Identifying the critical phosphorus balance for optimizing phosphorus input and regulating soil phosphorus effectiveness in a typical winter wheat-summer maize rotation system in North China

XU Meng-ze, WANG Yu-hong, NlE Cai-e, SONG Gui-pei, XlN Su-ning, LU Yan-li, BAl You-lu, ZHANG Yin-jie, WANG Lei#

1 State Key Laboratory of Efficient Utilization of Arid and Semi-arid Arable Land in Northern China/Key Laboratory of Plant Nutrition and Fertilizer of Ministry of Agriculture and Rural Affairs, Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, Beijing 100081, P.R.China

2 College of Resources and Environment, Henan Agricultural University, Zhengzhou 450002, P.R.China

Abstract Phosphorus (P) is a nonrenewable resource and a critical element for plant growth that plays an important role in improving crop yield.Excessive P fertilizer application is widespread in agricultural production, which not only wastes phosphate resources but also causes P accumulation and groundwater pollution.Here, we hypothesized that the apparent P balance of a crop system could be used as an indicator for identifying the critical P input in order to obtain a high yield with high phosphorus use efficiency (PUE).A 12-year field experiment with P fertilization rates of 0, 45,90, 135, 180, and 225 kg P2O5 ha-1 was conducted to determine the crop yield, PUE, and soil Olsen-P value response to P balance, and to optimize the P input.Annual yield stagnation occurred when the P fertilizer application exceeded a certain level, and high yield and PUE levels were achieved with annual P fertilizer application rates of 90-135 kg P2O5 ha-1.A critical P balance range of 2.15-4.45 kg P ha-1 was recommended to achieve optimum yield with minimal environmental risk.The critical P input range estimated from the P balance was 95.7-101 kg P2O5 ha-1, which improved relative yield (>90%) and PUE (90.0-94.9%).In addition, the P input-output balance helps in assessing future changes in Olsen-P values, which increased by 4.07 mg kg-1 of P for every 100 kg of P surplus.Overall, the P balance can be used as a critical indicator for P management in agriculture, providing a robust reference for limiting P excess and developing a more productive, efficient and environmentally friendly P fertilizer management strategy.

Keywords: yield of winter wheat and summer maize, phosphorus balance, phosphorus use efficiency, Olsen-P, critical phosphorus application rate

1.Introduction

Phosphorus (P) is a scarce mineral resource for which there is no substitute, and a life-supporting nutrient that is essential for various assimilatory and metabolic processes in plants, thereby playing an important role in improving crop yield and quality in agricultural production (Roberts and Johnston 2015; Wanget al.2021; Weisset al.2021).However, P fertilizer is commonly over-applied to croplands in agricultural production in order to meet the growing demand for food (Satoet al.2005; Liet al.2011).The amount of P fertilizer applied in China has been steadily increasing from 2.73 million tons in 1980 to 6.82 million tons in 2019 (NBSC 2020).The recovery of inseason P fertilizer by crops is low due to P fixation by the soil and the limited ability of the crops to utilize P, which results in a large accumulation of soil P (Tanget al.2008;Yang and Fang 2015).Previous studies have reported the widespread P accumulation and P pollution in intensive vegetable cropping systems (Kalkhajehet al.2017; Zhanget al.2021).However, in this context, few studies have investigated the effect of long-term P application on soil P accumulation in wheat-maize rotation systems.

When soil P accumulation exceeds a certain level, the risk of the loss of dissolved and particulate P in arable soil increases with a “sudden change” pattern (Hartet al.2004), resulting in wasted P fertilizer resources and an increased risk of water pollution (Chenet al.2008;Powerset al.2016).Previous research has indicated that approximately 80% of the phosphate rock used each year is used to make P fertilizer, and that phosphate resources are expected to be completely depleted in the next century (Roberts and Johnston 2015; Dhillonet al.2017).Hence, there is an urgent need to identify a key indicator of the P fertilizer application threshold to maximize productivity and phosphorus use efficiency (PUE, the ratio of P output to P input in a defined system), and to reduce the negative environmental impacts of excessive P application.

