Computers and Electronics in Agriculture 28 (2000) 51–65 A nearest neighbour approach to the simulation of spread of barley yellow dwarf virus T.J. Chaussalet a, *, J.A. Mann b , J.N. Perry b , J.C. Francos-Rodriguez a a Caendish School of Computer Science, Uniersity of Westminster, 115 New Caendish Street, London W1M 8JS, UK b IACR-Rothamsted Experimental Station, Harpenden, Herts AL52JQ, UK Received 1 November 1999; received in revised form 10 February 2000; accepted 16 February 2000 Abstract The spread of insect vectored plant virus diseases such as barley yellow dwarf virus has traditionally been depicted as disease progress curves which represent an integration of the interactions between virus, host plant and vector. In this paper, virus spread is described by the probability of a plant becoming infected conditioned on the number of infected plants neighbouring it. This has the advantage that the influence of aphid movement can be incorporated into the definition of the probability of a neighbour becoming infected. Data were collected from an experimental field of barley (cv. Alexis) sown at a row spacing of 12 cm on 29 March 1993 at Rothamsted Experimental Station. Twelve plots, each approxi- mately 84 ×70 cm were divided into a 7 ×7 grid of cells. The plots were arranged in four blocks, each with two treatments T1 (S. aenae adults) and T2 (S. aenae nymphs) and an untreated control. The probabilities of a plant, or rather a cell which is a group of plants, becoming infected conditioned on the number of its infected neighbours was estimated from this experiment considering various neighbourhoods. These probability estimates were then used to develop visual interactive simulation models of spread on a 51 ×51 grid of cells. In all simulation models, the central cell was set as infected at the start of the simulation to match the experimental design for treatments T1 and T2. The simulations were run for a 15-week period, replicated 50 times, and the resulting infection counts were averaged. These simulations were used to estimate the rate of spread of BYDV and to perform a range of sensitivity analyses. www.elsevier.com/locate/compag * Corresponding author. Tel.: +44-171-9115000; fax: +44-171-9115089. E-mail address: chausst@wmin.ac.uk (T.J. Chaussalet) 0168-1699/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved. PII:S0168-1699(00)00116-2