Environmental influences on Aedes aegypti catches in Biogents Sentinel traps during a Californian “rear and release” program: Implications for designing surveillance programs

Abstract

As Aedes aegypti continues to expand its global distribution, the diseases it vectors (dengue, Zika, chikungunya and yellow fever) are of increasing concern. Modern efforts to control this species include “rear and release” strategies where lab-reared mosquitoes are distributed throughout the landscape to replace or suppress invasive populations. These programs require intensive surveillance efforts to monitor their success, and the Biogents Sentinel (BGS) trap is one of the most effective tools for sampling adult Ae. aegypti. BGS trap catches can be highly variable throughout landscapes, so we investigated the potential impacts of environmental factors on adult Ae. aegypti capture rates during a “rear and release” program in California to better understand the relative contributions of true variability in population density across a landscape and trap context. We recorded male and female Ae. aegypti catches from BGS traps, with and without CO2, throughout control sites where no mosquitoes were released and in treatment sites where males infected with Wolbachia were released. BGS trap catches were positively influenced by higher proportions of shade or bushes in the front yard of the premises as well as the presence of potential larval habitats such as subterranean vaults. In contrast, an increase in residential habitat within a 100 m radius of trap locations negatively influenced BGS trap catches. For male Ae. aegypti, increased visual complexity of the trap location positively influenced capture rates, and the presence of yard drains negatively affected catch rates in control sites. Lastly, for BGS traps using CO2, higher catch rates were noted from traps placed greater than one meter from walls or fences for both male and female mosquitoes. These results have important implications for surveillance programs of Ae. aegypti throughout the Californian urban environment including adult monitoring during “rear and release” programs.

About the Authors:

Kyran M. Staunton

Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

Affiliations College of Public Health, Medical and Veterinary Sciences, James Cook University, Smithfield, Australia, Australian Institute of Tropical Health and Medicine, James Cook University, Smithfield, Australia

ORCID logo http://orcid.org/0000-0001-6812-3320

Jacob E. Crawford

Roles Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing

Affiliation: Verily Life Sciences, South San Francisco, California, United States of America

Devon Cornel

Roles Conceptualization, Data curation, Investigation, Writing – review & editing

Affiliation: Consolidated Mosquito Abatement District, Parlier, California, United States of America

Peter Yeeles

Roles Data curation, Formal analysis, Writing – original draft, Writing – review & editing

Affiliation: College of Science & Engineering, James Cook University, Smithfield, Australia

Mark Desnoyer

Roles Data curation, Formal analysis, Writing – review & editing

Affiliation: Verily Life Sciences, South San Francisco, California, United States of America

Josh Livni

Roles Data curation, Formal analysis, Writing – original draft, Writing – review & editing

Affiliation: Verily Life Sciences, South San Francisco, California, United States of America

Jodi Holeman

Roles Conceptualization, Investigation, Project administration, Resources, Writing – review & editing

Affiliation: Consolidated Mosquito Abatement District, Parlier, California, United States of America

F. Stephen Mulligan

Roles Conceptualization, Investigation, Project administration, Resources, Writing – review & editing

Affiliation: Consolidated Mosquito Abatement District, Parlier, California, United States of America

Nigel Snoad

Roles Funding acquisition, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing

Affiliation: Verily Life Sciences, South San Francisco, California, United States of America

Scott A. Ritchie

Roles Conceptualization, Funding acquisition, Methodology, Project administration, Supervision, Writing – original draft, Writing – review & editing

Affiliations College of Public Health, Medical and Veterinary Sciences, James Cook University, Smithfield, Australia, Australian Institute of Tropical Health and Medicine, James Cook University, Smithfield, Australia

Introduction

Aedes aegypti, the main vector of dengue, Zika, chikungunya and yellow fever viruses continues to increase its global distribution invading urban habitats outside of tropical climates, including many locations within the USA [1]. Aedes aegypti spread rapidly throughout North America being recorded in 220 counties in 28 states and the District of Columbia between 1995 and 2017 and was even responsible for a Zika virus outbreak in Florida in 2016 [2]. Within California, Ae. aegypti were introduced multiple times over the last few decades [3] with breeding populations being established in 2013 in Madera and Fresno in the central valley and San Mateo on the coast [4]. Based on typical trends [5], we could expect California to experience local outbreaks of diseases vectored by Ae. aegypti within the next decade. As Ae. aegypti have yet to reach their projected distributional limit within the USA [1, 5], this species is of great concern to many mosquito abatement districts.

With the relatively recent invasion of Aedes into California, traditional native mosquito control methods in this state have been heavily revised [6] to include treatments targeting these species. Recently, there has been significant global interest in the development of “rear and release” [7] programs to control mosquito populations [8–12]. These control methods involve mass-rearing mosquitoes in laboratories and releasing them throughout a landscape to either replace (using transinfections of symbiotic bacteria, such as Wolbachia; [12, 13]) or suppress (using the sterile insect technique or incompatible insect technique (IIT)) local populations [8, 9]. Verily Life Sciences (a Google affiliate) have been successfully performing an IIT “rear and release” program in Fresno County, California using male Ae. aegypti mosquitoes infected with Wolbachia which mate with local females and produce non-viable offspring due to cytoplasmic incompatibility [14]. By releasing 14.4 million males infected with Wolbachia across three neighbourhoods in 2018 Verily Life Sciences were able to demonstrate a suppression rate during the peak mosquito season of 95.5% [14].

Rear and release programs commonly deploy the Biogents Sentinel (Regensburg, Germany; hereinafter BGS) trap to survey mosquito populations and monitor success [8, 9, 12, 15]. The BGS trap is known as the gold standard for sampling Aedes mosquitoes and has been shown to outperform many other traps [16, 17]. This trap can be deployed unbaited, relying on its dark colouration to visually attract Aedes or with olfactory attractants such as BG-Lures and/or CO2 [18, 19]. It is common practice to use CO2 in combination with the BG-Lure in BGS traps as studies indicate that such methodology significantly increases Ae. aegypti catches [20, 21].

Despite the success of mosquito traps, including the BGS trap, Ae. aegypti collection rates in these traps can be highly variable [6]. Trap catches may be influenced by trap placement within premises and the attraction of mosquitoes to nearby oviposition and blood-feeding opportunities [6]. The distribution and dispersal of Ae. aegypti throughout the urban landscape may also be influenced by the availability of breeding habitats, barriers such as unshaded areas, busy roads or water bodies [22–24] which in turn could potentially impact collection rates. It is clear that identifying the causes of variation in BGS trap capture rates is vital to understanding and monitoring population trends across landscapes.

