Genetic studies of fall armyworm indicate a new introduction into Africa and identify limits to its migratory behavior

The fall armyworm, Spodoptera frugiperda (J.E. Smith) is native to the Americas and a major pest of corn and several other crops of economic importance. The species has characteristics that make it of particular concern as an invasive pest, including broad host range, long-distance migration behavior, and a propensity for field-evolved pesticide resistance. The discovery of fall armyworm in western Africa in 2016 was followed by what was apparently a remarkably rapid spread throughout sub-Saharan Africa by 2018, causing economic damage estimated in the tens of billions USD and threatening the food security of the continent. Understanding the history of the fall armyworm invasion of Africa and the genetic composition of the African populations is critical to assessing the risk posed to different crop types, the development of effective mitigation strategies, and to make Africa less vulnerable to future invasions of migratory moth pests. This paper tested and expanded on previous studies by combining data from 22 sub-Saharan nations during the period from 2016 to 2019. The results support initial descriptions of the fall armyworm invasion, including the near absence of the strain that prefers rice, millet, and pasture grasses, while providing additional evidence that the magnitude and extent of FAW natural migration on the continent is more limited than expected. The results also show that a second entry of fall armyworm likely occurred in western Africa from a source different than that of the original introduction. These findings indicate that western Africa continues to be at high risk of future introductions of FAW, which could complicate mitigation efforts.

The specimens were of variable quality and in many cases a single PCR amplification did not produce sufficient product. In these cases, a nested PCR protocol was performed. For COIB analysis the first PCR amplification was performed with the primer pair CO1-891F and CO1-1472R. One microliter of this first reaction was amplified using primers CO1-924F (5′-TTA TTG CTG TAC CAA CAG G-3′) and CO1-1303R (5′-CAG GAT ART CAG AAT ATC GACG-3′). For the Tpi marker, the first amplification was performed using primers Tpi469F (5′-AAG GAC ATC GGA GCC AAC TG-3′) and Tpi1195R (5′-AGT CAC TGA CCC ACC ATA CTG-3′). One microliter of the first reaction was then amplified using primers Tpi412F and Tpi1140R. Relative locations of the primers are described in Fig. 2. Determination of strain-identity using COI and Tpi. To facilitate the analysis of large numbers of samples, strain identity was initially defined by a single site in COI or Tpi as previously described 26 , with the identification confirmed by comparisons with nearby strain-specific SNPs ( Fig. 2A,B). Designation of the COI and Tpi gene sites are preceded by an "m" (mitochondria) or "g" (genomic), respectively. This is followed by the gene name, number of base pairs from the predicted translational start site for COI, or the 5′ start of the exon for Tpi and, when relevant, the nucleotide observed. The COIB segment lies in the 3′ half of the COI gene. COI-based strain identity was determined by mCOI1164, with an A or G signifying C-strain and a T indicating R-strain ( Fig. 2A). Sites mCOI1176 and mCOI1182 are similarly strain specific. Strain identity by Tpi is defined by gTpi183 found in the fourth exon (TpiE4) of the predicted Tpi coding region (Fig. 2B). Site gTpi168 shows the same strain-specific pattern as gTpi183. Two other sites display hemispheric differences in their association with strain identity due to differences in the R-strain. The nucleotide T is found at gTpi165 (gTpi165T) that correlates with gTpi168C and gTpi183T in Western Hemisphere FAW populations but not in FAW surveyed in Africa 26,28 . Site gTpi180 shows the reciprocal pattern with gTpi180G correlating with R-strain associated gTpi183T in Africa FAW but not with those in the Western Hemisphere.
A complication of using the Tpi marker is that male FAW (Z/Z) carry two copies of the Tpi gene and so can be heterozygous (TpiR/TpiC) for the strain markers. These heterozygous specimens (TpiH) are identified by DNA sequencing chromatographs that show overlapping C and T curves at gTpi183, as well as overlapping C/G  www.nature.com/scientificreports/

