Get access

Practical tool for landscape planning? An empirical investigation of network based models of habitat fragmentation


  • Erik Andersson,

  • Örjan Bodin

E. Andersson ( and Ö. Bodin, Dept of Systems Ecology, Stockholm Univ., SE–106 91 Stockholm, Sweden, and Stockholm Resilience Centre, Stockholm Univ., SE–106 91 Stockholm, Sweden


This study presents a graph-theoretical modelling approach using daily movements and habitat demands of different target bird species in an urban context to assess: 1) habitable land cover types, 2) threshold distances between patches of habitat, 3) the required minimum accessible habitat areas and 4) the effects of barriers and stepping stones. The modelling approach is tested using empirical data from field surveys in the urban area of Stockholm, Sweden.

The results show that groups of small habitat patches can house the same species as larger contiguous patches as long as they are perceived as functionally connected by the inhabitant organisms. Furthermore, we found that binary habitat/non-habitat representations of the landscape could roughly explain the variation in species occurrence, as long as habitat was properly defined. However, the explanatory power of the landscape models increased when features of matrix heterogeneity such as stepping stones and barriers were accounted for.

Synthesis and application: in a world where forest ecosystems are becoming increasingly fragmented there is an urgent need to find comprehensive and scientifically relevant methods for managing and planning ecosystems. This study shows that: 1) groups of well placed small habitat patches can, together, be sufficient to attract birds in intensively developed areas, 2) the presented modelling approach can help identify such groups of patches, 3) matrix heterogeneity should preferably be accounted for, and 4) proper assessments of habitable land cover types are important. Finally, we argue that the modelling approach applied here may substantially improve landscape management and planning at scales ranging from whole landscapes down to neighbourhoods.