Island of misfit tortoises: waif gopher tortoise health assessment following translocation

evidence of infection. The presence of nasal discharge was an important, but imperfect, predictorof Mycoplasma spp.infectionintranslocatedtortoises.Hemogramdatawerecomparablewithwildpopulations.Our study is the first comprehensive effort to assess pathogen prevalence and hemogram data of waif gopher tortoises following translocation. Although caution is warranted and pathogen screening necessary, waif tortoises may be an important resource for establishing or augmenting isolated populations when potential health risks can be managed.


Introduction
Translocation is the intentional movement of animals from one location to another. The practice has become an important tool for proactively managing imperilled wildlife species (Johnson et al., 2010;Biggins et al., 2011;Kraus et al., 2017). Despite the potential for success and its common use (Fischer and Lindenmayer, 2000;Germano and Bishop, 2009), translocation also carries many potential risks to wildlife health (Deem et al., 2001;Kock et al., 2010). In the United States, several pathogens and parasites have spread through the translocation of wild turkeys (Meleagris gallopavo; Castle and Christensen, 1990), raccoons (Procyon lotor; Schaffer et al., 1981) and white-tailed deer (Odocoileus virginianus; Davidson et al., 1996;Cohen et al., 2018). Additionally, translocation can induce physiological stress (Teixeira et al., 2007). In addition to the stress associated with capture and transport, translocated individuals may experience behavioural changes following release, such as increased movement and diet shifts, that result in more prolonged physiological stress (Dickens et al., 2010). Increased stress can cause mortality (Hartup et al., 1999) or indirectly affect health through immune system suppression and consequent downstream effects (Teixeira et al., 2007).
Because herpetofauna, including chelonians, have declined in recent decades (Gibbons et al., 2000;Stuart et al., 2004;Stanford et al., 2020), conservationists are increasingly tasked with balancing the potential benefits of management interventions such as translocation with potential adverse effects (Jacobson, 1994;Walker et al., 2008;Aiello et al., 2014). Numerous translocation efforts have been successful for herpetofauna species (Fitzgerald et al., 2015;Jarvie et al., 2016;Bell and Herbert, 2017;Kraus et al., 2017). However, emerging infectious diseases pose challenges for both the conservation of in situ populations (Daszak et al., 2000;Tompkins et al., 2015) as well as translocation programs (Pessier, 2008;Walker et al., 2008). The gopher tortoise (Gopherus polyphemus) is a long-lived species endemic to the southeastern United States and is among the most commonly translocated reptile species (Tuberville et al., 2008). The species has been federally listed as threatened in the western portion of its range since 1987 (United States Fish and Wildlife Service [USFWS], 1987) and has more recently become a candidate for federal listing throughout the remainder of its range (USFWS, 2011). Historically, gopher tortoises have been translocated as mitigation for construction activities (Florida Fish and Wildlife Conservation Commission [FFWCC], 2006;FFWCC, 2012;Sullivan et al., 2015). As development continues to fragment tortoise habitat, there is growing interest in using translocation as a tool to manage population viability (Gopher Tortoise Council, 2014).
Although habitat loss is the primary driver of declines for gopher tortoises (Smith et al., 2006), disease is also considered a threat (Jacobson, 1994;McLaughlin, 1997). Upper respi-ratory tract disease (URTD) caused by pathogenic bacteria in the genus Mycoplasma (with possible revision to genus Mycoplasmopsis; Gupta et al., 2018), has historically been the focal disease of management concern (Jacobson et al., 2014). Experimental trials confirmed M. agassizii as the etiologic agent of URTD for gopher tortoises (Brown et al., 1999), and a second etiologic agent, M. testudineum, was later identified (Brown et al., 2004). URTD has been a suspected factor in declines for both Mojave desert tortoises in California (Gopherus agassizii, Jacobson et al., 1991b;Berry et al., 2020) and gopher tortoises in Florida (Gates et al., 2002;Seigel et al., 2003). Infections with Ranavirus spp. in gopher tortoises have also been documented (Westhouse et al., 1996;Johnson et al., 2008;Cozad et al., 2020b), and although the effects on tortoise health are poorly understood, infections are often lethal in other chelonians (De Voe et al., 2004;Sim et al., 2016). Recent work has also revealed infection with a novel Anaplasma sp. (i.e. Candidatus Anaplasma testudinis), which may contribute to anaemia in gopher tortoises (Crosby et al., 2021). In addition to previously documented pathogens, there are ongoing efforts to identify novel pathogens that may impact tortoise health (Desiderio et al., 2021).
There are risks associated with any translocation; however, waif tortoises-tortoises that have been collected illegally, injured and rehabilitated or have unknown origins (FFWCC, 2012)-are typically excluded from translocation efforts due to heightened concerns of introducing pathogens. In the case of the congeneric desert tortoise, released pets have been implicated in spreading pathogens, including the causative agents for URTD (Schumacher et al., 1993;Jacobson, 1994;Johnson et al., 2006;Berry et al., 2015). Additionally, stress is an important concern with waif tortoises due to their history of captivity. Even short-term captivity can be stressful to wildlife (Gregory et al., 1996), and prolonged captivity can alter behaviour, physiology or nutritional status (Mason, 2010;DeGregorio et al., 2013DeGregorio et al., , 2017, potentially making the transition back to the wild more stressful. However, as the species continues to decline (Smith et al., 2006), waif animals could provide the needed individuals to bolster isolated wild populations. Additionally, an understanding of the suitability of formerly captive individuals for release will be important for the conservation of other long-lived reptiles that face similar challenges following their illegal collection and subsequent confiscation (Rosen and Smith, 2010;Mendiratta et al., 2017).

