Introduction

Along with other ecological and socio-economic functions, forests and their soils play a significant role in global biogeochemical cycles. However, industrially emitted waste products, such as sulphur (S) compounds, metals and other potentially toxic chemical substances, can have wide-ranging negative impacts on forest vitality and ecosystem services (Helmisaari et al. 1995; Frey et al. 2006; Kandziora-Ciupa et al. 2016; Novotný et al. 2017; Vacek et al. 2021; Pecina et al. 2022). Regarding the environmental persistence of metals, a long-term perspective is crucial to determine their critical loads (Helmisaari et al. 1995), risks and behaviour in forest soils.

While forest floor contamination with metals will strongly reflect local sources and influences (Sucharová et al. 2011), differing environmental conditions can have a strong influence on their dynamics in mineral soils (Sucharová et al. 2011, 2012), with parent rock, soil processes and forest composition amongst the most important factors (Ranger and Nys 1994; Augusto et al. 1998, 2002; Andersen et al. 2004; Svendsen et al. 2007; Mihaljevič et al. 2010; Augusto et al. 2015; Bai et al. 2022). The vertical distribution of metals in forest soil will depend both on the particular metal and a range of soil properties (Hernandez et al. 2003), with pH as a key parameter affecting metal mobility in soil (Römkens and Salomons 1998; Kelly et al. 2003; Taiz and Zeiger 2010). Likewise, the soil texture (Tang et al. 2021) and the quantity and quality of soil organic matter (OM) will also have an important effect (Andersen et al. 2002, 2004; Svendsen et al. 2007; Liénard and Colinet 2016; Wieczorek et al. 2018). However, factors such as the carbon-to-nitrogen ratio (C:N) are rarely evaluated in relation to metal mobility in forest soils, despite being one of the most important OM quality indicators. Metal-specific interactions with OM components affect their behaviour in soil, with lead (Pb) and copper (Cu), for example, having a greater affinity to fractions with a high molecular mass, causing them to accumulate predominantly in humus-rich topsoil. Owing to its strong bonding to OM, therefore, Pb mobility is reduced (Probst et al. 2003; Ruan et al. 2008). In comparison, zinc (Zn) and cadmium (Cd) migrate together with soluble organic fractions of medium molecular weight, allowing them to penetrate more easily into the deeper soil layers (Hernandez et al. 2003; Li et al. 2009). In general, conventional metals can be arranged depending on their mobility in forest soils as follows: Zn > Cd > Cu > Pb (Borůvka et al. 2015).

Owing to their greater overall leaf area (Fang et al. 2019; Andronikov et al. 2021) and their evergreen nature, conifers intercept more atmospheric gaseous and dust pollution per year than deciduous broadleaves (de Schrijver et al. 2007). Several studies have addressed this issue by comparing Norway spruce (Picea abies (L.) H. Karst.) and European beech (Fagus sylvatica L.), these being two of the most important tree species in Europe (Bosela et al. 2021; Vacek et al. 2021). The substances captured more efficiently in spruce stands are then flushed into the soil by precipitation (Vannier et al. 1993; Augusto et al. 2002), resulting in increased throughfall deposition of contaminants (e.g. S and N compounds) compared with beech stands (Rothe et al. 2002; Berger et al. 2008; Andronikov et al. 2021). In addition to deposition-related contaminant levels, tree species can also influence other soil properties and processes, with leaching of N and S compounds generally being higher in spruce stands and pH and OM decomposition being lower (Rothe et al. 2002; Andersen et al. 2004; Berger and Berger 2012). While the vertical distribution of C has been shown to differ between beech and spruce forests (Ransedokken et al. 2019), the influence of beech and spruce forests on the content and vertical distribution of metals in forest soil in general has still not been sufficiently described. In most related studies, the conclusions are insufficiently supported statistically, they draw conclusions based on few samples or sites and they either fail to describe sufficiently or homogenise forest stand conditions.

