IMB177.fm Page 179 Thursday, March 23, 2000 3:41 PM Insect Molecular Biology (2000) 9(2), 179–184 MitochondrialBlackwell Science, Ltd DNA sequence divergence among greenbug (Homoptera: Aphididae) biotypes: evidence for host-adapted races K. A. Shufran,1 J. D. Burd,1 J. A. Anstead2 and G. Lushai3 Introduction 1 USDA-ARS, Plant Science and Water Conservation Research Approximately 50% of the recognized insect biotypes on 2 Laboratory, Stillwater, OK, USA, Department of Entomology agricultural crops are aphids (Saxena & Barrion, 1987). One and Plant Pathology, Oklahoma State University, Stillwater, such aphid, the greenbug, Schizaphis graminum (Rondani) OK, USA, 3University of Southampton, School of Biological (Homoptera: Aphididae) consists of multiple biotypes that Sciences, Biodiversity and Ecology Division, Southampton, UK have hindered its management by host plant resistance (see Porter et al., 1997 for a comprehensive review). Since Abstract 1961, eleven greenbug biotypes have been described and given the letter designations A–K. All designations (except The full complement of known greenbug, Schizaphis D, which was erroneously based on insecticide resistance) graminum (Rondani), biotypes found in the USA were were based on plant responses (dead or alive) to greenbug subjected to a molecular phylogenetic analysis based feeding on an array of host plant differentials containing on a 1.2-kb portion of the cytochrome oxidase I mito- various resistance genes (Porter chondrial gene. In addition to these nine biotypes et al., 1997). The ability of (B, C, E, F, G, H, I, J and K), a probable isolate of the a greenbug biotype to kill a plant with a specific resistance enigmatic biotype A (NY), a ‘new biotype’ collected gene or genes is known as virulence. Although the occurrence of greenbug biotypes in the from Elymus canadensis (L.) (CWR), and an isolate USA is well documented, their origin and evolution are still from Germany (EUR) were included. Schizaphis controversial subjects (Porter rotundiventris (Signoret) was included as an outgroup. et al., 1997). Various hypo- Genetic distances among S. graminum biotypes theses have been proposed and include mutations (Starks ranged from 0.08% to 6.17% difference in nucleotide & Schuster, 1976), selection by resistant cultivars (e.g. substitutions. Neighbour-joining, maximum parsimony Eisenbach & Mittler, 1987), introductions (Blackman, 1981), and maximum likelihood analyses all produced sexual recombination (Puterka & Peters, 1989), and exploita- tion of crops by pre-existing biotypes (Porter dendrograms revealing three clades within S. graminum. et al., 1997). Clade 1 contained the ‘agricultural’ biotypes com- The utilization of modern, molecular genetic techniques monly found on sorghum and wheat (C, E, K, I, plus J) helped gain new insights into understanding this problem. and there were few substitutions among these biotypes. Three independent studies, focused on different regions Clade 2 contained F, G and NY, and Clade 3 contained of the genome of a few greenbug biotypes, yielded similar B, CWR and EUR, all of which are rarely found on results. Some biotypes were found to be distinct genetic crops. The rarest biotype, H, fell outside the above entities that probably diverged through years of reproductive isolation, predating the world history of wheat ( clades and may represent another Schizaphis species. Triticum S. graminum biotypes are a mixture of genotypes aestivum L.) cultivation (about 10 000 years ago). Using et al. (1989) belonging to three clades and may have diverged restriction enzyme patterns of mtDNA, Powers estimated a nucleotide sequence divergence of 1.2% and as host-adapted races on wild grasses. 1.0% between biotypes B vs. C, and B vs. E, respectively. Keywords: biotype, cytochrome oxidase I gene, host- Biotypes C and E were more closely related, with only adapted races, Schizaphis graminum, molecular about 0.17% sequence divergence. Based on these estim- phylogenetics. ates and the use of a molecular clock, they suggested biotypes B and C shared a common ancestral mitochondrial Received 4 October 1999; accepted 22 November 1999. Correspondence: genome approximately 0.3–0.6 million years ago during Dr Kevin A. Shufran, USDA-ARS, 1301 N. Western Rd, Stillwater, OK 74075, USA. Tel.: (405) 624–4141 × 240; fax: (405) 624–4142; e-mail: the Pleistocene era. [email protected] Black et al. (1992) demonstrated the usefulness of random © 2000 Blackwell Science Ltd 179 IMB177.fm Page 180 Thursday, March 23, 2000 3:41 PM 180 K. A. Shufran, et al. amplified polymorphic DNA (RAPD) to detect genetic poly- morphisms among greenbug biotypes. All biotypes tested (B, C, E, F, G, H and I) could easily be differentiated by RAPD profiles, except biotypes C and E which