Abundances of mosquito larvae and associated invertebrate communities were assessed in 27 temporary ponds during the spring season in wetland areas of Germany. Four genera of mosquitoes were identified: Aedes, Anopheles, Culex, and Culiseta. We focused our analyses on Aedes spp. because this genus was the most abundant (92% of total abundance) and frequently encountered mosquito (present in 65% of investigated sites). The abundance of Aedes spp. was negatively associated with the abundance of competitors for food, and to a lesser extent with those of intraguild predators and strict predators. The influence of these natural antagonists on larvae of Aedes was stronger in ponds with higher levels of dissolved oxygen (53 ± 4%) than in ponds with lower levels (16 ± 1%). The overall abundance of antagonists explained 42% of the variation in abundance of Aedes spp. at sites with higher levels of dissolved oxygen. Of this explained variation, competitors accounted for 34.7%, whereas the abundance of intraguild predators and strict predators accounted for only 6.8 and 0.5%, respectively. Therefore, the promotion of competing species might be an appropriate ecological approach for the control of Aedes spp. in temporary ponds in these areas.
Numerous vector-borne diseases are transmitted to humans and animals by mosquitoes in both tropical and temperate regions. For instance, the transmission of malaria involves mosquitoes of Anopheles species, whereas the transmission of arbovirus infections and filariasis mostly involves species of Aedes and Culex (Becker et al. 2003). Owing to the importance of mosquitoes as vectors for diseases in terms of public health, the ecological and environmental conditions that influence the abundances of these species are of great interest (Chaves and Koenraadt 2010).
A wide variety of aquatic environments (e.g., marshes, ponds, wells, drainage channels, lakes, and rivers) serve as breeding sites for mosquito larvae (Becker et al. 2003). However, many other invertebrate taxa (e.g., Crustacea, Acaria, and insect larvae) share the same habitats (Campos et al. 2004, Bambaradeniya et al. 2004) and interact with mosquito larvae through competition and predation.
In this investigation, we examined the aquatic communities of typical mosquito breeding sites, focusing on the mosquito larvae and accompanying invertebrates of temporary ponds within natural wetlands of Central Germany. The primary aim of the study was to determine the influence of antagonist species on larval populations of mosquitoes. We hypothesized that the abundance of mosquito larvae is inversely related to the densities of associated invertebrates that are potential antagonists. We investigated these associations at three locations in natural wetlands: 1) a flood plain of the middle Elbe (Rosslau), 2) a flood region (Spreewald), and 3) a flood plain of the River Parthe (Leipzig). Abiotic parameters (temperature, dissolved oxygen, pH, turbidity, emergent vegetation cover, and water depth and surface area) were also assessed to investigate the influence of these parameters on the abundances of mosquito larvae and their antagonists, as well as on their relationships.
We also attempted to distinguish the respective effects of competitors and potential predators on mosquito larvae. The abundance of mosquito larvae is limited mostly by competitors in natural temporary ponds and by predators in permanent ponds (Chase and Knight 2003). However, some predator species are also adapted to temporary ponds (Brendonck et al. 2002, Kumar and Ramakrishna 2003, Becker et al. 2003), and therefore may also play a role in the regulation of larval populations of mosquitoes.
MATERIALS AND METHODS
The study sites were located in wetlands in three different federal states of Central Germany where mosquitoes were known to be prevalent: Rosslau (Saxony-Anhalt), Spreewald (Brandenburg), and Leipzig (Saxony). The sites in Rosslau (51° 52' N, 12° 14' W) were located in the floodplain of the river Elbe. The sites in Spreewald (51° 02' N, 13° 53'W) were located in flood areas in a region of traditional irrigated agriculture that contained more than 200 small channels within an area of 484 km2. The sites in Leipzig (51° 20' N, 12° 21' W) were located in the flood plain of the river Parthe. Twenty-seven sites in total were investigated. The numbers of sites per location were eight, ten, and nine for Rosslau, Spreewald, and Leipzig, respectively.
