Enhancing Chelonian Health Assessments by Determining Agreement in Pathogen Detection between Sampling Sites Using Quantitative PCR

Adenoviruses (Rodríguez et al. 2018; Marschang 2019; Franzen-Klein et al. 2020), herpesviruses (Allender et al. 2009; Sim et al. 2015; Kane et al. 2017; Okoh et al. 2021), and Mycoplasmopsis spp. (Feldman et al. 2006; Jacobson et al. 2014; Adamovicz et al. 2018) are important causes of morbidity and mortality in chelonians. Reliable diagnostic methods are essential to characterize pathogen epidemiology and facilitate diagnosis (McArthur et al. 2002; Okoh et al. 2021). However, optimal anatomical sampling sites for antemortem pathogen detection have not been determined in chelonians.

Eastern box turtles (Terrapene carolina carolina) are small terrestrial chelonians found in the eastern US (Ernst et al. 1994; Dodd 2002). This species is currently listed as vulnerable by the International Union for Conservation of Nature due to habitat destruction and fragmentation, traffic mortality, legal or illegal collection for personal use, nest predation, and disease (Van Dijk 2011). Eastern box turtles are long-lived inhabitants of both aquatic and terrestrial habitats with high site fidelity, which makes them useful sentinels of ecosystem health (Rose and Allender 2011). Understanding the effects of pathogens in this species is therefore important for improving individual, population, and ecosystem health assessment.

Like many other chelonians, eastern box turtles can be infected with herpesviruses (e.g., Terrapene herpesvirus 1), adenoviruses (e.g., Terrapene adenovirus 1), and Mycoplasmopsis sp.; antemortem diagnosis typically relies upon quantitative (q)PCR of swabs of the oral cavity (OS), cloaca (CS), or both (OCS). However, which sampling site(s) maximize detection of box turtle pathogens have not yet been identified. Our study aimed to analyze the agreement between three anatomic sampling sites (OS, CS, and OCS) for detecting three chelonian pathogens via qPCR. We hypothesized that OCS samples would provide the greatest probability of pathogen detection.

Five study sites in Illinois were surveyed for free-ranging eastern box turtles from May through July 2017 and 2018, as previously described (Boers et al. 2017). Separate sterile cotton-tipped plastic-handled swabs (Fisher Scientific, Pittsburgh, Pennsylvania, USA) were collected from the OS, then the CS, then the oral cavity followed by the cloaca (OCS) in 88 turtles. Swabs were placed into separate sterile 2 mL Eppendorf tubes (Fisher Scientific) and stored in a cooler on ice packs in the field (<4 hr). Upon return to the laboratory, samples were frozen at –20 C. All animal activities were approved by the University of Illinois Institutional Animal Care and Use Committee (protocol 18000).

Swab DNA was extracted using a DNeasy Blood and Tissue Kit (QIAGEN Inc., Valencia, California, USA) following the manufacturer's protocol. Concentration and purity of DNA were assessed using a spectrophotometer (Nanodrop, Thermo-Fisher Scientific Inc., Waltham, Massachusetts, USA). We performed qPCR using existing assays for Terrapene herpesvirus 1 (TerHV1), Terrapene adenovirus 1 (TerADV1), and Mycoplasmopsis sp., as previously described (Kane et al. 2017; Franzen-Klein et al. 2020). Sterile deionized water was used as a negative control for all assays. Positive controls included plasmid-derived standard curves consisting of seven 10-fold serial dilutions for each pathogen. Quantities for positive samples were normalized based on DNA concentration, with final quantities reported as pathogen copies per ng of DNA.

Level of agreement for pathogen presence was compared between OS, CS, and OCS using the Cohen kappa. Interpretation of the Cohen kappa values follows McHugh (2012); which tailors recommendations conservatively based on health research goals (Table 1). Differences in DNA concentration and pathogen copy numbers between anatomic sampling sites were tested using paired samples ttests. All statistical analysis was performed using MedCalc software version 20.011 (MedCalc, Ostfeld, Belgium).

Concentration of DNA retrieved from swabs differed significantly between anatomical sampling sites (P<0.0001): the highest concentrations in OCS (mean 43.6 ng/µL, 95% confidence interval (CI), 38.6–48.7 ng/µL), followed by OS (mean 23.7 ng/µL, 95% CI, 18.7–28.7 ng/µL), then CS (mean 16.5 ng/µL, 95% CI, 11.4–21.5 ng/µL).

A total of 38/88 turtles (70/264 samples) tested positive for TerHV1 in at least one anatomic sampling site, 27/88 turtles (44/264 samples) tested positive for TerADV1 in at least one site, and 10/88 turtles (24/264 samples) tested positive for Mycoplasmopsis sp. in at least one site. Normalized copy numbers for positive samples were greatest for TerHV1 (mean 7,166 copies/ng DNA, range 0.053–181,819 copies/ng DNA) followed by TerADV1 (mean 409 copies/ng DNA, range 0.61–5,759 copies/ng DNA) then Mycoplasmopsis sp. (mean 252 copies/ng DNA, range 0.11–1,944 copies/ng DNA). There was no significant difference in normalized copy number between anatomic sampling sites for any pathogen (P>0.05).

