Abstract
Social representations (SRs) are systems of values, ideas, and practices that characterize individuals' and social groups' relationships to both their social and natural environment. Comparing SRs between places, social groups, and through time is critical to understanding how social-ecological systems (SESs) and their diverse uses are perceived, interpreted, and understood. This knowledge needs to be taken into account to achieve efficient land use management of SESs such as agricultural landscapes. People's perceptions of the landscape are increasingly studied in sustainability sciences and a growing number of studies use the SR framework for analyzing differences in SRs between stakeholders and localities or for detecting changes over time. Robust methodologies able to compare SRs are required for this purpose. In this paper, we propose a modular approach to studying SRs from words collected from free listing tasks. This approach relies on standardizing definitions of frequency thresholds commonly used to assess SR content, consensus level, and structure. We then illustrate the value of this methodological approach through a comparative study of farmers' social representations of the agricultural landscape among four contrasted social-ecological contexts in France. We show how our comparative method allows for characterizing spatial variations in SRs and identifying social-ecological factors that influence the structuration and content of SRs. Finally, we discuss our methodological progress and the implications of our results for public policies aimed at managing SESs and in particular agricultural landscapes for conservation.