Predictors of vascular plant species richness in Poland

Using machine learning techniques, we modelled vascular plant species richness (SR) at a landscape scale (10 × 10 km squares) in Poland, for four plant groups: native species, red-list species, archaeophytes, and neophytes. As explanatory variables, we used environmental characteristics (soil, land relief, and climate), human footprint measures (land use, agricultural intensity, human population density, communication network), and palaeoecological factors (glaciation range). The influence of the individual factors on SR varied considerably, and the different species groups also differed in their reaction to particular drivers. The best predictors of SR, regardless of the species group examined, were potassium content and clay fraction in the soil. This is the result of Pleistocene geological processes: most soils in Poland are of glacial origin, and thus sandy and K-poor. The SR of native taxa increased almost linearly with increased K and clay content, up to ~150 mg·kg^–1 K, and ~15% clay, and then stabilized, with further SR increases depending on other factors. Regarding these characteristics, the soils in Poland are more similar to those in Fennoscandia than to those in neighbouring countries like Czechia and Germany. The positive correlation of the SR of all species groups with population density results directly from anthropogenic influence for neophytes, but for native species, it results rather from humans intentionally choosing habitats promoting high plant SR when settling in prehistory. As a result, the environmental and palaeoecological variables explain native SR, while the anthropogenic variables were necessary for neophyte and archaeophyte richness. Besides the edaphic variables, a positive influence on native and red-list species richness was land relief diversity. The influence of pH variability, CaCO₃ content, water bodies presence, and other typical factors beneficial to biodiversity was rather minor. We did not find any direct effect of intensive agriculture on SR. It could be related to the dataset characteristics: the basic spatial units (100 km²) are not suited for tracing the agricultural effect, and/or the SR data represented mainly the situation at the end of the 1990s and did not reflect the rapid agricultural intensification resulting from Poland’s accession to the EU.