The intention of this study, therefore, is to address these weaknesses by providing a reliable methodology for assessing the influence of spruce and beech on the vertical distribution of Cd, Cu, Pb and Zn in forest soils. Comparing these species, we hypothesise that (1) metal accumulation will be higher in the OM-rich organic F + H horizons and topsoil of beech stands despite the lower deposition potential and (2) metal enrichment will be increased in deeper soil layers of spruce stands due to spruce-related acidification and increased metal mobility. The study is part of a research project focused on the influence of beech and spruce on metals in forest soils. The aim of the project is to understand the effect of the species on the accumulation and redistribution of metals in forest environment regarding their toxicity-related long-term threat to humans and forest biota. This study complements the already published studies on content of Cd, Cu, Pb and Zn in surface horizons of forest soils (Novotný et al. 2021), pollution of forest soils and edible mushrooms with these metals (Pecina et al. 2022) and vertical distribution of Hg in forest soils and it´s transfer to edible mushrooms (Pecina et al. 2021).

Material and methods

Study area

This study was carried out in the Jeseníky Mountains in the Czech Republic (Fig. 1), a region marginally affected by long-term atmospheric deposition of metals from heavy industry (Novák et al. 2003; Zapletal et al. 2007; ČHMÚ 2022a, 2022b). The region has an average annual precipitation of 810 mm and a mean annual temperature of 7.8 °C (1989–2019; CHMI 2021a). The study sites themselves (see below) were located at altitudes of 587–990 m. They had mainly northern exposure (Novotný et al. 2021), similar climate conditions defined by the conditions of the region and comparable levels of atmospheric deposition (available data for Cd = 0.3–0.6 ng/m3 and Pb = 5.1–5.5 ng/m3; CHMI 2021b) without extreme deviations. Furthermore, based on a preliminary field survey, all study sites had a similar geology with the parent rock consisting of acidic metamorphic rocks (phyllites and gneiss). Forest stands in the area are dominated by spruce and beech. They were mostly established as man-made even-aged monocultures and due to long-term intensive commercial cultivation, spruce stands occur across the entire altitudinal gradient of the study.

Fig. 1
figure 1

A Study area in the Jeseníky Mountains, northern Czech Republic, with sampling sites indicated on the background layer derived from regional forest development plans (forest areas in green), based on a digital shading model of the Czech Republic; PL = Poland; B location of the Czech Republic within Europe; C location of the study area in the Czech Republic; D example of sampling design within a forest stand

Full size image

Sampling design

Six beech- and six spruce-dominated (representation > 80%), even-aged (80–100-years-old) stands of > 1 ha were selected for soil sampling. Soil samples were collected at seven plots distributed over a hexagonal design (Fig. 1D) in each stand. For each plot, we obtained composite soil samples comprising individual samples from three soil pits spaced 4–8 m from each other, each soil pit being situated under the projection of different tree. In each case, samples were taken from the F + H organic layer and from different depths in the mineral layers (i.e. 0–2, 2–10, 10–20, and 20–30 cm) and then placed into labelled polyethylene bags. Overall, a total of 420 composite soil samples were collected from 84 plots.

To prevent any influence from the surrounding stands, the samples were collected at a minimum distance of 25 m from the edge of the stand being investigated. Likewise, to prevent any distortion of the results through local contamination, sites situated near potential sources of pollution (e.g. roads, mines, built-up areas) and sites visibly affected by surface erosion were eliminated from sampling during a preliminary field survey.

Sample preparation and laboratory analysis

All samples were dried at room temperature and then processed into a fine fraction (< 2 mm; ČSN ISO 11464), with organic soil samples being further ground to dust in a ball mill for metal content analyses. Subsequently, each sample was analysed for soil properties characterising quantity (C, N and S) and quality (pH and C:N) of OM (Cools et al. 2014).

A suspension of both mineral and organic soil in water (1:5 ratio) was used to measure pH, while total C and N content was determined using a VarioMAX CNS elementary analyser (Elementar Analysensysteme GmbH, Germany), following standard method ČSN ISO 10694. In mineral soil samples, the texture of the main grain fractions was determined according to USDA standards by dry sieving and pipette analysis (Burt 2004).

