ECOLOGY AND POPULATION BIOLOGY Identification and Comparison of Fall Armyworm (Lepidoptera: Noctuidae) Host Strains in Brazil, Texas, and Florida

ROD N. NAGOSHI,1,2 PIERRE SILVIE,3 ROBERT L. MEAGHER,1 JUAN LOPEZ,4 5 AND VILMAR MACHADO

Ann. Entomol. Soc. Am. 100(3): 394Ð402 (2007) ABSTRACT Fall armyworm, Spodoptera frugiperda (J.E. Smith) (Lepidoptera: Noctuidae), is a major economic pest throughout the Western Hemisphere. Studies of populations in the southern United States and the Caribbean demonstrated the existence of two morphologically identical but genetically distinct host strains. These races can be distinguished by using polymorphisms in the mitochondrial cytochrome oxidase I gene that deÞne two distinct maternal lineages that correlate with strain-speciÞc behaviors in Florida populations. Although there is evidence of different biotypes in Brazil, it has not been demonstrated that these biotypes are equivalent to the U.S. strains. Sampling from Brazil indicates that its fall armyworm populations consist of the two strain-speciÞc haplotypes found in Florida and also display the expected biases in plant host distribution. The same genetic markers also were present in samples from Texas, a major source of migrating fall armyworm in North America. These results indicate that the biology and behaviors of Brazilian fall armyworm populations are similar to those found in North America.

KEY WORDS fall armyworm, Spodoptera frugiperda, haplotype

Spodoptera frugiperda (J.E. Smith) (Lepidoptera: phisms (Pashley 1986), genetic polymorphisms (Lu et Noctuidae) is a pest of a number of agricultural crops. al. 1992, McMichael and Prowell 1999, Prowell et al. In the United States, fall armyworm is responsible for 2004), and mitochondrial haplotyping (Pashley and substantial economic damage to Þeld and sweet corn, Ke 1992, Lu and Adang 1996). The molecular differ- Zea mays L., sorghum, Sorghum vulgare Pers., and ences correlate with differences in physiology, devel- several turfgrass varieties (Sparks 1979, Pashley 1988b, opment, and behavior that are consistent with genet- Foster 1989). It also is considered a sporadic, but ically distinct populations (Pashley 1988a, Whitford et important, late-season pest of cotton, Gossypium hir- al. 1988, Pashley et al. 1995, Veenstra et al. 1995). An sutum L., and sugarcane, Saccharum spp. L. (Hall 1988, important consideration for pest management is that Pashley 1988b). In Brazil, fall armyworm has long been the two strains differ in their susceptibility to chemical considered one of the primary pests of maize and has and biological agents. R-strain larvae were more sus- more recently been observed infesting cotton (Mar- ceptible to transgenic Bacillus thuringiensis (Bt) Ber- tinelli et al. 2006). liner cotton and to the insecticides diazinon and car- The highly polyphagous behavior of fall armyworm baryl, whereas the reverse was true for carbofuran in the United States and the Caribbean is in part due (Pashley et al. 1987, Adamczyk et al. 1997). In addi- to the presence of two strains that differ in host pref- tion, several Bermuda grass, Cynodon dactylon (L.) erence (Pashley 1986, Prowell et al. 2004). The corn Pers., cultivars showed differential resistance to the (C) strain is associated with maize and sorghum, two strains (Leuck et al. 1968, Lynch et al. 1983, Pa- whereas the rice (R) strain is preferentially found in shley et al. 1987, Quisenberry and Whitford 1988, rice and turfgrass. The two strains are morphologically Jamjanya et al. 1990). identical, so they can only reliably be distinguished by Evidence that the same two strains exist in South molecular methods, most notably allozyme polymor- America is suggestive but inconclusive. AmpliÞed fragment length polymorphism (AFLP) analysis of 1 Center for Medical, Agricultural and Veterinary Entomology, larvae collected from maize and rice plants in sev- Agricultural Research Service, U.S. Department of Agriculture, eral locations in Brazil was consistent with the ex- Gainesville, FL 32604. istence of two distinct populations speciÞc to each 2 Corresponding author, e-mail: [email protected]ß.edu. 3 Centre de Coope´ration Internationale en Recherche Agronomique plant host (Busato et al. 2004). Fall armyworm lines pour le De´veloppement, UPR syste`me cotonniers en petit paysannat, derived from larvae collected from either maize or CIRAD-Montpellier, F 34398, France, Brasilia-DF, Brazil. rice also differed in physiology, development, and 4 Areawide Pest Management Research Unit, Agricultural Research insecticide susceptibility (Busato et al. 2005a,b, Service, U.S. Department of Agriculture, College Station, TX 77845. 5 Universidade do Vale do Rio dos Sinos-UNISINOS, Laborato´rio de 2006). However, another AFLP study comparing Biologia Molecular, Sa˜o Leopoldo, RS, Brazil. populations from Brazil, Argentina, Mexico, and the May 2007 NAGOSHI ET AL.: FALL ARMYWORM HOST STRAINS 395

