Four shades of red: optimizing the use of oviposition traps to monitor mosquito populations of arbovirus vectors in La Reunion

doi:10.21203/rs.3.rs-1980849/v1

Download PDF

Article

Four shades of red: optimizing the use of oviposition traps to monitor mosquito populations of arbovirus vectors in La Reunion

https://doi.org/10.21203/rs.3.rs-1980849/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Recently, several epidemics caused by mosquito-borne viruses - including dengue viruses (DV) have occurred in La Reunion and other Indian Ocean islands. Aedes albopictus and Ae. aegypti - two important DV vectors, co-occur in La Reunion. Aedes albopictus is the major DV vector, preponderant and widespread in the island. Aedes aegypti is localized in certain areas and shows a peculiar ecology, compared to other tropical populations of the same species. This study aimed at providing users (researchers, mosquito control operators, health managers) with a locally-validated oviposition traps (ovitraps) to monitor Aedes populations.

A field experiment was achieved in Saint-Joseph - where both mosquito species occur in sympatry, to assess the effect of different ovitrap settings on their detection probability and apparent density. Preliminary surveys indicated black plastic ovitraps were the best traps. Vacoa trees (Pandanus utilis) were the only observed breeding sites for Ae. aegypti, shared with Ae. albopictus. They were the experimental units in a design considering three factors: trap position in the trees (ground vs canopy), the oviposition surface used in the trap (blotting paper vs. vacoa leaf), and addition of organic matter (fish food) to the trap water.

The addition of organic matter to trap water was not an important factor in this study. On the ground, Ae. aegypti eggs were only found when vacoa leaves were used for oviposition. Their detection and apparent density increased greatly when ovitraps were located in the tree canopy. The main factor for Ae. albopictus was the oviposition surface, with a strong preference for blotting paper. However, with vacoa leaves, the detection was also close to 100%. Further laboratory studies indicated larvae survival was lower for a high egg density, combined with blotting paper as the oviposition surface.

When monitoring mixed Aedes populations in La Reunion, we recommend the use of black plastic ovitraps, filled with tap water, placed at 1.50-to-2.00-m high in vacoa trees, with vacoa leaves as the oviposition surface.

1.1 A changing epidemiology of mosquito-borne viruses

Dengue is the most widely distributed mosquito-borne viral infection. Dengue virus (DV) is primarily transmitted by Aedes mosquitoes, with an annual incidence of cases increasing from ca. 29 millions in 1990, to ca. 57 millions in 2019^[1]. DV is prevalent in 128 countries - mostly located in tropical regions, where rainfall, temperature, relative humidity and rapid unplanned urbanization favor DV mosquito vectors^[2–7].

Worldwide, Aedes aegypti is the most important and efficient DV vector, followed by Aedes albopictus^{[7, 8]}. Nevertheless, in several south-western Indian Ocean islands (SWOI) - especially in La Reunion, Ae. albopictus is considered as the main DV vector^[9]. Several dengue epidemics have been reported since 1977/78, when a large type-2 DV outbreak infected ca. 30% of the population^[10]. However, only 3 to 31 sporadic DV cases were reported yearly, until the occurrence of a new type-1 DV outbreak, with 228 reported cases^[11]. Since 2015, DV outbreaks yearly occurred^[12]. More than 16,000 cases were reported in 2020^[13]. At the end of July 2021 - i.e., at the end of DV transmission season, public-health authorities reported 27,213 confirmed dengue cases and 15 deaths as of January 2021^[14]. The consecutive circulation of several DV serotypes is accompanied by an increased risk of clinical severity, a possible sign of DV endemisation^[15].

Besides DV, Aedes albopictus is also the main vector of the chikungunya virus (CV). In July 2004, CV firstly spread from outbreaks on the East African coast to the Comoros Archipelagos, and all the other SWIO islands thereafter. In La Réunion, CV transmission was intense, with 300,000 cases recorded during a first wave (2005), and more than 250,000 cases during a second wave (2006), with more than 200 deaths, despite much efforts devoted to the control of vector populations^{[16, 17]}.

1.2 Aedes aegypti in La Reunion: uncovering peculiarities

Soghigian et al.^[18] found the common ancestor Ae. aegypti s.s., after diverging from Ae. mascarensis seven million years ago, colonized the African continent less than 85,000 years ago, long after it had colonized La Reunion and other SWIO islands, including Madagascar. Kotsakiozi et al.^[19] reported the Ae. aegypti strain found on La Reunion was genetically unique, and close to Ae. mascarensis. Thus, Ae. aegypti mosquitoes found nowadays in this island might have diverged from Ae. mascarensis populations. This last species is absent from La Reunion, but endemic to Mauritius^[20]. Indeed, Ae. aegypti mosquitoes described in La Reunion, are peculiar. They show the morphological traits of Ae. aegypti aegypti, sharing ecological traits with Ae. aegypti formosus. Both mosquitoes are considered as subspecies of Ae. aegypti s.s. spreading on the African continent less than 1,000 years ago^[18]. In La Reunion, the distribution of Ae. aegypti mosquitoes seems to be restricted to a few rural areas^[21]. This tiny distribution is attributed to the competition with Ae. albopictus mosquitoes, which are predominant in urban areas^[22]. The same authors also identified evidences of introgression events from Asian strains of Ae. aegypti aegypti, which were probably moped up from urban environments during DDT campaigns against malaria vectors^[18].

Since the first description of Ae. aegypti in Saint-Paul, La Reunion in 1907^[23], little information on its preferred development sites were obtained, unlike Ae. albopictus. The latter shows a strong ecological plasticity. It is present on almost all along the island’s coastline, up to an altitude of 1,200 m (800 m during the dry season), both in urban and rural sites. It uses natural and anthropogenic water collections such as rock holes and saucers under flower pots as larval sites^{[21, 24]}. Following the discovery of Ae. aegypti in 1907, a discontinuous distribution up to 600 m altitude, especially in the western part of the island, was observed in the 1950’s^[25]. Thereafter, observations of Ae. aegypti mosquitoes were not reported for more than 30 years: the population seemed to have disappeared from La Reunion. It was actually restricted to a few natural habitats,^[21] such as rock holes that it shares with Ae. albopictus and is never found in artificial sites^[22], unlike the urban pan-tropical Ae. aegypti aegypti. In La Reunion, the distribution of Ae. aegypti remains poorly known. Individuals of this species are rarely caught by the oviposition trap network implemented by the Regional Health Agency (ARS) to monitor Ae. albopictus populations. Thus, only a few Ae. aegypti eggs - vs. thousands of Ae. albopictus eggs, were collected by the ARS ovitrap network in the months preceding our study: 5 vs. 4,063 in 15 ovitraps set in our study site from 28 July 2020, to 6 October 2020, G. Dupuy, unpublished data).

