Host‐plant associated genetic divergence of two Diatraea spp. (Lepidoptera: Crambidae) stemborers on novel crop plants

Abstract Diatraea lineolata and Diatraea saccharalis (Lepidoptera: Crambidae) are moths with stemboring larvae that feed and develop on economically important grasses. This study investigated whether these moths have diverged from a native host plant, corn, onto introduced crop plants including sorghum, sugarcane, and rice. Diatraea larvae were collected from these four host plants throughout the year in El Salvador and were reared on artificial diet until moths or parasitoids emerged. Adult moths were subsequently identified to species. Amplified fragment length polymorphisms (AFLPs) and mitochondrial DNA cytochrome oxidase I (COI) were used to examine whether or not there was genetic divergence of D. lineolata or D. saccharalis populations on the four host plants. Percent parasitism was also determined for each moth on its host plants. D. lineolata was collected from corn in the rainy season and sorghum in the dry season. D. saccharalis was most abundant on sugarcane in the rainy season and sorghum in the dry season. The AFLP analysis found two genetically divergent populations of both D. lineolata and D. saccharalis. Both moths had high levels of parasitism on their dominant host plant in the rainy season, yet had low levels of parasitism on sorghum in the dry season. The presence of two genotypes of both Diatraea spp. on sorghum suggest that host‐associated differentiation is occurring on this novel introduced crop plant.