Monitoring communities of aquatic invertebrates and mosquito larvae
The sampling was performed once a week in each study area (Rosslau, Spreewald, and Leipzig) during the spring of 2007 (April 11th– June 13rd). Five subsamples of one to three liters (depending on the pond size and the load of suspended matter in the water) were collected at different points (with and without vegetation) in each pond, pooled into one sample, and filtered through a 55-µm mesh (Turner and Trexler 1997). The filtrate was conserved in 200 ml of distilled water and transported in plastic flasks to the laboratory to determine the taxa that were present. Micro-invertebrates (<5 mm in length) were identified and counted before further treatment to avoid the distortion of the shape of the ciliates and some rotifers by exposure to ethanol, which was used as a conservative agent (ethanol). Macro-invertebrates (>5 mm in length) were conserved with a mixture of ethanol:distilled water (70:30). The following identification keys were used: Ward and Whipple (1959), Durand and Lévêque (1980), Schwab (1995), Narchuk and Glukhova (1999), Becker et al. (2003), Tachet et al. (2003), and Streble and Krauter (2006).
Abiotic parameters were measured between 09:00 and 11:00 on each day the mosquito larvae and invertebrates were sampled. The percentage of emergent vegetation cover and the surface area of the water were estimated visually; water temperature and pH were determined with an electronic pH meter (HANNA, Woonsocket, U.S.A.); dissolved oxygen (DO) with an electronic oxymeter (ExStik DO600, Extech, Walthman, U.S.A.), and turbidity with a turbidity meter (Turbiquant 1100 IR, Merck, Darmstadt, Germany). The depth of the water was assessed with a ruler, and the mean value of two to five random measurements (depending on the size of the pond) taken at different points in each pond was used.
The sites that were investigated were all temporary ponds that dried out at least once during the period of the study. Samples that were collected from the same site before and after it had dried out were considered to be independent samples (IS). This is because in general (i) the abundance of mosquito larvae and the size of the ponds varied before and after drying out (Table 1) and (ii) these two variables, abundance of mosquito larvae and size of ponds, were not correlated. As a consequence, we collected a total of 77 independent samples from the 27 sites that were investigated. The numbers of independent samples per location and per site are shown in Table 1.
Table 1. The abundance of mosquito larvae and water surface area (size) for independent samples (IS) from sites at Rosslau, Spreewald, and Leipzig.
The datasets at the sites of Rosslau, Spreewald, and Leipzig were analyzed both separately and in combination. The results from data on specific locations confirmed those on combined locations. Hence, only results obtained from combined data are shown in this paper (unless otherwise indicated). Data from independent samples (as explained above) were utilized for statistical analyses. They were subjected to log (x +1) transformation prior to all analyses.
Analysis of variance (ANOVA) was used to examine differences in the abundances of mosquito larvae between Rosslau, Spreewald, and Leipzig. Principal component analysis (PCA) was used to highlight the relationships between taxa that were identified during the present study. PCA is a linear unconstrained multivariate ordination method that is appropriate for describing variations in complex systems that are characterized by many species (Leps and Smilauer 2003). We chose a linear method because the length of the gradient determined by preliminary detrended correspondence analysis (DCA) was short (i.e., 2.005). The associations that were detected in the PCA between mosquito larvae and antagonists were tested for significance using the hierarchical model of multiple linear regressions. The hierarchical model with change statistics was used to assess the partial contribution of competitors, intraguild predators, and strict predators to the predictive capacity of antagonists with respect to the abundance of larvae of Aedes spp. Competitors, Cyclopoida, and predators were entered as first, second, and third blocks in this analysis, respectively. This order represents the increasing importance of the respective antagonists in temporary ponds (Chase and Knight 2003). The variance that is explained by each additional group of antagonists (R2 change and the corresponding F test) is given at each stage of the regressions.
The influence of the abiotic parameters on the species abundance of mosquito larvae and antagonists was determined using redundancy analysis (RDA), which is a constrained form of PCA (Leps and Smilauer 2003). In addition and using a t-test, we compared abiotic parameters between sites that simultaneously contained low abundances of mosquito larvae and antagonists and the other sites. This made it possible to check whether the abiotic parameters could explain the low occurrence of both mosquito larvae and antagonists in some study sites.
Outliers were detected using the Grubbs' test that detects one outlier at a time in a univariate dataset. The outlier identified is expunged and the test is iterated until the results show no outliers in the dataset. The Grubb's test uses the procedure of the extreme studentized deviate method. The ratio Z, which is the ratio of the difference between the extreme value under analysis and the mean to the standard deviation (SD) from all values, including the extreme one, is calculated using the formula:
(GraphPad Software, Inc. 2005). Critical values for Z are given according to sample sizes in an extra table in the software. If the calculated value of Z is greater than the critical value for the sample size, the P value is less than 0.05 and the extreme value is considered to be an outlier.