This study was the first step of a research project aiming at providing methods and tools for controlling Aedes mosquito populations in La Reunion, with the boosted Sterile Insect Technique (SIT). This mosquito-control strategy is derived from classical SIT, which consists in large-scale release of sterile male mosquitoes to reduce wild mosquito populations. Mosquitoes are mass-reared in the laboratory, sexed, and then sterilized through a mixture of X- and \(\gamma\)-ray irradiation before being released in the wild. Wild virgin females mate with sterile males and therefore have no offspring. In boosted SIT, sterile males are coated with a biocide before release. They are used as carriers of the biocide to females and larval sites. The boosted SIT approach may allow considerable cost savings by reducing the numbers of sterile males to be released, as well as the frequency of releases^{[26, 27]}.

A pilot trial of boosted SIT was conducted in 2021 against Ae. aegypti in Saint-Joseph in the south of La Reunion. An accurate knowledge of the ecology and phenology of target mosquito populations is needed before starting any SIT-based control program.^[28] For instance, egg sterility - combined with the ratio of released sterile to wild males (available from the outcome of adult traps, such as CO₂baited BG traps), allow assessing the competitiveness of sterile vs. wild males^[29]. In addition, competitive release - i.e., better population dynamics, is expected for Ae. albopictus following the population reduction of its competitor, i.e. occupying the same breeding sites. Should such changes occur in the relative dynamics of these populations, they would only be detected if an efficient trapping method provided accurate apparent densities estimates for both species. Thus, it was necessary to assess and optimize the sensitivity and efficiency of oviposition traps which are widely used to assess the density of eggs laid down by gravid Aedes mosquito females^{[30, 31]}. Preliminary observations in the study area revealed rock holes, human-created micro-habitats, and tree holes exclusively hosted Ae. albopictus larvae. Aedes aegypti larvae were only found in water collections at the inner basis of vacoa-tree leaves (Pandanus utilis) (Fig. 1a, 1b, and 1c, and fig. S3), together with Ae. albopictus larvae.

1.3 Goals

The aim of this field experiment was to optimize oviposition trap settings - including height position, baiting scheme and oviposition surface, for the detection of Ae. aegypti mosquito population, and assessment of their density.

The main question addressed in this field experiment was:

What it the best setting for the ovitrap - including for survival of immature stages from hatching eggs up to the fourth larval instar (L4) occurring during the laboratory-rearing step, to maximize the probability to detect Ae. aegypti mosquitoes - and to provide accurate apparent density estimates for Ae. aegypti and Ae. albopictus populations ?

The study was conducted in the municipality of Saint-Joseph, in the south of La Reunion. Two sites were selected after preliminary investigations to assess the presence of well-established Ae. aegypti and Ae. albopictus mosquito populations:

a 10-ha tree orchard located along the Langevin River, planted with vacoa trees (Pandanus utilis), coconut trees and other indigenous or endemic trees (latitude 21°23′04.2″S, and longitude 55°38′45.3″E),
an isolated “ravine” (i.e, a narrow valley) with a riparian forest, covering ca. 2 ha, lying between the main road and the Indian Ocean (latitude 21°22′48.2″S, and longitude 55°39′34.7″E) in the Vincendo district of St Joseph municipality.

1.4 Study design

Black plastic ovitraps were used in the field experiment (Fig. 1b): Ae. aegypti female mosquitoes prefer to lay their eggs in black containers.^[30,32−34] Each ovitrap was filled with 250 mL of tap water. A Latin square design was adopted to control the effect of external factors possibly influencing the outcome: air temperature, relative humidity, wind speed and direction, habitat suitability for feeding, breeding, and resting. At each site, two well-developed vacoa trees were selected and kept all along the experiment. Three bimodal factors were crossed on each tree (i.e., eight traps / tree) during a given trapping session:

1. Height position (covariate “pos”) in the vacoa tree: ground level (“ground”) vs. tree-canopy level (“canopy”): between 1.5 and 2.0 m high,

2. Oviposition surface (covariate “surf”): blotting paper strip, innocuity assessed for Aedes eggs (“paper”) vs. piece of vacoa leaf (“leaf”): see Fig. 1d,

3. Organic matter (covariate “om”): addition (“added”), or not (“none”) to trap water. We used commercial fish food (JBL™ Novo Cuppy).

With two vacoa trees per site, a total of 32 outcomes were recorded on each trapping session: two sites * two trees / site * eight traps / tree. Each ovitrap had four possible (surf, om) combinations. Thus, four trapping sessions were needed to complete a replicate, i.e., 128 outcomes / replicate. Each replicate lasted two weeks. Two replicates were implemented in this experiment, for a total of 256 outcomes.

The study spanned from November 30, 2020 to January 14, 2021. A total of 11,826 eggs were collected from 256 trapping sessions. Identified mosquito L4 summed up to 6,992 (i.e., overall survival = 6992 / 11826 \(\simeq\) 59.1%), finally splitting into 975 Ae. aegypti, and 6,017 Ae. albopictus larvae.

2.1 Optimization of ovitrap settings in the field

2.1.1 Overview

A graphical presentation of the main results is provided in Fig. 2 (detection probability) and Fig. 3 (apparent density): population means and pairwise tests for differences in population means (H₀: no difference in population means). The apparent-density estimates are shown in tables S1 (Ae. aegypti) and S2 (Ae. albopictus). Estimates of the MMA model coefficients are available in tables S3 (Ae. aegypti) and S4 (Ae. albopictus).

2.1.2 Detection probability

• Aedes aegypti mosquitoes

The most important variable in the multi-model averaging (MMA) hurdle model - combining detection probability and apparent density, was the trap position in vacoa trees (fig. S1). It was selected in 100% of the plausible models kept for the MMA, with each of the 1,000 resampled datasets. Its coefficient - in the detection probability sub-model, was strongly negative and highly significant (table S3: -3.38 for the position on the ground, P < 0.001). The two other covariates were less important. Their coefficients were small and not significant (P > 0.05, fig. S1, table S3).

The detection probability was low when the ovitraps were set on the ground, irrespective of the addition of organic matter to the trap water (Fig. 2a, table S1):

• the estimate was close to 0 when the oviposition surface was a strip of blotting paper (Fig. 2a). This configuration (ovitrap on the ground + blotting paper strip as the oviposition surface) should not be used to monitor Ae aegypti populations in La Reunion.

