In situ phytoremediation of copper and cadmium in a co-contaminated soil and its biological and physical effects

Phytoremediation is a potential cost-effective technology for remediating heavy metal-contaminated soils. This method was used to evaluate the biomass and accumulation of copper (Cu) and cadmium (Cd) of plant species grown in contaminated soil and their biological and physical effects on the soil. In co-contaminated soils with copper (Cu) and cadmium (Cd), a three-year field experiment was conducted by planting four plant species in the co-contaminated acidic soil treated with hydroxyapatite. The four species produced different amounts of biomass in this order: Pennisetum sp. > Elsholtzia splendens > Setaria lutescens > Sedum plumbizincicola. Over three growing seasons, the best accumulators of Cu and Cd were Elsholtzia splendens and Sedum plumbizincicola, respectively. Overall, Pennisetum sp. was the best species for Cu and Cd removal when biomass was considered. The four plant treatments could improve the content of >0.25 mm mechanically stable (DR0.25) and water-stable (WR0.25) aggregates and significantly improve the aggregate mean mass diameter (MWD) and the geometric mean diameter (GMD). The largest increase was with the treatment of Pennisetum sinese, while the fractal dimension (FD) of mechanically stable aggregates could be significantly reduced by the treatment of Pennisetum sp. Hydroxyapatite and phytoremediation could improve the soil enzyme activity, and Elsholtzia splendens had the best effect in this respect. This study will provide a better understanding of the remediation of heavy metal contaminated soil.


Introduction
Soil contamination, particularly, agricultural soil contamination, has become a severe environmental problem as it poses a grave threat to human health by entering food chains and to environmental safety by leaching into groundwater. 1 Cadmium is a non-essential element that can cause great harm to plants and animals at very low concentrations. Excessive cadmium can not only cause serious pollution to soil, reducing crop yield and quality but can also be taken up by humans through the way of soil/crop/food migration, thus endangering human health. 2 In fact, some pollutants are nowadays never isolated in the soil system. Cd can be brought into the soil when other heavy metal elements enter the soil. For example, in the process of copper smelting, not only copper but cadmium will also enter the soil together, thus leading to some contaminated soils. 3 Various remediation methods, such as land lling, xation, and leaching, may be benecial to the remediation of Cu and Cd polluted soils. 4 However, these methods are usually expensive, and some of them can impose adverse effects on the biological activity, structure, and fertility of soils. 5 Phytoremediation is considered an environmentally friendly, gentle method of managing polluted sites as it uses biological processes to treat the pollutant. 6 In contrast to most other remediation technologies, phytoremediation has signicant environmental advantages. 7 The application of native plants for phytoremediation is particularly important, because they can better adapt to the soil properties, toxicity levels and climatic conditions of the contaminated site. 8 Gramineae species usually adapt faster to these conditions than other plant species because their shorter life cycles allow them to produce various genotypes in a shorter time. 9 Some experiments have been carried out to investigate the phytoremediation potential of various plants in greenhouse experiments. At the same time, hyperaccumulators are oen used, which can grow normally in soils contaminated with heavy metals and accumulate these metals in the harvested parts over the course of phytoremediation. 10 In fact, in addition to hyperaccumulators, some plants with large biomass, good adaptability and fast growth rates also show good remediation potential for heavy metals, such as Giantreed and Pennisetum sp. 11 However, in the process of remediation, because of the low soil pH, high metal toxicity may impede the growth of plants. 12 Results have shown that some materials have good adsorption and stability to Cu and Cd such as ligno-cellulosic materials, Saccharomyces cerevisiae, biosorbents, limestone and hydroxyapatite. [13][14][15] For example, the addition of hydroxyapatite can lead to an increase of soil pH; at the same time, it can reduce the toxicity of heavy metals through ion exchange reactions, surface complexation reactions, co-precipitation and precipitation. 10,16 All these processes will promote the growth of plants in polluted areas. For example, Xu et al. found that Pennisetum sp. could grow normally only when lime was applied in the copper and cadmium contaminated soils. 17 The aim of phytoremediation should not only be to remove heavy metals but also to improve soil quality. 18 Therefore, the comprehensive evaluation of soil quality needs to be considered when evaluating the phytoremediation effect. Soil enzyme activity is the most active constituent; a large number of studies have shown that soil enzyme activity is sensitive to heavy metal pollution because of its signicance in nutrient cycles, organic matter turnover, soil characteristics, microbial activity and biomass. 19,20 The most frequently measured activities include b-1,4-glucosidase (BG, which catalyzes the terminal reaction in cellulose degradation), b-1,4-N-acetylglucosaminidase (NAG, which catalyzes the terminal reaction in chitin degradation), and acid or alkaline phosphatase (AP, hydrolyze phosphate esters including phosphomonoesters, phosphodiesters and, in some cases, phosphosaccharides that release phosphate), which are frequently linked to the rates of microbial metabolism and biogeochemical processes. 21 At the same time, as a basic unit of soil structure, aggregate stability and its inuencing factors are important for maintaining good soil structure and soil fertility. 22 During phytoremediation, plants increase soil organic matter content through the decomposition of root exudates and litter, thus promoting the formation and stability of aggregates, improving soil physical and chemical properties. 23 Therefore, it is necessary to evaluate the composition and structure of soil aggregates which can reect the changes in soil physical properties during phytoremediation.
The goal of this study was to evaluate the metal removal efficiency of Pennisetum sp. compared with three other kinds of plants in combination with hydroxyapatite application in Cu and Cd contaminated soil. The objectives of this study include: (1) evaluating changes in soil chemical properties and heavy metal availability before and aer the combined remediation; (2) determining removal efficiency and removal amount of heavy metals by different plants; (3) measuring the composition and stability of soil aggregates; and (4) measuring the soil enzyme activity.