Table 1. Source locality and host information

Sample Location Collection Plant/habitat Date Source (in Fig. 2) Gainesville, FL Colony ArtiÞcial Oct. 2004 R.L.M. GN-FL1-4 Homestead, FL Pheromone Corn July 2002 R.L.M. HM-FL1 Homestead, FL Pheromone Corn Feb. 2004 R.L.M HM-FL2 Homestead, FL Pheromone Corn Oct.ÐNov. 2004 R.L.M HM-FL3-6 Homestead, FL Pheromone Corn Nov. 2005 R.L.M HM-FL7-10 Homestead, FL Larva Corn Jan 2005 R.L.M. HM-FL11 Ona, FL Pheromone Pasture AprilÐMay 2005 R.L.M. ON-FL1-4 Avon Park, FL Larva Corn Nov. 2003 Ña AP-FL1-2 Belle Glade, FL Larva Corn Nov. 2003 Ña BG-FL1-2 Okeechobee, FL Larva Sorghum (sorghum/ Oct.ÐNov. 2003 R.L.M. sudan hyb.) Weslaco, TX Pheromone Corn April 2004 B. WarÞeld WL-TX1-3 College Station, TX Pheromone Corn AprilÐOct. 2004 J.L. CS-TX1-13 College Station, TX Larva Corn Sept. 2004 J.L. CS-TX14-16 Palotina, Parana, Brazil Larva Corn Feb. 2005 P.S. PN-BR1-7 Palotina, Parana, Brazil Larva Corn Dec. 2005 P.S. Campo Verde, Mato Grosso, Brazil Larva Pasture Oct. 2005 P.S. MT-BR1-5 Campo Verde, Mato Grosso, Brazil Larva Poaceae Nov. 2005 P.S. MT-BR6-9 Campo Verde, Mato Grosso, Brazil Larva Millet Nov. 2005 P.S. MT-BR10-12 Campo Verde, Mato Grosso, Brazil Larva Corn Jan. 2005 P.S. MT-BR13-14 Campo Verde, Mato Grosso, Brazil Larva Sorghum (BF80) Mar.ÐApril 2005 P.S. MT-BR15-23 Primavera do Leste, Mato Grosso, Brazil Larva Sorghum (SARA) Oct.ÐNov. 2005 P.S. Primavera do Leste, Mato Grosso, Brazil Larva Amaranthus Nov. 2005 P.S. Primavera do Leste, Mato Grosso, Brazil Larva Millet Nov. 2005 P.S. MT-BR24-26 Primavera do Leste, Mato Grosso, Brazil Larva Cotton Dec. 2005 P.S. Primavera do Leste, Mato Grosso, Brazil Larva Corn Oct.ÐNov. 2005 P.S. Rio Grande do Sul, Brazil Larva Rice, corn Mar. 2005 V.M.

a Nagoshi and Meagher (2004).

United States found high genetic variability in lar- Materials and Methods vae collected from corn, with some samples showing Specimen Collections and Sites. Fall armyworm close genetic similarity to larvae collected from turf- specimens were obtained at several locations in the grass in the United States (Clark 2005). Allozyme southern United States (Table 1). Adult males were and mitochondrial haplotype analyses of larvae collected using pheromone traps as described previ- from French Guiana identiÞed the two strains, but ously (Meagher and Gallo-Meagher 2003). Standard there was no evidence of plant host speciÞcity as the plastic Unitraps were baited with a commercially R-strain predominated in both corn- and turf-dom- available fall armyworm pheromone (Scenturion Inc., inated habitats (Prowell et al. 2004). The seeming Clinton, WA), and contained insecticide strips (Her- inconsistencies in the results from the different con Environmental Co., Emigsville, PA). Collections groups, the wide geographical distribution of the from traps were made at various intervals, ranging tested populations, and the different methodologies from 1 to 14 d. After collection, specimens were stored used to identify strains make it difÞcult to assess at Ϫ20ЊC. Larvae were collected from host plants and whether the strains identiÞed in U.S. populations identiÞed by morphological criteria. These were then are genetically comparable with the biotypes de- preserved in 100% ethanol until DNA isolation or were scribed in South America. placed individually in 22.5-ml (0.75-oz.) plastic cups In this study, we used the mitochondrial cyto- with artiÞcial diet (Heliothis Premix, Stoneßy Indus- chrome oxidase I (COI) gene to identify and com- tries, Bryan, TX) to complete development. DNA was pare haplotypes present in Brazil with those in isolated from either adults or late (postfourth) instars. Texas and Florida. Fall armyworm in these two DNA Preparation. Individual specimens were ho- mogenized in 4 ml of phosphate-buffered saline (PBS; states are the primary source of the migratory pop- 20 mM sodium phosphate and 150 mM NaCl, pH 8.0) ulations that infest central and eastern North Amer- in a 15-ml test tube by using a tissue homogenizer ica (Sparks 1979), but relatively little is known (PRO ScientiÞc Inc., Oxford, CT). Cells and tissue about the haplotypes present in the Texas popula- were pelleted by centrifugation at 6000 ϫ g for 5 min tion. We further tested whether the haplotypes at room temperature. The pellet was resuspended in present in Brazil showed comparable biases in host 800 ␮l of cell lysis buffer (0.2 M sucrose, 0.1 M Tris- plant distribution to that observed for the two U.S. HCl, pH 8.0, 0.05 M EDTA, and 0.5% sodium dodecyl host strains. The implications of these results on the sulfate), transferred to a 1.5- or 2.0-ml microcentrifuge use of the strain-speciÞc haplotype markers to study tube, and incubated at 55ЊC for 5 min. Proteins were geographically separated fall armyworm popula- precipitated by the addition of 100 ␮l of 8 M potassium tions were discussed. acetate. The supernatant was transferred to a Zymo- 396 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 100, no. 3

