Aquatic predators of mosquito larvae in Kamphaeng Phet Thailand


Aquatic mosquito predators play a vital role in the natural control of mosquito populations. Therefore, we investigated the predation rates, prey preference, and other feeding characteristics of aquatic mosquito predators that occur within our region of activity, Kamphaeng Phet, Thailand. First we sampled aquatic habitats to find out which aquatic predators are common. Subsequently, we conducted a series of experiments to find out these before mentioned feeding characteristics of those common species (Weterings et al. 2014a,b, Weterings et al. 2015, Weterings 2015). Afterwards the data from these experiments were combined into a single dataset that we will present here.

Figure 1. Visualization of the best model based on the grouped experiments. In each graph the parameters that are not displayed are kept constant at their median. The dots display the residual variance after correcting for all other variables. The grey areas display the confidence interval. A) the effect of predator density on the per capita predation rates, B) the effect of prey density on the per capita predation rates, C) the effect of predator size on the per capita predation rates and D) the effect of predator species/order on the per capita predation rates.


The best model based on the combined dataset was a model that included predator densities, prey densities, predator size and predator species/order. This model had an AICc score that was 15.5 points lower than the second best model (Table 1). Predator size explained most variance in the per capita predation rates and had the strongest effect (see standardized estimate Table 2). From those models that only contained one explanatory variable the model with predator size explained 58% of the variation in contrast to 25% for predator species, 23% for prey density and 19% for predator densities. The winning model that contained all these variables explained 81% of the variation in the per capita predation rates. The relationship between predation rates and predator densities was log-linear (Figure 1). When predator densities increased the per capita predation rates decreased in a decelerating manner. Prey densities were also log-linear related to per capita predation rates. Predation rates increased with higher prey densities but this effect leveled off as prey densities became higher. Predation rates were highest for Diplonychus rusticus and Toxorhynchites splendens (Figure 2) . These species were also the largest of all predators in these experiments. In contrast, the predation rates of Anisops breddini were very high in relation to their size. Toxorhynchites splendens also effectively reduced mosquito larvae populations in the field. Containers that were placed in the field (see here) and were colonized by T. splendens showed a clear reduction in Aedes larvae densities (Figure 3) .

Table 1. Comparison of models for the data of all predation experiments in which aquatic invertebrates were fed Aedes aegypti larvae. P = predator density, M = prey density, S = Predator size, Sp = predator species/order. Null = the null model with only an intercept, k = number of model parameters, AICc = small sample size corrected Akaike Information Criterion, Δi = difference in AICc in comparison to the best model, W i = model weight.

Table 2. Normal and standardized parameter estimates and standard errors for the winning model of the combined predation experiments.

Predation rates for Corixidae, Pleidae and Veliidae were not experimentally tested. Instead predation rates were taken from existing literature (Figure 2). The predation rates for species from these families are generally much lower than those for the other aquatic predators. When comparing their co-occurrence with Aedes in contrast to other predators, Veliidae and Corixidae in particular appear to be not very efficient predators (Table 3). In containers without predators, Aedes larvae occurred in 61.7% of the sampled containers. For Veliidae 18.9% of the containers in which they occurred still had Aedes larvae, for Corixidae this was 25.0%. In contrast, Notonectidae and Anura co-occurred only with Aedes in 5.0% and 3.8% of the containers respectively. This is an indication that these two predators more efficiently reduce Aedes larvae. This low co-occurrence of Anura with Aedes larvae is remarkable because when predation by tadpoles was experimentally tested no predation was observed. The number of mosquito larvae was in most cases identical before and after the experiment (LINK). Predator size / age did not have an effect on predatory behavior in Anura. In a few cases the number of mosquito larvae were reduced with one to three larvae. Direct attacks of tadpoles on mosquito larvae were not observed. Neither large and old nor small and young tadpoles displayed any signs of predatory behavior.

Figure 2. Mean predation rates and standard errors for all aquatic predators (. Predation rates were based on experimental results from this study except for predation rates for Corixidae (Amrapala, et al., 2009) and Pleidae (Takahashi, et al., 1979). The numbers on the top of the graph are the sample sizes for each calculated mean.

