Soil nutrient stoichiometry impacts on soil organic carbon stocks in long-term phosphorus fertilisation experiments

Soil nutrient stoichiometry impacts on soil organic carbon stocks in long-term phosphorus fertilisation experiments Parag Bhople a,* , David Wall a , Karl Richards a , Timothy Clough b , Fiona Brennan a , Gary Lanigan a , Mart Ros c , Anke M. Herrmann d , Ingeborg F. Pedersen e , Lars Elsgaard e , Naoise Nunan e,f , Christoph Müller g,h,i , Kristina Kleineidam g , Sergio E. Morales j,k , Daniel Goll l , Giulia Bondi a a Crops Environment and Land Use Department, Teagasc, Johnstown Castle, Y35 TC97 Wexford, Ireland b Department of Soil and Physical Sciences, Lincoln University, Lincoln 7647 Christchurch, New Zealand c Wageningen Environmental Research, Wageningen University & Research, PO Box 47, 6700 AA Wageningen, the Netherlands d Swedish University of Agricultural Sciences, Department of Soil and Environment, PO Box 7014, SE-750 07 Uppsala, Sweden e Department of Agroecology, Aarhus University, Blichers Alle 20, DK-8830 Tjele, Denmark f Institute of Ecology and Environmental Sciences, CNRS-Sorbonne Universit´ e-IRD-UPEC-P7-INRA, Paris, France g Institute of Plant Ecology (IFZ), Justus Liebig University Giessen, Heinrich-Buff-Ring 26, 35392 Giessen, Germany h School of Biology and Environmental Science and Earth Institute, University College Dublin, Dublin 4 Belfield, Ireland i Liebig Centre for Agroecology and Climate Impact Research, Justus Liebig University Giessen, Germany j Department of Microbiology & Immunology, University of Otago, 720 Cumberland Street, PO Box 56, 9054 Dunedin, New Zealand k MPG Ranch, Florence, MT 59833, USA l Universit´ e Paris Saclay, CEA-CNRS-UVSQ, LSCE/IPSL, Gif-sur-Yvette, France ARTICLE INFO Handling Editor: Dr. Andrew Margenot Keywords: Soil carbon storage C:N:P stoichiometry Long-term field experiments Agroecosystems management Climate change ABSTRACT Managing agroecosystems to enhance soil organic carbon (SOC) storage is important for mitigating climate change. However, the transformation of SOC is intimately connected to nutrient cycling, particularly nitrogen (N) and phosphorus (P). While P constraints on plant growth are known, their effects on carbon (C) and N cycling remain uncertain. The study uses several long-term experiments (LTEs) to determine the importance of N-P interactions and the optimal C:N:P stoichiometry for long-term SOC stocks in managed agricultural systems. The aim was to determine the influence of multi-decadal P fertilisation on SOC stocks and stoichiometric interactions of C, N and P in agricultural soils (up to 50 cm) across different soil textural classes and land uses. For this, the soils were sampled at three depths 0–10, 10–30 and 30–50 cm from six LTEs in Europe (three grasslands and three arables) to determine soil physico-chemical properties. The results showed comparable SOC stocks in contrasting P treatments across land uses. In grasslands, SOC stocks at 0–50 cm depth ranged from 9.7 to 40.6 t C ha −1 while in arable sites, they were between 11.0 and 48.3 t C ha −1 . The SOC stocks did not vary significantly across P treatments indicating that long-term P fertilisation did not affect C storage. Grassland sites had higher SOC stocks in the 0–10 cm, while at arable sites they were higher at 10–30 cm depths. The maximum predicted SOC stock of 30.9 t C ha −1 was with SOC/TN (total nitrogen) ratio of 10.1 and SOC/TP (total phosphorus) ratio of 32.6 in grassland sites, while these ratios were 10.9 and 29.4, respectively, in arable sites, where the predicted maximum SOC stock was 33.3 t C ha −1 . Overall, the study shows that the long-term phosphorus fertilisation of grassland and arable soils did not affect SOC stocks at the studied LTEs. 1. Introduction In the terrestrial biosphere, soils are the largest reservoir of carbon (C) and agroecosystems function as one of the most important C sinks (Lal, 2018). Capturing atmospheric CO 2 via plant photosynthesis and storing organic C in soils can mitigate atmospheric greenhouse gas concentrations and contribute to climate regulation (Coonan et al., 2019). However, the benefits of soil organic carbon (SOC) sequestration * Corresponding author at: Teagasc, Crops Environment and Land Use Department, Johnstown Castle, Wexford Y35 TC97, Ireland. E-mail address: [email protected] (P. Bhople). Contents lists available at ScienceDirect Geoderma journal homepage: www.elsevier.com/locate/geoderma https://doi.org/10.1016/j.geoderma.2025.117538 Received 7 March 2025; Received in revised form 11 September 2025; Accepted 1 October 2025 Geoderma 463 (2025) 117538 Available online 29 October 2025 0016-7061/© 2025 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). go beyond reducing atmospheric CO 2 concentrations alone. Sequestering more C in soils as soil organic matter (SOM) contributes to enhancing soil structure by fostering aggregate formation, increasing porosity, increasing soil stability and cation exchange capacity, along with improved water infiltration and retention (Lal, 2018). These physico-chemical improvements enable soils to retain nutrients essential for plant growth while the enhanced SOM also supports diverse soil microbial communities, which play a vital role in nutrient cycling and organic matter decomposition (Kirkby et al., 2011). However, several factors, such as the climate, soil type and land use change, can influence the process of C sequestration in agroecosystems (Lal, 2018). Furthermore, management practices related to fertiliser use, residue inputs, and tillage intensity also affect C sequestration (Lal, 2018; Stockmann et al., 2015). Soil C sequestration is closely linked to the cycling of nitrogen (N) and phosphorus (P), which are the main nutrients in the soil that regulate plant productivity and microbial activity. Furthermore, the stoichiometric ratios of C, N and P (C:N:P ratio) influence the formation of SOM (Kirkby et al., 2014). An optimal balance of these elements is therefore crucial for maintaining soil fertility and the long-term storage of C. The availability of P has a crucial impact on soil C dynamics (Kirkby et al., 2014), particularly when it is a limiting nutrient (Spohn, 2020). However, elevated P levels may also be associated with an increase