For all pathogens the highest number of positive qPCR tests were from OCS samples, followed by OS samples, then CS samples (Table 2). The Cohen kappa results are presented in Table 3. Although OCS samples had the most detections, a small number of TerADV1 and TerHV1 positives were detected using OS but not OCS. Compared to the other positive samples, the normalized copy numbers of these OS positive but OCS negative samples were relatively low (2.78 copies/ng DNA for TerADV1 and 21.33–140.96 copies/ng DNA for TerHV1).

The development and wider availability of sensitive and specific diagnostic tests validated for use in nondomestic species has dramatically improved our understanding of disease dynamics, and our ability to diagnose infections. However, there are still many unknowns regarding appropriate antemortem diagnostic sampling locations for chelonian pathogens. Our study provides guidance on anatomic sample site selection for common box turtle pathogen detection using qPCR.

All pathogens were most frequently detected in OCS samples, indicating that molecular detection is generally maximized by this sampling. This might be partially explained by the significantly higher DNA yield in OCS vs. either OS or CS. However, biological explanations relating to the inclusion of two sampling sites in OCS swabs are also possible. For example, pathogens can be detected in the oral cavity of a turtle if they are shed from the respiratory tract, oral mucosa, nasal cavity, or esophagus (Origgi and Jacobson 2000; Feldman et al. 2006; Okoh et al. 2021). Mycoplasmopsis spp. and herpesviruses are thought to be primarily upper-respiratory tract pathogens in chelonians; therefore, higher rates of detection would be expected from oral cavity swabs than cloacal swabs (Origgi and Jacobson 2000; Okoh et al. 2021). We detected TerHV1 more commonly in OCS (34) and OS (33) vs. CS samples (three). This was also observed for Mycoplasmopsis sp. (10 in OCS, nine in OS, five in CS). Pathogens can be detected in the cloaca if they are shed from the gastrointestinal (GI) tract, urinary tract or reproductive tract, or if respiratory pathogens are swallowed and pass through the GI tract. Herpesviruses and Mycoplasmopsis spp. are not thought to primarily impact the GI, reproductive, or urinary tracts, but might be detected from the cloaca as pass-through DNA (Origgi and Jacobson 2000). Our data appear to support this with limited detection by cloacal swabs alone.

In tortoises, intestinal lesions are common with adenovirus infection (Marschang 2019); we therefore expected a higher cloacal detection of these viruses, compared to TerHV1 and Mycoplasmopsis sp. However, we detected TerADV1 more commonly in OCS (26) and OS (14) vs. CS (one). This might indicate a different tissue tropism for TerADV1 in box turtles or reflect a different history of coevolution compared to the historically more virulent tortoise adenoviruses. It is also possible that pathogen shedding sites and patterns might change during the course of an infection, so the optimal anatomic sampling site might vary based on a turtle's clinical status. Challenge studies could assess this possibility.

Despite pathogen detection occurring most commonly in OCS samples, we detected TerHV1 in four OS samples and TerADV1 in one OS sample without corresponding OCS positives. Several body fluids, such as urine, serum, feces, and saliva, contain PCR inhibitors (Ochert et al. 1984). Such PCR inhibitors within the combined OCS sample might have contributed to false negative OCS results. We did not incorporate internal qPCR positive controls to test for PCR inhibitors; such controls could be considered in future studies to rule out this possibility. Alternatively, the order of sample collection might have played a role for at least a few of the low-positive TerHV1 and TerADV1 samples. We collected OS and CS first, followed by OCS. If a small amount of pathogen DNA was present, it might have been mostly removed by the separate OS or CS before the OCS was collected.

The major limitation of our study is the lack of a gold-standard test for determining true pathogen status in each turtle. Because viral and bacterial culture methods for these pathogens have historically been unsuccessful, molecular detection remains the optimal antemortem diagnostic modality. The ability to compare our findings to a gold-standard diagnostic test would enable a clearer assessment of the sensitivity of qPCR at each anatomic sampling site; this is an important avenue of future research.

Despite these limitations, we recommend use of combined oral/cloacal swabs to detect TerADV1, TerHV1, and Mycoplasmopsis sp. in box turtles for diagnosis and disease epidemiology studies. Future directions are to test for pathogen DNA in other anatomic sites such as blood, nasal flushes, and postmortem tissues, and to broaden the pathogen scope for detection of diseases in box turtles and other species.

The authors would like to thank all members of the Illinois College of Veterinary Medicine turtle surveillance team members for assistance with sample collection and processing. We thank John Rucker for the use of his dogs and his assistance with turtle search efforts. We thank Bayer for providing heartworm and flea and tick preventatives for the turtle dogs.