An emergent infectious disease is one whose incidence has increased within the past 2 decades or threatens to increase in the near future (Lederberg et al. 1992). Emergence may be the result of the spread of a new agent, of the recognition of an infection that has been present but has gone undetected, of the realization that an established disease has an infectious origin, or of the reappearance of a known infection after a decline in incidence (Lederberg et al. 1992). These standards apply equally well to diseases of wildlife as to diseases of humans and domesticated animals (Daszak et al. 2000; Williams et al. 2002).

Upper respiratory tract disease (URTD) is an emergent disease of tortoises that has moved rapidly from obscurity to major concern. The principal causal agent of the disease is a mycoplasma, Mycoplasma agassizii (sensu lato) (Schumacher et al. 1993; Brown et al. 1994; McLaughlin 1997; see Baseman and Tully 1997). The mycoplasma was first recorded in the United States from the desert tortoise (Gopherus agassizii), a species that is confined to the Southwest (Jacobson et al. 1991, 1995), and, shortly thereafter, it was recorded from the gopher tortoise (Gopherus polyphemus), a species that is confined to the southeast (Beyer 1993). The mycoplasma is easily transmitted by direct contact between individuals (McLaughlin 1997; Brown et al. 2002), and URTD seemingly can result in die-offs of individuals (e.g., Jacobson et al. 1991, 1995; Berry 1997; Seigel et al., 2003, and included references).

Many questions about M. agassizii, its relationship to URTD, and the effects of URTD on the gopher tortoise still remain (Brown et al. 2002). The consequences for individuals of previous exposure to the mycoplasma (i.e., seropositive according to an enzyme-linked immunosorbent assay (ELISA) test for antibodies), or even of an active mycoplasmal infection, have not been determined. The connections between percentages of individuals in a population that are seropositive or that display clinical signs of URTD and consequent demographic change also have not been determined. Comparing demographic data before and after emergence of URTD could help to supply some of the missing information, but such data are rare for most diseases. We gathered demographic data on more than 60 gopher tortoise populations throughout Florida before the presence of URTD was widely recognized (McCoy and Mushinsky 1988, 1991). Subsequent screenings for antibodies revealed the presence of seropositive individuals in several of the populations that we had previously surveyed (Diemer Berish et al. 2000). Because we had demographic data from 8 to 10 years before the emergence of URTD, the opportunity was available to resurvey some of these populations with seropositive individuals to document changes in demographic profiles that may have resulted from the presence of M. agassizii.

METHODS

Study Sites

Surveys of gopher tortoise populations residing on a large number of protected sites in Florida were conducted in the late 1980s to early 1990s (McCoy and Mushinsky, 1988, 1991). Surveys were duplicated at 10 sites in northern peninsular Florida in 2000–2001 (Table 1). Site locations are presented in McCoy et al. (2006). Before our resurveys, screenings for antibodies at 4 of the sites had yielded seropositive individuals (Table 1). The study design, therefore, was to compare the results of demographic surveys before and after exposure to M. agassizii where the mycoplasma was known to have been present (4 sites) with contemporaneous demographic surveys and resurveys taken where the mycoplasma was not known to have been present (6 sites).

Table 1. Study sites (with acronyms).

View Fullscreen

Antibody Screening

To confirm the findings of previous screenings for antibodies, we hand captured individuals to extract blood samples during our resurveys and at other times, as needed. We captured individuals over as broad an area as possible at each site to attempt to evaluate the local distribution of M. agassizii. Because the blood samples also were to be used to measure corticosterone levels (McCoy et al, in prep.), bleeding was accomplished within 2–10 minutes of capture.

Captured individuals were placed in a turtle-restraining device, modeled after the one used successfully by J. Berish (Florida Fish and Wildlife Conservation Commission, pers. comm. 1998). By using aseptic techniques, a butterfly catheter or a syringe coated internally with 0.06 g heparin/mL saline solution was used to draw approximately 1 mL blood from the brachial vein of a foreleg. The blood sample was transferred to a 1.5-mL Eppendorf tube, labeled, and stored on wet ice. All individuals were released back to original capture locations within hours of capture. The turtle-restraining device, calipers, and field workers' hands were thoroughly washed with a 10% bleach solution.