Studies have demonstrated host-associated differentiation (HAD) for insects on native plants (Abrahamson et al., 2003;Dickey & Medina, 2010;Scheffer & Hawthorne, 2007;Stireman, Nason, & Heard, 2005) as well as for those on novel host plants (Feder, Hunt, & Bush, 1993;Forbes et al., 2009). Natural enemies of herbivores such as parasitoids may undergo cascading or sequential genetic divergence as they follow herbivores onto novel host plants (Stireman et al., 2005;Forbes et al., 2009), or herbivores may escape parasitism on their newly colonized host plants. Formation of host plant-associated insect populations has This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. been suggested to be more likely with longer associations between insects and novel host plants (Siemann, Rogers, & Dewalt, 2006). However, rapid adaptation and formation of host-associated insect populations have occurred on introduced cultivated crop plants (Dres & Mallet, 2002;Vialatte et al., 2005;Feder & Forbes, 2010). Additional examples of HAD on novel plants are likely to be uncovered given the large number of plants and insects, which have been introduced outside of their native range.
Host plant-associated differentiation of insect populations has been demonstrated in a number of systems. Native insects colonizing introduced crop plants have diverged in the course of two hundred years or so (Feder et al., 1993;Vialatte et al., 2005;Feder & Forbes, 2010). Rhagoletis pomonella (Walsh) (Diptera: Tephritidae) has shifted from the native plant hawthorne onto apple, an introduced crop transplanted from the Old World into the United States. Differences in phenologies between these two host plants contribute to isolating the host races of this insect (Feder & Forbes, 2010). The aphid Sitobion avenae Fabricius (Homoptera: Aphididae) has divergent genotypes on cultivated and uncultivated hosts (Vialatte et al., 2005), formed presumably in the last 100 years since its introduction into the United States. In Europe, corn was introduced in the last 500 years.
Currently, there are two host plant races of Ostrinia nubilalis Hübner (Lepidoptera: Crambidae), the European corn borer, which have unique pheromone blends and oviposition preference for their natal hosts (Bethenod et al., 2005). Similarly, the fall armyworm Spodoptera frugiperda J.E. Smith (Lepidoptera: Noctuidae) has been found to have two host plant strains, one feeding on corn and one on the introduced host rice (Pashley, Hardy, Hammond, & Mihm, 1990). The moths mate at different times of night and have unique pheromone blends, which contribute to ecological isolation (Groot, Marr, Heckel, & Schofl, 2010).
Larvae are endophagous, developing in host plant stems until they pupate and emerge as adults. This moth is considered an introduced species in the southern United States and can be a pest of cultivated plants including corn (Zea mays L.), sorghum (Sorghum bicolor L.), and rice (Oryza sativa L.) in various parts of its range (Cherry & Nuessly, 1993;Fuchs, Huffman, & Smith, 1979;Gifford & Mann, 1967;Vargas, Lastra, & Solis, 2013;White et al., 2001). Few population genetic studies of this insect have been conducted (Joyce et al., 2014;Lange, Scott, Graham, Sallam, & Allsopp, 2004;Pashley et al., 1990). Pashley et al. (1990 found that populations of D. saccharalis from the southern United States and Mexico were divergent from those of Brazil. D. saccharalis has been considered to be a single species with a wide distribution, but a recent study suggests that at least three or more species may exist (Joyce et al., 2014). In the southern United States, at least two putative species were found to occur, one in Texas and Louisiana, and a divergent genotype in Florida (Joyce et al., 2014). Exploring the genetic variability of D. saccharalis in a potential region of origin such as Central America would permit investigation of whether divergence has occurred between populations occurring on native crops such as corn or introduced crops. In Central America, another Diatraea species, D. lineolata (Walker) (Lepidoptera: Crambidae) also feeds on corn (Box, 1951;Quezada, 1978;Solis, 2004;Solis & Metz, 2015), a crop plant considered native in Mexico and Central America and domesticated in the last 5,000-10,000 years (Matsuoka et al., 2002). Because both D. saccharalis and D. lineolata are associated with corn and also feed on introduced crops, they may be subject to disruptive selection from crop plants introduced in the last few hundred years.
El Salvador has a tropical climate with a pronounced dry season and is classified as tropical savannah (Peel, Finlayson, & McMahon, 2007). During the rainy season from May until October, about 95% of annual rainfall occurs, with the months of April and November serving as transition months to and from the dry season. A widespread cropping rotation pattern in El Salvador is corn cultivation in the rainy season, followed by sorghum (an introduced crop) in the dry season (Quezada, 1978;Serrano-Cervantes et al., 1986). Rice and sugarcane, both nonnative cultivars, are also planted in the rainy season, while plantings of sugarcane persist throughout the year during all 12 months, essentially being cultivated as a perennial crop plant. As the tropical climate transitions from the rainy to the dry season and plants dry out, developing larvae of both moth species become quiescent and shift to a dormant form where they no longer feed and larvae lose their pigmented colored spots (pinacula) on the cuticle (Dyar & Heinrich, 1927;Kevan, 1944;Quezada, 1978). The climatic extremes between the rainy and dry season along with novel host plant availability could exert strong selective pressure on insect populations, contributing to insect populations adapting to novel host plants.
The objective was to determine whether either of two moth species, D. saccharalis and D. lineolata, has genetically divergent host plant-associated strains feeding on a native host, corn, and the introduced Old World crop plants, sugarcane, rice or sorghum. To test this, D. lineolata and D. saccharalis larvae were collected from available host plants in the rainy season and the dry season in El Salvador, Central America. If host-associated differentiation had occurred on novel host plants, insects from two host plants (such as corn and sorghum) in the same location should be more genetically divergent than insects collected from a single host plant species such as corn from distant geographic locations.

| Geographic location of study and field collection of larvae from host plants
The study was conducted in El Salvador, Central America, from August 2011 to May 2012 with an additional collection in rice in October 2013. We only used adult moths reared from field-collected, host plant-associated larvae. At each field site, host plants were searched for evidence of larval feeding in stems, indicated by insect frass exuding from a hole in the plant stem. Host plant stems with larvae were cut, and larvae were removed and placed individually in 60-ml plastic cups on artificial diet (Southland Products, Lake Village Arkansas) in order to rear larvae to adult moths. Larvae were transported to the laboratory for rearing at the University of El Salvador and observed at least twice a week. Any adult moths or parasitoid wasps or flies that emerged from larvae were preserved by freezing or by storage in 80% ethanol. We used the emerged parasitoids to determine the parasitism rate for each Diatraea species on each plant type by dividing the number of parasitoids emerged by the sum of moths and parasitoids emerged.

