Analysis of Long-Term Biological Data Series: Methodological Problems and Possible Solutions

Abstract

Long-term (multiyear) series of biological data have some specific features. The analysis of such data is treated as a statistical problem of the extraction of long-term signal from the general variance, which is determined by a complex of factors. An attempt is made to systematize the key points that should be considered in this problem. The main sources of data variability are considered, including the temporal and spatial variability of various scales and methodical and statistical sources. The methodological approaches to the assessment of these effects and the corresponding data correction are briefly discussed. This requires careful and reasoned data preparation and the choice of the appropriate statistical model and analytical methods. Some of these approaches are illustrated by case studies. However, the problem is too complex to have a single, universal solution that is suitable for any situation.