The “Auto-Dissemination” Approach: A Novel Concept to Fight Aedes albopictus in Urban Areas

Background The main constraint to the fight against container-breeding mosquito vectors of human arboviruses is the difficulty in targeting the multiplicity of larval sources, mostly represented by small man-made water containers. The aim of this work is to assess the feasibility of the “auto-dissemination” approach, already tested for Aedes aegypti, as a possible alternative to traditional, inefficient control tools, against Ae. albopictus in urban areas. The approach is based on the possibility that wild adult females, exposed to artificial resting sites contaminated with pyriproxyfen, can disseminate this juvenile hormone analogue to larval habitats, thus interfering with adult emergence. Methodology We carried out four field experiments in two areas of Rome that are typically highly infested with Ae. albopictus, i.e. the main cemetery and a small green area within a highly urbanised neighbourhood. In each area we used 10 pyriproxyfen “dissemination” stations, 10 “sentinel” sites and 10 covered, control sites. The sentinel and control sites each contained 25 Ae. albopictus larvae. These were monitored for development and adult emergence. Principal Findings When a 5% pyriproxyfen powder was used to contaminate the dissemination sites, we observed significantly higher mortality at the pupal stage in the sentinel sites (50–70%) than in the controls (<2%), showing that pyriproxyfen was transferred by mosquitoes into sentinel sites and that it had a lethal effect. Conclusions The results support the potential feasibility of the auto-dissemination approach to control Ae. albopictus in urban areas. Further studies will be carried out to optimize the method and provide an effective tool to reduce the biting nuisance caused by this aggressive species and the transmission risk of diseases such as Dengue and Chikungunya. These arboviruses pose an increasing threat in Europe as Ae. albopictus expands its range.


Introduction
Aedes albopictus (Skuse, 1894) is native to Southeast Asia. In recent decades this mosquito has invaded and efficiently colonized temperate areas of the US and Europe, thanks to passive transportation of eggs in used tires and the ability to produce diapausing eggs [1], [2]. In its tropical range the species is a secondary vector of human arboviruses, such as Dengue, while in temperate regions its impact on public health is mostly due to its aggressive and diurnal biting behavior. However, in recent years, Ae. albopictus has been the sole vector of large epidemics of Chikungunya virus in La Reunion, France [3], [4] and Kerala, India [5]. In August 2007 the species was responsible for a Chikungunya virus outbreak in the Province of Ravenna in northeastern Italy where more than 250 human cases were confirmed [6] and where the mosquito has repeatedly been identified as infected by Chikungunya [7], [6], [8] and Usutu viruses [9]. In 2010 Ae. albopictus was responsible for 2 and 31 cases of endemic transmission of Dengue in France and China, [10], [11]. The report of a human blood index $80% in Rome raises particular concern for the potential of this species to vector pathogens in urban areas where humans represent the major blood-meal source [12].
In Italy, Aedes albopictus was detected for the first time in Genoa in 1990 and the next year in Padua [13], [14]. In the following years it spread to 19 out of 20 Italian regions and over 82 out of 107 Italian provinces [15]. It was first detected in Rome in 1997 [16] and since then it has colonized the whole urban area through three phases ''a first massive spread, a following maintenance of infestation, and the colonization of alternative winter breeding sites with favourable climatic conditions'' [17]. The successful invasion of Rome and of other urban areas in Italy and worldwide is driven by the ability of this mosquito to exploit a large variety of water containers as larval breeding sites, including the catch basins of storm drains, used tires, domestic containers, vases, etc. The difficulty in identifying and treating all these sources makes control extremely difficult. A survey carried out in Rome's zoological gardens, which estimated the larval density of mosquitoes breeding in catch basins, showed that a large number of productive alternative larval biotopes exist [18]. Despite this, the most commonly utilized strategy to reduce Ae. albopictus densities in Italy is the treatment of catch basins with larvicidal compounds [19]. This only targets an unknown percentage of the overall aquatic habitat. Indoor insecticidal treatments and house screening are not suitable due to the largely exophilic behavior of the species, and outdoor adulticidal treatments are recommended only in case of emergencies (e.g. virus outbreaks or very high nuisance in sensible locations such as outdoor recreational areas) due to their environmental impact and their low cost-efficiency ratio [19].
The aim of this work is to assess the feasibility of a new approach for the control of Ae. albopictus in urban areas, inspired by results obtained on the tropical Dengue vector, Ae. aegypti, in Thailand [20] and in Peru [21]. This approach, named ''autodissemination'', is based on the possibility that wild adult mosquitoes exposed to artificial resting sites contaminated with pyriproxyfen (PPF, a juvenile hormone analogue), can disseminate insecticide to larval breeding sites, thus preventing adult emergence. This strategy is facilitated by the oviposition behaviour of both Ae. aegypti and Ae. albopictus, that typically scatter the eggs from a single gonothrophic cycle among several temporary sites. This increases the probability of at least some larvae reaching the adult stage [22]. Second, extraordinarily low doses of PPF (Ae. aegypti: LC 50 = 0.011 p.p.b., [23], 0.0039 p.p.b. [24]; Ae. albopictus: LC 50 = 0.11 p.p.b. [25], are needed to interfere with the metamorphosis of juvenile stages [26], and/or to cause morphological and functional aberrations in emerging adults, such as decreased fertility in males and females [27], [28]. Third, evidence from laboratory experiments shows that females either forced to walk on PPF-treated paper or topically contaminated can contaminate larval sites and significantly inhibit adult emergence [29], [30]. Thus, the ''auto-dissemination'' approach can be proposed as a 'pull' (i.e attraction of wild mosquitoes to PPF-contaminated sites for contamination) and 'push' (i.e. dispersal of contaminated mosquitoes and dissemination of PPF to larval habitats) control strategy with the potential to target the myriad of cryptic larval breeding sites that cannot be reached by traditional larvicidal applications.
We here present the results of four ''auto-dissemination'' experiments carried out in two areas of Rome with high infestations of Ae. albopictus: a cemetery and an enclosed garden.

