The Nuances of Plant Disease Severity Estimation Using Quantitative Ordinal Scales-Lessons Learned over Four Decades.

We revisit the foundations of the Horsfall-Barratt (HB) scale, a widely cited and applied plant disease visual assessment tool introduced in 1945, a full 37 years prior to T. T. Hebert's 1982 critique that raised concerns regarding the scale's rationale, particularly its reliance on the Weber-Fechner law and visual perception assumptions. Although use of the HB scale and similar ordinal scales persists, comprehensive studies have revealed that direct visual estimation using percentage scales often proves more accurate and reliable. Challenges remain, such as biases due to estimator subjectivity and the potential for misclassification. The logarithmic assumptions of the HB scale have been debunked, and the importance of choosing appropriate interval sizes and numbers of classes in developing ordinal scales is emphasized. Analyzing ordinal scale data appropriately is crucial, and recent advances offer promising methods that reduce type II error rates. The closely related disease severity index is noted to have its shortcomings and potential for misuse. The letter underscores the need for continuous refinement and critical evaluation of disease assessment methodologies.

From Detection to Protection: The Role of Optical Sensors, Robots, and Artificial Intelligence in Modern Plant Disease Management.

In the past decade, there has been a recognized need for innovative methods to monitor and manage plant diseases, aiming to meet the precision demands of modern agriculture. Over the last 15 years, significant advances in the detection, monitoring, and management of plant diseases have been made, largely propelled by cutting-edge technologies. Recent advances in precision agriculture have been driven by sophisticated tools such as optical sensors, artificial intelligence, microsensor networks, and autonomous driving vehicles. These technologies have enabled the development of novel cropping systems, allowing for targeted management of crops, contrasting with the traditional, homogeneous treatment of large crop areas. The research in this field is usually a highly collaborative and interdisciplinary endeavor. It brings together experts from diverse fields such as plant pathology, computer science, statistics, engineering, and agronomy to forge comprehensive solutions. Despite the progress, translating the advancements in the precision of decision-making or automation into agricultural practice remains a challenge. The knowledge transfer to agricultural practice and extension has been particularly challenging. Enhancing the accuracy and timeliness of disease detection continues to be a priority, with data-driven artificial intelligence systems poised to play a pivotal role. This perspective article addresses critical questions and challenges faced in the implementation of digital technologies for plant disease management. It underscores the urgency of integrating innovative technological advances with traditional integrated pest management. It highlights unresolved issues regarding the establishment of control thresholds for site-specific treatments and the necessary alignment of digital technology use with regulatory frameworks. Importantly, the paper calls for intensified research efforts, widespread knowledge dissemination, and education to optimize the application of digital tools for plant disease management, recognizing the intersection of technology's potential with its current practical limitations.

Disease Incidence-Severity Relationships on Leaflets, Leaves, and Fruit in the Pecan-Venturia effusa Pathosystem.

The most destructive disease of pecan in the southeastern United States is scab, caused by Venturia effusa. Incidence (I)-severity (S) relationships have not previously been characterized in this pathosystem, but incidence measures can save time and should have higher accuracy compared with estimates of severity. Ten scab-susceptible cultivars and seedling trees were assessed for I and S of scab on fruit (1,972 trees) and foliage (compound leaves and leaflets, 1,129 trees) between 2010 and 2014. Samples were assessed on a tree basis, and sample size ranged from 10 to 100 specimens per tree. The range in mean I and S was different depending on the organ (fruit I = 0 to 100%, S = 0 to 100%; compound leaves and leaflets, I = 0 to 100%, S = 0-10.1%, respectively). However, mean I could be 100% at a mean S < 2.0% for fruit, compound leaves, and leaflets. Both I and S data were transformed by complementary log-log prior to linear regression analysis. A linear regression model described the relationship between transformed I and S per tree for fruit (P ≤ 0.0001, R2 = 0.61), compound leaves (P ≤ 0.0001, R2 = 0.82), and leaflets (P ≤ 0.0001, R2 = 0.91) for all cultivars. The regression analysis showed significant effects of cultivar and year on the relationship between I and S; therefore, separate analyses were performed for each cultivar and year. Back-transformed predicted severity values showed that the differences among cultivars and years were negligible at severity <80%, and were generally numerically small at severity >80%. The observation that low severity persists until a high incidence of scab is achieved may limit the ability of incidence data to clearly differentiate treatment effects, even when based on the CLL transformation. But if found to be effective, and if used, it would reduce the labor requirements and result in more accurate data being obtained, as incidence estimates do not tend to suffer from the same subjective biases as do visual estimates of severity.

Plant Disease Severity Assessment-How Rater Bias, Assessment Method, and Experimental Design Affect Hypothesis Testing and Resource Use Efficiency.

