MODELING OF LONG-TERM DYNAMICS OF ABUNDANCE OF LARVAE POPULATIONS OF CHIRONOMUS PLUMOSUS (L.) AND CH. ANTHRACINUS ZETT. WITH THE USE OF ARTIFICIAL NEURAL NETWORKS (LAKE KRASNOYE, KARELIAN ISTHMUS, LENINGRAD REGION)

Abstract

This article presents the results of a study of long-term series (1964 -1983) of population dynamics of larvae of two species of chironomid of Chironomus plumosus L. and Ch. anthracinus Zett, which dominate the zoobenthos community of the Krasnoye Lake. The main goal of this work is to develop multiple regression models of time-series to predict e dynamics of the number and biomass of populations of two types of chironomids and to assess their predictive ability. The long-term dynamics series were analyzed using the methods of spectral (Fourier) analysis, which made it possible to determine their main harmonic components and evaluate the statistical relationships with the harmonics of the dynamics of chlorophyll concentration “a” in the lake. The article presents the results of forecasting the dynamics of abundance and biomass of chironomid larvae by two types of regression models: using linear multiple regression and regression models using the technology of artificial neural networks. As the results showed, to simulate and predict the time series of the population dynamics of chironomid larvae in the Krasnoye Lake, regression models based on the technology of artificial neural networks turned out to be much more accurate in predictions compared to linear multiple regression models.