Michael H. Reiskind, Department of Entomology and Plant Pathology, Oklahoma State University, Stillwater, OK 74078, U.S.A. E-mail: email@example.com
1. Natal habitat preference induction (NHPI) is a behavioural phenomenon in which offspring show a change in preference in adult oviposition choice as a function of experience as an immature.
2. Although well known in certain systems, such as herbivorous insects, this behaviour has not been well studied in aquatic insects.
3. The container–breeding mosquito, Aedes albopictus (Skuse) was used to test if NHPI occurs in aquatic insects under natural conditions of two leaf species as a nutritive base (Juniperus virginiana L. and Quercus virginiana Mill) and two larval densities.
4. Significant effects of leaf species and density on adult mosquito attributes were found, with J. virginiana and low larval density associated with more, faster developing, larger and more fecund mosquitoes. However, no evidence for NHPI was found. Instead a canalised behavior was found that included spreading eggs between high– and low–quality oviposition choices in the same proportions regardless of larval experience.
Across a wide variety of animal taxa natal site cues may influence subsequent habitat choices in nesting and/or oviposition, a phenomenon called natal habitat preference induction (NHPI) (Davis & Stamps, 2004; Davis, 2008). However, in insects, many of the previous studies of this phenomenon have been restricted to herbivorous insects (Barron, 2001), with less work on holometabolous insects that have an aquatic larval stage. Nevertheless, the same evolutionary forces that drive NHPI in herbivorous insects may apply to aquatic insects.
Natal habitat preference induction assumes oviposition behaviour is plastic, and thus responsive to variation in the environment, whether early in development (e.g. NHPI) or as adults (e.g. adult experience) (West–Eberhard, 2003). Alternatively, oviposition behaviour could be canalised, such that individuals make identical oviposition choices in response to variation in the environment (Waddington, 1942). Canalisation of behaviour does not necessarily require a single response, just that there is little variation in the response (Dobzhansky, 1970). For example, behaviour may be canalised to make one choice a certain percentage of times, and thus may appear plastic. This latter category may include risk–spreading strategies where choices are spread to avoid a single catastrophic mistake (Seger & Brockmann, 1987).
Most of the work examining NHPI in aquatic insects has been with two mosquitoes: Aedes aegypti L. and Culex quinquefasciatus Say (McCall & Eaton, 2001; McCall & Kelly, 2002; Kaur et al., 2003; Hamilton et al., 2011). These previous studies used a chemical cue that repels non–exposed (control) mosquitoes from oviposition sites. The repellency is subsequently relaxed in mosquitoes reared in the presence of the chemical. McCall and Eaton (2001) exposed larval C. quinquefasciatus to two substances often found in their larval habitats: skatole and p–cresol. They found that control and p–cresol–reared mosquitoes avoided skatole–laced habitats as adults, whereas larvae reared in the presence of skatole preferred skatole–laced habitats. Similarly, A. aegypti larvae reared under control conditions avoided habitats containing the repellent DEET, whereas larvae reared in the presence of DEET did not discriminate against DEET–laced habitats as adults (Kaur et al., 2003). Further examination of this system found that it was critical that the mosquito was exposed to the DEET signal from the larval stage through emergence to adulthood, so–called ‘adult experience reinforced natal habitat preference induction (AER–NHPI)’ (Hamilton et al., 2011).
