Mosquito larval habitat determines the fitness, survivorship, fecundity, and vector capacity of emerging adults. We manipulated larval density and food provisioning in the laboratory to determine their effects on a number of life history parameters of Anopheles gambiae. Larval mortality was positively correlated with larval density. In addition, increased density skewed sex ratios that favored females.
Density is another regulator of adult body size and survivorship. Larval development rate, adult survivorship, and adult body size all decrease as larval density increases (Schneider et al. 2000, Gimnig et al. 2002, Wiwatanaratanabatr and Kittayapong 2009). The intensity of density- dependant development in an ephemeral habitat that dries out quickly, such as water-filled hoof prints, will thus increase over time and eventually affect adult fecundity and vector capacity of the emerging cohort.
There are few studies that evaluate the effects of larval density on adult survivorship wherein a fixed amount of food per capita is employed. This design element is important because without it, one cannot say if larval density per se is the factor affecting larvae or if observed effects are simply due to decreases in the amount of food per individual. Put another way, information on the relationship between density-dependent or density-independent food availability and their interactions is lacking in most studies. There is at least one known field study in Kenya, where the effect of an additional input of food, or of 1st instar larvae, to pupal production was measured (Subra and Mouchet 1984). There, it was found that adding food promoted a significant increase in the number of pupae but adding larvae did not.
For years, ecologists have attempted to determine the impact of intraspecific competition during the larval stage on adult fitness and performance at varying densities (Paaijmans et al. 2009). From the perspective of medical entomology, the outcome of intraspecific competition for a vector species is critical to disease prevalence because density-dependent intraspecific competition may alter vector capacity of the adults (Takken et al. 1998). Hence, laboratory and field studies should focus on the importance of crowding in the larval stage to elucidate the effects on life history parameters of larvae and adults.
In this study, the interactions among same-age larvae in a given environment with specific amounts of food were evaluated. A series of experiments was conducted to determine the optimum amount of food for mosquito larvae to reach adulthood. The effects of larval density on some of the life table attributes, such as mortality, development, body size, etc. were measured. Here, food was provided on a per capita basis and also on an instar mass basis. The near optimum amount of food for a solitary larva to reach adulthood was determined. This amount was used in the subsequent experiments with different larval densities and container sizes to elucidate the interactions, competition, and cannibalism among the same age larvae of Anopheles gambiae s. s.
MATERIALS AND METHODS
All the experiments were conducted in a walk-in Conviron™ growth chamber at 30° C ± 2° C and 75–80% RH. Photoperiod was 12:12 (L:D). The Anopheles gambiae s.s. colony was brought from Njage, Tanzania, in 1997 and had been reared in the laboratory for several years for other experiments. Larvae were reared in plastic trays and fed Nutrafin fish food. A plastic pipette was used to transfer pupae to water-filled glass bowls, which were then placed inside 30×30×30 cm Plexiglas™ cages with mesh on five sides and cotton sleeves to access the cage on one side. Adults were fed human blood once per week and were provided with (5%) sugar solution.
Experiment 1: Determining optimal food level for a solitary larva
In this experiment, a single larva was reared per container. The container size was 3.75cm × 3.75cm × 6 cm and each larva had full access to 14.06 cm² space (the surface area) in its tub. Food was provided in two different amounts: low food (LF) and high food (HF). Each class had three different sub-classes, for example, low food treatment had LF1, LF2, LF3.
The amount of food was doubled in every set from the previous set. The amount also increased across instars because size increased with age (1st instar = 2–2.5 mm, 2nd instar = 3.5–5 mm, 3rd instar = 7–9.5mm, 4th instar = 10–12 mm). Assuming 1st and 2nd instar larvae consume less food; the total amount of food per capita, per biomass, in each tub, was less than that of 3rd and 4th instars, where the amount was increased as larvae grew to 80–90% of their body size and most biosynthesis took place during this stage (Briegel 2002).
A preliminary set of experiments was conducted with a very broad range of food provision (min-0.003mg to max 5 mg per larva) (Table 1). Predetermined amounts of Nutrafin® Basix Staple Tropical Fish Food were crushed, measured, and given to the larvae every day. The food was measured with a Cahn Electrobalance in microgram units. Food was provided on a per-instar basis assuming two days for each instar ± two days (there are four larval instars). For instance, for treatment LF1, 0.003 mg per larva*10 larvae for two days and so on. For each treatment value, there were five replicates. Therefore, there were 6×5 = 30 replicates in total. We also included a control with no food for all the treatment levels. The amounts that worked best were chosen for treatments 2 and 3.