Dissemination stations
Dissemination stations (DS, Figure 1) were adapted from modified sticky traps (ST), previously developed by our group [31]. The four sticky surfaces were replaced by four 1268 cm black cotton cloths (Figure 1b). A thick net was placed over the water to prevent mosquitoes from ovipositing ( Figure 1c). Before each experiment, each DS was filled with 700 ml of tap water and each cloth was dusted with 1 g of powdered PPF. This was obtained by manually grinding 0.5% or 5% PPF tablets (Proxilar, I.N.D.I.A. Industrie Chimiche S.p.A.) to an granule average size of 40-80 micron.

Sentinel sites and control sites
Larval/pupal mortality in sentinel sites (SS) that were potentially contaminated with PPF by wild mosquitoes was compared to mortality in uncontaminated control sites (CS). Sentinel sites were assembled by inserting a 600 ml Pyrex beaker into a standard ovitrap (i.e. a black vase). This was done to allow easy decontamination after the experiments (PPF adheres to plastics, but not to glass). The border of the beaker was covered with black tape to avoid reflections. Control sites were similar to SS but closed with white nets to prevent mosquitoes from entering the beakers and transferring PPF. Each SS and CS contained 200 ml of tap water, 0.07 g of cat biscuits and 25 third-instar Ae albopictus larvae. These were obtained from eggs collected from the ''Sapienza'' University campus by ovitraps during the weeks before the experiments. After hatching, the larvae were reared outdoors on the terrace of the Department of Public Health and Infectious Diseases of ''Sapienza'' University at a density of about 1 larva/ml.

