**3. Results**

### *3.1. The Dynamics of Fish Ponds across the GBA*

Figure 7 shows the overview of pond dynamics over the 10 periods in GBA. Based on the water area change index (*W*) in Figure 8, it can be seen that the total pond area firstly showed an incipiently increasing trend, followed by a decreasing trend. The fluctuation of W is basically consistent with the evolution of fish ponds in the GBA. From 1986 to 1994, fish ponds demonstrated a continuous rise from 106,603 hm<sup>2</sup> to 187,153 hm2, which is an enlargement of with a total of area of 80,550 hm<sup>2</sup> or a relative increase of 75.5%. Accordingly, *W* also performed an accelerated expansion and peaked in 1994. Since then, the areal change fluctuated within a small range until 2009, which means that fish ponds were under a relatively stable status during that period. In 2009, the area fell slightly to 184,589 hm2, accompanied by an increase, and reached the maximum value of 196,326 hm<sup>2</sup> in 2013. After that, there was a decreasing trend from 2013 to 2019, with an average annual shrinkage of 10,242 hm2; meanwhile, *W* also maintained a negative growth. By 2019, the

area had shrunk to 134,874 hm<sup>2</sup> with an overall decrease of 31.3%, but the total area was still larger than 1986.

**Figure 7.** Spatial distribution of fish ponds in the GBA in the 10 different study periods.

**Figure 8.** Water area change index and pond areal fluctuation in the GBA from 1986 to 2019.

According to the *SD* change for each period, the spatial distribution of fish ponds in the whole GBA could be divided into three phases (1986–1994, 1994–2013 and 2013–2019 in Figure 9, and Table 1). The increasing phase between 1986 and 1994 indicated that the differences in spatial distribution of the fish ponds increased during this phase. Geographically, it could also be seen in Fig. 4 that the fish ponds were mainly located in the middle of the GBA in early period, but gradually spread to the east and south of the GBA. The *SD* index dropped to 63,026.68 in 2009, showing fish ponds turned into a diffused spatial distribution. This can also be seen in Figure 7, the pond density increased, and the spatial distribution expanded outward. After they rebounded in 2013, the index kept decreasing to 52,474.643 in 2019, illustrating that the distribution shows a trend of initial agglomeration and subsequent diffusion between 2009 and 2019.

**Figure 9.** Fluctuation of three statistical indices compared to the areal changes of fish ponds from 1986 to 2019. (**A**): Theil index (*T*); (**B**): coefficient of variation (*CV*). (**C**): standard deviation (*SD*).


**Table 1.** Statistical results of fish ponds in GBA from 1986 to 2019.

The coefficient of variation and Theil coefficient in each year reflected that, there were large differences in the size and area of fish ponds in the early stage, but the ponds became increasingly homogeneous in the later period. From 1986 to 2009, the coefficient of variation dropped significantly from 1.358 in 1986 to 0.768 in 2009, and the Theil coefficient dropped from 0.428 to 0.282, indicating that the differences in pond size gradually decreased during this period. In the next stage 2009–2019, the Theil coefficient fluctuated basically in a relatively stable status. the variation coefficient also changed consistently to the Theil coefficient. The results proved that human activities have continuously influenced the distribution and size of fish ponds in the past 40 years. The fish ponds had transformed from an early near-natural ponds with different sizes and a near-natural random distribution into an artificial distribution and an artificial shape.

### *3.2. The Areal Variation Trends in the Prefectural Cities*

In 1986, Foshan had the largest number of fish ponds, followed by Zhongshan and Zhaoqing. However, by 2019, Jiangmen has become the largest, followed by Foshan and Zhaoqing. Considering the area of the fish ponds and the overall areal variations at the initial and the end of the study period, the prefectural cities can be roughly divided into the following three categories:

The growing cities (Figure 10) include Guangzhou, Huizhou, Zhuhai, Zhaoqing, Jiangmen, and Zhongshan. The trend of areal variation from 1986 to 2019 shows a fluctuating growing trend, but the peaks appeared at slightly different times. The year 2013 was the peak of Guangzhou, Zhuhai, Zhaoqing, and Jiangmen. As for Zhongshan, in addition to the peak in 1994, it also reached a second peak in 2013. All the five prefectural cities experienced a decline after 2013. Similarly, Huizhou began to decline after a slight rise in the 2013–2015 period.

**Figure 10.** The trends for three representative changes in fish ponds in the GBA.

The shrinking fish ponds were observed in Foshan and Dongguan (Figure 10). In 2019, the area of fish ponds in Foshan decreased by 39.43%. In comparison, the area of fish ponds in Dongguan decreased by 27.90%. In terms of the variation trends, both cities maintained a fluctuated growth first and then a gradual shrinkage. The maximum area of the fish ponds in two cities happened in the 1990s. In 1994, Foshan's fish ponds reached the maximum area of 77,729 hm2; in 1999, the area of Dongguan reached the highest value of 9995 hm2. The minimums in both appeared in 2019, indicating that the shrinking trends will remain.

The rapid shrinking cities (Figure 10) include Shenzhen and Hong Kong. Overall, fish ponds in both cities have almost disappeared. By 2019, the area of fish ponds in Shenzhen decreased by 97.1%; the counterpart in Hong Kong also decreased by 63.8%. The areas of fish ponds in the two regions still maintained a decreasing trend with slight fluctuations, with the maximum value in 1986 and the minimum value in 2019.