Nutrient balance is the difference between the nutrient inputs and outputs of an agricultural system, and it has been employed as an agro-environmental indicator for assessing the performance of nutrient management by providing information on nutrient use efficiency, soil fertility, and the environmental risk of nutrient losses from agricultural systems (McLellanet al.2018).A positive nutrient balance with a low nutrient use efficiency indicates a build-up of nutrients in a cropland, which poses an environmental risk; whereas a nutrient balance deficit and a very high nutrient use efficiency indicate the mining of soil nutrient reserves, which will reduce soil fertility over time and is unsustainable over the long term(Schr?deret al.2003; Thomaset al.2020).All the P that is not directly taken up by plants on agricultural land is accumulated in the soil and may be either taken up in the following growing season or accumulated deeper in the ground, and P accumulation is an unavoidable problem in intensive cropping systems (Pedersenet al.2020).Therefore, the P input should be only slightly higher than P output to achieve sustainable production and maintain soil fertility (Jordan-Meilleet al.2012;Ameryet al.2021).However, in soils with a high P balance, crops do not show an increase in yield and P losses increase with increasing P balance (Powerset al.2016; Svanbacket al.2019; Fanet al.2022; Mayer and Kaltschmitt 2022).The average soil P surplus in China increased from 4.6 kg P ha-1in 1980 to 42.1 kg P ha-1in 2012, and the P surplus on arable cropland showed considerable room for improvement (Maet al.2018).Although there are various methods for preventing water runoff and maintaining nutrient availability in the soil(e.g., certain tillage techniques (Hartet al.2004) or the application of slow-release fertilizers (Volf and Rosolem 2020)), some portion of the P is still wasted on fields without being utilized by the plants.Many researchers have highlighted the P surplus in agricultural systems and the environmental problems it causes, and have called for reducing the excess P balance and increasing PUE to achieve sustainable production in agriculture (Chenet al.2008; Penuelaset al.2013; Pedersenet al.2020;Thomaset al.2020; Jagdeep and Brar 2022).However,the specific P balance thresholds remain unclear, and standards for limiting the amount of P fertilizer application per unit area are mostly lacking.Furthermore, one challenge for optimizing fertilizer application rates in China is that producers do not have a clear understanding of the relationship between nutrient input and nutrient output in local agricultural systems (Heet al.2022).Given the robust relationship between nutrient balance and nutrient input (EUNEP 2016; Dinget al.2021), the P balance provides a viable method for identifying critical P fertilizer inputs by comprehensively considering the effects of P application on crop systems.

Previous studies have shown that each crop has a critical soil Olsen-P value (CPV) and that crop yields perform poorly when the Olsen-P is below the CPV;however, when the Olsen-P in the soil exceeds the CPV,crop yields do not significantly improve with increasing P inputs (Baiet al.2013; Xiet al.2016; Ameryet al.2021).Therefore, studying the dynamic characteristics of soil Olsen-P values is important for improving the crop yield and PUE.A significant positive correlation was found between the soil Olsen-P value and P balance (Shenet al.2014), and the same results were found in the fluvoaquic soils of China (Zhanget al.2019).In addition,the relationship between the soil Olsen-P value and P balance in agricultural systems has been shown to vary with differences in the environment, crop systems and soil physicochemical properties (Caoet al.2012).Therefore,once we clarify the relationship between the soil Olsen-P value and P balance, we can adjust the P balance to approach a specific CPV by changing the P input.

Given the above considerations, we conducted a 12-year field fertilization experiment on fluvo-aquic soil in the North China aiming to: (1) identify the P balance threshold for a winter wheat-summer maize rotation system; (2) evaluate the effects of P inputs with an optimized P balance on P output and PUE; and (3)scientifically guide the P input by linking the Olsen-P value to the P balance.This research is of great significance for improving crop yields, reducing the waste of phosphate ore resources, and reducing the risk of P fertilizer pollution.It also provides reasonable, scientific indicators for optimizing the application rate of P fertilizer and promoting the sustainable development of agricultural production.

