Introduction

Soil impacted by mining restricts the land’s agronomic use, as it is prone to soil erosion and toxic contamination (Timsina et al. 2022), contains low levels of soil organic matter (SOM), and as a consequence has low nutrient concentrations. Moreover, biodiversity loss, extreme soil pH, high bulk density, and lack of soil structure also pose a problem in mined soils (Sheoran et al. 2010; Asmarhansyah 2016). Therefore, reclamation to regenerate the health of soils after mining is a major challenge, in particular in the tropics where high rates of weathering and decomposition make the process become more complex (White and Blum 1995; Rafdinal et al. 2021).

Recently, it has been recognized that SOM, inherently high in carbon (C), is the basis for a “healthy soil”, and that increasing soil SOM worldwide is essential to combat climate change, land degradation and food insecurity and subsequently safeguarding human health (Lal 2011). SOM fosters both plant growth and soil microbial activity primarily through the provisioning of nutrients and organic C. Previous studies revealed increases in SOM improved productivity of highly degraded soil after mining (Sinha et al. 2009; Asmarhansyah 2016).

Several management options have been proposed and applied to promote soil organic C (SOC) and consequently improve soil health of highly degraded lands, such as the use of soil amendments, revegetation, conservation agriculture, and mechanical measures (Sheoran et al. 2010; Bhattacharyya et al. 2015; Timsina et al. 2022). Among those options, the application of soil amendments was considered as a more feasible strategy for cultivation purposes due to the diversity of materials and their availability on all continents (Garbowski et al. 2023). Organic soil amendments, which are usually C rich materials, evidently not only improve SOC and crop productivity but also have the potential to enhance other physico-chemical and biological properties (Kimetu et al. 2008; Sukartono et al. 2022; Garbowski et al. 2023).

Among the organic soil amendments, compost has demonstrated its benefits over and above the other potential sources, due to its ability to supply nutrients, promote soil microbial activity, and facilitate nutrient mineralization (Garbowski et al. 2023). Conventional knowledge suggests that unlike compost, addition of the wide carbon to nitrogen (C:N) ratios (> 100) of other amendments, such as biochar and lignin or cellulose rich plant residues (e.g., sawdust), potentially results in nutrient immobilization (Garbowski et al. 2023). However, negatively charged surfaces of organic soil amendments, in particular compost and biochar, are known to enhance nutrient retention (Agegnehu et al. 2017; Garbowski et al. 2023) especially in tropical soils. Compared with other organic soil amendments, the higher porosity and greater specific surface area of biochar could lead to higher water holding capacity, available water content, niches for soil microbes, and chelation (El-Naggar et al. 2019; Leng et al. 2009; Briedis et al. 2023).

There are few studies on the effects of soil amendments on the improvement of soil health in tropical highly degraded soils, particularly those in post-mining areas, although they are necessary to avoid risks and to reclaim post mining sites efficiently and effectively. In Indonesia, tin mining significantly contributes to the national economy but is a major cause of severe soil degradation (Maftukhah et al. 2022, 2023). Consequently, affordable and practical land reclamation measures, such as applying local soil amendments, have been planned and developed to restore soil health, combat food insecurity, and generate income, particularly for local communities, as tin mining on land becomes less profitable. Therefore, the current study aims to assess the effects of various soil amendments on soil health improvement. It examines the quantity and chemical composition of SOC, serving as indicators of soil health, and soil productivity during the pioneer stages of soil reclamation in a post-tin mining area in Indonesia.

Materials and methods

Study site

The study site was a mine tailing site of a post-tin mining area located on Bangka Island in Sungailiat district, Bangka Regency, Kepulauan Bangka Belitung Province, Indonesia (1° 47’ 22.9085 S and 106° 5’ 47.0461 E). The climate is humid tropical with a mean annual precipitation of 2199 mm and mean annual temperature of 27 °C during the study period (Suppl. Fig. 1). The native soil, developed from granite parent material and with a high sand fraction, is classified as Dystric Cambisol (IUSS Working Group WRB 2022). However, active terrestrial tin-mining over the past two centuries in the area has caused drastic landform changes, including severe soil degradation and tendency for soil erosion at the study site. The initial soil is thus classified as Technosol (IUSS Working Group WRB 2022) and exhibits restricted plant growth. The soil sand fraction is more than 70%, with pH < 6, low organic matter (approximately 1%), soil nutrients, and low plant available water as well as low water holding capacity (Suppl. Table 1).