Conservation Physiology • Volume 10 2022
Research article tortoises from a variety of origins were released in the preserve (McKee et al., 2021). No tortoises with clinical evidence of infection were released, but resource limitations prevented pathogen screening prior to release. Because waif tortoises present an ongoing management challenge and their potential contribution to recovery of wild populations is unknown, it is important to assess the health of released individuals. The objective of this study was to perform comprehensive health assessments of gopher tortoises released at the AGTHP, including visual examination, pathogen screening using traditional and quantitative PCR and haematological evaluation. Collectively, these data will serve as a monitoring baseline for comparison with future surveillance studies of this unique population, as well as inform a retrospective evaluation of the disease risks associated with translocations of waif gopher tortoises.  ), and any ticks were removed. Groups of tortoises (x = 13 adults) were soft-released into 1-ha circular pens for at least 10 months (following methods of Tuberville et al., 2005) to promote site fidelity and development of social relationships (see McKee et al., 2021 for details on release groups and tortoise origins).

Animal collection
This study was conducted in accordance with state permits (SCDNR Scientific Collection Permit Number #SC-04-2017, #SC-06-2018) and approved by University of Georgia IACUC protocols (AUP# A2017 05-022-Y1-A0). To locate tortoise burrows at AGTHP, we walked parallel transects spaced 15 m apart in all suitable habitat throughout the preserve during May-June 2017 and Feb-May 2018. We used a burrow camera to determine occupancy (Smith et al., 2005). If a burrow was occupied, we placed a wire-live trap covered in shade cloth at the burrow opening (Aresco and Guyer, 1999). To prevent tortoises from overheating, traps were checked multiple times daily (roughly every 4 h during 09:00-17:00). We also opportunistically captured any tortoise encountered outside of a burrow. Tortoises were transported in individual bins to the Savannah River Ecology Laboratory (located 38 km from AGTHP) where they were evaluated. Following their assessments, tortoises were kept overnight and returned to their point of capture within 24 h. During 2017 and 2018, wildlife agencies provided additional waif tortoises for release at AGTHP. Prior to their release, we processed these individuals similarly to captured tortoises.
We recorded the mass of tortoises to the nearest 2 g and measured to the nearest 1 mm the midline carapace length (MCL) from the nuchal scute to the supracaudal scute, the width at the widest point of the carapace, and the shell height at the highest point of the shell. We calculated tortoise body condition by dividing the mass (g) by the body shell volume (Loehr et al., 2004;Daly et al., 2018). To approximate shell volume, we used the formula for the half ellipsoid: shell volume (cm 3 ) = (π × MCL × width × height)/6000. Tortoises were classified as hatchlings, juveniles, subadults, adult males or adult females by their MCL and secondary sex characteristics (McKee et al., 2021). We considered female tortoises <230 mm to be mature adults if they were determined as gravid via palpation.

Visual health assessments
Because many chelonian pathogens target the respiratory tract (Origgi and Jacobson, 2000), we examined tortoises for evidence of upper respiratory infection. Individuals exhibiting nasal discharge, eroded nares, ocular discharge, ocular swelling or conjunctivitis were flagged as having clinical evidence of infection for known gopher tortoise pathogens, such as Mycoplasma spp. or Ranavirus spp. (Brown et al., 1999;McLaughlin et al., 2000;Brown et al., 2004;Johnson et al., 2008). Additionally, we assessed each tortoise's carapace, plastron, skin, cloaca, oral cavity, nares, eyes, tympanum and respiration and noted any abnormalities (see McKee, 2019 for full criteria). During examination, tortoises were also checked for ectoparasites.

