Application of artificial neural network for description of the Earth remote sensing data to identify natural habitats used by European elk, Alces alces (Linnaeus, 1758), in autumn-winter period 2023

Abstract

The paper presents the results of using neural networks of pixel-by-pixel image classification to decipher multichannel space images obtained by the Landsat 8 satellite. Natural habitats were identified based on differences in phytocenotic characteristics of land areas. The level of use of areas by elk was determined by the occurrence of elk presence indicators, which were the density of winter excrement piles (in spring) and the abundance of winged forms of the monoxenous ectoparasite of elk, Lipoptena cervi (Linnaeus, 1758), as measured by the number of individuals attacking the census taker (in the summer-autumn period). The material that forms the basis for deciphering was obtained by the route census method in different areas of the Leningrad Province. High values of the measured parameters were recorded in reforestation areas, in wet pine and birch forests, in aspen forests, on the edges of raised bogs and in overgrown fields – typical elk winter feeding habitats. Based on the data on the maximum values of the measured parameters, a positive correlation was established for the first time between the abundance of winged forms of L. cervi and the physiological activity of the host (the density of the piles of “winter” excrement). The paper also provides a valid reference to the material of schematic maps developed for a part of the territory of the Leningrad Province and Saint Petersburg. The scheme of the use of areas by elk is compiled on the basis of the relative index calculated from the measured parameters of the survey. Field observation areas and territories close to them were most reliably identified (90–95% accuracy). For areas not covered by field observations, the paper provides comments on the main inaccuracies of the schemes and the reasons for the identified discrepancies, based on an analysis of literary sources devoted to similar research topics. The results of the study indicate that the frequency of attacks of winged forms of L. cervi can be used as an additional criterion for assessing the stationary preferences of elk in the winter period using information on the period of active flight, and the decoding method mentioned in the paper can be used in limited areas of specific territories, including reserves, areas of hunting farms and district forestries.