Long-term carriage of Mycoplasma ovipneumoniae in bighorn sheep and implications for test and remove protocols

Data information: Each data fils is associated with a model in the manuscript "Long-term carriage of Mycoplasma ovipneumoniae in bighorn sheep and implications for test and remove protocols" and is explained in detail below. Movi.csv contains the data used in the model evaulating the seasonal effects on testing positive for Mycoplasma ovipneumoniae (M. ovi). Movi_PrevMovi.csv contains the data used in evauating the effects of previous M. ovi status, season, and age on the probability of testing positive for M. ovi. ChronicCarrier.csv contains the data used in evaulating effects of M. ovi status at a single point in time based on the probability of being a chronic carrier. Scenarios.csv contains the data used in hypothetica scenarios of test and remove. "Repeated" indicates that the animal would have been removed at that capture event under a repeated positive removal criteria and "Single" indicates that the animal would have been removed under a single positive removal criteria. For each dataset "Movi" indicates if an animal tested positive for M. ovi, "AID" indicates the unique animal identifier, "Herd" indicats the population the animal belongs to, "Season" indicates when the test was taken (Spring tests were taken in March and Winter tests were taken in December), "age" indicates the age of the animal when the test was taken, and "CC" indicates if an animal was classifed as a chronic carrier. Vairables that start with "prev" indicate the value of that variable at the previous capture event. Methods: Study areas We used data from four populations of bighorn sheep in the Greater Yellowstone Ecosystem including the Jackson population in the Gros Ventre Range (43°38′N, 110°29′W), the Upper Shoshone population in the Absaroka Range (44°21′N, 109°49′W), the Whiskey Mountain population in the eastern Wind River Range (43°18′N, 109°37′W), and the West Side population in the western Wind River Range (43°13′N, 109°49′W). The Whiskey Mountain and West Side population were migratory (Whiskey Mountain) and high elevation resident (West Side) segments of the same metapopulation, whereas the other populations were spatially and genetically distinct (Love Stowell et al. 2020). The Jackson, Whiskey Mountain, and Upper Shoshone populations were elevational migrants (Lowrey et al. 2020). All four populations had histories of pneumonia epizootics and were infected with the same suite of bacterial pathogens in the Mycoplasmataceae and Pasturellaceae families including M. ovi (Butler et al. 2018). During our study there was some pneumonia-related mortality in juveniles in all four populations (Wagler et al. 2023, T. Mong, personal observation). Although there was occasional mortality of adults caused by pneumonia, none of the populations experienced all-age pneumonia die-offs during this study. All four populations were native and have never been reintroduced or augmented with transplants. Animal capture and handling We captured adult female bighorn sheep in the Jackson, Upper Shoshone, and Whiskey Mountain populations beginning in 2015 and beginning in 2021 for the West Side population using helicopter net gunning (Jackson, Whiskey Mountain, and West Side; Wagler et al. 2022) or ground darting with chemical immobilization (Upper Shoshone; Kock et al. 1987). Each March and December following the initial capture event, we recaptured as many collared individuals as possible and captured new animals to maintain sample sizes in each population until spring 2023. Recapture rates were generally high, although some recaptures were not possible because of weather conditions or animals being in dangerous terrain or Wilderness areas. We had permission to capture animals in the Bridger Wilderness (West Side) and the Glacier Addition of the Fitzpatrick Wilderness (Whiskey Mountain); however, we missed recapture of some animals in the Whiskey Mountain and Jackson populations because of restrictions in those Wilderness areas. December captures occurred during the end of the mating season and when most of the migrants were on low-elevation winter ranges. March captures occurred in late winter when animals were in poorer nutritional condition (Smiley et al. 2022) and before spring migration and parturition (Argov et al. 2024). Upon capture, we assigned a unique identifier (Animal ID) to each animal fitted animals with a GPS collar (Advanced Telemetry Systems, Isanti, MN, USA; VECTRONIC Aerospace Gmbh, Berlin, Germany). We assessed nutritional condition using body palpation to estimate a body condition score and measured rump fat with ultrasonography (5-MHz transducer; Ibex Pro, E.I. Medical Imaging, Loveland, CO, USA; (Stephenson et al. 2020). We calculated percent ingesta-free body fat (hereafter body fat) for all animals using an equation developed for bighorn sheep (Stephenson et al. 2020). We estimated the age of animals based on incisor replacement, tooth wear, and horn annuli (Valdez and Krausman 1999). We cleaned capture equipment between animals with a solution of Virkon S Disinfectant (Lanxess, Cologne, Germany) or Re-Juv-Nal (Hillyard, Denver, Colorado, USA) to prevent pathogen transmission. All animal and handling was approved by Institutional Animal Care and Use Committee (20150316KM00148, 20180305KM00296, 20200305KM00412-03), the Wyoming Game and Fish Department (Chapter 33-1278), and were in accordance with the guidelines of the American Society of Mammalogists (Sikes 2016). We collected one nasal swab at each capture event to test for the presence of M. ovi by culture and PCR. We inserted nasal swabs deep into each of the nares and rotated them against the mucosa prior to removal. We stored polyester swabs (Puritan Medical Products, Pittsfield, Maine, USA) in tryptic soy broth (TSB; Hardy Diagnostics, Santa Maria, California, USA) with 15% glycerol (Butler et al. 2017); Alfa Aesar - Thermo Fisher