Phosphorus accumulation and leaching risk of greenhouse vegetable soils in South

Yusef KIANPOOR KALKHAJEHBiao HUANG*Helle SØRENSENPeter E.HOLM and Hans Christian B.HANSEN

1 Anhui Province Key Laborator y of Far mland Ecological Conser vation and Pollution Prevention，School of Resources and Environment，Anhui Agricultural University，Hefei 230036(China)

2 Depar tment of Plant and Environmental Sciences，University of Copenhagen，Thor valdsensvej 40，Frederiksberg DK-1871 C(Denmark)

3 Key Laborator y of Soil Environment and Pollution Remediation，Institute of Soil Science，Chinese Academy of Sciences，Nanjing 210008(China)

4 Sino-Danish Center for Education and Research(SDC)，Beijing 101408(China)

5 Data Science Lab，Department of Mathematical Sciences，University of Copenhagen，Universitetsparken 5，DK-2100 Copenhagen East(Denmark)

ABSTRACT Over-fertilization has caused signif icant phosphorus(P)accumulation in Chinese greenhouse vegetable production(GVP)soils.This study，for the f irst time，quantif ied prof ile P accumulation directly from soil P measurements，as well as subsoil P immobilization，in three alkaline coarse-textured GVP soil prof iles with 5(S5)，15(S15)，and 30(S30)years of cultivation in Tongshan，Southeast China.For each prof ile，soil samples were collected at depths of 0—10(topsoil)，10—20，20—40，40—60，60—80，and 80—100 cm.Phosphorus accumulation was estimated from the difference in P contents between topsoil and parent material(60—100 cm subsoil).Phosphorus mobility was assessed from measurements of water-soluble P concentration(PSol).Finally，P sorption isotherms were produced using a batch sorption experiment and f itted using a modif ied Langmuir model.High total P contents of 1 980(S5)，3 190(S15)，and 2 330(S30)mg kg-1 were measured in the topsoils versus lower total P content of approximately 600 mg kg-1 in the 80—100 cm subsoils.Likewise，topsoil PSol values were very high，varying from 6.4 to 17.0 mg L-1.The estimated annual P accumulations in the topsoils were 397(S5)，212(S15)，and 78(S30)kg ha-1 year-1.Sorption isotherms demonstrated the dominance of P desorption in highly P-saturated topsoils，whereas the amount of adsorbed P increased in the 80—100 cm subsoils with slightly larger P adsorption capacity.The total P adsorption capacity of the 80—100 cm subsoils at a solution P concentration of 0.5 mg L-1 was 15.7(S5)，8.7(S15)，and 6.5(S30)kg ha-1，demonstrating that subsoils were unable to secure P concentrations in leaching water below 0.5mg L-1 because of their insufficient P-binding capacity.

Key Words: greenhouse vegetable production，Langmuir model，P adsorption capacity，P desorption，P immobilization，P mobility，subsoil，topsoil

