Concentration of fifteen elements in herbaceous stems of Ephedra intermedia and influence of its growing soil

Mineral nutrients play important roles in the growth and metabolism of Ephedra intermedia, and are affected by soil factors. Fifteen elements were measured from wild E. intermedia as well as their growing soils using inductively coupled plasma mass spectroscopy to investigate the influences and characteristics of herb elements. The pH, cation exchange capacity, humus and soil mechanical composition were also determined in rhizosphere soils. Results showed that E. intermedia stems contained high N, low P concentrations in macronutrients and high Fe in micronutrients, and enriched N, S, Cl, P and Sr from soils. The 15 herb elements were affected by one or more soil factors, and K, P, Zn, Fe and Mn were important soil elements that influenced the mineral accumulation of E. intermedia. This study was useful for the artificial cultivation of wild E. intermedia.


Material and methods
Material. Samples Table S1). The herb samples were taken from green stems of wild E. intermedia. All plant materials were identified as herbaceous stems of authentic E. intermedia by Professor Hong Sui. Five quadrats were selected for collection at each sampling point (100 g plant per quadrat) to ensure that all plant samples were representative. The sampling area was 1 m × 1 m for every quadrat, and the distance between two quadrats was over 200 m. All voucher specimens were deposited in Room 207, Ningxia Research Centre of Modern Hui Medicine Engineering and Technology (Yinchuan, Ningxia, China). Soil samples were collected at the root growing location of collected plants in each sampling point (500 g soil per quadrat), and surface mulch was removed. Next, 2,500 g of soil samples was obtained at a topsoil depth of 20 cm, mixed fully, packed and labeled in each sampling point.
Standard stock solutions as well as internal standard solution were purchased from Beijing General Research Institute for Nonferrous Metals in China for inductively coupled plasma mass spectroscopy (ICP-MS). HNO 3 , HF and H 2 O 2 were of analytical reagent grade. Ultrapure water (Millipore Co., Ltd., U.S.) was used at a conductance of 18.25 MΩ cm. All glassware and plastic containers were thoroughly washed with HNO 3 and cleaned with ultrapure water prior to use. www.nature.com/scientificreports/ Sample preparation and measurement. The samples were prepared using the following protocols.
Herbaceous stems of E. intermedia were carefully rinsed with ultrapure water to remove dust particles. Soil samples were dried in an oven to reach a constant level. Dried samples were triturated into a homogeneous fine powder with a uniform particle size prior to analysis. For microwave-assisted digestion technique, a programmable 1,200 W microwave (MARS 5, CEM Co., Ltd., U.S.A) served as the digestion system, and 5 ml of HNO 3 along with 3 ml of H 2 O 2 were mixed with 1.000 g of plant samples in Teflon digestion vessels. Then, the following procedure was performed in order: 20 min in 120 ℃, 20 min in 160 ℃ and 45 min in 180 ℃. For analysis of elements in soil, 0.5000 g of sample was placed in Teflon digestion vessels with a mixture of HNO 3 (15 mL), HF (1 mL) and H 2 O 2 (2 mL). The following procedure was performed in order: 25 min in 150 ℃, 30 min in 170 ℃ and 80 min in 200 ℃. After digestion, the solution was dried at 150 ℃. The dry digest was cooled to room temperature, attenuated with ultrapure water and transferred to volumetric flasks to a final volume of 10 mL. Blank samples were produced using a uniform analytical process without the plants or soils.
Concentrations of mineral factors in the plant and soil samples were measured by ICP-MS (NexION 300D, PerkinElmer Instrument Co., U.S.) using JIS K0133-2007 Method 30 . All determinations were conducted in triplicate.
