1748-9326/20/11/114090

Abstract

Recycling organic waste products (OWPs) is known to influence soil physical, chemical, and biological properties, yet few studies have compared the long-term effects of different OWP type across multiple sites. This study examined the impacts of repeated OWP application on soil properties in two French long-term field experiments: QualiAgro and PROspective (20 and 18 years, respectively). The OWP included dehydrated urban sewage sludge (SLU), green waste and SLU compost, biowaste compost from source-separated municipal organic waste co-composted with green waste, municipal solid waste compost, farmyard manure from a dairy cow farm (FYM), and composted FYM from open-air composting on a concrete platform. The application of OWP led to increased soil nutrient levels and trace element availability, and stimulated microbial biomass and enzyme activities, while the response of nematode varied depending on site and OWP type. Biological properties were less affected than physico-chemical properties, though the OWP application enhanced soil microbial biomass and specific enzyme activities. The impact on soil nematode communities varied depending on OWP type and site. Strong correlations were observed among soil property changes, with exogenous carbon and nutrient inputs from OWP identified as key drivers. Larger changes were noted in QualiAgro, where OWP application rates were higher and initial soil quality lower. These findings highlight that OWP applications, depending on their type, rate, and initial soil conditions, can significantly alter soil properties.

Export citation and abstractBibTeXRIS

Organic waste (OW) is increasingly generated worldwide due to population growth, urbanization, and intensified industrial activities, with current estimates exceeding 2.2 billion tons of municipal solid waste (MSW), 13 billion tons of livestock manure, and 53 million tons of dry sewage sludge (SLU) annually, and further increases projected (Nanda and Berruti 2021, Feng et al 2023). The growing volume of OW products (OWP) offers significant opportunities for sustainable agriculture, as its application can sustain crop productivity by supplying nutrients while reducing the environmental impacts of mineral fertilizers (Chen et al 2022). While OWP enhance soil organic matter, improve the availability of major nutrients (N, P, K), and promote soil aggregate stability (Diacono and Montemurro 2010, Peltre et al 2017), they may also introduce trade-offs such as increased trace element accumulation and leaching (Araújo et al 2019, Chen et al 2025), highlighting the need for comprehensive evaluation of their impacts on soil health and ecosystem functions.

Soil biological indicators, including microbial communities, enzymes, and nematodes, are supposed to be more sensitive to environmental changes than physico-chemical properties and play crucial roles in nutrient cycling and soil ecosystem functions (Böhme et al 2005, Lazarova et al 2021). Numerous studies from various regions have confirmed that soil biological indicators are greatly influenced by fertilization management (Nahar et al 2006, Chang et al 2007, Li et al 2008, Puissant et al 2021). Assessing the dynamics of soil biological and physico-chemical properties under OWP application provides valuable insights into ecosystem responses to agricultural practices and ensures the preservation of soil ecosystem services (Garbuz et al 2021).

The benefits of OWP vary with type, application rate, climate conditions, and initial soil properties (Gómez-Muñoz et al 2017, Sharma et al 2019). However, few studies have systematically assessed how different OWP types and application rates jointly affect soil biological and physico-chemical properties over the long term across contrasting soils. Understanding these effects is critical for optimizing nutrient recycling, enhancing soil fertility, and minimizing risks such as trace element accumulation. This study analyzes two long-term field experiments in contrasting soils in France to quantify the impacts of repeated OWP applications on soil chemical, physical, and biological properties. By linking cumulative OWP fluxes with observed soil responses, the study identifies both site-specific and generalizable effects, providing actionable guidance for sustainable OWP recycling and soil management.

2.1. Field experiments

This study utilized soil samples from two French long-term field experiments within the SOERE-PRO network: PROspective (Colmar, northeastern France, established in 2000) and QualiAgro (Feucherolles, northwestern France, established in 1998). PROspective features Calcic Cambisol (IUSS Working Group WRB 2014) with a semicontinental climate and a maize-wheat-sugarbeet-barley rotation. QualiAgro comprises Horthic Luvisol soil with an oceanic climate. Crop rotation is based on various cereals (wheat, barley, rye) and maize rotation in half of the experiment (QUAΔN), while leguminous crops (alfa alfa, faba bean) are also used in the other half (QUAΔLEG) since 2014. Both sites implemented annual plowing (28 cm depth). All crop residues were incorporated in PROspective, while cereal straw were harvested until 2013 in QualiAgro. Although the two sites differ in location, climate, soil type, and crop rotation, their inclusion offers complementary contexts for evaluating the long-term effects of OWP under contrasting agro-environmental conditions.

