BSPP Presidential Meeting 1999

Biotic Interactions in Plant-pathogen Associations

Session IX - Methodology and modelling

Development of methods and models and their application to disease problems in the perennial citrus crop system
1
G. Hughes, 2T.R. Gottwald & S.M. 2S. M. Garnsay

1Institute of Ecology and Resource Management, University of Edinburgh, Scotland, U.K.
E-Mail:
ghughes@srv0.bio.ed.ac.uk
2
U.S. Horticultural Research Laboratory, USDA-ARS, Orlando, Florida, U.S.A.

The ancestral home of citrus is south-eastern Asia, but citrus is now grown throughout the world wherever the climatic moisture and temperature conditions are appropriate. Major centres of citriculture today include Argentina, Australia, Brazil, China, Cuba, Egypt, India, Israel, Italy, Japan, Mexico, Morocco, South Africa, Spain and the United States. For vector-borne citrus pathogens, quarantine situations may arise when a vector is present but the pathogen is absent; when the pathogen is present but a vector is absent; when both a pathogen and its vector(s) are absent; or when severe isolates of a pathogen or efficient vectors are absent. Quarantine procedures such as confiscation of material at ports of entry, inspection of ships cargoes at ports, and inspections at border crossings and airports can reduce the rate of introduction of exotic species that are harmful to agriculture, but international borders will always be leaky as far as pathogens and pests are concerned and efficiency of detection of introductions is often very low. Once an exotic species is introduced into an agro-ecosystem, the ability to detect it at very low incidence is fundamental to its containment. This involves the development of survey strategies that are often unique to the species being surveyed, and requires knowledge of the biology and spatial distribution of the exotic species within the crop.

The assessment of citrus tristeza virus (CTV) incidence by sampling involves laboratory assay of plant material collected in the field. CTV incidence may be assessed by sampling groups of citrus trees, recording the groups as CTV-positive (one or more infected trees) or CTV-negative (no infected trees), and then calculating disease incidence at the individual tree scale by means of a formula involving incidence at the group scale and the number of trees per group. This procedure works well when the CTV status of a tree can be regarded as independent of the CTV status of other trees in the same group. This is the case when the main vector species is Aphis gossypii and groups comprise four adjacent trees, because the spatial pattern of CTV incidence at the within-group scale can be regarded as random. However, when the main vector species is Toxoptera citricida, this simple procedure is not appropriate because the spatial pattern of CTV incidence at the within-group scale cannot be regarded as random. An alternative procedure for assessment of CTV incidence when the main vector species is Toxoptera citricida is operationally identical to that used when the main vector species is Aphis gossypii, but the calculation of CTV incidence at the scale of the individual tree is based on incidence at the group scale and effective sample size.

Virus and virus-like diseases of citrus are a dynamic system. Dispersal of existing pathogens to new areas, changes in the properties of existing viruses or vectors, and introduction of new cultivars are all factors. New problems may involve ingress of a pathogen from other hosts, or changes in vector dynamics associated with other crops. Citrus is a long-lived perennial reservoir into which viruses from outside sources can be introduced and accumulated. The impact of these ingress events is determined by the potential for secondary spread, and the pathogen and vector reservoirs in other crops. The development of detection methodology, and the deployment of this methodology in sampling protocols founded on epidemiological models, will play an important part in the management of citrus disease problems as they continue to arise.


The limits to predictability: chaos, stochasticity and hidden variables
Mike W. Shaw.

Department of Agricultural Botany, Plant Sciences, University of Reading, UK.
E-Mail:
m.w.shaw@rdg.ac.uk

Biological and environmental interactions which make pathogen populations intrinsically unpredictable will be briefly reviewed. Two causes of unpredictability will be discussed through examples: chaotic dynamics and evolutionary change resulting in different responses to the same environmental feature ("hidden variables"). The trophic relationships, spatial distributions and functional forms likely to lead to such unpredictability will be discussed.


Modelling the epidemiology of Dutch elm disease.
1
Jonathan Swinton & 2Chris A. Gilligan.

1
Department of Zoology & 2Department of Plant Sciences, University of Cambridge, Downing Street, Cambridge, CB2 3EA UK. 
E-Mail:
cag1@cam.ac.uk

Epidemiological analysis of Dutch elm disease is complicated by the interactions amongst the host, beetle vector and the parasitic and saprophytic phases of the fungal pathogen. Further difficulties arise from the long time scale of many of the demographic processes and by large distances over which epidemics occur, often in heterogeneous regions. We describe some recent modelling of the temporal and spatial dynamics of Dutch elm disease in order to: (i) identify key epidemiological parameters; (ii) predict the fate of the elm in the U.K. and (iii) analyse strategies for biological control of the disease. The model is first formulated as a simple deterministic system to describe the temporal dynamics of saplings, healthy trees, recently killed trees and dead trees colonised by the saprophytic phase of the pathogen. We use data from the detailed surveys of the Forestry Commission of the 1970s epidemic of Dutch elm disease in England to estimate key parameters for the transmission of infection and pathogenicity. The model is capable of mirroring observed data over short timescales, such as a decade, but also demonstrates over century-long timescales a wide variety of outcomes ranging from pathogen extinction to substantial loss of elm, depending on the relative magnitudes of the transmissibility and pathogenicity parameters. Some consequences of these results for the impact of the combined saprophytic and parasitic life cycle on the evolution of pathogenicity are described and some inferences about the competitive exclusion of Ophiostoma ulmi by O. novo-ulmi are explored. Further progress requires construction and analysis of a spatial version of the model and we illustrate this in the analysis of the putative control of Dutch elm disease by d-factors, hyperparasites in the form of dsRNA elements. We characterise the range of outcomes likely to follow the introduction of such an agent by modelling the resultant population dynamics as an ecological interaction between the wild-type, 'target' fungus and the hyperparasitised fungus. Finally, we discuss how the Forestry Commission data can be used to estimate the spatial scale of disease dispersal.