INTRODUCTION

Freshwater eutrophication is a signif icant concern in China as it causes degradation of water quality and results in unsustainable development(Huanget al.，2012;Niet al.，2016).Agricultural production through high rates of fertilization and manure application has led to nitrogen(N)and phosphorus(P)discharges that threaten the aquatic reservoirs with eutrophication(Carpenteret al.，1998).Chinese greenhouse vegetable production(GVP)systems represent very intensive agricultural production systems with high rates of fertilization and manure amendment，and therefore these systems may act as hot spots for nutrient release to fresh water(Chenet al.，2013;Yanet al.，2013).This challenge is growing because of the rapid expansion of GVP in China.In 2013，more than 2 500 000 ha of agricultural land in China were dedicated to GVP(MAC，2014).Regional P mobilization under often very high topsoil P contents in Chinese GVPs is well documented(e.g.，Qinet al.，2010;Kalkhajehet al.，2017)，although the actual P accumulation in these soils has seldom been measured directly.Previous studies have assessed the P balance in topsoil through differences between estimated P input(fertilizers and manure)and crop P removal.Such balance indicates annual P accumulations of 588 and 186 kg ha-1year-1in Chinese GVP soils used for fruity and leafy vegetables，respectively(Yanet al.，2013).Extreme P accumulations of more than 1 000 kg ha-1year-1have been reported for GVPs in Beijing，China(Chenet al.，2004);extreme P accumulations ranging from 80 to 5 000 kg ha-1year-1have been reported with manure as the main P input in Shandong，East China(Yuet al.，2010).Such accumulation rates of P are 10—200 times higher than those for excessively fertilized European soils，for example，25 kg ha-1year-1in Danish agricultural soils(Rubæketal.，2013)and 15 kg ha-1year-1in UK agricultural soils—f igures that are still considered high(Witherset al.，2001).High P accumulation coupled with frequent tillage and irrigation may stimulate soil P loss(Kinget al.，2015)，particularly in Southeast China with its relatively high annual precipitation of more than 1 000 mm(Shiet al.，2009).If subsoils do not have sufficient P-bonding affinity，P may leach to groundwater and/or surface water.The subsoil P-holding capacity was introduced in an earlier study with respect to restricting P mobility in acidic Templeton and Lismore soils in New Zealand with f ine sandy loam and stony silt loam textures，respectively(Sinajet al.，2002).Likewise，P leaching from Swedish clayey to sandy soils was found to be inf luenced by subsoil P adsorption capacity along with preferential f low pathways bypassing adsorption in subsoils(Djodjicet al.，2004;Anderssonet al.，2013).As Chinese GVP plots are mostly located in planar landscapes with little topography，subsurface leaching is considered to be the primary pathway of soil P loss(Kalkhajehet al.，2017).For measuring GVP soil P accumulation based on estimated inputs of fertilizers and manure，the P contents of these amendments have often been obtained only through interviews with the farmers.Furthermore，soil P accumulation and immobilization and the corresponding soil test P(STP)values and P adsorption capacities have been poorly studied for alkaline coarsetextured Chinese GVP soils(Kalkhajehet al.，2017，2018).Data for soil prof ile P and P-holding capacity and strength are critical for understanding P accumulation in different soil layers and for predicting P saturation and mobilization.Hence，this study aimed to i)compare prof ile P accumulation and saturation in three alkaline coarse-textured heavily fertilized GVP soils from Southeast China and ii)assess the capacity of subsoils to prevent P leaching from topsoils using P sorption isotherms.Results of this study can be used to develop sustainable agricultural management practices through revising fertilization and irrigation schemes to minimize P accumulation and leaching and to protect freshwater resources.

MATERIALS AND METHODS

Site selection and soil collection

Tongshan County in Jiangsu Province was selected as the study area because it represents a typical GVP base in Southeast China.The annual mean temperature and precipitation in this area are 14.5°C and 832 mm，respectively(Yang L Qet al.，2014).Soils of the greenhouse plots in this region are Ustic Cambosols(Entisols)with soil texture ranging from sandy loam to silty loam(CRGCST，2001).Illite and smectite are the dominant minerals in these soils(Xiong and Li，1987).The greenhouses in Tongshan County have been used for vegetable cultivation for almost 30 years and cover an area of approximately 670 ha.The dominant vegetables in Tongshan GVPs are cabbage(Brassica oleraceavar.capitata)，pakchoi(Brassica chinensisL.)，celery(Apiumgraveolensvar.dulce)，tomato(Lycopersicon esculentum)，cucumber(Cucumis SativusL.)，and pepper(Capsicum annuumL.).According to the interviewed farmers，the main fertilizers in Tongshan GVP are pig manure，cow manure，compound fertilizer，and manufactured water-soluble P fertilizer at estimated P application rates of 187，579，96—256，and 132—440 kg ha-1year-1，respectively.Three GVP plots of 5(S5)，15(S15)，and 30(S30)years of vegetable cultivation were sampled in July 2016.In each GVP plot，a bulk soil sample composed of f ive subsamples was taken from each of six different depths，0—10(topsoil)，10—20，20—40，40—60，60—80，and 80—100 cm，using a stainless steel auger.Soil samples were placed in plastic bags and transferred to laboratory.

Soil analysis

Collected soil samples were air dried and sieved(＜2 mm)for further analysis.Soil pH was measured at a 1:2.5(weight/volume)soil:water ratio using a glass electrode pH meter(PHS-3C，Shanghai Precision Scientif ic Instruments Co.，Ltd.，Shanghai，China).Soil electrical conductivity(EC)was determined at a 1:5(weight/volume)soil:water ratio using a conductivity meter(DDS-307，Shanghai Precision Scientif ic Instruments Co.，Ltd.，Shanghai，China).Soil organic matter(SOM)content was measured using the Walkley-Black wet oxidation method(Nelson and Sommers，1996).Laser diffraction(Coulter LS230，Beckman Coulter，Inc.，Brea，USA)was employed to determine soil particle size distribution.Soil total N(TN)content was measured using the Kjeldahl method(Pageet al.，1982).Our archive data suggested bulk densities(BD)of 1.08，1.52，1.60，and 1.62 kg m-3for the depths of 0—10，10—20，20—40，and 60—100 cm in the Tongshan GVP soils，respectively.To determine total P(TP)，soil samples were digested with concentrated H2SO4，HNO3，and HClO4at 180°C for 1 h(Lu，2000)，and TP content was measured using the molybdenum blue method(Murphy and Riley，1962).

