In the rapid urbanization process of China, the total amount of farmland rapidly decreased by approximately 11.1 million hectares from 1998 to 2005. As China has a relatively low cultivated land area per capita (Chien, 2015), this loss of farmland has attracted increasing attention from policy makers and researchers. Previous studies have shown that the encroachment of cultivated land first occurred in the eastern coastal area of China and then gradually spread to the central and western inland regions (Zhong et al., 2011; Liu et al., 2014; Jiang et al., 2020a; Wang et al., 2020). Environmental pollution and ecological deterioration have also contributed to the decline of farmland productivity (Jiang et al., 2017; Han et al., 2020). As the basic resource of agriculture, cultivated land destruction has weakened the national economy’s sustainability, social stability, and ecological security (Yang et al., 2000; Skinner et al., 2001; Wu et al., 2017).

Since 1986, the Chinese government has gradually introduced the farmland preservation law system and proposed preservation targets in the Master Plan Outline of Land Use (2006-2020), however cultivated land protection is still under great pressure. Urbanization and industrialization, which are considered to be the main driving forces of cultivated land loss (Song, 2014; Xiao et al., 2018; Yu et al., 2019), are still rapidly developing, promoting the spread of construction land from 26.0 million hectares in 2006 to 31.8 million hectares in 2016. To retain the total amount of farmland, the central government of China released the “No net loss of cultivated land policy”, through which, land spatial distribution was adjusted as well as land boundaries to provide potential areas for construction land (Liu et al., 2017). Thus, the landscape alterations of cultivated land, such as land fragmentation or land marginalization (Liu et al., 2019; Wang et al., 2020), caused the reduction of cultivated land use efficiency by impacting the lands’ spatial utilization conditions (York et al., 2011; Sklenicka et al., 2016). In addition, the land consolidation system also ensured the adjustment of the farmland spatial distribution as a standard and legal administrative procedure, which further promoted cultivated landscape change and became a huge challenge for cultivated land protection, especially in the rapid urbanization areas after 2006. Therefore, it is necessary to pay more attention to the landscape variations of farmland.

In rapid urbanization and industrialization, there are various reasons for cultivated land landscape variations, which can be generalized into natural and anthropogenic factors. Natural factors include the external environment, such as the topography, natural disasters, and accessibility (Liu et al., 2019; Liang et al., 2020); and the conditions of the cultivated land, such as the land productivity and land degradation (Jiang et al., 2020b). Conversely, anthropogenic factors cover a wider range of factors that include both political and socio-economic factors. Regarding political factors, researchers have mainly focused on the impact of the household land allocation system on farmland spatial distribution and the influences of property inheritance and ownership reform (Sklenicka et al., 2008; Lu et al., 2011). In terms of socio-economic factors, population, industry, and investment are considered regardless of their scale or structural characteristics (Jiang et al., 2019; Zhang et al., 2020). These factors cumulatively lead to variations in cultivated land landscape in different topographic regions (Xiao et al., 2018; Liang et al., 2020). The above mentioned factors mostly include regional indicators to cover the overall situation of cities and villages in order to evaluate the impacts of anthropogenic activities on cultivated land. Nevertheless, due to the leading role of urban areas in all aspects of regional socio-economic development, indicators such as the population structure are often more representative of urban development. This makes previous analysis of anthropogenic factors conducted via regional census data unable to distinguish the impact of rural development on cultivated land landscapes. In China, the rural shrinkage has lasted approximately 20 years during rapid urbanization development (Liu et al., 2017; Li et al., 2019). Due to the differences in the job opportunities and living environments between cities and villages, a large number of the young and middle-aged people have moved to urban areas, resulting in an age structure imbalance in the rural population (Li et al., 2014). This population loss in “hollow villages” has led to agricultural labor shortages, which causes farmland to go uncultivated (Long et al., 2012; Zhang et al., 2019). A rural area population influx exists in parts of the northern and eastern coastal provinces in China. Here, migrant populations prefer to gather in villages surrounding megacities to promote the development of the rural housing market (Hao et al., 2011; Zeng et al., 2019). While it is clear that the population fluctuation in rural areas directly influences the economic development and land use of rural areas, it is logical to wonder how it will further affect the cultivated land landscape.

Therefore, the main goal of this work is to investigate the relationship between rural population change and farmland landscape from a topographic perspective and discuss the impacts of anthropogenic disturbance caused by rural population change on cultivated lands. Taking Zhejiang Province as the empirical study area, this study aims to help the local government formulate an eco-friendly rural development policy and propose more effective cultivated land protection measures.

From 2000 to 2010 (stages 1 and 2), most of the county-level administrative units in Zhejiang experienced rural population decreases, and there were more rural areas with a net population outflow than inflow. Furthermore, the average cultivated land landscape index exhibited an upward trend, increasing from 0.48 in 2005 to 0.50 in 2010. After 2010, more than 60% of the total county-level administrative units showed a significant change in their rural population flow trend. However, 19 counties showed spatial agglomeration features with negative RSPV values in stage 3. The main reason for this spatial pattern is the topography, which was also significantly related to the CLI in the POOL regression model. Combining Figures 2 and 4, the counties with relatively large average slopes tend to have more complex cultivated land landscapes. Farmers who lived in those mountainous counties always used the terrace system, which involved small pieces and were separated compared to the paddy fields in the plain areas, as the main method for cultivating land. Although the terrace system is the invention of ancient farmer wisdom, it is unable to adapt to modern large-scale agricultural cultivation, leading to relatively lower agricultural income in those counties (RDI).

