Sustainable control of mosquito larvae in the field by the combined actions of the biological insecticide Bti and natural competitors



Integrated management of mosquitoes is becoming increasingly important, particularly in relation to avoiding recolonization of ponds after larvicide treatment. We conducted for the first time field experiments that involved exposing natural populations of the mosquito species Culex pipiens to: a) application of the biological insecticide Bacillus thuringiensis israelensis (Bti), b) the introduction of natural competitors (a crustacean community composed mainly of Daphnia spp.), or c) a combined treatment that involved both introduction of a crustacean community and the application of Bti. The treatment that involved only the introduction of crustaceans had no significant effect on mosquito larval populations, while treatment with Bti alone caused only a significant reduction in the abundance of mosquito larvae in the short-term (within 3–10 days after treatment). In contrast, the combined treatment rapidly reduced the abundance of mosquito larvae, which remained low throughout the entire observation period of 28 days. Growth of the introduced crustacean communities was favored by the immediate reduction in the abundance of mosquito larvae following Bti administration, thus preventing recolonization of ponds by mosquito larvae at the late period (days 14–28 after treatment). Both competition and the temporal order of establishment of different species are hence important mechanisms for efficient and sustainable mosquito control.


The control of mosquitoes is becoming increasingly challenging because climate change and global trade favor the spread of invasive mosquito species (Roiz et al. 2008, Schäfer and Lundström 2009) and strongly increase the associated risk of vector-borne diseases (Weaver and Reisen 2010). Most strategies for mosquito control are based on the use of insecticides. However, intensive use of insecticides has unwanted effects on non-target species (Suma et al. 2009, Mommaerts et al. 2010) and increases the risk of target species developing resistance (Akiner et al. 2009, Melo-Santos et al. 2010). Furthermore, treated populations can recover after application of the insecticide (Seleena et al. 1999). As a consequence, integrated pest management, using biological antagonists either alone or in combination with insecticides, has drawn increasing interest (Gurr et al. 2003, van den Berg and Takken 2009, Gentz et al. 2010).

Strategies for mosquito control that are based on biological methods have been used since the early 20th century (Legner and Sjogren 1984). In particular, the use of predators (e.g., Gambusia affinis, Dystiscidae, Odonata) has been investigated intensively (Bence 1988, Blaustein et al. 1995, Fincke et al. 1997, Kumar and Hwang 2006, Beketov and Liess 2007). However, difficulties in colonization and the management of predators, as well as a lack of synchrony between the life cycles of predator and prey, have impeded their deployment (Bay 1974, Kindlmann and Dixon 2001, Kumar and Hwang 2006). As a consequence, we have changed our focus to the use of natural competitors (i.e., crustaceans) to control mosquito larvae. Many crustacean species show similar biotope preferences (i.e., early colonization of temporary ponds) and similar feeding behavior (i.e., filter feeding) to that shown by mosquito larvae (Williams 2006). Given that crustaceans are found in nearly all types of pond, they could be used as agents for mosquito control without disturbing the natural biotope fauna. Previous field investigations found that competing communities of crustaceans had a negative impact on populations of mosquito larvae (Chase and Knight 2003, Beketov et al. 2010, Meyabeme Elono et al. 2010). In semi-field experiments, Cladocerans (e.g., Daphnia magna) negatively affected mosquitoes (i.e., Culex pipiens or Anopheles quadrimaculatus) by reducing the amount of oviposition (Duquesne et al. 2011), increasing time to pupation (Knight et al. 2004, Stav et al. 2005), and reducing total abundances of mosquito larvae (Knight et al. 2004, Duquesne et al. 2011). However, the negative effect of Cladocerans was only found for well-established populations, that is, Cladoceran populations that had developed for more than one week before colonization by mosquito larvae (Knight et al. 2004, Stav et al. 2005, Duquesne et al. 2011). Hence, under field conditions, control of mosquito larvae by crustaceans will be limited during the initial phase of community development, when abundances of crustaceans are still low. Thus, the use of insecticides may still be required to control mosquito larvae until populations of crustaceans have become established. In this respect, the use of the biological insecticide Bacillus thuringiensis israelensis (Bti) in combination with the introduction of a crustacean community might be an appropriate integrated approach to mosquito management. Indeed, Bti has no negative effect on crustacean populations (Rivière et al. 1987, Becker et al. 1992). However, its ability to eliminate mosquito larvae is only temporary (Boisvert and Boisvert 2000).

