Genetic characterization of fall armyworm (Spodoptera frugiperda) in Ecuador and comparisons with regional populations identify likely migratory relationships

Fall armyworm, Spodoptera frugiperda (J. E. Smith), is an important agricultural pest native to the Americas that has recently been introduced into the Eastern Hemisphere where it has spread rapidly through most of Africa and much of Asia. The long-term economic consequences of this invasion will depend on how the species and important subpopulations become distributed upon reaching equilibrium, which is expected to be influenced by a number of factors including climate, geography, agricultural practices, and seasonal winds, among others. Much of our understanding of fall armyworm movements have come from mapping genetically defined subpopulations in the Western Hemisphere, particularly in North America where annual long-distance migrations of thousands of kilometers have been documented and modeled. In contrast, fall armyworm mapping in much of the rest of the hemisphere is relatively incomplete, with the northern portion of South America particularly lacking despite its potential importance for understanding fall armyworm migration patterns. Here we describe the first genetic description of fall armyworm infesting corn in Ecuador, which lies near a likely migration conduit based on the location of regional trade winds. The results were compared with populations from corn habitats in select locations in the Caribbean and South America to investigate the possible migratory relationship between these populations and was further assessed with respect to prevailing wind patterns and the distribution of locations with climate favorable for fall armyworm population establishment and growth.


Introduction
The noctuid moth Spodoptera frugiperda (J. E. Smith) (Lepidoptera: Noctuidae), commonly called fall armyworm, is native to the Western Hemisphere. It is the primary insect pest of PLOS  FL-type/TX-type groupings [9,33,34]. The nuclear Tpi gene is also used as a marker of strain identity and may be more accurate than COIB [28]. Because of its location as a nuclear marker (as opposed to the mitochondrial COI), the Tpi marker can be used both in combination with COI and separately to explore the possibility of hybridization between strains [34,35]. Fall armyworm recently invaded Africa, where it rapidly spread throughout the continent and has subsequently been found in India and southeastern Asia [36][37][38][39]. The potential economic consequences are significant and there is much interest in using the Western Hemisphere information to better understand how fall armyworm is moving in the Eastern Hemisphere and its eventual distribution at equilibrium. Both will be largely determined by seasonal wind patterns, regional and local environmental conditions, and the availability of host plants, particularly corn. With respect to climate, the situation in Africa will likely reflect that of South America more than North America.
The northern portion of South America has not been studied using our suite of markers yet represents a potentially important region for understanding fall armyworm migration patterns. The two American continents are linked by two obvious pathways for natural migration between the continents for flying insects. One is the land connection comprised of Central America and the Isthmus of Panama, and the other is the chain of islands in the Caribbean Sea known as the Greater and Lesser Antilles. Our previous studies indicated that large population movements in the Caribbean were unlikely with little mixing observed between fall armyworms from the two continents, but we have no data from Central America or Panama.
We are in the process of extensive surveys of fall armyworm in Ecuador, which is the closest location we have analyzed to date to the Isthmus of Panama and Central America. We present in this study the first genetic description of fall armyworm from cornfields in Ecuador, which was then compared with those from select corn-growing locations in the Caribbean and South America to investigate possible migratory relationships. The results were further assessed within the context of prevailing wind patterns during the collection period and the distribution of climate favorable for fall armyworm growth and establishment. We further demonstrate that COIB can differentiate fall armyworm from other Spodoptera species, eliminating the need to sequence COIA and thereby simplifying the genetic analysis.

Species identification using COIB
The mitochondrial COI and nuclear Tpi genes contain polymorphisms useful for the analysis of fall armyworm populations (Fig 1). The 5' segment of the COI locus, COIA, contains the barcode region capable of species and host strain identification [32], while the more 3' segment, COIB, produce haplotypes that also identify strain and differentiate between two geographically separated populations [9]. To streamline the analysis by COI we tested whether COIB could substitute for COIA for species identification.
Sequence information from the COIB region were obtained for 13 Spodoptera species from GenBank. These were found to share a 259-bp COIB segment (designated COIB259). Despite the short length this sequence was sufficient to distinguish fall armyworm of both strains from the other Spodoptera species by Neighbor-joining analysis (Fig 2). Spodoptera cosmiodes (Walker) and S. descoinsi ((Lalanne-Cassou & Silvain) were identical in their COIB259 sequence, while S. albula (Walker) and S. dolichos (F.) appear to be closely related. A total of 143 specimens from Manabi, Ecuador were analyzed by COI sequencing with 22 COIB259 haplotypes observed. One Ecuador haplotype clustered with the consensus rice-strain COIB259 while the remainders were most similar to the consensus corn-strain haplotype (Fig 2).