To date, few studies have evaluated the effects of environmental influences on BGS trap collection rates. An exception is recent work by Staunton, Yeeles (15), which evaluated the influences of premises condition and trap location factors during the Debug Innisfail “rear and release” program in northern Australia (https://research.csiro.au/mozzieproject/). Staunton, Yeeles (15) found that, in control sites (where male mosquitoes were not released), the condition of a house and whether or not it had screened windows influenced trap catch rates, as well as trap location factors such as the shelter above the trap, distance from outside sitting areas, and the visual complexity of the trap environment. In suburbs where Wolbachia-infected male mosquitoes were released, Staunton, Yeeles (15) detected influences in BGS trap catches, relative to control sites, from the amount of shade within a premises, as well as the house design and building materials.

While the findings of Staunton, Yeeles (15) are informative for the northern Australian urban environment, they may not directly apply to the Californian Ae. aegypti habitat. Unlike the urban environment of northern Australia, California displays cold wet winters and the summer is dry and hot, with low humidity—environmental factors once thought to inhibit the establishment of Ae. aegypti in California [6]. Since rainfall in the Central Valley of California is restricted to cold months of the year when temperatures are unsuitable for larval development, larval habitats are reliant on water from residential watering [25]. However, similar to northern Australia, larval habitats for Ae. aegypti in California may also consist of surface water-holding containers as well as subterranean sites [6, 26]. Adult mosquito harbourage in northern Australia is often associated with “Queenslander” styled homes which may be unscreened, in relatively poor condition and contain ground-level open access rooms [27, 28]. As such houses are uncommon in California, adult mosquito harbourage may instead be linked to favourable microclimates provided by vegetative features and residential watering, as suggested by Hayden, Uejio [29] for Ae. aegypti in Arizona.

We investigated the influences of environmental factors on male and female Ae. aegypti catches in BGS traps, with and without CO2, during Verily’s 2018 “rear and release” program in Fresno County, California. The environmental factors assessed relate to vegetation characteristics, potential larval habitats, trap placement characters, and the amount of residential environment surrounding the trap. This investigation not only gives insight into the ecology of this species, but also valuable information for designing surveillance programs in such urban habitats with or without male releases.

Methods

Male mass rearing and release

Infected Ae. aegypti were mass-reared in the Verily facility in South San-Francisco as described by [14]. Female mosquitoes were removed prior to release using Verily’s automated sex-sorting process. On average, 78,000 males were consistently released daily between 17 April– 17 October 2018 across all treatment areas from a specially modified van that tracked the GPS location and size of each release. See Crawford, Clarke [14] for further details regarding this program.

Climate

Fresno County summers are generally characterised as hot with no rainfall and low humidity. Between July and October 2018, temperatures ranged from 56 to 108 F. The total rainfall during this period was 0 mm (NOAA weather station, Fresno Airport).

Response data collection and selection

The premises and trap evaluation protocol involved visually inspecting 180 BGS traps throughout six communities in Fresno County (for map refer to Crawford, Clarke [14]) between 12 and 14 September 2018. To maintain consistency of assessment methodology, a single staff member was responsible for scoring predictors for all traps and only traps within domestic-styled housing locations were analysed. BGS traps assessed within control sites (n = 68) were located in locations labelled C1 (n = 25), C2 (n = 28) and C3 (n = 15) which were 66 ha, 66 ha and 46 ha in area, respectively. These control sites were geographically paired with treatment sites (n = 112) in T1 (n = 44), T2 (n = 33) and T3 (n = 35), which were 130 ha, 74 ha and 89 ha in area, respectively (for further details including a map see Crawford, Clarke [14]). Each site was subdivided into 7-acre grids, and BGS traps were randomly set within grid cells in order to maintain even coverage and consistent trap densities among sites [14]. As the program aim was to suppress Ae. aegypti populations, an operational decision was made to release more males in locations with higher housing densities, assessed in this study as residential habitat, as such locations were thought to contain larger populations of this mosquito. Additionally, trap locations were biased towards houses which contained vegetation in their front yards (a predominant characteristics of such neighbourhoods) and traps were often positioned behind vegetation to improve security from passers-by. The data analysed were selected from six months of trapping, which coincides with state-wide peaks in activity [6], when both male and female mosquitoes were most active between 1 May– 1 November 2018. At all times BGS traps were baited with BG-Lures (Biogents, Regensburg, Germany), but within each weekly cycle all BGS traps were run simultaneously for six nights without CO2 and then one night with CO2, (deployed using 2 pounds of dry ice per trap held in an ‘Insulated Dry Ice Bucket for EVS Trap’ purchased from Bioquip, California, USA). Therefore, the four datasets analysed were total male and female Ae. aegypti collections from BGS traps set with and without CO2.

Environmental factor selection

Through consultation among mosquito experts, including local abatement staff, we selected and recorded a range of environmental and trap location factors considered relevant to this urban habitat (S1 Table). First, we recorded factors characterising the vegetative structure within a property. These included: the level of shade present in the yard (similar to that proposed by Tun-Lin, Kay [30] where scores 1–3 represented the proportion of yard covered in vegetative shade as <25%, 25–50% or >50%, respectively; S1 Fig), the proportion of yard that contained bushes as either being (<33%, 33–66% and >66% with scores 1–3 respectively; S2 Fig) and the number of neighbours surrounding the premises which contained high levels (>50%) of shade. Unlike Tun-Lin, Kay (30) we did not score house condition due to the consistently good condition of dwellings in the study sites [25].

Second, we recorded the within-premise factors characterising the traps’ locations. These included: whether the visibility of the trap from the road was clear, partially obscured or completely obscured (S3 Fig), the distance the trap was positioned from a wall or fence structure (<1 m, 1–2 m & >2 m; S4 Fig) and the visual complexity surrounding the trap from either competition with surrounding dark objects [18] and/or general visual obstruction (low, medium and high; S5 Fig). Third, we recorded various types of potential larval habitats (through exposure to water from irrigation) found throughout the premises. We noted the number of yard containers such as: buckets, tires, potted plant bases and tarpaulins within a premises as well as the presence of Pacific Gas and Electric (PG&E) subterranean vaults, yard drains and catch basins (S6 Fig). Finally, we investigated the influence of the amount of residential habitat within a 100 m radius of the trap location. These proportions were calculated by first buffering building footprints by 5 m, and then for each trap calculating the percentage of a 100 m radius circle around it that was covered by the buffered building area (S7 Fig).