| Identification of adult moths by morphology
Reared adult Diatraea were identified to species by examining the adult male and female genitalia and comparing them to the key by Dyar and Heinrich (1927). Abdomens of adult moths were prepared for study by soaking the abdomen in cold 10% potassium hydroxide (KOH) overnight to be able to study the sclerotized structures for identification of the genitalia (Robinson, 1976). They were then placed in polyethylene genitalia vials with glycerin for future study and slide mounting. Voucher specimens are at the National Museum of Natural History, Smithsonian Institution, Washington, DC.

| DNA extraction
Following the identification of adult moths, the six legs of each moth were used for DNA extractions using the Qiagen DNeasy Blood and Tissue kit (Venlo, Netherlands) following the protocols for animal tissue with an incubation time of 2 h at 65°C (Qiagen 2006). Final products were eluted in 100 μl of AE buffer. The DNA quantity was measured using the Qubit ® dsDNA HS Assay kit (Life Technologies).
The quantity of DNA in samples averaged 2-5 ng of DNA per μl. Both male and female adults were used for molecular work.

| Population genetics: amplified fragment length polymorphisms (AFLPS)
Amplified fragment length polymorphisms were produced as described by Vos et al. (1995) and Joyce et al. (2010). Two primer combi- Prior to capillary electrophoresis, 0.4 μl of the GeneScan LIZ 500 size standard and 0.9 μl of HiDi formamide (all Life Technologies) were added to 1 μl of the final product of each sample. Sample fragments were separated using automated capillary electrophoresis by the ABI 3730 XL automated capillary DNA sequencer. GeneMapper version 5.0 (Life Technologies) was used to determine presence or absence of fragments. The peak detection threshold was set for each primer combination and was typically 100 luminescent units. Each AFLP marker was considered a locus and assumed to have two possible alleles (0 = absent, 1 = present). Bands not present in more than one individual were eliminated (i.e., private alleles) prior to further analyses, as they were not considered informative. Structure 2.3.4 software (Pritchard, Wen, & Falush, 2007) was used to group individuals with similar genotypes within each species. Structure uses a Bayesian algorithm to cluster individuals into K, which is defined as the number of genetically distinct populations in a data set. Parameters used for the analyses include the following: no a priori assignment of individuals to a known population, analysis for diploid insects, a burn-in of 10,000 iterations, an admixture model, and independent loci.
If collection locations were fewer than 5 miles apart, they were considered one location for the population genetic analysis of AFLPs with Structure software (Figure 1). The following two sites were combined and considered one collection site for genetic analysis with Structure; Cooperative El Nilo, and Santa Cruz Porrillo (D. saccharalis sites 1, 5). Five collecting locations were used for Structure genetic analysis for both D. lineolata and D. saccharalis ( Figure 1). The number of potential populations for K was estimated as the number of geographic sampling locations (5) plus 4 (K = 9) for both species as suggested by Pritchard, Stephens, and Donnelly (2000), and each iteration was run 20 times. At the completion of Structure runs, K was calculated for each species using Structure Harvester (Earl & VonHoldt, 2012;Evanno, Regnaut, & Goudet, 2005), to determine the most likely number of population clusters (K) for each species.
Analyses of molecular variance (AMOVA) tests were run using the AFLP data using GenAlEx 6.0 (Peakall & Smouse, 2006). For D. lineolata, all individuals collected were included in a comparison of host plant-associated individuals collected from corn and sorghum (Table 1).
The individuals collected from corn were obtained in the rainy season, and those from sorghum were collected in the dry season; the AMOVA of season consists of the same individuals and produced identical results as that obtained when comparing host plant populations. A separate AMOVA was run to examine genetic variation among four host plant-site populations. We did not include sites with less than five individuals in AMOVA; therefore, the one adult D. lineolata from site 3 in San Miguel was not included. The four host plant-site populations compared were (1) corn at UES-EE, (2) corn at La Union, (3) sorghum at La Union, and (4) sorghum at Santa Cruz Porillo. The host plant-site analysis was chosen as there was an unbalanced design with different numbers of host plants at each site (Sword, Joern, & Senior, 2005