Strain-identity based on specific COI and Tpi polymorphisms
The two strains can also be distinguished by polymorphic sites in the COI and Tpi genes (Fig 1). The site COIB 1164 is strain-specific with a T 1164 indicating rice-strain (COI-RS) and an A 1164 or G 1164 diagnostic of the corn-strain (COI-CS). The results from this single polymorphic site were consistent with the phylogenetic comparisons as the single Ecuador specimen clustering as COI-RS expressed T 1164 , while the remaining Ecuador haplotypes clustering with the consensus COI-CS sequence expressed either A 1164 or G 1164 . Overall, the COI-CS haplotype predominated at all locations as expected since collections were limited to cornfields ( Fig 3A).
The COI strain data were confirmed by the Tpi marker, where strain identification is defined by the gTpi183Y polymorphism in exon4 ( Fig 3B). The Z-chromosome-linked Tpi gene is present in two copies in males (ZZ) and one copy in females (ZW), which means that a portion of the larval collections, which were not sexed, have the potential to be heterozygous for Tpi. As a result, three gTpi183Y sequencing chromatographs are observed that derive from homozygosity (ZZ) or hemizygosity (ZW) for C 183 (TpiC or corn-strain), T 183 (TpiR or ricestrain), or the presence of both C and T (TpiH or hybrid). The number of TpiC and TpiR chromosomes were estimated and both Ecuador locations were over 95% TpiC (Fig 3B).

COI-h haplotype proportions
Near the COIB 1164 strain-diagnostic marker is a polymorphic site, COIB 1287 , that is partially strain-specific. The COI-RS T 1164 variant is always associated with A 1287 , while the corn-strain A 1164 or G 1164 polymorphisms can be found with either A 1287 or G 1287 . As a result, COI-RS is identified by the combination T 1164 A 1287 while COI-CS includes four variants designated the COI-h haplotypes, A 1164 A 1287 (h1), A 1164 G 1287 (h2), G 1164 A 1287 (h3), and G 1164 G 1287 (h4) (Fig 1A). The relative proportions of h2 and h4 as measured by the metric (h4-h2)/(h2+h4) vary with location in such a way that they subdivide fall armyworm populations in the Western Hemisphere into two groups (FL-type and TX-type) [33]. Examination of specimens from Ecuador were of the TX-type, consistent with locations in South America and different from that found in the Greater Antilles and Florida (Fig 4).

Haplotype network comparisons of Tpi variation
Network analysis was used to compare the genetic variation in the Tpi gene of Ecuador fall armyworm with those from four regional collection sites. The Tpie4i4 haplotypes identified from Ecuador were similar in sequence to each other, with no more than two mutations difference between one haplotype and the most closely related variant (Fig 5). There was considerable overlap in the haplotypes found in Bolivia, Peru, and Trinidad-Tobago with those in Ecuador and with each other, again with little variation between them. This was in marked contrast with the collections from Puerto Rico. Seven haplotypes unique to Puerto Rico were identified that each differed by at least eight mutations from those at the other locations. These results suggest that the Puerto Rico population is effectively isolated from the other locations and is in the process of genetic divergence. In contrast, there appears to be significant interactions between the fall armyworms in Ecuador, Peru, Bolivia, and Trinidad and Tobago despite the thousands of kilometers that separate each of these locations.

