Effects of basalt, concrete fines, and steel slag on maize growth and toxic trace element accumulation in an enhanced weathering experiment

Terrestrial enhanced silicate weathering is a CO₂ removal technology involving the application of ground silicate materials to agricultural soils. Next to its CO₂ removal potential, it can improve soil fertility and crop growth, but silicate materials can also contain toxic trace elements. In a mesocosm experiment, we investigated the effect of basalt, concrete fines, and steel slag on biomass, nutrients, and the toxic trace element concentration of Zea mays, using a dose–response approach.

Plant biomass increased with basalt but not with concrete fines or steel slag. Generally, plant Ca, Mg, and corn Si concentrations increased with an increasing silicate application amount as a result of increased availability to plants. In contrast, plant N, P, and K concentrations were hardly affected by silicate application. Besides increased leaf Pb concentrations with steel slag application, which did not exceed the maximum limit set by the World Health Organization (WHO) and Food and Agriculture Organization (FAO) (0.05 mg Pb kg⁻¹ ww, where “ww” denotes wet weight), the toxic trace element concentrations in aerial plant tissues mostly decreased with increasing silicate application amount, presumably because of an increased soil pH and accumulation in plant roots. Thus, our study indicates mixed effects of silicate application on maize while also suggesting that the risk of toxic trace element contamination after a one-time application of the tested silicates is limited.

1 Introduction

Recently, there has been a growing interest in the use of silicate rock powder in agriculture for carbon dioxide (CO₂) removal (Edwards et al., 2017; Swoboda et al., 2022). When silicates react with water and CO₂, (bi)carbonates are formed that can be transported to the ocean via leaching into the groundwater, possibly storing carbon (C) for centuries or longer (Moosdorf et al., 2014). While the naturally occurring silicate rock weathering process has been important for stabilising climate at geological timescales, its pace is insufficient to substantially reduce the current rise in atmospheric CO₂ (Berner, 2004; Walker et al., 1981). Enhanced silicate weathering (EW) aims to accelerate this natural process through the mechanical grinding of the rocks into a fine powder.

In addition to its atmospheric CO₂ removal potential, applying silicate minerals to soils holds promise for improving agricultural practices. When silicate minerals weather, protons are consumed and weathering products, such as calcium (Ca²⁺) and magnesium (Mg²⁺), are released (Kelland et al., 2020; Ramos et al., 2022). This can improve soil chemical properties, such as increasing the soil pH and cation exchange capacity (CEC), and soil water retention (Anda et al., 2015; Calabrese et al., 2022; Taylor et al., 2017). Soil acidification and nutrient leaching are pervasive issues in agriculture, and EW can (in this way) contribute to soil health and improve crop growth (Tilman et al., 2002). Additionally, even though not considered an essential plant nutrient, the process of EW releases silicon (Si), which can improve plant resistance to pests and diseases, thereby improving crop health and productivity in general (Calabrese et al., 2022; Swoboda et al., 2022). Because of these benefits, silicate rock powder has been used as a fertiliser for many years (Van Straaten, 2006), particularly in tropical regions, where the release of base cations from these rocks can significantly enhance crop productivity (e.g. Swoboda et al., 2022). Nonetheless, EW also holds certain risks that need to be considered. Silicate materials typically contain toxic trace elements that are released into the environment during weathering, posing the risk of uptake by plants. The amount of toxic trace elements varies strongly among rocks and industrial silicates (Dupla et al., 2023; O'Connor et al., 2021). For instance, basalt, a naturally occurring and globally abundant silicate rock, generally exhibits lower concentrations of potentially harmful metals such as nickel (Ni) and chromium (Cr) compared to other silicate rock types and is, therefore, a preferred rock source to consider in agriculture (Kelland et al., 2020). Basalt is a mafic igneous rock that contains substantial amounts of Ca- and Mg-rich silicate minerals (Lewis et al., 2021). In the review of Swoboda et al. (2022), all trials with (ultra)mafic rocks on agricultural soils improved yields. Recent studies further corroborate these findings; Luchese et al. (2023), for instance, documented improved corn biomass upon basalt application in both clay and sandy clay loam soils. Similarly, basalt addition led to notable increases in dry mass, height, and stem diameter as well as the accumulation of macronutrients (nitrogen, N; phosphorus, P; potassium, K; sulfur, S; Ca; and Mg) in corn and beans grown on tropical soils (Conceição et al., 2022).

Most EW experiments have been conducted in tropical regions on highly weathered, acidic soils, while studies in a temperate climate are still scarce (Swoboda et al., 2022). However, Skov et al. (2024) demonstrated an increase in spring oat yield after application of basalt on direct-drill and ploughed plots in a temperate climate. These increases were assigned to a modest increase in pH that resulted in reduced manganese (Mn) and iron (Fe) uptake. This study also reported higher tissue Ca content and increased grain and tissue K upon basalt application. Contrastingly, the addition of volcanic rock dust to soils did not influence wheat growth in a mesocosm experiment in Sweden (Ramezanian et al., 2013). Furthermore, the aboveground biomass of potato plants tended to increase with basalt application in a mesocosm experiment in Belgium, growing on an alkaline soil (Vienne et al., 2022). Nonetheless, it is important to keep in mind that the apparent dominance of positive EW effects on crop growth might be affected by publication bias, as negative or non-significant results are less likely to be published compared to positive and expected outcomes (Dieleman and Janssens, 2011).

While the use of geological silicates requires energy for the mining and grinding of rocks, thereby partially offsetting the CO₂ removal potential (Beerling et al., 2020; Goll et al., 2021), this energy cost may be strongly reduced by using silicate materials derived from industrial processes. For instance, Ca-rich silicates generated as by-products from steel manufacturing (steel slag) or waste concrete fines represent alternatives that circumvent the need for additional mining (Renforth et al., 2011). These industrial by-products contain minerals that dissolve faster than basalt and are already produced as fine particles, thereby diminishing the energy cost of grinding (Beerling et al., 2020).

Concrete waste has previously been applied to soils to improve plant growth; however, the way in which it affects plants is currently poorly understood (Ho et al., 2021). Concrete by-products are produced in large quantities because concrete is a popular product throughout the construction industry (dos Reis et al., 2020). Concrete fines also contain silicate minerals and other cations, such as Fe, Ca, and Mg (Table 2), but almost 18 % of the concrete fines used in this study comprise calcite (CaCO₃) (Table S1). Calcite dissolution does not lead to the net uptake of CO₂, because all CO₂ that is consumed during dissolution is returned back to the atmosphere by the precipitation of carbonates in the ocean (Liu et al., 2011). Therefore, weathering of concrete fines will probably be less efficient for carbon capture. In contrast, the application of steel slag as a fertiliser in agriculture is already a common practice (Gao et al., 2020). Various trials have demonstrated increases in biomass upon steel slag application, for instance for soybean (Castro and Crusciol, 2013; Deus et al., 2020), maize (Castro and Crusciol, 2013; Wang and Qing-Sheng, 2006), and rice (Makabe-Sasaki and Sasaki, 2015; Wang et al., 2015). At three different locations in Germany, long-term field trials with steel slag application have been running since 1954, 1990, and 1993, respectively. At all locations, the yield of grass or arable crops increased significantly with the application of different kinds of steel slag, despite increases in the Cr and vanadium (V) contents in the soil (Branca et al., 2014). Nonetheless, as with other silicate materials, research about the influence of steel slag on crops in temperate climates is lacking, and the fate of potential toxic elements is unknown.

