Main

Urban soils are the foundation of green infrastructure initiatives. However, soils in cities are often compacted, polluted, sealed1 or otherwise ill suited to provide ecosystem functions2. As a solution, suitable soil materials are historically imported from surrounding ecosystems—a practice known as ‘land-take’3. However, in recent decades there has been a shift towards a more sustainable practice of re-utilizing surplus sediments and construction wastes4 as ‘constructed soils’. The re-use of these mineral wastes, accounting for 40% of the annual waste generated in the European Union5, supports a circular economy approach within cities to provide soil-based ecosystem services4.

In assessing the functionality of such an approach, studies have shown that incorporating mineral structural components such as track ballast, demolition rubble6,7,8, recovered concrete9 or sorted brick and mortar10,11 into urban constructed soil mixtures does not acutely diminish nutrient storage and can even improve water cycling in urban green. Moreover, the use of excavated subsoil, the deep soil horizons displaced during construction projects, may favor soil structural development12,13, a central process underpinning many soil functions and services. Yet, organic amendments (OAs) are needed for these mineral parent materials to more closely replicate a fertile topsoil. The use of OAs generated from the urban waste stream—such as biosolids, compost and biochar—can reduce reliance on less sustainable amendments such as peat and chemical fertilizer14 while still enhancing fertility.

Biochar is a particularly advantageous amendment capable of enhancing soil fertility15 and potentially replacing benchmark pollutant-retaining materials such as zeolite and activated carbon16,17, whose production and use in urban stormwater processing is often energy intensive and costly, contributing to urban carbon emissions17,18. As a (by)product of pyrolysis or gasification19, biochar can be sustainable alternative, valorizing organic waste while co-producing energy17,20. Biochar-amended constructed soils could provide further services in an urban context21. Pyrolyzing and redepositing urban waste products, such as woody tree clippings, into soils augments soil carbon longevity22. This direct form of belowground carbon accrual, combined with any positive influences on urban greening productivity, could represent an offset of 0.3 to 1.2% of annual emissions at the urban scale21. However, despite research showing that biochar can improve carbon storage, increase yield, retain pollutants, save water, carry nutrients and reduce acidity, these properties are rarely found all within one source material15,22. This discrepancy justifies the incorporation of multiple OAs in urban constructed soils to ensure a broader range of functionalities.

Our research experimentally evaluated the potential of local, circular economy-derived parent materials for constructing multifunctional urban soils. Based in Munich, Germany’s most densely populated city, we sourced a regionally typical subsoil—a skeletal sandy clay loam with a carbonate concentration of roughly 50% (Extended Data Table 1a)—from the construction waste of the city’s old military base being revitalized as an eco-housing district. Another subsoil from a neighboring urban area served as an additional case study. We aimed to maximize soil properties and functions linked to the socio-environmental benefits of urban green’s fertility, runoff infiltration, stormwater contaminant immobilization and carbon accrual by amending the subsoils with combinations of two types of waste-based OA—compost and biochar (Fig. 1). We then compared the OA combinations with a pollutant–adsorption benchmark, granulated activated carbon (GAC)19 and a low-cost option of only municipal green-waste compost (mGWC).

Fig. 1: Constructing and assessing soil mixture properties and functions.
figure 1

The top section in red details the production of the constructed soil mixtures including the processing of the excavated mineral waste, characteristics of the organic amendments and the nine constructed soil mixtures tested in each subsoil. The bottom section in blue details the analyses conducted on the individual parent materials, conducted on the constructed mixtures, monitored across the 30-day incubation and measured on the harvested, incubated samples. NMR, nuclear magnetic resonance. Figure created using BioRender.com.

Full size image

We evaluated three biochars, two wood-waste (WW) chars of high and low temperature (WW | 850 and WW | 540) and one green-waste (GW) char (GW | 680), sourced from a local provider (Extended Data Table 1b; Carbuna AG, Memmingen, Germany). We tested an OA addition of 4% per mass (~20% by volume), to address economic feasibility and a swift transition of research results to praxis, complying with the upper limit of government mandates for organic matter within constructed soils23. Twenty percent by volume has also been suggested as an important threshold in maintaining beneficial hydro-structural functioning while minimizing expensive organic parent material24. For each upcycled mineral subsoil, we created nine soil mixtures (Fig. 1)— three controls: (1) subsoil without amendment, (2) a mixture with only mGWC and (3) a mixture incorporating GAC—and six constructed experimental mixtures: (4)–(6) three containing compost and one type of biochar and (7)–(9) three containing compost with two biochars of varying qualities.

Specifically, we asked: does combining OAs of different qualities in waste-based constructed soil enhance soil multifunctionality? We predict the mGWC, and biochar processed at lower pyrolysis temperatures, to substantially increase soil fertility, whereas the (micro)porosity associated with high-temperature wood-waste biochar will augment pollutant retention. Furthermore, although some biochar lend to accrual of stable carbon—containing condensed aromatic carbon compounds similar to activated carbon—these low-reactivity OAs inherently contribute less to organo-mineral associations and, therefore, less to the transition from constructed mixtures to structurally developed soil. We propose that individual biochars are limited in the breadth of their application due to properties unique to their innate feedstock and processing. Using a cascade model, we translate improvements in properties to enhancements in ecosystem functioning and predict that combinations of biochars, along with compost, allows greater synchronization of ecosystem services, which we quantify in a multifunctionality score.

