Humans and water are closely connected in large river basins and form social-ecological systems (SESs). However, cross-scale effect in SESs make it difficult to identify the key forces driving human-water connections at the community scale when ignoring the landscape context. Focusing on the incongruous human-water relationships in the lower reaches of the Yellow River, we built local resident perception-based networks linking the agricultural subsystem, environmental subsystem, and cultural subsystem by distributing farmer household questionnaires and extracted 13 indicators from 7 kinds of network metrics to indicate human-water connections. We applied analysis of variance (ANOVA), random forest (RF) and multilevel linear model (MLM) methods to identify the driving forces of perception-based human-water connections among 20 factors at both the community and landscape scales. The results showed that the perception-based network indicators were mainly directly influenced by community-level driving factors, especially the accessibility of information, such as the frequency of going out, the frequency of accessing the Yellow River channel, and the information source for the national policy on the Yellow River. The influences of community-level driving factors on network indicators were affected by landscape-level driving factors, e.g., the nighttime light, population density, gross domestic product and proportion of artificial land, thus indicating indirect influences from the landscape context. These analyses and findings can enrich the methods by which social, ecological and hydrological elements are structurally linked in sociohydrologic research and highlight the cross-scale effect of the landscape context on human-water systems at the community level.
Keywords: Multilevel linear model; Network; Resilience; River basin; Social-ecological systems.
Copyright © 2022 Elsevier Ltd. All rights reserved.