Introduction

Malaria is a vector-borne disease caused by parasites of the genus Plasmodium that results in over 500,000 deaths annually [1]. Malaria is transmitted by female Anopheles spp. mosquitoes. Throughout Tanzania, malaria is one of the most wide-spread mosquito-borne diseases, with at least 2.5 million cases reported in 2022 [1–3]. Tanzania contributes to 4% of the total malaria deaths worldwide [1].

Amid changing climate conditions and an increase in tourism, Zanzibar City - the largest city of the Zanzibar archipelago that includes Stone Town - has experienced a recent increase in malaria cases. In 2009, the Zanzibar Ministry of Health initiated a set of policies to eliminate malaria on the archipelago, including increased indoor spraying, a systematic distribution of insecticidal nets, and a surveillance and case-detection system [4,5]. Zanzibar has therefore been able to reduce malaria transmission between 2005 and 2015 [6]. However, the yearly malaria incidence rate of 2.7 cases per 1,000 people reported in 2017 had increased to 3.6 by 2021 [5,7]. The increase in cases is thought to be due to an expansion of tourism and a greater influx of people infected with Plasmodium from mainland Tanzania, inadvertently providing blood meals for the local mosquitoes [8,9]. Additional vector control interventions are needed to reduce the resident mosquito population and help to eliminate malaria. Understanding the local ecology of the Anopheles mosquitoes is crucial to inform such control strategies. Due to their low mobility and fully-aquatic nature, mosquito larvae that develop in freshwater bodies are effective targets for reducing the abundance of adult vector mosquitoes and overall malaria transmission [10]. However, the breeding habits of Anopheles mosquitoes in Zanzibar City have not been characterized.

While urbanization was originally thought to decrease the incidence of malaria, unplanned, rapid urbanization in sub-Saharan Africa has been accompanied by sustained malaria transmission [11–14]. Zanzibar City has expanded 3.8% geographically per year since 2004, leading to unplanned urbanization with poor integration of water drainage into the city infrastructure, potentially increasing mosquito breeding sites [15]. Therefore, urbanization has likely contributed to enhanced malaria transmission in recent years [16]. With a population of 800,010 people in 2023, and a population growth rate of 4.48%, this trend is likely to continue [17].

Every year, particularly during the long rains between March and May, the pooling of stagnant freshwater in the urban environment creates breeding sites, thus contributing to greater numbers of disease vectors. Despite some existing drainage of the streets of Zanzibar City, excess water runoff often leads to the expansion or flooding of natural water pools (see S1 Fig). Additionally, Zanzibar City, as a tourist destination, features aesthetic ponds and nonfunctional fountains that often contain stagnant water, creating breeding sites for mosquitoes (particularly within historic Stone Town). Over recent years, climate change has resulted in shorter but heavier rains during the March-May rainy season, likely increasing the number of potential urban mosquito breeding sites in Zanzibar City [18,19]. For larvicidal strategies to be most effective against Anopheles mosquitoes, knowledge of the defining characteristics of Anopheles breeding sites is important. However, Anopheles breeding habits vary by geographical location and season, making it difficult to rely solely on studies conducted in other areas or at other times [20–22].

Currently, larvicidal strategies are rarely used in Africa [23,24]. However, larvicidal strategies may complement existing vector control measures in Zanzibar City as indoor residual spraying (IRS) and bednets lose effectiveness. This decrease in effectiveness is due to a shift in the malaria vector population from mainly Anopheles gambiae to the more adaptable Anopheles arabiensis [25]. The latter commonly bites humans outdoors, thereby avoiding bed nets and IRS [26]. Additionally, Anopheles arabiensis is becoming increasingly resistant to pyrethroid insecticides [27,28]. Notably, an Aedes aegypti-focused larval survey was conducted in Zanzibar City in 2018, which did not report any Anopheles larvae [29]. However, the study focused on artificial, temporary breeding site types known to harbor Aedes larvae [30]. To our knowledge, no preexisting data regarding Anopheles breeding sites in Zanzibar City are publicly available. We set out to address the lack of data by conducting larval surveys and breeding site characterizations for Anopheles larvae in Zanzibar City during the long rainy season. We assessed whether physicochemical parameters, hydro-period, and predator presence could be used to predict whether a water body contained Anopheles larvae. We found consistently high Anopheles larvae abundance in concrete, semi-permanent structures with high dissolved oxygen levels. Additionally, we found high Anopheles larvae abundance in natural semi-permanent areas with high dissolved oxygen levels after heavier rains. Our findings suggest that Anopheles mosquito breeding sites can be identified using specific characteristics in Zanzibar City, allowing for targeted larviciding as a strategy to reduce malaria transmission.

