A case for comprehensive analyses demonstrated by evaluating the yield benefits of neonicotinoid seed treatment in maize (Zea mays L.). (August 2018)

• Record Type: Journal Article
• Authors: Del Pozo-Valdivia, Alejandro I.; Reisig, Dominic D.; Arellano, Consuelo; Heiniger, Ron W.
• Is Part Of: Crop protection. Volume 110(2018)
• Page Start: 171
• Page End: 182
• Publication Date: 2018-08
• Subjects: Agronomic practice -- Clothianidin -- Corn -- Decision tree analysis -- Imidacloprid -- PROC GLMSELECT -- SAS enterprise miner work station -- Thiamethoxam; Plants, Protection of -- Periodicals; 632.05
• DOI: 10.1016/j.cropro.2017.10.021
• Languages: English
• ISSNs: 0261-2194

Abstract

Abstract: With increased scrutiny of the neonicotinoid class of chemistry and its negative impact on the pollinator community, ecological cost/benefit analyses of agronomic crops that use these insecticides are increasingly important. This study initially sought to address the question of yield benefit due to neonicotinoid seed treatment in maize ( Zea may s L.), using North Carolina yield contest data from 2002 to 2006, the time period from initial neonicotinoid seed treatment adoption to nearly ubiquitous adoption. However, we recognized that several agronomic practices, including planting date, hybrid selection, and fertilization, could affect the yield of this crop; moreover, they could be collinear with one another and the analysis could be skewed by early adopters of new technology. Hence, we used all available data to compare among traditional approaches and a data-mining approach for analyzing the impact of neonicotinoid seed treatment on maize yield. At-planting insecticide treatment was not an important predictor of maize yield. When analyzed using the traditional approach (T-test), yields were significantly higher for fields planted with neonicotinoid treated seed compared to seed without neonicotinoid; however, data-mining approach (Decision tree analysis) that took into account other factors contributing to yield did not identify seed treatments as important. The contrast in these results highlights the need for future carefully designed studies that target to minimize inter- and intra-site variation; and include measurements of additional factors that may influence yield, such as seeding rate, tillage, and herbicide applications, as input variables that are largely lacking in current approaches on the subject. Graphical abstract: Image 1 Highlights: This study is an example of using preexisting data to address ecological-based questions in agriculture. Initial goal was to assess yield benefit of neonicotinoid seed treatment in maize, using NC yield contest data (2002-2006). Data were analyzed using both a T-test and stepwise model selection, as well as a decision tree analysis (data mining). Observed trends on how neonicotinoid seed treatment affected maize yield were different among statistical approaches.