On the basis of the findings mentioned above, we hypothesized that the long-term effects of crustaceans will strengthen the impact of the insecticide, owing to additive and complementary effects of these two stressors on larval populations of mosquitoes. In the present study, this hypothesis was evaluated under field conditions by attempting to eliminate mosquito larvae that had colonized ephemeral ponds by either introducing only natural competitors, applying Bti alone, or applying Bti in combination with introduced natural competitors. The aim of the study was to investigate whether the integrated control approach was more effective and sustainable than the common strategy for mosquito control, which involves Bti treatment alone.



Field investigations were conducted in a forested area in Leipzig (51°18′15.60″ N 12°21′44.39″ E) during the period from 16 June, 2008 until 14 July, 2008. We monitored 18 natural ponds with surface areas that varied between 1 m2 and 8 m2 and a water depth that varied between 3 cm and 25 cm. No vegetation was present inside the ponds, but trees and bushes surrounded them. A natural community of mosquito larvae and other insects had colonized the ponds before treatment. However, these insect populations consisted mainly of mosquito larvae (Culicidae). Other insects, such as members of Ephemeroptera, Chironomoidae, and Notonectidae, were found only sporadically, and thus were excluded from the analysis. The results obtained from ponds that provided less suitable breeding conditions for mosquitoes as indicated by less than 50 mosquito larvae per liter on the day of treatment (N = 3) are not reported.


Three different experimental treatments were tested. For the control condition, no treatment was applied. For the first treatment, which is referred to as “Competition,” a natural community of crustaceans was introduced. The organisms that formed these crustacean communities were collected from a lake near Rosslau (51°53′08″ N, 12°19′11″ E) using a plankton net (mesh size, 500 μm). Organisms were introduced such that their final density in the treated ponds was approximately five individuals per litre. The introduced community comprised Ceriodaphnia spp. (74.7%), Simocephalus spp. (7.5%), Daphnia spp. (3.6%), Scapheloberis spp. (2.7%), Ostracoda (9.7%), and Cyclopoida (1.7%).

For the second treatment, which is referred to as “Bti,” ponds were treated with 1,000 μg Bti/L, which is the concentration used routinely in mosquito control programs (Becker 2003). The liquid stock solution used (Vectobac 12 AS, Valent BioScience Corporation, Lyon, France) had an activity of 1,200 International Toxin Units (ITU) per milligram.

For the third treatment, which is referred to as “Bti+Competition,” Bti was applied at a concentration of 1,000 μg Bti/liter, and a community of natural crustaceans was introduced as in the treatment “Competition.” Four replicates were considered in the data analysis for the treatments “Bti,”“Bti+Competition,” and control, and three replicates for the treatment “Competition.”


A water sample with a total volume of 0.5–2 liters (2 liters for ponds with a surface area >5 m2, 1 liter for ponds with a surface area 1–5 m2, 0.5 liter for ponds with a surface area <1 m2) was collected twice a week from each pond. Each water sample consisted of several 300 ml subsamples, which were collected with a scoop from different parts of the pond. The samples were filtered through a plankton net (55 μm mesh size) and preserved in 70% ethanol (approximately 30 ml). The abundances and composition of the zooplankton were then analyzed using a binocular Leica S6D microscope (Wetzlar, Germany). Mosquito larvae were characterized to the species level using the determination key of (Becker et al. 2010). All other invertebrates were characterized to suborder or family level using the determination keys of Brauer (1909), Stresemann (1957), Einsele (1993), and Klausnitzer (2009).