Methods

Sampling locations

Between April 11, 2024, and May 1, 2024, 16 permanent or semi-permanent sites and 30 temporary sites were sampled for Anopheles mosquito larvae (Fig 1A). Sites were initially found through reconnaissance or asking community members during the first week of the study. Because only publicly available water bodies were surveyed (and few such water bodies were found in the city’s Eastern sprawl), most study sites were near the more developed Stone Town and the densely populated neighborhoods nearby. Semi-permanent sites were defined as permanent structures that are occasionally drained (but rarely during the rainy season), while permanent sites were defined as structures that contain water continuously throughout the year. Semi-permanent and permanent sites sampled included unfinished or nonfunctional fountains (n = 3), artificial ponds (n = 4), roadside ditches (n = 3), and wetlands (n = 5) (S1 Fig). Temporary sites included tires (n = 10), plastic water containers (n = 18), and small concrete wash basins (n = 2). Each permanent or semi-permanent site larger than 4m in perimeter was divided into semi-randomized 0.5m x 0.5m subsites. A random number was generated that indicated the 0.5m x 0.5m site to sample. Before sampling, however, the 0.5m x 0.5m subsite was inspected to ensure that it could physically house mosquito larvae (e.g., adequate, stagnant water present), and that it was accessible with a larval dipper. If a randomly generated subsite could not be sampled, the nearest possible subsite was sampled instead. Once the subsite was identified, a 0.5m x 0.5m quadrat was placed into the water. At each site, data was collected at five subsites unless the site was smaller than five 0.5m x 0.5m quadrats (referred to herein as “subsite”). Because mosquito larvae distribution is not evenly distributed across larger habitats [33], each subsite was treated independently during analysis. Notably, all Anopheles species on Unguja are capable of transmitting P. falciparum [9,34], so mosquito larvae were identified to genus level. Mosquito genera on Unguja also include Culex and Aedes [35].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Map of Zanzibar City, Tanzania, showing sites sampled for mosquito larvae.

Map generated using a shapefile provided by GADM under a CC BY license. The shapefile can be found at https://gadm.org/download_country.html.

https://doi.org/10.1371/journal.pone.0313248.g001

Results

Characterization of physical properties of Anopheles breeding sites

To test whether Anopheles prefer permanent/semi-permanent sites or temporary sites in Zanzibar City during the rainy season, we compared the number of sites with Anopheles larvae between permanent/semi-permanent and temporary sites. We found that 56.25% of the permanent or semi-permanent sites contained Anopheles larvae at least once over the study duration, while only 10% of the temporary sites contained Anopheles larvae (Fisher’s Exact Test, P = 0.0013, OR = 11.57, 95% CI = 2.359–44.85) (S1 Table). The three temporary sites with Anopheles larvae included two plastic water containers and one rubber container. The permanent or semi-permanent Anopheles breeding sites were in both Stone Town and the greater Zanzibar City area, while the temporary Anopheles breeding sites were only found in the Zanzibar City area outside of Stone Town (Fig 2A). Pools were not included in the statistical analysis because they are subject to potentially confounding anthropogenic variables. However, three pool drains across two hotel pools in Stone Town were sampled for mosquito larvae, and no larvae were found. Additionally, six large rain tanks for drinking water were visually inspected for larvae (but not sampled due to ethical issues surrounding conducting larval dips in drinking water). Among the three tanks inspected that were left open, two had Culex larvae present. The three closed tanks inspected did not have any larvae.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Characterization of physical qualities of Anopheles breeding sites in Zanzibar City, Tanzania.