For Olsen P determination，soil samples 2.5 g each were extracted for 30 min using 50 mL 0.5 mol L-1sodium bicarbonate solution(pH 8.5)(Sims，2009a)，and the P concentration in the extract was measured using the molybdenum blue method(Murphy and Riley，1962)after supernatant f iltration through a 16-μm f ilter paper(Whatman No.43).Oxalate P，aluminium(Al)，and iron(Fe)were determined by extracting 2.5 g soil with 50 mL 0.2 mol L-1ammonium oxalate solution(pH 3.0)for 2 h in darkness(Schoumans，2009).Mehlich III P，Al，Fe，and calcium(Ca)were extracted with 20 mL Mehlich III extracting solution per 2 g of soil for 5 min(Sims，2009b).After f iltration through 16-μm f ilter paper(Whatman No.43)，metals and P in oxalate and Mehlich III extracts were measured using inductively coupled plasma(ICP)-mass spectrometry(ICP-MS)(X7，Thermo Scientif ic，Madison，USA)with the limits of detections of 0.01，0.01，0.01，and 0.02 mg L-1for P，Ca，Al，and Fe，respectively.Olsen and Mehlich III are suitable methods for extracting P from alkaline soils because of lowering of Ca activity and acidity of the solution，respectively(Sen Tranet al.，1990;Iatrouet al.，2014).In addition，Olsen and Mehlich III P values can be used to compare the P accumulation with those of other studies in Europe and North America，as the two methods are used in many studies.The ammonium oxalate method is suitable for extracting P associated with amorphous Fe and Al oxides(Hoodaet al.，2000).Furthermore，Mehlich III and ammonium oxalate are multi-element extractants that can be analyzed by ICP to reduce the analysis time，and they are also widely used to determine the degree of soil P saturation(e.g.，Hoodaet al.，2001;Wanget al.，2012).

The water-extractable P concentration(PSol)was determined by adding deionized water to 50 g air-dried soil until 50%of the maximum water-holding capacity was reached，and the mixture was left to equilibrate for 48 h.Afterwards，soils were wetted to 100%maximum water-holding capacity，mixed well，and left for an additional 48 h.Finally，the moist soil samples were transferred to 50 mL polyethylene vials and centrifuged at 3 580×gfor 10 min.The supernatant was collected and f iltered through a 0.45-μm polyethersulfone syringe f ilter(Nanjing Ronghua Scientif ic Equipment Co.，Ltd.，Nanjing，China)，and the P concentrations in the supernatant were measured colorimetrically using the molybdenum blue method(Murphy and Riley，1962;Kalkhajehet al.，2018)with a P detection limit of 0.02 mg L-1.All the analyses were performed in triplicate.

Phosphorus sor ption experiment

A batch sorption experiment was conducted for each soil prof ile，using samples from three selected depths of 0—10，20—40，and 80—100 cm to gain insight into P sorption and P saturation as a function of soil depth.Air-dried soil samples 2.5 g each were placed in 50-mL polyethylene centrifuge tubes and equilibrated with a 25 mL solution with initial P concentration of 0，5，15，30，50，or 100 mg L-1.After two drops of 0.1%chloroform were added as microbial inhibitor(Detenbeck and Brezonik，1991)，each centrifuge tube was shaken at 180 r min-1for 24 h using an end-over-end shaker at room temperature(25±1°C).After equilibration，the suspensions were centrifuged at 3 580×gfor 10 min，and P was determined in the collected and f iltered supernatant as described above.All the analyses were performed in triplicate.

Fitting of P sor ption isotherms

The Langmuir sorption isotherm equation was used to describe the relationship between the sorbed and solution P concentrations:

whereQtis the total amount of P sorbed per unit soil mass(mg kg-1)，Ceqis the solution equilibrium P concentration(mg L-1)，Qmaxis the maximum P that can be sorbed(mg kg-1)，andKis the binding constant representing P sorption strength(L mg-1).The amount of P sorbed for each dosing of P(Q，mg kg-1)was calculated as follows:

whereCadis the P concentration added in the solution(mg L-1)，Vis the solution volume(mL)，andWis the soil mass(g).

