The tortoise genus Gopherus consists of 4 species widely distributed across North America. One species, the desert tortoise (Gopherus agassizii), inhabits the Sonoran and Mojave deserts in 4 southwestern US states and the more mesic regions of Sinaloa, Mexico. Another species, the gopher tortoise (Gopherus polyphemus), inhabits the coastal plain in 6 southeastern US states. These 2 species belong to lineages that probably separated about 17–19 million years before present (mybp) (McCord 2002). Further separation of the desert tortoise into Sonoran Desert and Mojave Desert forms probably occurred about 5–6 mybp (Lamb et al. 1989; McCord 2002). A variety of behavioral differences characterize the desert tortoise forms, mostly reflecting responses to the variation in seasonal rainfall between and across the 2 deserts (Averill-Murray 2002; Morafka and Berry 2002; Van Devender 2002). In its xeric desert environment (average annual precipitation, 34–310 mm; USGS 2005), the Mojave Desert tortoise must endure water shortage and high dietary potassium, which together present considerable difficulty for an herbivore without salt glands (Dantzler and Schmidt-Nielsen 1966; Minnich 1977; Nagy and Medica 1986; Oftedal et al. 1994, 2002; Oftedal 2002, 2003).

Because it does not face the same environmental constraints as the desert tortoise, the gopher tortoise displays a different suite of behavioral traits associated with resource acquisition than either of the desert tortoise forms (Minnich and Ziegler 1977; Mushinsky et al. 2006). In its more mesic environment (average annual precipitation, 1140–1650 mm; Minnich and Ziegler 1977) and because of its burrowing ability, the gopher tortoise is able to obtain water readily and is not overburdened with potassium, although severe dehydration can alter this condition (Minnich and Ziegler 1977; Ross 1977). The relative humidity of the ambient air and the atmosphere within the burrow, and the water content of food plants, are typically high (Minnich and Ziegler 1977; Mushinsky et al. 2006).

Body condition indices (e.g., mass per unit volume) have suggested seasonal patterns of resource accumulation by the desert tortoise in the Mojave Desert (Nagy et al. 2002). Parallel studies of the gopher tortoise have not been performed, but a comparison of the patterns exhibited by the 2 related species inhabiting disparate hydrological conditions can provide insight into how the 2 species are affected by and respond to the different environments (Minnich and Ziegler 1977; Germano 1994). To this end, we compared the seasonal body condition indices of the desert tortoise in the Mojave Desert with the seasonal body condition indices of the gopher tortoise in central Florida. In Florida, the gopher tortoise inhabits upland habitats with deep, excessively well-drained sands (Diemer 1986; Mushinsky et al. 2006). The state is located within the humid subtropics (Winsberg 2003) and is among the wettest regions in the United States. Although droughts are possible during the dry season (typically, November–April), much of Florida typically escapes the serious and long-lasting droughts found in other parts of the southeastern United States (Chen and Gerber 1990).

METHODS

Data on body mass and volume for the gopher tortoise were collected over 3 periods, from 2 populations in central Florida. Data were collected during 1983–1985 (hereafter period 1) and 1988–1994 (period 2) at the Ecological Research Area (ERA) of the University of South Florida (lat 28°05′N, long 82°00′W) using the same population studied by Macdonald and Mushinsky (1988), Mushinsky et al. (1994, 2003), and Wilson et al. (1994); and during 2001–2004 (period 3) at Brooker Creek Preserve, northeast Pinellas County (lat 28°08′N, long 82°39′W) using the same population studied by Riedl et al. (2008). Field measurements were recorded in most months each year. Individuals were captured by hand and by using bucket traps situated at the mouths of active burrows; they were weighed and measured for midline carapace length, maximum shell height, and maximum shell width (e.g., Loehr et al. 2004, 2007). Nagy et al. (2002) used a different method for measuring shell dimensions of the desert tortoise. We compared the 2 methods for the gopher tortoise by measuring a sample of shells, and found that they yielded nearly identical calculations of volume, computed by simple multiplication (Nagy et al. 2002). We did not weigh individuals before they were able to defecate and urinate, as did Nagy et al. (2002), so to make the mass data from the gopher tortoise data comparable to those from the desert tortoise, we increased the estimated upper limits of gopher tortoise body condition by 3.4%, the mean difference between mass before and after excretion. The relationship between mass of excrement and initial body mass is strong (r²  =  0.756; F_1,12  =  37.096, p < 0.001, n  =  14) in the gopher tortoise. Only adults (individuals with carapace lengths > 240 mm; Landers et al. 1982; Mushinsky et al. 1994) were included in our analyses. For each period, we selected one record per individual at random from the multiple captures, to avoid pseudoreplication. We obtained independent data on body condition for 184 adults, including 22 males and 26 females in period 1, 48 males and 52 females in period 2, and 17 males and 19 females in period 3. Data on rainfall for the 3 periods were obtained from the Temple Terrace reporting station (National Climatic Data Center 2009), the nearest station to the ERA.

