We collected as many different turnip landraces as possible from the main planting areas in China and then numbered them. From these, the landrace “KTRG-B19” from Mianning Country, Sichuan Province of China was randomly selected to investigate the tolerance ability to Cd stresses. The seeds were sown in a seedling-raising plate under natural conditions with appropriate watering. When the seedlings grew to the trefoil stage, they were transplanted to uniform watertight pots (length: 30 cm; width: 20 cm; height: 15 cm) in the greenhouse (12-h light/12-h darkness, 22°C, 50–60% relative humidity). The humus soil was used. Each pot was planted with nine seedlings with consistent growth.

After a 2-week recovery period, the plants were treated with CdCl₂ solution. In order to select an appropriate Cd concentration for the later material treatment, we set a soil Cd²⁺ concentration gradient of 0, 1.22, 3.06, 6.12, 12.24, 30.60, and 61.20 mg kg^-1 (DW), respectively. After treatment for 3 weeks, the plant phenotype was recorded and the AG part of the three plants for Cd²⁺ treatment concentrations of 1.22, 6.12, 30.60, and 61.20 mg kg^-1 was collected to measure the Cd content.

According to the results of above Cd treatment, we selected the concentration of 50 mg kg^-1 Cd²⁺ in the mucky soil (DW) to explore the Cd accumulation differences among different landraces. A total of 18 turnip landraces from different regions were used for the experiments (Table 1). These landraces showed clear differences in morphological characteristics (Supplementary Figure S1), which might indicate various genetic resources. Two uniform watertight boxes (length: 64 cm; width: 44 cm; height: 26 cm) were prepared for each landrace. One box was set as the control group and the other was used for the treatment group. We used CdCl₂ to provide Cd²⁺ and it was mixed thoroughly with the soil. Each box was equipped with 15 kg of dried soil.

Seeds of each turnip landrace were regularly sown in nine dots (3 × 3) in both the control and treatment boxes. The boxes were placed under natural light and temperature, with appropriate watering. After seedling emergence, a total of nine seedlings with consistent growth remained (one seedling at each dot). After growing for a month, the AG and UG parts of the plants were collected, respectively, and the roots were cleaned using ultrapure water. The samples were dried under 80°C for 48 h and their weights were recorded. The AG and UG parts of three plants of each landrace were then used to measure the Cd content.

To draw the standard curve of Cd content, the Cd standard solution (1 mg mL^-1) was diluted with 5% HNO₃ to 10 mg L^-1 stock solution. The stock solution was then prepared into 0, 0.1, 0.2, 0.4, and 1 mg L^-1 standard solutions and mixed for detection. The solutions were detected by an Inductively Coupled Plasma Spectrometer (ICP-OES optima 8000, Perkin Elmer, US) using the wavelength of 214.44 nm. The standard curve was drawn only when the linear correlation coefficient was greater than 0.99.

Approximately 0.2–1.0 g dried samples were added to polytetrafluoroethylene digestion tanks, then, 5 mL HNO₃ was injected and the tanks were left to stand. When the reactions finished, the tanks were sealed with caps and put into a microwave digestion instrument (WX-8000) using the following digestion procedure: 100°C, 3 min; 140°C, 3 min; 160°C, 3 min; 180°C, 3 min; 190°C, 15 min. When the temperature cooled below 50°C, the digestion tanks were taken to the fume hood. The digestion solutions were transferred to 50 mL volumetric flasks and fixed volume to 50 mL by rinsing three or four times using ultrapure water. The blank control was treated using the same method. The sample solutions were detected under the same wavelength and the Cd contents were calculated according to the standard curve.

Statistical analyses were performed using SPSS version 18.0. One-way ANOVA was used to analyze significant differences among multiple samples and an independent-samples t-test was used between each pair of samples. All significant differences were identified at 0.05 levels. Linear regression analysis was used to identify the correlations.

When treated by Cd at different concentrations for 3 weeks, the morphology of turnip plants showed no significant differences compared with the control materials (Figure 1A). As the Cd concentration added to the soil increased, the Cd content accumulated in the AG part of plants was improved (Figure 1B). Furthermore, these two indexes showed a significant linear relationship (F = 160.3184, P < 0.001; Figure 1B).