A) Map showing site permanence and whether site contained Anopheles larvae on either visit. Map generated using shapefile provided by GADM under a CC BY license. The shapefile can be found at https://gadm.org/download_country.html. B) Principal component analysis graph showing loadings of physical parameters. C) Distribution of Anopheles larvae abundance in artificial and natural subsites. D) Distribution of Anopheles larvae abundance in semi-permanent and permanent subsites. E) Distribution of Anopheles larvae abundance in subsites of different site types in the presence or absence of Culex larvae. F) Distribution of Anopheles larvae abundance in subsites of different site types in the presence or absence of Aedes larvae. Statistical significance determined using a Mann-Whitney Test. For all pairwise comparisons: * P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

https://doi.org/10.1371/journal.pone.0313248.g002

Anopheles presence was associated with concrete, artificial, and permanent/semi-permenent subsites, in addition to water temperature and outdoor temperature (Fig 2B). Culex presence was associated with natural subsites and the presence of visible trash, while Aedes presence was associated with subsites made of plastic or rubber (Fig 2B). There was also statistical evidence for the permanent/semi-permanent artificial subsites harboring more Anopheles larvae than permanent/semi-permanent natural subsites (Two-tailed Mann-Whitney U test, P = 0.0022, U_{(93, 54)} = 1953) (Fig 2C). Significantly more Anopheles larvae were caught at semi-permanent subsites than permanent subsites (Two-tailed Mann-Whitney U test, P < 0.0001, U_{(84, 63)} = 1701) (Fig 2D). Cohabitation with other mosquito larvae genera was uncommon, and we found no statistical evidence of fewer Anopheles larvae in habitats shared with Culex or Aedes larvae (Fig 2E-F). Last, principal component analysis regression (PCAR) showed parameters with significant positive predictive capabilities, including permanence/semi-permanence, artificial, and concrete. Parameters with a significant negative predictive capability included Culex cohabitation, Aedes cohabitation, vegetation cover, the presence of a large dump within 3m, plastic/rubber, and natural. While the intercept was nonsignificant, the overall PCA regression was significant by analysis of variance (F_{(3, 173)} = 3.288, P = 0.0221) (S2 Table).

Characterization of chemical parameters of Anopheles breeding sites

Chemical parameters were associated with the presence of mosquito larvae of each genus. The percentage of dips per subsite with Anopheles was associated with dissolved oxygen concentration, dissolved oxygen saturation, and pH. The percentage of dips per subsite with Culex was associated with salinity, and the percentage of dips per subsite with Aedes was associated with relative turbidity and estimated chlorophyll content (Fig 3A). Although the distribution of the number of Anopheles, Culex, or Aedes larvae caught plotted against dissolved oxygen levels was not normally distributed (Anderson-Darling Test, P < 0.0001), Gaussian curves were used to visualize which dissolved oxygen levels were associated with a higher larval abundance for each genus in permanent/semipermanent subsites. A least squares regression comparison of the Gaussian fits of larvae abundance plotted against dissolved oxygen saturation/concentration suggested that distributions differ between genera (Extra sum of squares F test, F_{(6, 293)} = 6.897, P < 0.0001). The distribution of Anopheles larvae abundance was shifted towards higher dissolved oxygen levels than the distribution of Culex larvae abundance, while Aedes larvae were found in low abundance across a wide range of dissolved oxygen levels (Fig 3B-C). While a Gaussian curve could not be fit for larvae abundance plotted against pH, the distribution of Anopheles larvae abundance was shifted towards higher pH levels than Culex (Fig 3C). Aedes larvae were found in low abundance across a wide range of pH levels. Turbidity appeared to have little effect on the distribution of larvae abundance from different genera (Fig 3E).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Characterization of chemical parameters of Anopheles breeding sites in Zanzibar City, Tanzania.

