Submitted to: Computers and Electronics in Agriculture
Publication Type: Peer reviewed journal
Publication Acceptance Date: 6/15/2004
Publication Date: 3/1/2005
Citation: Johnson, C.K., Eskridge, K.M., Corwin, D.L. 2005. Apparent soil electrical conductivity: Applications for designing and evaluating field-scale experiments. Computers and Electronics in Agriculture. 46:181-202. Interpretive Summary: For farmers to adopt scientific concepts to their agricultural operations, evidence of positive outcomes at a scale of operation similar to their own is required; thus, field-scale systems experiments may hasten the adoption of precision agriculture and sustainable management practices. An approach is presented that uses new technologies of GPS, GIS, and mobile electromagnetic induction measurements to design and statistically evaluate field-scale experiments. The approach was first shown in semiarid northeastern Colorado, and then for a contrasting arid zone site in central California's San Joaquin Valley. The approach is referred to as within-field blocking and is based on apparent soil electrical conductivity (ECa) classification. It is a statistical method that enables the study of intact agroecosystems that are practical in terms of scale, management, farm equipment, and soil and yield heterogeneity. Because farmers can relate to experimental goals examined in systems similar to their own, this statistical method for field-scale experimental design can lead to greater acceptance and implementation of best and sustainable management practices and of site-specific crop management.
Technical Abstract: On-farm field-scale research has become increasingly common with the advent of new technologies. While promoting a realistic systems perspective, field-scale experiments do not lend themselves to traditional concepts of statistical design. In foundational research, a farm-scale dryland experiment in northeast Colorado evaluated apparent electrical conductivity (ECa) classification (within-field blocking) as a basis for estimating plot-scale experimental error. Comparison of mean-square (MS) errors for several properties and surface residue mass measured at this site, with those from a nearby plot-scale experiment, revealed that ECa-classified within-field variance approximates plot-scale experimental error. In the present study, we tested these findings at a second and disparate experimental site, Westlake Farms (WLF) in central California. This 32-ha site was ECa mapped and partitioned into four and five classes using a response-surface model. Classification based on ECa significantly delineated most soil properties evaluated (0 to 0.3 and/or 0 to 1.2 m) and effectively reduced MS error (P<0.10). The MS's for several soil properties evaluated at the site were then compared with those of an associated plot-scale experiment; most MS's were not significantly different between the two levels of scale (P<0.05), corroborating results from the Colorado experiment. These finding support the use of within-field ECa-classified variance as a surrogate for plot-scale experimental error and a basis for roughly evaluating treatment differences in non-replicated field-scale experiments. This alternative statistical design may promote field-scale research and encourage a reversal in research direction wherein research questions identified in field-scale studies are pursued at the plot-scale.