The study sites are located in the western side of South Korea (36°00′48″ N, 126°39′56″ E) (Fig. 1A), and the privately-owned paddy-growing field nearby the metal smelter had been extensively contaminated with heavy metals (arsenic, cadmium, and lead) [31]. Korean Government projects were implemented to remediate metal-contaminated paddy soils, and the acid washing (0.5 M H₂SO₄ and 0.5 M H₃PO₄) technique was applied to remove metal contaminants in the privately-owned paddy field soils. In the present study, the metal contaminated and remediated paddy soils were collected, and the in situ test plots were constructed in an area of 5 m (width) × 15 m (length) × 1 (height) m (Fig. 1B). To construct the reclaimed paddy soil plot, a 30-cm layer of clean soil was used to cover the remediated soil of depth 70-cm (total height, 1 m). This layer-covering is the Korean Government-recommended technique to restore privately-owned paddy fields. This reclaimed soil represented the privately-owned agricultural fields since the remediated soils have reconveyed to each site and used for the crop cultivation. The clean soil was obtained from a different area (36°03′17″ N, 126°40′54″ E), which is 4.97 km distant from study area. The soils were fertilized with 4.5 kg of nitrogenous manure, 4.5 kg of phosphorus manure, and 6.1 kg of potassium manure per hectare, and irrigated using stream water.

The history and location of the upland soils were the same as that of the paddy soils. The upland soils had been contaminated with heavy metals and remediated by soil washing using 0.1 N NaOH and Na₂CO₃ (alkali reduction or oxidation-reduction). The upland soils were classified as contaminated, remediated, and reclaimed soil plots. The reclaimed upland soil plot was prepared by covering a 30-cm layer of clean soil over the remediated soil of a 70-cm depth (total height, 1 m). The in situ test plots were constructed with same design of paddy plots (Fig. 1C). Each plot was treated with 1.2 kg/m² of fertilizer (Saengjiwang, Yeongmin Tech., Hwoaseong, South Korea) consisting of 30% peat, 20% vermiculite, 20% biotite, 10% zeolite, 5% treacle, and 15% other substances (e.g., amino acids).

Approximate 2 kg of surface soil samples (0–10 cm) were collected from each plot at the end of crop cultivation, and each soil sample (<2 mm) was air-dried (24°C) for two weeks. Soil physical, chemical, biological, and ecotoxicological indicators were classified according to previous study [30]. Thirteen soil physical, chemical, and biological indicators were chosen, and the five ecotoxicological indicators were selected (Table 1). The aggregate stability (AS), available phosphate (AvP), available silicate (AvS), bulk density (BD), cation exchange capacity (CEC), enzyme activity (EA), electrical conductivity (EC), exchangeable cation (ExC) including Ca, K, and Mg, organic matter (OM), pH, total nitrogen (TN), total phosphorus (TP), and water holding capacity (WHC) have been commonly used in previous studies [3, 11, 13, 32, 33]. The catalase (CAT), urease (UA), and β-glucosidase (βGA) activities were selected as main indicator EAs for soil quality assessment [34, 35].

Earthworm survival (E), nematode reproduction (NR), plant germination and growth (PG), soil algal biomass (SAB), and soil algal photosynthetic capacity (SAP) were measured using standard ecotoxicity bioassay as follows. The earthworm survival (indicator E) was measured according to standard test guidelines [36, 37], and NR bioassay was conducted according to the ISO [38] and Kim et al. [39, 40]. The indicator PG included barley (B, Hordeum vulgare), rice (R, Oryza sativa), sorghum (S, Sorghum biocolor), and wheat (W, Triticum aestivum) [31, 41], and two soil algal species (Chlorococcum infusionum, Ci; Chlamydomonas reinhardtii, Cr) were used to measure the SAB and SAP [42]. To normalize the ecotoxicological indicators, LUFA 2.2, which is widely used as control in soil ecotoxicity studies [43], was selected as reference soils.

The soil quality assessment consisted of indicator classification, measurement, scoring, weighting, and SQI calculation [30]. We used the indicator measurement data of each test plot, and indicator scoring was performed using the following assumptions: “more is better,” “less is better,” and “optimum is the best.” Each type included the standard scoring function (SSF) equations [3, 11, 30], and the SSF equations for “more is better” (Eq. (1)), “less is better” (Eq. (2)), and “optimum is the best” (Eq. (3)) can be described as:

The f(x), x, L, and U indicate the scores (ranged from 0.1 to 1.0), measurement data, and lower and upper threshold values of each indicator, respectively. The lower and upper thresholds were determined according to previous studies (Table 2). According to Kim et al. [30], the changeable thresholds can be applied to the specific indicators including AvP, Bd, EA, ExC (K), and ExC (Ca) (Table 3). These indicators have various threshold values, because the optimal values of AvP, ExC (K), and ExC (Ca) could differ with crop type [32], and the BD indicator has a correlation with soil texture [33, 44]. The EA indicators, including CAT, UA, and βGA, were normalized with the corresponding values in the control soils, such as the international standard soils and non-contaminated field soils [34, 35].

