Thousands of field trials are conducted annually to evaluate the usefulness of various techniques in weed control. Conventional data collection and statistical methods lead to relatively scant information from trials because of the spatial heterogeneity and temporal changes in weed abundance. To evaluate whether additional information could be drawn from new methods in on-farm trials, two experiments were carried out to compare different data collection and statistical methods. First, we compared conventional sampling method using biomass estimate of weed abundance to repeated visual assessment of the percentage ground cover. Biomass was sampled once after the treatment whereas ground cover was repeatedly sampled once before plus several occasions after the weed control. Secondly, we contrasted outcomes from ANOVA taking samples from a single point in time with repeated measures ANOVA and a multivariate method (the constrained ordination pRDA). It was concluded that ground cover estimate of weed abundance was as reliable as biomass estimate because the outcomes and conclusions drawn were relatively similar. The repeated measures ANOVA enabled to follow the temporal dimensional trend and the initial flora differences. Multivariate analysis went even further by displaying species-wise the impact of each control tool in the frame of the environmental gradients.

Keywords: Ground cover, Multivariate methods, Repeated measures, Weed control