Data for 3 populations of the desert tortoise in the Mojave Desert of California collected during the relatively wet years of 1992–1993 were kindly supplied by Kenneth Nagy. Again, only adults (as identified in Nagy et al. 2002) were included in our analyses. Whereas Nagy et al. (2002) calculated the “prime body condition” for the desert tortoise based on the highest recorded ratio of each recaptured individual, we selected one record per individual at random from the multiple captures, as we did for the gopher tortoise. We obtained independent data on body condition for 163 adults, including 86 males and 77 females.

Body condition can be assessed indirectly with body condition indices derived from the relationship between body mass and size. Several indices have been developed and employed on a broad range of taxa, including birds (Tella et al. 1995; Blanco and Tella 1997), mammals (Schulte-Hostedde et al. 2001, 2005; Arnould and Warneke 2002; Blackwell 2002), fish (Neumann and Willis 1994), reptiles (Jacobson et al. 1993; Hailey 2000; Willemsen and Hailey 2002; Salvidio and Delaugerre 2003), and invertebrates (Streissl and Hödl 2002; Arcos et al. 2003). Although body condition indices have been used as measures of relative energy or fat content, they can be influenced by variation in all constituents of body composition (Krebs and Singleton 1993; Virgl and Messier 1993; Schulte-Hostedde et al. 2001). In turtles, body size as measured by shell dimensions tends to be relatively constant in the short term (Brisbin 1972; Nagy et al. 2002), and variation in body mass largely reflects tissue hydration, water in the bladder, and food in the gut (Ross 1977; Peterson 1996a; Nagy et al. 2002; Willemsen and Hailey 2002). Some of the variation also reflects egg production by females. Loss of mass during egg laying may be a relatively small component of the overall variation, and the mass that is lost may be recovered quickly (Luckenbach 1982; Turner et al. 1984; Wallis et al. 1999; Moon et al. 2006; Mushinsky et al. 2006). In the present case, loss of mass during egg laying does not appear to strongly influence the patterns that we found. For example, potential loss of mass by desert tortoise females following production of a first clutch of eggs in late April–early May (Wallis et al. 1999) was not reflected in the body condition values observed at that time (see Results).

Previous research used the ratio (mass per unit volume) method to derive an index of body condition (Nagy et al. 2002; also see Bjorndal et al. 2000). The ratio method has statistical shortcomings, however (Atchley et al. 1976; Atchley 1978; Packard and Boardman 1988), and the more statistically sound residuals method has been advanced to account for the confounding effects of body size (Krebs and Singleton 1993; Schulte-Hostedde et al. 2001, 2005; Arnoud and Warneke 2002; Willemsen and Hailey 2002). The residuals method in turn has been criticized on statistical grounds (Garcia-Berthou 2001; Green 2001; Freckleton 2002; McCoy et al. 2006), particularly when the residuals are used as data in subsequent analyses. However, it is likely that the overall conclusions derived using the residual method are the same as those derived from other methods for most real data sets if certain key underlying assumptions are met (Green 2001; McCoy et al. 2006).

We assessed body condition with the ratio method and with 4 versions of the residuals method: 1) ordinary least squares regression performed on body mass against body volume (OLS1); 2) ordinary least squares regression performed on body mass against principal component one (PC1) of a principal components analysis of carapace length, width, and height (OLS2); 3) model II (reduced major axis [RMA]) regression performed on body mass against body volume (RMA1); and 4) model II RMA regressions performed on body mass against PC1 (RMA2) (Green 2001). Prior to the regression analyses, we tested the assumptions that mass increases linearly with the measure of body size (Green 2001) and that the first principal component is shared among data sets (McCoy et al. 2006). PC1 accounted for 88.9%–92.0% of variance in body measurements for the gopher tortoise and 94.7% of variance for the desert tortoise. Relationships between mass and PC1 departed significantly from linearity (p ≤ 0.01) for the gopher tortoise in period 2 and for the desert tortoise. Regressions were performed only on linear data.

Four datasets were used in the species comparison: data from 3 time periods for the gopher tortoise and from 1 time period for the desert tortoise. Duplicate analyses were run for body condition indices calculated by the ratio method (body mass vs. volume, as described by Nagy et al. 2002) and by the best version of the residuals method (see Results). Within each data set, body condition was compared among years using using analysis of variance (ANOVA) or a nonparametric equivalent, and between sexes, using a t-test or nonparametric equivalent, to determine whether the data could be pooled. Differences among seasons (winter  =  December–February, spring  =  March–May, summer  =  June–August, autumn  =  September–November) were then tested and using ANOVA or a nonparametric equivalent.