• The estimate was still less than 20% when the oviposition surface was a piece of vacoa leaf (Fig. 2a).

Symmetrically, the highest detection probability (Fig. 2a: ca. 0.8) was observed for ovitraps lying in the canopy of vacoa trees, with similar values (Fig. 2a: low differences, Fig. 2b: P > 0.05) when the oviposition surface was a blotting paper strip, or a piece of vacoa leaf.

The predominance of height position in vacoa trees - over the two other covariates, is highlighted in Fig. 2b: the differences in detection probability between the tree canopy and the ground level were large (from 50–80%), and highly significant (P < 0.001), whatever the combination with the other covariates. Hence, the cell block crossing canopy with ground categories is uniformly colored in dark red (P < 001). No other differences than between traps in tree canopy, and traps on the ground, were significant (P > 0.05).

• Aedes albopictus mosquitoes

The most important variable in the MMA hurdle model was the oviposition surface (fig. S1). It was selected in 100% of the plausible models kept for MMA, with each of the 1,000 resampled datasets. Its coefficient - in the detection probability sub-model, was strongly positive and highly significant (table S3: 1.597 for the blotting paper strip coefficient, P = 0.001). The two other covariates were less important. Their coefficients were small and not significant (P > 0.05, fig. S1, table S3).

The detection probability of Ae. albopictus mosquitoes was always high, whatever the ovitrap setting. It ranged from ca. 70% for ovitraps lying on the ground, with a piece of vacoa leaf as the oviposition surface, and organic matter added to trap water), to nearly 100% when the ovitraps were set on the canopy of vacoa trees (irrespective of the type of oviposition surface), or when they were set on the ground, with a blotting paper strip as the oviposition surface (Fig. 2c).

For Ae. albopictus, the effect of oviposition surface on the detection probability was lower than the effect of height position on the same outcome for Ae. aegypti, hence the “holes” (the lightest of the four shades of red) in the cell block crossing columns for the “paper”, with rows for the “leaf” oviposition surfaces (Fig. 2d) - to compare with the uniform aspect of the corresponding block in Fig. 2b (“canopy” vs. “ground” height positions): all dark-red cells (P < 0.001).

2.1.3 Apparent density

The effects of covariates on the apparent density (Fig. 3) were quite similar to those reported for the detection: importance of the height position of ovitraps for Ae. aegypti, vs. importance of oviposition surface for Ae. albopictus (fig. S1).

The estimated apparent density (Fig. 3, and tables S1 for Ae. aegypti, and S2 for Ae. albopictus) reinforced the detection results.

Setting ovitraps on the ground did not produce robust estimates for Ae aegypti mosquito populations - whatever the other settings. Thus, the estimated apparent density (table S1) ranged from 0.0 to 1.1 mosq. / trap / day (blotting paper as the oviposition surface, without or with organic matter in trap water), and from 0.3 to 0.4 mosq. / trap / day (vacoa leaf as the oviposition surface, without or with organic matter in trap water). In comparison, the estimates for Ae. albopictus (table S2) ranged from 12.7 to 15.1 mosq. / trap / day (vacoa leaf as the oviposition surface, without or with organic matter in trap water), and from 43.4 to 53.4 mosq. / trap / day (blotting paper as the oviposition surface, without or with organic matter in trap water).

For Ae aegypti, setting ovitraps on vacoa canopy much increased the estimates - and very significantly (P < 0.001), w.r.t. the ground level (Fig. 3a and 3c, table S1).

For Ae. albopictus, the highest estimates were obtained at the canopy level with blotting paper as the oviposition surface, and organic matter in trap water: 57 mosq. / trap / day, with 95% credible interval (CI) [39, 69]. These estimates were similar to those obtained at the ground level - or at the canopy level, with blotting paper as the oviposition surface, without or with organic matter in trap water. Using blotting paper as the oviposition surface was thus the main factor for Ae. albopictus apparent density: all pairwise differences were strong and highly significant (P < 0.001) for blotting-paper vs. vacoa-leaf comparisons (Fig. 3b and 3d, table S2).

2.2 Proportion of Aedes aegypti mosquitoes in L4

A specific analysis was done to select the best ovitrap setting to obtain balanced densities of Ae. aegypti and Ae. albopictus (where both populations are present). The previous step of the analysis provided evidences Ae.aegypti rarely goes to the ground. Therefore, we discarded data collected at the ground level, and estimated the proportions of Ae. aegypti vs. the oviposition surface, and the addition of organic matter in the trap water. Figure 4a and 4b show the addition of organic matter had little and non significant (P > 0.05) influence on the Ae. aegypti proportion.

With a blotting paper strip as the oviposition surface, the proportion always remained low: 14% [9, 20], for traps without, and 20% [14, 28] for traps with, organic matter added to the trap water. Conversely, using a piece of vacoa leaf as the oviposition surface provided more balanced estimates: 40% [27, 53] without organic matter, and 38% [27, 50] with organic matter.

2.3 L4 survival during the laboratory-rearing step

Figure 5 gives an overview of the available data (232 trapping outcome). L4 survival is represented by the colored level plot: the darker the plot regions, the lower the survival. The variation in observed survival was huge, ranging from 0.12 to 1.00 (mean = 0.68, median = 0.70). The lowest survival occurred in the right panel (eggs collected on blotting paper strips as the oviposition surface), more precisely in the upper right region, where the egg density (\(E\) ranging from 2 to 619, mean = 69, median = 41) was high, as was the proportion of Ae. albopictus in L4 (\(b\) ranging from 0 to 1, mean = 0.84, median = 1.00).

In the MMA beta-binomial logistic regression model, the most important covariate (fig. S2) was the oviposition surface (present in ca. 90% of the selected models), followed by the logarithm of egg density (ca. 65%), and the proportion of Ae. albopictus in L4 (ca. 55%). The interaction term between the proportion of Ae. albopictus and the logarithm of egg density was rarely selected (20%).

The estimated model coefficients are shown in table S5. Besides model intercept, the only significant coefficient was the oviposition surface (P = 0.003). Its value for the blotting paper strip category was negative (-0.57), i. e., a lower survival was observed for L4 larvae emerging from eggs collected on blotting paper strips, holding constant the other covariates.

To facilitate model interpretation, and to allow comparisons of interesting population means, we built a grid of predicted values, taking care to predict within the space defined by the observed covariates (Fig. 6).

