Download - The effect of soil compaction and soil physical properties on the mechanical resistance of South African forestry soils

Ž .Geoderma 78 1997 93–111

The effect of soil compaction and soil physicalproperties on the mechanical resistance of South

African forestry soils

C.W. Smith a,), M.A. Johnston b, S Lorentz c

a Institute for Commercial Forestry Research, UniÕersity of Natal, P.O. Box 100281, ScottsÕille 3208, SouthAfrica

b Department of Agronomy, UniÕersity of Natal, PriÕate Bag X01, ScottsÕille 3209, South Africac Department of Agricultural Engineering, UniÕersity of Natal, PriÕate Bag X01, ScottsÕille 3209, South

Africa

Received 10 May 1996; accepted 12 February 1997

Abstract

High soil strengths frequently develop in compacted forestry soils thus limiting root develop-ment of commercial timber species. The effect of soil compaction on mechanical resistance was

Ž .assessed for 29 South African forestry soils ranging widely in texture 8 to 66% clay and organicŽ .carbon content 0.26 to 5.77% . Models are presented which show that clay content and organic

Ž .carbon strongly influence the relationship between penetrometer soil strength PSS , bulk densityand water content. Organic carbon levels were not high enough to have a large effect on the

Ž .PSS–compaction–water content relationship for finer-textured soils )30% clay . However,evidence is presented to suggest that organic matter becomes more important in influencing PSSfor a range of water contents and bulk densities as clay content decreases. For cohesive soilsŽ .generally those with more than approximately 20% silt plus clay increasing clay content reducesthe rate at which PSS increases with decreasing water content. Similar to clayey soils, non-cohe-

Ž .sive soils -20% silt plus clay displayed slow rates of decrease in PSS as the soil dried althoughthe magnitude of PSS was considerably less for similar matric potentials and relative bulk

Ž .densities than for finer-textured soils. Thus the PSS at wilting point y1500 kPa increased withincreasing clay content. At a matric potential of y10 kPa PSS was most strongly related toloss-on-ignition indicating that organic carbon and clay mineralogy rather than soil texture aloneaffect soil strength in wet soils.

Keywords: soil strength; penetrometer; mechanical resistance; forestry soils; soil compaction

) Ž . Ž .Corresponding author. Tel.: q27 331 62314; Fax: q27 331 68905; E-mail: [email protected]

0016-7061r97r$17.00 q 1997 Elsevier Science B.V. All rights reserved.Ž .PII S0016-7061 97 00029-3

( )C.W. Smith et al.rGeoderma 78 1997 93–11194

1. Introduction

Measurements of soil strength to estimate mechanical resistance experienced by rootsystems have been widely used in commercial timber plantations to characterise the

Žcompacted state of forestry soils Sands et al., 1979; Jakobsen and Greacen, 1985; Grey.and Jacobs, 1987 . Reductions in rooting depth for various tree species grown in

Žcompacted soils have been attributed to increases in mechanical resistance Sands and.Bowen, 1978; Zisa et al., 1980; Tuttle et al., 1988 . A commonly accepted technique is

to measure mechanical impedance with a penetrometer. Although the measurement ofŽmechanical impedance in this way differs from that experienced by a root Barley and

.Greacen, 1967 , penetrometers have been used widely as comparative measures of soilŽstrength and as rapid appraisals of soil compaction in the field Campbell and O’Sulli-

. Ž .van, 1993 . Bengough 1993 has stated that the best indirect method of estimating soilresistance to root growth is by measuring soil resistance to a probe or penetrometer.

A factor limiting the use and interpretation of penetrometers in the field or as anindex of excessive compaction is that there is usually an insufficient data base for

Žpenetrometer resistance as a function of soil water content and bulk density Gupta and.Allmaras, 1987; Bennie and Burger, 1988 . A widely accepted norm has been to

measure penetrometer resistance at field capacity, which is essentially a reference point.However, it is very difficult to infer strength characteristics throughout the wholeavailable water range from a single measurement.

As soil strength is strongly related to water content it may vary considerablyŽ .throughout the year with wetting and drying cycles Spain et al., 1990 . The relationship

between strength and water content for a particular soil is strongly influenced by theŽ .degree of compaction Mirreh and Ketcheson, 1972 and this may affect forestry

management practices, such as the timing of tillage and site preparation practices. Whileit is clear that mechanical resistance is related to water content and bulk density theliterature is conflicting on the role of soil physical and chemical properties on thedevelopment of soil strength. Soil strength has been shown to be influenced by soil

Ž . Ž .texture Gerard, 1965; Byrd and Cassel, 1980 . Gerard 1965 , for example, showed thatthe strength of remoulded briquets, at a range soil water contents, increased with

Ž .increasing silt and clay contents. Similarly, Bennie and Burger 1988 illustrated therelationship between penetration resistance and silt plus clay content for a number of

Ž .predominantly sandy-textured soils clay plus silt -20% . Penetration resistance for thesame water content and bulk density increased with increasing amounts of silt plus clay.Increasing amounts of clay may also be responsible for increases in soil strength and

Ž .retardation of root growth Gerard et al., 1982 . Utilising reconstituted soil corescontaining between 66 and 83% sand, all with a similar bulk density, Byrd and CasselŽ .1980 reported a direct correlation between soil texture and mechanical resistance.

Ž .However, Stitt et al. 1982 studied the relationship between physical, chemical andmineralogical properties of a group of Atlantic coastal plain soils in the U.S.A. andfound no direct correlation between mechanical impedance and soil texture nor anyevidence that cementing agents were affecting cone index values.

Identification of factors affecting strength development is important for evaluating theeffects of compaction on soil properties and tree growth, soil trafficability and timing of

( )C.W. Smith et al.rGeoderma 78 1997 93–111 95

tillage operations. The objective of this study was to characterise the effect of soilŽ .compaction as reflected by bulk density on the relationship between penetrometer soil

strength and water content for a range of South African forestry soils, and to examinewhether these relationships could be related to commonly measured soil properties.

