Ecoregionalization assessment: Spatio-temporal analysis of net primary production across Ontario

An ecoregionalization can be defined as a partitioning scheme that captures landscape patterns by dividing an area into hierarchically nested ecounits based on similar physiographic and ecological characteristics. In this paper, we introduce new spatio-temporally explicit methods to characterize spatio-temporal variability of static, a priori defined, ecounits by their dynamic "spatio-temporal signatures" (STSs) and assess the strength of the ecoregionalization boundaries using this information. To analyze the spatial and temporal patterns of net primary productivity (NPP) at the ecozone, ecoregion, and ecodistrict levels of the National Ecological Framework of Canada (NEF) ecoregionalization, we compute a 15-year monthly series of NPP for Ontario at 8-km by 8-km resolution based on satellite images (NOAA-AVHRR) and a light-use efficiency model. At each level of the NEF hierarchy, within-unit homogeneity of the monthly, annual, and 15-year average NPP of ecounits is characterized by the Getis statistic, and between-unit heterogeneity of these variables is characterized by the boundary contrast (squared difference across the boundary). Similarities across the levels of the hierarchy are assessed by the sum-of-squared differences of monthly, annual, and 15-year average NPP of nested ecounits. Temporal trends of NPP per ecounit are measured using Kendall's correlation coefficient. The seasonal and annual variations in the growing season, as captured by a time series of NPP aggregated to the ecodistrict, ecoregion, and ecozone level. are shown to vary across Ontario. These results indicate the potential of our spatio-temporal approach for ecoregionalization assessment based on dynamic and spatially distributed data.