The University of Southampton
University of Southampton Institutional Repository

Simulating Rodent Populations using Neural Networks

Dickman, jon and Garratt, paul (2003) Simulating Rodent Populations using Neural Networks At MT'2003 Conferencia Internacional de Gerencia y Tecnologia, Cuba.

Record type: Conference or Workshop Item (Paper)


Agricultural businesses must manage pests effectively. Populations of pests and other economically important species e.g. microtine rodents can exhibit violent fluctuations whose causes are poorly understood. The Periodic Lethal Toxin Production, PLTP, hypothesis of Doncaster and Plesner-Jensen attempts to explain the precipitous decline phase of rodent cycles. Indirect and anecdotal observations support the hypothesis that the rodents are poisoned by plants. This explanation remains untested experimentally. This paper reports our design, development and use of a simulator to test the PLTP hypothesis in a virtual environment. The simulator incorporates six features important in the natural environment of the rodent rodents: breeding and death of the individual animals and their predators, growth of their food plants, selection of food plants by the rodents, predation on the rodents, migration of the rodents and predators and evolution of toxin resistance in individuals. A mathematical treatment of so many factors would be impossible. The simulator models the behaviour of individual animals active in a grid of zones where changing densities of rodents, plants and predators comprise the virtual environment. The individuals use a neural network to make important decisions such as choice of food and migration routes through the grid. The neural network responds to factors such as food availability and predator threat. The behavioural response to the environment evolves by means of a genetic algorithm applied to the virtual chromosomes of the individual rodents. The individuals grow and breed at rates proportional to their food intake. Predation reduces the population to provide a turnover of individuals within the system. The inclusion in the model of toxic plants and toxin resistance allows testing of the PLTP hypothesis. An individual rodent can spontaneously gain or lose resistance to plant toxins by mutation. An individual can inherit resistance if either parent has it. Modelled plants produce toxins when over-grazed, as observed in nature. The discrete event simulator executed successfully exploring the dynamics of rodent populations to test hypotheses which are pre-conditions to PLTP such as: do the rodents feed on toxic plants when the density of preferred plants is below a critical thresh hold? The model supported and refuted several similar hypotheses in the virtual environment. The simulation exercise replicated the cyclical population changes observed in natural rodent populations. For some cases involving migration across the zones in the system the neural network appeared to give authentic results: the cyclical fluctuations in population density becoming synchronised. However in more complex situations the model tended to a non-cyclic equilibrium. The novel simulation design combining neural networks and genetic algorithms proved to be very powerful. The model programmed in an extensible object-oriented manner allowed simulation of a variety of situations. While results fell short of validating the PLTP hypothesis in full, the use of a neural network to model decision-making by animals under study holds great promise for future investigations.

Microsoft Word garrattp.doc - Other
Download (6MB)

More information

Published date: 2003
Additional Information: Event Dates: april2003
Venue - Dates: MT'2003 Conferencia Internacional de Gerencia y Tecnologia, Cuba, 2003-04-01
Organisations: Electronic & Software Systems


Local EPrints ID: 262654
PURE UUID: b6f06f30-4683-40eb-aa15-cd4ec17f0abc

Catalogue record

Date deposited: 01 Jun 2006
Last modified: 18 Jul 2017 08:49

Export record


Author: jon Dickman
Author: paul Garratt

University divisions

Download statistics

Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.

View more statistics

Atom RSS 1.0 RSS 2.0

Contact ePrints Soton:

ePrints Soton supports OAI 2.0 with a base URL of

This repository has been built using EPrints software, developed at the University of Southampton, but available to everyone to use.

We use cookies to ensure that we give you the best experience on our website. If you continue without changing your settings, we will assume that you are happy to receive cookies on the University of Southampton website.