It should be noted that, in 2013, except for Huizhou, Shenzhen, and Hong Kong, the other eight cities all reached a peak before showing a significant drop. Research on this milestone is meaningful to understand the development of fish ponds over the past 10 years.

### *3.3. The Spatio Patterns in Areal Dynamics among Study Units*

The expansion coefficients (*E*) were calculated for the 25 study units in the GBA in each period. The study periods were divided into two stages–the first stage (1986–2013) and the second stage (2013–2019).

### 3.3.1. The First Stage (1986–2013)

At the first stage, the types of the study units were divided as follows:

Fast growing units, included Sihui, Sanshui, Panyu, Nansha, Xinhui, Taishan, Zhuhai, mainly located in central and southern GBA. Such cities obtained an expansion coefficient larger than 89.0, expansion rate greater than 5.6%, and internal structure transition coefficient, spatial structure transition coefficient both greater than 2.0. From 1986 to 2013, the total area of fish ponds in this units increased from 12,595 hm<sup>2</sup> to 80,369 hm2, with a total of 538.10%, which made the areal proportion of these units increased from 11.81% to 40.94%.

Growing unites, included Gaoyao, Gaoming, Heshan, Nanhai, Zhongshan, and Combined units B, C, and E, mainly located in the western GBA. Such units received a expansion coefficient higher than 1.4, expansion rate greater than 0.8, and internal structure transition coefficient, spatial structure transition coefficient both greater than 0.3. From 1986 to 2013, the area of ponds in these units increased from 40,033 hm<sup>2</sup> to 77,009 hm2, with a total of 92.36%, making the areal proportion of fish ponds in these units increased from 37.55% to 39.22%.

Stable unites, included five units in total, namely, Huizhou, Dongguan, Pengjiang, Jianghai, and Combined Unit B. Such units are relatively stable without significant increase: all of the expansion coefficient values were stable between 0 and 1, expansion rate and internal structure transition coefficient were both relatively low. During the first period, Dongguan maintained a relatively stable with slight growing, the apparent shrinkage of the area appeared in 2019, which lied in the second period, from 1986 to 2013. Although the area of fish ponds in this units increased from 12,937 hm<sup>2</sup> to 20,494 hm2, a total of 58.41%, the pond areal proportion of the units decreased from 12.14% to 10.44%. Therefore, compared to other units, this increase was very insignificant.

Shrinking units, included Chancheng, Shunde, Shenzhen, Combined Unit D, and Hong Kong, mainly located in north-western and south-eastern GBA. These units obtained negative expansion coefficient and internal structure transition coefficient, with continued shrinking in total pond surface area. From 1986 to 2013, the total area of ponds in these units decreased from 41,038 hm<sup>2</sup> to 18,454 hm2, a drop in 55.03%. The total pond areal proportion of the units decreased from 38.50% to only 9.40%.

### 3.3.2. The Second Stage (2013–2019)

Slight growing units, included only one unit of Taishan. From 2013 to 2019, the total area of ponds in this unit increased from 11,233 hm<sup>2</sup> to 13,430 hm2, with a total of 19.5%. The relative areal proportion of this unit increased from 5.72% to 9.96%.

Stable unites, included Huizhou, Combined Unit E, and Hong Kong (Figure 11b). From 2013 to 2019, the total area of fish ponds in these units increased from 10,922 hm<sup>2</sup> to 11,725 hm2, with a total of 7.35%. The relative areal proportion of these units increased from 5.56% to 8.69%.

Shrinking units, included Sanshui, Nanhai, Shunde, Gaoming, Chancheng, Panyu, Nansha, Dongguan, Shenzhen, Zhuhai, Zhongshan, Gaoyao, Sihui, Heshan, Xinhui, Jianghai, Pengjiang, Combined Units A, B, C and D, 21 units in all. It can be seen, during the stage, most of them were shrinking units. From 2013 to 2019, the area of fish ponds in these units dropped from 174,171 hm<sup>2</sup> to 109,719 hm2, a total of −37.00%, making the relative areal proportion of this units decreased from 88.72% to 81.35%. At the same time, the water area change index (Figure 12) also showed that the average water body change index of each unit from 2013 to 2019 was −5.67%, and the average annual change for each unit was −4.11 hm2. Among them, Shunde, Zhongshan and Sanshui, which were the top three in absolute annual shrinking in fish ponds, with an average annual shrinkage of 1164 hm2,

1055 hm<sup>2</sup> and 969 hm2, respectively. In Chancheng, Shenzhen and Heshan, where the average annual relative shrinking in fish ponds ranked the top three. The water body change index was −11.44%, −9.81% and −9.77% respectively.

(**A**) 

(**B**) 

**Figure 11.** The spatiotemporal patterns in areal dynamics in the study units in the two different stages: (**A**) the first stage from 1986 to 2013; (**B**) the second stage from 2013 to 2019.

**Figure 12.** Distribution of each unit for waterbody area change index in GBA from 2013 to 2019.

It can be seen that taking 2013 as the milestone, the spatial dynamic patterns of fish ponds in the GBA have changed significantly. Except for the four units of Taishan, Hong Kong, Huizhou, Combined Unit B, which are classified as weakly expansionism or relatively stable, the other 21 units all experienced shrinking.