Table 1. Values of various food provisions (mg per larva) for Anopheles gambiae when reared in individual containers. LF1, LF2, and LF3 refer to low food provision and HF1, HF2, and HF3 refer to high food provision.
Experiment 2: Different numbers of larvae/container with fixed per capita amount of food
In this experiment, the per capita amount of food and container size was fixed. Two different larval densities were chosen, namely low density or LD and high density or HD. In the LD treatment, the number of larvae was ten, whereas in the HD treatment, there were 200 larvae in a tub.
The size of the containers was chosen such that in low-density treatments, each larva had access to approximately 14.06cm² surface spaces as in experiment 1. Therefore, the container size was 12 cm × 12 cm × 12 cm. Hence, in low density tubs each larva had access to14.4 cm² surface area, which is close to 14.06 cm², whereas in high density tubs the surface space was 0.72 cm² for each larva. Each density had three replicates and an internal control per replicate, which was one tub in each sub classes, such as, LF1, LF2, etc. where no food was provided.
Experiment 3: Different numbers of larvae with fixed amount of food in larger size containers
Here, the above-mentioned experiment was repeated but with much larger containers to determine whether density-treatment effects were container-size specific. This experiment provided more space via more water volume and greater surface area for the larvae. In this experiment, the fixed variables were volume of food (from experiment 1) and number of larvae (from experiment 2). The container size was 25.4×38.1×6.35 cm. Therefore, in low density treatments, each larva was provided 96.77 cm² surface areas, whereas in high density treatments a larva was provided 4.83 cm² surface space in the tub. The density effect of this larger surface area was then compared with the result of experiment 2.
The water level in each rearing vessel was held constant at 5 cm deep per container. The water level was checked daily and evaporated water was replaced by the addition of distilled water whenever needed. The water was changed if it appeared sufficiently polluted to impact larval survival (based upon prior experience by Khandaker Jannat). The pollution occurred when larvae did not consume as much food as expected in the later instars, or the number of larvae was reduced by cannibalism. When this occurred, uneaten-provisioned food often created a biofilm on the surface of the water, which caused larval death. The containers were placed in a random order to minimize possible position effect.
The total-larval period was set to ten days; larvae that failed to pupate in that time period were discarded. Emerged adults were collected whenever observed and placed in small, labeled cuvettes. All of the cuvettes were placed in a refrigerator at 4° C for two h after adult collection, to kill the adults for further study. Larval mortality rate for each treatment was measured by recording all dead larvae every day, per container. Dead larvae were removed whenever they were observed. Missing larvae from each treatment were also recorded as dead.
Mean age at pupation was expressed by t, where, t= No. of days to pupation × No. of larvae pupated / total larvae pupated (Lyimo et al. 1992). Collected pupae from each treatment were kept in a small cage marked with the treatment names until emergence. Pupae that failed to hatch were discarded. The pupae were assessed twice per day for eclosion. Upon emergence, adults from each treatment were separated by sex and placed in the previously labeled microcentrifuge tubes. All tubes were placed in a refrigerator at 4° C.
Adult body size was determined by measuring the wing length (both wings) of the adult mosquitoes. The wing length measure was the distance from axial incision to the apical margin, excluding fringe of scales (Rohani et al. 2004). The mean of both wings was calculated. Wing length is considered as an indicator of body size because it is directly proportional to dry body weight. Differences in wing length by sex were measured with an ocular micrometer set at 0.64 mm/ocular unit.
The effect of nutrient provision on mortality rate, mean age at pupation, survivorship, and adult body size was evaluated. All the data showed a normal distribution, therefore, parametric tests (Student's t-test, analysis of variance, and Tukey's honestly significant difference test) were used to test the significance of the treatments. Statistical analyses were conducted using JMP 8.0 (SAS Institute Inc. 2005). Graphs were generated using GraphPad Prism 5.00 (GraphPad Software Inc. 2005).
In the second and third experiments, larval mortality was higher in high-density treatments. High density increased mean age at pupation in experiment 3 but not in experiment 2 but produced small-bodied adults in both the second and third experiments. In both low- and high-density treatments, males were protandous, emerged first, and had a smaller body size.
Experiment 1: Solitary larvae provisioned with different amounts of food
Results showed no significant differences among the three different amounts of food in the low food (LF) category (p value =0.2385, F ratio = 0.1847) and HF category (p value = 0.3982, F ratio =0.7649). Therefore, results of LF1, LF2, and LF3 were pooled to calculate the average and termed as LF. Similarly, the average for HF was used. In the case of the internal control, mortality was 100% at 1st instar. For low food treatments, the mortality rate was 100% for all three treatment values for LF (LF1, LF2, and LF3) before pupation. For high food treatments, the mean mortality was 20%. Mean age at pupation (t) for the HF treatment was 8.22.