According to the classic Langmuir model，if zero P is added(Cad=0)，then the equilibrium P concentration is also zero(Ceq=0)and henceQt=0.In our study，the equilibrium P concentration with no P added(C0)was signif icant，owing to the presence of native sorbed P(Qn)particularly in topsoils.Given that part of the native soil P(Qn)also participates in the sorption equilibrium，it should be included in Eq.1(Zhouet al.，2005).Therefore，Qtis the sum ofQnandQas follows:?

Combining Eqs.1—3，we can get Eq.4:

andQncan then be calculated as follows:

A combination of Eqs.1，3，and 5 results in the following modif ied Langmuir model，which in turn was used to f it P sorption isotherms:

where 10 is the value of solution:soil ratio(volume/weight)in this study.The values ofC0were measured for the samples with zero P added.The values of the Langmuir sorption parameters(QmaxandK)and their standard errors were estimated through nonlinear regression using Eq.6 in R(R Core Team，2017).Separate regression was conducted for each of the nine measurement series.

The point of zero net sorption(EPC0)indicates neither P adsorption nor P desorption(that is，Cad=Ceq).On the basis thatCad=Ceq=EPC0，the EPC0values were calculated using the following equation derived from the modif ied Langmuir model(Eq.6):

RESULTS

Soil properties

Selected physicochemical properties of the GVP soil prof iles are presented in Fig.1 and Tables I and S1(see Supplementary Material for Table SI).No signif icant differences were found in the pH，SOM，and TN values among the three GVP soil prof iles(Fig.1).Overall，these GVP soils had pH values of approximately 8.0 or higher with lower pH in topsoils than in subsoils(as high as 9.1 at a depth of 40—60 cm in S15).High SOM contents of 33.1±0.51(S5)，27.8±1.2(S15)，and 30.4±0.4(S30)g kg-1for the topsoils were followed by substantial SOM decreases for the subsoils below the depth of 40 cm.Likewise，TN was higher in the topsoils than in the subsoils with topsoil contents of 2.17±0.06(S5)，2.01±0.04(S15)，and 1.77±0.05(S30)g kg-1.Topsoil EC was much higher for S5(1 260μS cm-1)than for S15(225μS cm-1)and S30(269 μS cm-1)，and soil EC values decreased with depth only for S5.

The clay and sand contents showed a f luctuating trend with increasing soil depth(Fig.1，Table S1).Contrary to the rather uniform subsoil clay and sand contents of the GVP prof iles，the subsoils were evidently coarser textured for S15 and S30 with sand contents of 400 g kg-1，whereas the subsoil of S5 had a much lower sand content of 86 g kg-1.The clay contents of the three GVP topsoils and subsoils ranged between 50 and 100 g kg-1.

Fig.1 Selected physicochemical properties at depths of 0—10(topsoil)，10—20，20—40，40—60，60—80，and 80—100 cm in three greenhouse vegetable production soil prof iles with 5(S5)，15(S15)，and 30(S30)years of vegetable cultivation.Values are means with standard deviations shown by vertical bars(n=3).The value for each depth is given for the middle of that depth.OM=organic matter;EC=electrical conductivity.

The content of topsoil Mehlich III Ca ranged from 5 420±22 to 9 800±490 mg kg-1and was a quarter to a third of the subsoil contents of 24 200±470 to 28 400±190 mg kg-1(Table I).Fluctuations were apparent in the contents of oxalate Fe and Al.For S5，oxalate Fe increased from 897±42 mg kg-1in topsoil to 1 530±45 mg kg-1in subsoil.For S30，topsoil oxalate Al and Fe contents of 467±6 and 1 100±20 mg kg-1were substantially higher than subsoil contents of 173±9.4 and 385±20 mg kg-1，respectively.In S15，topsoil and subsoil had similar contents of oxalate Fe and Al.Signif icant positive and negative correlations were found between oxalate Fe and Al and clay and sand contents，respectively(P＜0.01).

Soil P contents and pools

The GVP soil prof ile distributions of TP as well as STP values are illustrated in Fig.2.In general，very high topsoil P contents were followed by dramatic decreases in the subsoils.In the topsoils，the sequence for TP and STP was S15＞S30＞S5，with TP contents of 3 580，2 750，and 2 450 mg kg-1，respectively.Almost the same low TP contents of approximately 600 mg kg-1were measured in the subsoils of the three prof iles.The topsoil Olsen P contents were between 26 and 138 times higher than those in the subsoils.Likewise，topsoil Mehlich III P contents were more than 200 times higher than those in the subsoils.

