Digital Phytoindication of Soil Salinity in Dry Steppes (Republic of Kalmykia)

Soil and geobotanical studies were carried out in a landscape district within the Northern Sarpa lowland of the Caspian Depression in a zone of light-chestnut soil. The collected data made it possible to assess the variation of soil-salinity values typical for dry-steppe plant species and communities in the Republic of Kalmykia and to produce digital models for a geobotanical indication of soil salinity. Geobotanical plots, soil test pits, and boreholes up to 2 m deep were established along a 64-m transect at 1 m intervals. The pNa salinity index was measured in aqueous suspensions of soil samples (1 : 5). As a result, it became possible to determine the salinity at depths of 0–30, 0–50, and 0–100 cm for 12 plant species and 7 plant communities registered on the transect and to distinguish three plant groups based on their tolerance to soil salinity. The first group includes species confined to nonsaline soils (salt concentrations vary within a narrow range). Plants belonging to the second group tolerate a broad range of salinity values and prefer nonsaline soils. The third group consists of salt loving species (halophytes) confined to saline and highly saline soils. Of the seven plant communities registered on the transect, two occur on nonsaline (down to a depth of 2 m) soils (Stipa lessingiana + Festuca valesiaca + Artemisia lerchiana and Stipa lessingiana + Festuca valesiaca + Tanacetum achilleifolium); two others tend to occur in soils that are not saline to a depth of 50 cm (Festuca valesiaca + Artemisia lerchiana + Tanacetum achilleifolium and Artemisia lerchiana + Tanacetum achilleifolium + Artemisia pauciflora); and three communities occur only on soils that are saline from a depth of 25–50 cm (Kochia prostrata + Artemisia pauciflora, Artemisia pauciflora, and Poa bulbosa + Anabasis aphylla). The Classification and Regression Tree (CART) method makes it possible to predict the soil salinity based on the occurrence of plant species identified as predictors. The prediction accuracy is 80% for the 0- to 30-cm layer, 81% for the 0- to 50-cm layer, and 64% for the 0- to 100-cm layer. The following plants have been identified as important (rank > 60) predictors: Kochia prostrata, Tanacetum achilleifolium, Artemisia austriaca, and Festuca valesiaca. Other species feature low prediction importance (validity) values and therefore cannot be used as predictors.