Fig. 1. Diagram of the portion of the mitochondrial COI gene analyzed along with relevant primers. The putative translational start site of the COI gene was arbitrarily designated as coordinate 0. Vertical lines within the gene indicate sites of the strain-speciÞc nucleotide substitutions described in the text. Arrows indicate location and direction of primers used for the PCR ampliÞcation and DNA sequencing. Primers JM-76 and JM-77 were described previously (Levy et al. 2002).

Spin III column (Zymo Research, Orange, CA) and DNA Sequence Analysis. DNA sequencing was processed according to manufacturerÕs instructions. performed by Northwoods DNA, Inc. (Bemidji, The DNA preparation was increased to a Þnal volume MN). Templates were either the fragment-isolated of 40 ␮l with distilled water. Each polymerase chain PCR product (0.5 ␮g) or the fragment subcloned reaction (PCR) reaction required 1 ␮l of the DNA into pGEM-T-Easy (1.0 ␮g). Direct sequencing of preparation. the fragment isolations was initially performed us- PCR Analysis and Cloning. PCR ampliÞcation of the ing JM-76, COI-1058R, and COI-914R (5Ј-ACACCT ␮ mitochondrial COI gene was performed in a 30- l GTTAATCCTCCTACAG-3Ј) (Fig. 1). Sequencing reaction mix containing 3 ␮lof10ϫ reaction buffer, 1 Ј ␮ ␮ ␮ ␮ of the subclones used the SP6 Promoter (5 -GAT l of 10 mM dNTP, 0.5 lof20 M primer mix, 1 l TTAGGTGACACTATAG-3Ј), the T7 Promoter (5Ј- of DNA template (0.05Ð0.5 ␮g), and 0.5 ␮lofTaqDNA TAATACGACTCACTATAGGG-3Ј), and COI-914R. polymerase (New England Biolabs, Beverly, MA). The sequences obtained were then aligned, and a The thermocycling program was 94ЊC (1 min), fol- lowed by 33 cycles of 92ЊC (45 s), 56ЊC (45 s), 72ЊC(1 consensus sequence was derived for each strain hap- min), and a Þnal segment of 72ЊC for 3 min. The lotype. appropriate restriction enzyme was diluted in 1ϫ of To optimize the accuracy of the DNA sequences the the manufacturerÕs reaction buffer to a concentration following strategy was used. The sequence of each of1U/␮l, and 5 ␮l was added to 8 ␮l of each PCR sample was compared with the consensus sequences reaction. Digestions were at 37ЊC for 1Ð2 h. Two mi- of both the R-strain and C-strain COI regions. If vari- croliters of 6X gel loading buffer was added to each ations from the consensus were found, the region in sample, which was run on a 1.8% PCR grade agarose question was sequenced again using one or more of (Fisher, Hampton, NH) horizontal gel. Typically, 40 these primers as needed: COI-545 F, 5Ј-TTTGAGCT or 96 PCR ampliÞcations and restriction digests were GTAGGTATTACYG-3Ј; JM77, 5Ј-ATCACCTCCWC performed at the same time by using either 0.2-ml tube CTGCAGGATC-3Ј; and COI-914R, 5Ј-GCWGATG strips or 96-well microtiter plates. Primers were syn- TAAAATAWGCTCGWG-3Ј. If the variation from thesized by Integrated DNA Technologies (Coralville, consensus was conÞrmed, a second PCR reaction IA). AmpliÞcation of the COI region used the primer was performed from the sample genomic DNA, and pair JM76 (5Ј-GAGCTGAATTAGGRACTCCAGG-3Ј) the DNA sequence was analyzed. It is highly un- and COI-1058R (5Ј-ACACCTGTTAATCCTCCTACAG- likely that the same random ampliÞcation error will Ј 3 ) to produce a 1.0-kb fragment (Fig. 1). occur in two independent reactions. If discrepan- ␮ For fragment isolations 6 l of 6X gel loading cies between the two sequences were found, a third buffer was added to each ampliÞcation reaction, and PCR reaction was performed and compared. By this the entire sample was run on a 1.8% agarose hori- strategy, each variation from the consensus se- zontal gel containing GelRed (Biotium, Hayward, quence was conÞrmed by two independent PCR CA) in 0.5X Tris-borate buffer (45 mM Tris base, 45 reactions, and the DNA sequence was analyzed. All mM boric acid, and 1 mM EDTA, pH 8.0). Fragments were visualized on a long-wave UV light box and cut other DNA sequences were obtained from National out from the gel. Fragment isolation was performed Center for Biotechnology Information GenBank. using Zymo-Spin I columns (Zymo Research, Or- DNA comparisons, alignments, and restriction site ange, CA) according to manufacturerÕs instructions. mapping were performed using the DS Gene pro- When necessary, fragments were subcloned into gram (Accelrys, San Diego, CA). Phylogenetic trees pGEM-T-Easy vector (Promega, Madison, WI) ac- were generated using neighbor-joining analysis and cording to manufacturerÕs instructions. DNA was Tajima-Nei distance by using the DS Gene program isolated from subclones using the GenElute 5-min (Tajima and Nei 1984, Felsenstein 1985). Trees were Plasmid Miniprep kit (Sigma-Aldrich, St. Louis, unrooted, and nodes were supported by 10,000ϫ MO) according to manufacturerÕs instructions. bootstrap analysis. May 2007 NAGOSHI ET AL.: FALL ARMYWORM HOST STRAINS 397

