Why the loess-covered terraces? pre-Qin period settlement patterns driven by agricultural soil suitability, geohazards, and climate in the upper Minjiang River arid valley, Southwest China

Prehistoric settlement patterns in mountain-valley systems offer critical insights into long-term human-environment interactions. This study investigates why pre-Qin period settlements (primarily 5500–4500 a B.P.) in the upper Minjiang River arid valley were exclusively distributed on loess-covered terraces rather than on more fertile but hazardous valley bottoms or slopes. We integrated extensive pedological fieldwork, high-resolution paleoclimatic reconstruction (temperature, precipitation, and flood frequency), and an Analytic Hierarchy Process-Fuzzy Logic (AHP-Fuzzy) model to quantitatively assess soil cultivation suitability across three geomorphic units. Results show that: (1) During the Neolithic peak, the region experienced a warmer and more humid climate than today, with a MAT of ∼12.8 °C (ranging from 11.5 to 14.3 °C) and a MAP of ∼610 mm (ranging from 535 to 685 mm), yet was punctuated by high-frequency flooding (∼79 events/century). (2) Loess-covered terraces achieved the highest agricultural suitability score (0.614), attributed to their deep solum thickness (mean 76 cm), low gravel content (mean 10.9%), gentle slopes (mean 15°), and stable geomorphology, despite having lower inherent fertility than slopes and valley bottoms. In contrast, valley bottoms (0.457) and slopes (0.389) were severely constrained by frequent geohazards and poor physical properties (shallow soils, high gravel content, steep gradients). (3) The strategic preference for loess-covered terraces reflects a proactive human adaptation, demonstrating a multi-dimensional macro-trade-off that intertwined biophysical baselines (climatic shifts, geohazards, and soil workability) with complex socio-cultural dynamics (e.g., technological limitations, social identity, and defensive needs). Our findings demonstrate that under pre-industrial technological conditions, soil workability and hazard avoidance often outweighed pure soil fertility in determining long-term settlement sustainability. This integrated approach offers a replicable model for assessing human-environment dynamics in other tectonically active and climatically sensitive mountain regions.