Physicochemical characterisation of the soils

The test sites differed significantly in terms of selected physicochemical properties of soils. They were divided based on soil texture, i.e. silty (RD, RZ), and sandy soils (GM, MI). The above division was reflected in a significantly lower content of soil fractions < 2 mm in sandy soils (0.8, 0.6 respectively). The second parameter differentiating the site was the pH of the soil. Based on the obtained pH measurement results in H2O, it was found that soils at MI and GM sites (sandy soils) differ from silty soils in their lower active acidity (4.72 and 4.95 respectively). A similar regularity can be observed by analysing the results of pH measurement in KCl, where the exchange acidity for sandy soils reached lower average values: 3.94 for MI and 4.19 for GM. Taking into account the division of sites based on soil texture, no differences were found in terms of carbon content, while the RD site was characterised by the highest C content (3.85%) compared to the others (Figs. 2A, 3, 4, 5, 6 and 7A; Additional file 1).

The concentrations of certain heavy metals in the soil, their transport to and in the plant in the context of the physical and chemical soil parametersZinc

The zinc content in soils was one of the largest among the tested metals and ranged from 35.59 to 68.76 mg/kg. The differences between the sites were statistically significant, with lower values for MI and GM sites with sandy soils, where the zinc concentration was almost half the background value of 67 mg/kg [38, 39]. In RZ, the zinc content in the soil was very close to the value considered typical (PI Zn = 1.03) (Additional file 2). A positive relationship was shown between the zinc content of the soil and its acidity (pH H2O r = 0.664; pH KCl r = 0.796) and soil fraction (r = 0.584), while soil organic carbon was not an important parameter (Additional file 3).

In plant material, the content of this metal differed significantly statistically, and in this respect the sap was found to have the lowest concentration (Fig. 2A, B; Additional file 2). The zinc content in pollen differed significantly statistically from the content in the leaves, in which a record concentration was found, 50 times higher than in the sap. Regardless of the soil texture, the lowest contents were found in the sap, followed successively by pollen, inflorescences, and leaves (Fig. 2B; Additional file 2).

Direct ordination analysis explains the relationship between soil parameters and zinc content in plant material with more than 40% efficiency, while pointing to clear differences between silty and sandy sites. With the exception of %C, the physical and chemical properties of the soils strongly influenced the zinc content of the birch. The influence of these parameters was negative, as evidenced by long vectors with opposite returns (Fig. 2A) and correlation analysis with significant correlation coefficients for the aerial parts of birch (Additional file 3).

The transfer of zinc from the soil to the plant (MR) depended on the physicochemical characteristics of the soil. A statistically significant negative dependence on soil pH (from r = -0.783 for L to r = -0.883 for C2) and fraction (from r = -0.671 for Sa to r = -0.822 C1) (Additional file 3) was found. Birch sap was characterised by the lowest MR values (from 0.05 to 0.10); the transfer of this element to the leaves was the most intense (from 2.51 to 6.54). Regardless of the type of plant material, sandy sites (GM and MI) were characterised by the best bioavailability of this metal. RDA analysis explains very well the relationship between soil parameters and MR values for plant material. The silty and sandy sites are very well separated by the second ordinate axis (PEV = 76.0%, p = 0.002) (Fig. 2A; Additional file 2).

For sTF, the PEV value is very low and the probability value for the first and all ordination axes is above > 0.05 (Fig. 2A). The value of this parameter for inflorescences increases most strongly at low pH (Additional file 3). Statistical analyses have shown that the values of this indicator differ for a particular type of plant material. The lowest values were found for birch pollen and the highest for leaves without division and by soil species (Fig. 2B; Additional file 2).

Copper

The Cu content in the soil was the lowest at the MI site, reaching an average of 3.52 mg/kg, while at other sites the values were in the range of 10.34 – 17.82 mg/kg, with the highest for RZ (Additional file 2). The copper content in the soil was also lower than the value considered typical, i.e. 39 mg/kg [38, 39], especially at sandy sites (average PI value Cu = 0.18) (Fig. 3B). It has been shown that the content of this element in the soil significantly depends on its fraction (r = 0.625), and is lower on acidic soils (pH H2O r = 0.644; pH KCl r = 0.618). Silty sites are statistically significantly higher in Cu content than sandy sites (Fig. 3A, B; Additional file 3).