Fig. 2. Strict consensus tree based on neighbor-joining anal- ysis of 73 COI sequences repre- senting fall armyworm popula- tions from Florida (FL), TX (TX), and Brazil (BR). The R- strain (SFU72977) and C-strain (SFU72974) sequences were obtained from GenBank and used to identify the host strain haplotypes. The COI sequences from Pseudaletia unipuncta (Ha- worth) were used as an outgroup. The sequences were obtained from GenBank (AF549698) and from a specimen isolated from the Þeld (our unpublished data). Numbers under branches denote bootstrap values (%) based on number per 10000 replicates. Only conÞdence values above 75% are indicated and nodes be- low 50% are collapsed. Scale bar equals 0.2 substitutions per site. 398 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 100, no. 3

Table 2. Distribution of polymorphisms at the strain-specific loci in the mitochondrial COI gene for fall armyworm strains collected from Florida, Texas, and Brazil

AciI ͓R͔ SacI ͓R͔ MspI ͓C͔ MspI ͓R͔ 171 207 489 BsmI͓C͔ 570 634 663 258 600 738 564 Ca TATTCCTCTTCTCCTATGA FL-R 12 0 12 0 12 0 12 0 12 0 12 0 12 0 12 0 12 0 12 0 FL-C 0 11 0 11 0 11 0 11 0 11 0 11 0 11 0 11 0 11 0 11 BR-R 14 0 14 0 14 0 14 0 14 0 14 0 14 0 14 0 14 0 14 0 BR-C 0 16 0 16 0 16 0 16 0 16 0 16 2 14 0 16 0 16 0 16 TX-R 9 0 9 0 9090909090909090 TX-C 0 9 0 8b 0909090909090909

759 768 795 798 816 823 912 987 1044 CTTCTCAGTCTCTCCTAT FL-R 12 0 12 0 12 0 12 0 12 0 12 0 12 0 12075 FL-C 0 11 0 11 0 11 0 11 0 11 0 11 0 11 0 11 0 11 BR-R 14 0 14 0 14 0 14 0 14 0 14 0 14 0 14 0 11 3 BR-C 0 16 0 16 0 16 0 16 0 16 0 16 0 16 0 16 1 15 TX-R 909090909090909054 TX-C 091809090909090909

a C, T, A, and G represent the cytosine, thymine, adenine, and guanine nucleotides, respectively. b In one sample, a C was present.

