a

.‘o—r—l ',
“.4; . 4'

';...~,

2:»
1: '
33::

I
53%.

'nr

1;“?

:i’

. nwu

:2.

‘ 'Z‘yfa'fl 23‘1"?
\‘fi‘rlmql

unfl'm
Raga]?

Wkgfi v5

3““!
Hfid “

Eii‘l

‘ n
‘s-EJZZ'
‘t ‘:

.5

Vii-"5‘3
n‘
. 21.1.VA‘M i“! 1 I
1”“: 23m
.1“;th
‘ ‘ l‘ ‘q z I: Y
‘ «W‘égp' Q1331 2:“

' -"...-..-.-
- M"-
.,
”a
‘ - I

o .

4,33%:

yum‘.’

|““ . ,A
'1‘ _ m1! “:3:

51"?
J

£qu """‘§§'s§";‘ 1'3; ..
. {13" Egg," 3 '11. 1.
oli £11" .121}: «11 3b 1“

-' :3”...

"A

:T'
$3.1
132:; .. .

n
"1.4
.-

”Hid-{u

‘1

- A
42"3'21,“

A ”A
2 ‘\ i E‘;
:.¢_
I '2'“ .
, .3}?

«3*
M5." LPW'gf“
SEE?“ +1.}.

V: .E.

fie.

“aka ‘th.

I ‘ ‘ ii an“
'1': .igé'zfi’l‘kgi‘mflfi “‘3 ‘
4x ”11%;, *1“ -.
*'* ".fiir’sa‘
Lp-

“""m-L . _
may h‘:. 'f 2' . ““37:
mg T r

a

‘?

fi-

7‘

E
,"v
3.-

M
via .

 

m J IllllllllllllllllllllllllIlllllllllllHllllllllllllllllllllll

/) 302058 1 991

(7K

00 f)

This is to certify that the

dissertation entitled

MICRO-SCALE SPATIAL VARIABILITY OF CROP ROOTS ,WATER
CONTENT ,SOIL TEXTURE AND THEIR INFLUENCE ON SOIL WATER
EXTRACTION RATES.

presented by

CARLOS M . PAGLI S

has been accepted towards fulfillment
of the requirements for

 

Ph.D. degreein Crop and Soil Sciences

/27-/7/77

 

 

MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771

 

LIBRARY

Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MICRO-SCALE SPATIAL VARIABILI'IY OF CROP ROOTS, WATER CONTENT,
SOIL TEXTURE AND THEIR INFLUENCE ON SOIL WATER EXTRACTION
RATES

By

Carlos M. Paglis

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Crop and Soil Sciences

1999

ABSTRACT

MICRO-SCALE SPATIAL VARIABILITY OF CROP ROOTS, WATER CONTENT,
SOIL TEXTURE AND THEIR INFLUENCE ON SOIL WATER EXTRACTION
RATES

BY

Carlos M. Paglis

Water extraction by roots is an important component in crop growth
simulation models. The capability of plants to survive in limited conditions of
water supply largely depends on the uniformity and depth of its root system. If
roots are not present in some parts of the soil, it is likely that areas where water
is available will be left behind during root growth in the soil environment. Roots
frequently bypass these soil regions as a consequence of root clumping.

Several models simulate water extraction based on the assumption that
roots are uniformly distributed in the soil. However, considering the natural
heterogeneity of the soil environment and genetic factors inherent to plant
species, the assumption of uniformity of root distribution is not valid. In addition,
other factors such as anthropogenic action on soil formation, competition among
plants, and soil physio-chemical and biological stresses will also contribute to a
non—uniform distribution of the root system. Knowing that this non-uniform root
distribution should affect simulation of water extraction by roots, the objectives of
this study were to describe the micro-scale spatial variability of crop roots, water

content, soil texture and their influence on soil water extraction rates.

To describe such variability, a greenhouse experiment was conducted at
Michigan State University, MI, and another one in the field, at Maricopa
Agricultural Center, AZ. The plants were grown in a terminal drought condition
until they were severely impacted by soil water deficits. Measurements of
volumetric water content within soil layers in the greenhouse and in the field
experiments were done with a neutron-probe gauge. A small TDR probe was
used to determine volumetric water contents in 2.5 cc soil samples collected in a
grid pattern in each soil layer at the termination of the both experiments. For
each soil sample the amount of roots and soil texture was determined. A
functional model for root water uptake was used to simulate water depletion for
each soil layer in the greenhouse experiment.

The results revealed that roots were non-uniformly distributed within soil
layers in both experiments, and that roots bypassed soil areas leaving water
without being extracted. Thus plants could show signs of water deficit despite an
appreciable amount of available water in the bulk soil.

A critical value constant, K, in a functional water uptake model
representing the daily fraction of extractable water was determined for each soil
layer to predict the water extraction based on the non-uniform root distribution.
Once these values were estimated, the functional model for root water uptake

was able to closely predict the water extraction in each soil layer.

to Janaina,

ACKNOWLEDGMENTS

I wish to express my gratitude and appreciation to:

Dr. Joe T. Ritchie for his guidance throughout my course and research.

The graduate committee:

Dr. A.J.M. Smucker, Dr. K.Poff, and Dr. J. Flore

Brian Graff and his crew at Michigan State University Research Farm.

Federal University of Lavras — Minas Gerais - Brazil

CAPES-Brazil

Nowlin Chair Group

My wife and my family

All friends from Argentina, Australia, Brazil, Greece, Hungary, India, Italy,

Lebanon, Jordan, Portugal and USA.

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................ vii
LIST OF FIGURES ............................................................................................. viii
LIST OF ABBREVIATIONS .................................................................................. x
Introduction .......................................................................................................... .1

References ...................................................................................................... 13

Chapter 1. Spatial variability of soil texture, root distribution and water content in
different soils. l: Toward micro-scale description.

Introduction ..................................................................................................... 19
Material and Methods ...................................................................................... 22
Results and Discussion ................................................................................... 30
Summary and Conclusions ............................................................................. 42
References ...................................................................................................... 43

Chapter 2. Spatial variability of soil texture, root distribution and water content in
different soils. II: A geostatistical approach to describe micro-scale spatial

variability.
Introduction ..................................................................................................... 64
Material and Methods ...................................................................................... 66
Results and Discussion ................................................................................... 68
Summary and Conclusions ............................................................................. 77
References ...................................................................................................... 78

Chapter 3. Root water extraction by com plants under conditions of limited water

supply.
Introduction ..................................................................................................... 93
Material and Methods ...................................................................................... 95
Results and Discussions ................................................................................. 98
Summary and Conclusions ........................................................................... 102
References .................................................................................... 103
General conclusions ........................................................................... 110

vi

LIST OF TABLES

Table 1.1. Soil texture analysis, pH and organic matter content (%) within soil
depths for two locations ...................................................................................... 58

Table 1.2. Descriptive statistics. Mean and standard deviation (SD) for some soil
properties at different soil depths. Greenhouse experiment at Michigan State
University, Ml ...................................................................................................... 59

Table 1.3. Descriptive statistics. Mean and standard deviation (SD) for some soil
properties at different soil depths. Field experiment at Maricopa, AZ. ................ 61

Table 1.4. Statistical correlation for volumetric water content (VWC), lower limit
(LL), available water content (AWC), root length density (RLD) and clay within
soil depths at two different locations ................................................................... 62

Table 1.5. Root length (RL) for five width classes, total root length (TRL), root
surface area (RSA) and root volume (RV) within soil depths. Data for corn in
Michigan and barley in Arizona. .......................................................................... 63

Table 2.1. Descriptive statistics for root length density (RLD), volumetric water
content (VWC), available water content (AWC), clay, sand, silt and lower limit
(LL) within soil depths. Greenhouse experiment (C1) at Michigan State
University, Ml ...................................................................................................... 81

Table 2.2. Descriptive statistics for root length density (RLD), volumetric water
content (VWC), available water content (AWC), clay, sand, silt and lower limit
(LL) within soil depths. Greenhouse experiment (CZ) at Michigan State
University, Ml ...................................................................................................... 83

Table 2.3. Descriptive statistics for volumetric water content (VWC), available
water content (AWC), root length density (RLD), clay, sand, silt and lower limit
(LL) within soil depths. Field experiment at Maricopa Agric. Center, AZ ............. 84

Table 2.4. Results for geostatistical analysis in the greenhouse experiment at
Michigan State University and field experiment at Arizona. Available water
content (AWC) and root length density (RLD) within soil layers .......................... 86

Table 3.1. K values for different soil layers in C1 and C2. Greenhouse
experiment at Michigan State University. .......................................................... 105

Table 3.2. Day of the year (DOY) for water extraction within soil layers for C1 and
CZ. Greenhouse experiment at Michigan State University, Ml .......................... 105

vii

LIST OF FIGURES

Figure 1.1. Grid sampling used in the greenhouse experiment (a) and in the field
experiment (b). .................................................................................................... 47

Figure 1.2. Daily maximum and minimum air temperature in the greenhouse at
Michigan State University, Ml. ............................................................................ 48

Figure 1.3. Air temperature (°C), solar radiation (MJ m'z) and rainfall (mm) at
Maricopa Agricultural Center, AZ. ....................................................................... 49

Figure 1.4. Plant height (cm) for three maize plants, C1(a) and 02(b).
Greenhouse experiment at Michigan State University, Ml .................................. 50

Figure 1.5. Cumulative leaf expansion from clay of the year (DOY) 13 to the
termination of the experiment at greenhouse. Data for C1 (a) and CZ (b) .......... 51

Figure 1.6. Daily leaf rolling (fraction of blade exposed) for different leaves and
cylinders, from day of year (DOY) 35 and 40, to the end of the experiment at
Michigan State University, Ml ............................................................................. 52

Figure 1.7. Daily evapotranspiration (mm day“), for C1 and C2, from day of year
(DOY) 10 to 90. Greenhouse experiment at Michigan State experiment, Ml ...... 52

Figure 1.8. Day of year (DOY) and volumetric water content within soil depths
(cm) for C1 (a) and C2 (b) in the greenhouse experiment at Michigan State
University, Ml. ..................................................................................................... 53

Figure 1.9. Volumetric water content (cm3cm'3) at different soil depths (cm) for
day of year (DOY) 15, 80 and at LL (C1) ............................................................ 54

Figure 1.10. Root length density (cm cm'3) (RLD) as function of available water
content (%)(AWC) for two different experiments. Greenhouse experiment (a)
(C1) at Michigan State University, MI; and field experiment (b) at Arizona. ........ 55

Figure 1.11. Volumetric water content (VWC) as function of clay content (%) at
30 cm, 45 cm and combined 30+45 cm depth in the greenhouse experiment, and

at 45 cm depth in the field experiment ................................................................ 56
Figure 1.12. Day of year (DOY) and volumetric water content within soil depths

(cm) at Maricopa Agricultural center, AZ ............................................................. 57
Figure 2.1. Hypothetical semivariogram .............................................................. 87

viii

Figure 2.2. Root length density (RLD)(cm cm'3) and available water content
(AWC) within soil depths (C1). Greenhouse experiment at Michigan State
University, Ml ...................................................................................................... 88

Figure 2.3. Root length density (RLD)(cm cm'3) and available water content
(AWC) within soil depths (C2). Greenhouse experiment (C2) at Michigan State
University, MI. ..................................................................................................... 90

Figure 2.4. Root length density (RLD) )(cm cm'3) and available water content
(AWC) within soil depths. Field experiment at Maricopa, AZ. ............................. 91

Figure 3.1 Volumetric water content (VWC)(cm3cm'3) within soil depths (cm) as
function of day of year (DOY). Data for C1 in the greenhouse experiment at
Michigan State University, MI ........................................................................... 106

Figure 3.2. Volumetric water content (VWC)(cm3cm'3) within soil depths (cm) as
function of day of year (DOY). Data for the greenhouse experiment (C2) at
Michigan State University, Ml ........................................................................... 107

Figure 3.3. Logarithm of the difference between AWC and LL as function of day
of year (DOY). The slopes of straight lines represent the K value for each soil
depth. Data for C1 and CZ at greenhouse experiment at Michigan State
University, Ml .................................................................................................... 108

Figure 3.4. Measured and simulated VWC (%) at different soil depths for C1 and
C2 in the greenhouse experiment at Michigan State University, Ml ............. 109

AWC

C1
C2
cm
DOY
EFV

ha

k9
LL

MAC
mm
MJm’Z
ml
MSU
RLD
RSA
RV

USDA

LIST OF ABBREVIATIONS

Available water content

Degree Celsius

Cylinder one

Cylinder two

Centimeters

Day of the year

Extraction front velocity (cm day")
Hectare

lntemal diameter

Kilograms

Lower limit

Meters

Maricopa Agricultural Center, AZ
Millimeters

Megajoules per square meter
Milliliters

Michigan State University, Ml
Root length density (cm cm'3)
Root surface area (cmz)

Root volume (cm3)

United States Department of Agriculture

Volumetric water content

Introduction

During growth and development of roots into the soil matrix, plant genetic
characteristics and soil environment govern the way roots penetrate and exploit
the soil around them. Genetic characteristics will determine root branching and
growth patterns, which are crop species dependent (Hamblin, 1985; Schieflbein
and Benfey, 1991; Zobel, 1996). Soil environment will affect not only the way
roots grow, but also other related processes such as water absorption, nutrient
transport and biological activities in the vicinity of root system (Kramer and
Boyer, 1995).

