*3.3. Metal Aging and Dissociation Time in Soil*

The time for the dissociation of metals from the solid-liquid interface was examined with respect to the effect of the time of soil exposure to vehicular emitted metal contamination (i.e., road age) on metal dissociation in soil (Figure 1). Negative marginally significant (*p* < 0.1) linear correlations were found between Tc and the following soil metal concentrations: Cd (R<sup>2</sup> = 0.456, *p* = 0.096), Cu (R2 = 0.560, *p* = 0.083) (Figure 1) and Zn (R<sup>2</sup> = 0.578, *p* = 0.072) (Figure 1). In contrast, the Tc of Mn increased significantly (*p* < 0.05) with the length of time for contamination (R<sup>2</sup> = 0.617, *p* = 0.002), and Ni did not show a linear relationship between Tc and road age (R2 = 0.084, *p* = 0.874).

Optimised linear models between wheat shoot concentrations, i.e., bioavailable metal concentrations and other soil properties for Mn, Ni and Zn all had R<sup>2</sup> > 0.9, but for Cd and Cu, no relationships were able to be developed (Table 4). Total metal concentrations in soil showed a consistent positive relationship with Ni and Zn metal bioavailability, but a negative relationship for Mn. Overall, the metal bioavailability was best explained by the total metal concentration in soil and the road age, while soil properties such as TOC and soil moisture were also important.

**Table 4.** Optimised linear models for the relationship between metal concentration in wheat, and soil chemical and physical properties. Models were determined by stepwise regression using list wise forward selection (criteria of α = 0.05 for entry).


**Figure 1.** The relationship between roadside soil metal dissociation time Tc (seconds) and the age of the adjacent roads representing soil exposure to vehicular emissions (years). R2 = linear correlation. Please note that in total 9 data points are included for each element, although some points overlap and are not visible.

#### **4. Discussion**

Wheat grown in roadside soils accumulated Cd, Mn, Ni and Zn significantly in shoots compared to the control site grown wheat (Table 2). These findings are an indication of the presence of bioavailable metals in the roadside soils tested. In general, out of all the soil assays tested, correlations between metals in wheat shoots and optimized linear models (Table 4) were the best representation of the bioavailability of metals found in the studied roadside soils. Other studies have shown that the toxicity and mobility of metals in soils are as dependent on the total concentrations of metals in soil [31,32].

The magnitude of the r-values for soil concentration (CSOL) relationships in this current study were weak (Tables 3 and 4) and demonstrates that only a small proportion of the variation in metal bioavailability to wheat could be explained via soil solution concentrations. That is, only about 30% of these metals found in wheat shoot could be explained by CSOL for metal concentrations like Zn and Ni. Thus, the CSOL metal concentrations in soil do not appear to be an appropriate measure of bioavailability in roadside soils where the metals have aged in the soil over a number of years. In comparison, CSOL concentrations have previously been reported as a good measure of bioavailability for Zn in spiked roadside soils [33].

The CaCl2 extraction was also found to be a poor measure for metal bioavailability in roadside soils, with the only significant correlation being for Cu (r = −0.582, *p* = 0.077). It should be noted that the r value for Cu in the current study, while significant is not strong, suggesting that CaCl2 extraction is a poor predictor of the variability of the metal bioavailability in roadside soil (aged). In comparison, Menzies et al. [34], in a meta-analysis of the literature, reported CaCl2 as a good measure of Cu bioavailability for grasses in soil.

The concentration of metals in roadside soils as measured by DGT did not correlate significantly with metal concentrations in wheat shoots (Table 3). This indicates that, in roadside soils where metals have aged for multiple years, DGT measurements of soils were not a good measure of the

bioavailability of metals to wheat shoots. While Zhang et al. [33] found good correlations between DGT measurements and Zn plant uptake, their results were for soils that had been recently spiked with metals collected from roadsides, rather than soils where the metals had been present for extended periods of time.

Even though there has been considerable research into metal bioavailability in soil (including Naidu et al. [2], Smolders et al. [35], Zogaj and Düring [11] and many others) there remains a lack of ability to robustly predict bioavailable metal concentrations in soil. Many soil properties such as pH, organic matter content, clay contents/soil composition, and contaminant physicochemical properties such as oxidation state and aqueous solubility are responsible for controlling the behaviour of metals in soil [4]. These properties can be used to predict the fraction of contaminants that may be bioavailable in a given soil, such that a model which can incorporate many of these individual effects on bioavailability may be more effective. The optimised linear models we developed using soil physiochemical properties, different metal concentrations representing different metal extracting techniques and age of the roadside soil showed very strong relationship for Mn, R<sup>2</sup> = 0.936; Ni, R<sup>2</sup> = 0.936 and Zn, R<sup>2</sup> = 0.931 (Table 4). While all the models developed were dependent on total soil metal concentrations, models developed for Mn and Zn clearly demonstrated the effect of road age, i.e., how long metals had aged in situ on metal bioavailability. Thus, optimised linear models have the potential for use in robustly predicting bioavailable metal concentrations in field soils where the metal concentrations have aged in situ.

The low R (CDGT/CSOL) values determined in this research (Table 3) may be as a result of metals being only slowly resupplied from the soil solid phase [36]. The reasons for the low resupply of metals could include limited metal reservoirs, a slow rate constant of the metal resupply process (Kdl) and/or slow desorption metal kinetics [7]. The results suggest a decrease in the R value in soil metals after the wheat assay compared to prior growing wheat. The low R values may be as a result of a reduction in labile metal concentration in the solution due to plant uptake [37].

The biological impact of metals will depend on the rate of the metal dissociation (desorption) from the soil solid to liquid phase and its subsequent availability to organisms. The Tc marginally significantly (*p* ≤ 0.1) decreased with the age of the roads for Cd (*p* = 0.096), Cu (*p* = 0.08) and Zn (*p* = 0.072), while for Mn, Tc significantly increased (*p* = 0.002) with the age of the roads (Figure 1). Therefore, Cd, Cu, Ni and Zn showed a slower release rate from the solid to solution phase compared to Mn in newer roads and a higher rate of release for old roads (Figure 1). This is counterintuitive for all metals except Mn, as it would generally be expected that metals that had aged longer in soils would be more tightly bound and therefore dissociate at lower rates. However, correlation coefficients analysis (Table 3) showed that Tc was positively related to soil pH for Cd (r = 0.630, *p* = 0.028, Ni (r = 0.661, *p* = 0.027) and Zn (r = 0.807, *p* = 0.004) (Table 2). The trend of decreasing soil pH with increasing age of roads (Table 1) is likely to have lowered the Tc for these metals. This is in contrast to Ernstberger et al. [8], who reported no clear trend for Ni between Tc and pH. The decrease of soil pH with the age of the road is possible due to sulphur contained in vehicular emissions, resulting in acid generation in soil [38]. The high Mn concentrations present in the newest roads could have also controlled the labile metal pools of other metals by binding to Mn-oxide. For example the presence of high Mn has been shown to effect the binding ability of Cd with Fe-Mn oxides and organics [39], resulting in lower solubility of the corresponding metal in new roadside soil metals. In addition, Temminghoff et al. [40] observed increases in solubility with increases in the organic carbon content of the soil for Cu and Zn in soil, and a similar scenario may have been operating in the roadside soils measured in the current experiment. Thus, a more alkaline pH, low Mn content and higher TOC in older roadside soils (Table 1) may have caused the higher availability of Cd, Cu and Zn compared to soils besides newer roads. Tc and pH could be used as an important tool in comparing soil metal aging effect in soil solutions, thus metal bioavailability.