In the laboratory, blood was centrifuged, and hematocrit levels, as well as cell and plasma levels, were measured. A 220-μL aliquot of plasma was sent to the BEECS laboratory, University of Florida, Gainesville, Florida, for ELISA testing for antibodies to M. agassizii. There, each sample was characterized as seronegative, suspect, or seropositive (see Schumacher et al. 1993). Remaining plasma was stored at −80°C for subsequent analysis of corticosterone levels (McCoy, et al. in prep.).

Sample Size for Antibody Screening

We found seropositive individuals at sites where previous screenings (Diemer Berish et al. 2000) had not found them (see Results). Some of the previous screenings were of small numbers of individuals (see Davidson et al. 1994), which suggests that detection of seropositive individuals may depend a great deal on sample size and that blood should be obtained from as many individuals in a population as possible. We examined the possibility that detection of seropositive individuals is a function of sample size by plotting the percentage of individuals determined to be seropositive and serologically suspect against the number of individuals tested for a large number of examples. These examples came from both published (Beyer 1993; Epperson 1997; Lederle et al. 1997; Schumacher et al. 1997; Smith et al. 1998; Diemer Berish et al. 2000) and unpublished (J. Diemer-Berish and L. Wendland, C. Heise and D. Epperson, E. McCoy and H. Mushinsky, unpubl. data, 2002) studies. To determine what sample sizes might be adequate to detect at least 1 seropositive individual in a population, we calculated the posterior probabilities of detection of seropositive individuals at the sites for which we had relatively large sample sizes (Boyd Hill Nature Reserve [BH], Wekiwa Springs State Park [WS], Lake Louisa State Park [LL], University of South Florida Ecological Research Area [RA]), based on the distribution of seropositive and seronegative individuals in the full sample.

Demographic Surveys

Surveys and resurveys used indirect assessment of population status via the conspicuous burrows that the gopher tortoise digs (McCoy and Mushinsky 1991, 1992; Mushinsky and McCoy 1994; Wilson et al. 1994; Mushinsky et al. 1997). We compared burrow number, condition, and size (width, see below) distribution between surveys and resurveys. Specifically, we compared estimated numbers of active and inactive burrows (definitions in McCoy and Mushinsky 1992), estimated numbers of inactive burrows relative to estimated numbers of active burrows, estimated numbers of abandoned burrows (definition in McCoy and Mushinsky 1992) relative to numbers of active plus inactive burrows, and size distributions. A decline in population size would be expected to be accompanied by a reduction in the absolute number of active plus inactive burrows, a decrease in the relative number of inactive burrows, an increase in the relative number of abandoned burrows, and an increase in the relative number of large burrows (McCoy et al. 2006). Methodological details may be found in McCoy et al. (2006). Results of the demographic surveys were related to results of antibody screening by simple matching.

The sizes (carapace length [CL]) of individuals that were captured away from their burrows for collection of blood samples (see below) did not appear to be a random sample of sizes even of adults. To determine if the size distribution of captured individuals was expected from the size distribution of all individuals in the same population, we compared the 2 distributions, the latter one as indicated by the size distribution of burrows. The width of a burrow is strongly positively correlated with the CL of the tortoise resident in the burrow (Martin and Layne 1987; Wilson et al. 1991; McCoy et al. 1993). We reduced burrow widths by 30% to account for the average difference between burrow width and CL (although it did not change the result if we reduced them by any value from 0% to 30%). We assumed that individuals of different sizes within a population used the same number of burrows, on average.

