Genetic manipulations of insect populations for pest control have been advocated for some time, but there are few cases where manipulated individuals have been released in the field and no cases where they have successfully invaded target populations. Population transformation using the intracellular bacterium Wolbachia is particularly attractive because this maternally-inherited agent provides a powerful mechanism to invade natural populations through cytoplasmic incompatibility. When Wolbachia are introduced into mosquitoes, they interfere with pathogen transmission and influence key life history traits such as lifespan. Here we describe how the wMel Wolbachia infection, introduced into the dengue vector Aedes aegypti from Drosophila melanogaster, successfully invaded two natural A. aegypti populations in Australia, reaching near-fixation in a few months following releases of wMel-infected A. aegypti adults. Models with plausible parameter values indicate that Wolbachia-infected mosquitoes suffered relatively small fitness costs, leading to an unstable equilibrium frequency <30% that must be exceeded for invasion. These findings demonstrate that Wolbachia-based strategies can be deployed as a practical approach to dengue suppression with potential for area-wide implementation.

The wMel infection of Drosophila melanogaster was successfully transferred into Aedes aegypti mosquitoes where it has the potential to suppress dengue and other arboviruses. The infection was subsequently spread into two natural populations at Yorkeys Knob and Gordonvale near Cairns, Queensland in 2011. Here we report on the stability of the infection following introduction and we characterize factors influencing the ongoing dynamics of the infection in these two populations. While the Wolbachia infection always remained high and near fixation in both locations, there was a persistent low frequency of uninfected mosquitoes. These uninfected mosquitoes showed weak spatial structure at both release sites although there was some clustering around two areas in Gordonvale. Infected females from both locations showed perfect maternal transmission consistent with patterns previously established pre-release in laboratory tests. After >2 years under field conditions, the infection continued to show complete cytoplasmic incompatibility across multiple gonotrophic cycles but persistent deleterious fitness effects, suggesting that host effects were stable over time. These results point to the stability of Wolbachia infections and their impact on hosts following local invasion, and also highlight the continued persistence of uninfected individuals at a low frequency most likely due to immigration.

Dengue-suppressing Wolbachia strains are promising tools for arbovirus control, particularly as they have the potential to self-spread following local introductions. To test this, we followed the frequency of the transinfected Wolbachia strain wMel through Ae. aegypti in Cairns, Australia, following releases at 3 nonisolated locations within the city in early 2013. Spatial spread was analysed graphically using interpolation and by fitting a statistical model describing the position and width of the wave. For the larger 2 of the 3 releases (covering 0.97 km2 and 0.52 km2), we observed slow but steady spatial spread, at about 100–200 m per year, roughly consistent with theoretical predictions. In contrast, the smallest release (0.11 km2) produced erratic temporal and spatial dynamics, with little evidence of spread after 2 years. This is consistent with the prediction concerning fitness-decreasing Wolbachia transinfections that a minimum release area is needed to achieve stable local establishment and spread in continuous habitats. Our graphical and likelihood analyses produced broadly consistent estimates of wave speed and wave width. Spread at all sites was spatially heterogeneous, suggesting that environmental heterogeneity will affect large-scale Wolbachia transformations of urban mosquito populations. The persistence and spread of Wolbachia in release areas meeting minimum area requirements indicates the promise of successful large-scale population transformation.

BackgroundIntroduced Wolbachia bacteria can influence the susceptibility of Aedes aegypti mosquitoes to arboviral infections as well as having detrimental effects on host fitness. Previous field trials demonstrated that the wMel strain of Wolbachia effectively and durably invades Ae. aegypti populations. Here we report on trials of a second strain, wMelPop-PGYP Wolbachia, in field sites in northern Australia (Machans Beach and Babinda) and central Vietnam (Tri Nguyen, Hon Mieu Island), each with contrasting natural Ae. aegypti densities.MethodsMosquitoes were released at the adult or pupal stages for different lengths of time at the sites depending on changes in Wolbachia frequency as assessed through PCR assays of material collected through Biogents-Sentinel (BG-S) traps and ovitraps. Adult numbers were also monitored through BG-S traps. Changes in Wolbachia frequency were compared across hamlets or house blocks.ResultsReleases of adult wMelPop-Ae. aegypti resulted in the transient invasion of wMelPop in all three field sites. Invasion at the Australian sites was heterogeneous, reflecting a slower rate of invasion in locations where background mosquito numbers were high. In contrast, invasion across Tri Nguyen was relatively uniform. After cessation of releases, the frequency of wMelPop declined in all sites, most rapidly in Babinda and Tri Nguyen. Within Machans Beach the rate of decrease varied among areas, and wMelPop was detected for several months in an area with a relatively low mosquito density.ConclusionsThese findings highlight challenges associated with releasing Wolbachia-Ae. aegypti combinations with low fitness, albeit strong virus interference properties, as a means of sustainable control of dengue virus transmission.Electronic supplementary materialThe online version of this article (doi:10.1186/s13071-015-1174-x) contains supplementary material, which is available to authorized users.

BackgroundDengue infection spread in naive populations occurs in an explosive and widespread fashion primarily due to the absence of population herd immunity, the population dynamics and dispersal of Ae. aegypti, and the movement of individuals within the urban space. Knowledge on the relative contribution of such factors to the spatial dimension of dengue virus spread has been limited. In the present study we analyzed the spatio-temporal pattern of a large dengue virus-2 (DENV-2) outbreak that affected the Australian city of Cairns (north Queensland) in 2003, quantified the relationship between dengue transmission and distance to the epidemic's index case (IC), evaluated the effects of indoor residual spraying (IRS) on the odds of dengue infection, and generated recommendations for city-wide dengue surveillance and control.Methods and FindingsWe retrospectively analyzed data from 383 DENV-2 confirmed cases and 1,163 IRS applications performed during the 25-week epidemic period. Spatial (local k-function, angular wavelets) and space-time (Knox test) analyses quantified the intensity and directionality of clustering of dengue cases, whereas a semi-parametric Bayesian space-time regression assessed the impact of IRS and spatial autocorrelation in the odds of weekly dengue infection. About 63% of the cases clustered up to 800 m around the IC's house. Most cases were distributed in the NW-SE axis as a consequence of the spatial arrangement of blocks within the city and, possibly, the prevailing winds. Space-time analysis showed that DENV-2 infection spread rapidly, generating 18 clusters (comprising 65% of all cases), and that these clusters varied in extent as a function of their distance to the IC's residence. IRS applications had a significant protective effect in the further occurrence of dengue cases, but only when they reached coverage of 60% or more of the neighboring premises of a house.ConclusionBy applying sound statistical analysis to a very detailed dataset from one of the largest outbreaks that affected the city of Cairns in recent times, we not only described the spread of dengue virus with high detail but also quantified the spatio-temporal dimension of dengue virus transmission within this complex urban environment. In areas susceptible to non-periodic dengue epidemics, effective disease prevention and control would depend on the prompt response to introduced cases. We foresee that some of the results and recommendations derived from our study may also be applicable to other areas currently affected or potentially subject to dengue epidemics.