In HF treatments, the male and female sex ratio was 49:51, respectively. Females were significantly larger than males (Figure 1), (p = <.0001*; F ratio = 22.693). Since larvae from low food treatments did not survive to pupation, the food provision from high food (HF) treatments was chosen for experiments 2 and 3. In experiments 2 and 3, larval densities were higher than experiment 1. When food was provided on a per capita basis, the larger amounts of food from the HF treatments caused larval death by polluting the water when larvae reached the later instars. Therefore, HF1 amount, which was the lowest food amount among the three HF amounts was determined to be the optimal food provisioning rate and provided the basis chosen for experiment 2 and 3.
Experiment 2: Different numbers of larvae/container with fixed amount of per capita food
Mean mortality rate in low density LD (LD1, LD2 and LD3) was 60.55% and high density HD (HD1, HD2 and HD3) was 96.44%; this difference was significant (p value 0.0449*) (Tables 2, 3). Mean age at pupation in the case of LD was 8.38 and HD was 7.16. Statistical analysis showed that larvae in the greater space treatment performed significantly better (p value 0.0049*) (Tables 2, 3) with regard to mean age at pupation (Tables 2, 3).
Table 2. Effect of two larval densities, low density (LD) and high density (HD), on mortality rate, mean age at pupation, and adult body size of Anopheles gambiae s.s. * indicates a statistically significant effect (ANOVA α= 0.05).
Sum of squares
2Experiment 2: Different numbers of larvae/container with fixed amount of food.
3Experiment 3: Different numbers of larvae with fixed amount of food in larger size containers.
Mean age at pupation2
Mean age at pupation3
Adult body size2
Adult body size3
Table 3. Effect of two larval densities, low density (LD) and high density (HD), on mortality rate, mean age at pupation, and adult body size of Anopheles gambiae s.s. Means with different letters are significantly different (Tukey's HSD α= 0.05).
Mean ± SE
1Experiment 2: Different numbers of larvae/container with fixed amount of food. Experiment 3: Different numbers of larvae with fixed amount of food in larger size containers.
96.258 ± 1.579 (A)
65.444 ± 14.076 (B)
6.666 ± 2.795 (A)
52.444 ± 2.795 (B)
Mean age at pupation
9.678 ± 2.279 (A)
2.295 ± 0.899 (B)
6.245 ± 0.384 (A)
9.886 ± 0.412 (B)
Adult body size
3.456 ± 0.032
3.406 ± 0.052
3.758 ± 0.085 (A)
3.281 ± 0.039 (B)
Male and female sex ratio at both densities was 22:78 and males emerged before females. There was no statistically significant difference in male vs female body size, but there was a trend for males to be smaller than females. Adults from LD treatment were bigger than HD treatment (Figure 2) but the difference was not statistically significant (p = 0.4209) (Table 2). Lower density produced bigger adults with longer developmental time.
Experiment 3: Different numbers of larvae with fixed amount of food in larger size containers
The mean larval mortality from LD treatment groups was 6.66%. For HD treatment groups it was 42.33% (Figure 3). The difference was significant (p value <.0001*) (Tables 2, 3). Mean age at pupation in LD was 6.27 and HD was 9.8 days (Figure 4), which was significant (p value <.0001*) (Tables 2, 3). Adults from LD treatments were significantly (p = <.0001*) (Tables 2, 3) larger than those of HD treatments (Figure 5).
For species with complex life histories such as mosquitoes, larval environment always plays a vital role in determining adult fitness. Changes in larval density and food availability can regulate the vectorial effectiveness by changing adult body size (Takken et al. 1998). In this study, crowding produced significantly smaller adults of both sexes with elevated mortality and lower survivorship. More importantly, because our experiments separated crowding per se from the effects of crowding via reduced food consumption, it appears that larval crowding can act both directly and indirectly on larval performance. Similar results have been found in studies that focused on culicines (Hawley 1985b, Lawrie 1966, Macia 2009). This crowding effect is common in other organisms that have aquatic larval stages, such as marbled salamanders, Ambyostoma opacum, where low density produced larger adults with more lipid storage and a higher survival rate than high larval density (Scott 1994). Tadpoles of Rana sphenocephala metamorphosed earlier at high density than tadpoles at low density whereas, amount of food did not have any impact (Richter et al. 2009). In weakfish, Cynoscion regalis, larval density affects larval growth dramatically but larval mortality was independent of larval density (Duffy and Epifanio 1994).