The effect of rater bias and assessment method on hypothesis testing was studied for representative experimental designs for plant disease assessment using balanced and unbalanced data sets. Data sets with the same number of replicate estimates for each of two treatments are termed "balanced" and those with unequal numbers of replicate estimates are termed "unbalanced". The three assessment methods considered were nearest percent estimates (NPEs), an amended 10% incremental scale, and the Horsfall-Barratt (H-B) scale. Estimates of severity of Septoria leaf blotch on leaves of winter wheat were used to develop distributions for a simulation model. The experimental designs are presented here in the context of simulation experiments which consider the optimal design for the number of specimens (individual units sampled) and the number of replicate estimates per specimen for a fixed total number of observations (total sample size for the treatments being compared). The criterion used to gauge each method was the power of the hypothesis test. As expected, at a given fixed number of observations, the balanced experimental designs invariably resulted in a higher power compared with the unbalanced designs at different disease severity means, mean differences, and variances. Based on these results, with unbiased estimates using NPE, the recommended number of replicate estimates taken per specimen is 2 (from a sample of specimens of at least 30), because this conserves resources. Furthermore, for biased estimates, an apparent difference in the power of the hypothesis test was observed between assessment methods and between experimental designs. Results indicated that, regardless of experimental design or rater bias, an amended 10% incremental scale has slightly less power compared with NPEs, and that the H-B scale is more likely than the others to cause a type II error. These results suggest that choice of assessment method, optimizing sample number and number of replicate estimates, and using a balanced experimental design are important criteria to consider to maximize the power of hypothesis tests for comparing treatments using disease severity estimates.

What interval characteristics make a good categorical disease assessment scale?

Plant pathologists most often obtain quantitative information on disease severity using visual assessments. Category scales have been used for assessing plant disease severity in field experiments, epidemiological studies, and for screening germplasm. The most widely used category scale is the Horsfall-Barratt (H-B) scale, but reports show that estimates of disease severity using the H-B scale are less precise compared with nearest percent estimates (NPEs) using the 0 to 100% ratio scale. Few studies have compared different category scales. The objective of this study was to compare NPEs, the H-B midpoint converted data, and four different linear category scales (5 and 10% increments, with and without additional grades at low severity [0.1, 0.5, 1.0, 2.0, 5.0, 10.0, 15.0, 20.0…100%, and 0.1, 0.5, 1.0, 2.0, 5.0, 10.0, 20.0, 30.0…100%, respectively]). Results of simulations based on known distributions of disease estimation using the type II error rate (the risk of failing to reject H0 when H0 is false) showed that at disease severity ≤ 5%, a 10% category scale had a greater probability of failing to reject H0 when H0 is false compared with all other methods, while the H-B scale performed least well at 20 to 50% severity. The 5% category scale performed as well as NPEs except when disease severity was ≤ 1%. Both the 5 and 10% category scales with the additional grades included performed as well as NPEs. These results were confirmed with a mixed model analysis and bootstrap analysis of the original rater assessment data. A better knowledge of the advantages and disadvantages of category scale types will provide a basis for plant pathologists and plant breeders seeking to maximize accuracy and reliability of assessments to make an informed decision when choosing a disease assessment method.

Understanding the ramifications of quantitative ordinal scales on accuracy of estimates of disease severity and data analysis in plant pathology.

The severity of plant diseases, traditionally defined as the proportion of the plant tissue exhibiting symptoms, is a key quantitative variable to know for many diseases but is prone to error. Plant pathologists face many situations in which the measurement by nearest percent estimates (NPEs) of disease severity is time-consuming or impractical. Moreover, rater NPEs of disease severity are notoriously variable. Therefore, NPEs of disease may be of questionable value if severity cannot be determined accurately and reliably. In such situations, researchers have often used a quantitative ordinal scale of measurement-often alleging the time saved, and the ease with which the scale can be learned. Because quantitative ordinal disease scales lack the resolution of the 0 to 100% scale, they are inherently less accurate. We contend that scale design and structure have ramifications for the resulting analysis of data from the ordinal scale data. To minimize inaccuracy and ensure that there is equivalent statistical power when using quantitative ordinal scale data, design of the scales can be optimized for use in the discipline of plant pathology. In this review, we focus on the nature of quantitative ordinal scales used in plant disease assessment. Subsequently, their application and effects will be discussed. Finally, we will review how to optimize quantitative ordinal scales design to allow sufficient accuracy of estimation while maximizing power for hypothesis testing.

Probability of detecting nematode infestations for quarantine sampling with imperfect extraction efficacy.

For quarantine sampling, it is of fundamental importance to determine the probability of finding an infestation when a specified number of units are inspected. In general, current sampling procedures assume 100% probability (perfect) of detecting a pest if it is present within a unit. Ideally, a nematode extraction method should remove all stages of all species with 100% efficiency regardless of season, temperature, or other environmental conditions; in practice however, no method approaches these criteria. In this study we determined the probability of detecting nematode infestations for quarantine sampling with imperfect extraction efficacy. Also, the required sample and the risk involved in detecting nematode infestations with imperfect extraction efficacy are presented. Moreover, we developed a computer program to calculate confidence levels for different scenarios with varying proportions of infestation and efficacy of detection. In addition, a case study, presenting the extraction efficacy of the modified Baermann's Funnel method on Aphelenchoides besseyi, is used to exemplify the use of our program to calculate the probability of detecting nematode infestations in quarantine sampling with imperfect extraction efficacy. The result has important implications for quarantine programs and highlights the need for a very large number of samples if perfect extraction efficacy is not achieved in such programs. We believe that the results of the study will be useful for the determination of realistic goals in the implementation of quarantine sampling.