A proposal for an integrated modelling framework to characterise habitat pattern

  • Authors:

  • Christine Estreguil;Daniele De Rigo;Giovanni Caudullo

  • Affiliations:

  • European Commission, Joint Research Centre, Institute for Environment and Sustainability, Via E. Fermi 2749, I-21027 Ispra, VA, Italy;European Commission, Joint Research Centre, Institute for Environment and Sustainability, Via E. Fermi 2749, I-21027 Ispra, VA, Italy and Politecnico di Milano, Dipartimento di Elettronica e Infor ...;European Commission, Joint Research Centre, Institute for Environment and Sustainability, Via E. Fermi 2749, I-21027 Ispra, VA, Italy

  • Venue:
  • Environmental Modelling & Software
  • Year:
  • 2014

Quantified Score

Hi-index 0.00

Visualization

Abstract

Harmonized information on habitat pattern, fragmentation and connectivity is one among the reporting needs of the biodiversity policy agenda. This paper presents a generic, reproducible and integrated characterisation of patterns into one modelling framework. Three available conceptual landscape model components are customised, revisited and partly combined to derive a set of indices organized into four families: general landscape composition, habitat morphology, edge interface and connectivity. A harmonized mathematical description is provided for known and suggested new indices. Their unambiguous and easy computability is ensured with the integrated use of publicly available software (GUIDOS free-download software, Conefor Sensinode free software) and of newly programmed tools. An edge interface tool combining morphological analysis and a moving window landscape mosaic tri-dimensional model is presented; a ''Power Weighted Probability of Dispersal'' (PWPD) function is proposed to make connectivity indices sensitive to the landscape resistance. The methodology is demonstrated for the focal forest habitat, by using sixty-five in-situ based habitat maps from the EBONE project (''European Biodiversity Observation NEtwork''). Twelve indices are applied. A statistical analysis is then conducted using classical linear correlation and nonlinear Brownian Distance Correlation (Mastrave free software modelling library) as alternative to traditional dimensionality-reduction techniques and with an effort towards reusability in other contexts and reproducible research, by means of concise semantic array programming codelets. The results highlight the less correlated and fundamental pattern components, corroborating the hypothesized hierarchical organization of the indices into four families, and also the feasibility of reducing further the number of indices within each category.