Soil characterisation measurement. All dried soil samples were ground and then sifted for analysis of physicochemical properties. Soil features studied included pH, CEC, humus and soil mechanical composition (SMC, involving silt, clay and sand). These characteristics were measured using the Chinese national standard methods in connection with soil 32-34 . Statistical analysis. The experimental data were processed using Excel 2010 and SPSS 21.0 statistical package. Multivariate data analysis techniques were applied, including correlation analysis (CA), coefficient of variation (CV), principal component analysis (PCA), hierarchical cluster analysis (HCA), multivariate analysis of variance (MAV) and discrimination analysis (DA).
CA and MAV were used to analyse the correlations and differences between the data by Pearson criteria and least significant difference (LSD), respectively. Statistical significance was recognised as a P value of less than 0.05, and high statistical significance had a P value of less than 0.01. CV was adopted to compute the deviation of element concentration levels between the soil and the plants 35 . PCA was conducted to extract factors from multivariate data, which made the data matrix less dimensional; original variables were converted into principal component data points 36 . HCA is used for grouping original variables and is an unsupervised pattern recognition technique 37 . HCA was performed using between-group linkage and squared Euclidean distance as the clustering method in this study. DA is a supervised pattern recognition technique to categorise samples by creating new variables 38 , thereby making the variance maximum between classifications and minimum within classifications.

Results and discussion
Soil characterisations. Soil characterisations including pH, CEC, SMC (silt, clay and sand) and humus from different samples were measured and listed in Supplementary Table S2. The soil samples were neutral or slightly alkaline with the pH ranging from 7.22 to 8.34. The pH was related to the presence of soil protons, which affected the ionic exchange of soil ingredients such as humic acids. The soil CEC varied from 12.05 to 44.06 mmol kg −1 . The sample with the maximal pH had the highest CEC (44.06), whereas the lowest CEC (12.05) was established in the neutral soil with a pH of 7.39. The CEC was the supreme quantity of cations at a certain soil pH level 39 and contributed to the soils' potential information to adsorb or release cations 19 .
The SMC distribution (0.05 ≤ sand < 2, 0.002 ≤ silt < 0.05 and clay < 0.002 mm) varied significantly in the soils (6.42-24.06% clay, 0.64-9.96% silt and 67.34-90.3% sand). The average humus concentrations were 0.71% in the soil samples (in the range of 0.24-2.93%). They differed in various sampling areas. All of them were important for estimating the fitness of the soil and provided specific environment for plant growth.  Supplementary Table S5.
The average element concentrations in E. intermedia samples followed the order Mo < Cu < Zn < B < Mn < Na < Sr < Cl < Fe < P < Mg < S < Ca < K < N. The elements of P, K, N, Ca, S and Mg were the macronutrients in the plants, and concentrations of N and K were much higher than those of other macronutrients and more than 6,000 ppm. The elements of Fe, Cl, Sr, Na, Mn, B, Zn, Cu and Mo were the trace elements in the plants, and concentrations of Fe, Cl, Sr and Na were much higher than those of other trace elements and more than 133 ppm. Levels of mineral nutrients differed evidently according to plant sampling sites 40,41 . The average total concentration of elements was 45,148.58 ppm with a variation from 34,628.23 ppm to 55,123.86 ppm in 16 plant samples. The maximal concentration was obtained for N, which ranged from 18,863.61 ppm to 34,283.81 ppm and possessed 59.77% of the elemental total concentration in E. intermedia. The ascending order of element concentrations in E. intermedia was consistent with that in E. sinica; nevertheless, the order of S was after that of Ca in E. sinica 31 . The average total concentration of elements in E. intermedia was more than that in E. sinica (34,863.55 ppm), and the N ratio in the elemental total concentration was higher than that in E. sinica (55.98%) 31 . As an essential macronutrient, P showed a lower concentration (mean 843.20 ppm) compared with the other macronutrients (mean ≥ 2,258.57 ppm). Iron had the maximum concentration in micronutrients, which varied from 365.03 ppm to 1,497.56 ppm; its mean was 780.98 ppm in E. intermedia, which was higher by 337.61 ppm than that in E. sinica 31  intermedia to obtain more Fe from its growing soil. Molybdenum concentration was less than that of other elements in E. intermedia, which was probably connected with its low availability and high affinity in soils. The