Study areas
The experiments were carried out in the following two sites located in a highly urbanized area of Rome (Italy) adjacentto the University ''Sapienza'' campus.
Site 1-Verano Cemetery. Verano is the largest (98 ha) and the oldest cemetery in Rome that is still in use. It is a green area within an urban neighbourhood less than 500 m away from ''Sapienza'' University campus. Very high densities of Ae. albopictus have been reported, as is common in cemeteries due to the high number of potential larval breeding sites represented by flower vases [12]. The site chosen for the experiments (41u54915.430 N and 12u31934.890) is a shallow underground crypt formed by eight 2262 m corridors with walls of tombs on both sides. Ten DS and 10 SS were placed in a single corridor, in pairs, as shown in

Author Summary
Aedes albopictus (the Asian Tiger mosquito) is one of the most invasive and aggressive disease vectors in the world. It is a serious public nuisance and a public health risk, due to its ability to transmit pathogens to humans. The control of this mosquito is complicated by the difficulty in targeting either juveniles (due to their ability to colonise myriad and often cryptic domestic and natural water containers) or adults (due to their diurnal and outdoor resting and biting patterns). The aim of our work was to assess the feasibility of a novel approach to control Ae. albopictus in urban areas. This was based on the possibility that wild adult females, exposed to artificial resting sites contaminated with an insect growth regulator, could subsequently contaminate their breeding sites and kill both the larvae and pupae developing therein. The results obtained from field experiments carried out in highly infested areas in Rome demonstrate that this approach has potential as a valid alternative to traditional (and ineffective) larval control efforts in urban areas. We therefore introduce a new perspective in the fight against the Asian Tiger Mosquito.
(41u54923.320N and 12u30957.350E). The experiments were carried out in the 1 ha garden (Figures 3a and 3b), where 10 DSs dusted with 5% pyriproxyfen were placed at a distance of about 2 m right from an equal number of SS; 10 CS were also placed about 2 m from SS. Two replicates were carried out from 28 July to 11

Experimental procedures
During the experiments, SS and CS were monitored every two days. For each larval cohort, we derived cumulative totals of dead larvae, dead pupae and emerged adults, as follows: i) live larvae were left to develop further; ii) dead larvae and pupae were counted and removed ( Figure 2b); iii) live pupae were counted and transferred by pipette to a separate disposable cup containing uncontaminated water; these cups were covered with netting and maintained under semi-natural conditions until adult emergence or pupal death; iv) the water level in the SS and CS was maintained at 200 ml volume; v) temperature was recorded using one data logger for each corridor in Site 1, and by individual data loggers located close to each SS in Site 2. Monitoring continued until all the original 25 larvae were either dead or emerged (7-12 days).

Statistical analysis
Mortality in SS and CS was analysed by nonparametric alternatives to the t-test, since the distribution was not normal. We used the Kolmogorov-Smirnov (K-S) two-sample test (twotailed), which is powerful even when the distributions differ in terms of dispersion We used Wilcoxon signed-rank tests (twotailed) to compare the mortality in SS between experiments within same field site: this is a non-parametric paired difference test that compares two related samples or repeated measures on a single sample.
Two mixed-effect logistic regression models (one per study area) were applied to quantify PPF-related mortality. Each model includes, as independent variables, the number of replicates (''replicate effect''), the treatment (i.e. potential contamination in SS by wild mosquitoes vs. no contamination in CS) and the interaction between the replicates and the treatments, which allows an evaluation of the effect on the mortality in the sentinel sites in each replicate. The code identifying each SS and each CS was included as random effect to correct for the repeated measures within sites (''site-effect''). Table 1 shows the overall mortality observed in SS and CS in the four experiments. Overall mortality in SS was $50% in the 3 experiments carried out with 5% PPF concentration, and 20.8% in the first experiment in Site 1, with 0.5% concentration. In all  experiments, .90% of the mortality recorded in SS occurred at the pupal stage. In Site 2 however, about 13% of deaths occurred during adult emergence. This has been frequently noted in insects treated with PPF [32], but never in mosquitoes. The overall mortality in CS was #2.5%.
In Site 2, both experiments (2.1 and 2.2) utilised the 5% concentration. The difference in mortality in SS vs. CS was significant only in Exp. 2.2 (K-S, p = 0.015; Figure 3 The results of the mixed-effect logistic regression model showed that mortality was always significantly higher in SS than in CS. In Site 1, mortality was 9-and 66.5-fold higher in SS than in CS-in Exp.1.1 and in Exp 1.2, respectively, although mortality observed in SS was approximately 3 times lower in Exp 1.1 than in Exp 1.2 ( Table 2). In Site 2, mortality was 49-and 37-fold higher in SS than in CS in Exp.2.1 and Exp 2.2, respectively (Table 3). Mortality was similar in the two replicates and no significant differences were found in the interaction between ''replicate'' and ''treatment''. Figure 4 shows mortality in SS in the four experiments. In Exp 1.1 mortality was concentrated in 3 out of 10 SS, while in Exp 1.2 it ranged between 40 and 100%. In both Site 2 experiments mortality .76% was observed in 5 out of 10 SS, 4 of which were located in the same position in both replicates.