Study site
The study site is located in Guixi City, Jiangxi Province, China (117 12 0 E, 28 19 0 N). The area is located near a large copper smelter and fertilizer plant. There are about 130 hectares of soil in the area that are contaminated by heavy metals (mainly Cu and Cd) due to 30 years of sewage irrigation. 24 At present, most of the soil in the area has been abandoned due to severe pollution, and desertication has appeared in some areas. The soil texture in this area is of sandy loam; the primary pollutants in the soil are Cu and Cd, with concentrations of 632 and 0.41 mg kg À1 , respectively. Moreover, the site soil is very acidic (pH ¼ 4.35), having soil organic carbon (SOC) content and cation exchange capacity (CEC) of 28.5 mg kg À1 and 8.31 cmol kg À1 , respectively. Total N and total P in the soil were 1.11 g kg À1 and 0.190 g kg À1 , respectively.

Reagent and plants
Hydroxyapatite (particle size ¼ 0.25 mm, pH ¼ 8.40, Cu and Cd concentrations were 9.54 mg kg À1 and 1.18 mg kg À1 ) was purchased from the Nanzhang Lihua mineral powder factory (Hubei, China). In this experiment, three species of plants were selected based on the high concentrations of Cu and Cd in the soil. We chose a copper tolerant plant (Elsholtzia splendens), a cadmium hyperaccumulator plant (Sedum plumbizincicola), and a kind of energy plant (Pennisetum sp.) that had a good tolerance to Cu and Cd in our previous study. All of the plants were derived from indoor-grown seedlings. . Weeds (mainly S. lutescens) were cleared from all plots before planting every year, and no weeding was carried out thereaer. The planting density was 20 cm Â 20 cm for E. splendens and S. plumbizincicola plants, and 50 cm Â 50 cm for Pennisetum sp. plants. All plots were managed using the same eld management.

Sample collection
S. plumbizincicola was harvested in mid-July every year, while the upper part of the other plants were cut out of the ground by a sickle in mid-December every year. All the plant samples were taken to the laboratory and washed with tap water and then rinsed with ultrapure water. They were then put into an oven at 80 C and dried until the weight no longer changed. Then, the sample was smashed with a grinder and put into a plastic bag.
Soil samples were collected from each plot from an area of 20 cm Â 20 cm Â 17 cm, with three samples taken from each plot and then mixed together to form a mixed sample. These samples were air dried and sieved using a 5 mm sieve, and the resulting samples were used for the analysis of soil aggregates. Soil samples were collected from the top 17 cm at ve representative locations per plot and then mixed together to form a composited sample aer the plant harvest. The soil samples were divided into two parts: one part was dried and sied for the analysis of soil physical and chemical properties, and the other part was kept at À80 C for soil enzyme activity analysis.