Soil environment has a major influence on root growth, and is reflected by
several types of stresses that roots encounter. These stresses are classified as
chemical, biological and physical (Russel, 1977; Jones et al., 1991; Kramer and
Boyer, 1995). Chemical stresses are usually attributed to toxic substances,
shortage of nutrients or unbalanced distribution of nutrients in the soil
(Schiefelbein and Benfey, 1991). Biological stresses induced by flora and fauna
of soil, e.g., root diseases, may cause a decrease in water and nutrient
absorption by plants. Finally, soil physical stress will limit root growth as a
function of variations in soil texture, structure, lack or distribution of water and
aeration. In summary, the influence of physical, chemical and biological stresses
in the soil can result in an environment that causes non-uniform root distribution.

Uneven distribution of root growth may vary in such a way that roots may
be found clustered in the soil. Root clustering is caused not only by the soil

environment, but also by plant characteristics, i.e., root geometry and branching

(Tardieu, 1988), and assimilate partitioning between root and shoot (Tardieu,
1989). In addition, competition and chemical interaction among roots will also
affect the way roots colonize the soil (Russel, 1977).

Researchers have also observed other causes for root clustering. It has
been observed that roots grow clustered inside bio-pores (wormholes, old root
channels), soil cracks, around peds and planes of weakness in the soil
(Passioura, 1983; Whiteley and Dexter, 1983; Hewitt and Dexter, 1984; Tardieu
and Manichon, 1987; Tardieu, 1989; Smucker and Aiken, 1992).

Plants with clustered roots growing in limited supply of water content, will
die or become dormant despite water availability in soil (Lafolie et al., 1991;
Bruckler et al, 1991a). In this sense, water is left in soil as a result of an
incomplete water extraction (Passioura, 1983, 1991). The reasons are once roots
are clumped (e.g. inside natural openings) they dry the soil in the vicinity, which
will increase the resistance of soil to water movement to the root surfaces
(Passioura, 1983). For instance, soil hydraulic conductivity will decrease around
roots and water movement from wet regions to dry regions where roots are
clumped will be low or impaired. If no more water is added to recharge the soil
profile, water availability for root absorption will decrease (Tardieu et al., 1992).

Once roots are clumped, low contact of their surface with soil may occur.
For instance, roots growing inside a natural opening may present a discontinuity
of contact between their surface and the soil surface. This discontinuity will

decrease water absorption in that location, consequently decreasing the amount

of water a plant can extract from soil (Passioura, 1983;1991; Tardieu et al.,
1992).

Hence, clumping will cause roots to be spatially distributed in the soil and
which is of considerable importance in that it can influence the efficiency with
which the soil is exploited. This spatial distribution of roots and depth of the root
system will be important for water extraction by plants (Robertson et al., 1993a;
Dardanelli et al., 1997). It is probable that an incomplete exploitation of water and
nutrients will occur if roots are not occupying some soil regions. Under conditions
of soil water deficit, this incomplete exploitation may affect crop yield (Boyer,
1982). The responses of crops in such conditions will vary depending upon plant
phenological stage, duration and severity of the water deficit period (Asseng et
at,1998)

Analyses of the water uptake pattern by plants indicate that many crops
do not fully utilize water found at deep soil layers (Passioura, 1983). Several
authors verified an incomplete water extraction in the soil profile (Bland et al.,
1989; Robertson et al., 1993a; Passioura, 1983). The reasons for incomplete
extraction were attributed to a low root density at depth of water extraction
(Robertson et al.,1993 b), onset of crop maturity before deep water extraction
commences (Robertson et al. 1993b), uneven root distribution (Robertson et al.,
1993a), root clustering (Passioura, 1983), high root axial resistance to water flow
(Passioura, 1983; Hamblin and Tennant 1987), and roots growing towards lose
regions in the soil after they have been deflected by more dense regions or other

kind of obstacles (Dexter and Hewitt, 1978). Water simulation models do not

account for these patterns of root growth and the incomplete water extraction.
This may be a drawback (McIntyre et al., 1995), because it can lead to spurious
results of water extracted by plants. Considering the spatial distribution of roots in
the soil matrix as a natural fact and the phenomena of incomplete water
extraction, we can assume that roots naturally bypass some soil regions when
they are colonizing the soil. This subject needs to be better investigated and

incorporated into new models of soil water extraction in the future.

Water Models

Numerous soil water simulation models have been developed in the past
with enormous degrees of sophistication. One of the earliest attempts to simulate
soil water was the simple water balance of Thornthwaite (1948). This model used
average monthly precipitation, soil water storage and estimated potential
evapotranspiration.

A microscopic approach was presented later to describe water uptake by
root systems (Gardner, 1960; and Cowan, 1965). It considered roots as an
infinitely long cylinder of uniform radius and water-absorbing properties, and
water flowing in the radial direction towards the root.

In a general sense, water simulation models are classified as budget,
semi-dynamic and dynamic models (Jong and Bootsma, 1996). Single budget
models, treat the soil profile as a bucket into which water flows until it is full, and
the excess of water is considered as runoff or drainage (Aase et al., 1973). In

other more advanced models the soil is divided in layers and water flows in a

cascading approach from upper to lower zones. For instance, water from
precipitation or irrigation flows to lower zones into the soil profile as soon as the
upper layer reaches the field capacity (O’Leary et al., 1985). They are useful in
characterizing soil climate zones, soil water regimes in arid and semi-arid regions
and in irrigation scheduling. They require minimum amount of input data, such as
drainage upper and lower limit of water content, daily precipitation and potential
evapotranspiration (Jong and Bootsma, 1996).

In semi-dynamic models, water movement within the root zone is also
simulated by transferring water among layers with transfer rates based on the
water in each layer (Ritchie, 1985). These rates depend on the soil hydraulic-
properties. The amount of water exceeding the drained upper limit drains into the
next layer.

In dynamic models, physical factors that control water movement in the
soil control infiltration and redistribution of water (Ross and Bristow, 1990).
Dynamic models use Darcy’s law combined with the equation of continuity. The
resulting equation is commonly referred as Richard’s equation with an added
term for root water extraction (sink term). The mathematical description of this
sink term is hindered by the complex mechanisms involved in water extraction
allied to the heterogeneity of the root-soil system (Molz, 1981; Li et al, 1999).
Richard’s equation is a nonlinear partial differential equation which can be solved
analytically only for specific initial and boundary conditions. The approach
adopted here is known as macroscopic approach. It is a complex equation to be

used, but it has been incorporated not only into more comprehensive crop growth

models but also into soil degradation and solute transport models (Jong and
Boostma, 1990).

Several water simulation models developed in the past decades are based
on the relationship between roots and soil (Hector, 1993). Most water uptake
models use the amount of roots per unit volume of soil to calculate water
extraction. They assume that roots are uniformly distributed in each soil layer,
presenting an exponential decrease in the amount of roots with soil depth
(Dwyer, 1988; Klepper, 1991). Based on the complex interactions of the root-soil
environment, researchers have questioned the validity of using root length
density (RLD) (cm roots cm‘3 soil) in water extraction models (Hamblin and
Tenant, 1987; Bland and Dugas, 1989;Lafolie et al., 1991; Dardanelli et al.,
1997). Soil heterogeneity and the dynamic process of the root system as a living
organism makes difficult the prediction where roots are and the amount of water
being extracted.

Water absorption is not uniform along the root surface. The root surface
capable of water extraction moves toward different soil zones during the root
growth period. It has been observed that water absorption is higher in new roots
compared to old ones (Hamblin, 1985). Variation in water extraction may also
occur in the same root, consequently the resistance to water absorption along
the roots will be different (Moreshet et al., 1996).

The assumptions of uniform root distribution in simulation models are
often not realistic. To overcome this limitation and to better predict water uptake

by plants, water extraction should be based only on water depletion with time.

Due to the difficulties and errors in measuring the amount of roots involved in the
water absorption process, the quantification of the amount of roots present in the
soil is not necessary (McIntyre et al., 1995). A functional model describing root
water uptake that does not take into account root length density was proposed by
Ritchie et al (1999). This model uses generalizations from measured soil water
content changes in the soil profile to predict root water uptake during plant
development. The model assumes a minimum threshold of root density at each
soil layer to extract water; that transpiration is greater than total water root supply
and that values of lower limit water content are accurately determined. Above the
lower limit, the root water uptake rate will decline exponentially with soil water
content in the same way as described by Passioura (1983). This approach leads
to predictions closer to reality, eliminates the uncertainties in complex root

system behavior, and it accounts for soil spatial variability.

Spatial Variability of Soil and Roots

When describing soil properties, soil is usually assumed as a
homogeneous material Merrick, 1998). This is a convenient assumption when
performing soil tests in laboratory or in conducting field experiments in small
plots, for instance, with different fertilizer rates. However, the degree of
homogeneity will depend on the scale used in these kinds of studies and the
objectives of the researcher (Trangmar et al., 1986).

Soil, in reality, is a heterogeneous medium with physical, chemical and

biological properties varying in space and time (Trangmar et al., 1986; Warrick et

al, 1986). This heterogeneity has its origin in the natural process of soil-fon'nation
factors and in the anthropogenic action (Ahuja and Nielsen, 1990; Tsegaye and
Hill, 1998; a,b; Warrick, 1998).

The natural processes of soil formation will form a nested structure with
climate operating over large spaces and soil weathering operating in long time
periods. During the genesis of the parent material, some variation in chemical
composition may occur. This may be natural or due to the presence of different
gases or cooling period. Such variation associated to a local action of chemical,
physical and biological reactions during the weathering process will lead to a
formation of a variable substrate. Other processes, such as erosion and
deposition of the parent material may contribute to the soil spatial variability
operating more locally or more frequently, e.g., weather (Beckett and Webster,
1971; Trangmar et al., 1986).

Geomorphical processes, for instance, alluvial deposition; geochemical
gradients, loss, retention or burial of land surfaces will also contribute to the soil
variability in the landscape ( Beckett and Webster, 1971).

Anthropogenic action, in addition to the natural processes mentioned
above, will affect soil structure and cause variability in several ways, e.g., by
intensive tillage practices, localized application of fertilizers and irrigation (Ahuja
and Nielsen, 1990; Warrick, 1998).

Soil heterogeneity will affect both soil physical and hydraulic properties.
Several works were done in the last few years with the objective of describing

such spatial variability. Some of them show spatial variations of pH, soil texture,

bulk density, conductivity, ranging from 1 meter to hundreds of meters (Yost et
al., 1982; Warrick, 1986; Yeh et al., 1986). Others, show the spatial variability in
the distribution of hydraulic conductivity (Moutonnet et al., 1988; Zhang and
Berndtsson, 1991; Mallants et al., 1996), water tension ( Saddiq et al.,1985; Yeh
et al., 1986), spatial and temporal soil water storage capacity (Or and Hanks,
1992), soil water content (Moutonnet et al., 1988), soil water retention ( Shouse
et al., 1995).

While considering macro-scale variability, few studies dealt with micro-
scale variability of soil. Bruckler et al. (1991b) that showed the spatial variability
of water potential at small scale. In his work, isocontour lines of water potential
and clumped roots were described based on samples taken 0.4 m far from the
plant stem. Amato (1991), demonstrated that soil water content varied spatially
on soil peds layers at a distances around 0.02 m. This variation in water content
affected significantly the amount of roots around these peds.