Despite the evident benefits, there remain substantial gaps in our understanding of the broader implications of EW in agriculture. Specifically, its influence on nutrient cycling and the potential for contamination with toxic trace elements, particularly with industrial by-products such as concrete fines and steel slag, warrants further investigation. Most research so far has been conducted in a tropical climate (Swoboda et al., 2022), often on highly weathered and acidic soils, but EW is also currently being considered for application in other climate regions and, thus, a better understanding is needed.

This study aims to quantify the influence of basalt, concrete fines, and steel slag on Zea mays growth, yield, and nutrient and toxic trace element concentrations in a mesocosm experiment in a temperate climate. The three types of silicates were applied using a dose–response approach ranging from 0 to 200 t ha⁻¹ of basalt, 0 to 31 t ha⁻¹ of concrete fines, and 0 to 10 t ha⁻¹ of steel slag. We hypothesised that (I) the availability of weathering products would increase with increasing silicate application amount; (II) plant biomass and plant concentrations of nutrients present in the silicates would increase with increasing silicate application; and (III) toxic trace element concentrations would increase with increasing silicate application, especially with basalt and steel slag, as they contain Cr, Ni, and V and Cr and V, respectively, while these are absent in the concrete fines.

2 Materials and methods

2.1 Experimental set-up

A total of 30 mesocosms (0.6 m height and 0.25 m radius) were constructed at the experimental site at the Drie Eiken campus of the University of Antwerp (51°09^′ N, 04°24^′ E), Belgium, and placed outdoors to receive natural rainfall (Fig. 1a). In May 2021, the bottom 40 cm of each mesocosm was filled with a sandy loam soil obtained from a pasture in Zandhoven, Belgium (Table 1). The upper 20 cm was filled with the same soil, either unamended in the control treatment (five replicates) or amended with one of three different types of silicates (basalt, concrete fines, or steel slag) mixed into the topsoil layer on 10 May 2021.

Figure 1 Panel (a) presents an overview of the experiment. Panel (b) shows one mesocosm with two maize plants.

Table 1 Soil characteristics of the original soil, before the start of the experiment.

Eifel basalt, comprised of a particle size at which 80 % of the material's mass was smaller than 310.78 µm (P80 = 310.78 µm), was obtained from Durubas (https://www.rpbl.de, last access: 6 June 2025) (Tables S1, S2). Blast-oxygen furnace slag (BOF-slag, hereafter referred to as steel slag), with P80 = 201.65 µm, was sourced from ArcelorMittal Ghent (https://belgium.arcelormittal.com, last access: 6 June 2025) (Tables S1, S2). Concrete waste was obtained from a local concrete recycling company (Recybo, http://www.recybo.be, last access: 6 June 2025) and dry sieved over a 2 mm sieve (Tables S1, S2). The fraction of material <2 mm is further referred to as “concrete fines”. The elemental composition of the silicate materials can be found in Table 2.

Table 2 Elemental composition (C – concentration) of basalt, concrete fines, and steel slag.

A dose–response experiment was established in which the silicates were amended using different concentrations: seven application amounts of basalt (10–200 t ha⁻¹), seven application amounts of concrete fines (7–31 t ha⁻¹), and seven application amounts of steel slag (1–10 t ha⁻¹) (Table 3). For basalt, there were five replicates of 50 t ha⁻¹, as this material has commonly been used in previous EW studies (Gillman et al., 2001; Swoboda et al., 2022). The application amounts were selected based on typical applications found in the literature, while also aiming for similar neutralising effects among silicate materials (based on the CaO and MgO in the materials and a small lab test with the materials) (e.g. Kelland et al., 2020; White et al., 2017). Concrete fines could weather faster than 10⁻⁶ mol m⁻² s⁻¹, depending on their carbonation (Palandri and Kharaka, 2004), and the dissolution rate of steel slag ranges between 10⁻⁹ and 10⁻⁷ mol m⁻² s⁻¹ (De Windt et al., 2011). This is faster than the weathering rates of basalt (10⁻¹²–10⁻¹⁰ mol m⁻² s⁻¹) (Gudbrandsson et al., 2011). Therefore, the application rate of basalt covers a wider range than concrete fines and steel slag.

Table 3 Application amounts of basalt, concrete fines, and steel slag and the number of replicates for each application rate. The five replicates for which no silicate material was applied are the same mesocosms for the three treatments, i.e. the experiment contained five control treatments.

Each mesocosm was equipped with a 2 cm diameter hole at the bottom for leachate collection, and a root exclusion mat covered the bottom of the mesocosm to prevent soil export through leaching. Glass collectors with a volume of 2.3 L were connected to the mesocosm via polyurethane tubing. On 3 June 2021, two sweet corn seedlings (variety Tom Thumb, purchased at Le Grenier) were planted in each mesocosm (Fig. 1b), and all pots were fertilised with nitrogen–phosphorous–potassium (NPK) fertiliser (96–10–79 kg ha⁻¹) by adding Ca(NO₃)₂, triple superphosphate (TSP, 45 % P₂O₅) and K₂SO₄. The fertilisation amount was similar to that used in Ven et al. (2019), except for P, which was halved to avoid overfertilisation in combination with the added silicates.

Soil water content and temperature were monitored using Campbell Scientific sensors (CS616) in each mesocosm (Fig. S2). Precipitation and air temperature for the duration of the experiment were retrieved from Visual Crossing Weather (https://www.visualcrossing.com/, last access: 3 September 2024) (Fig. S3). Mesocosms were also watered manually (Table S2) at the beginning of the experiment and during dry periods. Analysis to determine the CO₂ removal potential in this study were performed, and an in-depth assessment of the weathering rates and CO₂ removal is presented in Vienne et al. (2025).

2.2 Plant measurements

Throughout the growing season, plant height was monitored at 10 times. On 15 July and 11 August 2021, the leaf area index (LAI) was determined for each mesocosm, following the approach of Ven et al. (2019). For every plant, the number of leaves was counted (n_leaf) and the width (W_leaf) and length (L_leaf) of each leaf were measured. First, individual leaf area (LA) was calculated as follows:

$$\begin{array}{}\text{(1)}& \mathrm{LA}=L_{\mathrm{leaf}}\times W_{\mathrm{leaf}}\times A_{\mathrm{leaf}},\end{array}$$

using a conversion factor of A_leaf=0.75 for maize (Montgomery, 1911). The LAI was then calculated for each mesocosm as follows:

$$\begin{array}{}\text{(2)}& \mathrm{LAI}=\mathrm{LA}\times n_{\mathrm{plant}}\times n_{\mathrm{leaf}}\times A^{-\mathrm{1}},\end{array}$$

where A is the area of the mesocosms (0.20 m²).

