Skip to main content
ARS Home » Southeast Area » Byron, Georgia » Fruit and Tree Nut Research » Research » Publications at this Location » Publication #307013
Title: How many standard area diagram sets are needed for accurate disease severity assessment

Author
item Bock, Clive
item Hotchkiss, Michael - Mike
item Wood, Bruce

Submitted to: Phytopathology
Publication Type: Abstract Only
Publication Acceptance Date: 5/9/2014
Publication Date: 11/1/2018
Citation: Bock, C.H., Hotchkiss, M.W., Wood, B.W. 2018. How many standard area diagram sets are needed for accurate disease severity assessment. Phytopathology. 104:S3.15.

Interpretive Summary:

Technical Abstract: Standard area diagram sets (SADs) are widely used in plant pathology: a rater estimates disease severity by comparing an unknown sample to actual severities in the SADs and interpolates an estimate as accurately as possible (although some SADs have been developed for categorizing disease too). Most SADs have 3 to 10 images. We compared SADs with 3, 5, 7 and 10 images of known pecan scab severity on valves of pecan fruit; 12 raters used these to 1) estimate disease by interpolation, and 2) apply the image disease value to the sample as if it were a category. Raters also assessed the disease without any SADs. Agreement (Lin’s Concordance Correlation Coefficient, 'c) increased for raters using 3, 5, 7 or 10 SADs ('c=0.895 to 0.933) compared to estimates without SADs ('c=0.868) when interpolating an estimate. When based on categories, the estimates using 3, 5 or 7 SADs ('c=0.554 to 0.838) had less agreement to actual values compared to using the 10 SADs ('c=0.916), or when not using SADs ('c=0.868). Inter-rater reliability followed a similar pattern. Specific disease severity ranges over which SADs are chosen in relation to the range of severity in a sample might affect agreement. These results suggest as few as 3 SADs improve a rater’s ability to estimate disease by interpolation, and add to the evidence that category scales do not provide an accurate basis for estimating disease compared with a rater estimating to the nearest percent severity.