Interacting effects of land use and climate on rodent-borne pathogens in central Kenya

Abstract

Understanding the effects of anthropogenic disturbance on zoonotic disease risk is both a critical conservation objective and a public health priority. Here, we evaluate the effects of multiple forms of anthropogenic disturbance across a precipitation gradient on the abundance of pathogen-infected small mammal hosts in a multi-host, multi-pathogen system in central Kenya. Our results suggest that conversion to cropland and wildlife loss alone drive systematic increases in rodent-borne pathogen prevalence, but that pastoral conversion has no such systematic effects. The effects are most likely explained both by changes in total small mammal abundance, and by changes in relative abundance of a few high-competence species, although changes in vector assemblages may also be involved. Several pathogens responded to interactions between disturbance type and climatic conditions, suggesting the potential for synergistic effects of anthropogenic disturbance and climate change on the distribution of disease risk. Overall, these results indicate that conservation can be an effective tool for reducing abundance of rodent-borne pathogens in some contexts (e.g. wildlife loss alone); however, given the strong variation in effects across disturbance types, pathogen taxa and environmental conditions, the use of conservation as public health interventions will need to be carefully tailored to specific pathogens and human contexts.This article is part of the themed issue 'Conservation, biodiversity and infectious disease: scientific evidence and policy implications'.

Keywords: dilution effect; disease; diversity; land-use change; landscape ecology; susceptible host regulation.

Figure 1.

Four land-use types across Laikipia, Kenya were included in this analysis: (a) conserved sites, (b) experimental exclosure sites (where large mammals were excluded with electrified fences), (c) pastoral land use with heavy grazing of domestic stock and (d) cropland sites with small- or large-scale agricultural crops. Community composition of small mammals varied strongly across land-use types (e). For visual clarity, rare species (less than five individual captures) are not depicted in panel (e).

Figure 2.

Panel (a) shows the presence per site of each small mammal species. Rows are species and columns are sites; the presence of each species at a given site is depicted with a red vertical bar. (Nestedness plot constructed using the ‘nestedtemp’ function in the ‘vegan’ package.) This reveals a strongly nested assemblage structure, with some species (e.g. M. minutoides and G. robustus) present at both higher diversity sites (on left) and lower-diversity sites (on right), and other species present only in a subset of relatively high-diversity sites (e.g. L. striatus). The grey ‘boundary line’ shows expected distribution if species were perfectly nested, such that absences above and to the left of the line, and presences below and to the right of the line are both unexpected with a perfectly nested distribution. If species that were highly generalist across sites also tended to have more pathogens, we would expect that the nested structure would correlate with the proportion of infected individuals for each of the pathogens detected (high prevalence expected at top of the graph); however, we see no evidence of this (b). The five species most characteristic of cropland and pastoral habitats are highlighted in gold, whereas the five species most characteristic of conserved and exclosure habitats are highlighted in green.

Figure 3.

The net effect of three types of disturbance (relative to conserved landscapes) on number of infected small mammals per hectare for each of the five most commonly detected pathogens across all sites (±s.e.). Pathogens are abbreviated as follows: Bo, Borrelia; An, Anaplasma; Ba, Bartonella; Th, Theileria and He, Hepatozoon. Note differences in scale across panels.

Figure 4.

For three of the five pathogens examined (Anaplasma (a), Bartonella (b) and Theileria (c)), there were significant interactions between the effect of land-use type and that of mean annual rainfall on number of infected small mammals per hectare. For Anaplasma and Bartonella, the number of infected animals decreased with rainfall in pastoral (yellow) sites but were uncorrelated or positively correlated with rainfall in conserved (dark green) sites. For Theileria, the number of infected animals increased with rainfall overall, but this effect was stronger in cropland (orange) than conserved sites and was not detected in wildlife exclosure (light green) sites. Shading represents 95% confidence intervals.