Pathogen screening
We screened for 14 pathogens commonly associated with either wild or captive chelonian populations (Table 1). Due to the possibility that waif gopher tortoises had been exposed to conspecifics or other chelonian species while in captivity, our pathogen panel included some that have not specifically been documented in gopher tortoises but have been known to affect other species (Table 1). Separate oral and cloacal swabs were collected from each tortoise using sterile Copan Diagnostic Flocked Swabs ® (Copan Diagnostics Inc, Murrieta, CA, USA). We also collected a nasal swab if a tortoise exhibited nasal discharge during assessment. Unlike previous research that used antibody tests to determine exposure (McLaughlin et al., 2000;McGuire et al., 2014), swabs were used for molecular testing to determine the presence of pathogen DNA.   Wellehan et al., 2004 Swabs were analysed at the Wildlife Epidemiology Laboratory at the University of Illinois where they were tested for 14 pathogens, including 10 pathogens tested in both 2017 and 2018 using conventional or quantitative polymerase chain reaction (PCR, qPCR respectively). Supporting citations are provided for pathogens that have been previously documented in gopher tortoises. This screening method may result in an individual's status changing over sampling events Burgess et al., 2021). As such, if an individual captured in 2017 was recaptured in 2018, we collected oral swabs in both years and calculated prevalence annually. We analysed cloacal swabs in 2017 but did not detect pathogens (including individuals for which pathogens were detected in oral swabs), thus we did not analyse additional cloacal swabs for tortoises recaptured in 2018. However, we continued to collect and analyse cloacal swabs for tortoises first captured in 2018. We stored swabs at −80 • C until analysis.
We extracted the DNA in oral and cloacal swab samples with a QIAamp Blood mini Kit (QIAGEN Inc., Redwood City, CA, USA), following the manufacturer protocol. Quantity (ng/μl) and quality (A260:A280 ratio) of DNA were evaluated using a Nanodrop spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA, USA). We performed qPCR in a multiplex format using published or in-house primer-probe assays to evaluate 13 of the pathogens-10 were assessed in both years, 2 were assessed in 2017 only and 1 was assessed in 2018 only (Table 1; Detection method = 'qPCR'). Initially, specific target amplification was performed on each sample with pooled pathogen Taqman assays and preamp mastermix (Thermo-Fisher, Waltham, MA, USA). Each reaction was performed under the following cycling program on an MJ Tetrad thermocycler: 95 • C (10 min), 14 cycles of 95 • C (15 sec) and 60 • C (4 min). The qPCR assay was then performed in triplicate using 2.25 μl of amplified DNA from the first reaction on a Fluidigm 96.96 Gene Expression IFC and amplified on the Fluidigm Biomark HD Real Time PCR thermocycler (Fluidigm, South San Francisco, CA, USA) using the following cycling protocol: 70 • C (30 min), 25 • C (10 min), 95 • C (1 min), followed by 35 cycles at 96 • C (5 sec) and 60 • C (20 sec). Serial dilutions of positive controls for FV3-like Ranavirus, M. agassizii and M. testudineum were prepared from 10 7 to 10 1 copies per reaction. A non-template control was included on each plate. All reactions were then analysed using Fluidigm Real Time PCR analysis software (Fluidigm, South San Francisco, CA, USA). Following Fluidigm analysis, all positive samples were verified in a simplex reaction. Briefly, qPCR was performed in triplicate on a QuantStudio3 real time thermocycler. Samples were considered positive if all three replicates had a lower cycle threshold (C t ) value than the lowest detected standard dilution.
We used conventional PCR with a previously characterized two-step consensus assay for adenovirus detection for 2017 samples (Wellehan et al., 2004; Table 1; Detection method = 'PCR'). Products were electrophoresed on a 1% agarose gel and compared with positive and negative controls and a 100 bp DNA ladder. The PCR products producing appropriately sized bands (approximately 320 base pairs) were treated with Exo-SAP-IT (USB Corporation, Cleveland, OH 44128, USA), sequenced in both directions (W.M. Keck Center for Comparative and Functional Genomics, University of Illinois at Urbana-Champaign, Urbana, IL) and compared with known sequences in GenBank using BLASTN to confirm accurate detection.

