M6 AM Suspended and Bedded Sediments and Nutrients: Exposure-response Relationships
Monday, 14 November 2005: 8:00 AM - 11:40 AM in 324-326

(KUR-1117-734799) Nutrient load-response models: improving predictions using classification and normalization.

Kurtz, J¹, Latimer, J², ¹ U.S. Environmental Protection Agency, ORD/NHEERL/GED, Gulf Breeze, FL, USA² U.S. Environmental Protection Agency, ORD/NHEERL/AED, Naragansett, RI, USA

ABSTRACT- Linking nutrient loading to ecological response measures is important for developing water quality management strategies and tools. Ecological response measures should represent entire systems, relate to nutrients, and be comparable among systems. Candidates include loss of submerged aquatic vegetation (SAV) and trophic level or food web responses, such as biomass (e.g. indicated by chlorophyll a). Estuaries have similarities that may provide clues to their sensitivity to nutrient stress, including flushing rates. ORD has developed an estuarine classification that groups US estuaries based on physical and hydrological factors that are related to flushing rates. The classification divides estuaries into eleven groups based on principal component analysis to reduce the number of variables, followed by average-linking cluster analysis on normalized Euclidean distances, or similarity indices. Box-plots for each variable were compared to determine primary variables contributing to the clusters or classes. Size of the estuary and area were the primary variables. In addition, volume, flow and salinity contributed to class separation. Progress has also been made toward developing load-response relationships. Based on data from a group of Northeastern estuaries, seagrass loss (SAV % of shoreline) was related to annual nitrogen loading r² = 0.28). Variables identified through classification were logical candidates for normalizing the data to improve the fit of load response models among estuaries. When normalized by volume class, model fit was improved (r² = 0.99) for low and medium volume estuaries, respectively. Fit of high volume estuaries was not improved by normalization, where higher flushing rates are more likely to result in less sensitivity to nutrient over-enrichment. Additional relationships between load and response will be developed and refined with classification and normalization to provide effective tools for management of our coastal waters at larger scales.

Key words: nutrients, load response models, estuaries, classification