DIVISION S-4—SOIL FERTILITY AND PLANT NUTRITION

Measurements of Phosphorus Availability In Acid Soils of Pennsylvania1

DALE E. BAKER AND JON K. HALL2

ABSTRACTThe objective of this investigation was to compare extracting

procedures which might be superior to the Morgan method forpredicting the phosphorus status of soils in Pennsylvania.

The first experiment involved the Ap horizons of 161 soil types.Of these, 44 soils were selected for additional experiments.Some of the soils tested either very high or very low in P by allmethods, some tested high by one or more methods but not byother methods, and some fixed either high or low percentages ofthe added P against extraction by one or more of the methods.

With respect to the amounts of P removed from soils, the ex-tractants ranked in the order Purdue > Bray no. 1 > Morgan >Experimental > CaCI2. The respective extractants removedabout 52, 38, 8, 4, and 1% of the P which was added and allowedto react with the soils. In general, the test values obtained by onemethod were not very closely related to values obtained by othermethods for these soils.

The Bray no. 1 method was most accurate for predicting plantuptake of P while the Purdue method proposed by Al-Abbas andBarber was least accurate.

Additional Key Words for Indexing:isotopic dilution.

acid soils, organic P,

SEVERAL EXTRACTING procedures have been used to testsoils for available phosphorus by investigators working in

areas where the soils are generally acid (3, 6, 7, 12, 13). In1964 Al-Abbas and Barber (1) proposed a new method basedon results obtained from the use of the fractionation procedureof Chang and Jackson (4). The multiple regression analysesthey obtained indicated that the plant-available P was morehighly correlated with the iron phosphate fraction of the

soil P than with any other fraction. However, the water-soluble P also correlated with plant-available P.

Baker (2) reported that for widely different acid soils, plantgrowth and uptake of P were closely related to both theconcentration of P in the soil solution and the amount oflabile P (P of the soil solution and solid phase which equi-librated with 32P). From this work it was concluded thatplants respond to solid phase P by removing it in a mannerwhich cannot be predicted from its solubility alone.

Dunbar and Baker (5) found that for widely different acidsoijs, aluminum and iron! phosphates (extracted with NH4Fand NaOH, respectively) accounted for most of the nativeand added P. Iron phosphate predominated in the untreatedscils while aluminum phosphate accounted for most of thefixed P in soils treated with fertilizer P. By an isotopicdilution technique, the P fractions were ranked in the followingorder with respect to amounts of labile P removed per unitof solid phase P: water soluble > aluminum phosphate > ironphosphate > calcium phosphate. They hypothesized that inacid soils recently fertilized with P, aluminum phosphate(P soluble in NH4F) liberates more P to growing plants thanthe other fractions while iron phosphate (P soluble in NaOHafter the NH4F extraction) liberates more P to plants growingon soils which have not been fertilized recently with P.

Phosphorus fertilization and liming of soils have long beenconsidered the two most perplexing fertility problems inPennsylvania (11, 14). Hunter (8) reported recently that Pmay be substituted for lime on Hagerstown soil. Results ofprevious investigations (2, 5) indicated that fertilizer P reactsrapidly with Al in soils. Thus, either P or limestone should beeffective in reducing the concentration of soluble Al.

Growers in Pennsylvania have complained that the Morgantest presently used in Pennsylvania does not reflect theavailability of residual fixed P. The experience of growers andthe results reported by Baker (2) suggested that a soilextractant which removes a constant percent of the surface-

BAKER AND HALL: PHOSPHORUS AVAILABILITY IN ACID SOILS 663

bound P might be more valuable for use on soils ofPennsylvania.

The objective of these experiments was to compare extrac-ting procedures which might be superior to the Morganmethod for predicting the phosphorus status of soils inPennsylvania.

EXPERIMENTAL METHODSFor the initial experiment, samples were taken from the Ap

horizons of 161 soil types. The soils were from Pennsylvania,except the Marshall and Loring from Missouri, the Cecil fromGeorgia, and a clay latosol from Jamaica, W. I. Each of the soilswas tested in triplicate by the Bray no. 1 method. Assumingthat the Bray no. 1 extractant (0.025JV HC1, O.OSN NH4F) wouldremove about 20% of the added P from each soil (2), every soil(20 g) was treated with P as KH2PO4 to attain test levels of 100and 200 ppm of P, respectively, by this method. The averageamounts of P added to the soils was 402 and 902 ppm, respectively.The standard deviation for both rates of P was 88 ppm.

