Ecological and genetic relationships of the Forest-M form among chromosomal and molecular forms of the malaria vector Anopheles gambiae sensu stricto

Abstract

Background: Anopheles gambiae sensu stricto, one of the principal vectors of malaria, has been divided into two subspecific groups, known as the M and S molecular forms. Recent studies suggest that the M form found in Cameroon is genetically distinct from the M form found in Mali and elsewhere in West Africa, suggesting further subdivision within that form.

Methods: Chromosomal, microsatellite and geographic/ecological evidence are synthesized to identify sources of genetic polymorphism among chromosomal and molecular forms of the malaria vector Anopheles gambiae s.s.

Results: Cytogenetically the Forest M form is characterized as carrying the standard chromosome arrangement for six major chromosomal inversions, namely 2La, 2Rj, 2Rb, 2Rc, 2Rd, and 2Ru. Bayesian clustering analysis based on molecular form and chromosome inversion polymorphisms as well as microsatellites describe the Forest M form as a distinct population relative to the West African M form (Mopti-M form) and the S form. The Forest-M form was the most highly diverged of the An. gambiae s.s. groups based on microsatellite markers. The prevalence of the Forest M form was highly correlated with precipitation, suggesting that this form prefers much wetter environments than the Mopti-M form.

Conclusion: Chromosome inversions, microsatellite allele frequencies and habitat preference all indicate that the Forest M form of An. gambiae is genetically distinct from the other recognized forms within the taxon Anopheles gambiae sensu stricto. Since this study covers limited regions of Cameroon, the possibility of gene flow between the Forest-M form and Mopti-M form cannot be rejected. However, association studies of important phenotypes, such as insecticide resistance and refractoriness against malaria parasites, should take into consideration this complex population structure.

Figure 1

Map of collection sites and genotype distributions. Relative proportion of molecular form and 2La chromosome inversion karyotypes for collection sites in Mali and Cameroon are shown in pie charts. Colours and corresponding land cover types for all 17 International Geosphere-Biosphere Programme (IGBP) global vegetation classes are shown in the legend at the bottom of the figure.

Figure 2

Structure clustering results. A: Membership coefficients of all individuals for K = 7. Bayesian clustering based on seven markers, which include molecular form, the 2La, 2Rj, 2Rb, 2Rc, 2Rd, and 2Ru inversions. Each vertical line represents a single individual. Different colours represent different clusters (= forms). The green colour corresponds to Forest-M form, light-blue to Forest-S form, yellow to Godola-S form, blue to Savanna-S form, purple to Bamako-S form, red to Mopti-M form bc/bc homozygotes, and orange to the "other" Mopti-M forms. Individuals are ordered such that the latitude of collection sites increase from left to right. The latitude and longitude of each collection site are indicated on top of the bar plots and the names of the collection sites are indicated below the bar plots. The associated karyotypes are summarized in Table 2. B: Membership coefficients for K = 4 clustering based on microsatellite markers. The green colour corresponds to Forest-M form, light-blue to Forest-S form, Blue to all other S forms, and red for Mopti- M form. Note that separation of Godola-S, Savanna-S and Bamako-S were not supported at K = 4. C: Membership coefficient for K = 5 clustering based on microsatellite markers. The color scheme is similar to Figure 3B, with the addition of the Bamako-S form indicated in purple.

Figure 3

Posterior probabilities of karyotype and microsatellite based Bayesian cluster analyses. The boxplots were drawn after collecting posterior probabilities of 5 independent runs. The medians were drawn as thick black lines. Some boxes appear as a single line because the range of the posterior probabilities is too narrow to show as a box. Outliers are shown as open circle (○). A: Clustering of individuals into seven groups (K = 7) was the most probable solution for molecular form and karyotype data. B: Analysis of clustering of genotypes at 20 microsatellite loci yielded several outcomes with relatively high likelihood. Liklihoods for K = 4, K = 5 and K = 7 are similarly distributed (Wilcox twosample test, P > .05).

Figure 4

Unrooted phylogenetic tree (neighbour joining) of the seven groups of An. Gambiae identified by Bayesian analysis (using Structure software), as illustrated in Figure 2A. Tree is based on pair-wise F_ST values derived from allele frequencies at 20 microsatellite loci on chromosomes 2 and 3. Distances between all branches are significant (see Table 3).

Figure 5

Precipitation and molecular form. Correlation between the proportion of M molecular forms and the average annual precipitation in (A) Cameroon and (B) Mali. Average annual precipitation was calculated using annual precipitation at multiple weather stations at or near collection sites (within a 15 km radius) and over multiple years. ○ = average annual precipitation at each site. Details of precipitation data are provided in Table S2. A. The proportion of the M forms increases as precipitation increases in Cameroon. B. The opposite trend was observed in Mali. These seemingly conflicting results illustrate that the Forest-M form and Mopti-M form differ with respect to habitat preference as well as being differentiated genetically. The solid line is a linear model (y~ax+b) of the data. (A) For the Cameroon data, a = 0.0395 ± 0.0184 (P = 0.121) and b = -50.5 ± 37.8 (P = 0.274). (B) For the Mali data, a = -0.108 ± 0.0150 (P = 0.00197) and b = 122 ± 10.3 (P = 0.000283).

Figure 6

Precipitation and chromosome inversion. A: Inverse correlation between 2La inversion frequency and average annual precipitation. Average annual precipitation was calculated as for the data represented in Figure 5. ○ = average annual precipitation for each site. Details of precipitation data are provided in Additional file 1. The solid line is a linear model (y~ax+b) of data with coefficients of a = -0.0434 ± 0.00395 (P = 1.65 × 10^-6) and b = 120 ± 5.85 (P = 7.39 × 10^-9). B: Inverse correlation between 2Rb inversion frequency and annual precipitation. For a linear model, a= -0.0385 ± 0.00491 (P = 2.60 × 10^-5) and b = 101 ± 7.25 (P = 2.16 × 10^-7).