Hemogram evaluation and hemoparasite quantification
Prior to collecting blood, the collection site was disinfected with alcohol wipes. For subadult and adult tortoises, we collected 0.5-1.5 ml of blood via the brachial vein using a 25gauge heparinized needle. For juveniles, we used a 29.5-gauge needle to collect 0.3-0.6 ml of blood from the subcarapacial vein. Volume of blood collected did not exceed 0.5% of any tortoise's body weight. Blood was not collected from hatchling tortoises. To minimize handling time for tortoises, we only collected blood from tortoises during their first capture, resulting in one blood sample per tortoise. Blood samples with overt lymph dilution (as evidenced by the sample coloration) were removed from hemogram analysis. For each tortoise, a capillary tube was filled with whole blood and centrifuged (LW Scientific, Zipocrit; 3636×g) to determine packed cell volume (PCV), a metric that quantifies the red blood cell (RBC) column relative to plasma. We used a refractometer (Reichert Technologies, Depew, NY, USA) to estimate total protein (TP; gL −1 ). Following blood collection, we immediately prepared three to five blood smears that were then fixed with methanol and allowed to air dry prior to storage.
Blood films were stained using Wright-Giemsa (Harleco ® , EMD Millipore, Billerica, MA, USA) and screened for hemogregarines by counting the number of gametocytes in RBCs per 100 RBC. White blood cells (WBC), including WBC estimate (Weiss, 1984) and 200 WBC differential count, and blood cell morphology (WBC, RBC and thrombocytes) were assessed by blood film evaluation, an approach that has been used for many wildlife species (Davis et al., 2008), including gopher tortoises (Goessling et al., 2016;Cozad et al., 2020a). Heterophils were categorized as either immature or mature (Stacy et al., 2017;Stacy et al., 2022). We calculated heterophil:lymphocyte ratios (H:L ratios) by dividing total heterophils (mature and immature) by total lymphocytes for each individual.

Modelling predictors of infection with URTD-associated pathogens
We used logistic regression models to identify individual attributes useful in predicting infection with M. agassizii or M. testudineum. Individuals were considered 'infected' if they tested positive for either pathogen. Previous research has documented relationships between gopher tortoise infection status and sex/life stage (Karlin, 2008;Wendland et al., 2010), body condition (Cozad, 2018;Goessling et al., 2019) and PCV (Cozad, 2018). In reptiles and birds more generally, increases in heterophils may be indicative of inflammation (Stacy et al., 2011;Juul-Madsen et al., 2014;Stacy et al., 2017). Therefore, we included tortoise stage class immature heterophils (×10 3 μl −1 ), absolute total heterophils (immature and mature heterophils; ×10 3 μl −1 ), evidence of nasal discharge (present/absent), evidence of nasal erosion (present/absent) and ocular evidence of infection including ocular discharge, swelling or conjunctivitis (present/absent) as model parameters. We considered a null model containing no effects, all single parameter models, and all but two models formed from additive combinations of two parameters: because of evidence of strong collinearity between variables, we excluded the two models pairing immature with total heterophils and pairing PCV with total heterophils (Spearman's correlation test rho =0.34, S = 203 680, P< 0.001; rho= 0.23, S = 23 726, P < 0.01). Although a larger list of candidate models could be generated through additional combinations of variables, we limited our consideration to a smaller set of candidate models that (1) related to previously suggested hypotheses or documented patterns and (2) had simple structure given the small number of positive animals. Because tortoises were placed in soft-release pens on site in groups prior to release, we originally included release pen as a random effect in all candidate models. However, because the estimated variance parameter for pen was essentially zero (indicating no detectable difference in infection probability among pen groups), we subsequently removed release pen from all models.
To compare candidate models, we fit models to data that contained a complete set of individual attributes for each tortoise. Due to few complete records for juveniles (n = 2) and subadults (n = 14), we combined these stages into a single group classified as 'immature' in all candidate models. To more easily make comparisons across models with different continuous variables, we centred and scaled continuous variables (PCV, body condition, immature heterophils, total heterophils) by subtracting the mean and dividing by the standard deviation (SD) of each variable. We used Akaike's Information Criteria corrected for small sample sizes (AICc) to compare candidate models (Hurvich and Tsai, 1989;Akaike, 1998). We report the results from all candidate models with a delta AIC < 2. For all averaged observations we report the means ± SD. We conducted all analyses in R version 4.0.2 (R Development Core Team, 2020) and used package Multimodel Inference ('MuMIn') version 1.43.17 for model selection (Barton, 2020).