Experimental design and field management

A field experiment was established in 2018—the fourth year after tin mining activities ceased. Six soil amendment treatments comprising of: (1) control (no soil amendment application), (2) dolomite, (3) compost, (4) charcoal, (5) charcoal with compost (charcoal + compost), and (6) charcoal with sawdust (charcoal + sawdust) were arranged in a randomized complete block design with 4 replicates (blocks) (Suppl. Fig. 2). Each experimental plot (experimental unit) was 2 m x 2 m with 0.75 m width border (ridge) between the plots to avoid cross contamination by runoff. The soil amendments used in this study were chosen as they were domestically available. Their chemical properties are documented in Suppl. Table 2.

The soil amendments were applied manually in July 2018 with respect to the experimental design by broadcasting and then hoeing them down to 20 cm depth. The application rates were 10 t ha−1 for dolomite, charcoal, and compost treatments, and 20 t ha−1 with ratio of 1:1 (w/w) for charcoal + compost and charcoal + sawdust treatments. All soil amendments except charcoal were reapplied to their respective treatments in July 2021 at the rate of 10 t ha−1 to maintain crop productivity. Details are given in Suppl. Table 3.

During the study period, edible and forage crops were cultivated in an intercrop** system (Suppl. Fig. 1). Crop species were selected according to their potential benefits and by co-creation with local farmers (Suppl. Table 4). Cassava (Manihot esculenta Crantz) was planted as the main crop with 1 m x 1 m spacing (4 plants per experimental plot). Its tubers, stems and leaves were harvested at the end of the growing season in July. Along with the main crop, early (August–December/January) and late (January-July) season intercrop** plant seeds were sown in a 25 cm x 25 cm grid. A leguminous species was primarily selected as the inter-crop to increase nitrogen (N) content via N fixation, and to prevent soil erosion in accordance with local recommendations (Maftukhah et al. 2022). However, elephant grass (Pennisetum purpureum) was naturally regrown after its first harvest in July 2020, due to its low-maintenance complying with hygienic restrictions during the COVID-19 pandemic. Accordingly, monoculture elephant grass was cultivated in 2020/2021 with spacing 25 cm x 25 cm and harvested in July 2021.

The plot maintenance followed organic regenerative agricultural practices so that none of the machinery was used on the plots, and neither irrigation nor pest control was conducted during the study period. The soil received N inputs via the leguminous species and intrinsic N contained in soil amendment materials, as detailed in Suppl. Table 3, without an explicit fertilization program. Aboveground crop residues and weeds were removed after each harvesting to prepare the seedbed for the subsequent crops grown.

Soil sampling and soil property analysis

Soil sampling was conducted annually, taking place immediately after harvesting the main crop in July/August from 2019 to 2022. In every experimental plot, a composite soil sample at 0–20 cm depth was obtained by combining sampled soils from three random sampling points on the plot. The composite samples were then air-dried and sieved < 2 mm prior to soil physico-chemical property analysis. The dry-sieved soil samples were homogenized by ball milling to obtain finely ground soil powder for measuring total C and total N concentrations, C isotope ratio, and Fourier transform infrared (FTIR) spectra. Undisturbed soil samples were also collected using a coring method.