The PROspective experiment comprised a control without OWP application (CON) and five OWP treatments: dehydrated urban SLU, green waste and SLU compost (GWS), biowaste compost (BIO), farmyard manure (FYM), composted FYM (FYMC). All OWP were applied every 2 years since 2000 at the same nitrogen rate per application (170 kg N ha−1) and combined with two mineral nitrogen supplementation levels: (1) no mineral nitrogen (COLΔN−) and (2) optimal mineral nitrogen (COLΔN+). The QualiAgro experiment comprised a CON and four OWP treatments: BIO, GWS, FYM, and MSW compost, which were applied since 1998 every 2 years at the same C amount per application (4 t C ha−1), with two fertilization levels: (1) optimal mineral nitrogen (QUAΔN) and (2) low mineral nitrogen (QUAΔLEG). Both experiments were arranged in a randomized block design with four replicates, with each block containing all OWP treatments and the corresponding control (CON) (figure S1). Changes in the experimental designs occurred in 2014, with detailed descriptions of management and the implemented modifications available in the supplementary material and cited (Cambier et al 2019, Chen et al 2022, 2024).

Specifically, at the PROspective site, SLU, FYM and FYMC had significantly higher C content than GWS and BIO (370, 399, 348, 291 and 258 g kg−1 DM, respectively) (table S3). SLU had also a significantly higher total N content compared to other OWP (60 g kg−1 DM and 24 g kg−1 DM for SLU and the mean of the four other OWP, respectively). SLU and GWS had significantly higher P content compared to other OWP, while FYM and FYMC had significantly higher K content.

At the QualiAgro site, MSW and FYM had significantly higher C content than GWS and BIO (314, 327, 266 and 227 g kg−1 respectively). GWS had also a significantly higher N and P content, while FYM was more enriched in N and K, compared to BIO and MSW. Detailed physicochemical properties and application rates of the OWP are provided in the supplementary material (tables S1–S5).

2.2. Soil analysis

Detailed soil analysis methodologies are available in supplementary material. In brief, composite soil samples were collected from the 0–28 cm (plow layer) in both sites, with sampling conducted in March at the PROspective site and in April at the QualiAgro site. For each composite sample, at least five individual cores were collected randomly using an auger and homogenized. Chemical properties, including pH, total C and N, cation exchange capacity (CEC), Olsen-P (AP), and exchangeable K (EK), were analyzed using ISO standard methods. In PROspective, soil organic carbon (SOC) was specifically measured instead of total carbon (Chen et al 2022). Active and stable SOC proportions were determined using the PARTYsoc v2.0 model (Cécillon et al 2021) based on thermal analysis performed with Rock-Eval® 6 (Vinci Technologies, France). Extractable trace elements (Cd, Cr, Cu, Ni, Pb, Zn) were analyzed using EDTA chelated, as they better reflect soil quality and toxicity concerns (Gupta and Sinha 2007).

As physical soil properties, soil aggregate stability was assessed through fast wetting, wet stirring, and mechanical breakdown tests (Le Bissonnais 1996).

Concerning biological properties, enzymatic activities (phosphatase, arylsulfatase, β-glucosidase, urease, arylamidase) were measured per ISO 20130 and expressed as mU g−1 corresponding to nanomoles of product released per minute per gram of dry soil (Cheviron et al 2022). Microbial diversity (bacteria and fungi) was analyzed using DNA sequencing. Microbial DNA was extracted following the standardized procedure ISO 11063 (Plassart et al 2012), and the DNA concentrations were used as estimates of soil microbial biomass (Dequiedt et al 2011). Bacterial and fungal diversity was characterized by OTUs richness, evenness, and Shannon index (Constancias et al 2015). Soil nematodes were extracted by Oostenbrink elutriation coupled with cotton-wool filter method (extraction ISO 23611-4) and identified by morphological identification under light microscope into: (1) bacterial feeders, (2) fungal feeders, (3) omnivore-predators, and (4) plant feeders following Yeates et al (1993). Enrichment index, structure index, maturity index, plant parasite index, nematode channel ratio and Shannon index were calculated to assess the nematode community (Ferris et al 2001, Yeates 2003).