The only significant differences were found (Fig. 6a and 6b for the survival, and 6c for the P values) between the patterns surf = "leaf" + logE = 2 + b = 1 (survival \(\simeq\) 0.8) on the one hand, and surf = "paper" + logE = 4.5 + b = 0.35 (survival \(\simeq\) 0.58), or b = 1 (survival \(\simeq\) 0.60), on the other hand.

In conclusion, L4 survival was lower when eggs were collected on a blotting paper strip (vs. a piece of vacoa leaf). However, the difference in survival was only significant at high egg densities.

At this stage, we can fully answer the question asked:

What it the best setting for ovitraps - including for the L4 laboratory-rearing step, to maximize the probability to detect Ae. aegypti mosquitoes - and provide balanced apparent density estimates for Ae. aegypti and Ae. albopictus populations ?

1. Place the ovitraps in vacoa trees at man’height, not on the ground, to allow the detection of Ae. aegypti

2. Use a piece of vacoa leaf as the oviposition surface, not a blotting paper strip, to provide balanced estimates of apparent densities of both Ae. aegypti - and Ae. albopictus, and avoid falsely high egg density associated with density dependent mortality.

3.1 Trapping device and baiting

The efficiency of black plastic pots as ovitraps was demonstrated once again in our study. They are commonly used by the public-health and research agencies in La Reunion. Yap et al.^[32] found the black color of pots was attractive to Ae. albopictus.

The amount of water filling our oviposition traps, reached about 1/3 of the total height of the trap. It was very attractive for Ae. albopictus. Delatte et al.^[24] reported an average depth of 6 cm of clear water was an optimal filling for Ae. albopictus. Only tap water was used throughout our study. It was the most efficient to collect eggs from Ae. albopictus in another study^[33].

3.2 Organic matter in trap water

No strong preference was found for the trap location, or the addition of organic matter. This can be considered as an adaptive behaviour of females, that select various larval habitats relying on the higher competitiveness of their larvae. The same findings were made by Yoshioka et al.^[35], along with the evidence that the presence and density of con-specific larvae were the most important factors influencing site selection in a medium-to-low diet gradient.

The lack of effect of organic matter on the detection probability and apparent density of Ae. aegypti L4 was also reported by Wong et al.^[36] for which Ae. aegypti density was increased by the presence of con-specific larvae and pupae in the oviposition device, its size, and its exposure to sunlight.

Conversely, Hawley^[37] reported the availability of organic matter and food was essential for the survival of Ae. albopictus larvae and for their adult size, all the more in natural vs. artificial containers. Females were likely to choose oviposition sites loaded with organic matter. Delatte et al.^{[24, 38]} observed oviposition sites presenting organic matter (medium to high levels) in shaded locations provided ideal habitats for larval and pupal development of Ae. albopictus. In addition, Daugherty et al.^[39] found invertebrate carcasses favored larval development. Nevertheless, we did not find strong differences in apparent densities with the addition of organic matter (fish food: by-products of fishes, mollusks and crustaceans) at a standardized dose. It might be interesting to increase the doses of organic matter to test possible increase in attractiveness.

3.3 Trap position

Female mosquitoes select specific and appropriate habitats to lay their eggs, so that resource-mediated interactions are minimized, and offspring health is preserved^[40–42]. In this study, female Ae. aegypti and Ae. albopictus had contrasted oviposition behaviors. The ovitrap height position - at the ground level or in the vacoa canopy, had a strong and positive effect on the probability to detect Ae. aegypti L4, as well as on the estimate of their apparent density. We are not aware of previous mentions of Ae. aegypti oviposition preferences for vacoa trees in La Reunion, as we observed during preliminary screening of the larval habitats (fig. S3 and S4).

The strong preference of local Ae. aegypti populations for vacoa trees - and other natural oviposition sites, is probably an adaptation to their environment. As a matter of fact, water resources available at the basis of vacoa leaves are low (Fig. 1c). Consequently, so are food resources and larval charge capacity associated with this breeding site. Local strains of Ae. aegypti have been co-evolving much longer with vacoa trees – an endemic tree species from Mascareign islands, much longer and may thus be better adapted than Ae. albopictus mosquitoes which were only recently introduced in La Reunion, originating from south-eastern Asia^[21]. Therefore, they can compete with Ae. albopictus in this niche^[22]. This is consistent with higher Ae aegypti density found in vacoa tree canopy, where breeding sites are smaller, and more prone to desiccation, than at the ground level. Indeed, previous studies reported local Ae. aegypti mosquitoes were more resistant to desiccation and high temperature than Ae. albopictus: a drier environment favors them.^{[43, 44]} Although this knowledge seems to be new to scientists, local inhabitants are well aware of the abundance of Aedes larvae in vacoa trees, as observed during the communication campaign of this study. Managing vacoa trees in private properties might thus be considered as part of the integrated control of Aedes mosquitoes in La Reunion, taking into account the importance of these trees for other endemic species like the Manapany day gecko Phelsuma inexpectata .

3.4 Oviposition surface and egg density

The proportion of L4 Ae. aegypti emerging from eggs collected in ovitraps depended on (i) the oviposition surface, and (ii) egg density. Ovitraps placed in vacoa trees - with vacoa leaves as the oviposition surface, were the most favorable to a high proportion (ca. 40% in our study). The actual egg ratio of Ae. aegypti was likely under-estimated by the proportion of Ae. aegypti in L4 because of the higher mortality affecting their larvae when competing with Ae. albopictus larvae.

The better larvae survival found for oviposition surface with lower egg density is not surprising. Braks et al.^[45] and Moore & Fisher^[46] showed the survival of Ae. albopictus and Ae. aegypti mosquitoes decreased as their density increased - either for single or mixed populations. Besides the direct depredation of Ae. aegypti larvae by Ae. albopictus, we also observed - during the laboratory-rearing step, the higher the egg density, the longer the larval development time, and the smaller the larvae size at the L4 stage, and the higher the mortality. This is consistent with the better survival observed with the vacoa-leaf oviposition surface, because of the lower egg density associated with this surface. However, results of the beta-binomial logistic regression analysis (table S5) provided an evidence of a negative effect of the blotting paper on L4 survival, independently of the egg density (P = 0.002). This might be due to the direct, positive effects of specific compounds released in the trap water by vacoa leaves on Aedes larvae, or indirect effects through a microbiota-mediated impact. Further research would be needed to conclude.

