*3.2. Soil Microanalysis*

Micro-XRF was used to map and compare elemental distributions in soil thin sections of both the control and the polluted soil (Figure 1). The major elements, e.g., Al, Si, P, S, K, Ca and Fe, occurred with similar abundances and correlations in the two soils and could be attributed to the soil mineral constituents. Based on XRPD analysis, the overlapping distribution of Al, Si and K testifies to the presence of aluminium silicates, while the dark blue Si-containing particles in Figure 1 (middle panels) are instead related to quartz grains. Calcium is mainly associated with P (purple particles, Figure 1, right panels), presumably forming calcium phosphate phases, such as apatite; in other particles, Ca occurs uncorrelated with other elements, hence it is likely attributable to calcite (pure red particles, Figure 1, right panels), since C and O cannot be detected by XRF.

As expected, Pb was detected only in the polluted soil (S2), concentrated in particles ranging in size from hundreds to thousands of microns, visible as green areas (Figure 1b, right). Except for P, no other element was correlated with Pb. Several Pb-containing particles were then selected on the PTE-polluted soil thin sections and analysed again with increased resolution and accuracy in order to better investigate the nature of these Pb-containing particles as well as Pb vs. P correlations.

In Figure 2, the results of the μXRF hyperspectral data analysis performed on three Pbcontaining particles, selected as the most representative, are shown. For each particle, the optical image is coupled with its false-colour map, which depicts the correlation between Pb and P fluorescent signals detected at the particle section surface.

From the scatterplot of the intensities (expressed as counts) of the Pb L-lines and P K α-lines recorded in each pixel of the three micrographs shown in Figure 2, five main P/Pb correlations were identified and grouped with different colours (Figure 2a). The green region of the scatterplot contains those pixels containing almost only Pb, while the yellow, magenta, red and blue regions contain those pixels with an increasingly higher content of P, beside Pb. By applying such colour-based segmentation to the Pb-particle micrographs (Figure 2b–d, right), it can be generally observed that, starting from the inner only-Pbcontaining core (green areas of the particles), the P amount gradually increases towards the particle boundaries, with the higher P/Pb signal ratio (blue) observed at the border of the particle. Such a pattern is clearly visible for the particle shown in Figure 2c (right), while such distribution is slightly more structured in the other particles (Figure 2b,d, right).

This depends on the particles shape and on the cutting section, which was reasonably diametrical for particle (c) and more external for the others. By looking at the composition of the particles and at their morphology with a light microscope, they appear as metallic Pb slivers deriving from bullet residues. Indeed, many bullet residues are still present in the investigated soils. Furthermore, the comparison between the correlation maps and their corresponding light micrographs (Figure 2b–d, left) shows other additional interesting features: in the case of Figure 2c, for instance, the sliver shows a Pb-rich nucleus (sliver core) which appears pale in visible light, becoming darker as the P/Pb ratio increases moving toward the borders of the particle.

**Figure 1.** Thin section images (left) and corresponding μXRF maps (middle, right) of S1 (**a**) and S2 (**b**) soils. ((**<sup>a</sup>**,**b**) middle panels) Overlay of Al (green), Si (blue) and K (orange) distribution maps. ((**<sup>a</sup>**,**b**) right panels) Overlay of P (blue), Ca (red) and Pb (green) distribution maps. Brighter pixels correspond to higher element concentrations. The co-presence of two or more elements in the same pixels gives rise to different degrees of colour combinations.

**Figure 2.** (**a**) P (K fluorescent signal) vs. Pb (L fluorescent signal) scatterplot obtained using spectral μXRF deconvolution data of (**b**–**d**) particles. (**b**–**d**) particles micrographs (**left**) and related correlation maps (**right**) coloured according to the different P/Pb correlations identified through the scatterplot (**a**)).

Other interesting correlations observed after spectral deconvolution are those between Cl and P and between Ca and P, shown in the scatterplots in Figure 3a,c. Although with low intensity, the Cl–K signal is correlated with that of P (Figure 3a) and is not detected where only the Pb signal is present (without P; grey regions in Figure 3b). Chlorine is located only in the portions of the particles where P was also detected. The scatterplot in Figure 3c instead shows three main Ca–P correlations grouping pixels whose Ca content is yellow < green < red. By looking at the corresponding false-color micrographs (Figure 3d), the primary evidence is that such Ca–P correlations occur in the same pixels where also P–Pb and Cl–P correlations were observed. Moreover, it can be noticed that the higher Ca intensities are located at the very external borders of the Pb-containing particles whilst they decrease as they move towards the inner part (green > yellow).