Results
Regional distribution of genetic markers. Two earlier surveys found evidence for FAW population structure with respect to the frequencies of the mitochondrial COI strain markers and a Tpi exon variant designated TpiCa2 (Fig. 2B) 16,26 . FAW collected from the western African nations of Togo and São Tomé and Príncipe   16 , and this difference was still observed with the inclusion of data from five additional eastern locations (Table 2c,d) 26 .
In the current study the survey was expanded by 24 previously uncharacterized collections from 2016 to 2019 representing an additional 11 nations. This total of 34 collections were subdivided into two groups, western Africa (21 collections) and eastern Africa (13 collections, Fig. 3), based on the distribution of major agricultural areas that tend to concentrate in regions near the Atlantic and Indian oceans (Fig. 1). A statistically significant difference was observed in the COI-CS frequency between the groups (Table 2e). On average a majority of 58% of the FAW tested from the western African collections were COI-CS compared to a mean of 22% in the eastern group (Fig. 1). Multi-year data were available for six countries, Senegal (b), Ghana (c), Togo (d), Benin (e), Tanzania(r), and South Africa (w, Fig. 3). The COI-CS frequency was generally consistent over time at each location with two major exceptions. In Ghana 2016 and Benin 2017, COI-RS was the majority haplotype while COI-CS predominated in the subsequent 2 years.
With respect to the TpiCa2 haplotype, the previously observed significant differences between TogS&P versus other African collections was preserved even with the addition of new eastern collections (Table 2f). When the TogS&P pool was expanded to include other collections from the western region, the mean TpiCa2 frequency was still higher than in the east, 32% compared to 27% (Fig. 4A). However, this difference was no longer statistically significant (Table 2g).

Distribution of FAW strains and hybrids.
All the African specimens tested were collected from C-strain preferred host plants or habitats (corn or sorghum), leading to the expectation that the C-strain should be the majority in all collections. However, the COI-CS data suggests a differential distribution of the strains by geography, with the C-strain the majority in western African countries and the R-strain predominating in the more eastern and southern locations. This possibility was tested by examining the distribution of the Tpi strain markers in Africa. The TpiC haplotype, which is diagnostic of the C-strain, predominated in all locations with an overall mean frequency of 87% and a range of 74%-100% (Fig. 4B), with no significant difference in the mean TpiC frequency observed between western (91%) and eastern (95%) African FAW populations (Table 2h). Therefore, unlike the COI markers, the observed TpiC haplotype frequencies agree with expectations from host plants.
A subset of the specimens was analyzed with both COI and Tpi strain markers. Figure 5A describes examples of the crosses both within and between strains that can explain the observed COI and Tpi haplotype combinations. The COI marker is mitochondrial and so is inherited maternally, while the Tpi gene is on the Z-chromosome and displays sex-linked segregation. Figure 5B describes the results from 1197 specimens (404 from eastern Africa, 793 from western Africa) where both COI and Tpi sequence data were available. Two genotypes predominated in Africa. The COI-CS TpiC configuration that is concordant for both C-strain markers is the most common configuration in western Africa with a mean frequency of 50% that is not significantly different from the 36% frequency of the discordant COI-RS TpiC group (Fig. 5B, Table 2i). In contrast, the discordant COI-RS TpiC configuration predominates in eastern Africa at a frequency (73%) significantly higher than the COI-CS TpiC haplotype (18% , Table 2j). Overall, COI-CS TpiC frequencies in the western Africa group was significantly higher than in the eastern group (Table 2k) and, conversely, COI-RS TpiC was significantly more frequent in the east than west (Table 2l).
Consistent with previous studies the frequency of TpiR remains very low 14,27 . From a total of 1722 specimens tested only 18 were TpiR. These were found in Benin (3 specimens), the Democratic Republic of Congo (2), Ghana (2), South Africa (2), Senegal (2), and Togo (6), representing less than 5% of the specimens from each location. All 18 TpiR were identical in sequence for the 237 bp TpeI4 segment previously described and deposited into GenBank (MH729873) 14 . Another 211 African specimens produced sequence chromatographs with overlapping curves at certain polymorphic sites that are indicative of a TpiC/TpiR heterozygote (TpiH).  www.nature.com/scientificreports/ The mean TpiH frequency was higher in western Africa (13%) than eastern (8%), but the difference was not statistically significant (Table 2m). The TpiH specimens were found in all countries except for Ethiopia, Gabon, and Zimbabwe (where sample sizes were relatively low, < 15). This indicates that TpiR is broadly distributed in Africa but is rare and usually found in combination with TpiC as a likely hybrid.