Based on the degree of spatial variability in the soil, it is expected that
plants growing in such conditions are variable too (Brown and Scott, 1984;
Saddiq et al., 1985). Considering the relationship between soil and roots, one of
the first effects of such variation will probably occur in root growth. We already
saw that several types of soil stress may affect root growth. So, considering the
natural and/or man-made processes involved in soil spatial variability, these
stresses probably will not be uniformly distributed. This means that during the
period of root growth in the soil, roots will experience all sorts of different

environments in a small area. Without doubt, this may cause an uneven

distribution of roots along the soil profile or clumping of roots in some soil
regions. Roots tend to preferentially colonize lose zones because they are
deflected at the interface between loose and strong zones (Dexter and Hewitt,
1978). Consequently, the soil may present patches of high root density, and
others with only a few roots (Tardieu, 1994). Shrinkage cracks or spaces
between clods resulted in zones for root clumping (Tardieu and Manichon, 1987;
Tardieu, 1989, Passioura, 1983). Moreover, soil biopores may also contribute to
root clumping and consequently to the spatial variability of the root system
(Passioura, 1983; Whiteley and Dexter, 1983).

However, even considering the heterogeneity of soil physical and
chemical properties as the main cause of root spatial distribution in the soil
(Tardieu, 1994), other factors are also involved, such as plant characteristics
(Hamblin, 1985), competition among plants (Russel, 1977), partitioning of
assimilates within the plant (Smucker, 1984; Tardieu, 1989) and possible
interactions of all these factors with the soil environment.

Understanding the spatial variability of field soil is important for crop
production (Saddiq et al., 1985), but more important than that is to describe this
variability. Soil spatial variability will largely depend on the scale or frequency of
observation. This means that depending on the level of detail needed,
heterogeneity that may have been previously defined as random may be found to
contain a systematic component (Trangmar et al., 1986). Thus, considering the
scale used, soil spatial variability may occur in a few meters to several meters

(Warrick et al, 1986).

10

Researchers have demonstrated that spatial variability of roots in the
plane x-y has a high coefficient of variation, but no efforts were attempted to
explain this variability (Pettygrove et al. 1989). Researchers only explained the
variability existing in the vertical dimension. A possible reason for this may be
related to the statistical approach used in such studies (Trangmar et al.,1986).

Most studies have tried to explain the variability of soil properties by using
classical statistical methods, such as the analysis of variance (Saddiq et al.,
1985; Robertson and Gross, 1994). A few other studies tried to explain the soil
and root arrangement, by using indices to describe deviations from an expected
distribution (Grieg-Smith, 1983,). Spatial autocorrelation (Tardieu and Manichon,
1986; Tardieu, 1988) and the square root of number of roots per sampling unit
(Pettygrove et al. 1989) were also used to describe root arrangement in the soil.
Nonetheless, these methods do not provide a complete description of the
variability of soil properties or root distribution, because the variance does not
take into account the distance between observations (Saddiq et al., 1985). Pure
statistical treatment of the heterogeneity ignores the existence of spatial
correlation (Mallants et al., 1996).

Considering that there is some degree of dependence among
observations taken in a grid pattern, geostatistic analysis may be an adequate
tool to describe spatial variability (Matheron, 1973; Journel and Huijbregts, 1978;
lsaaks and Srivastava, 1989; Goovaerts, 1992, 1994, 1997). Geostatistics can
calculate the correlation between observations made at different neighboring

locations and the observed correlation structures can then be used to estimate

11

values in unsampled locations by means of Kriging or co-kriging techniques
(Isaaks and Srivastava, 1989; Goovaerts, 1992, 1997).

Geostatistic has already been used to study spatial variability of physical
and hydraulic soil properties in large scale. But, the possibility to use this
technique in studies of micro-scale variability of the soil environment associated
with roots, open a new perspective to understand the complex relationship
between soil and root in the process of water extraction.

The difficulty involved in determining the amount of water extracted by
plants in crop models will be a matter of discussion and study for a long time.
New studies must be proposed to fully understand the spatial variability of the
soil environment and the factors that govern root growth and development in
such environment.

The study proposed in this dissertation does not have the intention of
covering all topics mentioned here, but to contribute with a small parcel of
knowledge to improve future soil-water extraction models.

To provide such a contribution this study was divided into three chapters.
The first chapter has as objectives the characterization of distribution of roots and
soil properties at different soil depths. The main objective of the second chapter
is to study the spatial distribution of roots and soil properties in different soil
depths by using geostatistical analysis. Studies described in these chapters are
an attempt to understand and demonstrate that roots bypass soil regions during
the growing period. The objective of the third chapter is to characterize water

extraction by roots if the bypassing of soil regions is likely to occur.

12

References

Aase, J.K., J.R. Vlfight, and TH. Siddoway. 1973. Estimating soil water content on
native rangeland. Agric. Meteorol. 12:185-191.

Ahuja, LR, and DR. Nielsen. 1990. Field Soil-water relations. p. 143-190. In B.A.
Steward and DR. Nielsen (ed) Irrigation of Agricultural crops. Agron.
Monogr. 30. ASA, CSSA and SSSA, Madison, WI.

Amato, M. 1991. Spatial distribution and water uptake of roots in structured
growth media. Ph.D. diss. Michigan State Univ., East Lansing.

Asseng, S., J.T. Ritchie, A.J.M. Smucker, and M.J. Robertson. 1998. Root growth
and water uptake during water deficit and recovering in wheat. Plant Soil
201:265-273.

Beckett, P.H.T., and R. Webster. 1971. Soil Variability: a Review. Soils Fert.
3421-42.

Bland, W.L., and W. Dugas. 1989. Cotton root growth and soil water extraction.
Soil. Sci. Soc. Am. J. 53:1850-1855.

Boyer, JS 1982. Plant productivity and environment. Science. 218:443—338.

Brown, D., and HD. Scott, 1984. Dependence of crop growth and yield on root
development and activity. p.101-136. In Barber et al. (ed.) Roots, Nutrient
and Water Influx, and Plant Growth. ASA Spec. Publ. 49. SSSA,
Madison, WI.

Bruckler, L., F. Lafolie, and F. Tardieu. 1991a. Modeling root water and soil-root
water transport: ll. Field Comparisons. Soil Sci. Soc. Am. J. 55:1213—
1220.

Bruckler, L., F. Lafolie, and F. Tardieu. 1991b. Modeling the root water potential
and water extraction in the two-dimensional case. ll Field comparisons.
Soil Sci. Soc. Am. J. 55:1213-1220.

Cowan, LR. 1965. Transport of water in the soil-plant—atmosphere system. J. Appl.
Ecol. 2:221-239.

Dardanelli, J.L., O.A. Bachmeier, R. Sereno, and R. Gil. 1997. Rooting depth and

soil water extraction patterns of different crops in a silty loam haplustoll.
Field Crop Res. 54:29-38.

13

 

 

 

 

Dexter, AR, and JS Hewitt. 1978. The deflection of plant roots. J. Agric. Eng.
Res. 23:17-22.

Dwyer, L.M., D.W. Stewart, and D. Balchin. 1988. Rooting characteristics of corn,
soybeans and barley as a function of available water and soil physical
characteristics. Can. J. Soil Sci. 68:121-132.

Gardner, W.R. 1960. Dynamics aspects of water availability to plants. Soil Sci.
89(2):63-73.

Goovearts, P. 1992. Factorial kriging analysis: a useful tool for exploring the
structure of multivariate spatial soil information. J. Soil Sci. 43:597-619.

Goovaerts, P. 1994. Study of spatial relationships between two sets of variables
using multivariate geostatistics. Geoderma 62:93—107.

Goovearts, P. 1997. Geostatistics for natural resources evaluation. Oxford
University Press, New York, NY.

Grieg-Smith, P. 1983. Quantitative plant ecology. Univ. of Calif. Press. Berkeley,
CA.

Hamblin, AP. 1985. The influence of soil structure on water movement, crop root
growth, and water uptake. Adv. Agron. 38295-158.

Hamblin, A. and D. Tennant. 1987. Root length density and water uptake in
cereals and grain legumes: How well are they correlated? Aust. J. Agric.
Res. 38:513-527.

Hector, D.J., K. Gregson, and M. McGowan. 1993. A computer simulation
describing water extraction by crop root systems. Field Crop Res. 32:287-
304.

Hewitt, J.S., and AR. Dexter. 1984. The behavior of roots encountering cracks in
soil. "Development of a predictive model. Plant Soil 79:11-28.

lsaaks, E.H., and RM. Srivastava. 1989. An Introduction to applied geostatistics.
Oxford University Press, New York, NY.

Jones, C.A., W.L. Bland, J.T. Ritchie, and JR. Williams. 1991. Simulation of root
growth. p.91-120. In John Hanks and J.T. Ritchie (ed.) Modeling Plant
and Soil Systems. Agron. Monogr. 31. ASA, CSSA and SSSA, Madison,
WI.

Jong. R., and A. Bootsma. 1996. Review of recent developments in soil water
simulation models. Can. J. Soil. Sci. 76:263-273.

14

Journel, AG, and C. Huijbregts. 1978. Mining Geostatistics. Academic Press,
New York, NY.

Klepper, B. 1991. Crop root system response to irrigation. lrrig. Sci. 12:105—108.

Kramer, P.J., and J.S. Boyer. 1995. Water relation of plants and soils. Academic
Press. San Diego, CA.

Lafolie, F., L. Bruckler, and F. Tardieu. 1991. Modeling root water potential and
soil-root water transport: l. Model presentation. Soil Sci. Soc. Am. J.
55:1203—1212.

Li, K.Y., J.B. Boisvert, and R. De Jong.1999. An exponential root-water-uptake
model. Can. J. Soil. Sci. 79:333-343.

Mallants, 0., BF. Mohanty, D.Jacques, and J. Feyen. 1996. Spatial variability of
hydraulic properties in a multi-layered soil profile. Soil Sci. 161(3):167-
181.

Matheron, G.1973. The intrinsic random functions and their
applications.Adv.Appl.Prob.5:239-465.

McIntyre B.D., S.J. Riha, and DJ. Flower. 1995. Water uptake by pearl millet in a
semiarid environment. Field Crop Res. 43:67-76.

Molz, F.J. 1981. Models of water transport in soil-plant-system: A review. Water
Resourc. Res. 17(5):1245-1260.

Moreshet, S., B.Huang, and m. Huck. 1996. Water permeability of roots. p. 659—
678. In Yoav Waisel et al. (ed.) 2'“1| ed. Plant Roots: The Hidden Half.
Marcel Dekker, New York, NY.

Moutonnet, P., E. Pluyette, N.L.Mourabit, and P. Couchat. 1988. Measuring the
spatial variability of soil hydraulic conductivity using an automatic neutron
moisture gauge. Soil Sci. Soc. Am. J. 52:1521-1526.

O’Leary, G.J., D.J. Connor, and DH. White. 1985. A simulation model of the
development, growth and yield of the wheat crop. Agric. Syst. 17:1-26.

Or, 0., and R.J. Hanks. 1992. Spatial and temporal soil water estimation
considering soil variability and evapotranspiration uncertainty. Water
Resourc. Res. 28(3):803—814.

Passioura, J.B. 1983. Roots and drought resistance. Agric. Water. Management
7:265—280.

15

Passioura, J.B. 1991. Soil Structure and Plant Growth. Aust. J. Soil Res. 29:717-
28.

Pettygrove, GS, MP. Russelle, P. Becken, and P. Burford. 1989. Number,
surface area and spatial distribution of corn roots in minirhizothron
images. p. 250. In W.E. Larson et al. (ed.) Mechanics and related
processes in structured agricultural soils. Kluwier. London, England.

Ritchie, J.T. 1985 A user-oriented model of soil water balance in wheat. p. 292-
305. In W.Day and R.K. Alkin (ed) Wheat Growth and modeling. Plenum
Publishing, New York, NY.

Ritchie, J.T., J.L. Dardanelli, M. Calmon, and J.M. Andriani. 1999. A functional
model for root water uptake. Agron. J. (submitted).

Robertson, GP, and K. Gross.1994. Assessing the heterogeneity of below
ground resources: Quantifying pattern and scale. p. 237-253. In
M.M.Caldwell (ed.) Exploitation of Environmental heterogeneity by plants.
Academic Press, New York, NY.

Robertson, M.J., S. Fukay, M.M. Ludlow, and G.L. Hammer. 1993a. Water
extraction by grain sorghum in a sub-humid environment. I. Analysis of
the water extraction pattern. Field Crop Res. 33:81-97.

Robertson, M.J., S. Fukai, M.M. Ludlow, and G.L. Hammer. 1993b. Water
extraction by grain sorghum in a sub-humid environment. ll. Extraction in
relation to root growth. Field Crop Res. 33299-112.

Ross, P.J., and KL. Bristow. 1990. Simulating water movement in layered and
gradation soils using the Kirchoff transformation. Soil Sci. Soc. Am. J.
54:1519-1524.

Russel, RS. 1977. Plant root systems: Their function and interaction with the soil.
McGraw-Hill. England.

Saddiq, M.H., P.J. Wlerenga, J.M. Hendrickx, and MY. Hussain. 1985. Spatial
variability of soil water tension in an irrigated soil. Soil Sci. 140(2):126-
132.

