Laboratory drying of organic-matter rich soils- Phosphorus solubility

Laboratory drying of organic-matter rich soils:Phosphorus solubilityeffects,influence of soil characteristics,and consequencesfor environmental interpretationDavid Styles a,c,⁎,Catherine Coxon b,caDepartment of Botany,Trinity College,University of Dublin,Dublin 2,Ireland bDepartment of Geology,Trinity College,University of Dublin,Dublin 2,Ireland cCentre for the Environment,Trinity College,University of Dublin,Dublin 2,IrelandReceived 31August 2005;received in revised form 3March 2006;accepted 10March 2006Available online 21April 2006AbstractAgronomic soil test P (STP)and soil type data may be used to predict diffuse soluble P loss from soil to water (e.g.Withers,P.J.A.,Dils,R.M.,Hodgkinson,R.A.,1999.Transfer of phosphorus from small agricultural basins with variable soil types and land use.p.41.In:Impact of Land use change on nutrient loads from different sources.IAHS publication no.257;Daly,K.,Mills,P.,Coulter,B.,McGarrigle,M.,2002.Modelling phosphorus concentrations in Irish rivers using land use,soil type,and soil phosphorus data.J.Environ.Qual.,31:590–599).However,previous studies quantifying the effect of STP and soil type on P solubility have usually relied on laboratory analyses of dried soil samples to quantify P solubility.Drying soil samples has been found to substantially increase P solubility (e.g.Bartlett,R.,James,B.,1980.Studying dried,stored soil samples —some pitfalls.Soil Sci.Soc.Am.J.,44:721–724;Turner,B.L.,Haygarth,P.M.,2001.Phosphorus solubilization in rewetted soils.Nature,411:258–258),and microbial biomass and organic matter (OM)have been implicated in this (e.g.Chepkwony,C.K.,Haynes,R.J.,Swift,R.S.,Harrison,R.,2001.Mineralization of soil organic P induced by drying and rewetting as a source of plant-available P in limed and unlimed samples of an acid soil.Plant Soil,234:83–90;Turner,B.L.,Driessen,J.P.,Haygarth,P.M.,McKelvie,I.D.,2003.Potential contribution of lysed bacterial cells to phosphorus solubilisation in two rewetted Australian pasture soils.Soil Biol.Biochem.,35:187–189).Many grassland soils in western Ireland contain high OM and moisture contents,and may therefore be particularly susceptible to drying-induced soluble P release.This study quantifies drying-induced changes in P solubility,by measuring water-extractable P (P w )and desorbable P (P feo )contents in moist and dried samples of a range of 33,mainly grassland,soils from western Ireland.Molybdate analyses of P w extracts before and after acid digestion were used to quantify dissolved reactive P (P w DRP),taken to represent predominantly readily soluble inorganic P,and dissolved molybdate-unreactive P (MUP),taken to represent predominantly readily soluble organic P.Dried soil samples were analysed for a wide range of general and P-specific characteristics,including STP (Morgan P and Olsen P)and the degree of sorption saturation (DPSS).Based onOMGeoderma 136(2006)120–135Abbreviations:Alox,ammonium-oxalate extractable aluminium;DDW,double distilled water;DPSS1,degree of phosphorus sorption saturation calculated using P ox and PSC1;DPSS2,degree of phosphorus sorption saturation calculated stoichiometrically using P ox and PSC2;Fe ox ,ammonium-oxalate extractable iron;MC,moisture content;OM,organic matter;P feo ,iron-oxide strip desorbable phosphorus;P i ,inorganic P;P o ,organic P;P ox ,ammonium-oxalate-extractable phosphorus;PSC1,phosphorus sorption capacity calculated as sum of PSI and P ox ;PSC2,phosphorus sorption capacity calculated stoichiometrically from Al ox and Fe ox ;PSI,phosphorus sorption index (remaining phosphorus sorption capacity);P w DRP,water-extractable dissolved reactive phosphorus;P w MUP,water extractable (dissolved)molybdate-unreactive phosphorus;P w TDP,water-extractable total dissolved phosphorus;STP,soil test phosphorus;SOC,soil organic carbon;TEB,total exchangeable bases.⁎Corresponding author.Department of Botany,Trinity College,University of Dublin,Dublin 2,Ireland.Tel.:+35316083068.E-mail address:dstyles@tcd.ie (D.Styles).0016-7061/$-see front matter ©2006Elsevier B.V .All rights reserved.doi:10.1016/j.geoderma.2006.03.017content,soils were divided into peat(>30%OM)and predominantly mineral(<30%OM)soils after Daly et al.