2.3. Data analysis and statistical analysis

The statistical differences between each OWP treatment and its CON within the four experiments (COLΔN−, COLΔN+, QUAΔN, and QUAΔLEG) were evaluated using one-way ANOVA followed by Duncan’s test. Prior to analysis, data were checked for normality (Shapiro–Wilk test) and homogeneity of variance (Levene’s test); non-parametric alternatives were employed if the assumptions were violated. To reduce spatial heterogeneity, OWP treatment was compared with the CON within the same block, and Pearson correlation coefficients were calculated based on these differences to explore interrelationships among variables.

The gradient boosting model (GBM) was applied to evaluate the influence of OWP on soil characteristics. GBM is an ensemble machine-learning method that builds multiple regression trees sequentially and combines them to improve prediction accuracy. It captures nonlinear relationships and interactions among variables without requiring strict distributional assumptions (Friedman 2001). The cumulative budget of C, N, P, and K (ΔCΔbudget, ΔNΔbudget, ΔPΔbudget, ΔKΔbudget) between the OWP treatment and the CON within the same block were included as predictors, while the relative changes in soil properties served as response variables. The cumulative budgets were calculated as the difference between inputs (OWP, mineral fertilizers, and residues) and outputs (crop uptake and harvested residues). Other potential outputs, such as gaseous emissions and leaching, were not included due to the lack of the measurements and their relatively smaller magnitude compared with fertilization and crop uptake (Kros et al 2011, Welikhe et al 2023). An individual GBM model was calibrated for each soil property to specifically assess its response to the nutrient budgets. Hyperparameters were optimized via grid search using cross-validation with RMSE as the selection metric, while model performance was ultimately evaluated through cross-validated (CV-R2). All analyses were implemented using the ‘caret’ and ‘GBM’ packages in R (v.4.2.1). Detailed statistical methodologies are available in supplementary material.

Long-term OWP application significantly influenced soil chemical, physical and biological properties at both sites compared to controls (figure 1). OWP application significantly increased the TC, TN, AP, EK, CEC, trace elements and alter soil pH in most treatments, aligning with numerous previous studies (Maillard and Angers 2014, Gross and Glaser 2021). For example, TC increased by a maximum of 73.0% with GWSΔLEG and TN of 71.6% with GWSΔN in QUA. Both active and stable C increased after OWP application, except for certain specific treatments, with a more pronounced rise in active C. Such an increase led to a decrease in the proportion of stable C, particularly at QUA, where reductions ranged from 16.7% to 29.9%, reflecting the stronger effect of OWP on labile C pools. This likely reflects OWP enhancing active C by providing labile C and stimulating microbial activity to convert plant residue into active C (Poirier et al 2013, Li et al 2018).

Figure 1. Overview of the effects of OWP applications on soil enzymes (n = 4), microbial communities (n = 3), nematodes (n = 4), and physical and chemical properties (n = 4) at the COL and QUA. Each panel shows the relative change of OWP treatments compared with their corresponding control plots sampled in 2018. Significance tests were conducted separately within each experiment (COLΔN−, COLΔN+, QUAΔN, and QUAΔLEG). Colors indicate the magnitude and direction of change (red = increase; blue = decrease), with darker shades representing larger changes. * indicates a significant difference from the control treatment without OWP (p < 0.05). ‘N.A.’ indicates that data were not measured. Mean characteristics of soil properties in the control treatments of QUA and COL were shown in table S8 and the value of the plots are shown in table S9 for COL and table S10 for QUA.

Download figure:

Standard image High-resolution image

The increase in aggregate stability was limited compared to previous studies (Zhang et al 2014, He et al 2018, Chen et al 2021), particularly the findings by Annabi et al (2011) at site QUA. This discrepancy may be attributed to the revised experimental protocol implemented at QUA, which involved the incorporation of all crop residues into the soil since 2014. Crop residue return is widely recognized to enhance aggregate stability by increasing organic matter inputs, stimulating microbial activity, and improving soil structure, which may have masked or diluted the contribution of OWP application (Baloch et al 2025). Additionally, aggregate stability is highly sensitive to seasonal moisture fluctuations (Algayer et al 2014). Treatment contrasts are typically more evident when soil water content is close to field capacity (Bottinelli et al 2017). Since the sampling in this study was conducted in early spring, when soil water content and water repellency were relatively low, potential differences among treatments may have been obscured by variations in soil water content. These results indicate that the response of aggregate stability to OWP application is dynamic, shaped by interactions among climate, management, temporal factors, and OWP inputs. Selecting the optimal sampling time is inherently challenging, as it requires a compromise between capturing the distinct seasonal dynamics of various soil properties (Wuest 2015). Consequently, the findings should be interpreted within the specific environmental and temporal context of the study.

