Comparability and Transferability in Ecosystem-Assessment Techniques and Tools: An International Case Study.

Abstract

As environmental degradation now reaches around the globe, ecosystem-assessment techniques and tools (EATTs) are needed in new places and at physical scales that lie outside the previous boundaries of our accumulated technical experience. To meet this need many developing and less developed countries have adapted existing EATTs from the more developed world. In this case careful evaluation is required for their suitability in a new ecological context. I refer to this issue as tool “transferability.” A related issue arises in the context of inter-regional or very large-scale assessments. Since assessments occur in specific ecoregional settings, meta-analysis of accumulating national or regional assessment datasets must be free of contextual bias inherent in statistical data gathered using different methodologies, constrained by differing geographic particularities, and reflecting the responses of locally adapted biota. This is an issue I refer to as assessment data “comparability.”

My dissertation consists of six chapters treating various issues that arise when one tries to compare ecological assessment data from two very different parts of the world: in this case Michigan and South Korea. Chapter 1 introduces general background of EATT issues and case study regions. In chapters 2-5, I analyzed transferability of hydrologic modeling, biological field sampling techniques and indicator metric development. The analysis in chapter 6, used hydrologic modeling (chapters 2 and 3) and sampling method calibrations (chapters 4 and 5) to correct regional biases in both datasets. I then used residualization techniques to correct covariate biases and directly compare the response of biological communities to urban and to agricultural land use gradients. I found (1) South Korean methods were less efficient for fish sampling but more efficient macroinvertebrate sampling; (2) methodological calibration functions were required to account for these regional differences in sampling method; (3) regional ecological normalization (residualization) and rescaling proved necessary for an unbiased comparison of LU stressor-response relationships across regions. Overall, my study suggests that EATT transferability and assessment comparability are significant but under-appreciated problems in ecological assessment and that explicit correction of regional biases are necessary for comparative analysis.