Schiefelbein, J.W., and RN. Benfey. 1991. The development of plant roots: New
approaches to underground problems. The Plant Cell 321147-1154.

Shouse, P.J., W.B. Russell, D.S. Burden, H.M. Selim, J.B. Sisson, and M. van

Genuchten. 1995. Spatial variability of soil water retention functions in silt
loam soil. Soil Sci. 159(1): 1-12.

16

Smucker, A.J.M. 1984. Carbon utilization and losses by plant root systems. In
S.A.Barber and RD. Boulden (eds) Roots, nutrient and water influx, and
plant growth. Am. Soc. Agron. Madison, WI.

Smucker, A.J.M., and RM. Aiken. 1992. Dynamic root responses to water deficits.
Soil Sci. 154(4): 281-289.

Tardieu, F. 1988. Analysis of the spatial variability of maize root density. Il.
Distance between roots. Plant Soil 107:267-272.

Tardieu, F. 1989. Root system response to soil structural properties: micro and
macro scale. p. 153-172. In W.E. Larson et al. (ed.) Mechanics and
related processes in structured agricultural soils. Kluwier. London,
England.

Tardieu, F. 1994. Growth and functioning of roots and root systems subjected to
soil compaction. Towards a system with multiple signaling? Soil Tillage
Res. 30:217-243.

Tardieu, F., and H. Manichon, 1986. Characterisation en tant que capteur d’eau
de I’enracinement du mais en parcelle cultivee. ll. Une methode d’etude
de la repartition verticale et horizontale des racines. Agronomie 6(5): 415-
425.

Tardieu, F., and H. Manichon. 1987. Etat structural, enracinement et alimetation
hydrique du mais. I. Modelisation d’etat structuraux de la couche
labouree. Agronomie 7(2): 123-131.

Tardieu, F., L. Bruckler, and F. Lafolie. 1992. Root clumping may affect the root
water potential and the resistance to soil-water transport. Plant Soil
140:291-301.

Thornthwaite, CW. 1948. An approach toward a rational classification of climate.
Geog. Rev. 38:55-94.

Trangmar, B.B., R.S. Yost, and G. Uehara. 1986. Spatial dependence and
interpolation of soil properties in West Sumatra, Indonesia; l.Anisotropic
variation. Soil Sci. Soc. Am. J. 50:1391-1395.

Tsegaye, T. and R.L.Hill. 1998a. Intensive tillage effects on spatial variability of
soil physical properties. Soil Sci. 1 63 (20): 143-154.

Tsegaye, T., and R.L. Hill. 1998b. Intensive tillage effects on spatial variability of

soil test, plant growth, and nutrient uptake measurements. Soil Sci.
163(2): 155-165.

17

Warrick, AW. 1998. Spatial Variability. p.655-675. In D.Hillel (ed) Environmental
Soil Physics.

Warrick, A.W., D.E. Myers, and DR. Nielsen. 1986. Geostatistical methods
applied to soil science.p.53-82. In Klute(ed.). Methods of soil analysis.
Part 1. 2"d ed. Agron. Monogr.9. ASA and SSSA, Madison, WI.

Whiteley, GM, and AR. Dexter. 1983. Behavior of roots in cracks between soil
peds. Plant Soil 74:153-162.

Yeh, T.C.J., L.W. Gelhar, and P.J. Wlerenga. 1986. Observations of spatial
variability of soil-water pressure in a field soil. Soil Sci. 142(1): 7-12.

Yost, R.S., G. Uehara. and R.L. Fox. 1982. Geostatistical analysis of soil chemical
properties of large land areas. I. Semi-variograms. Soil Sci. Soc. Am. J.
46: 1 028-1 032.

Zhang, T., and R. Berndtsson. 1991. Analysis of soil water dynamics in time and
space by use of pattern recognition. Water Resourc. Res. 27(7): 1623-
1636.

Zobel, R.W. 1996. Genetic control of root systems. p. 21-30. In Yoav Waisel et al.
(ed.) 2"d ed. Plant Roots: The Hidden Half. Marcel Dekker, New York,
NY.

18

Chapter 1

Spatial variability of soil texture, root distribution and water content in
different soils. I: Toward micro-scale description.

Introduction

Present water extraction models assume that roots are uniformly
distributed in the soil or that root density decreases exponentially with soil depth
(Molz, 1981; Dwyer, 1988; Klepper, 1991). Such assumptions simplify not only
how roots are distributed in the soil but they also underestimate quantities of
water extracted by plants.

Uniform distribution of roots in the soil does not represent reality
(Hamblin and Tenant, 1987; McIntyre et al. 1995). The soil environment is a
heterogeneous medium for root growth and its properties may vary in space and
time (Trangmar et al., 1986; Ahuja and Nielsen, 1990; Warrick, 1998).
Consequently, the restrictions imposed on root growth in such a medium may
also vary by the same order of magnitude and will affect the way roots colonize
and explore the soil. Root growth is determined by several intrinsic factors
related to the plants. Genetic characteristics will govern root branching patterns
and morphogenesis, while competition and chemical interaction among roots will
play a role in its distribution in the soil (Hamblin, 1985; Zobel, 1996). The
distribution becomes more complex when one considers the interaction between
soil and roots. As a consequence of these interactions, roots should be

considered more uneven than uniformly distributed in the soil (Tardieu, 1988).

19

Uneven root distribution will affect water extraction. Roots are found
clustered in the soil due to their growth into natural biopores, planes of
weakness, around soil peds, or because roots were deflected towards regions of
the soil with a more loose structure (Passioura, 1983; Whiteley and Dexter, 1983;
Hewitt and Dexter, 1984; Tardieu and Manichon, 1987; Tardieu, 1989; Amato,
1991; Smucker and Aiken, 1992). Root clustering implies in high concentration
of roots occupying and exploring a small volume of soil. If roots are clustered in
some parts of the soil it is expected that they will explore these regions as much
as possible; thus, an incomplete exploration of other soil regions is more likely to
occur

Incomplete soil exploration may be explained assuming that, once
the soil region around the clustered roots was fully explored and all available
water was extracted it is quite probable that a dry region will develop around
these clustered roots. The hydraulic conductivity will decrease in this part of the
soil and the movement of water from wet regions to this dry region will be
impaired. If the soil profile is not replenished, water extraction by roots will cease
and plants may die or become dormant from a lack of water, even when a supply
of water is still available in regions near by. This explanation is acceptable for
roots growing around peds, planes of weakness or roots deflected towards soil
regions with a loose structure. In case of roots clustered inside biopores, the
same explanation could be used; however, in this situation the possibility of lack
of contact or irregular contact between roots and the internal surface wall of the

biopore will cause some gaps to occur and has to be taken into account. Also,

20

some mechanical impedance to root penetration in the Internal pore wall may
occur as a function of soil dryness around the roots.

Considering the effects of root clustering in water absorption, it is possible
that during root growth some soil layers or some soil regions will be bypassed,
leaving regions with water without being extracted (Bruckler et al. 1991; Lafolie et
al. 1991). The consequences of that for root water uptake simulation models are
that a bias or an overestimation may occur in the calculation of the total amount
of water extracted by crops. This may happen because such models do not take
into account the non-uniform distribution of roots in the soil. This is not an easy
task, considering all the uncertainties involved in the root-soil relationship and the
natural spatial variability of the soil. Nevertheless, due to the importance of root
distribution in water simulation models for predicting the amount of water
extracted by crops, an effort in understanding such relationship is necessary. It
must be understood why roots bypass some regions in the soil in detriment to
others. Most of the possible causes mentioned in the literature are speculations
and there are no experimental evidences for such phenomenon.

The objective of the work described here was to characterize, in micro-
scale, root, soil texture, and soil water content distributions within soil layers for
two different soils as a first step to understand the reasons why roots are

bypassing some sections in the soil.

21

Material and Methods
Greenhouse Experiment - Michigan

This experiment was done at Michigan State Research Farm, MSU, East
Lansing. The soil is classified according to USDA, Soil Conservation Service, as
Metea B (Loamy, mixed, mesic Arenic Hapludalfs). Metea series consists of deep
and well drained soils that are rapidly permeable in the upper part of the solum
and moderately or moderately slowly permeable in the lower part of the solum
and in the underlying material. Some results of the soil analysis are presented in
Table 1.1. In October of 1997 two steel cylinders with dimensions of 60 cm
diameter, 180 cm high and 0.6 cm of wall thickness were pushed into soil to
obtain undisturbed soil samples.

The cylinders with contained soil were removed and placed in greenhouse
at Pesticide Research Center, Michigan State University, East Lansing, and
covered with plastic to maintain their initial moisture contents. Three corn
seedlings (variety Cargill 333), with 4 leaves were transplanted on DOY 336 in
1997 in C1 and 20 days later in C2. Plants in each cylinder were fertilized with a
NPK formula (19—19-19) at rate of 500 kg he“1 at transplanting day and side
fertilized with nitrogen (100 kg ha“) 20 days later. After transplanting, plants were
watered daily until DOY15 in 1998, and water was withheld afterwards.

Daily minimum and maximum temperatures were obtained from a
minimum data set recorder (Model Ll-1200, Li-cor, Lincoln, NE) located inside
the greenhouse. Temperature data from DOY 23 in 1998 to DOY 91 in 1998 are

shown in Figure 1.2.

22

Field Experiment - Maricopa Agricultural Center, Arizona

A field experiment was conducted at Maricopa Agricultural Center (MAC),
University of Arizona research and demonstration farm (elevation 358 meters)
Soil at MAC is classified as Casa Grande series (fine-loamy, mixed, hyperthermic
Typic Natrargids). This soil is a deep, well-drained slowly permeable soil formed
in old alluvian. The soil typically has brown reddish to brown sandy loam or
sandy clay loam surface horizon from 0-30 cm deep. The subsoil horizon from 30
to 60 cm is usually a reddish brown sandy clay loam, which increases in calcium
carbonate content with depth. Below this horizon at depth of 60 to 100 cm is a
horizon enriched with calcium carbonate (calcic horizon), which also has a sandy
clay loam texture. The depth to the calcic horizon varies from 25 to 100 cm in
depth, but commonly occurs between 50 — 80 cm in depth (Post et al., 1988).
Some selected soil properties are shown in Table 1.1.

An area with approximately 450 square meters was chosen as a site of
experimental plot in March 1999. This area was disced and planted on DOY 22
with barley. Fertilizers were applied on DOY 39 in the water at a dose of 47 kg of
N/ha. Another dose of nitrogen, 93 liters ha‘1 of UAN3Z was applied on DOY 69.
At planting date the field was irrigated, by siphon system, with approximately 20
cm of water. The next irrigation dates were DOY 23 and 67, with 15 cm of water
in both occasions. The irrigation was withheld on DOY 67 in the chosen area,
and plants were grown on stored water. This area was chosen due to the non-

unifonnity among plants when compared to the neighbor areas.

23

Weather data collected at MAC farm is presented on Figure 1.3.

Plant measurements

Greenhouse experiment

After transplanting, two plants were randomly chosen and marked for
nondestructive measurements throughout the growing period in each cylinder.
Growing period here is referred as the period after irrigation has been withheld to
the end of the experiment (DOY80). General plant aspects such as leaf rolling,
leaf growth and dormancy aspect were considered as indicators of severe water
deficit. Evapotranspitation for both cylinders were calculated based on water
changes in the soil.

Leaf expansion and plant height measurements were taken every two
days. For leaf rolling characterization, leaf width measurements were taken at
middle portion of leaf blade and at ‘A far from the leaf tip in a fully developed leaf
(Carlesso, 1993). The area of each leaf was determined from measurements of
leaf length and maximum leaf width multiplied by 0.75 (Stickler et al., 1961).
Plant height was taken from the soil surface up to the last visible ligule. Leaf
expansion was considered from the period when the leaf emerged from the whorl
until the leaf ligule appeared.

No plant measurements were made at MAC farm.

24

Water measurements

Greenhouse Experiment

Soil water content was measured by neutron scattering technique (CPN,
Model 503 DR). One access tube (aluminum, 50mm i.d.) was seated in the
center of each cylinder. Readings were made every two days at 15 cm.
increments to a depth of 1.20 m. Reading period was from the transplanting date

on DOY 336 in 1997 until DOY 81 in 1998.

Field experiment

In the experimental plot, water content was also measured by using the
same technique described in greenhouse experiment. Three access tubes
(aluminum, 50 mm id.) were installed in the field in positions were barley plants
were showing different degrees of development. Readings were taken on daily

basis from DOY 91 until DOY 103 at 15cm increments to a depth of 1.25 m.