The concentration of dissolved oxygen, temperature, pH, and conductivity of water samples were measured twice a week, between 09:00 and 14:00, using an Oxi340 oxygen meter (WTW, Weilheim, Germany) and a pH/EC/TDS Combo testing meter (Hanna Instruments, Kehl am Rhein, Germany). Water parameters did not differ significantly among treatments (data not shown), and thus were not included in the subsequent analysis.


All data were log transformed before all data analyses, which were performed using three steps. First, data on abundances were analyzed for variance between experimental (“Competition,”“Bti,”“Bti+Competition”) and control treatments at each single sampling day using Student's t-test. Significant differences were denoted by asterisks in graphical representations of the data. Data were tested for a normal distribution (using the Shapiro–Wilk Normality Test) and homogeneity (F-test) to verify that underlying statistical assumptions were not violated. Second, the changes in population size (i.e., slope) over a period of time of mosquito larvae and crustaceans following the different treatments were compared with the changes in population size under control conditions. The time was separated in two periods, the early (days 3–14 after treatment) and late (days 14–28 after treatment) time periods. The separate data sets collected for each time period were analyzed using a generalized least squares (GLS) mixed model approach. Our response variable was “mosquito larval abundance” or “crustacean abundance.” The predictor variables were “day of development” and the “type of treatment” (“Competition,”“Bti,” or “Bti+Competition”). The analysis used multiple measures over time in multiple ponds, thus violating the statistical assumption of independence of observations for standard testing (Pinheiro and Bates 2000, West et al. 2006). As a consequence, we used the factorial variable “pond” as a random effect for a first model. Given that temporal autocorrelation of subsequent measures in the same pond was to be expected, we calculated a second additional model using an autocorrelation structure (AR1: autoregressive model of order 1) (Zuur et al. 2009). We then compared these two models using Akaike's information criterion (AIC) and chose the model structure with the lowest AIC (Zuur et al. 2009). As a result, the first model was used to analyze crustacean development, whereas the second model was considered to be appropriate for analyzing mosquito larval development. The final models were presented using the restricted maximum likelihood (REML) calculation. All models were validated by plotting theoretical quantiles vs standardized residuals (Q–Q plots) to assess the normality of residuals. Homogeneity of variance was evaluated by plotting residuals vs fitted values, and influential data points were identified using Cook's distance method (Quinn and Keough 2002). Third, the impacts of antagonists (Cladocera, Ostracoda, and Cyclopoida) on the abundance of mosquito larvae were investigated on the last day of the observation period (day 28 after treatment) using multiple linear regression. The abundances of antagonists with a significant impact were plotted against the abundances of mosquito larvae and the linear regression line was added.

Analyses were performed using the “R” statistical and programming environment (R Development Core Team 2010) and the “nlme” and “lattice” packages.


Invertebrate communities were dominated by mosquito larvae and crustaceans. Other taxa, including Notonectidae and larvae of Chaoberidae, Chironomidae, Ephemeroptera, and Megaloptera, were only sporadically observed and therefore excluded from analyses. The pattern of development of the introduced natural community of crustaceans over time indicated the existence of two distinct periods: an early time period (days 3–14 after treatment), which was characterized by increasing abundances of crustaceans, and a late time period (days 14–28 after treatment), which was characterized by stable crustacean populations (see below). Hence, the following analyses were performed separately for these two time periods.

Effect of treatments on abundances of mosquito larvae

Populations of mosquito larvae consisted mainly of Culex pipiens (>99%) and initial mosquito abundances were similar in all treatments (238 ± 96 individuals/liter).

The “Competition” treatment failed to decrease the number of mosquito larvae significantly. Indeed, although both the abundance of mosquito larvae at each sampling time and the overall increase in the population size of mosquito larvae during the early (days 3–14) and late (days 14–28) time periods were slightly lower for the “Competition” treatment than the control treatment (Figure 1, Table 1), the differences were rarely statistically significant.

Figure 1.

Changes in the abundance of mosquito larvae (mean ± SE) in different treatment groups [Control (N=4) = no treatment; Competition (N=3) = treatment with introduction of a crustacean community; Bti (N=4) = treatment with Bti (1,000 μg/liter); Bti+competition (N=4) = treatment with Bti (1,000 μg/liter) and the introduction of a crustacean community] over time [P0 = before treatment; P1 = 3 – 14 days after treatment; P2 = 14 – 28 days after treatment]. *Significant differences compared to control (t-test, p < 0.05).