Total C and total N concentrations together with isotopic C composition were measured by using an elemental analyzer isotope ratio mass spectrometer (IRMS; Delta PLUS, Thermo Finnigan, Bremen, Germany), connected via a ConFlo III interface (Thermo Fisher, Bremen, Germany) to a Thermo Delta V (Bremen, Germany). More details can be found in the Supplemental materials and methods. In addition, inorganic C concentration in dolomite amended soil was determined following ÖNORM L1084 (Austrian Standards 1989). SOC concentration was obtained by subtracting inorganic C concentration from the total C concentration. SOC concentration was then used to calculate SOC stock in Mg ha−1 by multiplying soil mass within the 0–20 cm depth. SOC stock balance of every treatment in each year was also computed separately from the difference between its SOC stock and that in control in the respective year. Net SOC stock balance of each treatment was then determined by the sum of its SOC stock balance during the study period.

Dissolved organic carbon (DOC) was extracted from 5 g dry soil in 50 ml ultrapure water (resistivity 18.2 MΩ⋅cm at 25 °C; total organic C ≤ 2 ppb). The 0.45 μm-filtrated solutions were measured in specific UV plates with spectrophotometric plate reader (Enspire Multimode 2300, Perkin Elmer Inc., USA) at 254 nm wavelength. DOC concentration was calculated from the absorbance according to Brandstetter et al. (1996). DOC concentration was reported in mg L−1 and its contribution to SOC was computed and expressed as DOC: SOC ratio in percent (%). Other soil physico-chemical properties e.g., pH, electrical conductivity (EC) and bulk density were analyzed by standard methods, with more details provided in the Supplemental materials and methods.

Fourier transform infrared (FTIR) spectroscopic measurement

FTIR spectroscopy is a technique to determine functional groups as well as to obtain a fingerprint of a sample (Parikh et al. 2014). In the current study, attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy was used to determine SOC chemical composition as a non-destructive and high throughput method. The finely ground soil samples were sieved through 63 μm prior to the measurement to increase homogeneity and SOC proportion in the studied soil as the sand fraction was > 70%. The samples were then measured at mid-infrared (MIR) wave region at wavenumber ranging 4000–400 cm−1 with an optical diamond crystal (Tensor 27 SN 1683; Bruker ®). Each sample was scanned 5 times (measurement replicates) at rate 32 scans per sample and resolution of 4 cm−1 against ambient air as a background.

Plant sampling and plant biomass determination

Fresh weights of the plant parts were recorded separately for the harvested inter-crops and main crops in every experimental plot during 2019–2022. These were stem, leaf, tuber and grain depending on the crop species (for details see Suppl. Table 5). Dry weights were recorded after drying at 65 °C for 48 h and moisture content (dry weight basis) was calculated. Biomass (dry weight) of each part was then estimated by multiplying its fresh weight by its moisture content with respect to plant species, soil amendment treatments and harvesting time. Annual plant biomass were computed by summing up biomass of every plant part obtained from the inter-crops and the main crop harvested during the study period and expressed in t ha−1.

Data analysis

Two-way repeated measures analysis of variances (ANOVA) was performed to determine significant effects of soil amendments and time on SOC characteristics comprising SOC stock, DOC concentration, DOC: SOC ratio, and C isotope ratio followed by post hoc pairwise t-test (p < 0.05, detailed in the Supplemental materials and methods). A linear mixed model (LMM) was fitted to determine effects of soil amendments and time as well as their interaction on annual biomass production. Estimated marginal means from the obtained LMM were separated by post hoc Tukey test (p < 0.05). Furthermore, one-way ANOVA was performed to determine significant effects of soil amendment treatments on the annual biomass production within the same year (each time point) followed by pairwise t-tests of multiple comparison (p < 0.05). More details on the LMM and ANOVA analyses performed on annual plant biomass are given in the Supplemental materials and methods.

For FTIR spectral data, the obtained spectra were preprocessed in Spectragypth v1.2.16.1 software (Menges 2023). After removing outlier spectra, the spectra were normalized based on the maximum-minimum method and baseline corrected to eliminate measurement artifacts and further averaged prior to data evaluation. According to SOC focus in the current study, the organic regions were selected for data evaluation as follows: (I) 3400–2600, (II) 1750–1550, and (III) 1310–1230 cm−1 (Tatzber et al. 2011; Le Guillou et al. 2015; Olaleye et al. 2020; Briedis et al. 2023). Their potential functional groups in mineral soils are given in Table 1. Principal component analysis (PCA) was performed to reduce variable dimensions of FTIR spectra. The top two bands loaded on PC1 and PC2 were selected to simply explain the dominant chemical composition of SOC.

