Water Retention by Core and Sieved Soil Samples1

PAUL W. UNGER2

ABSTRACTMechanical analyses and water retention by core and sieved

soil at —Va and —15 bar matric potentials were determinedfor samples from 26 sites ranging in texture from sand to clay.

Objectives were to obtain a basis for identifying which soilsmay be influenced by deep tillage and profile modification withrespect to water storage capacity and to determine the magni-tude of errors possible when using sieved soils to establish fieldsoil water contents.

At — Va bar potential, cores retained more water than sievedsoil when the water content was below 11%. The opposite oc-curred at higher water contents. At —15 bars potential, corescontained about 1 percentage point more water than sievedsoils throughout the water content range encountered. Theseresults show that treatments which thoroughly disrupt thenatural soil structure may decrease and increase the storagecapacity of coarse- and fine-textured soils, respectively.

When expressed as a percent of the core water content, dif-ferences between core and sieved soil contents at —Vb barpotential ranged from —40 to +25% at 5 and 40% core watercontents, respectively. At —15 bars potential, the range wasfrom —52 to —4% at 5 and 25% core water contents, re-spectively. These differences indicate caution should be usedwhen using sieved soils to infer water retention by field soils,regardless of texture.

This study suggests deep tillage and profile modification maydecrease and increase water storage in coarse- and fine-texturedsoils, respectively. However, because of structural development,organic matter content, and possibly clay size and type, resultsfor individual soils may differ from those indicated by the rela-tionships established. To more accurately evaluate the possibleeffects of deep tillage and profile modification on water reten-tion, the soil in question must be analyzed.

Additional index words: soil available water, soil water con-tent.

A SOIL'S WATER storage capacity may greatly influencehow efficiently precipitation and irrigation water are

used for crop production. For high efficiency, especiallywhere crops are dependent on precipitation, a large capacityis required for soils to store as much water as possible forplant use. The amount of water storable in soil is influenced,among other factors, by soil texture, structural develop-ment, profile characteristics, and organic matter content.

The storage capacity of some soils has been increased byprofile modification or deep tillage (Mech et al., 1965:Saveson and Lund, 1958).

In another phase of this study, relationships were estab-lished between the water content, texture, density, and or-ganic matter content of soils ranging in texture from sand toclay (Unger, 1975). Many relationships were significantat the 0.1 % probability level. One of the highest correlationcoefficients was for the relationship between soil core watercontent on a volume basis and soil clay content, which sug-gested that considerable information concerning soil waterwas available if the soil's clay content was known. Calcula-tions using assumed clay contents (assumed values were inthe middle of the range for the textural classes used) re-sulted in average water contents almost identical to thosepublished for the soils. However, calculated water contentsfor individual soils varied greatly from the published watercontents, which led to the conclusion that each soil's waterretention characteristics must be determined to obtain reli-able data for that soil.

A soil's water storage capacity is often approximatedfrom — Vs and — 15 bar matric potentials determined in alaboratory with pressure plate or membrane equipment.Generally, sieved soil is used because it is easier to deter-mine soil water storage capacity with sieved soil than withsoil cores. However, the amount of water retained by sievedsoil may differ markedly from that retained by undisturbedsoil cores or soil in the field (Bruce, 1972; Elrick and Tan-ner, 1955; Jamison and Kroth, 1958; Kroth and Jamison,1960; Mathers et al., 1963; Young, 1962; Young andDixon, 1966).

The objectives of this study were (i) to obtain a basis foridentifying soil textural classes that may be influenced bydeep tillage, profile modification, or other profile manipu-lating operations with respect to water storage capacity, and(ii) to estimate the possible magnitude of error involved inusing laboratory determinations with sieved samples to

1198 SOIL SCI. SOC. AMER. PROC., VOL. 39, 1975

approximate the water storage capacity under field condi-tions for different soils.

MATERIALS AND METHODS

For this study, samples were obtained from 26 different loca-tions or profile horizons in West and South Central Texas toobtain soil materials ranging in texture from sand to clay. Mostsamples were obtained from cultivated cropland, the remainderfrom rangeland. Details of the soils sampled and sampling loca-tions and depths have been presented elsewhere (Unger, 1975).

At each sampling site, six soil cores 5.4 cm in diameter and 3.0cm tall and some bulk soil were obtained. Cores were trimmedon one end before equilibrating them at —Ys bar matric poten-tial with pressure plate equipment. After equilibration, the coreswere trimmed to the cylinder volume, weighed, rewet, and equil-ibrated at —15 bar matric potential with pressure plate equip-ment. The cores were not loaded to prevent swelling. However,soil swelling was not a problem at —l/3 and —15 bar potentialsbecause the cores were trimmed at —1A bar potential. Swellingwas observed at saturation for some soils and has been pre-viously discussed (Unger, 1975).