Visual health assessment
During 2017-2018, we captured 130 individuals at the AGTHP and received 13 additional waif tortoises for release at the preserve. In total, we visually assessed 143 individual tortoises (including 46 individuals that were assessed in both 2017 and 2018), comprising 68 females, 50 males, 17 subadults, 5 juveniles and 3 hatchlings (Table 2). Of the 63 tortoises assessed in 2017, 2 (3.1%) exhibited ocular abnormalities (discharge, swelling or redness), 3 (4.8%) had nasal discharge and 4 (6.3%) had nasal erosion. In 2018, we assessed 126 individuals, of which 11 (8.7%) exhibited ocular abnormalities, 11 (8.7%) had nasal discharge and 14 had nasal erosion (11.1%). Of these animals with abnormalities, four exhibited both nasal discharge and erosion at the time of assessment, three exhibited both ocular abnormalities and nasal discharge, two individuals had both ocular abnormalities and nasal erosion and no individual exhibited all three abnormalities simultaneously. We did not observe any other substantial clinical abnormalities (e.g. recent shell or external injuries) or ectoparasites (including ticks) in either year. We did not see any individuals that appeared distressed at time of sample collection or that required veterinary intervention at the time of study. In addition to live animals, we also recovered the shells of nine deceased individuals over the course of the study. However, at the time of recovery, no soft tissue remained on the carcasses and no necropsies were possible.

Pathogen screening
In total, we screened 143 individuals with at least one swab type and obtained oral swabs from 139 of those individuals. Although we detected Mycoplasma spp. in both years, we did not detect any other pathogens of concern (including four species of Ranavirus) in our screenings (Table 1). In 2017, we tested 63 individuals using oral (n = 59), cloacal (n = 62) and/or nasal swabs (n = 2) for 13 pathogens (Table 1). In 2017, M. agassizii was the only pathogen detected, with a prevalence of 10.2% in the oral swabs [95% confidence interval (CI): 3.8-20.8%; n = 6 positive swabs]. We also detected M. agassizii in one nasal swab. This individual's oral swab was also positive, with the amount of DNA detected from the oral swab (15.1 copies/μl) over 10 times greater than the amount detected from the nasal swab (1.4 copies/μl; Table 3). In 2018, we tested 124 individuals using oral (n = 122), cloacal (n = 82) and/or nasal swabs (n = 4) for 11 pathogens (Table 1). We detected M. agassizii with a prevalence of 13.9% (95% CI: 8.3-21.4%; n = 17 positive) and M. testudineum with a prevalence of 0.8% (95% CI: 0-4.5%; n = 1 positive) in oral swabs. No tortoise was concurrently infected with both pathogens. All four of the nasal swabs that tested positive for M. agassizii were from individuals whose oral swabs also tested positive for the pathogen. We only detected M. agassizii in cloacal swabs from three individuals that also had positive oral swabs (Table 3). In 2018, we also detected emydid Mycoplasma in the oral swab of a single tortoise, a naturally recruited hatchling (i.e. not a released animal), that showed no clinical evidence of infection. Swabs collected in 2017 were not screened for this pathogen.  If multiple records of the same individual occurred during the sampling period, they were averaged prior to their incorporation into the stage class summary statistics. SDs and number of individuals (n) in each calculation are also provided.   URTD-associated pathogens in 2018-three for M. agassizii and one for M. testudineum.

Hemogram evaluation and hemoparasite quantification
In total, we collected blood from 132 individuals. All haematological data are presented by life stage in Table 2. Blood films from 57 individuals in 2017 and 72 individuals in 2018 were reviewed. In 2017, no infectious agents, including no evidence suggestive of Anaplasma spp., were detected. In 2018, rare hemogregarine infections were identified (<1 infected RBC/100 RBC) in two adult individuals (prevalence = 2.6%) with PCV of 29 and 31%, respectively. Anisocytosis (variation in red blood size) was characterized as 'absent' in 94 individuals, 'mild' in 33 individuals and 'moderate' in two individuals. Polychromasia (presence of immature red blood cells) was considered 'absent' in 99 individuals, 'mild' in 28 individuals and 'moderate' in two individuals. Overt morphological abnormalities in WBCs were not detected in any of the tortoises. Thrombocytes were considered adequate in all individuals.

Modelling predictors of infection with Mycoplasma pathogens
For 123 tortoises with complete sets of predictor variables [107 negative, 16 positive for either M. agassizii (n = 15) or M. testudineum (n = 1)], we fit data to 35 candidate models (Table 4). We found the presence of nasal discharge to be the most important predictor of infection, as it was the sole parameter in the top model (22.4% model weight) and the only factor identified as a significant predictor of infection (P < 0.001; Table 4; Fig. 1). Additionally, nasal discharge appeared in the top 8 models, which collectively accounted for >96.6% of model weight. We observed nasal discharge in 31.3% of positive animals and 3.7% of negative animals. The second most supported model (18.5% model weight) included the additive effects of tortoise stage class and nasal discharge (Table 4). Although the point estimate for the infection probability in immature tortoises was slightly higher than the infection probability of adult males and adult females, there was substantial variation in infection probability (Fig. 1). The third best model (12.2% model weight) included the additive effects of nasal discharge and immature heterophils (×10 3 μl −1 ), and model estimates indicated a positive relationship between absolute immature heterophils and infection probability (Table 4; Fig. 2). Models with the additive effect of nasal discharge and body condition, nasal discharge and PCV, nasal discharge and nasal erosion, nasal discharge and total heterophils and nasal discharge and ocular evidence of infection all received a similar weight, suggesting a high level of uncertainty among alternative models containing the nasal discharge predictor (Table 4)

