Implication of modelling choices on connectivity estimation: A comparative analysis

TL;DR

This study compares different landscape connectivity modelling approaches, highlighting how choices in landscape characterization, estimation methods, and indices influence habitat connectivity assessments and landscape planning.

Contribution

It provides a comprehensive comparison of multiple connectivity modelling approaches and emphasizes the importance of landscape characterization and stochasticity in connectivity analysis.

Findings

Landscape characterization greatly influences connectivity outcomes.

Different models reflect distinct assumptions about movement pathways.

Using multiple indices captures various facets of connectivity.

Abstract

We focus on connectivity methods used to understand and predict how landscapes and habitats facilitate or impede the movement and dispersal of species. Our objective is to compare the implication of methodological choices at three stages of the modelling framework: landscape characterisation, connectivity estimation, and connectivity assessment. What are the convergences and divergences of different modelling approaches? What are the implications of their combined results for landscape planning? We implemented two landscape characterisation approaches: expert opinion and species distribution model (SDM); four connectivity estimation models: Euclidean distance, least-cost paths (LCP), circuit theory, and stochastic movement simulation (SMS); and two connectivity indices: flux and area-weighted flux (dPCflux). We compared outcomes such as movement maps and habitat prioritisation for a rural landscape in southwestern France. Landscape characterisation is the main factor influencing connectivity assessment. The movement maps reflect the models' assumptions: LCP produced narrow beams reflecting the optimal pathways; whereas circuit theory and SMS produced wider estimation reflecting movement stochasticity, with SMS integrating behavioural drivers. The indices highlighted different aspects: dPCflux the surface of suitable habitats and flux their proximity. We recommend focusing on landscape characterisation before engaging further in the modelling framework. We emphasise the importance of stochasticity and behavioural drivers in connectivity, which can be reflected using circuit theory, SMS or other stochastic individual-based models. We stress the importance of using multiple indices to capture the multi-factorial aspect of connectivity.