forest, which in turn abuts low bluffs at the edges of terrace forest, leading to adjoining uplands. The uplands include wooded areas as well as the 24 ha area of buildings and lawns that is the headquarters of the Center. Turtles were located by intensive systematic search, both of the primary study area and of the surrounding areas, in 1945, 1955, 1965, and 1975. Effort was more variable in other years, but turtle captures numbered 100-600 in each of these years between 1944 and 1958 (400-600 in 5 of the years) and 100 in 1974. Turtles were marked for individual recognition by filing notches in the marginal scutes according to the code system used by Cagle (1939). Measurements of carapace dimensions and growth annuli, descriptions of other physical characteristics, and data concerning mating, nesting, hibernation, and aggression were recorded in the field at the time of capture (Stickel, 1950, 1978). Locations were recorded with reference to permanent bronze-cement markers at 330 ft (100.6 m) intervals which were subdivided further by temporary markers at 110 ft (33.5 m) intervals. Stability of home ranges was evaluated by comparing locations of the geographic centers of the ranges in different years. Centers were taken as the mean position of capture sites in east-west and north-south directions. Comparisons among years were made by analysis of variance and Duncan's multiple-range test, using a significance level of P < 0.05. Sizes of home ranges in the bottomland forest were measured by an adaptation of the bivariate normal method of Jennrich and Turner (1969) and Mazurkiewicz (1969), which provides a density-probability estimate. Adaptation was necessary because bivariate-normal analysis requires significant correlation between points on the x and y axes, which occurred (at the P < 0.05 level) in only 26% of the cases in the present study. Normality of the distributions along the x and y axes was tested by Snedecor's method for small samples, using a significance level of P < 0.01 (Snedecor, 1956:199). When the locations along the x and y axes were not correlated, the dimensions of the range were estimated in the east-west and north-south directions, using two standard deviations on either side of the mean, thus encompassing 95% of the expected points. When the locations on the two axes were correlated, the dimensions were estimated along the regression line (using the longest dimension as the x axis) and along a line perpendicular to it (y axis). These procedures provided estimates of the length and width of the turtle's range. Average range dimensions were calculated for adult individuals 20 or more years old and captured in 2 or more years, during which the location of the range had not changed significantly. To avoid the bias of underestimating range size of turtles whose ranges were largely outside the primary study area, records for those whose geographic centers were 30 m or less from the border were excluded. This distance was based on comparisons that showed significantly (AOV; P < 0.05) lower figures for this group than for those with centers at greater distances from the borders. Individual records more than four standard deviations from the mean of the others in a set were eliminated as outliers; most of these (12 of 14) were for females and appeared to represent egg-laying excursions, although the less intensive collecting outside the study area prevented full evaluation. Mean capture distances, which provide an index to range size, were used to determine whether average home range size in the population changed with time. Distances between capture sites (in random order) were measured and the means were calculated for each individual captured five or more times in a year (1945, 1955, 1965, 1975). Means for each year were calculated from these data. Randomization of capture sequence helps reduce the bias of sequential captures, but the method otherwise is that of Fitch (1958).