Assessing fitness costs associated with genes endowing herbicide resistance is of importance in predicting resistance dynamics and proposing resistance-management strategies (Maxwell, Roush & Radosevitch 1990; Jasieniuk, Brûlé-Babel & Morrison 1996). This is the first study where weed individuals with a characterized genotype at a locus endowing herbicide resistance have been used to assess fitness costs associated with herbicide resistance, and where fitness cost dominance was assessed.
Mutant ACCase alleles have been shown to appear independently in distinct black-grass populations (Délye et al. 2004). Given that there may be occurrences when some of the differences observed could be due to linkage disequilibrium of a mutant ACCase allele with deleterious genes, and given that fitness cost associated with a resistance gene can also vary depending on the genetic background in which it occurs (Bergelson & Purrington 1996), a total of five, three and two segregating populations were used across each of two field experiments to assess possible fitness costs associated with Leu-1781, Asn-2041 and Gly-2078 ACCase alleles, respectively. Fitness cost was also reported to vary with environment (Bergelson & Purrington 1996; Jordan et al. 1999; Martin & Lenormand 2006), which is the reason why we chose to assess it in the situation where resistance genes are selected, in competition with a crop plant rather than in the absence of competition.
fitness cost depends on the mutant accase allele
We did not observe any fitness cost on plant growth and seed production associated with Leu-1781 ACCase in 10 different black-grass populations segregating for this allele. Our results confirmed and extended those of previous studies where the fitness cost associated with Leu-1781 ACCase on plant growth and seed production was investigated in rye-grass in the absence of competition (Vila-Aiub, Neve & Powles 2005). These authors observed a fitness cost associated with enhanced herbicide metabolism, and an absence of additional fitness cost in plants with both enhanced herbicide metabolism and Leu-1781 ACCase, which enabled them to conclude indirectly that there was an absence of fitness cost associated with Leu-1781 ACCase in the rye-grass population they studied (Vila-Aiub, Neve & Powles 2005). They also found that Leu-1781 ACCase may confer differences in germination in plants from the particular population they studied (Vila-Aiub, Neve, Steadman & Powles 2005), but this finding needs to be confirmed across a broader range of genetic backgrounds before its general value can be established. Literature suggests that the absence of fitness cost associated with Leu-1781 ACCase is a general feature. First, Leu-1781 ACCase enzyme activity did not differ from that of wild-type ACCase in the grass species where it has been studied (rye-grass, Yu et al. 2007; green foxtail Setaria viridis, Shukla, Leach & Devine 1997; confirmation that the mutant ACCase studied was Leu-1781 was provided by Délye, Wang & Darmency 2002). Second, Leu-1781 ACCase is clearly the most widespread mutant ACCase allele in black-grass populations in France (Menchari et al. 2006). Third, Leu-1781 ACCase is the mutant ACCase allele most frequently identified across grass species where populations evolved mutant ACCase (five species, Délye 2005; Zhang & Powles 2006a; Liu et al. 2007; Yu et al. 2007). Finally, this allele became fixed in at least three grass species, annual blue-grass Poa annua, prostrated blue-grass Poa supina and red fescue Festuca rubra (Délye & Michel 2005). Among these species, prostrated blue-grass and red fescue are not arable weeds, and have consequently not been subjected to the selective pressure of ACCase-inhibiting herbicides. All these data are clearly in favour of the absence of fitness cost associated with Leu-1781 ACCase.