A) Principal component analysis showing loadings of chemical parameters. B) Distribution of Culex abundance (left axis) and Anopheles/Aedes abundance (right axis) at each subsite plotted against dissolved oxygen concentrations. C) Distribution of Culex abundance (left axis) and Anopheles/Aedes abundance (right axis) at each subsite plotted against percentages of dissolved oxygen saturation. D) Distribution of Culex abundance (left axis) and Anopheles/Aedes abundance (right axis) at each subsite plotted against pH. E) Distribution of Culex abundance (left axis) and Anopheles/Aedes abundance (right axis) at each subsite plotted against relative turbidity.

https://doi.org/10.1371/journal.pone.0313248.g003

Incorporating rainfall data into model to predict Anopheles presence

Each permanent/semi-permanent site was visited twice during the sampling period, with at least one week separating each visit. Therefore, differences in Anopheles larvae abundance between visits were analyzed. Significantly more Anopheles larvae were caught per subsite during the second visit of fountains (Two-tailed Mann-Whitney U test, P = 0.0169, U_(15,15) = 67.5) and wetlands (Two-tailed Mann-Whitney U test, P = 0.0485, U_{(22, 22)} = 187.0) (Fig 4A). No significant difference in Anopheles larvae abundance between visits of artificial ponds and ditches was observed (Fig 4A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. The effect of rainfall on Anopheles breeding site preferences.

A) Distribution of Anopheles larvae abundance at subsites between site types and visits. B) Amount of rain within 48hrs of visiting each site, grouped by site types. Statistical significance determined using multiple Wilcoxon tests for paired data. C) Individual comparisons of Anopheles larvae abundance at each visit to each site. X axis shows the amount of rain within 48hrs of visiting that site. D) Distribution of percentage dissolved oxygen saturation across subsites of different site types, split into whether Anopheles larvae were caught. E) Distribution of dissolved oxygen concentrations across subsites of different site types, split into whether Anopheles larvae were caught. Statistical significance was determined using either a Mann-Whitney U Test or Kruskal-Wallis Test with Dunn’s uncorrected step-down tests, unless otherwise noted. For all pairwise comparisons: * P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

https://doi.org/10.1371/journal.pone.0313248.g004

There was a significant difference in rainfall between visits of artificial ponds (Wilcoxon matched-pairs signed rank test, P = 0.00061, W = -108.0) and fountains (Wilcoxon matched-pairs signed rank test, P = 0.000061, W = -120.0) (Fig 4B). However, because the visits for each type of permanent/semi-permanent site did not occur on the same day (e.g., not all artificial ponds were sampled on the same day), we also compared rainfall between individual sites. Overall, wetlands had significant differences in Anopheles abundance between visits but nonsignificant differences in rainfall. However, one wetland had strong, rain-associated increases in Anopheles larvae abundance (Fig 4C).

Because PCA indicated an association between Anopheles breeding sites in Zanzibar City and higher dissolved oxygen levels, we hypothesized that rain-independent breeding sites might be preferred by Anopheles mosquitoes due to consistently higher levels of dissolved oxygen. Additionally, the semi-permanent subsites sampled had significantly greater levels of dissolved oxygen (Welch’s T test, t_(94.69) = 6.912, P < 0.0001), lower levels of salinity (Two-tailed Mann-Whitney U test, U_{(84, 62)} = 2053, P = 0.0146), and fewer predators present than permanent subsites (Χ² = 5.227, df = 1, P = 0.0222) (S2 Fig, S3 and S4 Tables). Commonly observed predators of mosquito larvae included dragonfly/damselfly larvae (order Odonata), backswimmers and water skaters (order Hemiptera), and stonefly larvae (order Plecoptera) (S3 Table). The subsites from artificial ponds and ditches where Anopheles larvae were caught had significantly higher levels of dissolved oxygen than the subsites where no Anopheles larvae were caught (Fig 4D-E) (S5 and S6 Tables). Subsites from fountains where Anopheles larvae were caught had significantly lower levels of dissolved oxygen than subsites from fountains without Anopheles larvae, and there was no significant difference in dissolved oxygen levels between either group of subsites from wetlands (S5 and S6 Tables). Subsites with Anopheles larvae from artificial ponds had significantly higher levels of dissolved oxygen than subsites with Anopheles larvae from fountains and wetlands (Fig 4D-E) (S7 and S8 Tables). Additionally, subsites with Anopheles larvae from ditches had significantly higher dissolved oxygen levels than subsites with Anopheles larvae from wetlands (S7 and S8 Tables).