The air-dried bulk soil was passed through a sieve with 2-mmdiameter openings before determining the water content at —1/3

Table 1—Soil sand, silt, and clay content; texture; and watercontent at — Vs and —15 bar matric potential

on a percent by weight basis

Soil water content

Sampleno. Sand Silt Clay

-1/3 bar potential -15 bars potential

Texture Core Sieved Core Sieved

123456789

1011121314151617181920212223242526

21181816555060665694917797905439244350726961324238 :55

% -4341.3029202217172811803

243136291718162148433627

36415255252823171658

1537

223040283310151820152618

Clay loamSilty clayClayClaySandy clay loamSandy clay loamSandy clay loamSandy loamSandy loamSandSandSandy loamSandSandSandy clay loamClay loamClay-clay loamClay loamSandy clay loamSandy loamSandy loamSandy loamLoamLoamLoamSandy loam

25.522.629.132.417.318.319.513.88.24.99.69.23.5

13.917.325.724.017.423.5

8.912.113.122.117.123.415.2

— %by29.530.736.939.719.621.318.514.46.23.75.5

11.92.7

14.517.330.430.821.222.510.713.915.424.517.825.118.8

weight —

16.816.922.224.211.312.113.29.33.92.54.45.62.17.9

11.719.616.710.420.36.09.09.1

12.38.7

19.211.6

17.118.422.024.210.811.99.47.53.22.33.37.61.57.8

10.613.517.79.9

14.14.57.38.9

11.99.2

12.810.6

and —15 bar matric potentials with pressure plate equipment.Retainer rings for the sieved soil were about 1.5 cm tall. Sievedsoil subsamples were used for making particle-size analyses bythe hydrometer method (Day, 1965).

Relationships among various determined and calculated varia-bles were established by simple regression techniques (Ezekieland Fox, 1959) to ascertain the possibility of estimating corewater contents from those measured on sieved samples.

RESULTS AND DISCUSSION

Soil sand, silt, and clay content; texture; and water con-tents on a weight basis at — Ya and — 15 bar matric poten-tials for core and sieved samples are given in Table 1.

Soil Texture

The sand content of samples ranged from 18 to 97%,while that of silt ranged from 0 to 48% and that of clayfrom 3 to 55%. Textural names were obtained from a tex-tural triangle. One sample was given two names becausethe point represented by the sand, silt, and clay content wason a line between two groups.

Relationships between soil sand, silt, or clay content andsoil water content were previously reported (Unger, 1975).The sand, silt, and clay percentages were included in Table1 to provide information concerning the amounts of thesematerials in the soils.

Core-sieved Soil Water Content Relationships

Relationships between core water (— Ys or — 15 bars)3

and sieved soil water (— Ys bar), and between core andsieved soil water (— 15 bars), all on a weight basis, weresignificant at the 0.1% level (Eq. 1, 2, and 3, Table 2).The correlation coefficient for the relationship between coreand sieved soil water (— 15 bars) was slightly lower thanthe one obtained for the relationship between core water(— 15 bars) and sieved soil water (—Ys bar). In addition,the standard error of estimate was greater for core andsieved soil water (— 15 bars) than for core water (— 15bars) and sieved soil water (— Vs bar). Hence, core water(—15 bars) can be inferred from sieved soil water (— Y)bar) with no greater error than using sieved soil water(— 15 bars) to infer core water (—15 bars).

The relationship between sieved soil water contents on

3 Denotes core water contents at —Vi and —15 bars matricpotential. Similar notations are used for sieved soil water con-tent at —!/3 and —15 bar matric potentials.

Table 2—Regression equations, correlation coefficients, and standard errors of estimate associated with the relationships between onedependent and one independent variable in simple regression analyses

Correlation coefficient

Equationno.

1234567

8

F-variable, %

Core water (-1/3 bar)Core water (-15 bars)Core water (-15 bars)Sieved soil water (-15 bars)Core available waterSieved soil available water(Sieved soil - core water

-5- core water X 100); (-1/3 bar)

(Sieved soil - core water)(-1/3 bar)

*-variable, %

Sieved soil water (-1/3 bar)Sieved soil water (-1/3 bar)Sieved soil water (-15 bars)Sieved soil water (-1/3 bar)Sieved soil water (-1/3 bar)Sieved soil water (-1/3 bar)

Core Water (-1/3 bar)

