BRIEF COMMUNICATION doi:10.1111/evo.12531 Fitness cost due to herbicide resistance may trigger genetic background evolution Henri Darmency, 1,2 Yosra Menchari, 1,3 Val ´ erie Le Corre, 1 and Christophe D ´ elye 1 1 INRA, UMR1347 Agro ´ ecologie, Dijon 21000, France 2 E-mail: darmency@dijon.inra.fr 3 Current Address: Institut Sup ´ erieur de Biotechnologie de B ´ eja, 9000 B ´ eja, Tunisia Received September 3, 2013 Accepted September 8, 2014 This article investigates the possible existence of mechanisms counterbalancing the negative pleiotropic effects on development and reproduction that are conferred by alleles responsible for herbicide resistance in the weed Alopecurus myosuroides. We considered three herbicide-resistant, mutant acetyl-coenzyme A carboxylase (ACCase) alleles, Leu1781, Asn2041, and Gly2078, found in eight resistant populations. Of these, Gly2078 is the only allele with a known fitness cost. We compared plants homozygous for wild-type ACCase alleles that were siblings of plants carrying a given mutant resistant ACCase allele with plants from three populations where resistance did not evolve. In each of two series of experiments, we measured germination dynamics, seedling vigor, plant height, vegetative biomass, and seed production. The wild-type siblings of plants carrying Gly2078 performed better in the field, on average, than wild-type plants that were sibling of plants carrying other mutant ACCase alleles, and particularly those carrying Leu1781. We propose that rapid evolution of the genetic background of plants from the populations where the Gly2078 allele originally arose could partially counterbalance Gly2078 fitness cost, enhancing the spread of the resistant genotypes. KEY WORDS: ACCase, Alopecurus myosuroides, compensatory evolution, fitness cost, herbicide resistance, weeds. The repeated use of herbicides in arable fields has led to weed adaptation. Selection has favored resistance mechanisms that enable weed plants to withstand herbicide applications in 237 species so far (Heap 2014). The first cases of herbicide resistance studied in weeds were associated with lower relative fitness of the resistant plants in the absence of herbicides (Holt 1990). The results of these early studies contributed to the widespread idea that any new mutation conferring herbicide resistance would be associated with deleterious pleiotropic effects, also known as fit- ness costs (reviewed in Vila-Aiub et al. 2009). More generally, mutations that confer resistance to xenobiotics are usually thought to carry an associated fitness cost in the absence of the selection pressure (Coustau et al. 2000). This idea has recently been chal- lenged by studies demonstrating that herbicide resistance is not consistently associated with reduced fitness. Rather, expression of any fitness cost depends on the resistance mechanism, mutation, genetic background, and environmental conditions (Bergelson and Purrington 1996; Menchari et al. 2008; Vila-Aiub et al. 2009; D´ elye et al. 2013). However, resistant plants selected by herbicide applications in agricultural fields (i.e., not laboratory mutants) have generally been collected after the spread of resistant weeds has caused visible weed control failure. This happens several gen- erations after the original resistant mutation has appeared. Thus, by the time farmers notice herbicide resistance, adaptive selec- tion could have already reduced the fitness cost associated with herbicide resistance alleles in the field. It is therefore not possi- ble to separate the possible pleiotropic effects of the resistance allele itself from the outcome of possible subsequent evolution- ary processes that may have counterbalanced these effects. The potential for compensatory evolution has never been considered in relation to herbicide resistance. Compensatory evolution has been observed in several situations in which alleles responsi- ble for antibiotic resistance (Andersson 2003; Schulz zur Wiesch et al. 2010) or insecticide resistance (Correa et al. 2011) have an associated fitness cost. Compensatory evolution has been most closely studied in bacteria—the high generation turnover allows for evolutionary studies both in the laboratory and in in vivo clinical situations. 271 C  2014 The Author(s). Evolution C  2014 The Society for the Study of Evolution. Evolution 69-1: 271–278