The soils were wetted and dried alternately six times at 40Cto hasten equilibration. Following the equilibration with 31P,each sample was treated with 50 ml of a solution to supply1.5 AIC of ffiP containing 10 ppm of 31P as a carrier. After standingin this solution for three days, the samples were dried at 40C.

The soil testing methods compared in this investigation arepresented in Table 1 (1, 3, 6, 7, 9, 10). The method of Al-Abbasand Barber (1) will be referred to in this report as the "Purdue"method. Since results of other experiments (2, 5) indicated thatboth water soluble and fluoride extractable P were related to theavailability of P, the "experimental" method of Table 1 was de-veloped and included to determine if an extractant intermediatebetween CaClj and Bray no. 1 extractant would be superior toeither alone for use in estimating the availability of P. Pre-liminary work indicated that the experimental method removed asmall but more constant percent of the added P than did C&Ck.

From the 161 soils used in the initial experiment, 44 were se-lected for a greenhouse study. Each soil was selected on the basisof one or more characteristics including: (i) the availability of Pby all tests was either very low or very high, (ii) the availabilityof P was low by one or more methods, but not by another method,(iii) the soil fixed a high or low percentage of the added P againstextraction by one or more of the extracting methods. Thus,problem soils from a soil testing point of view were selected forthe greenhouse study. The Ap horizons from each of the 44soils were sampled early in the spring of 1965. Two replicationsof each soil (15kg of < 2-mm fraction) were placed in 3.8-liter(4-gallon) pots and treated with 50 ppm N as NH4NO3 and 50ppm K as KsSO4. No variable treatments were included, and nolime or P was added. Four different corn hybrids were plantedin each pot on May 13, 1965. The soil moisture was maintainedat about 25% by frequent additions of distilled water plus bring-ing each pot to constant weight once each week. The plants wereharvested individually on June 10, four weeks after planting.The dry weights of each plant, the percent P in plants, and themilligrams of P in each plant were determined. In the seedling

Table 1—Extracting procedures used to studyrP in different soils

Method orfraction

Extracting solutionSoil

solutionratio

Shaking time

CaCb O.OlAfCaCli 1:7 15 min (Stand 8 hr)Bray no. 1 0.025AT HC1, 0.03JV NH<F 1:7 1 minExperimental 0.002AT HC1, 0.002JV NH.F 1:10 10 minPurdue 0.075AT NasCjO^ 0.255JV

NaOH 1:20 5 minMorgan 0.735JV Na CsH«O> , 0.5N

HCiHiO, @ pH 4.8 1:5 30 minInorganic P Cone. HC1 Followed by 0. S N NaOH (combined solutions)Total P Combined solutions plus HC1O« digestionOrganic P Total P — Inorganic P

stage, the varieties were not different3 with respect to growth andP uptake; thus, averages for the four varieties were used for com-paring soils. Each soil was tested for availability of P by each ofthe methods described in Table 1.

RESULTS AND DISCUSSIONDetailed results of the initial experiment involving 161 soiJs

will not be presented in this report. The amounts of total P(10) in the different soils ranged from 213 to 1928 ppm,Table 2. With respect to the amount of available P removed,the extractants ranked in the order Purdue > Bray No. 1 >Morgan > Experimental > CaCl2. The P removed by eachof these extractants was considered to represent a portion ofthe available P because each one removes a portion of thesurface bound P which is 32P labile (5). The authors believethat a distinction should be made between available, fixed Pand the availability of fixed P (2).