Clay

Regression equation

y = 2.944 + 0.737*y = 0.308 + 0.594*y = 1.196 + 0.992*y = -0.498 + 0.578*y = 2.636 + 0.143*y= 0.437 + 0.431*

y=-16.0'39+ 1.405*

y= -2.032 + 0.180*

Value

0.9730.9470.9350.9780.6640.962

0.542

0.784

Level ofsignificance

0.0010.0010.0010.0010.0010.001

0.010

0.001

Standard errorof estimate

1.7462.0242.2341.2271.6131.219

16.518

1.959

UNGER: WATER RETENTION BY CORE AND SIEVED SOIL SAMPLES 1199

S40H-Z030oa:UJ20

uj 10a:oO r>

Y« 2.944 + 0.737 Xr = 0.973

0 10 20 30 40SIEVED SOIL WATER CONTENT-%

Fig. 1—Core water content plotted as a function of sieved soilwater content, both on a weight basis at — Vs bar matric po-tential. The 1:1 line is also plotted.

a weight basis at — Vs and — 15 bars was significant at the0.1% level (Eq. 4, Table 2). The corresponding r2 valuewas 0.956.

Although relationships between core water (— Vs and— 15 bars) and sieved soil water (— Va bar) resulted incorrelation coefficients greater than 0.900 (Eqs. 1 and 2,Table 2), the coefficient for the relationship between coreavailable water and sieved soil water (—1/3 bar) was only0.664 (Eq. 5, Table 2). While significant at the 0.1%level, the r2 value indicated that only 44% of core availablewater content variation was attributable to factors influenc-ing the sieved soil water content variation at — Vs bar.(Available water is arbitrarily defined as water held in soilat matric potentials between — V s and — 15 bars. Whilesands, loamy sands, and sandy loams may be somewhatdrained at — Vs bar, this value was chosen to have a com-mon reference point for all samples.) For the relationshipbetween sieved soil available water and sieved soil water(— '/3 bar), the correlation coefficient was 0.962 (Eq. 6,Table 2). The corresponding r2 value (0.926) indicatedthat a high per cent of sieved soil available water contentvariation was associated with factors that influenced sievedsoil water content variation at — l/s bar.

The higher correlation coefficient and lower standarderror of estimate for the relationship between sieved soilavailable water content and sieved soil water (— Vs bar)than between core available water and sieved soil water(— Vs bar) showed better agreement between the formerand sieved soil water (— Vs bar) than between the latterand sieved soil water (— Vs bar). Field soil available watercontents (volume basis) can be approximated from coreavailable water or by multiplying sieved soil available water(weight basis) by soil bulk density. For the latter, addi-tional errors are involved in relating sieved soil water tofield soil water because of error involved in determiningbulk densities. However, if bulk density errors are small,overall errors may still be less for obtaining available watercontents for field soil from sieved soil than from soil coreavailable water contents.

Relationships between core and sieved soil water con-tents on a weight basis at — Vs and — 15 bars were signifi-cant at the 0.1% level (Eqs. 1 and 3, Table 2). However,the regression line slope for the relationship at — Vs bar

38

z-25

£20z8 15(r£ 10

UJ w

DC

8 o,

l-l line

Y = 1.196 + 0.992Xr =0.935

0 5 10 15 20 25SIEVED SOIL WATER CONTENT—%

Fig. 2—Core water content plotted as a function of sieved soilwater content, both on a weight basis at —15 bar matric po-tential. The 1:1 line is also plotted.

was markedly different from the slope at — 15 bars. At— Va bar, the slope was 0.73, showing a marked deviationfrom the 1 to 1 line (Fig. 1). Based on the figure, corescontained more water than sieved soil when the water con-tent was below about 11 % and less when above about 11%.

The effects of soil disruption (crushing and sieving) onwater retention were related to soil structure. The low watercontent soils were sandy soils. Structural development,although slight, in the undisturbed sandy soils increased theporosity and, hence, resulted in slightly higher water con-tents at a high (— Vs bar) matric potential as compared withthe sandy sieved soils. Crushing and sieving sandy soils re-sulted in essentially single-grained conditions, which resultedin more complete water removal at the high potential.

Water contents increased with clay contents for the coreand sieved soil samples (Unger, 1975). All soils had gener-ally high bulk densities as previously reported (Unger, 1975).Therefore, poor structure (low porosity and aggregation)apparently was not a major factor in water retention in highclay content undisturbed soils. Crushing these soils increasedtheir porosity and thus resulted in greater water contentsthan in the undisturbed cores at a high matric potential.

At — 15 bars, the regression line slope for the relationshipbetween core and sieved soil water contents was 0.992 (Eq.3, Table 2), indicating that the regression and 1 to 1 lineswere almost parallel (Fig. 2). Fig. 2 illustrates that corescontained slightly over 1 percentage point more water thansieved soil throughout the water content range encountered.