Mosquito larvae were found in 71% of the independent samples studied. Four genera of mosquito were identified: Aedes, Anopheles, Culex, and Culiseta. Aedes spp. were the most abundantly collected (92% of the total numbers of larvae) and the most frequently encountered (present in 65% of the total number of sites). The other genera of mosquitoes (Anopheles, Culex, and Culiseta), which represented the remaining 8% of the total number of larvae, were present in 22% of the sites (Table 2). The mean abundances of mosquito larvae showed no significant difference between the Rosslau, Spreewald, and Leipzig regions (the means and standard errors (medians) were 6 ± 2 (1), 11 ± 6 (1), and 5 ± 1 (4) individuals per liter, respectively; data not shown; ANOVA, P > 0.05). Owing to the fact that Aedes spp. were the dominant mosquito species at the study sites (Table 2), we focused our analyses on these mosquitoes. Whenever the mosquito species that were characterized by low abundances were considered, they were classified as the single group “other mosquitoes.”
Table 2. Abundances (mean ± SE (median)) and frequencies of mosquito larvae and invertebrate taxa for all samples (n = 77) collected at the three locations of Rosslau, Spreewald, and Leipzig.
aLarvae of Diptera other than mosquitoes.
Aedes spp. larvae
7 ± 3 (0.6)
Larva of other mosquitoes
0.6 ± 0.2 (0)
63 ± 44 (0)
35 ± 8 (13)
7 ± 2
488 ± 104 (104)
302 ± 58 (78)
362 ± 45 (260)
8 ± 2 (0.6)
0.2 ± 0.05 (0)
0.1 ± 0.02 (0)
0.1 ± 0.05 (0)
0.1 ± 0.03 (0)
1 ± 0.5 (0)
3 ± 1 (0)
7 ± 2 (0)
Abundances and distribution of associated invertebrate taxa
The associated invertebrate taxa that were identified included Ciliata, Rotifera, Microcrustacea (Cladocera, Copepoda, and Ostracoda), Isopoda (Asellus), Annelida, Nematoda, Planaria, and larvae of the insects Dytiscidae, Hydrophilidae, Scirtidae (Hydrocyphon), Diptera (mosquitoes, Dixa, Mochlonyx, Chironomidae, Chaoboridae, and Stratiomyidae), and Odonata (Anisoptera and Zygoptera). Microcrustacea were the most abundant and the most frequently encountered of the invertebrate taxa among the samples (Table 2). Insects other than mosquitoes, with the exception of Diptera, were the least abundant. The abundances and frequencies of all invertebrate taxa among the sites are shown in Table 2.
Influence of associated invertebrate taxa on Aedes larvae
The results of the principal component analysis (PCA) showed that the abundance of larvae of Aedes spp. was associated negatively with the abundance of antagonists and more specifically of (a) some competing taxa (Ceriodaphnia, Chydorus, Daphnia, Simocephalus, Calanoida, and larvae of Chironomidae), (b) some strict predatory taxa (insect larvae of the groups Chaoborus, Dytiscidae, Hydrophilidae, Anisoptera and Zygoptera), and (c) intraguild predators (Cyclopoida species) (Figure 1). Larvae of Aedes spp. were also associated negatively with the abundances of larvae of other mosquitoes (Figure 1). Slight and positive associations were found between the abundance of larvae of Aedes spp. and those of other potential competitors (Ostracoda, Harpacticoida, Annelida, and the Scirtidae Hydrocyphon) or predators (Figure 1).
Interactions between associated invertebrate taxa and Aedes spp. in relation to the abiotic parameters
In a first analysis, RDA was carried out on data for all sites to test for the influence of abiotic parameters on the abundances of mosquito larvae and associated invertebrate communities. The results showed that larvae of Aedes spp. and their antagonists (identified in Figure 1) did not correlate significantly with the abiotic parameters that were investigated (data not shown). Only Ostracoda which had a slightly positive association with Aedes spp. (Figure 1), showed significant correlations with turbidity, emergent vegetation cover, pH, and water surface area (data not shown). The means (medians) of these parameters for all sites were as follows: temperature, 15.2 ± 0.4° C (14° C); DO, 32 ± 3% (27%); pH, 7.06 ± 0.08 (7.2); turbidity, 18 ± 3 NTU (8.3 NTU); depth, 0.12 ± 0.07 m (0.11 m); water surface area, 400 ± 40 m2 (300 m2); and emergent vegetation cover, 50 ± 4% (70%).