The principal component analysis (PCA) was employed for indicator weighting. Only the principal components (PCs) of eigenvalues ≥ 1 were considered for weighting, and each eigenvector within the available PC was squared and summed [3, 30, 45, 46]. The communality was calculated, and the ratio to total communality was determined as weights [3, 30]. By the end of the process, we used the integrated quality index (IQI) equation approach [2, 3, 6, 13, 47], and the SQI was calculated to quantify the soil class at each plot:

where Wi, Si, and n are the weights of each indicator, indicator scores, and number of indicators, respectively. The SQI can classify soils into five classes: very good (> 0.85), good (0.70–0.85), moderate (0.55–0.69), bad (0.40–0.55), and very bad (< 0.40) [13, 30, 48]. To evaluate whether the ecotoxicological indicators affected the SQI value, the soil quality assessment was performed with or without ecotoxicological indicators.

Crop productivity data was generated by in situ cultivation under actual field conditions. Rice plants (Oryza sativa) were grown for 108 days on paddy soil plots. Before planting, the rice seeds were sterilized for 1 day using a germicide, washed for 2 days, and germinated at 30–32 °C. They were transplanted into the paddy soil plots, watered, and the weight of 100-grains was measured after 108 days. The detailed description of in situ cultivation on paddy soil plots can be found in the study of Kim et al. [31]. On upland soil plots, perilla (Perilla frutescens) and soybean (Glycine max) were used, and nursery crops were planted with 80 replications for each upland soil plot. After 120 days, the grains of perilla and soybean were threshed, and the weight of total grains (for perilla) and single crop-grains (for soybean) was determined. The in situ crop productivity of each plot was compared with the soil classes (very bad, bad, moderate, good, and very good) estimated by SQI with or without ecotoxicological indicators.

The indicators in reference and paddy soils (contaminated, remediated, and reclaimed) were measured as shown in Table S1 (Supporting Information). The data were scored using SSF equations (Table 2 and 3), and they were listed as shown in Table S2 (Supporting Information). As the soil textures of reference and paddy soils (contaminated, remediated, and reclaimed) were determined as sand, clay, clay, and silty clay loam respectively, the different SSF thresholds were applied for BD indicator scoring. In addition, we chose “rice” as the target crop, and the SSF thresholds of AvP and ExC (K) were selected as 120 mg/kg and 0.3 cmol⁺/kg for rice group, respectively. ExC (Ca) was also scored using SSF threshold (6 cmol⁺/kg) of chilly, Chinese cabbage, apple, Asian pear, rice, sesame, sweat potato group. PCA was conducted using scores of each indicator, and the three available PCs (eigenvalue ≥ 1) were determined (Table S3, Supporting Information), and the communality and weights were calculated as shown in Table S4 (Supporting Information). The highest weights with or without ecotoxicological indicators were determined as PG (S) and pH, respectively.

The SQIs of each paddy soil were exhibited as shown in Table S5 (Supporting Information). The SQI with ecotoxicological indicators were calculated as 0.770, 0.714, 0.716, and 0.589 on reference, contaminated, remediated, and reclaimed soils, respectively, and these values indicated “good,” “good,” “good,” and “moderate” soil classes (Fig. 2A). We run a soil quality assessment system without ecotoxicological indicators (Fig. 2B). The SQI showed 0.678 (moderate), 0.673 (moderate), 0.735 (good), and 0.520 (bad) on reference, contaminated, remediated, and reclaimed soils, respectively. The reference and contaminated soils descended to “moderate” from “good” of soil class, the reclaimed soil decreased to “bad” from “moderate” of soil class, and the remediated soil showed no changes.

The indicator measurement data of upland soils were scored using SSF function as shown in Table S6 (Supporting Information). The soil textures of reference, contaminated, remediated, and reclaimed soils were determined as sand, sandy loam, loamy sand, and sandy loam respectively, and the different SSF thresholds were used in BD indicator scoring. As “bean” was chosen as one of the target crops, the SSF thresholds of AvP, ExC (K), and ExC (Ca) were selected as 250 mg/kg, 0.55 cmol⁺/kg and 7 cmol⁺/kg, respectively (Table 3). After the PCA, we obtained the three available PCs (eigenvalue ≥ 1) (Table S7, Supporting Information), and the highest weights with or without ecotoxicological indicators were determined as EA (CAT) (Table S8, Supporting Information).

