JIA-2018-09

2102 WANG Di et al. Journal of Integrative Agriculture 2018, 17(9): 2096–2106 spatial distribution of the sampling units stratified by their average ground slope. 3. Results 3.1. Spatial autocorrelation and its significance at different sampling unit sizes To demonstrate that there is spatial autocorrelation within the sampling units, Fig. 9 shows the values of the Moran’s I and Z -score for maize and rice areas, as shown in Fig. 3, at 25 different sampling unit sizes. For both maize and rice, the Moran’s I value decreased as the sampling unit size increased, while the Z-score value increased with the sampling unit size. Furthermore, when the sampling unit size was less than 1 000 m, Moran’s I value decreased rapidly, varying from 0.89 to 0.65. However, when the sampling unit size was greater than 1000 m, the decrease of Moran’s I value slowed down until it became stable, and varied from 0.65 to 0.49. This indicated that 1 000 m was the inflection point where the curve of the Moran’s I value changed with the sampling unit size. The Moran’s I values for maize and rice for the 25 sampling unit sizes ranged from 0.49 to 0.89. The corresponding Z -score values were all greater than 300, which indicated very significant spatial autocorrelation of maize and rice acreage between the sampling units. The values of Moran’s I and Z -scores for rice were greater than those for maize, regardless of the sampling unit size. The reason was that the growing region of rice was more concentrated than that of maize, in this study area. 3.2. Influence of CPI stratification on spatial autocor- relation among crop acreages in the sampling units To illustrate the spatial autocorrelation of CPI between sampling units, Fig. 10 shows the changes of Moran’s I values for the maize and rice sampling units for the five planting intensity levels. When the sampling unit size was less than or equal to 3 000 m, the change ranges and absolute values of Moran’s I were both very small. However, when the sampling unit size was greater than 3 000 m, the amplitudes and absolute values of Moran’s I significantly increased for both maize and rice. In addition, when the sampling unit size was less than 3 000 m, the Moran’s I values of the planting intensity class variable for both crops were less than 0.2, which denoted very little spatial autocorrelation of the CPI strata for either maize or rice sampling units. Thus, for this variable, conventional sampling methods can be used to select the sample units, and then extrapolate the results to population values. In addition, it can be also seen that the changes of Moran’s I values were relatively regular for maize and rice, when the sampling unit size was less than 3 000 m; however, the changes of the Moran’s I values become very irregular, when the sampling unit size was greater than 3000 m. This indicated that Moran’s I can be simulated when sampling units were limited within 3 000 m. 3.3. Influence of cultivated land fragmentation stratification on spatial autocorrelation among crop acreages in the sampling units Fig. 11 shows the changes in Moran’s I values of maize and rice sampling units for five levels of FRG, respectively. The profile of Moran’s I value could be classified into two types: the first was the profile when the sampling unit size was 4 000 and 5 000 m, and the second was the profile when the sampling units were sizes of 200, 500, 1000, 2000 and 3 000 m. For the first profile, when the sampling unit size was greater than 3 000 m, the Moran’s I values fluctuated irregularly; however, the Moran’s I values in the second 126°10´E 126°0´E 125°50´E 125°40´E 125°30´E 125°20´E 125°10´E 126°20´E 126°30´E 126°10´E 126°0´E 125°50´E 125°40´E 125°30´E 125°20´E 125°10´E 126°20´E 126°30´E 44°50´N 44°40´N 44°30´N 44°20´N 44°50´N 44°40´N 44°30´N 44°20´N N 0 10000 20000 m County boundary 1st class maize 2nd class maize 3rd class maize 4th class maize 5th class maize Fig. 7 Spatial distribution of the sampling units stratified by their fragmentation index (FRG) on a 1 000 m×1 000 m grid. 126°10´E 126°0´E 125°50´E 125°40´E 125°30´E 125°20´E 125°10´E 126°20´E 126°30´E 126°10´E 126°0´E 125°50´E 125°40´E 125°30´E 125°20´E 125°10´E 126°20´E 126°30´E 44°50´N 44°40´N 44°30´N 44°20´N 44°50´N 44°40´N 44°30´N 44°20´N N 0 10000 20000 m County boundary 1st class maize 2nd class maize 3rd class maize 4th class maize 5th class maize Fig. 8 Spatial distribution of the sampling units stratified by their average ground slope on a 1 000 m×1 000 m grid.