Weather effects on herbicide efficacy at reduced doses. A systems approach

The thesis summaries and discusses results of weather effects on herbicide efficacy in field experiments with spring barley during 1991-1994 at six different sites each year in southern Sweden. Treatments were four sub-normal doses (1/8, 1/4, 1/2 and 3/4 of recommended dose) of two herbicides, dichlorprop-P/MCPA and tribenuron-methyl, applied at three different application times. Weather stations, placed in the experimental fields, registered weather data. Regression analyses on individual weather factors revealed significant effects of some weather factors on the herbicide efficacy but it was not possible to disclose a totally consistent and causal relationship to weather. Thus, precipitation and high soil temperature decreased the effects of both herbicides regardless of dose while the influences of air temperature and global radiation were variable. When the influcence of weather was integrated in the weed growth rate as estimated by a simple mechanistic growth model, it was found that a herbicide application performed at an increasing weed growth rate gave the strongest effect. The conclusion was that environmental conditions that enhanced weed growth rate also increased the herbicide effect. The experiments also showed that herbicide dose and application time had significant influence on the herbicide effect. A higher dose and an early application gave a stronger treatment effect. The grain yield of barley was increased by herbicide treatment but not significant differences between doses or application times were detected. The yield variation was not significantly affected by the weed occurrence while the weather-induced yield variation, as analysed by a complex mechanistic growth model (WOFOST), was considerable. A lowering of herbicide dose did not increase the production risk of barley grain yield. It is concluded that a herbicide application should preferably be performed when the growth conditions are favourable in order to obtain a satisfactory treatment effect at sub-normal herbicide doses. These conditions may be predicted by a growth model, if on-line weather information is available.