While “hard” clustering may appear somewhat unrealistic in ecology, its application is of great help when one needs a typology or a decision-making tool
requiring unambiguous allocation of the sites. Fuzzy clustering is a more nuanced,
and therefore more realistic approach in most ecological situations if its purpose is to
describe similarity relationships among sites. Other very powerful methods exist that
are designed to reveal the structure of continuous data. These methods, simple and
constrained ordination, are explored in the following chapters.
4.16 Conclusion
This chapter has not covered all the possibilities offered by the large family of cluster
analyses, but you have explored its main avenues and seen how flexible this
approach can be. Every ecological research project has its own features and constraints; in many cases cluster analysis can provide rich insights into the data. The
clustering techniques themselves are numerous, as are also the ways of interpreting
the results. It is up to you to exploit this lore to optimise the output of your research.
150
4 Cluster Analysis