Soil physicochemical and heavy metal analysis
Soil pH was determined based on the method of Xu and was measured using a pH meter (PHS-3CW-CN, Bante instrument, Shanghai, China). 10 Soil organic carbon (SOC) was determined by the Walkley-Black procedure. 25 Soil available nitrogen (N) and phosphate (P) were determined in the same way as Bingham, and soil available potassium (K) was measured in accordance with Olsen. 26,27 Total soil Cu and Cd were measured in accordance with the method used by Cui, and a standard soil sample (GBW07405, National Research Center for Certied Reference Materials, China) was used to ensure the reliability of the experimental data. 28 The available Cu and Cd in soils were extracted with 0.01 mol L À1 CaCl 2 and determined in accordance with the method of Cui. 19

Aggregates analysis
Mechanically stable aggregates. The mechanically stable aggregates was measured by the dry sieve method. 29 The airdried soil was put on the sieve with apertures of 5 mm, 2 mm, 1 mm, 0.5 mm and 0.25 mm. Then the aggregates were divided into six grades: >5 mm, 2-5 mm, 1-2 mm, 0.5-1 mm, 0.25-0.5 mm, and <0.25 mm. Then the proportion of each grade aggregate to soil weight was calculated based on the results.
Water stable aggregates. The water stable aggregates were measured by the method of Bearden. 30 100 g of soil samples were prepared according to the dry sieve ratio, placed in a 5 mm soil sieve and soaked in distilled water for 10 min. The soil samples were then passed through 2 mm, 1 mm, 0.5 mm, and 0.25 mm soil sieves, respectively. The aggregates were separated by moving the sieve 3 cm upward and downward 50 times (2 min); and then, the soil particles on the sieves were rinsed into the aluminum box, dried at 50 C and weighed.
The contents of >0.25 mm mechanically stable aggregates (DR 0.25 ) and >0.25 mm water-stable aggregates (WR 0.25 ) were calculated based on the results. The mean weight diameter (MWD, mm), geometric mean diameter (GMD, mm) and fractal dimension (FD) of aggregates were calculated by Zhao's method. 31,32 Wi ln Xi X i is the average soil aggregate diameter at any level, equal to the average value of the adjacent two sieve hole stages. The upper limit of the diameter of >5 mm aggregates is 10 mm. X max is the average particle size of the maximum particle size, mm, M(r < X i ) is the weight of aggregates smaller than X i , and M is the aggregate weight. With lg i X max as abscissa and lg Mðr \X i Þ M as ordinate, the slope is calculated by linear tting with the least squares method. Finally, the fractal dimension (FD) of mass is calculated from the slope.

Soil enzyme activity
The activities of b-glucosidase (BG), N-acetylglucosaminidase (NAG), and acid phosphatase (AP) were measured by the method of Saiya-Cork. 33 Soils were assayed at ambient pH by suspending approximately 1 g of soil in 100 mL of 50 mm sodium acetate buffer. The microplates were incubated in the dark at 20 C for 4 h. During the incubation, the incubation plates were shaken every hour to ensure the homogeneity of the reaction mixtures. To stop the reaction, a 1 mL aliquot of 1 M NaOH was added to each well.

Statistical analysis
All the data were presented as mean AE standard error and were estimated using one-way ANOVA at a signicance level of 0.05 using SPSS 20.0 (IBM SPSS, Somers, NY, USA) when necessary. All the graphics in this article were made with Sigmaplot 12.5.