Compared to the mountainous counties, the research samples with small average slopes experienced less population outflow. The main reason for this phenomenon was bottom-up industrialization. In those counties, small- and medium-sized enterprises (SMEs) played leading roles during the industrialization process (Zhou et al., 2013; Xiang et al., 2019). Similar to the industrial parks developed in urban districts, SMEs gathered rapidly in small towns, forming economic agglomeration areas by cooperating with professional markets. However, these massive economic agglomeration areas did not appear in every research unit but were rather concentrated in the northern and eastern coastal areas of Zhejiang, represented by the “Wenzhou mode” (Lin et al., 2019). Some studies suggested that SMEs provided many local nonagricultural jobs for rural residents, and a portion of these SMEs did not engage in agriculture, nor did they move to large cities. Instead, they chose to work in the industrial sector in the surrounding small towns and lived in the original rural area (Zhu, 2015). Although this could not completely reverse the outflow of the rural population, it could moderately reduce the outflow amount in stages 1 and 2. Because the main employment sector in rural areas has changed from agriculture to industry, cultivated land is no longer the most important means of production for rural residents. Rural residents preferred to transfer the use rights of unused farmland to agricultural companies or specialized agricultural households through the market-oriented land transfer system (Ye, 2015). For a specific cultivated land area, the concentration of main bodies of agricultural production would greatly improve the land-use efficiency while the standardized production mode of modern agriculture could also promote anthropogenic interference with the cultivated lands (Chen, 2019). For example, the natural spatial distribution of cultivated lands will be gathered, and the irregular shapes of original farmlands will be geometrized to adapt to large-scale mechanical farming. Therefore, this part of the counties, which are located in the coastal areas and basin areas in Zhejiang, was more likely to have a higher per capita rural residential income (RDI) and proportion of primary industry (PIP) and to have a relatively strong anthropogenic impact on cultivated lands.

Furthermore, some county-level units had a positive RPSV after 2010 (stage 3), and most of them aggregated around the metropolitan districts of Hangzhou, Ningbo, Wenzhou, and Jinhua-Yiwu metropolitan areas. Due to the urban function spillover effect, the industrial sectors in these counties have been further developed, providing more jobs for the local people to attract more migrant workers to return home for employment (Zhang et al., 2017). However, the rural areas in those counties become leisure tourism destinations for citizens who live in urban districts (Gao et al., 2017; Xu et al., 2018). Rural tourism has not only attracted the original outflow of registered rural residents to return and start their own businesses, but it has also absorbed the original agricultural labor force into the service industry (Gao et al., 2020). The emergence of leisure tourism in rural areas has further changed the employment structure of rural residents and improved the RDI; however, rural tourism development and rural construction land expansion has aggravated human disturbances to cultivated lands, especially some rural tourism facilities directly occupying cultivated lands.

It is a long-term task to preserve the cultivated lands in China. After the Chinese central government proposed a strict target amount of cultivated land preservation, how to improve the cultivated land utilization efficiency is a critical issue. Previous studies clearly illustrated that the spatial distribution and landscape features of lands were highly correlated with land use efficiency, and the influencing factors also varied. This study focused on the relationship between cultivated land landscape characteristics and rural population variation to explore human anthropogenic disturbances to cultivated land from a spatial perspective. Based on Landscan data, we evaluated the population spatial variation in rural areas to replace the population scale indicator to express the rural population changes in the research samples. The empirical evaluation results showed that the rural population of 50 county-level administrative units in Zhejiang decreased from 2000 to 2010 while 19 counties had more rural population inflow areas than outflow areas after 2010. Using the CLI indicator as the independent variable and five socio-economic factors as the control variables, the regression results of the POOL model showed that the RPSV had a significant and negative influence on the farmland Landscape metrics. The cultivated land in the research samples with a larger rural population variation seems to have relatively less anthropogenic disturbance. In addition, the rural residential disposal income, the average slope, and the proportion of primary industry were all positively and significantly related to the CLI.

The empirical results implied that the variation in the rural population does have a significant impact on cultivated land use efficiency. However, the reasons for this influence vary in different regions. For counties with more rural population outflow areas, the loss of the agricultural labor force and the difficulty of sloped farmlands adapting to mechanized farming were unconducive to land use efficiency improvement. However, the research samples with more rural population inflow areas also had to face the problem that the increase in nonagricultural production activities in rural areas would increase the difficulty of cultivated land protection. Thus, local governments must pay more attention to farmland utilization efficiency improvements adapted to their own developmental conditions. For the former counties mentioned above, the key point of land protection is how to improve the production efficiency of fragmented land through modern agricultural techniques, thereby reducing the amount of abandoned farmland. For the latter counties mentioned above, local governments should strengthen the construction control of housing and facilities in rural areas.

Our study still has some limitations that provide potential directions for future research. First, previous studies have made considerable achievements regarding the impact of land systems on cultivated land preservation (Wu et al., 2017; Wang et al., 2010). However, it is difficult to incorporate political indicators into our empirical regression model because such types of basic data are difficult to obtain. Future studies focused on the influence of politics on cultivated lands could use some specific protection policies as the events and apply the difference-in-differences method to analyze the actual impacts of preservation policy. Second, this study focused on cultivated land preservation from a topographic perspective, regardless of the productive ability or pollution of the land. Although each of those research fields was fully studied, future works could pay attention to the comprehensive capacity of cultivated lands considering the spatial characteristics and quality of the land.