2. Materials and methods

2.1. Soils studied

Twenty-nine soils representing a wide range of soil textures and organic carboncontents from the main timber-growing regions in South Africa were used in thisinvestigation. The soils were selected from the main timber-growing areas in the easternseaboard of South Africa. This region is a predominantly summer rainfall region varyingin altitude from 0 to 1800 m. Rainfall is in the order of 700–1400 mm per annum andthe mean annual temperature ranges from 138 to 188C. Soil samples were taken fromfreshly cut faces in open pits. At the same time the soil profiles were described and

Žclassified according to the systems of South Africa Soil Classification Working Group,. Ž .1991 and the United States Soil Survey Staff, 1990 .

2.2. Sample preparation

Two sets of soil samples were prepared:Ž .1 ‘Repacked’ soil cores prepared for uniaxial compression tests in a separate study

Ž .Smith et al., 1997 for all the forestry soils. In this way it was possible to measurepenetrometer soil strength on a wide range of bulk densities and water contents for eachsoil. Field samples of soil were air-dried and passed through a 2 mm sieve. Each soilwas wet up to a range of water contents between saturation and wilting point. For eachsample approximately 2 kg of the air-dry soil was poured into a plastic tray and broughtto the desired water content by wetting with an atomizer followed by thorough mixing.The tray was then placed in a plastic bag and the sample was allowed to equilibrate for48 h. After equilibration, soil samples at each water content were placed in aluminiumcylinders, 77 mm in diameter and 50 mm long. As the cylinders had no attached base,they were placed on a perforated 5 mm metal base before the soil was added. Thecylinders were gently tapped to allow settling of the soil particles. Soil samples in thecylinder were then subject to a range of applied pressures by hydraulic press consisting

Ž .of a hydraulic ram connected to a piston Koolen, 1974 . The pressure was maintainedfor a few seconds and then released. The samples were allowed to ‘rebound’ before finalheights were measured for bulk density calculation.

Ž .2 ‘Repacked’ soil cores utilised during the determination of the water retentionŽ .characteristics of forestry soils Smith, 1995 . PSS was determined on all the soils

following equilibration at either y10 kPa or y1500 kPa after drying from saturation.The samples were prepared for the determination of full water retentivity curves from


saturation to y1500 kPa, following similar procedures to those of Hill and SumnerŽ . Ž . Ž .1967 and Mirreh and Ketcheson 1972 . Air-dry sieved soil -2 mm soil was used toprepare the soil cores. The mass of soil required to obtain the desired bulk density when

Ž .compressed into an aluminium cylinder 53.0 mm in diameter and 29.5 mm long wasmeasured out. The samples were then equilibrated in a plastic bag for two days at water

Žcontents approximating the critical water content for maximum compressibility Smith et.al., 1997 . These were then compressed into the cylinder using an Instron Universal

Ž . 1Testing Machine IUTM . For the higher bulk densities, the sample was compressed intwo increments. The first half was compressed into the cylinder and then the second halfwas compressed on top of that. Triplicate cylinders at three or four levels of compactionwere established for each soil. All the soil cores were saturated with water for severaldays to ensure as complete a saturation of the soil as possible. As the original weight ofair-dry soil was known, together with the dimensions of the core and the air-dry watercontent, it was possible to ascertain when the soil was close to saturation by weighing.For equilibration at y10 kPa soils were equilibrated on a tension table apparatus

Ž .consisting of diatomaceous earth over coarse sand Smith and Thomasson, 1974 . Soilwater contents were recorded at matric potentials of y1.0, y2.5, y5.0, y7.5 andy10.0 kPa, allowing a minimum of 48 h for equilibration at each matric potential. Thesoil cores were equilibrated at y1500 kPa, corresponding to ‘wilting point’, using a 15bar ceramic plate in a pressure chamber.

( )2.3. Penetrometer soil strength PSS

The compressed soil cores with an attached metal base were placed on a top-panbalance. PSS was then measured on the compressed cores using a 608, 2 mm basaldiameter cone penetrometer, relieved to 1.5 mm behind the tip in order to reduce thesoil–steel friction component of cone resistance as much as possible. Also the use ofsuch a small cone in 75 mm laterally confined soil cores avoids problems due to edgeeffects. Core diameter may affect cone resistance if the core diameter is less than 20

Ž .times that of the probe Bengough, 1993 . The penetrometer was mounted on an InstronŽ .Universal Testing Machine IUTM and penetrated the cores at a rate of 10 mmrmin to

a depth of 40 mm. Two replicate penetrations were carried out for each soil core.Because of a malfunction in the recording chart mechanism of the IUTM the soil coreswere placed on a balance which was tared to zero. Balance readings were recorded at 5mm depth increments, producing seven readings altogether, from which an average wascalculated. The initial reading at 5 mm depth was disregarded as PSS increased withdepth up to a certain depth. This depth, beyond which PSS became relatively constant, is

Ž .known as the critical penetration depth Bradford, 1986 and was achieved for most soilsbetween 5 and 10 mm. The force required for the penetrometer to penetrate the soil was

Ž .calculated by converting the balance reading kg into cone resistance or penetrometer

1 The use of trade names does not imply endorsement.


Ž . Ž . Ž 2 .soil strength MPa by dividing the load kg by the cone basal area m andŽ .multiplying by the acceleration due to gravity Bradford, 1986 . Thus:

PSS MPa sMgrp r 2 1Ž . Ž .y2 Ž .where: gs9.807 m s ; Msmass kg ; and rsbasal radius of penetrometer cone

Ž 2 .m .For a cone with a basal diameter of 2 mm, a mass of 1 g recorded on the balance

corresponded to a cone resistance of 3.12166 kPa. On completion of the penetrationmeasurements the cores were oven-dried and bulk density and water content weredetermined.