Because turnip plants could tolerate at least 61.20 mg kg^-1 Cd²⁺ in soil (Section Screening Test for Cd Concentration), we selected the soil concentration of 50 mg kg^-1 Cd²⁺ to investigate the tolerance and accumulation abilities of different turnip landraces. When growing for the same time after sowing (a month), we found that the 18 turnip landraces showed substantial differences in plant growth under the control condition (Figure 2). The average total biomass ranged from 0.73 (KTRG-B56) to 4.00 g (KTRG-B25). When treated by Cd, the biomasses of all 18 landraces decreased by diverse degrees compared with their respective control materials (Figure 2) and thus the mean total biomass of plants in the Cd-treated soil also varied between 0.41 (KTRG-B50) and 3.19 g (KTRG-B13; Figure 2). Among all landraces, the growth of nine landraces (KTRG-B03, KTRG-B16, KTRG-B19, KTRG-B22, KTRG-B25, KTRG-B28, KTRG-B31, KTRG-B36 and KTRG-B54) was significantly inhibited by Cd (Figure 2), whereas the remaining landraces (KTRG-B06, KTRG-B13, KTRG-B14, KTRG-B45c, KTRG-B48a, KTRG-B48b, KTRG-B50, KTRG-B56 and KTRG-B57) showed a stronger tolerance to the Cd concentration used in this study (Figure 2).

The Cd accumulation concentrations in both the AG and UG parts were different among the investigated landraces (Figure 3). The average Cd concentrations in the AG part ranged from 52.94 (KTRG-B25) to 146.95 mg kg^-1 DW (KTRG-B19) with a mean value of 99.48 mg kg^-1 DW (Figure 3), which were greater than the concentration in the soil. Specially, Cd accumulation exceed 100 mg kg^-1 in the leaves of KTRG-B14 (125.27 mg kg^-1), KTRG-B16 (141.25 mg kg^-1), KTRG-B19 (146.95 mg kg^-1), KTRG-B45c (116.63 mg kg^-1), KTRG-B50 (105.88 mg kg^-1), KTRG-B54 (118.47 mg kg^-1), KTRG-B56 (139.87 mg kg^-1) and KTRG-B57 (106.27 mg kg^-1), whereas most of the others were very close to this value (Figure 3). The average Cd concentrations in UG part ranged from 8.17 (KTRG-B54) to 81.52 mg kg^-1 DW (KTRG-B16; Figure 3) and the mean value was 44.03 mg kg^-1 DW (Figure 3). The scope of this variation was much greater than that of the AG part (Figure 3). For each turnip, the Cd concentration in the AG part was higher than that in the UG part (Figure 3). To understand the absorption and transportation characteristics, two parameters including EC and TF were introduced. In the present study, the ECs of AG part of all landraces were greater than 1 (Table 1), and the mean value was 1.99. In detail, KTRG-B16 (2.83), KTRG-B19 (2.84) and KTRG-B56 (2.80) showed high enrichment ability whereas KTRG-B13 (1.42), KTRG-B25 (1.06) and KTRG-B28 (1.22) had low ECs (Table 1). Similarly, the TFs of all landraces were also greater than 1 (Table 1), with values that ranged from 1.36 (KTRG-B25) to 4.82 (KTRG-B54) with a mean value of 2.61 (Table 1).

Regression analyses indicated that the Cd concentrations in the AG part of turnip plants were significantly correlated with the corresponding biomasses (F = 8.6456, P = 0.0049; Figure 4A). However, the correlations in UG part were not significant (F = 3.9475, P = 0.0523; Figure 4B). To further assess the Cd accumulation effects of different turnip landraces, we estimated the total Cd content accumulated by individual plant using the following formula: C_T= C_AG×B_AG+C_UG×B_UG where C_T is the total Cd content (μg); C_AG and C_UG is the Cd concentration in the AG and UG parts (μg g^-1), respectively; and B_AG and B_UG is the biomass of the AG and UG parts (g), respectively. The results showed that total Cd content of the single plant was different from each other among 18 landraces (Figure 5). Individual plants could enrich a minimum of 41.08 (KTRG-B50) and maximum of 215.92 μg (KTRG-B13) Cd element (Figure 5), and the mean accumulation content was 105.17 μg.