Results

Stormwater contaminant immobilization by parent materials

To assess capacity for pollutant removal, we conducted batch adsorption tests with the various parent materials for select heavy metal and organic pollutants in a synthetic stormwater matrix (Fig. 2a,b). Organic contaminants were represented by the biocides mecoprop and terbutryn—both commonly used in building façade coatings and noted for their high solubility and leaching potential. The removal capacity of the high-temperature wood-waste biochar (WW | 850 BC) was directly comparable to the GAC control, immobilizing 98% of the organic contaminants (Fig. 2a). Combinations of the WW | 850 BC with the medium-temperature green-waste biochar (GW | 680 BC) or the low-temperature wood-waste biochar (WW | 540 BC) showed similar capacities in immobilizing terbutryn (98 ± 0% removal) but reduced capacity for mecoprop (89 ± 3% and 88 ± 3% removal, respectively). Alternatively, the use of lower-temperature biochars individually exhibited limited adsorption of the organic contaminants.

Fig. 2: Chemical characteristics and pollutant removal capacity of the OAs.
figure 2

a, Individual pollutant-retention capacity (n = 3) by OAs for the four pollutants tested: heavy metals copper and zinc and biocides mecoprop and terbutryn. Error bars represent the mean ± one standard deviation. b, The aggregated pollutant-retention capacity (n = 12) of the singular and dual biochar treatments, with triangles representing heavy metals and circles representing organic pollutants. Dashed lines indicate non-significant differences between the singular biochar treatments and mixes containing 1:1 ratio of municipal compost. Error bars represent the mean ± one standard deviation. c, Results of 13C NMR spectroscopy (n = 1) displaying the relative intensities of the types of carbon bonding at chemical shifts of 0–45 ppm (alkyl C), 45–110 ppm (O/N-alkyl C), 110–160 (aryl-C) and 160–220 ppm (carboxyl-C) for the biochars and municipal compost. Figure created using BioRender.com.

Full size image

The GW | 680 BC achieved high heavy metal removal, immobilizing 90 ± 9% of copper and 83 ± 8% of zinc (Fig. 2a), surpassing the performance of the GAC control, which averaged 60 ± 8% removal across both pollutants. A 1:1 (w/w) dual mixture of WW | 850 and GW | 680 performed similarly well, immobilizing > 95% of copper and > 70% of zinc. Although WW | 850 BC also demonstrated adequate immobilization of copper (84 ± 5%) and zinc (62 ± 12%), removal of both analytes by the WW | 540 BC was decisively poor, 40 ± 14% and 12 ± 17%, respectively.

Compost alone was able to moderately adsorb heavy metals (65 ± 6% zinc; 51 ± 4% copper) but was ineffective in immobilizing organic pollutants (31 ± 9% terbutryn; 6 ± 1% mecoprop) (Fig. 2a). Mixtures of compost and individual biochars (1:1 w/w) diminished the immobilization capacity of WW | 850 and GW | 680 on heavy metals by 7 ± 12 and 4 ± 13 percentage points, respectively, while increasing the potential of WW | 540 ( + 23 ± 19% points). Addition of compost decreased WW | 850 and WW | 540 retention of the two organic biocides by an average of 27 ± 1 and 9 ± 9 percentage points, respectively.

Considering an aggregated pollutant removal capacity, GAC (79 ± 22%), WW | 850 (86 ± 17%) and the combination of WW | 850 and GW | 680 (89 ± 12%) resulted in considerably higher total contaminant immobilization than all other OA combinations (Fig. 2b).

The subsoils showed similar capacities to remove heavy metal pollutants, with the medium sand removing 26 ± 1% and 54 ± 4% of zinc and copper, respectively, and the sandy clay loam removing 36 ± 4% and 69 ± 5% (Extended Data Table 2). However, immobilization of trace organic compounds was low, with the sandy clay loam removing less than 20% of terbutryn and less than 10% mecoprop, while the medium sand removed effectively zero organic pollutants.

Boosted fertility parameters in constructed soil mixtures

In mixing the OA combinations with subsoil, we assessed their potential for boosting fertility in constructed soil mixtures. On average, OAs decreased the original bulk densities of the sandy and sandy clay loam mineral materials, 1.31 ( ± 0.06) and 1.36 ( ± 0.01), respectively, by 9% and 7% (Table 1). This OA-associated drop in bulk density was accompanied by a significant 33% and 12% improvement in the mineral soils’ water content at field capacity (Fig. 3). Interestingly, increases in water content associated with OA addition in the sandy subsoil material did not surpass the baseline water content of the sandy clay loam mineral subsoil.

Fig. 3: Gravimetric water content at field capacity.
figure 3

(n = 5). Fig. 1 provides soil mixture clarifications. Asterisk (*) indicates significant differences from the subsoil control that were analyzed using one-way ANOVA, with targeted post-hoc orthogonal contrasts restricted to predefined comparisons. Error bars represent the mean ± one standard deviation. Figure created using BioRender.com.