(2000).Almost all soils exhibited large increases in P solubility after drying,with the size of drying-induced soluble P increases increasing in the order P w DRP<P w MUP<P feo.Absolute drying-induced increases were statistically similar between peat and mineral soil groups(e.g.P feo mean increases of9.06and13.81for peat and mineral soil groups,respectively).However,when expressed as a proportion of moist sample P solubility,all soluble P increases were significantly greater for peat compared with mineral soils(e.g.P feo percentage increases735%and155%for peat and mineral soil groups,respectively).Consequently,some differentiation in P solubility between peat and mineral soil groups was lost after drying.Drying-induced P w increases appeared to be related to soil OM and moisture contents,indicating drying-induced P release through microbial cell lysis and OM destabilisation.Increases were not related to moist sample P solubility or STP,resulting in the loss of some differentiation in P solubility between high and low STP soils.Drying-induced P feo increases were positively related to DPSS,reflecting chemical equilibrium control of desorbable P released from dried samples.It was concluded that the standard practice of drying soils prior to P solubility analyses reduces the inferred importance of soil type and STP as risk factors in diffuse soluble P losses.©2006Elsevier B.V.All rights reserved.Keywords:Soils;Soil drying;Phosphorus solubility;Soil test phosphorus;Peat;Ireland1.IntroductionDiffuse losses of phosphorus(P)from soil to water have been implicated in eutrophication of freshwaters, and developing reliable risk indicators for diffuse soil P loss has become urgent for EU member states who must comply with strict water quality guidelines arising from the EU Water Framework Directive(CEC,2000). One important risk factor for diffuse P loss is soil test P (STP)as measured by agronomic soil P tests(e.g. Mozaffari and Sims,1994;Foy et al.,1995;Smith et al.,1995;Withers et al.,1999;Gburek et al.,2000).In the Republic of Ireland,Morgan P(sodium acetate at pH4.8)is the standard STP method,while in a number of other European countries,Olsen P(0.5molar sodium hydrogen carbonate at pH8.5)is the standard STP method.Both water-extractable P(P w)and iron-oxide strip P(P feo)were originally developed,and are still used in some countries,as agronomic STP methods;other countries use them for environmental testing.These tests may be taken to represent the risk of soluble P loss from soils,and have been successfully related to soluble P losses in the field(Sharpley et al., 1981;Yli-Halla et al.,1995;Pote et al.,1999b; McDowell and Sharpley,2001;Torrent and Delgado, 2001).P w was initially developed as a simple and soil-type universal agronomic P test by Van der Paauw(1971). Soluble P extracted in this test is usually operationally defined using filtration through0.45µm pores,though considerable quantities of fine colloidal P<0.45µm remain in solution and may be measured during subsequent P analysis(Haygarth et al.,1997;Matthews et al.,1998).P w may then be further classified according to reaction with acidic molybdate reagent in molybdate reactive P(MRP)and defined as‘dissolved reactive P’(DRP)or‘molybdate-unreactive P’(MUP)accordingly. DRP is considered to be predominantly soluble inorganic P(P i),though it may include significant quantities of organic P(P o)and condensed P hydrolysed during acidic MRP analysis(Rigler,1968),in addition to fine colloidal P.Total dissolved P(TDP)is determined by MRP analysis on filtered samples in which organic and condensed P(MUP)have been digested,and includes fine colloidal P.In this paper,P w fractions are referred to as P w DRP,P w MUP and P w TDP. Unfortunately,as outlined above,these operational definitions do not correspond precisely with theoretical definitions of soil P fractions.For comparative purposes in this paper,it is assumed that P w DRP represents predominantly water-soluble P i,P w MUP represents predominantly water-soluble organic and condensed P, and P w TDP represents the sum of these fractions.The P feo test was developed by H.A.Sissingh,and published by Van der Zee et al.