Indicators of the impacts of habitat loss on connectivity and related conservation priorities: Do they change when habitat patches are defined at different scales?

• Indicators of connectivity depend on the scale at which habitat patches are defined.
• We evaluate how indicators change when scale varies from patch to metapatch level.
• Indicators of habitat loss impact on connectivity are poorly consistent across scales.
• Key habitat patches for connectivity are robust at patch and metapatch levels.
• Habitat reachability indicators perform better than purely topological metrics.

Delivering indicators of habitat connectivity first requires identifying the habitat units that will be treated as individual entities for spatially explicit analyses. These units can be defined at different spatial scales or hierarchical levels, from single habitat patches to aggregations of multiple neighbor patches. Many studies have assessed the scale sensitivity of landscape-level pattern metrics when changing spatial resolution or extent. However, how patch-level connectivity indicators change across hierarchical levels (independently from modifications in resolution or extent) has been largely overlooked, despite the potentially strong and determinant effects on their outcomes and final uses. We evaluated how the hierarchical level at which habitat units are defined affects two types of outcomes frequently derived from connectivity indicators: (i) the importance values (or estimated amount of decrease in landscape connectivity that would be caused by the loss of certain habitat areas), and (ii) the priority ranking (key areas to conserve to minimize connectivity loss), as given by a selected set of widely used metrics (habitat availability, network centrality, metapopulation capacity). We found that importance values can largely vary depending on how habitat units are defined, suggesting that such results may be flawed by a particular a priori selection of hierarchical levels. However, the identification of which parts of the landscape contain the key connectivity providers (priority ranking) was robust, particularly for those metrics that account for the amount of connectivity within habitat units. We conclude that current connectivity indicators based on patch removals do not allow, considering their scale dependence, to consistently assess the magnitude of connectivity decrease resulting from large-scale habitat loss, but that they can be used with much more confidence for detecting those key areas that most contribute to maintain current connectivity levels.