The Burmese star tortoise (Geochelone platynota) is a medium-sized tortoise (carapace length to 30 cm) endemic to the dry zone of central Myanmar, an arid region (< 1000 mm of rainfall/yr) formed by the rain shadow of the western mountains (Platt et al. 2011b). Chronic overharvesting of G. platynota, mainly for domestic consumption by rural Burmese (Blyth 1863; Theobald 1868), coupled with habitat loss resulted in widespread, albeit gradual, population declines over many years (Platt et al. 2011b). Harvesting greatly intensified during the mid-1990s, driven by the burgeoning demand from wildlife markets in southern China (Behler 1997; Altherr and Freyer 2000) where G. platynota was sold as food and incorporated into traditional medicines (Platt et al. 2000). Illegal collecting for the high-end international pet trade soon eclipsed demand from other sources, resulting in the near extinction of wild populations (Platt et al. 2011b). By the late 1990s and early 2000s, G. platynota was considered ecologically and functionally extinct in the wild, although scattered individuals no doubt persisted in some areas (Platt et al. 2011a, 2011b). Even those populations within the national protected area network were decimated by rampant poaching (Platt et al. 2001, 2003). Consequently, G. platynota was designated a member of “extinction row” along with other high-risk chelonian taxa (Buhlmann et al. 2002), listed as Critically Endangered by the International Union for Conservation of Nature, and ranked among the 25 most endangered chelonians in the world (Rhodin et al. 2011, 2017).

Given the critical status of wild populations, captivebreeding was recognized as the only remaining option for preventing the biological extinction of G. platynota and ultimately restoring this species as a functional member of the dry-zone ecosystem (Horne et al. 2012; Platt S.G. et al. 2017). In 2004, the Nature and Wildlife Conservation Division of the Myanmar Ministry of Environment, Conservation and Forestry, in collaboration with the Wildlife Conservation Society (WCS) and Turtle Survival Alliance Myanmar Program established 3 ex situ assurance colonies (defined as demographically and genetically viable captive breeding groups of imperiled taxa maintained as a hedge against the extinction of wild populations; Buhlmann et al. 2002) of G. platynota at Lawkanandar, Minzontaung, and Shwe Settaw wildlife sanctuaries (LWS, MWS, and SSWS, respectively) in Myanmar (Platt S.G. et al. 2017). Sites in state-owned wildlife sanctuaries were selected based on environmental requirements of the tortoises, ability to maintain security, and the availability of staff to care for the animals. The assurance colonies are located in the central dry zone of Myanmar where annual precipitation is 500–1000 mm, most of which falls during a brief wet season (mid-June through September) followed by an extended dry season (October–June). Ambient temperatures range from 4°C at night during January and February to 43°C during the day in the dry season. All 3 sites are within the historical range of G. platynota (Platt S.G. et al. 2017) and where wild populations occurred (SSWS and MWS) until extirpation in the 1990s. Founder animals originated from previous confiscations of illegally trafficked tortoises and tortoises that had been collected from the wild in MWS and SSWS in the early- to mid-2000s (Platt S.G. et al. 2017). The best estimate is that the founders consisted of 175 tortoises of approximately equal sex ratios (Platt S.G. et al. 2017). The colonies were created to 1) ensure the biological survival of G. platynota, and 2) produce offspring for reintroduction into protected natural habitat (Platt S.G. et al. 2017). Husbandry, diet management, veterinary care, and neonatal care were standardized among the assurance colonies. Ex situ propagation proved a resounding success and, by July 2018, > 14,000 offspring had been produced in captivity (K.P., unpubl. data, 2018; Platt K. et al. 2017; Platt S.G. et al. 2017). Complimentary to ex situ efforts, a national action plan for star tortoise conservation was prepared, habitat suitable for reintroduction was identified, local community support and participation by Buddhist monasteries was secured, and a cadre of management, enforcement, and husbandry personnel was assembled and trained (Platt K. et al. 2014a, 2014b, 2017). Participation by Buddhist monks was solicited and obtained to afford additional protection by appealing to local religious customs.

