Two Decades of Landscape Connectivity Science: a Bibliometric Analysis of Thematic Structure, Temporal Trajectories, and the Ecological Security Divide

Landscape connectivity is a central organizing concept of spatial ecology and conservation, but the breadth of systems, methods, and disciplines invoking the term obscures its intellectual structure. This study maps that structure empirically to reveal the field’s principal divisions and their trajectories. I applied bag-of-words multivariate clustering, redundancy analysis (RDA), random forest machine learning, and multivariate trajectory analysis to 6,280 landscape-connectivity papers from the Web of Science Core Collection, examining hierarchical clusters at five levels (2, 3, 6, 12, and 18 clusters). The deepest division in the field is not methodological but framework-based: a compact, rapidly growing cluster of “ecological security pattern” planning papers—concentrated in China and built on a state-led, source–corridor–node, composite-index idiom—separates from the rest of connectivity science, distinguished by a random forest with 97.7% accuracy on words including ECOLOGICAL, SECURITY, SOURCES, CORRIDORS, PATTERN, and CHINA. Finer partitions resolved a coherent methodological core of cost-distance/resistance modeling and graph-theoretic network analysis, alongside applied bodies spanning forest fragmentation, urban green infrastructure, freshwater barriers, road ecology, dispersal theory, hydrological connectivity, climate-driven range shift, corridor design, wetlands, and large-landscape conservation. Trajectory analysis showed broadly stable thematic structure over time even as the field grew, with the ecological security cluster youngest and most rapidly emerging and the resistance/network core most lexically stable. The deepest fault line in connectivity science is one of national research framing rather than methodological dispute—a structural feature with direct implications for how the field develops.