Soil chemical characteristics and heavy metal availability
In the untreated soil (CK), the soil pH decreased from 4.24 to 4.20 during the three years ( Table 1). The soil pH increased signicantly from 4.24 to 5.17 when hydroxyapatite was added. This might be due to the high pH (8.40) of hydroxyapatite. This was consistent with Cui et al., who found that the soil pH could be improved from of the addition of hydroxyapatite. 28 At the same time, we found that plant growth did not signicantly affect the soil pH (Table 1), although the plants might have secreted some weak organic acid ions, amino acids, vitamins, and inorganic ions (HCO 3 À , OH À , and H + ) by the roots, which could change the soil pH. 34 In the same way as the CK treatment, we found that the soil pH treated with hydroxyapatite also decreased slightly over time. This may be due to the fact that our experimental area was located in a typical acid rain area and that H + in the atmosphere entered the soil, thus the reducing of the soil pH. The SOC concentration of the combination Table 1 Soil chemical characteristics after the harvest of four plant species during the three years. CK SOC, soil organic carbon; T-N, total nitrogen; T-P total phosphorus; CEC, cation exchange capacity; T-Cu, total Cu concentration; T-Cd, total Cd concentration. Different lowercase letters indicate significant differences between treatments at the same time (n ¼ 3, treatments (ME, MS, MP) increased signicantly compared with CK aer three years of remediation (Table 1). However, applying hydroxyapatite alone (MW) had no signicant effect on SOC. This might be due to how the growing of plants could increase the amount of litter and ne roots and change the structure of soil aggregates, leading to an increase in SOC content. 35 The phosphorus content in each treatment was signicantly higher than that in the untreated soil because of the addition of hydroxyapatite. In southern China, phosphorus is generally decient in soil, and the addition of hydroxyapatite is benecial to mitigate the adverse effects of phosphorus deciency in plants.
It is known that the harmfulness of heavy metals in soils is mainly determined by their availability and mobility, CaCl 2extractable heavy metals can be used as an index to measure the availability of heavy metals in soil. 36 Our experimental results indicated that untreated soil had the highest CaCl 2 extractability (Cu 81.6 mg kg À1 , Cd 0.125 mg kg À1 in 2013). The addition of hydroxyapatite signicantly reduced the available Cu and Cd in the soil; the lowest extractable Cu by CaCl 2 (Cu 43.5 mg kg À1 ) was in Sedum plumbizincicola plots and Cd by CaCl 2 (Cd 0.089 mg kg À1 ) was in Pennisetum sp. plots ( Table 1). The results showed that potential mobility of Cu and Cd in the control was higher than that in hydroxyapatite treated soils, which might be mainly due to the lower pH in the CK.

Biomass and metal accumulation
In our study, the native S. lutescens and the three phytoextractors were able to grow normally only aer hydroxyapatite application. Among the four plants grown, the biomass of Pennisetum sp. was the largest followed by Elsholtzia splendens, Setaria lutescens and Sedum plumbizincicola ( Table 2). As a hyperaccumulator of cadmium, Sedum plumbizincicola exhibited a high absorptive capacity for Cu (451.5 mg kg À1 ) and Cd (13.7 mg kg À1 ) in our study, which were 13.8 and 11.8 times that of Setaria lutescens, respectively. E. splendens (a Cutolerant plant) also showed a high absorption capacity for Cu and Cd, which reached 202 mg kg À1 and 2.59 mg kg À1 , respectively (Table 2, Fig. 1).
According to Pedron, phytoremediation can be considered as a successful treatment for soil contaminated with heavy metals if plants are able to reduce soil metal concentration over time through the uptake processes. 37 Thus, "removal efficiency" should be calculated using tissue concentration and biomass produced to illustrate the treatment effectiveness of phytoextraction. 38 With regard to the total accumulation of Cu and Cd, Pennisetum sp. showed the greatest advantage, with the threeyear cumulative amounts of 8.67 Â 10 3 g h m À2 and 121 g h m À2 , respectively ( Table 2). The integrated results of plant biomass and accumulation capability showed that E. splendens and S. plumbizincicola had similar absolute accumulations of Cu and Cd. The poor biomass and accumulation ability of S. lutescens showed that this native plant species had the lowest absolute accumulation. In terms of absolute accumulation concentration, the remediation efficiency of different plants ranked Pennisetum sp. > E. splendens > S. plumbizincicola > S. lutescens. Based on our data, we would suggest that intercropping S. plumbizincicola with Pennisetum sp. or E. splendens might be a useful approach to removing Cu and Cd from soils. Nevertheless, more experiments are still needed to verify this hypothesis.