Soil Sampling Procedures

Greenhouse experiment

Following the termination, when com plants presented severe signs of
water deficit and soil water contents approached LL, both cylinders were sliced
into 15 cm layers and each layer was analyzed by following two distinct
procedures. One procedure was the reading of volumetric water content by using
TDR technique (Topp, 1980, 1982) and the other one was the soil sampling per

SB.

25

A small TDR probe with 2 dieletric rods 2.1 cm long and 1.4 cm apart were
developed to determine VWC in the soil at small scale (Amato and Ritchie,
1995). This probe was calibrated by using two different soils and the equation
was used to transform the dielectric constant values observed in VWC. This TDR
probe was used with a Tektronix 1503B cable tester (Tektronix, Beaverton, OR),
and a computer connected to this cable tester read and stored all the readings
automatically. A total of 138 readings were made in a circular grid pattern with
each reading position being at 2.5 cm apart (Figure 1.1). A special template was
used to direct the TDR readings and also to keep all the samples in the same
position and alignment in all layers. In this way a 3 dimensional grid was
obtained, allowing further investigation of spatial distribution of the VWC. After
the dielectric constant readings were completed, soil samples were collected with
a soil sampler. (2.54 cm height x 2.54 cm i.d.). Each soil sample was taken in the
same position where dielectric constant readings were made. Each sample was
packed in whirlpak plastic bag and then stored in a freezer at — 20 °C for further

analysis.

Field Experiment

At MAC, a trench with approximately 4 m2 and about 2 m deep was
excavated in order to expose the soil profile to be sampled. This trench was
prepared where barley plants were showing severe signs of water deficit. The
procedure and equipment used for the dielectric constant readings were the

same as those adopted for the greenhouse experiment. The main differences

26

between both experiments are related to the amount of soil samples taken and
the positioning of sampling. In the sampling area, 100 soil samples were taken in
each soil layer on a square grid pattern with sampling positions located at 2.5 cm
intervals (Figure 1.1). In total, 550 soil samples were collected in 6 soil layers at
15 cm increment in depth from 45 cm up to 125 cm. Due to the dry conditions of
the soil at the sampling date, it was impossible to sample the first 300m in the
profile. Because of the high soil resistance at 45 cm depth, only 50 samples were
collected following the grid sampling.

In the field, each soil sample was packed in whirlpak plastic bag and
temporally stored in a container with ice; and at end of the day soil samples
where immediately frozen. Soil samples were transported frozen to MSU and

stored at —20 °C for further analyses.

Soil Texture Analysis

Soil texture analysis was performed by using the micro-pipette method
(Miller and Miller, 1987; Burt et al., 1993) and the textural soil classification
proposed by the USDA was followed (Gee and Bauder, 1986). This method was
chosen considering not only the small size of each sample and large number of
soil samples to be analyzed, but also the facility and fast procedure analyses in
comparison to the traditional hydrometer and pipette methods (Miller and Miller,
1987; Burt et al., 1993). The micro-pipette method was used for each soil sample
right after each sample was thawed. During the weighting process and washing

of sand fraction, each sample was carefully examined for the presence of roots.

27

This was done in order to recover any possible piece of root and to avoid
mistakes during the root analysis. This technique was used for all soil samples

collected in both experiments.

Root analysis

Roots were washed from the soil samples by successive wet sieving on
sieve 35 (50 um aperture) and sieve 60 (250 um aperture). After this procedure
roots were put in a methanol solution 15% (v/v) plus 5 ml of Malachite green
oxalate solution 1% (WM and preserved at 4 °C ( Smucker, 1990).

Stained roots were rinsed with distilled water on an ultra fine (25 um)
nylon screen and uniformly distributed on a small clear plastic tray where the
roots were covered with a thin film of water for image video recording. Once roots
were carefully spread in the tray to avoid overlapping, images were video
recorded by a computer controlled robotic camera on VHS videotape (Smucker,
1990). Video recorded images were processed by a Sun Ultra-based computer
algorithm in the Root Image Processing Laboratory (RIPL), at Michigan State
University (Dowdy et al., 1998; Pietola and Smucker, 1998). The Program WR-
RIPL V3.0 was used for measurements of root length. As a result, roots were
classified in 5 roots classes based on their diameter. The image resolution used
in this process was 276 pixels per 1 cm. Average root width classes were Class 1
(0.2 mm), Class 2 (0.5mm), Class 3 (0.9 mm), Class 4 (1.4 mm) and Class 5 (2.1
mm). All data are for roots without debris (Dowdy et al., 1998). RSA was

calculated as a sum of all image areas in all width classes, assuming a uniform

28

cylindrical root shape, with the formula A=2an, where r is root radius and L total
root length. RV were calculated as V= nr 2L.

The same procedure was adopted to analyze the samples collected at

MAC.

Statistical Approach

Statistical analysis were performed on the percent of clay, sand, silt, VWC
(%), AWC (%), and RLD (cm3cm’3) for all layers in both experiments by using
SAS (SAS Institute, 1990).

29

Results and Discussion

Greenhouse Experiment

Plant measurements

Several measurements were completed during the period when plants
grew on a limited supply of water. All these measurements were used as a basis
to decide the termination of the greenhouse experiment and the beginning of soil
water measurement with the TDR probe and soil sampling.

For each cylinder three corn plants were measured (Figure 1.4). In the
beginning of the growing period, plants in C1 were 12 to 20 cm tall. Initially the
growth in height was about 1 cm day", thereafter the growth rate was about 2 cm
clay’1 as long as the there was a adequate supply of water in the root zone. After
DOY 50 this rate decreased. All plants showed the same rate. Plants 1 and 2
stopped growing before plant 3; however plant 1 had a new growing stage after
no sign of growth for almost 24 days. It grew at rate of 1 cm day’1 approximately
and decreased again by DOY 100. As the other plants stopped growing, more
water probably became available for plant 1 and by DOY 100 the rate decreased
and no more growth was noticed. Plants in CZ presented similar patterns and
rates of growth. However, the rate of growth was lower than that observed in
plants growing in C1. Plant 2 stop growing on DOY 20 and plant 1 on DOY 40.
Only plant 3 kept growing at rates of approximately 1 cm day", probably because
more water became available after the other plants stop extracting it. However,
this plant reached the maximum height of 75 cm at DOY 80 and then stopped.

The reason for plants on CZ being shorter than plants on C1 was probably due to

30

their transplanting date. These plants were transplanted 20 days later than that
on C1, but the irrigation was withheld in the same period. In this case these
plants were subjected to the same period of water deficit observed in C1.

Leaf growth and rolling were monitored every other day to verify how
leaves of corn plants would be affected during the period of water deficit. Based
on the results of rolling index in Figure 1.6, it is clear that plant leaves were able
to recover the normal area while the water supply lasted; but with time this was
not possible anymore. With water deficit increasing, leaves kept rolled to avoid
more water loss. For both cylinders the rolling index decreased after DOY 45. In
C1 the rolling index reached its maximum values, around 0.6 after DOY 55 and
did not recover anymore. For C2 rolling index was more variable, with values
ranging from 0.3 to 0.6 after DOY 55. However, it reached values around 0.2 by
DOY 75. This occurred at same period when this plant stopped growing in DOY
80. At this time, the cumulative growth was less than that observed in the
previous leaves and the cumulative leaf growth for plants in both cylinders
decreased around DOY 40 to DOY 45 (Figure 1.5).

Approximately 30 days after the irrigation was withheld water deficit
became more severe (DOY 45). The daily evapotranspiration estimated for
plants growing in both cylinders decreased after the last irrigation on DOY 15
(Figure 1.7). Plants were able to resist water deficit for almost 30 days, but after
that it was evident that water deficit impaired plant growth. By DOY 80 in C1 and

DOY 92 C2, leaves showed a coriaceous aspect and a pale green coloration and

31

the experiment was terminated. Water measurements with neutron probe also

indicated that plants had reached their maximum of water deficits.

Water measurements with neutron probe

Based on the time course of water depletion for different soil layers, one
can see that for the first 4 layers in C1 (soil depths from 30 to 75 cm), the amount
of water extracted by plants was higher than that observed at deep layers. There
was an exponential decay in water depletion in 30-75 cm but it was not observed
at deep layers (from 90 to 120 cm). For C2 the same pattern was observed,
plants extracted more water in the upper layers than that at deep layers (Figure
1.8). It is likely that when water extraction stopped for upper layers, indicated by
the asymptote in the exponential decay, the lower limit of soil water content was
reached at those specific layers. This means that plants roots were not able to
extract water from soil, because either there was no more water available or the
potential at which water was being held was possibly greater than plant roots
could overcome to withdraw it from the soil. The resistance of soil to water
diffusion toward the root zone was probably high (Passioura, 1983). Another
possibility is that roots may have bypassed some soil regions without extracting
all available water in that part of the soil. If roots are not evenly distributed it is
not possible to extract all water from the soil. In this case, the neutron-probe can
not detect such water variability in the soil. The reason for this is that neutron-
probe readings are based on a volume of soil reached by the fast neutrons

emitted from the probe. The sphere of influence will vary according to the

32

wetness of the soil from 10 to 25 cm, which makes the neutron probe inadequate
for detection in water variability in small scale (Hillel, 1996). As a consequence,
patches of high water content will be averaged with dry ones in the sphere of
influence, which will bias the amount of water detected. It was assumed that
plants had stopped extracting water from deep layers at DOY 76 when the time
course graph shows no more variation in water content for several days in all
layers.

In this study, available water content was calculated as the amount of
water above LL for both experiments. As LL was presumably reached at upper
layers, but not at deep ones, its estimation for all soil depths was done based on
the volumetric water and clay content at upper layers.

The values of LL and VWC for C1 on DOY15 and on DOY 80 when the
experiment was terminated indicate that com plants extracted almost all the
available water in the upper layers (Figure 1.9). However, below 75 cm the
amount of water left above LL increased with depth. In this case the soil never
reached LL. Measurements with the neutron probe showed no more variation in
water content at deep layers. The average of VWC remaining for each layer
varied from 2% at 90 cm depth up to 6% at 120 cm and the amount of water left
without being extracted below 90 cm was around 15 mm. Based on these
values, it was evident that some water above LL was left in deep layers. It seems
that roots were not able to extract all water from these layers. Root bypassing of
some soil regions could have caused this poor water extraction. As it can be

seen in Figure 1.9, in C1, the rate of water extraction at deep layers were

33

occurring, but at low values. If the root system was not exploring these soil
layers, then the amount of water left might be explained by an uneven root
distribution or low root density at that depths.

A good correlation between roots and AWC was expected for all soil
depths. However, low values of R2 were found when RLD was plotted against the
AWC (Figure 1.10). A weak correlation was found, but there were evidences that
AWC increased when RLD (cm cm'3) decreased. This weak correlation was
probably caused by extreme values of RLD observed in some soil samples. Also,
different root regions can extract water at different rates. The correlation became
more evident at RLD ranging from 0 to 2 cm cm‘3. Similar situation was observed
for the correlation between RLD and AWC in the field experiment at MAC. Even
with a weak correlation between AWC and RLD, the results indicated that if there
are no roots in some soil regions the amount of water left is higher.

By knowing the clay and VWC content for each soil sample, it was
possible to establish a relationship among them. The results of this correlation
showed low values of R2 for the upper soil layers in both experiments (Figure
1.11). But despite the weak correlation, a cloud of values for VWC was
concentrated around a certain value of clay content. In C1 the average
percentage of clay content at 30 and 45 cm was around 6% and the VWC
observed was about 8 to 9%. The same analysis was adopted for C2 at 600m
depth and for the field experiment. In this way LL was estimated for all soil layers,

based on the clay content, for both experiments.

34

Water measurements with TDR probe

Measurement of VWC with the TDR probe began after the first part of the
experiment has been completed. At this time plants showed signs of water deficit
and no more water extraction was detectable at deep layers.

The statistical results for VWC determined by the TDR probe and for
AWC, calculated as the amount of water above LL, are depicted on Table 1.2.
VWC was higher at deep layers than that at upper layers. The mean and
standard deviation were calculated based on 966 TDR readings. All means were
significantly different (P=0.01), from each other with the exception of the VWC
observed at 15 and 60 cm.

Results for AWC showed significant difference (P=0.01), among soil
layers. Available water content was higher at deep layers; but different at 105 cm
and 120 cm, and was not different for layers 30, 45 and 60cm (Table 1.2). This
indicates that besides of plants are being affected by water deficit at the time
when the experiment was finished, some water was left in the soil. Similar results
were reported in other experiments (Passioura, 1983; Bland et Dugas, 1989;
Amato, 1991; Robertson et al., 1993). Also, the amount of roots found at deep
soil layers was low compared to the amount of roots at upper layers (Table 1.2),
meaning that there was not enough root to absorb water at deep soil layers.
Passioura (1983) argues that root length density lower than 0.05 cm cm’3 is not
enough to supply water demand and this can explain the water left at deeper

layers.