in microbial biomass and respiration (Spohn and Schleuss, 2019), which accelerate the decomposition of SOM and potentially offset C sequestration by releasing CO 2 back into the atmosphere (Kirkby et al., 2014; Coonan et al., 2019). This dual role of P in promoting SOM build-up and mineralisation highlights the complexity of interactions between nutrient-driven productivity and SOC persistence in soils. Long-term P fertilisation experiments (LTEs) offer the possibility to study the distinct relationships between P, N and soil C dynamics. For example, in the European context, moderate P inputs (20–30 kg P ha −1 yr −1 ) have been shown to enhance productivity, while excess P (>50 kg P ha −1 yr −1 ) increased SOM decomposition via microbial respiration in grassland soils in Ireland (Graça et al., 2022). It is important to note that short-term priming effects do not necessarily translate into higher longterm net SOM decomposition. Another Irish LTE demonstrated that organic P inputs stimulated microbial activity, leading to enhanced N mineralization and short-term positive priming responses. However, over the long term, these treatments resulted in lower net SOM decomposition compared with inorganic P inputs, which showed negative priming effects (Kelly, 2022). In a grassland study in the Netherlands, abundant N and continuous P enrichment stimulated microbial activity resulting in higher SOM turnover (Reijneveld et al., 2009). Conversely, LTEs in arable lands in Sweden showed that moderate P and N input supported plant growth and slower SOM decomposition, while additional P (>40 kg P ha −1 yr −1 ) amplified priming effects and SOM loss in presence of high N contents in soils (Bergkvist et al., 2011). In a Danish arable site under P and N inputs, SOC stocks decreased due to increased microbial biomass and respiration rates (Liang et al., 2019). Similar effects are also found outside of Europe. For example, a grazed pasture LTE study in New Zealand found that moderate P rates (20–30 kg P ha −1 yr −1 ) resulted in more stable SOM, but that excess P resulted in a net loss of C from soils (Condron and Goh, 1989). Although the understanding of interactions between nutrients (C, N and P) in topsoils has progressed, less is known about the processes in subsoils. The distribution of N varies throughout the soil profile depending on N inputs from fertiliser and residues, as well as processes such as N deposition, decomposition of litter, microbial N fixation, leaching and denitrification (Chen et al., 2014; Rumpel et al., 2015; Peixoto et al., 2021). Moreover, due to limited root activity, reduced microbial biomass and rhizodeposition, soil fertility usually decreases with increasing depth, as indicated by higher soil C/N ratios (Kirkby et al., 2011; Bhople et al., 2019, Liang et al., 2019). The availability of P in agricultural soils strongly depends on its presence in soil minerals and its adsorption to reactive surfaces in soil (Graça et al., 2022). Most of the microbial and plant activity is concentrated in topsoils, associated with readily available labile P pools. In subsoils, extending below 30 cm depth, P mobility and availability decreases due to stronger adsorption to soil particles leading to P stratification throughout the soil profile in combination with changes in soil pH determining the bioavailability of P in soils (Lorenz and Lal, 2005; Rumpel et al., 2015; Liang et al., 2019; Graça et al., 2022). Furthermore, the stoichiometric ratio of soil C:N:P is not constant with soil depth making the role of P in sub-soil C sequestration unclear. In this regard, long-term P fertiliser experiments have Table 1 Long-term experimental sites (LTEs) involved in the study and their main characteristics. Country Ireland The Netherlands Sweden Denmark Institute Teagasc Wageningen University Swedish University of Agricultural Sciences Aarhus University Trial name Cowlands (JC1) Dairy-cut (JC2) Lelystad (LE) Lanna Skara 1 (LS1) Lanna Skara 2 (LS2) Jyndevad (JY) Land Use Grassland Grassland Grassland Arable Arable Arable Soil disturbance Grazing regime silage Combination of silage and rotational grazing Crop rotation, annual inversion tillage Crop rotation, annual inversion tillage Annual ploughing Location 52.3058 N 52.2979 N 52.52 N 59.8761 N 59.8761 N 54.8896 N 6.5080 W 6.4994 W 5.55 E 17.9533 E 17.9533 E 9.1276 E Year of start 1968 1995 1989 1936 1941 1944 (liming 1942) Soil type-textural class Humic Gleysol − Sandy clay loam Stagnic Cambisol −Clay loam Gleyic Fuvisol Clay Udertic Haploboroll (USDA Soil; Taxonomy) – Silt clay loam Udertic Haploboroll (USDA Soil; Taxonomy) – Silt clay loam Orthic Haplohumod −Coarse sand Phosphorus (P) fertiliser Calcium superphosphate Triple superphosphate Cattle manure + Superphosphate Superphosphate Superphosphate Superphosphate (until 2012) and subsequently as Triple superphosphate P fertiliser management One application annually One application annually Two applications annually One application every sixth year One application every sixth year. One application annually P treatments Low High Low High Low High Low High Low High Low High P rates (kg ha −1 ) 0 30 0 45 23 41 0 105 0 105 0 15.6 P Morgan (mg L -1 ) 1.19 6.47 1.82 4.63 12.30 20.47 0.81 1.62 0.62 1.11 2.10 12.48 Typical average crop yield (kg DM ha −1 ) 3831 3785 4060 4540 9888 10,952 5484 6731 3266 5461 3674 4340 P contents of harvested biomass (g P kg −1 DM) 2.9 3.5 2.7 3.2 3.5 3.5 3.1 3.9 3.1 3.9 3.0 3.3 P. Bhople et al. Geoderma 463 (2025) 117538 2 demonstrated that P addition can increase labile P (based on the Olsen P index) in topsoil under grazed pasture management, while in subsoils P remained tightly bound to mineral surfaces thereby reducing its availability in free form (Coonan et al., 2019; Reijneveld et al., 2009). The N and P availability co-limits C mineralisation in deeper soils where the C/P ratio is also higher (Peixoto et al., 2021). High C/P ratio may cause microbes to immobilise P, while simultaneously, low N availability can limit microbial activity making both N and P co-limiting factors for C mineralisation in deeper soils. Conversely, with a low C/P ratio, the mineralisation of organic P seems to be stimulated, increasing P availability in soils and simultaneously accelerating both the microbial decomposition