were virtually identical. Using Southern analysis, Black (1993) also found biotype-specific patterns in the repeat structure of the intergenic spacer (IGS) region of the ribosomal RNA cistron. Similarly with mtDNA, biotypes B, C, E, F, G and H all showed substantial divergence and were easily separable, except C and E. He concluded that each biotype began as a single clone or population, followed by a lengthy period of reproductive isolation, and that this hypothesis was supported by the results of Powers et al. (1989). The above studies concluded that biotypes C and E (holocyclic biotypes) were so closely related that they were indistinguishable. Biotype B (an anholocyclic biotype) was the most divergent and zero to little variation within biotypes was found (based on mtDNA, RAPD, or certain regions of rDNA). These techniques were useful in providing markers for population studies. The above studies also provided supporting evidence that the evolution of at least some Figure 1. Maximum likelihood tree of greenbug biotypes, isolates and S. greenbug biotypes could be attributed to host-race formation rotundiventris produced from nucleotide sequences from a 1.2-kb portion independent of or prior to human agricultural practices of the COI gene. For both distance/neighbour-joining and maximum parsimony analyses, 1000 bootstrap replications were performed. The percentage of (Porter et al., 1997). replications supporting each branch are shown. The top value represents While the above studies were informative about the neighbour-joining, while the bottom number represents maximum parsimony. relationships among some biotypes, none included the full complement of known biotypes today, nor could the genetic markers used be analysed in a phylogenetic context. Results and discussion Lacking from all was the enigmatic biotype A. Wood (1961) described the first greenbug biotype and gave it Silent and replacement substitutions were found between the designation B when he discovered a population that S. rotundiventris and S. graminum, as well as among the killed resistant DS-28 A wheat. Biotype A then referred to twelve S. graminum biotypes and isolates. Among the S. that portion of the population that was unable to injure graminum biotypes and isolates tested (S. rotundiventris DS-28 A and predated the discovery of biotype B. However, excluded), there were 123 variable sites; ninety-five were the three above molecular studies (Powers et al., 1989; silent substitutions and twenty-eight were replacement Black et al., 1992; Black, 1993), could not determine substitutions. Transition to transversion ratios ranged from whether biotypes arose independently of one another, or if 0 to 7.33. The majority (74%) of transitions were thymine– they were selected from or diverged from existing biotypes. cytosine substitutions, while the bulk of the transversions Since these publications, two additional biotypes have were adenine–thymine (59%) and adenine–cytosine (25%). been described, J (Beregovoy & Peters, 1995) and K The third codon position was biased towards adenine (Harvey et al., 1997). (38%) and thymine (46%). In the present study we examined mtDNA nucleotide Genetic divergence was detected among greenbug sequences and estimated the degree of genetic related- biotypes and isolates. 1000 bootstrap replications were ness among all known greenbug biotypes, including a performed in the maximum parsimony, maximum likelihood, probable sample of biotype A, a grass collected isolate and distance/neighbour-joining analyses. Cladograms with unique virulence characteristics, an isolate from Ger- produced by all three methods had identical topologies, many, and S. rotundiventris (Signoret) as an outgroup. We therefore we only present the maximum likelihood tree attempted to infer the phylogenetic relationships among (Fig. 1). Three distinct clades were identified, which we biotypes by constructing a phylogeny based on these herein refer to as 1, 2 and 3 for simplicity and to avoid DNA sequences. We discuss that phylogeny with respect confusion with the biotype letter designations. Clade 1 to feeding reactions on host–plant differentials, ecology, contained the predominant ‘agricultural’ biotypes infesting and reproductive strategy. This is the first comprehensive sorghum and wheat, i.e. C, E, I and K. Biotype J was also study, and the first to identify phylogenetic, evolutionary included, which was found on wheat in Idaho but is a non- relationships among all identified S. graminum biotypes. virulent biotype (Beregovoy & Peters, 1995). © 2000 Blackwell Science Ltd, Insect Molecular Biology, 9, 179–184 IMB177.fm Page 181 Thursday, March 23, 2000 3:41 PM Mitochondrial DNA divergence among greenbug biotypes 181 Table 1. Sequence analysis of a 1.2-kb