Changes in soil organic carbon pools following long-term fertilization under a rain-fed cropping system in the Loess Plateau,China

WANG Ren-jie,SONG Jia-shan,FENG Yong-tao,ZHOU Jiang-xiang,XIE Jun-yu,Asif KHAN,CHE Zong-xian,ZHANG Shu-lan,YANG Xue-yun

1 Key Laboratory of Plant Nutrition and the Agri-environment in Northwest China,Ministry of Agriculture and Rural Affairs/College of Natural Resources and Environment,Northwest A&F University,Yangling 712100,P.R.China

2 Baoji Extension and Service Centre of Agricultural Technology,Baoji 721001,P.R.China

3 College of Natural Resources and Environment,Shanxi Agricultural University,Taigu 030801,P.R.China

4 Institute of Soil,Fertilizer and Water-saving Agriculture,Gansu Academy of Agricultural Sciences,Lanzhou 730070,P.R.China

Abstract Understanding the mechanism of soil organic carbon (SOC) sequestration is of paramount importance in sustaining crop productivity and mitigating climate change.Long-term trials were employed to investigate the responses of total SOC and its pools,i.e.,mineral-associated OC (MOC),particulate OC (POC,containing Light-POC and Heavy-POC),to fertilization regimes at Yangling (25-year),Tianshui (35-year) and Pingliang (37-year) under a rain-fed cropping system in the Loess Plateau.The fertilization regimes in each trial included three treatments,i.e.,control (no nutrient input,CK),chemical fertilizers (CF),and organic manure plus chemical fertilizers (MCF).Relative to the CK,long-term fertilization appreciably increased SOC storage by 134,89 and 129 kg ha–1 yr–1 under CF,and 418,153 and 384 kg ha–1 yr–1 under MCF in plough layer soils (0–20 cm),respectively,at the Yangling,Tianshui and Pingliang sites.The MOC pools accounted for 72,67 and 64% of the total SOC at the above three sites with sequestration rates of 76,57 and 83 kg ha–1 yr–1 under CF and 238,118 and 156 kg ha–1 yr–1 under MCF,respectively.Moreover,the MOC pool displayed a saturation behavior under MCF conditions.The POC accordingly constituted 27,33 and 36% of SOC,of which Light-POC accounted for 11,17 and 22%and Heavy-POC for 17,16 and 15% of SOC,respectively.The sequestration rates of POC were 58,32 and 46 kg ha–1 yr–1 under CF,and 181,90 and 228 kg ha–1 yr–1 under MCF at the three respective sites,in which Light-POC explained 59,81 and 72% of POC under CF,and 60,40 and 69% of POC under MCF,with Heavy-POC accounting for the balance.Compared with CK,the application of CF alone did not affect the proportions of MOC or total POC to SOC,whereas MCF application markedly reduced the proportion of MOC and increased the POC ratio,mainly in the Light-POC pool.The distribution of SOC among different pools was closely related to the distribution and stability of aggregates.The present study confirmed that organic manure amendment not only sequestered more SOC but also significantly altered the composition of SOC,thus improving SOC quality,which is possibly related to the SOC saturation level.

Keywords:chemical fertilizer,organic manure,mineral-associated OC,particulate OC,Light-POC,Heavy-POC

1.Introduction

Soil organic carbon (SOC) is the most important indicator of soil quality due to its significant effect on soil physical,chemical,and biological properties,thereby affecting crop production and yield variability (Panet al.2009).The enhancement of SOC sequestration also plays a vital role in targeting the French initiative of ‘4 per 1 000’ and thus helps to alleviate the potential global climate change (Baveyeet al.2018).Simultaneously,agricultural practices,such as fertilizer applications,have a profound impact on crop yields and carbon inputs,and then further influence the storage of SOC (Conget al.2012;Maillard and Angers 2014;Zhouet al.2017;Zhanget al.2019).

The SOC storage reflects the long-term balance between the C input and decomposition (Maillard and Angers 2014;Liuet al.2019),and its stabilization mechanisms include physical,chemical and biochemical aspects that are related to soil texture,soil aggregation and the chemical composition of SOC (Sixet al.2002;Duet al.2014).A linear relationship between an increase in SOC content and C input has been documented in many agricultural soils,especially those with low to moderate SOC contents (Zhanget al.2010;Liuet al.2013;Fenget al.2014;Heet al.2018).However,soils with high SOC levels showed no further gain in SOC even after receiving an enhanced C addition,and there exists a so-called soil C saturation (Sixet al.2002;Guldeet al.2008;Duet al.2014;Liu Cet al.2014).The physical fractionation method is a commonly used approach to study SOC stabilization mechanisms (Degryzeet al.2004;Guldeet al.2008;Eet al.2012).The total SOC is generally separated into two major pools,namely labile and stabilized pools (Haynes 2005).The labile pool,like POC,influences soil function in specific ways and is believed to be much more sensitive to changes in soil management practices(Janzenet al.1997).The decomposition of the labile pool drives the flux of CO2from soils to the atmosphere,and this process is highly important as it fuels the soil food web and strongly influences nutrient cycling and crop productivity(Chanet al.2001;Mandalet al.2010).For these reasons,the labile pool of organic matter is regarded as an important index of soil organic matter quality,i.e.,a higher proportion of labile organic carbon to SOC implies an improved quality of OC in the soil (Haynes 2000;Malhiet al.2005).The stabilized pool,like the mineral-associated OC (MOC),is highly resistant to microbial decomposition because of the physical and/or chemical protection of soil minerals(Lützowetet al.2007) and makes up the majority of the SOC (Haynes 2005).Many previous studies found that with an increase in C input,the stabilized carbon pool of a soil saturates;as a result,additional C input can only be accumulated in labile C pools that have a relatively faster turnover (Koolet al.2007;Guldeet al.2008;Duet al.2014).This implies that changes in SOC quality could be related to the degree of SOC saturation,especially the MOC pool to a certain extent.

