Some reasons why the latent period should not always be considered constant over the course of a plant disease epidemic. (29th June 2018)

• Record Type: Journal Article
• Authors: Suffert, F.; Thompson, R. N.
• Is Part Of: Plant pathology. Volume 67:Number 9(2018)
• Page Start: 1831
• Page End: 1840
• Publication Date: 2018-06-29
• Subjects: latent period -- mathematical modelling -- plant disease epidemiology -- Zymoseptoria tritici; Agricultural pests -- Periodicals; Plant diseases -- Periodicals; 632
• DOI: 10.1111/ppa.12894
• Languages: English
• ISSNs: 0032-0862

Abstract

Abstract : The latent period is a crucial life history trait, particularly for polycyclic plant diseases, because it determines how many complete infection cycles could theoretically occur during an epidemic. Experiments in controlled conditions are generally used to assess pathogenicity and host susceptibility, and also provide the opportunity to measure the distribution of latent periods in epidemiological systems. Once estimated for one or several pairs of host–pathogen genotypes, the mean value of this trait is usually considered to be fixed and is often used 'as is' in models. This review contends that the latent period can display non‐negligible variability over the course of a disease epidemic, and that this variability has multiple sources, some of which have complex, antagonistic impacts. Arguments are developed for four sources of variation challenging the assumption that the latent period remains constant: (i) daily fluctuations in host temperature (or other organ–environment factors); (ii) nature of inoculum; (iii) host stage or age of host tissues; and (iv) intrapopulation competition and selection for aggressiveness traits. The review is focused on the wheat pathogen Zymoseptoria tritici, making use of empirical datasets collected during the first author's research projects and a targeted literature review. Such empirical epidemiological knowledge is potentially important for epidemiological modellers. While some studies have demonstrated that the distribution of latent periods around the mean value has consequences for epidemiological dynamics, it is shown here that it might also be important for modellers to account for changes in this mean value during an epidemic. These results may be critical for improving epidemic forecasting.