(1987).It is used as an indicator of desorbable P i,and involves shaking soil solutions with an iron-oxide-impregnated paper strip, which acts as a constant sink for P i.However,some P o species also adsorb onto the iron-oxide strips,such as inositol hexakisphosphate,and these may be hydrolysed during acid extraction of the strip and subsequent MRP analysis of the extracts.Robinson and Sharpley(1994) found that considerable quantities of P o were extracted by the P feo test.It is assumed here that P feo measures predominantly desorbable P i,though also includes some desorbable P o.For convenience,soils are usually dried in prepara-tion for laboratory analyses,despite evidence that drying and rewetting soils alters their physical and biological characteristics and releases soluble P(Bartlett and121D.Styles,C.Coxon/Geoderma136(2006)120–135James,1980;Pote et al.,1999a;Turner and Haygarth, 2001;Styles et al.,2006).Large decreases in the number of viable microbial cells present in soils immediately after drying and rewetting(Stevenson,1956;Sparling et al.,1985;Kieft et al.,1987;Van Gestel et al.,1991; Turner et al.,2003)have been attributed to cell lysis induced by osmotic shock upon rewetting with a low solute solution such as water(Kieft et al.,1987). Increases in P w after drying have been found to be dominated by the MUP fraction(Turner and Haygarth, 2001;Turner et al.,2003),and increases in this fraction have been found to correlate with soil microbial P content(Sparling et al.,1985;Turner and Haygarth, 2001;Turner et al.,2003).Microbial activity has also been implicated in the release and transformation of P associated with organic matter(OM),through mineral-ization flushes observed after dried soils were rewetted (Stevenson,1956;Bartlett and James,1980;Van Gestel et al.,1991).These mineralization flushes were attributed to high concentrations of dissolved soil organic C(SOC)in rewetted soils.Bartlett and James (1980)found that SOC became increasingly soluble after drying,with increasing storage time,and attributed this effect to oxidation of organic compounds.Thus, oxidation of SOC caused by soil drying and storage may directly release associated P o,and indirectly release P i via stimulated mineralization.The size and composition of drying-induced bicar-bonate-extractable P(similar to Olsen P)increases (Brookes et al.,1982;Sparling et al.,1985;Srivastava, 1997;Turner and Haygarth,2003)led Turner and Haygarth(2003)to conclude that mechanisms other than microbial P release were responsible—specifically, solubilisation of OM through physical disruption of OM coatings.Other physicochemical processes such as oxidation and hydrolysis have been implicated in drying-induced release of SOC and elements associated with P-binding(Al,Ca,Fe and Mn)in soils(Raveh and Avnimelech,1978;Bartlett and James,1980;Ross et al., 1994;Phillips and Greenway,1998).Drying-induced increases in the inositol hexakisphosphate concentra-tions of water extracts from five Australian pasture soils also suggest physicochemical mechanisms of P release (Turner and Haygarth,2003).This organic P fraction behaves similarly to P i in forming strong complexes with Al,Fe and Ca compounds.Drying soils may also result in dissolution and desorption from specific surface areas exposed by aggregate and OM breakdown (Raveh and Avnimelech,1978;Olsen and Court,1982; Chepkwony et al.,2001).Some studies have found that acidic STP extraction methods are unaffected by drying soils(Pote et al.,1999a;Olde Venterink et al.,2002).Large drying effects on P solubility analyses could obscure relation-ships with STP and soil type,as used to develop soil soluble P loss indices.Although samples maintained at sampled moisture content may undergo some biochemical changes,and physical changes after moist sieving(e.g.Chapman et al.,1997b),they are likely to better represent soils in original field conditions compared with artificially dried samples. Irish grassland soils,with a high proportion of peats and gleys,typically maintain high moisture contents (MC)throughout much of the year and may be particularly