Premises condition and trap location data analysis

Models were run within the R statistical environment ver. 3.3.3 (R Core Team 2017) using a methodology previously reported by Staunton, Yeeles (15). We created separate models for male and female response data (total counts per trap), each including both the pooled treatment (n = 112) and control (n = 68) sites, using generalised linear models (GLM) in the lme4 package [31]. Differences between continuous predictors were scaled (2*SD) and centred to avoid convergence issues [32]. Initial count-data models fit using Poisson distributions exhibited overdispersion, therefore final models were estimated using negative binomial distributions with log-link functions. We used the ‘corvif’ function in the AED package [33] to calculate variance-inflation factors and found that multicollinearity wasn’t likely to confound models. Subsequently, we analysed the effect of 11 parameters (shade, proportion of bushes in front yard, number of neighbours with high shade level in their premises, the visibility of the trap from the road, the distance of the trap from a wall or fence, the visual complexity of a trap, the number of yard containers, the presence of PG&E vaults, yard drains and catch basins and the amount of residential habitat within a 100 m radius of the trap) that might influence trap capture rates and their interactions with treatment (the releases of males). To account for trap fails we included an offset parameter in the model of the count of the total number of trap successes during the sampling period analysed. Finally, to account for the bias created by releasing more males in locations with higher housing densities we included a covariate of the total number of males released within 100 m radius of each BGS trap during the collection period as a fixed factor in our models. This factor was calculated for each release using a sensor on board the release vehicle that counts each individual mosquito as it is released and creates a record with GPS coordinates. In the rare instances where the sensor failed, the data was backfilled with a point placed at the triggered release location equal to the number of mosquitoes placed into the release device at the factory.

A single model would therefore take the basic form of, for example, Total female count per trap ~ (11 Parameters) * Treatment + Males released within 100 m of BGS trap locations + offset (trap successes). Using this approach, influences from parameters on response variables of interest from control sites could be modelled with interactions with male releases in treatment sites. By incorporating these interactions into the models we could examine how changes in environmental characteristics influenced catch rates per trap within Control sites, and whether they were relatively different in Treatment sites. We used backward selection to simplify models [34] by comparing AIC (Akaike’s Information Criterion) scores. Finally, we sequentially reintegrated removed terms back into the final model to determine their effect sizes and prevent bias associated with otherwise assigning effect sizes of zero [35, 36]. As the model used a negative bionomial distribution with a log-link function, Pseudo-R2 (McFadden) values [37] were calculated for final models using the modEvA package [38] to indicate model fit. The McFadden Pseudo-R2, calculated as 1 –(the maximum of the log likelihood function divided by the log likelihood of the null model), describes an excellent model fit being between 0.2 and 0.4 [39].

A linear regression was performed (in GraphPad Prism ver. 7.03) to confirm our suspected bias between the proportion of residential habitat within a 100 m radius of trap locations and the total number of males released per BGS trap assessed. For this analysis, data from both trap types (with and without CO2) were combined as the released males potentially lived longer than 24 hours and therefore males may have been caught by either trap type.

Results

Trap catches

In total, 604,262 Ae. aegypti were sampled over the six month period from the 180 BGS traps analysed in this study (Table 1).

[Figure omitted. See PDF.]

Table 1. BGS trap collections of Ae. aegypti in Fresno County from 1 May—1 November 2018.

Traps ran for a total of 27 weeks in this analysis and during that time were serviced either after one (traps with CO2) or six days (traps without CO2) per week. Male Ae. aegypti infected with Wolbachia were not released in control sites, but were released within treatment sites.

https://doi.org/10.1371/journal.pntd.0008367.t001

Bias associated with releasing more males in areas with greater residential habitat

There was a significantly positive relationship (R2 = 0.37, F1,110 = 64.1, P < 0.0001) between the proportion of residential habitat within a 100 m radius of trap locations and the total number of males released per BGS trap (S8 Fig).

Influences of environmental factors on male Ae. aegypti catches in BGS traps with CO2

Releasing males positively influenced (Treatment: 1.92 ± 0.26 [ß ± S. E.] & Males released within 100 m radius: 1.04 ± 0.18; Table 2) male catches in BGS traps baited with CO2. Additionally, BGS traps located in premises with high (>50%) levels of shade caught more males (0.33 ± 0.13; Table 2; Fig 1A) than those located in premises containing low shade (<25%). Vegetation further influenced BGS catches with more males being sampled in traps at premises with medium (33–66%) or high (>66%) proportions of bushes in the front yard in control sites (0.32 ± 0.14 and 0.8 ± 0.22 respectively; Table 2; Fig 1B). However, the positive influence from a high proportion of bushes on BGS trap catches was not supported in treatment sites (-0.66 ± 0.27; Table 2; Fig 1B).

[Figure omitted. See PDF.]

Fig 1. Total male Ae. aegypti catches per BGS trap baited with CO2 (ß ± S. E.) for each predictor with confidence intervals that don’t overlap zero.

Predictors grouped as those from sites which didn’t have males released (Control) and those which interacted with males released (Treatment (interactions)). Stars indicate modelled predictor levels with confidence intervals that don’t overlap zero. Note that treatment interaction data may be biased by greater male releases occurring in areas with higher residential habitat densities which is accounted for in the model. A) Shade levels 1, 2 & 3 represent <25%, 25–50% & >50% coverage of yard by shade, respectively. B) Bushes in front yard levels 1, 2, & 3 represent <33%, 33–66% and >66%, respectively. C) Distance of the trap from a wall or fence levels 1, 2 & 3 represent <1 m, 1–2 m, & >2 m, respectively. D) Visual complexity levels 1, 2 & 3 represent clear, partial and high amounts of obstruction and/or visual competition from surrounding features, respectively. E) If PG&E vaults were present at the premises. F) Whether yard drains were present at the premises. G) Proportion of residential habitat (m2) within a 100 m radius of trap locations.

https://doi.org/10.1371/journal.pntd.0008367.g001

[Figure omitted. See PDF.]