In the plant, the copper content differed significantly, with the lowest concentrations in the sap, i.e. from 0.13 in RD to 0.20 mg/kg in GM. The Cu content in inflorescences and pollen (7.90 – 10.66 mg/kg) was characterised by a tendency to accumulate the element at the GM and MI sites, with sandy soils (Fig. 3B; Additional file 2). Negative correlations with pH were found for the pre-pollination inflorescence (r = -0.527) and for pollen (r = -0.551). No similar relationships were found for leaves, and the Cu content ranged from 6.16 mg/kg in MI to 7.18 mg/kg in RZ (Additional file 2, 3). The RDA triplot indicates that the copper content in generative organs, especially in pollen and inflorescences after pollination, is lower on soils with a higher pH and a higher content of this element in the soil (Fig. 3A).

The accumulation of copper in the plant (MR) depended on the physical and chemical characteristics of the soil. A statistically significant relationship was found with pH (pH H2O from r = -0.632 for Sa to r = -0.731 for L) and fraction < 2 mm (from r = -0.476 for C1 to r = -0.686 L) (Additional file 3). The MR index was characterised by the lowest values in the sap, within the range of 0.01 – 0.05. A clearly higher accumulation was found in the leaves, i.e. MR from 0.41 to 2.00. In the generative parts of the plant, in the inflorescences and pollen, bioaccumulation was greatest, with less mobility at the RZ site and the largest in MI (Additional file 2). Regardless of the plant material, metal accumulation was statistically significantly higher at sandy sites (Fig. 3B). Direct ordination analysis explains the relationship between soil parameters and metal transport to the plant (MR index) with more than 84% efficiency indicating clear negative relationships (Fig. 3A).

Much higher values were characterised by the sTF ratio in relation to MR. Among the plant material studied, the leaves had the lowest values of this indicator (from 42.08 in GM to 60.16 in MI) and the highest in inflorescences in the range from 54.01 for P to 101.66 for C1 (Additional file 2). Soil textural group was not a parameter differentiating the transport of copper in the plant (Fig. 3A, B), nor was it shown that the physicochemical properties of the soil affect the volume of this transport (Additional file 3).

Lead

In none of the tested soil samples did the concentration of lead in the soil exceed 25 mg/kg [45], which have allowed us to conclude that the soils were not contaminated with lead. What is more, they were close to the typical ones, as indicated by the PI index values from 0.74 in GM to 1.46 in RD. The lead content in the soil was highly variable (SD standard deviation values) and ranged from 13.37 mg/kg (MI) to 24.96 mg/kg (RD). The existing differences were not statistically significant (Additional file 2). Soil texture was also not a statistically significant parameter differentiating these sites, although as indicated by multidimensional analyses at many sandy sites, the lead content in the soil is relatively low (Fig. 4A, B). A dependence of the lead content in the soil on its physicochemical parameters was also not demonstrated (Additional file 3).

With the exception of sap, the lead content of plant material and its bioavailability was strongly negatively correlated with soil pH and clay fraction (Additional file 3). The significant impact of these parameters is confirmed by RDA analyses (p < 0.05). For the average metal content PEV = 54.1% and for MR PEV = 75.6%. In the case of the sTF indicator, correlations with individual physicochemical parameters are much weaker or statistically insignificant, as is the result of the RDA multivariate analysis (PEV = 26.7%; p > 0.05) (Fig. 4A; Additional file 3).

The lowest lead content in the tested plant material was shown by sap, on average 0.03 mg/kg. In the aerial parts of the plant, lead content ranged from 0.17 mg/kg in pollen collected at silty sites to 1.33 mg/kg in inflorescences collected after pollination at sandy sites (Additional file 2). These values differed significantly statistically if the parameter grouping the site was soil texture. It can then be concluded that at sandy sites the lead content was always higher, especially in the case of generative parts (Fig. 4A, B). It is worth noting the fact that the lead content in the material collected in RZ (silty soil) is significantly lower than that collected in GM (sandy soil) (Additional file 2).

MR values indicate poor lead availability to the plant, from 0.001 for Sa in RD to 0.11 for C1 in MI (Additional file 2). When comparing the soil species, with the exception of sap, the availability of lead to the plant is higher at sandy sites, and is the same for generative and vegetative parts (Fig. 4A, B). At these sites, the variability of the obtained results was high (relatively high SD; Additional file 2). At the silty sites, significant differences concerned pollen and inflorescences after pollination (Fig. 4B). The accumulation of copper in the plant (MR) depended on the physical and chemical characteristics of the soil. For aerial parts of plant, a statistically significant relationship with pH (e.g. for pH H2O from r = -0.634 for C1 to r = -0.835 for P) and fraction < 2 mm (from r = -0.585 for C1 to r = -0.828 for P) (Additional file 3) was found. Very large fluctuations (SD) were found for the sTF index, from 2.47 for P in RZ to 185.60 for C2 for GM (Additional file 2). The differences between the silty/sandy sites were not statistically significant (Fig. 4B), but in many cases at the RZ position the values of this indicator were the lowest and in MI the highest (Additional file 2).

