impingement the assessment of impact is straight- 

 forward and consists of the direct enumeration of 

 fish losses. However, for eggs and larvae subject 

 to entrainment, measurement of primary loss is 

 less direct and assessment of the resulting impact 

 at the population level (as adult fish) is much 

 more difficult. This type of assessment involves 

 two basic tasks: the estimation of organisms en- 

 trained as a fraction of the annual production; 

 and the measurement of any resulting population 

 changes. The first task involves sampling at the 

 plant discharge and often the use of hydrodynam- 

 ics models to predict the spatial distribution of 

 fish eggs and larvae in the vicinity of the plant; 

 and the second requires measuring the size of the 

 adult population in either absolute or relative 

 terms. Because fish populations are complex dy- 

 namic entities strongly influenced by the physical 

 environment, the determination of their size and 

 spatial distribution is not easy. It has long been 

 known (Gushing 1977) that marine fish stocks 

 undergo very large fluctuations in abundance, 

 both annual and long-term, primarily induced by 

 climatic and hydrographic factors. This natural 

 and erratic variability creates serious problems for 

 measuring anything but the largest changes. In 

 addition, when knowledge of the range and fre- 

 quency of long-term fluctuations is lacking, short- 

 tcnn monitoring data may show a misleading and 

 alarming population decline if observations started 

 at the begining of a downturn. ITiis rules out 

 the use of short-term empirical assessment meth- 

 ods because they address impact on a single co- 

 hort or year-class in the sense of Goodyear's 

 (1978) adult-equivalent method. An additional 

 problem inherent of the latter method and of 

 most fishery stock-assessment methods is that 

 they assume equilibrium population conditions, 

 which implies a regularity in nature that is contrary 

 to all evidence. All these problems point to the 

 need for parsimonious impact assessment ap- 

 proaches leading to the creation of a substantial 

 data base on which to build a comprehensive 

 fishery assessment model capable of predicting 

 long-term effects at the population level. Prefer- 

 ably, this model should incorporate key biological 

 processes governing recruitment, the effect of en- 

 vironmental factors known to affect year-class 



strength, and it should not require that the pop- 

 ulation be at equilibrium. 



In assessing the possible impact of MNPS op- 

 eration on the local winter flounder, NU has rec- 

 ognized the importance of larval losses on the 

 long-term stability of the adult population. The 

 problems of estimating the fraction of larvae en- 

 trained and of measuring the resulting population 

 change were examined taking into account the 

 geographic and hydrographic features of Niantic 

 Bay and the spawning and nursery areas in the 

 Niantic River. Therefore, the approach for impact 

 assessment consisted of a combination of sampling 

 programs and analytical methods leading to the 

 development of a comprehensive simulation 

 model which included hydrodynamics and popu- 

 lation dynamics submodels. The sampling pro- 

 grams and methods for estimating population pa- 

 rameters, describing larval behavior, and deter- 

 mining the stock-recruitment relationship have 

 been discussed in previous sections. The simula- 

 tion model components (larval dispersal and en- 

 trainment, and population dynamics submodels) 

 and the probabilistic risk analysis methodology 

 for long-term impact assessments are described 

 below. 



Larval dispersal and entrainment model 



As mentioned previously, one of the two basic 

 tasks in assessing entrainment impacts is to esti- 

 mate the fraction of total larval production lost 

 to entrainment. The problem here is that, given 

 the location of the plant intakes relative to the 

 spawning and nursery areas in Niantic River and 

 the prevailing tidal currents in Niantic Bay, the 

 use of the number of entrained larvae as a direct 

 loss to the locally spawning stock is not appropiate 

 for impact assessment at MNPS. An early hy- 

 drodynamics model of the area used by Saila 

 (1976) predicted that, if larvae behaved as passive 

 particles, most of those flushed out of the river 

 by tidal action would progress towards Millstone 

 Point and would continue moving in an east- 

 southeast direction untU they left the area via the 

 Twotree Island Channel. Saila and his coworkers 

 estimated that only 30% of the organisms entering 



208 