Previously,contradictory results have been reported on the accumulation of total SOC and its pools (POC and MOC)after application of recommended chemical fertilizers (CF).For example,some studies have shown that the long-term application of CF significantly increases the total amount of SOC and its POC and/or MOC pools (Eet al.2012;Mannaet al.2013;Chaudharyet al.2017).Meanwhile,some other researchers have reported that the long-term application of CF has no effects on SOC,POC or MOC contents in soils under different cropping systems and climate conditions(Divitoet al.2011;Yanet al.2013;Chatterjeeet al.2017;Weiet al.2017).Nevertheless,most of the previous studies documented that application of CF combined with organic manure can greatly increase the SOC and POC and/or MOC contents (Eet al.2012;Mannaet al.2013;Baiano and Morra 2017;Liet al.2018),although the magnitudes vary among studies.In addition,long-term manure incorporation,especially under continuous high C input conditions,leads to stable soil C pool (MOC) reaching saturation levels (Guldeet al.2008;Duet al.2014).These results suggest that stabilization of SOC and its pools under various fertilization regimes for a given soil depends largely upon climate,soil types,cropping systems and fertilization practices.

The Loess Plateau in north-western China,covering 63 000 km2,is an important dryland farming area in the country.Soils derived from the loess materials are widely distributed in this region,and these soils have more favorable particle-size distributions and mineralogical compositions than those developed from other sediments(Catt 2001).A single crop per year,such as winter wheat,spring maize or potato,is the prevailing cropping system under rain-fed conditions in the region.Previous studies mostly investigated the effects of long-term fertilization on total SOC storage or labile carbon pools in an individual experiment (Yanget al.2011,2012;Liuet al.2013;Liu E Ket al.2014;Xieet al.2018),however,very little information is available about the role the SOC pools plays in SOC stabilization or modification of SOC quality.Filling this knowledge gap will help to establish judicious field management practices to improve both SOC level and quality,thereby serving the sustainable development of regional agriculture.Therefore,this study aimed to investigate the mechanisms of SOC stabilization and changes in SOC quality based on long-term experiments at multiple sites in the Loess Plateau.We hypothesized that long-term application of chemical fertilizers either alone or in combination with organic manure could increase total SOC,but the distribution of sequestered OC into the C pools or the roles of OC pools in stabilizing SOC may differ between these two fertilization regimes due to the variations in their C input and soil environmental conditions;and these differences may subsequently further alter the SOC quality.

2.Materials and methods

2.1.Study sites and experimental design

The basic information and soil properties of the three long-term experiments under rain-fed cropping systems in the Loess Plateau,northwestern China are given in Table 1.There are seven treatments in total at Yangling site,eight at Tianshui and six at Pingliang.For this study,three treatments were selected at each site.Details of the precise quantities of fertilizers/organic manure applied in the treatments are presented in Table 2.Each treatment has three replicates except for at Yangling,where only one was used for practical reasons (Yanget al.2011).

Table 2 Experimental treatments and rates of fertilizers applied at the three sites in the Loess Plateau,China

Table 1Informationof experimental sitesandsome properties of theploughlayersoils(0–20cm)priorto thestartof thelong-termexperimentsin theLoessPlateau,China1)Claycontent(%)31.62 28.04 25.35 Siltcontent(%)43.67 39.2 38.79 Sandcontent(%)24.71 32.76 35.86 pH 8.62 8.54 8.2(H2O)TN(gkg–1)0.94 0.82 0.95 SOC(gkg–1)7.33 9.14 5.2 Meanannual precipitation (mm)562 500 474 Meanannual temperature (°C)13 11.5 9.5 Experimental period(year)Latitudeand longitude 34°17´N,108°00´E 25 (1990–2015)34°05´N,104°05´E 35 (1981–2016)35°16´N,107°30´E 37 (1979–2016)Experimentalsite Yangling Tianshui Pingliang 1) The contents of sand,silt andclay presentedaremeansof unfertilized controls at each site.

