Complex Systems Lab

WIST
(Weather-driven, Individual-based, Spatially explicit, Terrestrial ecosystem model)

WIST is a generally configurable terrestrial ecosystem model that is under continual development by members of the Complex Systems Laboratory, Université de Montréal. In the model, the behaviour of individual plants and animals of different species is simulated at a detailed spatiotemporal resolution, allowing for the study of the cross-scale relationships between structure and function in ecosystems. The model is unique in the field of ecosystem modelling due to its degree of configurability and the level of detail and breadth of scope with which an ecosystem can be represented.

Research with WIST is supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) and the RQCHP, Réseau québécois de calcul de haute performance. WIST has recently been ported to a parallel computing platform, permitting us to do simulations on much larger spatial scales than previously possible. We will be amongst the first users to make use of the RQCHP's new shared memory supercomputer!

Current projects / Publications / Overview of model

Current projects using WIST

Exploring the relationship between disturbance frequency and ecosystem complexity This project shows that ecosystems subject to persistent disturbance reach a critical level of complexity just before system collapse.

Study of the factors contributing to the successful colonisation of an invasive species Current research on invasive species suggests that successful invaders are often superior competitors in their new environment, however landscape configuration and disturbances may also have an impact, by creating corridors by which species can invade and/or creating new opportunities for seedling. Recent simulations with WIST show that even moderate levels of disturbance on a landscape can facilitate the colonisation of a species that would not otherwise be a sucessful invader.

Publications related to our work with WIST

Parrott, L. 2004. Analysis of simulated long-term ecosystem dynamics using visual recurrence analysis. Ecological Complexity 1(2): 111-125. Pre-print (pdf).

Parrott, L. and R. Kok. 2002. A generic, individual-based approach to modelling higher trophic levels in simulation of terrestrial ecosystems. Ecological Modelling 154: 151-178.

Parrott, L. and R. Kok. 2001. A generic primary producer model for use in ecosystem simulation. Ecological Modelling 139(1): 75-99.

Parrott, L. and R. Kok. 2000. Use of an object-based model to represent complex features of ecosystems. Proceedings of the Third International Conference on Complex Systems, New England Complex System Institute, Nashua, NH, May 21-26, 2000. InterJournal Manuscript #371. Download article here (pdf).

Overview of the model

WIST is constructed of simple, object-based representations of most of the major biotic and abiotic components that are found in any ecosystem, including a spatially explicit terrain, an atmosphere, and various plants and animals of different species. The behaviour of many of these components is driven by climate-related external forcing functions (temperature, radiation & rainfall). Each component is represented as an object, the state of which is described by properties, and the functionality of which is described by rule-based expressions. During simulation, the state of each object is computed at regular time increments, and the system’s global-level comportment arises as the aggregation of object-level events. Simulations based on the model has been shown to reproduce many of the complex features and dynamics common to all large ecological assemblages.

The model is largely mechanistic, and is based on a representation of mass flows within and between different ecosystem objects. A complete mass balance is performed during a simulation, such that all mass exchanges, however miniscule, are accounted. The main categories of objects and sources of mass flows in the modelled ecosystem are shown in Figure 1.

Figure 1: Overview of mass flows represented in the WIST model

Plant objects may represent individual plants or “lumps” of plants in the case of grasses and herbs (covering a maximum area equal to 1 terrain cell, see below). The growth of each plant object over its life cycle is modelled in a mechanistic fashion in which each plant photosynthesizes according to available light and temperature conditions and then allocates photosynthates to its various parts (leaf, stem, root, etc.) according to seasonal requirements. While plant form is not explicitly represented, light is limited according to a simple 3-layer canopy model that takes into consideration the leaf area index of plants in the canopy in the calculation of available radiant energy for any given plant. Growth may be limited if there is not enough available water or nutrients to synthesize photosynthates. The plant sub-model can be parameterized to represent plants of different species, including trees, bushes, grasses and herbs.

Figure 2: General representation of plant objects in the WIST model

Animal objects represent individual animals (currently only mammalian species have been modelled) that interact with one another and with their environment. Animals must eat in order to meet their metabolic requirements (calculated as a function of body mass, current activity, gestation, etc.). Ingested food is digested into lean or fatty mass and indigestible portions are excreted. A starving animal may metabolise its fatty mass reserves as required to stay alive. Animals may prey upon plants or other animals according to feeding preferences. The community food web is a directed graph of trophic relationships in which each species in the community, as well as detritus, carrion and seeds are represented by nodes. Edges connect two nodes for which there is a predator prey relationship. The strength of each relationship is given a relative value between 0 and 1, which is intended to represent the delectability of a prey item, where a low value means that the predator will eat the prey species very rarely (generally only if nothing else exists to eat) and a value close to 1 represents nearly exclusive predation. In addition to trophic relationships, a number of simple social rules are included. For example, reproduction is currently modelled after mammals, thus young animals stay with and are fed by their mother. Also, prey will flee from predators, and burrowing animals must select a place to build a burrow, and the choice is affected by the presence of other members of their species (allowing for density-dependent rules).

The terrain or landscape is modelled as a matrix of cells. Each terrain cell is a rectangular column composed of soil (various layers of decomposing material mixed with inorganic substrate) and water (saturated and unsaturated). The area of the terrain cells is a model parameter and determines the spatial resolution of the model (a resolution of 10 m x 10 m is currently used). While all cells have the same surface area, the height may differ according to landscape topography. During each time step, organic material in each terrain cell is decomposed, rainwater may infiltrate, and subsurface water may flow in or out to neighbouring cells.

The atmosphere of the ecosystem is also modelled as a discrete entity. The atmosphere is represented as mixture of water, oxygen, carbon dioxide and nitrogen gases in the same proportions as they are found in the Earth’s atmosphere. Animals and plants exchange gas with the atmosphere due to respiration and photosynthesis. The model ecosystem also contains a number of hypothetical mass storage chambers (“material storage”, Figure 1) in which about 60% of the total system mass resides. These chambers are intended to mimic the buffering and regulating effects of the enormous material reserves available to the Earth’s natural ecosystems (e.g., in the ocean and the Earth’s crust). Material is moved in and out of these chambers as required to maintain the partial pressure of the atmospheric gases at their set values. These chambers also serve as the source of water for rainfall and as the basin into which surplus surface runoff is collected (to avoid flooding events).

The model does not include a climate sub-model. Rather, an artificial weather generator is used to generate a realistic series of temperature, rainfall and radiation values, at a temporal resolution equal to the simulation time step. Alternatively, historic weather data from a real site may be used in place of the simulated values. The various biological activities in the ecosystem (photosynthesis, decomposition, etc.) are thus functions of the current weather conditions provided by the weather generator or data source. Weather is, therefore, a forcing function of the ecosystem, and there is no feedback between climate and ecosystem level processes.

WIST allows for the parameterization of any number of different plant and animal species, the initial population sizes of which may be defined for a given simulation. To date, the model has been parameterized to represent a hypothetical temperate climate ecosystem (annual temperature range between 2-38 ¾C) composed of up to about 25 plant and animal species interacting on a gently sloping 500 m x 500 m terrain. The time step used is 10 minutes, or 144 iterations per simulated day. Simulations based on these model configurations have demonstrated the capability of the model to generate emergent spatial heterogeneity on the terrain (starting from random species distributions), emergent and persistent quasi-cyclic temporal population dynamics, as well as many other non-trivial structures and dynamics that are characteristic of complex systems (Parrott and Kok, 2000).