of SOM and the metabolic utilisation of C and P (Fontaine et al., 2007; Kirkby et al., 2014; Peixoto et al., 2021). Although subsoils are critical long-term sinks of C (Coonan et al., 2019), the SOC stock changes as a function of depths and the potential interactions with other nutrient ratios remain poorly explored, highlighting the need for an integrated approach to studying soil nutrients and how their stoichiometric ratios influence the C storage potential of soils. The current study investigates the distribution of SOC stocks and C: N:P stoichiometry in soil profiles from six long-term experimental sites (LTEs) in Europe. The six LTEs represent distinct phosphorus (P) fertilisation levels (high and low) and two land use types (grasslands and arable). The use of LTEs and associated P legacy data are useful for investigating C storage dynamics in response to long-term P inputs as well as to explore topsoil and sub-soil SOC storage potential (Lorenz and Lal, 2005; Kell, 2011; Lynch and Wojciechowski, 2015). We tested the following hypotheses: 1) The long-term P fertilisation inputs influence SOC stocks across soil depths, with higher P fertilisation leading to greater accumulation of SOC compared to the low P inputs; 2) The variations in SOC stocks are associated with soil C:N:P stoichiometry, reflecting potential interactions among C, N and P in studied agroecosystems. The specific objective of the study was to evaluate the effects of long-term P fertilisation regimes on SOC stocks and to examine their relationships with soil C:N:P stoichiometry across different landuses and soil depths. 2. Materials and methods 2.1. The long-term experimental sites and P management This study was carried out using LTEs in Europe with contrasting P fertiliser treatments that have been running continuously for 29–89 years (Table 1). The LTEs covered six field sites located in Ireland (JC1, JC2), The Netherlands (LE), Sweden (LS1, LS2) and Denmark (JY). All sites are in the cool temperate climatic zone of the Northern hemisphere. The LTEs included three grassland sites (JC1, JC2, LE) and three arable sites (LS1, LS2, JY). The grassland sites were subjected to grazing only, cutting only or a combination of both. Soil texture at the grassland sites ranged from sandy clay loam to clay. The arable sites were tilled inversely or ploughed annually, and the soil texture ranged from coarse sand (JY) to heavy clay soils (LS1 and LS2). The LTEs have been subjected to controlled P fertiliser applications. At site JC1 and JC2, calcium superphosphate and triple superphosphate were applied as P fertilisers with one application per year generally in early spring. There were also P returns through grazing at JC1. At the LE site, a combination of cattle manure and superphosphate was applied twice per year, in early spring and after every harvest. A high dose of P fertiliser was applied every sixth year at sites LS1 and LS2, while at the JY site, mineral P fertiliser was applied once per year as super phosphate until 2012 and subsequently as triple superphosphate. In this study, we have used total P (TP) as an indicator of legacy P accumulation from long-term P fertilisation inputs, to provide a consistent measure for cross-sites comparison in the LTEs. Two P levels were selected from each site were categorized nominally as low and high P. In our study, the ‘low’ and ‘high’ P also refers to the P content of the soil rather than the amendments. At all sites, the low P application consisted of zero P (0 kg ha −1 ) except at the LE grassland site where the low P application was 23 kg ha −1 yr −1 . The high P application rate varied across different LTEs (Table 1) due to the experimental designs, differences in soil types and the P requirements for plant growth. At the three grassland sites (JC1, JC2 and LE), the high P application consisted of 30, 45 and 41 kg ha −1 yr −1 respectively. At the three arable sites, the high P application consisted of 105 kg ha −1 applied every 6th year at the LS1 and LS2 sites, and 15.6 kg ha −1 yr −1 for JY. The JY site also received one application of 156 and 263 kg P ha −1 in 1944 and 1972, respectively to improve their overall P status (Pedersen et al., 2025). At the JY site soils were sampled in treatment plots receiving highest liming rates of 12 Mg ha −1 every 6 to 10 years. All P treatments were replicated three times across the sites JC1, JC2 and JY. At site LE, only one treatment replicate was available and at LS1 and LS2 two replicates were available for high and low P treatments. 2.2. Soil sampling Samples from the two P fertilisation treatments (high and low) were collected at the six LTEs in September 2022, within each plot (30 m × 30 m =900 m 2 size, except for the JY site where the plot area was 90 m 2 ) a representative sampling area was selected. By tracing the diagonals of the sampling area, a soil pit (approx. 60 ×60 ×60 cm) was excavated at the identified mid sampling point. The profile face in the pit was cleaned with a knife/trowel and a composite sample of 500 g was taken from three defined depths, namely 0–10 cm, 10–30 cm and 30–50 cm. Within a plot, two composite samples were pooled as a representative of a specific depth and treatment. This resulted in 84 soil samples in total. Soil samples for bulk density (BD) were collected by inserting steel rings (Ø 4.5 cm; height =4.5 cm) into the soil at depths of 0–10, 10–30 and 30–50 cm for all P treatment replicates. 