| dc.description |
Infectious diseases of wildlife are typically studied using data on antibody and pathogen levels. In
order to interpret these data, it is necessary to know the course of antibodies and pathogen levels after
infection. Such data are typically collected using experimental infection studies in which host individuals are
inoculated in the laboratory and sampled over an extended period, but because laboratory conditions are
controlled and much less variable than natural conditions, the immune response and pathogen dynamics may
differ. Here, we compared Morogoro arenavirus infection patterns between naturally and experimentally
infected multimammate mice (Mastomys natalensis). Longitudinal samples were collected during three months
of bi-weekly trapping in Morogoro, Tanzania, and antibody titer and viral RNA presence were determined. The
time of infection was estimated from these data using a recently developed Bayesian approach, which allowed
us to assess whether the natural temporal patterns match the previously observed patterns in the laboratory. A
good match was found for 52% of naturally infected individuals, while most of the mismatches can be
explained by the presence of chronically infected individuals (35%), maternal antibodies (10%), and an
antibody detection limit (25%). These results suggest that while laboratory data are useful for interpreting field
samples, there can still be differences due to conditions that were not tested in the laboratory. |
|