| Mitochondrial DNA-COI
A 658-base pair region (the "bar code") of the mitochondrial COI gene region was sequenced from 26 D. lineolata and 23 D. saccharalis including a few individuals from each collection site (Tables 1 and 2, Figure 1).
The DNA used for sequencing COI was extracted as described above.
The bar-code region of the COI gene was amplified using primers for the mitochondrial DNA "bar code" of Lepidoptera (Hajibabaei, Janzen, Burns, Hallwachs, & Hebert, 2006). The forward primer LepF was 5_-ATTCAACCAATCATAAAGATATTGG-3, and the reverse primer sequence of LepR was 5_-TAAACTTCTGGATGTCCAAAAAATCA-3 (Hajibabaei et al., 2006). The touchdown PCR program consisted of an initial 2 min at 95°C, then 12 cycles of 95°C for 10 s, 58-46°C for 10 s with a lowering of 1°C temperature each cycle, and 72°C for 60 s. Following PCR and confirmation of amplification on an agarose gel, samples were cleaned up using a USB Exo-sapit pcr cleanup kit (Affymetrix, Inc., Santa Clara, Cal.

| RESULTS
Diatraea lineolata and D. saccharalis larvae were collected at sites throughout El Salvador (Tables 1 and 2, Figure 1) Adult identification was confirmed using genitalia and the key by Dyar and Heinrich (1927) (Figure 2a,b). Diatraea lineolata was collected from corn in the rainy season, and sorghum in the dry season, but was not found in any collections from sugarcane or rice. Diatraea saccharalis was collected from sugarcane and rice in the rainy season, and sorghum and corn in the dry season. Although D. saccharalis was collected in rice, another stemborer Rupela albinela (Cramer) (Lepidoptera: Pyralidae) was much more abundant in rice.

| Population comparisons using AFLPs
We produced 125 unique AFLP markers using two primer combinations for 65 D. lineolata moths. The 65 individuals consisted of 22 adults reared from corn and 43 adults reared from sorghum. Individuals from corn were reared from two sites (Table 1), while individuals from sorghum were obtained from three sites (Table 1). Structure Harvester found that K = 2, indicating two genetically distinct groups of D. lineolata ( Figure 3). Collections from corn were assigned primarily to one cluster (green, Figure 3), while collections from sorghum had two genetically distinct clusters (red and green, Figure 3). To visualize geographic variation in genotypes, the individual bars from structure, Samples from rice from two sites were all assigned to the green cluster, as were those from corn. Most individuals reared from sugarcane in the rainy season were assigned to the green cluster; collections from sorghum had a mix of two genotypes ( Figure 5).

| Analysis of molecular variation (AMOVA)
AMOVA of host plant-associated populations of D. lineolata found that host plants had a significant effect on genetic variation ( 3.4 was run using the following parameters: diploid individuals, 10,000 iterations, admixed data, and independent loci. Structure Harvester found that K = 2. The number above the bars represents the collection site for larval collections (see Table 1, Figure 1). The host plant that each larva was collected from is below the bars

| Mitochondrial DNA-COI sequences
We obtained mitochondrial COI sequences from 26 individuals of D. lineolata including a few from each collection site. There was little genetic variation between pairs of sequences of mitochondrial DNA.
Most individuals of D. lineolata were 99% or more similar, with 1-2 base pairs differing among some pairs. No previously sequenced COI   (Table S1). An additional small group of four individuals from East and South Texas were also 1% divergent from other D. saccharalis in that clade. The second large grouping in the phylogenetic tree consists of individuals from South America, including Brazil, Argentina, and Bolivia. This group of moths was 2-3% divergent from the other two clades (Table S1); similarly, the third group of D. saccharalis from Florida was 2-3% divergent from the other two clusters.

