How space and society shape Linpan village landscapes in Sichuan

China’s most celebrated rural landscape — a mosaic of farmhouses tucked inside cool, dense islands of bamboo and broadleaf trees — is being redrawn, and a new study claims to have identified the invisible machinery doing the drawing. Writing in the journal Regional Environmental Change, a team led by Lei Zhang of the College of Landscape Architecture at Sichuan Agricultural University reports that the famous linpan villages of western Sichuan are not simply dissolving under the wheels of urbanization, as is often assumed. Instead, they are being deliberately reorganized by an interacting web of actors — government agencies, enterprises, village cooperatives and individual landholders — whose negotiations over land rights, money and authority are translated, almost line by line, into the arrangement of fields, forests and buildings on the ground. By fusing thirty years of satellite imagery with landscape ecology metrics and the sociology of networks, the researchers have delivered something relatively rare in landscape science: a combined quantitative and conceptual demonstration that the shape of a rural landscape is the visible fingerprint of how it is governed. The work also turns a decades-old sociological theory into a practical instrument for reading maps — and, potentially, for steering them.

Linpan, a term best rendered as “forest-enclosed” settlements, are the signature cultural landscape of the Chengdu Plain, one of the world’s oldest continuously cultivated irrigated farming regions. In the classic linpan pattern, each farmstead sits within or beside a dense clump of trees, and the settlement as a whole scatters across the plain as a fine-grained patchwork of woodlots, cultivated fields, irrigation ditches and homesteads, rather than clustering into a compact nucleated village. That architecture is far more than picturesque. The groves buffer temperature and humidity, shelter birds and other wildlife, slow and intercept runoff, and anchor a sense of local identity that has survived centuries of agricultural intensification. But the plain is also the demographic and agricultural heart of Sichuan Province, and Pidu District, where the new study is set, lies squarely in the expanding metropolitan orbit of Chengdu, one of China’s largest megacities. The researchers frame the resulting pressure as a triple force: urbanization, policy adjustments and ecological restructuring, all acting on the same patchwork at the same time. Understanding which force wins, and how, has become an urgent question for anyone hoping to keep this landscape alive.

To disentangle those forces, the team focused on a single emblematic case: Anlong Village in Pidu District, Chengdu, spanning the period from 1995 to 2025. The design stacks three analytical layers on top of one another. The first is multitemporal remote sensing, which converts satellite data into land-use maps for successive benchmark years, allowing researchers to measure precisely how much land moved among cultivated fields, construction footprints and forest cover across three decades. The second is a suite of landscape pattern indices — numerical descriptors borrowed from landscape ecology that capture not merely how much of each land type exists, but how it is spatially arranged. The third is actor-network theory, a sociological framework that treats social order as the evolving product of networks linking human and non-human actors, rather than as the output of fixed institutions. It is this deliberate coupling of spatial measurement with social theory, the authors argue, that gives the study its analytical teeth: for one real village, it becomes possible to trace how policy, power and money become geography, and how geography in turn stabilizes the social arrangements that produced it.

The technical heart of the analysis lies in those pattern indices. Remote sensing classification first sorts every pixel of a satellite scene into categories such as cultivated land, forest, water and built-up area; change detection then compares classified maps across epochs to quantify gains and losses class by class. But raw areas alone can conceal what is really happening, which is where landscape metrics earn their keep. Composition metrics, such as the proportional share of each land class, describe the overall resource mix, while configuration metrics — patch density, edge density, mean patch size, aggregation and connectivity — describe the geometry of that mix. The distinction matters because a landscape can lose the same total area of forest in two ecologically opposite ways: through the attrition of many small groves, which fractures habitat corridors, multiplies disruptive edges and severs the movement of species, or through the consolidation of wooded land into fewer, larger and more contiguous blocks, which restores functional connectivity. The indices separate these trajectories with numerical precision, and they are sensitive enough to register the fingerprints of planning decisions — zoning boundaries, land consolidation projects, road alignments — directly in the geometry of the land.