Scientific, Ward Hill, Massachusetts, USA) and frozen on dry ice for transport. We placed swabs in enrichment broth (modified TSB-1 Jennings-Gaines et al. 2016) and culture (Wood et al. 2017) and PCR (Manlove et al. 2019) were performed at the Wyoming Game and Fish Wildlife Health Laboratory (Laramie, Wyoming, USA). For PCR, DNA was extracted from enrichment broth using the DNeasy Blood and Tissue Kit (Qiagen, Germantown, Maryland, USA). We classified the chronic carriage status of each animal for each time it was captured (Appendix S1: Figures , S1, S2, S3, and S4). Garwood et al. (2020) defined chronic carriers as animals that consistently tested positive for M. ovi after being captured at least twice over a 20-month sampling period. Similarly, we defined chronic carriers as animals that tested positive two times within 14 months (a three-capture sequence) with captures occurring in December and March (Figure 1). We allowed for one negative test within the three-capture sequence in the definition of chronic carriers because detection of M. ovi is imperfect (Butler et al. 2017). Because of the long duration of our study, we allowed for changes in individual carriage status within the duration of our study (i.e. carriage status of an animal could change to or from the chronic carrier state based on the result of each three test window). For example, Animal ID 22 tested negative for M. ovi each time she was captured from March 2015 to December 2021 and was thus classified as uninfected during those capture events (Figure 1). She tested positive in March and December 2022, which was when her carriage status changed to chronic carrier. Animals classified as chronic carriers could subsequently be classified as uninfected if they had two negative tests within 14 months. At least two tests within a three-capture sequence were required to be classified. In instances where the first two subsequent tests had opposite results (i.e. one positive and one negative), a third test was required for classification. Statistical analysis We first evaluated the effects of the seasonal timing of testing (i.e. March or December) and age on the probability of testing positive for M. ovi and remaining positive for M. ovi using two mixed-effects binomial regression models. With the first model we assessed the probability of testing positive for M. ovi regardless of previous infection status and included M. ovi status as the response and age and timing of test as explanatory variables. We included a quadratic term for age because younger and older animals are more likely to test positive for M. ovi than middle-aged animals (Plowright et al. 2017). With the second model we assessed the probability of testing positive for M. ovi but also included their M. ovi status from the previous capture event as an explanatory variable. We retained age, a quadratic term for age, and the timing of test as predictor variables. We used this second model to inform the likelihood of remaining positive; consequently, we subset data to animals where their previous M. ovi status was known. We included a random intercept for animal ID nested within population. To assess the probability of being a chronic carrier based on M. ovi status at a single point in time, we used a mixed-effects binomial regression model. The response was chronic carrier status (0 or 1), and the explanatory variables were M. ovi status and age. We tested for an interaction between seasonal timing of test and M. ovi status, which allowed us to test for differences in the effect of M. ovi status on the probability of being a chronic carrier between December and March. We did not include the parent term for timing of test in the model because there was no biological reason for the probability of being a chronic carrier to change between months. We included a random intercept for population but not for Animal ID because being a chronic carrier was mostly a repeatable characteristic of the animal (i.e. Animal ID and chronic carrier status were highly correlated) and our goal was to model the probability of being a chronic carrier with M. ovi tests during different times of year. We fit all three regression models using the glmmTMB package (Brooks et al. 2017) in Program R version 4.3.2 (R Core Team 2023). We built a global model consisting of all variables of interest and used Akaike information criterion adjusted for small sample size (AICc) to compare every possible combination of variables from the global model (Burnham and Anderson 2002). We evaluated all models within 2 AICc of the top model and retained the model with the most variables because we were interested in the effects of all predictors, even if they were marginally informative. To understand how different removal criteria influence the number of animal removals, we compared the proportion of animals that qualified for removal from four hypothetical scenarios of test and remove occurring over two consecutive capture events that were one year apart. We subset this analysis to animals that were ≥2 years old because most test and remove efforts are focused on adults. The four scenarios included two different removal criteria (single positive test and repeated positive test) implemented with captures (i.e. testing for M. ovi) occurring in two different seasons (December and March). For each scenario, we subset our data to two consecutive capture events within the respective months (i.e. one year apart) and only included animals that were captured at each capture event. Under the single positive removal criteria, we quantified the proportion of animals that tested positive once out of the two consecutive M. ovi tests. Under the repeated positive removal criteria, we quantified the proportion of animals that had two positive M. ovi tests. We quantified the proportion of animals that qualified for removal under all four hypothetical scenarios for each consecutive two capture window (n = 8 for March, n = 7 for December).