Thesis (BS Biology) -- University of the Philippines Mindanao, 2010

A long term experiment comparing manual and chemical weed management in a banana plantation was conducted to test the hypothesis that weed control can be done without sacrificing the environment. The weed count, biomass and seed bank of seven monocot weed species Cyperus brevifollius, cynodon dactylon, eleusine indica, Imperata cylindrica, Paspalum conjugatum, Digitaria cilliaris and Digitaria longiflora were monitored. Based on the seed bank and field counts, weed population was not significantly higher in manual plots than chemical plots. Trends and exceptions were explained in terms of the biological characteristics of weed species and their differential response to weed control treatments. Species diversity was higher in chemical plots because the herbicide kept the roots intact, reducing the soil and weed seed erosion. Laboratory experiment were conducted to determine if parquet resistance exists among the weed species, and if the incidence of resistance would increase with continuous treatment with paraquat over four generations. It was found out that resistant types occur at a fairly high frequency (42%, 24% and 74% in C. brevifolius, C. dactylon and E. indica). Species E. indica seedlings, which was studied more closely, were sprayed with paraquat, saved seeds from the survivors, planted the seeds and repeated the process over four generations. Species, E. indica showed an insignificant increase of frequency of parquet resistant plants over four generations, but taller and faster maturation of resistant individuals with generation advance. The experiments led to the following conclusions: 1) reduction in total weed population did not differ between manual and chemical weed management, but chemical weed management resulted in a statistically higher weed species diversity; 2) E. indica has a high proportion of paraquat-resistant individuals, and 3) repeated paraquat application increased the fitness of resistant types