2.2.Cropping systems

The crops planted at Yangling site were winter wheat (TriticumaestivumL.) followed by soybean(GlycinemaxL.) in 1991–1998.Because drought frequently occurred at the time of soybean planting,which caused crop failure in most of the years,only winter wheat as a single crop per year,has been planted since 1999.At Pingliang site,cropping started with two years of single maize(ZeamaysL.),then 4–5 years of winter wheat from 1979 to 1998,soybean in 1999 and sorghum in 2000 (SorghumbicolorL.).Since 2000,winter wheat monoculture has been continuously practiced.At Tianshui site,the winter wheat has been planted with break crops of winter oil seed rape (BrassicanapusL.) in 1987,2000 and 2003,and flax (LinumusitatisimumL.) in 1991.

2.3.Soil sampling

Soil samples were collected in September 2015 at Yangling site,and in June 2016 at Pingliang and Tianshui sites.At Yangling site,each plot was divided into three parts of equal size as three pseudo-replicates.Three undisturbed soil cores (10 cm in inner diameter and 10 cm in height)were collected with an increment of 10 cm down to 20 cm from each replicated plot and bulked to give a composite sample,and three composite samples were generated at each depth for each treatment at all sites.Meanwhile,the soil bulk density was measured for each sample.The bulk sample was screened through an 8-mm sieve and air-dried,then subsamples were further sieved through 2-mm and 0.15-mm for analysis.

2.4.Soil organic carbon fractionation and analysis

The procedure of soil organic carbon fractionation was adopted from Degryzeet al.(2004).Briefly,a 60-g air-dried sample (<2 mm) was dispersed into 100 mL of 5% sodium hexametaphosphate solution and shaken for 15 h.The soil suspension was poured into a 0.053 mm sieve and the retained material (POC) was rinsed with distilled water until the water ran clear,and then oven-dried at 60°C and weighed.One portion of POC was used for determination of total OC.The other was placed in a 250-mL centrifuge tube with 50 mL of NaI solution with a density of 1.85 g cm–3and thoroughly mixed,then centrifuged to collect Light-POC.The collected Light-POC was rinsed thoroughly with 75 mL 0.01 mol L–1CaCl2and deionized water to remove NaI,transferred into a dry pan,and oven-dried at 60°C.The Light-POC was also used for organic C determination.The OC concentration of the Heavy-POC was estimated as differences in OC concentrations between total POC and Light-POC pools.The concentration of MOC was calculated by the difference between total OC and POC.

2.5.Determination of soil properties

Organic C in bulk soil was determined by potassium dichromate (K2Cr2O7) oxidation at 170–180°C followed by titration with 0.1 mol L−1ferrous sulfate (Bao 2005).Organic C in the POC and Light-POC pools were measured with a Vario MACRO cube elemental analyzer (Elementar Analysensysteme GmbH,Hanau,Germany).Total N concentration was measured by the Kjeldahl method after concentrated H2SO4digestion in the presence of K2SO4-CuSO4-Se as a catalyst (Bremner 1996).Olsen-P was determined with the ammonium molybdate method described by Murphy and Riley (1962).Soil exchangeable K was extracted with 1 mol L–1ammonium acetate (pH 7.0)and measured with a flame photometer (Li 1983).Soil inorganic carbon was measured with a titrimetric method as described by Bundy and Bremner (1972).Soil pH was determined with a pH electrode at a soil-to-water ratio of 1:2.5.Soil particle-size distribution was determined with the pipette method as described by Miller and Miller (1987).Aggregates were separated with wet sieving by passing the air-dried soil through three stacked sieves with decreasing mesh sizes of 2,0.25 and 0.053 mm to isolate four aggregate size fractions following the method described by Elliott(1986).Briefly,a 50 g sample of each soil passed through an 8-mm sieve was submerged into deionized water on top of a 2-mm sieve for 5 min prior to sieving.Sieving was carried out by gently raising the sieve by 3 cm and lowering it again 50 times over a 2-min period.The fraction that remained on each sieve was collected in a pre-weighed aluminum pan and dried overnight (50°C).

2.6.Data analysis

The storage of the SOC and its pools were calculated using the following equations:

where SOCcontentis SOC content in bulk soil (g C kg–1soil),h is soil depth (m),BD is soil bulk density (Mg m–3) and 10 is a unit conversion.The MOCcontent,POCcontent,Light-POCcontent,Heavy-POCcontentare OC contents (g C kg–1) in the corresponding SOC pools.Mratiois mass proportion of OC in the given pool to SOC (g OC pools g–1soil).

The sequestration rates of SOC pools were calculated as follows:

where OCfertilizerand OCcontrolare OC storage in fertilized (CF or MCF) and control treatments,respectively,in bulk soil and its pools,and t is the period of experimentation (year).

Aggregate stability was expressed as mean weight diameter (MWD):

where W is the proportion of aggregates in each size class.

One-way ANOVA was used to evaluate the effects of fertilization and sites on storage of SOC and its pools,and proportions of OC pools to SOC.WhenF-values were significant,multiple comparisons of means were performed with the least significant difference method(LSD) at 0.05 probability.The above statistical analyses were completed with the SPASS 18.0 Software (SPSS Inc.,Chicago,IL,USA).Moreover,redundancy analysis(RDA) was employed to analyze the relationships between the proportions of OC pools and soil properties across the three sites with Canoco ver.5.0,and only those parameters with significant relationships between them are presented in the figures.