Discussion
Our study is the first to assess the health of gopher tortoises in a population established almost entirely through the release of waif animals. Overall, most captured individuals appeared clinically normal. Despite the varied, often unknown histories of individual animals, we detected very few pathogens in this unique and isolated population. We documented the presence of only two known tortoise pathogens, M. agassizii with 10.2% and 13.9% prevalence and M. testudineum with 0.0% and 0.8% prevalence in 2017 and 2018, respectively. These pathogens appear to be common in wild gopher tortoise pop-ulations (McCoy et al., 2007;Berish et al., 2010;Goessling et al., 2019), and prior exposure has been documented in wild populations throughout much of the tortoise's range, including Florida (Karlin, 2008;Ozgul et al., 2009), Georgia (McGuire et al., 2014), Louisiana (Diaz-Figueroa, 2005, Mississippi (Smith et al., 1998) and Alabama (Goessling et al., 2019). However, prevalence varies substantially among populations, with some populations having either very low (0-3%) or very high seroprevalence (96%-100%; McGuire et al., 2014). In a recent study that used qPCR to evaluate M. agassizii infection in seven Alabama tortoise populations, prevalence rates ranged from 0% to 6%  (Goessling et al., 2019). For a translocated population in Florida that also used qPCR, M. agassizii prevalence was estimated to be 20% (Cozad et al., 2020b). Prior research specifically on waif tortoises is limited, but the few existing studies that have identified M. agassizii in waif tortoises have also observed a relatively low number of positive individuals (Whitfield et al., 2018;Elbers and Taylor, 2019).
We found that presence of nasal discharge was an important (but imperfect) clinical predictor of current infection with URTD-associated pathogens, as has been previously reported in both gopher tortoises and desert tortoises (Schumacher et al., 1997;Karlin 2008;Berish et al., 2010). Previous research has shown that nasal discharge may appear more quickly post-infection than other metrics such as antibody presence (Brown et al., 1999;Brown et al., 2002;Aiello et al., 2019;Drake et al., 2019); thus, presence of nasal discharge may be a useful screening tool for identifying candidates to quarantine, determining suitability of tortoises for release or prioritizing individuals for pathogen testing. Nasal erosion and ocular evidence of infection appeared to be less important predictors of active infection. The presence of nasal discharge appeared in the eight best models, while nasal erosion and ocular evidence of infection appeared only in the sixth and eighth model, respectively (Table 4). Only three of the 21 animals positive for either pathogen exhibited ocular evidence of infection and these individuals also had nasal discharge. Similarly, ocular abnormalities were found to be unreliable predictors of Mycoplasma spp. infection in Mojave desert tortoises (Burgess et al., 2021). Although nasal erosion was a weaker predictor of current infection, it likely provides useful information on prior pathogen exposure, as it has been shown to be positively correlated with the presence of M. agassizii antibodies (Goessling et al., 2019). Because URTD is a recrudescent disease where tortoises can begin shedding bacteria again , evidence of prior infection may still be informative for management or for screening acquired waif gopher tortoises for potential release.
Despite the attention URTD has received in scientific literature, the overall effects of the disease in Gopherus tortoises are still debated (Sandmeier et al., 2009;Berish et al., 2010). In many cases, URTD appears to have only limited effects on tortoise survival (Karlin, 2008;Ozgul et al., 2009) and population persistence (McCoy et al., 2007). Rarely, however, URTD has been implicated in mortality events (Jacobson et al., 1991b;Gates et al., 2002;Seigel et al., 2003). In our study, the majority of infected individuals were carriers of the pathogen without any evidence of clinical disease and were clinically normal. However, we noticed a high rate of infected tortoises in a single soft-release pen (pen 10), with 50% (n = 6/12) of sampled individuals testing positive in either 2017 or 2018 for M. agassizii. Of the nine shells recovered during the study period, eight were recovered from this pen. Considering that 22 animals (12 adults) had been released during 2016-2017, there was at least a 36% mortality rate for this pen. Because seven out of eight of these individuals died before the study, we cannot conclude with certainty that the higher mortality in pen 10 was the result of a specific pathogen, and which other factors may have played a role. Non-infectious factors, such as stress associated with release density, have been linked to increased mortality in other translocated populations (Cozad et al., 2020a). However, at AGTHP, release densities and conditions were similar for all cohorts, and we did not observe high mortality rates for the population as a whole (McKee et al., 2021). Therefore, we suspect M. agassizii may have contributed to the higher mortality rate observed in pen 10. Although we captured seemingly healthy animals positive for URTDassociated pathogens, similar variability in host response has been noted in desert tortoises experimentally infected with the pathogen . Prior research suggests that high morbidity is related to high prevalence, indicating an effect of pathogen load in populations . Because there was a higher prevalence observed in pen 10 relative to the rest of the population, individuals in this pen may have exhibited worse outcomes than infected individuals elsewhere on the preserve. Variability in pathogen strain can also result in different virulence patterns (Perez et al., 2020), and additional work is likely needed to better characterize the genetic diversity of URTD-associated pathogens as it relates to virulence (Weitzman et al., 2017).
Given prior findings that immature gopher tortoises are less likely to be infected with Mycoplasma spp. than adults (Karlin, 2008;Wendland et al., 2010;Page-Karjian et al., 2021), it is unexpected that our model estimated infection probability to be higher for younger animals. Because this population is comprised of waif animals that were previously housed in captive facilities, normal exposure patterns could have been altered by increased contact associated with shared housing and provisioning for individuals across life stages while in captivity. Additionally, prior studies have used antibody presence to determine the exposure status of individuals, which captures prior exposure but is not always indicative of current shedding of the pathogen (McCoy et al., 2007;Wendland et al., 2010;McGuire et al., 2014;Aiello et al., 2019). As pathogen screening using qPCR analysis of swabs becomes more common in health assessment studies, we will be able to better compare our results to other gopher tortoise populations.
