Inheritance of multiple herbicide resistance in wild oat (Avena fatua L.)

Transcription

1 Inheritance of multiple herbicide resistance in wild oat (Avena fatua L.) Jocelyn D. Karlowsky 1, Anita L. Brûlé-Babel 1, Lyle F. Friesen 1, Rene C. Van Acker 1, and Gary H. Crow 2 1 Department of Plant Science, University of Manitoba, Winnipeg, Manitoba, Canada R3T 2N2 ( Lyle_Friesen@umanitoba.ca); and 2 Department of Animal Science, University of Manitoba, Winnipeg, Manitoba, Canada R3T 2N2. Received 18 January 2005, accepted 3 August Karlowsky, J. D., Brûlé-Babel, A. L., Friesen, L. F., Van Acker, R. C. and Crow, G. H Inheritance of multiple herbicide resistance in wild oat (Avena fatua L.). Can. J. Plant Sci. 86: To gain some insight into the surprisingly frequent occurrence of multiple herbicide resistant wild oat in western Canada, the inheritance of multiple herbicide resistance was studied in two wild oat (Avena fatua L.) populations, UMWO12-01 and UMWO12-03, from Manitoba, Canada. Both populations are resistant to each of three distinct herbicides, imazamethabenz-methyl, flamprop-methyl, and fenoxaprop-p-ethyl (hereafter referred to as imazamethabenz, flamprop, and fenoxaprop-p, respectively). Crosses were made between each resistant (R) population and a susceptible (S) wild oat population (UM5) (R/S crosses), and between the two resistant populations (R/R crosses). Subsets of parental, F 2 plants, and F 2 -derived F 3 ( ) families were treated separately with each of the three herbicides and classified as R or S for individual plants, and homozygous R, segregating, or homozygous S for families. F 2 plants and families from R/S crosses segregated in 3R:1S and 1 homozygous R:2 segregating:1 homozygous S ratios, respectively. These ratios indicate that a single dominant or semi-dominant nuclear gene controls resistance to each of these herbicides in each population. F 2 plants and families from R/R crosses segregated for resistance/susceptibility when treated with either imazamethabenz or flamprop. Therefore, the genes for resistance to these two herbicides are different in each R population. Individual family response demonstrated that the genes were not independent of each other, indicating possible linkage between the genes for resistance to each herbicide. Genetic linkage could explain how the wild oat populations developed multiple resistance in the absence of selection by two of the herbicides, imazamethabenz and flamprop. Key words: Wild oat, Avena fatua, herbicide resistance, genetics of resistance, multiple resistance Karlowski, J. D., Brûle-Babel, A. L., Friessen, L. F., Van Acker, R. C. et Crow, G. H Hérédité de la résistance multiple aux herbicides chez la folle avoine (Avena fatua L.). Can. J. Plant Sci. 86: Les auteurs ont étudié la résistance multiple aux herbicides chez deux populations de folle avoine (Avena fatua L.) du Manitoba (Canada), UMWO12-01 et UMWO12-03, dans l espoir d éclaircir la surprenante fréquence de la résistance multiple aux herbicides chez cette espèce dans l Ouest canadien. Les deux populations en question résistent à trois herbicides : l imazaméthabenz-méthyle, le flamfrop-méthyle et le fénoxaprop-p-éthyle (ci-après appelés l imazaméthabenz, le flamfrop et le fénoxaprop-p, respectivement). Les auteurs ont croisé chaque population résistante (R) avec une population sensible (S) (UM5) (croisements R/S) et les deux populations résistantes entre elles (croisements R/R). Des sous-groupes composés de plants parentaux, de plants de la F 2 et de familles F 3 issues de la F 2 ( ) ont été traités séparément avec chacun des trois herbicides, puis chaque plant a été classé R ou S ou R homozygote, ségrégant ou S homozygote, pour les familles. Les ratios de ségrégation des plants de la F 2 et des familles des croisements R/S s établissaient respectivement à 3R:1S et à 1 homozygote R :2 ségrégant:1 homozygote S. Ces ratios indiquent qu un seul gène dominant ou mi-dominant dans le noyau commande la résistance à chaque herbicide au sein de chaque population. Le traitement des plants de la F 2 et des familles des croisements R/R à l imazaméthabenz ou au flamfrop entraîne la ségrégation pour la résistance/sensibilité. On en conclut que les gènes codant la résistance à ces deux herbicides diffèrent dans chaque population R. La réaction distincte des familles révèle que les gènes ne sont pas indépendants, signe qu il pourrait exister un lien entre ceux responsables de la résistance à chaque herbicide. De tels liens expliqueraient comment les populations de folle avoine ont acquis une résistance multiple sans avoir été sélectionnées par les herbicides imazaméthabenz et flamfrop. Mots clés: Folle avoine, Avena fatua, résistance aux herbicides, génétique de la résistance, résistance multiple The evolution of weeds with multiple herbicide resistance is a concern on the Canadian prairies, particularly the frequent occurrence of multiple herbicide resistant wild oat (Avena fatua L.). In 1997, 27% of cereal fields surveyed in Manitoba had populations of wild oat that exhibited resistance to more than one herbicide mode of action (Beckie et al. 1999). Four of these wild oat populations were resistant to ACCase inhibitors (Group 1), ALS inhibitors (Group 2), triallate (Group 8), difenzoquat (Group 26), and flampropmethyl (flamprop, Group 25). No herbicides are registered 317 for use in western Canada to selectively control these wild oat populations resistant to five different modes of action in wheat, leaving producers with fewer cropping/weed control options for infested fields. Multiple herbicide resistance also was identified in populations of green foxtail [Setaria viridis (L.) Beauv.] in Manitoba and Saskatchewan (Morrison and Abbreviations: Fenoxaprop-P, fenoxaprop-p-ethyl; flamprop, flamprop-methyl; imazamethabenz, imazamethabenz-methyl; R, resistant; S, susceptible