Table 1.  Effects of treatments [Competition (N=3) = introduction of crustacean community; Bti (N=4) = treatment with Bti (1,000 μg/liter); Bti+competition (N=4) = treatment with Bti (1,000 μg/liter) and the introduction of a crustacean community] as compared with control conditions (N=4) on the changes in population size (i.e., slope) of zooplankton (mosquito larvae and crustaceans) over different time periods (early = 3–14 days after treatment; late = 14–28 days after treatment).
  1. a) AR-1: autoregressive model of order 1.

  2. b) Random effect: generalized least squares (GLS) model with pond as a random effect and allowing for unequal variance in the variance-covariate terms (see Methods section for details).

Mosquito larvaeCompetitionearly0.442–1.0990.283AR-1a)
 Btiearly0.2812.452 0.021 AR-1a)
 Bti + competitionearly0.3361.3070.202AR-1a)
  late0.024–2.019 0.047 AR-1a)
 Bti + competitionearly0.3022.699 0.012 Randomb)
 Bti + competitionearly0.0690.2460.807Randomb)
  late0.0262.253 0.030 Randomb)
  late0.082–2.125 0.041 Randomb)
 Bti + competitionearly0.0480.0180.986Randomb)

The “Bti” treatment significantly reduced the number of mosquito larvae in the short-term, but not in the long-term. Indeed, during the early time period, at 3 days, 7 days, and 10 days after treatment, the “Bti” treatment resulted in significantly lower abundances of mosquito larvae than the control treatment (Figure 1). However, after day 3, when the lowest value was observed, the abundance of mosquito larvae increased significantly (days 3–14, Table 1), and by day 14 had reached a similar abundance to that recorded in the control (Figure 1). In the longer term (days 14–28), there were no significant differences between the control and “Bti” treatments in terms of either the abundance at specific days or changes in population size over that period of time (Figure 1, Table 1).

The “Bti+Competition” treatment caused a significant decrease in the abundance of mosquito larvae in the short term, with significant differences in relation to the control treatment evident on days 3, 7, and 10 after treatment (Figure 1). In contrast to the “Bti” treatment, mosquito larval population size did not increase significantly under the combined treatment at the end of the early time period but rather decreased further during the late time period (Table 1). Consequently, under the “Bti+Competition” treatment, the abundance of mosquito larvae was significantly lower than for the control treatment on almost all sampling days (Figure 1).

Effect of treatments on crustacean populations (abundances and composition)

For all treatments, the crustacean communities were comprised of organisms in three Orders: Cladocera (77.1%± 33.6%), Cyclopoida (14.7%± 31.1%), and Ostracoda (8.25%± 19.0%). Organisms within each of the three Orders were analyzed separately. Cladocera were dominated by Daphnia spp. (82.8%± 25.6%), with all other species each accounting for less than 3% of the total population of Cladocera. Abundances of crustaceans were initially very low (Cladocera at 7.20 ± 19.2 individuals/liter, Cyclopoida at 1.80 ± 6.69 individuals/liter, and Ostracoda at 0.13 ± 0.52 individuals/liter), and were similar in all treatments before the competitive crustacean communities were introduced (Figure 2).

Figure 2.

Changes in the abundances of (a) Cladocera, (b) Ostracoda, and (c) Cyclopoida. Abundances (mean ± SE) in different treatment groups [Control (N=4) = no treatment; Competition (N=3) = treatment with the introduction of a crustacean community; Bti (N=4) = treatment with Bti (1,000 μg/liter); Bti+competition (N=4) = treatment with Bti (1,000 μg/liter) and the introduction of a crustacean community] over time [P0 = before treatment; P1 = 3 – 14 days after treatment; P2 = 14 – 28 days after treatment]. *Significant differences compared to control (t-test, p < 0.05).