These previous attempts at showing NHPI for mosquitoes demonstrate an increase in relative attraction to an otherwise repellent, artificial chemical. They do not examine the behavioural response to the natural variation in nutritional inputs into mosquito larval habitats, which can include variation in leaf species, vegetative parts, and animal carcasses (e.g. Lounibos et al., 1993; Barrera et al., 2006; Yee et al., 2007; Kaufman et al., 2010; Reiskind et al., 2010; Reiskind & Zarrabi, 2011). Variation in inputs into larval habitats makes the container–dwelling mosquito Aedes albopictus Skuse a good system for testing for NHPI in an aquatic insect under natural conditions. Aedes albopictus invaded North and South America, Africa and Europe from Asia, and has adapted from a phytotelmata using an ancestor into an anthropophilic, artificial container–using vector of human disease (Hawley, 1988). Different amounts or types (species or parts) of plant litter inputs into container habitats can alleviate density dependent reductions in growth and survival of A. albopictus larvae and change population growth rates (Sota, 1993; Murrell & Juliano, 2008; Reiskind et al., 2010; Reiskind & Zarrabi, 2011). As such, both organic inputs and larval density determine the quality of a larval habitat (Juliano, 2009). Furthermore, container–dwelling mosquitoes show variation in oviposition choice based upon the perceived suitability of the larval habitat for larval development (Bentley & Day, 1989; Trexler et al., 2003; Yee, 2008; Reiskind et al., 2009; Obenauer et al., 2010; Ponnusamy et al., 2010; Reiskind & Zarrabi, 2012). Prior larval experience may help shape the oviposition choice behaviour of A. albopictus by providing a signal to habitats that are of high quality, or at least sufficient for development. Alternatively, the large amount of variation in larval habitat quality may suggest a less plastic, canalised response would be advantageous to avoid making poor oviposition choices when based upon larval experience.
To examine whether NHPI operates for an aquatic insect under natural conditions, we tested the hypothesis that A. albopictus exhibits NHPI with a larval habitat of different leaf species. To partially control for differences in leaf quality for developing larvae, we tested adults generated at two larval densities, with a lower density acting as a signal of a higher quality habitat (Stamps et al., 2009). We predicted that mosquitoes would show NHPI by ovipositing in containers provisioned with their natal site leaf species, and that the effect of natal site signal on adult oviposition would be more pronounced in females emerging from low larval density habitats relative to high larval density.
Materials and methods
Mosquitoes used in this experiment were F3A. albopictus from approximately 2000 eggs collected in the summer of 2010 from Tulsa, Oklahoma. The first two generations in the lab were reared on an infusion of water with 4 g/l live oak (Quercus virginiana Mill) leaves and 0.15 g/l of yeast albumin at a density of approximately 100 larvae/l of infusion. Mosquitoes were provided access to blood on the arm of a volunteer (M.H.R., exempted from the Institutional Review Board, 28 August 2008) approximately weekly to produce eggs.
To test the impact of larval environment on adult oviposition choice, mosquitoes were reared in cohorts at 2 densities (15 and 40 first instar larvae) in 2 leaf types (live oak, Quercus virginiana and eastern red cedar, Juniperus virginiana), to be tested in separate oviposition trials, with 8 replicates of each treatment combination for a total of 32 observational units. Owing to limitations in incubator space, the eight replicates were split into two trials of four replicates, conducted sequentially.
Each larval cohort was provided with 2 g of dried leaf material in 500 ml of tap water and 100 µl of locally collected pond water to provide a microbial inoculum in 1–l food grade plastic containers (Instawares, Inc., Naperville, Illinois), aged 3 days before the addition of larvae. Senesced live oak leaves were collected in Florida as described previously (Reiskind et al., 2010). Senesced cedar leaves were collected opportunistically in Stillwater, Oklahoma, the previous fall. Although the leaf material was not collected in the same area, there is no evidence that mosquitoes are adapted to leaves from a particular location (Sota, 1993). All leaf material was air dried over several weeks in a low humidity environment until their weight stabilised. Eggs were hatched in tap water, and the appropriate number of first instar larvae was added to each microcosm within 36 h of hatching. Cohorts were monitored daily for pupation, and pupae were removed when observed and placed into adult cages. Adult cages were modified plastic containers of 33 cm × 46 cm × 26.5 cm for a total internal volume of 40.2 l. To avoid pseudoreplication, each larval container was matched to an adult cage, for a total of 32 adult cages. Each replicate of each treatment combination was housed in a separate incubator (such that each incubator had four larval containers and four adult cages, one from each treatment combination), set to 28 °C, LD 14:10 h with an open pan of water to ensure high humidity.
