I present here methods for describing how spatial distribution changes as population abundance varies. Four models for biomass-dependent spatial dynamics are described and characterized by geostatistical aggregation curves. These curves provide a simple way to choose between models when characterizing spatio-temporal variability of survey data. A test of significance is proposed based on a bootstrap resampling algorithm. The analysis is applied to two spatio-temporal series of monitoring surveys; a groundfish bottom trawl survey and a pelagic echointegration survey. Relative to the population mean, the relative histograms in both series are time invariant for medium and high observed abundances. But for low population abundance, the relative histogram is more skewed. I then discuss the use of commercial CPUE data for deriving time series of comparable abundance indices when the density histogram changes with abundance.