Statistical comparison of scales for measuring the severity of wheat tan spot disease

Introduction
Sarri and Prescott scale (A) is one of the most popular scales for measuring foliar diseases of small-grain cereals, which is expressed as 0-9 scores. Double digit scale (AB) was created from incorporating A and B (disease severity on top leaves) scales. A and AB scales are considered as ordinal rank scales, and their recorded data (as descrete variables) can not be analyzed by parametric statistical methods. If they are suitably transformed to disease severity index (or percent of disease index = PDI), then can be analyzed by parametric methods as continous variable. The objectives of this study were to compare different methods and scales for measuring the disease severity, statistical analysis, advantages and disadvantages, and finally to introduce the appropriate method for evaluating the wheat tan spot disease caused by Drechslera tritici-repentis and other foliar diseases of small grain cereals.
Materials and methods
In this experiment, the intensity of tan spot disease on ten wheat cultivars was measured in a randomized complete block design with three replications and recorded as four variables, A, B, AB and PDI. In this study, the first variable (A) was used to show the severity of the disease based on the Surrey and Prescott scale (equivalent to the first digit of the double-digit scale), and the second variable (B) to express the contamination level of the disease severity on top leaves (equivalent to the second digit of the double-digit scale). The third variable (AB) was obtained from the combination of the first and second variables, and the fourth variable was the percentage of disease severity index (PDI). To compare variables and sselect the best ones, statistical analysis of data was performed using descriptive statistics, analysis of variance, comparison of means by LSD method, as well as correlation coefficients and regression analysis between the studied variables.
Research findings
Evaluating the skewness and kurtosis statistics of the studied four variables in this research showed that only the B and PDI variables had a normal distribution. Shapiro-Wilk and Kolmogorov-Smirnov statistics and data frequency distribution chart also showed the normality and continuity of two variables B and PDI, while the variables A and AB did not have a normal distribution and due to the gaps created between different data groups in frequency distribution chart, they can be considered as discrete variables. The results of analysis of variance indicated a significat difference between the studied treatments (wheat cultivars) in terms of two variables B and PDI, but the difference between wheat cultivars was not significant for the variables A and AB. Estimating the relationships between four variables using correlation coefficients and linear regression analysis revealed a significant relationship between the variables A and AB as well as the variables B and PDI. The results of linear regression showed that although the double-digit scale is a composite scale, there was a very strong regression relationship between its values and the variable A, while the relationship between the double-digit scale and the variable B was weak. On the other hand, PDI, which is the most complete variable to describe the disease severity, showed a strong regression relationship with the second digit of the double-digit scale (variable B), but its relationship with the first digit or numerical value of the double-digit scale (variable A) was weak.
Conclusion
The results of the statistical analysis of this research showed that the expression of disease severity (untransformed data) in terms of the variables B (disease severity on the upper leaves) and PDI (disease severity index) had more statistically appropriate than the variables A (Sari and Prescott scale) and AB (double-digit scale). Therefore, improving and upgrading the Sari and Prescott scale to a double-digit scale can only be effective in practice if it is correctly transformed into a comprehensive disease index (such as PDI), otherwise there will not be a big change in the quality of disease assessment. It is concluded that in assessing the severity of wheat leaf diseases, after determining the disease severity based on the A or AB scale, the data must firstly be converted to the variable PDI and then the analysis of variance or regression analysis should be performed.