• fishing;
  • population;
  • quasi-extinction;
  • simulation model;
  • stochasticity

1. We explored harvested fish populations in which an extinction risk exists as a consequence of a notable drop in abundance. An adequate formulation of the optimal harvesting problem must account for several criteria. Economic maximization should be balanced against minimization of extinction risk. In addition, environmental stochasticity should be taken into consideration to obtain a realistic description of the population dynamics.

2. The approach to multicriteria optimization of the harvesting policy is illustrated by models of two commercially harvested planktophagous fish found in the Azov Sea in Russia. A stochastic simulation model for the community of competing anchovy Engraulis encrasicholus and sprat Clupeonella delicatula was developed and investigated. The parameters of the model were estimated on the basis of time series of population abundance and environmental factors that influence reproduction.

3. The model was used to assess numerically the extinction risk of this exploited community. A Pareto approach was used to treat the optimization problem with two criteria: maximizing the total catch and minimizing the extinction probability. No single solution exists but a set of Pareto-optimal fishing strategies was defined, i.e. a set of trade-off solutions. These various solutions can easily be compared by fishery managers and experts before making final decisions.

4. It was found that the harvesting strategy that is currently applied is quite efficient since it is very close to the Pareto-optimal set of solutions. Our recommendation is that these two populations should be considered as a community. Specifically, by increasing the harvest of endemic sprat and decreasing the harvest of the less productive migratory anchovy, fishermen will obtain a higher total harvest with only a slightly higher extinction risk.