The results of this updated survey of the African FAW are in general agreement with past studies and lead to three observations with potentially important implications concerning the entry and migration of FAW within the continent. The first is that previously observed regional differences in the distribution of the COI-CS haplotype persists even after the inclusion of additional sites and later collections. The second is that the African FAW population is dominated by the C-strain and a lineage that appears to be derived from an interstrain hybridization event, with the R-strain continuing to be mostly absent. The third is evidence of a new incursion of FAW into western Africa that appears to be from a different source than the previous introduction. The potential ramifications of these observations are discussed below.
A map of the distribution of farmland in Africa shows concentrations along the western and eastern coastal nations separated by large covers of natural forest with relatively little agricultural activity (Fig. 1). This pattern suggests limited availability of preferred host plants for FAW in the central area that could restrict west-east movements of large populations by natural migration, thereby explaining the persistence of the COI-CS asymmetry first observed in collections from 2016-201716 and still detected in this study. However, such limitations in migration run counter to the prevailing invasion scenario of a recent introduction of FAW into western Africa followed by the rapid migration to the rest of sub-Saharan Africa in the next two years. Population movements of that scale occurring on a regular basis should lead to a genetically homogeneous African FAW population. Therefore, the persistence of the observed east-west difference in COI-CS frequency indicates that transcontinental movements of large numbers of FAW by natural migration is unlikely and suggests that human-assisted migration through trade probably played a significant role in the rapid spread of this pest across Africa.
Other moth species also show evidence of a divide between populations in western and eastern Africa. In particular, the noctuid moth Busseola fusca (Fuller) and the pyralid Eldana saccharina (Walker) both show clades defined by mitochondrial haplotypes that are geographically separated in a manner sharing broad similarity to that displayed by FAW32–34. Both B. fusca and E. saccharina are native to Africa with the observed segregation attributed to geological and climatic events dating back to the Miocene and Pleistocene eras. The persistence of these phylogeographic patterns to the present day suggests the existence of significant physical barriers to natural migration on the African continent that impede homogenization and would be expected to impact the distribution and mixing of FAW populations.
The differential distribution of the COI-CS and COI-RS haplotypes in Africa is particularly interesting because these are commonly used markers for identifying the FAW host strains in Western Hemisphere populations. The strains differ in their association with plant hosts in the field, with the C-strain preferentially found in corn, sorghum, and cotton while the R-strain preferring turf and pasture grasses, alfalfa, millet, and rice20,21,35. The determination of what strains are present is critical to assessments of what crops are at risk of significant FAW infestations. However, such assessments are complicated by the fact that the strains are morphologically indistinguishable and so can only be identified by a small number of molecular markers that have so far been limited to genetic elements that map to mitochondria (such as COI17) or the sex chromosomes (i.e., Tpi18, FR36, esterase19). The association of the COI and Tpi strain markers with FAW collected from different plant host species has been consistently observed in surveys from both Americas, indicating that the strains are broadly distributed and a general characteristic of the species22,23,37. However, this correspondence is not absolute. For example, typically about 20% of FAW collected from corn hosts in the Americas display R-strain markers, and there are multiple FAW collections from corn or rice host plants where a majority will display the opposing strain markers21,24,38. These observations suggest that the association between FAW strain and plant host is more of a preference than a requirement, consistent with laboratory feeding studies indicating that both strains can successfully develop on the same set of plant hosts39,40. In addition, while reproductive barriers between the strains have been observed, they are incomplete, with successful hybridization between strains demonstrated in the laboratory and evidence of significant hybrid frequency found in field populations19,41−44. The hybrids appear to differ from the parental strains with respect to mating behavior and reproductive compatibility41–43, but the impact on plant host preference remains uncertain.
Because gene flow between strains is directly dependent on the formation of interstrain hybrids, we expect that the amount of strain divergence at any location will be impacted by whether and to what degree differential plant host preferences deter mating between strains. If this factor is significant, then divergence should tend to increase in locations where the primary hosts for each strain are abundant and separated, as under these conditions the strains can remain segregated. In contrast, in habitats with less host variety or abundance the two strains are more likely to overlap out of necessity, increasing the likelihood of cross hybridization. Given these considerations, it is likely that FAW displays a complex population structure made up of the C-strain, the R-strain, and inter-strain hybrids, where the proportion of each group and the frequency of mating between and within groups will depend upon the types and distributions of local plant hosts. If correct, then the degree of divergence between the two strains as measured by genetic differentiation could vary significantly by location. This scenario could explain the contradictory results from recent studies where differences between the strains at the whole genome level were detected in some comparisons45, but not in others46.