Full size image
Table 1 Physical, chemical and biological parameters of the constructed soil mixtures underlining different soil functions
Full size table

Furthermore, all OA combinations significantly increased total nitrogen content over the initial soil mineral material (Table 1). mGWC and addition of 2% GW | 680 BC displayed the largest influences, + 0.77 ± 0.07 and + 0.87 ± 0.08 mg g−1, respectively, across the two mineral soils—a roughly 100% increase from the sandy clay loam and 340% and 380% increases for the nutrient-poor sandy soil; significantly higher than the GAC-amended soil mixtures.The pH of the medium sand was neutral (7.0 ± 0.0), whereas the calcareous sandy clay loam (CaCO3 content 51.2 ± 1.8%) was mildly alkaline at 8.3 ± 0.1 (Table 1). The organic materials were all alkaline (Fig. 1), ranging from the compost at 8.1 ± 0.0 to GW | 680 BC at 10.7 ± 0.2, inducing significant increases in pH values over the pure sandy mineral material, and over the 4% mGWC control. OA types showed large differences in improvements to nutrient availability (Table 1). In the sandy clay loam material, 4% mGWC addition induced a significant 25% increase in total cation exchange capacity (CEC); all other amendments had little to no impact, only sustaining the effective increase from the 2% mGWC within the mixtures. As the initial CEC of the sandy subsoil was quite low, an average of just below 5 cmol•c kg−1 soil, the increase induced by even 2% mGWC reflected significant changes in several of the amended mixtures. Consequently, a 4% addition of mGWC afforded an 80% increase in the sandy subsoil’s total CEC, significantly higher than all mixtures containing biochar.

Mineral surface associations and aggregate development across 30 days

We conducted a 30-day incubation on all soil mixtures, per the experimental set-up of Bucka et al. 202125, which documented the rapid transformation of artificial parent materials into structurally stable soil. Although short-term incubations cannot completely predict soil development under field conditions, they can provide an assessment of a soil’s pedogenic potential.

Although we observed a small shift towards macro-aggregate formation ((i) in Fig. 4a) in the sandy material, increases in the mean weight diameter of water-stable soil aggregates ((i) in Fig. 4b)—an indicator for stronger soil structural resistance—were marginal and insignificant. Comparatively in a month’s time, we noted an average 41% increase in small macroaggregates ((ii) in Fig. 4a) and a subsequent significantly larger mean weight diameter ((ii) in Fig. 4b) for the sandy clay loam material. Interestingly, in both soil materials, soil structural development in OA mixtures did not significantly differ from that of the excavated subsoil alone (Fig. 4), despite preceding significant increases in microbial respiration across the 30 days of incubation (Table 1).

Fig. 4: Soil structural development.
figure 4

a, Development of water-stable aggregate size distribution, represented by four size fractions, as seen in the primary particle size of the subsoil without aggregates, the excavated subsoil pre-incubation (n = 3) and the development of the subsoil and soil mixtures after 30 days of incubation (n = 4) b, Changes in water-stable aggregate mean weight diameter (MWD) from the pre-incubation time point (n = 4); asterisks indicate significant differences in mean weight diameter from the excavated subsoil before incubation investigated using one-way ANOVA, with targeted post-hoc orthogonal contrasts restricted to predefined comparisons. Fig. 1 provides soil mixture clarifications. Error bars represent the mean ± one standard deviation. Figure created using BioRender.com.

Full size image

The available specific mineral surface area (SSA) of the calcareous sandy clay loam and medium sand parent materials were both low at 8.23 m2 g−1 and 7.21 m2 g−1, respectively (Table 1). Dry mixing with 4% mGWC supported early organo-mineral associations in both mineral materials, a calculated 0.95 m2 g−1 of OM-covered mineral surface in the sandy clay loam and 0.32 m2 g−1 in the medium sand. Biochar and activated carbon are used as pollutant-immobilizing OAs on behalf of their high porosity and ensuing surface areas20. Therefore, we could not rule out that these amendments could both associate with mineral material, covering mineral surfaces, and add surface area. Accordingly, we did not calculate OM-coverage for these mixtures. However, we did observe soil material-dependent associations of biochar with mineral soil surfaces, with the medium sand showing larger increases in SSA in all dry-mixed biochar-based OA combinations. Addition of 2% GAC resulted in the largest increase in surface area for the mixtures adding approximately 16 m2 g−1 to each mineral soil, a roughly 210% increase. Whereas the 30-day incubation led to a ubiquitous decline in free mineral surface area of all soil mixtures, OA-amended sandy clay loam mixtures showed strong similarities in reduction to the subsoil mineral material alone (Extended Data Fig. 1a). Alternatively, OA-amended medium sand mixtures showed equal to or greater reduction in available mineral surface area to that of the pure subsoil after incubation (Extended Data Fig. 1b).