2.Materials and methods

2.1.Experimental site

A 12-year field experiment on a winter wheat-summer maize crop rotation site was carried out from October 2008 to October 2020 in the Guangyang District in the city of Langfang, Hebei Province, China (116°35′16′′E,39°47′35′′N).This area has a warm-temperate continental monsoon climate in a midlatitude zone.The monthly average temperature and precipitation data are shown in Appendix A.The mean annual temperature at the site is 11.9°C, and the mean annual precipitation is 571 mm.The average annual sunshine duration is 2 660 h, and there are 183 frost-free days per year.The soil at the experimental site was classified as fluvoaquic soil and the soil texture was loamy sand (78.7%sand, 13.5% silt and 7.78% clay) with a bulk density of 1.43 g cm-3.Prior to the experiment, the soil (0-20 cm)physiochemical properties were as follows: pH 8.1,organic matter content 11.7 g kg-1, available N 56.7 mg kg-1, Olsen-P value 11.9 mg kg-1, and available potassium 43.1 mg kg-1.

2.2.Experimental design and management

The trial was conducted at six P levels (P0, P1, P2, P3,P4, and P5), and the fertilizer application rates are shown in Table 1.Plots (100 m2, 8 m×12.5 m) were arranged in an equally spaced sequence.In the winter wheat season,chemical fertilizer in the form of 1/2 N and full P2O5and K2O was applied as a basal fertilizer.The remaining N was applied in two applications (1/3 N at greening; 1/6 N at jointing).In the summer maize season, the same amount of chemical N fertilizer was applied as in thewheat season, with no phosphate or potassium, and in each treatment, 2/5 N was applied at the seedling stage and 3/5 N at the opening stage.All straw was shredded and returned to the field when the winter wheat and summer maize matured.

Table 1 Fertilizer application rates for each treatment on winter wheat-summer maize

The data used in this study ran from the start of the winter wheat planting season in 2008 until the summer maize harvest in 2020.The winter wheat varieties tested were Baofeng 104 in 2008 and Jingdong 17 from 2009 onward, while the summer maize variety was Zhengdan 958.The fertilizers used were urea (containing 46% N), diammonium phosphate (containing 18% N and 46% P2O5), calcium superphosphate (containing 12%P2O5), and potassium sulfate (containing 50% K2O).

2.3.Sampling and measurements

Winter wheat was sown in October each year and harvested in June the following year, while summer maize was planted after the winter wheat harvest and harvested in October.The crops were machine harvested, and the yields were measured as the actual harvest.Three plant samples were collected from representative sampling points in each treatment before winter wheat and summer maize were harvested.Soil samples were collected at 0-20 cm depth after each rotation cycle.

After the crop was harvested and air-dried, grain and straw samples were oven-dried for 30 min at 105°C and then heated at 70°C to a constant weight to determine the dry matter and nutrient contents.The oven-dried straw and grains were ground and digested with H2SO4-H2O2at 300-360°C.The plant P contents were measured according to the vanadium-molybdate yellow colorimetric method.Soil Olsen-P values were measured according to the methods of Lu (2000).

2.4.Data analysis

Agronomic indices of P fertilizer use efficiencyThree indicators of P fertilizer use efficiency were calculated as follows:

Recovery efficiency of P (REP; %)=(Pup-P0up)/Prate×100

Agronomic efficiency of P (AEP; kg kg-1)=(Pyield-P0yield)/Prate

Partial factor productivity of P (PFPP; kg kg-1)=Pyield/Pratewhere Pupand P0upindicate the P uptake from the P and P0 treatments (kg P2O5ha-1), respectively; Pyieldand P0yieldindicate the grain yields from the P and P0 treatments (kg ha-1), respectively; and Prateis the P application rate (kg P2O5ha-1).

Calculation of P balanceThe soil P balance was calculated as the P input minus the P output.In contrast to the nutrient balance determined through all nutrient input and output pathways, the apparent nutrient balance determined by fertilizer inputs and crop outputs is precise and easily compared across different agricultural systems(Zickeret al.2018; Jagdeep and Brar 2022).To facilitate communication and comparison of the experimental results, the production system was treated as a “black box”.We did not use crop straw as a P input and output pathway, as it did not leave the production system.The apparent P balance was calculated as follows:

P balance (kg P ha-1)=P input-P output

where the P input is chemical P fertilizer (kg P ha-1) and P output is the removal of P in the harvested grain (kg P ha-1).