2.4. Analytical methods

Ž .Soil texture was described in terms of the percentage of clay -0.002 mm , fine siltŽ . Ž . Ž .0.002–0.02 mm , coarse silt 0.02–0.05 mm , fine sand 0.05–0.25 mm , medium sandŽ . Ž .0.25–0.50 mm , and coarse sand 0.50–2.00 mm . Soil samples were pre-treated with

Ž .hydrogen peroxide 30%, vrv and the size fractions were determined by the pipetteŽ . Žmethod Day, 1965 after treatment with calgon sodium hexametaphosphate and sodium.carbonate and ultrasound. Organic carbon was determined by the Walkley–Black wet

Ž . Ž .oxidation method Walkley, 1947 . Loss-on-ignition LOI was determined by the lossŽ .in mass after ignition at 4508C and expressed as a percentage of oven-dry 1058C soil

mass. At this temperature, and for a similar range of South African forestry soils,Ž . ŽDonkin 1991 showed that LOI was strongly correlated with organic carbon Walkley–

.Black method and soil texture. Soil pH was determined in a suspension of 10 g soil inŽ .25 ml deionised H O, and in 1 M KCl. Effective cation exchange capacity ECEC was2

calculated as the sum of exchangeable basic and acidic cations and was expressed incmol kgy1 soil.c

3. Results and discussion

A feature of the soils was the wide range of textures and organic carbon levels. ClayŽ .contents ranged from 8 to 66% and organic carbon from 0.26 to 5.77% Table 1 . In

general, 2 : 1 clays were lacking as evidenced by the low ECEC values which rangedfrom 0.85 to 12.96 cmol kgy1. The highly weathered nature of the soils is reflected byc

some 85% of all soils having an ECEC of less than 6.00 cmol kgy1 soil and low tocŽ . Ž .very low pH KCl values Table 1 . The soils studied comprised mainly of Oxisols,

Ž .Ultisols, Alfisols and Entisols Soil Survey Staff, 1990 .

( )3.1. Effect of soil compaction on penetrometer soil strength PSS Õersus water contentrelationships for Õarious forestry soils

Ž .Penetrometer soil strength PSS versus water content relationships for a range ofbulk densities for selected forestry soils are illustrated in Fig. 1. The graphs on the


Tab

le1

Soil

iden

tific

atio

nan

dsi

tech

arac

teri

stic

sof

the

soils

stud

ied

Part

icle

size

Soil

clas

sifi

catio

nSi

teL

oss-

on-

Org

anic

EC

EC

pHŽ

.di

stri

butio

n%

igni

tion

carb

onb

a,b

Soil

form

and

Soil

taxo

nom

yPa

rent

mat

eria

ly

1Ž

.Ž

.Ž

.Ž

.N

o.%

%cm

olkg

KC

la

ccl

aysi

ltsa

ndfa

mily

1AL

usik

i111

0T

ypic

Kan

hapl

usta

lfT

illite

2247

316.

624.

026.

773.

952A

Car

tref

2100

Typ

icH

apla

quep

tC

oars

e-gr

aine

dar

kosi

c13

1770

2.29

1.42

3.51

3.95

sand

ston

e2E

1115

740.

810.

321.

424.

253A

Lus

iki2

120

Aqu

icK

anha

plus

talf

Till

ite24

3441

5.86

2.37

4.04

3.85

4AM

agw

a12

00U

stic

Kan

hapl

ohum

ult

Arg

illac

eous

shal

e41

527

14.1

82.

106.

383.

955A

Inan

da12

00U

stic

Kan

dihu

mul

tM

icac

eous

sand

ston

e30

1555

6.73

2.15

3.94

n.d

6AK

rans

kop

1100

Hum

icH

aplu

stox

Dol

erite

5144

518

.49

5.77

4.82

4.45

6B46

2925

13.9

82.

642.

264.

756B

265

2312

11.2

71.

091.

645.

157A

Kra

nsko

p11

00H

umic

Hap

lust

oxD

oler

ite56

2133

11.3

04.

223.

564.

208A

Inan

da12

00T

ypic

Kan

hapl

ustu

ltA

rgill

aceo

ussh

ale

4044

168.

313.

589.

664.

558B

5037

134.

471.

0610

.27

4.59

9AN

oman

ci22

00Pa

chic

Hap

lum

brep

tG

neis

srsc

hist

2722

518.

383.

8312

.96

n.d.

10A

Car

tref

2200

Typ

icH

apla

quep

tT

illite

1634

502.

150.

953.

013.

8014

AH

utto

n21

00T

ypic

Kan

dius

tult

Coa

rse-

grai

ned

129

791.

370.

431.

163.

90sa

ndst

one


15A

Fern

woo

d21

10T

ypic

Ust

ipsa

mm

ent

Aeo

lian

sand

911

803.

131.

651.

894.

1016

AN

oman

ci12

00L

ithic

Ust

ochr

ept

Bio

tite

gran

ite24

1759

4.41

2.36

4.75

4.95

17A

Hut

ton

1200

Typ

icK

andi

ustu

ltB

iotit

egr

anite

2610

642.

751.

493.

254.

3017

B29

665

2.94

1.09

3.22

4.25

18A

Inan

da12

00T

ypic

Kan

hapl

ustu

ltL

euco

crat

icgr

anite

2816

564.

012.

422.

553.

7019

AK

rans

kop

1100

Hum

icX

anth

icD

iaba

se66

286

15.3

34.

133.

674.

10H

aplu

stox

20A

Clo

velly

1100

Lith

icH

aplu

stox

Med

ium

-gra

ined

3516

494.

031.

371.

763.

95sa

ndst

one

20B

3119

503.

580.

921.

434.