2.3. Soil analyses The composite soil samples were transferred to the laboratory facilities in Teagasc Johnstown Castle, Ireland. Roots and stones were removed from all soil samples. Later, the samples were dried at 40 ˚C and sieved (<2 mm mesh). The soil pH was measured using deionised water (1:5 w/v) as described by Graça et al. (2022). A sub-sample of soil (0.2 g of dried ballmilled soil) was used to measure soil total carbon (TC), total nitrogen (TN) and total inorganic carbon (TIC) contents using a TrueSpec C/N analyser (LECO, USA). The SOC contents were calculated as the difference between TC and TIC. Dried soil (5 g) was used for Morgan’s P extraction in a solution of 15 mL 0.62 M NaOH and 1.25 M CH 3 COOH (pH 4.8). The soil solution was shaken for 30 min, before filtering and quantifying on a spectrophotometer (Camspec, UK) to indicate the plant available P index (Graça et al., 2022). The Mehlich 3 extraction was performed using 20 mL Mehlich 3 reagent consisting of 0.2 M CH 3 COOH, 0.25 M NH 4 NO 3 , 0.015 M NH 4 F, 0.013 M HNO 3 and 0.001 M EDTA and 2 g of dried soil (Mehlich, 1984). The mixture was shaken at 180 rpm for 5 min and total available soil P was measured using Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES). From a 0.5 g dried soil sample, the total elements including P (TP) were measured using a reverse aqua regia digestion method followed by spectrometry measurement of the digestate (ISO 11466, 1995). In this method deionised water (2 mL), HCl (37 %, 2.5 mL) and HNO 3 (69 %, 7.5 mL) were added to the tube containing the dried soil sample and placed in a digester heated at 180 ◦C for 40 min. Upon cooling the samples were filtered (Whatmann no. 42), before ICP–OES analysis. In this study, the role of P was central and therefore we employed multiple approaches to quantify soil P. The plant-available P form was determined using Morgan’s and Mehlich-3 extractions, which provide indices of labile P forms relevant for uptake by plants. Furthermore, total P (TP) was measured using aqua regia digestion followed by ICP-OES, as a proxy to capture both organic and mineral-bound P forms. In this P. Bhople et al. Geoderma 463 (2025) 117538 3 study, TP was therefore selected as the principal metric for soil stoichiometric analysis considering that it integrates the cumulative impacts of long-term fertilisation and management on soil P status across depths and LTEs. While organic P (OP) is often considered to be more directly associated with biological processes, it also contains both labile and stable fractions with varying availability (Spohn, 2020; Barrow et al., 2021). Compared to this, TP consists of both organic and inorganic pools and thus reflects legacy P accumulation due to long-term fertilisation inputs (Kirkby et al., 2011; He et al., 2023). We indeed acknowledge that TP is not a direct substitute for OP, especially when assessing effects of short-term microbial processes and this represents a limitation of our study. Nevertheless, TP provided an analytically consistent and historically comparable measure across studied LTEs, which allowed us to evaluate cumulative P enrichment effects in relation to SOC stock changes (see Discussion for further caveats L513-535). The soil BD was calculated using dry weight (drying at 105 ◦C for 24–48 h) of the soil core samples divided by the volume of the soil core. The volume of stones or >2-mm fragments were subtracted from the total soil core volume to calculate the volume of soil (Walter et al., 2016). Using the recorded weights the BD (g cm −3 ) was calculated with following equation: BD (g cm −3 ) =[(NetDW −weight of stones −weight of wooden debris)/(V −Vs −Vw)] (1) Where, V is the inner volume of the soil BD ring (98.17 cm 3 ), Vs is the volume of >2 mm stones (g of stones/2.50 g cm −3 (avg. density of stones) and Vw is the volume of >2 mm wooden debris (b of wood/0.8 g cm −3 (avg. density of wood). 2.4. Calculations and statistical analysis The SOC stocks were calculated as described in Necp´ alov´ a et al. (2014) and represented in t C ha −1 (Equation 2). SOC stocks (t C ha −1 ) =SOC concentration (mg C g −1 dry soil) ×fine soil BD (g cm −3 ) ×Soil depth (thickness in cm) (2) Two-way analysis of variance (ANOVA) was carried out on all samples with P treatments as replicates and used to determine whether P fertiliser treatments and soil depths had significant (p <0.05) effects per individual site on: (i) soil physico-chemical properties; and (ii) SOC stocks (t C ha −1 ). Tukey HSD post-hoc test allowed multiple comparison of means at 95 % confidence interval, using R-package “multcomp” (Hothorn et al. 2008). The normality and heteroscedasticity in data were checked using residual distribution and, if violated, data were log/BoxCox transformed (R-package “MASS”; Ripley et al., 2011). Correlation analyses were conducted to determine the relationships between total N and P in the grassland and arable sites at both P levels, and between SOC stocks and total N and P contents in three soil depths for each land use. Principal component analysis (PCA) was performed using the Rpackage “vegan” (Oksanen, 2015). Two-way ANOVA was used for testing differences between depths and other parameters by log transforming the data when required. We fitted a polynomial regression (up to the third degree) to model the non-linear relationships between the predictor variables (SOC/TN and SOC/TP ratios) and the response variable (SOC stocks). The model included SOC stocks, SOC/TN and SOC/ TP data from all sites, P levels and depths. Following the model estimation, we applied a numerical optimisation method to identify the SOC/TN and SOC/TP ratios that would predict maximised SOC stocks with 99 % confidence intervals, considering all interactions in the data set. The optimisation process began with the formulation of an objective function intended to predict SOC stocks based on the polynomial regression model. We chose the “L-BFGS-B” method (Zhu et al., 1995)), which is appropriate for problems that have boundary constraints on the variables. To quantify the uncertainty associated with these optimal ratios, we calculated the standard errors using the inverse of the Hessian matrix obtained from the optimisation process. All statistics were performed using R version 4.4.1 (R Core Team, 2021). Table 2 Soil physico-chemical properties at grassland sites. Site P Depth BD pH SOC TN TP Mehlich3 P Morgan’s P SOC/TN SOC/TP SOC stocks treatments cm g cm −3 g kg −1 mg kg −1 mg L -1 t C ha −1 JC1 High 0–10 1.11 (0.08) a 5.24 (0.04) bc 26.94 (0.98) a 2.67 (0.10) a 1.06 (0.06) a 0.88 (0.62) b 11.37 (1.36) a 10.08 (0.01) a 25.53 (1.80) ab 29.86 (3.20) a 10–30 1.32 (0.01) a 5.40 (0.16) ac 10.54 (1.34) b 1.09 (0.12) bc 0.33 (0.07) b 29.78 (4.08) a 1.10 (0.24) bc 9.62 (0.53) a 31.84 (2.72) ab 27.74 (3.80) a 30–50 1.23 (0.17) a 5.78 (0.17) a 5.17 (1.82) c 0.54 (0.26) cd 0.23 (0.09) bc 15.23 (1.36) ab 0.64 (0.47) cd 10.13 (2.05) a 23.22 (6.61) b 12.28 (2.44) b Low 0–10 1.09 (0.07) a 5.02 (0.05) b 23.79 (2.20) a 2.44 (0.25) ab 0.35 (0.06) b 9.06 (7.84) ab 1.65 (0.35) b 9.77 (0.83) a 68.94 (17.91) a 25.99 (3.99) a 