The situation in Africa differs from the Americas in that while both COI-CS and COI-RS are detected at high frequency, the TpiR haplotype is very rare in Africa, present in less than 1% of all specimens even in collections from habitats dominated by R-strain preferred plant hosts27. The Africa population is dominated by two COI Tpi configurations, COI-CS TpiC present in 41% of specimens and COI-RS TpiC at 47%, with the former preferentially found in the western Africa grouping of collection sites and the latter in eastern Africa (Fig. 3). The COI-CS TpiC configuration is normally representative of the C-strain and is the predominant haplotype found in specimens from cornfields in the Western Hemisphere16,21,22. The COI-RS TpiC is a configuration predicted to arise from an interstrain cross between a R-strain female and C-strain male with the daughters then back crossed to C-strain males (RC hybrid, Fig. 3). This configuration is found at variable frequencies in the Western Hemisphere, with the highest frequencies associated with C-strain preferred host plants18,24. These observations suggest that the COI-RS TpiC population is behaving as the C-strain with respect to plant host use and this seems to be the case in Africa as reports of agricultural damage by FAW has consistently been limited to C-strain plants like corn and sorghum7,27.
Given these results, we believe that the COI-RS TpiC population should be considered part of the C-strain group despite its derivation from an interstrain hybrid mating. Justification for this assumption comes from consideration of the effectiveness of Z-linked markers such as Tpi to distinguish strains, which can best be explained if the primary determinants for strain identity are also on the Z-chromosome and therefore physically linked to the Tpi gene. This proposition is consistent with observations that genetic differences between lepidopteran species are disproportionately sex-linked47. Based on this reasoning, it is likely that strain identity is defined primarily, if not solely, by the Z-chromosome, and we note that the TpiC marker indicates that the Z-chromosome of both COI-CS TpiC and COI-RS TpiC is of the C-strain. If these assumptions are correct, then we anticipate no significant differences in the behaviors of the COI-CS TpiC and COI-RS TpiC groups and suggest that their differential distribution across western and eastern Africa is probably due to chance.
The evidence of a second FAW introduction into Africa comes from what appears to be an influx of the TX-type COI-h2 haplotype in 2018 into Ghana, Togo, and perhaps Benin (Fig. 6B). In collections from October to December 2018, the COI-h2 haplotype made up 32% (51/162) of specimens from Ejura, Ghana, with a peak in October where it was the majority form, 58% (25/43). COI-h2 was only detected in October in Vogan, Togo, where it made up 61% (11/18) of specimens, while Benin showed 14% (5/37) during the same month. In contrast, in all other African locations, including collections from Togo, Ghana, and Benin from other years, the pooled COI-h2 frequency was only 0.5% (9/1501) with a range from 0.00-0.02%. Once introduced, the COI-h2 haplotype would be expected to disperse into the much larger COI-h4 population and become increasingly difficult to detect. This appears to be what occurred as the COI-h4 haplotype again predominated in Ghana and Togo after October 2018 (Fig. 5B).
If this incidence of COI-h2 does represent a new incursion it appears to be from a different source than that of the original introduction that gave rise to the predominantly FL-type COI-h4 composition of the Africa population. The possibility of a second incursion of Western Hemisphere FAW into western Africa is troubling as it suggests that the conditions that allowed for the first introduction are still in place despite efforts to improve monitoring and food security. This means that FAW subpopulations of concern thought to be currently rare or absent in Africa could be introduced at any time. This includes the R-strain, which would put important crops like rice, millet, and forage grasses at risk, and lines resistant to the Cry1F Bt-pesticide that have compromised certain Bt-products in Puerto Rico48.
In summary, the results for this genetic survey of FAW in Africa demonstrate the value of continued surveillance of pest populations at the continental level. FAW is in the process of becoming established in Africa with the distribution of permanent populations and pattern of regional migrations still to be determined. Identification of genetic structure as found for the COI-CS haplotype can define the magnitude and limits of natural migration. Evidence of a second incursion of FAW, most likely from the Western Hemisphere, indicate that continued introductions are plausible, which could rapidly alter the composition of the Africa population with respect to pesticide resistance and host range.