Table 3. Percent of containers that contained Aedes in the presence of certain predators. In absence of predators Aedes occurred in 61.7% of the containers (n=47). n shows the number of containers in which each specific predator was present.


The experiments on aquatic predators all showed consistent results for all tested species, except for tadpoles. Predation rates were all non-linearly related to prey density and predator density, as is seen in a Holling type II functional response (Holling, 1959). Predator size was a factor of major importance, even more so than taxonomic group, prey density or predator density. These results are consistent with previous studies were predator size was also shown to be more important than species-specific differences (Rall, et al., 2011).

Toxorhynchites splendens

Toxorhynchites splendens was one of the predators with the highest predation rates. These rates were dependent on prey density as well as prey type (see here). Predation rates for fourth instar T. splendens larvae feeding on fourth instar prey ranged between 10 and 23, with higher rates in high prey density treatments and low rates in low prey density treatments. Other studies have shown similarly high predation rates for this species (Pramanik and Raut, 2003). The total predation of mosquito larvae during the larval stage of Toxorhynchites species has previously been estimated to approximately 5,000 first instar larvae and 300 fourth instar larvae (Steffan and Evenhuis, 1981; Focks, 1982), which is a very impressive number. Although predation is relatively high for Toxorhynchites species, these rates do vary depending on what prey are available and predator-specific traits such as age. Predation rates differ significantly between early and late instar Toxorhynchites larvae (Pramanik and Raut, 2003). Predation rates also differ among prey species. Pramanik and Raut (2003) showed that predation rates were highly dependent on the size of prey which they used to explain differences in rates among prey type. However, the we showed different results, where predation rates were measured (see here) using mixed species. In this case, T. splendens did not feed on more smaller prey, as Pramanik and Raut (2003) suggested, but fed more on the most active prey, which in this case were the slightly larger Aedes larvae instead of the smaller Culex larvae. Similar results have been reported for Notonectidae, where the behaviour/activity of mosquito prey also had a strong influence on their vulnerability (Sih, 1986).

Overall, Toxorhynchites is an important genus of mosquito predators, especially for container breeding mosquitoes such as A. aegypti and A. albopictus. The high predation rates and beneficial behavioral traits, such as killing prey before pupating and surviving dry periods as fourth instar, as well as the use of specific breeding habitats, make this genus of high interest. There is often a strong reduction in Aedes larvae associated with the presence of Toxorhynchites species (Figure 3). One major drawback is that most, if not all, species are associated with forests and therefore do not occur in those areas where the contact between humans and vector mosquitoes is highest: highly urbanized areas (Focks, et al., 1983).

Figure 3. Aedes reduction caused by T. splendens larvae in four different containers. The only predators in these containers were T. splendens larvae. A, B and C display a clear reduction in Aedes larvae when the number of T. splendens larvae increase. D shows a stable Aedes larvae population in the absence of predators. Figure A also displays an increase in Aedes larvae after the number of T. splendens larvae decreases.


Notonectidae have a strong effect on mosquito larval densities and on aquatic community structure in general (Blaustein, 1998; Eitam, et al., 2002). Several mechanisms are underlying this effect including predation but also other indirect effects. Anisops breddini is very common in Thailand (Zettel, et al., 2012) and shows great potential as a biocontrol agent for mosquito larvae. However, there are many other Notonectidae species in the region that are just as common, if not more (Zettel, et al., 2012). Notonectidae as a family are therefore of great interest, especially because they occur in all kind of areas and in most types of mosquito breeding habitats.

Small water bugs

Corixidae, Veliidae and Pleidae feed generally on very low numbers of mosquito larvae in comparison to other aquatic mosquito larvae predators. Their small size is probably one of the main reasons for these low rates. All three families feed more efficiently on first and second instar mosquito larvae than on third and fourth instar mosquito larvae. Hence, they need to be early colonizers in order to be effective in the control of container-inhabiting mosquito larvae. All three families were very abundant in all types of aquatic habitats, except for storm-drains. In addition, Corixidae and Veliidae were often found in urban containers. Hence, the role of these three families as mosquito predators could be more important than their size and predation rates superficially indicate.