Determining the range of P balanceIn this study, we attempted to determine the upper limit of the P balance for a winter wheat-summer maize rotation system based on the response curve of the yield to the P balance.To determine the variations in the winter wheat and summer corn yields that were caused by climatic differences during 2008-2020, the data were analyzed using the relative yield (RY) instead of the actual yield.We used the model correlation coefficient (R2) as an indicator to select an applicable linear-platform model for winter wheat and summer maize (Singhet al.2016; Huanget al.2021):

RY (%)=Yi/Ymax×100

where RY is the relative yield of the crop grain (%),Yiis the grain yield of each treatment in each year (kg ha-1),and Ymaxis the maximum grain yield of all treatments each year (kg ha-1).

Y=ax+b (x

Y=c (x≥xi)

whereYis the relative yield (%),xis the P balance (kg P ha-1), and a, b, and c are the fitting constants.

Phosphorus use efficiency (PUE) indicatorPUE was calculated according to the principle of mass balance, i.e.,using P input and P output (Syerset al.2008):

PUE (%)=P output/P input

where P input is the chemical P fertilizer (kg P ha-1) and P output is the removal of accumulated P in the harvested grain (kg P ha-1).

Relationships between the Olsen-P value and relativeyield and P balanceWe determined the relationship between Olsen-P and relative yield using the Mitscherlich equation described by Singhet al.(2016) and Zhanget al.(2019) in order to diagnose the P required in a winter wheat-summer corn rotation system and identify the CPV.The Mitscherlich equation is as follows:

Y=A×(1-Exp(-bx))

whereYis the relative yield of the prediction (%), A is the maximum relative yield (%), and b is the coefficient of the relative yield of the soil Olsen-P (x) value.The soil Olsen-P content is the CPV when the Mitscherlich model simulates a Y of 90%.

We also performed linear regression analysis to describe the relationship between the soil Olsen-P value and the P balance using a linear model.The details are as follows:

Cumulative P balance (kg P ha-1)=∑(Apparent P balance of crops in a year)

Y=ax+b

whereYis the Olsen-P (mg kg-1) value,xis the cumulative P balance (kg P ha-1), and a and b are the fitting constants.The linear model was determined by the highest coefficient of determination (R2) for the model.

2.5.Data statistics

Excel 2016 was used for data collation.Variance analysis was performed to determine the differences among the obtained data, and least significant difference (LSD) tests were run in SPSS version 20.0.Plots were made using Sigmaplot 12.5 and Origin 2021.

3.Results

3.1.Grain yield, P uptake, and P fertilizer use effi-ciency

The long-term application of P fertilizer significantly affected the yields of winter wheat and summer maize(Fig.1-A and B).The average yields of winter wheat in treatments P0 to P5 were 906, 3 427, 4 974, 5 104, 5 129,and 5 124 kg ha-1, respectively, and those of summer maize were 5 593, 8 127, 8 907, 8 922, 8 542 and 8 324 kg ha-1, respectively (Appendix B).Compared to those in the P0 treatment, the yields of winter wheat and summer maize significantly increased by 278-466 and 45.3-59.5%,respectively, in the P1 to P5 treatments.Compared to the P1 treatment, the yields of winter wheat and summer maize were significantly increased in P2 to P5.In addition, the RY of P0 and P1 showed a decreasing trend as the crop rotation cycle was extended (Fig.1-C).The mean RY values of the P0 to P5 treatments were 45.7, 79.9, 96.2,96.3, 93.7, and 92.3%, respectively (Fig.1-D).Among them, the RY difference was not significant when the annual P fertilizer application was increased to 90-225 kg P2O5ha-1.

Fig.1 Effect of long-term phosphorus (P) fertilizer application on yield (A and B) and relative yield (C and D) of winter wheatsummer maize.P0, 0 kg P2O5 ha-1; P1, 45 kg P2O5 ha-1; P2, 90 kg P2O5 ha-1; P3, 135 kg P2O5 ha-1; P4, 180 kg P2O5 ha-1; P5,225 kg P2O5 ha-1 per year.Error bar is SD (n=12).The lower and upper edge lines represent the 5 and 95% thresholds of the data, respectively, and the solid points represent the vertical outliers (A, B, and D).The lower and upper quartiles of the boxplots represent the 25 and 75% thresholds of the data, respectively, the red solid lines represent the average values, and the black dashed lines represent the median values.Different lowercase letters above the boxplots indicate significant differences in the mean values of the different treatments for the 12-year rotation cycle at the 0.05 confidence level.