Results of the C-strain), and at site 768 one C-strain sample carried the R-strainÐassociated A instead of the ex- A 937-bp portion of the mitochondrial COI gene was pected T. The most polymorphic site was at coordi- isolated and sequenced from a total of 73 individuals nate 1044, which was speciÞc for the C-strain (with from Brazil, Florida, and Texas (Fig. 1). The samples were obtained by either pheromone trapping or larval one exceptional sample from Brazil) but variable in collection from a variety of plant types, including the R-strain for populations from all three locations. In maize, cotton, sugarcane, millet, sorghum, and pasture the latter case, the exceptions resulted from the sub- grasses (Table 1). To determine whether the se- stitution of the normally C-strainÐspeciÞc T instead of quences could be partitioned into two groups associ- the expected A. None of the base changes in these ated with the known host strains, neighbor-joining strain-speciÞc sites alter the conceptual COI amino reconstruction analysis was performed, and a phylo- acid sequence. genetic tree was constructed (Fig. 2). The sequences Additional polymorphisms were observed in 24 from all areas could be subdivided into two major other nucleotide loci that were not speciÞc to a strain. groups that corresponded with the consensus R-strain In combination with the variations in the strain-spe- and C-strain sequences (GenBank accession nos. ciÞc sites these identiÞed a total of 28 different hap- U72977 and U72974, respectively). Subsets from the lotypes, 14 each in the C-strain and R-strain subpopu- Texas and Brazil samples were present in each, indi- lations. We designated the RS1 and CS1 haplotypes as cating that the strain-speciÞc haplotypes character- the consensus sequences for the R-strain (Table 3) ized in Florida populations are also present in fall and C-strain (Table 4), respectively, and these dif- armyworm from these other locations. fered by no more than three nucleotides from the Sequence Characterization of the Strain-Specific other variants within each strain. Six of these poly- Sites. The strain-speciÞc subpopulations were deÞned morphisms alter the conceptual peptide sequence by by 19 polymorphic nucleotide sites (Table 2). Three a single base substitution. of the sites (their coordinates of 258, 564, and 600 RS1 was the most common R-strain haplotype over- indicate nucleotide distance from the presumptive all, representing 42% (15/36) of the R-strain popula- translational start site of the COI gene) were de- tion (Table 5). Similarly, CS1 was predominant in the scribed previously in sequence, and restriction en- C-strain, making up 51% (19/37) of this subgroup. zyme analyses of an internal 568-bp region was de- However, there was substantial variability in the dis- Þned by the JM76/JM77 primers (Levy et al. 2002, tribution of the haplotypes between geographical ar- Nagoshi et al. 2005). An additional strain-speciÞc MspI eas. For example, in the Florida samples the propor- site was observed at site 738 that is present in the tion of the RS3 haplotype approximated that of the R-strain but not the C-strain. In 16 of the 19 sites, there RS1, whereas RS3 was not found in Brazil or Texas. In was 100% strain speciÞcity for fall armyworm samples addition, although the C-strain samples from Brazil from all locations (Table 2). Two sites showed single consisted of nine different haplotypes, each present in exceptions; at site 207 one C-strain sample from Texas roughly equal proportions, CS1 was the predominant had a C instead of the expected T (found in the rest C-strain haplotype in Texas and Florida. The majority May 2007 NAGOSHI ET AL.: FALL ARMYWORM HOST STRAINS 399

Table 3. Haplotypes identified from fall armyworm collected from Brazil, Texas, and Florida for the R-strain

Haplotype 132 133 143 204 268 357 430 444 570 596 606 633 666 839 848 879 891 1044 RS1 A A A A G A A T T T T T A T T A T G RS2 A RS3 CA RS4 Ca RS5 G A RS6 Ga RS7 G RS8 G T A RS9 C A RS10 CA RS11 G RS12 G RS13 G C Ca RS14 Aa Ga

a Causes an amino acid substitution. of haplotypes of either strain, 22 of 28, were repre- between the two collection periods, with the C-strain sented by only a single specimen. percentage changing from 0 to 60 in a 2-wk period. Brazilian Strains Display Plant Host Preferences Similar to U.S. Populations. In the Rio Grande do Sul Discussion state, contemporaneous larval collections from corn and rice plants showed a deÞnitive strain-speciÞc host Fall armyworm populations in Brazil consist of the preference, with the C-strain predominating in corn same two mitochondrial lineages that deÞne the two and absent in rice (Fig. 3). The C-strain also predom- host strains in North American populations. The Bra- inated in corn from the Mato Grosso state throughout zilian lineages also displayed similar asymmetries in the year, making up Ͼ80% of the larvae collected as plant host use, with the C-strain predominating in corn well as in spring-grown cotton (in Brazil, spring is and cotton and the R-strain in millet and a pasture approximately from September to November). The grass. The speciÞcity of C-strain larvae to corn and R-strain was the predominant population found in R-strain to turf/pasture grasses was demonstrated pre- millet and in two common pasture plants, one of the viously for populations in the Caribbean, Florida, and Amaranthus genus, the other of the Poaceae family. A Louisiana (Pashley 1986, Meagher and Gallo-Meagher more variable pattern was observed in sorghum. Dur- 2003, Meagher and Nagoshi 2004, Prowell et al. 2004). ing two collection periods in March and April, almost That cotton is preferred by the C-strain was demon- 90% of the larvae were of the C-strain, consistent with strated for fall armyworms collected in Louisiana and the C-strain bias reported previously for Honduras the Mississippi delta (Pashley 1988a; our unpublished and Georgia (U.S.) sorghum (Pashley 1988b, Lu and data). These data indicate that the same two strains Adang 1996). However, as the year progressed, the exist in both Brazil and the United States and display proportion of C-strain larvae declined (Fig. 3). To similar behaviors with respect to host plant choice. investigate whether a similar pattern could be found A more complicated behavior is exhibited with re- in U.S. populations, we analyzed larvae collected from spect to sorghum, which was presumed to be pre- sorghum in the Okeechobee region of Florida in 2003 ferred by the C-strain based on studies of samples from (Fig. 4). Variability in strain proportions was observed Honduras and Mississippi (Pashley 1988b, Lu and

