Long-Range Dispersal Behaviour and Spatial Distribution Modelling of Adult Mosquitoes in the Winnipeg Region

Abstract

Mosquitoes are present in virtually every nation worldwide, acting as both a vector for many serious pathogens, and as a nuisance because of their blood-feeding behaviours. As a principle of Integrated Mosquito Management (IMM), pre-emptive rather than reactive mosquito control measures are recommended and have been shown to be effective in suppressing mosquito populations. However, pre-emptive actions require insight into the spatial dynamics of mosquitoes to be effective, as mosquito dispersal behaviour is broadly influenced by local environmental conditions and species physiology. This research project was designed to investigate the dispersal behaviour and landscape ecology of adult mosquitoes in the Winnipeg region in central Canada. In Manitoba, mosquitoes primarily present an annoyance rather than a public health risk, though a risk of exposure to several mosquito-borne diseases persists in southern Manitoba. As part of the Winnipeg’s long-standing IMM program, the city maintains a mosquito control buffer zone extending approximately 10 km beyond the city limits. Within this zone, a surveillance program is implemented for adult and larval mosquitoes and larviciding operations occur when necessary. However, there is a lack of local evidence to justify the size of this zone, and literature addressing the establishment of effective buffer zones for mosquito control is nearly non-existent. Assessment of the potential effectiveness of a buffer zone requires information on mosquito flight ranges and dispersal behaviours, as well as knowledge of their overall distribution. This study 1) demonstrates two approaches that are commonly used to characterize mosquito behaviour for the purpose of optimizing control measures: mark-release-recapture (MRR) experiments and spatial distribution modelling, and 2) uses the results of these approaches to infer the potential effectiveness of the mosquito control buffer zone surrounding Winnipeg. Mark-release-recapture studies have long been used to measure mosquito flight distances and to investigate the environmental drivers of dispersal movement. In this study, field-sourced larval mosquitoes were reared to adulthood and marked with fluorescent dust prior to release. I hypothesized that mosquitoes would actively disperse towards areas with higher moisture profiles, such as those with dense canopy cover or near bodies of water. Additionally, I predicted that females would orientate toward areas with high densities of Mosquitoes are present in virtually every nation worldwide, acting as both a vector for many serious pathogens, and as a nuisance because of their blood-feeding behaviours. As a principle of Integrated Mosquito Management (IMM), pre-emptive rather than reactive mosquito control measures are recommended and have been shown to be effective in suppressing mosquito populations. However, pre-emptive actions require insight into the spatial dynamics of mosquitoes to be effective, as mosquito dispersal behaviour is broadly influenced by local environmental conditions and species physiology. This research project was designed to investigate the dispersal behaviour and landscape ecology of adult mosquitoes in the Winnipeg region in central Canada. In Manitoba, mosquitoes primarily present an annoyance rather than a public health risk, though a risk of exposure to several mosquito-borne diseases persists in southern Manitoba. As part of the Winnipeg’s long-standing IMM program, the city maintains a mosquito control buffer zone extending approximately 10 km beyond the city limits. Within this zone, a surveillance program is implemented for adult and larval mosquitoes and larviciding operations occur when necessary. However, there is a lack of local evidence to justify the size of this zone, and literature addressing the establishment of effective buffer zones for mosquito control is nearly non-existent. Assessment of the potential effectiveness of a buffer zone requires information on mosquito flight ranges and dispersal behaviours, as well as knowledge of their overall distribution. This study 1) demonstrates two approaches that are commonly used to characterize mosquito behaviour for the purpose of optimizing control measures: mark-release-recapture (MRR) experiments and spatial distribution modelling, and 2) uses the results of these approaches to infer the potential effectiveness of the mosquito control buffer zone surrounding Winnipeg. Mark-release-recapture studies have long been used to measure mosquito flight distances and to investigate the environmental drivers of dispersal movement. In this study, field-sourced larval mosquitoes were reared to adulthood and marked with fluorescent dust prior to release. I hypothesized that mosquitoes would actively disperse towards areas with higher moisture profiles, such as those with dense canopy cover or near bodies of water. Additionally, I predicted that