What should a good epidemiological question look like?
School of Biological Sciences - Plant Science Laboratory - L4 TOB2 - Earley Gate - Whiteknights - Readings - Berks RG6 6AS – UNITED KINGDOM
The question defining epidemiology is perhaps 'Why has my crop become diseased? To answer this question requires good etiology and well-defined questions about pathogen life-cycle, host, environment and management, susceptible to answer by experiment. Assumptions about etiology and life-cycle based on circumstantial evidence are often accepted and appear often to be wrong: examples from my own work include Mycosphaerella graminicola - assumed to be trash-borne, but with air-borne ascospores critical - and Botrytis cinerea - assumed to be airborne and localised, but in some crops seed-borne and systemic. With correct etiology and life cycle it is frequently assumed that all we then need to know for management is some simple function of the environment, usually as measured by commercial instruments. In practice this is extremely hard to investigate, because there are logical and/or statistical gaps between the information available and the inferences we seek; it will only be true if environment (and not host, other organisms, or intrinsic biology) is what limits pathogen populations. Statistically, there are major problems due to the correlation of environment in different locations, and the corresponding lack of degrees of freedom in what otherwise seem large experiments. Logically there are problems to do with what we measure and how we reduce an infinity of dimensions in environmental records to manageable dimensions. My examples will include the same pathogens as before: in which long-term changes in M. graminicola may be mediated by acid rain; while conditions leading to disease in plants infected by B. cinerea are as much to do with the host physiology as the environment, but the cumulated environment determines the host physiology. My conclusion is that scientifically we should recognise that in some areas our knowledge is distressingly limited; and in giving advice and in politics we should plan for the unexpected, and constantly look for the unexpected assumption we did not know we had - but which has now proved unfounded.