The problem of the extinction risk assessment for harvested populations was formalized as a multicriterion optimization problem, which made it possible to apply mathematical models, simulation and optimization methods.
Field data on Azov Sea fish populations were statistically processed and computer models for extinction risk analysis were developed. A set of detailed simulation models for the Microsoft Windows operating systems have been created. In the models (the kernel part of the software), the influence of population parameters - fecundity, maturity, mortality rates, intra and interspecific competition, lower critical size, etc. were considered. Some regulated external factors, such as the intensity of river inflow and salinity (both are highly important for the brackish Azov Sea), average annual water temperature, etc., are incorporated into the models. The environmental stresses, characterized by amplitude and frequency of stochastic factors, both natural and anthropogenic, were taken into account.
All model parameters were estimated using least squares criterion, using observed time series of catches, population abundance, environmental factors.
The used modelling technique and the software developed make it possible to solve the following problems:
to evaluate the model parameters, to simulate retrospectively the actual population dynamics and thus to check the model's adequacy;
to simulate population dynamics in a stochastic environment and to perform a quasi-extinction analysis;
to maximize harvest under extinction risk constraints;
to minimize extinction risk under harvest constraints;
to carry out a Pareto analysis and to present graphically trade off between alternative optimizing criteria;
to find out the Pareto optimal intervals of fishing efforts in the two-population community model;
to work out recommendations for rational harvesting policies.
The analysis of the competing planktophagous populations demonstrated that the harvesting rate for both of them can be increased without affecting seriously the extinction risks. The optimal anchovy-sprat harvesting ratio is suggested.
The analysis of the pike-perch population showed a relatively high extinction risk value under the current harvesting rate. Other result concerns to the problem of recruits catch. It was obtained that diminution of recruit catch improves both criteria, namely total catch and quasi-extinction risk. At the same time, regarding to the extinction risk more intensive harvesting of recruits is not so dangerous as more intensive harvesting of adults.
The models are useful not only for research and fishery analysis but also for teaching. During the course of the project, the software was intensively used in a seminar for the applied mathematics students of the Rostov State University