0521

AK

rans

kop

1100

Typ

icH

aplu

stox

Gra

nitic

gnei

ss44

1541

9.01

4.23

2.84

4.15

21B

4616

384.

931.

052.

324.

3522

AH

utto

n12

00R

hodi

cK

anha

plus

tult

Hor

nble

nde

biot

ite32

860

3.98

1.21

2.24

4.25

22B

528

404.

860.

342.

694.

3523

ASw

artla

nd12

11R

hodi

cK

anha

plus

talf

Dia

base

3420

465.

271.

575.

745.

2024

ASw

artla

nd21

11T

ypic

Kan

hapl

usta

lfH

ornb

lend

ebi

otite

3017

532.

871.

947.

885.

9024

B38

1844

2.80

1.46

2.25

4.30

25A

Inan

da12

00T

ypic

Kan

dihu

mul

tB

iotit

egr

anite

3411

556.

372.

851.

984.

0526

AM

agw

a11

00H

umic

Xan

thic

Hap

lust

oxB

iotit

egr

anite

3011

5910

.04

3.42

2.74

3.85

aŽ

.So

ilC

lass

ific

atio

nW

orki

ngG

roup

1991

.b

Ž.

Soil

Surv

eySt

aff

1990

.


Ž .Fig. 1. a–f The effect of compaction on the penetrometer soil strength–water content relationship for selectedŽ .forestry soils. The symbol WP is the water content corresponding to wilting point y1500 kPa . FCL and

Ž .FCH are the water contents corresponding to field capacity y10 kPa at low and high levels of compaction,Ž y1 . Ž y3 .respectively. u sgravimetric water content kg kg , r sbulk density Mg m .m b


Ž .Fig. 1. continued .


left-hand side have been transposed from the original data on the right-hand side forŽ .clarity Smith, 1995 . All the graphs in Fig. 1 have been drawn at the same scale to

Ž .enable a visual comparison. Water retention data from Smith 1995 has been superim-posed on the x-axis so that an initial interpretation of the PSS values with respect toavailable water capacity criteria can be made. As the matric potential at a given water

Ž .content depends upon the compaction level Larson and Gupta, 1980 , three values ofŽ .matric potential, those corresponding to wilting point y1500 kPa and field capacity

Ž .y10 kPa , are given in the graphs corresponding to the lowest and highest bulkŽ .densities WP and FCL and FCH, respectively . Only one water content is presented for

WP since the changes in water content at WP for different levels of compaction weretoo small to be represented on the graphs. It should also be pointed out that the PSSreadings were not carried out on soils dried under tension during the determination ofthe water retention characteristic. Gravimetric water content has been used becausevolumetric water content or matric potential may change during penetration due to

Žparticle rearrangement whereas gravimetric water content remains constant Koolen and.Kuipers, 1983 .

For all soils PSS increases with increasing bulk density and decreasing water content,except at lower levels of compaction when there is usually a decline in PSS as the soilbecomes very dry. A feature of all the graphs is that, for a range of bulk densities, onlysmall differences in PSS occur at water contents approaching field capacity and wetter.As the soils dry the lines diverge, illustrating that differences in soil strength fordifferent bulk densities are greater at lower rather than high water contents.

Ž . Ž .For the loamy sand 15A Fig. 1a only small differences in strength developmentwere noted across a wide range of water contents. In contrast to the other soils in Fig. 1,PSS for the loamy sand does not undergo an abrupt change at any bulk density as watercontent changes. This can be primarily related to the contribution of frictional ratherthan cohesion forces to PSS. Surface tension forces may contribute to resistance atintermediate water contents, termed annular bridges by Panayiotopolous and MullinsŽ .1985 , which bind soil particles together forming bonds of sufficient strength to providea degree of resistance. These forces will be negligible when the sandy soil becomes verywet or very dry.

South African forestry soils with between 12 and 20% clay and derived from eithersandstone sediments or glacial tillite are known to pose a number of establishmentproblems for forestry management due to their poor consistence. Two such soils, the

Ž . Ž . Ž .sandy loam 14A and the loam 10A Fig. 1b and c, respectively illustrate thesensitivity of PSS to water content especially at high bulk densities. Pronounced

Ž .increases in PSS from 1 to 5 MPa for the sandy loam occurred over a range of waterŽ .contents, from a range of as little as 4% by mass . Similar results have been reported by

Ž .Mullins et al. 1989 for hardsetting soils. The proximity of the iso-stress lines for theŽ .sandy loam 14A, Fig. 1b compared to the more widely spaced iso-stress lines for the

Ž .loam 10A, Fig. 1c is probably related to the slightly greater cohesion and availableŽ .water capacity which reflects a more even pore size distribution of the loam due to its

Ž .higher clay plus silt content 50% as opposed to 21% for the sandy loam . Alsoobservations in the field and the laboratory show that both soils slump readily when theyare wet up rapidly. As the soil approaches field capacity, the bulk densities for


uncompacted soil increases. This is due to the smaller pores filling with water thusseriously reducing the effectiveness of the high-tension water bridges binding the

Ž .particles together Akram and Kemper, 1979 . Consequently, particles succumb togravitational forces and, in the case of the more sandy-textured soils, slumping is oftenobserved.

Ž .The PSS of the sandy clay loam soil 22A developed from granite displays a strongŽ .dependence on water content and compaction level Fig. 1d . The PSS levels at high

water contents are very low but increase rapidly as the soils dry at moderate levels ofcompaction. With higher clay contents changes in PSS take place across a wider range

Ž . Ž . Žof water contents for the clay 19A and silty clay 4A 66% and 41% clay,.respectively, Fig. 1e and f . As these soils approach wilting point the PSS values

increase less rapidly at higher levels of compaction than the less clayey soils. These soilsalso differ from the soils with a lower clay content in that PSS at field capacity is higherat all levels of compaction.