average element concentrations in the soil samples followed the order Mo < Cu < B < Cl < Zn < Sr < S < P < N < Mn < Mg < Fe < K < Na < Ca. This order was different from that in the plants. The average total concentration of elements was 92,635.34 ppm with the differentiation from 41,553.26 ppm to 132,697.48 ppm in the soils. The soil element of maximal concentration was Ca, its mean was 66,026.66 ppm, and it accounted for 71.28% of the total concentration of elements. The top six element concentrations in the soil were Mn, Mg, Fe, K, Na and Ca; those in E. intermedia were P, Mg, S, Ca, K and N. Concentrations of Ca, Mg and K were much high in the soils or E. intermedia. As the plant required macronutrients, the demands of N, P and S were large for E. intermedia. Supplementary Tables S3 and S4 show that the concentrations of N, P and S in E. intermedia were higher than those in the soils; thus, E. intermedia were able to transport these three elements actively to meet their own needs. Concentrations of Mo and Cu in E. intermedia and its growing soil were lower than those of other elements, and their sequences of elements were the same in E. intermedia and its growing soils. However, concentrations of Mo and Cu in the soil were higher than those in E. intermedia. Thus, the absorption of Mo and Cu in E. intermedia should not be the result of passive transport.
Element variation in E. intermedia and its growing soil samples. For visual observation, the concentrations of plant and soil elements were converted logarithmically (Fig. 2a), and their CV (ratio of standard deviation to mean of all samples) was presented (Fig. 2b). Elemental concentrations varied greatly between the plants and the soils. Concentrations of Sr, Cl, P and S as well as N were significantly higher in E. intermedia (P < 0.01) than in the soils, and other elements in E. intermedia were significantly lower (P < 0.05) than those in the soil samples except for Mg. The concentration of Mg was not significantly different between the plant samples and the soil samples. intermedia were clearly unstable, and the CVs were large. By contrast, their concentrations in the soil were stable, and the CVs were small. Therefore, the absorption of these elements might be greatly affected by the aboveground environmental factors in the growth location of E. intermedia. The bottom six elements of CVs in E. intermedia were Cl, S, Ca, B, Na and Zn. The top six elements of CVs in the soil were Na, S, Mo, B, Sr and Ca. Thus, concentrations of S, Ca, B and Na were stable, and their CVs were small in E. intermedia. By contrast, their concentrations were not stable, and the CVs were large in the soil samples. Therefore, the absorption of S, Ca, B and Na might be mainly related to biological characteristics of E. intermedia rather than its soil environments. The largest CV and ratio of maximum to minimum (RMM) were observed in Fe of E. intermedia (Supplementary Table S5). Iron shows an important influence on animal energy metabolism, and it is the component of human hemoglobin and myoglobin 43 ; Iron might be a curative indicator for E. intermedia. Phosphorous was similar to Fe in E. intermedia, that is, its RMM and CV were lower than those of Fe but higher than those of other elements. The maximum and minimum concentrations were higher for N, S, P, Cl and Sr in E. intermedia than in the soil samples. Potassium and magnesium minimum concentrations in herbs were also more than both in the soil samples. Overall, most elements' maximum or minimum concentrations and CVs in the soils were higher than those in E. intermedia. CVs displayed a high element dependence for E. intermedia or soils. CVs of Fe and P varied by > 34% in E. intermedia, whereas CVs of Zn, Na, B, Ca, S and Cl varied by < 15%, which meant they were relatively constant. However, CVs of Na, S, Mo and B were > 41% in the soils, whereas CVs of Mn and Fe were < 20%. CV was consistent with the stability of limiting elements in E. intermedia or the soil sample. The CV of Na in the soils was approximately 70%, which was the largest among soil elements; thus, it was the most sensitive element in the soil. However, the CV of Na in E. intermedia was steady and only 14.15%. Element accumulation in E. Intermedia. Enrichment coefficients were computed to indicate the uptake and storage behaviour of elements in wild E. intermedia, which reflected the soil-plant system element migration rate to some extent, and investigate the capacity of E. intermedia to store elements from their growing soil. Enrichment coefficient (EC) was described in the following formula for the soil-plant system 27 : where [M plant ] represents the average concentration of an element in the herbaceous stem of E. intermedia, and [M soil ] represents the same element average concentration in the soils.