Aboveground P uptake was significantly higher in the winter wheat and summer maize treated with P fertilizer compared to that without P application during the 12-year rotation (Table 2).The average values of P uptake per year in the P1-P5 treatments were 35.5, 49.1, 49.7, 53.7,and 52.6 kg P ha-1, respectively, which were significantly higher by 119-231% compared to that in P0.Seasonal and annual differences in P uptake were not significant when the P fertilizer application ranged from 90 to 225 kg P2O5ha-1per year.In addition, we found that the P fertilizer use efficiency (REP, AEPand PFPP) decreased with increasing P application, and this phenomenon could be found in each crop rotation cycle (Table 3).The REP,AEP, and PFPPvalues reached higher levels when the P application did not exceed 135 kg P2O5ha-1per year.

Table 2 Aboveground phosphorus (P) uptake of the winter wheat-summer maize rotation systems under long-term P fertilizer application

3.2.P balance and PUE

The P balance increased with increasing P fertilizer rates(Fig.2-A).After 12 years of crop rotation, the P input (P fertilizer) was lower than the P output (grain P uptake) in the P0 and P1 treatments, and the P balance was always in a deficit, with mean values of -12.5 and -7.50 kg P ha-1, respectively, and the crop mainly absorbed P fromthe soil.The P input and output were nearly in equilibrium in the P2 treatment, with a mean P balance of 1.55 kg P ha-1.The P3 to P5 treatments all had P surpluses, in the amounts of 20.6, 37.5, and 58.7 kg P ha-1, respectively.

The variations in PUE for the different P fertilizer application rates are shown in Fig.2-C, indicating the PUE decreased with an increasing P fertilizer application rate.The average PUE values for the P1 to P5 treatments were 138, 96.0, 65.1, 52.3 and 40.3%, respectively(Appendix C).Both the P balance and PUE were exponentially related to the P fertilizer application rate(Fig.2-B and D).

3.3.Upper limit of P balance based on relative yield determination

The response of RY to the increasing P balance was determined by the best fit using a linear-plateau model(P<0.01; Fig.3).The minimum P balance needed to achieve the maximum RY (94.3%) in the winter wheatsummer maize rotation system was 4.45 kg P ha-1per year.Below this level, the RY increased linearly (slope of 1.87).A plateau in the RY occurred when the P balance exceeded 4.45 kg P ha-1.The P balance at a 90% RY was 2.15 kg P ha-1, and maintaining a P balance in the range of 2.15-4.45 kg P ha-1ensured that the RY did not fall below 90%.Based on the above analysis, the critical P balance range for optimizing the yield was identified as 2.15-4.45 kg P ha-1, and a P balance above 4.45 kg P ha-1would waste phosphate resources.

3.4.Range of P inputs and PUE

The relationships between P input (x) and P output (y)were best regressed by the linear-plateau model across all the data (P<0.01; Fig.4-A).The minimum P input needed to achieve the maximum P output (39.7 kg P ha-1) was 42.2 kg P ha-1.Up to this amount, the P output increased linearly (slope of 0.64).Beyond this amount,the attainable P output reached 39.7 kg P ha-1.The intercept in Fig.4-A shows that the P output was 13.2 kg P ha-1when the P input was 0 kg P ha-1, and the P input was 36.6 kg P ha-1when the P balance was 0 kg P ha-1.

The P balance (x-y) and the PUE (y/x) were calculated in this study from the P input and P output values.According to the critical P balance range of 2.15-4.45 kg P ha-1, an expression between P input and P output can be obtained as 2.15≤x-y,x-y≤4.45 (Fig.4-A).The intersection of this equation in the P input-P output model falls on the plateau line of the P output, and the calculated P input threshold range was 41.8-44.1 kg P ha-1(95.7-101 kg P2O5ha-1).Therefore, the target range of the PUE was calculated as 90.0-94.9% based on the key P input and P output coordinate points in Fig.4-C, i.e.,x=41.8,y=41.2 andx=44.1,y=41.2, respectively.