3.5 Ecology and DV transmission

Females of Ae. albopictus mosquitoes lay their eggs in two types of habitat: natural sites, and artificial sites.^[21] The two sites where we conducted our study were both in urban or peri-urban areas, the houses being only a few meters away from the nearest oviposition traps. Domestic and peri-domestic potential breeding sites, such as wasted containers, flower cups, or used tires, are favorable oviposition sites for Ae. albopictus. Peri-domestic and natural sites represent more random habitats, as they are more influenced by seasonal changes in climate and rainfall than domestic sites whose water supply is more dependent on anthropogenic activities^[21]. Thus, in difficult conditions for Ae. albopictus, its competitive advantage over Ae. aegypti is mostly abolished. After the initial decline in the distribution of Ae. aegypti following the introduction of Ae. albopictus in La Reunion and the subsequent changes affecting the island (urbanization, anthropization of natural environments, malaria vector control campaigns), it seems likely that Ae. aegypti populations were able to survive in places where such natural larval habitats are present.

In a field experiment to compare different baiting schemes for adult traps - implemented in the same sites and vector season, we reported the same order of magnitude for the apparent densities of these populations, when adult traps were baited with CO₂^[47], or when we set the ovitraps in vacoa canopy, with vacoa leaves for oviposition. The two studies were done - by chance, during a large DV outbreak which affected Saint-Joseph and many other municipalities. Therefore, Ae. aegypti mosquitoes might play a role in DV transmission in such ecosystems.

Aedes mosquitoes show a great ecological plasticity, and have a global distribution covering different ecosystems, at different altitudes and latitudes, and under a large range of environmental conditions. Our results extend the current knowledge on the ecology of Ae. aegypti and Ae. albopictus, by quantifying the preferences of female mosquitoes for three oviposition parameters.

This study of oviposition preferences, as well as the determination of the biological and ecological factors regulating the natural Aedes populations in La Reunion, will be useful for the development of sound monitoring and control strategies, especially for genetic control methods. It also shows local knowledge on Aedes ecology may be accounted for when designing integrated control strategies, which will in turn facilitate their appropriation by local populations

In addition to the specific results presented here, a particular attention was paid to the methodological aspects of monitoring mosquito population - particularly data analysis. Methods and tools for monitoring mosquito populations - like this ovitrap assessment, are selected and co-developed together with local and regional public-health practitioners, to identify and address their actual needs. In turn, regional meetings and training courses are being organized and will be extended as new results are available.

Indeed, past events highlighted the utmost importance of being able and ready to prevent, detect, mitigate, and monitor disease outbreaks linked to Aedes mosquito vectors, such as the DV - as well as chikungunya, Zika, or any other / new emerging mosquito-borne virus.

5.1 Biological data

5.1.1 Harvesting and processing mosquito eggs

Each day, all oviposition surfaces were collected from the ovitraps. In addition, their water was filtered to collect possible Ae. aegypti eggs as these mosquitoes also lay their eggs on the water surface of breeding sites^[48].

Collected oviposition surfaces were placed in a controlled-atmosphere chamber (air temperature: 28 ± 2°C, relative humidity: 70 ± 10% RH, and photo-period: 12:12 h (light:dark) at Cirad laboratory, Saint-Pierre (La Reunion) for one week, to dry the oviposition surfaces.

5.1.2 Rearing mosquito larvae

Eggs were counted and placed in 400-mL plastic containers covered with mosquito netting, and filled with 250 mL of tap water. Larvae were grown in the same greenhouse. They were fed with crushed dry cat food (Friskies Adults™)^[49]. As they grew, they were fed with an increasing amount of food. This amount was kept low to avoid spurious bacterial growth - a possible cause of larval death^[30]. The amount of food was computed on the basis of four mg / larvae throughout the rearing stage, with 10% at d₀ + 1, 45% at d₀ + 2, and 45% at d₀ + 4. Larvae reaching the fourth instar at 4–5 days were stored in a freezer (+ 4°C), identified using an ad-hoc procedure^[50], and counted using a binocular. All collected data were compiled in a database.

5.2 Data analysis

Three questions were addressed in this field experiment - and consecutively, three different datasets were defined:

1. What is the best trapping setting for ovitraps to maximize the probability to detect Ae. aegypti (and Ae. albopictus as well), and assess its apparent density ?

2. What is the best trapping setting for ovitraps to provide accurate estimates for the proportion of mosquitoes from both species ?

3. What is the best laboratory-rearing setting to maximize the probability of survival of immature-stage up to the fourth instar (L4 larvae) ?

5.2.1 Question-specific models

• Detection probability and apparent density

We chose a hurdle model^[51] - made of two sub-models which were jointly fitted: (i) a logistic Bernoulli model for the presence data (detection probability), and a zero-truncated negative binomial model for the apparent-density counts. The same covariates - defined above, were used to model the detection probability and the apparent density.

• Proportion of Ae. aegypti in L4

The previous steps of the analysis provided evidences that Ae.aegypti rarely goes to the ground. Therefore, we discarded data collected at the ground level, and compared the proportions of Ae. aegypti according to the oviposition surface, and the addition of organic matter in trap water (covariates). We did not use models: we used the proportion of Ae. aegypti in the original dataset as an estimate of population means, and used the procedure described below to estimate its credible intervals and test proportion differences between groups defined by the covariates.

• L4 survival at the laboratory-rearing stage

The survival was the ratio \(L⁄E\), where \(E\) was the number of collected eggs, and \(L\) the overall number of surviving L4 on the day the were counted. We discarded records with \(E=0\). We considered three covariates (1) the proportion b of Ae. albopictus in the final L4 count (ranging from 0 to 1), included to assess the difference in survival between Ae. albopictus and Ae. aegypti, (2) the logarithm of the egg count (logE, ranging from 0.69 for \(E=2\), to 6.43 for \(E=619\))) to assess possible density-dependence effect on larvae mortality; (3) the oviposition surface (surf: blotting paper strip vs. piece of vacoa leaf), to assess the effect of oviposition surface. We used a beta-binomial logistic regression model to analyze survival.

5.3 Estimates, credible intervals, and tests

5.3.1 Population means

In the analyzed datasets, the outcome (L4 count, or proportion) showed a large variance. Beyond the usual overdispersion - with respect to Poisson or binomial distributions, met in many studies using insect counts^[52], this large variance was related to the peculiarities of Aedes mosquitoes ecology. At the single exception of one count in the apparent density dataset which was spuriously influential in model coefficients - and thus, predicted population means (\(m=45\) Ae. aegypti L4, ovitrap on the ground), we did not discard the corresponding data because they are an important feature of Aedes ecology. Therefore, we did not expect to select a single best model adapted to an unrealistic “average” situation.

