ALMaSS, an agent-based model for animals in temperate European landscapes
Introduction
The animal, landscape and man simulation system (ALMaSS) was designed as a predictive tool for answering policy questions regarding the effect of changing landscape structure or management on key species in the Danish landscape. The goal of correctly predicting local interactions between animals and their environments in detail is perhaps impossibly remote. Paradoxically, unlike other disciplines, e.g. economics, the idea of prediction using best available knowledge, albeit imprecisely, is often an anathema to biologists. In contrast, in applied sciences answers are often required and hence the creation of ALMaSS. ALMaSS is an ecological simulation system, and is an attempt to make use of the best available knowledge to develop predictions under the understanding that the reliability of predictions may be hard to quantify. ALMaSS was designed to be comprehensive enough to tackle some of the questions typically posed by government agencies and pressure groups to applied research institutes. These questions often consist of wanting to evaluate the effect of changing land-use or management. However, these questions will usually implicitly incorporate questions of spatial and temporal dynamics, which given the complex structures of most European landscapes, are difficult to handle. A crucial element of this approach is the idea that local conditions in space and time exert a major influence on the success of individual animals, and that this has a population-dynamic impact. This justification supports the use of individual-based models (IBMs; Huston et al., 1988), and thus, ALMaSS is an individual-based modelling system. However, the extent to which the individuals in ALMaSS react to their environment and remember past events (if often only physiologically), make the term IBM a little imprecise. Rather we prefer the term agent-based model (ABM). The term agent comes from computing science and has been defined variously, however, the definitions of Huhns and Singh (1998) “agents are active, persistent (software) components that perceive, reason, act, and communicate”, and Franklin and Graesser (1997) “An autonomous agent is a system situated within and a part of an environment that senses that environment and acts on it, over time, in pursuit of its own agenda so as to effect what it senses in the future”, embody the basic properties. It is in this way that animals in ALMaSS act, via sensing information from their local surroundings and making behavioural decisions. They can communicate (e.g. via breeding, or territory disputes), reason (e.g. habitat selection, movement), and act to pursue their own agenda of survival and reproduction. Thus, ALMaSS is a system based on the behavioural ecology of the animals in question. The use of agents or multi-agent systems (Ferber, 1999, Bousquet et al., 2001) requires the development not only of the agents themselves, but also of their environment. It is typical of the IBM literature that the emphasis is placed on the animals being modelled (e.g. Railsback, 2001), with the environment into which the models are placed receiving scant attention. But, if we follow the IBM paradigm and accept that changes in environmental conditions over space and time exert an impact on our results, then it is important to model these changes accurately. If we can combine both behaviour and environment at sensible scales, we will approach what Lima and Zollner (1996) term “a behavioural ecology of ecological landscapes”, and perhaps improve our predictive ability.
This paper describes ALMaSS emphasising its landscape representation and presents some example simulations of field vole dynamics investigating the effect of altering landscape structure and management.
Section snippets
Model construction
ALMaSS was constructed in C++ using the OOP paradigm (Harmon, 1993, Booch, 1994). Agent-based toolkits such as SWARM (Minar et al., 1996) were originally considered for the construction of the system, as was a specially designed visual programming language, Viola (Topping et al., 2003). However, both approaches were abandoned due, in the case of toolkits because of a lack of efficiency and flexibility, and in the case of Viola due to difficulties of usage. In both cases, the learning curves for
Model testing
The sensitivity analysis was carried out in two phases. The first phase (plausibility) consisted of using the visual interface to the model to monitor individual vole behaviour in the landscape and to use vole ecologists to verify this. The second phase was a sensitivity analysis and comprised of altering the main vole parameters by ±5, ±10 and ±20% and assessing the effect on the population. This procedure identified HQT 1–3 to be the most sensitive parameters. In particular changes in HQT 1
Discussion
Parrott and Kok (2002) in developing a very detailed physiological model of a generic ‘mammal’ in a spatial environment argue that when animals have been in focus, it is usually at the expense of a detailed environment (e.g. Henein et al., 1998, Hendry et al., 1997). They characterise ABMs as usually representing animals as abstract entities with little resemblance to biological organisms (e.g. Holland, 1975, Hraber and Milne, 1997). A detailed physiological model is probably not needed when
Acknowledgements
This research was sponsored by ‘Changing Landscapes—Centre for Strategic Studies in Cultural Environment, Nature and Landscape History’ and by the ARLAS centre under the ‘Area usage: the farmer as a landscape manager’ programme. Our thanks to Jørgen Olesen for the crop modelling, to Poul Nygaard Andersen for GIS work, Peter Lange for the field verification and digitising, and to Ditte Holm Andersen for running the sensitivity analyses.