susceptible to laboratory drying effects. Further understanding the effects of soil drying on P solubility is important in order to reliably relate abundant STP data,and soil management plans,to diffuse P loss risk.While the primary aim of work conducted in this study was to develop soil P desorption coefficients for use in the Mask catchment of western Ireland(Styles et al.,2006),a secondary aim evolved to examine the importance of soil drying.Therefore,although no experiments were designed to identify specific causes of drying effects on soil P solubility,the substantial quantity of data collected in this study enabled some empirical observations to be made.Thus,the aims of this paper are:(1)To quantify the effects of sample drying on Psolubility,as measured by P w and P feo,for highOM and MC Irish grassland soils.(2)To relate drying induced changes in differentsoluble P fractions with soil type and soilcharacteristics,in order to infer causal mechan-isms of drying-induced P release and assess thepredictive power of soil characteristics for drying-induced soluble P release.(3)To assess the implications of drying soil samplesfor the interpretation of environmental laboratoryP tests,and relationships with STP.2.Materials and methods2.1.Soil samplingThirty grassland fields and three forested areas were selected within the Mask catchment of County Mayo, western Ireland,to include a range of soil types and soil P levels(Table1).At each sampling location,a composite soil sample was systematically collected in a‘W’pattern across the area,composed of approxi-mately25cores of2cm diameter and5cm depth.122 D.Styles,C.Coxon/Geoderma136(2006)120–135Samples were stored at 4°C prior to being sieved through a 4mm brass sieve within 3days of collection.Sieved,homogenised soils were then divided into two sub-samples,‘moist ’and ‘dried ’.Moist samples were stored in sealed foil trays at 4°C in order to maintain them at sampled moisture content and minimise biochemical reactions prior to analyses over the succeeding few days,while the remaining sub-samples were spread within folded paper to dry at approxi-mately 25°C for at least 10days.Dried soils were then 2mm mesh sieved and stored in foil trays prior to analyses.Immediately upon sieving,sub-samples of the moist soils were analysed for MC by weight loss after 24h at 105°C.Moist soils were weighed out on an equivalent oven-dry weight basis for analysis,while air-dried soils were weighed out and small adjustments made in final calculations to correct for remaining moisture content.2.2.Expression of resultsAll values are the mean of duplicate analyses,except pH,total organic carbon (TOC)and total nitrogen (TN)(no replicates).All results were corrected for moisture content and expressed on an oven-dry weight basis,prior to conversion to a volume basis using the empirical relationship between OM and bulk density reported by Jeffrey (1970):BD ¼1:482−0:6786log X ðr ¼−0:905Þ;where BD is expressed as g ml −1and X represents percentage ignition loss at 500°C (OM content).While it is more usual for soil concentrations to be expressed on a weight,rather than volume,basis,volume-based expression is more appropriate for ecological compar-isons (Jeffrey,1970),especially among soils of differing bulk densities.It also more accurately reflects soil –Table 1Information on soil group (P=peat,M=mineral),major characteristics (organic matter,moisture content,total exchangeable bases,pH),soil test P (Morgan P and Olsen P)and P saturation (DPSS1and DPSS2)for the studied soils Soil no.Group OM (%)MC (%)TEB (meq)pH Pm (mg l −1soil)Olsen P (mg l −1soil)DPSS1(%)DPSS2(%)1P 63.269.98.8 5.1 2.9 4.114.0 3.42M 13.441.4 4.6 5.27.553.843.711.23M 18.650.012.2 6.419.043.363.221.94M 18.255.521.07.5 6.511.663.814.15M 13.045.812.9 6.57.015.046.37.16P 50.576.2 3.3 4.8 2.5 3.97.6 1.07M 20.444.817.8 5.9 2.313.822.5 2.68M 11.440.919.77.413.926.551.110.69P 69.473.610.0 5.6 4.6 2.418.3 3.710M 18.148.79.8 5.710.137.658.614.611M 11.748.316.57.6 1.526.039.38.712M 10.241.9 5.3 5.4 2.621.532.1 5.514M 16.550.911.1 5.6 3.613.734.5 6.315M 17.345.015.0 6.7 2.7 6.228.4 4.616M 16.747.312.8 6.0 4.210.734.9 6.517P 38.961.425.77.4 2.7 2.618.6 