Ecology, 83(3), 2002, pp. 669±679 q 2002 by the Ecological Society of America OVIPOSITION HABITAT SELECTION BY MOSQUITOES (CULISETA LONGIAREOLATA) AND CONSEQUENCES FOR POPULATION SIZE MATTHEW SPENCER,1,3 LEON BLAUSTEIN,1 AND JOEL E. COHEN2 1Community Ecology Laboratory, Institute of Evolution, University of Haifa, Haifa 31905 Israel 2Laboratory of Populations, Rockefeller University, 1230 York Avenue, Box 20, New York, New York 10021-6399 USA and Columbia University, USA Abstract. Many kinds of adaptive behavior, including responses to risk of predation, have been documented, but there have been few attempts to translate these behaviors into consequences for populations. We present one of the ®rst models to predict the consequences of adaptive behavior for population size in a speci®c natural system. Larvae of the mosquito Culiseta longiareolata (Diptera: Culicidae) develop in freshwater pools. They are vulnerable to predation by the backswimmer Notonecta maculata (Hemiptera: Heteroptera), and to intraspeci®c competition. Adult female C. longiareolata usually avoid ovipositing in pools that contain N. maculata. This is presumably an adaptive response that increases individual ®tness, but it is also likely to affect the size of the population. We take a novel approach to understanding the relationship between adaptive behavior and population dynamics in C. longiareolata. We use a nonlinear stage-structured population model to predict the evolutionarily stable oviposition strategy and its consequences for the size of the C. lon- giareolata population. Our model predicts that female C. longiareolata should always avoid ovipositing in pools with N. maculata. Such avoidance will increase the equilibrium size of the C. longiareolata population, relative to a population in which oviposition is indis- criminate with respect to N. maculata. The qualitative effect on population size is the same even if, as observed, C. longiareolata occasionally oviposit in pools containing N. maculata. These predictions have important practical implications for assessing the effectiveness of predators as biological control agents. Key words: adaptive oviposition behavior; biological control; Culiseta longiareolata; evolution- arily stable strategy; habitat selection; Notonecta maculata; population dynamics; risk of predation; stage-structured population model. INTRODUCTION be profoundly in¯uenced by social behavior, through its effects on competition for resources (Clutton-Brock Adaptive behavior is generally believed to have im- and Albon 1985). However, only a few attempts have portant consequences for population dynamics (Suth- been made to model adaptive behavior and population erland 1996, Fryxell and Lundberg 1998). Behavioral dynamics in speci®c natural systems. Diehl et al. (2000) assumptions are implicit in the parameters of most pop- showed that in a model of an open system containing ulation dynamic models (Hassell and May 1985), and predators, grazers, and producers, reduced foraging and models in which individuals are allowed to modify their increased emigration by grazers in response to in- behavior so as to maximize their ®tness can show very creased predator densities (assumed to be adaptive re- different dynamics from models in which behavior is sponses) could result in increased producer biomass. ®xed (Parker 1985). For example, optimal foraging by These predictions were in broad agreement with the predators affects the functional response linking prey results of an arti®cial stream study using trout, inver- density to consumption rate (Chesson and Rosenzweig tebrate grazers, and benthic algae. Smith et al. (2000) 1991). Host selection behavior may be an important developed a model for the population dynamics of a factor in the effectiveness of parasitoids as biological ®sh (the bitterling) whose larvae are parasitic on fresh- control agents (Luck 1990). Many empirical studies water mussels, and predicted that avoidance of already- have shown that behavioral changes such as reduced parasitized mussels by ovipositing females would re- foraging activity in response to increased risk of pre- duce the total ®sh population relative to a model with dation can alter per capita interaction strengths in food nondiscriminatory oviposition, while maximizing the webs (e.g., Peacor and Werner 1997, Anholt et al. ®tness of individual ®sh. Here, we develop a model for 2000). Population regulation in mammals is likely to the population dynamics of a mosquito able to choose adaptively between oviposition sites with and without Manuscript received 5 June 2000; revised 30 March 2001; predators. accepted 9 April 2001. 