For metal contents (Cd, Cu, Pb and Zn) analysis, the pre-prepared soil samples were digested in aqua regia (HNO3 and HCl at a ratio of 1:3) in an ETHOS EASY microwave digestion device (Milestone, Italy). After digestion, the samples were quantitatively transferred using Milli-Q water (Merck, Germany) to a volume of 25 ml. The concentration of elements in the digested sample was measured using a ContrAA 800D atomic absorption spectrophotometer (Analytik Jena, Germany) with a continuous source of radiation (Xe lamp), a high-resolution monochromator and an optical system based on a high-resolution Echelle monochromator and a sensitive charge coupled device (CCD) detector. Content of Cd, Cu and Pb was determined using the electrothermal atomiser (ET-AAS) method, while Zn content was determined using the dual flame system (FAAS) method. All measurements were controlled using blanks, triplicate measurements and matrix reference materials, i.e. METRANAL 31 (light sandy soil), METRANAL 33 (clay loam soil) and METRANAL 34 (loamy soil) (Analytika, Czech Republic). The detection limits for Cd, Cu, Pb and Zn were 0.003, 0.012, 0.033 and 0.007 mg/kg, respectively.

Statistical analysis

Linear mixed models were constructed using the ‘lme4’ package (Bates et al. 2015) in the R software package v.4.0.5 (R Core Team 2021). To enable direct comparison of model regression coefficients, soil data were centred and standardised to unit variance prior to analysis. Model assumptions were assessed by visual inspection of model residuals. The validity of the fixed effect terms was assessed based on the likelihood ratio (LR) test. The F-value statistics for fixed effects were used to select among competing models. All statistical differences were considered significant at p < 0.05. To assess the separate effect of tree species on metal contents, tree-species identity was introduced as the factor variable of fixed effects, and sample plot ID was used as the factor variable of random effects for each separate model of different soil layers. Random effect covered site-specific differences enabling the interaction of many factors, such as topography or atmospheric deposition. In contrast, for comparing metal contents between different soil layers, sample plot ID was introduced as a factor variable of random effects and the soil layer identity was used as the factor variable of fixed effects. The multiple comparisons between soil layers were further assessed with Tukey’s post-hoc test using the glht function in the R ‘multcomp’ package (Hothorn et al. 2008). To identify reliable predictors of metal contents across the study sites in specific soil layers, the site ID was used as the factor variable of random effects and the independent soil variables such as C, N, and S contents, the C:N ratio, pH, or the proportions of sand, silt, and clay in mineral soil were used as fixed effects in separate competing models. Finally, whenever the interaction between tree-species identity and the explanatory soil variable was significant based on the likelihood test, the interaction term was included into fixed effects of the final model.

Results

Vertical distribution of metals in the soil profile

A significant difference in overall metal content between beech and spruce stands was only confirmed for Zn, with levels being significantly higher in all soil layers in beech stands (Fig. 2). Significant differences in vertical distribution of metals between forest types were observed for Cd and Cu, with no significant differences with increasing profile depth in beech stands but significant accumulation in the F + H horizons compared to mineral soil in spruce stands. Cu showed a significant vertical decline down to 10 cm in spruce stands, after which levels did not change significantly. The vertical distribution of Pb between beech and spruce stands only differed in local maximum levels, with highest values being found in the F + H layer in spruce stands and in the 0–2 cm layer in beech stands. Subsequently, Pb content then decreased gradually to a depth of 20 cm in both forest types. Conversely, lowest Zn content was found in the upper soil profile layers and increased with depth (Fig. 2).

Fig. 2
figure 2

Vertical distribution of metals in the soil profile. Different uppercase letters indicate significant differences between forest types, while different lowercase letters indicate significant differences between sampling depths within a forest type; boxes represent quartiles, red crosses represent arithmetic means, points represent mean values at a site (differentiated by colour), with point size proportional to the in-site standard deviation (n = 7)

Full size image

Effect of soil properties on soil profile metal distribution

The vertical distribution of metals was influenced by the various properties found in the individual soil layers (Table 1), especially those linked with the quantity (C, N, S) and quality (pH, C:N) of OM. Quantity of OM had a significant effect on Cd (positive) and Zn (negative) content in the F + H layer, and on Pb (positive) in mineral soil (Table 1), while OM quality affected the content of all metals in the F + H layer, Cu at depths of 0–2 cm (negative effect of C:N) and 20–30 cm (positive effect of pH) in mineral soil, and Zn content throughout the profile (Fig. 2). The physical properties of mineral soil (sand, silt and clay content) were of limited significance for Pb and Zn content, with no effect on Cd or Cu.

