Influence of farmland marginalization in mountainous and hilly areas on land use changes at the county level.

Agricultural works alter earth's surface at the largest scale among human-driven activities. Previous studies have focused more on the reclamation of natural land, however, farmland marginalization (FM), emerging as an important mean of land use changes in mountainous and hilly areas (MHAs) has always been overlooked in the background of production efficiency improvement along with urbanization and population migration. This paper examined the characteristics of the spatial-temporal distribution and conversion of marginalized farmland in the MHAs of China at county level (excluding Hong Kong, Macau, and Taiwan) from 1990 to 2020, regarding farmland in MHAs converted into non-built-up land as FM. The results showed that: (1) The total area of marginalized farmland in the MHAs was 1.03 × 10 km. The counties with larger area of marginalized farmland were concentrated around the Hu Line, and those with higher ratio were distributed in southern mountainous areas. (2) The area of marginalized farmland in each stage exhibited a fluctuating trend from 1990 to 2020. Forests and grasslands were prioritized as the desirable types in land conversion, and had prominent spatial agglomeration. (3) The influence of FM in MHAs on land use changes at county level demonstrated significant spatial-temporal heterogeneity, with wide range and low intensity from 1990 to 2000 and 2015 to 2020, and narrow range and high intensity from 2000 to 2015, and the counties with high intensity were distributed in the Loess Plateau and Sichuan-Chongqing hilly region. (4) The slope of marginalized farmland exhibited a prominent rule of spatial distribution, but an insignificant temporal trend under the influence of governmental policies. The larger the slope was, the higher the degree of marginalization was, but not necessarily earlier it occurred. The results can provide a reference for the formulation and implementation of farmland protection policies.