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.

Background

Rodents are important hosts of zoonotic pathogens, including plague which is caused by the bacteria Yersinia pestis [1]. In Madagascar, where 81% of global plague cases occurred between 2013 and 2018 [2], most cases occur in the central highlands [3]. The main epidemiological cycle involves black rats (Rattus rattus) and two flea species, Xenopsylla cheopis, which parasitizes rats inside houses, and Synopsyllus fonquerniei, which is prevalent on outside rats and the distribution of which closely matches that of human plague cases, highlighting its importance [4]. Plague mitigation strategies largely focus on preventing transmission to humans and, therefore, the human–flea interface, whilst the risk posed by infected fleas leaving dead rodents has led to mixed advice on the value of rodent control [5]. Nevertheless, since reducing rodent abundance can lead to vector population reductions, effective rodent management could be an important strategy in tackling plague risk [3].

On the face of it, localized control seems a sensible strategy for reducing rodent abundance in and around human habitation; however, rats are well known for being neophobic (avoiding new things in their environment), and capable of compensatory reproduction and immigration which can counteract the effects of control. Indeed, in a previous study of intermittent control within houses we showed that reduction in abundance was only very short-term [3]. Studies elsewhere have found similar results [6]. Furthermore, although some studies have shown that intensive community-led trapping can be effective at reducing rodent populations in rural African households, most previous research has focussed on food security benefits [7, 8], whereas others have shown that spatially limited control can potentially exacerbate disease transmission within reservoirs due to increased infected host movement towards areas populated by susceptible hosts [9, 10] and possibly by changes in population demographics and disease susceptibility [11]. Therefore, the complex and context-specific epidemiology of rodent-borne diseases necessitates careful evaluation of control measures targeting rodents and/or vectors [5].

In rural Madagascar, black rats are prevalent both inside houses and outside, in peri-domestic areas and in agricultural fields. Studies have shown that there is migration of black rats between these habitats. We are therefore faced with the challenge that removal efforts inside houses may be compensated for through immigration. Furthermore, there are major concerns that increased immigration may increase exposure to the outside flea S. fonquerniei, potentially increasing plague exposure risk to people. We addressed these knowledge gaps by assessing the impact of intensive trapping inside households on key indices of plague risk: rodent abundance, rodent flea infestation (i.e. proportion of rodents that carry a flea), and flea abundance.

Main text

We conducted a rodent management experiment in six villages in Analavory Commune-Miarinarivo District-Itasy Region, Madagascar, during 2022–2023 (Additional file 1: Figure S1, Table 1). In three (intervention) villages, communities themselves conducted intensive daily rodent trapping inside all houses. Two traps were distributed per household, with snap-traps (Romax Snap-R, 14 L × 7.5 W × 6.5 H cm) used during the non-plague season (May–August 2022) and live-capture wire mesh traps (BTS Company 30 L × 10 W × 10 H cm) used during the plague season (September 2022–April 2023). Community agents recorded the number of rodents captured per day. Live-caught rodents were euthanized by cervical dislocation. In three (control) villages, householders could continue any rodent management activities they usually used. Monitoring of rodent and flea abundance was conducted at 4-month intervals (March 2022–March 2023) within 16 houses per village in four sampling areas, using tracking tiles and trapping inside houses. Tracking tiles recorded activity (rodent footprints, scratches and tail swipes) using ceramic tiles (20 × 20 cm) painted with a mixture of 32% blue chalk powder, 64% white spirit, and 4% motor oil [3]. Three tracking tiles were distributed in each house and checked daily, with tiles set for two nights in July and November 2022 and one night in March 2023. After tiles were removed, one live-capture wire mesh trap (30 L × 10 W × 10 H cm) and one Sherman trap (H.B. Sherman Traps Inc., 23 L × 7.5 W × 9 H cm) were set in each house for four nights (reduced to two nights in March 2023) and checked daily. Captured animals were euthanized by cervical dislocation and identified to species. Animals were brushed to remove fleas which were stored in 95% alcohol and later identification was performed using a binocular microscope (Leica, Wetzlar, Germany) and available morphological keys [12]. Blood samples were collected through cardiac puncture. Whole blood was centrifuged to separate the serum and kept at 4 °C in the field and − 20 °C on return to laboratory. Detection of anti-F1 IgG antibodies was performed using an enzyme-linked immunosorbent assay (ELISA), with samples tested in duplicate and a mean optical density of 0.15 used as a threshold for IgG detection [13]. In each plate a negative and a positive rodent sample were included as controls.