Table 2. Parameter estimates from the GLMs fitting male and female Ae. aegypti catches per BGS trap baited with CO2.

Effects with ‘: Treatment’ refer to interactions between parameters and the treatment (male Ae. aegypti infected with Wolbachia release) effect. Pseudo-R2 (McFadden) values = 0.28 and 0.3 for males and female models, respectively.

https://doi.org/10.1371/journal.pntd.0008367.t002

BGS traps with CO2 that were positioned within premises either 1>2 m or >2 m from a wall or fence displayed higher catch rates of male Ae. aegypti in control sites (0.65 ± 0.16 and 0.49 ± 0.18 respectively; Table 2; Fig 1C). Both these relationships were not supported with treatment sites (-0.46 ± 0.21 and -0.55 ± 0.25, respectively; Table 2; Fig 1C). Additionally, BGS traps located in highly visually complex areas caught more males than those set in locations characterised by low visual complexity (0.63 ± 0.21; Table 2; Fig 1D).

PG&E vault presence positively influenced male catches in BGS traps with CO2 (0.47 ± 0.20; Table 2; Fig 1E), although this relationship was not supported in treatment sites (-0.52 ± 0.23; Table 2; Fig 1E). Yard drain presence negatively influenced BGS trap (with CO2) catches of male Ae. aegypti in controls sites (-0.27 ± 0.13; Table 2; Fig 1F), but again this relationship was not supported in treatment sites (0.50 ± 0.16; Table 2; Fig 1F). Finally, the amount of residential habitat, within a 100 m radius of each trap, negatively influenced (-0.37 ± 0.12) male catches in control sites (Table 2; Fig 1F).

Influences of environmental factors on female Ae. aegypti catches in BGS traps with CO2

Releasing Wolbachia-infected males negatively influenced (-2.46 ± 0.11; Table 2) female catches in BGS traps baited with CO2. BGS traps located in premises with medium (25–50%) or high (>50%) levels of shade caught more females (0.33 ± 0.12 and 0.53 ± 0.15 respectively; Table 2; Fig 2A) than those located in premises with low shade (<25%).

BGS traps distanced either 1–2 m or >2 m from a wall or fence caught more female Ae. aegypti than those within 1 m of such structures (0.58 ± 0.14 and 0.64 ± 0.17 respectively; Table 2; Fig 2B). Again, the presence of PG&E vaults positively influenced BGS trap catches (0.36 ± 0.13; Table 2; Fig 2C). Lastly, the amount of residential habitat within a 100 m radius of each trap negatively influenced (-0.57 ± 0.13) female catches (Table 2; Fig 2D).

[Figure omitted. See PDF.]

Fig 2. Total female Ae. aegypti catches per BGS trap baited with CO2 (ß ± S. E.) for each predictor with confidence intervals that don’t overlap zero.

Predictors grouped as those from sites which didn’t have males released (Control) and those which interacted with males released (Treatment (interactions)). Stars indicate modelled predictor levels with confidence intervals that don’t overlap zero. Note that treatment interaction data may be biased by greater male releases occurring in areas with higher residential habitat densities which is accounted for in the model. A) Shade levels 1, 2 & 3 represent <25%, 25–50% & >50% coverage of yard by shade, respectively. B) Distance of trap from a wall or fence levels 1, 2 & 3 represent <1 m, 1–2 m, & >2 m, respectively. C) If PG&E vaults were present at the premises. D) Proportion of residential habitat (m2) within a 100 m radius of trap locations.

https://doi.org/10.1371/journal.pntd.0008367.g002

Influences of environmental factors on male Ae. aegypti catches in BGS traps without CO2

Releasing males positively influenced (Treatment: 2.29 ± 0.33 & Males released within 100 m radius: 1 ± 0.24; Table 3) the numbers of males caught in BGS traps that were not baited with CO2. BGS traps located at premises with medium (33–66%) or high (>66%) proportions of bushes in the front yard in control sites caught more males than those with low (<30%) proportions of bushes (0.56 ± 0.18 and 0.77 ± 0.27 respectively; Table 3; Fig 1A). However, the positive influence from placing traps in locations with medium proportions of bushes (33–66%) was not supported in treatment sites (-0.59 ± 0.22; Table 3; Fig 3A).

[Figure omitted. See PDF.]

Fig 3. Total male Ae. aegypti catches per BGS trap without CO2 (ß ± S. E.) for each predictor with confidence intervals that don’t overlap zero.

Predictors grouped as those from sites which didn’t have males released (Control) and those which interacted with males released (Treatment (interactions)). Stars indicate modelled predictor levels with confidence intervals that don’t overlap zero. Note that treatment interaction data may be biased by greater male releases occurring in areas with higher residential habitat densities which is accounted for in the model. A) Bushes in front yard levels 1, 2, & 3 represent <33%, 33–66% and >66%, respectively. B) Distance of the traps from a wall or fence levels 1, 2 & 3 represent <1 m, 1–2 m, & >2 m, respectively. C) Visual complexity levels 1, 2 & 3 represent clear, partial and high amounts of obstruction and/or visual competition from surrounding features, respectively. D) If PG&E vaults were present at the premises. E) Yard drains present or absent at premises. F) Proportion of residential habitat (m2) within a 100 m radius of trap locations.

https://doi.org/10.1371/journal.pntd.0008367.g003

[Figure omitted. See PDF.]

Table 3. Parameter estimates from the GLMs fitting male and female Ae. aegypti catches per BGS trap without CO2.

Effects with ‘: Treatment’ refer to interactions between parameters and the treatment (Wolbachia-infected male Ae. aegypti releases) effect. Pseudo-R2 (McFadden) values = 0.22 and 0.25 for males and female models respectively.

https://doi.org/10.1371/journal.pntd.0008367.t003

BGS traps positioned at 1–2 m from a wall or fence displayed higher catch rates of male Ae. aegypti than those within 1 m of such structures (0.32 ± 0.14; Table 3; Fig 3B). Furthermore, BGS traps located in highly visually complex areas caught more males than those in locations characterised by low visual complexity (0.66 ± 0.27; Table 3; Fig 3C).