Significant interspecific differences in plant absorption and accumulation of HMs have been demonstrated. Arthur et al. (2000) divided plants into three categories based on their HM absorption and accumulation capabilities: high accumulation type, moderate accumulation type and low accumulation type. Turnip has previously been described as belonging to high Cd accumulation plants (Arthur et al., 2000); however, supporting data and evidence were limited. According to a previous survey, the mean Cd concentration in soil around the world was about 0.35 mg kg^-1 (He et al., 1998), and the Cd concentration in normal soil usually did not exceed 1 mg kg^-1. In this study, a Cd concentration gradient from 1.22 to 61.20 mg kg^-1 was used to treat turnip plants, which could represent mildly, moderately and severely contaminated soil. The unscathed morphology compared with the control samples indicated that turnip plants had very strong tolerance to Cd stress. Furthermore, Cd content absorbed by turnip increased with the increasing Cd concentration in soil. At the highest Cd treatment concentration, Cd concentration in turnip plants (69.80 mg kg^-1) was close to the critical value of Cd hyperaccumulators (Baker and Brooks, 1989). Our results indicated that this turnip landrace investigated (KTRG-B19) had strong accumulation ability for soil Cd²⁺.

We next examined a total of 18 turnip landraces to further confirm their absorption capacities. The Cd concentrations accumulated in AG parts of all turnip landraces were more than 50 mg kg^-1, and those of eight landraces exceeded 100 mg kg^-1 (Figure 3). The EC value is usually used to reflect the ability of plants to accumulate environmental pollutants, which is indicated as the ratio of pollutant concentration in plants to that in the environment (He, 2013). In the present study, Cd ECs of AG parts of all turnip landraces were greater than 1 (Table 2). In addition, another notable finding of our study was that the accumulation time was within a month, which indicated that Cd absorption and accumulation of turnip might be further enhanced as the plant continues to grow. Previous studies provide evidence to support this speculation; for example, Niu (2012) reported that Cd accumulation concentrations in AG parts of six rape landraces gradually increased from the seedling stage to maturity stage. In addition, Chen (2013) found that Cd concentration in Bidens pilosa roots increased during the 30- to150-day life stage under various Cd treatment concentrations.

Overall, our results have clearly shown that turnip has a high Cd accumulation ability, which supports the conclusion of Arthur et al. (2000). However, the maximum value of Cd accumulation in turnip still requires further study. In addition, whether turnip has strong ability in absorbing other HMs is unclear and also requires additional research.

A noticeable intraspecific variation in plant accumulation of Cd has also been reported. High and low Cd accumulation cultivars have been identified in many crops, such as rape (Niu, 2012; Wu et al., 2015), pakchoi (Zhou et al., 2016), watercress (Wang et al., 2015), and Chinese leaf mustard (Dai et al., 2012). Differences in Cd absorption and accumulation among different turnip cultivars or genotypes also have been indicated in studies around the world (Bingham et al., 1975; Molahoseini and Feizi, 2012; Parveen et al., 2013). The results of the present study further demonstrated that there are intraspecific differences of Cd tolerance and accumulation in turnip. Figure 2 shows that plant growth varied significantly among different turnip landraces under normal conditions. Under the same Cd conditions, plant growth of different turnip landraces was inhibited to various degrees (Figure 2). The results suggested that these turnip landraces had diverse tolerances to Cd stress and different Cd accumulation abilities. The maximum Cd concentration (KTRG-B19) in the AG part was 2.78 times the minimum value (KTRG-B25) while the ratio of the highest Cd concentration (KTRG-B19) and the lowest value (KTRG-B28) was 4.33 in the UG part. Interestingly, we found that the AG biomass was significantly negatively correlated with the corresponding Cd concentration (Figure 4A). This finding indicates that inhibited plant growth of some turnip landraces might result from large absorption of Cd. However, this was not observed in the UG part (Figure 4A), which could be attributed to the large TFs that represented a strong transfer capability of Cd from the plant UG part to the AG part (He, 2013).