Table 1 Intervention status, number of households and coordinates of the villages selected for this study

Generalized Linear Mixed Models were used to test for differences between control and intervention villages in: (1) rodent (R. rattus and Mus musculus) relative abundance from capture and tracking tile data; (2) probability of being infested by a flea (for each flea species separately); and (3) relative abundance of each flea species. Flea analyses focussed on data from R. rattus as M. musculus carried few fleas. We checked for differences between intervention and control villages prior to the experiment, using data from March 2022 for rodents and X. cheopis and, due to seasonality of S. fonquerniei [4], data from pilot sampling conducted in August–October 2021 (peak abundance period for this flea). Intervention analyses examined the effectiveness of treatment using monitoring data from July 2022–March 2023. Rodent and flea abundance can vary seasonally, whilst the effectiveness of management may accumulate over time or differ seasonally due to changes in rodent reproduction or movement [1, 14]. Therefore, intervention period analyses evaluated month and treatment, either individually, additively or with an interaction. For count analyses, the response was captures per household with an offset of sampling effort (calculated as the number of traps containing rodents or not sprung plus half the number of traps which were sprung or had bait removed but which had not caught a rodent) [15]. We used Akaike’s information criterion (AIC) to compare Poisson, negative binomial, zero-inflated Poisson and zero-inflated negative binomial models, selecting the distribution with the lowest AIC. For analyses of tracking tile and flea infestation data, presence or absence per tile or per rat, respectively, was modelled with a binomial distribution. To capture spatial variation not related to treatment, models included random effects of house (for intervention analyses), nested in sampling area, nested in village. We present treatment effects from the model with the lowest AIC that included treatment. Models were run using the glmmTMB package in R software (version 4.2.0; R Foundation for Statistical Computing, Vienna, Austria, https://cran.r-project.org/).

Preliminary analysis confirmed that, prior to our treatment, intervention and control villages did not differ in rodent or flea abundances (Table 2, Fig. 1). Between end April 2022 and end February 2023, 2013 R. rattus and 1297 M. musculus were removed by daily trapping from the three intervention villages, with daily captures declining rapidly and then remaining low (Fig. 2). Evidence from monitoring indicated that treatment effectively reduced R. rattus abundance inside houses, but not M. musculus (Fig. 3A, Table 2). There was evidence that R. rattus were less likely to be infested by X. cheopis in intervention villages, and this tendency combined with the large decline in the numbers of R. rattus led to a considerable decline in the relative abundance of X. cheopis in intervention villages (Fig. 3B). Importantly, we found no evidence that inside house rats in intervention villages were more likely to be infested with S. fonquerniei, the flea typically found on outside rats [4], nor that its relative abundance was greater inside houses in intervention villages (Fig. 3B). Moreover, the total flea index, a commonly used measure of risk in plague studies [3], was reduced by 46% in intervention villages compared to control villages (Table 2). Our analyses also highlighted seasonal variation in R. rattus and M. musculus abundance. We found little evidence of interactions between treatment and month except for tracking tile data for R. rattus, where there was evidence that the inside house rat population in intervention villages partially recovered in March 2023 (Tables 3 and 4). The 133 R. rattus with serum samples were all seronegative.

Table 2 Rodent captures, on-rodent flea abundance, rodent flea infestation and flea index at intervention and control villages before and during intervention

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

Table 3 Final competitive model(s) for rodent relative abundance inside houses during the intervention period (July 2022-March 2023)

Table 4 Final competitive model(s) for the probability of being infested by a flea species and relative abundance of rodent fleas inside houses during the intervention period (July 2022-March 2023)

Conclusions

Our results show that intensive trapping inside houses can reduce rodent populations and maintain low numbers, and significantly reduce the abundance of a key flea vector of plague (X. cheopis). These results highlight that communities working together can impact the risk from rodent-borne diseases, despite the high reproduction potential of rodents. Furthermore, in the context of plague epidemiology in Madagascar, we found no evidence that removing rats from inside houses had an effect on the abundance of S. fonquerniei inside houses. Thus, there is no evidence of counterproductive effects of intensive trapping on vector movement and disease transmission, such as was found for Lassa virus infection following intensive rodent trapping in Guinea, West Africa [11]. This is of major concern for health authorities in Madagascar, where the human plague season takes place from September to April. The start of this period appears to coincide with high abundances of S. fonquerniei on R. rattus outside houses, with previous studies indicating high abundances from September to January, peaking in October [16]. There is therefore apprehension about movement of R. rattus carrying S. fonquerniei into houses from surrounding areas [3]. Our findings therefore have important implications for plague mitigation strategies in Madagascar.

In terms of R. rattus abundance, the effects of community-led intensive trapping were rapid rather than cumulative. However, the slight recovery of house rat populations in March may reflect immigration following the peak reproduction period for rats outside [6]. This further emphasizes the need for sustained control in order to overcome the compensatory responses of rodent populations. We therefore recognize that the duration of interventions must be extended to keep rodent abundance low and prevent their increase as has occurred in short-term interventions [3, 6]. We believe the lack of an effect on M. musculus populations is related to trap type, with the trigger mechanism being less sensitive to smaller-sized animals; additional studies are assessing whether smaller snap-traps and live-capture traps with finer wire mesh may be more effective at removing these individuals. Finally, in follow-up surveys, participating communities expressed their commitment to continuing community-led intensive rodent trapping, whilst local authorities indicated a desire for the strategy to be expanded to all villages within the commune. Maintaining community engagement and motivation will be crucial in determining the long-term sustainability of this approach, whilst it will also be important to evaluate how social and economic aspects influence its applicability to other contexts.