Weed vegetation response to chemical and manual selective ground cover management in a shaded coffee plantation

Summary Total weed control with machete and herbicides was compared in a field experiment from 1994 to 1999, with four selective ground cover management tactics (± herbicides and ± planted ground cover legume) in the inter-rows of coffee plants. The aim of the selective management was to control weeds in patches, leaving uncontrolled species considered suitable to protect the soil from erosion and compaction, and to suppress more aggressive weed growth. Fresh biomass was sampled early and late in the rainy seasons. Weed data from the different years, treatments and blocks were analysed with a multivariate technique, partial redundancy analysis (pRDA), using different combinations of independent variables and covariables and resulting in a multivariate anova . Weed biomass and number of species drastically decreased over time as coffee and shade trees aged. The five treatments also significantly affected the composition of the weed vegetation. The combination of partial slashing and application of herbicides in patches was more effective in reducing unwanted weed biomass and also enhanced the spread of the ground cover legume, whereas the use of only partial slashing enhanced the spread of the 'weed' Oplismenus burmannii . This species was considered a suitable ground cover species to protect the soil, as it emerges at the onset of the rainy season, is more persistent late in the rainy season and sustains growth under the shade of coffee in production.