In a second analysis, we divided the data set into two categories: (1) sites with lower abundances of both Aedes larvae and antagonists (<1 larva/liter and <100 antagonists/liter); these sites had low levels of DO (16 ± 1%) (n = 42), and (2) sites with higher abundances of organisms (Aedes larvae and antagonists) that had higher levels of DO (53 ± 4%) (n = 35). The difference in DO between these two categories was significant (t75 assuming equal variances, P = 0.018). The thresholds of one larva/liter and 100 antagonists/liter were selected, as these values, were close to the medians for larvae of Aedes spp. (0.6 individual/liter) and antagonists (89 individuals/liter) over all sites. The sites with high levels of DO contained approximately 65% more antagonists than the sites with low DO (medians of 127 and 77 individuals/liter, respectively). We used linear regressions to test for the significance of the associations that we found between larvae of Aedes spp., other mosquito larvae, and antagonists (Figure 1). This analysis was carried out for both categories (sites with higher and sites with lower DO) and the results obtained are detailed below.
At sites with higher levels of DO
In this subset (n = 35), one sample was found to be an outlier (Figure 2A) by the Grubb's test (P<0.05). The results that were obtained both including and excluding the outlier are presented in this section because of the substantial influence of this single outlier on the analyses.
When the outlier was not considered, the abundance of the pooled antagonists explained 42% of the variation observed in the abundance of larvae of Aedes spp (100 ×ΣR2 change, Table 3). With respect to the explained variation, the abundance of the competitors accounted for 34.7% (Figure 2B), whereas the abundances of the intraguild and strict predators accounted for 6.8 and 0.5%, respectively (Table 3). The influence of the other mosquitoes on the effects of the antagonists was not significant (P>0.05) and accounted for only 0.8% (data not shown).
Table 3. Multiple linear regressions (hierarchical model) showing the negative relationships between the abundances of larvae of Aedes spp. and antagonists in sites with high levels of dissolved oxygen and low dissolved oxygen. The partial contribution (R2 change) of competitors, intraguild predators, and strict predators to the predictive capacity of antagonists with respect to the abundance of larvae of Aedes spp. is provided.
When the outlier was considered, the abundance of the pooled antagonists explained 22.4% of the variation observed in the abundance of larvae of Aedes spp. (Table 3). With respect to this explained variation, the competitors, intraguild predators, and strict predators accounted for 11.2, 7.2, and 3.9%, respectively (Table 3). The influence of the other mosquitoes on the effects of the antagonists was not significant (P>0.05) and accounted for only 1.9% (data not shown).
Overall, the findings obtained for the sites with higher levels of DO (“with” and “without” the outlier, Figures 2A and 2B, respectively) revealed that among the groups of antagonists studied, the abundance of competitors was the main factor that affected the abundance of larvae of Aedes spp. (R2 change, Table 3). The partial contributions of the abundance of intraguild and strict predators were small (R2 change, Table 3) and were not significant (P change, Table 3). Factors that could explain the difference between the outlier and the other observations were not clear. The difference could be related to parameters that were not investigated during the present study.
At sites with lower levels of DO
No outliers were identified in this subset (n = 42). At sites with lower levels of DO, the abundance of the antagonists explained only 5.4% of the variation observed in the abundance of Aedes spp. (Table 3). Therefore, the negative influence of antagonists on larvae of Aedes spp. was small and not significant at these sites (Table 3).
Abundance of Aedes species
Aedes species were the most abundant of the mosquitoes recorded in this study, which was carried out during the spring of 2007 in wetlands of Rosslau, Leipzig, and Spreewald. This finding with respect to larval stages corresponds to that of Schäfer et al. (1997), who reported a predominance of Aedes species, including Ae.communis, Ae. rusticus, Ae. punctor, Ae. cantans, and Ae. dianteus, among adult mosquitoes collected from late April to early June, 1993, in Bienwald, Germany. Therefore, Aedes species seem to be the predominant mosquitoes in Germany during the spring. Members of this genus are involved potentially in the transmission of arboviral infections and filariasis in Germany (Becker et al. 2003). Therefore, the control of Aedes species as potential vectors of disease is very important.