Regression analysis

The data suggested an interaction between dissolved oxygen levels (both saturation and concentration), rainfall within 48hrs, and/or site type, so we fit a logistic regression model to all data from permanent/semi-permanent sites and temporary sites. The outcome used for the logistic regression model was either “Anopheles caught” or “no Anopheles caught”. We included physical parameters that were significant in the original PCA regression analysis, as well as dissolved oxygen concentration and depth. We omitted data that were included in the initial PCA regression analysis, but for which dissolved oxygen data was unavailable. Dissolved oxygen concentration was included in the model instead of dissolved oxygen saturation because dissolved oxygen concentration was more significantly associated with Anopheles presence than saturation (as shown in Fig 4D-E). Interaction terms included dissolved oxygen saturation and semi-permanence, dissolved oxygen concentration and concrete, dissolved oxygen concentration and rainfall within 48hrs, concrete and rain within 48hrs, and natural and rain within 48hrs. Significant physical predictors included the percentage of dips with Aedes larvae (lower odds of catching Anopheles larvae), increased depth (lower odds of catching Anopheles larvae), the presence of a predator in the quadrat (lower odds of catching Anopheles larvae), and more rain within 48hrs (lower odds of catching Anopheles larvae) (Table 1). Significant chemical predictors included increased salinity (lower odds of catching Anopheles larvae) (Table 1). Significant interaction terms included increased concentration dissolved oxygen in semi-permanent subsites (lower odds of catching Anopheles larvae), increased concentration dissolved oxygen in concrete subsites (higher odds of catching Anopheles larvae), and increased concentration of dissolved oxygen after more rainfall within 48hrs (higher odds of catching Anopheles larvae) (Table 1). The area under the receiver operating characteristic (ROC) curve was 0.9034, and the positive predictive power was 65.7% (S3 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Logistic regression model of Anopheles larvae caught vs. not caught.

https://doi.org/10.1371/journal.pone.0313248.t001

Discussion

In this study, we characterized the breeding sites of Anopheles mosquitoes during the rainy season in Zanzibar City. Our results suggest that Anopheles breeding sites in Zanzibar City tend to be concrete, semi-permanent water bodies with high dissolved oxygen saturation regardless of heavy rainfall. However, Anopheles breeding sites can expand to natural semi-permanent water bodies with high dissolved oxygen concentration after heavier rains.

Semi-permanent, artificial sites are likely preferred by Anopheles mosquitoes over temporary breeding sites due to their longer developmental times compared to larvae from different mosquito genera. Anopheles mosquitoes develop from egg to adult in 10–14 days, while Aedes and Culex mosquitoes develop from eggs to adults in 7–10 days [40]. Since Anopheles larvae take longer to develop, they may either be outcompeted in temporary environments where Aedes mosquitoes are more suited to breed in, or the container is emptied too frequently for Anopheles larvae to develop [30]. This finding is consistent with studies in Ethiopia and elsewhere where more Anopheles larvae were found mostly in artificial, permanent/semi-permanent sites [22,41–43]. However, this finding contradicts other studies where Anopheles larvae were found mostly in either natural or temporary environments [20,44,45], suggesting either context-dependent adaptations to use semi-permanent sites for breeding, or variation in the definition of “temporary” between studies.

