This research examines the use of agent-based modelling for the simulation of clonal growth and recruitment in seagrasses. The Swarm Simulation System is used to build a model of spatial interactions between three seagrass species on Success Bank, Western Australia. The primary application for the model is the investigation of processes generating spatial patterns in seagrass meadows. The features of interest include the shape and structure of seagrass patches, and species composition within these patches.

The model incorporates processes of clonal growth in seagrasses, including the spatial pattern of rhizome growth, rate of horizontal rhizome extension, and the development of mono-specific patches or meadows. Recruitment from seed is also incorporated into the model, as are seagrass losses due to winter storms. Each of these characteristics can be manipulated in the model through variables set by the user. This allows the independent control of clonal growth and recruitment for each simulated species. For each species, the response to the changes in local densities of other species is also controlled through variables. The primary focus of the model at this stage is inter-specific competition for space between Posidonia coriacea, Amphibolis griffithii and Heterozostera tasmanica.

Early simulation results support the hypothesis that local interactions between individuals are important in the emergence of local and landscape-scale spatial patterns in seagrass beds. Changes in spatial patterns within simulations are similar to those recorded over a three decade period on Success Bank, Western Australia. Changes in the species composition of simulated seagrass meadows are found to be a function of the characteristics of clonal growth in each species making up the meadows. Sexual reproduction and recruitment generate a second, more diffuse, spatial pattern which is superimposed on the more dominant pattern of rhizome growth.

Preliminary sensitivity testing indicates that the model is robust, and variables can be manipulated to produce subtle changes in patterns of growth. More rigorous testing of the model will be possible in future work investigating variation between the structure of seagrass beds at different sites in Western Australia. In the present study, simulations have been run for individual species alone, and for two competitive scenarios involving the three species. The simulations have been effective in reproducing the characteristic linear pattern of growth found in Posidonia species in Western Australia. The degree of agreement between simulation outcomes and field observations is discussed with a view to future applications of the model and improvements to the model where inadequacies have become apparent.