The prediction of occurrence of weed resistance to herbicides is limited by the scarcity of informat ion on weed genetics. The risk of finding a resistant weed biotype depends on two aspects : the probability of selecting a herbicide resistant mutant, and the probability of spreading the resistance in the area. The objectives of this work were to study factors that control the probability of selecting a herbicide resista NT mutant, and to predict the risk of selecting a resistant plant. The factors that control the probability of selecting a herbicide resistant mutant are grouped in three categories: those related to the genetics of resistance, those related to the number of plants under selection, and thoserelated to the herbicide. The genetics of resistance is governed by the initial frequency of the resista nt allele, dominance of the gene, and type of species pollination. The number of plants under selection depend on size of the herbicide treated area and weed density. The factors related to the herbicide are its mechanism of action, and frequency of use of the same mechanism of action in the area. Computer simulations indicate great risk of occurrence of a resistant mutant in áreas with high weed density, mainly those sprayed with herbicides whose mechanisms of action present history of high initial frequency of the resistant trait on the weed populations.
Resistance; simulation; prediction; genetics