3.Results

3.1.Soil properties

At 0–10 cm soil depth,compared to CK,the application of CF or MCF showed a significant impact on aggregate distribution.Moreover,CF markedly increased MWD by 9 and 15% at Yangling and Pingliang;and MCF enhanced MWD by 7,16 and 16% at Yangling,Tianshui and Pingliang,respectively(Table 3).At 10–20 cm soil depth,long-term fertilization also greatly influenced aggregate distribution at all sites (except for Yangling) but had no significant impact on MWD.

Table 3 Aggregate distribution (%) and mean weight diameter (MWD) of each treatment at 0–10 and 10–20 cm depths across three sites in the Loess Plateau,China

Soil pH was lower under CF treatment than under CK by 2 and 3% at 0–10 and 10–20 cm depths at Tianshui and by 3% at 10–20 cm depth at Pingliang.But the application of MCF significantly decreased pH relative to CK by 2,2 and 3% at 0–10 cm depth,and by 5,3 and 3% at 10–20 cm depth,respectively,at Yangling,Tianshui and Pingliang sites (Table 4).In addition,CF markedly increased soil total N relative to CK by 32,9 and 16% at 0–10 cm,and by 31,13 and 12% at 10–20 cm,respectively,at Yangling,Tianshui and Pingliang sites.MCF also greatly increased total N by 77,23 and 72% at 0–10 cm,and by 64,27 and 71% at 10–20 cm at the above respective three sites.Compared to CK,CF treatment also considerably enriched Olsen-P by as much as 4.35,2.02 and 5.15-fold at 0–10 cm,and by 2.80,1.71 and 5.52-fold at 10–20 cm depth,respectively,at Yangling,Tianshui and Pingliang sites.The application of MCF drastically enhanced Olsen-P by as much as 15.43,5.86 and 14.84-fold at 0–10 cm,and by 16.94,6.94 and 19.11-fold at 10–20 cm at the above three sites.The MCF treatment also enhanced soil exchangeable K (AK) by 129,30 and 139% at 0–10 cm,and by 101,10 and 148% at 10–20 cm,respectively,at Yangling,Tianshui and Pingliang sites.Long-term fertilization also influenced the total inorganic carbon (TIC) content to some extent at Yangling and Pingliang sites (Table 4).

TIC (g kg–1)11.12 ab 11.85 a 10.47 b 6.51a 5.79a 6.79a 9.07b 6.77c 10.26 a 167.94b 171.17b 337.72a 155.00b 130.74b 171.17a 133.98b 121.04b 332.87a Table 4Some soil chemical properties of thetreatments involved at 0–10 and10–20cm depths across threesites in theLoessPlateau,China1)10–20 cm 5.96c 22.66 b 106.91a 4.53c 12.29 b 35.99 a 2.63c 17.14 b 52.90 a TN(g kg–1)Olsen-P(mgkg–1)AK (mgkg–1)0.83c 1.09b 1.36a 1.12c 1.26b 1.42a 0.85c 0.95b 1.45a pH 8.23a 8.08a 7.86b 8.25a 8.05b 7.98b 8.24a 8.09b 8.01c TIC (g kg–1)10.33 b 12.21 a 10.20 b 6.22a 6.51a 5.91a 8.38a 6.64b 9.09a 201.89c 269.81b 462.23a 158.23b 143.68b 205.13a 167.94b 137.21b 400.78a 0–10cm Olsen-P (mgkg–1)AK (mgkg–1)7.38c 39.49 b 121.23a 6.05c 18.28 b 41.49 a 4.81c 29.60 b 76.21 a TN(g kg–1)0.96c 1.27b 1.70a 1.18c 1.29b 1.45a 0.86c 1.00b 1.48a pH 8.10a 8.11a 7.93b 8.21a 8.11b 8.02b 8.21a 8.15a 7.98b Treatment2)CK CF MCF CK CF MCF CK CF MCF Site Yangling Tianshui Pingliang 1) TNrefers to totalnitrogen;AK denotesexchangeable potassium;TICmeanstotalinorganiccarbon.2) CKdenotescontroltreatmentwithoutreceivinganyfertilizers;CF refers to treatmentreceivingchemical fertilizers;andMCFstands fortreatmentreceivingorganicmanure plus chemical fertilizers.Different lowercaseletterswithin each column foraspecific site indicate significantdifferencesbetweentreatments at agivensoil depthat P<0.05.

3.2.SOC and its pools

Compared to CK,CF considerably raised the SOC storage by 11,14 and 35% at 0–10 cm depth at Yangling,Tianshui and Pingliang,and by 17 and 12% at 10–20 cm depth at Yangling and Pingliang,respectively (Table 5).Consequently,the sequestration rates of SOC were 55,51 and 88 kg ha–1yr–1at 0–10 cm,and 79 and 41 kg ha–1yr–1at 10–20 cm soil depth at the corresponding sites mentioned above (Fig.1).On average,the sequestration rates of SOC were 65 and 53 kg ha–1yr–1at 0–10 and 10–20 cm depths across the sites for the CF treatment.Addition of MCF greatly enhanced SOC storage by 43,27 and 74% at 0–10 cm depth,and by 43,17 and 59% at 10–20 cm depth,respectively,at Yangling,Tianshui and Pingliang sites (Table 5).Correspondingly,SOC was sequestered at rates of 213,92 and 187 kg ha–1yr–1at 0–10 cm and 205,61 and 197 kg ha–1yr–1at 10–20 cm depth at the above three sites (Fig.1).On average,the sequestration rates of SOC were 164 and 154 kg ha–1yr–1at 0–10 and 10–20 cm depths across the sites for the MCF treatment.

