*3.3. Land-Type Change in Paddy Field and Dryland under Different Topographic Conditions* 3.3.1. Land-Type Change in Paddy Field and Dryland at Different Elevations

Land conversion between paddy field, dryland, and other land from 1990 to 2020 was mainly concentrated at an elevation of 0–200 m (Figure 6). Among them, the greatest area of land conversion occurred between dryland and other land, followed by the conversion between paddy field and other land. The area of the conversion between paddy field and dryland was the lowest.

**Figure 6.** Land conversion between paddy field, dry land, and other land at different elevations from 1990 to 2020. P, paddy field; D, dryland; O, other land. The numbers in the figures refer to the area among the land changes (km2).

It is worth noting that when looking at the interconversion between paddy field and dryland, the results show that at a 0–200 m elevation, the area changed from dryland to paddy field was much greater than that of paddy field to dryland. On the contrary, at a >200 elevation, the change from paddy field to dryland was more than that of dryland to paddy field.

In terms of the conversion between paddy field and other land, the area of paddy field to other land at each elevation was more than that of other land to paddy field. Unlike paddy field, the conversion between dryland and other land was not only active at the 0–200 m elevation but also significant at the 1000–1500 m elevation (the "second ladder" of China). This discovery confirms the importance of further analyzing the change in dryland and other lands at higher elevations in western China.

### 3.3.2. Land-Type Change in Paddy Field and Dryland at Different Slopes

From 1990 to 2020, land conversion among paddy field, dryland, and other land was mainly located on 0–2◦ slopes, and the area of land-use change decreased along with the increase in slope (Figure 7). Among them, the land conversion area between dryland and other land was the largest, followed by the conversion between paddy field and other land, whereas the conversion between paddy field and dryland was the lowest. In terms of the interconversion between paddy field and dryland, the area changed from dryland to paddy field at a 0–2◦ slope was more than that of paddy field to dryland, but on other slopes, the area changed from paddy field to dryland was larger. In terms of the conversion between dryland and other land, the dryland area changed to other land on a >25◦ slope was more than that of other land to dryland, but on other slopes, the area changed from other land to dryland was higher. In terms of the conversion between paddy field and other land, the area changed from paddy field to other land was more than that of other land to paddy field on slopes, which indicates that the decrease in the paddy field area was significant whether the terrain conditions were steep or flat.

**Figure 7.** Land conversion between paddy field, dryland, and other land at different slopes from 1990 to 2020.

#### 3.3.3. Land-Type Change in Paddy Field and Dryland on Different Slope Aspects

On the eastern slope, the conversion from paddy field to dryland, the conversion from dryland to other land, and the conversion from paddy field to other land were the largest. On the western slope, the change from dryland to paddy field, the change from other land to dryland, and the change from other land to paddy field were the greatest.

The conversion between dryland and other land was the most prominent, the conversion between paddy field and other land was the second-most prominent, and the conversion between paddy field and dryland was the least prominent.

It is worth noting that, when comparing the southern slope (with better photothermal conditions) and the northern slope (with worse light conditions), the results show that the loss of paddy field and dryland was greater on the southern slope than on the northern slope, while the increase in paddy field and dryland was greater on the northern slope than on the southern slope. In other words, the paddy fields and drylands with better light conditions in mountainous and hilly areas decreased (Figure 8).

**Figure 8.** Land conversion between paddy field, dryland, and other land at different slope aspects from 1990 to 2020.

#### *3.4. The Landscape Characteristics of Paddy Field and Dryland under Different Topographic Conditions*

The landscape indexes of paddy field from 1990 to 2020 are shown in Table 3. In terms of the intensity indices, the NP and PD increased, which indicates that the landscape of the paddy field became more fragmented. The LPI increased, which indicates that the dominance of the main paddy field patches increased. Specifically (Tables S6–S8 in the Supplementary Materials), the largest NPs appeared at 200–500 m, on a 2–6◦ slope, and on the western slope aspect; the largest increase in the NP appeared at 200–500 m, on a 15–25◦ slope, and on the northern slope aspect; the largest decrease in the NP appeared at 0–200 m, on a 2–6◦ slope, and on the southern slope aspect. The largest PDs were distributed at 2500–3500 m, on a 15–25◦ slope, and on the northeastern slope aspect. The largest increase in the PD appeared at >3500 m and on the southern slope aspect, and the largest decrease in the PD appeared on a 6–15◦ slope and the northwestern slope aspect. The largest LPIs were distributed at >3500 m DEM, on a 0–2◦ slope, and on the southeastern slope aspect; the largest increase in the LPI was distributed at 0–200 m and on the southern slope aspect; the largest decrease in the LPI was distributed at >3500 m, on a 6–15◦ slope, and on the southeastern slope aspect.