Discussion
A number of experiments carried out under laboratory conditions have shown that adult Ae. albopictus topically contaminated with PPF can transfer enough material to water containers to exert a significant lethal effect on pre-imaginal stages developing therein [23], [20], [29], [30]. For the first time we tested the hypothesis that wild Ae. albopictus adults can act as ''autodisseminators'' of PPF and inhibit adult emergence from sentinel sites. Although it was not possible to measure PPF concentration in those sites (the low doses of the product are beneath the limits of detection for any published method), the evidence supports the working hypothesis and suggests that auto-dissemination could represent a valid, novel approach for reducing Ae. albopictus densities in urban temperate areas.
We observed significantly higher mortality in our sentinel sites than in the controls. This shows that PPF was transferred by mosquitoes into sentinel sites and elicited a lethal effect. Mortality was almost exclusively limited to the pupal stage, i.e. the stage on which PPF is known to have its major effect [26]. Note that control sites were exposed to exactly the same experimental conditions as sentinel sites, except for the fact that contact with potentially contaminated mosquitoes was prevented by a net cover in the controls.
Mortality was not uniformly distributed among sentinel sites. This strongly suggests that PPF contamination occurred at some sites, but not others -presumably as a result of differences in the frequency of visits made by contaminated mosquitoes. This is particularly evident in Site 2, where mortality was observed in 5 out of 10 SSs, 4 of which were located at the same position in both replicates (Figures 3 and 4), leading to hypothesize that these sites were more attractive for the mosquitoes than the remaining ones. The same applies for the first replicate in Site 1. In the second replicate, however, mortality ranging from 40 to 100% was observed in all SSs suggesting that these were all visited and contaminated but with differing frequency.
In the cemetery experiments, a 10-fold increase in PPF concentration resulted in a 3-fold increase in mortality (Table 2).
Median mortality in sentinel-sites, in the three experiments carried out with a 5% PPF concentration, was higher in Site 1 than in Site 2 (18 and 13 deaths/25 larvae, respectively). In the mixed-effect logistic regression model, the variance due to the ''site effect'' was 0.12 (Standard deviation = 0.35) and 1.74 (Standard Deviation = 1.32), respectively. This is the likely consequence of the fact that, as mentioned above, in the Site 2 most mortality was  observed only in 50% of the sentinel sites, while in Site 1 40-100% mortality was observed in every sentinel site. This presumably reflects the different ecology of the two experimental areas. Site 1, being an underground corridor, is ecologically very homogeneous and it is reasonable to hypothesize that all sentinel sites were equally attractive to the flying wild mosquitoes. The enclosed garden in Site 2 was far more heterogeneous and sentinel sites were located outdoors in a wider area compared to Site 1. It is reasonable to hypothesise that under these conditions, sentinel sites vary in attractiveness depending on contrast against background, shade, humidity, etc. However, even under such heterogeneous conditions the overall mortality in sentinel sites was about 37-49 fold higher than in controls. It is also important to emphasize that the experiments were not designed with the aim of identifying environmental factors that might optimise impact, although the effect of sun exposure has already been hypothesized in autodissemination experiments carried out against Ae. aegypti in Peru [21]. In fact, it is relevant to highlight that the major/minor attractiveness of sentinel sites to mosquitoes is likely to affect the results of the experiments, but has a relatively low practical relevance. In fact, if the approach is applied to reduce Ae. albopictus adult densities, the mosquitoes themselves will disseminate PPF to the most attractive (i.e. most productive) natural breeding sites. Further studies are required to assess the ideal conditions for the location of dissemination stations, as well as to evaluate the persistence of the lethal effect under different environmental (e.g. temperatures, sun exposures) and ecological (e.g. air flow, presence of animals, of humans, other disturbances/attractions) variables.