ECs of mineral nutrients of E. intermedia varied widely (Fig. 3). Element ECs followed the order Na < Mn < Ca < Cu < Fe < Zn < B < K < Mo < Mg < Sr < P < Cl < S < N. ECs of N, S, Cl, P and Sr were greater than 1 and higher than those of other elements. Concentrations of these elements in E. intermedia were all higher than those in the soil, which was the result of the active transport of E. intermedia. ECs of Na, Mn and Ca were less than 0.06 and smaller than those of other elements. Concentrations of these elements in the soil were much higher than those in E. intermedia. The ascending order of ECs in E. intermedia was nearly consistent with that of E. sinica other than P, Cl, K, Mo, Ca and Fe 31 . E. intermedia accumulated the highest amount of N with an EC of 50.52 and showed a good tendency to accumulate S, Cl, P and Sr, whose ECs were 22.42, 5.45, 3.21 and 1.60 respectively, which might be involved in meeting the specific nutrition needs of E. intermedia. ECs of other elements were low (< 1), and Na accumulation was the lowest with a logarithm EC of − 2.

Correlation analysis between plant elements and soil compositions. CA between E. intermedia
and the soil samples was carried out to investigate the influence of the soil on the plant. The results are shown in Tables 1 and 2. Of the 225 element CAs between E. intermedia and the soil samples (Table 1), 49 had a significant correlation; P, K, Mn and Ca in E. intermedia were positively correlated with those in the soils. In E. sinica, K, Mn and Ca were also related to those in the soils, and the same CAs showed 102 significant correlations; thus, www.nature.com/scientificreports/ elements of E. intermedia were affected to a lesser extent by soil elements than those of E. sinica 31 . Chlorine and strontium in the plants were not related to any element in the soil samples. Nitrogen as well as Cu in E. intermedia were correlated with seven elements in the soils, and the correlative number of elements was maximal in the plants. Molybdenum, potassium and phosphorous in the herbs were related to six, five and five elements in the soils, respectively. All correlative elements in the plants were significantly positively related to the designated elements in the soils except S. Only S in the plants was significantly negatively correlated with Zn in the soils. Iron in the soil samples was associated with eight elements in E. intermedia and held the maximal number of correlative elements. Manganese and zinc in the soils were correlated with seven and six elements in the herbs, respectively. Sulfur and chlorine in the soils were not related to any element in the plants. Therefore, N, P, K, Cu and Mo in E. intermedia were easily influenced by elements in the soils, and Fe, Mn and Zn in the soils were important for E. intermedia. The uptake of mineral nutrients in a plant was controlled by soil characterisations and plant biological properties 20 . CAs between herb elemental concentrations, and soil characterisation factors demonstrated that humus, CEC and sand were more important than other characterisation factors in the soils (   Tables 3 and 4. According to Table 3, the first six principal components (PC1 ~ 6) explained 87.511% of the soil data, and contributed 33.639%, 17.337%, 14.445%, 9.576%, 6.943% and 5.571% to the total variance. Soil PCA proved that P, K, Zn, B, silt and Fe were dominantly incorporated in PC1. Humus was highly weighted in PC2. Nitrogen along with Mn contributed to the maximum loading for PC3. PC4 was formed by variable Mo, and all element weights were low in PC5 and PC6. Therefore, the factors of silt, humus, N, Zn, K, B, P, Fe, Mo and Mn were considered characteristic factors in the growing soil of wild E. intermedia, and their loading weights were all more than 0.7 in PC1 ~ 6. These factors might be recognised as the most potent markers to illustrate the influence of soil on plant.