As the procedures for captive propagation were honed and the numbers of individuals burgeoned, plans for reintroduction were developed (Jacobson et al. 1999; Platt K. et al. 2017). These included securing enhanced protection for release areas, building soft-release pens (1.0 ha of natural grassland and forest habitat), recruiting and training personnel to radiotrack released animals, and performing prerelease health assessments. The initial release procedures involved collecting individual morphometric data, applying permanent identifiers and radios on all tortoises, and performing health assessments on all tortoises prior to placing the tortoises in the soft-release enclosures. The soft-release enclosures had shallow basins to provide drinking water to the tortoises. The plan was to keep groups of 50 tortoises in each of 3 enclosures. For the period that the tortoises were in the enclosures, no provisioning with food was made, in order to acclimate the tortoises to foraging on natural foods. Animals were radiotracked weekly from outside the enclosures to prevent damaging vegetation by walking on it. Initially, water was provided, but the frequency and amount of water was decreased until there was no provisioning prior to release. The releases were performed by making openings at the base of the fencing at 6-, 12-, or 18-mo intervals to allow the tortoises to self-release (Platt et al. 2015).

Although no known infectious disease had been observed in the captive population, care was taken to ensure management practices did not inadvertently introduce unknown infectious diseases into the wild via released animals. Obtaining baseline data at both the individual and population levels and identifying the presence or absence of potential pathogens are important when considering risk versus benefit of translocation programs (Wedland et al. 2009; Seimon et al. 2017). Ideally, free-ranging animals would be sampled and tested, and then the same testing would be performed on captive animals to compare health status of free-ranging tortoises with the captive population. However, there were no tortoises remaining in the wild to sample (Platt et al. 2011a); therefore, baseline data for Burmese star tortoises are limited to the captive population. Nevertheless, this information is important if importation of tortoises from other countries or other captive facilities in Myanmar were to be considered, or release of tortoises of unknown provenance into newly produced free-ranging populations were contemplated.

This report details the health assessment exams and testing that were performed on a large group of captive G. platynota in 2013 and 2014, by veterinarians from the Wildlife Conservation Society/Bronx Zoo (WCS/BZ), Turtle Survival Alliance, and Behler Conservation Center, using mobile laboratories from WCS/BZ, in Myanmar. The protocols followed in this program can serve as constructs for health assessments of other chelonian translocations and introductions in low resource settings.

METHODS

Animal handlings and samplings were performed at MWS and LWS (Platt S.G. et al. 2017). Those centers had been chosen to be the initial source populations for reintroduction, and the first releases were to be made adjacent to the MWS assurance facility, into the MWS. The SSWS colony had not been expanded sufficiently to provide tortoises for the initial releases. All tortoises that were sampled were hatched in captivity and head-started. They had been fed similar diets, ad lib, at both facilities throughout their rearing and during the period of data collection. Dietary items included water spinach (Ipomoea aquatica), various locally available grasses, carrot roots and tops, cucumber fruit, and Roselle (Malvaceae) foliage, augmented with banana leaves and stems, gourd foliage, prickly-pear cactus pads, and fruits of tomato, gourd, cantaloupe, papaya, and watermelon, depending on availability. Some tortoises had been hatched and reared at MWS while others had been moved to MWS from LWS ≥ 1 mo prior to the health assessments, in preparation for initial release into the MWS sanctuary.

From the time of inception of the facilities there was veterinary oversight and care at each facility. Preventive medicine and husbandry practices included limited routine fecal analysis, routine deworming, and segregated housing based on size, mass, and year of hatching. Individual animal treatments were performed for occasional conditions such as trauma or cloacal prolapses. Gross necropsies were performed on animals that died. The first releases to soft release enclosures were planned for late 2013 with subsequent releases to follow yearly. In preparation for release, comprehensive health assessments were performed in 2013 and 2014. The numbers of tortoises assessed at MWS in 2013 and 2014 were 152 and 151, respectively (Table 1). In 2013 at LWS, 84 tortoises (50 5–7-yr-old tortoises, and 34 < 2-yr-old tortoises) were sampled as part of the overall health surveillance program. In 2014 at LWS, another group (n = 152) of 5–7-yr-old animals was sampled as part of the release program (Table 1).