3.5.The responses of yield to Olsen-P, and Olsen-P to cumulative P balance

The application of P had a significant effect on the soil Olsen-P content (Fig.5-A).Compared to the treatment without P, the average Olsen-P value increased significantly in the P fertilizer application treatments and increased with the P fertilizer application rate (Fig.5-B).The relationship between the P balance and the Olsen-P value was established using a linear model (Fig.5-C,P<0.01).The Olsen-P value showed a positive linear relationship with the cumulative P balance.The Olsen-P value increased by 4.07 mg kg-1for every 100 kg P ha-1increase in the cumulative P balance of the soil.The relationship between the RY and Olsen-P content could be simulated by the Mitscherlich model, and the correlation between these two variables reached a significant level(P<0.01; Fig.5-D).Our predicted maximum RY of 95.3%could be achieved with an adequate Olsen-P content, and a RY above 90% could be guaranteed when the Olsen-P content exceeded 7.92 mg kg-1.

Fig.5 The changes in Olsen-P (A and B) from October 2008 to June 2020, the response of Olsen-P to cumulative phosphorus(P) balance (C), and the relationship between the relative yield and Olsen-P (D) under long-term P fertilization treatments in a fluvo-aquic soil.P0, 0 kg P2O5 ha-1; P1, 45 kg P2O5 ha-1; P2, 90 kg P2O5 ha-1; P3, 135 kg P2O5 ha-1; P4, 180 kg P2O5 ha-1; P5,225 kg P2O5 ha-1 per year.Error bar is SD (n=12).The lower and upper edge lines represent the 5 and 95% thresholds of the data, respectively, and the solid points represent the vertical outliers (B).The lower and upper quartiles of the boxplots represent the 25 and 75% thresholds of the data, respectively, the red solid lines represent the average values, and the black dashed lines represent the median values.Different lowercase letters above the boxplots indicate significant differences in the mean values of different treatments for the 12-year rotation cycle at the 0.05 confidence level.

4.Discussion

4.1.Upper limits of P balance

With the efficient use of P fertilizer, crop systems will achieve high yields and PUE with a limited P surplus.Several foreign researchers have developed P balance thresholds after considering agricultural and environmental results.For example, Thomaset al.(2020) found through meta-analysis that a P balance of 4 kg P ha-1led to high productivity in cropping systems, and P balance values between 1.5 and 4.5 kg P ha-1in grassland systems maintained optimal Olsen-P concentrations in these systems.Ameryet al.(2021) found that maintaining a P balance in the range of 1-10 kg P ha-1through fertilization measures could guarantee a steady crop yield.Hunteret al.(2017) noted that discussion around sustainable agricultural intensification should focus on the goal of food production.In this study, we proposed that the threshold for P balance in northern China from the consideration of yield maximization ranged from 2.15 to 4.45 kg P ha-1in the winter wheat-summer maize rotation system(Fig.3), which would apply only to situations where the soil fertility is low or where the maximum yield is sought.These results indicated that the differences in the optimal P balance can be mainly attributed to differences in P losses under different soil types and cropping systems(Jianget al.2019), as well as the systematically defined nutrient input and output pathways.Furthermore, we found that the optimal P balance ranges obtained by different researchers were all greater than zero, which was also confirmed by our results.We found that when the P balance was less than 0, the RY was below 90%(Fig.3), resulting in lower productivity of the crop system.This phenomenon occurs because the P that is applied to the soil can be easily fixed by metal cations (such as calcium, iron, and aluminum) or absorbed onto mineral surfaces, leading to low P availability (Torrentet al.1992),and P losses through runoff and leaching are unavoidable in agricultural systems (van Middelkoopet al.2016;Svanbacket al.2019).In general, P balance calculations should rely on long-term field experience, rather than short-term experience, in order to avoid changes in P application and the interference of climatic factors.However, more research is needed to identify the critical P balances for other agricultural systems.