The MOC storage rates were statistically the same between CF and CK treatments at both soil depths at all sites except for 10–20 cm soil at Yangling,where MOC was sequestered at a rate of 57 kg ha–1yr–1(Fig.1).However,the application of MCF significantly increased MOC storage relative to CK by 29,26 and 46% at 0–10 cm soil depth at Yangling,Tianshui and Pingliang;and by 37 and 33% at 10–20 cm depth at Yangling and Pingliang,respectively (Table 5).The MOC was sequestered at rates of 109,57 and 77 kg ha–1yr–1at 0–10 cm at Yangling,Tianshui and Pingliang,and of 128 and 79 kg ha–1yr–1at 10–20 cm at Yangling and Pingliang (Fig.1).On average,the sequestration rates of MOC were 81 and 90 kg ha–1yr–1at 0–10 and 10–20 cm depths across all three sites.

The POC storage rates were also equal for CF and CK treatments at both soil depths at all sites except for 0–10 cm soil at Pingliang,where POC was sequestered at 28 kg ha–1yr–1(Fig.1).While the application of MCF markedly improved POC storage over CK by 86 and 127% at 0–10 cm depth at Yangling and Pingliang,and by 60,45 and 125%at 10–20 cm depth at Yangling,Tianshui and Pingliang,respectively(Table 5).The sequestered POC was 104 and 110 kg ha–1yr–1at 0–10 cm at Yangling and Pingliang;and 76,55 and 117 kg ha–1yr–1at 10–20 cm at Yangling,Tianshui and Pingliang,respectively (Fig.1).On average,the POC was sequestered at 84 and 83 kg ha–1yr–1under MCF treatment at 0–10 and 10–20 cm depths.

Long-term application of CF significantly increased the storage of Light-POC relative to CK at 0–10 cm depth by 49 and 67% at Yangling and Tianshui,but showed no effect on soils at 10–20 cm at all sites (Table 5).The sequestered Light-POC was 24 and 15 kg ha–1yr–1at 0–10 cm,respectively,at Yangling and Tianshui (Fig.1).By sharp contrast,the addition of MCF greatly increased the storage rates of Light-POC by 124,80 and 142% at 0–10 cm depth at Yangling,Tianshui and Pingliang sites,and by 142% at Yangling and 154% at Pingliang for 10–20 cm soil depth (Table 5).Correspondingly,the sequestration rates of Light-POC were 60,20 and 77 kg ha–1yr–1at 0–10 cm,and 49 and 81 kg ha–1yr–1at 10–20 cm at the respective sites stated above (Fig.1).On average,the sequestration rates of Light-POC were 48 and 50 kg ha–1yr–1at 0–10 and 10–20 cm depths under the MCF treatment.

Storage of Heavy-POC had no change on CF treatment relative to CK at either soil depth at any of the sites.However,MCF treatment drastically enhanced Heavy-POC storage over CK by 60 and 105% at 0–10 cm soil depth at Yangling and Pingliang,and by 85% at 10–20 cm soil depth at Pingliang site (Table 5).The sequestered Heavy-POC was 45 ha–1yr–1at Yangling and 34 kg ha–1yr–1at Pingliang at 0–10 cm,and 36 kg ha–1yr–1at Pingliang at 10–20 cm soil depth (Fig.1).

Fig.1 Sequestration rates of soil organic carbon (SOC) and its pools in soils at 0–10 and 10–20 cm depths receiving chemical fertilizers either alone or in combination with organic manure at Yangling,Tianshui and Pingliang sites in the Loess Plateau,China.CF,treatment receiving chemical fertilizers;MCF,treatment receiving organic manure plus chemical fertilizers.Different lowercase letters inside,at the top,and near the left side of the bars indicate significant differences in OC pools,SOC storage and POC storage,respectively,between treatments at P<0.05.The error bar represents standard deviation (n=3).