A feature of all the graphs presented here is the slight convexity of the PSS–watercontent relationship at low bulk densities. The PSS commonly decreases as the soil driesbeyond wilting point. This slight drop in PSS as the soils become dry at relatively low

Ž .bulk densities has also been recorded by Mirreh and Ketcheson 1972 and Akram andŽ . Ž .Kemper 1979 . Mirreh and Ketcheson 1972 developed relationships between pen-

etrometer resistance, bulk density and matric potential for a clay loam soil and noted thatat low bulk densities mechanical resistance of the soil passed through a maximum asmatric potential decreased from y100 to y800 kPa. It was suggested that the convexityof the soil resistance surface along the matric potential axis at low bulk densities, similarto the water content–PSS relationships in Fig. 1, was caused by inter-particle moisturebonds increasing in strength as water is drained from larger pores. Further drainageresults in a larger number of broken bonds and a net decline in resistance results.

A number of authors have placed great emphasis on the role of effective stress in theŽgeneration of high levels of soil strength, particularly in hardsetting soils Mullins and

. Ž .Panayiotopolous, 1984; Mullins et al., 1989; Ley et al., 1993 . Mullins et al. 1989emphasised that since hardsetting is usually accompanied by slumping, which itself is aprocess of compaction, it is difficult to separate the effects of compaction from those ofeffective stress in the development of high levels of soil strength. Although no direct

Ž .evidence was presented in this study effective stresses were not calculated , the natureof the relationships in Fig. 1, and in particular those for the two hardsetting soils in the

Ž .study, the sandy loam and the loam Fig. 1b and c suggests that compaction rather thaneffective stress, is primarily responsible for the magnitude of PSS. This was borne out

Ž .by Vepraskas 1984 who reported a close correlation between cone index and effectivestress but the contribution of effective stress to the overall soil strength was very smalland depended more on factors such as bulk density and soil type. The nature of therelationships presented in Fig. 1 show that effective stress influences the rate at whichPSS increases but only as the soil dries at higher compaction levels for all the soils. Thisphenomenon is particularly noticeable for soils with low organic matter contents andclay contents of between about 12 and 30%, i.e., those categorised by Mullins et al.Ž .1989 as being susceptible to hardsetting.


( ) ( )3.2. Penetrometer soil strength PSS at wilting point y1500 kPa and field capacity( )y10 kPa

As the normal range of bulk densities and water contents varies depending upon soiltexture, establishing the actual effect of clay content on PSS at a constant bulk densityand water content would have limited value. It was decided, therefore, to compare PSSvalues at wilting point and field capacity for a range of selected soils at two levels of

Ž .compaction: high, corresponding to a relative bulk density RBD of 0.9; and moderate,Ž .corresponding to a RBD of 0.8 Figs. 2 and 3 . RBD is simply the actual bulk density

divided by maximum bulk density, the latter being measured by the Proctor Impact testŽ .American Society for Testing and Materials, 1985 . A simple linear and multipleregression was carried out for each compaction level to test the significance of therelationship between PSS and soil physical properties.

A significant relationship exists between PSS and clay content for both levels ofŽ .compaction when soils are at a matric potential of y1500 kPa Fig. 2 . A power

function showed a better fit of the data than a simple linear regression. This demon-strated that at both moderate and high levels of compaction, PSS at wilting pointincreases with increasing clay content up to a point after which PSS declines. These

Ž .results tend to support the results of Ball and O’Sullivan 1982 who showed thatpenetrometer resistance increased with a higher proportion of small particle sizes atconstant water content and bulk density. Similarly, for soils with less than 20% silt plus

Ž .clay, Bennie and Burger 1988 reported an increase in PSS with increasing clay contentfor soils at a water content of 0.1% by volume. However, as both of these studies had alimited range of soil textures, it was not clear whether PSS would continue to riseindefinitely with increasing clay content.

Ž .At field capacity y10 kPa a significant correlation was found between clay contentand PSS for both compaction levels but the correlation was poor, being significant atP-0.05. Interestingly, the correlation improved substantially when loss-on-ignitionŽ . Ž .LOI replaced clay as the independent variable Fig. 3 . The significant relationship

Fig. 2. Relationship between penetrometer soil strength and clay content at a matric potential of y1500 kPafor a range of selected forestry soils at two levels of compaction corresponding to 0.8 and 0.9 of relative bulkdensity.


Ž .Fig. 3. Relationship between penetrometer soil strength and loss-on-ignition LOI at a matric potential ofy10 kPa for a range of selected forestry soils at two levels of compaction corresponding to 0.8 and 0.9 ofrelative bulk density.

between LOI and PSS at wilting point for two levels of compaction was also recordedbut these correlations were not as strong as the correlations between PSS and claycontent at wilting point. This suggests that soil texture and organic carbon content which

Ž .are reflected by LOI Donkin, 1991 are more important in strength development athigher water contents than at lower water contents where the effect of clay content alone

Ž .is dominant. Causarano 1993 found that organic matter increases the strength of wetsoil and decreases the strength of dry soils.

3.3. A quantitatiÕe description of penetrometer soil strength

The graphs presented in the previous section showed that PSS is strongly related tobulk density and water content. An appropriate multiple regression equation was soughtwhich could explain the variation in PSS by using bulk density, mass water content and

Žsoil physical properties as independent variables. Stepwise multiple regression back-. Ž .ward option was employed Draper and Smith, 1981 and various combinations of the

variables were tested based on the nature of the relationships presented in Fig. 1. Asmentioned in the introduction, previous studies have either concentrated on developing

Žthese types of relationships for one soil only Mirreh and Ketcheson, 1972; Cassel et al.,. Ž1978 or a limited range of textures Byrd and Cassel, 1980; Bennie and Burger, 1988;

.Henderson et al., 1988 .Two soils, selected at random, were left out of the regression for model validation

purposes later. The final model selected illustrates that the relationship between PSS andbulk density and water content is strongly influenced by clay content and organic carbonŽ Ž ..Eq. 2 . The regression coefficients, standard errors and multiple correlation coefficientŽ . Ž .R are presented in Table 2. The model in Eq. 2 explains almost 82% of thevariability in PSS which is a considerable achievement considering the large number ofsoils in the study. The standard error of estimate was 0.2436 with a total of 1102 degrees

Ž .of freedom Table 2 .