Plants were harvested on 26 August 2021, starting with the top leaf of each plant for further analysis. The aboveground biomass was then harvested and separated into stems, leaves, tassels, and corn. A week after harvesting, roots were sampled from topsoil (0–20 cm) and subsoil (20–60 cm) to estimate the root biomass within each mesocosm. One soil core (100 cm³) was taken below the shoot of each plant (resulting in a total of two cores per mesocosm for each soil layer, as each mesocosm contains two plants), and one core was taken for each soil layer at the centre of the mesocosms. A similar method was used in Ven et al. (2020), who were able to close the carbon (C) balance using a similar approach, demonstrating its accuracy. Subsequently, soil was carefully rinsed over a 2 mm sieve positioned above a receptacle. Roots were collected from the sieve, including those floating on the water in the receptacle. After drying for 48 h at 70 °C, the dry weight (dw) of each plant part was determined. For roots, the average dry root biomass (g cm³) was calculated assuming that the core from the centre of the mesocosm represented 50 % of the root distribution across the surface area, while each of the two cores underneath the plants accounted for 25 % each. Plant samples (aboveground and root biomass) were dried at 70 °C for 48 h.

Leaves, stems, tassels, corn seeds, and roots were ground using a centrifugal mill (model ZM 200, Retsch GmbH, Haan, Germany) with a sieve of 0.25 mm mesh size. All ground samples were analysed for nutrients (Ca, Fe, K, Mg, P, and Si), using inductively coupled plasma optical emission spectroscopy (ICP-OES; iCAP 6300 Duo, Thermo Scientific), and C and N content, via dry combustion based on the Dumas method using an elemental analyser (model FLASH 2000, Interscience, Louvain-la-Neuve, Belgium). Leaves, stems, tassels, and corn seeds were also analysed for toxic trace elements (cadmium, Cd; Cr; Ni; lead, Pb; and V). Due to limited root samples that were harvested, toxic trace element concentrations in roots were not analysed in our study. For each plant sample, 0.3 g was weighed and digested with H₂SO₄, salicylic acid, H₂O₂, and selenium to determine Ca, Fe, K, Mg, and P and the toxic trace elements listed above according to Walinga et al. (1995). Si was determined by digestion of a 30 mg plant sample with 25 mL 0.5 N NaOH. Pb and Ni concentrations were compared to the maximum allowable levels. As these are reported per gram wet weight (ww) of plant material, we converted our concentrations from dw to ww. For corn Pb, the limit of quantification (LOQ) (0.1 mg kg⁻¹ dw) was used, as all Pb concentrations in corn were lower.

2.3 Soil measurements

Starting from 19 May 2021, soil porewater samples were collected weekly for chemical analysis using Rhizon samplers (Rhizon Flex, Rhizosphere Research Products B.V., Wageningen, the Netherlands) installed at 5 cm depth in each mesocosm. Soil porewater pH was measured with a Metrohm 914 pH/Conductometer. For chemical analysis, samples were filtered through a 0.45 µm polyethylene terephthalate (PET) filter. Concentrations of Ca, Cd, Cr, Fe, Mg, Ni, Pb, Si, and V were analysed using ICP-OES (iCAP 6300 Duo, Thermo Scientific). Dissolved inorganic carbon (DIC) was measured with a Skalar (FORMACS^HT). Ca, Fe, K, Mg, Si, and DIC were analysed weekly, whereas Cd, Cr, Ni, Pb, and V were analysed once on 18 August 2021, near the end of the growing season.

Topsoil pH was measured on five occasions (25 May, 10 and 24 June, 19 July, and 4 August) by collecting and pooling three subsamples per mesocosm from right underneath the soil surface (+1 cm). Subsequently, 4 g of soil was suspended in 10 mL of deionised water and shaken before analysis using a Metrohm 914 pH/Conductometer. Plant Root Simulator probes (PRS^®, Western Ag Innovations, Saskatoon, Canada) were inserted into the soil twice: on 24 June and 2 August 2021. They were retrieved 1 week later, on 1 July and 9 August 2021, respectively. More information about these ion exchange membranes is provided by Qian and Schoenau (2002). Soil CEC was determined following the protocol of Brown (1943) for which approximately 2.5 g of air-dried soil was extracted with 1 M NH₄ acetate for 1 h at room temperature with continuous agitation. Extracts were measured using ICP-OES (iCAP 6300 Duo, Thermo Scientific).

2.4 Statistical analysis

All statistical analyses were conducted in R (RStudio, 2021.09.0.0). The level of significance for all analyses was set at p≤0.05. Data normality of residuals was checked using the Shapiro–Wilk test, and the homoscedasticity of the residuals was examined by plotting them against the fitted values. If one of these assumptions was violated, a logarithmic, inverse or square root transformation was applied to the data. Plots were designed with the ggplot2 package (Wickham, 2016), ggpubr package (Kassambara, 2023) and the Rmisc package (Rmisc, 2024).

Linear regression analyses were performed to explore the relationship between the application amount of each silicate material separately and the variables of interest, i.e. plant biomass, nutrient and toxic trace element content of the different plant parts, toxic trace element content of the porewater, and the soil CEC. If the normality or homoscedasticity of the residuals of the linear model was not met after data transformation, a non-parametric Kruskal–Wallis test was used. To assess changes over time and differences among application amounts for each silicate type, a linear mixed model (nlme package; Pinheiro et al., 2013) was applied for plant height, soil pH, soil porewater pH, soil porewater DIC, soil porewater nutrient concentrations, and nutrient availability from the PRS probes. Silicate application amount, time (days after sowing, or burial date for the nutrient availability from the PRS probes), and their interaction were included as fixed effects, while mesocosm was treated as a random effect. To account for heteroscedasticity in the data across different time points, the weights = varIdent component was incorporated into the model. Differences among silicates were investigated using Method S1 in the Supplement.

To investigate which soil parameters are linked with differences in biomass, a linear model was constructed based on the Akaike information criterion (AIC). All soil parameters (soil pH, CEC, porewater pH, porewater nutrients, and porewater toxic trace elements) were included in the original model, whereafter the best fit was selected with the lowest AIC score. Hereafter, a two-way ANOVA was performed to search for soil parameters that significantly influence plant biomass. To reduce the complexity of the data, a principle component analysis (PCA) was done to explore differences in soil variables (soil and porewater pH, CEC, porewater nutrients, and porewater toxic trace elements) among the three silicate treatments. Linear regression analysis was performed with each principal component (PC) as a function of silicate treatment and as a function of plant biomass.

3 Results

3.1 Soil chemistry

Application of basalt, concrete fines, and steel slag resulted in significant increases in soil pH, with the most pronounced increases for concrete fines and steel slag (Fig. 2, Table S3). Porewater pH and DIC also generally increased along with the basalt, concrete fines, and steel slag application amount (Fig. 2, Table S3), with a more pronounced DIC increase with increasing steel slag application amount compared to basalt and concrete fines (Fig. 2, Table S4). The basalt effect on DIC increased over time, whereas the steel slag effect slightly decreased as the growing season progressed (Fig. 2).