In terms of the shape indices, the AWMSI and LSI experienced fluctuated growth, which indicates that the shape of the paddy patches tended to be complex and irregular. Specifically (Tables S6–S8 in the Supplementary Materials), the largest LSIs were distributed at 0–200 m, on a 2–6◦ slope, and on the western slope aspect; the largest increase in the LSI appeared at 200–500 m, on a 15–25◦ slope, and on the northern slope aspect; the largest decrease in the LSI appeared at 0–200 m, on a 2–6◦ slope, and on the southern slope aspect. The largest AWMSIs were distributed at 0–200 m, on a 0–2◦ slope, and on the southeastern slope aspect; the largest increase in the AWMSI was distributed at 0–200 m; the largest decrease in the AWMSI was distributed at 200–500 m, on a 0–2◦ slope, and on the southeastern slope aspect.

In terms of the vergence indices, the AI decreased while the DIVISION and SPLIT increased, which meant that the landscape of paddy field tended to be more dispersed. Specifically (Tables S6–S8 in the Supplementary Materials), DIVISION decreased at 0–200 m and 1000–1500 m but increased at other elevations, slopes, and slope aspects. The largest SPLITs were distributed at 200–500 m, on a 2–6◦ slope, and on the western slope aspect, whereas the largest increase in the number of SPLIT was distributed at 200–500 m, on a 6–15◦ slope, and on the northern slope aspect, and the largest decrease in the number of SPLIT was distributed at 1000–1500 m, on a 2–6◦ slope, and on the southern slope aspect. The largest AIs were distributed at 0–200 m, on a 0–2◦ slope, and on the western slope aspect, the largest increase in the AI was distributed on a >25◦ slope and on the northwestern slope aspect, and the largest decrease in the number of SPLIT was distributed at >3500 m, on a 0–2◦ slope, and on the southern slope aspect. After 2015, the AI increased while the DIVISION and SPLIT decreased slightly, which indicates that along with the land consolidation projects, the landscape of paddy field appeared to be more concentrated.

**Category Index 1990 1995 2000 2005 2010 2015 2020** Intensity NP 63,269 62,748 63,935 64,902 65,098 65,753 64,574 PD 0.1339 0.1331 0.1349 0.1393 0.14 0.1414 0.1405 LPI 12.55 12.23 13.65 13.80 13.70 13.46 14.23 Shape LSI 334.23 334.05 337.28 340.51 342.46 345.18 337.89 AWMSI 14.85 15.00 16.03 15.93 15.93 15.65 16.05 Vergence AI 51.44 51.39 51.06 50.15 49.85 49.44 50.23 DIVISION 0.9775 0.9789 0.9762 0.9763 0.977 0.978 0.9776 SPLIT 44.35 47.34 42.08 42.25 43.41 45.38 44.52

**Table 3.** Results of the landscape pattern analysis of paddy field.

The intensity indices of dryland are shown in Table 4. The NP and PD increased while the LPI decreased, which indicates that the landscape fragmentation of dryland became more significant. Specifically (Tables S9–S11 in the Supplementary Materials), the largest NPs were distributed at 500–1000 m, on a 2–6◦ slope, and on the eastern slope aspect; the largest increase in the NP appeared at 500–1000 m, on a 6–15◦ slope, and on the northern slope aspect; the largest decrease in the NP appeared at 0–200 m, on a 0–2◦ slope, and on the southeastern slope aspect. The largest PDs were distributed at >3500 m, on a >25◦ slope, and on the western slope aspect; the largest increase in the PD was distributed at 1500–2500 m and on the eastern slope aspect; the largest increase in the PD was distributed at >3500 m, on a 15–25◦ slope, and on the northern slope aspect. The largest LPIs were distributed at 0–200 m, on a 0–2◦ slope, and on the northern slope aspect; the largest increase in the number of LPIs was distributed at 0–200 m, on a 6–15◦ slope, and on the northern slope aspect; the largest decrease in the number of LPIs was distributed at >3500 m, on a 0–2◦ slope, and on the southern slope aspect.

In terms of the shape indices, the AWMSI and LSI increased, which indicates that the shape of the dryland tended to be more complex and irregular. Specifically (Tables S9–S11 in the Supplementary Materials), the largest LSIs were distributed at 1000–1500 m, on a 2–6◦ slope, and on the eastern slope aspect; the largest increase in the LSI was distributed at 500–1000 m, on a 15–25◦ slope, and on the northern slope aspect; the largest decrease in the LSI was distributed at 1000–1500 m, on a 2–6◦ slope, and on the southeastern slope aspect. The largest AWMSIs were distributed at 0–200 m, on a 0–2◦ slope, and on the northern slope aspect; the largest increase in AWMSI was distributed at 0–200 m, on a 0–2◦ slope, and on the northern slope aspect; the largest decrease in the AWMSI was distributed at 1000–1500 m, on a 2–6◦ slope, and on the southern slope aspect.