3 Present address: Department of Biochemistry, University Many insects, including mosquitoes, have a dispers- of Cambridge, Tennis Court Road, Cambridge, CB2 1QW, ing adult stage and nondispersing immature stages. The England. E-mail: [email protected] habitats in which immature insects develop are often 669 670 MATTHEW SPENCER ET AL. Ecology, Vol. 83, No. 3 PLATE 1. Adult backswimmers (Notonecta maculata) are common predators of mosquito larvae in temporary pools in Israel. Photograph by Leon Blaustein and Eden Orion. discrete, for example, separate water bodies for mos- by natural selection, what behavior would we expect quitoes, and fruit, fungi, or carrion for other species. (Parker and Maynard Smith 1990, Orzack and Sober When habitat patches differ in their suitability for im- 1994)? To answer this question, we estimate parameters mature growth and survival, adult females may choose for the model from a range of experiments and obser- the patches in which they oviposit to maximize their vations. We then use the model to predict the evolu- ®tness. Several vertebrate and invertebrate taxa, in- tionarily stable oviposition strategy (ESS) at equilib- cluding mosquitoes, have been shown to avoid ovi- rium, and its sensitivity to changes in parameter esti- positing in patches that contain predators (reviewed by mates. We compare the predicted ESS strategy with the Blaustein 1999). We de®ne oviposition habitat selec- observed strategy in an experiment. We predict the con- tion (OHS) as a pattern of oviposition that is selective sequences of both the ESS strategy and the observed with respect to some measurable attribute(s) of the hab- strategy for the equilibrium adult population size, rel- itat. Like many other aspects of behavior, OHS is likely ative to a strategy in which female C. longiareolata do to be in¯uenced by selection acting on individual or- not discriminate between pools with and without pred- ganisms, but to have consequences at the level of the ators when ovipositing. population. The mosquito Culiseta longiareolata Macquart (Dip- METHODS tera: Culicidae) is common and abundant in temporary Assumptions pools in the Middle East and Africa (van Pletzen and van der Linde 1981, Ward and Blaustein 1994). The 1) Several separate pools of equal quality (apart from larvae are extremely vulnerable to predation by the differences in the numbers of predators and immature aquatic hemipteran (backswimmer) Notonecta macu- C. longiareolata they contain) are available to ovipos- lata Fabricius (Hemiptera: Heteroptera), which is also iting females. Temporary rockpools in Israel often oc- common in temporary pools (see Plate 1). C. longi- cur in wadis or exposed areas of bedrock, with man- areolata avoids ovipositing in pools containing N. ma- ypools in an area of a few hundred square meters. Max- culata (Blaustein et al. 1995, Blaustein 1998). Ovi- imum ¯ight distances of most mosquitoes are several position rates may be slightly lower in pools already kilometers (Service 1993), so all pools in a small area containing C. longiareolata larvae, and adding nutri- are likely to be available to any female. ents (increasing the growth of microorganisms and al- 2) Females choose oviposition sites so as to maxi- gae on which larvae feed, and thus potentially reducing mize their ®tness. We de®ne ®tness as the number of exploitation competition) increases oviposition (Blau- offspring surviving to adulthood under equilibrium stein and Kotler 1993). It seems plausible that natural conditions. Fitness decreases as the numbers of pred- selection played some role in the evolution of this OHS ators and C. longiareolata larvae in a pool increase. behavior. Female C. longiareolata are less likely to oviposit in Here, we describe a simple population model for C. pools with predators than pools without predators longiareolata in a system of pools with and without N. (Blaustein et al. 1995, Blaustein 1998, Stav et al. 1999). maculata. We address the question: if females oviposit The presence of predators (Notonecta maculata) re- so as to maximize their ®tness, and OHS is explained duces survival of larval C. longiareolata (Blaustein March 2002 OVIPOSITION BEHAVIOR AND POPULATION SIZE 671 1998). High larval densities of C. longiareolata are where e, l, and u are the numbers of eggs, immatures likely to reduce survival and increase development (including both larvae and pupae), and adults respec- time as they do in the congeneric Culiseta incidens tively, at times indicated by subscripts. H is the pro- (Barr 1985) and many other mosquitoes (e.g., Cham- portion of eggs that hatches each day (dimensionless, bers 1985, Mogi et al. 1985, Roberts 1998). Older lar- and assumed to be always greater than zero), F is the vae of C. longiareolata are also cannibalistic on youn- number of eggs laid per adult each day (dimensionless), ger larvae (L. Blaustein, personal observation).