Table 1. Geochelone platynota sampled at captive assurance colonies at Minzontaung Wildlife Sanctuary (MWS) and Lawkanandar Wildlife Sanctuary (LWS), Myanmar, during 2013 and 2014.

View Fullscreen

Complete assessments consisted of body mass determination, physical examination, blood sampling, and swabbing of the choana and cloaca. Physical exams followed the protocols of Berry and Christopher (2001), but were adapted for use in captive tortoises. Any abnormalities found during physical exam were noted. The 34 juvenile tortoises at LWS in 2013 only underwent physical examinations and swabbing. Prior to sampling, tortoises chosen for potential release received temporary and permanent marking for identification (notching, microchip placement, carapace tattoos) and were outfitted with radiotransmitters (Ri-2B; Holohil Systems Ltd, Carp, ON, Canada) glued to the carapace, and a subset of that group had temperature monitors (iButtons, model DS1921G; Thermochron, Whitewater, WI) affixed to their carapaces. The time from marking until sampling varied but was within 2–3 hrs.

In 2013, portable gasoline generators were used to provide power for laboratory functions at MWS. However, gasoline generators were unable to operate continuously for the time required to perform polymerase chain reaction (PCR). Therefore, a larger, older diesel generator, capable of operating nonstop without a cool-down period, was obtained from an area village and transported to the laboratory with a water buffalo (Bubalus bubalis)–drawn cart for temporary use. Laboratory testing at MWS was performed in a building within 100 m of the tortoise handling site. Similar facilities were lacking at LWS so, after collection, samples were processed at a local inn approximately 2 km away, in rooms with a permanent electrical supply that had been outfitted with the laboratory equipment. This eliminated the need for portable generators. In 2014, some of the MWS, all the LWS hematology, and all the molecular testing from both MWS and LWS were performed at that same inn.

Hematology. — Blood sampling (n = 342; Table 2) was performed in the mornings before 1200 hrs, using 23–30-gauge needles on 0.3–1.0-mL syringes. Anatomical sites accessed included the subcarapacal sinus, dorsal tail vein, and brachial vessels. Blood collection sites were cleaned with alcohol, and a minimum of 0.2 mL and maximum of 1 mL of whole blood was drawn into heparinized syringes. The total blood volume collected was always < 0.6% of total body weight (Beaupre et al. 2004; Sykes and Klaphake 2008). In the immediate area where tortoises were handled, whole blood was thoroughly mixed and aliquots were placed into plain microcentrifuge tubes and onto each of 2 filter papers (FTA, and 903 protein-saver; Whatman Inc., Clifton, NJ). The blood samples were kept on coldpacks in insulated coolers until transported to the temporary hematology laboratories, which had been outfitted with a centrifuge, blood rotator, microscope, refractometer, and slide staining area. Filter papers were air-dried and then placed in plastic bags with desiccant for transport, long-term storage, and possible future analysis. The following parameters were determined: packed cell volume (PCV), total solids (TS), total white blood cell (WBC) count, red blood cell (RBC) count, and differential WBC determination following standard procedures (Campbell 2006; Strik et al. 2007; Sykes and Klaphake 2008). For PCV, microhematocrit tubes were filled with whole blood, centrifuged at 12,000 × g for 5 min and then percent volume of packed cells read off a scale. Total solids, sometimes referred to as total protein, were determined by placing the plasma portion of the microhematocrit tubes onto a hand-held refractometer and the resulting reading of grams/decaliter were recorded. The WBC and RBC counts were obtained by using Nat and Herrick's solution and hemocytometers. A dilution of 1:100 was made using a mixture of 1980 µL Natt-Herricks stain and 20 µL whole blood. A hemocytometer chamber was filled on both sides. For the RBC count, all the cells in the 4 corner and central squares within the multi-gridded central square on both sides of the hemocytometer were counted, divided by 2, and then multiplied by 5000. For the WBC count, all the WBCs in all the squares of each side of the hemocytometer were counted, divided by 2, and then multiplied by 1100 to get the total WBC count. For differential counts, blood smears were made on glass slides, dried, and subsequently stained with a 3-step modified Romanowsky staining method (DiffQuik; Baxter Diagnostics, Inc, McGaw Park, IL). Differential counts were obtained by microscopically counting 100 WBCs and identifying the different types of WBCs (lymphocytes, heterophils, eosinophils, basophils, and mono-azurophils). Veterinary technicians performed all the counts and differentials. Not all hematology tests were run on all samples, depending on the available sample volumes and the ability to run appropriate quality controls.