Table 1 Assignments for selected mid-infrared bands related to potential organic composition in mineral soils
Full size table

Partial least squared (PLS) regression analyses were conducted to determine relationships among available physico-chemical soil properties and annual plant biomass as well as climatic conditions. The PLS regression model performance was evaluated by the leave-one-out cross-validation method and the coefficient of determination (R2). Stepwise variables selection based on variable importance on projection (VIP) scores, and model modification was conducted to minimize root mean squared error of prediction (RMSEP). The predictors with VIP > 1 were considered to be able to strongly influence the response variables (Andersen and Bro 2010). The explanatory power of the PLS regression models was reported by the sum of explanatory power of the components for X. The relationships among variables were examined by the loadings on the biplot: variables that appear close to each other have positive correlations while variables with negative correlations are distant to each other (Samad and Bertilsson 2017). More details of PLS regression analyses are given in the Supplemental materials and methods.

All data and statistical analyses were conducted in R environment (R version 4.1.3 and RStudio 2023.09.0), unless mentioned otherwise. The following R packages were used: lme4 (Bates et al. 2015) and lmerTest for LMM analysis (Kuznetsova et al. 2017), rstatix for ANOVA and multiple comparison (Kassambara 2023), stats for PCA (R Core Team 2022), and pls for PLS regression analysis (Liland et al. 2023).

Results

Soil organic carbon (SOC) characteristics under different soil amendment treatments

SOC stocks

Soil amendments, the time since imposition of treatment (from herein referred to as time), as well as their interaction, significantly changed SOC stocks (p < 0.05) in the post-tin mining soil (Fig. 1a). The organic soil amendments (compost, charcoal, charcoal + compost and charcoal + sawdust) significantly increased SOC stocks (p < 0.05). Soils amended with compost and charcoal + compost contained higher SOC stocks than any other treatments over the study period. In the charcoal and charcoal + compost treatments SOC stocks decreased significantly (p < 0.05) on average by 60% from the second to the third year. In contrast, soils amended with dolomite had relatively low but stable SOC stocks across the 4 years even in the year 2022 after dolomite was reapplied. Unlike control, SOC stocks in all amended treatments were > 17% higher on average than the initial value in the first year of the study. SOC stocks in the control increased over time but exceeded the initial SOC stock from the third year onwards.

Fig. 1
figure 1

a Soil organic carbon (SOC) stock in 0–20 cm depth soil and b SOC balance compared with control in each respective year (ΔSOC) under different soil amendment treatments. Presented bars and error bars are means and SE (n = 4), respectively. Different upper- and lowercase letters, and * indicate significant differences determined by two-way repeated measures ANOVA followed by post hoc test at p < 0.05. Uppercase letters: significant differences among treatments; lowercase letters: significant differences among time points at each treatment; and *: significant differences between treatments and control at each time point. Dashed line presents average value before amendment

Full size image

The SOC stock balances (gain vs. loss) in Fig. 1b shows the highest SOC gain (> 25 Mg ha−1) in the second year in soil amended with organic materials compared to the control. Specifically, the results showed that compost application led to an SOC gain over time. Applying charcoal lessened the magnitude of SOC gain during the third and the fourth years compared to pure compost. However, applying charcoal and charcoal + compost resulted in higher net SOC gains, which were > 50 Mg ha−1, during the first three years after amendment (2019–2021; Suppl. Table 6). After reapplication of soil amendments (except charcoal) in 2021, net SOC gains were observed in the compost and the charcoal + compost treatments and net SOC loss was observed in the dolomite treatment.