Most of the total P was in the inorganic form, but sub-stantial amounts of 'organic' P were removed from some soils.Calculations made for the amounts of organic and total Premoved from these soils indicated that an average of about27 % of the total P was 'organic' P. Since an average of 35.6 %of the added 32P was recovered with the 'organic' P, it issuggested that the so called 'organic' P (10) consisted largelyof inorganic P which was associated with the organic matterremoved by NaOH extraction. The average specific activities(% added 32P removed per ppm of 31P) were 0.246, 0.198, and0.208 for the organic, inorganic, and total P, respectively.From the results of Dunbar and Baker (5) it may be concludedthat the P removed by NaOH extraction as 'organic' P washeld predominantly as aluminum phosphate, which equi-librated more rapidly with 32P, giving it a higher specificactivity than other solid phase phosphates. The specificactivities of P removed by the other extractants were 0.54,0.70, 1.85 and 8.50 for the Purdue, Bray no. 1, Experimentaland CaClj extractants, respectively. No 32P equilibration wascarried out for the Morgan extractant.

By subtracting the values obtained for each untreated soilfrom those obtained at each rate of added P, the percent ofthe added P removed by each extractant was calculated,Table 3. Only the Purdue method and the Bray no. 1 methodremoved appreciable amounts of the added P. The 38%removed by the Bray no. 1 method was much higher thanwas expected; consequently, the soil test values were higherthan would normally be encountered. The 20% removalobtained in previous work (2), included only that removedfrom soil after a crop had been grown. The crop removed upto 18% of the added P, so the higher removal should havebeen expected. In addition to the Purdue and Bray no. 1methods removing a larger percent of the added P, the lowercoefficients of variation indicated that these percentages weremore nearly constant.

The low removal of added P by the Morgan and other weakextractants limits their usefulness in measuring the residualeffect of rotation fertilization. The removal of relatively

»Baker, D. E., R. R. Bradford and W. I. Thomas. Sept. 1966.Accumulation of Ca, Sr, Mg, P and Zn by genotypes of corn underdifferent soil fertility levels. Paper Sm-77/22 presented atSymposium on the use of isotopes in plant nutrition and physi-ology. FAO/IAEA, Vienna, Austria.

664 SOIL SCI. SOC. AMEH. PKOC., VOL. 31, 1967

Table 2—Amounts of P and percentages of the added 32P removed by different extractants from untreated Ap horizons of 161 soils

Extractant

CaCbBray no. 1PurdueExperimentalMorganLabile PInorganic POrganic PTotal P

Range

0.025- 2.11.2 - 163.2

24.9 - 283.00.10 - 7.210.11 - 18.91.18 - 131

95 -15180 - 723

213 -1928

Mean

0.1325.882.6

1.064.41

23.9428157584

S.D.

0.1921.043.0

1.083.2]

20.4221126268

Range

0.05- 3.262.4 - 29.3

20.0 - 67.00.16- 0.91

——

41 - 800 - 61

67 -120

Mean

- % "P —————

0.99713.838.4

1.60——

67.535.6

103

S.D.

0.6575.9

82.61.28——

7.113.713.0

Table 3—Percentages of the added P which were removed bydifferent extractants from 161 soils receiving 2 variable

rates of P

Rate of P added

Method 402 ±88 902

, ppm

±88 Average %P Removed

CaChBray no. 1 (3, 6)ExperimentalPurdue (1)Morgan (7)Average

%

1.038.73.6

51.57.7

21.8

CV %

6631502453

%

2.138.33.7

54.97.6

22.6

CV %

8126352459

1.538.553.23.07.7

* Each percentage value represents an average for 161 soils. Honestly signifi-cant difference (HSD 0.05) for comparing:

Average % P removed by extractants = 1.9.Average % P removed at different rates of P =0.2.Rate of P X extractant interaction = N.S.

greater percentages of the added P by CaCl2 with an increasein the rate of P applied, was not observed with the otherextractants. The experimental extractant was included in aneffort to find an extractant intermediate between the 0.01MCaCl2 and Bray' no. 1 methods which would remove watersoluble and readily replaceable solid phase P. It was assumedthat neither the H+ or F~ ions in the experimental extractantwould be specific for P, but would react with other constituentsaccording to their prevalence in the soil. Thus, when all ofthe reagent reacts with the soil, the P should be released inproportion to its abundance in the soil. The presence of freeCaCOs, or other compounds more soluble in the extractant

-tfeftH^fised^^^rould-cause-the-extractant to-remove less P-thanit should. This is a weakness of most P extractants. Unlessan excess of the extractant is used, compounds which do notcontain P and which are not present in equal amounts indifferent soils, will react with and "use up" the extractant.