There is growing interest in using readily available diagnostic tools such as haematology to understand inflammation as it relates to stress, immune functions and infection status in reptiles (Sandmeier et al., 2018;Rosenberg et al., 2018a;Neuman-Lee et al., 2019;Sandmeier et al., 2019a). Although inference from our results is limited by a low number of pathogen-positive animals, we observed a positive correlation between absolute immature heterophils and probability of infection with URTD-associated pathogens (Fig. 2). This finding supports previous reports that immature heterophils play an important role in assessing clinical status and active inflammation in Gopherus spp. tortoises (Stacy et al., 2017; .  Rosenberg et al., 2018a;Sandmeier et al., 2019a). In the congeneric Mojave desert tortoise, haematological abnormalities were present in tortoises with clinical signs of infection with M. agassizii (Christopher et al., 2003). Future studies that include a larger sample size of Mycoplasma spp.positive animals could help clarify the relationship between infection status, number of immature heterophils and other haematological and immune function analytes. Understanding these analytes and their significance for population health in endangered species will become increasingly important for identifying stressors and other specific impacts that could require targeted mitigation strategies and conservation efforts (Ohmer et al., 2021).
In addition to the two known tortoise pathogens, we also documented the presence of emydid Mycoplasma in a single individual (an unmarked, naturally recruited hatchling in 2018). This pathogen was first documented in eastern box turtles (Terrapene carolina carolina; Feldman et al., 2006). Since its discovery, it has been documented in many North American chelonians and usually does not cause disease (Ossiboff et al., 2015;Sandmeier et al., 2019b). The pathogen copy numbers in this individual were very low (0.45 copies/ng DNA) and the tortoise showed no evidence of clinical disease. Given the low copy numbers, it is possible the individual was not actually infected by the pathogen and we detected it due to contamination of the sample or non-infected passage of a Mycoplasma from a shared environment with an emydid host (e.g. box turtle). However, no other known tortoise pathogens, including Ranavirus spp., were detected in our study.
Stress has often been cited as a concern with translocation efforts (Teixeira et al., 2007;Dickens et al., 2010), particularly in circumstances that involve the release of formerly captive individuals. The H:L ratios documented in the AGTHP population (1.35 ± 0.81) were similar to values reported in other translocated populations (1.98 ± 0.96; Cozad, 2018) and for wild gopher tortoises in Florida (0.32-2.88; Page-Karjian et al., 2021) and Mississippi (1.15 ± 0.87 and 2.08 ± 1.31, Holbrook, 2015). Although there are few studies for comparison, particularly from wild in situ populations, comparison with reported values suggests that on average, translocated waif tortoises have similar H:L ratios to their wild counterparts.
Despite our efforts to check traps multiple times throughout the day, blood sampling occurred at the end of the day after animals were transported to the laboratory, resulting in a time lag of several hours between capture and blood collection; these aspects in addition to handling of animals could have affected hemogram results (e.g. caused some degree of distress). A recent study that collected blood samples in the field within 3 mins of handling observed an average H:L ratio below 1.0 (Goessling and Mendonça, 2021), with lowest H:L ratios in less distressed groups (e.g. untrapped, hand-captured tortoises). This finding suggests that H:L ratios we observed, although similar to those reported in the literature, may be elevated due to acute capture effects and not reflective of a true physiological baseline. Indeed, only 42% of the calculated H:L ratios fell below 1.0. Because temperature, season, capture techniques, individual variation, sample collection protocols and analytical methods can influence haematological data (Goessling et al., 2016;Rosenberg et al., 2018a;Rosenberg et al., 2018b;Sandmeier et al., 2019a), caution is always warranted when making comparisons among studies. However, quantifying immature heterophils can aid in interpreting H:L ratios and distinguishing between stress (characterized by increased mature heterophils) and inflammation (as indicated by increased immature heterophils and/or presence of toxic change) (Stacy et al., 2022).
Outside of its potential relationships with inflammation, concern for underlying infection, and/or stress, haematological analysis can provide additional insight into the overall health of individuals. Despite the diversity of backgrounds and unique nature of this population, the hemogram data we observed in this study closely aligned with previously reported reference intervals for a wild gopher tortoise population in southeastern Florida (Page-Karjian et al., 2021). The hemogram data of the majority of adult tortoises (>85%) fell within reference intervals for PCV, TP, WBC count, H:L ratio and absolute heterophils, lymphocytes, monocytes and basophils. Interestingly, adult gopher tortoises from southeastern Florida had comparatively higher eosinophils than our waif tortoises, which could suggest antigenic stimulation in the Florida population, as it also documented a higher prevalence of URTD and individuals infected with Anaplasma spp. (Stacy et al., 2011;Page-Karjian et al., 2021); although the latter pathogen was not evaluated by PCR in our study, there was no evidence by blood film evaluation.
The health assessment data we present for waif gopher tortoises released to a wildlife preserve over the course of a decade are comparable with data for their wild counterparts. Despite the population being largely comprised of waif tortoises-individuals that are considered at higher risk due to factors such as time in captivity or unknown origin-we did not detect any unexpected pathogens and the prevalence of URTD-associated pathogens was comparable with the prevalence reported in wild in situ populations. Although the AGTHP population appears healthy overall, it is important to consider that health assessments and pathogen screenings occurred up to 11 years following release. Of the 203 juvenile, subadult and adult waif tortoises released into the population prior to 2017, we recaptured 107 of them (53%; McKee et al., 2021). Certainly, mortality accounts for some portion of the animals we failed to recapture, and we acknowledge that some individuals infected with pathogens may have died prior to our survey. However, long-term survival estimates from this population are comparable with wild in situ populations

