^{1}

^{*}

^{2}

^{3}

^{1}

The authors have declared that no competing interests exist.

Conceived and designed the experiments: GW NB CD KY. Performed the experiments: GW NB CD KY. Analyzed the data: GW. Contributed reagents/materials/analysis tools: GW NB CD KY. Wrote the paper: GW.

Oviposition site selection by gravid females is an important determinant of the distribution, abundance, and dynamics of dipteran hematophagous insects. The presence of conspecific immature stages in a potential oviposition site could, on the one hand, indicate the suitability of that site but on the other hand could indicate the potential for intraspecific competition. In this paper, we present a graphic model suggesting that the trade-off between these two opposing forces could result in a hump-shaped density-dependent relationship between oviposition rate and conspecific immature stage density (hereafter, the “Hump-shaped regulation model”) with positive effects of aggregation prevailing at low densities and negative effect of intraspecific competition prevailing at higher densities. We field-tested the predictions of this model at both the egg- and the larval levels with

For organisms lacking parental care and where larval dispersal is limited, oviposition-site selection decisions by gravid females are critical fitness-enhancing choices with critical implications to the distribution, abundance and dynamics of those populations

The effect of conspecific immature stages on the oviposition site-selection of gravid mosquitoes has received a lot of attention, however, the results are highly conflicting with some studies reporting no effect _{max}) occurring at some intermediate level of conspecific immature stage density, which we term ‘optimal density’ (N_{opt}) (

(A) Cost-benefit model of the relationship between conspecific immature stage density and the reassurance benefit due to egg or larvae aggregation and cost of competition in the absence and presence of resources. Two-headed smoothed and dashed arrows represent the maximal net benefit in the absence and presence of resources, respectively. (B) The trade-off between the benefit of reassurance and the cost of intra-specific competition should result in a hump-shaped relationship between conspecific immature density and fitness (G). G_{max} indicates the maximal fitness and N_{opt} indicates the density at which it occurs. The switching preference density threshold (SPDT) indicates the conspecific density at which this fitness line crosses the neutral-preference line. This neutral-preference line dissects the state-space into regions of conspecific attraction (G > 0, light blue) and repellence (G < 0, pink). Hence, at densities below the SPDT attraction to conspecific immature stages should be exhibited while above it repellence from conspecific immature stages should be exhibited. Red dashed arrows indicate the expected shift in N_{opt}, G_{max}, and SPDT due to resource addition.

Although, to date, only five studies involving mosquitoes have described such a hump-shaped relationship _{opt} and SPDT to higher conspecific densities and increase in overall oviposition rate (

Peabody Park is a thirty-four acre recreational and research deciduous forest on the northern side of the University of North Carolina at Greensboro campus. Its average elevation is 241 meters above sea level and soil texture is loamy. A system of several creeks, part of the Haw River Basin, flows throughout the park. In this park the predominant container breeding mosquito is ^{th} – September 20^{th}, 2012. All experiments were conducted using oviposition cups distributed, at 10 m intervals, along straight transects running through the forest with distance between transects ≥ 20 m (

Transect are represented by lines and blue circles represent the oviposition cups. Numbers inside the circles represent the initial number of larvae introduced (experiments 1 and 2). Numbers underneath transects represent the number of spatial replicates.

The effect of conspecific larvae Two experiments were conducted. One tested the effect of larval density on mosquito oviposition response and the second tested the effect medium enrichment on this relationship. In experiment 1 (^{st} -2^{nd} instar ^{st}/2^{nd} larvae to complete metamorphosis) and then collected for egg counting in the lab. At each collection, larvae and cup-water were also collected and transferred in a large (1242 mL) nylon bag (Whirlpack, Nasco, Fort Atkinson, WI) to the lab where larvae and pupae were counted. Cups were replenished with fresh dechlorinated water and a new ovistrip was inserted to each cup.

In experiment 2 (

The effect of conspecific eggs In experiment 3 (

If the presence of pre-existing conspecific eggs does not affect oviposition behavior, then there should be no difference between the cumulative number of eggs in the daily-replacement- and the continuous-exposure cups throughout egg density range (the neutral hypothesis) (

To confirm lack of confounding effect due to age of the ovistrip (in the daily replacement cups ovistrips are fresh whereas in the continuous exposure they age with exposure time) we placed ovistrips in water-filled oviposition cups and aged them under simulated field conditions in an environmental chamber (27°C, 80%RH, 12∶12 hr. photoperiod) for 0, 2, 5, and 10 days. Then, along two 100 m long transects, we deployed 20 pairs of oviposition cups: one cup with the aged germination paper and the other with a fresh germination paper. Ovistrips were collected five days later and eggs were counted in the lab.