Table 5Soil bulk density (Mg m–3) andstorageof soil organic carbonand its pools(Mg ha–1) in bulksoils at 0–10 and10–20cm depths across three sites in the Loess Plateau,China1)2.37a 2.36a 3.06a 1.62a 1.57a 2.66a 1.57b 1.78b 2.91a Light-POC Heavy-POC 0.86b 0.96b 2.08a 2.22a 2.62ab 2.91a 2.16b 2.62b 5.48a POC 3.22b 3.32b 5.14a 3.84b 4.19b 5.57a 3.73b 4.40b 8.39a 10–20 cm MOC 8.78c 10.71 b 12.00 a 9.70a 9.54a 9.66a 8.75b 9.55b 11.68 a SOC 12.00 c 14.00 b 17.14 a 13.04 b 13.73 b 15.23 a 12.44 c 19.76 a BD 1.45a 1.37b 1.35b 1.23a 1.15a 1.15a 1.47a 1.33b 1.84b 2.13b 2.94a 2.73a 2.26a 2.73a 1.31b 1.70ab 1.45 ab 13.97b 2.69a Light-POC Heavy-POC 1.19c 1.77b 2.67a 1.56b 2.60a 2.81a 1.89b 2.53b 4.57a 0–10cmPOC 3.02b 3.91b 5.61a 4.30a 4.86a 5.53a 3.20c 4.23b 7.26a MOC 9.27b 9.76b 11.99 a 7.88b 9.06ab 9.90a 6.17b 8.39ab 8.99a SOC 12.29 c 13.67 b 17.60 a 12.17 c 13.92 b 15.43 a 9.36c 12.62 b 16.25 a BD 1.36a 1.15b 1.13b 1.10a 1.13a 1.16a 1.11b 1.29a 1.06b Treatment2)CK CF MCF CK CF MCF CK CF MCF Site Yangling Tianshui Pingliang 1) BDstands forbulk density;SOCmeanssoil organiccarbon;MOCrefers to mineralassociated organiccarbon;POCstands forparticulateorganiccarbon.2) CKdenotescontroltreatmentwithoutreceivinganyfertilizers;CF refers to treatmentreceivingchemical fertilizers;andMCFstands fortreatmentreceivingorganicmanure plus chemical fertilizers.Different lowercaseletterswithin each column foraspecific site indicate significantdifferencesbetweentreatments at agivensoil depthat P<0.05.

3.3.Distribution of SOC in OC pools

The MOC,Light-POC and Heavy-POC pools accounted for 68–75,10–15 and 15–17% of the total SOC at 0–10 cm soil depth,and 70–77,6–12 and 17–20% at 10–20 cm soil depth,respectively,at Yangling site (Fig.2).Those pools accounted for 64–65,13–19 and 16–22% at 0–10 cm and 63–74,14–20 and 12–17% at 10–20 cm at Tianshui site;and 55–66,20–28 and 14–17% at 0–10 cm and 59–70,17–27 and 13–15% at 10–20 cm at Pingliang site,respectively.In general,the MOC was the largest OC pool,accounting for 55 to 77% of the total SOC (Fig.2);the POC accounted for the remaining 23–45%,of which Light-POC and Heavy-POC constituted 6–28 and 12–22% of SOC across all sites.

The application of CF dramatically increased the proportion of Light-POC to SOC at 0–10 cm soil depth by 3 and 6% at Yangling and Tianshui,and by 5% at 10–20 cm depth at Tianshui but showed no effect on proportions of either MOC or POC to SOC as compared with CK at either soil depth at any of the sites (Fig.2).The MCF treatment drastically decreased the proportion of MOC to SOC relative to CK by 7% at Yangling and 11% at Pingliang at 0–10 cm and by 11% at 10–20 cm at both Tianshui and Pingliang sites (Fig.2).In contrast,MCF markedly improved the proportion of POC to SOC over CK by 7% at Yangling and 11% at Pingliang for 0–10 cm depth,by 11%at Tianshui and by 12% at Pingliang for 10–20 cm depth(Fig.2).Moreover,MCF treatment markedly increased the proportion of Light-POC relative to CK by 5,5 and 8% at 0–10 cm depth,and by 5,6 and 10% at 10–20 cm,respectively,at Yangling,Tianshui and Pingliang.Across the sites,the reduction of the MOC proportion wasca.8 and 12%,and the increase of the POC proportion was 8 and 12%,of which the Light-POC contributed 75 and 58%,respectively,at 0–10 and 10–20 cm soil depths over CK.

3.4.Relationship between OC pools and soil properties

Correlation analysis showed that the relationship between MOC and SOC was best fitted by the quadratic function,while those between the remaining OC pools and SOC could be described well by the linear function at each site and across sites except for Heavy-POC at Tianshui (P<0.05,Fig.3).The analysis of the correlations between proportions of OC pools to SOC and soil properties showed that axis 1 and axis 2 explained 43.6 and 10.6% of the total variation,respectively (Fig.4).There were positive correlations between proportion of MOC and bulk density (BD),MWD,silt plus clay content and proportion of macroaggregates(>0.25 mm).While proportions of POC and Light-POC had positive correlations with SOC content and proportion of microaggregates (0.053–0.25 mm).

Fig.3 Relationships between soil organic carbon (SOC) storage of the bulk soil and its pools at soil depths of 0–10 and 10–20 cm subjected to different long-term fertilization treatments at three sites and across sites in the Loess Plateau,China (P<0.05).MOC refers to mineral associated organic carbon;POC stands for particulate organic carbon.