Table 2. Total numbers of individual hematology and polymerase chain reaction (PCR) pathogen testing done on captive Geochelone platynota at Minzontaung Wildlife Sanctuary and Lawkanandar Wildlife Sanctuary, Myanmar, during 2013 and 2014. TINC indicates intranuclear coccidian parasite of Testudines.

View Fullscreen

Hematology Statistical Methods. — To determine whether differences existed between sex or site for any of the hematology parameters, any individual that did not have sex identified was excluded from the analysis, leaving n = 318 sexed individuals. Linear mixed-effects models, with the interaction of sex and site (sex × site) and calendar year set as a random effect (Bates et al. 2015), were performed on all the hematology parameters. Full-model residual plots of the hematology data were visually inspected for normality (R Core Team 2017). Total solids and lymphocyte percentages (Lymph) were deemed to be normally distributed, while WBC counts, RBC counts, and PCV were log transformed to meet the assumptions of normality. For percentage of data not deemed normal (monocytes-azurophils [Mono-Azur], heterophils [Het], eosinophils [Eos], and basophils [Baso]), generalized linear mixed-effects models (glmer function; binomial family with logit link function; lme4 package) were performed (Crawley 2007). For all hematology parameters, the candidate models were the interaction of sex and site (sex × site), sex and site (sex + site), sex alone (sex), site alone (site) and were each compared with the null model (random effects only) using a likelihood ratio test (analysis of variance [ANOVA] function; stats package; R Core Team 2017). If the full model (sex × site) was significantly different from the null, the performance of candidate models was compared via Akaike's Information Criteria (AIC_c) corrected for small sample size (aictab function: AICcmodavg package: Mazerolle 2017). For models significantly different from the null model, but not distinguishable from other candidate models (< 2 ΔAIC_c), model-averaging was performed to assess the effect of the parameters given model uncertainty. Notched box-and-whisker plots were used to help clarify the differences of sex and site taking into consideration yearly differences. All hematology data were pairwise tested for correlation using Spearman's rho statistic (cor function; stats package; R Core Team 2017).

Molecular Testing. — Combined choanal and cloacal swab samples for PCR testing were collected from 539 tortoises (Tables 1 and 2) and placed into individual plastic tubes. The swabs (sterile MW113; Medical Wire and Equipment, Devizes, UK) were cooled and then transported to the portable molecular laboratory for immediate testing.

The molecular testing in Myanmar was performed to the same standards as in the laboratory facility at WCS/ BZ, including appropriate positive, negative, and inhibition controls. DNA was extracted by adding 150 µL of PrepMant (Thermo Fisher Scientific, Carlsbad, CA) to each swab, according to the manufacturer's protocol and extracts were heated to 95°C for 10 min and diluted 1:10 in RNAse/DNAse-free water prior to PCR. Positive control material consisting of a plasmid containing the primer or probe binding sites for each PCR assay was used and diluted to a range of concentrations to generate a standard curve and test the sensitivity for each assay. All PCR assays had a sensitivity of detection below 50 copies/µL in a PCR reaction. All qPCR assays were also tested for PCR inhibition using a synthetic positive control. PCR inhibition was observed in a subset 25 of the 539 samples tested by qPCR in the ranavirus qPCR assay (n = 18) and intranuclear coccidian parasite of Testudines (TINC) qPCR assay (n = 7). In these situations, the sample was further diluted (1:10) and retested. PCR inhibition was not observed in any of the retests except for one sample that needed to be diluted 1:100 before PCR inhibition was overcome.

For herpesvirus, samples (n = 539) were screened using a nested PCR that amplifies a 220-bp fragment of a conserved region of the herpesvirus DNA polymerase gene, using conditions and primers previously described (Ossiboff et al. 2015a).