2.918M 24.051.114.3 6.158.961.787.636.019M 17.249.422.37.423.634.556.113.120M 17.444.019.0 6.124.853.379.719.721M 17.242.018.1 6.529.265.176.422.923M 15.138.419.6 6.4 1.0 4.613.6 1.624P 47.162.217.2 5.5 2.7 3.924.1 4.525M 15.945.613.3 5.7 2.514.929.2 5.126P 38.072.012.4 6.0 2.910.930.1 5.827M 23.059.2 6.8 5.2 5.532.243.48.028M 16.146.98.8 5.514.063.466.622.129M 12.344.8 6.8 5.10.5 5.5 5.40.830M 10.437.9 6.0 5.2 2.922.133.1 5.031M 15.352.6 4.7 5.29.642.641.210.333P 54.374.49.9 5.3 4.18.117.1 3.134M 23.256.223.87.1 1.1 5.923.4 3.036M 27.059.016.4 6.3 6.717.942.08.637M18.952.317.86.833.050.475.422.3Concentrations expressed on an oven dry,volume basis (except MC,expressed on a wet-weight basis).123D.Styles,C.Coxon /Geoderma 136(2006)120–135water interaction,which occurs predominantly within a few centimeters of the surface(Sharpley,1985)and adjacent to preferential flow routes.Daly et al.(2000) found that soil OM content corresponded well with major Irish soil types,and in this study their definition of peat soils(OM>30%)is used.This boundary effective-ly divided the soils into predominantly mineral soils (≤27%OM)and peat soils(≥38%OM)(Table1). However,the mineral soil group does contain soils with relatively high OM contents and that,while predomi-nantly composed of mineral material,may be defined as peaty complexes.2.3.General soil analysesSoil pH was determined by stirring10g of dried soil with25ml double distilled water(DDW)and standing for30min prior to pH measurement with a calomel-glass electrode.Carbonate content was determined titrimetrically using NaOH following equilibration with HCl(Allen,1989).TOC and TN were analysed by dry combustion on a LECO CNS-1000elemental analyser after pre-treatment with sulphurous acid to burn off any carbonates.OM was determined by loss on ignition,after combustion at550°C for4h. Exchangeable bases and Mn were determined by extracting soils with ammonium acetate at pH7.0 (Allen,1989),while ammonium oxalate-extractable aluminium(Al ox),iron(Fe ox)and phosphorus(P ox) were quantified following Uusitalo and Tuhkanen (2000).Extraction solutions were filtered through Whatman No.1filter papers prior to measurement of elemental concentrations using a Perkin-Elmer®atom-ic absorption spectrophotometer.2.4.Soil P testsOlsen P was measured according to the method of Olsen et al.(1954).Soils were extracted at a soil/ solution ratio of1:20with0.5M NaHCO3at pH8.5, filtered through Whatman No.1filter papers,and the extracts acidified to pH5with H2SO4.Morgan P was determined by extracting6ml soil with30ml sodium acetate at pH4.8for30min,and measuring MRP in Whatman No.2filtrates(Morgan,1941).Remaining P sorption capacity of the soils(P sorption index:PSI)was determined according to the method of Freese et al. (1992).MRP remaining in solution was measured after shaking0.5g dried soil for24h with50ml0.02M KCl solution containing15mg P l−1as KH2PO4.Phosphorus sorption capacity(PSC)was calculated in two ways.Firstly,P ox was added to the PSI for each soil to give PSC1(mg l−1soil).Secondly,the sum of molar Al ox and Fe ox concentrations were multiplied by an activity factor of0.5to give PSC2(mmol l−1)(Freese et al.,1992).Degree of phosphorus sorption saturation (DPSS)was calculated using the PSC values.P ox, expressed on a volume and molar basis,was divided by volume and molar PSC1and PSC2values,respectively, to give DPSS1and DPSS2(Freese et al.,1992).2.5.P solubility analysesMoist and dried samples of all soils were analysed for P w by shaking1g of dry-weight equivalent soil with 100ml of DDW for30min,filtering through0.45µm filter papers,and analysing solution for MRP before and after acid-persulphate digestion(Grasshoff et al.,1999) to give soil contents of P w DRP and P w TDP,respectively. P w MUP was determined as the difference between these two values.P feo was determined according to the method of Menon et al.