Soil aggregates structure and stability
As a basic component of soil, aggregates play an important role in the transportation of water, nutrients and air in soil. The improvement of soil aggregate stability is conducive to the progress of these processes. 39 In the process of evaluating the stability of soil aggregates, the average mass diameter, geometric mean diameter and aggregate stability rate are commonly used indicators. Using these indicators, we can objectively evaluate the stability of soil aggregates. 40 The content of the soil with mechanically stable aggregates was 69.4-76.6% before the harvest of vegetation in 2015 (Table 3), which was less than a lot of previous studies. 41,42 This indicated that the physical structure of the soil in this area was poor, which might Paper be related to serious soil pollution and difficult growth of plants in the area, resulting in the slight desertication of soil and the deterioration of soil structure. 43 The content of DR 0.25 in soil signicantly increased aer 3 years of remediation for all three kinds of plants; moreover, the 3 plant treatments mainly increased the mechanically stable aggregate content of >5 mm, 2-5 mm and 0.5-1 mm, while especially increasing the content of >2 mm aggregates. This showed that the remediation of the three plants had a signicant promoting effect on the formation of soil with >0.25 mm mechanically stable aggregates, which was mainly achieved by increasing the content of the >2 mm mechanically stable aggregates. Among the 3 plants, the maximum increase of >0.25 mm aggregates was Pennisetum sp., especially for the >5 mm and 2-5 mm aggregates, which increased by 29.0% and 10.1% compared with the CK treatment, respectively. The results showed that Pennisetum sp. had the largest increase in mechanically stable aggregates and had the best effect on improving soil physical structure. The percent content of waterstable aggregates for different treatments were 44.7-52.6% (Table 3). Similar to mechanically stable aggregates, all three kinds of plants could signicantly increase the content of waterstable aggregates. The increase was mainly concentrated in the >5 mm aggregates, and it was not signicant in other particle  sizes. This was the same as that of Zheng, who found that the vegetation restoration process mainly increased the content of >5 mm aggregates. 44 The mean weight diameter (MWD) and geometric mean diameter (GMD) are important indexes for evaluating the stability of soil aggregates. The increase of MWD and GMD values can represent the increase in soil aggregate stability. 45 The MWD and GMD of mechanically stable and water-stable aggregates increased signicantly aer the 3 years of remediation in our study ( Table 4). The MWD and GMD values of both mechanically stable and water-stable aggregates increased to the greatest extent. In terms of MWD, MP treatment increased the amount of mechanically stable and water-stable aggregates by 16.2% and 24.2% compared with CK treatment. In terms of GMD, the GMD of mechanically stable and water-stable aggregates, compared to CK, increased 29.1% and 25.0% by MP treatment, respectively. The results proved again that the application of hydroxyapatite and Pennisetum sp. in the phytoremediation of degraded heavy metal contaminated soil could improve the stability of soil aggregates and the physical structure of the soil. Moreover, the MWD and GMD of the mechanically stable aggregate were greater than those of water-stable aggregates in all the treatments, which indicated that the mechanically stable aggregates were the main aggregate type in the soil. These results were consistent with those of Zhang et al. 46 Fractal dimension (FD) is a new index to used evaluate the comprehensive soil structure. It can reect the stability of soil aggregates while reecting the uniformity of soil texture. The lower the fractal dimension, the more benecial to the improvement of soil nutrient circulation and structure. 47 The study found that the fractal dimension (FD) of mechanically stable aggregates could be signicantly reduced by the restoration of 4 planting. The range of the reduction range was 1.59-3.81% (Table 4). This showed that aer 3 years of vegetation restoration, the particle size composition of the soil aggregates was more homogeneous and the physical structure of soil had been improved.
Soil organic carbon is an important index for evaluating soil quality, which has an important inuence on the formation and cementation of aggregates. As an existing place for soil organic carbon, aggregates play an important role in the storage of organic matter and the transport of water vapor. Therefore, soil organic carbon and aggregates are inseparable. 48 Aer 3 years of vegetation restoration on this heavy metal contaminated soil, the content of soil organic matter and >0.25 mm aggregates increased signicantly, while the improvement of Pennisetum sp. treatment which had the highest biomass per unit area was the most signicant. Regression analysis showed that soil organic matter content was positively correlated with >0.25 mm mechanically stable aggregates (DR 0.25 ) and water-stable aggregates (WR 0.25 ) (R 2 ¼ 0.550*, 0.504*). This result was in accordance with the results of Zhu et al., who found that the main reason for the formation and increase of large aggregates was the increase in organic matter content. 49 The restoration of vegetation improved the content of organic matter and organic residues in the soil, and the smaller aggregates in the soil formed a larger aggregate by cementing organic carbon, mycelium nuclei and plant residues in the soil. 50