PCA results of E. intermedia showed that 15 original variables were converted to five PCs that totally contributed 84.442% of all element data sets ( Table 4). The highly weighted factors under these PCs were Cu, B, Zn, Mo, N, Mn, K, Fe, Ca and Cl. PC1 explained 41.375% of the data variation and was mainly affected by Cu, B, Zn, Mo, N and Mn as the highly weighted factors. PC2, which contributed to 14.855% of the studied variances, displayed the main loading factor for Fe. PC3 and PC4 accounted for 10.712% and 10.126% of the total variance, respectively. Calcium was the major loading factor in PC3, and Cl was in PC4. Moreover, PC5 revealed 7.374% of the total variance, and all element weights were low in PC5. Thus, N, K, Mn, Zn, Cu, B, Ca, Mo, Fe and Cl were considered the characteristic minerals in E. intermedia, and their loading weights were all more than 0.7 in PC1 ~ 5.
Hierarchical cluster analysis of plant and its growing soil samples. HCA was conducted for E.
intermedia and soil samples based on the between-group linkage method. The results are shown in Fig. 4. According to the concentrations of 15 elements, 16 herb samples were clustered into three at a distance of 7.5 (Fig. 4a),   www.nature.com/scientificreports/ which showed no difference from the cluster result according to 10 characteristic element concentrations from PCA (Fig. 4b). Based on 21 soil factors, 16 soil samples were grouped into four at a distance of 9.5 (Fig. 4c), which was not consistent with those of E. intermedia samples. However, based on the 10 soil characteristic factors from PCA, 16 soil samples were clustered into three at a distance of 9.5 (Fig. 4d), which approximately agreed with the clustered result of E. intermedia samples. Sixteen soil samples were grouped into three for E. intermedia samples in HCA to reveal the effect of soil factors on herb mineral nutrients. Then, MAV was performed in three groups of soil samples. The difference was not significant (P > 0.05) among three soil groups according to 21 soil factors for analysis. However, based on 10 characteristic factors from the soil PCA, the three groups had a significant difference (P < 0.05). Thus, multiple comparisons of LSD were carried out. The results indicated that concentrations of K, P, Zn, Fe and Mn had a significant difference in the three groups of soil samples. The five elements of K, P, Zn, Fe and Mn in the soil samples were related significantly to 11 elements in E. intermedia samples, namely, N, K, S, Mg, P, Na, Mn, B, Zn, Cu and Mo, which all increased with the increase of soil K, P, Zn, Fe and Mn concentration except for S in E. intermedia. Thus, these soil elements might be the important factors that influenced the element concentrations in E. intermedia.

Discrimination analysis of plant samples based on soil elements. The samples of E. intermedia
were classified into three based on the HCA, and their growing soil samples were divided into the same three groups artificially. In the three soil groups, Zn, K, P, Fe and Mn showed significant differences. Therefore, differences of mineral nutrients and classification of E. intermedia samples were probably related to differences of Zn, K, P, Fe and Mn in the soil because these soil elements could affect the elemental absorption in E. intermedia. Discriminant analysis (DA) was conducted according to concentrations of Zn, K, P, Fe and Mn in the soils to test this relationship. The results are shown in Fig. 5 and Table 5. Distribution patterns of E. intermedia samples are displayed in the plot (Fig. 5a). The samples of E. intermedia from 16 origins were grouped into three in different spaces with a precision rate of 93.8%, which was consistent with the HCA results. Figure 5b shows the correlation chart of loading for the five variables (Zn, K, P, Fe and Mn) from the soil samples. Discriminant function 1 revealed 88.2% of the variance, contributing to primary separation for E. intermedia samples and having a high positive correlation with soil K and Mn. Discriminant function 2 (representing 11.8% of the variance) was posi-   Total correct (%) 93.8