For the Ranavirus qPCR assays (n = 236), samples were analyzed by real-time quantitative PCR targeting a conserved region of the Ranavirus major capsid protein gene using a consensus primer and probe set (CON F and CON R; Gilbert et al. 2012). Twenty-five-microliter reactions containing 5 µL of 5× Quantipath Master Mix (QuantiFast Pathogen PCR + IC Kit; Qiagen Inc., Germantown, MD), 900 nM each primer, 250 nM probe (Life Technologies, Grand Island, NY), 2.5 µL of 10× exogenous internal positive control primers and probe, 2.5 µL of 10× exogenous internal positive control DNA, DNase/RNase-free water, and 1 µL of diluted DNA were added to each well of a 48-well plate.

For the TINC, qPCR assays (n = 303) samples were analyzed by real-time quantitative PCR targeting a conserved region, the 18S rDNA region, using the primer and probe set (INCqPCRF and INCqPCR2R; Alvarez et al. 2013). Twenty-microliter reactions containing 10 µL of 2× Taqman Environmental Master Mix (Life Technologies), 900 nM each primer, 250 nM probe (Life Technologies), 2.5 µL of 10× exogenous internal positive control primers and probe, 0.5 µL of 50× exogenous internal positive control DNA (TaqMant Exogenous Internal Positive Control kit; Life Technologies), DNase/ RNase-free water, and 1 µL of diluted DNA were added to each well of a 48-well plate. The exogenous internal positive control reagents served as inhibition controls in the PCR reactions.

For Mycoplasma testing (n = 539), conventional PCR was performed to amplify a 451-bp segment of the 16S-23S intergenic spacer region of Mycoplasma using primers and conditions that have been previously described (Ossiboff et al. 2015b).

RESULTS

Hematology. — Mean hematological values (Table 3), when all tortoises were combined, were within biological ranges reported for other tortoises (Christopher et al. 1999; Zaias et al. 2006; Strik et al. 2007; Eshar et al. 2014; Lopez et al. 2017). The only hematological parameters that significantly correlated were Lymph negatively correlating to Het (r = -0.83). For Mono-Azur or Lymph percentages, there was no significant difference of sex, site, or the combination of sex and site. For the remaining hematology parameters, the effect of sex and site (via the determination of best models) are reported in Supplemental Table 1 (all supplemental material is available at http://doi.org/10.2744/CCB-1353.1.s1). For parameters where multiple models performed similarly, model-averaged coefficients and confidence intervals of the effect of sex and site are reported in Supplemental Figure 1. Taking into consideration yearly differences, for total WBC count, males had higher values than females, and tortoises at MWS had higher values than those at LWS. For RBC count, MWS males had higher values than MWS females or LWS males. There was no clear pattern for TS, PCV, and Eos when effects were averaged across their respective best models. For Het, MWS had lower values compared with LWS, but MWS males had higher values than MWS females or LWS males. For Baso, males had higher values than females and MWS had higher values than LWS, but MWS males had lower values than any other grouping.

Table 3. Hematological values of the Burmese star tortoise sampled at Minzontaung Wildlife Sanctuary and Lawkanandar Wildlife Sanctuary, Myanmar, during 2013 and 2014, presented as mean ± standard deviation (range), and sample size organized by sex, site, and year. Percentages for Mono-Azur, Het, Baso, and Eos were derived from 100 WBC counts from blood smear slides and their respective counts were derived from each individual's WBC count multiplied by the ratio of their respective WBC type.

View Fullscreen

Table 3. Extended

View Fullscreen

Molecular. — All samples screened, in both years, were PCR-negative for all target pathogens (Ranavirus, herpesvirus, Mycoplasma spp., and TINC).

DISCUSSION

The importance of health assessments of captive populations that serve as a source of animals for introduction into the wild cannot be overstated. In situations, as in the present conditions, where there were no wild animals from which to determine wild tortoise baseline data, due diligence required rigorous testing of the source population, as was performed. Once the health status of the source populations is determined, additional animals should not be incorporated into the source captive populations without rigorous quarantine and health assessment. For Burmese star tortoises, after successful reestablishment in the wild, health screening of the wild tortoises for baseline comparison with potential release animals should be conducted.