Applied to Anlong Village, the metrics revealed a land-use transformation unfolding in three sharply distinct phases. In the first phase, the village operated under traditional agriculture: the inherited linpan mosaic remained largely intact, cultivated land dominated the landscape matrix, and change proceeded at the slow tempo of customary farming. In the second phase, specialized horticultural operations took hold — production was reorganized around market-oriented, specialized planting rather than subsistence field patterns, rewriting both the economic base of the village and the spatial texture of its fields. In the third and most recent phase, the landscape entered what the authors call multifunctional landscape integration, in which agricultural production, ecological space, construction land and new rural functions coexist in a deliberately curated arrangement rather than by historical accident. Across all three phases, the directional trends were unambiguous. Cultivated land area declined continuously, pressed by construction land that expanded steadily throughout the entire study period. Forest land, by contrast, did not decline monotonically: it fluctuated, expanding and contracting in step with the restructuring programs of each successive phase — a signature, the researchers argue, of active landscape management rather than passive, piecemeal loss.

The second half of the analysis explains those spatial signatures through actor-network theory, an approach forged by sociologists of science such as Bruno Latour and Michel Callon. Its central premise is deliberately radical: agency is not the monopoly of humans or institutions but is distributed across a network of actants — people, organizations, documents, contracts, money, land parcels, even trees and irrigation channels — whose interactions generate effects. Networks are assembled through translation, the process by which an actor defines a problem in terms that enroll others, aligning their divergent interests so that a collective project can move forward; a stable network, once formed, behaves like a black box that reliably produces outcomes. The researchers applied this lens to Anlong Village by reconstructing its key events, and two emerged as decisive network-forging moments: the reform of property rights and the transfer of land use rights. In an actor-network reading, these were never merely administrative paperwork. Each event redrew who could act, on which land, with what expected returns — and in doing so physically reorganized the village, patch by patch.

What emerged from that reconstruction is a core network system composed of the government, enterprises and village cooperatives, into which a wider circle of stakeholders — including individual villagers — became progressively incorporated. Crucially, the network did not run on goodwill alone. The authors identify three mechanisms that kept it functional through three decades of upheaval. The first is smooth power flows: directives, funding and information could move through the network without structural blockages, so decisions taken at one node could actually be executed at another. The second is effective interest negotiation, through which the divergent goals of officials, investors and farmers were reconciled into shared projects rather than left to collide. The third is a working set of mechanisms for resolving conflicts, which kept disputes over land, compensation and boundaries from hardening into stalemate. Where much earlier research has described rural landscape change in China as the chaotic by-product of fragmented, weakly coordinated actors, Anlong Village displays the opposite configuration: a sufficiently well-connected network that converted contention into coordination, and coordination into a coherent spatial plan.

The spatial end product of that coordination is what the authors call an orderly restructuring characterized by ecological centralization and industrial decentralization — a compact phrase that encodes a great deal of landscape logic. Ecological centralization means that the ecological substance of the linpan, its forests and associated green infrastructure, is being consolidated into larger, more contiguous and more centrally organized blocks, reversing the chronic fragmentation that ordinarily accompanies rural development. Industrial decentralization means that commercial and production functions are dispersing into distributed nodes across the village rather than piling into one dense zone. Read together, the two trends produce a landscape in which ecology gains coherence while economic activity spreads outward — an inversion of the spiral typically seen in unplanned peri-urban growth, where scattered construction shreds habitat while industry concentrates chaotically. From this, the study derives its central conceptual claim: the observed landscape pattern is, quite literally, the spatial manifestation of the governance efficacy of actor networks. Geography, in this framing, is governance made visible, and landscape metrics become an audit instrument for institutional performance.

To generalize beyond a single village, the team distilled its findings into a dynamic, coordinated socio-spatial model that binds policy intervention, spatial pattern and ecological value into one continuously coupled system, rather than treating them as separate research silos. Within the framework of China’s rural revitalization strategy, the model is presented as a decision-making basis for balancing environmental sustainability against rural development — telling planners not only what a landscape should contain, but which configurations of actors and institutions make that landscape achievable and stable over time. The study also joins a wider global conversation about traditional rural landscapes, from the hedged farmlands of Europe to other heritage mosaics across Asia, all of which face the same collision of urbanization, policy shifts and ecological restructuring. The research was supported by the National Natural Science Foundation of China under a project dedicated to identifying and conserving the historical landscape characteristics of linpan settlements, and was published on 27 July 2026 in Regional Environmental Change as Volume 26, Article 160. Its takeaway is disarmingly simple: if you want to save a landscape, first map the network that builds it.