Abstracts

Diversity of the Fauna

We analyzed birds population density data in logariphmic scale (335 species from 1163 habitats of West Siberia) from zoomonitoring laboratory data base. Environment characteristics of all habitats expressed in nominal scale such as flora (geobotanical map divisions), till, building, productivity, damping, relief, zonality, provinciality, wooded rate, forest structure were analyzed also. The sample was divided on two parts: learning (863) and control (300 random chose habitats). Approximation of every species density in learning sample by multiple regression was calculated on every environment factor, which in fact coincide with average density value for every factor gradation (Kullback, 1958). The set of these approximations was processed without normalization by the principal components method for extracting approximation integrated on all factors. Approximation coefficients were used in control sample for calculation estimated species density by means of environment factors. As a quality measure of approximation and estimation the correlation coefficient square - explained variation- between estimated and source data was used.We analyzed birds population density data in logariphmic scale (335 species from 1163 habitats of West Siberia) from zoomonitoring laboratory data base. Environment characteristics of all habitats expressed in nominal scale such as flora (geobotanical map divisions), till, building, productivity, damping, relief, zonality, provinciality, wooded rate, forest structure were analyzed also. The sample was divided on two parts: learning (863) and control (300 random chose habitats). Approximation of every species density in learning sample by multiple regression was calculated on every environment factor, which in fact coincide with average density value for every factor gradation (Kullback, 1958). The set of these approximations was processed without normalization by the principal components method for extracting approximation integrated on all factors. Approximation coefficients were used in control sample for calculation estimated species density by means of environment factors. As a quality measure of approximation and estimation the correlation coefficient square - explained variation- between estimated and source data was used.

Average explained variation of approximation for 50 most widespread species is equal 51.2% (r=0.72), of estimation - 44.0% (r=0.66). Thus satisfactory density estimation of basic part of West Siberia birds population is possible by expert environment characteristics only without special investigations of habitats.

Besides the abundance matrix of these 50 species from all 1163 habitats was processed by the principal components method. First five components of birds population density occupied, respectively, 29.0%, 11.5%, 9.7%, 5.5% and 3.5% of total variation i.e. three intrinsic factors defined more than 50% West Siberia birds population density. Every component was processed as independent variable by the same scheme as species. Correlation coefficients between input values of components and estimated by expert environment characteristics on control sample was equal 0.86, 0.86, 0.71, 0.83 and 0.47, respectively. It once more emphasized that environment characteristics are the factors defined West Siberia birds population density.

LITERATURE

Kullback S. Information theory and statistics. N.Y. - John Wiley, London - Chapman & Hall.

Note. Abstracts are published in author's edition

|Home Page| |English Part|

Go to Home