PS 30-158 - Portfolio Theory, Management, and Macrosystems Biology of U.S. Forests

Abstract: Issues of scale are one of the great unsolved problems in ecology. In this research, we approach scaling of a forest ecosystem process through Modern Portfolio Theory (MPT). MPT hypothesizes that as the domain of interest grows larger, a more diverse array of forests, management types, edaphic factors, climate, and disturbances are encountered. Interannual variability of ecosystem processes is hypothetically lower when measured at coarser scales, i.e., the portfolio effect. We have limited understanding of how factors driving this phenomenon operate and interact at different scales. We must quantify the relative importance and interactions of these factors as they change across scale. We use MPT to examine how annual variability of C uptake at multiple spatial extents: site, landscape, regional, and national, is affected by the distribution and relative abundance of forest management approaches. C uptake is quantified by 22 years of Terra MODIS Yearly Net Primary Production gap-filled product (MOD17A3HGF.006) at 500-m spatial resolution for the southeastern U.S. coastal plain and piedmont (SEUS) and the forested areas of the U.S. Pacific Northwest (PNW). Four different approaches: production, preservation, ecological, and passive forest management comprise the portfolio. Earlier maps of forest management set the baseline for computational experiments that examine how the interannual mean-variance-covariance relationship changes across scales and portfolio makeup. We isolated the effect of portfolios (management) at different levels of aggregation by comparing MODIS NPP variability influenced by four experimental treatments on the abundance and spatial configurations of forest management: 1) the actual distribution of forest management types determined by continental-scale management mapping; 2) spatially random distribution of management types with statistical equivalence to the actual map, i.e., the same size-frequency distribution of management units; 3) fractional amounts of management proportions, e.g., ¼, ½, and 2x the actual management distributions; and 4) no management, i.e., removing any information about management type, effectively creating regions of passive management. We vary spatial scale in two ways: using three levels of the hierarchical Bailey ecoregions (levels 2, 3, and 4) and creating tessellations of hexagons with three sizes to cover the two regions. We calculated a 22-year mean-variance combinations of MODIS NPP at each of the different scales for each of the different experimental treatments. Preliminary outputs indicate that interannual variance is reduced as the extent of measurement increases but only very slightly, casting doubt on the predictive power of portfolio theory for modeling regional- to continental-scale ecosystem processes.