Semi-permanent subsites sampled had more favorable conditions for Anopheles than permanent sites, including higher levels of dissolved oxygen, fewer predators, and lower salinity (S2 Fig). Chemical qualities of water bodies influence Anopheles ability to breed due to impacts on nutrient availability, toxicity, selection for predators, algae growth, exposure to sunlight, and other factors [46–48]. In Zanzibar City, dissolved oxygen, low abundance of predators, and low salinity are associated with higher Anopheles larvae abundance, as seen in some locations but not in others [49,50]. While the reasons behind these differences in breeding site chemical characteristics are likely to be site-specific, permanent sites may have more artificial introduction of predators or competing species, more accumulated pollution than the more frequently drained semipermanent sites, or more nutrients that support a naturally higher population of predators. Higher predator abundance has been associated with fewer Anopheles larvae in permanent sites in Ethiopia [48], which is consistent with our data, except semipermanent sites are used as breeding sites in Zanzibar City instead of temporary sites. Therefore, the data presented in this study support a location-dependent hypothesis for Anopheles breeding site preferences, where Anopheles prefer artificial, semipermanent breeding sites during the rainy season in Zanzibar City.

As high dissolved oxygen is an indicator of overall clean water [51], it is unclear from our data whether Anopheles larvae are found in sites with relatively higher dissolved oxygen concentrations/saturation because of the requirement for dissolved oxygen itself, or because of the lack of pollutants that negatively covary with dissolved oxygen levels. Data from experiments on Culex and Aedes larvae suggest that larvae from these genera can use dissolved oxygen for respiration when reaching the water surface to access atmospheric oxygen is difficult [52,53]. Although formal dissolved oxygen utilization experiments with Anopheles larvae are lacking, some authors suggest that dissolved oxygen might itself be especially important for Anopheles larvae development because Anopheles larvae lack breathing tubes [54]. Alternatively, as higher dissolved oxygen is an indicator of a less-polluted environment [51], the tendency for Anopheles breeding sites to have higher dissolved oxygen levels may be solely because those sites are less likely to have harmful pollutants and more likely to have the nutrients required for Anopheles development. Due to the lack of a strong, direct correlation between dissolved oxygen levels and Anopheles larvae abundance, the data from this study more strongly support the latter hypothesis (especially since the water surface was accessible to larvae in all breeding sites). However, more laboratory-based experimental studies are necessary to determine the role of dissolved oxygen in Anopheles larvae development.

Although we observed cohabitation between Anopheles and Culex larvae, this cohabitation was relatively uncommon, suggesting that each genus occupies distinct niches in Zanzibar City. This lack of common cohabitation is consistent with data from other studies, suggesting that Culex and Anopheles larvae compete for nutrients [48,55]. The site in our study with the most cohabitation between Anopheles and Culex larvae was a large, flooded wetland only after heavy rain, which could mean that the rain-associated expansion of this breeding site minimized the competition between mosquito genera. Culex larvae are well known to tolerate heavily polluted sites and are often able to avoid competition at these sites [56,57], while the tolerance of Anopheles larvae for polluted sites is more debated. While some study locations support the hypothesis that Anopheles primarily breed in clean sites [49,50], studies on mainland Africa have provided evidence of Anopheles adapting to more polluted environments [41,44,58,59]. With Aedes larvae dominating temporary breeding sites and Culex larvae dominating more polluted breeding sites with salinity levels above 0 ppt, it is possible that there is little pressure for Anopheles to use more polluted sites for breeding during the rainy season in Zanzibar City. However, as breeding site characteristics change during the dry season - when the semipermanent sites may not be available for breeding [42] - it is likely that Anopheles will need to find other breeding sites and may be forced into more polluted environments.

Additionally, it is likely that larval density decreases during the dry period, as other studies have shown higher densities of Anopheles mosquitoes during or immediately after rainy seasons [60–62]. However, a recent analysis of Anopheles stephensi breeding patterns revealed that seasonal shifts in breeding patterns are also dependent on the degree of urbanization [63], highlighting how seasonal breeding patterns in one location cannot be extrapolated to other locations. Studies of Anopheles breeding sites in Zanzibar City during other seasons are, therefore, crucial for a complete understanding of breeding dynamics.