Discussion
The characterization of fall armyworm from Ecuador provides an important data point from a region in the Western Hemisphere where fall armyworm has not been studied extensively with our suite of molecular markers. Ecuador is of particular interest because of its location relative to potential intercontinental migration pathways and air transport systems that are likely driving fall armyworm migration in South America. The South American Low Level Jet is the predominant lower altitude wind system in the region, forming an easterly trade wind from the equatorial Atlantic that is deflected sharply south-southeastward by the higher elevations of the Andes Mountain Range [41]. This occurs just south of Ecuador and most likely directs migration of fall armyworm populations to the southern and eastern portions of the continent. With respect to the survey locations, these observations suggest that populations in Trinidad and Tobago could contribute to fall armyworm in Ecuador, which in turn acts as a migratory source for Peru. However, seasonal winds in Ecuador during two periods of high fall armyworm levels are generally of low velocity and inconsistent direction (Fig 6). Such conditions raise the possibility that the fall armyworm populations in Ecuador could be relatively isolated from the rest of the continent.
Our findings of strong similarities in the COIB ( Fig 5) and Tpi (Fig 6) haplotype compositions of fall armyworms from Ecuador, Peru, Bolivia, and Trinidad and Tobago are consistent with extensive mixing of the fall armyworm from these four locations, while those from Puerto Rico appear to be isolated. This result has important implications if true as Puerto Rico is only 100 km further from Ecuador than Trinidad and Tobago based on direct flight comparisons. It Fall armyworm in Ecuador suggests that despite the known ability of fall armyworm to migrate thousands of kilometers per season in North America, the approximately 800 km of water separating Puerto Rico from the South American mainland (compared to only 50 km for Trinidad and Tobago) is a significant barrier preventing regular migration of fall armyworm from one site to the other.
To identify the most likely drivers of population movements in this region we examined two environmental factors that are known to significantly impact fall armyworm migration and dispersion patterns. These are wind vectors and the distribution of habitats with suitable climates. Substantial populations of fall armyworm were observed in cornfields in Ecuador during the spring and fall seasons. Wind vectors during these periods in most of Ecuador and Peru exhibit low velocity and inconsistent direction (Fig 6A), which are not compatible with the wind-dependent long-distance migration behavior observed in North America. This can be visualized by HYSPLIT analysis where we projected the direction and distance of dispersion promoted by wind vectors averaged for the months of March and September over a single 12-hour nocturnal flight period (Fig 6B). Dispersion projections from Trinidad and Tobago show a strong westward bias consistent with the strong prevailing wind vectors there, which contrasted with the more localized distributions originating from Ecuador and Peru.
More favorable to fall armyworm movements were the extensive areas with supportive habitats for fall armyworm populations (Fig 7). The CLIMEX analysis was set to show shading in map location with values equal or greater than 33, with an ecoclimatic index (EI) value of 30 or above considered highly favorable [42]. The EI climate suitability map describes a contiguous pathway of favorable habits from Trinidad and Tobago that extends through the eastern sections of Ecuador, Peru, and Bolivia. From these observations, if the widely dispersed fall armyworm from Trinidad and Tobago, Ecuador, Peru, and Bolivia represent a single interbreeding population as suggested by their genetic similarities, then movement between populations is likely to be dominated by short flights to adjacent favorable habitats rather than the long-distance wind-directed migration observed in North America. This migratory behavior should also be compatible with host plant availability as agricultural activity in the region, including corn production, largely overlap the areas of favorable climate for fall armyworm [43]. Less certain are the population densities that can be supported. The large migratory populations observed and modeled in North America are generally associated with extensive acreages of contiguous and high intensity corn plantings [4], conditions that tend to optimize population density and may not be typical of corn production in the northern region of South America. While speculative, these observations when considered in total suggests that fall armyworm movements in this region of South America may be associated with smaller migratory populations covering shorter distances in more variable directions.
In summary, it is important to note that these studies are limited to the fall armyworm infesting corn as collections from other host plants, particularly those associated with the rice- strain have been difficult to obtain in this region. Based on this sampling we conclude that the COI and Tpi compositions of fall armyworm found in corn in Ecuador are consistent with populations in the northern regions of South America extending from Trinidad and Tobago to the east, Peru to the south west, and Bolivia to the south undergoing substantial mixing due to migration and population dispersion. This pattern is consistent with the seasonal wind patterns consisting of strong eastward trade winds over Trinidad and Tobago that are deflected southward by the Andes mountain range in Peru as well as a favorable distribution of suitable habitats that could support localized movements of fall armyworm from Ecuador into eastern Peru and Bolivia. Evidence of substantial introgression of fall armyworm populations from Central America or the Greater Antilles into Ecuador have yet to be observed.