Fig.4 Redundancy analysis (RDA) between proportions of soil organic carbon (SOC) pools and properties of soils at 0–10 and 10–20 cm depths across three sites in the Loess Plateau,China.CK,control treatment without receiving any fertilizers;CF,treatment receiving chemical fertilizers;MCF,treatement receiving organic manure plus chemical fertilizers.The length of each arrow indicates the contribution of the corresponding property to the structural variation.MWD denotes mean weight diameter;silt+clay (%) is the summation of silt and clay contents in particle size analysis;(0.053–0.25)% and >0.25%represent the proportions of aggregates of 0.053–0.25 mm(microaggregates) and >0.25 mm (macroaggregates) to bulk soil by wet sieving,respectively;BD stands for bulk density;MOC refers to mineral associated organic carbon;POC stands for particulate organic carbon.

4.Discussion

4.1.Effects of long-term fertilization on total SOC sequestration

We observed that both CF and MCF treatments could significantly increase SOC sequestration at the two soil depths tested.The annual mean sequestration rates of SOC in plough layer soils (up to 20 cm) were 89–134 and 153–418 kg ha–1yr–1,respectively,for CF and MCF treatments over CK among sites for the single cropping system under rain-fed conditions on the loess soil.These results were within the reported values ranging from–4 to 177 kg ha–1yr–1for CF treatment and from 104 to 641 kg ha–1yr–1for MCF treatment under much the same conditions (Fanet al.2008;Liuet al.2010).Whereas,our values were lower than those reported at the global scale,where a meta-analysis showed that the annual mean SOC sequestration rates were 211 and 522 kg ha–1yr–1for the addition of chemical fertilizers and the incorporation of organic supplements across cropping systems (Maillard and Angers 2014).However,our results were also much lower than those reported under double cropping systems,which were 53–356 and 242–1 158 kg ha–1yr–1for CF and MCF treatments (Bhattacharyyaet al.2008;Conget al.2012;Chaudharyet al.2017;Liet al.2018;Wanget al.2020).The variations in SOC sequestration among studies could be ascribed to the factors which may affect C input and decomposition,such as cropping system,crop yield and exogenous organic supplements,soil properties (texture,aggregation and initial OC level),soil tillage,experimental duration and climate conditions (Guldeet al.2008;Liu Cet al.2014;Maillard and Angers 2014;Xiaet al.2018;Guanet al.2019;Berhaneet al.2020).In this study,enhancement of SOC storage by CF application may be attributed to its significant contributions to the increasing crop yield,thereby more C input to soil from the roots,root exudates and stubble (Fanet al.2008;Eet al.2012;Liet al.2018;Wanget al.2020),and also to the beneficial effects from the improved aggregation as a result of the abovementioned reasons (Table 3).More SOC sequestration under MCF treatment might be directly related to the larger proportion of recalcitrant organic compounds contained in the applied organic manure (Drinkwateret al.1998;Limaet al.2009;Liu E Ket al.2014),and the additional increase in C input might also be linked to the greater crop yield,as well as the enhanced aggregation,which is similar to that for CF as stated above (Table 3).For the three sites tested here,the SOC sequestration rates were in a descending order of Yangling>Pingliang>Tianshui under both fertilized treatments.This variation may be caused by their differences in organic C inputs.For example,Yangling site had the highest yield level (unpublished data) because of its better climate (Table 1) and soil conditions (Tables 3 and 4),and the other two sites showed similar yields (Fanet al.2005;Eet al.2012).In addition,the quality/purity of the organic manure applied was much better at Yangling than at the other two sites (as inferred from Table 2).Moreover,the highest clay plus silt content at Yangling site (Table 3)might imply greater capacity to preserve C (Sixet al.2002;Zhanget al.2010).However,low precipitation and relatively high temperature as well as the greater initial SOC level at Tianshui site may lead to a high decomposition rate and low SOC sequestration rate (Zhanget al.2010).The observed higher annual SOC sequestration rate at Yangling site might also be due to its relatively shorter experimental time of 25 years versus the 35 or 37 years for the other two sites.

4.2.Effects of long-term fertilization on sequestration of OC pools

Overall,the long-term application of CF had no effect on MOC or POC stocks compared to CK at either soil depth at any of the sites (Table 5),but greatly increased Light-POC stocks at 0–10 cm depth for both Yangling and Tianshui sites (Table 5).Some previous studies reported that 15-year application of CF significantly increased POC and/or MOC contents relative to CK (Eet al.2012;Mannaet al.2013;Chaudharyet al.2017),whereas others found that CF had no effects on POC or MOC contents of various soils under different cropping systems and climate conditions (Divitoet al.2011;Yanet al.2013;Chatterjeeet al.2017;Weiet al.2017).The variation of results in different studies or sites like in this study might be ascribed to the comprehensive conditions including the C input level and the related decomposition environment (Bruunet al.2010;Wanget al.2020).In general,the C inputs are likely to accumulate in labile organic carbon rather than in MOC (Fenget al.2014).Our results indicated that the C inputs in the CF treatment were not enough to modify the MOC pool in loess soils under a dry-land cropping system.