In terms of the vergence indices, the AI decreased while the DIVISION and SPLIT increased, which proves the fragmentation and diversification characteristics of dryland landscape. Specifically (Tables S9–S11 in the Supplementary Materials), DIVISION was larger at other elevations, slopes, and slope aspects; the largest increase in the DIVISION was distributed at >3500 m, on a 0–2◦ slope. The largest SPLITs were distributed at 2500–3500 m, on a 15–25◦ slope, and on the western slope aspect; the most significant increase in SPLIT was distributed at >3500 m, on a >25◦ slope, and on the western slope aspect; the largest decrease in the SPLIT was distributed at 200–500 m, on a 6–15◦ slope, and on the northern slope aspect. The largest AIs was distributed at 0–200 m, on a 0–2◦ slope, and on the northern slope aspect.


**Table 4.** Results of the landscape pattern analysis of dryland.

In general, the landscape fragmentation of paddy field and dryland increased, but dryland showed a higher aggregate degree and more obvious change towards complex shapes and fragmentation of land plots. Furthermore, most of the fragmentation and complexity in the shape of paddy field and dryland were concentrated in low-elevation areas with flat terrain, which emphasizes the loss of high-quality arable land with superior topographic conditions. In addition, the dryland in high-elevation areas tended to be more dispersed and more complicated, which needs further observation for long-term agricultural production.

#### **4. Discussion**

#### *4.1. Changes in Microterrain Factors Led to a Decrease in Paddy Field and Dryland with Good Photothermal Conditions*

As the primary resources of arable land use, water, heat, light, and other natural factors impact the growth of crops. Scholars have found that as a microterrain factor, the slope aspect has a significant effect on the growth and spatial distribution of plants [30]. The slope aspect determines the photothermal conditions of vegetation through the reception of solar radiation and hydrological processes which, in turn, affects crop growth. In countries in the Northern Hemisphere, the northern slope aspect has lower temperatures, lower light intensity, higher relative humidity, and more abundant soil nutrients [31–33], while the southern slope aspect has sufficient light, less moisture, and large diurnal temperature differences. Therefore, most studies indicate that plants on the southern slope aspect have a higher photosynthetic level and are more productive [34–36]. Similarly, this paper showed that 40.56% of paddy fields and 38.48% of drylands in China were distributed on the southern, southwestern, and southeastern slope aspects, which further proves the higher suitability of paddy field and dryland use on southern slopes.

However, the results of this paper found that the area of paddy field and dryland with better light conditions in China was decreasing. The area of paddy fields and drylands on the southeastern, southern, and southwestern slopes generally decreased from 1990 to 2020 but increased on the northwestern, northern, and northeastern slopes. Furthermore, conversion from paddy field and dryland to other land was more active on the southern, southeastern, eastern, and northeastern slopes. These results indicate the worsened solarthermal conditions of paddy field and dryland, which may cause an explicit potential decline in the productivity of arable land. Though some studies have paid attention to the change in arable land area and spatial distribution, the change in productivity caused by

the microterrain changes in arable land is rarely discussed, which needs further attention in future research.

#### *4.2. In Addition to the "Third Ladder", the Changes in Paddy Field and Dryland Have Become Active on the "Second Ladder" of China*

The terrain of China is high in the west and low in the east. According to the elevation from west to east, the whole country can be roughly divided by three ladders including the first ladder (>4000 m) in the west, the second ladder (1000–2000 m) in the middle, and the third ladder (<500 m) in the east. This study found that the land conversion of paddy field and dryland in China mainly occurred in the eastern plain on the third ladder. Similarly, relative studies have supported this conclusion such as the large-scale conversion from paddy field to dryland and construction land in the Beijing–Tianjin–Hebei Region [37], the conversion from other land to dryland and paddy field in the Northeast China Plain [4,15], and the conversion from paddy field and dryland to construction land in Yangtze Plain areas [27].

It is worth emphasizing that this article has a new finding. In addition to the third ladder, the transition between paddy field, dryland, and other land on the second ladder was also active (such as at an elevation of 1000–1500 m). Coincidentally, Chi [38] and Dong [39] also found that a large area of the land conversion of paddy field, dryland, and other land occurred on the second ladder, such as Inner Mongolia, in recent years. Therefore, the land-use change in paddy field and dryland became active on the second ladder. Although climatic conditions, population density, and economic development on the second ladder are poorer than that on the third ladder, the use of arable land at higher elevations is proven to have increased.