Specimen collections and DNA preparation
Ecuador specimens were obtained in 2018 as larvae from corn plants from two cantons in Manabi province, Portoviejo (March and September) and Tosagua (March) ( Table 1). Collections from Peru, Puerto Rico, Trinidad and Tobago, Bolivia, multiple locations in Florida, and the Dominican Republic were described previously (Table 1). For these collections, the data for COIB were previously reported [33,44], while data for Tpi were from this study. No endangered or protected species were involved in this study. We obtained permission from private farmers for access and data collection in their fields.
Collected specimens were stored either air-dried or in ethanol at room temperature. A portion of each specimen was excised and homogenized in a 5-ml Dounce homogenizer (Thermo Fisher Scientific, Waltham, MA, USA) in 800 μl Genomic Lysis buffer (Zymo Research, Orange, CA, USA) and incubated at 55˚C for 5-30 min. Debris was removed by centrifugation at 10,000 rpm for 5 min. The supernatant was transferred to a Zymo-Spin III column (Zymo Research, Orange, CA, USA) and processed according to manufacturer's instructions. The DNA preparation was increased to a final volume of 100 μl with distilled water. Genomic DNA preparations of fall armyworm samples from previous studies were stored at -20˚C. Species identity was initially determined by morphology and confirmed by sequence analysis of the COIB region.
For fragment isolations, 6 μl of 6X gel loading buffer was added to each amplification reaction and the entire sample run on a 1.8% agarose horizontal gel containing GelRed (Biotium, Hayward, CA) in 0.5X Tris-borate buffer (TBE, 45 mM Tris base, 45 mM boric acid, 1 mM EDTA pH 8.0). Fragments were visualized on a long-wave UV light box and manually cut out from the gel. Fragment isolation was performed using Zymo-Spin I columns (Zymo Research, Orange, CA) according to manufacturer's instructions. The University of Florida Interdisciplinary Center for Biotechnology (Gainesville, FL) and Genewiz (South Plainfield, NJ) performed the DNA sequencing.
DNA alignments and consensus building were performed using MUSCLE (multiple sequence comparison by log-expectation), a public domain multiple alignment software incorporated into the Geneious Pro 10.1.2 program (Biomatters, New Zealand, http://www. geneious.com) [45]. Phylogenetic trees were graphically displayed in a neighbor-joining (NJ) tree analysis also included in the Geneious Pro 10.1.2 program [46]. Phylogenetic networks were estimated by the TCS statistical parsimony algorithm ( [47]) incorporated in the software program PopArt [48].

Characterization of the CO1 and Tpi gene segments
The genetic markers are all single nucleotide substitutions. Sites in the COI gene are designated by an "m" (mitochondria) while Tpi sites are designated "g" (genomic). This is followed by the DNA name, number of base pairs from the predicted translational start site (COI), 5' start of exon (Tpi), or 5' start of the intron (TpI4) and the nucleotides observed using IUPAC convention (R: A or G, Y: C or T, W: A or T, K: G or T, S: C or G, D: A or G or T).
The COI markers are from the maternally inherited mitochondrial genome. The COIB segment was amplified by CO1 primers 891F and 1472R (Fig 1A). Overlapping COIB segments for various Spodoptera species were obtained from GenBank and included, S. abula (HQ177407), S. litura (F.)(HQ177375). These sequences had in common a 259-bp segment designated COIB259, which was used for species identification by phylogenetic analysis.
Variants in the Tpi e4 exon segment can also be used to identify host strain identity with results generally comparable with the CO1 marker [28]. The gTpi183Y site is on the fourth exon of the predicted Tpi coding region and was PCR amplified using the Tpi primers 412F and 1140R (Fig 1B). The C-strain allele (TpiC) is indicated by a C 183 and the R-strain (TpiR) by T183 [28]. The Tpi gene is located on the Z sex chromosome that is present in one copy in females and two copies in males. Because the genomic DNA was directly sequenced, males heterozygous for Tpi alleles will simultaneously display both alternatives at polymorphic sites, which if different are easily identified by overlapping sequencing chromatographs. Heterozygosity at site Tpi 183 was limited to C/T and was denoted as TpiH.
The Tpie4i4 segment includes a 52 bp portion of the e4 exon followed by approximately 172 bp of the TpI4 intron, the latter of which is of variable length due to frequent insertions and deletions (indels). The segment was sequenced with primer 891F for the initial sequencing reaction and 1140R for 2 nd strand sequence confirmation when needed in cases of ambiguity. This segment was chosen for analysis because it empirically had the most consistent sequence quality with the given primers. A variable but often high percentage of specimens were heterozygous for frameshift mutations in the intron that could be identified by overlapping chromatographs immediately after the polymorphism. These were not further analyzed.

