Tracking surveys of the introduced plant community in Baiyun Lake were conducted in 2013 and 2016. In the tracking surveys, the community composition changes of the introduced single dominant plant community were recorded. The introduced single dominant plant community included Cyperus papyrus, Thalia dealbata, Thalia geniculata, Vallisneria natans, Nelumbo nucifera, Nymphaea mexicana, Ludwigia adscendens, Myriophyllum aquaticum, Phragmites australis, Arundo donax, Pontederia cordata, Acorus calamus, Colocasia esculenta, Canna glauca and Echinodorus grisebachii communities.

The wetland plant species have been divided into 4 kinds of lifeforms: emergent plants (EP, include the plant which part stems and leaves pierces the water surface with root attached to the substrate and the plant which live beside water but the root system is low than the high water level), submergent plants (SP), floating plants (FP) and floating leaf plants (FLP), according to the position of the roots relative to the soil and the position of the leaves relative to the water surface (Sculthorpe 1967). Species abundance was recorded with a 5-level classification: Soe (Sociales, there are a lot of plant and the overground plants are closed), Cop (Copiosae, is numerous), Sp (Sparsal, is small and scattered), Sol (Solitariae, is small and sparse), Un (Unicum, individual or single), quantified as 5, 4, 3, 2 and 1 in Table S2, respectively according to Drude abundance; the species not seen in the survey were recorded as 0 in Table S2. In Table S2, Total Frequency (TF) is the sum of the frequency of the species in the 8 wetlands, while Total Abundance (TA) is the sum of the abundance of the species in 8 wetlands. In this study, plant species that were introduced but not found in the 8 surrounding wetlands were excluded from analysis. The data analysis and figure plot were performed using R language (Definition Version 3.3.3 for Windows). The relationship between Baiyun Lake and the 8 wetlands was evaluated via Jaccard index [30] and Clustering analysis (hclust, ward. D2).

Jaccard index = c/(a + b − c)

a: The number of species in the A communities; b: The number of species in the B communities; c: The number of species both in the A and B communities.

In the vegetation survey result, 147 species of wetland and aquatic plants were recorded, belonging to 44 families and 90 genera; 120 species of native plants accounted for approximately 40% of the records of Guangdong Province aquatic and wetland plants, while exotic plants included approximately 27 species, accounting for approximately 18.4% of the total survey of the plants. The number of species of each wetland (except Baiyun Lake) was 44.5 ± 9.93 species, and the proportion of exotic plants in each wetland varied from 2.27% to 24.4%. Under the influence of artificial introduction, the number of species and exotic species of Baiyun Lake are higher than the natural wetland (Table 1 and Table S2).

In the 147 identified species, 9 species were submergent plants, 6 species were floating leaves, 5 species were floating plants and 127 species were emergent plants. Among which, H. verticillata, V. natans, P. wrightii, Potamogeton crispus and Chara vulgaris were the relatively high abundance and frequency submergent plants; Commelina communis, Polygonum lapathifolium, Leersia hexandra, Paspalum paspaloides, Colocasia esculenta, Alternanthera philoxeroides, Polygonum chinense, Bidens pilosa var. radiata and Arundo donax, were the relatively high abundance and frequency emergent plants. In addition to the floating plants, Eichhornia crassipes and Pistia stratiotes, two invasive plants were relatively high in abundance and frequency, and of floating leaf plants, only Ludwigia adscendens were more common (Table S2).

Figure 3(a) (Clustering analysis) and Figure 3(b) (Jaccard community similarity analysis) showed that in 2013, the wetland plant community composition of Baiyun Lake was more similar to that of the Shenzhen Longgang riverine wetland, which is near Baiyun Lake and in the early stage of wetland vegetation restoration; after three years of recovery, it was more similar to the Liuxi River, which is closest to and interconnected with Baiyun Lake. In general, the community species composition of Baiyun Lake was more similar to that of the Liuxi river, the Shenzhen Longgang riverine wetland, the Tong Lake watershed wetland and the Dashahe Reservoir wetland, the four wetlands near Baiyun Lake; it

is less similar to He Jiang upstream, Wu Jiang upstream, Zhen Jiang upstream and Dong Jiang upstream wetlands, those four wetlands farther away from Baiyun Lake.

Figure 4 showed that in the natural wetlands, the most common wetland plants are emergent plants (82.49% ± 5.25%), followed by submergent plants (6.6% ± 3.85%), floating plants (6.42% ± 2.69%) and floating leaf plants(4.48% ± 1.81%). In the process of vegetation restoration and succession of the new wetland, Baiyun Lake also maintains similar percentages for the four types of plant life form in the natural wetland.

