Mining spatial co-location patterns plays a key role in spatial data mining. Spatial co-location patterns refer to subsets of features whose objects are frequently located in close geographic proximity. Due to spatial heterogeneity, spatial co-location patterns are usually not the same across geographic space. However, existing methods are mainly designed to discover global spatial co-location patterns, and not suitable for detecting regional spatial co-location patterns. On that account, an adaptive method for mining hierarchical spatial co-location patterns is proposed in this paper. Firstly, global spatial co-location patterns are detected and other non-prevalent co-location patterns are identified as candidate regional co-location patterns. Then, for each candidate pattern, adaptive spatial clustering method is used to delineate localities of that pattern in the study area, and participation ratio is utilized to measure the prevalence of the candidate co-location pattern. Finally, an overlap operation is developed to deduce localities of (k+1)-size co-location patterns from localities of k-size co-location patterns. Experiments on both simulated and real-life datasets show that the proposed method is effective for detecting hierarchical spatial co-location patterns.