Seasonal Inactivity of the Desert Box Turtle, Terrapene ornata luteola, at the Species’ Southwestern Range Limit in Arizona

—At its southwestern range limit in southeastern Arizona, the Desert Box Turtle Terrapene ornata luteola, takes refuge in subterranean mammal burrows, primarily those of the Kangaroo Rat Dipodomys spectabilis, from November through June. I inferred temporal and physiological characteristics of turtles in burrows from temperatures collected with miniature data loggers buried in the soil and attached to turtles from August 2001 through June 2003. In the fall, entrance into hibernation varied among individual turtles and between years from 18 October to 6 December. Mean daily body temperatures at the beginning of hibernation were about 14–188C. From December through February, body temperatures averaged about 98C and were slightly less than and highly correlated with the 30-cm soil temperature. By April, body temperatures of turtles approached levels exhibited by surface-active turtles (minimum approximately 188C for moving turtles) but turtles did not emerge from burrows until the summer monsoons began about three months later. Thus, the five-month hibernation period of T. o. luteola was contiguous with a three-month estivation period for a total annual period of subterranean refuge of about eight months. Except for extending into warm months, the annual inactive period of T. o. luteola in southeastern Arizona was not remarkably different from that in other parts of the species’ range. In arid southeastern Arizona, populations of Terrapene ornata, primarily a prairie-adapted form (Legler, 1960; Dodd, 2001), are at the species’ southwestern range limit and may be active for only a few weeks out of the year. In seasonally cold or dry environments, reptiles commonly spend a substantial portion of the year awaiting favorable weather in hibernation or estivation (Gregory, 1982). Furthermore, in areas such as southeastern Arizona, contiguous periods of cold and hot dry seasons may produce an extended period of weather conditions unfavorable for surface activity. With low annual reproductive output and the lowest field metabolic rates reported for any reptile, Terrapene have the potential to be highly successful in unpredictable resource environments (Penick et al., 2002). Legler (1960) hypothesized that extreme southwestern populations of T. ornata, formally recognized as Terrapene ornata luteola, were likely to be physiologically adapted for living in rigorous arid environments. However, activity and thermal biology of T. o. luteola during the activity season in southeastern Arizona did not differ substantially from that in more central prairie portions of the species’ range (Plummer, 2003). In this paper, I show that certain temporal and thermal aspects of hibernation of T. o. luteola in southeastern Arizona are also not exceptional; however, hibernation is followed by an estivation period that results in a major portion of the annual cycle being spent inactive in underground refugia. MATERIALS AND METHODS The study area was located in semiarid grassland habitat in the Sulphur Springs Valley, 44 km southeast of Willcox, Cochise County, Arizona, at 1400 m elevation. The Valley is a southwestern extension of plains grassland that is heavily influenced by both Sonoran and Chihuahuan Desert flora (Rosen et al., 1996). The study area is active ranchland periodically grazed by cattle. Most of the annual surface activity of T. ornata occurs during a distinct 8–12-week summer monsoon period normally beginning in early July. When not active on the surface, turtles usually take refuge in various subterranean retreats (Plummer, 2003). Burrows of the bannertail kangaroo rat (Dipodomys spectabilis), a keystone species in southwestern desert grasslands (Brown and Heske, 1990; Hawkins and Nicoletto, 1992), provided the primary subterranean retreat for T. ornata on the study area throughout the year. Burrows of other mammals (e.g., Taxidea, Sylvilagus, Thomomys, Spermophilis) also provided underground retreats. Turtles were collected as they came to drink at a 0.2-ha cattle tank during their early morning and late afternoon summer activity periods (Plummer, 2003). Captured turtles were measured (carapace length in millimeters, mass in grams), sexed, and given a unique mark by filing notches in the marginal scutes. I attached a radiotransmitter (AVM Model SM1; 16350 mm, 16 g; size N alkaline cell) to the posterior edge of the carapace with a short piece of stainless steel wire so that it could freely pivot about the attachment point. I also firmly attached a miniature iButton temperature data logger (Dallas Semiconductor Corp.; 6 3 16.5 mm, 3 g) to the left posterio-lateral portion of the carapace with stainless-steel wire such that the sensor measured the immediate surface temperature of the carapace (Tb). The combined mass of the transmitter and data logger was approximately 20 g, , 5% of turtle body mass. Once in the ground, Tbs of hibernating reptiles should track ground temperatures (Gregory, 1982); therefore, iButton temperatures should be representative of core body temperatures. Turtles were monitored from July 2001 to August 2003. Data were downloaded from the loggers and transmitters refurbished on 21–25 July in 2002 and 1–3 August in 2003. To obtain soil temperatures, I buried iButton loggers at depths of 1 cm, 30 cm, and 60 cm on level ground about 100 m from the tank near known D. spectabilis burrow hibernacula of several turtles. The summer home range of each turtle encompassed the tank (unpubl. data). All loggers recorded temperatures every four hours beginning at 0000 h. Based on known Tbs of turtles in burrows during their summer activity period (minimum 208C, range , 28C; Plummer, 2003), I defined entrance into hibernation for an individual turtle as a consecutive 5-day period in the fall in which the daily Tb mean was , 208C and range was , 28C. Because this procedure might not detect turtles just beginning their hibernation in shallow burrow mouths, it should produce conservative (i.e., err toward a later date) estimates of the date of entrance into hibernation. Data were analyzed with SYSTAT 10.2 (SYSTAT Software, Inc., Richmond, CA, 2002). Correlates of Tb were examined with Pearson correlation analysis with Bonferroni probabilities. Body temperatures of individual turtles between years were compared with a Wilcoxon signed rank test. Unless otherwise stated, means are given 6 1 SE.