35

Soil texture analysis

Results for clay, sand and silt content for each of the 670 soil samples
analyzed are presented in Table 1.2 for both cylinders. Soil texture was
significantly different (P=0.01) for each soil depth. Clay and silt content were
higher at deep layers than the sand content, which was higher in upper layers.
The higher concentrations of clay content were observed at 90 and 120 cm
followed by the clay content at 105 cm, slightly similar to that at 75 cm in C1. This
can explain why the water content was high in deep layers than that observed at
upper layers, where sand content was found to be 73 to 75 %. The low capacity
of sand to hold water and the water extraction by plants explain the lower water
contents observed in these layers as observed in the previous sections. Despite
the averages of clay, sand and silt content observed in Table 1.2, for each soil
layer in C1, these averages tell us nothing about the clay content for each

individual soil sample.

Root analysis

In addition to water content measurements and soil textural analysis, the
amount of roots were also determined for each soil sample. In C1 the average of
RLD for each soil layer was obtained from approximately 138 soil samples (Table
1.2). The results were highly significant (P=0.01) and the greatest values of RLD
means were found at 30 and 45 cm, followed by lower densities at deep layers.
Below 45 cm RLD means were statistically the same. Again, these results

showed that RLD varied with soil depth but it did not take into account the

36

Variability found when each sample was individually analyzed. For C2, the
average RLD was considered statistically equal and unfortunately, due to soil
dryness at top layers, no RLD estimation was possible at 15 and 30 cm depth.

Root surface area (RSA) (cm2) and RV (cm3) for C1 followed the same
tendency of decreasing with soil depth observed in RLD (Table 1.5). This result
was expected, considering that these values were calculated based on RL (cm),
which also decreased with soil depth. Class 1 (0.2 mm) and 2 (0.5 mm), were the
major contributors to the total root length for each soil layer (Table 1.5) Their
contribution was on average 73% of the total root length for all layers. This
indicates that in great majority the RLD was comprised of fine roots with small
diameters. This certainly had a great impact on the values of water content left at
deep layers.

In order to understand the spatial variability of roots, volumetric soil water
content and soil texture, a correlation analysis was done (Table 1.4) for the
greenhouse experiment. Lower limit and clay content was the best correlation
found (Table 1.4). For each layer R2 (Pearson correlation) was as high as 0.99;
however, for RLD versus LL or even RLD versus clay content the correlation was
not significant. Other examples of significant correlation with low values of R2 are
found. VWC shows a good correlation with clay content and the same was
observed for AWC in some layers. But even for these variables low values of R2
were found. This means that the statistical approach was not adequate to show

the possible relationship among the variables within soil layers.

37

 

‘—~u

 

Field Experiment at MAC

Water measurements

The results of VWC observed during the time course of 10 days for
different soil depths are depicted in Figure 1.12. The amount of water did not
vary during the period of monitoring. Except for the upper layer at 15 cm where
rainfall caused an increase in water content on DOY 92 (Figure 1.3). There were
three rainfall events that contributed for the elevation in soil water content for the
next few days. After that period, VWC decreased again to the same values
observed at the time of installation of the tubes in the field. No more variations in
VWC were observed for the other soil depths. Even with high temperatures (30 -
35°C) water content for all layers did not vary (Figure 1.3).

Within soil layers VWC was significantly different (P=0.01) (Table 1.3).
Higher values (14% of VWC) were found at bottom layers, 105 and 120 cm,
compared to the upper layers that were slightly different. Only the upper layer (45
cm), was the driest one with 9% of VWC. Tube 2 was chosen since plants
showed more symptoms of water deficit when compared to the other plants in
tubes 1 and 3. AWC was different within layers, with values ranging from about

6% at 75cm depth to 1% at 45 cm depth.

Soil Results
Descriptive statistics are presented on Table 1.3 for the soil
samples taken where the access tube number 2 was located. Sand and silt

content were statistically different (P=0.01) within soil depths; however, the range

38

of variation was around 6% for sand and 3% for silt. Clay content had the highest
variation, ranging from 4% at 75 cm depth up to 12 % at 125 cm depth. Clay
content was higher at bottom layers than at upper layers. However, at 45 cm clay

content was as high as it was found at 105 cm.

Root analysis

High root concentration was observed at 45 cm, where RLD values were
around 1 cm cm'3. RLD decreased with soil depth (P=0.01). RSA (cm2) and RV
(cm3) also decrease with soil depth (Table 1.5). In the field experiment, class 5
(2.1 mm) had an additional influence on the results obtained. This class
contributed with 30% of the total RL (cm) and class1 and 2 together accounted
for 41%. This contrasted with the results observed in the greenhouse
experiment, where the major classes of root diameter were class 1 (0.2 mm) and
class 2 (0.5 mm). Yet at 60 cm depth class1 and 2 were responsible for 56%
whereas class 5 accounted for just 19% of the total RL. Interesting to observe is
that at 105 cm depth, despite of 1 sample has been found with roots, they were
thick, varying from 1.4 to 2.1 mm of diameter. Contrary to this finding, roots were
thin at 90 cm having diameter in between 0.2 and 0.5 mm.

Correlation analysis among clay, VWC, AWC, LL and RLD were
performed (Table 1.4). Among these variables only the correlation between LL
and clay content showed a high value for R2 (P=0.01). Correlation between AWC

and clay content was low, but still significant (P=0.01). RLD showed no

39

correlation with clay or LL, and volumetric water content had poor correlation with
clay content at 45 cm and 125 cm depth.

The results presented in both experiments were based on soil samples
collected by following a grid pattern for each soil depth. This was done to detect
spatial variability of water, roots and soil texture at each layer studied. However,
from these results no spatial variability was detected, and therefore it could not
explain why plants left water in the soil without being absorbed (Figure 1.9).

The natural processes involved during the soil formation cause the spatial
variability of soils (Beckett and Webster, 1971;Trangmar et al., 1985; Warrick et
al., 1986; Warrick, 1998), and the degree of variability will be different for each
soil and the scale used to detect it (Trangmar et al., 1986). This variability will
determine not only the physical and chemical properties of the soil along the
profile, but also water properties, soil biology, and even root distribution.

In this study it was possible to see that some variables differed by several
units at same soil depth (Tables 1.2 and 1.3). The values of standard deviation
and the range of occurrence of the variables studied within soil layers for both
experiments support this conclusion. The statistical analysis showed that
variables differed among themselves when compared within soil layers.
Nevertheless, this type of analysis does not take into account the distance
between samples. The only way to verify that some spatial variability occurred
was based on the standard deviation and the range for each variable. This
variability was assumed because samples were taken in a grid pattern. In this

case, not only volumetric water content but also roots and soil texture would be

40

different in some parts of the soil within the same layer. Assuming that samples
are different for each position from where they were taken, this could explain why
plants left water behind. Usually, it is assumed that macropores, fissures or other
natural openings in the soil cause non-uniform distribution of roots (Passioura,
1983; Hewit and Dexter, 1984; Tardieu and Manichon, 1987, Tardieu, 1989). In
this study the clay and water content had an important role in the root distribution
at each depth. The possible non-uniform distribution of clay may have affected
root distribution by directing root growth toward soil zones were water content
was lower than that observed where clay content was high. As observed by
Dexter and Hewitt (1978), roots tend to colonize loose zones in the soil because
they are deviated at the interface between loose and strong zones. As a
consequence, soil profile will present patches of high root density and others with
only few roots. Since clay can hold high amount of water, if clay content is so
high that can cause some impediment to root growth, they will not penetrate that
soil region and will bypass it leaving water without being extracted. Chapter 2 of
this dissertation will prove that the concept explained here is true. Depending on
the resistance of this barrier, roots will growth together in regions where soil
penetration and even water extraction is easier. In this way, a natural clumping
will occur in that region with severe consequences. For instance, drying the soil
region around them could increase the soil resistance to water movement
towards the roots (Passioura, 1983). In this case, there is water in the soil, but its
hydraulic conductivity is low to allow water movement and consequently no water

can be extracted. The only way for plants to survive is by producing new roots in

41

soil regions where conditions are more favorable or to increase absorption in
other soil regions (Passioura, 1983); nevertheless, if plants are already stressed
by lack of water, production of new roots will further drain the limited carbon
reserves of a stressed plant (Smucker, 1984, 1992). New roots at deep layers do
not mean a guaranty of water supply, because the amount of roots produced can
be insufficient to absorb water or supply the total demand by the plant, for

instance, RLD low than 0.05 cm cm‘3 (Passioura, 1983).

Summary and Conclusions

Based on the results obtained in this study, classical statistical analysis
could not provide the basis to detect spatial variability in the soil. It just
demonstrated that volumetric water content, roots and soil texture varied within
soil layers, and that there was a poor correlation among them at each soil depth.
It is not known whether these variables are spatially distributed in each soil layer.
Moreover, if such variability exists its degree is not known. Because of the grid
pattern adopted during the sampling process and existence of different values
found for each soil sample studied, it was assumed that this variability might
exist. To find more details about this supposed spatial variability and how it
affects root growth and water extraction, a new approach is necessary. This will

be the subject of the next chapter in this thesis.

42

References

Ahuja, LR, and DR. Nielsen. 1990. Field Soil-water relations. p. 143-190. In
B.A. Steward and DR. Nielsen (ed.) Irrigation of Agricultural crops. Agron.
Monogr. 30. ASA, CSSA and SSSA, Madison, WI.

Amato, M. 1991. Spatial distribution and water uptake of roots in structured
growth media. Ph.D. diss. Michigan State Univ., East Lansing.

Amato, M., and J.T. Ritchie.1995. Small spatial scale soil water content
measurements with time-domain reflectometry. Soil Sci. Soc. Am. J.
59:325-329.

Beckett, P.H.T., and R. Webster. 1971. Soil Variability: a Review. Soils Fert.
34:1-15.

Bland, W.L., and W. Dugas. 1989. Cotton root growth and soil water extraction.
Soil. Sci. Soc. Am. J. 53:1850-1855.

Bruckler, L., F. Lafolie, and F. Tardieu. 1991. Modeling root water and soil-root
water transport: ll. Field Comparisons. Soil Sci. Soc. Am. J. 55:1213-1220.

Burt, R., T.G. Reinsch, and WP. Miller. 1993. A micro-pipette method for water
dispersible clay. Commun. Soil Sci. Plant Anal. Z4(19&20):2531—2544.

Carlesso, R. 1993. Influence of soil water deficits on maize growth and leaf area
adjustments. Ph.D. diss. Michigan State Univ., East Lansing.

Dexter, AR. and J.S. Hewitt. 1978. The deflection of plant roots. J. Agric. Eng.
Res. 23;17-22.

Dowdy, R.H., A.J.M. Smucker, M.S. Dowlan, and JC. Ferguson. 1998.
Automated image analysis for separating plant roots from soil debris
elutriated from soil cores. Plant Soil 200:91-94.

Dwyer, L.M., D.W. Stewart, and D. Balchin. 1988. Rooting characteristics of corn,
soybeans and barley as a function of available water and soil physical
characteristics. Can. J. Soil Sci. 68:121-132.

Gee, G.W., and J.W. Bauder. 1986. Particle-size analysis. p. 383-411. In A.Klute

(ed.) Methods of soil analysis. Part 1.2"GI ed. Agron. Monogr. 9. ASA and
SSSA, Madison, WI.

43

Hamblin, A. and D. Tennant. 1987. Root length density and water uptake in
cereals and grain legumes: How well are they correlated? Aust. J. Agric.
Res. 38:513-527.

Hamblin, AP. 1985. The influence of soil structure on water movement, crop root
growth, and water uptake. Adv. Agron. 38:95—158.

Hewitt, J.S. and AR. Dexter. 1984. The behavior of roots encountering cracks in
soil. II. Development of a predictive model. Plant Soil 79:11-28.

Hillel, D. 1997. Introduction to soil physics. Academic Press. San Diego, CA.

Klepper, B. 1991. Crop root system response to irrigation. lrrig. Sci. 12:105—
108.

Lafolie, F., L. Bruckler, and F. Tardieu. 1991. Modeling root water potential and
soil-root water transport: I. Model presentation. Soil Sci. Soc. Am. J.
55:1203—1212.

McIntyre B.D., S.J. Riha, and DJ. Flower. 1995. Water uptake by pearl millet in a
semiarid environment. Field Crop Res. 43:67—76.

