David Alvarez-Ponce

Mojave desert tortoises ( Gopherus agassizii ), a threatened species under the US Endangered Species Act, are long‐lived reptiles that experience a chronic respiratory disease. The virulence of primary etiologic agent, Mycoplasma agassizii , remains poorly understood, but it exhibits temporal and geographic variability in causing disease outbreaks in host tortoises. Multiple attempts to culture and characterize the diversity of M. agassizii have had minimal success, even though this opportunistic pathogen chronically persists in nearly every population of Mojave desert tortoises. The current geographic range and the molecular mechanisms of virulence of the type‐strain, PS6 T , are unknown, and the bacterium is thought to have low‐to‐moderate virulence. We designed a quantitative polymerase chain reaction (qPCR) targeting three putative virulence genes annotated on the PS6 T genome as exo‐α‐sialidases, enzymes which facilitate growth in many bacterial pathogens. We tested 140 M. agassizii ‐positive DNA samples collected from 2010 to 2012 across the range of Mojave desert tortoises. We found evidence of multiple‐strain infections within hosts. We also found the prevalence of these sialidase‐encoding genes to be highest in tortoise populations surrounding southern Nevada, the area from which PS6 T was originally isolated. We found a general pattern of loss or reduced presence of sialidase among strains, even within a single host. However, in samples that were positive for any of the putative sialidase genes, one particular gene (528), was positively associated with bacterial loads of M. agassizii and may act as a growth factor for the bacterium. Our results suggest three evolutionary patterns: (1) high levels of variation, possibly due to neutral changes and chronic persistence, (2) a trade‐off between moderate virulence and transmission, and (3) selection against virulence in environmental conditions known to be physiologically stressful to the host. Our approach of quantifying genetic variation via qPCR represents a useful model of studying host–pathogen dynamics.