Dissolved organic carbon (DOC)

The quantity of DOC, an SOC pool easily available to microbes, differed significantly among soil amendment treatments and across the duration of the experiment (p < 0.01; Fig. 2a). The highest DOC concentration over time was found under charcoal + compost treatment while the lowest was recorded in the control. DOC concentration in all treatments tended to decrease over the study period. DOC concentrations in the first and the third year were comparable and significantly higher than in the second and the fourth year (p < 0.05). Unlike other treatments, amending soil with compost resulted in a more gradual decrease in DOC concentration over time. Similar to DOC concentrations, soil amendments and time as well as their interaction significantly affected DOC:SOC ratio (p < 0.05, Fig. 2b). The highest ratio over time was found in the dolomite treatment followed by that in the control.

Fig. 2
figure 2

a Dissolved organic carbon (DOC) concentration and b DOC:SOC ratio in soil under different soil amendment treatments. Presented bars and error bars are means and SE (n = 4), respectively. Different upper- and lowercase letters, and * indicate significant differences determined by two-way repeated measures ANOVA followed by post hoc test at p < 0.05. Uppercase letters: significant differences among treatments; lowercase letters: significant differences among time points at each treatment; and *: significant differences between treatments and control at each time point. Dashed line presents average value before amendment

Full size image

Fourier transform infrared (FTIR) spectra

FTIR spectra from 96 soil samples were analyzed by using PCA. Two principal components (PCs) were obtained and explained 77% of the total variance of the organic composition in the studied soil (Fig. 3a). The primary PC (PC1) contributed 58.8% and the secondary PC (PC2) 28.2% to the total variance explained. PC1 was heavily loaded with wavenumber centered at 1634 and 2918 cm−1 (band III and I in Fig. 3b, respectively). According to Table 1, the band I (wavenumber 2920–2850 cm−1) refers to functional group C–H of CH2 in long chain hydrocarbons and the band III (wavenumber 1650–1600 cm−1) C=O in protein, C=C aromatic in lignin, and C–O in COO−. PC2 was chiefly characterized by band IV (wavenumber 1280–1200 cm−1) and II (centered at 1700 cm−1). The band IV indicated C–N of amide III and C–O single while the band II indicated C=O of esters.

Fig. 3
figure 3

a Score plot and b loadings of two principal components (PC1 and PC2) from the principal components analysis (PCA) of Fourier transform infrared (FTIR) spectra of soil under different amendment treatments across selected soil organic matter (SOM)-related spectral range. Wavenumbers with the top two loadings on PC1 and PC2 are indicated by band number I to IV

Full size image

The PCA results (Fig. 3a) show a clear separation between dolomite and other treatments regardless of time and its wide distribution along PC2. PC1 segregated soils in the second and the third years. In the second year, a tight cluster was observed from the charcoal treatment while other organic amendment treatments loosely grouped along PC1. Charcoal, charcoal + compost and charcoal + sawdust treatments in the third year clustered closely together. Instead, the control and the soils amended with organic materials in the first and the fourth years distributed around the origin.

Annual plant biomass

During the 4 years of study, annual plant biomass significantly differed among treatments (p < 0.001) not including time effect (Fig. 4). Charcoal + compost amended treatments produced significantly higher annual plant biomass than other treatments. As expected, the lowest amounts of annual plant biomass were recorded in the control. Compared to the control, soil amended with organic materials significantly improved the annual plant biomass only in the first year (p < 0.05), particularly the charcoal + compost treatment. Charcoal + compost treatment increased the yield five-fold in the first year and on average two-fold throughout the study period compared to the control. Applying dolomite or charcoal as a single treatment did not show any significant increase (p ≥ 0.05) in the annual plant biomass during the study period.

