Comparison of single-point alfalfa yield models based on visual disease intensity and remote sensing assessments

Abstract

Field experiments were conducted to compare alfalfa yield models based on visual disease intensity and remote sensing assessments. A broad range of foliar disease levels in alfalfa was achieved by applying either one or two sprays of five different fungicides at Ames and Nashua, Iowa, United States. Disease incidence, disease severity, percent defoliation, and the percentage of sunlight (λ = 810 nm) reflected from alfalfa canopies were assessed weekly for a total of 16 alfalfa growth cycles. Single-point yield models based upon each assessment method were constructed and compared for assessments performed on the date of alfalfa harvest (H), as well as 1 to 5 weeks prior to harvest (H-1 to H-5). Significant relationships between alfalfa yield and assessment of disease incidence, disease severity, percent defoliation, and percent reflectance were obtained for 2, 2, 10, and 13 harvest dates, respectively. Significant single-point yield models based on disease incidence, disease severity, percent defoliation, and percent reflectance explained 43% to 55%, 40% to 50%, 44% to 66%, and 53% to 91% of the variation in alfalfa yields, respectively. Thus, this study demonstrated that remote sensing assessments had a better relationship with yield compared with the three visual disease assessment methods. Other key findings in this study were that damage coefficients relating percent defoliation to alfalfa yield increased nonlinearly as the site-specific attainable yield increased. This indicates that different locations, alfalfa growth cycles, and years each required a different yield model. Moreover, as alfalfa potential yield increases, the greater the return on investment would be from applying a fungicide or biocontrol agent that effectively controls foliar diseases of alfalfa. When data were combined across locations, alfalfa growth cycles, and years, standardized defoliation explained 52% of the variation in standardized yield, whereas standardized reflectance explained 70% of the variation in standardized yield.