To date, there has been no published study addressing the possible fitness costs associated with Asn-2041 or Gly-2078 ACCase. Asn-2041 ACCase from black-grass displayed a moderately reduced enzyme activity when compared with wild-type ACCase (Délye et al. 2003). It was significantly less frequent than Leu-1781 ACCase in France (Menchari et al. 2006). Asn-2041 ACCase has been reported in only two other weed species so far: rye-grass (Délye et al. 2003; Zhang & Powles 2006b; Yu et al. 2007) and wild oat Avena sterilis (Liu et al. 2007). However, we did not detect any fitness cost on plant growth or seed production associated with Asn-2041 ACCase across six black-grass populations segregating for this allele grown in competition with wheat. Because our study did not address all key parameters of the black-grass life cycle, a possible deleterious pleiotropic effect of Asn-2041 ACCase on seed germination or survival in the soil cannot be excluded, and needs to be investigated. Additionally, the broad phenotypic plasticity of black-grass and the phenotypic variability observed across the 2 years in which the field experiments were conducted (Table S1) may render undetectable any subtle fitness cost, nevertheless having a cumulative, deleterious effect in the field over many growing seasons.
Sequencing showed that two distinct Gly-2078 ACCase haplotypes occurred in both populations G2078-D41-04 and -05. Both haplotypes were distinct from the one occurring in populations G2078-D83-04 and -05 (not shown). This enabled us to rule out the occurrence of an unknown gene linked to Gly-2078 ACCase that would be responsible for the significant reduction in plant height, vegetative biomass and seed production observed in M/M Gly-2078 ACCase plants across both field experiments (Table 4; Fig. 1). Gly-2078 ACCase was the mutant, resistant ACCase allele detected least frequently across black-grass populations in France (Menchari et al. 2006). It has been reported only in black-grass and rye-grass so far (Délye et al. 2005; Yu et al. 2007). Its enzyme activity was about half that of wild-type ACCase (Délye et al. 2005; Yu et al. 2007). The data suggested that there may be a fitness cost associated with Gly-2078 ACCase, which our study confirms: we found that the fitness cost associated with Gly-2078 ACCase would be recessive or weakly semidominant. Characterizing the dominance of fitness cost is very important because the spread and persistence of resistance genes initially depends on the fitness of the first mutant resistant individuals to appear, which are M/W plants in diploid species such as black-grass. The diffusion of a resistant ACCase allele will be much faster if the associated fitness cost is recessive (Jasieniuk, Brûlé-Babel & Morrison 1996). As a consequence, the fitness cost associated with Gly-2078 ACCase should only moderately limit the frequency and alter the initial evolutionary dynamics of this type of alleles in black-grass populations.
The low (<4%) genetic contribution to variation in vegetative biomass, plant height and seed production observed here across genotypes and populations suggests that it is unlikely that black-grass will be selected for higher or lower vegetative and reproductive productivity. Thus any fitness cost reducing vegetative and reproductive productivity is expected to have a dramatic effect on the survival and reproduction of the plants. Large individual plants gain more resources (water, light, nutrients) and produce more seeds than smaller ones. Smaller individuals thus tend to be eradicated in plant populations (Weiner 1986). In addition, the reduction in M/M, Gly-2078 plant height observed (7 cm, about the average length of an inflorescence) may prevent inflorescences from emerging from the wheat canopy, reducing the possibility of pollen exchange that is crucial for seed production in the allogamous black-grass. M/M, Gly-2078 plants are therefore expected to be at a disadvantage in competitive crops or in competition with plants containing Leu-1781 ACCase, Asn-2041 ACCase or, in the absence of herbicide treatment, with wild-type plants. This disadvantage might be exacerbated by a segregation distortion against M/M, Gly-2078 ACCase plants (Table 2), although the deficit observed in this genotype may also reflect differences in the dynamics of germination among genotypes. ACCase being a key enzyme in lipid biosynthesis, and lipid being important storage compounds in black-grass seeds (Darmency, Landry & Mossé 1981), pleiotropic effects of the reduced enzyme activity of Gly-2078 ACCase on seed germination and/or survival in the soil may exist, and need to be investigated.