Fig. 4
figure 4

Annual plant biomass under different soil amendment treatments. Presented bars and error bars are means and SE (n = 4), respectively. Different uppercase letters indicate significant differences among treatments during 2019–2022 determined by linear mixed model based on standardized annual plant biomass data and followed by post hoc test at p < 0.05. * indicates significant differences between treatments and control at each time point determined by one-way ANOVA followed by post hoc test at p < 0.05

Full size image

Relationships between SOC characteristics, plant biomass, and environmental conditions

For SOC stocks, the obtained PLS regression model had RMSEP of 0.30 with R2 = 0.39 and could explained 70.1% of the variance in the selected environmental variables (PC1 from FTIR spectra, N concentration, EC and maximum temperature (Tmax)). PC1 and N concentrations presented their influence on SOC stocks (VIP > 1) and on primary component (component 1) (Fig. 5a). SOC stocks positively correlated to N concentration while had negative correlation with PC1.

Fig. 5
figure 5

Score and loading biplot for the partial least squares (PLS) regression analysis of a soil organic carbon stocks (SOC) and b SOC chemical composition determined by PC1 and PC2 from PCA of FTIR spectra, as influenced by environmental descriptor variables. Variables in bold indicate variable importance in projection (VIP) scores > 1.0. Soil amendment treatment scores are represented on the primary axes, and loadings on the secondary axes. Variable loadings show the relationships between predictor environmental variables (black circles) and the response variables (black diamonds). δ13C: carbon isotope ratio; SOC: soil organic carbon stock; %N: total nitrogen concentration; pH: soil pH; EC: electrical conductivity; DOC:SOC: DOC to SOC ratio; PC1 and PC2: scores of principal components (PC1 and PC2) of FTIR spectra; Tmax: maximum temperature in each cultivation year

Full size image

By performing PLS regression of PC1 and PC2 from FTIR spectra, a model with 2 components was obtained to determine factors that affected SOC chemical composition as assessed by using FTIR spectra (Fig. 5b). Seven independent variables (SOC stock, DOC:SOC ratio, carbon isotope ratio (δ13C), N concentration, pH, EC and Tmax) were included in the model. The obtained model had RMSEP of 0.04 for PC1 and 0.03 for PC2. It could explain 69.7% of the variance in independent variables (selected environmental variables) with R2 = 0.66 for PC1 and 0.57 for PC2. SOC stock and EC were found to influence PC1 and PC2 (VIP > 1) over the other independent variables. However, all independent variables had VIP > 0.8. Primary component (component 1) was determined by SOC descriptors (SOC stock and DOC:SOC ratio), while secondary component (component 2) was determined by soil chemical descriptors (pH and EC) and δ13C.

Similar to the score plot (Fig. 3a), the biplot (Fig. 5b) presents a separation of SOC chemical composition (as determined by PC1 and PC2) between dolomite and other amendment treatments by component 2 regardless of time. SOC chemical composition in soil amended with dolomite was linked to higher pH, EC and δ13C. SOC chemical composition in control and compost amended soil were negligibly explained by the model as they distributed around the origin. Instead, component 1 segregated SOC chemical composition under charcoal and charcoal + compost in the second from the third year. SOC chemical composition in the second year, in particular under charcoal and charcoal + compost treatments, was influenced by higher SOC stock associated with higher N concentration and temperature. Furthermore, higher DOC:SOC ratio shaped a dense cluster of SOC chemical composition under charcoal, charcoal + compost and charcoal + sawdust treatments in the third year.

However, the PLS regression model negligibly explained relationships between annual plant biomass and the soil parameters and the climatic conditions (R2 = 0.07) and their weak correlations were observed (Suppl. Figs. 3 and 4). SOC stocks, DOC and N concentrations as well as PC1 showed their influence on annual plant biomass (VIP > 1) in the current study (Suppl. Fig. 3).