The predation experiments showed that tadpoles of the species B. melanostictus, K. pulchra and H. raniceps do not feed on mosquito larvae. Predation of mosquito larvae did not dependent on tadpole age, nor on mosquito larvae size (see here). Although, some species do prey on mosquito larvae (eg. Petranka and Kennedy, 1999), the great majority of anuran tadpoles is considered to be strictly herbivorous (Willems, et al., 2005). Although the tested tadpoles did not feed on mosquito larvae, it does not mean that these species have no effect on the control of mosquitoes in nature. Tadpoles of one of the studied species (B. melanostictus) has been shown to feed on A. aegypti eggs (Bowatte, et al., 2013). The most common tadpole in the field was P. leucomystax, which is closely related to P. cruciger, a species for which egg predation has also been observed (Bowatte, et al., 2013). It is likely that, of the other tested species, at least K. pulchra is also an egg predator because during field surveys mosquito larvae were often absent from containers that were colonized by this species (Table 7). Nevertheless, there are other mechanisms that could underlie this pattern of low co-occurrence in Aedes larvae and tadpoles, which include competition, competition avoiding oviposition site selection and the presence of other predators (Blaustein and Margalit, 1994; Mokany and Shine, 2006a,b). Most mosquito larvae and tadpoles feed on detritus. Therefore, competition may be of more importance in comparison to predation (Blaustein and Margalit, 1994; Mokany and Shine, 2003). For particular mosquito-tadpoles systems it has been shown that both groups reduce growth in each other’s presence regardless of resource availability. Mokany and Shine (2003) suggest that certain fungi in the feces of tadpoles are causing this reduced growth rate. In addition, certain mosquito species are capable of detecting tadpole presence and avoid these habitats for ovipositing, possibly reducing competition for their offspring (Mokany and Shine, 2003).

Odonata and Nepomorpha are important naturally-occurring predators of mosquito larvae. However, they were rarely found in Aedes compatible breeding habitats. For this reason, they will not be further discussed in here.


Some of the most important aquatic predators of Aedes larvae in Kamphaeng Phet are Toxorhynchites splendens, Notonectidae, tadpoles and micro-Heteroptera. Toxorhynchites splendens is important because it is one of the first species to colonize new breeding habitats (see here) and it is specialized in feeding on mosquito larvae. In addition, it is able to reduce Aedes mosquitoes more effectively than, for example, co-occurring Culex larvae. Nevertheless, this species is mainly arboreal and is only found in urban areas when vegetation is dense enough such as in parks. Notonectidae is a family that contains many very efficient mosquito larvae predators that can quickly reduce mosquito larvae densities. It is not a trophic specialist therefore it can also sustain viable populations in the absence of Aedes larvae. Notonectidae are regularly found in urban containers (17.6%) which make it a relative important predator in these areas where the Aedes-human contact is highest. Tadpoles occur equally often in urban and agricultural habitat as Notonectidae but are even more often found in forest habitats. Tadpoles of several common species were found to not feed on mosquito larvae at all. However, in the field the co-occurrence in breeding habitats with Aedes larvae was very low suggesting that they may have an effect on Aedes. Whether this effect is based on egg predation, competition or Aedes ovipositing site selection remains unsolved and should be addressed in future studies. Micro-Heteroptera were not experimentally tested for their predatory ability, but field observations indicate that they do negatively affect Aedes larvae populations. This effect is much lower than for the other aquatic predators, which is possibly due to their small size (allometry). However, they are the most widely distributed of all predators in Aedes breeding habitats and are very often present in urban habitats. Therefore their role as a predator in the biological and natural control of Aedes larvae cannot be neglected.