Contributing factors to carbon stabilization

Per definition, amending the mineral soil material with 4% w/w carbonaceous matter led to significant increases in organic carbon that were nearly additive in nature (Table 1 and Extended Data Table 1b). The 13C nuclear magnetic resonance (NMR) spectra revealed dominantly aromatic structures (Fig. 2c), indicating the high chemical recalcitrance of the incorporated char materials that increases soil organic carbon residence times. mGWC added significantly less organic carbon per gram amendment to the mixtures (Extended Data Table 1b), moreover, the ratio of aromatic carbon was far lower (< 25%), with easily mineralizable O-alkyl groups representing 40% of the chemical profile (Fig. 2c). Even so, microbial respiration data did not show soil mixtures containing only compost to have higher mineralization rates than those containing biochar or activated carbon (Table 1). Contrarily, the combinations containing 2% GW | 680 BC respired significantly more than the mGWC and GAC controls, holding true for both mineral materials (Table 1). In the sandy soil, this singular biochar mixture also lent to significantly higher cumulative mineralization than either of the 1% w/w dual biochar combination counterparts (Table 1). When auditing our values for carbonate dissolution via normalization to grams of organic carbon, both of our nitrogen-rich mixtures, those containing GW | 680 or 4% GWC, indicated more active microbial communities (Extended Data Fig. 2a).

Discussion

Our work shows the ecological value of amending urban mineral wastes with OAs of different qualities, promoting a circular soil economy4. Ecological value is derived from synergizing an increasing number of soil services, which we calculate using a simple standardized indicator- and average-based multifunctionality score (0 to 1) for our constructed soil mixtures amended with mGWC and biochar (Fig. 5; Supplementary Section A). We find that while addition of mGWC alone has limited multifunctionality (0.48 ± 0.05), dual biochar mixtures containing high-temperature wood-waste biochar (WW | 850 BC) achieve a multifunctionality balance (0.81 ± 0.03) equivalent to soil mixtures containing GAC (0.80 ± 0.06). However, the singular WW | 850 BC mixture holds the highest multifunctionality score (0.88 ± 0.05), challenging our hypothesis that biochars of different qualities are needed to enhance multifunctionality. Nevertheless, our results make evident that biochar and compost control different soil functions.

Fig. 5: Conversion of sediment construction wastes to multifunctional soil mixtures via addition of organic amendments.
figure 5

Limitations and synchronization of ecosystem services. Figure created using BioRender.com.

Full size image

The low N contents in our mixtures are comparable to other, unsealed urban soils26,27 and to B or C horizons of local grasslands28, with CECs similar to other waste-based constructed soils6. Of the OAs tested, only mGWC contributed significant amounts of primary and secondary macro-nutrients (Table 1; Extended Data Table 3) to the soil mixtures, advocating for its inclusion as the base OA in otherwise infertile mineral constructions. Biosolids, a waste-based OA not explored in this study, also offer a fast-releasing supply of nutrients, particularly phosphorus14. However use of biosolids on soils in Germany, and probably other European countries to follow, is being phased out, making composting pathways a safer investment for municipalities. Adequate water and air supply are also important features for urban greening29. Despite differences in OA type, particle size distribution and pyrolysis production temperatures (Extended Data Table 1b), we found no significant differences between the bulk densities and water-holding capacities of the amended soils (Table 1 and Fig. 3)15,30. This supports the universality of OAsʼ positive effects on plant-available water retention24,31 and bulk density, suggesting practitioner choice in OA combinations can be made based on other parameters. Function scoring, weighed evenly between increases sustained in CEC, N content and water content at field capacity, revealed mixtures with 4% mGWC (0.91 ± 0.16) and 2% green-waste biochar (GW | 680 BC; 0.75 ± 13) added the most to soil fertility.

In our scoring the function of aggregated pollutant immobilization capacity, all biochar mixtures containing WW | 850 achieved scores equal to or greater than the GAC control (0.89 ± 0.04; Supplementary Section A and Fig. 2b). The capacity of GAC, virgin coal that is physically or chemically ‘activated’ to increase its porosity, to immobilize pollutants is often attributed to its high SSA—particularly pertaining to organic pollutants20. Accordingly, the GAC-amended mixtures depicted over a 200% increase in SSA (Table 1). However, similar organic pollutant immobilization by biochars were not accompanied by the same drastic increases in amended soil mixtures SSA. We hypothesize the European Biochar Certification-certified measure for SSA, N2 gas diffusion, may underestimate biochar surface area by up to two orders of magnitude32, and this unmeasured range of micropores, often associated with high-temperature biochars, could contribute to the highly effective organic pollutant-retention capacity of WW | 850. Furthermore, the presence of high aromaticity, evidenced by the aryl carbon content in the 13C NMR spectra, could instigate hydrophobic interactions offering an additional pollutant-stabilizing mechanism33. These phenyl structures, together with surface functional groups containing nitrogen34, provide electrostatic attraction33 with the potential to bind the heavy metal cations, leading to stronger retention by the two higher temperature biochars. We also acknowledge the highly alkaline pH values of the OAs, particularly the green-waste-based biochar (Extended Data Table 1b), which suggest further stabilization of heavy metals via precipitation as hydroxides or carbonates33. Of note, the pH of an evolving soil system should be taken into consideration when relying on precipitation as an immobilization mechanism, as acidifying mineral material would hinder long-term retention of some heavy metals. Furthermore, mGWC—a nutrient-rich compound—and combinations thereof, displayed weaker pollutant immobilization capacities, particularly of organic pollutants. This indicates a trade-off in pollutant retention when augmenting mixtures for fertility, as contaminant sorption is defined by the available binding sites in a soil, sites for which nutrients and pollutants may directly compete.

