Critical information in the assessment of disease is the amount of disease that is present. This can be measured as the proportion of a plant community that is diseased (disease incidence) or as the proportion of plant area that is affected (disease severity). Often, disease has to exceed a certain threshold before it reduces the yield of a crop, but it is usually difficult to accurately estimate the yield reduction caused by a specific disease. For example, many diseases occur on senescing tissue that would not have contributed to the yield anyway. Easier diseases to assess are those that kill whole trees in orchards or plantations, and those that destroy the actual harvested product, such as fruit or grain.
Disease and crop loss assessments are necessary for evaluating the economic impact of a disease and the benefit of particular control strategies. There is no point in implementing a control measure if it will cost more that the increased crop yield will return. The growth of the crop, its yield potential, the development of the disease and its impact on yield all have to be measured to predict the impact on yield of particular levels of disease. This information can be combined with predictions of likely disease levels to determine whether preventative treatments are worthwhile.
In the past, most crop loss assessment has been qualitative, producing vague, inaccurate and sometimes misleading data. One major problem with this is the complex nature of disease development. Rarely can disease be attributed to just one factor. Assessment of the effect of disease on crop yield normally involves five steps:
The first step to quantify the effect of disease is to develop a key that describes the growth and development of healthy plants during the growing season. It should describe development, either from sowing to harvest, in the case of annual plants, or from season to season in perennial plants. Details drawings or photographs, showing characteristics of the various stages of development, including leaf formation, flowering, fruiting and senescence are needed. Standardised growth keys have been developed for a number of crop plants, enabling comparison between different countries and different conditions.
Disease assessment methods need to easily provide objective measurements for a specific growth stage of a crop so that data from different sources is comparable, and provide an adequate sample of the crop for assessment. Whether it is disease incidence, or disease severity, or both, that are measured, depends on the nature of the disease. An "all or nothing" disease, for example, that inevitably kills any plant it infects, could be measured just by counting the number of plants that infected (disease incidence). However, in the case of a disease that causes varying degrees of damage to plants throughout the crop, a more complex measurement is needed, that assesses disease severity.The disease incidence for biotrophic pathogens can be measured by counting the number of plants, leaves, flowers etc that are infected, but the disease severity is assessed by estimating the proportion of total photosynthetic area that is diseased. While this is generally less precise and less controllable than counting individual plants, it is usually a better predictor of crop loss. Because judging the proportion of diseased leaf by eye is unreliable, disease assessment keys, showing different disease severities as blackened areas, have been devised for various crops.
To produce a disease assessment key, the development of disease over the whole disease cycle and at different stages of plant growth must be studied to make prototype standard diagrams and/or descriptions. The accuracy of the key then needs to be tested, by assessing disease severity in the field using the key, and then assessing the same samples using accurate measurement techniques in the laboratory. There are also computer-programs designed to train observers in disease severity assessment, by presenting images of diseased leaves, which the observer assesses, and comparing their result with the known level of disease in the diagram. This aims to reduce variation in results caused by different observers.
These disease assessment methods will only be accurate if performed on a representative sample of the crop. Samples of crop units (plants, leaves, fruit etc) can be taken randomly from a crop, or standard quadrats can be placed in the crop and all plants within the quadrat assessed. In a test plot, a part of the plot is usually assessed for disease. Taking samples only from the edge of a plantation will not necessarily be representative of the bulk of the crop. To determine how many samples need to be taken, it is possible to sample a number of times with progressively more samples, and find the point at which the standard error is low. A disease that is uniformly spread throughout a crop will require fewer samples for an accurate assessment than a disease that has a patchy distribution throughout the crop.
Once the amount of disease has been determined, the next step is to assess, either experimentally or statistically, the effect of different levels of disease on the crop yield. Experimental assessment involves setting up test crops in which the level of disease is controlled. Monitoring crops with different levels of disease allows the comparison of epidemic progress and crop yield under different disease conditions. A relationship between disease parameters and yield can then be formulated, allowing prediction of crop loss for a certain level of disease at a particular point in the crop's growth. This method is dependent upon the assumption that the treatments used to keep disease at a certain level have no effect on crop yield themselves. This might not always be the case. Similarly, assessments using susceptible and resistant cultivars rely on both cultivars having a similar yield to start with. Again, this might not be true, and must be determined fist in a disease-free environment. All aspects of experimental design need to be carefully considered in order to gain an accurate assessment of the effect of disease on crop yield. For example, harvesting methods under experimental conditions are often more efficient than under field conditions, giving the appearance of a higher crop yield. From crop loss assessment studies, models can be devised. Crop loss models are usually based on one of three types of disease assessment: disease at a critical point in development, disease at multiple points in development, and disease throughout crop development.
The statistical approach to assessing disease involves statistical analysis of crop yields under different levels of disease that occur naturally in the field. The levels of disease and the crop yields are monitored, but not manipulated, and then yields from different seasons or areas with different levels of disease can be compared to determine the effect of disease on crop yield. The advantage of this method is the use of outcomes of real cropping situations in the field, not experimentally manipulated crops. The disadvantage of this approach is that in the field, the conditions are not controlled, and there are many factors that could affect yield besides disease. An alternative statistical approach to crop loss assessment is the use of questionnaires, filled in by the farmers, that allow disease and pest incidence and severity to be related to the yields they produce.