2 318 CANADIAN JOURNAL OF PLANT SCIENCE Devine 1994), false cleavers (Galium spurium L.) in Alberta (Hall et al. 1998), and populations of pigweed (Amaranthus spp.) in Ontario (Ferguson et al. 2000; Diebold et al. 2003). Multiple herbicide resistance can occur by either enhanced metabolism of more than one herbicide mode of action (Preston et al. 1996) or by multiple mechanisms of resistance in an individual plant. Accumulation of resistance mechanisms explained the development of populations of green foxtail resistant to Group 1 and dinitroaniline (Group 3) herbicides (Heap and Morrison 1996). These resistant green foxtail biotypes were initially selected with dinitroaniline herbicides, and after resistance had developed, frequent application of Group 1 herbicides selected for a second mechanism of resistance. In wild oat, frequent use of Group 1 herbicides appears to precede the development of multiple herbicide resistance (Beckie et al. 1999). Wild oat has relatively low levels of outcrossing in cropping situations (Murray et al. 2002); therefore, there is a low probability that multiple herbicide resistance in wild oat would develop through accumulation of resistance alleles/mechanisms due to pollen flow. This type of evolution of multiple resistance is more probable in highly outcrossing species. The relatively high number of populations of wild oat that have exhibited resistance to multiple herbicide modes of action in the absence of specific selection (i.e., no use of a specific herbicide) suggests that multiple resistance due to enhanced metabolism controlled by a single gene is more likely than accumulation of multiple resistance genes (Somody et al. 1984; Beckie et al. 1999; Friesen et al. 2000). The probability of multiple mechanisms of resistance in an unselected population is the product of the probabilities of natural resistance (resulting from mutations) to each herbicide and therefore should be rare in the absence of selection (Wrubel and Gressel 1994). Many factors affect the development of herbicide resistance in weed species. These factors include the number of alleles involved in the expression of functional resistance, the frequency of resistance alleles in unselected populations, the reproduction and breeding characteristics of the species, the longevity of seed in the soil, the intensity of selection that differentiates resistant individuals from susceptible, the relative fitness of resistant and susceptible genotypes, and the mode of inheritance of the resistance allele(s) (Cousens and Mortimer 1995). The genetics of single mode of action herbicide resistance has been studied in a number of weed species. The majority of these studies have indicated that resistance is due to one dominant or partially dominant nuclear encoded gene (Jasieniuk et al. 1996). The inheritance of multiple herbicide resistance is not as well understood. To date, there have been few genetic studies involving weeds that exhibit resistance to more than one herbicide mode of action. Letouzé and Gasquez (2001) reported that both fenoxaprop-p-ethyl (fenoxaprop-p, Group 1) and flupyrsulfuron (Group 2) resistance in a population of blackgrass (Alopecurus myosuroides Huds.) from France is controlled by a single nuclear dominant gene, possibly conferring enhanced metabolism of both herbicides. A second mechanism of fenoxaprop-p resistance in this blackgrass population was identified as an ACCase mutation controlled by a different nuclear dominant gene. Letouzé and Gasquez (2001) concluded that there are two distinct, nuclear dominant genes controlling the two mechanisms of resistance in this blackgrass population. Interestingly, the population was never exposed to flupyrsulfuron; therefore, the intensive use of fenoxaprop-p for 6 consecutive years also selected for metabolism-based resistance to flupyrsulfuron (which has a completely different target site from fenoxaprop-p). In a multiple herbicide resistant rigid ryegrass (Lolium rigidum Gaud.) population from Australia, Preston (2003) concluded that resistance is controlled by at least five distinct genes. This population of rigid ryegrass is resistant to at least nine chemical groups, representing six different herbicide modes of action. The broad spectrum resistance in this rigid ryegrass population is conferred by both target site mutations and enhanced herbicide metabolism mechanisms. Both Letouzé and Gasquez (2001) and Preston (2003) concluded that the outcrossing nature of these weed species aids in the accumulation of alleles that control the eventual expression of multiple herbicide resistance in individual plants. Van Eerd et al. (2004) investigated the inheritance of multiple herbicide resistance in a biotype of false cleavers originating from Alberta (Hall et al. 1998) and reported that resistance to quinclorac and ALS-inhibitor herbicides is conferred by two distinct genes that are not tightly linked. At field dosages, the inheritance of resistance to quinclorac is mediated by a recessive allele at a single nuclear gene, while the inheritance of ALS-inhibitor resistance is controlled by a dominant allele at a different single nuclear locus. The false cleavers population had not been exposed to selection specifically by quinclorac, but Van Eerd et al. (2004) hypothesized that previous use of other auxinic herbicides may have selected for the resistant biotype (the population had been exposed to selection by ALS inhibitors). False cleavers is a predominantly self-pollinated broadleaved weed (Malik and Vanden Born 1988). Kern et al. (2002) reported that triallate resistance in triallate/difenzoquat herbicide resistant wild oat was controlled by two recessive nuclear genes. However, they provided no information regarding the inheritance of difenzoquat resistance or possible cosegregation with triallate resistance. Elucidating the genetic components of resistance remains important after resistance has developed as this information may provide some insight into the mechanism of resistance, and may assist in developing strategies for long-term control and/or to delay the evolution of herbicide resistant weed populations (Cousens and Mortimer 1995; Christoffers 1999). The two multiple herbicide resistant wild oat populations used in this study originated from the northwestern agricultural region (Swan River region) of Manitoba, Canada (Friesen et al. 2000). The populations were first identified when producers reported unsatisfactory control of wild oat in spring wheat crops sprayed with imazamethabenz-methyl (imazamethabenz). Characterization of resistance indicated that the populations were resistant to imazamethabenz, flamprop, and fenoxaprop-p. Imazamethabenz had not been previously used on either population. The objectives of this study were to: (1) determine the mode of inheritance of