Because this sampling event took place during April (i.e., in rainy season), when consistent rain over many days is followed by days of intense sun, we hypothesized that rain might play a role in the changes in breeding site productivity between visits. Studies have shown that rainfall contributes to either the expansion of Anopheles breeding sites, or the flooding of breeding sites and increased mortality of Anopheles larvae [64,65]. However, these two outcomes are clearly dependent on the amount of rainfall and the specific shape of the site. The sites in our study most subjected to incoming drainage that could potentially transfer Anopheles larvae from other breeding sites were those classified as drains. Additionally, the wetland located at the eastern extreme of the study area also accepted an influx of drainage. While a small number of Anopheles larvae were found at the wetland on the eastern extreme of the study area, it is unlikely that the presence of Anopheles larvae at this site was dependent on drainage influx because the Anopheles were found during the visit with less rain. All other sites were filled mostly by rainwater and did not accept runoff from other locations, so any increase in Anopheles larvae can be attributed to breeding at that site. Our data, therefore, suggest that higher rainfall levels are associated with more Anopheles larvae being found in wetlands and fountains, while artificial ponds and ditches serve as Anopheles breeding sites regardless of rain levels.

Many insects observed alongside mosquito larvae were known predators of mosquito larvae [66–68]. Because predator abundance decreased the odds of observing Anopheles larvae in a subsite, promoting the development of predatory larvae may also help reduce Anopheles mosquito abundance. Although their abundance was not recorded in this study, Cyclopoid copepods are also known to be predators of mosquito larvae and have been shown to complement the activity of antilarval phytochemicals [69,70]. Future studies aiming to assess whether copepod abundance correlates with less Anopheles breeding site productivity could also help inform antimalarial strategies. Introduction of predatory fish (e.g., Gambusia spp.) into artificial ponds has shown to be an effective antilarval strategy in some sites [71]. However, Gambusia spp. target other prey over mosquito larvae and are destructive towards the natural ecosystem when introduced [72]. The presence of other prey besides mosquito larvae in a breeding site also affects the practicality of introducing predatory invertebrates into water bodies as an antimalarial strategy [72,73]. Therefore, while introducing natural predators of mosquito larvae may seem like a promising antimalarial strategy, each site’s ecosystem must first be analyzed to ensure specific targeting of mosquito larvae. Nevertheless, the introduction of predators could be effective when used in combination with other larvicidal strategies, especially because adult Anopheles mosquitoes may be able to detect the presence of predators in water bodies and avoid breeding in such sites (mostly through chemosensory mechanisms) [69]. Future studies are necessary to determine the most effective larvicidal strategy in Zanzibar City.

Although larvicides are rarely used in Africa [23,24], the data from our study suggest that semipermanent ponds or obstructed concrete ditches with high levels of dissolved oxygen saturation may be effective targets for larvicidal strategies in Zanzibar City. Additionally, preventing the flooding and expansion of wetland sites during rain by increasing drainage from those sites might also help prevent Anopheles from using these as breeding sites. Because mosquito larvae require stagnant water to breathe [74], larvae abundance in fountains could be minimized by ensuring that these fountains remain functional, and larvae abundance in drains could be minimized by ensuring that these drains remain free of stagnant water. The use of synthetic larvicides should be avoided, as their use can result in harmful, off-target effects to the environment, and mosquito larvae can develop resistance to the larvicides [75]. However, phytochemicals from plant extracts may be a useful alternative, as they are more benign to the environment, more biodegradable, and contain a diverse array of larvicidal secondary compounds that minimize the risk of resistance [76,77]. Additionally, bacterial larvicides using larva-specific toxins from Bacillus thuringiensis israelensis or Bacillus sphaericus may be another safe and effective larvicidal strategy [78,79].