Results obtained by each extractant are related to thoseobtained by the Bray no. 1 method in Table 4. The correla-tion coefficients reported in Table 4, indicated that valuesobtained by the Bray no. 1 method and the experimental

method were highly associated (r2 = 86%). The valuesobtained by the different methods are related closely enoughto those obtained by the Bray no. 1 method to be of some valuein extrapolating calibration data obtained by one method toestablish standards for use with other methods.

A total of 44 soils was selected from the 161 for use in agreenhouse experiment. The soils were predominantly acidwith only three having a pH above 7. The precision of eachtest over replications is indicated by the magnitude of thecoefficient of variation, Table 5. The Bray no. 1 method wasmost precise while the precision for the Purdue method wasvery poor. The Purdue method'gave variable results becauseof the colored extracts resulting from the extraction with analkaline solution. The charcoal treatment recommended byAl-Abbas and Barber (1) was satisfactory for some soils, but,in addition, many required an acid treatment to remove thedispersed humus. With CaCl? the concentration of P was solow in m&ay soils that a spectrophotometer with a 2.5 cm cellwas required to measure the concentration of P precisely.A Beckman D.U. Spectrophotometer with a test tube adapterwas used for this purpose.

The percent P in the plants ranged from 0.09 to 0.41 whichwas more than a 400% difference, while plant growth variedby more than 2500%. A matrix of correlation coefficientsrelating the various measures of P availability is presented inTable 6. The high correlation coefficient (0.82) relating plantweight to percent P in the plants indicated that dry matterand concentrations of P in the plants were highly correlated.Results for the "experimental" method (r = 0.70) were mosthighly correlated with plant weight, while those for the Brayno. 1 method (r = 0.74) were most highly correlated with the.concentration of P in the plants. It was concluded that thesoil test methods ranked as follows with respect to their valuefor predicting the availability of P in these soils: Bray no. 1 S:"Experimental" £ CaCl2 > Morgan > Purdue. The valuesfor Bray no. 1 and the "experimental" methods were highly.correlated (r = 0.86) and the values for the CaClo equilibrationand the experimental method were highly correlated (r =

Table 4—Linear regression coefficients, polynomial coefficients and coefficients of determination (r2 & R2) calculated by methods of leastsquares to relate test values by different soil test methods to those obtained by the Bray no. 1 method

Linear regressionY = A + bx

CaCPPurdueExperimentalMorgan

a

- 3.377

- 1.61.1

b

0.0581.2570.0970.191

r2 %

51.875.786.556.2

Polynomial by least squaresY =a+bx+cx^

a

0.8329.4

- 1.260.017

b

—0.0112.0500.0900.209

c

0,00015-0.00175

0.0000130.800041

* %

59.178.985.957.0

BAKER AND HALL: PHOSPHORUS AVAILABILITY IN ACID SOILS 665

Table 5—A statistical summary of different soil tests and plantanalyses for evaluating the availability of soil P

Determination

CaCU, M m/literCaCla, ppmBray's no. 1ExperimentalPurdueMorganPlant P, %Plant wt-. gPlant P, mgSoil pHSoil L.R.'

Range for44 soils

0.24- 10.0O.OS- 2.21.7 -1120.32- 22

20 -5320.38- 620.09- 0.410.35- 8.80.40- 384.3 - 7.00.0 - 9.8

Mean

1.620.3525.83.4

1067.50.192.145.465.574.19

CVf(%)

15.615.65.27.3

32.211.312.415.023.11.5

15.0

* L.R. = Lime requirement in meq/100 g by method of Woodruff,t CV = Coefficient of variation calculated from A.O.V. error mean square.

Table 6—Correlation coefficients* (r) relating soil test valuesplant growth and P uptake by corn on 44 soils

Coefficient for Corn on 44 Soils t

n21S)4)5)6)7)S)

CaCkBray's no. 1ExperimentalPurdueMorganPlant P. %Dry matter gPlant P, mg

1

0.790.920.500.750.610.640.68

2

0.860.790.740.740.630.66

3

0.570.710.640.700.73

4

0.500.560.410.43

5 6

0.560.55 0.820.55 0.84

7

0.97

* A correlation coefficient of 0.27 is required for significance at the 1% levelof probability.

t The 1, 2, 3t 4i etc. of the horizontal portion of the matrix correspond to thesame numbers as defined in the vertical portion.