Table 1 Regression coefficients of fixed effects in mixed models of metal content and soil properties at different soil depths
Full size table

The best regression models of metal contents in Table 2 revealed that the marginal R2 (attributed to the selected soil properties) were often relatively low compared to the conditional R2 (including random effects). This indicated that the distribution of metals was strongly site-specific (high contribution of random R2, Table 2). In the F + H layer for Pb and Zn contents and in the 20–30-cm depth for Cu contents, the fixed effects also included the interaction with tree species, as the regression model with explanatory soil variable was significant either for spruce or beech (Table 2). Overall, random effects explained 15.3–26.8%, 36.9–46.8%, 13.3–35% and 44.3–63.6% of the variation in Cd, Cu, Pb and Zn contents, respectively, across all assessed layers. For Cu and Zn, the influence of random effects was greater than that of fixed effects (soil variables) across the soil profile. For Pb, however, the relative importance of random effects decreased continuously with soil depth.

Table 2 The best models for metal content as a response variable, with selected independent soil variables (covariates) as fixed effects
Full size table

Discussion

Metal content depending on soil properties

Cadmium

Both the quantity and quality of OM (Table 1) were shown to affect Cd content in the F + H layer, with the positive correlation of OM quantity and Cd content related to the high affinity of Cd to OM (Andersen et al. 2002; Holm et al. 2003; Kabała and Singh 2006; Luo et al. 2019). However, we could find no published work linking Cd and C:N in forest soils, despite a thorough examination of the literature. In our study, the positive correlation between C:N and Cd indicated increasing accumulation of Cd as OM quality decreased, most likely in relation to potential fixation of soil N and deceleration of decomposition (Cools et al. 2014). In such cases, differing levels of metal biosorption are caused by variations in the composition of extracellular polymeric substances (EPS; Yuncu et al. 2006), which in turn is significantly affected by the C:N ratio (Durmaz and Sanin 2001). Furthermore, Yuncu et al. (2006) noted that the affinity of Cd to carboxylic groups in EPS increased as C:N increased, resulting in an increase in Cd biosorption.

The solubility of Cd, and its ability to migrate to lower soil layers, increases as pH decreases (McBride 1989), which is generally expressed as a positive correlation between Cd and pH. The unusual negative correlation between pH and Cd found in the F + H layer in this study (Table 1) may be related to its origin from acidic atmospheric depositions (see “Lead” section ).

Copper

Cu content was affected by factors related to the quality and quantity of OM (Table 1). Strong absorption of Cu by OM has been confirmed by several studies (e.g. Bruemmer et al. 1986; Kabata-Pendias and Mukherjee 2007). In the present study, Cu content in OM-rich layers (F + H and 0−2 cm) was explained and negatively correlated with the C:N ratio (Table 1, Fig. S1). Wang et al. (2019) showed that the negative correlation of Cu with C:N in forest soils was caused by immobilisation of Cu as the level of OM decomposition increased, while Yuncu et al. (2006) explained the increase in Cu biosorption as C:N declined as being due to the affinity of Cu to amine groups in EPS. This was also confirmed by Andersen et al. (2004), who stated that spruce stands, which have a significantly higher C:N ratio, and consequently a lower degree of OM decomposition (Cools et al. 2014), have a significantly greater ability to increase Cu solubility than beech stands.

The content of Cu at depths > 2 cm in mineral soil was also explained, though to a lesser degree, by the quantity of OM (R2 = 6−7% for S) and pH (R2 = 9%, Table 2). However, based on the low Cu content and low coefficient of determination for covariates, we consider the impact of these factors on Cu content in mineral soils as negligible.