In the density-dependent experiments, mortality was higher in the high-density treatments. Although food was given at optimal rates based upon experiment 1, density still played a role in higher mortality. Even though larvae had access to the entire amount of resources (larval food provided) in the environment, the food provision could have been polluting water quickly enough to kill the wrigglers and competition, cannibalism, etc. also played some role in the larval death. But in experiment 1 (one larva per container) nutritional stress was the reason of high mortality. Here, larvae did not face any competition for food but the amount was not sufficient for it to reach to pupation. Legros et al. (2009) stated that density-dependent mortality occurs when all individuals have access to the food in the environment but the amount is such that each individual obtains, or is likely to obtain, a suboptimal share; a form of scramble competition (Smith and Smith 2006).
Mean age at pupation
Life-history theory suggests that the timing of metamorphosis is of major importance (Rowe and Ludwig 1991, Abrams et al. 1996), because individuals experience a strong reduction in fitness if they fail to reach a specific stage before a set time (Rudolf and Rodel 2007). In most of the low-density treatments, larvae pupated in seven to eight days, whereas in most of the high-density treatments some larvae pupated early but the rest remained in the 3rd or 4th instar until the tenth day. These larvae never pupated even when they were kept in the container for longer periods of time; they remained in the 3rd or 4th instar for over a month and eventually died without pupating. Arrivillaga and Barrera (2004) found that 3rd instar Aedes aegypti can survive in a food-stressed condition for a significant number of days. The larvae that never pupated received less food than required per day and were forced to draw on energy reserves but never could accumulate sufficient energy to develop to older instars or pupae stages. Day and Rowe (2002) published a model, which is an extension of the Wilbur-Collins model for amphibian development, where they included a threshold below which maturity is not possible.
In experiment 2, mean age at pupation in the case of LD was 8.38, and HD was 7.16. High density will increase larval competition but it is not clear why this leads to decreased larval time. Since the larvae were not suffering from low per capita food provision, perhaps they were attempting to escape a difficult situation and had sufficient nutritional reserves to do so. Of course, this situation is somewhat artificial, with very high density but adequate food provisioning. However, in experiment 3, the high-density treatment generated a greater mean age at pupation. This result supported the observation of Gimnig et al. (2002). Further experiments that would titrate pupation age against density might explain the discrepancy between experiments 2 and 3.
Larvae that did not pupate on time might start to consume other larvae in the environment to reach pupation. Thus, some larvae were capable of successfully completing the larval stage later. Optimum time of metamorphosis thus expanded under several limiting factors. This kind of plasticity of growth and age at metamorphosis can lead to differences among individuals of the same cohort in survival, vector capacity, and fecundity.
In experiment 2, the emerged male to female ratio was 22:78, which varies greatly from the typical 50:50 ratios. The reason for such a high female ratio is unknown. One possible reason could be cannibalism; male larvae were cannibalized more than female larvae. Adult body size showed that females were larger than the males. In such cases, males were also smaller than females during the larval or pupal stages. Smaller larvae are always more susceptible prey than larger larvae. Hence, most of the male larvae were cannibalized by the larger larvae.
On the other hand, a sex ratio skewed to females has also been observed in field populations in the case of Anopheles gambiae s.l. in a study in Western Kenya (Mutuku et al. 2006). Again, in some insect families, female-biased sex ratios at high larval densities have been observed before. Cipollini (1991) stated that male-biased mortality at higher larval densities produced a female-biased sex-ratio in the case of a weevil, Acanthos celidesobtectus. In this study, the magnitude of density-dependent larval development and mortality thus caused the distorted sex ratio. One caveat from our study is that the primary source of food was high-protein fish food whereas in nature, algae may be a significant portion of the diet (Kaufman et al. 2006). However, there is no reason to believe that the results from the current experiments would have differed had we used a mixed diet and that crowding per se would be mitigated by an algal addition to the fish food diet.
In conclusion, the experiments suggested that there is a cost per se from being crowded, either through stress or perhaps through direct energetic costs if larvae expend extra energy interacting with each other. Amount of food consumption was crucial for a larva to successfully complete larval life. Determining optimum amount of food per larva limits all other possibilities, e.g., scarcity of food, crowding effect, etc., of affecting adult fitness in density-dependent experiments. This is the first study of the effect of larval food on adult life where the optimal amount of food was provided to each larva, which may eliminate the possibility of prolonged larval stage and less exposure to predation by conspecifics.
This work was supported by an NSERC Discovery grant to BDR. We thank Dr. Gerhard Gries and Dr. Martin Adamson for comments on an earlier version of this paper.