ABSTRACT
The seedling stage is a common target of many weed mechanical control and herbicide methods because of its high vulnerability (Fenner, 1987). The success of control methods targeted at weed seedlings depends upon reach ing the highest number of individuals at this developmental stage. However, it is practically impossible to determine which proportion of a certain weed population is being reached by a control method. Indeed, the number of emerged seedlings can be counted, but we do not really know which frac tion of the population they represent. Construction of weed seed germina tion models that predict which proportion of the seed bank germinates at a certain time would be useful tools for determining the most suitable time for seedling control and, consequently, should result in a higher efficacy of controls methods (Benech-Arnold and Sanchez, 1995). Although many models that successfully predict seed germination have been developed, one of the most important limitations for the formulation of such models in many common weed species is the existence of dormancy. The lack of de tailed research intending to understand and quantify how environmental factors regulate dormancy status in field situations probably prevented the elaboration of an adequate theoretical framework for the construction of predictive models addressing dormancy changes in weed seed bank popula tions.