3 KARLOWSKY ET AL. INHERITANCE OF MULTIPLE RESISTANCE IN WILD OAT 319 resistance to imazamethabenz, flamprop, and fenoxaprop-p in the multiple herbicide resistant wild oat populations, and (2) determine whether the gene(s) conferring resistance in the two populations occur at the same locus (loci). MATERIALS AND METHODS Plant Material Producers collected seed of the multiple herbicide resistant wild oat populations UMWO12-01 and UMWO12-03 from wild oat plants/patches in their fields in mid-august 1994 and 1995, respectively (Friesen et al. 2000). Plants grown from each population were screened with 100 g a.i. ha 1 of fenoxaprop-p applied at the three-leaf stage. For each population, seed was harvested from 20 surviving plants and pooled to form a base population. The susceptible biotype (wild type) used in this study was UM5. It was collected from Portage la Prairie, Manitoba and has been used and characterized in numerous studies (Devine et al. 1993; Heap et al. 1993; Murray et al. 1995; Murray et al. 1996). Growing Conditions All wild oat seed used in this study were dehulled by hand in order to stimulate germination. Seed was germinated in 9-cmdiameter plastic petri dishes lined with Whatman No. 1 filter paper (Whatman No. 1, Whatman Int. Ltd. Maidstone, UK). Dishes were moistened with a 0.1% wt/vol KNO 3 solution and transferred to a refrigerator at 4 C in the dark for 48 h. Petri dishes were then moved to room temperature (20 to 25 C) in the dark for a further 48 h prior to planting the germinated seed. To ensure more uniform germination, seed was pierced with a dissecting needle after an initial 24-h period at 4 C in the dark. Seed was pierced on the dorsal side near the embryo and kept at 4 C in the dark for an additional 48 h, followed by 48 h at room temperature in the dark prior to planting. Only obviously germinated wild oat seeds with a visible radicle and shoot were planted in the pots or trays. Germinated wild oat seed was planted in a potting mixture of clay loam/sand/peat moss in a 2:1:1 ratio by volume. Plants used to make initial crosses and produce subsequent generations were grown in 15-cm-diameter pots. Seedlings that were screened with herbicides were grown in horticultural trays (53 cm 26 cm 6 cm) divided into 48 cells (one seedling per cell) containing the same soil mixture. All plants were grown in growth rooms or cabinets with a 16-h photoperiod and approximate irradiance of 200 µmol m 2 s 1 photosynthetic photon flux for plants grown to maturity, and 600 µmol m 2 s 1 for seedlings sprayed with herbicide. The temperature regime was set at 21/15 C day/night. Plants were watered as required and fertilized weekly with a dilute solution of water soluble fertilizer (Peters Professional Water Soluble Fertilizer with chelated micronutrients, W. R. Grace & Co., P.O. Box 238, Fogelsville, PA 18051) at a rate of approximately 2.4 g L 1 product (approximately equivalent to 40 kg ha 1 N). Population Development Prior to F 1 production, UMWO12-01 and UMWO12-03 wild oat parental base populations were screened with imazamethabenz at 350 g a.i. ha 1 to ensure that each parent plant was resistant. Furthermore, to determine if each parent was homozygous for resistance, florets from each surviving plant were self-pollinated by enclosing portions of individual panicles in glassine crossing bags. Depending on seed availability, up to 20 progeny produced by selfing resistant parents were tested for segregation of resistance to each of the herbicides, imazamethabenz, flamprop, and fenoxaprop- P. Herbicides were applied separately (i.e., not tank mixed) to all populations tested such that individual seedlings received only one herbicide application. This was done because prior observations (data not shown) indicated that tank mixing these herbicides does not result in normal herbicidal activity. Only seed from crosses made with parental plants that were homozygous for resistance to each of the three herbicides was retained and used in the study. In an attempt to maximize wild oat seed production, two crossing methods were used. The first method involved hand pollination of emasculated panicles (Brown 1980). Wild oat anthers have been reported to dehisce in the afternoon (Raju et al. 1985); therefore, growth room lights were switched on and the daytime temperature cycle began at 0200 so that hand pollination could begin at about Anthers began to dehisce approximately 6 h after the lights in the growth room switched on and continued to dehisce for about 3 h. The second crossing method used was the approach method described by McDaniel et al. (1967). The approach method involved placing a common perforated plastic bag (very small perforations) at an appropriate growth stage over an emasculated female parent panicle and a male (pollen donor) panicle. Panicles were still attached to their respective plants so that growth and development would continue. The bags were shaken daily to maximize pollen movement within the crossing bag. After 1 wk, the male panicles were removed so that seed developing on the paternal plant would not be mixed with seed from the maternal plant. Forty-two parental pairs were crossed using the hand pollination method and 19 pairs were crossed using the approach method. During all stages of seed production panicles of interest were covered with glassine or perforated plastic bags to prevent contamination by stray pollen. Reciprocal crosses were made between each of the resistant biotypes and the susceptible biotype (UMWO12-01/UM5, UM5/UMWO12-01, UMWO12-03/UM5 and UM5/UMWO12-03), as well as between the resistant biotypes (UMWO12-01/UMWO12-03 and UMWO12-03/UMWO12-01) (female parent listed first for each cross). Crosses were made with a minimum of 10 parental plant pairs for each combination. Four F 1 hybrids from each reciprocal cross were allowed to self-pollinate to produce the F 2 generation. However, only three F 2 families from each cross were screened with the herbicides due to time and growth room space limitations. These families were chosen based upon results obtained from screening selfed parental seed for homozygosity of the resistance trait. Due to the same limitations, only one F 1 family from each reciprocal cross was advanced to the generation, chosen again by results from screen-