0.92). Results for the Morgan test and the Purdue test werenot as highly correlated with the results for the other tests.A correlation coefficient of 0.50 relating test values for theMorgan method to those of the Purdue method indicated thatonly 25% of the variation in test results by one method couldbe accounted for by those of the other method.

Scatter diagrams relating average percent P in plants tosoil test values are presented in Fig. 1 to 5. The line througheach diagram represented the best fit for a quadratic equationas calculated by the method of least squares. For CaCl2 thepercent P in plants increased until 0.4 to 0.6 ppm (2.6 to3.9 /urn/liter) of soil P was recovered, and then did not changefor soils containing higher test levels, Fig. 1. The calculatedequation fit to these results indicated that only 48% of thevariation in percent P in the plants was explained by the soiltest results. This is not a good correlation, and from apractical point of view, one would prefer that the scatteraround the line be much less, especially at the lower levels ofplant and soil test P.

Results for the "experimental" method were associatedwith plant uptake of P in a manner similar to that observedfor CaCl2 and indicated that plants would not respond to Padded to soils testing more than about 6 ppm.

There was less scatter about the line relating percent plantP to soil test values obtained by the Bray no. 1 method. Soilstesting 40 ppm or more produced plants containing the higherconcentration of P. The R* value of about 61 % indicated thatthe plant and soil test results for P were associated.

The scatter diagram relating plant P to soil test P by the

0.20 0.40 0.60 ~OBO 1.00 1.20 1.40P.P.M. Soil P(0.0!MCoCI2)

1.80 2.00 2.20 2.40

Fig. 1—Scatter diagram and polynomial equation relating plantP to amounts of P removed from 44 soils by 0.01M CaCl2.

Y = 0.106 •0.0045X-0.000022X2

SY.x = 0.0556R2 = 60.8 %

u ILO ' 210 31.0 41.0 51.0 61.0 71.0 81.0 91.0 101.0 N 1.0 121.0PPM Soil P (0.025NHC!,0.03NNH,,F)

Fig. 2—Scatter diagram and polynomial equation relating plant Pto amounts of P removed from 44 soils by the Bray no. 1method.

Y =O.I34»0.0243X-0.00068X;

SY .X =0.0645R2 = 47.4%

24.0

Fig. 3—Scatter diagram and polynomial equation relating plant Pto amounts of P removed from 44 soils by the "experimental"method.

Purdue method illustrates the poor relationship for these soilsbetween availability of P and fixed available P. The soilstesting highest by the Purdue method were not highest inavailability of P. Although highly significant, the degree ofassociation between plant P and amounts of P extracted bythe Purdue method was not good relative to that obtainedwith the other methods (excluding Morgan's), and this isconfirmed by the scatter diagram, Fig. 4.

666 SOIL SCI. SOC. AMEH. PKOC., VOL. 31, 1967

0.56

0.48

0.40t/5

"c

51 0.32c

Q. 0.24

"HS 0.16

.0.08

Table 7—Yields of grain for 6 corn hybrids grown on Huntingtonsilt loam soil at 4 rates of P

Y =0.077 + O.OOI7X-0.0000026X2

SY.x = 0.0548R2 = 62.3 %

100 180P.P.M. Soil P

260 340 420(075NNa,C

500 580 6602 w 2 0 4 , 0 .255NNaOH)

Fig. 4—Scatter diagram and polynomial equation relating plant Pto amounts of P removed from 44 soils by the Purdue method.

(The horizontal axis should read: (.075AT Na2C204, 0.255.ZV NaOH)

0.56

0.48

0.40in

10.32CT

1-0.24CO)

S 0.16<D

Q_

0.08

O

Y = O.I46*0.0094X-O.OOOI2X2

Sy.x = 0.0671R2 = 42.6 %

8.0 16.0 24.0 32.0 40.0 48.0 560 64.0 72.0 80.0P.P.M. Soil P (0.735 N Na Ac, 0.5 N HAc atpH48)

Fig. 5—Scatter diagram and polynomial equation relating plant Pto amounts of P removed from 44 soils by the Morgan method.

