Harvest weed seed control (HWSC) is an effective technique for managing wild radish (Raphanus raphanistrum L.), a weed that retains its seed until harvest. However, earlier flowering time (leading to increased seed shedding before harvest) is a risk to HWSC effectiveness. This study investigated the effects of repeated HWSC on the evolution of R. raphanistrum flowering dates, using two methods: an adaptation of the SOMER model that included flowering genes (called SOMEF); and a mathematical calculation of the endpoints of flowering date evolution utilizing the relevant life-history equations. In weed management systems with highly effective herbicides, the additional use of HWSC predicted R. raphanistrum population extinction. Low weed numbers and rapid extinction meant that any gradual evolution in days to first flower (DFF) was insufficient to lead to HWSC evasion. In alternative management systems with less vigorous herbicide control and using HWSC, modeling predicted a maximum 2- to 3-d reduction in DFF. In contrast, mathematical calculations of the phenotypes maximizing seeds returned to the seedbank predicted an endpoint to evolution of 12-d earlier flowering, which matched field observations. However, genetic change postulated by the mathematical calculations was not hampered by a restriction to changing DFF allele frequencies. Unknown accompanying genetic changes could affect germination dates or flowering triggers.

Simulation modeling that included only flowering genes failed to predict the magnitude of an observed 12-d reduction in DFF. Differences between the 12 d observed in the field (and predicted using mathematical calculations) and the modest changes demonstrated in this field-based modeling study are postulated to be due to unaccounted evolutionary changes in R. raphanistrum.