Figure 2 Soil pH, porewater DIC, and porewater pH during the experiment for the basalt, concrete fines, and steel slag treatments. The control treatment (0 t ha⁻¹) and basalt treatment with an application rate of 50 t ha⁻¹ are averages of five replicates and are shown with their standard error. The other treatments each have one replicate. P values and F values are shown from linear regression analysis with soil pH, porewater pH, or DIC as the fixed effect and the silicate application amount, time, and the interaction as the covariables. If the interaction was not statistically significant, this was deleted from the model and is, thus, not shown here. Statistically significant relationships are indicated by an asterisk (^*).

Figure 3 Ca, Fe, K, Mg, and Si in the porewater during the experiment for the silicate treatments (basalt, concrete fines, and steel slag). The control treatment (0 t ha⁻¹) and basalt treatment with an application rate of 50 t ha⁻¹ are averages of five replicates and are shown with their standard error. Note that the error bars are smaller than the symbol in some cases. The other treatments have one measurement each. P and F values are shown in Table S3.

The influence of silicate application on soil porewater Ca concentrations varied among silicate material types (Fig. 3, Table S4). Increases were observed with increasing application amount of concrete fines and steel slag (Fig. 3, Table S3), which is similar to plant-available Ca from PRS probes (Fig. S3). Conversely, porewater Ca concentrations decreased with increasing basalt application amount. However, data for the highest application amounts of basalt (150 and 200 t ha⁻¹) are only available for the last two sampling dates, during which Ca concentrations increased with increasing basalt application amount (p<0.01, F=26.4, statistical analysis only performed on the last two sampling dates). This increase is supported by data retrieved from the PRS probes, where plant-available Ca increased with basalt application amount (Fig. S3). Nonetheless, plant-available Ca was lower with basalt compared to concrete fines and steel slag application (Table S5).

A significant time × application amount interaction effect was found for the Mg and Si concentrations in the soil porewater (Fig. 3, Table S3). Both increased with basalt, concrete fines, and steel slag application amount, yet the positive effect on Mg concentrations diminished towards the end of the growing season (Fig. 3, Table S3). Plant-available Mg from the PRS probes also increased along with the application amount of basalt and concrete fines (Fig. S4). Increases in Si with basalt and steel slag were higher at the beginning of the experiment, whereas increases in Si with the application amount of concrete fines were relatively constant over time (Fig. 3, Table S3). In contrast, Fe concentrations decreased with an increasing application amount of concrete fines, whereas these were not affected by basalt or steel slag application (Fig. 3, Table S3). However, this was not supported by the bioavailability of Fe retrieved from the PRS probes, which was not affected by silicate application (Fig. S3). Porewater K concentrations significantly increased with the application amount of concrete fines, whereas no effects were observed with basalt and steel slag (Fig. 3, Table S3).

Figure 4 Concentrations of Cr, Ni, Pb, and V in the porewater at the end of the growing season for the basalt, concrete fines, and steel slag treatments. The control treatment (0 t ha⁻¹) and basalt treatment with an application rate of 50 t ha⁻¹ are averages of five replicates and are shown with their standard error. The other treatments each have one replicate. P values and F values are shown from a linear regression analysis with the toxic trace element (Cr, Ni, Pb, and V) concentration as the fixed effect and silicate application amount as the covariable. Statistically significant relationships are indicated by an asterisk (^*).

In over 73 % of the porewater samples, Cd concentrations were below the LOQ (Table S6), rendering statistical analysis unfeasible. Cr concentrations remained unaffected by basalt, whereas a decrease was observed with increasing application amount of concrete fines, in contrast to increasing Cr concentrations with increasing steel slag application amount (Fig. 4, Table S4). Porewater Ni and Pb concentrations increased with basalt application amount, yet remained unaffected by concrete fines or steel slag application. Moreover, these concentrations were higher with basalt compared to concrete fines and steel slag application (Fig. 4, Table S4). These changes in Pb concentrations contrast with PRS probe data, where Pb increased with concrete fines application but remained unchanged with basalt and steel slag application amount (Fig. S4, Table S5). Porewater V concentrations were higher with steel slag application compared to basalt and concrete fines and increased with the application amount of basalt, concrete fines, and steel slag (Fig. 4, Table S4).

Figure 5 PC1 and PC5 of the principal component analysis with porewater nutrients (Ca, Fe, K, Mg, and Si), toxic trace elements (Cr, Ni, Pb, and V), porewater pH, soil pH, and soil CEC. Differences in PC1 and PC5 were found among silicate materials (B – basalt, C – control, CF – concrete fines, and S – steel slag) by linear regression analysis with PC1 or PC5 as the fixed variable and the silicate treatment as the covariable. Similar letters denote no statistically significant differences among the materials, while different letters mean that PC1 or PC5 differed significantly among silicate materials, using a Tukey post hoc test. P and F values are shown in Table S7.

The PCA analysis showed a significant effect of silicate material on PC1 (p<0.01, F=32.9) and PC5 (p<0.01, F=8.18) (Fig. 5). PC1 is negatively correlated with porewater Pb and Ni concentrations but positively correlated with soil pH, CEC, DIC, and porewater Ca and V concentrations. PC1 is significantly lower for basalt compared to concrete fines and steel slag (both p<0.01; Table S7). PC5 is negatively correlated with soil pH but positively correlated with concentrations of Fe, Cr, and V in the porewater. PC5 is significantly higher for steel slag compared to basalt (p=0.03) and concrete fines (p<0.01). While PC2 (positively correlated with Ca, Fe, Mg, Si, and K porewater concentrations but negatively correlated with soil CEC, porewater Cr, and DIC) did not differ among the treatments, PC3 (negatively correlated with CEC and porewater Ni, Pb, and Mg) and PC4 (negatively correlated with porewater Cr and pH but positively correlated with porewater V concentrations) showed significant differences among treatments (Fig. 8, Table S7, Fig. S5). PC3 is significantly lower with basalt compared to the control, while PC4 is significantly higher with steel slag compared to concrete fines (Table S7, Fig. S5). The differences among concrete fines and steel slag were more prominent for PC5, which only explained 3.1 % of the variance less than PC4 (Figs. 5, S5; TableS7).

3.2 Plant growth parameters

Although basalt application did not affect plant height or the LAI, plant biomass increased with basalt application amount (Figs. 6, 7; Table S8). This increase was significant for the stem, tassel, and total aboveground and root biomass, whereas it was borderline significant for the leaves (Fig. 7). Total aboveground biomass was significantly higher with basalt application compared to concrete fines and steel slag (Fig. 7, Table S9). Contrastingly, plant height increased over time with an increasing concrete fines (almost statistically significant) and steel slag (statistically significant) application amount, but the LAI and the stem, leaves, corn, and root biomass remained unaffected; hence, total aboveground biomass did not change (Figs. 6, 7, Table S8). Nonetheless, tassel biomass increased significantly with both increasing application amounts of steel slag and concrete fines (Fig. 7).

Figure 6 Plant growth during the experiment for the basalt, concrete fines, and steel slag treatments. The control (0 t ha⁻¹) and basalt treatment with an application rate of 50 t ha⁻¹ are averages of five replicates and are shown with their standard error. The other treatments have one replicate. P values and F values are shown from a linear mixed model with the plant height as the response variable and silicate application amount, time, and interaction as the covariable. If the interaction was not statistically significant, this was deleted from the model and is, thus, not shown here. Statistically significant relationships are indicated by an asterisk (^*).