TABLE I Mehlich III Ca，Al，and Fe(Ca M 3，Al M 3，and Fe M 3，respectively)and oxalate Al and Fe(Al Ox and Fe Ox，respectively)at depths of 0—10(topsoil)，10—20，20—40，40—60，60—80，and 80—100 cm in three greenhouse vegetable production soil prof iles with 5(S5)，15(S15)，and 30(S30)years of vegetable cultivation

Fig.2 Total P，Olsen P，Mehlich III P，and oxalate P at depths of 0—10(topsoil)，10—20，20—40，40—60，60—80，and 80—100 cm in three greenhouse vegetable production soil prof iles with 5(S5)，15(S15)，and 30(S30)years of vegetable cultivation.Values are means with standard deviations shown by vertical bars(n=3).The value for each depth is given for the middle of that depth.

Consistent with the soil TP and STP，very high topsoil PSolvalues of 17.0(S5)，13.0(S15)，and 6.4(S30)mg L-1were followed by values below the detection limit in subsoils of all GVP soil prof iles(Fig.3).However，for S15 and S30，subsoil PSolvalues of 0.26 and 0.29 mg L-1were observed down to the depth of 80 cm，whereas it was approximately 0.10 mg L-1below the depth of 20 cm for S5.

Fig.3 Water-extractable P concentration at depths of 0—10(topsoil)，10—20，20—40，40—60，60—80，and 80—100 cm in three greenhouse vegetable production soil prof iles with 5(S5)，15(S15)，and 30(S30)years of vegetable cultivation.Values are means with standard deviations shown by vertical bars(n=3).The value for each depth is given for the middle of that depth.

Soil P accumulation

After a signif icant decrease in TP and STP within the top 40 cm in all GVP soil prof iles，negligible changes were observed from 60 to 100 cm(Fig.2).Hence，the mean TP and STP of the 60—100 cm subsoil were taken as reference values representing the P contents in the parent material(PM).With these values，soil TP or STP accumulation at the depth of 0—10，10—20，or 20—40 cm(Ad，kg ha-1)was calculated as follows:

whereCdand CPMrepresent the TP or STP content(mg kg-1)at the depth of 0—10，10—20，or 20—40 cm and in PM，respectively，BDdis the BD of the depth of 0—10，10—20，or 20—40 cm(kg m-3)，Dis the depth(m)，and 104in m2ha-1，103in L m-3，and 106in kg mg-1are for gettingAdon the basis of kg ha-1.

The TP and STP accumulations were divided by the duration of GVP to calculate the average annual P accumulations in the GVP soil prof iles(Fig.4).The topsoil TP accumulations of S5，S15，and S30 were 1 980，3 190，and 2 330 kg ha-1，respectively，resulting in average annual TP accumulations of 397，212，and 78 kg ha-1year-1，respectively.The TP accumulations at the depths of 10—20 and 20—40 cm were also very high，though lower than those in the topsoils.The TP accumulations in the top 40 cm were 3 330(S5)，7 040(S15)，and 5 200(S30)kg ha-1，resulting in average annual TP accumulations of 666，470，and 174 kg ha-1year-1，respectively.In the topsoils，the average annual Olsen P accumulations were 115(S5)，25(S15)，and 8(S30)kg ha-1year-1，the average annual Mehlich III P accumulations were 207，65，and 22 kg ha-1year-1，respectively，and the average annual oxalate P accumulations were 300，93，and 45 kg ha-1year-1，respectively.

Fig.4 Mean annual total P(TP)and soil test P(STP，including Olsen，Mehlich III，and Oxalate P)accumulations at the depths of 0—10(topsoil)，10—20，and 20—40 cm，calculated using Eq.8，taking the TP and STP contents at the depth of 60—100 cm as reference values representing the P contents in the parent material，in three greenhouse vegetable production soil prof iles with 5(S5)，15(S15)，and 30(S30)years of vegetable cultivation.