RESULTS

Blood samples were obtained from 152 individuals, and the blood was tested for the presence of antibodies to M. agassizii (Table 2). Seropositive individuals were identified at the 4 sites known previously to have seropositive individuals (BH, Gold Head Branch State Park [GB], Ichetucknee Springs State Park [IS], WS), as we expected. However, seropositive individuals also were identified at 4 sites not previously known to have seropositive individuals (LL, O'Leno State Park [OL], RA, Suwannee River State Park [SR]). Relatively small numbers of samples taken previously at 2 of these sites failed to yield seropositive individuals (Diemer Berish et al. 2000). We note that we were able to obtain only relatively small numbers of samples at the 2 sites where no seropositive individuals were identified (Fort Cooper State Park [FC], San Felasco Hammoch State Preserve [SF]) and suggest that more individuals be screened there. No significant difference in the proportional representation of seropositive (plus suspect) individuals between our study and the Diemer-Berish et al. (2000) study could be detected at IS or WS, but significant differences were detected at BH (Fisher's Exact test, p = 0.05) and GB (Fisher's Exact test, p = 0.04). In the first case, we identified relatively more seropositive individuals and, in the second case, relatively fewer than Diemer Berish et al. (2000).

Table 2. Results of antibody screenings.

View Fullscreen

Detection of seropositive individuals is strongly related to sampling effort. Plots of percentages of individuals determined to be seropositive (plus suspect) against the number of individuals tested revealed a strong linear relationship (Fig. 1). The relationship was not changed substantially if only seropositive individuals were used. None of the sites at which > 15 individuals were tested failed to yield seropositive and/or serologically suspect individuals (although some other sites not included in our analysis at which samples sizes were > 15 have failed to yield them; J. Berish, pers. comm., 2002). The probability that a seropositive individual was not sampled by the first, third, and fifth individual (at various sites) were, 0.25, 0.02, < 0.01 (BH), 0.52, 0.14, 0.04 (WS), 0.95, 0.85, 0.76 (LL), and 0.47, 0.10, 0.02 (RA), respectively. The first seropositive individual actually was either the second (BH, WS) or third (LL, RA) individual sampled. Such rapid detection was to be expected at BH, WS, and RA, where the percentage of seropositive individuals was 50% or greater but not at LL, where the probability that a seropositive individual was not sampled was still 0.44 by the 15th individual.

View Figure

• Download as Powerpoint
• Download Figure

A complete analysis of the demographic survey data is presented in McCoy et al. (2006). Comparisons of burrow number, condition, and size distribution between surveys uniformly suggest that gopher tortoise populations had declined at 5 sites (BH, FC, GB, OL, WS), and 4 of these sites either were known to have seropositive individuals before the resurveys or seropositive individuals were detected for the first time during the resurveys. Comparisons of burrow attributes between surveys also suggest, but less uniformly, that populations declined at another 3 sites (LL, RA, SR), and all of them have seropositive individuals. Finally, comparisons of burrow attributes between surveys suggest that populations were at least stable at 2 sites (IS, SF), and one of these sites has seropositive individuals.

Sizes of the individuals that were captured away from their burrows for collection of blood samples were not as expected based upon the burrow-size distributions at all sites. The numbers of individuals in size classes 3–4 vs. size classes 5–6 (Alford 1980) could not be distinguished from the expected numbers for BH or LL, but more individuals than expected were in size classes 3–4 at WS (χ² = 6.28, p = 0.01) and at RA (χ² = 3.92, p = 0.05).

DISCUSSION

When this study was designed, we did not anticipate either the widespread decline in gopher tortoise populations at the study sites, which are all conservation areas, or the widespread occurrence of seropositive exposure to M. agassizii. These 2 findings clearly weakened the ability of the study to reveal any connection between population change and the presence of the mycoplasma. Nevertheless, the results of our field study provide important cautions about erecting generalizations. The results do not unambiguously implicate the exposure of individuals to M. agassizii in the decline of these gopher tortoise populations. Some sites that showed no indication of decline had relatively high percentages of seropositive individuals (e.g., IS). Likewise, some sites that exhibited substantial decline either had no seropositive individuals or extremely low percentages of them (e.g., FC). The results reinforce the conclusion that the significance of exposure of gopher tortoise populations to mycoplasmal infections may be variable and often transient (see Brown et al. 2002).