10–30 1.18 (0.29) a 5.38 (0.20) ab 10.87 (1.69) b 1.17 (0.18) ac 0.27 (0.04) bc 10.32 (8.14) ab 0.39 (0.01) cd 9.29 (0.44) a 40.21 (2.42) b 24.96 (3.51) a 30–50 1.56 (0.11) a 5.83 (0.23) a 3.13 (0.39) c 0.34 (0.19) d 0.15 (0.04) c 14.87 (6.95) ab 0.25 (0.12) d 10.90 (4.55) a 20.91 (4.30) c 9.69 (0.64) b JC2 High 0–10 1.11 (0.03) b 6.16 (0.02) a 22.95 (2.96) ab 2.27 (0.32) ab 0.74 (0.07) a 41.22 (19.75) a 2.84 (1.14) a 10.12 (0.27) a 30.76 (1.10) cd 25.57 (3.88) b 10–30 1.30 (0.12) ab 6.11 (0.08) a 13.02 (1.75) bc 1.38 (0.23) ac 0.47 (0.06) bc 40.54 (12.29) a 0.61 (0.37) b 9.51 (0.59) ab 27.58 (0.59) cd 33.80 (4.29) ab 30–50 1.29 (0.28) ab 6.21 (0.09) a 8.03 (3.97) cd 0.93 (0.47) cd 0.32 (0.13) bd 41.19 (22.78) a 0.27 (0.06) b 8.87 (2.10) ab 24.28 (2.66) d 19.77 (6.92) bc Low 0–10 1.11 (0.08) b 6.24 (0.23) a 27.45 (6.65) a 2.68 (0.57) b 0.59 (0.07) ac 18.84 (6.76) ab 2.15 (1.04) a 10.19 (0.30) a 46.35 (6.67) a 30.37 (7.64) ab 10–30 1.27 (0.05) ab 6.10 (0.09) a 14.22 (1.85) bc 1.57 (0.16) bc 0.41 (0.06) cd 2.48 (0.17) b 0.50 (0.15) b 9.04 (0.35) ab 34.87 (1.76) ab 36.16 (5.98) a 30–50 1.48 (0.06) a 6.21 (0.10) a 5.05 (0.38) d 0.69 (0.09) d 0.23 (0.01) d 35.91 (20.65) a 0.21 (0.02) b 7.32 (0.49) b 21.52 (1.11) e 14.92 (1.19) bc LE High 0–10 1.25 7.52 32.40 3.27 0.93 65.02 25.30 9.91 34.84 40.55 10–30 1.42 7.93 11.06 1.25 0.49 7.60 2.00 8.85 22.71 31.51 30–50 1.25 7.95 8.00 0.83 0.41 2.21 0.98 9.63 19.45 20.02 Low 0–10 1.34 7.63 29.74 2.96 0.76 33.47 13.70 10.05 39.10 39.92 10–30 1.41 7.89 9.45 1.09 0.45 2.48 1.05 8.67 20.95 26.70 30–50 1.20 7.96 11.31 1.25 0.49 5.80 1.65 9.04 23.22 27.17 In Table 2, values represent sample means (for each depth; n =3 at JC1, JC2; n =1 at LE). Different letters of significance according to Tukey’s posthoc test (p <0.05) follow standard deviations in parentheses. The comparisons were made between depths at high and low P treatments in each individual sites. Same letters indicate no statistical difference. P, phosphorus; BD, bulk density; SOC, soil organic carbon; TN, total nitrogen; TP, total phosphorus; SOC/TN and SOC/TP, ratios. P. Bhople et al. Geoderma 463 (2025) 117538 4 3. Results 3.1. Soil properties We estimated net phosphorus (P) balances (P input – P export in biomass) across all experimental sites to check if we can provide a more functionally relevant comparison than nominal “low” versus “high” treatments. Based on site-specific P inputs, observed data, and literaturederived values for dry matter yields and biomass P contents in studied agroecosystem (Graça et al., 2022; Poeplau et al., 2016; Bergkvist et al., 2011), the results of this analysis show that low P treatments were consistently in deficit across grassland and arable systems (−11 to −21 kg P ha −1 yr −1 ), indicating long-term depletion of soil P. The high P treatments led to surpluses in grasslands (ranging from +6 to +32 kg P ha −1 yr −1 ) and were nearly balanced in arable systems (ranging from −4 to +3 kg P ha −1 yr −1 ). This analysis highlights that net P addition may lead to legacy fertilisation effects and nutrient constraints influencing SOC storage patterns which can be explored more in future studies (Supplementary Table 2, Supplementary Fig. 1). With the exception of the JC2 grassland and the JY arable sites, the soil BD did not differ significantly between P fertiliser treatments at any given depth (Tables 2 and 3). In both management systems (grassland and arable), the BD ranged from 1.1 to 1.6 g cm −3 . The soils were generally acidic, with the pH ranging from 5.0 to 6.9 in all grassland and arable sites, except for the LE grassland (both P treatments, Table 2) and the JY arable site (Table 3), which had neutral to alkaline pH. These plots had been limed (12 t ha −1 every 7–10 years). The SOC, TN and TP contents were significantly higher in topsoil (0–10 cm) and decreased with depth at the grassland sites (JC1, JC2, LE). Overall, the SOC contents ranged from 32.4 g kg −1 in the topsoil (0–10 cm) to 3.1 g kg −1 at a depth of 30–50 cm across grassland sites (Table 2). The TN and TP contents also decreased significantly with depth at all grassland sites and ranged from 3.3 to 0.3 g kg −1 TN and 1.0 to 0.2 g kg −1 TP. At the arable sites, contents of SOC and TN were significantly lower in the deepest soil layer (30–50 cm). The SOC contents ranged from 18.1 to 4.4 g kg −1 and TN contents from 1.7 to 0.5 g kg −1 through all depths for arable sites (Table 3). The TP content varied inconsistently through the soil depths across the arable sites in the deepest soil layer. At the JY site, the TP ranged from 0.15 to 0.2 g kg −1 in 30–50 cm depth across P treatments. At the LS sites only 2 soil replicates were available and represented a much wider range (0.85–1.6 g kg −1 ) in TP at 30–50 cm depth at LS1 and LS2 sites. At these sites, the low P treatments in particular, had much higher TP levels than expected in the 30–50 cm soil layer, which could be due to uneven distribution of P to this layer as a result of successive full inversion tillage events (Table 3). No consistent trend was observed in the available P contents (Mehlich3 extract) in the grassland sites JC1 and JC2 or in the arable sites LS1 and LS2, even with different P treatments. However, available P consistently decreased with increasing soil depths at LE and JY sites. In the grassland sites, the available P contents ranged from 0.9 to 65.0 mg kg −1 , while in arable sites they were between 1.6 and 205.2 mg kg −1 (Tables 2 and 3). The plant available P (Morgan’s extract) contents consistently decreased from top soil (0–10 cm) to deeper soil depths (10–50 cm) at both grassland and arable sites except for sites JC2 and LS2. The available P contents were generally higher in high P treatments compared to the low P treatments across all sites (Tables 2 and 3). The discrepancies between Mehlich3 and Morgan P with soil depth indicate the distinct extraction chemistry of these two methods and their target fractions rather than differences in the discrete in-situ P pools. Compared to Morgan’s extractant which targets a narrower range of readily plant-available P forms especially under acidic soil conditions, Mehlich3 can mobilise both labile and moderately bound P forms due to higher extraction strength. Our results on Mehlich3 P not following the