RESULTS

The 5 body condition indices produced results that were highly significantly correlated when compared within species and time period (Spearman's rank correlation, p < 0.01). Relationships between body mass and body condition are displayed in Table 1.

Table 1 Results (p-value) of regressions of body condition index, derived by 5 methods (OLS  =  ordinary least squares regression, RMA  =  reduced major axis regression), against body mass.^a

View Fullscreen

For the gopher tortoise, significant (p < 0.05) positive effects of body mass on body condition were detected for females in period 1 with the index derived from the ratio method, and for females in period 2 with both the index derived from the ratio method and residuals method OLS1. No other relationships were significant at p < 0.05, although the relationship could be considered borderline for females in period 1 with the residuals method OLS2.

For the desert tortoise, a significant (p < 0.05) positive relationship between body mass and body condition was found in males using the ratio method and regression OLS1. Residuals from the RMA regression produced a body condition index that exhibited seasonal variation comparable to that calculated using a ratio of body mass to volume (Fig. 1) but was the only measure of body condition that unambiguously displayed no significant size effects in either sex or either species.

View Figure

• Download as Powerpoint
• Download Figure

The species comparisons using the ratio method and the residuals (from RMA regression) method yielded virtually identical results, and we report only the statistics from the ratio method (see Nagy et al. 2002). For the gopher tortoise, body condition varied among periods, being significantly higher in period 1 than in period 2 (t  =  4436.5, p < 0.01) or 3 (t  =  2.672, p < 0.01). No significant difference was detected in body condition among years within period 1 (F_1,2  =  0.048, p  =  0.953), period 2 (H  =  10.713, p  =  0.098), or period 3 (F_1,3  =  0.980, p  =  0.415). No significant difference was detected in body condition between sexes in period 1 (t  =  1.028, p  =  0.310), period 2 (t  =  2351.0, p  =  0.617) or period 3 (t  =  0.417, p  =  0.680) (Table 2). Data were pooled between years and sexes for further analyses. For the desert tortoise, no significant difference was detected in body condition between years (t  =  −1.348, p  =  0.178) or between the sexes (t  =  0.138, p  =  0.891) (Table 2). Data were pooled between years and sexes for further analyses.

Table 2 Mean body condition (ratio method) ± SD between sexes.

View Fullscreen

Body condition of the gopher tortoise did not differ significantly among seasons in period 1 (H  =  3.646, p  =  0.302), period 2 (H  =  2.513, p  =  0.285), or period 3 (H  =  4.087, p  =  0.130) (Fig. 2). Body condition of the desert tortoise, however, increased from March to May and then steadily declined through the remainder of the active season, resulting in a significant difference among seasons (H  =  19.282, p < 0.001) (Fig. 2). Body condition of the gopher tortoise reached a maximum slightly in excess of 0.60 in winter, and body condition of the desert tortoise reached a maximum at 0.63 in spring (Fig. 2). Mean body condition was significantly higher in the desert tortoise than in the gopher tortoise for periods 1, 2, and 3 (for all, t ≥ 2025.0, p ≤ 0.002). Body volume of the gopher tortoise tends to be somewhat greater than that of the desert tortoise, for the same body length (unpubl. data; see Woodbury and Hardy 1948, McRae et al. 1981a).

View Figure

• Download as Powerpoint
• Download Figure

Because body condition of the gopher tortoise did not vary among seasons, no relationship to rainfall was detectable within periods. This result contrasts markedly with the strong correlation between body condition and rainfall displayed by the desert tortoise over the short term (Wallis et al. 1999; Nagy et al. 2002). On the other hand, body condition of the gopher tortoise did vary over the long term, among periods. Because the data for periods 1 and 2 came from the same location, the lower body condition observed in period 2 may be related to the prolonged episode (7 years) of generally low rainfall (Fig. 3).

View Figure

• Download as Powerpoint
• Download Figure

DISCUSSION

The difference in seasonal body condition between the gopher tortoise and the desert tortoise and in the relationship of body condition to rainfall are consistent with what is known about the physiology and behavior of the 2 species. The desert tortoise possesses a variety of traits that allow it to survive in an arid environment with very occasional rainfall. When it does not rain in the active season, individuals tend to remain in their burrows and do not eat or void urine or feces; when it subsequently rains, individuals tend to become active, drink rainwater, void urinary wastes (including stored potassium), store fresh water in their bladders (as much as 30% of body mass), and resume feeding (Minnich 1976, 1977, 1979; Minnich and Zielger 1976; Ross 1977; Shoemaker and Nagy 1977; Medica et al. 1980; Nagy and Medica 1986; Peterson 1996b; Henen et al. 1998; Jørgensen 1998; Wilson et al. 2001; Henen 2002; Longshore et al. 2003). Thus, the elements that are principal contributors to short-term mass gain—tissue hydration, bladder water, and food in the gut—are tied closely to rainfall.

