Use this URL to cite or link to this record in EThOS:
Title: Analysis of aggregated plant disease incidence data
Author: Samita, Sembakutti
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 1995
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
If diseased plants (or plant units) are randomly dispersed, the frequency distribution of diseased plants (or plant units) per sample may be described by a binomial distribution, and statistical analyses may be based on the linear logistic model. Since most disease incidence data do not have a random spatial pattern, the binomial distribution can hardly ever, in practice, be used to describe observed frequencies. In this study, the use of conditional probability distributions, such as the logistic-normal binomial distribution, for such data is illustrated. Both descriptive distribution fitting and statistical modelling are discussed. The study evaluates several methods for analysis of incidence data which do not exhibit a random spatial pattern. Some of these methods are applied to plant disease data for the first time. A method of choosing between the different analyses is discussed. All the techniques are illustrated using examples and, as an application, survey data collected on pineapple wilt disease in Sri Lanka are extensively studied. As an alternative method of describing disease incidence data with a non random spatial pattern, the use of two-dimensional distance class (2DCLASS) analysis was evaluated using the same survey data. 2DCLASS analysis is widely accepted in plant disease epidemiology as a method of analysing non-random spatial patterns when the observations are made as presence or absence of the disease on individual plant basis. We demonstrate the possibility of using quadrat-based data in 2DCLASS analysis. We investigate the use of 2DCLASS analysis as a methodology and find some drawbacks with this technique, which are discussed in detail. Moreover, this study introduces a new parameter in the 2DCLASS analysis called Scaled Core Cluster size, that may be more suitable to use for comparison of datasets of different sizes.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available