The presence of PG&E vaults positively influenced male catches in BGS traps without CO2 (0.61 ± 0.26; Table 3; Fig 3D). Whereas, the presence of yard drains within premises negatively influenced male catch rates (-0.33 ± 0.16; Table 3; Fig 3E). Both relationships, regarding these potential larval habitat types, were not supported in treatment sites (-0.6 ± 0.3; Table 3; Fig 3D and 0.57 ± 0.2; Table 3; Fig 3E for PG&E vaults and yard drains, respectively). Finally, the amount of residential habitat, within a 100 m radius of each trap, negatively influenced (-0.33 ± 0.16) male catches (Table 3; Fig 3F).

Influences of environmental factors on female Ae. aegypti catches in BGS traps without CO2

Releasing males negatively influenced (-2.32 ± 0.15; Table 3) female catches in BGS traps not baited with CO2. BGS traps located in premises with medium (25–50%) or high (>50%) levels of shade caught more females (0.35 ± 0.14 and 0.42 ± 0.18, respectively; Table 3; Fig 4A) than those in premises characterised by low shade (<25%).

[Figure omitted. See PDF.]

Fig 4. Total female Ae. aegypti catches per BGS trap without CO2 (ß ± S. E.) for each predictor with confidence intervals that don’t overlap zero.

Predictors grouped as those from sites which didn’t have males released (Control) and those which interacted with released males (Treatment (interactions)). Stars indicate modelled predictor levels with confidence intervals that don’t overlap zero. Note that treatment interaction data may be biased by greater male releases occurring in areas with higher residential habitat densities which is accounted for in the model. A) Shade levels 1, 2 & 3 represent <25%, 25–50% & >50% coverage of yard by shade, respectively. B) Bushes in front yard levels 1, 2, & 3 represent <33, 33–66% and >66%, respectively. C) If PG&E vaults were present at the premises. D) Proportion of residential habitat (m2) within a 100 m radius of the trap location.

https://doi.org/10.1371/journal.pntd.0008367.g004

Premises containing a high proportion (>66%) of bushes in the front yard in treatment sites negatively influenced female Ae. aegypti capture rates (from BGS traps without CO2), relative to such locations in control sites (-0.75 ± 0.38; Table 3; Fig 4B). The presence of PG&E vaults again positively influenced BGS trap catches of female Ae. aegypti (0.48 ± 0.14; Table 3; Fig 4C). Finally, the amount of residential habitat within a 100 m radius of each trap negatively influenced (-0.55 ± 0.15) female catches in BGS traps without CO2 (Table 3; Fig 4D).

Discussion

Key findings from this study for consideration by staff setting BGS traps for Ae. aegypti surveillance in similar urban habitats include: 1) BGS traps placed within premisses with high proportions of shade, or medium to high proportions of bushes, caught more Ae. aegypti than those placed in locations where such features were scarce, 2) BGS traps deployed with CO2 displayed higher catch rates of Ae. aegypti when distanced >1 m from structures such as walls or fences, 3) BGS traps placed in locations characterised by high visual complexity, generally due to being placed behind bushes, demonstrated higher catch rates of male Ae. aegypti than those placed out in the open, 4) BGS traps placed within premisses with subterranean vaults, such as PG&E vaults, displayed higher catch rates of Ae. aegypti than those placed in premisses without such features and 5) BGS traps placed within locations surrounded by higher proportions of residential habitat displayed reduced catch rates of Ae. aegypti. Specific influences of the environmental factors mentioned above on BGS trap catch rates varied according to mosquito sex, if traps were deployed with or without CO2 and whether the population surveyed was impacted by male releases and are discussed in detail within the sections below.

Influence from vegetation

Higher proportions of shade within premises positively influenced BGS trap catches of male Ae. aegypti when CO2 was used and females with or without the use of CO2. This association between shade and collection rates of Ae. aegypti has been previously recorded using ovitraps in Australia [24, 40] and in Arizona [29]. Staunton, Yeeles [15] found a positive association between high shade levels and both male and female Ae. aegypti catch rates in treatment sites for a similar rear and release program, although unlike this current work the relationship they detected wasn’t supported in the control sites. Well shaded areas have been suggested to provide favourable microclimates for resting Ae. aegypti, and may act as dispersal corridors and provide sustenance via floral nectaries to adults and decomposing organic matter for larvae [24, 29].

Our current study also found that medium or high proportions of bushes in the front yard of premises also positively influenced male Ae. aegypti catches in control sites. These relationships were less evident with BGS trap catches in treatment sites. Additionally, female catches using BGS traps without CO2 were negatively influenced by a high proportion of front yard bushes in the treatment sites, relative to the control sites. Like heavily shaded locations, bushy yards may provide suitable resting habitats for males which subsequently increase the catch rates of BGS traps located in such areas. However, when males are released from vans, a higher proportion of bushes may interfere with the movement of males into the BGS trap and thereby reduce catch rates relative to locations with comparatively fewer bushes. Additionally, BGS traps were commonly placed within or behind bushes in front yards. Therefore, a yard with fewer bushes may attract the males to the specific area containing this vegetation and subsequently, the BGS trap. It is unclear why comparatively fewer females were caught in treatment locations containing a greater proportion of bushes when using a BGS trap without CO2. Perhaps this was due to more effective suppression in such locations, but further investigations are certainly warranted to better clarify the influence of this environmental factor on Ae. aegypti catches.

Distance from wall or fence

BGS traps with CO2 were generally more effective when placed further than 1 m from a wall or fence. The addition of CO2 to the BGS trap provides an olfactory attractant which increases the catch rates of BGS traps [21]. As the CO2 deployed with BGS traps was a gas likey carried by wind it is logical that placing the trap close to a barrier may have reduced the gas’s ability to permeate throughout a region and therefore decreased the trap’s attractive quality. The additional findings that trap location, regarding distance to a wall or fence, didn’t strongly influence catch rates when CO2 was not used, provide additional support for this hypothesis.

Visual complexity

Placing the BGS traps in locations characterised by high levels of visual complexity positively influenced male catch rates. This finding contradicts previous work performed in northern Australia that suggests that higher visual complexity reduces the effectiveness of the BGS traps [15, 18]. While Staunton, Yeeles (15) suggested that this negative influence may derive from competing dark objects and obscured visual cues of the trap, the habitat in which the traps were placed in California is different to that of northern Australia. Unlike northern Australian premises, which may contain many manufactured objects in their yards [30], the premises in Fresno displayed front yards that generally contained manicured gardens. As stated above, traps were generally placed behind vegetation in the yard to reduce visibility to passers-by. Subsequently, the traps which were located within highly visually complex environments were generally also placed within bushy gardens. Although no collinearity between these factors was detected, traps which were hard to see may still have benefitted from being placed close to bushes which are likely to be attractive microhabitats.