Miller, WP, and OM. Miller.1987. A micro-pipette method for soil mechanical
analysis. Commun. Soil Sci. Plant Anal. 18(1):1-15.

Molz, F .J. 1981. Models of water transport in soil-plant-system: A review. Water
Resourc. Res. 17(5): 1 245-1 260.

Passioura, J.B. 1983. Roots and drought resistance. Agric. Water. Management
7:265-280.

Pietola, L. and A.J.M. Smucker. 1998. Fibrous carrot root responses to irrigation
and compaction of sandy and organic soils. Plant Soil 200:95—105.

Post, OF, C. Mack, P.D. Camp, and A. S. Suliman. 1988. Mapping and
characterization of the soils on the University of Arizona Maricopa
Agricultural Center. p. 49-60. In Hydrology and water resources in Arizona
and southwest. Proceedings of the 1988 Meetings of the Arizona Section
American Water Resources Association. University of Arizona, Tucson,
AZ.

Robertson, M.J., S.Fukai, M.M. Ludlow, and G.L. Hammer. 1993. Water
extraction by grain sorghum in a sub-humid environment. I. Analysis of the
water extraction pattern. Field Crop Res. 33:81-97.

SAS Institute 1990. SAS/STAT user’s guide. Version 6. SAS Inst.,Cary, NC.

44

Smucker, A.J.M. 1984. Carbon utilization and losses by plant root systems. In
S.A.Barber and RD. Boulden (eds) Roots, nutrient and water influx, and
plant growth. Am. Soc. Agron’. Madison, WI.

Smucker, A.J.M. 1990. Quantification of root dynamics in agroecological
systems. Remote Sens. Review 52237-248.

Smucker, A.J.M., and RM. Aiken. 1992. Dynamic root responses to water
deficits. Soil Sci. 154(4):281-289.

Stickler, F.C., S. Wearden, and A.W.Pauli. 1961. Leaf area determination in grain
sorghum. Agron. J. 53:187-188.

Tardieu, F. 1988. Analysis of the spatial variability of maize root density. II.
Distance between roots. Plant Soil 107:267-272.

Tardieu, F. 1989. Root system response to soil structural properties: micro and
macro scale. p. 153-172. In W.E. Larson et al. (ed.) Mechanics and
related processes in structured agricultural soils. Kluwier. London,
England.

Tardieu, F., and H. Manichon. 1987. Etat structural, enracinement et alimetation
hydrique du mais. l. Modelisation d’etat structuraux de la couche labouree.
Agronomie 7(2):123—1 31.

Tektronix. 1987. 1503B. Metallic time domain reflectometer. Operator Manual.
Tektronix, Beaverton, OR.

TOpp, G.C., J.L. Davis, and AP. Annan. 1980. Electromagnetic determination of
soil water content: Measurements in coaxial transmission lines. Water
Resour. Res. 16:574-582.

Topp, G.C., J.L. Davis, and AP. Annan. 1982 Electromagnetic determination of
soil water content using TDR: l. Applications to wetting fronts and steep
gradients. Soil Sci. Am. J. 46:672-678.

Trangmar, B.B., R.S. Yost, and G. Uehara. 1985. Application of geostatistics to
spatial studies of soil properties. Adv. Agron. 38:45-94.

Trangmar, B.B., R.S. Yost, and G. Uehara. 1986. Spatial dependence and
interpolation of soil properties in West Sumatra, Indonesia; l.Anisotropic
variation. Soil Sci. Soc. Am. J. 50:1391-1395.

Warrick, AW. 1998. Spatial Variability. p.655—675. In D.Hillel (ed) Environmental
Soil Physics.

45

Warrick, A.W., D.E. Myers, and DR. Nielsen. 1986. Geostatistical methods
applied to soil science.p.53-82. In Klute(ed.)?. Methods of soil analysis.
Part 1. 2nd ed. Agron. Monogr.9. ASA and SSSA, Madison, WI.

Whiteley, GM, and AR. Dexter. 1983. Behavior of roots in cracks between soil
peds. Plant Soil 74:153-162.

Zobel, R.W. 1996. Genetic control of root systems. p. 21-30. In Yoav Waisel et

al. (ed.) 2nd ed. Plant Roots: The Hidden Half. Marcel Dekker, New York,
NY.

46

, , L-L-L._._..___L___L _ LL, 2

50'; .000000. ;
I0 00......... 000
40l0000 ... 0000
00000 ..0 000
(00000 00000

3000000 0000
C... O...

20 i0.0.0.0.0 . . 0.0.0.0.0
O O O
101,0.0.0.0 . . . . 0.0.0.0
‘ . 0 0 0 0 0 . , (a)
l 0 0 0 0 0 0 l

1 - ~ .~~--T0+r~~—.--—-— - ~'
10 20 so 40

250 0 0— r+4 0 0- 0—
0 0 0 0 0 0 0 3

20+ 0 0 0 0 0 0 0

3‘ 0 0 0 o 0 0 0 0

150 0 0 0 0 0 0 0 0 A
I“ 0 0 0 0 0 0 0 0 +
10+ 0 0 0 0 0 0 0 0 I
0 0 0 0 0 0 0 0 0

50 0 0 0 0 0 0 0 0 (b)
40444444 0;

5 1O 15 20 25
(cm)

Figure 1.1. Grid sampling used in the greenhouse experiment (a) and in the field
experiment. (b).

47

5O

 

 

 

0
0 9. .
40- ’ 0 . . ~
. . O. . ’
a 0 9 ' 9 0 '
o . C . Q ~ ~ . . . .
V 0 O 9 0 0
9 30 r ’0 ' ' ' a 0
3 ’ Q r 0. 0.0
‘5 00 0
Q (9 oo o o O OcD
O O CID 0 C9
5 20 - o o o o (93% (9% o 00 o
o o
.2: O O (P O 0
<1: 0 c9 0 O Q)
o
10 - O 0%
0 Maximum
0 Minimum
0 f I I I l I I I I I I I I r f

 

20 25 30 35 4O 45 50 55 60 65 70 75 80 85 90 95 100
DOY

Figure 1.2. Daily maximum and minimum air temperature in the greenhouse at
Michigan State University, MI. (1998).

48

4O

 

 

0 Maximum
0 Minimum . 9
A 30. ..' I...
U '0 I O O
Q, 0. 0’00 o
J 0 . . fl
9 0"‘f ~ 0 0 .' . ’ 0
1:6 20- w. . 0 .0 0 '.. O
8. . . ' ' 60
GE, 0 Oco
o-o 10- O 0
h <90 0 o 8 q? (90
< ,. . mesa. ewesii .
0 O 0 C5“) ‘3) (90 O o
(56> 0000 o 0 oo
0 O
Q) d) OQECD (53

 

'10 I T l

 

35-

 

 

 

 

 

 

 

NA
E 304
3 .Q' “I
V 25— A 0.
C . O
'2 20‘ 43’“ ° -
‘3 a? ' 0
E 15 “‘w o. . O
E 10— O . . .
o _ 0
a) 5 ' .
0 T
12
10~
E
e 8‘
“E 64
1T!
II
4—
2‘
0 I T IIIIIII I II In T
O 20 4O 60 80 100 120
DOY

Figure.1.3. Air temperature (°C), solar radiation (MJ m'2) and rainfall (mm) at

Maricopa Agricultural Center - AZ.(1999).

49

140

100

 

 

 

 

 

 

 

 

 

 

 

 

3
,\ 1
E 2
(J
V
4—0
.c
.5?
d)
3:
4—0
(I
E
Q.
(a)
0 r T T r ‘1'
0 20 40 60 80 100 120
DOY
lastnngafion
100
90*
E5 80~ 3
0 fl—‘V
\—I 70-4 ,/
'0-0
.c
.9 60- "fl
0 1
J: 50~
4.1
5% 4o -
CL 30,K’Wooormooocxxnoooooo 0000 o 2
o
20~
10‘
(b)
0 I T T
0

20 4o 60 80 100 120
\I/ DOY

lastfingafion

Figure 1.4. Plant height (cm) for three maize plants, C1(a) and CZ(b).
Greenhouse experiment at Michigan State University , MI.

50

 

80

60 ~

40-

20 a

Cummulative leaf length (cm)

 

leaf number
-—C»-12

O 14
—v- 16
—v 18
+ 20
-—G 22
—O 24

 

OOOOOOOOOOOOOOOOOOOOO-‘ O
O

(a)

 

100

 

20 30 40 50 60 70 8O

DOY

 

80 n,

Cummulative leaf length (cm)

60 4

4o-

20 a

 

leaf number
+ 12
AC) 14
-‘V-16
—*7 18
—I- 20
-{} 22
4 24

    

meH-vwm — —v

,/

JV] vyVV'V‘VV‘V'VV-V n—v
f
I

I-I——-I
m
b0000050000095500000~ g

(b)

 

 

10

20 30 40 50 60 70 80

DOY

Figure 1.5. Cumulative leaf expansion from day of the year (DOY) 13 to the
termination of the experiment at the greenhouse. Data for C1 (a) and C2 (b).

51

 

 

Ieaf13-C2

 

 

 

 

 

1.0 4 leaf 16-C1 1.0 .
S x
a; 0.8 g 0.8
g 5
3 0.6 I .E’ 0.6«
E S
3 rt
04 i 04
0.2 0.2

 

303540455055606570 3035404550556065707580

DOY DOY

Figure 1.6. Daily leaf rolling (fraction of blade exposed) for different leaves and
cylinders, from day of year (DOY)35 and 40, to the end of the greenhouse experi-
ment at Michigan State University, Ml.

 

 

 

 

 

 

 

 

:8 €‘8

>‘ >

m (U

Es Es

.§4« .54-

’6.

m, I”,

«52 £2

2 8

go a... 30 a....eefi

”102030405060708090 102030405060708090100
DOY DOY

Figure 1.7. Daily evapotranspiration (mm day"), for C1 and CZ, from day of
year (DOY) 10 to 90. Greenhouse experiment at Michigan State University,Ml.

52

0.35

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Soil depth (cm)
0.30 ~ 0
or so
00 q 45
"7" 025 H T”
s - “ m . + ..
O 0. {SI-90
‘2 t ‘ +106
, I
g 0.20( _ , 0120
v
0
'o
E 0.15 ‘ ‘
0 ~.
0, %N
00.00 V9": 00k
0.10- 000000 _ L 90
0.05 I I T T T T T
10 20 30 40 50 60 70 80 90
DOY
0.30
Soil depth (cm)
-0-ao
0.25 ~ ‘0 ‘5
4' 00
A +75
0? -I- 90
E 0.20- ‘0'“
”0 +120
E
8
0 0.15—
0.10 a
(b)
0.05 T l T I T f T T T TI
10 20 30 4o 50 60 7o 80 90 100 110 120

DOY

Figure 1.8. Day of year (DOY) and volumetric water content within soil depths (cm)
for C1(a) and CZ(b) in greenhouse experiment. Michigan State University,MI.

53

Volumetric water content (cm3 cm'3)

 

0.05 0.10 0.15 0.20 0.25 0.30 0.35
0 I l I I J
l l l l |
,,___ L___I___J__-J __J____
TI | l i‘\ l
3m___ i___i___i___i_;\i____
r T T 7 an

I l i /
4::‘*ti———i—“*i“——i’c——

El 60 -—--— i-———+——--t-—-¥-l ————————
g l I / I

8

E l | / l

9°‘“‘"Il*f“i"‘i T 1““

I

I

I

I

7e——— L———+———+— ab4———4————

I

mse——Ih&— —L———i—+_J___fl____
I

l I \ l l | l
uw———4l— \bL———+—r—% ————————
LLi MI") I | ”(WI
135 "'-———T__—_:I“___"I'—__—l'_'_—‘_l_—__

 

 

 

I I I I I

Figure 1.9. Volumetric water content (cm3cm'3) at different soil depths (cm) for day
of year (DOY) 15, 80 and at lower limit (LL) (C1).

54

 

20

 

 

 

 

 

 

 

 

 

 

 

 

i Y=1 .06+3.616xp(-XIO.67)
R2= 0.10
(g) .
< ' .
(a)
8 10 12 14 16 18
RLD
20
Y=1 .57+3.79exp(-XIO.56)
R2= 0.17
154’
l
I)
‘z’
<
II . 9 .0 o (b)
'5 T I T
0 1 2 3 4
RLD

Figure 1.10. Root length density (cm cm'3) (RLD) as function of available water
content (%) (AWC) for two different experiments. Greenhouse experiment (a)
C1 at Michigan State University, MI; and field experiment (b) at Arizona.