Limitations to this study include other potentially important parameters (e.g., algae cover) not being analyzed, limited geographical range, potential differences in Anopheles pupae abundance being overlooked, and a lack of species-level resolution of mosquito breeding habits. Differences in microbiota composition between sites were also not analyzed, which could also impact the ability for Anopheles larvae to develop [80]. While the purpose of this study was to analyze the Anopheles genus, as all Anopheles species in Zanzibar are capable of transmitting malaria, individual species have distinct breeding habits in other locations [21,62,81]. Determining which Anopheles species are occupying each type of breeding site in Zanzibar City would be useful for monitoring adaptations to more polluted breeding sites and the relative importance of each species in malaria and filarial disease transmission [35,44]. Additionally, larvicidal strategies tend to be more effective in areas where Anopheles adults are exophilic, as residual indoor spraying or bed nets are less effective in those areas [82]. Therefore, more recent studies on the behavior of adult Anopheles mosquitoes in Zanzibar City are necessary and can also help further assess breeding site productivity [83].

Nevertheless, the short-term nature of the study provides evidence of rapid, rain-associated changes in Anopheles breeding site preferences in Zanzibar City, while also revealing sites that are likely to be more consistent breeding sites if they remain filled with water. The small geographical range included provides a targeted analysis of Zanzibar City-specific breeding habitats of Anopheles mosquitoes. Additionally, while our study did not include all potentially relevant parameters, the parameters measured are easy to monitor (e.g., more easily than site-specific microbiota) and can therefore be put into practice for larvicidal strategies targeted against Anopheles. This study can serve as a baseline for future longitudinal studies on Anopheles breeding sites in Zanzibar City, which could be used to develop a larvicidal strategy to combat persistent malaria in the Zanzibar archipelago.

Conclusion

The incidence of malaria in Zanzibar City has remained between 1–2% since 2015, with a slight but steady increase due to continued import of Plasmodium spp. from other malaria endemic areas. Therefore, targeting the resident population of Anopheles vector mosquitoes in the largest city of the Zanzibar archipelago is a crucial strategy to combating malaria on the islands. With pyrethroid resistance in adult Anopheles mosquitoes increasing on Unguja, larvicidal strategies could serve as an alternative strategy to decrease the population numbers of resident Anopheles mosquitoes. Although the geographic range of this study focused on the core of Zanzibar City, Anopheles larvae appeared to be evenly distributed between core Stone Town and the surrounding Zanzibar City. However, as most beautification projects and artificial, semipermanent structures are in the more tourist-trafficked area of core Stone Town, careful monitoring of these projects is necessary to prevent these concrete basins in parks from becoming nutrient and oxygen-rich breeding grounds for Anopheles mosquitoes.

Our study provides the first systematic survey of Anopheles breeding sites in Zanzibar City. Our findings that Anopheles prefer either semipermanent, concrete habitats with high dissolved oxygen levels, but can expand to sites with lower dissolved oxygen levels after heavy rains, suggest that Anopheles larvae may be targeted as an antimalarial strategy. However, because our study is the first of its kind conducted in Zanzibar City, future longitudinal research is required to characterize the breeding sites of Anopheles mosquitoes during dry periods. Additionally, the productivity of each breeding site should be studied by analyzing which breeding sites are producing Anopheles pupae and mature mosquitoes. Given the current targetability of Anopheles larvae, potential public health policies that may reduce the population of Anopheles mosquitoes during the rainy season could include introducing phytochemical larvicides into artificial ponds, ensuring that fountains are functional, and preventing the flooding of wetlands. Through a combination of strategies targeting larval and adult Anopheles, the resident population of Anopheles mosquitoes could be depleted sufficiently to eliminate malaria in Unguja.

Supporting information

Acknowledgments

We would like to acknowledge and thank the faculty and staff at the School for International Training office in Zanzibar for providing logistical and material help for this project. We would like to thank Dr. Mohamed Jiddawi at the Zanzibar Medical Group for the thoughtful advice when conceptualizing the study. Additionally, we thank Dr. Mohammed Maalim at the University of Dar es Salaam Institute of Marine Sciences for lending us the water quality measurement devices.