As a vegetable plant, intraspecific differences of Cd accumulation might be of great relevance to human life. On the one hand, for food safety, cultivars with low capacity of Cd accumulation or Cd pollution-safe cultivars are an economical and effective strategy to restrict Cd transfer into the food chain. On the other hand, cultivars with high capacity of Cd accumulation are excellent candidates for the phytoremediation of contaminated soil. For turnip, all the landraces measured in the present study belong to the high Cd accumulation type. This suggests that it is not advisable to consume turnips cultivated in environments where safe Cd levels are exceeded. Moreover, based on the different results from different countries (Page et al., 1972; Bingham et al., 1975; Mani and Kumar, 2004; Mani et al., 2007; Molahoseini and Feizi, 2012; Parveen et al., 2013), we should consider comparing different turnip cultivars or genotypes worldwide and screening for low and high-Cd-accumulating cultivars. As mentioned before, the large intraspecific differences of Cd accumulation in turnip might be caused by various genetic backgrounds. However, the environmental Cd concentrations also influenced the Cd uptake in turnip plants in different studies (Page et al., 1972; Mani and Kumar, 2004). In addition, environmental conditions including medium, pH, organic matter and other ion contents could affect the absorption of turnip to Cd to different degrees (Mani and Kumar, 2004; Mani et al., 2007; Parveen et al., 2013). Therefore, experiments performed under the same environments are necessary in future studies.

As shown in Figure 3, the Cd concentrations in the AG parts of most landraces exceeded or approximated 100 mg kg^-1 DW. However, it was still not rigorous to classify these landraces as Cd hyperaccumulators. Based on the current definition, Cd hyperaccumulators must satisfy a further three requirements including that both the EC and TF should greater than 1 (He, 2013). Furthermore, the plant growth should not be significantly restrained when subjected to Cd-polluted soil (He, 2013). Based on these requirements, we identified five turnip landraces (KTRG-B14, KTRG-B45c, KTRG-B50, KTRG-B56 and KTRG-B57) that may be considered Cd hyperaccumulators (Figures 2 and 3, Table 2). In addition, three landraces (KTRG-B13, KTRG-B48a and KTRG-48b) were expected to be classified in Cd hyperaccumulators; their biomasses were not significantly reduced under the Cd treatment but their Cd concentrations did not exceed 100 mg kg^-1. Nonetheless, the Cd concentrations in their AG parts were expected to be more than 100 mg kg^-1 DW under a longer growth period.

Phytoremediation is often considered as the most promising remediation approach of HM contaminated soil as it is a cost-effective and environmentally friendly technology (Kramer, 2005; Long et al., 2013). Phytoremediation can mainly be divided into phytoextraction and phytostabilization according to their respective principles (Ferraz et al., 2012). Plant extraction is directly dependent on plants absorbing and transferring HMs to ground parts to remove them (Pilon-Smits, 2005). The key to plant extraction is to find HM hyperaccumulators with fast growth, large biomass and high accumulation capability (Ferraz et al., 2012). According to a Chinese review article (He, 2013), more than 500 HM hyperaccumulators have been found or reported around the world, but most of them (about 300 species) were nickel hyperaccumulators. Compared with these data, the number of Cd hyperaccumulators remains limited. There were just 20–30 Cd-hyperaccumulation plants that have been reported by 2010 (He, 2013). Thus, our findings for turnip should provide great resources for phytoremediation for Cd polluted soil. Several studies have shown that some species, such as Convolvulus arvensis (Gardea-Torresdey et al., 2004) and Salsola kali (de la Rosa et al., 2004), have strong Cd accumulation capabilities but do not strictly conform to the traditional definition of a Cd hyperaccumulator, as they showed higher Cd accumulation concentrations in UG parts than in AG parts. However, these plants were still suitable for soil remediation.

Besides Cd accumulation capability, plant biomass is also a key factor that decides the Cd removal efficiency, which is supported by our results. The total Cd contents in individual plant of the five Cd-hyperaccumulation turnip landraces (KTRG-B14, KTRG-B45c, KTRG-B50, KTRG-B56, and KTRG-B57) were not the highest (Figure 5). Conversely, those landraces with lower Cd concentrations (e.g., KTRG-B13 and KTRG-B28) in plants enriched much larger masses of Cd (Figures 3 and 5). Indeed, many Cd hyperaccumulators, especially some wild plants, usually grow slowly and have smaller biomasses, which severely limits their phytoremediation efficiencies (Rungruang et al., 2011; Mani et al., 2016). Like many other cruciferous vegetables, turnip has the biological characteristics of rapid growth and high biomass and is easy to harvest. Together with our results in the present study, turnip has great potential in phytoremediation for Cd polluted soil. In addition, territory restriction is another reason for preventing the application of some Cd hyperaccumulators in environmental remediation (Wei and Chen, 2001). However, turnip has the advantage of extensive cultivation regions, even including the Tibetan Plateau. Therefore, although it has not yet been applied in practical soil remediation, turnip is a promising candidate that requires further research attention.