Beside P, Ca and Cl, no other chemical element has been detected within these Pbcontaining fragments. The ubiquitous occurrence of P and Ca should indicate the presence of Ca phosphates (or even Cl apatite), while the presence of P and Cl, together with Pb should be attributed to the formation of Pb phosphates over the metallic slivers, such as secondary and tertiary Pb phosphates (Pb2HPO4 and Pb3(PO4)2, respectively) and/or pyromorphite. The latter refers to three species, depending on the anion X in the structure Pb5(PO4)3X (namely, chloropyromorphite, where X = Cl<sup>−</sup>; hydroxypyromorphite, where X = OH<sup>−</sup>; and fluoropyromophite, where X = F−). Unfortunately, neither F nor O and H can be detected by μXRF, therefore none of these three forms can be excluded, chloropyromorphite most likely being present.

Lead orthophosphates, above all pyromorphite, are highly insoluble phases. In 1932, Jowett and Price [33] determined an extremely low solubility product (Ksp) of 10−79.115 for Pb5(PO4)3Cl, which is the most insoluble Pb orthophosphate known. In the same study, the authors demonstrated that secondary and tertiary orthophosphates transform to pyromorphite even in the presence of low chloride concentrations, thus explaining the fact that, while the first two phases are not found in nature, pyromorphite is rather common.

Subsequent studies [34,35] proposed for Pb5(PO4)3Cl the Ksp of 10−25.05 for soil with a pH in the usual range. However, even considering the higher value of <sup>10</sup>−25.05, pyromorphite can be still considered several orders of magnitude less soluble than the most common Pb minerals in soils [2]. Indeed, the use of phosphates for amending Pb-polluted soils is a well-known and common remediation practice. By promoting the formation of Pb insoluble species, such as pyromorphite, phosphate addition leads to the reduction of Pb bioavailability and toxicity [2,36].

**Figure 3.** (**a**) Cl (K fluorescent signal) vs. P (K fluorescent signal) and (**c**) Ca (K fluorescent signal) vs. P (K fluorescent signal) scatterplots obtained using spectral μXRF deconvolution data of the three particles of Figure 2. (**b**,**d**) The three particles colored according to the different Cl–P and Ca–P correlations identified through the scatterplots (the background grey micrographs corresponding to Pb–L signals).

SEM-EDX results (Figure 4) also confirmed μXRF observations and allowed us to investigate additional features of the Pb particles and P-bearing encrustations. In the backscattered electron (BSE) image in Figure 4a (left), the darker grey regions, mainly present at the particle boundaries, indicate an average lower Z, thus the presence of lighter elements in addition to Pb. As shown by the corresponding EDX layered image (Figure 4a, right) and spectrum (Figure 4d1), this latter being relative to the white point analyses reported in the EDX image (Figure 4a, right), these low Z elements are P, Ca and Cl. In Figure 4b, which depicts a detail of the P-bearing regions, a cluster of darker grains (identified mostly as K-feldspars) ranging from a sub-micrometric scale to several micrometres in size is dispersed in a binding light grey matrix. Such matrices contain Pb, P, Ca and Cl together with Si and Al, these latter attributable to silicate fragments (region 2 and related spectrum). In Figure 4c (right), the ubiquitous occurrence of Pb, P, Ca and Cl is noticed in a compact region (spectrum d3) and within an aggregate-like structure, as well as in Figure 4b. In both Figure 4b,c, inner portions show almost only the presence of Pb (spectrum d4).

**Figure 4.** (**Left panel**) BSE micrographs of a Pb-containing particle. (**b**,**<sup>c</sup>**) Higher magnification details of the weathering crust of the particle shown in (**a**). (**Middle panel**) EDX layered images. (**d**, **right panel**) EDX spectra of the points (white in (**a**)) and/or regions (yellow numbers in (**b**,**<sup>c</sup>**)) indicated within the middle panel: (1) typical EDX spectrum of the P-bearing coating at particle boundaries; (2) Pb–Ca phosphates matrix in a silicate environment forming an aggregate-like structure; (3) compact Pb and Ca phosphate solid solution; (4) metallic or weakly weathered Pb.