Literature search was conducted using ISI-Web of knowledge for all years using search code: Topic = (mosquito* and (oviposition or egg laying) and (conspecific or habitat selection or competition)). A total of 194 papers were found and an additional 11 were added based on relevant references mentioned in the reference list of relevant papers. Only papers considering the effect of conspecific eggs, larvae, or pupae on oviposition response of mosquitoes were included in the analysis (

Due to the nature of the data (count) and its high degree of overdispersion, we analyzed it using negative-binomial (NB) generalized-linear models

Variable | Coefficient | SE | z-value | P-value |

Intercept | 4.212 | 0.211 | 19.791 | |

Larvae | 5.414E-3 | 2.295E-3 | 2.359 | |

Larvae^{2} |
-1.563E-5 | 7.759E-6 | –2.015 | |

Transect B | -9.95E-1 | 0.260 | –3.818 | |

Transect C | -8.64E-2 | 1.993E-01 | –0.434 | 0.6656 |

Date 9/20 | -9.719E-01 | 2.590E-01 | 3.752 | |

Date 8/25 | -4.296E-01 | 2.825E-01 | –1.521 | 0.1283 |

Date 9/27 | 4.869E-01 | 3.892E-01 | 1.251 | 0.2109 |

Date 8/30 | -7.757E-02 | 2.570E-01 | 0.302 | 0.7627 |

Date 9/6 | -1.452E-02 | 2.620E-01 | –0.055 | 0.9558 |

Variable | DF | Deviance | Res. DF | Res. Dev. | P value |

NULL | 61 | 136.69 | |||

Larvae | 1 | 40.56 | 60 | 96.13 | |

Enrichment | 1 | 16.61 | 59 | 79.52 | |

Transect | 4 | 5.45 | 55 | 74.07 | 0.244 |

Larvae x Enrichment | 1 | 10.89 | 54 | 63.18 |

Despite high variability in these data, after controlling for the effect of transect location and session date, the relationship between number of larvae and number of new eggs laid was consistent with a negative second-order polynomial relationship (^{st}-derivative of the regression line), and then gradually decreasing (

Least-squares 2^{nd}-order polynomial regression plot of this relationship is presented.

Conspecific larva number had a, general, positive effect on egg deposition (^{2} = 2129.5, P<0.0001). This resulted in larval number range being substantially lower in the control medium (range: 4–41, mean: 14.65 larvae/cup) compared with the enriched medium (range: 0–352, mean 91.96 larvae/cup). Hence, since larval number in the control and the treatment cups differed, we could not evaluate our original hypothesis. Nonetheless, we tested the HSR model predictions by fitting a second-order polynomial NB regression for data of each treatment. For the control medium, only the linear term (positive slope) was significant (eggs = 4.735+0.015Larvae, z = 4.653, P<0.0001) with the linear model fitting the data better compared with the second-order polynomial model (ΔAIC = 2) (^{nd}-order polynomial regression was significant (

Least-squares regression plot of the relationship between conspecific larvae number of number of mosquito eggs laid per cup for enriched (red squares) and the water (blue diamond) media.

Coefficient | SE | z value | P value | |

Intercept | 5.313 | 0.133 | 39.725 | |

Larvae | 6.711E-3 | 2.5E-3 | 2.684 | |

Larvae^{2} |
-1.650E-05 | 8.098E-06 | –2.037 |

As expected, the cumulative number of eggs increased with exposure time (

(* P <0.05, ** P <0.01, *** P < 0.001, **** P < 0.0001)

Variable | DF | Deviance | Res. DF | Res. Dev. | P value |

NULL | 319 | 594.87 | |||

Treatment | 1 | 34.44 | 318 | 556.43 | |

Exposure time | 1 | 116.39 | 317 | 440.04 | |

Transect | 1 | 1.27 | 316 | 438.77 | 0.259 |

Session | 6 | 79.05 | 310 | 359.72 | |

Treatment x exposure time | 1 | 5.17 | 309 | 354.55 |

The results of the control experiment, suggest that this preference for the daily-replacement cups is not due to aversion from aged ovistrips. Number of eggs did not differ between aged and fresh ovistrips for 0, 2, and 5 aging days. However, for ovistrips aged for 10 days number of eggs was actually higher in the aged ovistrips, which is exactly the opposite of what would have been expected due to a confounding effect of aversion from aged ovistrips (

Paper aging time (days) | No. eggs in aged ovistrip (mean ± SE) | No. eggs in non-aged ovistrip (mean ± SE) | Paired-t test |

0 | 49.87±4.54 | 52.93± 5.44 | t = –0.68, P = 0.50 |

2 | 71.95±11.04 | 73.7410.23 | t = –0.14, P = 0.89 |

5 | 47.38±4.42 | 44.56± 6.88 | t = 0.51, P = 0.61 |

10 | 50.0± 8.15 | 38.3± 4.72 | t = 2.28, P = 0.03 |

The regression of ΔE (the difference between the cumulative number of eggs in the replacement and the exposure cups) against the cumulative number of eggs in the daily-replacement cups (minus 1) revealed a line with a significant negative intercept (±se) (–26.10±5.08, t = –5.14, P<0.0001) and positive slope (0.59±0.04, t = 16.31, P<0.0001) (