Calculation of haplotype numbers
The mitochondrial COI markers are calculated directly as the number of specimens exhibiting the COI haplotypes divided by total specimens. Because Tpi is a sex-linked nuclear gene, the number of Tpi genes present will differ by sex. Specimens with a TpiC or TpiR haplotype could be either a homozygous male or hemizygous female. We assumed a 1:1 sex ratio for the larval specimens so that the average number of Tpi genes per specimen is given as 1.5 as calculated: (2 in males + 1 in females)/2. Based on this reasoning the number of TpiC and TpiR specimens were multiplied by 1.5 to estimate the number of chromosomes carrying each marker. In comparison, TpiH specimens are heterozygous and so carried one of each marker. The estimated number of TpiC chromosomes was calculated by 1.5TpiC + TpiH, and TpiR chromosomes = 1.5TpiR + TpiH.

CLIMEX climate suitability analysis
CLIMEX is a dynamic simulation model that estimates the potential geographical distribution and relative abundance of a species according to climate [42]. In principle, extrapolating locations with favorable climate based on the biological parameters of a species should provide a reasonable estimate of geographical distribution, and for pests identify locations at high risk of infestation. CLIMEX projections for fall armyworm has been performed at continental and global scales and while there were a few variations in the biological parameters for fall armyworm used between studies these generated only modest differences in the distribution maps [49][50][51]. We performed a CLIMEX suitability analysis focused on the northern portion of South America, using fall armyworm parameters from du Plessis et al (2018) [49] (Table 2). Climate information was imported from Climond (www.climond.org) [52], using historical data from 1961-1990 at a resolution of 10'.
The Ecoclimatic Index (EI) is calculate from the annual Growth Index (GI) and the Stress Index (SI). The GI combines the minimum limit, optimum lower, optimum upper, and maximum limit for fall armyworm temperature and moisture indices calculated on a weekly basis, then averaged for the year to provide a measure of population growth potential. This is counterbalanced by the SI, which is a measure of unfavorable conditions focused primarily on temperature and precipitation. The EI integrates the GI and SI and is presented on a 0-100 scale, where 100 represents 100% suitability throughout the year (as would occur in an incubator). EI values greater than 30 are considered to be favorable for the long-term survival of the species in the region [42].
For this study, the Compare Locations (1 species) function in the CLIMEX program was used with the Grid Data simulation file. The species' known distribution data was imported from CABI's Invasive Species Compendium (www.cabi.org/isc). The location component was set to CM:10 South America. No climate change scenario or irrigation components were set. An EI map was created from the simulation, with the map specifications set at 0.16 diameter circles, showing zeros with three legend items.

HYSPLIT air trajectory projections
Air transport trajectories for select locations were estimated using the Hybrid Single Particle Lagrangian Integrated Trajectory Model at the Air Resources Laboratory (ARL) READY web site run by NOAA (http://ready.arl.noaa.gov/HYSPLIT.php) [53]. Projections were made for air transport conditions averaged for the 30-day period from March 1-30, 2018 and September Fall armyworm in Ecuador 1-30, 2018. Fall armyworm migrates nocturnally so the duration of continuous flight was limited to a 12-hour period beginning at dusk, with a starting altitude of 500 m AGL and a maximum altitude of 1500 m AGL [4]. The pathways of the projections from each location were averaged and displayed as a frequency distribution with percentages reflecting the proportion of trajectories entering a given grid.