Background
Concerted efforts towards malaria control and elimination have led to a global decline in malaria cases by about 40% between 2000 and 2015 [
1]. Almost 80% of this reduction is attributed to vector control by widespread distribution of insecticide-treated nets (ITNs) and indoor residual spray (IRS) [
1]. Despite these successes, malaria reduction in the World Health Organization (WHO) African region has stalled [
2], which may in part reflect limitations of ITNs and IRS. These limitations include plasticity in behaviours including early biting and outdoor resting, and feeding on animals allowing adult malaria vectors to avoid exposure to insecticides [
3,
4], widespread insecticide resistance [
5,
6], and high operational costs of IRS in particular which limit coverage [
2]. Implementation of supplementary vector control tools is required to further reduce malaria transmission in a trajectory toward elimination [
7,
8].
Larval source management (LSM) is a tool for further development because all mosquitoes need to lay eggs in an aquatic habitat irrespective of their biting or resting preferences [
9]. However, the uptake of this intervention is impeded by the management effort required [
10] and the lack of knowledge of aquatic habitats that are the most preferred for egg laying, which would allow a more spatially-targeted approach. The oviposition behaviour of
Anopheles mosquitoes, and specifically the preference of particular aquatic habitats for egg-laying, has been studied in order to better target LSM and to develop novel attract and kill strategies for vector control [
11,
12]. Habitat preferences are frequently inferred from the abundance of early instar larvae in a habitat [
13]. This relies on the assumption that higher early instar larval density results from a greater number of gravid females selecting the site for oviposition [
11,
12]. However, to date only indirect tests of this assumption exists [
14]. Furthermore, whilst there is strong indication from cage experiments that species of the
Anopheles gambiae complex frequently distribute their eggs in more than one egg-laying sites (skip-oviposition) [
15], this behaviour has not yet been widely accepted in
Anopheles due to the few studies providing supporting evidence [
14].
An LSM tactic currently being explored is the auto-dissemination of insect growth regulators (IGRs) where adult mosquitoes naturally transfer the IGR from resting to breeding sites [
16,
17]. Large numbers of females visiting multiple breeding sites would aid transfer of IGR among habitats to achieve biologically-relevant mortality of the immature stages. However, skip oviposition has been widely accepted to occur habitually in
Aedes mosquitos unlike in
Anopheles [
18‐
20]. In typical habitats of western Kenya early instar densities vary from 160/m
2 in puddles to between 0.3 and 10/m
2 in open drainages, cultivated swamps, and river fringes [
13,
21].
Anopheles females can lay up to 200 eggs [
15], and it is unclear whether the typically-low densities of early instars are a result of few gravid females visiting most habitats or of high mortality of the eggs or larvae from many females.
Here genotype-based family reconstructions methods were used to enumerate the number of female Anopheles arabiensis depositing eggs in single and multiple habitats in relation to overall larval densities. It was hypothesized that relative high number of early instar An. arabiensis larvae in a habitat is an indicator of high number of females laying eggs in a site; therefore, differences in early instar abundance across similar habitat types of comparable size correlate with the number of females laying eggs.
Discussion
Understanding mosquito oviposition behaviour is important to inform the development and implementation of breeding habitat-targeted mosquito control strategies. However, direct observational studies of oviposition behaviour are challenging due to its nocturnal occurrence. As such, limited evidence exists to support phenomena such as habitat preference, variations in larval densities across breeding habitats and frequency of skip oviposition by An. gambiae complex females.
In this study, genetic clustering (BAPS) and sibship reconstruction (COLONY) methods were applied to enumerate the number of female
An. arabiensis depositing eggs in single and multiple habitats in relation to overall larval densities. The findings show that relative high number of early instar larvae in a breeding habitat is indicative of an increased number of ovipositing females. This supports the hypothesis that early instar larval abundance of
An. arabiensis is an indicator of the number of females laying eggs in a habitat and can hence be used to investigate oviposition preferences. This finding is consistent with a study that previously found a positive correlation between the number of mosquito families and number of larvae in a breeding site [
29]. The family sizes detected were generally low, concordant with previous studies [
14,
29], and the discrepancy between potential an actual family sizes maybe attributable to low egg hatchability and first instar survival rates [
30].
This study also found that
An. arabiensis females often display skip oviposition, distributing their eggs in multiple habitats. On average, more than half of the families investigated from the experimental and natural habitat surveys resulted from skip-oviposition. The ponds and natural habitats were in a small spatial scale, which could have contributed to the high skip oviposition rates. Since gravid
Anopheles females tend to hover over habitats before deciding to lay eggs, it was not surprising that skip-oviposition occurred over the entire range. This finding is similar to a study on
An. gambiae s.s., which found that 57% of females had skip-oviposited in habitats [
14]. In contrast, a much lower skip-oviposition frequency of 26% was observed in two-choice cage assays on
An. gambiae s.s. oviposition in the laboratory [
15]. The low frequencies observed in this cage assay could have been due to the individual mosquitoes responding to a similar oviposition substrate reducing preference to skip oviposit. There was also a relationship between mosquito family size and the number of habitats in which they were distributed in both the artificial ponds and natural habitats implying that females that laid more eggs were more likely to exhibit skip oviposition.
Frequent skip-oviposition behaviour could contribute positively to the success of
An. arabiensis mosquito control methods using auto-dissemination of insect growth regulators (IGRs) where skipping adult females naturally transfer the IGR from resting to breeding sites.
Anopheles arabiensis have been shown to experimentally transfer pyriproxyfen from their resting sites to aquatic habitats leading to significantly reduced larval emergence [
16,
17]. Our results suggest that skip oviposition behaviour should also be considered as an important factor in such studies and in the ecology of these vectors.
The number of families estimated using BAPS were consistently much lower than COLONY both in the artificial ponds and natural habitats. This could been in part due to a tendency of COLONY to over-split large families as previously noted in simulation studies [
31,
32]. Also, whilst both estimators supported the relationship that high larval density in a breeding habitat is indicative of an increased number of ovipositing females in the ponds, only BAPS did so in the natural habitats. However, despite contrasting family sizes between the two software, individual pairwise sib-ship assignments made by both agreed strongly in the ponds, although the correspondence was more moderate in the natural habitats. This may reflect a problem with ‘self-referencing’ in COLONY, whereby in the natural habitats the same genotype set was used as those assigned, rather than a wider set fully representing genetic diversity of the population. A similar observation was made in the two families of known full-siblings and in the pond samples when ‘self-referencing’ with COLONY.
Conclusion
The current study demonstrates the potential use of An. arabiensis larval abundance in breeding habitats as an indicator of oviposition site preference and provides empirical evidence of frequent skip-oviposition behaviour, which should be considered when studying their ecology and applying larval control methods. These findings support previous findings for An. gambiae s.s., but at present little is known about oviposition behavior in the other major East African malaria vector Anopheles funestus, and studies are now warranted.
Acknowledgements
Thanks to Elizabeth Masinde, Paul Ouma, and Rose Imelda for assistance with the field collection of adult and larval mosquitoes, setting up the artificial ponds, and morphological identification of the mosquito larvae.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.