The μXRF and SEM-EDX results described so far sugges<sup>t</sup> that a weathering shell formed over time around the metallic slivers arising from the past firing range activities. Such shells consist, perhaps, of a mixture of Ca phosphate and Pb orthophosphate species, among which it has been possible to distinguish chloropyromorphite. Previous studies demonstrated that Pb stabilization as pyromorphite occurs through an intermediate Pb and Ca solid solution upon hydroxyapatite (HA) dissolution. Such an intermediate phase (Pb(10-x)Cax(PO4)6(OH)2–PbCaHA–) transforms into pyromorphite over time through Ca2+ exchange by Pb2+ [37–39]. In the studied slivers (Figures 2 and 3), the different correlations found between P–Pb and Ca–P seem to sugges<sup>t</sup> a layering within the weathering crust, the lower Ca signal intensity indicating higher rates of Pb2+ substitution in the inner layers and hence a higher presence of pyromorphite. The external layers, which feature the highest Ca–P and P–Pb correlations, sugges<sup>t</sup> early stages of PbCaHA formation. These sliver coatings can be further considered rather stable under the soil pH conditions. Indeed, the pH value of 7.4 measured in the polluted soil (Table 1) falls well within the pH range 3–11 in which chloropyromorphite has been found to be the most stable Pb species [36].

It could be argued that neutral to sub-alkaline pHs are ye<sup>t</sup> inhibitory for pyromorphite formation due to the reduced solubility of P, which lowers both the kinetics and rate of Pb stabilization [40]. Conversely, more acidic conditions and hence soluble orthophosphate ions are required to react with Pb and precipitate pyromorphite [41]. However, a high amount of available P was measured in the soil (Table 1), and soil micro-XRF mapping also confirmed a high occurrence of P, either correlated with Ca or not: such P could have contributed to pyromorphite formation and, consequently, to Pb stabilization. Among the most common P forms in soil, apatite and phosphate rock are quite insoluble at neutral pHs. By contrast, P fertilizers, such as monocalcium and dicalcium phosphate (Ca (H2PO4)2 and CaHPO4, respectively), decrease soil pH upon dissolution, increasing the solubility of both P and Pb [40]. Based on these observations, it can be hypothesized that in the studied soil, which was used for agriculture after being dismissed as a shooting range, stabilization processes were triggered by soil phosphatic fertilization. The temporary soil acidification would have favoured pyromorphite precipitation at the fragment surfaces, thus protecting metallic Pb from further weathering processes and, at the same time, strongly limiting the possible mobilization of this hazardous element. Indeed, despite in alkaline and subalkaline conditions, the most commonly weathering phases of Pb bullet residues found in shooting ranges are lead carbonates and oxides (e.g., cerussite and litharge) [40], other previous studies showed the natural formation of pyromorphite in Pb-rich soils with high concentrations of available phosphates [42,43].

#### *3.3. Bioassays with E. andrei*

Considering the similarity of the physicochemical characteristics of S1 and S2 soils (Table 1), it may be hypothesized that the earthworm testing results could be only minimally influenced by the different soil properties (mainly the OM and carbonate content). In addition, the pH values of S1 and S2 (7.8 and 7.4, respectively) allow us to exclude pHshock effects for earthworms, since they prefer circumneutral soils, even if *E. andrei* could also tolerate acidic conditions [44].