same depth trends as Morgan’s P are in line with the previous study which showed different extractants responded variably during fertiliser Table 3 Soil physico-chemical properties at arable sites. Site P Depth BD pH SOC TN TP Mehlich3 P Morgan’s P SOC/TN SOC/TP SOC stocks treatments cm g cm −3 g kg −1 mg kg −1 mg L -1 t C ha −1 LS1 High 0–10 1.18 (0.16) a 6.09 (0.10) a 18.07 (0.16) a 1.58 (0.05) a 0.62 (0.01) b 97.09 (4.15) a 2.05 (0.56) a 11.48 (0.46) a 29.27 (0.05) a 21.35 (3.05) c 10–30 1.39 (0.04) a 6.13 (0.11) a 17.11 (1.17) a 1.58 (0.11) a 0.60 (0.03) b 69.12 (29.15) b 1.78 (0.04) ab 10.87 (0.01) a 28.38 (0.48) a 47.49 (1.89) a 30–50 1.12 (0.13) a 6.15 (0.19) a 4.92 (0.06) b 0.57 (0.05) b 0.40 (0.04) c 33.18 (24.62) b 0.48 (0.05) cd 8.63 (0.94) bc 12.33 (1.45) ab 11.04 (1.12) d Low 0–10 1.26 (0.08) a 6.19 (0.01) a 17.98 (0.28) a 1.66 (0.01) a 0.54 (0.05) ab 37.43 (27.07) b 0.87 (0.04) ce 10.83 (0.26) a 33.22 (2.21) a 22.74 (1.75) b 10–30 1.37 (0.08) a 6.08 (0.01) a 17.59 (1.75) a 1.65 (0.13) a 0.54 (0.00) ab 54.93 (34.86) b 1.00 (0.30) be 10.69 (0.19) ab 32.54 (3.35) a 48.26 (7.50) a 30–50 1.33 (0.04) a 6.29 (0.28) a 5.25 (0.47) b 0.69 (0.01) b 0.86 (0.63) a 31.62 (3.22) b 0.41 (0.03) d 7.66 (0.54) c 8.11 (5.35) b 13.95 (1.65) bd LS2 High 0–10 1.40 (0.05) a 6.37 (0.17) a 15.35 (0.37) a 1.44 (0.05) a 0.49 (0.01) b 12.84 (4.74) b 1.18 (0.25) ab 10.70 (0.11) ab 31.46 (1.25) a 21.51 (1.21) b 10–30 1.41 (0.05) a 6.31 (0.23) a 15.12 (0.38) a 1.32 (0.00) ab 0.58 (0.05) ab 20.51 (2.72) a 1.50 (0.83) a 11.45 (0.29) a 26.20 (0.65) ab 42.49 (0.51) a 30–50 1.35 (0.01) a 6.66 (0.23) a 4.36 (0.20b 0.57 (0.05) c 0.54 (0.01) ab 1.79 (0.48) c 0.41 (0.03) b 7.72 (1.18) bc 8.02 (0.53) bc 11.79 (0.50) cd Low 0–10 1.30 (0.02) a 6.42 (0.14) a 15.45 (1.40) a 1.48 (0.05) a 0.55 (0.13) ab 7.58 (2.24) bc 0.68 (0.09) ab 10.43 (0.65) ab 28.57 (4.25) a 19.99 (1.45) ab 10–30 1.34 (0.07) a 6.37 (0.25) a 15.26 (2.10) a 1.45 (0.10) a 0.49 (0.05) b 7.90 (2.46) bc 0.76 (0.15) ab 10.54 (0.78) ab 31.30 (1.90) a 40.70 (3.58) ab 30–50 1.38 (0.07) a 6.77 (0.18) a 4.57 (1.45) b 0.66 (0.30) bc 1.59 (1.45) a 1.56 (0.33) c 0.42 (0.03) b 7.19 (1.00) c 4.11 (2.75) c 12.52 (3.48) c JY High 0–10 1.28 (0.06) b 7.09 (0.19) a 13.78 (0.45) a 1.02 (0.08) a 0.51 (0.06) a 205.14 (33.93) a 19.30 (3.20) a 13.49 (0.70) a 27.53 (4.24) b 17.56 (0.25) c 10–30 1.44 (0.02) a 7.12 (0.10) a 15.11 (0.19) a 1.08 (0.02) a 0.44 (0.05) a 159.63 (22.36) a 14.23 (0.80) a 13.99 (0.05) a 34.42 (3.88) ab 43.56 (0.42) a 30–50 1.34 (0.06) ab 7.10 (0.11) a 9.73 (0.63) b 0.73 (0.12) b 0.21 (0.02) bc 38.97 (2.85) c 1.03 (0.44) b 13.51 (1.83) a 46.40 (6.89) a 26.19 (2.56) b Low 0–10 1.36 (0.01) ab 6.88 (0.28) a 14.75 (0.35) a 1.22 (0.03) a 0.30 (0.03) b 78.60 (11.91) b 2.42 (0.45) b 12.09 (0.05) a 49.00 (4.87) a 20.07 (0.32) ab 10–30 1.44 (0.02) a 7.08 (0.12) a 13.61 (0.97) a 1.04 (0.05) a 0.28 (0.05) b 62.33 (11.24) b 1.40 (0.85) b 13.11 (0.69) a 49.01 (5.69) a 39.11 (3.05) a 30–50 1.39 (0.02) ab 6.86 (0.05) a 6.68 (0.47) c 0.50 (0.09) c 0.15 (0.03) c 22.11 (0.28) c 1.13 (0.86) b 13.47 (1.75) a 46.40 (9.87) a 18.61 (1.55) c In Table 3, values represent sample means (for each depth; n =2 at LS1, LS2; n =3 at JY). Different letters of significance according to Tukey’s posthoc test (p <0.05) follow standard deviations in parentheses. The comparisons were made between depths at high and low P treatments in each individual sites. Same letters indicate no statistical difference. P, phosphorus; BD, bulk density; SOC, soil organic carbon; TN, total nitrogen; TP, total phosphorus; SOC/TN and SOC/TP, ratios. P. Bhople et al. Geoderma 463 (2025) 117538 5 drawdown (Khomenko et al., 2024). The results are also in line with the view of Barrow et al. (2021) who emphasised, P extraction methods provide operational indices of plant-available P and should not be interpretated as measuring specific, discrete pools of P in-situ. This underscores the need for extractant-specific interpretation when evaluating soil P dynamics across management regimes or soil depths. Overall, high ratios of SOC/TN and SOC/TP occurred at all grassland sites except LE and ranged between 7.3–10.9 for the SOC/TN ratio and 19.5–69.0 for the SOC/TP ratio across different soil depths (Table 2). The same ratios decreased significantly with increasing depths at arable sites especially at LS1 and LS2 sites, while the overall range for SOC/TN was between 7.2 and 14.0 and that of SOC/TP from 4.1 to 49.0 (Table 3). 3.2. Soil organic carbon stocks as affected by land use and soil depth The P fertiliser addition had no effect on SOC stocks (Fig. 1). In grassland sites, the SOC stocks ranged from 40.5 t C ha −1 in topsoils to 9.7 t C ha −1 at the 30–50 cm depth (Fig. 1, Table 2). The SOC stocks differed significantly between the 0–30 cm and 30–50 cm depths. In arable soils, the SOC stocks ranged from 48.3 to 11.0 t C ha −1 (Fig. 1, Table 3). The highest SOC stocks in arable sites occurred at 10–30 cm depth, which was significantly greater than in 0–10 and 30–50 cm layers. In both grassland and arable sites, the lowest SOC stocks were found at 30–50 cm depth (Fig. 1). 3.3. Relationships between soil total N, total P and SOC stocks The soil TN and TP contents were generally higher in grassland sites than in arable sites (Fig. 2). The correlations between these nutrients were stronger at the grassland sites (R 2 >0.65) compared to the arable sites (R 2 <0.27, Fig. 2). This pattern was similar for both P treatments within each land use. As soil TN content increased so did SOC (Fig. 3A). The strongest and most significant correlations between SOC stocks and TN contents occurred at 0–10 cm and 30–50 cm depth (R 2 >0.79, Fig. 3A) of the grassland sites. In arable soils the correlations between SOC stocks and TN were weaker compared to grasslands sites reaching up to R 2 >0.57 with higher SOC stocks correlating with higher TN contents at the 10–30 cm soil depth. While these relationships were weakest in deeper soils at 30–50 cm (R 2 <0.18, Fig. 3A). The SOC stocks and TP contents correlated significantly (R 2 >0.50, Fig. 3B) at 10–30 and 30–50 cm soil depths in both land uses while these relationships were weak and not significant in top 0–10 cm at both sites (R 2 <0.12, Fig. 3B). 