Simple models for complicated epidemics: Epidemiologically-based surveillance for early detection of invading plant diseases

Taylor, Thomas

Abstract

Plant disease epidemiology is the study of plant pests and their related diseases in populations of plants over time and space. In plant disease epidemiology, mathematical and computer simulation modelling is often deployed to provide insight into the factors that drive spread, as well as to identify effective intervention strategies. A key intervention for the successful eradication and control of invasive plant pests is surveillance for early detection of invading populations. Models have been used to address outstanding research questions, such as what prevalence will a pest have reached when first detected and how much surveillance resources need to be allocated for early detection. This thesis addresses how generalisable early detection models are when applied to realistic epidemiological scenarios. Here I show that epidemiological parameters including the dispersal ability of the pest, landscape heterogeneity, detection assay sensitivity and surveillance intensity influence the degree to which a simple epidemic model, termed the ‘rule of thumb’, can be used to predict detection-prevalence in complex epidemics. I also apply both the rule of thumb and a spatially-explicit stochastic epidemiological model to the case study of Oak Processionary Moth (Thaumetopoea processionea). These results indicated in particular that the rule of thumb is less accurate for short wavefront (spatially-compressed epidemic spread), highly virulent (infectious) pests where the frequency between surveillance rounds is long. These findings show that the rule of thumb benefits from increased surveillance frequency because the rule of thumb assumes constant sampling efforts in a non-spatial context with 100% detection rate. The findings also indicate that the effects of landscape heterogeneity are largely mitigated in the context of high distance pest dispersal and that a modification to the rule of thumb can increase generalisability drastically. I anticipate that the findings of this thesis will demonstrate that early detection models can be broadly applicable to a range of diseases when dispersal distances of pests are high, virulence is low, landscapes are largely homogenous and detection sensitivity is high.

• Thesis Type: Thesis
• Deposit Date: Oct 23, 2024
• Publicly Available Date: Nov 25, 2024
• Keywords: Plant, Disease, Epidemiology, Modelling
• Award Date: Oct 24, 2024