Adult mosquitoes were provided a 20% sucrose solution ad libitum from emergence. Three days after the last female emerged, adults were provided with the opportunity to feed on a human volunteer's forearm (M.H.R., exempted from Institutional Review Board at Oklahoma State University, 28 August 2008), and the number of females feeding was recorded. Three days after feeding, each cage was provided with two, separate 250–ml containers of 3–day–old 4 g/l leaf infusion of live oak and cedar in a 500–ml food grade plastic container lined with seed germination paper (Seedburo inc., Danville, Illinois). They were given 7 days to oviposit, after which the paper was removed, eggs counted, and mosquitoes collected and killed by freezing. All mosquitoes were identified to sex, counted, weighed, and had their wings measured. Male and female survival to adulthood, mean days to pupation, weight, wing length, and number of eggs laid per female blood–fed were determined for each replicate. The measurements of population performance (e.g. finite rate of increase) could not be calculated because it was impossible to associate a given female mosquito with her emergence date. Nevertheless, survival, mean development time, wing length, weight, and fecundity provide a relative measure of habitat quality that reflects population performance.
One incubator failed during the oviposition portion of the experiment, and that replicate was removed from the analysis. All outcome variables were assessed for normality, which could not be rejected for these data. Comparisons between larval outcomes (survival, days to emergence, weight, and fecundity per blood–fed female) were made by two–way, multivariate analysis of variance (manova) with a type I sum of squares. Significance was assessed using Pillai's trace, which is robust to assumptions of normality and appropriate given the small sample sizes (Scheiner, 2001). A preference for oviposition media was assessed by comparing the proportion of eggs laid in cedar (eggs laid in cedar/total eggs laid) for each replicate by a non–parametric, rank test (Kruskal–Wallis one–way analysis of variance).
Both the leaf type and the density of the larval environment significantly affected larval survival, time to pupation, female weight, and the mean fecundity of females (Fig. 1 and Table 1). In general, mosquitoes derived from cedar leaf had a higher survival, less time to pupation, weighed more, and had a higher fecundity than those from oak leaf habitats. Consequently, there was variation in the average number of mosquitoes blood–fed (means: cedar high: 4.75; cedar low: 3.25, oak high: 1.875, and oak low: 1.75). Mosquitoes from low–density habitats had a higher survival, shorter time to pupation, weighed more and had higher fecundity than those from high–density habitats. There was no significant interaction between density and leaf type. There was overlap in some of the outcomes between low–density, oak provisioned habitat and cedar provisioned habitats, particularly the high–density, cedar provisioned habitat (Fig. 1). However, the larval origin of the mosquitoes had no effect on their subsequent, adult oviposition choice, with females consistently preferring cedar habitats at a ratio of around 3:1 (Kruskal–Wallis: , P < 0.9620, Fig. 2).
Table 1. manova of outcomes from larval habitat.
Days to pupation
*No significant data.
Leaf × density
We can reject the hypothesis of NHPI for A. albopictus under these conditions. Adult females consistently preferred habitats provisioned with cedar to habitats provisioned with an equal mass of live oak, regardless of their larval origin. This preference may reflect the empirical quality of cedar habitats, as larvae grown in cedar had generally a higher survival to adulthood, larger mass, produced more eggs per female, and developed quicker than in oak provisioned habitats. Consequently, we conclude these mosquitoes do not have a variable response in oviposition choice as a result of natal site characteristics reflective of natural variation in larval environments.