The SQI were calculated as 0.743, 0.598, 0.634, and 0.408 on reference, contaminated, remediated, and reclaimed soils, respectively, and these values indicated good, moderate, moderate, and very bad soil class (Table S9, Supporting Information). With the soil quality assessment system without ecotoxicological indicators, the SQI was calculated as 0.579 (moderate), 0.528 (bad), 0.572 (moderate), and 0.539 (bad) in reference, contaminated, remediated, and reclaimed soils, respectively. The soil class of the contaminated soil became “bad,” and SQI of the reclaimed soil increased to 0.528 from 0.408 (Fig. 2C and D).

The in situ crop productivity of paddy soil plots were intimately reported in our previous study [31]. The weights of 100 rice grains were determined as 2.8 ± 0.3, 2.7 ± 0.2, and 2.9 ± 0.2 g on contaminated, remediated, and reclaimed plots, respectively. The dot data represented the data distribution of SQI versus in situ crop productivity (grains weight) on paddy soil plots (Fig. 3A). The dot data of each paddy soil plot showed dense distribution, because the SQI and in situ crop productivity were similar. In order to evaluate whether the ecotoxicological indicators influenced SQI calculation, the SQI without ecotoxicological indicators was calculated and compared with in situ crop productivity. The dot data of the reclaimed paddy soil plot were SQI sifted to the left side of the x-axis (red arrow in Fig. 3B).

The weight of total perilla-grains was determined as 1,111 and 1,742 g at the contaminated and remediated upland soil plots, respectively, and the weight of single soybean-grains was determined as 9.4 ± 6.1 and 33.9 ± 18.8 g at the contaminated and remediated upland soil plots. As the crops at the reclaimed upland soil plot were all withered since planting, the in situ data could not be obtained (Fig. S1). This can be attributed to the high soil EC level at the reclaimed upland soil plot (Table S1, Supplementary material). The SQIs of each upland plot were compared with in situ crop productivity (grains weight), and the dot data showed good linear shape (Fig. 4A and B). However, in the SQI without ecotoxicological indicators, the SQI of the contaminated and reclaimed upland soils was SQI sifted to the left (“bad” area) and right (“moderate” area) of the x-axis, respectively (red arrows in Fig. 4C and D). The SQIs of each upland soils showed a good correlation with in situ crop productivity when the ecotoxicological indicators are used. These results imply that ecotoxicological indicators play an important role in representing productivity function of the upland soil plots since they directly response to low functionality of soils for their survival. Although EC of the reclaimed upland soil showed higher level compared to other soils (Supporting Table S1), but SQI without the ecotoxicological indicators was similar levels with other upland soils (Fig. 2). However, the ecotoxicological indicators, especially P (plants) and E (earthworm), directly response to the amended soil (Supporting Table S1), and SQI showed good correlations with in situ crop productivity (Fig. 4).

The remediation techniques can disrupt physicochemical properties of soil and can be fatal to soil organisms because it changes the balance of the soil ecosystem [17–19]. These disruptions can affect the measurement of traditional biological indicators such as microbial C/N, plant cover, root quality rating, and earthworm/nematode abundances [9–13]. The biological monitoring methods such as QBS, EMI, TRIAD, which evaluate the biodiversity by actual field survey, were also suggested in previous studies [49–51], and these approaches enable to correctly decide the degree of soil organism’s vulnerability and biological activity. However, the biological monitoring methods could not evaluate severely contaminated or disrupted soils where life of soil organisms are not allowed [17–19]. Currently, for in situ assessment using laboratorial animals, the previous studies employed earthworm and soil algal species into field situation, and they found some possibility as rapid tool for soil quality evaluation on the contaminated or remediated sites [52–54]. The ecotoxicological approaches also investigated using field soil samples on laboratorial condition, and the bioactivities of earthworm, nematode, and soil algae were evaluated through the modified standard bioassay methods [31, 36, 40]. Kim et al. [30] suggested the soil quality assessment system using ecotoxicological indicators on contaminated and remediated sites, and they reported that the use of ecotoxicological indicator can serve the soil quality assessment even in contaminated or physically and chemically treated soil. We also speculated that the ecotoxicological indicators can show more sensitive response to the damaged soils compared with the other indicators, because the ecotoxicological indicators can provide direct evidence of biological response.