2) There is a need to determine the individual and synergistic impacts of 

 pollutants, habitat destruction, and natural processes on egg and early 

 larval mortality of fish and shellfish in our most impacted estuaries. 



3) We need to determine the extent to which various taxa (from bacteria 

 to marine mammals) can adapt or acclimate to contaminant stress before 

 suffering reproductive problems. 



4) The physical and biological coupling between estuarine and coastal 

 waters needs to be described with particular emphasis placed on the 

 flux of contaminants and particulate matter. 



5) The concentration levels or dosages of contaminants or mixtures of 

 contaminants which cause reproductive stress in the early life stages 

 of fish, shellfish, benthic fauna, and plankton need to be investigated 

 and defined. 



6) The uptake pathways and rates, biochemical processing, and depuration 

 pathways and rates of contaminants by marine biota need to be determined. 



7) The exchange of material between the water column and the sediments 

 needs to be better defined. 



8) Sediment-organism interactions need to be better quantified with emphasis 

 on the fluxes of pollutants. 



9) Definition of vertical exchange processes in the presence of density 

 gradients and tide and wind forcing remain key areas of weakness in 

 hydrodynamic modeling. 



VII. WEAKNESSES AND STRENGTHS IN ECOSYSTEM MODELING 

 Critical Weaknesses of Ecosystem Modeling 



1) Modeled representations of many biotic and abiotic interactions are 

 not ecologically strong nor fully understood. 



2) In general, models are not subjected to a rigorous and iterative 

 process of repeated applications to verify and refine model structure 

 (e.g., variables, rates, processes). 



3) The functional form of critical process definitions are often ill- 

 derived (e.g., Q10 functions for temperature forcing on biotic process 

 rates). 



4) There is a high degree of uncertainty about appropriate levels of space 

 and time aggregation, particularly relating to pollutant-ecosystem 

 interactions . 



5) Levels of uncertainty in overall model predictions are high (i.e., 

 confidence levels are low). 



6) Defining the bounds (what is to be included and how) of the modeled 

 system for a given management application is difficult and often 

 arbitrary. 



263 



