Releasing incompatible males drives strong suppression across populations of wild and Wolbachia-carrying Aedes aegypti in Australia

Abstract

Releasing sterile or incompatible male insects is a proven method of population management in agricultural systems with the potential to revolutionize mosquito control. Through a collaborative venture with the "Debug" Verily Life Sciences team, we assessed the incompatible insect technique (IIT) with the mosquito vector Aedes aegypti in northern Australia in a replicated treatment control field trial. Backcrossing a US strain of Ae. aegypti carrying Wolbachia wAlbB from Aedes albopictus with a local strain, we generated a wAlbB2-F4 strain incompatible with both the wild-type (no Wolbachia) and wMel-Wolbachia Ae. aegypti now extant in North Queensland. The wAlbB2-F4 strain was manually mass reared with males separated from females using Verily sex-sorting technologies to obtain no detectable female contamination in the field. With community consent, we delivered a total of three million IIT males into three isolated landscapes of over 200 houses each, releasing ∼50 males per house three times a week over 20 wk. Detecting initial overflooding ratios of between 5:1 and 10:1, strong population declines well beyond 80% were detected across all treatment landscapes when compared to controls. Monitoring through the following season to observe the ongoing effect saw one treatment landscape devoid of adult Ae. aegypti early in the season. A second landscape showed reduced adults, and the third recovered fully. These encouraging results in suppressing both wild-type and wMel-Ae. aegypti confirms the utility of bidirectional incompatibility in the field setting, show the IIT to be robust, and indicate that the removal of this arbovirus vector from human-occupied landscapes may be achievable.

Keywords: Aedes aegypti; arbovirus vector; biological control; incompatible insect technology; vector control.

Fig. 1.

Field trial sites in the Northern Cassowary Coast region. Control landscapes are white, and treatment landscapes are red. Urban landscapes are isolated by extensive sugar cane, and banana plantations and the proximity of all landscapes to each other are show in the upper panel, with more-detailed images of the three treatment (T1 to T3) and the three control landscapes (C1 to C3) in the lower panel. Red dots indicate positions of paired BGS traps and ovitraps used for monitoring.

Fig. 2.

Male release numbers and male overflooding ratios. The upper panel shows the wAlbB2-F4 release numbers into the treatment landscapes. During November to December 2017, 11 test releases were performed in into T1 (Mourilyan) over 3.5 wk, and two test releases were run into T2 and T3 (South Johnstone and Goondi Bend) over a single week. Then after a gap of 5 wk, continuous releases began on January 12, 2018, with 54 releases over 20 continuous weeks that ended on May 25, 2018. Fluctuations in release numbers reflect manual mass rearing productivity variation early in the releases. The lower panel shows the calculated overflooding ratio determined from the number of wAlbB2-F4 males divided by the total number of wild-type and wMel males in the traps. Gaps in the graphs during May 2018 manifest when no wild-type or wMel males were collected and overflooding ratios could not be calculated.

Fig. 3.

Population suppression summary for Ae. aegypti adults collected in BGS traps. The upper panels shows adult weekly mean BGS trapping rate and bootstrapped 95% CIs, with treatments shown as red bars (T1 to T3) and controls shown in blue bars (C1 to C3). The numbers above the bars represent the number of measurements contributing to the trapping rate estimate and includes pre-experiment test releases. The gray lines show the expected number of living released wAlbB2-F4 males assuming 30% death rate per day. Test releases through November and December 2017 (11 into T1 and two into T2 to T3) were followed by a 5-wk hiatus, and then 55 releases over 20 wk (three per week) were sustained into T1 to T3 (January 12 to May 25). The lower panels show aggregated treatment and control plots for the suppression experiment (red T1 to T3 and blue C1 to C3). The population suppression of Ae. aegypti is shown in the second bottom panel as relative trap ratio of treatments to controls in purple. The dashed lines delineate values of 1 and 0, between which the treatment landscapes are lower than in the controls (i.e., suppression). In the bottom panel, temperature is tracked by the red line, with rainfall shown as blue bars.

Fig. 4.

Larval productivity of Ae. aegypti in paired ovitraps. The upper panels show adult weekly mean larval productivity rate and bootstrapped 95% CIs, with treatments shown as red bars (T1 to T3) and controls shown in blue bars (C1 to C3). The numbers above the bars represent the number of measurements contributing to the trapping rate estimate. The gray lines show the expected number of living released wAlbB2-F4 males assuming 30% death rate per day. Test releases through November and December 2017 (11 into T1 and two into T2 to T3) were followed by a 5-wk hiatus, and then 55 releases over 20 wk (three per week) were sustained into T1 to T3 (January 12 to May 25). The lower panels show aggregated treatment and control plots for larval productivity (red T1 to T3 and blue C1 to C3). Ae. aegypti larval productivity is shown below that as relative trap ratio of treatments to controls in purple. The dashed lines delineate values of 1 and 0, between which the treatment landscapes are lower than in the controls (i.e., suppression). In the bottom panel, temperature is tracked by the red line with, rainfall shown as blue bars.

Fig. 5.

Ae. aegypti IIT suppression ongoing effect through the following season. Adult Ae. aegypti were monitored using BGS traps collected weekly over 26 wk through the following season in all three treatment landscapes (T1-Mourilyan, T2-South Johnstone, and T3-Goondi Bend) and one control lanscape (C1-Wangan). Rainfall is indicated by blue bars. Treatment 1 (Mourilyan) showed the strongest suppression effect across the following season with very few adults detected, most of which were collected late in the season. Landscapes T2 (South Johnstone) appeared to fully recover, and T3 (Goondi Bend) showed a reduced population through the following season.