While a deleterious effect of Gly-2078 ACCase was clearly detected in M/M plants in the 2004–05 field experiment, it was not evident in the 2005–06 experiment (Table S1). The 2005–06 field experiment was performed in a context of lower competition for water resources than the previous experiment, as can be seen clearly in the higher total vegetative dry biomass and seed production values (Table S1). Also, the mother plants used to produce the segregating populations were not the same as in the previous year. Fitness cost on plant growth and seed production associated with Gly-2078 ACCase is thus clearly dependent on the population and environmental conditions experienced during the growing season. This illustrates the roles of the genetic background and of the environment in the expression of fitness cost.
From the literature and our results, Leu-1781 ACCase appears to be the ‘best’ herbicide-resistant ACCase allele to be selected for in black-grass, and probably in other grass weed species. Given that mutant, resistant ACCase alleles arise by multiple, independent appearances in black-grass populations (Délye et al. 2004; Menchari et al. 2006), a replacement of the less fit mutant ACCase alleles by Leu-1781 ACCase in black-grass populations is expected, especially if herbicide selective pressure is maintained and all ACCase alleles confer a similar degree of protection against the selecting herbicide.
management of resistance endowed by mutant accase
The different mutant ACCase alleles do not confer identical cross-resistance patterns to ACCase-inhibiting herbicides. These herbicides essentially belong to two chemical families, ‘fops’ and ‘dims’ (reviewed by Délye 2005). Leu-1781, Asn-2041 and Gly-2078 ACCase alleles all confer resistance to the two fop herbicides (fenoxaprop and clodinafop) used most commonly in France over the past 15 years (Délye 2005; Menchari et al. 2006). However, Leu-1781 ACCase confers cross-resistance to other, but not all, fops and to some, but not all, dims. Asn-2041 ACCase confers cross-resistance to all fops assayed but not to dims. Gly-2078 confers resistance to all fops and to all dims assayed (reviewed by Délye 2005), which may be a reason why this allele is still present in a number of black-grass populations (Menchari et al. 2006) despite its association with a significant fitness cost. The broad cross-resistance spectrum of Gly-2078 ACCase makes it the most potentially troublesome resistant ACCase allele in terms of resistance management. This allele is fortunately associated with a recessive or weakly semidominant fitness cost such that the use of competitive crops (rye, wheat sown at higher density, wheat cultivars with a strong biomass development) should strongly hamper the development, seed production and pollen exchange of M/M plants, and could perhaps reduce those of M/W plants containing Gly-2078 ACCase. Alternatively, including a grass or alfalfa meadow in the crop rotation should have a similar, or more drastic, effect on the development and seed production of plants containing Gly-2078 ACCase.
To obtain a complete picture of fitness costs associated with mutant ACCase alleles, we need to study the whole life cycle of black-grass, and to investigate the possible pleiotropic effects of Cys-2027 and Ala-2096 ACCase alleles. This may reveal differences in the biology and ecology of plants containing these alleles that can be exploited by cultural practices to create or to maximize fitness costs, as demonstrated elsewhere (Jordan et al. 1999). However, given that fitness costs associated with resistant ACCase alleles may vary with black-grass genetic background and with the environment, as observed here for Gly-2078 ACCase, the continued success of such cultural practices is uncertain. Also, it is possible that, as is very likely the case for Leu-1781 ACCase, a significant part of the genes endowing resistance to ACCase-inhibiting herbicides are not associated with pleiotropic effects on the black-grass life cycle. A solution could then be to use diversified management practices, as has been suggested to manage triazine-resistant Amaranthus hybridus plants (Jordan et al. 1999). This would involve combining cultural practices such as ploughing, which disrupts black-grass proliferation (Chauvel et al. 2001), with cultural practices found to create or maximize the fitness cost for some resistant ACCase alleles, and herbicides with diverse modes of action. Such a diversified black-grass management programme should prevent selection for resistant ACCase alleles, or reduce the speed with which it spreads. However, as Jordan et al. (1999) noted, ‘many social, economic and agronomic barriers stand in the way of diversified integrated weed management’.