Linkages among agronomic, environmental and weed management characteristics in North American sweet corn
2009
Williams, Martin M. | Davis, Adam S. | Rabaey, Tom L. | Boerboom, Chris M.
Much of our understanding of weed communities and their interactions with crops comes from studies conducted at, or below, the spatial scale of individual fields. This scale allows for tight control of experimental variables, but systematically ignores the potential for regional-scale environmental variation to affect agronomic operations and thereby influence weed management outcomes. We quantified linkages among agronomic, environmental and weed management characteristics of 174 commercial sweet corn fields throughout the north central United States and evaluated crop and weed responses to these variables using classification and regression tree (CART) analysis. Multi-model selection indicated that characteristics of weed management systems, especially total cost and herbicide rate, were important predictors of weed diversity, interference and fecundity. Adding agronomic variables, such as planting date, or environmental variables, such as latitude, explained additional variation in weed floristic measures. We tested yield predictions of the most parsimonious CART model against a verification data set comprised of over 1500 published observations from 25 experiments conducted in the major North American regions where sweet corn is grown for processing. Yield values fell within the 95% confidence interval of observed data for most branches of the tree, suggesting the experimental and analytical approaches were reasonably robust. Several characteristics favoring sweet corn productivity and weed management sustainability were identified. This work resulted in easily interpretable models, both by scientists and producers, which place crop and weed responses within the context of regional-scale variation in agricultural management and the environment.
显示更多 [+] 显示较少 [-]