An integrated multilevel and geostatistical approach to study the impact of environmental factors on biodiversity at different spatial and temporal scales

Nowadays, biodiversity loss and climate variability are among the most pressing threats to human societies and ecosystems. Consequently, it becomes of fundamental importance to identify methodological approaches which can jointly analyse the dynamics of biodiversity in connection with climate variations. In this context, a combination of multilevel and geostatistical models is proposed to determine the influence of environmental factors on biodiversity occurring in a spatio-temporal domain. In particular, the present study is carried out in four interlinked phases, focusing on: a) the computation of the biodiversity multilevel indicator, which considers both the diversity of species and landscape features of the area being investigated, b) the spatio-temporal analysis of potential environmental factors (such as air pollutants and meteorological variables) which could alter the biodiversity, c) the selection of the triggering and driving determinants which can affect the biodiversity, through a dedicated multilevel analysis, d) the construction of probability maps which allow to identify low or high biodiversity areas within the domain under study, by non-parametric geostatistical techniques. The empirical findings reveal a statistically significant relationship between the reduction of biodiversity and the increasing air pollutantsconcentrations. These results can be extremely useful to devise strategies that aim to safeguard both the environment and the public health according to the Sustainable Development Goals reported in the 2030 Agenda of the United Nations.