The results of our field study, combined with our analysis of antibody screening data, strongly suggest that the sampling method can influence the judgment of whether a population has any seropositive individuals and the estimate of percentage of seropositive individuals in the population. Therefore, sequential samples of individuals taken from the same population could yield variable percentages because of actual differences in relative numbers of seropositive and seronegative individuals over time, because of uncertainties in the serological analyses (Brown et al. 2002; McCoy in prep.), or simply because of differences in sample sizes and/or techniques. Previous screenings for antibodies at some of our sites may have failed to reveal the presence of seropositive individuals either because the mycoplasma was not present at the time or because the previous screenings were of too few individuals. When relatively large samples of individuals are screened, relatively large numbers of seropositive individuals tend to be detected. However, even screening a large number of individuals at sites where the percentage of seropositive individuals is low still may not even detect exposure to M. agassizii. Capturing individuals that are above ground for collection of blood samples, rather than trapping individuals as they emerge from burrows, also may be problematic. The individuals available for easy sampling, because they are away from their burrows, were, on average, smaller than expected by chance, at some sites. If medium-sized individuals actually are more active above ground than larger individuals, then they also could have higher rates of infection with M. agassizii, either because of the increased chance of encounter with other individuals or because they are attempting to increase their body temperatures as a response to infection.

Overall, our findings indicate that exposure to M. agassizii is more widespread even than previously suspected. The occurrence of M. agassizii in the genus Gopherus appears to be ancient, predating the breakup of Laurasia, about 100 Mybp (Brown 2002). The immediate ancestor of the gopher tortoise and desert tortoise may have harbored the microorganism, and it was passed along to one or both daughter species when they diverged some 18 Mybp (Morafka et al. 1994; Berry et al. 2002; D. Morafka, pers. comm., 1999). The mycoplasma also could have been communicated between species during subsequent range expansions of the Xerobates and Gopherus grades (Bramble 1982) some 10 Mybp or later (Morafka et al. 1994; Berry et al. 2002; D. Morafka, pers. comm., 1999), or communicated between species by aboriginal man (“ancient pathogen pollution,” with subsequent spillover; MacPhee and Marx 1997; Daszak et al. 2000), or communicated between species relatively recently (and the mycoplasma is now spreading from one or a few focal points; Fischer et al. 1997; Ley et al. 1997; Dhondt et al. 1998; Hochachka and Dhondt 2000). Movements of infected individuals among populations over the past couple of decades cannot explain the current-wide distribution of the mycoplasma in the gopher tortoise in Florida (Diemer Berish et al. 2000), however, suggesting that microorganism and host may have coexisted for some time. The substantial geographical variation in the virulence of M. agassizii (Brown et al. 2002) and the possibly low genetic similarity among geographical isolates of the mycoplasma (L. Wendland, University of Florida, pers. comm., 2001) reinforce this suggestion.

Our findings also imply that the simplistic response that we have taken to the emergence of URTD may need to be reconsidered. Concerns about the apparent ease of transmission of the mycoplasma and the potential lethality of URTD naturally have led to attempts to minimize the risk of spread (Brown et al. 2002). The approach typically taken by regulatory agencies simply is to prevent contact between presumably infected and presumably noninfected individuals. For example, individuals at a donor site must be tested (ELISA test for antibodies to M. agassizii) in advance of translocation in Florida (Florida Fish and Wildlife Conservation Commission, 2001). The fact that the gopher tortoise is screened for exposure to M. agassizii before translocation clearly is appropriate (Cunningham 1996; Woodroffe 1999; Daszak et al. 2000), and caution that suggests that vigilance concerning transmission of M. agassizii should not be relaxed. However, if the mycoplasma already is very widespread and the population consequences of infection are not always dire, then the importance of M. agassizii and URTD within the general conservation strategy for the gopher tortoise may not be as clear as we have supposed (see McSweegan 1996; Daszak et al. 2000). For instance, if strains of M. agassizii are variable in virulence, as evidence now suggests (Brown et al. 2002), then careful isolation of high-virulence strains, on the one hand, and relaxation of the moratorium on translocation of the low-virulence strains, on the other hand, may be warranted. Furthermore, if the importance of M. agassizii and URTD are not as clear as we have supposed, then it becomes reasonable to ask if we have focused too much on minimizing the risk of spread of the mycoplasma, to the exclusion of considering other potential means of addressing URTD and also of focusing on the many other risks facing the gopher tortoise? The time probably has come for a more balanced conservation approach, taking into account the relative risks and associated costs involved not only of 1) transmitting M. agassizii among gopher tortoise populations but also of 2) transmitting the mycoplasma within populations or to other species, 3) failing to allow translocations, 4) dooming demographically valuable individuals because they are suspected of harboring mycoplasma, 5) underestimating the importance of other pathogens (such as tortoise herpesvirus), and 6) diverting attention and resources away from acquiring, managing, and restoring critical gopher tortoise habitat.