The abundances of Cladocera did increase in the control during the early time period (days 3–14), but stabilized at a higher level during the late time period (days 14–28) (Figure 2a). A similar pattern was observed for the “Bti” and “Competition” treatments (Table 1). Although in the “Competition” treatment, the abundances of Cladocera increased over time (Figure 2a), there was no significant difference as compared with the control (Figure 2a). In contrast, following the “Bti+Competition” treatment, the abundance of Cladocera increased significantly during the early time period (Table 1) and remained significantly higher than that of the control during the late time period (Figure 2a).

The abundance of Ostracoda increased slightly in the control during the early time period, but these species had apparently disappeared completely by the end of the late time period (Figure 2b). A similar observation was made following the “Bti” and “Competition” treatments (Table 1), with some significant differences in abundance evident between the “Competition” and control treatments at a few time points (Figure 2b). Following the “Bti+Competition” treatment, changes in the abundance of Ostracoda showed similar trends to those seen for the control treatment during the early time period. However, the abundance of Ostracoda increased significantly during the late time period (Table 1) and reached significantly higher values than those seen in the control at 21 days after treatment (Figure 2b).

Abundances of Cyclopoida remained stable over time for the control, “Competition,” and “Bti+Competition” treatments (Figure 2c). For the “Bti” treatment, the abundance increased slightly during the early time period (Figure 2c) and decreased significantly during the late time period (Table 1).

Correlations between changes in the abundances of crustaceans and mosquito larvae

The separate analysis of abundances of mosquito larvae and crustaceans described in sections 3.1 and 3.2 showed that during the late time period (days 14–28 after treatment), a low abundance of mosquito larvae was associated with a high abundance of crustaceans (Figures 1 and 2). The direct relationship between mosquito larvae and crustaceans was analyzed subsequently using the data from the last day of the observation period (day 28) to minimize the effect of the Bti treatment. The impact of crustaceans from the orders Cladocera, Ostracoda, and Cyclopoida on the abundance of mosquito larvae was analyzed using multiple linear regression. The results showed that only members of Cladocera, which was by far the best represented order in the crustacean population, had a significant impact on the number of mosquito larvae (ANOVA of multiple linear regression model, pCladocera= 0.004, pOstracoda= 0.374, pCyclopoida= 0.377) (Figure 3).

Figure 3.

Correlation between the abundances of mosquito larvae and Cladocera on day 28 for all treatment groups [Control (N=4) = no treatment; Competition (N=3) = treatment with the introduction of a crustacean community; Bti (N=4) = treatment with Bti (1,000 μg/liter); Bti+competition (N=4) = treatment with Bti (1,000 μg/liter) and the introduction of a crustacean community]. Linear regression (y = 2.30 – 0,004x, p = 0.003, R2= 0.512).


The results of the present field study, which was performed in temporary ponds in forested areas in Saxony (Germany), clearly show that the combined treatment of natural ponds with the biological insecticide Bti and the introduction of natural crustacean communities reduced the abundance of mosquito larvae more sustainably than single treatments that involved either Bti treatment or the introduction of crustaceans alone. This result is consistent with our earlier hypothesis that simultaneous application of Bti and introduction of crustaceans prolongs the effect of Bti application (Liess and Duquesne, unpublished data). The present study demonstrates the effectiveness of this approach under field conditions and revealed those mechanisms driving the positive effect of the combined approach.

The dominant mosquito species in all ponds analyzed was Cx. pipiens, which is a mosquito species that is found commonly in urban areas of Germany during the summer (Becker et al. 2010). Treatment of ponds with Bti alone almost completely eliminated the populations of Cx. pipiens larvae (as much as a 96% reduction in their sizes) within three days. However, these populations recovered after recolonization and had reached sizes similar to those of the control group within two weeks. This finding is consistent with other studies that showed Bti is active against mosquito larvae for only a few days (Karch et al. 1991, Aldemir 2009) and that repeated treatment is needed to ensure long-term reductions in the sizes of mosquito populations (Becker 2003).

