Clustering is fundamental to pattern recognition in science, but methods vary widely and can strongly affect the patterns we detect. Non-hierarchical (e.g. K-means) and hierarchical (e.g. Ward's) clustering methods are well known to ecologists, but both force a spherical structure on groups. New to ecology, density-based methods (e.g. DBSCAN) define group structure with local densities instead of centers. This allows analysts to perceive odd distributions in noisy data. Density-based methods are found to equal or surpass center-based methods when tested on simulated data. An important problem in every cluster analysis is the choice of parameters. We describe a new method, 1:1 consensus, that optimizes partitions at multiple scales. Density-based consensus clustering should find many applications in ecology, famous for odd distributions, noisy data, and multiple scales of interest. We use it to define home ranges and illustrate important concepts with the 2-dimensional spatial point patterns of tropical tree populations. The methods are equally well applied to multidimensional and mixed data, as required for studies of functional type, niche, and community.