Despite the significant increased microbial biomass, bacterial and fungal diversity remained largely unaffected, with the exception of a 37.8% decrease in fungal richness under FYMΔN− at COL. Enzyme activities, including β-glucosidase, urease, and arylamidase, were significantly enhanced by OWP treatments, with larger increases observed at QUA. Phosphatase activity increased only at QUA under GWS treatments, with rises of 35.4% and 46.7% observed in GWSΔN and GWSΔLEG, respectively. These results highlight the role of OWP in supplying labile C and modulating enzymatic functions (Tejada et al 2006). Nematode communities showed site- and OWP type-specific responses, and no significant changes for nematode-related indicators under COLΔN+. Bacterial-feeding nematodes increased under BIOΔN and FYMΔN in QUA, while fungal-feeding nematodes decreased by 51.8% under SLUΔN- in COL. Overall, nematode indices in COL remained largely stable, with maturity index, structure index, enrichment index, and Shannon index showing minimal variation. Notably, under FYMΔN in QUA, the enrichment index and Shannon index increased by 23.0% and 9.4%, respectively, while the maturity index declined slightly by 11.0%, suggesting that FYM application promotes a well-structured and complex soil nematode community (Liu et al 2016). Overall, most nematode indicators did not show significant responses to OWP application, partly due to high spatial variability. In Contrast, maturity index declined by 20.8%–24.3% in the QUAΔLEG experiment, accompanied by an average 89.2% reduction in the abundance of omnivorous-predatory nematodes while enrichment index increase (+40.7% in average). These results suggest that nutrient recycling function, especially nitrogen, is stimulated in this soil. This result is consistent with the presence of a legume crop on the plot at sampling time (DuPont et al 2009, Villenave and Chauvin et al 2018).

In general, more significant changes were observed in QUA than in COL (figure 2(a)), with the significant effects mainly occurred in soil chemical properties (85.7% of measured indicators changed at QUA), while changes in soil biological indicators and aggregate stability were less pronounced. The stronger effects in QUA likely resulted from its higher OWP input (4 t C ha−1 vs. ∼2 t C ha−1 in COL), supplying more exogenous carbon, while crop residue export further emphasized OWP as the main carbon source. Additionally, site-specific responses were influenced by native soil properties, such as soil C and pH. Soils in QUA, characterized by lower baseline soil C content, likely demonstrated greater sensitivity to changes in biological communities and structure, and the application of equal amounts of C from different OWP types further facilitated a clearer assessment of their impacts on soil biology and differentiation among OWP types (Yanardağ et al 2017, Sadet-Bourgeteau et al 2018). In contrast, soils in COL was rich in calcium carbonate leading to an alkaline pH, which could influence microbial communities and therefore result in the different biological responses (Lejon et al 2007). For instance, OWP inputs increased arylsulfatase activities in COL whereas not in QUA, probably due to the critical role of calcium in regulating arylsulfatase activity (Chen et al 2019).

Figure 2. (a) Proportion of cases where significant differences were observed between OWP treatments and their respective controls, expressed relative to the total number of comparisons for each soil property (a value of 100% indicates that all measured properties responded significantly to all OWP treatments). (b) Correlation matrix of soil physical, chemical, and biological properties. Positive correlations are shown in blue, negative correlations in red; color intensity indicates the strength of the correlation. P-values for each correlation coefficient are indicated: *P < 0.05, **P < 0.01, ***P < 0.001. (c) Proportion of statistically significant correlations among soil properties relative to the total number of relationships (100% indicates that all correlations were significant).

