A sampling methodology to facilitate biodiversity assessment in public green spaces

- Public urban green spaces reveal high levels of variability.
- We propose a sampling strategy for these spaces that reflects their diversity.
- The characterization of the spaces is based on variables relevant for biodiversity.
- The clustering procedure was finite mixture modelling.
- The method is replicable due to variables that are universal and easy to assess.

The growing concern with biodiversity loss has raised the attention on the importance of cities as habitats for a unique assemblage of plants and animals, particularly its public green spaces. Public green spaces, namely parks, gardens and green squares, are often too numerous to allow a detailed study of all of them. Due to their high heterogeneity, a random selection or a stratification based on few features would have consequences on the statistical validity of subsequent biodiversity analysis. Therefore, we aim to present a sampling methodology for public urban green spaces for the selection of a representative group that reflects the diversity of the original population. First, the stratification is based on a selection of variables considered relevant for biodiversity research and easy to evaluate, specifically total area, vegetation cover, impermeable area, water, age, dominant function and space character. Then, a clustering method, through finite mixture modelling, is applied to generate groups of similar green spaces. The application of the proposed sampling methodology was tested in Porto, Portugal. It aims to facilitate site selection for urban biodiversity surveys, in order to improve the accuracy and reliability of biodiversity analysis in public green spaces.