Potential larval habitats

Consistently, for both sexes and trap types, the presence of PG&E vaults positively influenced Ae. aegypti catch rates in control sites, potentially due to being productive larval habitats or effective harbourage features. The hot dry Californian summer climate was traditionally considered to inhibit Ae. aegypti establishment within this state [6]. However, within Los Angeles and Merida Mexico, Ae. aegypti were associated with storm sewers [6, 41] and research in northern Australia also emphasised the importance of subterranean containers, particularly telecommunication pits comparable to the PG&E vaults, as larval habitats [26]. In Australia, these underground larval habitats were increasingly used by Ae. aegypti during the dry winter when surface containers were dry [26]. Additionally, Ae. aegypti established within a highly xeric Indian town were noted to utilise subterranean cement water containers known as tankas [42], again indicating the potential importance of such features to this species’ survival.

Within treatment sites, the negative influence of PG&E vault presence on male catch rates, relative to control sites, may reflect the comprehensive release of males throughout neighbourhoods and even indicate effective suppression of such habitats. Clearly, further work is required to identify the extent to which these features are facilitating the establishment of Ae. aegypti within California and are able to be suppressed by male releases.

Unlike PG&E vaults, the presence of yard drains negatively influenced male Ae. aegypti catch rates in control sites. This finding was unexpected as, while not a direct focus of this study, our casual inspections of yard drains found them to generally contain water. Additionally, we noted that many wet yard drains contained yard clippings or other organic material and while we collected some Ae. aegypti larvae, in the few samples we collected, we predominantly found immature Culex. Although Ae. aegypti oviposition is known to increase in water containing an organic infusion [43], Culex are known to prefer to oviposit in water with a high organic content such as that from older infusions [44]. Therefore, perhaps yard drains are generally more suitable to sustaining Culex populations. As our observations regarding yard drains as larval habitats were relatively anecdotal, our BGS trapping results indicate that more formal studies into this larval habitat are warranted.

As with PG&E vaults, the influence of yard drain presence on male catch rates noted in control sites was not supported in treatment sites. Again, this difference in relationships between sites may relate to the comprehensive male releases throughout the treatment neighbourhoods. Female Ae. aegypti are thought to be attracted to the odours emanating from potential larval habitats [24] so it is possible that the males released from vans may also be attracted to such areas.

Residential habitat with 100 m radius of trap

The finding that greater proportions of residential area negatively influenced Ae. aegypti catch rates in BGS traps is counter-intuitive to previous research demonstrating that these mosquitoes prefer densely occupied areas [45, 46]. Perhaps, the good condition of screened houses throughout these areas force these mosquitoes to display relatively exophilic behaviours. Therefore, trap catches may be positively influenced by alternate landscape features such as large yards, parks, car garages, roads and other clearings and buildings. Large yards and parks may contain additional features such as bushes or trees which create attractive microhabitats for Ae. aegypti. Additionally, clearings and busy roads are suggested to be barriers to dispersal for Ae. aegypti [22, 24] and as such may direct the movement of these mosquitoes towards locations containing the traps. These findings reinforce how environmental factors may influence Ae. aegypti catch rates at larger landscape scales and not just at fine resolutions within premisses.

It is worth restating that all traps and associated environmental factors were assessed within the front yards of premises. Unlike northern Australian communities, Californian residents do not support staff access to their backyards. While back yard features, which may influence BGS trap efficacy, were not assessed, the results presented in this study remain relevant and informative to public health staff who will generally set mosquito traps within front yards in this urban environment. Additionally, the authors acknowledge that fine-scale weather factors such as temperature and humidity at trap locations may also influence trap catch rates, but were not included in these analyses. Determining the influences of such factors were outside the scope of this study which aimed to provide information regarding environmental characteristics that were likely correlated, such as proportion of shade, but more easily applied by staff when selecting trap locations.

Conclusions

The environmental factors that influenced BGS trap catches were generally consistent across mosquito sex and trap (with or without CO2) and site (control or treatment) type. Surveillance programs of Ae. aegypti within Fresno, should consider that BGS trap catches were positively influenced by the amount of vegetation within a premises and the presence of potential larval habitats or harbourage features such as PG&E vaults. Additionally, catch rates from BGS traps using CO2 were positively influenced when the traps were positioned >1 m away from walls or fences and male mosquito catches were strongly influenced by the proportion of bushes within premises, unlike females. Lastly, BGS trap catches were consistently negatively influenced by greater proportions of residential areas within 100 m of trap locations. Although it remains difficult to disentangle whether these environmental features are better indicators of local Ae. aegypti breeding and harbourage or better trap performance, such results indicate that male releases could be conditional, dependent on residential type and environmental characteristics such as vegetation structure and potential larval habitats.

These findings also suggest that environmental factors influence BGS trap catches of Ae. aegypti in California differently to tropical locations such as northern Australia and emphasise the highly adaptable nature of this species. Surveillance programs therefore need to strongly consider the fundamental requirements for each life stage of Ae. aegypti and how the environment they are surveying within may facilitate the survival of this species and consequently influence trap success.

Supporting information

S1 Table. Number of sites for each categorical predictor measured in control and treatment areas.

https://doi.org/10.1371/journal.pntd.0008367.s001

(DOCX)

S1 Fig.

Different amounts of shade A. <25%, B. 25–50% C. >50%.

https://doi.org/10.1371/journal.pntd.0008367.s002

(TIF)

S2 Fig.

Different amounts of bushes in the front yard A. <33%, B. 33–66% and C. >66%.

https://doi.org/10.1371/journal.pntd.0008367.s003

(TIF)

S3 Fig.

BGS trap visibility from the road A. clear, B. partially obscured and C. obscured.

https://doi.org/10.1371/journal.pntd.0008367.s004

(TIF)

S4 Fig.

BGS trap proximity to wall or fence A. <1 m, B. 1–2 m and C. >2 m.

https://doi.org/10.1371/journal.pntd.0008367.s005

(TIF)

S5 Fig.