3.4. Relationships between soil variables and CNP interactions The first two axes of the PCA ordination plot (Fig. 4) explained 92.7 % total variability in the samples (PC1, 68.8 %; PC2, 23.5 %). The grassland samples were more associated with SOC, SOC stocks, TN, ratios of SOC/TN, SOC/TP and TN/TP, highlighting more influence of soil chemical properties. In contrast, arable samples clustered along PC2 and were associated with physical soil properties such as BD, sand, silt and clay and also SOC/TN values. The PCA biplot also suggests that P fractions (Mehlich3 P, Morgan’s P) and liming further contribute to differentiating land use types together with soil depths (Fig. 4). For grassland sites, the optimisation algorithm converged to an optimal SOC/TN ratio of 10.1 and a SOC/TP ratio of 32.6, with 95 % confidence intervals of 8.8–11.4 and 26.0–39.1, respectively and predicted SOC stock of 30.9 t C ha −1 (Fig. 5A). For the arable sites, the optimisation algorithm converged to an optimal SOC/TN ratio of 10.9 and a SOC/TP ratio of 29.4 with 95 % confidence intervals of 9.7–12.1 and 25.0–33.8, respectively, and predicted SOC stock of 33.34 t C ha −1 (Fig. 5B). These values represent the conditions under which the predicted SOC stocks are maximised, given the modelled relationships. While the optimisation approach provides an integrated view of SOC/ TN and SOC/TP ratios, the pairwise relationships between SOC vs. SOC/ TP are also presented in Supplementary Table 1. These indicate broadly consistent optima but with weaker relationships due to the limited number of sites and intrinsic variations. 4. Discussion 4.1. The impacts of land use on SOC stocks and their response to longterm P fertilisation across land uses The results in our study do not indicate significant effects of land uses on SOC stocks. Although previous studies have shown that grassland soils can have higher SOC stocks, due to perennial deep-rooting vegetation and reduced disturbance (Kell, 2011; Poeplau et al., 2018), this was not the case here. This may be explained by other site-specific interacting factors, such as climate, management practices, soil types, and nutrient stoichiometry, which together can influence SOC storage (Kirkby et al., 2014; Lal, 2018; Poeplau et al., 2018). The SOC stocks in the arable sites may reflect compensatory mechanisms as seen in other studies such as enhanced residue inputs under high P conditions Fig. 1. Mean soil organic carbon (SOC) stocks at different P treatments (high and low) in grassland and arable sites. Different letters indicate significant difference between the means (p <0.05) when depths were compared across high and low P treatments. Same letters represent no statistical difference between the means of P treatments. P. Bhople et al. Geoderma 463 (2025) 117538 6 (Poeplau et al., 2018) or the interactions between liming and P availability as observed at the JY site (Pedersen et al., 2025). Our results do not confirm the first hypothesis, namely that higher P fertilisation would lead to higher SOC stocks. While P addition stimulates primary productivity, and thus the quantity of SOM inputs to soils (Spohn and Schleuss, 2019), our findings suggest that beyond a certain threshold additional P may not necessarily translate into measurable differences in SOC stocks. The lack of significant differences in SOC stocks between the high and low P treatments in both grassland and arable systems also align with the findings of Poeplau et al. (2016), who reported that P fertilisation under N-limited conditions can reduce SOC stocks. Thus, imbalances in nutrients may negate the potential benefits of P additions on long-term SOC storage. Previous studies in grassland systems reported moderate P additions enhanced SOC storage (Poeplau et al., 2016; Poeplau et al., 2018), but excessive P potentially triggered microbial priming effects leading to SOC loss (Griffiths et al., 2012). Additionally, the work of Kelly (2022) highlights the complexity of P effects on SOC dynamics and emphasizes the role of microbial-mediated organic matter turnover in grassland soils, which can explain why SOC remained stable despite varying P inputs in grassland soils. This indicates that priming effects represent short-term microbial responses rather than direct indicators of cumulative SOM losses. In Kelly (2022), organic P inputs induced positive priming and enhanced N mineralization, but these microbial shifts did not lead to higher long-term SOM decomposition. In fact, relative to inorganic P inputs, organic P inputs were associated with lower net SOM losses, underscoring that priming responses might be short lived and relatively small in magnitude and under specific conditions may exert limited effects on long-term SOM stability. Though stronger and more frequent priming effects could still influence SOC dynamics. In contrast, no expected P-induced SOC losses in arable soils were observed, which might be due to initial low SOC stocks limiting microbial decomposition (Lorenz and Lal, 2005) or increased crop residue returns mitigating C losses (Poeplau et al., 2018). Furthermore, increases in crop yield were much greater for arable than grassland sites, so the lack of an impact on SOC stocks may reflect increased decomposition due to increased P inputs at arable sites. However, it might also be possible that SOC inputs may not be directly related to P fertilisation (e.g., increases harvestable yield but not root growth or aboveground residue inputs). Additionally, the SOC stocks may not be sufficiently responsive to be simply detectable over the range of P inputs applied in the LTEs, given their intrinsic variability and longterm equilibration. Overall, these findings indicate that P availability alone does not dictate SOC stocks but interacts with other biotic and abiotic constraints, highlighting the necessity of balanced fertilisation strategies tailored to specific soil and ecosystem context for effective SOC storage. 