Bayesian network model informs on the impact of biotic and abiotic factors on the population characteristics of an endangered plant species

Preservation of the endangered plant species requires knowledge of the impact of environmental factors on their populations. This study was undertaken to evaluate the potential of a Bayesian network (BN) model to analyze the impact of habitat conditions on the current characteristics of Pulsatilla patens (L.) Mill. population in NE Poland. The model was based on field data collected in 2011–2015 from 47 sites in the four largest forest complexes of NE Poland. The causal network between (1) the biotic and (2) abiotic factors as well as morphological-developmental features of individuals and demographic features of populations (20 attributes, 1573 records) was made using the Bayesian Search Algorithm in GeNIe 2.0. The results indicate that the greatest impact on the population features had the number of competing species in forest undergrowth and exposition of slopes. Diagnostic testing and sensitivity analysis revealed that these factors were the major variables determining the size of the population, developmental phase, and the size of individuals. Tenfold cross-validation revealed that this model was the most effective to analyze the impact of habitat conditions on the presence of generative shoots, developmental phase, and juvenile shoots. The optimal levels of factors following from the BN modeling of the impact of habitat conditions on the characteristics of the population will allow regular growth (size of individuals) and development (life cycle of individuals) of a given endangered species.