The diagram relating plant P to soil test values by theMorgan method showed that the uptake of P by the plantswas associated with soil test levels between zero and five ppm.Thus, it was not possible to separate soils medium to highwith respect to availability of P from those which were low.Some soils (usually those near pH 7) tested high in P by theMorgan method when the availability of P was medium to low.

A field experiment was conducted on Huntington silt loam,a river bottom soil near Lock Haven, Pennsylvania during thesummer of 1965. The objective of this experiment was toobtain additional soil test calibration data. Phosphorustreatments of O, 33.3, 100 and 300 ppm of P as triple super-phosphate were applied. Hybrid 5, Table 7, a prolific hybrid,which gave a dry matter yield response to each increment ofphosphorus, produced an average of 56%, 73%, 87%, and93% of its maximum yield. The yield from the highestyielding plot for each hybrid was considered the maximumyield for that hybrid hi the experiment. The respectiveaverage test levels obtained from 3 replications at the time thecorn was harvested were 3.2,14, 36, and 80 ppm by the Brayno. 1 method. The average percent of maximum yield for all

jppm 1

033.3

100300Meankg/ha

3,5706,8974,8935,6465,018

X 0.01594

2

3,8275,4584,7055,9595,018

= mi/acre

3

4,2664,3284,2036,0224,705

Hybrid*

4

- kg/ha

2,1333,3254,2663,3873,262

5

4,9566,2737,1518,2186,649

6

5,1447,6537,5907,9677,088

Mean

3,9525,5205,4586,210

* Pedigrees of hydrids were as follows: 1. WH X Pa W703; 2. Pa 36 X A509;3. MS1334XR53;4. CMD5XCO 106; 5. C 103XP4533; 6. (Wf9XOH 51A)(Pa 54 X W 22).

hybrids obtained at the respective test levels were 55%, 73%,76 % and 86 %. The grain yield response was similar, Table 7,except for hybrid 6 which produced more grain per unit drymatter than did hybrid 5. The apparent lack of response tothe 100 ppm P treatment resulted from relatively poor growthof all hybrids grown on this treatment in one of the threereplicates.

From the results of this investigation and from the relation-ship between the Bray no. 1 test values and those for theMorgan method which is presently used in Pennsylvania thefollowing calibration values for the Bray no. 1 method seemjustified for use as guides on Pennsylvania soils.

Test values, ppm

8050-8030-5015-3015

Availability category

Very highHighMediumLowVery low

The test levels required for maximum rates of growth ofcorn in Pennsylvania are higher than those suggested byothers (3, 6); however, the higher fixation capacities of soils inPennsylvania cause the relationship between availability of Pand soil test levels to be different than for soils in themidwest (2).

CONCLUSIONSFrom the results presented it was concluded:1) The availability of soil P was not related closely with

the amounts of fixed P which would become available withtime. Thus, no single soil test method can be used to predictboth the availability of the soil P and the amounts whichshould be added to soils which are low with respect to Pavailability.

2) The Morgan method is not satisfactory for these soilsbecause it does not remove substantial amounts of the fixed P.It is not reliable for predicting the availability of P becausethe results are affected significantly by soil pH.

3) The "experimental" method or the CaCl2 equilibrationprovided the best measure of the availability of P while thePurdue method provided the best measure of the amount offixed P which would be available over a long period of time.

KUBOTA ET AL.: MOLYBDENUM TOXICITY IN GRAZING ANIMALS 667

If all soils to be tested are below pH 7, the "experimental"method is applicable. However, if the pH is greater than 7,the excess base would "use up" the extractant giving lowresults.

4) The Bray no. 1 method is recommended for testingPennsylvania soils where results for a single extraction are tobe used for predicting both the availability of P and theamount of P which should be added. This extractant removesa substantial amount of the fixed P and the results are moreclosely correlated with the uptake of P by corn plants. ForPennsylvania soils testing less than 50 ppm by the Bray no. 1method, the availability of P is likely to be below optimumfor seedling corn plants.