Water relationships of the gopher tortoise are less well-studied than those of the desert tortoise. Individual gopher tortoises generally are more active (i.e., out of their burrows) and feed more regularly throughout the year than individual desert tortoises. Gopher tortoises often forage during the hottest part of the day, become less active during rainfall, tend not to drink rainwater, and tend to store relatively little water in the bladder (Minnich and Ziegler 1977; Ross 1977; Douglas and Layne 1978; Jodice et al. 2006). Thus, they maintain a relatively constant water balance, receiving water via frequent foraging on plants with high water content (Minnich and Ziegler 1977). Although preferred foods of the gopher tortoise may be scarce during mid- to late summer (Garner and Landers 1981; McRae et al. 1981b; Mushinsky et al. 2003), any scarcity was not reflected in our body condition results. Individuals had unexpectedly high body condition during the low-activity cooler months, between September and March. Because we were not actively trapping during those months, any measurements came from individuals that were out of their burrows foraging during warm spells, so our body condition results may not have been typical. The forage available to individuals in the cooler months comprises plants with relatively low nutritional value, mainly grasses (McRae et al. 1981b; Mushinsky et al. 2003). Because of the thermal stability of their deep burrows (Ultsch and Anderson 1986), individuals may retain large quantities of these low-quality plants in their guts.

Researchers have become increasingly concerned about how changing rainfall patterns, particularly as a result of regional climate change, may affect the well-being of the desert tortoise (e.g., Oftedal 2002). They correctly are concerned that disruption of normal rainfall patterns, such as changes in the frequency of El Niño–Southern Oscillation events (Henen et al. 1998; Oftedal et al. 2002), may have serious adverse consequences for the desert tortoise. Plants that have high water and protein to potassium ratios (high PEP plants), which may be critical for the desert tortoise and preferentially selected (Oftedal 2002; but see Tracy et al. 2006) occur more frequently in high-rainfall years (Oftedal 2002). Juveniles may find it difficult to obtain sufficient nitrogen for growth when these plants are not readily available (Oftedal et al. 2002).

Although similar concerns for the well-being of the gopher tortoise under changing climatic conditions have not been expressed, we suggest that it may be at least as poorly equipped to deal with changes in rainfall as the desert tortoise. West-central Florida already may have generally reduced summer rainfall as a consequence of land use changes (Marshall et al. 2004), and rainfall in the region is predicted to decline substantially over the next 50 years, especially during the potentially crucial winter–spring cooler dry season (The Nature Conservancy 2009). The lack of what appear to be desert adaptations (but see Bradshaw 1997; Morafka and Berry 2002) in the gopher tortoise naturally leads to the assumption that water limitation has not been a particular problem, which, in turn, leads to complacency about the possibility of future water limitation. The virtually complete lack of data on water balance under natural drought conditions is problematic, however.

Data on the consequences of water restriction suggest an interesting convergence of gopher tortoise responses with those of the desert tortoise. Under water restriction, the gopher tortoise may drink rainwater (Ashton and Ashton 1991) and may retain increasing amounts of water in the bladder and store urates there (Ross 1977). Plasma potassium concentrations increase and the excess potassium also may be stored in the bladder (Ross 1977). Studies of water influx rates using doubly labeled water indicate that the gopher tortoise and desert tortoise can achieve similarly high rates: means of 1.8–3.1 ml · 100 g⁻¹ · day⁻¹ (Minnich and Ziegler 1977; Jodice et al. 2006) and 1.8–2.5 ml · 100 g⁻¹ · day⁻¹ (Henen et al. 1998), respectively. Water influx rates tend to be substantially lower in the desert tortoise when averaged over time (Henen et al. 1998), however, and extremely lower during seasonal or longer droughts (Minnich 1977; Peterson 1996a, 1996b; Henen et al. 1998). Similar data on water influx rates under drought conditions are not available for the gopher tortoise, but loss of mass during dehydration appears to be more rapid for the gopher tortoise (1.07 ml · 100 g⁻¹ · day⁻¹; Ross 1977) than the desert tortoise (0.28 ml · 100 g⁻¹ · day⁻¹; Minnich 1977). This difference may reflect a substantially higher rate of evaporative water loss in the gopher tortoise, meaning that it may need to adopt water conservation measures at a relatively high water influx rate (Minnich and Ziegler 1977; Ross 1977; Johnston 1996). Clearly, water budgets of individuals of all ages should be assessed in the field under a range of rainfall conditions.