In the only other auto-dissemination experiment ever carried out in the field outdoors (i.e. in Peru with a 0.5% PPF concentration [21]), the overall reductions in Ae. aegypti adult emergence was 49-84%, as opposed to a 7-8% mortality in controls. In the single experiment carried out in Rome with the same PPF concentration, the overall reduction in Ae. albopictus adult emergence was 20.8%, as opposed to a 2.4% mortality in controls. This is quite encouraging: in fact a lower overall effect could have been expected in our experiments, since PPF LC 50 reported for Ae. albopictus (0,11 ppb [25]) is about 10 times higher than that reported for Ae. aegypti (0.011 ppb [23]; 0.0039 ppb [24]). It is also worth noting that in Rome control sites were monitored concurrently to sentinel sites and were located at the same distance from dissemination stations, while in Peru they were separated in time. Thus, our results rule out the hypothesis that PPF could be passively transported by wind. If this had occurred, we would have expected higher mortality in at least some control sites.
Overall, the results from these small-scale experiments carried out in Rome strongly encourage further studies to evaluate the feasibility of the exploitation of the auto-dissemination approach to control Ae. albopictus densities in urban areas. In fact, the experimental design probably underestimates the overall impact of the approach, as other known effects of PPF on mosquitoessuch as sterilizing effects on adult females [28] and a decrease in male spermiogenesis [27] -were not taken into consideration. Moreover, the effect of auto-disseminated PPF was only monitored on third instar larvae (or later stages), while in natural breeding sites younger larvae will also be contaminated, presumably increasing the overall impact. The auto-dissemination approach has the potential to effectively counter the main challenge to conventional larviciding approaches by effectively targeting the myriad of cryptic breeding sites that these mosquitoes utilise. Based on results from mark-release-recapture experiments carried out in the campus of Sapienza University in Rome [33], it is possible to hypothesize that gravid, PPF-contaminated Ae. albopictus females could contaminate breeding sites in a 200-m radius area around a dissemination station. Other relevant advantages of the auto-dissemination approach are: i) a higher residual activity of PPF (4 months in water [34]) compared to that of other compounds commonly used for larval control and ii) no risk for human health, due to the low-toxicity of the product for vertebrates and the high sensitivity of mosquito larvae/pupae [35]. With regard to this latter point, although PPF is effective against many insects, the proposed approach targets containerbreeding species with such tiny amounts of compound, disseminated exclusively to their breeding sites, that impacts on non-target species are likely to be minimal. Finally, the auto-dissemination approach could be a very cost-effective control tool. Once deployed, the dissemination station does not require frequent maintenance, nor frequent toxicant applications, thanks to PPF's outstanding persistence [28]. It is also of note that other Culicidae species (e.g. Culex pipiens, a common nuisance species in urban Italy, sharing some Ae. albopictus breeding sites) may contribute to the dissemination of the product [36] and be affected by it themselves. Indeed, larger-scale field experiments are needed to evaluate more precisely the worldwide feasibility of the approach and to promote its use against Ae. albopictus.
With appropriate modifications to the dissemination stations, auto-dissemination tools may be simple enough to be deployed by members of the public. In the experiments carried out in Rome, PPF dissemination stations were adapted from existing sticky-traps for collecting Ae. albopictus and Ae. aegypti females [31], [37]. The sticky surfaces were replaced with cloth surfaces dusted with pulverized PPF. In Peru, dissemination stations consisted of 1-liter plastic disposable pots lined with damp black cloth, These dissemination tools could be optimized to increase their attractiveness for mosquitoes and their overall practicability. The fabric of the cloths could be optimized for the release of PPF to mosquito legs, and the method of application of the compound into the cloth could be optimized and standardized. Other methods to attract and contaminate adult mosquitoes could be developed (see, for instance, [30]). Finally, the auto-dissemination approach could be exploited to spread other mosquito toxic compounds, such as other juvenile hormones or fungi, which might have an even greater impact than PPF [38], [39], [40], [41].