Three empirical models were compared to evaluate the efficiency in predicting bean crop (Phaseolus vulgaris) yield loss (PRR) due to its coexistence with the weed (Brachiaria plantaginea). The population of the latter was described through density (Npd), relative leaf area index (Fpd) and relative ground cover (Spd). The model coefficients were estimated using data obtained from a field experiment carried out from October 1999 to January 2000 at the University of Sao Paulo (USP), Piracicaba, Sao Paulo, Brazil. The experiment was arranged in a randomized block design using split plots, with weed sowing dates (eight days before, on the same day, five and 10 days after crop sowing) as the main plots and weed densities (0, 25, 50, 100 and 200 plants m-2) as subplots. The crop yield losses in the treatment with zero density of B. plantaginea were compared to those predicted by fitting the data to the models. Although the model variable Npd could not predict crop yield loss due the relative time of weed sowing [PRR=0,0114.Npd/(1+0,0114.Npd ), r²=0,03], those models whith variables Fpd or Spd could [PRR=6,27.Fpd/(1+5,27.Fpd), r²=0,82; PRR=7,77.Spd/(1+6,77.Spd), r²=0,89]. The visually estimated weed relative ground cover is a potential variable to substitute the weed relative leaf area index.
weed-crop interactions; competition; weed density; relative leaf area index; relative ground cover