Conclusions
Collectively, our results suggest that waif gopher tortoises can play an important role in the recovery and sustainability of isolated populations that are far below the threshold of viability (McKee et al., 2021). As suitable tortoise habitat becomes increasingly fragmented, there may be additional opportunities to introduce waif tortoises to isolated sites without jeopardizing the health of resident or neighbouring populations. Quarantine practices, pathogen screening and visual assessment prior to release, along with post-release monitoring are important measures that can reduce the risks associated with the release of formerly captive individuals. Because the species is declining (Smith et al., 2006), the risk of potential adverse health impacts will likely need to be balanced with the need for increasing the number of populations and the number of reproductive adults in existing populations of this long-lived species, which may require up to 20 years to reach maturity (Diemer, 1986). Moreover, as the demand for turtles and tortoises has resulted in the ongoing illegal collection and confiscation of many other species around the globe (Rosen and Smith, 2010;Bush et al., 2014;Mendiratta et al., 2017), it is important to understand how to best use these formerly captive individuals (particularly reproductive adults) for conservation objectives. As one of the first studies to assess the health of a population augmented with formerly captive chelonians, our findings provide important insight for other species facing similar management dilemmas.

Data Availability
The data underlying this article will be shared on request to the corresponding author.

Author Contributions
R.M., K.B. and T.T. acquired funding for research, conducted fieldwork and visually assessed tortoises. C.M. assisted with statistical analysis and study design. N.S. analysed blood films and provided expertise in interpreting haematological results. M.A. performed pathogen screening. R.M. drafted the original manuscript. All authors revised, contributed to and approved the final manuscript.