Community-led intensive trapping reduces abundance of key plague reservoir and flea vector

Abstract

Background: Zoonotic pathogens transmitted by rodents are highly prevalent in low-middle income countries and effective control measures that are easily implemented are urgently needed. Whilst rodent control seems sensible as a mitigation strategy, there is a risk that disease prevalence in reservoir populations can increase following control due to impacts on movement and demographics. Additionally, removing rodents from the population does not necessarily lead to reductions in abundance as populations can compensate for removal through increased breeding and immigration. In a previous study of intermittent control within houses, we showed that reduction in rodent abundance was only very short-term. Working in rural settings within the plague-endemic area of Madagascar, this study explores whether community-led daily intensive rodent trapping within houses can effectively reduce long-term rodent and flea abundance.

Main text: A rodent management experiment was carried out in six rural villages of Madagascar during 2022-2023. Three villages were selected as intervention villages, where intensive daily rodent trapping inside houses was conducted. Surveillance of rodent and flea abundance using traps and tiles took place at 4-month intervals. We show that community-led intensive rodent trapping in rural Malagasy households effectively reduced abundance of the main rodent reservoir (Rattus rattus) and indoor flea vector (Xenopsylla cheopis) of plague. Importantly, indoor abundance of the outside flea vector (Synopsyllus fonquerniei) did not increase.

Conclusions: Community-based intensive rodent trapping inside houses is an effective methodology in controlling key reservoirs and vectors of plague, which can be implemented by the communities themselves. Co-ordinated and sustained rodent control should be considered as an important plague mitigation strategy.

Keywords: Madagascar; Plague; Rodent control; Rodent-borne diseases; Rural population; Siphonaptera; Zoonosis.

Fig. 1

A Effect of treatment on the relative abundance of R. rattus and M. musculus inside houses during the pre-intervention period (March 2022). Data were collected using trapping (captures). There was no data collected from tracking tiles during the pre-treatment period. Models for R. rattus and M. musculus used a negative binomial and Poisson distribution, respectively. B Effect of treatment on the probability of being infested by and relative abundance of X. cheopis and S. fonquerniei on R. rattus inside houses during the pre-intervention period (March 2022 for X. cheopis and August or October 2021 for S. fonquerniei). Models for the probability of being infested by a flea species used a binomial distribution, and models for relative abundance used a negative binomial distribution for both flea species. The forest plots illustrate the odd ratios (circles) and 85% confidence intervals (CI confidence interval; whiskers) for the effect of treatment. Confidence intervals overlap 1 (vertical dotted line) indicating no difference between intervention and non-intervention villages during the pre-intervention period. For all analyses, the model with the lowest AIC was the intercept only model

Fig. 2

Results from the daily trapping in each of the three intervention villages between May 2022 (i.e. the start of daily trapping) and February 2023. Plots show weekly averages for the number of (A) rats and (B) mice caught by daily trapping, standardised to reflect numbers per 20 houses. Live-capture traps were used from September 2022 to February 2023 (i.e. during the plague season)

Fig. 3

A Effect of treatment on the relative abundance of R. rattus and M. musculus inside houses during the intervention period (July 2022-March 2023). Data were collected using trapping (captures) and tracking tiles. Effects based on the model with the lowest AIC that included treatment. Models for capture data of R. rattus and M. musculus used a Poisson and negative binomial distribution, respectively. Models for tracking tile data used a binomial distribution for both rodent species. B Effect of treatment on the probability of being infested by and relative abundance of X. cheopis and S. fonquerniei on R. rattus inside houses during the intervention period. Effects based on the model with the lowest AIC that included treatment. Models for the probability of being infested by a flea species used a binomial distribution for both flea species. Models for relative abundance of X. cheopis and S. fonquerniei used a negative binomial and a Poisson distribution, respectively. Forest plot illustrates the odd ratios (circles) and 85% confidence intervals (CI confidence interval; whiskers). Effects with confidence intervals overlapping 1 (vertical dotted line) indicate no significant treatment effect (treatment was a non-informative parameter), whilst those with confidence intervals not overlapping 1 indicate a significant effect of treatment