In total, we identified 44 papers addressing the issue of the effect of conspecific immature stages on the oviposition response of mosquitoes (^{2} = 12.75, df = 1, P = 0.0004) and the proportion of “Density-dependent effects” (13.2%) was significantly lower than expected (proportion test: χ^{2} = 6.16, df = 1, P = 0.013). The proportion of “no effects” (25.2%) and “negative effect” (19.8%) did not differ significantly from the 25% expectation. Note, that the number of “No effect” reports (n = 23) might be an under-estimate due to reporting- or publication bias. This distribution of conspecific effects did not differ significantly among mosquito immature stages (contingency table: χ^{2} = 8.84, df = 6, P = 0.18). The distribution of conspecific density effects did differ significantly between lab and field studies (contingency table: χ^{2} = 17.91, df = 3, P = 0.0004). In lab studies (n = 61), the proportion of “positive effects” (52.4%) was significantly larger than expected (proportion test: χ^{2} = 23.08, df = 1, P<0.0001) and the proportion of “negative effects” (8.2%) was significantly smaller than expected (proportion test: χ^{2} = 8.3, df = 1, P = 0.004). In contrast, in field studies (n = 31), the proportion of “negative effects” (43.3%) was significantly larger than expected (proportion test: χ^{2} = 4.44, df = 1, P = 0.035) while the proportion of “positive effects” (23.3%) did not differ from the expected 25% (proportion test: χ^{2} = 0, df = 1, P = 1). In addition, the proportion of “density-dependent effects” differed between lab and field studies with significantly lower proportion (8.2%) than expected (proportion test: χ^{2} = 8.31, df = 1, P = 0.004) in the former and not significantly different (proportion test: χ^{2} = 0.71, df = 1, P = 0.40) in the latter (16.7%). The proportion of “no effects” did not differ significantly from expected for neither study types (30% and 17% for lab and field studies, respectively).

No effect | Positive effect | Negative effect | DD effect | N | |

Laboratory studies | |||||

Eggs | 6 | 10 | 2 | 4 | 22 |

Larvae | 7 | 17 | 3 | 2 | 29 |

Pupae | 5 | 4 | 0 | 1 | 10 |

N | 18 | 31 | 5 | 7 | 61 |

Field studies | |||||

Eggs | 4 | 3 | 3 | 2 | 12 |

Larvae | 1 | 4 | 10 | 3 | 18 |

Pupae | 0 | 0 | 0 | 0 | 0 |

N | 5 | 7 | 13 | 5 | 30 |

All studies combined | |||||

Eggs | 10 | 13 | 5 | 6 | 34 |

Larvae | 8 | 21 | 13 | 5 | 47 |

Pupae | 5 | 4 | 0 | 1 | 10 |

N | 23 | 38 | 18 | 12 | 91 |

We evaluated the HSR hypothesis with five mosquito species for which a sufficient amount of data was available on their oviposition response at a range of conspecific immature densities (

With

At the larval level, we observed hump-shape relationship, consistent with the HSR model (

Enrichment of the rearing medium was expected to increase overall oviposition rates and to shift optimal conspecific density and SPDT to higher larval densities due to suppression of the competitive effect (_{opt},

The other three studies reporting a HSR pattern with respect to conspecific larvae were those of Benzon and Apperson ^{nd} instar

At the eggs level, we also found strong support for the hump-shape regulation model. At low egg numbers

Only four previous studies have reported a HSR pattern with respect to conspecific eggs

In this paper, we suggested that the disparate and conflicting results concerning the effect of conspecific immature stages on the oviposition response of mosquitoes could be synthesized by the HSR model with positive-effect observations occurring at the lower range of densities, no-effect occurring at intermediate densities, and negative-effects occurring at high densities. We also suggested that the dearth of observations of the HSR pattern and the conflicting reports regarding these relationships are likely the outcome of the design of most previous studies that tended to explore only a limited range of conspecific densities. To evaluate this hypothesis, we conducted an exhaustive literature review (

Based on the Ideal-Free-Distribution (IFD) theory

(DOCX)

(DOC)

The authors thank Malcolm Schug (UNC-Greensboro) for reviewing an earlier draft of this paper. We also thank Burt Kotler (Ben-Gurion University of the Negev) for review of the model. The authors also wish to thank three anonymous reviewers for the useful comments and suggestions. We also thank the Landscape Ecology class of fall semester 2011 (UNC-Greensboro, Biology program), Laura White, and many other undergraduate student assistants for their help in field-work, data collection, and egg counting. No permits were required for the described study, which complied with all relevant regulations.