Since numerous biological and environmental factors are involved in a complex manner in ecological processes, data collected from field surveys or laboratory experiments in ecology are analytically complex. Appropriate understanding of ecological data, however, is critical in objectively characterizing ecological systems at issue (e.g., pollution, pest infestation) and in providing useful information for ecosystem monitoring and management.
Artificial neural networks, based on supervised and unsupervised learning, is an alternative tool for ecological data processing. While supervised learning is carried out for data estimation (e.g., prediction, revealing the environment-community causality relationships) based on a priori knowledge (i.e., templates), unsuper-vised learning is useful in extracting information from the data (e.g., ordination, classification) without previous knowledge. Especially, self-organizing maps (SOMs) based on the Kohonen network are extensively used in the extraction of information from ecological data. In this article, the principles and application of the SOM are outlined along with examples to demonstrate patterning and visualization resulting from the network.
Was this article helpful?