The application of MCF significantly increased storage rates of all OC pools at both soil depths and all sites.This observation is consistent with those reported by others (Eet al.2012;Mannaet al.2013;Fenget al.2014;Baiano and Morra 2017;Liet al.2018).Nevertheless,the sequestration rates were different from site to site.For example,the highest rate of MOC sequestration was at Yangling,followed by that at Pingliang,and then the lowest at Tianshui;while Pingliang site presented the highest rates for POC and Light-POC,followed by those at Yangling,and the lowest at Tianshui.The capacity of the MOC pool for a given soil is mainly related to its content of clay and silt particles (Degryzeet al.2004).The highest sequestration rate of MOC at Yangling was probably related to its highest clay and silt content and greater amount of C input (Tables 1 and 2) as discussed above in the SOC section.Nevertheless,we found no further increases in MOC storage with increasing SOC storage (Fig.3).These findngs demonstrated that soil mineral fractions under the tested conditions had possibly reached C saturation after 25–37 years of manure incorporation at all sites.Previous studies have also documented C saturation behavior for the silt-clay particles,and that once the capacity associated with clay-silt was saturated,further SOC accumulation would be found mainly in the labile C pools (e.g.,POC pools)(Guldeet al.2008;Duet al.2014).The linear relationship between POC pools and SOC further supported the above C saturation behavior of the MOC pool.This can help explain the greater rates of POC and Light-POC sequestration at Pingliang than at Yangling because of the lower capacity of MOC at the former site (Table 1).

4.3.Effects of long-term fertilization on the distribution of OC pools

We found that the application of CF to loess soils showed no effect on the distribution of SOC into MOC and POC pools at either soil depth at any site,but it significantly changed the OC distribution within POC pools at Yangling and Tianshui sites (Fig.2).The MCF markedly changed the distribution of SOC,which was manifested by the considerably lower MOC proportion and obviously greater POC and Light-POC proportions (Fig.2).These results confirmed our hypothesis that the role the OC pool plays in SOC sequestration was shaped by fertilization regimes,especially those that have the most striking impacts on soil properties.

Although the addition of chemical fertilizers could drastically promote crop performance and thus lead to an increase in C input into soil (Table 5),it was probably not sufficient to make changes in SOC distribution into MOC and POC pools (only modifying Light-POC to certain extent)(Table 5;Fig.2),or it need more time to do so (Balesdentet al.1998;Haiet al.2010;Lianget al.2012).It is worth noting that the crop residues were largely removed in our cases.Besides the C input,other factors such as physical protection by soil aggregation (Duet al.2014) or the decomposition mechanisms through enzymes and microbial communities might also impact the distribution of OC in different pools(Tianet al.2017).In contrast,sufficient C supply under the integrated application of chemical fertilizers and organic supplements promoted the carbon transformation into the MOC pool,which probably reached the saturation level(Fig.3),and more OC could be accumulated in the POC pool (Sixet al.2002;Guldeet al.2008;Duet al.2014),thereby modifying the SOC composition (Fig.2).The shift in SOC composition has been deemed as an indicator of SOC quality alteration,and soils with a larger proportion of labile C pool are believed to have better SOC quality(Haynes 2000;Malhiet al.2005;Lützowetet al.2007).Our results suggested that the long-term application of chemical fertilizers made no modification of SOC quality relative to the control (similar ratios of total POC to SOC);but the long-term integrated application of chemical fertilizers and organic manure improved both SOC quantity and quality.The change in SOC quality was tightly related to aggregate distribution,i.e.,significant positive correlations between the proportions of Light-POC and micro-aggregates(0.053–0.25 mm),and between proportions of Heavy-POC and macro-aggregates (>0.25 mm) (Fig.4),which might imply that the labile C was largely protected by both macroand micro-aggregates in the tested soils.Our results on the changes in SOC quality further support the concept of hierarchical carbon saturation (Guldeet al.2008),reflecting the additional C inputs being accumulated in labile C pools after the mineral C pool reached saturation (Guldeet al.2008;Duet al.2014).However,the saturation of the MOC pool might imply the likely saturation of SOC in bulk soil,in which case C stabilization efficiency can be very low (Wanget al.2020),and most of the C inputs into soils would largely emit as carbon dioxide,which is not environmentally friendly.

5.Conclusion

Long-term application of chemical fertilizers significantly increased SOC storage,while the distribution of SOC into POC and MOC pools was equal to that of control,thus it made no change in the SOC quality of the tested loess soils.In contrast,long-term integrated application of chemical fertilizers and organic manure substantially enhanced storage rates of SOC and its pools.Moreover,the MOC pool showed a saturation behavior,and more C accumulated into the POC pool,thus altering SOC composition and its quality.Hence,compared to the application of mineral fertilizers alone,integrated application of organic manure and chemical fertilizers could increase both SOC quantity and quality.Thus,further study may be needed to investigate the coordination of SOC quality and C stabilization efficiency to maintain the soil fertility and reduce C emission under a high C input environment.

Acknowledgements

This work was sponsored by the Ministry of Agriculture and Rural Affairs of China under Special funds for the Operation and Maintenance of Scientific Research Facilities(G202010-2).We are incredibly grateful to all the staff involved in the running and managing of the three long-term experiments.The authors also thank Dr.Zhigang Liu from AgSpace,UK for editing the English text.

Declaration of competing interest

The authors declare that they have no conflict of interest.