Lead

Content of Pb in the F + H layer was best explained by the content of S and the pH value (Table 1). The significance of S and pH in the F + H layer contrasts with previous studies showing a determining relationship between Pb and OM (Egli et al. 1999; Kushwaha et al. 2018). Owing to the naturally low content of Pb and Cd in most mineral soils (Kabata-Pendias and Mukherjee 2007), which was also confirmed in this study, one can assume an almost exclusively anthropogenic origin via S-rich acidic atmospheric deposition (ČHMÚ 2022a, 2022b). This is also supported by i) their close relationship in the F + H layer (Fig. S2), and ii) the positive correlation of Pb and Cd to S in the F + H layer (Table 1) which, for Pb, prevailed throughout the profile. This relationship was not observed for Cd in mineral soils, probably because of a high number of samples below the detection limit. Thus, the significant negative relationship between Pb and pH in the F + H layer may directly indicate a more decisive effect of acidic deposition load than OM-related biogeochemical processes. In contrast, a more balanced influence of these factors may be expected for Cd. In mineral soil, the significant impact of OM quantity (R2 = 27−44% for C, R2 = 25% for N; Table 2) predominated, in compliance with previous studies that have shown heavy absorption of Pb by OM (Bruemmer et al. 1986; Kushwaha et al. 2018).

Zinc

Content of Zn in the F + H layer was best explained by both the quality and quantity of OM, and almost exclusively by the C:N ratio and pH value in mineral soil (Table 1). The negative correlation between Zn and C, N, S and C:N in the F + H layer, and the negative correlation between Zn and the key explanatory variable C:N in mineral soil (Table 2), despite the high affinity of metals to OM, indicates that Zn content is not primarily affected by its affinity to OM but by pH, with which Zn correlates positively (Table 2, Fig. S2). While unusual, the more significant effect of OM quantity and C:N to the detriment of pH (Hernandez et al. 2003) in mineral soil is probably an expression of the well-known negative correlation between pH and C:N (Cools et al. 2014), also evident from PCA (Fig. S2). Yuncu et al. (2006) also noted that Zn biosorption displayed no increasing or decreasing trend with C:N.

Based on the predominantly low pH observed in this study (Fig. S1), and the known dominant effect of pH on Zn mobility in the soil (Hernandez et al. 2003), we suggest that pH is the crucial factor influencing Zn distribution. In an acidic environment, which in this study was represented especially by the upper soil horizons, these being the main recipients of atmospheric deposition (Andronikov et al. 2021), Zn is converted into an exchangeable form that is highly soluble and can migrate to lower layers in the soil profile more quickly (Billett et al. 1991; Hernandez et al. 2003; McLaren et al. 2004; Chrastný et al. 2012), where its migration is slowed by increasing pH, causing its gradual accumulation (Figs. 2 and S1). This is also confirmed by Hernandez et al. (2003), who reported an increase in Zn content with increasing depth. Other contributors to this phenomenon may include i) accelerated transport of highly mobile Zn from the strongly acidic topsoil through washout by heavy rain (Svendsen et al. 2007), ii) effective competition between Pb and Cu with Zn, which has lower affinity to OM, leading to its displacement and mobilisation (Svendsen et al. 2007), and iii) lowered emission inputs over recent years. In the case of variant iii, there would be no accumulation of Zn in the upper OM-rich horizons, as recorded by Svendsen et al. (2007) at a site very close to acute source of pollution. However, Pecina et al. (2022) found high Zn and Cd contents in mushrooms in our study area and linked this with precipitation-related atmospheric deposition and the high mobility and bioavailability of these metals. Consequently, we suggest that the low atmospheric deposition of highly mobile Cd in our study area (0.2−0.3 ng.m−3/5 years; 2016−2020; ČHMÚ 2022b) is probably almost completely absorbed by mushrooms or plants, while the higher deposition of Zn (Andronikov et al. 2021) probably allows for both its absorption (Pecina et al. 2022) and mobility into the deeper soil layers (Fig. 2).

The influence of beech and spruce on vertical distribution of metals in the soil profile

Despite it being generally accepted that metal contents tend to be higher in coniferous stands (Andersen et al. 2004), we were only able to confirm a significant impact of forest composition on content of Zn with higher values in beech stands (Fig. 2). Therefore, hypothesis 1 was confirmed only for this metal. The impact of C content of beech stands in the F + H layer (Table 2) was unlikely to be the primary factor determining Zn distribution in the soil, rather it was an expression of the negative correlation with pH value, which was also evident from PCA (Fig. S2). The lower Zn content recorded in spruce stands (as opposed to beech stands), despite higher Zn throughfall in conifers (Andronikov et al. 2021), was most likely caused by i) higher Zn solubility and mobility resulting from a lower soil pH (Sect. 4.1, Fig. S1), and ii) absorption of mobilised Zn by living organisms.