The reasons for this can be found by tracing the regional development of the second ladder. Since the implementation of the China Western Development Strategy in 2000, the Chinese government has increased its support for land consolidation and basic arable land construction in the west and has carried out a series of ecological restoration projects to prevent desertification and improve the ecological environment of arable land. Such policy supports resulted in a significant increase in new agricultural modernization, agricultural capital investment, and per capita arable land area [40]. After the agricultural tax was abolished by the No. 1 Central Document in 2006, "The Guidance on Promoting Sustainable Development of Agriculture and Animal Husbandry in Northwest Arid Regions" was issued, which increased agricultural subsidies in poor areas in West China. These have further motivated the enthusiasm for farm production on the second ladder.

It is worth mentioning that in 2017, the government began to permit trans-provincial "land-ticket transactions" to keep an arable land requisition–compensation balance (i.e., arable land occupied by urban construction in provinces with a land shortage can be replenished by provinces with abundant arable land reserve resources). Benefited by abundant land resources, the provinces on the second ladder developed and replenished a large amount of new arable land in "land-ticket transactions", which has also contributed to the dramatic land conversion between paddy field, dryland, and other land at the second ladder.

However, because of the special arid climate environment of the second ladder, the increase in arable land in this area may further intensify the contradiction of water use [41], which will cause the degradation of the ecosystem and the quality of arable land and lead to an increase in desertification over the long term [42–44]. Therefore, faced with the active conversion of paddy field and dryland on the second ladder, the local government needs to focus on pushing forward the delineation of the "three lines" (i.e., urban development boundary, permanent basic arable land, and ecological protection red line) and the "three areas" (i.e., urban space, agricultural space, and ecological space) in territorial space planning. Unsuitable arable land with high costs and ecological risks should be retired and the reclamation of arable land in ecologically fragile areas should be limited to rationalize the land-use structure at a higher elevation. Moreover, large-scale land use in Europe has

shown some adverse effects such as the persistence of pesticides and other agricultural inputs. Taking a cue from this, the government in China has new opportunities and challenges [45]. The government should further strengthen the arable land-use intensity and high-standard cropland construction and actively promote the protection of arable land such as by letting land lie fallow and through reasonable crop rotation [46]. To improve the acceptance of environmentally friendly techniques, the government should provide more subsidies for eco-friendly agriculture to farmers.

#### *4.3. Increased Paddy Field and Dryland on Slopes Exacerbated the Erosion Risk*

From the perspective of land suitability, arable land in low-elevation plains is rich in water and heat resources, which is more suitable for agricultural production. This study found that paddy fields at low elevations (0–200 m) accounted for 64.6% of the total paddy field area, and paddy fields on low slope (0–2◦) areas accounted for 67.62% of the total paddy field area. Drylands at low elevations (0–200 m) and low slopes (0–2◦) accounted for 39.78% and 61.46% of the total dryland area, respectively. Compared with dryland, paddy field is more sensitive to topography and have higher requirements for water retention. However, this study found that the paddy field and dryland went "up the hill", i.e., an increasing number of paddy fields and drylands appeared at higher elevations (>200 m) and steeper slopes (>6◦). Because of the large quick water flow on slopes, the phenomenon of soil and water erosion caused by the increase in paddy field and dryland on slopes needs more attention [47]. Rice cultivation in China and Southeast Asia is an important source of farmers' income. The increasingly unideal topographic conditions of paddy fields may bring about more socioeconomic uncertainties, which need further study on the consequences by researchers.

In addition, in terraced fields on slopes, paddy fields tend to exhibit a higher water and fertilizer retention capacity and lower soil erosion intensity than drylands [48]. For example, Gao [49] and Xiao [50] found that paddy fields were less affected by soil erosion than drylands in karst areas, and Chen [51] pointed out that the paddy fields in Taiwan's mountains were less affected by rainfall erosion than drylands. However, this paper found that the transition from paddy field to dryland within a >2◦ slope and >200 m elevation was greater than the transition from dryland to paddy field. Taking the mountainous southwest China (including Chongqing, Sichuan, Guizhou, and Yunnan) as an example, the area of paddy field to dryland within a >2◦ slope and >200 m elevation was 24,080 and 25,981 km2, respectively, which is much higher than the area from dryland to paddy field (15,022 and 16,561 km2). As a result, considering the massive change from paddy field to dryland, stronger water–soil conservation is highly needed to improve the productivity of mountainous arable land.
