In this study using a large sample set of mosquitoes from 29 independent infections across four malaria transmission seasons, there is no difference between the M and S molecular forms of A. gambiae in their genetic susceptibility to infection by P. falciparum, as measured by oocyst prevalence or intensity after exposure to blood meals from natural gametocyte carriers in Burkina Faso. Further, there is no difference in infection susceptibility between A. gambiae s.s. and its sister species, A. arabiensis. These findings are in accord with several previous studies that found no difference in natural infection rates between M and S form mosquitoes [23–25, 44]. As mentioned above, there are multiple potential reasons why measures of genetic susceptibility presented here, and natural infection rates presented in the cited work, might differ. Natural infection rates do not control for, and in fact summarize, additional factors beyond genetic susceptibility, such as differential human feeding rates, age at infection, longevity, and exposure to other pathogens that could influence immune status. The natural infection studies and this study of genetic susceptibility, taken together, query the effects of genetics, ecology and behaviour on vectorial capacity of the M and S molecular forms. The observation that the outcome does not differ between studies done in entirely different ways would appear to firmly reject the hypothesis of differences in infection or vectorial capacity of the M and S molecular forms, at multiple geographic locations in West and Central Africa. Thus, it is concluded that the M and S molecular forms, along with A. arabiensis, are all equally dangerous vectors of human malaria.
There are two published accounts that report a difference in genetic susceptibility between M and S forms, and are thus at odds with the current results and the previous wild collections. In one of the reports, the mosquitoes tested were from pure M or S form laboratory colonies . In that study, the S form colony was found to be more susceptible than the M form colony. However, colonization of mosquitoes is associated with strong selection pressure and genetic bottlenecks due to the founder effect . Studies of colonization and domestication in other organisms also reveal large random and non-random loss of genetic diversity . Thus, phenotype results derived from colonies pertain only to those specific colonies, and cannot serve as a model for the source population without adequate replication using multiple independent colonies of each source population or subgroup.
The other discrepant study found the opposite direction of difference, that is, higher susceptibility of M form mosquitoes . There are multiple differences between that study and the current one, including their use of RT-PCR for parasite detection, while here microscopy was used. There may also be differences in larval collection methods, since here small numbers of larvae were collected per site over a large number of sites, to avoid sibling bias due to genetically related mosquitoes, and any potential influence of larval site microbial flora is also controlled for (see Methods), in order to specifically query mosquito genetics. In addition, in the current study, tested mosquitoes were collected at multiple times of the season across four years to control for temporal bias. These methodological details are not described in Boissière et al., so it is not possible to evaluate if they could underlie the different results. Most importantly, the current study demonstrates that the infections that contained the largest representation of both molecular forms in the same infection displayed the least infection difference between molecular forms, while the infections that were less representative of both forms were the most likely to capture uncontrolled experimental noise and consequently deviate artifactually from the null average. Boissière et al. generated 18 infections from a sympatric zone that may have included both molecular forms in the same infection, but the sizes of infection samples or their relative M and S form composition was not reported. Consequently, it is not possible to identify potential sources of the different results between that study and the current report.
The results presented here indicate that an important requirement for comparative susceptibility studies is the inclusion of all groups being compared in the same parasite challenge, and at the largest feasible sample size per infection. Based on the principles of statistical sampling, comparisons involving the largest and most representative infection samples should detect the most robust, statistically significant differences in infection prevalence or intensity, if they exist. Conversely, even in the absence of a true biological difference in susceptibility between population groups, smaller and less representative infection samples, which suffer the most stochastic variation due to experimental noise, will still display the largest (but artifactual) differences in infection.
It cannot be ruled out that M and S forms in different geographic locations or ecological situations could display different infection results than those reported here. Different populations are under distinct selective pressures, including exposure to other pathogens, and this could yield local differences between M and S forms for malaria susceptibility. Also, the evolutionary and demographic history of the M and S forms is not yet clear, and complex population admixtures are observed [12, 17], as well as reproductively isolated founder populations with distinct ecological characteristics such as Goundry , Forest-M and Mopti-M forms . Thus, different populations of the M or S form might not all share the same evolutionary history, and could therefore display different response to malaria parasites. The Goundry form from Burkina Faso is significantly more genetically susceptible to P. falciparum than are the sympatric M and S forms combined , and some chromosomal forms of A. gambiae s.s. in Mali appear to display different levels of natural infection .
Populations of A. gambiae s.s. in at least Mali [20, 21] and Burkina Faso at the current study site  segregate for genetic variation at immune genes between the molecular forms. For two gene loci, TEP1 and APL1, signatures of positive selection and low levels of nucleotide variation are found within the M form, whereas sympatric S form mosquitoes do not display evidence of positive selection. However, despite the evidence of different evolutionary pressure on immune genes among molecular forms, the preponderance of reports including this one clearly demonstrates that the molecular genetic variation at the TEP1 and APL1 loci is not associated with any consistent molecular form susceptibility difference to malaria parasites in nature. It is nevertheless puzzling that the genetic differentiation between molecular forms at two genes that protect mosquitoes against P. falciparum in gene silencing experiments, TEP1 and APL1A[21, 47], is not reflected in a phenotypic difference between molecular form susceptibility in nature. Resolving this apparent paradox will require further work.
Anopheles arabiensis is often regarded as more exophilic and zoophilic than A. gambiae s.s. . However, these results indicate that when A. arabiensis is offered the same infectious human blood meal as A. gambiae s.s., there is no difference in the establishment and extent of P. falciparum infection. This equivalent genetic susceptibility of the two vectors is important in the context of reports of that A. arabiensis has replaced A. gambiae after vector control interventions [2, 3]. Even though A. arabiensis may not be the primary vector in many locations, it has a genetic susceptibility equivalent to that of A. gambiae s.s.