Stable soil structure makes it possible to use soils, rather than just sand or gravel, in green infrastructure systems intended for infiltration and drainage, with the coinciding benefit of greater carbon storage potential35. However, the development of a stable soil structure, at least via natural pedogenic processes, takes years to decades, directly conflicting with timeframes of urban initiatives where time from installation of a constructed soil to the use of a green space may occur in months, if not weeks. Considering this limitation, we tested the feasibility of rapid (30 days) water-stable aggregate formation, encouraged by successes in both artificial and constructed soils12,25,36. Within our 30-day experiment, we did not find water-stable structural development in the sandy soil but did observe greater mean weight diameter of aggregates in the sandy clay loam soil mixtures. Interestingly, our results present no significant differences between the excavated subsoils and the amended treatments (Fig. 4), downplaying the role of organic matter-mediated structural formation in the first 30 days. Our results imply that any rapid increase in aggregation is contingent on the subsoil parent material rather than the introduction of organic matter.

Furthermore, we observed a large amount of biochar floating after our wet-sieving treatment, suggesting a low bond strength between the biochar and mineral fraction37. For biochar, nucleation of structural aggregates, like many other functionalities, is dependent on surface chemistry. Results from our 13C NMR spectroscopy (Fig. 2), in combination with the H to Corg ratio (Fig. 1), suggest the absence of reactive functional groups. Although our findings are in line with studies that illustrate biochar application to be neutral or even antagonistic to aggregate formation36,38,39,40,41,42, other studies have demonstrated biochar’s ability to instigate soil particle agglomeration39,40,43,44. In reviewing the literature reporting biochar-instigated aggregation within short timeframes, we found biases in biochar chemical composition, with many studies either using chars processed at lower pyrolysis temperatures39,44 or char in mixtures with more labile material40,43. Therein, we conclude that municipal compost, granulated activated carbon and both green-waste and wood-waste biochars (>500 °C) are all unfit to initialize rapid structural formation in these soil materials. We recommend further research on constructing multifunctional soils for urban green initiatives focused on the intersection of infiltration, drainage and carbon cycling to investigate more intensive mixing approaches using labile, intermediate and recalcitrant organic parents materials to help stimulate both a rapid and sustainable development of soil structure.

Although biochar lends little to the protective process of stable aggregate occlusion, char’s complexity and the infrequent occurrence of its chemical form45 heightens microbial investment for mineralization, increasing its longevity in soils. Recent opinions suggest amending with this type of organic matter is the preferred approach for increasing C stocks in soils with a low proportion of reactive minerals46, such as our subsoil-based mixtures. Though our function scoring excludes the function of infiltration due to no significant differences in structural development between amended soils, we include the service of stable carbon accrual; here the mixtures with 2% WW | 850 BC (1.00 ± 0.01) and GAC (0.95 ± 0.11) demonstrate the highest potential (Supplementary Section A).

The results of our study emphasize the importance of different geogenic qualities on ecosystem service potential in constructed urban mixtures and subsequent soil. Although both mineral subsoils failed to retain organic pollutants, they exhibited natural attenuation of heavy metals (Extended Data Table 2). This effect was more pronounced in the sandy clay loam, probably due to its higher clay and carbonate content, which enhanced its capacity for heavy metal immobilization47. A high carbonate content also contributes to a more alkaline soil environment, which in turn influences both carbon and nutrient storage dynamics48. Furthermore, although our study lacked amendment-initiated water-stable aggregate development, we hypothesize the characteristic larger carbonate49 and clay fractions50 of the sandy clay loam parent material to induce abiotic gluing within the 30-day incubation, increasing the soil structural stability (Fig. 4). Geogenic properties also entail certain trade-offs. The calcareous cementing that allows for rapid structural formation may also diminish root space, while the increase in pH associated with higher carbonate content may limit nutrient availability for urban green. The influence of particle size is also clear, exemplified in the baseline difference in the water content of the excavated subsoils at field capacity (Fig. 3), as additions of reasonable quantities of OAs to sandy soil material could not overcome the inherent capacity of a finer soil material. These results raise the importance of knowledgeably constructing functional soil mixtures, choosing OAs to balance the weaknesses of the available mineral material.

Conclusion

Our findings highlight the functional performance of amended, circular-economy-based soil mixtures. Whereas all tested amendments improved soil physical and chemical properties of construction-derived subsoil mineral materials, select combinations of municipal green-waste compost and biochar outperformed both compost alone and mixtures containing granulated activated carbon, a common standard for pollutant immobilization. Using a cascade model, we translated these improvements in soil properties to quantitative scores for soil ecosystem services, comparing the functions of fertility for urban greening, stormwater contaminant immobilization and stable carbon accrual.