Table 4. Haplotypes identified from fall armyworm collected from Brazil, Texas, and Florida for the C-strain

Haplotype 173 207 600 666 711 717 768 788 851 858 925 974 1020 1044 CS1 A T T A T A T A T T A A A A CS2 G CS3 C CS4 G CS5 Ga CS6 G CS7 T CS8 C CS9 C CS10 G CS11 T CS12 GG CS13 G CS14 G

a Causes an amino acid substitution. 400 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 100, no. 3

Table 5. Distribution of haplotypes in different locations 92% (Fig. 3). Similar differences were observed in collections from sorghum in Florida, suggesting that Locality Haplotype n the strain preference for this plant type is signiÞcantly Florida Texas Brazil inßuenced by environmental factors (Fig. 4). The role RS-1 3 6 6 15 of seasonality was indicated by the observation that RS-2 0 3 3 6 the C-strain predominated (89%) in Brazil sorghum RS-3 4 0 0 4 RS-4 1 0 0 1 during the autumn (MarchÐJune), but it declined to RS-5 1 0 0 1 50% of the collections in the spring (SeptemberÐNo- RS-6 1 0 0 1 vember), with substantial variation in strain propor- RS-7 1 0 0 1 RS-8 1 0 0 1 tions between the three spring collections (Fig. 3). RS-9 0 1 0 1 Interestingly, we also observed seasonality in strain RS-10 0 0 1 1 use of corn in southern and central Florida (Nagoshi RS-11 0 0 1 1 RS-12 0 0 1 1 and Meagher 2004). Although the proportion of larvae RS-13 0 0 1 1 found in corn during the Florida spring season (Feb- RS-14 0 0 1 1 ruaryÐApril) was heavily biased to the C-strain, there Total 12 10 14 36 were periods in the fall (SeptemberÐNovember) CS1 8 9 2 19 when R-strain larvae were the majority collected (Fig. CS2 0 2 1 3 4). The reason for this seasonality is unknown, but it CS3 0 3 1 4 could reßect migration behaviors, seasonal differences CS4 1 0 0 1 CS5 1 0 0 1 in the availability of alternative plant hosts, or other CS6 1 0 0 1 season-speciÞc environmental factors that can over- CS7 0 1 0 1 rule strain-speciÞc physiological preferences. CS8 0 1 0 1 CS9 0 0 1 1 These observations demonstrate that conclusions of CS10 0 0 1 1 plant host preferences based on sampling over a lim- CS11 0 0 1 1 ited time can be misleading. This is particularly true CS12 0 0 1 1 for the R-strain, which seems to be able to use host CS13 0 0 1 1 CS14 0 0 1 1 plants normally associated with the C-strain, at least Total 11 16 10 37 during certain times of the year. This ßexibility is consistent with laboratory studies showing that al- though the two strains can differ in their use of, and Adang 1996). However, larvae collected from sor- development on, speciÞc plant hosts, they are both ghum in Brazil showed substantial variation in strain able to develop to adulthood on these hosts with proportions, with C-strain values ranging from 13 to relatively small differences in viability and no re-

Fig. 3. Strain proportions in larvae collected from different plant hosts in Brazil. Solid bar denotes R-strain, striped bar denotes C-strain. Numbers above bars indicate number of samples tested. All dates are from 2005. Note that seasons in Brazil are reversed from that observed in Florida such that spring occurs during SeptemberÐNovember and fall during MarchÐJune. RGdS, Rio Grande de Sol; P, Parana; MG, Mato Grosso; Amar., Amaranthus; and Poac., Poaceae. May 2007 NAGOSHI ET AL.: FALL ARMYWORM HOST STRAINS 401