4 320 CANADIAN JOURNAL OF PLANT SCIENCE ing parental seed. Approximately 80 F 2 plants from each F 1 family were grown out and self-pollinated to produce the generation. Seed from individual plants was harvested and stored separately at each stage of population development to maintain distinct pedigrees for each plant. Family sizes were determined based on a formula described by Jasieniuk et al. (1996). All parental plant pairs for each cross were recorded. The parental pairs were designated by a single letter. Each F 1 plant derived from crossing a designated parental pair was assigned a unique number. All pedigrees were maintained through the generations such that a specific family could be traced back to a specific F 2 plant, F 1 plant, and parental pair. Herbicide Screening Techniques For imazamethabenz and fenoxaprop-p, wild oat seedlings were sprayed at the two- to three-leaf stage, approximately 14 d after planting. For flamprop, seedlings were sprayed at the three- to four-leaf stage, approximately 21 d after planting. These growth stages follow commercial label recommendations. Herbicide dosages used in this study were based on a previous study using these multiple herbicide resistant wild oat populations (Friesen et al. 2000). These herbicide dosages provided a clear distinction between resistant (R) and susceptible (S) plants. Dosages of imazamethabenz, flamprop, and fenoxaprop-p used were 350 g a.i. ha 1 plus 0.25% vol/vol nonionic surfactant, 300 g a.i. ha 1, and 90 g a.i. ha 1, respectively. Recommended field dosages of imazamethabenz, flamprop, and fenoxaprop-p for wild oat control in western Canada are 400 g a.i. ha 1, 260 g a.i. ha 1, and 92 g a.i. ha 1, respectively. Commercial formulations of the herbicides were used throughout the study. Resistant and susceptible wild oat controls were planted in each tray to assist in assessing plant response. A research cabinet sprayer equipped with a TeeJet flat-fan nozzle (TeeJet SS80015 flat fan nozzle tip, Spraying Systems Co., Wheaton, IL 60188) calibrated to deliver 117 L ha 1 of spray solution at 310 kpa in a single pass was used to apply the herbicides. Forty F 2 seedlings from each cross and 15 to 20 seedlings from each F 2 plant were treated with each of the herbicides and segregation patterns were determined. A total of 5,655 F 2 seedlings and 686 families each consisting of 15 to 20 individuals were screened. Wild oat plants were visually assessed as R or S. Plants showing signs of recovery and new growth after herbicide treatment were considered R, while plants exhibiting no new growth and complete tissue death were considered S. Plants were assessed two, three, and 4 wk after treatment. Data presented are based on the assessment 4 wk after treatment. To account for some individual plant variability in response to herbicide treatment, families were classified as homozygous resistant when one or no individuals died, homozygous susceptible when two or fewer individuals survived, and all other families were considered to be segregating for the resistance trait. These criteria were set based on the formula outlined by Strickberger (1985), which was used to calculate the expected frequency of a specific (relatively rare) F 3 family structure derived from a heterozygous F 2 plant. The formula is: [n!/(x!y!)] [p x q y ] where n is the family size, x is the number of surviving individuals (homozygous dominant or heterozygous), y is the number of dead individuals (homozygous recessive), p is the expected frequency of the resistant phenotype (homozygous dominant + heterozygotes) = 0.75, and q is the expected frequency of the susceptible phenotype (homozygous recessive) = Using this formula given a heterozygous F 2 plant, a family size of 20 individuals, and assuming resistance is controlled by one dominant allele, the expected frequency of a 19R:1S family is and a 20R:0S family is Similarly, the expected frequencies of 2R:18S, 1R:19S, and 0R:20S families are , , and , respectively. Twenty germinated seeds were planted for each family; however, in some cases not all seedlings emerged. Only those families where at least 15 individuals emerged and were sprayed were used in determining segregation ratios. A worst case scenario can be calculated; that is, the expected frequency of a 14R:1S family derived from a heterozygous F 2 plant is The majority of families screened in this study had more than 15 individuals. Thus, using these criteria and summing the relevant expected frequencies, the possibility of misclassifying a genotypically segregating family as a homozygous resistant or homozygous susceptible family was relatively low. The criteria outlined above were used for all three herbicides. Statistical Analysis Chi-square tests were used to determine the goodness of fit of the segregation ratios to known genetic models (Strickberger 1985). Data were pooled where possible and appropriate based on homogeneity chi-square tests. The Yates correction factor was used where appropriate for all analyses. Linkage between resistance traits to the three different herbicides was determined based on methods described in Strickberger (1985). RESULTS AND DISCUSSION Parental Screening Forty-two parental pairs were crossed using the hand pollination method and 19 parental pairs were crossed using the approach method, with a total of 369 and 58 F 1 seeds produced, respectively. The average number of F 1 seeds produced per panicle using the hand pollination technique was 8.8, while an average of only 3.1 seeds were produced per panicle using the approach method. To confirm homozygosity for resistance, seed produced from self-pollinating a portion of the florets on each resistant parental plant was screened with each of the three herbicides. The number of seedlings screened with each herbicide was determined based on the amount of selfed seed produced by each parent, with 20 seedlings being screened per parent per herbicide for the majority of the crosses. For material advanced to the F 2 generation, the minimum number of selfed seedlings screened per parent per herbicide was five in several instances (due to poor seed set

5 KARLOWSKY ET AL. INHERITANCE OF MULTIPLE RESISTANCE IN WILD OAT 321 on several parent plants). For material advanced to the generation, the minimum number of selfed seedlings screened per parent per herbicide was eight. Based on the results of the parental screening, 14 F 1 populations were eliminated due to a lack of homozygosity for flamprop resistance [data tabulated in Karlowsky (2004)]. All parents were homozygous for imazamethabenz resistance. The results for screening the parental seed with imazamethabenz are not surprising, as the parents used to make the crosses had all survived treatment with imazamethabenz. Results from screening parental seed with fenoxaprop-p indicated that none of the UMWO12-01 parents were homozygous for resistance. As a result no crosses involving UMWO12-01 were characterized for inheritance of the fenoxaprop-p resistance trait. To allow the correct interpretation of segregation ratios in F 2 and generations, parental homozygosity is necessary and crucial. The fact that crosses were eliminated due to results obtained by this step demonstrates the importance of testing parental plants for homozygosity. However, the genetic base of these populations would have been narrowed somewhat by this step. The fact that not all parental plants were homozygous resistant to all three herbicides provided definitive evidence that resistance to the three herbicides in an