Nitrogen content and cation exchange capacity, two soil health indicators supporting plant fertility, were almost exclusively supplied by additions of compost, supporting its role as the base organic amendment in these mixtures. Despite its central role in fertility, we found a trade-off in compost’s pollutant immobilization functionality score. Here select biochar combinations exhibited high over overall retention of both organic and heavy metal pollutants, scoring higher in this functionality than granulated activated carbon. Although our scoring system showed the accrual of chemically stable carbon associated with the high-temperature wood-waste biochar mixtures to be equivalent to that of the activated carbon, none of the tested organic amendments initiated rapid water-stable aggregate formation that could both enhance hydraulic conductivity and act as an intermediate carbon storage pool. We affirm here the need for further mixing of organic parent material of different qualities—including more labile compounds as nuclear particulate organic matter to prompt aggregation and therefore further increase the multifunctionality of the mixture.

Overall, the multifunctionality score of mixtures contain compost alone was low, whereas biochar mixtures containing high-temperature wood-waste biochar achieved an equal or higher multifunctionality balance than those containing granulated activated carbon. Urban migration and densification place higher demands on city spaces, making it increasingly important to enhance the value of our green spaces by adding additional functions and benefits. Our findings emphasize that the highest joint multifunctionality score is obtained by mixing amendments of varying qualities and reactivities, though the necessity of an amendment is function dependent. We emphasize that practitioners should be knowledgeable of their city’s geological background, ecosystem service enhancement is dependent on the educated choice in amendments and their interactions with geogenically controlled variables.

Methods

Two subsoil materials of the study were provided by Bodeninstitut Prügl and Andreas Thaler, representing typical ‘waste soils’ excavated during construction projects. Sampled from the glaciofluvial gravel terrace deposits of the Alps, commonly referred to as the Munich gravel plain51, the first soil material is a transition horizon 20- to 40-cm thick between the topsoil and pure gravel, locally referred to as a ‘Rotlage’52. This skeletal sandy loam material with a high carbonate concentration (> 50%) is a Cw(T) horizon of a calcaric Regosol. Found on the Langweider high terrace along the river Lech and characterized by sandy loess53, the Augsburg soil material is a Bw horizon from a cambic Arenosol. The particle size distribution (Extended Data Table 1a) of these bulk soils naturally resist compaction and accommodate mandated infiltration rates set for urban green spaces intended for decentralized infiltration (DWA 2020). However, to conduct pertinent soil analyses, we investigate the ecological functions of a homogenized, active soil fraction (< 2 mm) (Fig. 1). The extent to which the bulk soils would be sieved is project dependent, but we conjecture that fertility and pollutant retention measured for the active soil fraction will be proportional to the ratio of the active to coarse fraction within a soil material, as often measured with carbon stocks54.

Biochars were chosen on the basis of availability from the local supplier; preference was given to feedstocks that were not manure- and sludge-based chars due to the lower potential of contamination and nutrient leaching55 and to biochars endorsed by the European Biochar Certification.

Analyses conducted on individual components

We determined particle size distribution of the mineral soils in duplicates, sieving to 2 mm then removing organic matter and carbonates by 30% H2O2 and 1 M HCl, respectively (Extended Data Table 1a). Coarse content data (> 2 mm) were obtained from testing agency factsheets. Samples were rotated overnight (> 16 h) with 0.025 M Na-Pyrophosphat. The sandy fractions (> 63 µm) were wet sieved, whereas silt and clay fractions were freeze dried and determined via X Ray sedimentation (Sedigraph III PLUS, Micromeritics). Biochar particle size distributions were provided to 0.33 mm by the company, however compost and granulated activated carbon were manually dry sieved (n = 2) to 0.5 mm (Extended Data Table 1b).

The chemical composition of the OAs was characterized using solid-state 13C NMR spectroscopy (n = 1; Bruker Biospin DSX 200 NMR spectrometer), where samples were spun in a magic angle spinning probe56 at a rotation speed of 6.8 kHz with an acquisition time of 0.4 ms. The obtained spectra were integrated according to four major chemical shift regions: 0–45 ppm (alkyl C), 45–110 ppm (O/N-alkyl C), 110–160 (aryl-C) and 160–220 ppm (carboxyl-C)57, with the alkyl C:O alkyl C ratio (−10–45/45–110 ppm) further computed to describe the degree of aliphaticity58. In measuring the high-temperature wood-waste biochar and granulated activated carbon, we experienced difficulties in obtaining a clean spectra, probably due to the high electrical conductivity that correlates with the stacking of aromatic sheets59. The H to Corg ratio for the organic amendments was calculated from an elemental analysis conducted by the TUM Catalysis Research Center in Garching, Germany (HEKATech EURO-EA).

We employed batch adsorption experiments (n = 3) using a synthetic stormwater matrix representing mixed urban runoff (composition as reported in Spahr et al. 202260) to assess the adsorption performance of the OAs and soil materials (Supplementary Section B). Mixtures of biochar (1:1) and mixtures of biochar and compost (1:1) were also tested. The adsorption performance of the biochars and the mixtures were compared with that of granular activated carbon. Soil and compost were also tested to evaluate their potential for enhancing the removal of both heavy metals and biocides. The testing vessels containing 0.5 g l−1 of adsorbent were spiked after a 24-hour pre-equilibration time to achieve a target concentration of 100 μg l−1 of the heavy metals copper and zinc and the biocides mecoprop and terbutryn. Three vessels of synthetic stormwater, spiked with pollutants but without adsorbents, were tested as control samples. The concentration of heavy metals and biocides after five days of contact with the adsorbents was measured through flame atomic absorption spectrometry (Varian Spectrometer AA-240FS). The limit of quantification (LOQ) was 5.0 µg l−1 for copper and 20.0 µg l−1 for zinc. The biocide concentration was determined through liquid chromatography coupled to mass spectrometry. The LOQ for all biocides was 25 ng l−1.