PSSsyau ybr qc log r qd CLAYye ORG C 2Ž .m b b


Table 2Ž . Ž .Regression coefficients and standard errors of the terms in Eqs. 2 – 4

Equation Regression coefficients R

a b c d e

2 y5.0483 y3.6162 32.1616 0.1051 0.2273 0.908))

Ž . Ž . Ž . Ž . Ž .0.6652 0.1985 1.2017 0.0041 0.03823 y17.5797 y4.7966 39.2046 0.1535 0.6207 0.914))

Ž . Ž . Ž . Ž . Ž .1.4174 0.2648 1.7024 0.0078 0.06814 y1.3329 6.9498 0.1549 y7.7943 0.926))

Ž . Ž . Ž . Ž .0.7925 0.2854 0.0109 0.4633

Note that all values of R are significant at P -0.01. Standard errors are shown in brackets.

The model has confirmed the features of the relationships presented so far in that soiltexture and organic carbon appear to have a considerable influence on the nature of thePSS–water content relationship for different compaction levels. The exclusion of otherparticle sizes, including clay plus silt, from the model has shown that clay content is theparticular textural component influencing the relationship. The results support those of

Ž .McCormack and Wilding 1981 who related penetrometer soil strength to bulk density,gravimetric water content and clay content and developed a multiple regression equationrelating these properties to PSS which explained 78% of the variation in soil strength.

Ž .Similarly, Gerard et al. 1982 reported that soil strength, measured by a small conepenetrometer, was significantly correlated with bulk density, void ratio and clay content.The exclusion of clay plus silt from the final model was at variance with a number of

Ž .authors Gerard, 1965; Byrd and Cassel, 1980 who found a significant relationshipbetween clay plus silt and PSS values in combination with compaction level and watercontent.

Ž .The inclusion of organic carbon in Eq. 2 is of interest since previous inspection ofthe PSS–water contentrbulk density relationships of the type presented in Fig. 1showed that organic carbon appeared to have a greater effect on these relationships formore coarsely textured soils. Therefore the stepwise regression procedure was carriedout separately on data sets for soils having above and below 30% clay. The final modelsselected are given below and the regression coefficients, standard errors and multiple

Ž .correlation coefficients R are presented in Table 2.For soils with less than 30% clay:

PSSsyau ybr qc log r qd CLAYqe ORG C 3Ž .m b b

For soils with more than 30% clay:

PSSsyau qbr qc CLAYyd log CLAY 4Ž .m b

The inclusion of organic carbon as an independent variable in the model for soilsŽ Ž ..with less than 30% clay Eq. 3 and its exclusion from the model for soils with more

Ž Ž ..than 30% clay Eq. 4 demonstrates the importance of organic carbon in affecting thePSS of the more coarsely textured soils. For more finely textured soils, organic carbonlevels may need to be much higher to have any effect on PSS. For example, Ohu et al.Ž .1986 reported that organic matter contents of between 10 and 17% decreased the


strength of three soils of varying clay content at any water content at various levels ofsoil compaction. Such high levels of organic matter are well in excess of the organicmatter levels reported in this study. Clay content alone, rather than in combination withorganic carbon, becomes increasingly important in affecting the PSS of the more finely

Ž Ž ..textured soils Eq. 4 . The fact that slightly more of the variation in PSS can beŽ .explained by splitting soils into categories based on clay content suggests that Eqs. 3

Ž .and 4 are most useful for the prediction of PSS from soil physical properties.

Ž .Fig. 4. Predicted relationships solid lines versus measured values of the relationship between penetrometerŽ . Ž . Ž .soil strength PSS and water content at a range of compaction levels for: a sandy clay loam, and b loam.

The broken lines correspond to the best fit line through the data points predicted by the models for theŽ . Ž .individual soils in Eqs. 3 and 4 , and Table 2.


The effect of organic matter on PSS values could be due to organic particlesdisrupting the surface tension forces which contribute to soil strength as the soils dryŽ . Ž .Mullins et al., 1989; Causarano, 1993 . Gupta et al. 1989 suggested that organicresidue particles are more effective in separating single-grain particles in sandy soilsthan in finer-textured soils due to the lower surface area of the former. Since the veryhigh organic matter levels in this study were usually associated with higher clay contentsorganic matter is therefore unlikely to have any mechanical effect on the more finely

Ž .textured soils due to the high surface area of the clays Gupta et al., 1989 . It isŽsuggested that the dominant effect on strength properties for fine-textured soils loam

.and finer is soil texture rather than organic matter.

3.4. Model Õalidation

The effect of compaction on the PSS–water content relationship is shown in Fig. 4Ž .for the two soils which were left out of the original stepwise regression, a loam 10A

Ž . Ž .and a sandy clay loam 23A . This was compared to the predicted values solid linesŽ . Ž . Ž . Ž . Žusing the model given in Eq. 3 loam, 10A and Eq. 4 clay loam, 23A Fig. 4a and

.b, respectively . The broken lines indicate the best fit through the data points from theŽ . Ž .individual model for each soil evaluated from Eqs. 3 and 4 and Table 2. In both cases

the model slightly underpredicts PSS at very low water contents and overpredicts PSS athigh water contents. However, the agreement between predicted and measured isremarkably good. It is clear that clay content and organic carbon can be effectivelyutilised to predict the effect of compaction and water content on PSS.