The aboveground $/$ belowground ratio was unaffected by any treatments (Table S8). The model selected by the AIC method to investigate which soil parameters affected biomass included the K, Ca, Pb, and Cr concentrations in the soil porewater and soil pH and CEC. Of these variables, total aboveground biomass was significantly positively correlated with soil porewater Ca (p<0.01, F=3.47) and Pb (p<0.01, F=4.19) concentrations and with the soil pH (p=0.03, F=2.81). The PCA analysis showed a significantly negative relationship between biomass and PC3, which is negatively correlated with CEC values and porewater Mg, Ni, and Pb concentrations (Fig. 8).

Figure 7 Biomass of the different plant parts (leaves, corn, tassel, and stem) and total aboveground biomass of maize plants for the basalt, concrete fines, and steel slag treatments. The control (0 t ha⁻¹) and basalt treatment with an application rate of 50 t ha⁻¹ are averages of five replicates and are shown with their standard error. The other treatments each have one replicate. P and F values are shown from a linear regression analysis with biomass as the response variable and concentration of the silicate material as covariable. A statistically significant regression is indicated by an asterisk (^*), with the equation and regression line.

3.3 Plant nutrients

An increased basalt application amount resulted in decreased Ca concentrations in the aboveground plant parts, although this was only significant for the stem and the tassel. Root Ca concentrations, on the other hand, increased with increasing basalt application amount (Fig. 9), but these were generally lower with basalt than with concrete fines or steel slag application. Mg concentrations increased in the plant parts with increasing basalt application, although these increases were not statistically significant for the top leaf and the corn (Fig. 9). Root Mg concentrations were significantly higher with basalt compared to concrete fines and steel slag (Table S10). Basalt application did not significantly affect the Fe and K concentration in any of the plant parts, and the P concentration remained unaffected except for a significant increase in the tassel (Fig. 9). Basalt application had no significant impact on the Si concentration in the top leaf, tassel, or roots. In stems, Si concentrations decreased with an increasing application amount of basalt, whereas the Si concentration increased in the corn (Fig. 9).

Figure 8 Scatter plot of PC1 and PC3 of the principal component analysis with porewater nutrients (Ca, Fe, K, Mg, and Si), toxic trace elements (Cr, Ni, Pb, and V), porewater pH, soil pH, and soil CEC. The sizes of the dots reflect the total aboveground biomass, which was significantly correlated with PC3. Different silicate treatments (B – basalt, C – control, CF – concrete fines, and S – steel slag) are displayed using different colours.

The $\mathrm{C}/\mathrm{N}$ ratio of plant tissues was usually not affected by basalt application, although the $\mathrm{C}/\mathrm{N}$ ratio decreased with increasing basalt application amount in the tassels, whereas the $\mathrm{C}/\mathrm{N}$ ratio increased in the top leaf due to reduced N concentration (Figs. 10, S6). Total plant C stocks increased with an increasing basalt application amount, whereas N stocks were not affected (Fig. S7). Furthermore, total plant Mg and K stocks increased with an increasing basalt application amount, whereas Ca, P, and Si stocks remained unaffected (Fig. 11).

Figure 9 Concentrations of Ca, Fe, K, Mg, P, and Si in corn, tassel, roots, stem, and top leaf of the basalt treatment. Data of 0 and 50 t ha⁻¹ of basalt are averages of five replicates with their standard error. Note that the error bars are smaller than the symbol in some cases. The other application amounts each have one replicate. P and F values are shown from the linear model with the elemental concentration as the response variable and the silicate concentration as the covariable. Statistically significant relationships are indicated with an asterisk (^*), with the equation and regression line.

Figure 10 The $\mathrm{C}/\mathrm{N}$ ratio in the corn, tassel, stem, top leaf, and roots of maize for the silicate treatments (basalt, concrete fines, and steel slag). Data without silicates (0 t ha⁻¹) and 50 t ha⁻¹ of basalt are an average of five replicates with their standard error. Note that the error bars are smaller than the symbol in some cases. The other treatments each have one replicate. P and F values are shown from the linear model with the $\mathrm{C}/\mathrm{N}$ ratio as the response variable and the silicate concentration as the covariable. Statistically significant relationships are indicated with an asterisk (^*), with the equation and regression line.

Figure 11 Stocks of Ca, K, Mg, P, and Si in the crops for the silicate treatments (basalt, concrete fines, and steel slag). Data without silicates (0 t ha⁻¹) and 50 t ha⁻¹ of basalt are an average of five replicates with their standard error. Note that the error bars are smaller than the symbol in some cases. The other treatments each have one replicate. P and F values are shown from the linear model with the stocks as the response variable and the silicate concentration as the covariable. Statistically significant relationships are indicated with an asterisk (^*), with the equation and regression line.

Figure 12 Concentrations of Ca, Fe, K, Mg, P, and Si in the corn, tassel, roots, stem, and top leaf of the concrete fines treatment. Data of the control treatment (0 t ha⁻¹) are an average of five replicates with their standard error. Note that the error bars are smaller than the symbol in some cases. The other application amounts each have one replicate. P and F values are shown from the linear model with the elemental concentration as the response variable and the silicate concentration as the covariable. Statistically significant relationships are indicated with an asterisk (^*), with the equation and regression line.

Figure 13 Concentrations of Ca, Fe, K, Mg, P, and Si in the corn, tassel, roots, stem, and top leaf of the steel slag treatment. Data of the control treatment (0 t ha⁻¹) are an average of five replicates with their standard error. Note that the error bars are smaller than the symbol in some cases. The other application amounts each have one replicate. P and F values are shown from the linear model with the elemental concentration as the response variable and the silicate concentration as the covariable. Statistically significant relationships are indicated with an asterisk (^*), with the equation and regression line.

The application of concrete fines did not affect plant Ca concentrations, except for a significantly increased root Ca concentration and a tendency toward decreased tassel Ca concentrations (p=0.08) (Fig. 12). Stem Ca concentrations were higher with concrete fines compared to basalt and steel slag (Table S10). Mg concentrations increased with an increasing concrete fine application amount, although this increase was not statistically significant in the corn or roots (Fig. 12). Corn Si concentrations also increased along with concrete fines, and they were higher than with basalt or steel slag (Fig. 12, Table S10). Concrete fine application did not affect plant Fe, K, or P concentrations. The $\mathrm{C}/\mathrm{N}$ ratio significantly increased in the top leaf, even though C and N concentrations were not significantly affected by concrete fines (Figs. 12, S6). In the other plant parts, the $\mathrm{C}/\mathrm{N}$ ratio remained unaffected (Fig. 10). Total plant C stocks increased significantly with an increasing application amount of concrete fines, whereas total plant N stocks were not affected (Fig. S7). Plant P and Si stocks were not influenced by the application amount of concrete fines, whereas Ca, K, and Mg stocks increased with concrete fines, although this increase was only statistically significant for Ca and K (Fig. 11).

