Local spatial autocorrelation statistics quantify multi-scale patterns in distributional data: an example from the Maya Lowlands

Three articles appearing in the Journal of Archaeological Science chronicle the application of spatial statistics to a set of 47 dated Classic Period monuments in the southern Maya Lowlands (J. Archaeol. Sci. 17 (1990) 197; J. Archaeol. Sci. 12 (1985) 377; J. Archaeol. Sci. 20 (1993) 705). Although previous studies demonstrate that tests of global spatial autocorrelation can inform interpretations of archaeological data, they do not provide the means to characterize local concentrations of spatially dependent values, or what Kvamme (J. Archaeol. Sci. 17 (1990) 203) called “the particular nature of [the global] trend”. Here, local Moran's Iiand Gi∗statistics are applied to the same set of distributional data to quantitatively characterize spatial autocorrelation of terminal monument dates at the local, not global, scale. The results raise a number of local-scale hypotheses that, though undetected by previous global spatial analyses, might lead to a refined interpretation of the spatial distribution of dated monuments and, by extension, the reorganization of the Classic Maya. In a more general sense, the present application to dated Maya monuments serves as one example of how local spatial statistics can further strengthen distributional archaeological interpretations.