Filling evidence gaps with expert opinion: The use of Delphi analysis in least-cost modelling of functional connectivity

Assessment of landscape functional connectivity is increasingly important for planning landscape scale conservation measures. However, measuring the functional connectivity of landscapes is challenging due to the lack of data on species landscape interactions and because connectivity is species-specific. We developed parameters for a connectivity indicator using Delphi analysis, and critically examine the use of Delphi analysis in this context. To calculate the connectivity indicator we used the following parameters: maximum dispersal distance, negative edge effects of different land cover, and relative permeability of different land cover.Delphi is a technique designed to numerically synthesise expert opinion in data-poor environments and is based on repetitive questionnaires interspersed with controlled feedback. Three panels of experts were assembled, one covering each of three habitats of interest. Experts found the process challenging especially fixing exact numbers given the potential range of values. However, panels generally assigned higher permeability and low edge effects to semi-natural land cover classes, assigning low permeability and high edge impacts to more modified land cover. During the Delphi process we found that experts were prepared to alter their answers in response to feedback from the previous round. Participants’ answers which did change between rounds generally changed to approach the group median, and when they did, the associated confidence score was more likely to rise than to fall. After three rounds, answers were generally stable. Delphi proved a useful method to use to generate parameter values for the connectivity indicator, with the method particularly acceptable to stakeholders of the indicator project.

► We used Delphi analysis to generate parameters for a connectivity indicator from expert opinion. ► Experts found the process challenging given the range of potential species and values involved. ► Panels generally assigned higher permeability and lower edge effects to semi-natural land cover. ► Experts did alter their answers in response to feedback, approaching the group median. ► Delphi generated broadly-accepted least-cost parameter values, in a transparent and credible way.