Phosphorus sor ption isotherms and parameters

The P sorption isotherms of the three GVP soil prof iles could be adequately f itted with the modif ied Langmuir model(Eq.6).As shown in Fig.5，P sorption isotherms differed signif icantly among the individual GVP soil prof iles and with soil depths.Isotherms for highly P-saturated topsoils in S5 and S15 showed much desorption，exhibiting desorption over the full solution concentration range tested and hence no ability to adsorb further P at solution P concentrations of 50—100 mg L-1.In contrast to isotherms of the topsoil，isotherms of the less P-rich subsoils showed higher adsorption affinity(steeper isotherm slopes)and higher maximum adsorption capacities particularly for S5 and S15 at the depths of 60—80 and 80—100 cm，although desorption was still observed at low solution P concentrations.

Signif icant differences were found among the modif ied Langmuir P sorption parameters for the individual GVP soil prof iles and as a function of depth(Table II).TheC0values of 8.64(S5)，5.13(S15)，and 2.90(S30)mg L-1in the topsoils were much higher than those of 0.00，0.06，and 0.04 mg L-1in the subsoils，respectively，ref lecting high desorption in the topsoils.Similar observations were obtained for EPC0with values below the limit of detection(desorption throughout)，54.2，and 11.3 mg L-1in the topsoils and values of 0.00，0.14，and 0.12 mg L-1in the subsoils for S5，S15，and S30，respectively.As indicated in Fig.5，the isotherms for the topsoils of S5 and S15 showed desorption over the full solution P concentration range tested;S30 did not adsorb P until the solution P concentration reached 11 mg L-1.The topsoil in S30 showed a lowQmaxvalue of 108 mg kg-1(Table II).For the 80—100 cm subsoils，theQmaxvalues decreased with longer history of vegetable cultivation with the meanQmaxvalues of 515，290，and 116 mg kg-1for S5，S15，and S30，respectively.The LangmuirKvalues f luctuated from topsoil to subsoils，ranging from 0.014 to 0.067 L mg-1.

Fig.5 Sorption isotherms of P f itted with the modif ied Langmuir model(Eq.6)for the depths of 0—10(topsoil)，20—40，and 80—100 cm in three greenhouse vegetable production soil prof iles with 5(S5)，15(S15)，and 30(S30)years of vegetable cultivation.

TABLE II Estimated P sorption parameters a)using the modif ied Langmuir model(Eq.6)and the P solution concentration at zero sorption(EPC0，Eq.7)for the depths of 0—10(topsoil)，20—40，and 80—100 cm in three greenhouse vegetable production soil prof iles with 5(S5)，15(S15)，and 30(S30)years of vegetable cultivation