  • Blaustein, L. and Margalit, J. (1994). Mosquito larvae (Culiseta longiareolata) prey upon and compete with toad tadpoles (Bufo viridis). Journal of Animal Ecology, 63(4), 841–850.
  • Blaustein, L. (1998). Influence of the predatory backswimmer, Notonecta maculata, on invertebrate community structure. Ecological Entomology, 23(3), 246–252.
  • Bowatte, G., Perera, P., Senevirathne, G., Meegaskumbura, S. and Meegaskumbura, M. (2013). Tadpoles as dengue mosquito (Aedes aegypti) egg predators. Biological Control, 67(3), 469–474.
  • Eitam, A., Blaustein, L. and Mangel, M. (2002). Effects of Anisops sardea (Hemiptera: Notonectidae) on oviposition habitat selection by mosquitoes and other dipterans and on community structure in artificial pools. Hydrobiologia, 485, 183–189.
  • Focks, D., Sackett, S., Dame, D. and Bailey, D. (1983). Ability of Toxorhynchites amboinensis (Doleschall) (Diptera: Culicidae) to locate and oviposit in artificial containers in an urban environment. Environmental Entomology, 12, 1073–1077.
  • Holling, C. S. (1959). Some characteristics of simple types of predation and parasitism. The Canadian Entomologist, 91(07), 385–398.
  • Petranka, J. W. and Kennedy, C. A. (1999). Morphology: Pond tadpoles with generalized is it time to reconsider their functional roles in aquatic communities? Oecologia, 120(4), 621–631.
  • Pramanik, M. and Raut, S. (2003). Occurrence of the giant mosquito Toxorhynchites splendens in drains and its predation potential on some vector mosquitoes of Kolkata(Calcutta), India. Medical Entomology and Zoology, 54(4), 315–323.
  • Mokany, A. and Shine, R. (2003). Competition between tadpoles and mosquito larvae. Oecologia, 135(4), 615–20.
  • Mokany, A. and Shine, R. (2006a). Oviposition site selection by mosquitoes is affected by cues from conspecific larvae and anuran tadpoles. Austral Ecology, 28(2003), 33–37.
  • Mokany, A. and Shine, R. (2006b). Pond attributes influence competitive interactions between tadpoles and mosquito larvae. Austral Ecology, 27(2002), 396–404.
  • Rall, B. C., Kalinkat, G., Ott, D., Vucic-Pestic, O. and Brose, U. (2011). Taxonomic versus allometric constraints on non-linear interaction strengths. Oikos, 120(4), 483–492.
  • Sih, A. (1986). Antipredator responses and the perception of danger by mosquito larvae. Ecology, 67, 434–441.
  • Steffan, W. and Evenhuis, N. (1981). Biology of Toxorhynchites. Annual Review of Entomology, 26, 159–181.
  • Weterings, R. (2015). The tadpoles of three Thai anuran species fo not prey on mosquito larvae. Journal of Vector Ecology, in press.
  • Weterings, R., Umponstira, C. and Buckley, H. L. (2014a). Container-breeding mosquitoes and predator community dynamics along an urban-forest gradient: the effects of habitat type and isolation. Basic and Applied Ecology, 15, 486– 495.
  • Weterings, R., Umponstira, C. and Buckley, H. L. (2014b). Predation rates of mixed instar Odonata naiads feeding on Aedes aegypti and Armigeres moultoni (Diptera: Culicidae) larvae. Journal of Asia-Pacific Entomology, 18(1), 1–8.
  • Weterings, R., Umponstira, C. and Buckley, H. L. (2015). Density-dependent allometric functional response models. Ecological Modelling, 303, 12–18.
  • Willems, K., Webb, C. and Russel, R. (2005). Tadpoles of four common Australian frogs are not effective predators of the common pest and vector mosquito Culex annulirostris. Journal of the American Mosquito Control Association, 21(4), 492–494.
  • Zettel, H., Lane, D. J. W. and Pangantihon, C. V. (2012). Notes on Notonectidae (Hemiptera: Heteroptera) from southeastern Asia, mostly from Brunei and the Philippines. Acta Entomologica Musei Nationalis Pragae, 52(1), 29–48.