Skip to main content
ARS Home » Pacific West Area » Maricopa, Arizona » U.S. Arid Land Agricultural Research Center » Water Management and Conservation Research » Research » Publications at this Location » Publication #175302
Title: TESTING APPROPRIATE ON-FARM TRIAL DESIGNS AND STATISTICAL METHODS FOR COTTON PRECISION FARMING

Author
item GRIFFIN, TERRY - PURDUE UNIVERSITY
item Fitzgerald, Glenn
item LAMBERT, DAYTON - PURDUE UNIVERSITY
item LOWENBERG-DEBOER, J - PURDUE UNIVERSITY
item BARNES, EDWARD - COTTON INC, CARY NC
item ROTH, ROBERT - UNIV OF ARIZONA

Submitted to: National Cotton Council Beltwide Cotton Conference
Publication Type: Proceedings
Publication Acceptance Date: 1/4/2005
Publication Date: 3/5/2005
Citation: Griffin, T., Fitzgerald, G.J., Lambert, D., Lowenberg-Deboer, J., Barnes, E.M., Roth, R. 2005. Testing appropriate on-farm trial designs and statistical methods for cotton precision farming. In National Cotton Council Beltwide Cotton Conferences, New Orleans, Louisiana, January 4-7, 2005. 1:383-392.

Interpretive Summary: Conventional statistical methods do not provide adequate information to farmers making management decisions when large scale, on-farm comparisons of or plant varieties or input treatments such as midseason insecticides are made. More valid comparisons using small replicated plots are too time-consuming and sometimes not logistically possible for a farmer to implement. An experiment with large treatment blocks was conducted on an Arizona farm with a conventional and three reduced tillage treatments using a cotton picker yield monitor and two methods of statistical analysis. One method applied a traditional Analysis of Variance and the other a Spatial Analysis of Variance. The second method was found to be more sensitive to treatment differences than the traditional analysis. This method of analysis should help farm managers obtain more reliable data to make better management decisions and, therefore, be more profitable. This method also has application to other scientists working with crops besides cotton harvested with yield monitors with limited replication data.

Technical Abstract: Adoption of precision agriculture in cotton has lagged behind the use of spatial technologies in grain and oilseed crops because the commercialization of cotton yield monitors (YM) occurred several years after the introduction of grain yield monitors. In 2001, 37% of U.S. corn acreage was harvested with a YM, but less than 2% of U.S. cotton was harvested with a YM. Now that cotton yield monitors are available, cotton farmers' interest in on-farm comparisons is growing. Cotton YM data can be collected on-the-go, and planned on-farm comparisons implemented, harvested, and analyzed without interfering with crop production. This is particularly important for inputs specific to cotton such as midseason insecticides, growth regulators, and defoliants applied with aerial applicators. If farmers want to compare input products or rates, larger treatment blocks would be easiest to implement. The objective of this study was to determine if spatial analysis could lead to better farm management decisions from the limited replication data farmers currently collect with cotton YM. To demonstrate how spatial analysis methods apply to on-farm cotton research, two regression methods were used on large block tillage comparisons. Four tillage treatments were applied to cotton at The University of Arizona's Maricopa Agricultural Center. Results indicate the ANOVA using a spatial regression model provides more accurate results compared to standard ANOVA. When standard AVOVA was used, significance levels indicated two treatment variables were different from the mean at the 10% level and one at the 5% level and one at the 1% level. These results indicate more information is gained about local variations over the production surface when spatial autocorrelation is taken into account. Using ordinary ANOVA, these effects would not be identified.