MODELING THE SPECIES DISTRIBUTION RANGE USING GEOSTATISTICAL TECHNIQUES (EXAMPLE OF SPHAGNUM MOSSES)

• Published in: Proceedings of the Karelian Research Centre of the Russian Academy of Sciences
• Main Author: Sergei Popov
• Format: Article in Journal/Newspaper
• Language: English; Russian
• Published: Karelian Research Centre of the Russian Academy of Sciences 2017
• Subjects: geostatistics; biogeography; sphagnum; kriging; continuous surface; Science; Q; Fennoscandia
• Online Access: https://doi.org/10.17076/bg558; https://doaj.org/article/7b8cd56bc46249718beb7b9393aa7cc6

Geostatistical techniques allow studying the distribution of objects in space and detecting the patterns of their spatial distribution. Among others, the kriging method is a powerful technique. The method allows creating continuous surfaces using point layers containing data on species occurrence. 168 points where local bryofloras have been investigated were placed on the map of the East European Plain and Eastern Fennoscandia. Continuous surfaces of Sphagnum palustre, S. centrale, S. magellanicum, S. papillosum, S. austini, S. affine occurrence were compiled. The occurrence of a species was determined on a 6-point scale: 0 - species absent, 1 - very rare, 2 - rare, 3 - sporadic, 4 - frequent, 5 - common, widespread. The types of the points' spatial distribution were determined by analyzing Thiessen polygons and values of variance between points. GRID-covers were constructed from these points for each of the species on the basis of values on the occurrence scale. The spatial resolution of the covers is 10 km per 1 pixel. GRID-covers were made by the method of ordinary kriging with spherical variogram. The integer scale of species occurrences was automatically translated into a continuous scale after the kriging procedure, and each pixel received a value. Kriging parameters were defined by the study of the experimental variograms for each species. The resultant continuous surfaces were verified by cross-validation. Reclassification of kriging-generated continuous surfaces in the integer form allowed to determine the species occurrence zones (optimum and pessimum). Zones of occurrence have identified for each of the mentioned species. Statistical and biological reliability of the simulated surfaces was estimated. The algorithm of mapping species ranges by geostatistical methods was described.