Discussion

Effects of soil amendments on SOC characteristics

The organic soil amendments significantly increased SOC stocks (Fig. 1) during the study period, especially in the compost and the charcoal + compost treatments. All organic soil amendments used in the current study contained high organic C concentrations (> 20%) with variable amounts of available nutrients, depending on the source materials (Suppl. Table 2). Among the organic amendments used, charcoal as well as sawdust tended to be main sources of SOC in their treatments possibly due to their lower turnover rates (Lorenz and Lal 2014). Furthermore, we posit that the high porosity of charcoal as well as sawdust would also benefit soil microbial niches (Leng et al. 2012) and potassium (K) (Sukkaew et al. 2022) particularly in the studied soil where nutrients were very limited (Suppl. Tables 1 and Suppl. Fig. 6). This combination would result in only slight gains in its SOC stocks (13.60 Mg ha−1) during the first 3 years. Decreasing SOC stock together with SOC stock loss after dolomite reapplication (Fig. 1b) indicated negative impact of excessive dolomite application on SOC accumulation. The net effect of lime application has been discussed and is seen as somewhat controversial. When sufficient nutrients are available for plant growth, liming can promote plant growth and lead to increases in SOM (Paradelo et al. 2015). Yet, in soil with limited SOM or nutrients, liming is likely only to increase the mineralization by neutralizing soil pH (Minick et al. 2017). Accordingly, dolomite application negligibly supported the crop growth to enhance SOM build-up in the current study where soil contained low level of SOM and nutrients. This was also supported by the positive and close relationship between SOC and N concentration and its negative relationship to chemical composition PC1 (Fig. 5a). These relationships demonstrated that SOC stock in the current study was importantly influenced by nutrient availability as organic substances indicated by PC1 were potentially released by plant roots and soil microbes to acquire nutrients (Richardson et al. 2009) and would promote decomposition of organic materials resulting in SOC stock decrease particularly in the dolomite treatment.

On the contrary, SOC chemical composition under dolomite treatment was distinct from other treatments (Fig. 3a). It was characterized by high contribution of amide III (band IV) incorporating with long chain hydrocarbon (band I), which were released from either plant roots or soil microbes or both to acquire nutrients (Garg and Geetanjali 2009; Richardson et al. 2009). This was supported by its relatively higher concentration of DOC and DOC:SOC ratio (Fig. 3). The chemical composition of SOC in dolomite amended soil was apparently regulated by abiotic factors, such as EC and pH (Fig. 5b). This result possibly implied outstanding effects of dolomite application on abiotic condition modification in the degraded soil, such as in the current study. Moreover, its negative correlation to DOC:SOC ratio (Fig. 5b) suggested that SOC with low degradability tended to influence the chemical composition in dolomite treatment, as its DOC could be protected in stable soil aggregates (Sae-Tun et al. 2023) due to Ca bridge as a result of lime application (Keiblinger et al. 2016). These observations were also supported by Minick et al. (2017) who demonstrated that soil pH neutralization, as a result of Ca application, enhanced mineralization of litter-derived C by soil microbes.

Effects of soil amendments on annual plant biomass

Organic soil amendments significantly improved the annual plant biomass in the first year after amendment (Fig. 4). This observation indicated a short-term effect of soil amendment on quantitative plant biomass improvement in the post-tin mining soils which is consistent with Islami et al. (2011). On average, soils amended with charcoal + compost had the highest annual plant biomass followed by those with compost. These results demonstrated the role of compost in supplying essential nutrients and charcoal in fostering nutrient retention, a phenomenon which has been observed in previous studies (Islami et al. 2011; Vijay et al. 2021). A study on the same experimental setup clearly showed strong positive correlations between crop yields and available soil nutrients: available P for centrosema (Centrosema pubescens) and available K for cassava (Maftukhah et al. 2022). In addition, N concentrations in soil during the study period were very low (< 0.2%, Suppl. Fig. 6a). Indeed, considerably lower plant biomass was recorded in the current study compared to other studies in Indonesia (Islami et al. 2011; Sukartono et al. 2022) suggesting that plant nutrients were among key limiting factors in this soil system, as we knew from the outset (Suppl. Tables 1 and Suppl. Fig. 6a). Accordingly, additional nutrient supplies would be required to successfully amend this highly degraded soil for crop production, as suggested by Kimetu et al. (2008), especially when using C-rich materials with high C:N ratio, such as charcoal and sawdust.