4. Conclusions

The relationships developed have provided an insight into the effect of soil com-Ž .paction and water content on penetrometer soil strength PSS . This study has shown

that PSS is principally related to bulk density, water content, clay content and organiccarbon. For all the soils in this study, PSS increases with increasing bulk density anddecreasing water content, except at lower levels of compaction when there is usually adecline in PSS as the soil becomes very dry. Only small differences in PSS occur atwater contents approaching field capacity and wetter for a range of bulk densities.Differences in PSS at various bulk densities are greatest when soils are dry rather thanwhen they are wet.

For cohesive soils, increasing clay content reduces the rate at which PSS increases asthe soil dries to wilting point and beyond. For non-cohesive soils in the studyŽ .approximately -20% clay plus silt , the rate of increase in PSS as the soils dry issimilar to that for soils with a high clay content, despite changes taking place at verydifferent water contents. At wilting point, higher PSS values are obtained with increas-ing clay content up to about 45 to 50% clay, after which PSS declines. With increasing

Ž .loss-on-ignition LOI contents, PSS values are higher at field capacity, especially athigher bulk densities.

The regression models showed that the relationship between PSS and bulk densityand water content was affected by clay content and organic carbon for soils with less


than 30% clay and by clay content alone for soils with greater than 30% clay. Organiccarbon levels, though ranging from 0.16 to 5.77%, were probably not high enough tohave an influence on PSS for the more clayey soils. A clay content of 30% is suggestedas a threshold value below which moderate changes in organic carbon content can havean appreciable effect on PSS. Although the results in this study relate to disturbedsamples it is believed that the results will reflect the behaviour of field soils since theclay fraction of the majority of South African forestry soils is dominated by kaoliniticclays and hydrous oxides resulting in a lack of macro-structure and the tendency todevelop strong micro-aggregation.

The models can be used to predict the likely effects of soil compaction on themechanical impedance of soils to root penetration and, as a result, to assist in a morerefined assessment of the compaction susceptibility of forestry soils. Coupled withinformation on critical root penetration criteria, the models could serve as an invaluabletool in soil waterrsoil fertility studies. In addition, the correlation between PSS and

Ž .trafficability Knight and Frietag, 1962 will enable a more meaningful appraisal of soiltype in forestry terrain classifications.

Acknowledgements

The authors are indebted to Mrs. Mary Galbraith and Mr. Innocent Mchunu for theirtechnical assistance throughout this project. Mrs. Galbraith is also acknowledged for herhelp in the preparation of the figures.

References

Akram, M., Kemper, W.D., 1979. Infiltration of soils as affected by pressure and water content at the time ofcompaction. Soil Sci. Soc. Am. J. 43, 1080–1086.

American Society for Testing and Materials, 1985. Standard test methods for moisture density relations ofŽ . Ž .soils using a 5.5-lb 2.5 kg hammer and 12 inch 304.8 mm drop. Annual Book of ASTM Standards, Part

19. ASTM, Philadelphia, Pa., pp. 157–162Ball, B.C., O’Sullivan, M.F., 1982. Soil strength and crop emergence in direct drilled and ploughed cereal

seedbeds in seven field experiments. J. Soil. Sci. 33, 609–622.Barley, K.P., Greacen, E.L., 1967. Mechanical resistance as a soil factor influencing growth of roots and

underground shoots. Adv. Agron. 19, 1–43.Bengough, A.G., 1993. The penetrometer in relation to mechanical resistance to root growth. In: Smith, K.A.,

Ž .Mullins, C.E. Eds. , Soil Analysis: Physical Methods, 1st ed. Marcel Dekker, New York, pp. 431–445.Bennie, A.T.P., Burger, R. du T., 1988. Penetration resistance of fine sandy apedal soils as affected by bulk

density, water content and texture. S. Afr. J. Plant Soil 5, 5–10.Ž .Bradford, J.M., 1986. Penetrability. In: Klute, A. Ed. , Methods of Soil Analysis, Part I. Physical and

Mineralogical Methods. Agronomy Monograph No. 9, 2nd ed., American Society of Agronomy—SoilScience Society of America, Madison, Wisc., pp. 463–478.

Byrd, C.W., Cassel, D.K., 1980. The effect of sand content upon cone index and selected physical properties.Soil Sci. 129, 197–204.

Campbell, D.J., O’Sullivan, M.F., 1993. The cone penetrometer in relation to trafficability, compaction andŽ .tillage. In: Smith, K.A., Mullins, C.E. Eds. , Soil Analysis: Physical Methods, 1st ed. Marcel Dekker,

New York, pp. 399–429.


Cassel, D.K., Bowen, H.D., Nelson, L.A., 1978. An evaluation of mechanical impedance for three tillagetreatments on Norfolk sandy loam. Soil Sci. Soc. Am. J. 42, 116–120.

Causarano, H., 1993. Factors affecting the tensile strength of soil aggregates. Soil Tillage Res. 28, 15–25.Ž .Day, P.R., 1965. Particle fractionation and particle size analysis. In: Black, C.A. Ed. , Methods of Soil

Analysis. American Society of Agronomy, Madison, Wisc., pp. 562–566.Donkin, M.J., 1991. Loss-on-ignition as an estimator of soil organic carbon in A-horizon forestry soils.

Commun. Soil Sci. Plant Anal. 22, 233–241.Draper, N.R., Smith, H., 1981. Applied Regression Analysis, 2nd ed. Wiley, New York, N.Y., 709 pp.Gerard, C.J., 1965. The influence of soil moisture, soil texture, drying conditions and exchangeable cations on

soil strength. Soil Sci. Soc. Am. Proc. 29, 641–645.Gerard, C.J., Sexton, P., Shaw, G., 1982. Physical factors influencing soil strength and root growth. Agron. J.