Steel slag application did not affect plant Ca concentrations, except for decreased stem Ca concentrations (Fig. 13). In leaves and tassels, Mg concentrations significantly increased with steel slag application amount, whereas these concentrations decreased in the stems (Fig. 13). Stem Mg concentrations were also lower with steel slag application than with basalt or concrete fines (Table S10). Corn and root Mg concentrations were not affected (Fig. 13). Steel slag application did not influence the Fe, K, P, or Si concentrations in the plants except for increased corn Si concentrations with an increasing steel slag application amount (Fig. 13). Leaf Si concentrations were higher with steel slag compared to basalt and concrete fines (Table S10). The $\mathrm{C}/\mathrm{N}$ ratios in the plant parts remained unaffected by steel slag application (Fig. 10), although stem N and top-leaf C concentrations decreased (Fig. S6). Total plant C and N stocks (Fig. S7) and Ca, Mg, P, K, and Si stocks (Fig. 11) remained unaffected by steel slag application.

3.4 Toxic trace elements

In all corn samples and in over 85 % of the stem samples, Pb was below the LOQ (0.1 mg kg⁻¹ dw) and, thus, could not be statistically analysed. Pb was only detected in one control treatment and three basalt treatments (Table S6). Corn Cd concentrations were below the LOQ (0.015 mg kg⁻¹ dw) in all samples of the steel slag treatment, and Cd was only detected once with concrete fines and once with basalt application (Table S6). Basalt application did not significantly affect the toxic trace element concentration in the plant parts, except for decreasing Cd (statistically significant), with the largest decrease of 54 % when 75 t ha⁻¹ of basalt was applied, and V (borderline significant) concentrations in the top leaf (Fig. 14).

Figure 14 Concentrations of Cd, Cr, Ni, Pb, and V in the corn, tassel, stem, and top leaf of the basalt treatments. Data of 0 and 50 t ha⁻¹ of basalt are an average of five replicates with their standard error. Note that the error bars are smaller than the symbol in some cases. For 10, 25, 75, 100, 150, and 200 t ha⁻¹ of basalt, there is one measurement each. P and F values are shown from the linear model with the toxic trace element concentration as the response variable and the silicate concentration as the covariable. Statistically significant relationships are indicated with an asterisk (^*), with the equation and regression line.

The application of concrete fines generally decreased plant Cr concentrations, albeit only significantly in corn and tassel (Fig. 15), with a maximum decrease of 81 % with the highest application amount and 46 % for 10.57 t ha⁻¹ of concrete fines, respectively. Additionally, tassel Cd concentrations decreased with an increasing application amount of concrete fines, while Cd concentrations in the other plant parts remained unaffected. Corn Ni concentrations decreased significantly, with a maximum decrease of 60.8 % when 21.15 t ha⁻¹ of concrete fines material was applied. A non-significant trend toward decreased stem Ni concentrations was observed with an increasing concrete fine application amount, while Ni concentration was unaffected in the other plant parts (Fig. 15). Neither plant Pb concentrations nor plant V concentrations were affected by the application of concrete fines, except for a decrease in V observed in the top leaf (Fig. 15).

Apart from decreased Cd concentrations in the stem (statistically significant) and in tassels (borderline significant), Cd levels were not affected by steel slag application (Fig. 16). Stem Cd concentrations were lower with steel slag compared to basalt and concrete fines (Table S11). Furthermore, steel slag application decreased corn Cr concentrations, by up to 83 % with an application amount of 3 t ha⁻¹ of steel slag. No influence was observed in the other plant parts. V and Pb concentrations were not affected by steel slag application, except for an increase observed in the top leaf (Fig. 16). With the highest application amount of steel slag, Pb concentrations increased by 47 % compared to the control treatment.

Figure 15 Concentrations of Cd, Cr, Ni, Pb, and V in the corn, tassel, stem, and top leaf for the concrete fines treatments. Data of 0 t ha⁻¹ of concrete fines are an average of five replicates with their standard error. Note that the error bars are smaller than the symbol in some cases. For 7.04, 10.57, 14.09, 17.62, 21.15, 24.67, and 31.07 t ha⁻¹ of concrete fines, there is one measurement each. P and F values are shown from the linear model with the toxic trace element concentration as the response variable and the silicate concentration as the covariable. Statistically significant relationships are indicated with an asterisk (^*), with the equation and regression line.

Figure 16 Concentrations of Cd, Cr, Ni, Pb, and V in the corn, tassel, stem, and top leaf for the steel slag treatments. Data of 0 t ha⁻¹ of steel slag are an average of five replicates with their standard error. Note that the error bars are smaller than the symbol in some cases. For 2, 3, 4, 5, 6, 7, and 10 t ha⁻¹ of steel slag, there is one measurement each. P and F values are shown from the linear model with the toxic trace element concentration as the response variable and the silicate concentration as the covariable. Statistically significant relationships are indicated with an asterisk (^*), with the equation and regression line.

4 Discussion

4.1 Silicate weathering and soil properties

The DIC and base cations such as Ca²⁺ and Mg²⁺ in the porewater provide an indication of the weathering rate of the added silicate minerals (Haque et al., 2019; Kelland et al., 2020). As expected, increases in DIC were lowest for basalt compared to concrete fines and steel slag, indicating a higher weathering rate for the latter two silicates. This was accompanied by a higher initial increase in soil and porewater pH for steel slag and concrete fines. Concrete fines used in this study contained about 18 % calcite, whereas steel slag contained about 9 % calcite. Calcite weathers faster than silicate minerals (Berner et al., 1983; Lehmann et al., 2023), potentially explaining the higher increase in DIC with concrete fines and steel slag compared to basalt, as basalt does not contain calcite. Even though the CO₂ removal of these silicates is not part of the current study, we want to emphasise that calcite weathering does not contribute to long-term carbon capture and has a risk of reversal if carbonates reprecipitate downstream after leaching (Berner et al., 1983; Lehmann et al., 2023). In other words, the higher increase in weathering products with concrete fines and steel slag in our experiment does not imply a proportionate increase in CO₂ removal. EW studies have also consistently shown increases in DIC and pH following silicate application, whether using a mixture of rock types (Guo et al., 2023), basalt (Buckingham and Henderson, 2024; Conceição et al., 2022), or steel slag (Buckingham and Henderson, 2024).

Moreover, the fact that the release of Mg and Ca increased with the application amounts of the silicates, as evidenced by soil porewater, PRS probes, and plant measurements, is in line with other experiments (Amann et al., 2020; Buckingham and Henderson, 2024; Guo et al., 2023; Vienne et al., 2022; Yan et al., 2023). Soil porewater Si concentrations also increased in our study (Fig. 3), similar to previous studies (e.g. Buckingham and Henderson, 2024; Guo et al., 2023; Vienne et al., 2022). The observed variability in nutrient concentrations, especially Si, for the control treatment and the 50 t ha⁻¹ of basalt is likely due to local differences in soil conditions and fluctuations in porewater chemistry. Despite this variability, significant differences among treatments were identified. However, because this study did not include replicates for the other application amounts, the extent of variability could not be confirmed.