Fig. 4. Strain proportions in larvae collected from corn and sorghum in Florida. Solid bar denotes R-strain, striped bar denotes C-strain. Numbers above bars indicate number of samples tested. The sorghum collections were from Okeechobee, FL. Larval collections from corn were from Belle Glade, FL. All samples were collected in 2003. ported differences in adult Þtness (Pashley 1988a, havior in such different environments should provide Veenstra et al. 1995). new insight into the biology underlying host plant Our analysis of the COI region showed Ͼ99% iden- choice and migration, and further characterization of tity between the Brazil, Texas, and Florida mitochon- genetic differences between strains and geographi- drial sequences for each strain haplotype. Further- cally distant populations will potentially identify new more, the strain speciÞcity of the 19 sites was markers for investigating these issues. conserved in all three populations, as was the high rate of polymorphism in site 1044 for the R-strain. There- fore, it seems that either the divergence of the two Acknowledgments mitochondrial lineages into the two strains occurred before the dissemination of fall armyworm into Texas, We thank Jane Sharp for excellent technical support. We Florida, and Brazil, where they are now reproduc- thank Robert Shatters (USDA-ARS) and Claudia Copeland (USDA-ARS) for helpful comments on the manuscript and tively isolated, or there is substantial gene ßow oc- William WarÞeld for samples from Weslaco, TX. The use of curring that has essentially homogenized these geo- trade, Þrm, or corporation names in this publication is for the graphically distant populations. Geographical isolation information and convenience of the reader. Such use does would be supported by the existence of genetic poly- not constitute an ofÞcial endorsement or approval by the morphisms that are speciÞc to the geographical areas. United States Department of Agriculture or the Agricultural Three candidate polymorphisms were observed. Site Research Service of any product or service to the exclusion 633 is present in one third of the Florida R-strain of others that may be suitable. tested, but it was not found in Brazil or Texas samples. Site 717 and site 666 were found in one third and two ninths of the Texas C-strain, respectively, but not in References Cited the Florida C-strain. More extensive surveys of these Adamczyk, J. J., Jr., J. W. Holloway, B. R. Leonard, and J. B. polymorphisms are being performed to more accu- Graves. 1997. Susceptibility of fall armyworm collected rately assess their geographical distribution. from different plant hosts to selected insecticides and The demonstration that the same strain-speciÞc mi- transgenic Bt cotton. J. Cotton Sci. 1: 21Ð28. tochondrial markers identify equivalent strains in both Busato, G. R., A. D. Gru¨ tzmacher, A. C. de Oliveira, E. A. Brazil and U.S. fall armyworm populations should fa- Vieira, P. D. Zimmer, M. M. Kopp, J. d. M. Bandeira, and cilitate research on this important agricultural pest in T. R. Magalha˜es. 2004. Analysis of the molecular struc- South America. Relatively simple and rapid methods ture and diversity of Spodoptera frugiperda (J.E. Smith) (Lepidoptera: Noctuidae) populations associated to the are available for characterizing these markers, allow- corn and rice crops in Rio Grande do Sul State, Brazil. ing for more extensive sampling of population distri- Neotrop. Entomol. 33: 709Ð716. butions, and the commonality in strains and mitochon- Busato, G. R., A. D. Gru¨ tzmacher, M. S. Garcia, F. P. Giolo, drial lineages legitimatize comparisons with North M. J. Zotti, and J. d. M. Bandeira. 2005a. Thermal re- American surveys. Comparisons of fall armyworm be- quirements and estimate of the number of generations of 402 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 100, no. 3