Crustacean communities were dominated largely by Cladocera (mainly Daphnia spp.), which are common species in all types of freshwater ponds (Williams 2006). Besides members of Cladocera, members of Cyclopoida and Ostracoda were also present, although only Cladocera affected the size of the population of mosquito larvae significantly. The dominant role of Cladocera in this regard is consistent with other field and outdoor mesocosm studies, which have also demonstrated the negative effect of Cladocera on the establishment of populations of larvae of Cx. pipiens and Aedes spp. (Chase and Knight 2003, Meyabeme Elono et al. 2010, Duquesne et al. 2011). A correlation between the abundances of mosquito larvae and Cladocera spp. at the end of the observation period (day 28 after treatment) demonstrated that competition between these antagonists was density dependent (Figure 3). This is consistent with findings from studies that focused only on mosquitoes that showed that under both laboratory and field conditions, an increasing density of competitors is linked to increased mortality of the mosquito species of concern, delayed maturity, reduced adult size, and reduced adult longevity (Renshaw et al. 1993, Teng and Apperson 2000, Agnew et al. 2002, Braks et al. 2004, Reiskind and Lounibos 2009). In studies of outdoor pond mesocosms, both oviposition and the development of Cx. pipiens larvae were reduced more in the presence of high densities of Cladocera than in the presence of low densities (Duquesne et al. 2011). Hence, together with previous studies, the present study demonstrates that only large numbers of competitors can control populations of mosquito larvae.

However, competition is not a one-way road, and interspecific competition works in both directions. Indeed, the abundance of Cladocerans increased less following the introduction of a crustacean population alone than following the combined treatment. One explanation for this is that the reduced number of replicates used for the treatment that involved the introduction of crustaceans alone reduced the statistical power of the results for this treatment compared with the results obtained for the combined treatment. Another explanation is that competition alters according to sequence of introduction in a way that the competitor arriving first gains advantages of the competitor arriving later (Lawler and Morin 1993, Blaustein and Margalit 1996, Stokes et al. 2009). Indeed, (Foit et al. 2012) showed that larvae of Cx. pipiens delay the development of offspring of Daphnia magna when the sizes of D. magna populations have already been suppressed by application of a chemical compound. In contrast, established populations of D. magna affect both oviposition and the larval development of Cx. pipiens negatively (Duquesne et al. 2011), as well as time to metamorphosis and the size of larvae at the time of metamorphosis (Stav et al. 2005).

The timing of the succession of different populations influenced the outcome of the competition between mosquitoes and crustaceans. Simultaneous administration of Bti and introduction of a crustacean community disturbed the normal competitive interaction as a result of the ability of Bti to cause an initial reduction in the number of prior colonizers. In fact, from an ecological perspective, the Bti insecticide acts as a stressor that alters the interactions between competing groups of species, in this case, weakening the population of mosquito larvae (Griswold and Lounibos 2005, Juliano 2007). The decline in the size of the population of mosquito larvae promoted propagation of the introduced communities of crustaceans, enabling them to become the dominant group within two weeks and thus prevent recolonization of ponds by additional mosquito larvae. However, when established populations of mosquito larvae were not eliminated by Bti, as in the case of the control treatment or the treatment that involved introduction of a crustacean population alone, the development of a natural crustacean community was largely inhibited owing to the increased abundance of mosquito larvae.

Our results showed that competition is an important determinant of the community structures of ephemeral ponds (Blaustein and Chase 2007, Juliano 2009). Furthermore, given that competitors can be affected by each other, the temporal order in which species enter a system is of major importance because it can affect competitive processes. Early establishment of crustacean communities can be highly effective in the prevention of outbreaks of mosquitoes, and hence potential outbreaks of mosquito-borne diseases. In cases in which larval populations of mosquitoes are already established, combined treatment that involves the administration of a biological pesticide, such as Bti, and introduction of a crustacean community ensures sustainable control of the sizes of mosquito populations.


We are grateful to our colleague Kaarina Foit at the Department of System Ecotoxicology for her indispensable help and support with the statistical analysis. This work was kindly supported by the Helmholtz Impulse and Networking Fund through the Helmholtz Interdisciplinary Graduate School for Environmental Research (HIGRADE).