(1989)by shaking1g of dry-weight equivalent soil with40ml of solution(0.02M KCl was used instead of0.01M CaCl2),containing an iron-oxide impregnated Whatman No.54filter-paper strip2cm wide by10cm long,for16h.The paper strip was then removed,rinsed,and shaken in0.1M H2SO4for2h,and the MRP in this solution was analysed.All solution MRP concentrations were measured using the modified ascorbic acid method of John(1970).On moist samples, solubility analyses were carried out within2days of sieving in order to minimise storage changes.2.6.Statistical analysesData were statistically analysed using Data Desk version 6.0,checked for normal distribution,and parametric or non-parametric statistics used accordingly. Pearson-Product Moment correlations and least-squares regressions were used to test for the strength of associations and suspected causal effects,respectively, while paired t-tests were used to compare treatments (moist or dried samples)and different soluble P fractions.Mann–Whitney U-tests were used to compare soil groups(peat or mineral soils).Stepwise multiple regression analyses were performed after Velleman (1993),to derive predictive models for drying effects.3.Results3.1.Water-extractable PThe quantities of P w DRP and P w TDP extracted from moist and air-dried soil samples,and the drying-induced124 D.Styles,C.Coxon/Geoderma136(2006)120–135changes,expressed on a volume basis and as a percentage (for change),are displayed in Table 2.Moist sample P w TDP ranged from 1.15to 33.74mg l −1soil,and air-drying increased the quantities of P w TDP extracted by between 2.01mg l −1(soil 17)and 23.64mg l −1(soil 3),with a mean change for all soils of 9.37mg l −1.Only soil 7displayed a decrease in P w TDP after drying,driven by a decrease in P w DRP content,and this same soil displayed the smallest drying-induced P feo increase of any soil.Fig.1displays mean moist and dried sample soluble P fractions.Drying samples resulted in signif-icantly (paired t -test p <0.0001)greater release of P w MUP than P w DRP (all-soil mean increases of 6.33and 3.04mg l −1for P w MUP and P w DRP,respectively).Expressed as a proportion of moist sample contents,the mean P w MUP increase was over three times greater than the mean DRP increase (469%compared with 144%).Consequently,the mean proportion of P w TDP accounted for by P w MUP rose from approximately one quarter in moist samples to almost half in dried samples (Fig.1).Expressed on a soil-weight basis for the purpose of comparison with other studies,mineral soil P w MUP increases ranged from 0.96to 27.07mg P kg −1,and averaged 10.13mg P kg −1.Moist and dried sample P w DRP and TDP contents were strongly linearly correlated (e.g.moist and dried sample P w DRP r =0.96,p <0.0001),but moist and dried sample P w MUP contents were not significantly correlated (Table 3).This is displayed graphically in Fig.2a,which also highlights the linearity of the relationship between moist and dried sample P w DRP.Relative increases for each soluble P fraction afterTable 2Mean and S.D.of 2replicates for moist sample P w DRP,and P feo ,with change after drying (expressed as percentage change in brackets)Soil no.P w DRP P w TDP P feo MoistChangeMoist ChangeMoistChange(%)(mg l −1soil)(%)(%)10.98±0.10 3.10(316) 1.93±0.079.26(480) 4.05±0.0513.76(340)27.51±0.15 4.01(53)10.34±0.0017.30(167)21.53±2.0413.65(63)311.09±1.589.20(83)13.65±1.8223.64(173)31.89±3.2125.97(81)4 2.27±0.37 3.41(150) 4.04±0.5420.38(505)9.13±0.4221.96(241)5 6.74±0.66 2.23(33)9.36±0.8712.81(137)15.34±1.0715.35(100)60.69±0.01 4.74(689) 1.15±0.1011.61(1010)0.29±0.089.85(3378)710.06±0.54−5.81(−58)14.51±0.37−5.24(−36)20.46±2.00 1.28(6)8 6.57±0.86 1.69(26)10.09±0.90 6.39(63)18.22±0.9711.35(62)9 2.01±0.30 3.77(188) 2.77±0.268.46(306) 6.66±1.117.60(114)1010.23±0.97 6.00(59)12.83±1.0112.01(94)32.48±0.8126.10(80)110.91±0.19 1.92(211) 2.13±0.087.15(336) 3.58±0.568.96(250)12 2.89±0.19 2.95(102) 5.30±0.32 6.97(131)8.95±0.1910.33(116)14 1.39±0.00 6.41(462) 3.30±0.3511.55(350) 5.66±1.3624.28(429)15 3.37±0.07 1.15(34) 5.31±0.329.11(172)13.30±0.75 2.74(21)16 2.79±0.81 4.04(145) 5.45±1.7910.60(195)17.12±0.9910.57(62)170.43±0.01 1.38(322) 1.47±0.07 2.01(137) 1.58±0.01 6.21(394)1831.19±0.87 2.39(8)33.74±0.808.55(25)72.51±0.39−32.85(−45)197.29±0.10 1.35(18)9.52±0.327.59(80)21.03±0.7112.39(59)2015.19±1.63 2.73(18)18.24±1.65 6.69(37)50.22±1.6416.62(33)2114.77±0.56 2.95(20)17.13±0.468.63(50)47.46±1.8215.91(34)230.89±0.19 1.34(151) 2.85±0.25 4.87(171) 1.44±0.00 