DISCUSSION

Soil P accumulation

In this study，P accumulation was determined directly from the difference in P contents between each depth of top 40 cm and parent material，assuming that the current P values at the depth of 60—100 cm represent the P contents at the time of parent material deposition.Previous studies have shown substantial annual P accumulations in Chinese GVP soils from 6 to 1 069 kg ha-1year-1(e.g.，Chenet al.，2004;Feiet al.，2011)，triple the range of this study(Table III，Fig.4).Unlike our study，these studies estimated P accumulation based on differences between P inputs and crop P removal.Prior to the greenhouse vegetable cultivation，the soils in this study were under conventional wheat-rice rotation with estimated mean topsoil TP content of 950±50 mg kg-1as measured in adjacent croplands.Hence，simple subtraction of this historic TP content from the current soil TP and use of Eq.8 resulted in net TP accumulations of 1 620，2 840，and 1 940 kg ha-1in the topsoils of S5，S15，and S30，respectively，equivalent to the calculated average annual topsoil TP accumulations of 324，190，and 65 kg ha-1year-1，respectively，after GVP establishment.The annual P amendment for S5，S15，and S30 may have varied over time owing to variations in types of manure and management used，meaning that the average annual accumulation may cover a higher variation in accumulation for specif ic years.Therefore，we cannot draw conclusions on annual P accumulation for the current fertilization scheme.However，the highest annual rate of accumulation was obtained for S5，the soil prof ile that had the shortest cropping history.This may indicate that P amendment rates increased over time.It should also be noted that our calculations did not account for P leaching loss and hence the accumulation or“surplus”P calculated represented minimum estimates.Our f indings suggested that Chinese GVP soils were massively P accumulated compared to other soils listed in Table III.For instance，annual P accumulation in the Richmond River catchment(0—30 cm)in Australia ranged from-0.32 to 4.46 kg ha-1year-1(McKee and Eyre，2000)，whereas the accumulation was 0.19，20，and 6 kg ha-1year-1in acidic sandy clay loam(0—5 cm)in Northern Ireland(Watson and Matthews，2008)，acidic soil(0—7.5 cm)from sheep grazing pasture in New Zealand(Tianet al.，2017)，and neutralalkaline GVP soils(0—20 cm)in Liaoning，Northeast China(Feiet al.，2011)，respectively.Nevertheless，the average annual P accumulations in our GVP soils were comparable with those of neutral-alkaline GVP soils in Beijing，China(Chenet al.，2004)and acidic-neutral agricultural soils in subtropical area of China(Menget al.，2018)of 11—1 069and 252—906 kg ha-1year-1，respectively.Although large quantities of manure and inorganic fertilizers led to signif icant topsoil(0—10 cm)P accumulation(Kalkhajehet al.，2017)，both TP and STP values were still high at the depths of 20—40 cm compared to values in the deeper subsoil(60—100 cm)(Fig.2).Frequent tillage and deep soil cultivation of the greenhouse plots have contributed to P enrichment in the 80—100 cm subsoils.This is verif ied by a high content of SOM consistent with manure amendment.Furthermore，the high P saturation and low remaining-P adsorption capacity of top soil layers may result in P leaching to the subsoils(Wanget al.，2015).The marked decreases of all STP values below the depth of 40 cm for all GVP soil prof iles suggested that P was not available and hence was probably present in primary minerals，strongly sequestered by Fe and Al oxides，or bound to Ca in the subsoil layers(Table I，Fig.1).Other studies have also reported high P adsorption in Ca-rich alkaline coarsetextured soils studied.For instance，the P adsorption capacity was signif icantly correlated to Mehlich III Ca in alkaline sandy soils in northern China(Xueet al.，2014).Similar results were obtained for sandy soils(pH＞8)in Padova，Italy(Pizzeghelloet al.，2011).This effect is ascribed to the abundance of exchangeable Ca ions and Ca solids such as calcite that contribute to P precipitation or chemisorption of P(Samadi and Gilkes，1999;Vepraskas and Faulkner，2001).Nevertheless，the impact of pH on P distribution between soil solution and solid phases is complex owing to presence of numerous sorbents and solutes at the same time(White，1980;Staunton and Leprince，1996).

TABLE III Examples of P accumulation in cropped soils

Phosphorus sor ption isotherms and P solubility

Large differences were evident in P sorption isotherms for the GVP soil prof iles，particularly S5 and S15 with pronounced topsoil P desorption and stronger subsoil P adsorption(Fig.4).These differences are related to the soil P saturation，which in turn depends on the soil P adsorption capacity.If oxalate P represents the existing mobilizable but already adsorbed P pool andQmaxrepresents the additional P that can be sorbed and estimated from the sorption isotherms，then the P saturation(%)can be estimated as follows:

According to this equation，the topsoil of S30 had a high P saturation of 92%.The topsoils of S5 and S15 were fully saturated as indicated by extensive desorption in the sorption experiment(Fig.5).Fully or almost fully P-saturated topsoils result in high solution P concentrations leaching to subsoils(Yli-Hallaet al.，2002).In contrast，the 80—100 cm subsoils，which contained less P，had P saturation of 25%(S5)，40%(S15)，and 42%(S30).Furthermore，a decrease in the 80—100 cm subsoil SOM content may cause less competition for binding sites of P sorbents，consequently increasing P adsorption capacities(Yang G Ret al.，2014;Andersenetal.，2015).Nonetheless，it appeared that subsoils in the GVP soil prof iles，particularly S15 and S30，did not have sufficient affinity for P adsorption at low solution P concentrations.Therefore，the 80—100 cm subsoils will not act as P traps at environmental P threshold concentrations of 0.05 or 0.10 mg L-1，as judged by their desorption at solution P concentration of 0 mg L-1in the sorption experiment(Fig.5).Our results of sorption isotherms together with the setting of an environmental solution P concentration can be used to estimate the total P adsorption capacity of subsoil layers.