Evidence of a new FAW introduction into western Africa. The COIB segment contains SNPs that
subdivide the COI-CS haplotype into four variants, COI-h1-4 ( Fig. 2A), that are found throughout the Western Hemisphere but consistently differ in their relative proportions in a manner that identifies two geographically distinct groups 32,33 . Specifically, the COI-h4 variant is the predominant form in populations that winter in Florida and most of the Caribbean while COI-h2 is the majority in South America and populations that winter in Texas and Mexico 18,21,32,34 . Using the metric m = (h4 − h2)/(h2 + h4) we consistently find that populations from Florida and Puerto Rico give positive values while those from Texas and South America are negative (Fig. 6A) 13,32 . Our past surveys of FAW from Africa found a predominance of COI-h4 in all tested locations and this was generally confirmed by the additional collections in this study. We found that all the collections outside of Ghana and Togo that span most of the African continent were strongly positive, with a mean of + 0.97 (Fig. 6A). Similar results were obtained for collections from multiple corn sites in Ghana in 2016 and 2017 where the COI-h2 haplotype was not found, resulting in an m-value of + 1.0 (Fig. 6B). In contrast, October 2018 collections from corn fields in Ejura, Ghana showed a preponderance of the COI-h2 haplotype (25 COI-h2 compared to 12 COI-h4) giving an m-value of − 0.35. Ejura is within 50 km of sites sampled in 2016 and 2017 and so these specimens are expected to be part of the same interbreeding population. Collections from Ejura in subsequent months show a return to positive m-values, though at levels below that observed in 2016-2017 (Fig. 6B).
A similar pattern was observed in collections from adjacent country of Togo. FAW from multiple locations in Togo (including the inland Vogan zone) in 2016 and from the more coastal Lomé area in 2017 gave positive m-values of + 0.96 and + 0.98, respectively, with only four specimens expressing the COI-h2 haplotype out of 513 tested (Fig. 6B). But in October 2018, collections from corn in the Vogan zone had an m-value of − 0.69, with 11 of the 18 specimens being h2 and four specimens COI-h1, a haplotype that has so far not been found elsewhere in Africa and is a minor haplotype in the Western Hemisphere. However, subsequent collections from the Vogan zone from November and December 2018 showed highly positive m-values with no COI-h1 or COI-h2 haplotypes

Discussion
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 to 2017 16 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. While the observation of a small number of identical COI and Tpi haplotypes throughout Africa strongly suggests a single, common source population 16 , regular and substantial migration at the continental scale should result in the homogenization of haplotype frequencies as well. Therefore, the persistence of the east-west difference in COI-CS frequency indicates that transcontinental movements of large numbers of FAW by natural migration is limited and suggests that smaller scale movements through trade probably played a significant role in the initial spread of this pest across Africa. www.nature.com/scientificreports/ 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 FAW [35][36][37] . 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 rice 20,21,38 . 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 COI 17 ) or the sex chromosomes (i.e., Tpi 18 , FR 39 , esterase 19 ). 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 species 22,23,40 . 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 markers 21,24,41 . 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 hosts 42,43 . 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 populations 19,[44][45][46][47] . The hybrids appear to differ from the parental strains with respect to mating behavior and reproductive compatibility [44][45][46] , 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 www.nature.com/scientificreports/ 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 comparisons 48 , but not in others 49 . 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 hosts 27 . 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 Hemisphere 16,21,22 . 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 (COI-RS TpiC hybrid, Fig. 5A). This configuration is found at variable frequencies in the Western Hemisphere, with the highest frequencies associated with C-strain preferred host plants 18,24 . 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 sorghum 7,27 .
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-linked 50 . 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 TXtype 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 to 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 Vogan, Togo and to a lesser extent in Ejura, Ghana after October 2018 (Fig. 5B). This persistence of lower m-values in Ejura in November and December 2018 is of potential interest as it could indicate that the COI-h2 haplotype is becoming established in the area, which might happen if the dispersion and mixing of populations in Ejura is less than occurs at the Vogan site. Continued monitoring of Ejura FAW is needed to determine if this represents a long-term shift in COI-h haplotype proportions.
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 FAW lines capable of developing on corn expressing the Cry1F Bt-product in Puerto Rico 51 .
How FAW reaches Africa from the Western Hemisphere is an important issue that remains unclear, with trade the most likely mechanism. Using the European Union (EU) as an example, assessments based on FAW intercept frequencies at EU ports of entry suggest that as many as a million FAW larvae could enter the EU annually 52,53 . A related study in Australia found evidence that sea cargo containers are routinely exposed to and often unintentionally carry economically important insects, potentially serving as a conduit for the spread of invasive pests 54 . These observations are consistent with our findings that the threat of new introductions of FAW into Africa is significant.
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.