The yield responses to inputs of fertilizer and herbicide were tested on lowland farms in the central Philippine province of Antique. Replicated experiments were superimposed on 138 rainfed rice crops during four cropping seasons over 2 years. The experiments tested the yield responses to recommended inputs of the herbicide butachlor and to N, P and K. In the latter part of the project, the experiments also tested the yield responses to Zn (25 experiments) and S (24 experiments). Mean yield responses to herbicide, N, P and K were significant in all seasons in which they were tested, and the mean yield response to Zn was significant in one of the three seasons in which it was tested. There was relatively little variation between farms in the yield responses to herbicide or N. However there was greater between-farm variability in yield responses to P, K, Zn and S; crops growing on most farms showed little or no yield response while those on some farms showed extremely large responses. Reasons for the between-farm variability were sought from measurements of soil organic matter, P, K and water status, and from the recent history of fertilizer application. None of the relationships between yield responses and these factors was sufficiently close to be useful in predicting the responses from a particular field. The results are discussed in terms of possible reasons for the variability, and the implications for extension advice in an edaphically heterogeneous region. The most appropriate means of identifying responsive crops where such deficiencies are suspected is to conduct an extensive program of test strips of nutrients applied to farm crops. © 1990.