Threshold solution P concentration and environmental implications

Continuous fertilization in excess of subsoil P adsorption capacity will cause subsurface P leaching and consequently eutrophication of receiving surface waters.This phenomenon may occur in coarse-textured Chinese GVP soils with very high potential for P mobilization(Kalkhajehet al.，2018).The column leaching experiment of this study showed an estimated annual P leaching between 7 and 12 kg ha-1year-1from the GVP soils(Ustic Cambosols)tested and that Olsen P above 41 mg kg-1led to exponential increase in P leaching from these soils(Kalkhajehet al.，2017).Hence，we expected to observe signif icant quantities of P leaching from the GVP topsoils.Hereby，the critical aspects are the subsoil capability to catch P mobilized from the topsoil and minimize overall leaching from the soils to ground-or surface waters.We set the threshold solution P concentration at 0.5 mg L-1based on a desire to limit P leaching and protect sensitive water bodies against eutrophication.The total amount of P adsorbed(kg ha-1)in the 80—100 cm subsoils at that threshold concentration can be calculated using Eq.10:whereQ′is the amount of P that can be adsorbed at the threshold concentration(read from isotherms for the deepest soil layer)(mg kg-1)，BD is set at 1.62 kg m-3(as for the depths of 60—100 cm)，Dis set at 0.20 m(from 0.8 to 1.0 m)，and 104in m2ha-1，103in L m-3，and 106in kg mg-1are for getting P sorbed on the basis of kg ha-1.Using Eq.10，the 80—100 cm subsoil P adsorption values for S5，S15，and S30 were 15.7，8.7，and 6.5 kg ha-1，respectively.Therefore，the 80—100 cm subsoil has little remaining capacity to adsorb P before the threshold is exceeded and P will leach at solution P concentrations exceeding 0.5 mg L-1.Field studies are needed to provide more precise values for the subsurface Pholding capacity.The transport routes for P through subsoils to drainage channels，creeks，and groundwater should be mapped，and the role of surface runoffand signif icance of preferential f low pathways should be examined.The subsoil that adsorbed P in the leachate water from the top soil layers may be thicker than 20 cm in our calculation，and hence more P could be sorbed.For example，assuming a thickness of 0.40 m(from 0.6 to 1.0 m)using the sorption isotherms for the deepest soil layers，then 31.4，17.3，and 13.0 kg ha-1of P could be sorbed by S5，S15，and S30 subsoils，respectively.These values are still low.Irrespective of this，the extremely P-enriched GVP topsoils were clearly a signif icant source of P export to the surrounding environments.

According to the f indings of this study，alternative management practices should be adopted to reduce soil P accumulation and mobilization.As such，fertilization particularly through manure application should be optimized or even stopped in heavily P-saturated soils.Simultaneously，attention must be paid to Ninputs as high P accumulation is partly due to soil N inputsviamanure amendments containing too much P.Promoting farmers’awareness with respect to the environmental consequences of their current P fertilization practices is crucial.Accordingly，the irrigation rate and frequency should be minimized through replacement of surface irrigation with drip irrigation.Finally，tile drains equipped with P-adsorbing materials such as dolomite，ocher，slag，concrete，and dedicated f ilter(McDowellet al.，2008;Lyngsie，2013;Vandermoereet al.，2018)may be established in GVP f ields to catch and recycle the mobilized P.These measures would improve the sustainability of GVP and protect freshwater resources.

CONCLUSIONS

This study addressed P accumulation and risk of P loss for three alkaline coarse-textured Chinese GVP soil prof iles of 5，15，and 30 years of cultivation.Overall，annual TP accumulation in top soil layers ranged from 70 to 400 kg ha-1year-1，suggesting signif icant P input to the GVP soilsversusscant crop P removal.However，P(TP and STP)accumulation values decreased by 7%to 60%below the depth of 40 cm compared with topsoil.Likewise，very high soil PSolvalues suggest a very high risk of P leaching.Regardless of cultivation history，sorption isotherms indicated the dominance of desorption in the topsoils and adsorption in the subsoils.However，subsoils were unable to catch P from a leaching solution when the threshold solution P concentration was set at 0.5 mg L-1，owing to insufficient P adsorption capacity or low P affinity of the subsoils.Our results point to the substantial risk of P leaching and hence degeneration of reservoir water from Chinese alkaline coarse-textured P-rich GVP soils if their subsoils do not have large P adsorption capacity or affinity.

ACKNOWLEDGEMENTS

The authors would like to acknowledge the f inancial support from the Sino-Danish Center for Education and Research(SDC).Furthermore，we are grateful for the f inancial support of the Special Research Foundation of the Public Natural Resource Management Department from the Ministry of Environmental Protection of China(No.201409044)and the National Natural Science Foundation of China(No.41473073).We are also thankful to the lab technicians of the Key Laboratory of Soil Environment and Pollution Remediation，Institute of Soil Science，Chinese Academy of Sciences，who assisted us with sample preparation and analyses.

SUPPLEMENTARY MATERIAL

Supplementary material for this article can be found in the online version.