Boundary detection in altitudinal treeline ecotones in the Spanish centralpyrenees

Spatial identification and description of ecological boundaries is fundamen tal to better understanding of treeline dynamics. Ecological boundaries acr oss two contrasting subalpine Pinus uncinata forest-alpine grassland ecoton es were delineated within the Central Pyrenees (Ordesa and Tesso sites). Bo undaries were delineated using an edge detection algorithm for two-dimensio nal data (lattice-wombling). Tree density, size-structure, growth-form, and estimated age were used to reveal spatial location of boundaries for sever al size and growth-form tree classes. Overlap statistics were applied to qu antify spatial relationships among boundaries determined for different sets of variables. The most significant and consistent boundaries were those fo r structural variables at the Ordesa site. At this site, the sequential dis position of bigger and unistemmed trees descending across the ecotone produ ced boundaries for size-structure and growth-form variables. These boundari es were located along an ordered spatial pattern (altitudinal diagonal). At the Tesso site, there were few consistent boundaries, most of which were d eveloped along the slope. Overlap statistics showed that boundaries at the Ordesa site were more spatially related than were those at the Tesso site. This result held when any set of variables was considered. The studied ecot ones describe sharp (Ordesa site) and gradual (Tesso site) structural chang es in tree populations, related to situations similar to the ecotone and ec cocline concepts, respectively. The possible environmental driving factors producing these patterns are the strong winds and reduced snow cover at hig her altitudes at the Ordesa site, and snow avalanches at the Tesso site. Bo undary detection through time in permanent plots might be a better tool for monitoring climate-change impact in the forest-alpine grassland ecotone th an the subjective location of treelines.