Spatial variability in the growth of direct-seeded rainfed lowland rice (Oryza sativa L.) in northeast Thailand

Large within-field variation in rice growth often causes production loss in broadcast-seeded (BC) rainfed lowland rice. The spatial variability of direct-seeded rainfed lowland rice was evaluated in 2004, 2005, and 2007 in on-station experiments at Ubon Ratchathani, northeast Thailand, in relation to soil water content and weed infestation, by adopting semivariogram and block kriging, including comparisons among BC with harrowing (BCH; no weeding), BC with no harrowing (BCNH; no weeding), and row-seeded (RS; interrow weeding once) fields. BCH and BCNH were also compared in 11 farmers' fields in 2006 and 2007, to assess the effect of harrowing on rice growth and weed infestation. During most of the rice growing periods, flooded and non-flooded portions existed simultaneously in the fields, with different proportions among years and among seeding methods in the on-station experiment. BCH and BCNH rice had large within-field variation in seedling density, heading date, shoot dry matter, grain yield, harvest index, panicle density, and filled spikelet per panicle, as well as in weed infestation, measured by a quick visual estimation. Many of the measured variables (except mean soil water content in RS in 2007, seedling density in BCH in 2005 and 2007, shoot dry matter in BCH and BCNH in 2007, and panicle density in BCH in 2007) were spatially dependent (i.e., data from nearby locations were most similar) by geostatistical analysis. Analysis of correlations using the 420 data sets of BCH plots in 2005 and BCH, BCNH, and RS plots in 2007 revealed a positive correlation between soil water content and grain yield and negative correlations between weed infestation and soil water content and grain yield. Compared with BCH, in 2007 BCNH had much lower grain yield because of lower soil water content after establishment and more weed infestation. BCH had higher grain yield than BCNH in weedy fields in the farmers' fields experiment. RS with interrow weeding resulted in a smaller coefficient of variation, smaller sill value, and higher grain yield than BCH, due to less weed infestation and a higher proportion of flooded water. These results indicated that reducing the spatial variability in rice growth requires careful field preparation, such as harrowing to level the soil surface and to reduce the uneven distribution of standing water and the variability in soil water content, combined with effective crop and weed management (i.e., harrowing and row-seeding). This is the first study that examined spatial variability in the growth of direct-seeded rice as a function of soil water content and weed infestation in a rainfed lowland environment.