4.2. Multifaceted role of P on subsoil C storage as influenced by long-term P fertilisation We show that long-term P fertilisation did not significantly increase SOC stocks in deeper soil layers (30–50 cm), which leads to the rejection of our first hypothesis. While plant productivity and organic matter input are generally associated with increased P input (Chen et al., 2014), this did not lead to higher SOC stocks in deeper layers for either land uses. Such an absence of significant effects of P treatments on SOC stocks suggest that SOC storage in subsoils might be regulated by additional interacting factors, such as microbial constraints, soil structure, and nutrient stoichiometry (Liang et al., 2019). It might also be because of the P fertilisation did not increase P inputs to the subsoils in any appreciable manner. In grasslands, the SOC stocks were stable up to 30 cm depth but reduced significantly at 30–50 cm, indicating limited subsoil C accumulation. This is in line with previous grassland-based studies demonstrating that SOC accumulation in topsoil is driven by continuous plant residue inputs along with minimal soil disturbance, while in the deeper soil layers, SOC storage is constrained by lower root density and microbial activity (Lorenz and Lal, 2005; Lynch and Wojciechowski, 2015). These results are also in line with those of Griffiths et al. (2012), who reported no significant SOC changes in grassland soils under different long-term P fertilisation treatments and that the system maintained stable C:N:P ratios. However, in the arable soils, increased microbial activity and organic matter turnover, as well as tillage-induced aeration and soil structural disruption could have resulted in lower SOC stocks in the topsoil (0–10 cm) as compared to the 10–30 cm soil depths. This apparent difference is likely due to the larger depth increment and mixing of residues due to ploughing, which resulted in SOC concentration being distributed over a greater soil volume. Furthermore, the middepth (10–30 cm) showed significantly higher SOC stocks compared to other depths (0–10 and 30–50 cm), which can be attributed to residue incorporation via ploughing. This is in line with previous observations suggesting that ploughing redistributed organic matter and enhanced SOC storage (Fontaine et al., 2007). Likewise, Coonan et al. (2019) also reported residue incorporation and SOC enhancement at intermediate depths and no significant SOC sequestration in arable subsoils. Overall, these findings are consistent with a recent study that demonstrated that the redistribution of C caused by tillage plays a more important role in SOC storage than P fertilisation (Peixoto et al., 2021). Additionally, some recent trials in Germany and Denmark (Kim et al., 2022; Pedersen et al., 2025) have also shown that microbial-mediated decomposition Fig. 2. Relationships between total nitrogen (N) and total phosphorus (P) contents across soil depths at different P treatments (high and low) in grassland and arable sites. P. Bhople et al. Geoderma 463 (2025) 117538 7 rates critically determine subsoil SOC dynamics as opposed to nutrient additions alone. Therefore, despite known contributions of P fertilisation towards plant productivity and nutrient cycling, its direct influence on deeper SOC stocks remains limited and further illustrates the multifaceted role of P to be explored via long-term experimental trials. 4.3. Soil C:N:P stoichiometry and its role in SOC storage across soil depths The results in our study showed nutrient interactions influenced SOC distribution across depths, but P effects alone were not significant. Instead, N contents were more closely related to SOC stocks, particularly in grassland systems. In grasslands, SOC stocks in the top soils varied in association with the availability of N as indicated by higher N contents. Stronger influences of N availability and microbial processes and decomposition dynamics were previously reported in grassland systems (Lorenz and Lal, 2005; Liang et al., 2019), supporting our inferences. However, in deeper layers (10–50 cm) higher C/N ratios exhibit signs of N limitation which can constrain microbial activity, leading to slower decomposition and potential SOC accumulation as previously seen in a long-term P fertilisation trial (Griffiths et al., 2012). Furthermore, Torres-Sallan et al. (2017) also highlighted the importance of mineral associations in subsoil SOC persistence, which might perhaps be the case for observations in SOC distribution and nutrient interactions in the Fig. 3A. Relationships between soil organic carbon (SOC) stocks and total nitrogen (N) at different P treatments (high and low) and soil depth gradients in grassland and arable sites. P. Bhople et al. Geoderma 463 (2025) 117538 8 grassland systems in our study. In contrast, arable soils showed different patterns of SOC stock distribution, where significant stocks occurred at 10–30 cm depth, potentially due to soil inversion and residue incorporation. Our observations of the trend in SOC stocks with depth in arable soils are consistent with the studies of Necp´ alov´ a et al. (2014) and Fontaine et al. (2007), who showed that ploughing redistributes organic matter to intermediate depths, which promotes SOC accumulation but does not necessarily lead to long-term sequestration. The impacts of P availability on SOC were ambiguous. Previous studies explored OP as a biologically relevant P pool to study stoichiometry (Spohn, 2020), particularly due to its microbial associations. However, TP remains a widely used indicator in soil biogeochemical research in the context of long-term studies (Kirkby et al., 2011), it was selected in our study to represent cumulative P enrichment in LTE soils. We infer it to be a relevant metric for soil stoichiometric assessments providing an integrative measure of long-term P status in soils. The OP includes biologically active forms but also comprises stable organic complexes that may not be readily plant or microbes-available (Barrow et al., 2021). In contrast, TP includes both organic and mineral-bound forms and provides a robust proxy for long-term P status under legacy fertilisation that can affect P availability and influence SOC stocks. The use of TP has ensured analytical consistency across sites and enabled comparisons with previous LTE-based stoichiometric studies (Kirkby et al., 2011). However, we do acknowledge its limitation for assessing Fig. 3B. Relationships between soil organic carbon (SOC) stocks and total phosphorus (P) at different P treatments (high and low) and soil depths in grassland and arable sites. P. Bhople et al. Geoderma 463 (2025) 117538 9