Influence of associated invertebrate taxa on larvae of Aedes species
Our results showed that the abundance of Aedes spp. was mainly negatively correlated with the abundance of food competitors. Therefore, when present in the same habitat, Ceriodaphnia spp., Chydorus spp., Daphnia spp., Simocephalus spp., Calanoida, and larvae of Chironomidae compete efficiently for food resources and affect the abundance of Aedes spp. Similarly, Chase and Knight (2003) demonstrated that non-mosquito competitors (larvae of Chironomidae and Cladocera) limited the abundance of Anopheles quadrimaculatus and Culex pipiens to a great extent in temporary ponds of Northwest Pennsylvania (U.S.A.). All the relevant species of competitors that we identified in our study are known to inhabit the littoral zones of ponds (Adamczuk 2006). Obviously, competition between these species and the mosquitoes that live mostly in shallow waters (Abdullah and Merdan 1995) is especially strong.
Our results also showed that predators influenced, although to a lesser extent than competitors, the larval populations of Aedes spp. in temporary ponds. Cyclopoida, which are intraguild predators of mosquito larvae, were found to act as antagonists in our study. This is in agreement with the results of a number of laboratory and field studies around the world that have shown that Cyclopoida prey on mosquitoes (Marten and Reid 2007). Strict predators of mosquitoes such as Zygoptera and Dytiscidae exerted the smallest influence in our study, which might have been due to the relatively low abundances of these taxa in temporary ponds. Banerjee et al. (2009) found no larvae of Aedes species in habitats that contained strict predators such as beetles and Odonata. The negative relationship reported by these authors might be a consequence of the fact that the investigated ponds had been in existence for long periods. Indeed, such biotopes shelter abundant predators, as mentioned by Schneider (1997).
Influence of the abiotic parameters
The abiotic parameters investigated, with the exception of dissolved oxygen (DO), did not influence the abundance of the antagonists identified in the present study. Given that the investigation took place during one spring season, the restricted variations in abiotic parameters between sites may not have been sufficient to elicit differences in species abundances.
Concerning larvae of Aedes spp., the results support the hypothesis that mosquito larvae, including Aedes spp., are not particularly sensitive to variations in the water-quality parameters measured here. For example, in a laboratory study, the survival of larval Ae. aegypti was not affected by pH values that ranged from 4 to 11 (Clark et al. 2004). Similarly, in a field study, water-quality parameters were found to be of minor importance for the abundance of mosquito larvae (Beketov et al. 2010).
However, the results of the study reported herein showed that the sites with simultaneously lower abundances of Aedes spp. and antagonists differed from the sites with higher abundances of these organisms in terms of the level of DO, and that the interactions that were detected between the two groups differed. In fact, the negative correlation between antagonists and larvae of Aedes spp. was much stronger in water with high levels of DO than in water with low DO. Therefore, the influence of antagonists on the abundance of Aedes spp. might be particularly strong in sites with high concentrations of DO. The present study highlights the importance of the indirect effect of an environmental parameter on interactions between mosquito larvae and their associated communities.
Our results showed that invertebrate species that are antagonists of mosquito larvae limit the larval populations of Aedes spp. during spring and thus reduce the suitability of temporary ponds as mosquito breeding sites in wetland areas in European temperate regions. Among these groups of antagonists, the food competitors for food influenced the larval populations more strongly than predators. Therefore, competitors might have potential as biological agents for use in the control of the abundance of larvae of Aedes spp. Such findings with regards to the ecological conditions related to biological interactions between mosquito larvae and associated species are important for the implementation of appropriate control measures and integrated management of areas infested with mosquitoes.
We are grateful to Mikhail Beketov and Kaarina Foit for help with the statistical analyses. We thank the anonymous reviewers whose comments and suggestions improved the present manuscript greatly. This study was funded by the DAAD (German Academic Exchange Service) and the Helmholtz Centre for Environmental Research – UFZ Leipzig, Germany.