The additive effects of applied charcoal would increase the plant biomass produced from charcoal + compost treatment over and above the other treatments by improving the physico-chemical and biological properties of the soil. It has been shown for example, that charcoal (or biochar) application could increase soil water holding capacity (Maftukhah et al. 2022) and soil available water content (Głąb et al. 2016), thus preventing nutrient losses (Gul and Whalen 2016) and promoting nodulation in legumes (Mia et al. 2014), possibly leading to improved root growth (Gul and Whalen 2016). However, the effect of the biochar is dependent on feedstock, dose usage, particle size, production, and soil characteristics (Aller 2016; El-Naggar et al. 2019). It has been suggested that application of charcoal/biochar alone could cause minimal N and P immobilization in nutrient limited soils (Gul and Whalen 2016; Hood-Nowotny et al. 2018), and this was observed in the fourth year in the current study, as plant biomass in the biochar only treatment was slightly lower than the control.

In contrast to organic amendments, dolomite negligibly improved plant biomass, although it stabilized and neutralized soil pH throughout the study period (Suppl. Fig. 6b). Other studies on wheat and legumes revealed improved grain yields by applying dolomite to tropical soils (Moreira et al. 2015; Raboin et al. 2016; Carmeis Filho et al. 2017). Contrastingly, dolomite might have a negative impact on cassava yield which was a main crop in the current study as Ca could decrease K availability especially in coarse textured soils (Sukkaew et al. 2022). This indicated that the effects of soil amendments depend on crop species and soil nutrient availability which also were reviewed by Vijay et al. (2021) and Wortman et al. (2017). In these nutrient limited soils of this study, it should be noted that plant biomass was predominantly constrained by nutrient availability rather than SOC improvement and the other measured environmental variables, which had weaker relationships with the annual plant biomass (Suppl. Figs. 3 and 4).

Overall, the effect of soil amendment on SOC stock and plant biomass improvement tended to be more promising than SOC chemical composition in the post-tin mining soil during the first 4 consecutive years of reclamation. The use of soil amendments, in particular organic ones, could accelerate soil health improvement in highly degraded lands as evidenced by increased SOC stocks and plant biomass production in the current study. Based on the observed results, the magnitude of amendment effects would tend to be co-influenced by time and crop species. However, the increase in SOC stocks over time in control suggested the benefits of minimum practices, that none of soil amendment, fertilization and irrigation was applied, to possibly increase SOC stock with the lowest cost. The reapplication of soil amendments was not necessary except compost application which underlined the necessity of nutrient management to sustainably improve the health of the highly degraded soil.

Conclusions

During the pioneer stage of soil reclamation, application of charcoal + compost as nutrient sources constituted the most promising amendment measure to improve soil health. The efficiency of this treatment is largely mediated by enhanced SOC dynamics as the key driver for restoring soil ecosystem functions in the post-tin mining site. Unlike our dolomite treatment, this charcoal + compost combination, along with other organic amendments applied in the study, fostered an increase in SOC stocks, thereby crucially regulating its chemical composition (e.g., through their effects on plant root and microbial growth and activity, and SOC retention). Furthermore, both the quantity and quality of SOC were evidently influenced by the interaction among the type of amendment, time, and crop species, with more noticeable effects observed on SOC chemical composition when associated with charcoal. While organic amendments generally contribute to an accrual of SOC stocks, the addition of compost together with charcoal simultaneously improved nutrient provision from organic sources with relevant sorption, storage and buffering characteristics attributed to recalcitrant SOC. This broad effect of a mixed organic amendment on soil health could not be achieved when limiting to mineral (dolomite) or recalcitrant C sources only. Consequently, restoration efforts as expressed in improved plant biomass were highest in the charcoal-compost mixture treatment. We therefore concluded that combined usage of soil amendments, potentially together with time and crop species grown, are essential to initiate restoration of soil ecosystem functionality in tin-mining effected soils in Bangka through their influence on characterizing SOC. Hence, tradeoff between food provisioning and other benefits should be considered for species selection during the reclamation campaign.