All earthworms survived after the 28-day exposure period, evidencing no acute toxicity effects, as previously experienced by other authors for similar background concentrations [5]. Moreover, the average weight of the earthworms increased after 28 days to a similar extent in both S1 and S2 mesocosms (*t*-test, *p* < 0.05). On the other hand, earthworms introduced into the polluted mesocosm accumulated significantly higher amounts of Pb in tissues and coelom fluid than the control (*p* < 0.01). Morgan and Morgan [45] found a positive correlation between Pb tissues and soil concentrations, different to other potentially toxic metals, such as Cu, Zn and Cd, whose tissue concentrations decreased proportionally as their soil concentrations increased. However, differently from these elements, Pb concentrations in earthworm tissues were generally below the soil Pb concentration, due to the higher retention of Pb by organic and inorganic soil components [46]. Even if the Pb tissue concentration of earthworms grown in S2 exceeds by ten times that of animals grown in the control soil, the bioconcentration factor (BF) obtained for S2 is very low (Table 2). Indeed, in a comprehensive study on Pb uptake in earthworms exposed to Pb-polluted soil, Langdon et al. reported a Pb tissue concentration of 266, 406 and 637 mg kg−<sup>1</sup> (with corresponding BFs of 0.27, 0.14 and 0.16) for *E. andrei* exposed to 1000, 3000 and 4000 mg Pb kg−1, respectively [5]. Based on these findings, an average uptake of approximately 300 mg Pb kg−<sup>1</sup> in earthworm tissues exposed to S2 (1575 mg Pb kg−1) should be expected, but it was only 95 mg kg−<sup>1</sup> (Table 2). However, in the OECD-style experiments carried out by Langdon et al., Pb mesocosm contamination was obtained by artificially spiking the testing soils with aqueous solutions of Pb(NO3)2: this would imply higher Pb availability (and hence possibly higher Pb accumulation rates) than in field-collected soils, in which Pb can be already at least partly stabilized through complexation or adsorption to soil components (therefore less bioaccesible). Nonetheless, even by applying the regression equation proposed by Morgan and Morgan (e.g., log10Pbworm = −1.073 + 1.042 log10Pbsoil), who instead worked with field-collected soils [45], the Pb uptake should be at least two times higher than our

experimental data. The value found for the control soil (11 mg kg−1) is instead consistent with Langdon's results and explainable as the normal trace amount of Pb in the organisms.

**Table 2.** Results of bioassays with *E. andrei* in the unpolluted (S1) and polluted (S2) soils. Ecotoxicology data (mean value ± standard deviation, n = 5) with different letters in the column are statistically different according to *t*-testing (*p* < 0.05).


ns: Not significant.

Total PTE concentration in coelom fluid has been proposed in previous studies as a bioavailability indicator [30,31,47]. Differently from the determination of total concentration in tissues, PTE quantification in fluids is not affected by possible soil residues entrapped in the gu<sup>t</sup> [48]; furthermore, it is a less laborious and time-consuming procedure, and it is unnecessary to kill the animals. Looking at our results (Table 2), the increase of Pb in coelom fluid would sugges<sup>t</sup> some metal sequestration mechanisms occurring within the chloragogenous tissue, which surrounds the earthworm gu<sup>t</sup> epithelium, by means of metal-binding metal-inducible cysteine-rich metalloproteins, namely, metallothioneins (MTs) [49]. Albeit Pb is not recognized as a MT-inducer [50], it was detected together with Cd and Zn in the chloragogenous tissue of earthworms collected from an abandoned Zn–Pb mine in Draethen (UK), by means of proton-induced X-ray emission analysis (PIXE) [51]. In such work, Morgan et al. evidenced the primary role of Cd (and much less of Zn) in inducing MT expression in earthworms living in heavily polluted soil. This would match the possible scenario of the firing soil studied in this work and could probably explain the Pb accumulation in coelom fluids. Indeed, besides Pb, high levels of Cd, Sb and Zn (which are elements related to the past firing activity) have been detected in S2 soil (Table 1) and could have worked as MT inducers, thus favouring Pb binding, also.

Overall, the results of earthworm tests seem to depict a very limited bioavailability for soil Pb. On the one hand, neither mortality nor weight decrease was observed; rather earthworms grew over the 28-day experimental period. On the other hand, Pb bioconcentration in tissues was lower than reported in the literature for similar soil Pb concentrations, whilst a relevant amount of Pb in the coelom fluids suggested a metal-induced metal-sequestration mechanism. Such findings appear in accordance with the results of the soil characterization and microanalyses, which showed a negligible NH4NO3-extractable Pb fraction (Table 1) and predominant Pb stabilization by phosphates. Indeed, significant correlations between promptly bioavailable metal fractions in soils (including Pb) and their bioconcentration in earthworms have been indicated in previous works [52]. In addition, the high amount of OM in S2 (Table 1) may have contributed to limiting Pb bioavailability. At the same time, the formation of pyromorphite makes Pb inaccessible in the gastrointestinal tract, since these phases remain insoluble even after ingestion [2]. Formation of pyromorphite may also occur inside the intestine in the presence of phosphate, thanks to gastrointestinal acidic conditions [40]. These latter considerations would further confirm a low Pb accessibility in the studied case on the base of P availability.
