We demonstrate how direct (constrained) ordination methods can be used to analyze differential control by different weed management tactics. Weed biomass data of individual weed species were compiled for 110 trials on potatoes, each of which included a mechanical control treatment, a herbicide treatment, and an untreated control. We used “trial id” as 110 categorical dummy covariables, thereby eliminating between-trial variation, and made direct comparisons between the data from numerous trials. Unweeded plots had the largest biomass of all species as expected. Contrasting the results for the standard herbicide and the mechanical control treatments in a separate analysis, we could rank species from those best controlled by the herbicide (several annuals) to those best controlled by the mechanical treatment (a perennial grass and a climbing annual).

Nomenclature: Potato, Solanum tuberosum L.

Additional index words: Metribuzin, multivariate analysis, redundancy analysis, selectivity, Sweden, weed community.

Abbreviations: pRDA, partial Redundancy Analysis; RDA, Redundancy Analysis.