In contrast, Fe was unaffected and even decreased with the application of concrete fines. The latter may be attributed to the increased pH, which can stimulate the formation of insoluble Fe compounds, reducing Fe availability (e.g. Lindsay et al., 2008; Shenker and Chen, 2005). Basalt and steel slag contain more Fe than concrete fines (see Table 2) and may have supplied sufficient Fe to counterbalance decreases in Fe as a result of increased pH.

The silicate materials in our study contained toxic trace elements which can also be released. Given their known toxicity for plants (Nagajyoti et al., 2010), it is imperative to assess the associated risks of toxic trace element pollution when considering EW in agriculture. Indeed, porewater concentrations of Cr, Ni, Pb, and V increased in some of our treatments. Increased porewater Ni concentrations with basalt and higher porewater Cr concentrations with steel slag can be explained by the material composition (Table 2). On the other hand, porewater V concentrations also increased with an increasing amount of concrete fines material, which did not contain V. Thus, porewater V concentrations were not only influenced by the release of V from the silicates (as they also increased with steel slag, which does contain V) but also by changes in soil properties. As V is more mobile with increasing pH (Chen et al., 2021), we presume that the higher pH (Fig. 2) increased porewater V concentrations. The availability of Pb is generally also controlled by soil pH. At a higher pH, Pb is less available for plants (Kushwaha et al., 2018). Despite basalt lacking Pb and the rise in soil pH, porewater Pb concentrations increased with increasing basalt application. This unexpected increase may be related to competition for soil binding sites, with the released Ca species increasingly outcompeting Pb due to their similar size and reactivity (Klitzke and Lang, 2009; Li et al., 2014).

4.2 Plant growth

Plants can benefit from EW because it can create a more favourable environment (Calabrese et al., 2022; Kelland et al., 2020). Indeed, basalt application increased aboveground and belowground biomass. These findings are consistent with observations of tropical systems (Swoboda et al., 2022). Moreover, in temperate climates, basalt has been reported to increase agricultural yield (Kelland et al., 2020; Skov et al., 2024), although other studies found no significant effect (Vienne et al., 2022). Furthermore, increased root biomass with basalt application may indicate higher belowground C inputs by plants; however, it can also increase microbial activity, leading to accelerated soil organic matter decomposition, which can impact soil organic C stocks (Fu and Cheng, 2002; Kögel-Knabner et al., 2022; Kuzyakov, 2002).

In contrast to basalt, the application of concrete fines and steel slag did not affect plant biomass, except for increased tassel biomass. Previously, the application of steel slag has been shown to have varying effects on crop growth. In a tropical climate, soybean yield increased with steel slag application in two consecutive crop cycles (Deus et al., 2020). Similarly, Wang et al. (2015) reported an initial increase in rice yield upon steel slag application on a paddy field, but this effect diminished after two growing seasons. Pistocchi et al. (2017), on the other hand, reported a decrease in wheat yield, while tomato yield increased over three growing seasons upon steel slag application in slightly alkaline loamy sand soils.

In our study, differences in plant biomass were significantly correlated with differences in soil pH; thus, pH may (partly) explain the biomass response. The optimal pH for maize growth is between 5.5 and 7.5 (Lizárraga-Paulín et al., 2011), a threshold that was surpassed early in the growing season for the highest application amounts of concrete fines and steel slag but not for basalt (Fig. 2). Therefore, we presume that the pH was more conducive across the spectrum of basalt application amounts, thereby increasing the plant biomass. This postulation is supported by the PCA analysis, as soil pH was one of the major variables affecting both PC1, which separated the control and basalt treatment from the concrete fines and steel slag, and PC5, which separated the concrete fines and the steel slag treatments. However, plant biomass only differed significantly between basalt and steel slag application but not compared to concrete fines. The high variation in biomass with the application of concrete fines precludes one from drawing firm conclusions. Even though the substantial pH increase may have limited the positive plant growth response with steel slag application, no adverse effects were detected.

4.3 Nutrient status of the plants

Nutrient release from silicate weathering is expected to stimulate plant growth and biomass, with nutrient release varying based on material composition. We expected the same differences in plant stoichiometry, and plant nutrient concentrations indeed aligned with the soil Si and Mg measurements: corn Si and plant Mg concentrations increased with silicate application. However, with basalt application, plant Ca concentrations even decreased in some plant parts and are, thus, not in line with plant availability of Ca, presumably because of competition between Mg and Ca. The Ca $/$ Mg ratio in soil porewater decreased with basalt (Fig. S8). This has previously been related to a reduced plant Ca concentration (Kopsell et al., 2013; Osemwota, 2007). A similar trend in increasing Mg and decreasing Ca concentrations was found with dunite application in wheat and barley (Rijnders et al., 2023) and in ryegrass (Ten Berge et al., 2012).

The three silicate materials contained Mg, resulting in higher porewater Mg and plant Mg concentrations. Mg is an essential plant nutrient that is crucial for various physiological functions, including protein and chlorophyll synthesis, enzyme activation and adenosine triphosphate synthesis, and regulatory processes of photosynthesis (Hawkesford et al., 2012). Mg deficiency often arises in agricultural systems with imbalanced crop fertilisation with N, P, and K, primarily due to competition between Mg and other ions, particularly K and Ca (Guo et al., 2016). The increased Mg availability and plant concentrations indicate that fertilisation with silicate minerals can aid in restoring Mg availability in agricultural soils. Our results correspond to previous experiments employing basalt rocks. Boniao et al. (2002), for example, showed that corn height and biomass increased after basalt amendment and that this was correlated with the Mg concentration in soil porewater. Similarly, in Cameroon, maize cultivation utilising three local basalts revealed the largest yield benefit for the basalt with the highest Mg and Ca content, despite its lower K and P content (Tchouankoue, 2014). Additionally, other minerals, like the Mg silicate dunite, also increased Mg concentrations in wheat and barley (Rijnders et al., 2023).

Even though not considered an essential plant nutrient, Si can protect the plant from (a)biotic stresses by making the plant structure more rigid and stronger (Bhatt and Sharma, 2018). Additionally, Si interacts with several key components of stress signalling pathways of plants (Majeed et al., 2019), which can stimulate crop growth especially when plants are experiencing stress. Indeed, studies in which steel slag increased rice (Ning et al., 2014) and soybean (Deus et al., 2020) yields also reported increased plant Si concentrations. In our study, the plants did not experience any noticeable stress, which may explain why Si concentrations were not related to increased plant biomass.

Unlike Ca, Mg, and Si, plant concentrations of P, K, and Fe were not much affected by silicate application. Despite increased K availability with concrete fines, plant K concentrations remained unaffected. The simultaneous increase in K stocks following the application of basalt and concrete fines suggests that K was sufficiently available across all treatments. While this contrasts with a few earlier studies that did report an increased plant K concentration upon silicate addition (Ten Berge et al., 2012; Skov et al., 2024), our result is in line with the study of Rijnders et al. (2023), in which the application of 220 t ha⁻¹ of dunite did not affect the P and K concentrations in wheat and barley. Plant Fe concentrations also remained unaffected by silicate application, despite a decrease in Fe in soil porewater with concrete fines. Fe is relatively immobile, and symptoms of Fe deficiency typically appear on the new leaves. However, the lowest leaf Fe concentration measured in our study was 0.07 mg g⁻¹ dw (70 ppm), which falls perfectly above the minimum threshold of Fe concentrations in young leaves (i.e. 30 ppm Fe on a dry-weight basis; Hochmuth et al., 2012). Therefore, we can assume that sufficient Fe was available, even though plant-available Fe decreased with concrete fines. In an experiment of Wang and Qing-Sheng (2006), Fe also did not increase with their lowest steel slag application amount (44 t ha⁻¹), using sandy loam soil.