Download figure:

Standard image High-resolution image

Changes in soil properties showed strong correlations (figure 2(b)). TN and TC were related to most of the soil biological properties. Notably, a significant positive correlation was observed between soil microbial biomass and active soil C, rather than stable C (figure 2(b)), highlighting active soil C serves as an energy source and a critical limiting factor for microbial growth and proliferation (Zhang et al 2022). Trace elements exhibited the strongest association with other soil properties (figure 2). This was likely primarily due to trace elements originating from OWP application, as the simultaneous increase with OWP input. However, contrary to previous reports suggesting that enzymatic activity diminishes with increasing trace elements concentrations and availability (Moreno et al 2003), our results indicated that trace elements did not negatively affect phosphatase, β-glucosidase, arylsulfatase, and urease, only arylamidase activities were negative correlated with trace elements availability (figure 2(b)), likely due to the concentration-dependent negative effects of trace elements being offset by the positive influence of OWP application (Albiach et al 2000).

The GBM model performance in predicting soil property changes based on element flux budgets and their relative contributions is shown in figure 3. While the model accurately predicted soil chemical properties, its performance was moderate to low for enzyme activities, microbial, and nematode indicators. This discrepancy likely reflects the more direct link between element flux budgets and soil chemical processes (Chen et al 2022). In contrast, biological indicators are shaped by a wider range of ecological and temporal factors, and therefore may not follow a single trend over time (Griffiths et al 2001, Chernov and Zhelezova 2020).

Figure 3. The R2 of test dataset in GBM and the relative contributions (%) of predictive variables (total P, K, N, and C budgets) for the GBM model of (a) soil chemical characteristics (b) soil nematodes (c) soil enzyme activities and (d) soil microbial.

Download figure:

Standard image High-resolution image

The C and N budgets were critical for predicting changes in soil chemical properties, explaining up to 87% of the variation in CEC and more than 50% of the variance in soil enzyme and nematode indices. Soil AP and EK were predominantly determined by the P and K budgets, respectively. For enzyme activities, the C and N budgets remained key factors: the C budget primarily influenced phosphatase and β-glucosidase, while the N budget was the major driver for urease and arylamidase. C and P budgets were the main explanatory factor for Shannon and Simpson’s diversity indices of enzyme activities. The C budget also played a dominant role in determining the nematodes enrichment index, maturity index, and accounting for the vast majority of the variation in soil microbial biomass. However, microbial richness and diversity showed a different pattern. While C and N budgets strongly predicted soil physicochemical properties and enzymatic activities, they accounted for little of the variance in bacterial richness and Shannon diversity. Instead, K and P budgets contributed more substantially to these microbial indicators, highlighting the nuanced nutrient dependencies across different soil biological domains.

Long-term OWP application significantly alters soil chemical, physical, and biological properties, with effects largely influenced by application rates, OWP types, and native soil conditions. While improvements in soil chemical properties were consistent, enhancements in biological properties and aggregate stability were more spatially variable and less statistically robust. Indicators of biological activity (enzyme activity, microbial biomass, nematodes abundance) were mostly improved by OWP application while indicators of biological diversity were less affected. This suggests that while OWP provides resources for soil organisms, it may not sufficiently improve habitat conditions to enhance biodiversity. Therefore, complementary practices such as no tillage and crop diversification are likely needed to promote soil structure and biological diversity. Element flux budgets of OWP effectively explained shifts in soil chemical properties, though the improvement effect of OWP was limited in soils with high initial SOM, particularly when crop residues were also incorporated. Despite being a common concern, the increased availability of trace elements still resulted in low concentrations that did not adversely affect soil biological activity.

We thank the technicians responsible of the QualiAgro and PROspective experiment management. The QualiAgro and PROspective field experiment is part of the SOERE-PRO (network of long-term experiments dedicated to the study of impacts of organic waste product recycling) integrated as a service of the ‘Investment in the Future’ infrastructure AnaEE-France, overseen by the French National Research Agency (ANR-11-INBS-0001). This work benefited of the use of the Biochem-Env platform (DOI 10.15454/HA6V6Y, UMR 1402 ECOSYS, INRAE IdF Versailles-Saclay centre), granted by the French government by the Agence Nationale de la Recherche under the France 2030 program, reference ‘ANR-24-INBS-0001’ AnaEE France. The QualiAgro experiment was founded and is still supported by INRAE and Veolia R&I. The PROspective experiment is supported by INRAE and SMRA68. This work was funded by Ademe (Graine PROTERR project). Haotian Chen gratefully acknowledged financial support from the Postdoctoral Fellowship Program of China Postdoctoral Science Foundation (GZC20252642).

The data that support the findings of this study are openly available at the following URL/DOI: https://doi.org/10.5281/zenodo.15584379.