74, 875–879.Grey, D.C., Jacobs, E.O., 1987. The impact of harvesting on forest site quality. S. Afr. For. J. 140, 60–66.Gupta, S.C., Allmaras, R.R., 1987. Models to assess the susceptibility of soil to excessive compaction. Adv.

Soil Sci. 6, 65–100.Gupta, S.C., Schneider, E.C., Larson, W.E., Hadas, A., 1989. Influence of corn residue on compression and

compaction behaviour of soils. Soil Sci. Soc. Am. J. 51, 207–212.Henderson, C., Levett, A., Lisle, D., 1988. The effects of soil water content and bulk density on the

compactibility and soil penetration resistance of some western Australian sandy soils. Aust. J. Soil Res. 26,391–400.

Hill, J.N.S., Sumner, M.E., 1967. Effect of bulk density on moisture characteristics of soils. Soil Sci. 103,234–238.

Jakobsen, B.F., Greacen, E.L., 1985. Compaction of sandy forest soils by forwarder operations. Soil TillageRes. 5, 55–70.

Koolen, A.J., 1974. A method for soil compactability determination. J. Agric. Eng. Res. 19, 271–278.Koolen, A.J., Kuipers, H., 1983. Agricultural Soil Mechanics. Springer-Verlag, New York, pp. 57–58.Knight, S.J., Frietag, D.R., 1962. Measurement of soil trafficability characteristics. Trans. Am. Soc. Agric.

Eng. 5, 121–132.Larson, W.E., Gupta, S.C., 1980. Estimating critical stresses in unsaturated soils from changes in pore water

pressure during confined compression. Soil Sci. Soc. Am. J. 44, 1127–1132.Ley, G.J., Mullins, C.E., Lal, R., 1993. Effect of soil properties on the strength of weakly structured tropical

soils. Soil Tillage Res. 28, 1–13.McCormack, D.E., Wilding, L.P., 1981. Soil properties influencing penetration resistances of Canfield and

Geeburg soils. Soil Sci. Soc. Am. J. 45, 1148–1152.Mirreh, H.F., Ketcheson, J.W., 1972. Influence of soil bulk density and matric potential on soil resistance to

penetration. Can. J. Soil Sci. 52, 477–483.Mullins, C.E., Panayiotopolous, K.P., 1984. The strength of unsaturated mixtures of sand and kaolin and the

concept of effective stress. J. Soil Sci. 35, 459–468.Mullins, C.E., Macleod, D.A., Northcote, K.H., Tisdall, J.M., Young, I.M., 1989. Hardsetting soils: Be-

haviour, occurrence, and management. Adv. Soil Sci. 10, 37–108.Ohu, J.O., Raghavan, G.S.V., McKyes, E., 1986. Shear strength prediction of compacted soils with varying

added organic matter contents. Trans. Am. Soc. Agric. Eng. 28, 201–424.Panayiotopolous, K.P., Mullins, C.E., 1985. Packing of sands. J. Soil Sci. 36, 129–139.Sands, R., Bowen, G.D., 1978. Compaction of sandy soils in radiata pine forests, 2. Effects of compaction on

root configuration and growth of radiata pine seedlings. Aust. For. Res. 8, 163–170.Sands, R., Greacen, E.L., Gerard, C.J., 1979. Compaction of sandy soils in radiata pine forests, 1. A

penetrometer study. Aust. J. Soil Res. 17, 101–113.Smith, C.W., 1995. Assessing the Compaction Susceptibility of Forestry Soils. Unpublished PhD thesis,

University of Natal, Pietermaritzburg, pp. 89–97, 118–131.Smith, C.W., Johnston, M.A., Lorentz, S., 1997. Assessing the compaction susceptibility of forestry soils, I.

The effect of soil type, water content and applied pressure on uniaxial compaction. Soil Tillage Res., inpress.

Smith, P., Thomasson, A.J., 1974. Density and water-release characteristics. In: Avery, B.W., Bascomb, C.L.


Ž .Eds. , Soil Survey Laboratory Methods. Tech. Monograph No. 6, Rothamstead Experimental Station,Harpenden, Herts., pp. 42–55.

Soil Classification Working Group, 1991. Soil Classification: a Taxonomic System for South Africa. Memoirson the Agricultural Resources of South Africa No. 15. Department of Agricultural Development, Pretoria,257 pp.

Soil Survey Staff, 1990. Keys to Soil Taxonomy, 4th ed. SMSS Technical Monograph, Blacksburg, Va., 422pp.

Spain, A.V., Prove, B.G., Hodgen, M.J., Lee, K.E., 1990. Seasonal variation in penetration resistance andshear strength of three rainforest soils from northeastern Queensland. Geoderma 47, 79–92.

Stitt, R.E., Cassel, D.K., Weed, S.B., Nelson, L.A., 1982. Mechanical impedance of tillage pans in AtlanticCoastal Plain soils and relationships with soil physical, chemical and mineralogical properties. Soil Sci.Soc. Am. Proc. 46, 100–106.

Tuttle, C.L., Golden, M.S., Meldahl, M.S., 1988. Soil compaction effects on Pinus taeda establishment fromseed and early growth. Can. J. For. Res. 18, 628–632.

Vepraskas, M.J., 1984. Cone index of loamy sands as influenced by pore size distribution and effective stress.Soil Sci. Soc. Am. J. 48, 1220–1225.

Walkley, A., 1947. A critical examination of a rapid method for determining organic carbon in soils—effect ofvariations in digestion conditions and of inorganic constituents. Soil Sci. 63, 251–264.

Zisa, R.P., Halverson, H.G., Stout, B.B., 1980. Establishment and early growth of conifers on compact soils inurban areas. US Department of Agricultural Forest Services, Northeastern Experimental Station, ResearchPaper NE 451, Radnor, Pa., 8 pp.

Download - The effect of soil compaction and soil physical properties on the mechanical resistance of South African forestry soils

Top Related