Cadmium

Among the elements studied, Cd was characterised by the lowest content in the soil, which ranged from 0.05 mg/kg in GM to 0.63 mg/kg in RZ, with a statistically significant trend of higher concentrations at silty sites (Additional file 2). These values depended on the chemical properties (pH H2O r = 0.738; pH KCl r = 0.751) and fraction of soil (r = 0.781) (Fig. 5A; Additional file 3). Based on the PI index, it was found that the cadmium content was on average 2.5 times higher than the background (0.1 mg/kg) [38, 39], and after taking into account soil texture, for silty soil this value was more than 4 times higher (Fig. 5B; Additional file 2).

The cadmium content of the plant depended on the chemical and physical properties of the soil. RDA analysis clarified these relationships with PEV = 64.6%. On acidic soils with a lower content of < 2 mm fraction, the content of this metal in the plant was higher, and the strongest, negative relationships occurred between pH H2O and the cadmium content in pollen (r = -0.838). The least cadmium was found in birch sap (from 0.003 mg/kg in RZ to 0.014 mg/kg in GM) and the differentiating factor was primarily soil texture. In the sap taken from the MI and GM sites (sandy sites), the Cd content was significantly higher. A similar relationship was found for generative organs, including pollen, where the cadmium content was highest (from 0.68 mg/kg in RZ to 6.53 mg/kg in MI). The leaves had a lower cadmium content (from 1.01 in MI to 1.47 mg/kg in GM) and soil species was not a differentiating factor in sites (Fig. 5A, B; Additional file 2, 3).

RDA triplot analysis for cadmium transport from soil to plant indicates that its parameters differentiate study sites very well in terms of soil species (PEV = 89.0%) and high MR values characterise sandy sites. Low pH values significantly favoured transport from the soil to the sap and aerial parts of birch, as evidenced by the values of correlation coefficients (below r = -0.7). The content of the < 2 mm fraction had an even stronger effect, especially for pollen (Fig. 5A; Additional file 3). The greatest mobility of cadmium from the soil was found for sites occurring on sandy soils with record bioaccumulation in pollen collected in MI (MR = 104.22). The accumulation of this element in the sap was the smallest, minimal in the samples collected in RZ (MR = 0.004) (Additional file 2).

Soil textural group was not a differentiating factor in the transport of this metal from sap to generative organs (pollen, inflorescences) and the PEV value for the RDA analysis was only 27.1% and was not statistically significant. Statistical analyses (U Mann–Whitney test) indicate better cadmium transfer to leaves at silty sites (Fig. 5A, B). In terms of this indicator, the 4 sites differed statistically significantly only for C1 (Additional file 2). Considering each soil parameter separately, it can be concluded that the pH and soil fraction had a statistically significant effect only on the transfer of cadmium from the sap to the leaves (Additional file 3).

Chromium

The average chromium content in the soil ranged from 6.85 mg/kg in MI to 38.39 mg/kg in RZ. These values were even significantly lower (more than 3 times) than the reference value (i.e. 69 mg/kg) [38, 39], and after taking into account the soil textural group, for sandy soil this value was almost 8 times lower than the background (Fig. 6B; Additional file 2). Soil species was a factor determining this parameter; at sites with sandy soils, the Cr content was significantly lower (Fig. 6B). With the exception of the organic carbon content of the soil, other physicochemical parameters significantly influenced the chromium content in it (Fig. 6A). The correlations were positive, the strongest for the soil species (r = 0.867) (Additional file 3).

In the plant material, the chromium content was many times lower than in the soil. Particularly low content was found in the sap (from 0.0038 mg/kg in GM to 0.0046 mg/kg in MI). Due to their large diversity (high SD), the differences between the 4 sites were not statistically significant (Additional file 2). In the aerial parts of birch, the content of Cr was similar; with the exception of pollen, differences were not noted (Fig. 6B). With the exception of pollen, the Cr content did not depend on the chemical and physical properties of the soil and soil texture, which is confirmed by statistically insignificant correlation coefficients (Additional file 3), RDA results, i.e. the arrangement and length of vectors on the triplot, PEV (26.4, p > 0.05) (Fig. 6A) and the U-Mann–Whitney test (Fig. 6B).