The absence of any significant difference between Cd and Cu contents (Fig. 2) was probably affected by the scale of the assessed area, which enabled the interaction of many uncontrolled factors, such as topography, atmospheric deposition or microclimate conditions. In the case of Cd, Cu and Zn, the contribution of site identity was even higher than the effect of explanatory soil variables, as suggested by the difference between fixed and random effects (Table 2), while the role of site-specific differences apparently declined in deeper soil layers for Pb. Varying natural conditions may have had a considerable influence on the process of OM accumulation and decomposition, i.e. the quality and quantity of OM (Coûteaux et al. 1995; Prescott and Grayston 2013), which has a direct impact on metal content in OM-rich horizons (Andersen et al. 2004).

The significant impact of tree species on metal content was also evident in their differing vertical distribution, particularly the contrasting accumulation of Cd, Cu and Pb in the F + H layer in spruce stands (Fig. 2). This differentiating effect of spruce on the accumulation of metals, which has been confirmed by other studies (Andersen et al. 2002, 2004; Svendsen et al. 2007), results especially from its ability to capture atmospheric pollution more effectively on an assimilation apparatus (Rothe et al. 2002). This was also indicated by the significantly higher content of S in the F + H layer in spruce stands (Fig. S1). Zapletal et al. (2007) were able to show that spruce stands enable wash out and foliar leaching of SO4 from atmospheric deposition of industrial pollutants, the primary source of S and metal deposition in the Jeseníky Mountains (Novák et al. 2003; Zapletal 2012). The exclusively spruce-related significant effect of S as the main explanatory variable of Pb content in the F + H layer (Table 2) confirms the supposition of a dominant anthropogenic origin, as well as the importance of the interaction between spruce and atmospheric deposition. Spruce stands also influence soil metal content through slow decomposition of litter, which then accumulates (Berger and Berger 2012), leading to a significantly higher C content in the F + H layer and reduced variability in the C:N ratio (generally higher than in beech stands; Fig. S1), a factor affecting the biosorption of Cd, Cu and Pb (Yuncu et al. 2006). However, this phenomenon seems to be significant only for the distribution of metals in the soils of spruce stands; it is not decisive for their total content in the studied scale of soil profile compared to beech stands as already partially mentioned and hypothesised (hypothesis 1). Accordingly, the second hypothesis is rejected. If increased metal enrichment in deeper soil layers of spruce stands due to spruce-related acidification and related metal mobility occurs, then at depths greater than 30 cm.

Increased accumulation of Pb in OM-enriched mineral soil horizons (0−2 cm) provided a clear example of the effect of beech stands on the mobility of metals (Fig. 2). This can be explained by either i) decomposition of metal-absorbing OM and its reabsorption as a solid phase, ii) the varying intensity of OM decomposition in OM-enriched horizons, or iii) washout by heavy rain (Svendsen et al. 2007). Similar conclusions were reached by Novotný et al. (2021), who explained the higher content of Pb in OM-enriched mineral soil horizons under beech stands as Pb being released from F + H layers as a result of more rapid OM mineralisation.

Conclusion

In this regional-scale evaluation of metals distribution in soil, we showed that Cd and Cu content in OM-rich horizons is directly linked to the intensity of OM decomposition (represented by C:N), regardless of forest type. Decomposition of OM is mainly driven by heterogenous environmental conditions, which were partly controlled in our models by the random effects. As the level of OM decomposition increased, Cd content decreased and Cu content increased. Content of Pb, on the other hand, was influenced by S, pH and random effects, regardless of forest type, indicating that atmospheric deposition was a primary factor affecting Pb content in soil, rather than OM-related biogeochemical processes. Zn was the only metal whose content was significantly affected by forest type, with Zn mobility being reduced and affinity to OM increased as a result of the higher pH under the beech stands, leading to accumulation of Zn downward in the soil profile. Despite only having unambiguous evidence for Zn, stands dominated by beech or spruce clearly had a differing influence on the distribution of metals in all cases.