individual plant is not conferred by the same gene. Imazamethabenz Results R/S Crosses F 2 Results. Data were pooled within families where acceptable, based on chi-square homogeneity values. When data could not be pooled, results derived from an individual F 1 plant are reported. For the UMWO12-01/UM5 crosses, the F 2 families from cross A fit a 3R:1S segregation ratio (Table 1). Three of the four F 2 families from cross B fit a 3R:1S ratio. Although pooled results from cross E did not fit a 3R:1S ratio, three of the four families fit a 3R:1S ratio when tested individually (data not shown). For the reciprocal UM5/UMWO12-01 crosses, only F 2 families from cross H were homogeneous and fit a 3R:1S ratio. F 2 families from crosses F and G were not homogeneous, but six of eight families fit a 3R:1S ratio. For the two families (F-4 and G- 3) that did not fit the 3R:1S ratio, probabilities were close to the acceptable 0.05 level. A 3R:1S segregation pattern is expected when one dominant gene is responsible for resistance. Lack of reciprocal cross differences indicated that inheritance of resistance is nuclear, and not cytoplasmic. In some crosses involving population UMWO12-01, the segregation ratio observed was reversed (i.e., 1R:3S) (Table 1). Crosses C and D were homogeneous and segregated in a 1R:3S ratio, as did F 2 family G-4. F 2 family B-2 did not fit a 1R:3S ratio, but more S than R individuals were observed. This reversal of dominance is believed to be the result of the herbicide dosage (activity) used to screen the populations (Seefeldt et al. 1998). When herbicide dosages/activity are near the threshold level, resistant heterozygous individuals may display a resistant phenotype the majority of the time, but may appear to be susceptible under some conditions. It may also be possible that the resistance trait does not exhibit complete dominance, as is the case for ALS inhibitor resistance in other weed species [reviewed in Tranel and Wright (2002)]. Differences in the phenotypic responses of the heterozygotes may also involve an interaction of genotype, herbicide dosage, and growth cabinet environment. While all growth cabinets were set to the same specifications, some variation in soil moisture status occurred since plants were watered as required based on visual determination. All F 2 families from crosses C and D were screened in the same growth cabinet in the same run, and six of these seven families segregated in a 1R:3S ratio. The particular environmental conditions during this specific run and their influence on herbicide activity may account for the reversal of dominance. Results from crosses involving UMWO12-03 also segregated in a 3R:1S ratio for the majority of crosses (Table 1). All F 2 families for crosses K and N were homogeneous and fit a 3R:1S ratio. F 2 families from crosses I, J, and M were homogeneous, but did not fit a 3R:1S ratio at the 0.05 probability level. Individually, the F 2 families from cross M fit a 3R:1S ratio (data not shown). F 2 families from cross L were not homogeneous, but individually all 4 families fit a 3R:1S ratio. The 3R:1S ratio observed in the majority of crosses indicates that one dominant gene is responsible for resistance to imazamethabenz in population UMWO As with UMWO12-01, the resistance gene appears to be nuclear in origin rather than cytoplasmic. Results. To confirm F 2 results, families were screened with imazamethabenz. families developed from UMWO12-01/UM5 and UM5/UMWO12-01 crosses and from UMWO12-03/UM5 and UM5/UMWO12-03 crosses segregated in a 1 homozygous resistant : 2 segregating: 1 homozygous susceptible ratio (Table 2). This confirmed the F 2 results and is in agreement with results from other studies (Tranel and Wright 2002), indicating that one dominant nuclear gene controls imazamethabenz resistance in wild oat populations UMWO12-01 and UMWO R/R Crosses F 2 Results. Crosses were made between resistant populations to determine if resistance to imazamethabenz was due to the same gene mutation. F 2 seedlings segregated into R and S individuals (Table 1). If the same gene were responsible for resistance in both populations then all F 2 seedlings should have been resistant. Segregation in the F 2 generation indicated that different genes are involved in the expression of imazamethabenz resistance in each population. The 15R:1S segregation ratio observed in the F 2 for some of the crosses (Q-2, S-2, and T-2) indicated that a different dominant gene controls resistance in each population. In this situation the presence of at least one dominant allele in an individual confers resistance. However, for the majority of cases the results did not conform to a 15:1 ratio. Some of the F 2 families fit a 9R:7S ratio (O, Q-1, R-1, S-1, T-1, and T- 4) and others fit a 11R:5S ratio (R-2, S-3 and S-4, and T-3). Both of these ratios can be explained individually involving one separate dominant nuclear gene in each population. For the families segregating in a 9R:7S ratio, a single copy of a dominant resistance allele from each population must be

6 322 CANADIAN JOURNAL OF PLANT SCIENCE Table 1. Segregation for imazamethabenz resistance in the F 2 generation of resistant/susceptible (R/S), susceptible/resistant (S/R), and resistant/resistant (R/R) wild oat crosses Resistant Susceptible Cross z number of plants Ratio χ 2 Probability y UMWO12-01/UM5 (R/S) A : B : B : B : B : C + D : E : UM5/UMWO12-01 (S/R) F : F : F : F : G : G : G : G : H : UMWO12-03/UM5 (R/S) I : J : K : UM5/UMWO12-03 (S/R) L : L : L : L : M : N : UMWO12-01/UMWO12-03 (R/R) O : P : Q : Q : UMWO12-03/UMWO12-01 (R/R) R : R : S : S : S : S : T : T : T : T : z Crosses not followed by a number designation indicate pooled results from three or four different F 1 plants. Data were pooled based on chi-square homogeneity tests. Crosses where a number designation follows the cross indicates that results are from an individual F 1 plant. Probability values of 0.05 or greater indicate that the data does not differ significantly from the test ratio. Table 2. Segregation for imazamethabenz resistance in the generation of resistant/susceptible (R/S) and susceptible/resistant (S/R) wild oat crosses Resistant Segregating Susceptible Cross z number of families Ratio χ 2 Probability y UMWO12-01/UM5 (R/S) A + B :2: UM5/UMWO12-01 (S/R) F :2: UMWO12-03/UM5 (R/S) K :2: UM5/UMWO12-03 (S/R) M :2: z Data for all crosses in this table are pooled results from three or four different F 1 plants. Data were pooled based on chi-square homogeneity tests. Probability values of 0.05 or greater indicate that the data does not differ significantly from the test ratio.