4.2.Upper limits of the P application rate and PUE

Due to differences among fertilization management measures, there are various combinations of P inputs and outputs in farmland systems (Medinskiet al.2018).The regression analysis model of P inputs and outputs can provide critical information for the efficient application of P fertilizer (Fig.4-A).In one study, P uptake by crops increased with increasing P fertilizer application in wheat-maize crop rotation experiments, and the P uptake gradually leveled off once the P application rate reached a certain value (Singhet al.2016).We found the same results in this study, with a maximum P output of 39.7 kg P ha-1for winter wheat-summer maize under an adequate P supply, and higher P inputs (above 42.2 kg P ha-1) did not contribute to a higher P output.The results obtained in this study indicated a P output of 13.2 kg P ha-1for a P input of 0 kg P ha-1, reflecting the soil P supply and P balance in the absence of P input under the experimental conditions, which provides critical reference information for P fertilizer application decisions.Previous studies have shown that wheat and maize yields do not consistently increase with increasing P fertilizer application (Denget al.2014; Jagdeep and Brar 2022).With a single application of P fertilizer in the winter wheat season, the annual P input threshold range of 41.8-44.4 kg P ha-1(95.7-101 kg P2O5ha-1) derived in this study can meet the P requirements for optimal yield as well as for maximal P output (Fig.4-A).These results are in accordance with those of a recent meta-analysis study by Liet al.(2021), who reported that annual P applications of 90-150 kg P2O5ha-1and a 1:0 ratio of P applications to winter wheat and summer maize showed the best increases in both the single season yield and annual yields in northern China.This could be due to the low soil temperatures during the winter wheat growing season resulting in a low soil P supply capacity, while high temperatures during the summer maize growing season would increase the soil P supply capacity by increasing P activity and microbial metabolism (Shaw and Cleveland 2020).Hence, applying P fertilizer during the winter wheat season and foregoing its application during the summer maize season is a measure that could improve yield and phosphorus fertilizer utilization while reducing phosphorus losses.

Many studies have shown that improving PUE is essential for the sustainable use of phosphate resources and the sustainable development of agriculture (Childerset al.2011; Johnstonet al.2014; Dhillonet al.2017; Zouet al.2022).For the calculation of PUE under long-termin situexperiments, Syerset al.(2008) proposed the use of equilibrium methods that include information about the seasonal effects and the cumulative release effects of soil P.This method avoids the errors caused by the decrease of yield and P output year by year in the no-fertilization treatments (Roweet al.2015).Previous studies have shown that PUE calculations using the equilibrium method generally exceed 80% (Roberts and Johnston 2015);and a PUE of 100% means that there is little change in P in the soil, a PUE of less than 100% means that the P balance is in surplus, and a PUE greater than 100%indicates that the P balance is deficient (Dhillonet al.2017).Therefore, the PUE should not exceed 100%;otherwise, the sustainability of crop productivity will be compromised.This relationship explains the need for proper external P sources in agricultural systems.In this study, the P balance and PUE were calculated from P input and P output so that the upper limit of the P balance was converted to a P input threshold while also predicting PUE.To achieve the goal of sustainable agricultural development, a reference PUE value range of 90.0 to 94.9% was proposed in this study (Fig.4-B).The PUE estimated in this study avoids extreme cases (PUE<50%,PUE>100%) and provides a possibility for sustainable P management in agriculture.Therefore, quantifying the relationship between P input and output can provide information on the P balance, PUE and potential P losses in agricultural systems to assess whether the fertilization patterns are reasonable.

4.3.Regional P management strategy using the relationship between the soil Olsen-P value and the P budget

The results of long-term experiments are important for the interpretation of P tests in soil, and the temporal variations in tested soil P and different responses of crops to P availability should be given more consideration in making P fertilizer recommendations (Xiet al.2016;Zickeret al.2018).Previous studies have shown that the P input-output balance helps in assessing the future changes in Olsen-P values and adjusting the existing P fertilizer application recommendations (Jagdeep and Brar 2022), and the soil Olsen-P levels in different regions are essential references for applying critical P thresholds (Wuet al.2020; O’Donnellet al.2021).Previous studies have shown that when the Olsen-P content exceeds a critical value, the CaCl2-P content increases sharply, leading to a significant increase in the risk of P loss, and this critical value is the critical environmental risk value (CEV)(Hesketh and Brookes 2000; Khanet al.2018).Many studies have shown that the CEV of the fluvo-aquic soil in North China is approximately 25 mg kg-1(Liuet al.2020;Qinet al.2020; Zhanget al.2020).The P management flow chart (Fig.6) shows the recommended P balance and P fertilizer application strategy for fluvo-aquic soil in North China at different soil Olsen-P levels.The relationship between the P balance and the soil Olsen-P value (Fig.5-C) was used to accurately predict the changes in the Olsen-P value in three different situations:(i) When the soil Olsen-P levels are below the CPV (7.92 mg kg-1; Fig.5-D), the P input should be higher than the P output in the system, and the P fertilizer application rate(95.7-101 kg P2O5ha-1) determined by the P balance threshold (2.15-4.45 kg P ha-1) can increase the crop yield and soil Olsen-P value (0.0814-0.163 mg kg-1per year).(ii) When the Olsen-P levels are between the CPV and CEV, the P balance should not exceed the upper limit.The recommended P balance is in the range of 0-2.15 kg P ha-1, which allows for a high crop yield and P use efficiency with less P input.(iii) A soil Olsen-P value above the CEV can provide sufficient P for the crop, and in a typical winter wheat-summer maize rotation in North China, the annual P output with adequate soil P was 39.7 kg P ha-1(Fig.4-A).In the absence of P fertilization,the soil accumulated P to maintain high crop yields for a certain amount of time (soil Olsen-P decreased by 1.62 mg kg-1per year), providing valid information for agricultural P management.

