Season of death, pathogen persistence and wildlife behaviour alter number of anthrax secondary infections from environmental reservoirs

Abstract

An important part of infectious disease management is predicting factors that influence disease outbreaks, such as R, the number of secondary infections arising from an infected individual. Estimating R is particularly challenging for environmentally transmitted pathogens given time lags between cases and subsequent infections. Here, we calculated R for Bacillus anthracis infections arising from anthrax carcass sites in Etosha National Park, Namibia. Combining host behavioural data, pathogen concentrations and simulation models, we show that R is spatially and temporally variable, driven by spore concentrations at death, host visitation rates and early preference for foraging at infectious sites. While spores were detected up to a decade after death, most secondary infections occurred within 2 years. Transmission simulations under scenarios combining site infectiousness and host exposure risk under different environmental conditions led to dramatically different outbreak dynamics, from pathogen extinction (R < 1) to explosive outbreaks (R > 10). These transmission heterogeneities may explain variation in anthrax outbreak dynamics observed globally, and more generally, the critical importance of environmental variation underlying host-pathogen interactions. Notably, our approach allowed us to estimate the lethal dose of a highly virulent pathogen non-invasively from observational studies and epidemiological data, useful when experiments on wildlife are undesirable or impractical.

Keywords: Bacillus anthracis; disease transmission; environmentally transmitted pathogen; host–pathogen contact; reproduction number.

Figure 1.

Model visualization for the estimation of the reproduction number R of anthrax in Etosha National Park, Namibia. Each simulation ran over 10 years and was repeated 100 times. Each year is separated into three seasons (k), the hot wet, cool dry and hot dry seasons. Each probability density function (PDF) is obtained using the camera trap data. During the hot wet season, the proportion of animals ingesting grass and soil ω is ω_wet, while it is ω_dry during the cool and hot dry seasons. The yearly reproduction number R_t is the sum of the number of infections occurring during each of the three seasons k of year t. The reproduction number R is the sum of the yearly reproduction number.

Figure 2.

Total number of individuals by age and sex recorded visiting (lighter colours) and grazing (darker colours) at anthrax carcass sites (a,b), and the time spent grazing (in seconds) by season, at control (blue) and carcass (red) sites (c,d), for plains zebra (Equus quagga: a,c) and blue wildebeest (Connochaetes taurinus: b,d) in Etosha National Park, Namibia.

Figure 3.

Decline in Bacillus anthracis spore concentrations at 40 anthrax carcass sites in Etosha National Park, Namibia, over time, on grass tops (green) and in surface soils (brown). Grass concentrations were obtained from extrapolation of the soil concentrations (equation 2.1).

Figure 4.

Variation in the reproduction number, R, for plains zebras (Equus quagga) at anthrax carcass sites, for a lethal dose threshold of 10⁷ Bacillus anthracis spores based on differences in seasonality of host visitation, season of reservoir formation, and percentage of individuals ingesting soil during grazing, estimated over the 10-year lifetime of a B. anthracis reservoir site. The percentage of animals ingesting soil while grazing is varied from 0% to 100%. In all cases, animals are ingesting 10% soil, 90% grass (by weight) if soil is consumed. Soil contact during foraging is considered for two seasons, wet season (numbers below the diagonal) and dry season (numbers above the diagonal). The R values falling along the diagonal represent no difference in soil exposure by season. The black line represents the most likely parameter space for the study system, according to host foraging literature [29,30]. The grey tiles represent an R < 1, then a colour gradient is used for R > 1. (a) reading key; (b) all host and reservoir data; (c) visitation rates under good rainfall, wet season reservoirs; (d) visitation rates under good rainfall, dry season reservoirs; (e) visitation rates under simulated drought, wet season reservoirs; (f) visitation rates under simulated drought, dry season reservoirs.

Figure 5.

Prediction of the number of anthrax cases recorded over 100 years depending on lethal dose thresholds of 10⁵, 10⁶, 10⁷ and 10⁸. The shaded area represents the standard deviation over the 100 simulations. The black dashed line represents the recorded mortality in Etosha National Park, Namibia from 1974 to 2020. Each simulation started with the recorded mortality data from a decade (2003–2013). The x-axis represents the total number of potentially infectious carcass sites present in the environment (each site ‘disappears’ after 10 years). Predictions were made using the simulation model output, with a limit of 10 000 new cases. The lethal doses 10⁵, 10⁶ and 10⁷ lead to an explosion of cases, while the lethal dose of 10⁸ leads to extinction. Lethal doses maintaining a relatively stable number of cases over time would be between 10⁷ and 10⁸. The y-axis is log-transformed.

Figure 6.

Variation in the average number of plains zebra (Equus quagga) secondary anthrax mortalities occurring per year after death at an anthrax carcass site, depending on the lethal dose threshold of Bacillus anthracis spores ingested. Opaque colours and dotted error bars represent the average number of infections within the assumed parameter space of our study system, i.e. the percentage of animals ingesting soil while grazing varied from 0% to 40%. The transparent bars and solid error bars represent the average number of secondary infections for the entire parameter space, i.e. the percentage of animals ingesting soil while grazing varied from 0% to 100%. The y-axis is square root transformed to better visualize small numbers. Error bars represent standard deviation.