7 KARLOWSKY ET AL. INHERITANCE OF MULTIPLE RESISTANCE IN WILD OAT 323 present for resistance to be expressed. For the families segregating in a 11R:5S ratio, more than one copy of a resistance allele from either population must be present for resistance to be expressed. Results. families from R/R crosses were screened with imazamethabenz to clarify F 2 results. Two genetic ratios were tested; 1 homozygous resistant: 8 segregating: 7 homozygous susceptible, which would confirm a 9R:7S ratio observed in the F 2 generation, and 7 homozygous resistant: 8 segregating: 1 homozygous susceptible ratio, which would confirm either a 15R:1S or an 11R:5S ratio in the F 2 generation. The pooled families did not fit either ratio (data not shown). The lack of fit may have been due to the family size used in screening. A family size of 20 individuals is considered to be adequate to confirm the segregation pattern of an F 2 population when only one gene is segregating, but is not sufficient to accurately separate homozygous resistant and segregating families when two or more genes are segregating. As an alternative to the 1:8:7 and 7:8:1 ratios, chi-square values were determined for 9:7 and 15:1 ratios, where homozygous dominant and segregating families were classified as the same phenotype (non-susceptible) (Table 3). With this re-classification, both crosses fit a 15:1 ratio. Based on F 2 and results, resistance to imazamethabenz is due to different resistance genes in each wild oat population, however the interaction between these genes and the resistant phenotype was not clearly determined. Flamprop Results R/S Crosses F 2 Results. For the crosses involving UMWO12-01, the F 2 families from C and D and the reciprocal cross F fit a 3R:1S ratio (Table 4). Crosses G and H were not homogeneous, however 5 of 7 F 2 families fit a 3R:1S ratio. Similar to imazamethabenz results, F 2 screening results for flamprop suggest that one dominant nuclear gene controls resistance in population UMWO Cross B segregated in the opposite direction and fit a 1R:3S ratio (Table 4). Two other families (A and E) had a higher proportion of susceptible than resistant individuals but did not fit a 1R:3S ratio. Similar to imazamethabenz results, this apparent reversal of dominance probably was related to the herbicide dosage (activity) used for screening. With reversal of dominance, heterozygotes exhibit the susceptible phenotype. Partial dominance of the resistance trait may also explain the occasional reversal of dominance. Crosses A and E were screened in the same run in the same growth cabinet. Specific conditions in this run may have influenced flamprop activity such that a reversal of dominance was observed. UMWO12-03/UM5 crosses were all homogeneous and segregated in a 3R:1S ratio (Table 4). For the reciprocal crosses, only M was homogeneous but did not fit a 3R:1S ratio. However, when considered individually, two of the four families from this cross fit the expected ratio (data not shown). Of the seven F 2 families that were not homogeneous, five fit a 3R:1S ratio. These results suggest that one Table 3. Segregation for imazamethabenz resistance in the generation of resistant/resistant (R/R) wild oat crosses Nonsusceptible y Susceptible Cross z number of families Ratio χ 2 Probability x UMWO12-01/UMWO12-03 (R/R) P : UMWO12-03/UMWO12-01 (R/R) T : z Data for all crosses in this table are pooled results from three or four different F 1 plants. Data were pooled based on chi-square homogeneity tests. y Non-susceptible includes both homozygous resistant and segregating families. x Probability values of 0.05 or greater indicate that the data does not differ significantly from the test ratio. dominant nuclear gene is responsible for flamprop resistance in UMWO Results. families were screened to verify F 2 results (Table 5). Crosses involving UMWO12-01 segregated in a 1 homozygous resistant: 2 segregating:1 homozygous susceptible ratio, which confirmed that a single nuclear gene is responsible for flamprop resistance in this population. Results from crosses involving UMWO12-03 were not as clear. Cross K was homogeneous and fit a 1:2:1 ratio, while M did not. Departure from the 1:2:1 ratio may be due to partial dominance of the resistance trait or due to the herbicide dosage (activity) used to screen these populations. The majority of F 2 results (Table 4) involving UMWO12-03 fit a 3:1 ratio, so it is very likely that one dominant nuclear gene is responsible for flamprop resistance in this population. R/R Crosses F 2 Results. Resistant by resistant crosses were made to determine if resistance to flamprop was controlled by the same gene in each population. Similar to imazamethabenz results, segregation of R and S individuals in the F 2 generation indicated that the two resistant populations carry different resistance genes to flamprop. Ratios tested included 9R:7S, 11R:5S, and 15R:1S for the F 2 results, and the ratio with the best fit is reported (Table 4). A 15R:1S ratio fit best for all of the F 2 crosses, which indicates that a different dominant nuclear gene controls resistance in each population. However, crosses O and Q did not fit the ratio at the 0.05 level of significance. This may be due to partial dominance of the resistance trait or the herbicide dosage (activity) used. Results. families from R/R crosses were screened to clarify F 2 results. The ratio used to test the results was 7 homozygous resistant: 8 segregating: 1 homozygous susceptible as this was the ratio that would be expected based on F 2 results of 15R:1S. Neither of the crosses tested fit a 7:8:1 ratio (data not shown). Similar to imazamethabenz results, the lack of fit of data to a 7:8:1 ratio may be due to the family size screened. A 15R:1S ratio was tested for the data by combining the homozygous resistant and heterozygous families as a non-susceptible phenotype (Table 6). Results from reciprocal