References (50)
- et al.
TIGMOD: an individual-based spatially explicit model for simulating tiger/human interaction in multiple use forest
Ecol. Model.
(2001) - et al.
Multi-agent simulation of group foraging in sheep: effects of spatial memory, conspecific attraction and plot size
Ecol. Model.
(2001) - et al.
A two-dimentional individual-based model of territorial behaviour: possible population consequences of kinship in red grouse
Ecol. Model.
(1997) - et al.
Community assembly in a model ecosystem
Ecol. Model.
(1997) - et al.
Towards a behavioural ecology of ecological landscapes
Trends Evol. Ecol.
(1996) Individual-based models and the individual-based approach to population ecology
Ecol. Model.
(1999)Landscape dynamics in farmland
TREE
(1988)- et al.
A generic, individual-based approach to modelling higher trophic levels in simulation of terrestrial ecosystems
Ecol. Model.
(2002) Concepts from complex adaptive systems as a framework for individual-based modelling
Ecol. Model.
(2001)- et al.
The extinction risk of a population exploiting a resource
Ecol. Model.
(1999)
Individual-based modelling in ecology: what makes the difference?
TREE
Landscape pattern and productivity effects on source-sink dynamics of deer populations
Ecol. Model.
Shifting spacing behaviour of male field voles (Microtus agrestis) over the reproductive season
Ann. Zool. Fennici
Counter-strategies to infanticide in mammals: costs and consequences
Oikos
A quantitative analysis of shoot phenology and dominance in herbaceous vegetation
J. Ecol.
Multiagent simulations of hunting wild meat in a village in eastern Cameroon
Ecol. Model.
Long and short term changes in gonadal activity of field voles and bank voles
Oikos
Effects of an attractive sink leading into maladaptive habitat selection
Am. Nat.
Predation as a regulating factor on small rodent populations in southern Sweden
Oikos
Density-related home-range size and overlap in adult field voles (Microtus agrestis) in southern Sweden
J. Mamm.
Predicting extinction times from environmental stochasticity and carrying capacity
Conserv. Biol.
Cited by (132)
A meta-analysis of ecotoxicological models used for plant protection product risk assessment before their placing on the market
2022, Science of the Total EnvironmentSupporting non-target arthropods in agroecosystems: Modelling effects of insecticides and landscape structure on carabids in agricultural landscapes
2021, Science of the Total EnvironmentEuropean hares do not avoid newly pesticide-sprayed fields: Overspray as unnoticed pathway of pesticide exposure
2020, Science of the Total EnvironmentCitation Excerpt :All statistical analyses were carried out in R 3.2.5 (R Core Team, 2013). We used one of the most comprehensive agent-based risk assessment models available, the Animal, Landscape and Man Simulation System (ALMaSS) (Topping et al., 2003), with addition of the ToxImpact module enabling detailed simulations of pesticide exposure and toxicology (Topping et al., 2005). ALMaSS combines a landscape simulation model and an animal population model, which interact with each other.
Mitigating bioenergy-driven biodiversity decline: A modelling approach with the European brown hare
2020, Ecological Modelling