females would orientate toward areas with high densities of their preferred hosts. With Winnipeg mosquito buffer in mind, these experiments were designed to a) establish the flight ranges of common mosquito species, and b) discern the influence of landscape-based variables on adult mosquitoes dispersing into the urban areas of Winnipeg from the peri-urban outskirts. Recaptured mosquitoes included primarily Aedes vexans, Culiseta inornata, and Coquillettidia perturbans, all of which were recaptured 3 km or more from a single release site located on the southern edge of the City of Winnipeg within a few days of release. Female Ae. vexans were found to commonly travel more than 3 km following release, and male recaptures were often observed several kilometres from the release site. A few female Ae. vexans were recaptured over 15 km away, but the lifetime flight range for most (90%) is estimated to be over 8 km. Female Cs. inornata movement was in part influenced by the presence of mammalian livestock. While too few Cq. perturbans were recaptured to estimate mean flight distance or total flight range, in two separate events, marked individuals were recaptured over 26 km from the release site. Generally, mosquitoes appeared to be more prevalent in areas with extensive vegetative cover, and findings suggest these areas may act as corridors that facilitate dispersal into urban areas. Remote sensing data and spatial models created with Geographic Information Systems (GIS) platforms are increasingly being used to help clarify landscape-wide patterns of mosquito distribution. The approach presented here used nine consecutive years of trap data from Winnipeg’s mosquito surveillance program to model mosquito distribution in unsampled areas. Landscape-based variables such as distance to nearest river, land cover and land use classes, as well as vegetation and wetness indices were extracted for the study area using Sentinel-2 satellite imagery at a resolution of 10 m. Circular zonal areas (“buffer zones”, though not to be confused with the mosquito control buffer zone surrounding Winnipeg) generated at varying distances around traps were used to characterize habitats in terms of these variables. These were then used as explanatory variables in random forest regressions iterated to identify key predictors of mosquito distribution. The maps produced from the final models identified “hotspots” within the Winnipeg area several common local mosquito species. I hypothesized that mosquito populations would be densest in areas with high moisture profiles, such as vegetated regions near rivers, as was implied by the MRR experiment outcomes. Hotspots for Ae. vexans and Cx. restuans confirmed this hypothesis, as these were clustered within riparian areas closest to rivers. Conversely, population hotspots for Cx. tarsalis were located near or beyond city limits. No consistent trends were identified for Ae. dorsalis. Influence of certain vegetation-based land cover classes such as cultivated land, grass and forest were more important in predicting mosquito distribution at larger scales (500 to 1000 m) in comparison to land use classes such as commercial or industrial areas which influenced mosquito distributions at smaller scales (50 to 100 m). These support the hypothesis that relationships between mosquitoes and their surroundings extend beyond the reach of their olfactory or visual senses and that larger scale landscape factors also influence mosquito movement to a significant extent. Based on these findings, we can infer that many mosquito species are capable of dispersing distances that justify the extent of the current buffer zone surrounding the Winnipeg city limits. The results from the MRR experiments demonstrated that these mosquitoes could disperse several kilometres into the city from outside its limits. However, the MRR experiments were not designed to fully address the directionality of their movements, and mosquitoes may be dispersing towards rural areas as well. The results of both the spatial distribution models and the MRR experiments provided similar insights regarding habitat preferences. Most prominently, the distribution and dispersal patterns of Ae. vexans suggests that mosquitoes may be using vegetated riverbanks as corridors for dispersal. Additionally, the risk of encountering the medically important species Cx. tarsalis increases with greater distances from the city center. Both these findings indicate that the diversity in habitat preference for Winnipeg’s mosquitoes would necessitate thorough monitoring and treatment of larval habitats to prevent most mosquitoes from immigrating into the city. This may be impractical, nay impossible given the nature of the breeding habitats preferred by Ae. vexans (soils with intermittent flooding) and Cx. tarsalis (small and discrete artificial containers) and the difficulty associated with treating these. Regardless, these results illustrate an improved understanding of the landscape ecology of mosquitoes in the Winnipeg region.