8 324 CANADIAN JOURNAL OF PLANT SCIENCE Table 4. Segregation for flamprop resistance in the F 2 generation of resistant/susceptible (R/S), susceptible/resistant (S/R), and resistant/resistant (R/R) wild oat crosses Resistant Susceptible Cross z number of plants Ratio χ 2 Probability y UMWO12-01/UM5 (R/S) A : B : C + D : E : UM5/UMWO12-01 (S/R) F : G : G : G : G : H : H : H : UMWO12-03/UM5 (R/S) I : J : K : UM5/UMWO12-03 (S/R) L : L : L : L : M : N : N : N : UMWO12-01/UMWO12-03 (R/R) O : P : Q : UMWO12-03/UMWO12-01 (R/R) R : S : T : z Crosses not followed by a number designation indicate pooled results from three or four different F 1 plants. Data were pooled based on chi-square homogeneity tests. Crosses where a number designation follows the cross indicates that results are from an individual F 1 plant. Probability values of 0.05 or greater indicate that the data does not differ significantly from the test ratio. Table 5. Segregation for flamprop resistance in the generation of resistant/susceptible (R/S) and susceptible/resistant (S/R) wild oat crosses Resistant Segregating Susceptible Cross z number of families Ratio χ 2 Probability y UMWO12-01/UM5 (R/S) A + B :2: UM5/UMWO12-01 (S/R) F :2: UMWO12-03/UM5 (R/S) K :2: UM5/UMWO12-03 (S/R) M :2: z Data for all crosses in this table are pooled results from three or four different F 1 plants. Data were pooled based on chi-square homogeneity tests. Probability values of 0.05 or greater indicate that the data does not differ significantly from the test ratio. crosses P and T both fit a 15R:1S ratio. results confirmed that flamprop resistance is due to different genes in UMWO12-01 and UMWO12-03, however the interaction between these genes and the resistant phenotype was not clearly determined. Fenoxaprop-P Results F 2 Results. Based on the results of the parental screening described previously, only crosses involving UMWO12-03 were used to study the inheritance of fenoxaprop-p resistance. All UMWO12-03/UM5 crosses were homogeneous and fit a 3R:1S ratio (Table 7). The reciprocal crosses also generally segregated following a 3R:1S ratio. F 2 families from cross M were homogeneous and fit a 3R:1S ratio. F 2 families from cross L also were homogeneous, but did not fit the tested ratio. However, when tested individually, two of the four families from this cross fit a 3R:1S ratio (data not shown). F 2 fam-

9 KARLOWSKY ET AL. INHERITANCE OF MULTIPLE RESISTANCE IN WILD OAT 325 Table 6. Segregation for flamprop resistance in the generation of resistant/resistant (R/R) wild oat crosses Non-susceptible y Susceptible Cross z number of families Ratio χ 2 Probability x UMWO12-01/UMWO12-03 (R/R) P : UMWO12-03/UMWO12-01 (R/R) T : z Data for all crosses in this table are pooled results from three or four different F 1 plants. Data were pooled based on chi-square homogeneity tests. Non-susceptible includes both homozygous resistant and segregating families. x Probability values of 0.05 or greater indicate that the data does not differ significantly from the test ratio. ilies from cross N were not homogeneous, but two of the three families did fit a 3R:1S ratio, and the family that did not fit had a probability value close to the acceptable 0.05 level. Since the majority of crosses fit a 3R:1S ratio, it appears that one dominant gene confers resistance to fenoxaprop-p in UMWO Reciprocal crosses segregated following the same ratio indicating the resistance gene is nuclear and not cytoplasmic. Similar results have been reported for ACCase inhibitor resistance in other wild oat populations (Murray et al. 1995, 1996; Kibite et al. 1995; Seefeldt et al. 1998). Unlike the results for imazamethabenz and flamprop, reversal of dominance was not observed. Results. results were analyzed to determine if they supported the hypothesis that a single dominant nuclear gene was responsible for conferring resistance to fenoxaprop-p in UMWO The results did not closely fit the expected 1 homozygous resistant: 2 segregating: 1 homozygous susceptible ratio (Table 8). The number of segregating families was close to the expected value, but there were too many R families and too few S families to fit the expected ratio. Reciprocal crosses segregated in a similar manner, confirming that the resistance gene is nuclear. While F 2 results supported the hypothesis that a single dominant nuclear gene confers resistance to fenoxaprop-p, results were not conclusive as there appeared to be an over-representation of R families in the generation. Closer examination of the F 2 results also indicated that there were more R individuals than expected (Table 7). Most of the F 2 families segregated in ratios greater than 3R:1S (average segregation ratio was 6R:1S). However, statistically these ratios did not deviate substantially enough to be classified as other than 3R:1S. This lack of fit may be due to the herbicide dosage (activity) used in screening. UMWO12-03 has a resistance factor of 2.9 for fenoxaprop-p (Friesen et al. 2000). This is a relatively low level of resistance, and as a result a relatively low herbicide dosage had to be used to test plant response. This low herbicide dosage may have led to a disproportionate number of escapes and misclassification of susceptible individuals as resistant in both the F 2 and generations. This could account for the small number of susceptible families observed and the lack of fit of the 1:2:1 ratio in the generation. Linkage Between Herbicide Resistance Genes Linkage in UMWO12-01 Data obtained from screening R/S and S/R families with the different herbicides were used to derive the genotypes of individual F 2 plants [data tabulated in Karlowsky (2004)]. The presence of non-parental recombinants, in addition to the parental screening results detailed above, indicated that the same gene was not responsible for resistance to imazamethabenz and flamprop in UMWO For R/S and S/R crosses, non-parental recombinants are those individuals that are resistant to one herbicide and not the other (as indicated previously, only crosses involving R parents homozygous for the resistance traits were screened). All other phenotypes, resistant to both herbicides or susceptible to both herbicides, are parental types. If the resistance genes are independent then the F 2 individuals should segregate in a 9 resistant to both imazamethabenz and flamprop: 3 resistant to imazamethabenz and susceptible to flamprop: 3 susceptible to imazamethabenz and resistant to flamprop: 1 susceptible to both imazamethabenz and flamprop ratio. The pooled data from crosses A, B, and F did