Standard Area Diagram Design Determines Accuracy, Speed and Sampling Efficiency in Coffee Leaf Rust Assessment

ABSTRACT

Although standard area diagrams (SADs) are widely used to reduce subjectivity in visual disease assessment, their design has often prioritised accuracy metrics. Here we developed a two‐colour, 10‐diagram SAD to aid severity assessment of coffee leaf rust caused by Hemileia vastatrix , and evaluated its performance in terms of accuracy and implications for assessment time and field sampling efficiency. Seventeen raters provided visual estimates that were either unaided, aided by an existing true‐colour 21‐diagram SAD and aided by the new SAD. Estimates using the new SAD achieved the numerically highest mean accuracy (Lin's CCC > 0.93) and bias correction coefficient (Cb = 0.97). CP (0.70) was higher than for unaided assessments (CP = 0.29) and the existing SAD (CP = 0.65). Inter‐rater reliability was consistently high across methods (> 0.90). Field evaluations confirmed that both SADs were similarly effective (Lin's CCC > 0.86). The minimum number of plants required to achieve a 95% confidence interval half‐width of ±2 percentage points increased with mean field severity, and depended on sampling design and SAD method, ranging from 3 to 8 plants in the low disease‐severity field to > 40 plants under the least intensive design in the high disease‐severity field. Predicted total field scoring time ranged from < 10 to 50 min across scenarios and the new SAD reduced assessment time by approximately 11%–13% while maintaining the same precision target. Overall, a simplified SAD can preserve statistical performance while improving operational efficiency, enabling the same level of precision with reduced in‐field disease scoring effort.