|Stone, Kenneth - Ken|
|BARNES, EDWARD - Cotton, Inc|
|JONES, DONALD - Cotton, Inc|
|Campbell, Benjamin - Todd|
Submitted to: The Plant Phenome Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/15/2017
Publication Date: 2/15/2018
Citation: Hugie, K.L., Bauer, P.J., Stone, K.C., Barnes, E.M., Jones, D.C., Campbell, B.T. 2018. Improving the precision of NDVI estimates in upland cotton field trials. The Plant Phenome Journal. 1(1):1-9.
Interpretive Summary: Imagery and remote sensing technologies are revolutionizing crop improvement programs by enabling larger numbers of breeding lines to be evaluated across time and space. For instance, simultaneous measures of plant spectral reflectance and height can be rapidly collected and used to estimate crop biomass throughout the growing season. As the size of field trials increases, controlling for environmental variability will be a critical factor in generating reliable measures of the performance potential of cultivars and breeding lines. In the current study, we compared the efficacy of different experimental designs and statistical procedures in the analysis of spectral reflectance data collected on upland cotton performance trials conducted at Florence, South Carolina, in 2014 and 2015. The use of more complex experimental designs and statistical methods increased the precision of spectral reflectance estimates and had substantial effects on the selection of superior breeding lines, particularly at selection intensities above 10%. These results suggest that the use of more complex experimental designs and statistical procedures should be considered to minimize experimental error introduced by environmental variability in spectral reflectance data. These findings also indicate that further research into the effects of spatial variability in spectral reflectance on the relationship with lint yield in upland cotton is warranted.
Technical Abstract: Controlling for spatial variability is important in high-throughput phenotyping studies that enable large numbers of genotypes to be evaluated across time and space. In the current study, we compared the efficacy of different experimental designs and spatial models in the analysis of canopy spectral reflectance data collected on upland cotton (Gossypium hirsutum L.). Canopy spectral reflectance, as measured by normalized difference vegetation index (NDVI), was measured at first bloom on three upland cotton performance trials conducted in Florence, SC during 2014 and 2015. The relative efficiency and estimates of genotype effects were compared among the commonly used randomized complete block design, an a-lattice incomplete block design, a nearest neighbor adjustment, and spatially correlated error models. The incomplete block design and spatial models improved the relative efficiency (i.e. precision) of genotype effect estimates. Spatial variability in canopy reflectance had substantial effects on the selection of superior genotypes, particularly at 5 and 10% selection intensities. These results suggest that the use of more complex experimental designs and a posteriori statistical procedures should be considered to minimize error introduced by spatial variability in data collected using high-throughput phenotyping systems. These findings also indicate that further research into the effects of spatial variability in NDVI on the relationship with lint yield in upland cotton is warranted.