Analyses conducted on dry-mixed soil combinations

Figure 1 lists the analyses conducted on the dry-mixed, homogenized soil combinations. Bulk density (g cm−3) was measured with three replicates by loading approximately 100 g soil into 100 cm3 soil cores. Each core was tapped three times to allow settling of soil but not to use over-head pressure for compaction, before being saturated overnight and placed on a suction plate (EcoTech Umwelt-Messsysteme) to determine volume at field capacity, here a water tension of − 15 kPa. Mixtures were then dried at 105 °C to determine the corresponding soil weight. Specific surface area (SSA) of the soil mixtures was measured with two replicates via multipoint N2-BET61 (m2 g−1; AUTOSORB-1, Quantachrome Instruments) using approximately 2–3 g of air-dried material outgassed with He (> 14 h at 40 °C under vacuum).

Calcium carbonate content of mixtures and individual components were quantified using a calcimeter (Eijkelkamp), and inorganic carbon contents were calculated as 12% of the measured calcium carbonate per the International Organization for Standardization formula ISO 10693 (n = 5; mg g−1). Total carbon, organic carbon and total nitrogen contents were then measured via dry combustion by a CN analyzer (HEKAtech) (n = 5; mg g−1). Organic carbon contents were obtained by adding sufficient HCl to destroy the inorganic carbon within the samples before combustion.

The nitrogen values, particularly of the unamended subsoil, approach the detection limits of the instrument, rendering significant differences in computed nitrogen balances ineffectual. However it should be noted, in an additive evaluation of our parent materials, our sandy substrate was amended with 2% WW | 850 BC and that with 2% GW | 680 BC ( + 0.30 and + 0.21 mg g−1, respectively) and the sandy clay loam with only compost (+0.20 mg g−1) to contain marginally more nitrogen than expected from the individual components (Supplementary Section C). In budgeting for the maximum total N in our system after incubation, including small nutritional inputs to not limit microbial growth, we observed compost (35 ± 19% medium sand, 19 ± 8% sandy clay loam) and singular biochar mixtures, most notably WW | 850 (44 ± 22%, 13 ± 6%), to contain less N than theorized; whereas the combination of WW | 850 and GW | 680 BC maintained theoretical levels (4 ± 19%, 4 ± 7%).

The pH value of the mineral material and soil mixtures were conducted with a pH Meter (Mettler-Toledo SevenEasy S20; n = 5) in a 1:2.5 soil to deionised water ratio, whereas the individual OAs were measured at more diluted ratios of 1:5 and 1:10 for compost and biochar, respectively. Due to immediate liquid–solid phase separation the pH value of the granulated activated carbon as an individual OA was not measured. Samples were shaken at 100 r.p.m. for 1 h, before allowing to settle for another hour before analysis. Total cation exchange capacity and effective cation exchange capacity (macro-nutrient ions K+, Na+,Mg2+, Ca2+) of the soil mixtures were obtained following the German Handbook for Forestry Soil Analysis (König et al. 2005) (n = 5; cmol•c kg−1). In short, 2.5 g of each sample was percolated first with a BaCl2 solution, followed by MgCl2. Both effluents were collected and the cation ion sum (Ca, Mg, K and Na) and the total Ba ion release were measured and compared via inductively coupled plasma (ICP) analysis (Vista-PRO CCD Simultaneous ICP-OES, Varian Inc.).

Incubation experimental set-up and concurrent analyses

The set-up of the incubation experiment followed the procedure of Bucka et al. 201962, shortly: 300 g of an air-dried soil mixture homogenized by repeated mixing was filled into a microcosm and placed onto a suction plate (polyamide membrane, pore size 0.45 μm, EcoTech Umwelt-Messsysteme) at a water tension of −15 kPa (corresponding to a pF value of 2.2, approximate field capacity of loamy sands) in a closed hydraulic system.

During the first three days of the experiment, 30 ml of a 1:10 diluted Hoagland’s solution (pH 5.5, Hoagland’s No 2 basal salt mixture, Sigma-Aldrich) was administered per day to ensure all pores were filled and that the samples could then equilibrate to −15 kPa. Thereafter, 10 ml were administered after each respiration measurement (every 48–72 h) to counteract evaporation, leading to a total input of 368 mg of nutrient powder per microcosm. Gravimetric water content was monitored by weight throughout the incubation and used as a single indicator of plant-available water content due to the high correlation between increases in water content at field capacity and plant-available water content with soil organic matter (SOM) addition63. Five replicates of each soil mixture were incubated for a total of 30 days in the dark at a constant temperature of 20 °C (Fig. 1, box 4; as an exception, one replicate of the sandy clay loam containing granulated activated carbon was lost during handling).