Generally close agreement between water contents at— 15 bars for disturbed and undisturbed soil was reportedby Taylor et al. (1963) and by Young (1962). These authors,along with Mathers et al. (1963), also recognized that watercontents of fine-textured soils were generally higher whensieved materials, rather than undisturbed cores, were usedto determine water retention at — l/3 bar.

Greater water retention by sieved than undisturbed soilat — ]/3 bar and little change in retention at — 15 bars forfine-textured soils suggests that deep tillage and profilemodification treatments may increase the water storagecapacity of fine-textured soils. The determinations weremade with finely divided soil (< 2.0 mm diameter), whichundoubtedly would not be practical to attain under large-scale field conditions. However, additional determinationswith disturbed soil that was passed through different sieveswith openings ranging from 1 to 25mm also showed that

1200 SOIL SCI. SOC. AMER. PROC., VOL. 39, 1975

sieved soil retained more water than cores. Hence, labora-tory determinations with cores and sieved soils can be usedto evaluate the potential of using deep tillage and profilemodification treatments for increasing a soil's water storagecapacity. The persistence of the change in soil conditionsresulting from the treatments is important in the field. How-ever, to evaluate the persistence, repeated determinationswould be necessary.

Error Analysis for Core-sieved Soil Water ContentsSince water retention by cores and sieved soil may be

markedly different, the following analyses were made todetermine the magnitudes of errors involved in determiningwater contents with sieved soil instead of undisturbed soilfor soils ranging in texture from sand to clay. Because ofthe range in texture, a range in water contents also resulted.

Based on the deviation of the regression line from the 1to 1 line in Fig. 1, percentage point differences between coreand sieved soil water contents at — Ys bar were less at lowthan at high water contents. Errors mentioned by Matherset al. (1963) and Taylor et al. (1963) were based on per-centage point differences. However, the differences ex-pressed as a percent of the core water content ranged from— 40.0 to + 25.0 at 5 and 40% core water contents, respec-tively. These results show that relatively large errors arepossible at low and high water content ranges, and thatcaution should be used when inferring field soil water con-tents from laboratory determinations with sieved soil re-gardless of texture, which is a major factor influencingwater retention. Confidence limits for the regression line inFig. 1 were calculated at the 95% probability level. Theselimits included the 1 to 1 line between 8.5 and 14.0% soilwater contents.

The percentage point differences between core and sievedsoil water contents at — 15 bars were essentially constantthroughout the water content range encountered. The dif-ferences expressed as a percent of core water content rangedfrom —52.0 to —4.0 at 5 and 25% core water contents,respectively. The differences, however, may not be real be-cause the 1 to 1 line was within the confidence limit ofthe regression line throughout most of the indicated range.Only between about 7.0 and 13.5% soil water was the 1 to1 line outside of the 95% probability confidence limit ofthe regression line shown in Fig. 2. Percentage errors maybe greater when sieved soil is used for water content deter-minations when the soil is coarse-textured than when it isfine-textured, especially at — Ya bar. However, from a prac-tical standpoint, amount differences are probably moreimportant than percent differences.

Using core water content as the base, the relationshipbetween the percent change (sieved soil minus core soilwater content divided by core water content times 100) andcore water content, all at — Y a bar, was significant at' the1.0% level (Eq. 7, Table 2). Although the correlation coef-ficient was significant, the standard error of estimate waslarge (16.518), showing limited usefulness of the regressionequation for predicting percent differences between sievedand core soil water contents from core water contents. Thisanalysis, however, illustrates that large errors can be intro-

duced when estimates of field soil water relations are basedon water retention measurements using sieved soil in alaboratory.

Close relationships between soil water content at — Ya or— 15 bars and clay content were previously reported(Unger, 1975). The actual water content differences (per-centage points) between sieved and core soil samples werealso closely related to soil clay content (Eq. 8, Table 2).Based on this equation, water retention was lower for sievedsoil than for cores if the clay content was below 11.3%. Athigher clay contents, the opposite was true. These resultssuggest that deep tillage and profile modification may re-duce water retention on coarse-textured soils and increasewater retention on fine-textured soils. These results alsosuggest that mixing of soils having horizons of differenttextures could increase the total amount of water held ina soil profile. While these results apply to soils in general,individual soils may behave quite differently due to differ-ences in structural development, organic matter content,and possibly clay size and type. Consequently, to preciselyevaluate the possible-effects of tillage and profile modifica-tion on soil water retention, the soil in question must beanalyzed.