Author Information

Radha G. Mohanty

Abstract

Papadakis analysis, originally proposed by Papadakis in 1937 belongs to a larger class of methodologies called the nearest neighbor analysis which is primarily based on the fact that plots in close proximity ("neighbors") are exposed to similar environmental conditions and therefore, for a given plot, information from its neighboring plots could be used for adjustment of its response for spatial variability. The basic theory behind the application of Papadakis methodology to field trials is relatively simple. It is based on an analysis of covariance where the covariate is an index of fertility environment), and the response is some observable trait (e.g., grain yield), which is adjusted up or down to reflect the effect due to spatial variability. There have been several references in the literature to application of Papadakis methodology to field trials where the analysis is routinely carried out on data coming from a replicated design within a testing location. The application that is presented here is an exception to the rule in that the analysis is conducted on multi-location data with single replication per location. In plant breeding industry, a recent trend has been to move towards one-replicate testing system to maximize the coverage of the testing environments. Note that for a one-replicate test, no design such as a Lattice, can be used for adjustment of the observations for spatial variability. We start with describing the theory and methodology behind the proposed Papadakis analysis for multilocation data. Several practical problems such as impact of missing values on Papadakis covariate, choice of homogeneous vs. heterogeneous slope coefficient, and effect of influential observations, etc. are discussed and solutions are proposed. Finally, results from several validation studies on com yield data, including comparison to lattice adjusted plot values and ANOV A on adjusted vs. unadjusted data are presented to demonstrate the benefit from the proposed procedure.

Keywords

Papadakis analysis, NNA, Nearest neighbor adjustment, Spatial analysis, Genotype by environment interaction, Plant breeding, Lattice analysis

Creative Commons License


This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Share

COinS
 
Apr 29th, 10:00 AM

PAPADAKIS NEAREST NEIGHBOR ANALYSIS OF YIELD IN AGRICULTURAL EXPERIMENTS

Papadakis analysis, originally proposed by Papadakis in 1937 belongs to a larger class of methodologies called the nearest neighbor analysis which is primarily based on the fact that plots in close proximity ("neighbors") are exposed to similar environmental conditions and therefore, for a given plot, information from its neighboring plots could be used for adjustment of its response for spatial variability. The basic theory behind the application of Papadakis methodology to field trials is relatively simple. It is based on an analysis of covariance where the covariate is an index of fertility environment), and the response is some observable trait (e.g., grain yield), which is adjusted up or down to reflect the effect due to spatial variability. There have been several references in the literature to application of Papadakis methodology to field trials where the analysis is routinely carried out on data coming from a replicated design within a testing location. The application that is presented here is an exception to the rule in that the analysis is conducted on multi-location data with single replication per location. In plant breeding industry, a recent trend has been to move towards one-replicate testing system to maximize the coverage of the testing environments. Note that for a one-replicate test, no design such as a Lattice, can be used for adjustment of the observations for spatial variability. We start with describing the theory and methodology behind the proposed Papadakis analysis for multilocation data. Several practical problems such as impact of missing values on Papadakis covariate, choice of homogeneous vs. heterogeneous slope coefficient, and effect of influential observations, etc. are discussed and solutions are proposed. Finally, results from several validation studies on com yield data, including comparison to lattice adjusted plot values and ANOV A on adjusted vs. unadjusted data are presented to demonstrate the benefit from the proposed procedure.