Visual competition to the BGS trap A. low, B. medium and C. high.

https://doi.org/10.1371/journal.pntd.0008367.s006

(TIF)

S6 Fig.

Potential larval habitats A. PG&E vault, B. yard drain, C. catch basin and D. yard container.

https://doi.org/10.1371/journal.pntd.0008367.s007

(TIF)

S7 Fig.

Residential habitat footprints (yellow areas) within 100 m radius of BGS traps (blue points) in T1, Fresno. Map created in QGIS version 3.4 with base layer sourced from The United States Geological Services (USGS).

https://doi.org/10.1371/journal.pntd.0008367.s008

(TIF)

S8 Fig. Total males released per BGS trap for collections with and without CO2 combined (R2 = 0.37, F1,110 = 64.1, P < 0.0001).

https://doi.org/10.1371/journal.pntd.0008367.s009

(TIF)

Acknowledgments

We thank the community of Fresno County and the Consolidated Mosquito Abatement District for their support during this project.

Citation: Staunton KM, Crawford JE, Cornel D, Yeeles P, Desnoyer M, Livni J, et al. (2020) Environmental influences on Aedes aegypti catches in Biogents Sentinel traps during a Californian “rear and release” program: Implications for designing surveillance programs. PLoS Negl Trop Dis 14(6): e0008367. https://doi.org/10.1371/journal.pntd.0008367

References

1. Kraemer MU, Sinka ME, Duda KA, Mylne AQ, Shearer FM, Barker CM, et al. The global distribution of the arbovirus vectors Aedes aegypti and Ae. albopictus. elife. 2015;4:e08347. pmid:26126267

2. Hahn MB, Eisen L, McAllister J, Savage HM, Mutebi J-P, Eisen RJ. Updated reported distribution of Aedes (Stegomyia) aegypti and Aedes (Stegomyia) albopictus (Diptera: Culicidae) in the United States, 1995–2016. J Med Entomol. 2017;54(5):1420–4. pmid:28874014

3. Pless E, Gloria-Soria A, Evans BR, Kramer V, Bolling BG, Tabachnick WJ, et al. Multiple introductions of the dengue vector, Aedes aegypti, into California. PLoS Negl Trop Dis. 2017;11(8):e0005718. pmid:28796789

4. Gloria-Soria A, Brown JE, Kramer V, Hardstone Yoshimizu M, Powell JR. Origin of the dengue fever mosquito, Aedes aegypti, in California. PLoS Negl Trop Dis. 2014;8(7):e3029. pmid:25077804

5. Kraemer MUG, Reiner RC, Brady OJ, Messina JP, Gilbert M, Pigott DM, et al. Past and future spread of the arbovirus vectors Aedes aegypti and Aedes albopictus. Nat Microbiol. 2019. pmid:30833735

6. Metzger ME, Hardstone Yoshimizu M, Padgett KA, Hu R, Kramer VL. Detection and establishment of Aedes aegypti and Aedes albopictus (Diptera: Culicidae) mosquitoes in California, 2011–2015. J Med Entomol. 2017;54(3):533–43. pmid:28399270

7. Ritchie S. Rear and release: a new paradigm for dengue control. Aust Entomol. 2014;53(4):363–7.

8. Carvalho DO, McKemey AR, Garziera L, Lacroix R, Donnelly CA, Alphey L, et al. Suppression of a field population of Aedes aegypti in Brazil by sustained release of transgenic male mosquitoes. PLoS Negl Trop Dis. 2015;9(7):e0003864. pmid:26135160

9. Harris AF, McKemey AR, Nimmo D, Curtis Z, Black I, Morgan SA, et al. Successful suppression of a field mosquito population by sustained release of engineered male mosquitoes. Nat Biotechnol. 2012;30(9):828. pmid:22965050

10. Lacroix R, McKemey AR, Raduan N, Kwee Wee L, Hong Ming W, Guat Ney T, et al. Open field release of genetically engineered sterile male Aedes aegypti in Malaysia. PLoS One. 2012;7(8):e42771. pmid:22970102

11. Marris E. Bacteria could be key to freeing South Pacific of mosquitoes. Nature. 2017; 548:17–8. pmid:28770863

12. O'Neill S, Ryan P, Turley A, Wilson G, Retzki K, Iturbe-Ormaetxe I, et al. Scaled deployment of Wolbachia to protect the community from Aedes transmitted arboviruses Gates Open Res. 2018;2(36). pmid:30596205

13. Flores HA, O’Neill SL. Controlling vector-borne diseases by releasing modified mosquitoes. Nat Rev Microbiol. 2018;16(8):508–18. pmid:29777177

14. Crawford JE, Clarke DW, Criswell V, Desnoyer M, Cornel D, Deegan B, et al. Efficient production of male Wolbachia-infected Aedes aegypti mosquitoes enables large-scale suppression of wild populations. Nat Biotechnol. 2020;38:482–92. Epub 06 April 2020. pmid:32265562

15. Staunton KM, Yeeles P, Nowrouzi S, Townsend M, Paton CJ, Trewin B, et al. Trap location and premises condition influences on Aedes aegypti (Diptera: Culicidae) catches using Biogents Sentinel traps during a ‘Rear and Release’ program: Implications for designing surveillance programs. J Med Entomol. 2019; Epub 28 Feb 2019.

16. Kröckel U, Rose A, Eiras ÁE, Geier M. New tools for surveillance of adult yellow fever mosquitoes: Comparison of trap catches with human landing rates in an urban environment. J Am Mosq Control Assoc. 2006;22(2):229–38. pmid:17019768

17. Williams CR, Long SA, Webb CE, Bitzhenner M, Geier M, Russell RC, et al. Aedes aegypti population sampling using BG-Sentinel traps in North Queensland Australia: Statistical considerations for trap deployment and sampling strategy. J Med Entomol. 2007;44(2):345–50. pmid:17427707

18. Ball TS, Ritchie SR. Evaluation of BG-Sentinel trap trapping efficacy for Aedes aegypti (Diptera: Culicidae) in a visually competitive environment. J Med Entomol. 2010;47(4):657–63. pmid:20695282

19. Ponlawat A, Khongtak P, Jaichapor B, Pongsiri A, Evans BP. Field evaluation of two commercial mosquito traps baited with different attractants and colored lights for malaria vector surveillance in Thailand. Parasit Vectors. 2017;10(1):378. pmid:28784149

20. Barrera R, Mackay AJ, Amador M. An improved trap to capture adult container-inhabiting mosquitoes. J Am Mosq Control Assoc. 2013;29(4):358–68. pmid:24551969

21. de Ázara TMF, Degener CM, Roque RA, Ohly JJ, Geier M, Eiras ÁE. The impact of CO2 on collection of Aedes aegypti (Linnaeus) and Culex quinquefasciatus Say by BG-Sentinel traps in Manaus, Brazil. Mem I Oswaldo Cruz. 2013;108(2):229–32. pmid:23579804.