Furthermore, plant-available N did not change with silicate application. While the plant N concentration slightly changed with basalt and steel slag application (Fig. S6), plant N stocks remained unaffected. This was expected because the silicates lack N and every treatment received the same input of N fertiliser. However, the $\mathrm{C}/\mathrm{N}$ ratio within leaf tissues increased with the application of basalt and concrete fines, suggesting a higher C assimilation rate per unit of N, indicating improved N use efficiency or N limitation (Schlüter et al., 2012). Given the biomass increase with basalt and no decrease with concrete fines, the elevated $\mathrm{C}/\mathrm{N}$ ratio does not indicate N limitation, particularly given that the mesocosms received the same amount of N fertilisation.

4.4 Plant toxic trace elements

In addition to nutrients, silicate weathering can be accompanied by toxic trace element release, raising concerns about toxic trace element contamination. The primary risks associated with basalt are Al, Cr, and Ni, while concrete fines contain only Al in a substantial amount, and steel slag contain various toxic trace elements, with notable risks linked to Al and Cr. Despite the release of toxic trace elements from silicate weathering, concentrations of these elements in plant tissues were not affected or even decreased with increasing silicate application amount. This is similar to a previous study in which dunite application did not affect most toxic trace elements and even decreased Ba and Sr concentrations in wheat and barley (Rijnders et al., 2023). Deus et al. (2020) also found that steel slag application did not affect toxic trace element concentrations in soybean, which was probably related to the increase in soil pH. Furthermore, basalt application also did not increase concentrations of Cd, Cr, Ni, and Pb in spring oats (Skov et al., 2024).

In this study, Cd, Cr, and V decreased in the plant parts with increasing silicate application. Decreases in the Cd concentration were presumably a consequence of reduced Cd solubility following the pH increase with silicate application. In keeping with our results, He et al. (2020) reported a decrease in Cd solubility and a subsequent reduction in Cd concentrations in rice following steel slag amendment. In contradiction to our hypothesis, plant V concentrations did not increase in our experiment (and even decreased with concrete fines), despite elevated V concentrations in the soil porewater. Although V availability increases with pH, its uptake by roots decreases as the soil pH increases (Roychoudhury, 2020). Moreover, V that is taken up by plant roots will be predominantly retained in root tissue, particularly in cell walls where multiple proteins and polysaccharides form chelates with heavy-metal ions (Aihemaiti et al., 2020; Roychoudhury, 2020), thereby reducing its accumulation in aerial plant parts (Aihemaiti et al., 2020). This is similar for Cr, which may explain the decreased Cr concentrations in the aboveground plant parts (Ertani et al., 2017). Pistocchi et al. (2017) demonstrated that V and Cr indeed accumulated in the roots of tomato plants without translocation to the aerial plant parts after the application of steel slag.

Furthermore, increased porewater Ni concentrations upon basalt application did not translate into increased Ni concentrations in the plant, presumably due to its uptake competition with other essential ions by plant roots (Sreekanth et al., 2013; Yusuf et al., 2011). This is similar to what was found by Vienne et al. (2022), in which basalt application did not increase Ni concentrations in potatoes. Ni is a crucial component of many enzymes; its deficiency can reduce plant growth, but it can also be toxic for plants at elevated concentrations (Yusuf et al., 2011). Nevertheless, the toxicity level for sensitive species (10 mg kg⁻¹ dw) was not exceeded in our study (Yusuf et al., 2011). In contrast, top-leaf Pb concentrations increased with steel slag application, but these remained below the maximum allowable level of 0.05 mg kg⁻¹ ww that has been set for corn (WHO and FAO, 2022), ranging from $\mathrm{0.58}\times {\mathrm{10}}^{-\mathrm{6}}$ to 0.03 mg kg⁻¹ ww for all of the plant parts and from $\mathrm{0.64}\times {\mathrm{10}}^{-\mathrm{4}}$ to $\mathrm{0.6}\times {\mathrm{10}}^{-\mathrm{3}}$ mg kg⁻¹ ww for corn grains. Thus, the one-time steel slag application up to 10 t ha⁻¹ in our experiment did not pose a problem with respect to toxic trace element contamination.

While reduced toxic trace elements in the plant may be beneficial for crop production at first sight, increased porewater concentrations, potential accumulation in the plant roots, and the immobilisation of these elements in the soil may pose an environmental risk in the longer term. Our short-term experiment did not allow for complete weathering of the silicate materials, and increased release of toxic trace elements may, thus, still occur over longer time periods (Dupla et al., 2023). Furthermore, future decreases in soil pH may re-release these elements into the environment (Kicińska et al., 2022). This way, the toxic trace elements could gradually leach through the soil into the groundwater, potentially posing a risk to water quality and human health (Li et al., 2021; Sbai et al., 2024). Our findings provide valuable insights into the impact of EW on Zea mays using basalt, concrete fines, and steel slag. However, the effectivity, benefits, and risks of EW can vary with soil characteristics and the specific silicate materials used (Abdalqadir et al., 2024; Lewis et al., 2021).

5 Conclusion

In our mesocosm experiment, we investigated the influence of a range of application amounts of basalt, concrete fines, and steel slag on the growth, nutrient status, and toxic trace elements in Zea mays. Basalt application increased plant biomass, whereas plant biomass was not affected by concrete fines and steel slag. Differences in plant biomass are presumably due to exceedance of the optimal soil pH early in the growing season for the highest application amounts of concrete fines and steel slag but not with basalt. However, despite the limited effect on crop growth, plants were not negatively affected by silicate application.

Silicate application increased the soil pH and the porewater Ca, Mg, and Si availability, which translated into increased plant Mg and Si concentrations. Plant Ca concentrations also increased with concrete fines and steel slag application, whereas a decrease in stem Ca was observed for basalt. This difference was probably related to the Ca $/$ Mg ratio of the rock. Consistent with previous studies, toxic trace element concentrations did not increase or even decreased in the aboveground plant parts with silicate application, even though the silicate materials contained and released these elements. Only leaf Pb increased with steel slag application and reached a concentration of up to 0.03 mg Pb kg⁻¹ ww, which is still below the maximum level of 0.05 mg kg⁻¹ ww for corn set by the World Health Organization (WHO). Moreover, the Pb concentration in the corn grains was always below the LOQ. Therefore, we conclude that, in our experiment, crops mostly benefited from silicate application, with the largest benefits observed for basalt application. We did not find concerning toxic trace element accumulation in this short-term experiment, but this effect requires further verification through long-term monitoring.