A parcel-based graph to match connectivity analysis with field action in agricultural landscapes: Is node removal a reliable method?

Patch-based graphs are widely used to display and quantify landscape connectivity. They are specially relevant for decision support in land planning and biological conservation. Matching connectivity analyses with practical actions in agricultural landscapes involves considering management units rather than habitat patches. However, at a local scale, the classical method for prioritizing graph elements (node removal) using connectivity indexes such as delta IIC could be viewed as a highly contrived approach with respect to the actual changes in land use. Here we address the relevance of this method compared to simulations likely to display these land-use changes in a more realistic way. Prioritization as determined by the removal method is tested here against simulated land use changes in four scenarios (e.g. replacing grasslands by croplands) for an agricultural area in the Jura massif (eastern France) where field actions are undertaken to combat the spread of grassland rodents. The results obtained by ranking all the parcels (''enumerative'' approach) show that the removal method provides rankings similar to those obtained with the land-use change scenarios, except for the planting of hedgerows. However, defining a limited number of key parcels (''cumulative'' approach) results in different rankings whatever the scenarios. This shows that when applying parcel-based graphs to practical actions, the reliability of the removal method depends on the way the connectivity analysis is conducted. Simulating land-use changes, which is more realistic but more time-consuming, proves relevant if only a few key parcels need to be identified for actions to be conducted in the field.