Instead, we adopted a multi-model averaging (MMA) approach, which is well suited when there is not a single plausible model with respect to the available data.^[53] A “full” model was defined, representing the most complex structure expected in the dataset. For the detection probability, as well as for the apparent density datasets, this was the additive model with the three covariates studied (oviposition surface, height position in the vacoa tree, and addition of organic matter). For the survival of L4, the full model had four terms: the oviposition surface, the logarithm of the initial egg count, the proportion of Ae. albopictus mosquitoes in the final L4-larvae count, and the interaction between these last two covariates.

From each full model, the set of all possible sub-models was defined. For the survival model, when the interaction term was considered, we also included the main effects involved in the interaction. For each dataset, all the models were fitted using a maximum-likelihood method. Then, they were ranked using AICc, the small-sample version of the Akaike information criterion^[54]. The AICc difference between consecutive ranked models was computed, and then divided by the AICc difference between the first and last models. This quantity is the Akaike weight^[55]. It was used as an indicator of relative model plausibility, and to compute the weighted means of the fitted values from each compared model, i.e., multi-model averaging^[55].

5.3.2 Credible intervals and tests

To estimate the 95% credible intervals of the statistics of interest (differences between population means), we used a bootstrap resampling procedure of the observed dataset.^[56] The combination of sites and dates (two sites, and four dates by replicate, and three replicates) was taken as the 24 resampling units of four trapping sessions each (i.e., the four dates). A random sample of these units was drawn with replacement in the initial data set, with the same size in sampling units, and 1,000 replicates. It was used to compute the simulated means and their differences.

• The 2.5% and 97.5% quantiles of the empirical distribution of each simulated mean were used as the 95% credible interval of population means.

• To assess the significance of differences in population means corresponding to ovitrap settings, or rearing conditions, we computed them from the original dataset. The null hypothesis to be tested for each difference was its credible interval included zero. To implement the test, we computed each difference for the series of bootstrapped datasets. Then, using a basic search procedure, we looked for the smallest \(\alpha\) value for each difference (series of B values) - starting from \(\alpha =0\), so that the credible interval defined by the two quantiles \(\alpha\)/2 and 1-\(\alpha\)/2 did not include zero (actually, we tested whether the product of these quantiles was > 0).

5.3.3 P-value adjustment for multiple comparisons, and visualization

In both analyses (detection / apparent density, and survival), we compared height population means. Therefore, we had to assess 8 * 7 / 2 = 28 differences, thus encountering the issue of multiple comparisons using the same sample - possibly leading to spuriously small P values and drawing erroneous conclusion (i.e., rejecting the null hypothesis too often). To address this situation, we adopted the Holm procedure to get P values adjusted for multiple comparisons.^[57]

To facilitate the interpretation of all these P values - and the identification of possible specific ecological patterns, we plotted them using level plots of P-value matrices. Each cell of the matrix contained a probability code for the P value category associated with a difference \(\varDelta\) (column j - row i) between the population mean indexed by column j, and the population mean indexed by row i. The color associated with \({\varDelta }_{i,j}\) was defined by two criteria:

1. the sign of \({\varDelta }_{i,j}\): red if \({\varDelta }_{i,j}>0\), blue if \({\varDelta }_{i,j}<0\);

2. a code (1, 2, 3, or 4) for the P value - defined by the following probability breaks: 0, 0.001, 0.01, 0.05, and 1. The code was associated with a shade of red or blue - according to the sign of \({\varDelta }_{i,j}\): the smaller the code (i.e., the smaller the P value), the darker the shade. Because the matrix was symmetric, with a diagonal of zero’s, we only represented its upper side (above the diagonal), omitting the diagonal.

In addition, matrix columns and rows were ordered according to the importance of covariates involved in the tested differences, to facilitate interpretation of the results.

The R software and computing environment^[58] was used for data analysis. In addition, we used add-on packages for the most specialized features: pscl for hurdle models,^[51] aods3 for beta-binomial logistic regression^[59], and MuMIn for multi-model averaging and inference^[60], implementing the information-theoretical approach and methods described in Burnham & Anderson^[55].

Data availability

Data - and R code needed to run data analysis, are available in a zip file (ovitrao.zip) found in the Supplementary Information material. The master R code is gathered in an R Markdown document^[61] located at the root of the unzipped file. Data are available as a csv file located in the sub-folder called data. Most of the statistical work is done by R functions automatically sourced in the running R session from an external file called functions.R. The list of add-on R packages needed to run the data analysis, - code to install and update them, are available in the file packages.R, also sourced in the R session.

Acknowledgments

This research work was partially funded by the European Union, (i) European Research Council, Grant Agreement No. 682387-REVOLINC, and (ii) Horizon 2020 program, Grant Agreement No. 874850-MOOD, both in the framework of the One Health Indian Ocean partnership network (www.onehealth-oi.org). It is cataloged as MOOD-048. The contents of this document are the sole responsibility of the authors and don’t necessarily reflect the views of the European Commission.

The Agence Régionale de Santé - Réunion (ARS-Réunion) - in particular its vector-control department, was strongly committed to defining the goals of this study, and its design. We are much grateful to Pr. Xavier Deparis, Head of ARS-Réunion, and Dr Nausicaa Habchi-Hanriot, vector-control department.

We thank Dr. Marc J.B. Vreysen, Head of the FAO-IAEA Insect Pest Control Laboratory (Vienna, Austria), for his comments on the draft version of this paper.

Author contributions

RB and IC equally contributed to the work.
JB is the corresponding author.
JB, TB, LCG, GD designed the study.
JB supervised the study.
RB, IC, MD, JB made the field and laboratory work.
RB, IC, RL, JB analyzed the data.
RB, IC, JB, RL wrote the paper.
RB, IC, RL, LCG, MD, GD, TB revised the draft versions of the paper.

Competing interests

The author(s) declare no competing interest.

