I think that the question should be sharpened to reflect the use of the assessment you intend to do. I assume that you are working with a fisheries expert (and a biologist) then the question becomes one that is common to environmental risk assessment. You will need a damage function (e.g., a concentration-response model), data to parameterize the function, an estimation method (e.g., the maximum likelihood estimator, MLE), and a computer program to do what I just summarized.
The problem is that the data may be limited and thus you will need to run some Monte Carlo simulation or bootstrap; make sure that the assumptions inherent to using an estimation method are met (e.g., data can be correlated, the variance may not be constant and so on). In other words, you need to have a statistician to help you.
What is the purpose of the assessment: guide policy, just do homework or project, or … ? The use of your analysis will increase the need for a protocol for the study, the choice of damage function, the data sets that you may want to use (unless they are your own data). All of these may require careful investigation and assessment of the available literature if any. If your study is novel, are you trying to identify a relationship between exposure and biological response in a fish species?
I shared with you a research proposing a multi-criteria approach based on hazard analysis which develops a case study on a real fish manufacturing process.