Since the beginning of agriculture, generations of farmers have been evolving practices for combating the various plagues suffered by our crops. Following our discovery of the causes of plant diseases in the early nineteenth century, our growing understanding of the interactions of pathogen and host has enabled us to develop a wide array of measures for the control of specific plant diseases.
From this accumulated knowledge base, we can distill some general principles of plant disease control that can help us address the management of new problems on whatever crop in any environment. One such set of principles, first articulated by H. H. Whetzel in 1929 and modified somewhat by various authors over the years, has been widely adopted and taught to generations of plant pathology students around the world. These "traditional principles", as they have come to be known, were outlined by a committee of the US National Academy of Sciences, 1968.
While these principles are as valid today as they were in 1929, in the context of modern concepts of plant disease management, they have some critical shortcomings. First of all, these principles are stated in absolute terms (e.g., "exclude", "prevent", and "eliminate") that imply a goal of zero disease. Plant disease "control" in this sense is not practical, and in most cases is not even possible. Indeed, we need not eliminate a disease; we merely need to reduce its progress and keep disease development below an acceptable level. Instead of plant disease control, we need to think in terms of plant disease management.
A second shortcoming is that the traditional principles of plant disease control do not take into consideration the dynamics of plant disease, that is, the changes in the incidence and severity of disease in time and space. (See: Disease Progress.) Furthermore, considering that different diseases differ in their dynamics, they do not indicate the relative effectiveness of the various tactics for the control of a particular disease. They also fail to show how the different disease control measures interact in their effects on disease dynamics. We need some means of assessing quantitatively the effects of various control measures, singly and in combination, on the progress of disease.
Finally, the traditional principles of plant disease control tend to emphasize tactics without fitting them into an adequate overall strategy.
Does this mean that we should abandon the traditional principles? Of course not! We merely have to fit them into an appropriate overall strategy based on epidemiological principles.
The important point to remember is that countless human undertakings, be they military operations, political campaigns, football games, or any other kind of organized effort, have failed, despite flawless tactics, for lack of a sound strategy. Any endeavor that requires a series of connected tasks for its completion also requires some kind of overall plan. Each individual task, no matter how skillfully executed or how successful its outcome, will not advance progress toward the final objective unless it has a coherent relationship with all of the other necessary tasks.
Examining these models, we can see that in both there are three ways in which we can reduce x at any point in the epidemic:
These, then, can be used as three major strategies for managing plant disease epidemics, and we can organize our plant disease control tactics under one or more of these overall strategies. Furthermore, by means of the model we can assess the quantitative impact of each strategy, not only by itself, but in its interaction with others.
It is easier to understand (and remember!) these concepts if we actually select different values for x0 and r, plug them into the model, and graph the outcome. This can be done easily with a calculator that has an exponential function, or with the accompanying simulation.
Clearly developing a sound disease management strategy requires enough knowledge of the biology of the pathogen and host to select the appropriate epidemiological model. It also requires at least "ball-park" estimates of the model parameters and the magnitude of the impact of each specific tactic on the initial inoculum or the apparent infection rate. Failure to adopt such a quantitative approach can lead to some embarrassing or even very costly errors. (Example)
Objectives tend to occur in hierarchies. The most general objective, in order to be realized, will have several sub-objectives that first must be successfully accomplished. Each of those sub-objectives can also have sub-objectives, and so on, in a hierarchy that can consist of several layers of objectives.
For example, suppose that our general objective (goal) is to reduce the losses caused by potato late blight in a particular field. We could construct a hierarchy of objectives as follows:
Note that this is just a slice out of a much larger hierarchy of objectives. What we, as plant pathologists or pest control specialists, might set as our most general objective would simply be an intermediary objective for a grower, who has to manage other pests, manage the crop, and be concerned with the productivity of the whole farm. At the other end of the scale, under some of our fourth-level objectives we could create a fifth level and perhaps even a sixth. For example, "Apply fungicides as necessary" would require some determination of the susceptibility of the variety that we had planted, an assessment of the inoculum available, and an assessment of the environmental conditions, perhaps requiring the use of a predictive model.
Several important pathogens of dry beans, including Pseudomonas syringae pv. phaseolicola (the causal agent of halo blight), Xanthomonas phaseoli (the common blight pathogen), and Colletotrichum lagenarium (the fungus responsible for anthracnose) are seedborne. Recommendations for the control of these diseases, therefore, always include the reduction of seed infection through some kind of "clean seed" program.
The seed for most of the dry bean production in the United States is grown in the semi-arid areas of the Pacific Northwest, where there is very little development of these important seedborne pathogens. In most years the seed produced in these areas has a vanishingly low incidence of seed infection.
In the dry bean producing areas of the central and northeastern US, however, the weather during most summers is at least moderately favorable for the development of epidemics of these diseases. By planting only western-grown seed, dry bean producers in the rest of the country can escape serious infection. Suppose, however, that for reasons of economics and politics the eastern growers decide to establish their own local certified bean seed production program. They know, of course, that they are likely to get some seed infection, but they can afford to invest a bit more in protecting the seed crop with fungicides and bactericides than they can the rest of their beans, and new technologies permit the detection of very low levels of seed infection in their certification program.
Despite the frequent use of the term "disease-free seed", zero infection is impossible, and so in any seed certification program it is necessary to establish an acceptable level of seed infection. Without getting into sampling error and sensitivity of the seed assay, which, of course, are important considerations, we can calculate the maximum allowable seed infection very roughly using our knowledge of the epidemiology of the disease(s) in question.
We begin by working backward from harvest, where we have to decide what level of disease we can allow at the end of the season. This is usually based on economic criteria and yield-loss models, and let us suppose for the sake of this example that we have determined that in the case of halo blight the final incidence of disease allowable is 25% of the plants infected.
We next have to decide which of the epidemiological models to use, and since halo blight clearly is polycyclic, we select the logistic model. Now we have to estimate the apparent infection rate of halo blight under the conditions to which the beans are likely to be exposed. (Ideally we would make several estimates of r, each under different environmental conditions, to calculate the acceptable level of seed infection under the whole range of conditions that we expect to encounter in the field.) This can be done by conducting a series of field trials or by looking up some published disease progress data. (See Estimating Model Parameters: Some Examples.) The rest is simply a matter of plugging in our estimates for r, the final disease incidence, and the length of the season into the simple exponential model and solving for initial disease incidence. (See Practical Uses of Epidemiological Models.)
What becomes painfully obvious in this case is that the maximum allowable level of initial disease incidence is so low that it is not practically achievable by seed selection alone. Our best tactic is to purchase seed produced in semi-arid environments where the level of seed infection is, in fact, exceedingly low. Many eastern bean producers could have saved themselves large sums of money by making these simple calculations.
Next: Management Simulation