Figure 5(a) showed that during 2013-2016, the total species in Baiyun Lake increased from 28 to 71, as the number of emergent plant species increased from 23 to 59, while the other three types of plant life form increased in smaller amounts. Similar to Baiyun Lake, as showed in Figure 5(b) of Caohai Lake, during 1983-2016, the total species number increased from 41 to 79, as emergent plants increased from 19 to 55. Therefore, in the process of the reconstruction and recovery of wetland plants, emergent plant numbers increased (Figure 5(a),

Figure 5(b)), while the submergent, the floating and the floating leaf plants also began to increase (Figure 5(a)), and in the end, they became balanced (Figure 5(b)).

According to the tracking survey, as Table 2 demonstrated that from 2013 to 2016, C. papyrus, T. geniculata, V. natans, N. nucifera, N. mexicana, L. adscendens,

M. aquaticum, P. australis and A. donax all continued to maintain a single dominant community. The advantages of P. cordata, A. calamus, C. esculenta and C. glauca in a partial single dominant community were replaced by the invasive species A. philoxeroides, W. trilobata and B. pilosa var. radiata. The advantages of E. grisebachii in a partial single dominant community were replaced by L. hexandra or C. diffusa. In addition, the community tracking survey also showed that the original single optimal P. cordata community and E. grisebachii community was also infiltrated by the exotic plants M. aquaticum and H. vulgaris.

Much research shows that the aquatic plant richness of the natural wetlands was increased, such as farmland wetlands in north-west Poland [31] , Changhu Lake [32] [33] , Lashi Lake [34] [35] , and Napa Lake [23] ; there are also reports of emergent plant coverage and wetland plant richness increases in restoration wetlands [36] , while our study indicates that the major contributors to the increase in species richness during wetland restoration are emergent, particularly in those wetlands that have been restored for years. In these study wetlands, the ratio of emergent plants to total wetland plants is high, and the richness of emergent plants increases the most; the composition changes in submergent, floating and floating leaf plants in the total wetland plant figures is small, and the richness changes are also smaller.

The aquatic plant germplasm resources of the target wetland were mainly sourced from surrounding wetlands. In 2013, there were only 23 species of emergent plants in Baiyun Lake, and this dramatically increased to 59 species in 2016. When the wetlands had just been built, most of the emergent plants may have come from the original soil seed bank, in general, and after the environment changes and most of the original species are cleared, those plants whose seeds are pre-existing in the seed bank are rapidly recruited [37] [38] . After several years, few new species still come from the original soil seed bank because only fresh seeds can greatly increase the possibility of a species settling [39] ; thus, we believe that most of the new increases in species numbers between 2013 and 2016 mainly come from the distribution of surrounding wetlands. In addition, during Baiyun Lake’s wetland plant community restoration process, wetland plant community composition and structure becomes more similar to the surrounding wetlands gradually over time. The closer to the wetland, the higher similarity of species composition; in particular, the Liuxi River wetland, which has river-lake connectivity with Baiyun Lake, has the highest similarity with Baiyun Lake. The result is confirmation that the plant communities in created marshes will become similar to those in the natural marshes, taking only a few years [25] . This also indicates that newly increasing plant species in Baiyun Lake may come from surrounding wetlands. The wetlands around Baiyun Lake have many more emergent plants, aside from the other three types of aquatic plants. This leads to the richness and abundance of emergent plant germplasm that enters in Baiyun Lake, more than the other three life forms of aquatic plants.

Rational utilization of self-design and biological engineering in wetland restoration/construction needs to differentiate between emergent plants and submergent, floating and floating leaf plants. For Caohai Lake, during the self-design of wetland vegetation, the richness of emergent plants was increasing, but the richness of submergent, floating and floating leaf plants changed less. Referring to Baiyun Lake, the low abundance of the submergent plant species and germplasm resources in the surrounding wetland limited the submergent plants’ spontaneous recovery in Baiyun Lake: no submergent plants were spreaded naturally in last 6 years, these results confirmed that species dispersal determine the number of species of submergent plants in restored wetlands [40] , also confirmed that the need of submergent plant germplasm’s introduction during the Baiyun Lake restoration. So, the input of submergent germplasm resources was increased in April 2016, including V. natans, H. verticillata, P. wrightii, M. spicatum and C. demersum.

The restoration of wetlands requires the re-establishment of a native wetland plant community [41] , where invasive species may cause unexpected changes [42] or dramatic ecological disturbances [9] because they have the ability to exclude other plant species [38] . The aquatic-terrestrial ecotone is vulnerable to invasive plants, and numerous invasive plants can harm this area. In Baiyun Lake’s aquatic-terrestrial ecotone, A. philoxeroides, B. pilosa var. radiata, W. trilobata and other invasive plants spread faster and excluded other plant species, greatly influencing the regional vegetation restoration. In addition, the introduction of M. aquaticum and H. vulgaris have a potential risk of invasion [43] [44] . Therefore, implementing restoration/construction of wetland is needed to prevent invasive plant hazards to the aquatic-terrestrial ecotone.