The decision to perform all the hematology and molecular testing in Myanmar was predicated on needing to provide timely results, rather than having to wait on the issuance of export and import permits for transporting samples to laboratories in the United States for testing. Additionally, tests for hematology requiring fresh whole blood (PCV, TS, hemocytometer counts) were, by necessity, completed in Myanmar. Challenges associated with setting up a temporary molecular laboratory were 2-fold: 1) physical transport of equipment, and 2) provision of constant electrical power. These did not affect the results of any tests, proved to be surmountable, and will become even less important in the future as equipment is refined.

As stated previously, mean hematological values were within biological ranges reported for other turtles or tortoises (Christopher et al. 1999; Berry and Christopher 2001; Campbell 2006; Zaias et al. 2006; Strik et al. 2007; Eshar et al. 2014; Lopez et al. 2017); but, for some hematology parameters, the differences between these values within groups may be explained by variations in environmental temperatures between sites, excitement level of animals being handled, and hydration status of individuals at the time of sampling (which may be affected by differences in water consumption). The interval between identification marking and time of sampling may also have had individual effects. There could also be variations between veterinarians, in terms of which venipuncture site was used (Gottdenker and Jacobson 1995) and amount of time animals were handled. In the future, those data should be recorded at the time of collection.

As with many baseline hematology data, the means ± standard deviations and ranges provide information that will prove helpful for future comparisons. By combining all the hematological data for 2 yrs, baseline hematology values have now been established for G. platynota. These data can be used for subsequent health screening of populations or individual animals. Variation from baseline may indicate stress (higher heterophils, lower lymphocytes), infection (increased heterophils or basophils in bacterial infections, increased or decreased lymphocytes in viral infections), malnutrition (lower packed cell volume, lower total solids), or other problems (lower RBC counts in trauma or hemoparasites) too numerous to recount here.

The numbers of tortoises screened for pathogens via PCR in 2013 (n = 236) was approximately 8.5% of the combined population at the 2 assurance colonies at the time. Assuming a prevalence rate of 5% for each pathogen of concern, only 90 animals would need to be sampled to have 99% confidence level that at least one animal would test positive for each pathogen (Samuel et al. 2003). All animals tested were PCR-negative for the target pathogens (based on the sensitivity of detection of the PCR tests). The tortoises sampled at LWS included 34 that were younger than 2 yrs old and that were subjectively assessed to be underweight for their age. They were being housed in enclosures that did not allow for adequate access to heat and sunlight and responded positively immediately to correction of those factors. They were included in PCR testing and were determined to be negative. We are confident that testing all tortoises destined for release was a good predictor for the presence or absence of those pathogens in the general population of the assurance colonies. However, all diagnostics tests have the potential for false-positive and false-negative results. We therefore cannot rule out the possibility of animals harboring levels of pathogen that are below the detection limit of the screening protocols, or that the animals may not be actively shedding the pathogen. In addition, if there had been any positive results on PCR, the product would have been sequenced to confirm whether those were true or false positives. The reported diagnostic results should be interpreted in the context of all other clinical and laboratory findings.

Previous health assessments of turtles in other countries have sometimes left conservationists with more questions than answers. In Myanmar, veterinary oversight of the assurance colonies, with concomitant physical exams and necropsies for years prior to release, has provided confidence that there have been no ongoing infectious processes affecting the health status of the tortoises in these colonies. Prior to development of PCRbased techniques for reptile pathogens, testing largely consisted of measuring antibody titers rather than determining presence or absence of a pathogen. Although antibody testing can often answer the question as to whether an animal has been exposed to an organism, antibody exposure (unlike PCR) will not reveal whether the animal currently harbors it or determine what strain or species may be present. Antibody titers can also be suggestive of subclinical infection for pathogens that are known to cause persistent latent infections (Brown et al. 2002; Marenzoni et al. 2018). However, there are few antibody tests, other than for Mycoplasma sp., which have proved useful in chelonians. Using PCR and DNA sequencing, it is possible to identify and compare different species and/or strains of organisms among different chelonians, although it does not distinguish between pathogenic or nonpathogenic status. This information allows researchers to make more informed decisions about the potential impact of those organisms on the overall health of the host. Recent investigations into Mycoplasma, herpesvirus, and ranavirus in Emydidae turtles in the Northeastern United States, as well as Batagur affinis in Cambodia, have shown that there are likely speciesspecific viruses and bacteria inhabiting animals naturally without causing disease (Ossiboff et al. 2015b, 2015c; Seimon et al. 2017). The presence of those organisms without signs of illness or population declines may suggest coevolution of host and potential pathogen. It is also possible that PCR-positive animals, while adapted to being carriers of some pathogens, may become clinically ill when immunocompromised. PCR-positive animals should be monitored over time to determine whether carrier states can cause active infections. A seemingly innocuous pathogen in one species may cause morbidity and mortality when transmitted to related species, as has been shown to happen with herpesvirus in painted and map turtles (Chrysemys picta and Graptemys geographica;Ossiboff et al. 2015a). Individual negative results should be evaluated with caution because it may only reflect lack of shedding, or sequestration in nonaccessible regions of the body of the pathogen at the time of sampling.