We suggest that maintaining or creating conditions necessary to minimize the chance of re-emergence of URTD, to prevent URTD from reaching epidemic proportions, and to allow populations to recover from URTD should be a concern that is treated at least on par with preventing transmission of M. agassizii. The necessary conditions most likely are those that promote reasonably large population sizes, prevent crowding, and moderate unpredictable environmental changes (see Vidaver 1996; Pounds et al. 1999; Diemer Berish et al. 2000; Daszak et al. 2000; Harvell et al. 2002). For the gopher tortoise, the best way to assure such conditions is with large areas of well-maintained habitat. By assuming that die-offs from URTD occurred well before we recognized the disease, then populations likely have recovered from it, when given the chance. For example, when we initially surveyed the SR site in 1990, we found 16 tortoise shells along the transects. Finding even 1 shell during a survey was a rarity at the time, so we had no ready explanation for finding so many and tentatively attributed it to poaching and predation, even though the shells largely were intact. In retrospect, we may have been surveying in the midst of an outbreak of URTD and, if so, any consequences of the disease were not particularly evident a decade later. As gopher tortoise habitat continues to decline in extent and quality, large populations in good habitat may continue to have a chance to recover, but small populations or populations in poorly managed habitat may be in serious danger of local extinction. Populations confined to small, isolated patches of habitat tend to occur at high densities, be affected by habitat degeneration, and, perhaps, be chronically stressed (Mushinsky and McCoy 1994; McCoy et al. 2007; McCoy and Mushinsky 2007). Furthermore, if the ultimate goal is for populations to be self-sustaining in the face of environmental pressures, including disease, then persistent veterinary intervention to keep small populations healthy, if such is even possible, may not be efficacious.

It seems strange in the 21st century to be arguing for the importance of habitat size and quality in conserving a species. Although habitat protection and, increasingly, habitat management normally form the cornerstone of a general conservation strategy, they seem to receive much less attention when disease considerations are prominent. For example, some gopher tortoise researchers have rejected the so-called protected lands paradigm (i.e., protect and manage the habitat and the population will remain viable) in the face of URTD (Seigel et al. 2003), despite the fact that habitat protection and management rather than direct veterinary intervention are becoming increasingly important means of coping with diseases of wildlife in general (Woodroffe 1999). We agree that the protected lands paradigm probably should be modified as a general principle. Protected lands are likely to allow for population recovery only if they are sufficiently large, although the definition of “sufficiently large” certainly is not clear in the case of the gopher tortoise (McCoy and Mushinsky 2007). We see at least 3 dangers in rejecting the protected lands paradigm out of hand, however. The first danger is that the forces for development will seize upon the rejection as a license to continue destroying habitat. The second danger is that the rejection leaves no reasonable alternative: we do not know how to manage small populations effectively in the absence of disease, let alone in the presence of disease. The third danger is that the rejection runs counter, in principle, to developing a sound general conservation strategy: a sound strategy requires protection of many large, heterogeneous areas of habitat (e.g., Soulé and Simberloff 1986). In our opinion, the role of habitat protection in conserving the gopher tortoise is enhanced rather than diminished in the face of URTD.