not fit this ratio (Table 9). There are two possible theoretical explanations for this lack of fit: (1) either the gene for imazamethabenz or flamprop resistance does not segregate in the expected 3R:1S ratio (but the data presented above indicate that they do), or (2) the two resistance genes are linked (Strickberger 1985). To determine if the genes are linked, chi-square linkage tests were calculated as described by Strickberger (1985). Segregation of each gene was first tested separately to determine if data fit a 3R:1S ratio (Table 9). Imazamethabenz and flamprop resistance genes both fit a 3R:1S ratio. The test for the independent assortment of the genes failed, indicating that the two genes are linked. By dividing the product of non-parental phenotypes by the product of parental phenotypes a z-value of was obtained. This value corresponds to a 10.0% recombination value (Strickberger 1985) indicating that the genes conferring imazamethabenz and flamprop resistance in UMWO12-01 are located relatively close to one another on the same chromosome (i.e.. they are linked). Linkage in UMWO12-03 Similar to UMWO12-01, and in addition to the parental screening results detailed above, the presence of nonparental recombinants for derived F 2 genotypes for imazamethabenz and flamprop resistance indicated that a different gene confers resistance to each herbicide in UMWO12-03 [data tabulated in Karlowsky (2004)]. The data did not fit a 9:3:3:1 ratio indicating that the genes for resistance may be linked (Table 10). Tests for adherence to the 3R:1S ratio

10 326 CANADIAN JOURNAL OF PLANT SCIENCE Table 7. Segregation for fenoxaprop-p resistance in the F 2 generation of resistant/susceptible (R/S) and susceptible/resistant (S/R) wild oat crosses Resistant Susceptible Cross z number of plants Ratio χ 2 Probability y UMWO12-03/UM5 (R/S) I : J : K : UM5/UMWO12-03(S/R) L : M : N : N : N : z Crosses not followed by a number designation indicate pooled results from three or four different F 1 plants. Data were pooled based on chi-square homogeneity tests. Crosses where a number designation follows the cross indicates that results are from an individual F 1 plant. Probability values of 0.05 or greater indicate that the data does not differ significantly from the test ratio. Table 8. Segregation for fenoxaprop-p resistance in the generation of resistant/susceptible (R/S) and susceptible/resistant (S/R) wild oat crosses Resistant Segregating Susceptible Cross z number of families Ratio χ 2 Probability y UMWO12-03/UM5 (R/S) K :2: UM5/UMWO12-03 (S/R) M :2: z Data for all crosses in this table are pooled results from three or four different F 1 plants. Data were pooled based on chi-square homogeneity tests. y Probability values of 0.05 or greater indicate that the data does not differ significantly from the test ratio. Table 9. Chi-square tests to detect the presence of linkage between imazamethabenz and flamprop resistance in UMWO12-01 wild oat based on derived F 2 genotypes from family screening results for all crosses advanced to the F 3 generation z. Data is summarized from Karlowsky 2004 Test ratio y (Observed) χ 2 Probability x 9:3:3:1 (41:2:3:9) :1 imazamethabenz (43:12) :1 flamprop (44:11) Independence z Data based on Crosses A, B, and F (Tables 2 and 5) where there was sufficient seed (and emerged seedlings) available to screen with the herbicides. Refer to Materials and Methods for family screening criteria. y The test ratios were 9 resistant to both imazamethabenz and flamprop: 3 resistant to imazamethabenz and susceptible to flamprop: 3 susceptible to imazamethabenz and resistant to flamprop: 1 susceptible to both imazamethabenz and flamprop; and 3 resistant: 1 susceptible. x Probability values of 0.05 or greater indicate that the data does not differ significantly from the test ratio. were acceptable for imazamethabenz resistance but not for flamprop resistance. The test for independence also failed, indicating linkage of the genes. However, since the test for 3R:1S segregation of flamprop resistance failed it was not possible to determine how much of the lack of fit to the 9:3:3:1 ratio was due to linkage and how much was due to the relative scarcity of flamprop-susceptible phenotypes. Similar to the situation for imazamethabenz and flamprop, the presence of non-parental recombinants in the derived F 2 genotypes indicated that different genes confer resistance to imazamethabenz and fenoxaprop-p in UMWO The data did not fit a 9:3:3:1 ratio, and the data for fenoxaprop-p did not fit a 3R:1S ratio (Table 10). However, the test for independence produced an acceptable chi-square value indicating that the lack of fit to the 9:3:3:1 ratio was not due to linkage in this Table 10. Chi-square tests to detect the presence of linkage between imazamethabenz, flamprop, and fenoxaprop-p resistance in UMWO12-03 wild oat based on derived F 2 genotypes from family screening results for all crosses advanced to the F 3 generation z. Data is summarized from Karlowsky 2004 Test ratio y (Observed) χ 2 Probability x Imazamethabenz/flamprop linkage 9:3:3:1 (81:0:6:10) :1 imazamethabenz (81:16) :1 flamprop (87:10) Independence Imazamethabenz/fenoxaprop-P linkage 9:3:3:1 (51:0:9:1) :1 imazamethabenz (51:10) :1 fenoxaprop-p (60:1) Independence Flamprop/fenoxaprop-P linkage 9:3:3:1 (59:0:6:2) :1 flamprop (59:8) :1 fenoxaprop-p (65:2) Independence z Data based on Crosses K and M (Tables 2, 5, and 8) where there was sufficient seed (and emerged seedlings) available to screen with the herbicides. Refer to Materials and Methods for family screening criteria. y The test ratios were 9 resistant to both herbicides: 3 resistant to herbicide- 1 and susceptible to herbicide-2: 3 susceptible to herbicide-1 and resistant to herbicide-2: 1 susceptible to both herbicides; and 3 resistant: 1 susceptible. x Probability values of 0.05 or greater indicate that the data does not differ significantly from the test ratio. case. The lack of fit of the fenoxaprop-p data to the 3R:1S ratio indicated that the lack of fit to the 9:3:3:1 ratio was due to the relative scarcity of fenoxaprop-p susceptible phenotypes. Again, due to the presence of non-parental types in the derived F 2 genotypes, linkage between flamprop and