Yang, X., Quam, M. B., Zhang, T. & Sang, S. Global burden for dengue and the evolving pattern in the past 30 years. Journal of travel medicine 28, taab146 (2021).
Simmons, C. P., Farrar, J. J., Vinh Chau, N. van & Wills, B. Dengue. New England Journal of Medicine 366, 1423–1432 (2012).
Brady, O. J. et al. Refining the global spatial limits of dengue virus transmission by evidence-based consensus. PLoS Neglected Tropical Diseases 6, e1760 (2012).
Morin, C. W., Comrie, A. C. & Ernst, K. Climate and dengue transmission: Evidence and implications. Environmental Health Perspectives 121, 1264–1272 (2013).
Brady, O. J. et al. Global temperature constraints on Aedes aegypti and Ae. albopictus persistence and competence for dengue virus transmission. Parasites Vectors 7, 338 (2014).
Betanzos-Reyes, Á. F. et al. Association of dengue fever with Aedes spp. abundance and climatological effects. Salud Pública de México 60, 12 (2017).
WHO. Dengue and severe dengue. (2022).
Gubler, D. J. Dengue and dengue hemorrhagic fever. Clinical Microbiology Reviews 11, 480–496 (1998).
Delatte, H. et al. Aedes albopictus, vecteur des virus du chikungunya et de la dengue à La Réunion : biologie et contrôle. Parasite 15, 3–13 (2008).
Kles, V., Michault, A., Rodhain, F., Mevel, F. & Chastel, C. A serological survey regarding Flaviviridae infections on the island of Reunion (1971-1989). Bulletin de la Societe de Pathologie Exotique (1990) 87, 71–76 (1994).
Pierre, V. et al. Epidémie de dengue 1 à la Réunion en 2004. Journal de Veille Sanitaire (2005).
Vincent, M. et al. From the threat to the large outbreak: dengue on Reunion Island, 2015 to 2018. Eurosurveillance 24, (2019).
Cellule Santé Publique France en Région, ARS. Situation de la dengue à La Réunion au 15 décembre 2020. https://www.lareunion.ars.sante.fr/avec-le-retour-de-lete-agissons-des-maintenant-contre-la-dengue (2020).
Agence Régionale de Santé. Communiqué de presse: dengue à La Réunion. Situation au 28 juillet 2021. https://www.lareunion.ars.sante.fr/system/files/2021-07/2021-07-28-Dengue-Situation à La Réunion_0.pdf (2021).
Hafsia, S. et al. Overview of dengue outbreaks in the southwestern Indian Ocean and analysis of factors involved in the shift toward endemicity in Reunion Island: A systematic review. PLOS Neglected Tropical Diseases 16, e0010547 (2022).
Soumahoro, M.-K. et al. The Chikungunya epidemic on La Réunion Island in 2005-2006: a cost-of-illness study. PLoS Neglected Tropical Diseases 5, e1197 (2011).
Larrieu, S., Balleydier, E., Renault, P., Baville, M. & Filleul, L. [Epidemiological surveillance du chikungunya on Reunion Island from 2005 to 2011]. Médecine tropicale : Revue du Corps de Santé colonial 72 Spec No, 38–42 (2012).
Soghigian, J. et al. Genetic evidence for the origin of Aedes aegypti, the yellow fever mosquito, in the southwestern Indian Ocean. Molecular Ecology 29, 3593–3606 (2020).
Kotsakiozi, P. et al. Population structure of a vector of human diseases: Aedes aegypti in its ancestral range, Africa. Ecology and Evolution 8, 7835–7848 (2018).
MacGregor, M. E. Aedes (Stegomyia) mascarensis, MacGregor: a new Mosquito from Mauritius. Bulletin of Entomological Research 14, 409–412 (1924).
Salvan, M. & Mouchet, J. Aedes albopictus et Aedes aegypti à l’Ile de La Réunion. Ann Soc Belg Med Trop 74, 323–6 (1994).
Bagny, L., Delatte, H., Quilici, S. & Fontenille, D. Progressive Decrease in Aedes aegypti Distribution in Reunion Island Since the 1900s. Journal of Medical Entomology 46, 1541–1545 (2009).
Vassal, J. J. Le paludisme à l’Ile de La Réunion. Per Gli Stud Della Maria 8, 18–27 (1907).
Delatte, H. et al. Geographic Distribution and Developmental Sites of Aedes albopictus (Diptera: Culicidae) During a Chikungunya Epidemic Event. Vector-Borne and Zoonotic Diseases 8, 25–34 (2008).
Hamon, J. Etudes biologique et systématique des Culicinae de l’Ile de La Réunion. Mem. Inst. Scient. Madagascar 4, 521–541 (1953).
Bouyer, J. & Lefrançois, T. Boosting the sterile insect technique to control mosquitoes. Trends in Parasitology 30, 271–273 (2014).
Haramboure, M. et al. Modelling the control of aedes albopictus mosquitoes based on sterile males release techniques in a tropical environment. Ecological Modelling 424, 109002 (2020).
Bouyer, J., Yamada, H., Pereira, R., Bourtzis, K. & Vreysen, M. J. B. Phased Conditional Approach for Mosquito Management Using Sterile Insect Technique. Trends in Parasitology 36, 325–336 (2020).
Bouyer, J. & Vreysen, M. J. B. Yes, irradiated sterile male mosquitoes can be sexually competitive! Trends in Parasitology 36, 877–880 (2020).
Organization, W. H. Guidelines for laboratory and field testing of mosquito larvicides. WHO/CDS/WHOPES/GCDPP/2005.13 (2005).
World Health Organization and Special Programme for Research and Training in Tropical Diseases and World Health Organization. Department of Control of Neglected Tropical Diseases and World Health Organization. Epidemic and Pandemic Alert. Dengue: Guidelines for diagnosis, treatment, prevention and control. (World Health Organization, 2009).
Yap, H. H. Preliminary report on the color preference for oviposition by Aedes albopictus (Skuse) in the field. Southeast Asian J Trop Med Public Health 6, 1–2 (1975).
Yap, H. H., Lee, C. Y., Chong, N. L., Foo, A. E. S. & Lim, M. P. Oviposition site preference of Aedes albopictus in the laboratory. Journal of the American Mosquito Control Association-Mosquito News 11, 128–132 (1995).
Marin, G., Mahiba, B., Arivoli, S. & Tennyson, S. Does colour of ovitrap influence the ovipositional preference of Aedes aegypti Linnaeus 1762 (Diptera: Culicidae). Int. J. Mosq. Res 7, 11–15 (2020).
Yoshioka, M. et al. Diet and density dependent competition affect larval performance and oviposition site selection in the mosquito species Aedes albopictus (Diptera: Culicidae). Parasites & Vectors 5, (2012).
Wong, J., Stoddard, S. T., Astete, H., Morrison, A. C. & Scott, T. W. Oviposition Site Selection by the Dengue Vector Aedes aegypti and Its Implications for Dengue Control. PLoS Neglected Tropical Diseases 5, e1015 (2011).
Hawley, W. A. The biology of aedes albopictus. Journal of the American Mosquito Control Association. Supplement 1, 1–39 (1988).
Delatte, H., Gimonneau, G., Triboire, A. & Fontenille, D. Influence of Temperature on Immature Development, Survival, Longevity, Fecundity, and Gonotrophic Cycles of Aedes albopictus, Vector of Chikungunya and Dengue in the Indian Ocean. Journal of Medical Entomology 46, 33–41 (2009).
Daugherty, M. P., Alto, B. W. & Juliano, S. A. Invertebrate Carcasses as a Resource for Competing Aedes albopictus and Aedes aegypti (Diptera: Culicidae). Journal of Medical Entomology 37, 364–372 (2000).
Papaj, D. R. & Rausher, M. D. Individual variation in host location by phytophagous insects. Herbivorous Insects: Host seeking behavior and mechanisms 77–127 (1983).
Valladares, G. & Lawton, J. H. Host-plant selection in the holly leaf-miner: Does mother know best? The Journal of Animal Ecology 60, 227 (1991).
Ellis, A. M. Incorporating density dependence into the oviposition preference-offspring performance hypothesis. Journal of Animal Ecology 77, 247–256 (2008).
Juliano, S. A., OMeara, G. F., Morrill, J. R. & Cutwa, M. M. Desiccation and thermal tolerance of eggs and the coexistence of competing mosquitoes. Oecologia 130, 458–469 (2002).
Costanzo, K. S., Kesavaraju, B. & Juliano, S. A. Condition-specific competion in container mosquitoes: the role of non-competing life-history stages. Ecology 86, 3289–3295 (2005).
Braks, M. A. H., Honório, N. A., Lounibos, L. P., De-Oliveira, R. L. & Juliano, S. A. Interspecific Competition Between Two Invasive Species of Container Mosquitoes, Aedes aegypti and Aedes albopictus (Diptera: Culicidae), in Brazil. Annals of the Entomological Society of America 97, 130–139 (2004).
Moore, C. G. & Fisher, B. R. Competition in Mosquitoes.1 Density and Species Ratio Effects on Growth, Mortality, Fecundity, and Production of Growth Retardant2. Annals of the Entomological Society of America 62, 1325–1331 (1969).
Claudel, I. et al. To bait or not to bait? Optimizing the use of adult mosquito traps for monitoring arbovirus vector populations in La Réunion Island. (2022) doi:https://doi.org/10.21203/rs.3.rs-1798972/v1.
Madeira, N. G., Macharelli, C. A. & Carvalho, L. R. Variation of the Oviposition Preferences of Aedes aegypti in Function of Substratum and Humidity. Memórias do Instituto Oswaldo Cruz 97, 415–420 (2002).
Bellini, R. et al. Use of the sterile insect technique against Aedes albopictus in Italy: first results of a pilot trial. in Area-wide control of insect pests 505–515 (Springer, 2007).
Boussès, P., Dehecq, J. S., Brengues, C. & Fontenille, D. Inventaire actualisé des moustiques (Diptera : Culicidae) de l’île de La Réunion, océan Indien. Bulletin de la Société de pathologie exotique 106, 113–125 (2013).
Zeileis, A., Kleiber, C. & Jackman, S. Regression Models for Count Data in R. Journal of Statistical Software 27, (2008).
Sileshi, G. Selecting the right statistical model for analysis of insect count data by using information theoretic measures. Bull. Entomol. Res. 96, 479–488 (2006).
Guthery, F. S., Burnham, K. P. & Anderson, D. R. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. The Journal of Wildlife Management 67, 655 (2003).
Hurvich, C. M. & Tsai, C.-L. Model selection for extended quasi-likelihood models in small samples. Biometrics 51, 1077–1084 (1995).
Burnham, K. P. & Anderson, D. R. Model selection and multimodel inference: a practical information-theoretic approach. 496 (Springer-Verlag, 2002).
Manly, B. F. J. Randomization, bootstrap and Monte Carlo methods in biology. 399 (CRC Press / Chapman & Hall, 2006). doi:10.1201/9781315273075.
Holm, S. A simple sequentially rejective multiple test procedure. Scandinavian journal of statistics 65–70 (1979).
R Core Team. R: A Language and Environment for Statistical Computing. (R Foundation for Statistical Computing, 2022).
Lesnoff, M. & Lancelot, R. aods3: analysis of overdispersed data using S3 methods. (2018).
Barton, K. MuMIn: Multi-Model Inference. (2022).
Xie, Y., Dervieux, C. & Riederer, E. R Markdown Cookbook. (Chapman; Hall/CRC, 2020). doi:10.1201/9781003097471.