The MR indicator indicated the lowest accumulation of chromium in the sap, where it reached values from 0.0001 for RZ to 0.0007 for MI (Additional file 2). Higher accumulation was found in leaves and generative organs, and soil species influenced this parameter. At sandy sites (GM and MI), these values were statistically significantly higher than at silty sites (Fig. 6B). Chromium accumulation in the plant was higher in acidic soils and the correlation coefficients for each type of plant material were statistically significant and varied for pH H2O from r = -0.604 for C1 to r = -0.695 for P. A stronger, negative correlation was found for soil fraction with the strongest correlation for Sa (r = -0.827) (Additional file 3). RDA analysis confirmed these relationships with PEV = 77.1% and p < 0.05 (Fig. 6A).

At the studied sites, the transport of chromium in the plant (sTF) differed and concerned generative parts (Additional file 2). The average values for plant material without and with grouping by soil species did not differ statistically significantly. Confirmation of the lack of dependence is the RDA result, namely the length and position of vectors on the triplot relative to the ordinate axes, low PEV value (18.8%) and p > 0.05 (Fig. 6A). A negative effect of soil pH (pH H2O) on this indicator was found for P and C2 (r values of -0.658, -0.451 respectively) and clay fraction of soil only for P (r = -0.574) (Additional file 3).

Nickel

The Ni content in the soil ranged from 5.43 to 22.31 mg/kg and assumed significantly lower values at sandy sites (GM and MI). At each studied site, the nickel content in the soil was significantly lower than the geochemical background, as evidenced by low values of PI indicators; the average indicator was 0.22 (Additional file 2). As in the case of other HMs, the Ni content in the soil was positively correlated with the pH of the soil (pH H2O r = 0.758; pH KCl r = 0.653), it also depended positively on the content of the clay fraction (r = 0.817) (Additional file 3).

In the plant, the lowest nickel content was found in the sap (from 1.11 mg/kg to 1.42 mg/kg) and the highest in inflorescences (from 9.6 mg/kg to 16.03 mg/kg) (Additional file 2). As shown by statistical analyses, soil texture was not a factor differentiating sites, which is also confirmed by the RDA chart. Unlike other metals, sandy and silty sites do not cluster relative to the ordinate axes. The position of the vectors and their return indicate a negative relationship between the organic carbon content of the soil and the content of this metal in C1 (Fig. 7A). This is significantly confirmed by the statistical correlation coefficient (r = -0.486) (Additional file 3). The remaining physicochemical parameters were less important, the RDA analysis explains only 21.0% of the variability and the result is statistically insignificant (Fig. 7A).

Based on the results of the Mann–Whitney U test, it can be concluded that on sandy soils, the availability of nickel from the soil to the plant (MR) was statistically significantly higher. On the RDA triplot, these sites are clearly separated by the second axis of the ordinance. The result of this analysis is statistically significant (p = 0.002) with a PEV of 84.6%. The MR index reached the lowest values for the RD and RZ sites, regardless of the plant material being tested, and the highest for GM and MI. It is worth noting the highest MR values for inflorescences (average value for C1 – 1.29 and for C2 – 1.51). With the exception of %C, the physicochemical properties negatively affect the availability of nickel to the plant, which is confirmed by negative, statistically significant correlation coefficients with the strongest relationship between the clay fraction and the mobility of nickel to sap (r = -0.830) (Additional file 3).

The mobility of Ni in the plant (sTF) expressed by the ratio of the content of the tested metal in the aerial parts to the sap was characterised by values from 3.12 to 11.57. The GM site was characterised by the highest values regardless of the plant material studied, assuming the lowest values for P (3.59) and the highest for inflorescences (9.17 for C1, 10.75 for C2). The ordinate axes did not discriminate against the sites according to soil texture (PEV = 25.4%, p = 0.262) (Fig. 7A); the lack of significance of this grouping variable is confirmed by statistical analyses (Fig. 7B).

The relationships between the content of heavy metals in the soil and the birch and their transport to and in the birch

To sum up, it should be stated that the relationships between the content of heavy metals in the soil and in different parts of birch are diverse. In the soil, the average zinc content was clearly the highest and the concentrations of the next highest elements, lead and chromium, which did not differ from each other statistically significantly, were more than twice as low. In soils, the cadmium content was very low at all sites, below 0.5 mg/kg. The sites differed in terms of the total content of heavy metals in the soil (PI total) with a clear tendency of their greater accumulation in silty soils. In plant material, zinc also reached the highest concentrations, but unlike soil, lead and chromium contents were generally the lowest (below 1 mg/kg). It should be emphasised that although the content of nickel and copper in the soil was low, in the case of plants these elements belonged to the group with the relatively highest content (Figs. 2B, 3, 4, 5, 6 and 7B; Fig. 8A; Additional file 2).