Mycoplasma, herpesvirus, Ranavirus, and TINC were chosen as the most relevant pathogens to screen for based on previous disease surveillance projects and reports of disease in tortoises (Une et al. 1999; Rivera et al. 2009; Jacobson et al. 2014; Agha et al. 2017; Ekatarina et al. 2017; Stilwell et al. 2017). Although there are likely to be potential pathogens for which tests have yet to be developed, based on what is currently known, the combination of hematology and physical examination together with PCR should be used to confirm health status and establish baseline information for G. platynota and other reintroduction programs (Raphael et al. 2016). The ongoing veterinary monitoring, including necropsies and preventive medicine, of the assurance colonies since their inception provides additional information on the overall health status of the tortoises. In the 14+ yrs since establishment, there has been no evidence of infectious disease in the captive-bred and head-started Burmese star tortoises. Now that these tortoises have been released into protected areas, have started to reproduce, and thus far have an extremely low death rate in the wild (Platt K. et al. 2017), it is reasonable to set their health status, derived from this study, as a benchmark for future releases. The animals that have been released and those in the assurance colonies are presumably naïve to the pathogens for which they were screened and as such, the introduction of any animals carrying those pathogens could have detrimental or even catastrophic consequences for the wild and captive populations. Moving forward, if no new animals are introduced to the assurance colonies, further prerelease health screening for G. platynota can be limited to individual animal examination by biologists and veterinarians. However, it is recommended that periodic monitoring for pathogens take place both in captive and free-ranging animals to assure that ongoing biosecurity practices are intact and that previously undetected or new pathogens are not identified. Molecular screening for select chelonian pathogens also sets a standard that can be used as an importation requirement for turtles or tortoises in Myanmar, even if these are not destined for release. Our results provide assurance that these specific pathogens are not circulating in tortoises in the Myanmar facilities and, therefore, we recommend that no additional tortoises be placed in those facilities from confiscations, wildlife trade, or animal imports without rigorous testing and quarantine.

As testing modalities improve and more pathogens become known, it will be important to look for those in the samples already collected and from new animals. Some of that testing could be performed on molecular products already extracted or on the whole blood stored on filter paper. Standards for best practices in reintroduction programs will continue to evolve to broaden the baseline knowledge of species. There is no universal agreement on the goals and breadth of prerelease testing, but the topic should remain open and be discussed in the context of specific programs and risk analysis (Brown et al. 2002). The protocols followed in this program can serve as a baseline model for other chelonian translocations and introductions, which can be modified dependent on the species, species medical history, and environment involved (Berry and Christopher 2001).

As of the end of 2017, almost 2000 tortoises had been released into MWS and SSWS. Reproduction has been documented in 7% of the released females, and 65% of the tortoises established home ranges in the release area (Platt K. et al. 2017). Future investigation into the physiology and health status of free-ranging Burmese star tortoises in the protected areas and assurance colonies should include fecal analysis, plasma biochemical determinations, nutritional evaluations with relevant mineral testing, screening for additional pathogens via molecular techniques and antibody testing, and establishing a condition index (Nagy et al. 2002; US Fish and Wildlife Service 2015). As resources become available, examining other chelonian species in the areas of release for pathogens is also recommended to determine what organisms exist that may have the potential to affect reintroduced G. platynota.