No competing interests reported.

Download PDF

Editorial decision: Major revision
15 Sep, 2022
Reviews received at journal
09 Sep, 2022
Reviewers agreed at journal
05 Sep, 2022
Reviewers agreed at journal
28 Aug, 2022
Reviewers invited by journal
25 Aug, 2022
Editor assigned by journal
25 Aug, 2022
Editor invited by journal
25 Aug, 2022
Submission checks completed at journal
25 Aug, 2022
First submitted to journal
20 Aug, 2022

You are reading this latest preprint version

Four shades of red: optimizing the use of oviposition traps to monitor mosquito populations of arbovirus vectors in La Reunion

Status:

Version 1

Abstract

Figures

1 Introduction

1.1 A changing epidemiology of mosquito-borne viruses

1.2 Aedes aegypti in La Reunion: uncovering peculiarities

1.3 Goals

1.4 Study design

2 Results

2.1 Optimization of ovitrap settings in the field

2.1.1 Overview

2.1.2 Detection probability

2.1.3 Apparent density

2.2 Proportion of Aedes aegypti mosquitoes in L4

2.3 L4 survival during the laboratory-rearing step

3 Discussion

3.1 Trapping device and baiting

3.2 Organic matter in trap water

3.3 Trap position

3.4 Oviposition surface and egg density

3.5 Ecology and DV transmission

4 Conclusion

5 Methods

5.1 Biological data

5.1.1 Harvesting and processing mosquito eggs

5.1.2 Rearing mosquito larvae

5.2 Data analysis

5.2.1 Question-specific models

5.3 Estimates, credible intervals, and tests

5.3.1 Population means

5.3.2 Credible intervals and tests

5.3.3 P-value adjustment for multiple comparisons, and visualization

Declarations

Data availability

Acknowledgments

Author contributions

Competing interests

References

Additional Declarations

Supplementary Files

Status:

Version 1