The fast spread of resistance to insecticides observed in the main malaria vectors, Anopheles gambiae sensu stricto, Anopheles arabiensis and Anopheles funestus [1, 2] suggests that the effectiveness of mass distribution of insecticide-treated nets (ITNs) and large-scale indoor residual spraying (IRS) in reducing the incidence of malaria in endemic countries [3, 4] will reach a plateau in the foreseeable future. There is an urgent need for development of not only new chemical compounds, but also of novel and alternative vector control approaches to complement pesticide-based strategies. This urgency explains the renewed interest in vector control using sterile male releases  and the rapid expansion of research focused on the release of genetically manipulated mosquitoes unable to transmit malaria . Implicit to these approaches is the necessity to produce large numbers of sexually competitive male mosquitoes from colonized strains in order to target wild vector populations [7, 8]. The current knowledge base for mosquito mass-rearing techniques has been accumulated over a number of sterile-male mosquito release programmes attempted during the 1970s [5, 7, 9]. Although some of those programmes significantly impacted the targeted vector populations, results were generally too poor to warrant their continuation and expansion . These projects generated valuable data about the relative mating competitiveness of laboratory-reared sterile males compared to wild males and putative negative effects of the chemical or radioactivity sterilization steps involved in producing sterile males . They were, however, generally unable to identify the exact genetic and environmental processes associated with colonization and laboratory rearing that negatively affected the reproductive phenotype of mass-produces males [5, 8, 9].
Colonized strains that are well adapted to the laboratory are able to mate and lay eggs reliably and predictably in the laboratory setting and as such are the starting point of all release control programmes. In the process of establishing a new laboratory colony, the mosquito population undergoes at least one, and possibly several, selective sweeps and genetic bottlenecks as only a fraction of wild captured individuals survive and reproduce in their new environment and the resulting newly colonized strain progressively adapts to the insectary rearing conditions. Therefore, notwithstanding the potential direct negative fitness effects of sterilization or transgenesis [10, 11], the genetic changes associated with colonization have the potential to affect the competitiveness and fitness of a candidate release strain [7, 8, 12]. As an example, the colonization of a wild population of An. gambiae s.s. resulted in six-fold decrease in microsatellite allelic richness and two-fold decrease in heterozygosity over a period of two years . Similar patterns have been reported from comparisons of isozyme allelic richness in field population versus laboratory strains of Aedes aegypti and Aedes formosus . Most of the strains used preferentially for genetic engineering of An. gambiae have been bred in the laboratory for over 25 years (G3, KIL, etc.)  and are most likely to be considerably inbred. Inbreeding is thought to negatively affect fitness by increasing the frequency of homozygotes at the expense of hererozygotes . Negative effects can occur either through the accumulation of deleterious recessive alleles leading to unfit homozygotes - the partial dominance hypothesis, or through the loss of favourable heterozygotes - the overdominance hypothesis [17, 18].
The broad causal relationship between inbreeding, decreased phenotypic quality and fitness is well documented from animal breeding studies . In addition, the availability of neutral molecular markers in an increasingly large number of organisms has resulted in a recent flurry of heterozygosity-fitness correlation (HFC) studies reporting correlations between estimates of genetic diversity and fitness components in a variety of wild and captive populations . Currently, none of these studies focus on mosquitoes. However there are some reports of negative effects of inbreeding on the reproductive success of An. gambiae laboratory populations (e g, ). Moreover, the loss of viability associated with severe inbreeding in attempts to isolate morphological genetic mutants and isogenic lines in Aedine and Anopheline mosquitoes is well documented [14, 22].
The expected negative effects of inbreeding on laboratory-reared mosquitoes have led to different schemes for reconstituting their genetic diversity prior to mosquito release programmes . These approaches require crossing and backcrossing laboratory strains with the progeny of field-collected individuals, and are thus not always practical to implement regularly and efficiently . Critically, these schemes ignore the independent contribution of selection for laboratory conditions, another genetic process that could impact the future mating competitiveness of released individuals. Consequently, such schemes can only be considered as hit-or-miss approaches. In addition, there is currently very little understanding of which reproductive traits are negatively impacted by colonization and of how these changes could potentially translate into decreased mating competitiveness in the field [8, 23]. Without that knowledge it is virtually impossible to improve on current breeding schemes and laboratory-rearing practices .
Here changes in sperm length, testes size and male accessory gland size of An. gambiae occurring at different stages of the colonization process were investigated through comparisons of the progeny of field-collected individuals and different laboratory strains aged two to 35+ years. Sperm length has been shown to be very variable in laboratory strains of An. gambiae  and one study reported that longer sperm were more likely to be stored in the female spermathecae upon mating than shorter ones . There is also limited evidence that sperm length could correlate with male reproductive success in An. gambiae . There are currently no studies focusing on variation in testes and male accessory glands size among laboratory or field anopheline populations. In anophelines, the size of both organs is known to increase with male mosquito age and culminate five to six days after emergence [27–29]. Testes size is expected to correlate with the size of the sperm reservoir, and thus could potentially affect the total number of females that males can inseminate. In addition to transferring sperm, male mosquitoes deposit a mating plug in the female atrium during copulation. The mating plug is produced by the male accessory glands and, once deposited in the female, acts as a physical barrier that decreases the likelihood of females mating with other males [21, 30]. These plugs also contain an array of sex-peptides that are responsible for inducing a cascade of behavioural changes in females [30–32]. These changes include refractoriness to further mating [30, 33, 34], host finding, feeding , and the initiation of oogenesis . Changes in the size of male accessory glands could affect the size and/or number of plugs that males are able to transfer to females, and therefore determine the number of females they can inseminate.
In addition to comparing those reproductive characters in relation to the age of mosquito colonies, these traits were compared in a colony used to produce two genetically-modified (GM) strains. These strains had been genetically-modified using a two-phase transformation system . The procedure required for genetic transformation leads to two successive genetic bottlenecks that could potentially affect the reproductive phenotype of these and other GM strains created using similar approaches. Finally, we performed crosses between strains and the progeny of field-caught females to create genetically-refreshed outbred strains for comparison with non-refreshed ones. Crosses were also made between old strains to generate heterotic hybrid males. Both types of crosses enabled us to better compare the effects of inbreeding from the effects of selection on the male reproductive phenotype.
This study is the first to describe broad phenotypic changes affecting sperm length, and the size of testes and male accessory glands during the colonization process of laboratory strains of An. gambiae and to shed light on the underlying genetic processes leading to these changes. The results have important implications for ecological studies focusing on mosquito reproductive success in the laboratory, as well as for protocols of mass mosquito rearing that are critical to the success of malaria control strategies relying on mosquito releases.