There are a few places in our experimental design that might bias the results. First, the microbial inocula, although equally applied to all treatments, may have some bacterial species that favour the breakdown of one leaf species versus the other. We do not suspect this to be a likely situation, as the pond from which the water was collected does not receive inputs from either species. The second possible bias has to do with the chronological age of mosquitoes. The oak and high density treatments yielded mosquitoes that had been larvae for longer, although the blood feeding and oviposition experiments used females approximately the same age since emergence. The effect of chronological age, therefore, was not able to be controlled for. However, the uniformity of the oviposition response suggests that any impact of this was minimal, and certainly not strong enough to make false conclusions. Finally, as mentioned in the methods, mosquitoes and leaf material were not collected in the same locations. However, there is no evidence mosquitoes are locally adapted to certain tree populations (Sota, 1993), and both leaf species are found from Oklahoma to Florida.
Previous work in similar systems has found evidence for some degree of NHPI in A. aegypti and C. quinquefasciatus (McCall & Eaton, 2001; Kaur et al., 2003; Hamilton et al., 2011). However, these studies did not show preference was induced by larval habitat. Instead, they found a reduction in oviposition deterrence to a particular semiochemical (skatole for C. quinquefasciatus and DEET for A. aegypti). Why larval habitat stimuli should relax but not stimulate an otherwise inflexible response remains to be determined. It is possible the chemical cue is retained in the imago, as has been proposed for herbivores, or that the stimulus is retained in the central nervous system (e.g. learned) (Barron, 2001). Interpreting our results in light of these previous studies suggest the mechanism may involve the loss of detection ability, as opposed to a gain in discrimination. This loss of detection ability may have important consequences, and it is critical to determine if larval exposure to a host repellant such as DEET affects the efficacy of this common repellant in host–seeking females. This may also differentiate between a learned, associative memory [between DEET and larval site, as proposed by Kaur et al. (2003) and a change in neural or sensory physiology larval exposure to DEET renders adults unable to detect DEET].
The uniformity in response to leaf species choice exhibited in our data regardless of larval condition suggests a canalised response to nutrient quality. In spite of the likely diversity of larval habitat quality in the field as a result of leaf species diversity, plant part diversity (e.g. leaves, flowers and/or fruits), and animal carcasses (Lounibos et al., 1993; Yee et al., 2007; Reiskind et al., 2010; Reiskind & Zarrabi, 2011), A. albopictus chose cedar habitats at a consistent ratio of 3:1 regardless of natal site leaf environment or larval density. We did not track individual females within each replicate, but previous data has shown skip–oviposition behaviour in A. albopictus (Trexler et al., 1998) suggesting that these mosquitoes may be bet–hedging, in which eggs were spread into both habitat types. A preference for certain leaf species over others in oviposition choice has been supported by several previous controlled studies, and may be determined by the diverse microbial community (and the chemical signals they produce) in these habitats (Trexler et al., 1998, 2003; Ponnusamy et al., 2008, 2010; Reiskind et al., 2009). The high diversity of larval habitat quality, dictated by variable plant and animal inputs, may make larval experience an unreliable predictor of habitat quality. Indeed, the strongest examples of NHPI come from organisms that have to make a choice of one host species versus another (e.g. herbivores and parasitoids), possibly a more predictable environment than small aquatic containers (Davis & Stamps, 2004). Although we focused on the differences in quality between one leaf species and another, most artificial container habitats from field studies contain a diversity of plant and animal resources, often in combination (Yee et al., 2007; Reiskind et al., 2010). Confronted with such diversity, a canalised behavioural response that includes spreading eggs in poorer oviposition sites may be less likely to make mistakes resulting in complete loss of progeny than a highly flexible approach dependent upon natal experience.
The authors wish to thank the Oklahoma Center for the Advancement of Science and Technology (OCAST: Grant HR09–157 to MHR) and the Oklahoma Agricultural Experiment Station (OAES Hatch Project: 2702) for their support of this research. The authors wish to thank two anonymous reviewers and the editorial staff at Ecological Entomology for their insightful inputs. We would also like to thank Richard Grantham and Justin Talley for reading an earlier version of this manuscript and Jacklyn Shelton for assistance in the laboratory.