Soil enzyme activities
Soil enzyme activity is a direct indicator of soil microbial activity in response to metabolic requirements and available nutrients; it is especially useful for evaluating the impact of heavy metal pollution in soil. 51 The untreated soil showed very low BG and NAG activities, indicating a poor functional ability to catalyze the decomposition and transformation of soil carbon and nitrogen. The activities of BG, NAG and AP increased signicantly by 205%, 114% and 17.4% in E. splendens soil, respectively, as compared with the control (Fig. 2). But the activities of these three enzymes were all at low levels in the Setaria lutescens plot. In contrast to BG and NAG, AP activity was only slightly affected by the treatments except for the E. splendens plots. This might be due to the fact that acid conditions were favorable to acid phosphatase activity. 52 Previous studies have reported that heavy metals in soils can inhibit enzyme activities (1) by their toxic effects on soil microora, (2) by combining with the active groups of the enzymes (3) through the complexation of the substrate, and (4) by reacting with the enzyme-substrate complex. 53  Interestingly, in addition to the signicant correlation between NAG activity and CaCl 2 -extractable copper, there was no signicant correlation among the other two enzyme activities and CaCl 2 -extractable copper and cadmium (Table 5). BG and NAG activity was mainly signicantly positively correlated with soil pH and CEC but negatively correlated with total Cu in the soil. The soil pH and cation exchange affect the immobilization of enzymes in the soil. 56 The potential for biological activity as indicated by the enzyme activity may be attributed to root residues and root exudates (such as amino acids, sugars, phenolics, polysaccharides and proteins). Such processes could improve soil physical and chemical properties and cause changes in the composition and function of soil microbial communities. 57 In our study, soil pH was an important factor that restricted plant growth; the increase of soil pH could promote the activity of soil BG and NAG activity. AP contributes to the transformation of organic phosphorus to inorganic phosphorus, thereby enhancing the absorption of inorganic phosphorus by plants. 58 The synthesis of new phosphatase or the phosphatase activity in the soil can be inhibited by inorganic phosphate. 59 In this study, the addition of hydroxyapatite led to the increase of phosphate, which could lead to the decrease of AP activity. On the other hand, the addition of hydroxyapatite led to the increase of soil pH and to the reduced toxicity of heavy metals in the soil, which could improve the activity AP. The combined effect resulted in a small increase in the activity of AP (Fig. 2c).

Conclusions
This study demonstrates the benets of combining hydroxyapatite application and phytoextraction for the improvement of soil quality when remediating heavy metals. Four plant species were successfully established in the hydroxyapatite-amended soil and produced different amounts of aboveground biomass in the order of Pennisetum sp. > Elsholtzia splendens > Setaria lutescens > Sedum plumbizincicola. Results indicated that Pennisetum sp. was the best species for Cu and Cd removal from the contaminated soils. The application of hydroxyapatite and four plant treatments could improve the content of >0.25 mm mechanically stable (DR 0.25 ) and water-stable (WR 0.25 ) aggregates and the stability of soil aggregates, the largest increase being with the treatment of Pennisetum sinese. In addition, hydroxyapatite and phytoremediation could improve soil enzyme activity, and Elsholtzia splendens had the best effect in this respect. In conclusion, Elsholtzia splendens and Pennisetum sp. may be the best choices for the remediation of this type of heavy metal contaminated soil.

Conflicts of interest
There are no conicts to declare. Table 5 Correlation coefficients among soil biological properties, soil chemical properties, and CaCl 2 -extractable Cu and Cd. (BG, b-1,4glucosidase; NAG, b-1,4-N-acetylglucosaminidase; AP, acid phosphatase; CEC, cation exchange capacity; T-Cu, total Cu; T-Cd, total Cd; C-Cu, CaCl 2 -extractable Cu; C-Cd, CaCl 2 -extractable Cd; SOC, soil organic carbon; T-N, soil total nitrogen; T-P soil total phosphorus. All samples of all plots included in these correlation analyses (n ¼ 18). ** Correlation is significant at the 0.01 level; * correlation is significant at the 0.05 level)