biotypes “corn” and “rice” of Spodoptera frugiperda. Pesq. in Florida using mitochondrial markers. Fla. Entomol. 86: Agropec. Bras. 40: 329Ð335. 450Ð455. Busato, G. R., A. D. Gru¨ tzmacher, M. S. Garcia, F. P. Giolo, Meagher, R. L., and R. N. Nagoshi. 2004. Population dynam- M. J. Zotti, and G. J. Stefanello, Jr. 2005b. Compared ics and occurrence of Spodoptera frugiperda host strains biology of Spodoptera frugiperda (J.E. Smith) (Lepidop- in southern Florida. Ecol. Entomol. 29: 614Ð621. tera: Noctuidae) populations in corn and rice leaves. Nagoshi, R. N., and R. L. Meagher. 2004. Seasonal distribu- Neotrop. Entomol. 34: 743Ð750. tion of fall armyworm (Lepidoptera: Noctuidae) host Busato, G. R., A. D. Gru¨ tzmacher, M. S. Garcia, M. J. Zotti, strains in agricultural and turf grass habitats. Environ. S. D. No¨rnberg, T. R. Magalha˜es, and J. d. B. Magalha˜es. Entomol. 33: 881Ð889. 2006. Susceptibility of caterpillars of the biotypes corn Nagoshi, R. N., R. L. Meagher, Jr., J. J. Adamczyk, Jr., S. K. and rice of Spodoptera frugiperda (J.E. Smith, 1797) (Lep- Braman, R. L. Brandenburg, and G. Nuessly. 2005. New idoptera: Noctuidae) to insecticides with different action restriction fragment length polymorphisms in the cyto- manners. Cienc. Rural 36: 15Ð20. chrome oxidase I gene facilitate host strain identiÞcation Clark, P. L. 2005. Population variation of Spodoptera frugi- of fall armyworm (Lepidoptera: Noctuidae) populations perda (J.E. Smith) in the Western Hemisphere. Ph.D. in the southeastern United States. J. Econ. Entomol. 99: dissertation, University of Nebraska, Lincoln, NE. 671Ð677. Felsenstein, J. 1985. ConÞdence limits on phylogenies: an Pashley, D. P. 1986. Host-associated genetic differentiation approach using bootstrap. Evolution 39: 783Ð791. in fall armyworm (Lepidoptera: Noctuidae): a sibling Foster, R. E. 1989. Strategies for protecting sweet corn ears species complex? Ann. Entomol. Soc. Am. 79: 898Ð904. from damage by fall armyworm (Lepidoptera: Noctu- Pashley, D. P., S. S. Quisenberry, and T. Jamjanya. 1987. idae) in southern Florida. Fla. Entomol. 72: 146Ð151. Impact of fall armyworm (Lepidoptera: Noctuidae) host Hall, D. G. 1988. Insects and mites associated with sugar- strains on the evaluation of bermudagrass resistance. J. cane in Florida. Fla. Entomol. 71: 138Ð150. Econ. Entomol. 80: 1127Ð1130. Jamjanya, T., S. S. Quisenberry, S. S. Croughan, and R. N. Pashley, D. P. 1988a. Quantitative genetics, development, Story. 1990. Comparison of bermudagrass lines grown in and physiological adaptation in host strains of fall army- different cultural conditions and the effect on screening worm. Evolution 42: 93Ð102. Pashley, D. P. 1988b. The current status of fall armyworm for all armyworm (Lepidoptera: Noctuidae) resistance. J. host strains. Fla. Entomol. 71: 227Ð234. Econ. Entomol. 83: 585Ð590. Pashley, D. P., and L. D. Ke. 1992. Sequence evolution in Leuck, D. B., C. M. Taliaferro, G. W. Burton, R. L. Burton, mitochondrial ribosomal and ND-1 genes in Lepidoptera: and M. C. Bowman. 1968. Resistance in bermudagrass to implications for phylogenetic analyses. Mol. Biol. Evol. 9: the fall armyworm. J. Econ. Entomol. 61: 1321Ð1322. 1061Ð1075. Levy, H. C., A. Garcia-Maruniak, and J. E. Maruniak. 2002. Pashley, D. P., T. N. Hardy, and A. M. Hammond. 1995. Host Strain identiÞcation of Spodoptera frugiperda (Lepidop- effects on developmental and reproductive traits in fall tera: Noctuidae) insects and cell line: PCR-RFLP of cy- armyworm strains (Lepidoptera: Noctuidae). Ann. En- tochrome oxidase subunit I gene. Fla. Entomol. 85: 186Ð tomol. Soc. Am. 88: 748Ð755. 190. Prowell, D. P., M. McMichael, and J.-F. Silvain. 2004. Mul- Lu, Y., M. J. Adang, D. J. Eisenhour, and G. D. Kochert. 1992. tilocus genetic analysis of host use, introgression, and Restriction fragment length polymorphism analysis of speciation in host strains of fall armyworm (Lepidoptera: genetic variation in North American populations of the Noctuidae). Ann. Entomol. Soc. Am. 97: 1034Ð1044. fall armyworm Spodoptera frugiperda (Lepidoptera: Noc- Quisenberry, S. S., and F. Whitford. 1988. Evaluation of tuidae). Mol. Ecol. 1: 199Ð208. bermudagrass resistance to fall armyworm (Lepidoptera: Lu, Y. J., and M. J. Adang. 1996. Distinguishing fall army Noctuidae): inßuence of host strain and dietary condi- worm (Lepidoptera: Noctuidae) strains using a diagnos- tioning. J. Econ. Entomol. 81: 1463Ð1468. tic mitochondrial DNA marker. Fla. Entomol. 79: 48Ð55. Sparks, A. N. 1979. A review of the biology of the fall ar- Lynch, R. E., W. G. Monson, B. R. Wiseman, and G. W. Burton. myworm. Fla. Entomol. 62: 82Ð86. 1983. Bermudagrass resistance to fall armyworm (Lepidop- Tajima, F., and M. Nei. 1984. Estimation of evolutionary tera: Noctuidae). Environ. Entomol. 12: 1837Ð1840. distance between nucleotide sequences. Mol. Biol. Evol. Martinelli, S., R. M. Barata, M. I. Zucchi, M.D.C. Silva-Filho, 1: 269Ð285. and C. Omoto. 2006. Molecular variability of Spodoptera Veenstra, K. H., D. P. Pashley, and J. A. Ottea. 1995. Host- frugiperda (Lepidoptera: Noctuidae) populations associ- plant adaptation in fall armyworm host strains: compar- ated to maize and cotton crops in Brazil. J. Econ. Entomol. ison of food consumption, utilization, and detoxication 99: 516Ð526. enzyme activities. Ann. Entomol. Soc. Am. 88: 80Ð91. McMichael, M., and D. P. Prowell. 1999. Differences in am- Whitford, F., S. S. Quisenberry, T. J. Riley, and J. W. Lee. pliÞed fragment-length polymorphisms in fall armyworm 1988. Oviposition preference, mating compatibility, and (Lepidoptera: Noctuidae) host strains. Ann. Entomol. development of two fall armyworm strains. Fla. Entomol. Soc. Am. 92: 175Ð181. 71: 234Ð243. Meagher, R. L., and M. Gallo-Meagher. 2003. Identifying host strains of fall armyworm (Lepidoptera: Noctuidae) Received 6 October 2006; accepted 30 November 2006.