CO2 release of the microcosms was measured every 48 to 72 h by placing the samples into air-tight containers for approximately 5 h, capturing the CO2 released in 15 ml of 0.1 M NaOH solution. Afterwards, 2 ml of BaCl2 were added to stop the reaction from reversing before the NaOH solution was titrated to pH 8.3 (Mettler-Toledo), according to Bimüller et al. 201464 and Luxhøi et al. 200665. Respiration was extrapolated for days during which respired CO2 was not measured (Extended Data Fig. 2b), allowing the overall CO2-C release to be determined. A cross-soil material comparison of the cumulative CO2 respiration values normalized to grams of organic carbon is used as evidence for the minimal to non-existent transformation of the carbonates present in the calcareous sandy clay loam mixtures to CO2 (Extended Data Fig. 2a).

After 30 days the experiment was terminated and the top centimeter of the microcosm was discarded before the samples were dissected into a fresh fraction and air-dried fraction. The former was used within two weeks of sampling, while the latter was dried for a minimum of seven days before further analysis.

Additional analyses conducted on incubated samples

As an indicator of soil structural stability against hydrological forces, we use the size distribution of isolated, water-stable aggregates as an indicator. The size distribution can be represented by a singular parameter, mean weight diameter (MWD)66, which is defined and calculated (equation (1)) as the sum of the mean diameter of each aggregate fraction (Supplementary Section B) multiplied by the mass contribution of that fraction to the total recovered weight of the soil.

$$\mathrm{MWD}=\mathop{\sum }\limits_{i=1}^{n}{x}_{i}{w}_{i}$$
(1)

All samples were fractioned into four aggregate size classes67: large macroaggregates (>2 mm), small macroaggregates (2–250 µm), large microaggregates (250–53 µm) and a silt and clay sized fraction (<53 µm). Approximately 10 g of soil sample were loaded into a sieve tower automated to rise and fall by 1 cm for 130 cycles while submerged in deionized water.

In analyzing the subsoil materials, we measured three states: (1) dispersed samples of the excavated subsoil to give context for the innate particle size distribution of the soil material (dispersion via Na-Pyrophosphate shaken overnight), (2) the sieved, excavated subsoil before incubation and (3) the incubated subsoil samples. The prior two treatments were lightly wetted 30 minutes before the analysis to bring the soil to a hydrated state. Otherwise, to minimize the effect of a drying and wetting cycle in structure formation, samples from the incubation, including all amended mixtures, were analyzed fresh within 2 weeks of harvest (n = 4).

Lastly, to assess the change in free mineral surface area, SSA was measured a second time (under the same parameters) using two incubated samples of each soil.

Multifunctionality score calculation

We conducted a quantitative scoring to evaluate the relative multifunctional potential of our organic amendment combinations. Using an averaging approach68, we compute an aggregate metric of multifunctionality after estimating scores for the individual functions of increased soil fertility, pollutant-retention capacity and contribution to the accrual of stable organic carbon. Individual function scores were calculated from a series of measured variables chosen based on their relation to the desired soil function. Each variable dataset and function score were standardized to the proportion of the maximum value within that dataset. The increased soil fertility attributed to each organic amendment combination was calculated based on the increase in nitrogen content, the increase in cation exchange capacity and the increase in water content at field capacity. In turn, the pollutant-retention capacity was calculated based on the retention capacity for heavy metals, zinc and copper and organic pollutants, tetrabutryn and mecoprop. Lastly, contribution to the accrual of stable organic carbon was determined by multiplying the input of each type of carbon by the inverse of the hydrogen to organic carbon ratio, a common determinant for the degree of aromatization and soil carbon chemical persistence22,69,70. As no significant difference was detected between the capacity of amended substrates and the subsoil material in inducing larger water-stable aggregates as an indicator for structural stability and subsequent influence on infiltration this function was left out. All functions were weighted equally within the index.

Statistical analyses

Version 2023.12.1.402 of RStudio71 was used for data handling (via writexl, v.1.5.4 and readxl, v.1.4.5)72,73,74,75, statistical analyses (via car, v.3.1 and lsmeans, v.2.30) and data visualizations (via ggplot2, v.3.5.2)76. All categorical factors were test via one-sided ANOVA after checking for homogeneity of variance and for normality within the residuals using Levenes test and Shapiro–Wilks tests, however final judgments were made via visual aids such as bar plots and QQ plots. A Tukey post-hoc analysis was used in comparing the pollutant-retention potential of organic parent materials. Due to the large number of possible statistical combinations, orthogonal contrasting was used as an alternative post-hoc test in all other cases (Supplementary Section B). We did this to add strength to our analyses, testing only the statistical combinations supporting our research questions: significant changes in mixture properties per addition of organic amendments compared to the subsoil; significant changes in biochar-amended mixture properties compared to the municipal compost control; significant changes in biochar-amended mixture properties compared to the granulated activated carbon control; and significant differences seen between a singular biochar mixture compared to the related dual biochar mixtures. Detailed results regarding the degrees of freedom, F-values, t-ratios and p-values are available for the ANOVA, Tukey and contrasts in Supplementary Section B. The dataset supporting the results of this study are available on request from the corresponding author.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.