6.21(433)24 1.29±0.26 4.12(319) 2.37±0.417.30(308) 3.10±0.6013.03(420)25 3.18±0.12 2.19(69) 5.72±0.097.15(125)10.93±0.278.78(80)26 1.44±0.48 1.22(85) 2.60±0.47 4.52(174) 3.78±0.24 4.21(111)278.30±0.64 5.54(67)11.10±0.4113.57(122)16.85±0.1322.96(136)2813.27±1.58 2.49(19)16.41±1.90 6.87(42)26.37±2.8023.26(88)29 1.01±0.180.81(81) 2.13±0.19 3.33(156) 1.19±0.19 2.52(211)30 4.28±0.38 1.64(38) 6.70±0.18 4.40(66)11.29±0.48 6.76(60)317.90±0.107.73(98)10.23±0.5812.03(118)16.49±0.4721.19(129)330.80±0.22 4.62(579) 1.32±0.049.08(688) 2.24±0.908.74(390)340.72±0.20 2.23(311) 4.44±0.857.86(177)0.79±0.017.46(948)36 5.38±0.64 1.51(28) 5.78±1.278.25(143)12.16±0.3511.76(97)3712.32±0.075.19(42)14.84±0.0914.29(96)36.93±1.0816.92(46)Concentrations expressed on an oven dry,volume basis.125D.Styles,C.Coxon /Geoderma 136(2006)120–135drying were significantly negatively correlated with moist sample contents,while absolute increases were significantly positively correlated with dried sample contents,most strongly for P w MUP (r =0.98,p <0.0001).Fig.3displays the non-linear inverse relationships between moist sample soluble P contents and drying-induced percentage increases,best described by power equations.The P w MUP fraction was unique in displaying a positive correlation between dried sample content and drying-induced percentage increase (r =0.47,p ≤0.0001),and between absolute and relative increases (r =0.60,p <0.01)(Table 3).3.2.Desorbable PA greater quantity of P was desorbed to the iron-oxide strip over 16h than was released to water over 30min.Moist sample mean P feo and P w TDP were 16.52and 9.37mg l −1,respectively.Moist sample P feo ranged from 0.79mg l −1for soil 34to 72.51mg l −1for soil 18(Table 2).When respective decreases (P w TDP,soil 7,and P feo ,soil 18)were omitted,the all-soil mean drying-induced P feo increase of 12.77mg l −1was significantly greater than the all-soil mean P w TDP increase of 9.82mg l −1(paired t -test p =0.002).From here on,these two decreases are omitted from statistical descrip-tions and figures.Mean drying-induced relative increases were proportionately similar (272%compared with 206%;paired t -test p =0.45).As with the P w fractions,relative drying-induced increases in P feo were inversely correlated with moist sample desorption (Fig.3c),while actual increases were positively correlated with dried sample desorption (Table 3).P feo contents inmoist and dried samples were strongly linearly related (Fig.2b).3.3.Relationships with soil characteristicsThere were no particularly strong correlations between soluble P increases and soil characteristics (Table 3).Table 3shows that the most consistent correlations between soluble P increases and general soil characteristics involved MC and OM content (and related variables,such as TOC and TN),although MC and OM were strongly co-correlated (r =0.90,p ≤0.0001).Relative increases were positively correlated with MC and OM content for all soluble P fractions except P w MUP (close to statistical significance).Moist sample P w MUP,P w TDP and P feo were negatively correlated with MC.V olume-expressed increases were not significantly correlated with either MC or OM content when all soils were considered,but when peat soils were omitted from statistical analyses,most volume-expressed increases were weakly but significantly correlated with MC and OM (Table 4).Other significant correlations of note in Table 3are relatively strong positive correlations for DPSS (1and 2)with P feo increases,and negative correlations for DPSS (1and 2)with relative increases in P w DRP and (for DPSS1only)P w TDP and P feo .These negative correlations were stronger for the empirical DPSS1(based on P ox and PSI),and may reflect positive associations between DPSS and moist sample soluble P content.P feo increases were significantly positively correlated with STP (Olsen P,Morgan P and P ox )and extractable K,and significantly negatively correlated with PSI,while P w DRP increases weresignificantlyFig.1.Water-extractable DRP and MUP fractions before and after drying,averaged for all soils,for mineral soils and for peat soils.Error bars indicate S.D.126 D.Styles,C.Coxon /Geoderma 136(2006)120–135。