Fig.6 Flow chart of the suitable phosphorus (P) fertilizer application management system.CPV, the critical soil Olsen-P value(7.9 mg kg-1 in this study); CEV, the critical environmental risk value (25 mg kg-1 in this study).

Previous studies have shown that the accumulated P in the soil can replace a large portion of otherwise applied P fertilizer and is a great potential source of available P for crops (Sattariet al.2012; Yang and Yang 2021).Moreover, when the soil P is sufficient, the crop demand for P fertilizer can be temporarily disregarded,and the crop yield can remain stable for many years after fertilizer application stops (Qaswaret al.2020;Vandermoereet al.2021).Previous researchers have predicted trends in the Olsen-P values based on the CPV and the soil Olsen-P values in relation to the cumulative P balance (Zhanget al.2019), which indicate that it would take approximately 9-11 years for the Olsen-P levels to decline from 286-320 mg kg-1to the average CPV of maize and wheat in the region.Roweet al.(2015)found that accumulated P in soils could meet global crop P requirements for approximately 9 to 22 years.Hence,the potential of accumulated P in agriculture to support future agricultural production needs to be more carefully considered, which is consistent with the findings reported by Powerset al.(2016) regarding accumulated P.Thus,to maintain sustainable crop productivity and higher PUE under intensively cultivated maize-wheat cropping systems, changes in the soil Olsen-P and P balance should be regularly monitored for regulating fertilizer P rates.

There is considerable uncertainty and variability in the chemical behavior of P fertilizers as they enter the soil (Syerset al.2008).The potential influences of local climatic conditions and soil properties should be considered in fertilization decisions for sustainable P management on croplands.The critical soil Olsen-P value, P balance and P input thresholds must be modeled for the different regions and cropping systems in China due to its large arable land area and the complexity of variations in soil fertility status.Here, our proposed method for determining the critical soil Olsen-P value,critical P balance and P application can help to improve crop yield and PUE while reducing P losses, which can provide a reference for developing agricultural P management measures in other regions and countries.

5.Conclusion

Field measurements of the winter wheat-summer maize yield, P input, P output, soil Olsen-P values and P balance over 12 years in northern China showed that the P balance is a robust indicator for identifying the turning points of maximum yields and is valuable for optimizing the P application rate.We observed stagnation in crops and a significant decrease in PUE when the P fertilizer application exceeded a certain level.The ranges of the critical P balance and the P input of the winter wheat-summer maize rotation system predicted in this study were 2.15-4.45 and 95.7-101 kg P2O5ha-1, respectively, which could increase crop yield and reduce the risk of P loss at a high PUE level(90.0-94.9%).In addition, the P input-output balance helps in assessing future changes in Olsen-P values and adjusting existing P fertilizer application recommendations.The CPV in the region and potential influences of the local climate, soil properties, and other factors should be adequately considered in P fertilizer application decisions.Overall, our results can provide an important reference for the efficient use of P fertilizer, P pollution risk reduction,and agricultural P management.

Acknowledgements

This study was funded by the National Key Research and Development Program of China (2021YFD1700900).

Declaration of competing interests

The authors declare that they have no conflict of interest.

Appendicesassociated with this paper are available on https://doi.org/10.1016/j.jia.2023.05.030