For each element, its content as well as the transfer from soil to plant (MR) were the smallest in the sap. The largest transfer from the soil to different parts of plants concerned cadmium (mainly catkins and pollen) and the lowest was lead and chromium, where MR did not exceed 0.1 (Figs. 2B, 3, 4, 5, 6 and 7B; Fig. 9). The transfer of elements in the plant depended on the metal, the highest was in the case of cadmium (from 233.53 for C2 to 344.8 for P) and the lowest for nickel (from 3.59 P to 10.75 C2). The sTF values for the other elements were at a similar level (from 27.58 Pb P to 74.93 Cu C1). In generative organs, the second sTF-valued element was Cu. The content of heavy elements in the soil as well as their transfer to different parts of the plant (MR) generally depended on the soil textural group in contrast to the sTF indicator indicating the transfer of elements in the plant itself (Figs. 2B, 3, 4, 5, 6 and 7B, Fig. 8B, C, Additional file 2).

Antioxidants

Antioxidant properties were measured by FRAP and DPPH methods and the analysis of the results indicated the highest antioxidant potential for catkins before and after pollination, while the birch sap showed the lowest antioxidant properties (Additional file 4). The analysis of the content of secondary metabolites also confirms such a diversity in the antioxidant potential of plant material. Phenolic compounds, measured using the Folin method, were the highest in catkins and also in leaves, and the lowest in birch sap. Quite low values were also observed for pollen. Leaves contained the highest content of flavonoids, while in birch sap the flavonoids content was very low (Additional file 4).

With the exception of two examples (C2 and the sap collected at GM), we did not observe statistically significant differences in antioxidant properties between materials sampled from different sites (Additional file 4). In several cases, the soil textural type was the factor determining the antioxidant potential measured using the DPPH, FRAP and Folin methods. According to the U Mann–Whitney, the antioxidant activity of leaves collected at sandy sites (GM and MI) was significantly higher than at silty sites (p = 0.014, p = 0.017, p = 0.031, respectively). Opposite relationships were found for C2 for the FRAP method (p = 0.037) (Additional file 4).

Statistically significant differences in the antioxidant properties of individual plant organs were performed by Kruskal–Wallis test (p < 0.000) (Table 1). The highest amounts of phenols were found in inflorescences and leaves, while sap and pollen had a significantly lower content. In terms of the content of flavonoids, three groups were distinguished. The leaves contained the highest concentrations, followed by inflorescences and pollen, and the least amount of flavonoids was found in the sap. The plant organs also differed in their antioxidant ability measured using the DPPH and FRAP methods. Regardless of the method used, two groups can be distinguished. The inflorescences and leaves were characterised by the similar highest properties, and the sap and pollen were significantly lower (Fig. 9).

Table 1 The homogeneous groups according to Dunn's post-hoc test

The antioxidant properties are influenced by the content of metals in the soil and in the plant itself. Based on the analysis of the results of multiple regression (Table 2), it can be concluded that the metals significantly affecting the content of polyphenols and antioxidant properties are primarily copper, followed by lead, cadmium, and chromium, and occasionally nickel and zinc. An increase in the concentration of lead in the soil positively affects these properties. Zinc (negative relationship) and cadmium (positive dependence) (DPPH) content are also important variables. The cadmium content in the soil is also an important factor in enhancing the antioxidant properties of pollen (FRAP method) and an additional important explanatory variable is the content of nickel (Folin method) or copper (DPPH method). The antioxidant potential also increases with low copper content in pollen (FRAP and DPPH methods). Copper is also an important factor influencing the antioxidant properties of other parts of birch. A positive correlation was shown between the concentration of copper in the pre-pollination inflorescences and antioxidants (measured by FRAP) and the content of polyphenols (Folin method) in these organs, as well as a negative relationship with the lead content in the soil. The ability to scavenge free radicals (DPPH) in inflorescences C1 increases with low copper content in the soil and higher chromium content in the soil. These two variables describe these relationships to the highest degree (R2 = 66.5). Similarly, a negative correlation was observed between the concentration of flavonoids in C2 and the concentration of Cu, but the coefficient of determination was at a low level, not exceeding 20%. In leaves, a decrease in Cu concentration and an increase in lead concentration significantly affects the antioxidant properties, measured by the FRAP method, and in the case of polyphenol content (measured by Folin method), the explanatory variable is only the copper content in the leaves. These relationships are explained at 30.4% and 35.9%, respectively (Table 2).

Table 2 The results of regression analysis showing the relationships between the concentrations of minerals in plants and soil and antioxidant activities. The results were regression formulas with multiple correlation coefficients and multiple determination coefficients (R2). The level of significance was α ≤ 0.05