22. Hemme RR, Thomas CL, Chadee DD, Severson DW. Influence of urban landscapes on population dynamics in a short-distance migrant mosquito: Evidence for the dengue vector Aedes aegypti. PLoS Negl Trop Dis. 2010;4(3):e634. pmid:20300516

23. Maciel-de-Freitas R, Lourenço-de-Oliveira R. Presumed unconstrained dispersal of Aedes aegypti in the city of Rio de Janeiro, Brazil. Rev Saude Publica. 2009;43:8–12.

24. Russell RC, Webb CE, Williams CR, Ritchie SA. Mark–release–recapture study to measure dispersal of the mosquito Aedes aegypti in Cairns, Queensland, Australia. Med Vet Entomol. 2005;19(4):451–7. pmid:16336310

25. Cornel AJ, Holeman J, Nieman CC, Lee Y, Smith C, Amorino M, et al. Surveillance, insecticide resistance and control of an invasive Aedes aegypti (Diptera: Culicidae) population in California. F1000Res. 2016;5. https://doi.org/10.12688/f1000research.8107.1.

26. Kay BH, Russell BM, Holt JS, Ryan PA, Foley PN, Lyons SA. The importance of subterranean mosquito habitat to arbovirus vector control strategies in north Queensland, Australia. J Med Entomol. 2000;37(6):846–53. pmid:11126539

27. Ritchie SA. Dengue vector bionomics: why Aedes aegypti is such a good vector. In: Gubler DJ, Ooi EE, Vasudevan S, Farrar J, editors. Dengue and Dengue Hemorrhagic Fever, 2nd Edition. Wallingford: CABI; 2014. p. 455–80.

28. Williams CR, Long SA, Russell RC, Ritchie SA. Field efficacy of the BG-Sentinel compared with CDC backpack aspirators and CO2-baited EVS traps for collection of adult Aedes aegypti in Cairns, Queensland, Australia. J Am Mosq Control Assoc. 2006;22(2):296–300. pmid:17019776

29. Hayden MH, Uejio CK, Walker K, Ramberg F, Moreno R, Rosales C, et al. Microclimate and human factors in the divergent ecology of Aedes aegypti along the Arizona, U.S./Sonora, MX border. EcoHealth. 2010;7(1):64–77. pmid:20232228

30. Tun-Lin W, Kay BH, Barnes A. The premise condition index: A tool for streamlining surveys of Aedes aegypti. Am J Trop Med Hyg. 1995;53(6):591–4. pmid:8561259

31. Bates D, Mächler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. 2015-10-07 ed2015.

32. Gelman A. Scaling regression inputs by dividing by two standard deviations. Stat Med. 2008;27(15):2865–73. pmid:17960576

33. Luštrik R, Stachelek J. AED: Package Accompanying 'Mixed Effects Models and Extensions in Ecology with R'. 2009. p. R package version 1.0.

34. Chandler RB, Royle JA, King DI. Inference about density and temporary emigration in unmarked populations. Ecology. 2011;92(7):1429–35. pmid:21870617

35. Hegyi G, Garamszegi LZ. Using information theory as a substitute for stepwise regression in ecology and behavior. Behav Ecol Sociobiol. 2011;65(1):69–76.

36. Hegyi G, Laczi M. Using full models, stepwise regression and model selection in ecological data sets: Monte Carlo simulations. Ann Zool Fennici. 2015;52(5–6):257–79.

37. McFadden D. Analysis of qualitative choice behavior. Zarembka P, editor. New York, NY: Academic Press.; 1974.

38. Barbosa AM, Brown JA, Jimenez-Valverde A, Real R. Model Evaluation and Analysis. 2016. p. R package version 2.1.

39. McFadden D. Quantitative methods for analysing travel behaviour of individuals: some recent developments. In: Hensher DA, Stopher PR, editors. Behaviour travel modelling. London: Croom Helm; 1979. p. 279–318.

40. Hoffmann AA, Goundar AA, Long SA, Johnson PH, Ritchie SA. Invasion of Wolbachia at the residential block level is associated with local abundance of Stegomyia aegypti, yellow fever mosquito, populations and property attributes. Med Vet Entomol. 2014;28(S1):90–7. pmid:25171611

41. Manrique-Saide P, Uc V, Prado C, Carmona C, Vadillo J, Chan R, et al. Storm sewers as larval habitats for Aedes aegypti and Culex spp. in a neighborhood of Merida, Mexico. J Am Mosq Control Assoc. 2012;28(3):255–7. pmid:23833907

42. Tyagi B, Hiriyan J. Breeding of dengue vector Aedes aegypti (Linnaeus) in rural Thar Desert, north-western Rajasthan, India. Dengue Bull. 2004;28:220–2.

43. Reiter P, Amador M, Colon N. Enhancement of the CDC ovitrap with hay infusion for daily monitoring of Aedes aegypti populations. J Am Mosq Control Assoc. 1991;7:52–5. pmid:2045808

44. Day JF. Mosquito oviposition behavior and vector control. Insects. 2016;7(4):65. pmid:27869724.

45. Dzul-Manzanilla F, Ibarra-López J, Bibiano Marín W, Martini-Jaimes A, Leyva JT, Correa-Morales F, et al. Indoor resting behavior of Aedes aegypti (Diptera: Culicidae) in Acapulco, Mexico. J Med Entomol. 2017;54(2):501–4. pmid:28011725

46. Salazar FV, Chareonviriyaphap T, Grieco JP, Eisen L, Prabaripai A, Ojo TA, et al. Influence of location and distance of Biogents Sentinel traps from human-occupied experimental huts on Aedes aegypti recapture and entry into huts. J Am Mosq Control Assoc. 2018;34(3):201–9. pmid:31442172