| DISCUSSION
Diatraea larvae were collected from host plants and reared to adults to provide definitive evidence of their host plant association. D. lineolata was the predominant Diatraea species feeding on corn. Diatraea saccharalis was common on sugarcane in the rainy season and occasionally found on rice. Both D. saccharalis and D. lineolata were abundant on sorghum in the dry season, with nearly twice as many larvae of D. lineolata collected from sorghum as on were collected on corn.
Diatraea lineolata is native to the Western Hemisphere and has had a much longer association with corn than with sorghum, which was introduced from Africa in the mid-1800s (Dillon et al., 2007). The as- F I G U R E 6 Diatraea saccharalis collected in El Salvador combined with individuals previously sequenced from GenBank and Bold. Bootstrap support values are based on 500 pseudoreplicates, and those above 80% are shown above supported nodes. Previously sequenced individuals have GenBank accession numbers or Barcode of Life identification numbers. El Salvador individuals are followed by an original collection number, followed by host plant of collection, and then collection site.
Collection sites correspond to the six sites listed in Table 2. Collection sites followed by a number (i.e., rice 1.5) correspond to the individual bars in the AFLP structure analysis in Figure 5 Two populations of D. lineolata from La Union from corn and sorghum were collected several months apart in the same field, yet the sorghum field had an additional genotype present which was not  (Via, 1999).
The mechanisms that may contribute to host-associated differentiation and maintenance of the isolation of insects on two or more host plant have been explored in other systems. Insects feeding on a novel host plants can develop host fidelity with oviposition preference for the natal plant (Diehl & Bush 1984;Wood et al., 1999), and assortative mating could occur between individuals from the same host plant. Host plant fidelity could be tested for D. lineolata populations from corn and sorghum, to determine whether adult moths prefer to oviposit on or prefer the odor of their respective host plants. Rhagoletis pomonella host plant-associated populations prefer the odors of their natal host plant and avoid odors of non-natal host plants (Forbes, Fisher, & Feder, 2005). Hybrids of two Rhagoletis host plant-associated populations did not prefer the host plant of either parent species; the authors suggest this could lead to inability to find suitable oviposition sites and failure to reproduce, contributing to further isolation of host plant-associated populations (Linn et al., 2004). McBride and Singer (2010) (Via, 1999). More genetic variability was found in the mitochondrial DNA sequences from D. saccharalis individuals than among those of D. lineolata. A recent study found evidence for several possible D. saccharalis species, but host plant-associated strains have not been previously investigated (Joyce et al., 2014). D. saccharalis mitochondrial DNA sequences from El Salvador had two groups (a large and small clade) that were 1% divergent from each other ( Figure 6,  (Figures 5 and 6).
The AFLPs used in the Structure analysis found that all D. saccharalis from rice were one genotype, most larvae from sugarcane were one genotype, and collections from sorghum consisted of two genotypes ( Figure 5).  (Quezada, 1978). Parasitoids in other systems are attracted to the odors associated with their host insects or odors associated with their host herbivores feeding on plants (Vet & Dicke, 1992). It is possible that the parasitoid flies (Tachinidae) attacking Diatraea spp. in El Salvador do not yet recognize volatiles emitted from sorghum because it is a novel plant, as this is the cases for some predators that have shifted host plants (Raffa, Powell, & Townsend, 2013).

CONFLICT OF INTEREST
None declared.

DATA ACCESSIBILITY
All mitochondrial DNA-COI sequences have been submitted to GenBank to request accession numbers.