Canoco

The distribution and abundance matrix of OTUs was randomly resampled using “daisychopper.pl” (Gilbert et al., 2009 (link)) to equalize sampling efforts. After normalizing, 4354 sequences were left for each sample. Alpha diversity measures including richness estimator Chao 1 (Chao and Bunge, 2002 (link)), diversity index Shannon (Magurran, 1988 ) and Good's coverage (Good, 1953 ) were calculated at a 3% dissimilarity level in Mothur. For the beta diversity, a multi-sample similarity dendrogram was constructed according to the OTUs composition and abundance based on Bray-Curtis dissimilarity. Pairwise analyses of similarities (ANOSIM) of bacterioplankton communities were calculated in PRIMER 5 (Plymouth Marine Laboratory, West Hoe, Plymouth, UK). A ternary plot was generated in R software (RDC Team, 2008 ) to compare the community composition of different groups at the genus level. Redundancy analysis (RDA) with Monte Carlo test was performed to calculate the relationship between bacterial clades and water properties at both order and genus levels, following the results of pretested detrended correspondence analysis using Canoco (Version 4.5, Microcomputer Power). A neighbor-joining phylogenetic tree comparing bacterial community compositions in different estuaries was constructed by MEGA 5 (Tamura et al., 2011 (link)).

The percentage of the difference of Cd in the soil before and after phytoremediation to the total amount of Cd before phytoremediation was calculated in a decrease ratio of Cd (MDR). The ability of the plants to extract Cd from the soil was indicated by the bioconcentration factor (BCF). The calculation formulas are as follows: $$\mathrm{M}\mathrm{D}\mathrm{R}=\frac{{\mathrm{C}}_0-{\mathrm{C}}_{\mathrm{m}}}{{\mathrm{C}}_0}\times 100\mathrm{\%}$$
In which:
C₀—the content of total Cd in soil before phytoremediation, mg kg⁻¹C_m—the content of total Cd in soil after phytoremediation, mg kg⁻¹ $$\mathrm{B}\mathrm{C}\mathrm{F}=\frac{{\mathrm{C}}_{\mathrm{p}}}{{\mathrm{C}}_0}\times 100\mathrm{\%}$$
In which:
C_p—the content of Cd in the plant, mg kg⁻¹Three parallel samples were tested for each indicator, and the final test result was represented by mean ± standard deviation. IBM SPSS Statistics (Version 25.0, Armonk, NY, USA: IBM Corp.) was used for statistical analysis of the experimental data, and the significant differences were analyzed by LSD (least significant difference) method at the level of p < 0.05. RDA analysis was performed by Canoco (Version 5, Microcomputer Power, Ithaca, NY, USA).