55

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

14 IT . T I'Misotzn“: 18 Ml45cm
12 I . .~. g i 16 ' o
10 ~: . 0 . .: .0...:.0. I. 144 .
e3 8: ' ~7'e’1-‘3‘7J €12 ~°~~° '.
l '. 0 O ’0 . l . 0 . . 3.0. .{0
g 6i . 1 lg“) ,f32f0‘03'30 , .
4 4‘ ' . L; 81 : ' . ° :
2 t 0 l 6 1 . .
0 I.Y=.0.034.X+soo .R?=oor2 ___ ___ _ _ ,1 4 ,Lgogsxie .R2=°-092 . To a
O 2 4 6 8 10 12 1 2 3 4 5 6 7 8 9
cla °/
N O) clay (%)
1: 4 MI30+45cm 14 A245em
1 4
8 0‘
5. 9 ‘
4 . 8 :
2 1 7 I 2 .
O _LDEZTX:881 00004 6 Y=O_4T26X+527 RMS Y fi

0 2 4 6 8 10 12 5 6 7 8 9 10 11 12
clay (%) clay (%)
Figure. 1.11. Volumetric water content (WVC) as function of clay content (%) at

30 cm, 45 cm and combined 30+45 depth in the greenhouse experiment, and
at 45 cm depth in the field experiment.

56

 

20
18 ~

16‘
%4~%—~%o**9%~———ew1m

 

 

 

"5‘ J
E 14
(‘00
.l F" __ __
E 12 + +++++ _.___. 13:
O
102 D—-CI’~D--——D—-D""D‘-D-—D——~-—_ ——{j
8 232*~V—#* ___thv 60
I ~7- __ + v
P):3:——:‘:I_.LT l ‘~*7~~1 45
15
4 T f I T T T T
90 92 94 96 98 100 102 104
DOY

Figure 1.12. Day of year (DOY) and volumetric water content within depths (cm) at
Maricopa Agricultural Center - AZ.

57

Table 1.1. Soil texture analysis, pH and organic matter (%) within soil layers

for two locations.

 

 

Site Soil depth Sand Silt Clay pH O.M.

(cm) ( "/- (%)

Michigan 0 - 30 74.4 18.9 6.7 7.1 0.9
30 - 45 77.1 16.2 6.7 5.4 0.8

45 - 60 79.1 14.2 6.7 5.9 0.8

60 - 75 71.1 20.2 8.7 6.0 0.3

75 - 90 59.1 22.2 18.7 6.5 0.6

90 - 105 71.1 18.2 10.7 7.6 0.2

105 - 120 57.1 30.2 12.7 8.1 0.2

Maricopa 0 -15 76.3 10.4 13.4 8.4 0.6
15 - 30 80.3 8.4 11.4 8.5 0.3

30 - 45 81.3 8.4 10.4 8.6 0.1

45 - 60 80.1 8.5 11.4 8.7 0.2

60 - 75 80.3 8.4 11.4 8.6 0.2

75 - 90 81.1 7.5 11.4 8.5 0.2

90 -105 79.9 8.7 11.4 8.3 0.3

105 -125 75.9 11.7 12.4 8.2 0.3

 

 

58

 

59

 

2.8.2 :43: item: :Bemm :52 tom. EN :29 see“.
Se 2.o mum med 2e B...” m: cm

e 2s a Be a 2.2 n 2.2 a See a 8.8 n 8.2 ewes. 232
Be Re New «E 8e 5e 8e am

e 8e 0 26 a Be a 2.2 a Dem a some a 2.2 ends. 82-8
26 and 8e Be he one 8e am

e 2s a Re e 8e 0 8.2 e 2. K. e 8.8 a 2.2 ends. 8-2
Re is Rs Be 2.4 one ems am

a Be e Be a EN _e 8.2 e 8.2 a BB e and sees. 2-8
8e :8 med 82 2a :2 «Na am

e mas e and Sue See a 9.2 a 8.2 e 8e. 58.2 8.?
8d 2d a: new 8.2 as 2.2 am

a new a Be .22 a one a ewes a 8.2 e 2e :82 218
8.0 Be. Be 2a 8.2 «3 :2 am

e was e 8e Leno ewes a 8.2 n 8.2 e Be :82 one

................................................................. 5
29
j 3m 02:. 03> “__m seem .66 £80

 

._5_.>L._mLm>_cD 2.3m 39:22 E. EoEtoaxm 8305590 .2306 :8
E9226 E magtoaoa oEom .8 .59 co=m_>o_c Emucfiw can 522 .mozwzflm 023:0me .Né 05m...

._e>e_ £5805 88 8 28588 :
.92 £888 mod n E8585 .
E3558 8: - m:

 

 

:88 28.0 :82 :88 :8: .88 :28 En".
Be 8.0 88 8.4 8.8 8.2 88 am

e 28 a Be a 2.8 a 8.2 a 8.2 a 8.8 a 8.2 :82 8-8
Se 88 2.8 «we Re 8.2 8.8 cm

a 8.8 n 28 a 8.8 a as e 8.2 an 8. E a 8.8 ends. 28
8.0 «no 88 8.2 8.2 8.2 8.2 am

e 8.8 n Be a 88 e 8.8 e 8.8 a 5.8 e Es :82 8-8

22
.3 9m 0>>< 25> Em 88 30 same

 

.553 .82 same

60

._e>e_ £8898 58 a 28588 t

 

: 888 : 5.8 : 2.3 : 8.8 : 8.2 : 8.8 : 888 .58“.
8.2 88 5.8 8.8 88 88 88 am

a 8.2 e 88 a 8.4 e 8.2 8 8.2 a 5.: a 8.2 :85. 82-82
28 88 8.8 8.8 88 88 8.2 cm

a 28 e 58 a 88 a :2 e 8.2 a 2.2 a 28 58.2 82-8
88 88 5.2 8.2 2.8 8.8 8.2 am

e 8.8 e 38 a 28 a 2.2 a 8.2 a 8.8 e 88 88.2 88.2
$8 28 t8 8.. 88 El. :2 am

e 2.2 e 88 a :8 a 8.2 a 8.2 a 84.8 a 8+ :82 2-8
2.8 :8 8.8 2.8 8.2 2.2 2.2 am

e 8.8 a 88 a 28 a 8.2 a 2.2 a 8.8 e 88 :82 8-84
88 88 88 8.2 83 8: 8.2 cm

a 88.8 a 888 e 88 e 88 a 8.2 e 8.2 a 88 e82 988

j Ba 03.... 95> 8 8:8 30 £88

 

.N< .3028: an EmEtoaxo 20K .2328 :8" EQoEv
«m moEeaoa :3 9:3 .8 .59 53.83% Emucflm new cums. .85sz 039.800 .02 038...

61

888K .8 885885 .

 

 

 

 

6.8K a 88588.8 >29: .-
5338 89me
- - m: 00.0 - .5 F00 - .3 00.0 00 7008
m: 3.0- m: 00.0 m: 08.0- m: 08.0- r. 00.0 m: 00.0 m: 00.0 00 8.00
m: 0 F .0 m: 00.0- t. 00.0 m: 0 F .0 3 00.0 m: 00.0- m: 3.0- 00-00
m: 00.0 m: 00.0- - 00.0 m: _.0.0 r. 00.0 m: 00.0- m: 00.0 00-00
m: 0 F .0 m: 80.0- t. 00.0 m: 0 _- .0- .3 00.0 m: 00.0- m: 3.0- 00-0w
m: 00.0 m: 00.0- m: 08.0 m: v0.0 1 00.0 m: 00.0- - 00.0 0Y00 m:o~:<
m: 0.0- m: 00.0- t 00.0 m: 08.0- t 00.0 m: 50.0- .; 00.0 00-00
m: 08.0 m: 00.0 5. v0.0 m: 0rd 3 00.0 m: 00.0 t 00.0 00-00 00
m: 010 m: 00.0- m: 3.0 mc 30 i 00.0 m: 00.0- m: 80.0- 00-0w cm0Eo=2
m: 00.0- m: V0.0 m: 80.0- m: 0 F .0 .3 00.0 m: v0.0 m: 00.0- 00 T00?
m: 3.0 m: 00.0- i 00.0 m: v0.0 3 00.0 m: 00.0- i. 80.0 00 8.00
.. 00.0 m: 0 F .0 .3 00.0 - 00.0 t. 00.0 m: 0.0 t {.0 00-00
m: 00.0 m: 00.0- t. 00.0 m: 00.0- .3 00.0 m: 00.0- t 00.0 00-00
m: 00.0 m: 00.0- m: 0F.0- m: 80.0 .3 00.0 m: 00.0- m: 00.0 00-0v
mc 00.0- m: 00.0- m: 00.0 m: 00.0- 3 00.0 mc 070- m: v0.0 9100 5
m: 0.0- 0: 00.0 m: :20 m: 80.0- t 00.0 m: 00.0- - 00.0 00-0 58030:).
> 58.00
>50xnjm 04mx0>>< ><-_Ox0>>< 04183.. ><-_Ox-_-_ 04m x 0>>> <40x0>>> __ow

 

8888888. 828:6 828 H 858.88 =8 25;, Eu 888 6va £858

£05. 89. .8530 22:8 8902, o_nm__m>m .3.: ES. 326. .8>>>v Emacoo 89.83 2095.9, .2 cozflmtoo $808020 4: 22m...

62

..0>0_ 0.5 00 08:8 Boo. oz C

 

 

 

 

 

- - - - - - - - «00F-00F
0F.0 00.V F0. FF 00.V 00.0 V0.F 00.F 00.0 00F-00
00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00-00
0F.0 00.0 00.00 0F.0 00.0 0F.0 F0.0 0V.0F 0000
V0... F0. F0 00.V00 0V.00 00.00 00.00 00.00 0 F.00F 00-0V
0V.0 00.00F 00.000 F0.00F 00.00 00. F0 00.VOF 00.00F 0V.00 0CONt<
00.0 00.00 00.00F 00.00 V0.0F F0.VF 00.00 00.00 00-00
0V. F 00.0V 00.00F F0.0V 00 .00 00.00 00.00 0F. F0 0000 A000
00.0 00. F0 00.0FF V0. FF 00.0F V0. F0 00.0V 00.0V 00-0V 00020.5.
00.0 V0.V 0V.0V 0V.0 00.0 0F.V 0V.00 0V.0F 00F-00F
VV.0 00.00 0F.V00 0F.0 00.0 F0.0F 00.00 00.00 03.00
0V.0 0V.00 F0000 V0.0 0F.0 V0.0F 00.00F 00.00F 00-00
00.0 00.0 00.00 00.0 00.0 F0.F F0.00 00.00 0000
00.0 0V.00 00.00V VF.00 V0.0V 00. F0 00.00 00.00F 00-0V
00.0 0F.0VV V0.0000 00.VOF F0.00F 0V.000 V0.000F 00. F FOF 0V.00 :00
00.0 00.0V0 00.0 F00 0V.00F 0V. FOF FV. F00 00.0F00 00.00V0 00-0 00020.5.
0880 0.880 .85. 8853-3. 8:83-880 €508-08 88508080 .55 08.80 0:80 5000 0:0

>1 <01 0m...- 0 000.0 V 000.0 0 000.0 0 000.0 F 000.0 :00
3.630.

 

055?. 0. >003 0:0 80030.5. :. Soc 8.. 0000 05000 :00 0.5.3 9.0 083.9

60.. 0:0 20% 00.0 000030 009 .30.... 50:0. .02 .009 00000.0 50.? 0>c .8 310 £000.. “com .0.F 0.8.0...

63

Chapter 2

Spatial variability of soil texture, root distribution and water content in
different soils. II: A Geostatistical approach to describe micro-scale spatial
variability.

Introduction

Soil environment is a complex media to host plant growth (Russel, 1977;
Jones et al., 1991; Kramer and Boyer, 1995). The inherent heterogeneity of soil
properties will affect not only root growth, but also .water content and the
interaction among them (Ahuja and Nielsen, 1990). Such complex environment
offers several restrictions to root growth through physical, chemical and biological
stresses, which may vary at different scale in space and time (RusSel, 1977;
Kuiper et al., 1988; Bennie, 1996). Thus, roots may experience different
environments when growing through the soil matrix (Passioura, 1991). The
heterogeneous nature of soil causes a high degree of variability in root growth,
which consequently affects the top portion of the plant (Brown and Scott, 1984).

Understanding the soil spatial variability is important for crop production
(Saddiq et al., 1985). Usually, studies of spatial variability involve distances that
can vary from meters to hundreds of meters (Yost et al., 1982; Warrick et al.,
1986), and the spatial variability of soil physical, chemical and hydraulic
properties will be treated according to the scale adopted (Trangmar et al., 1986).
Studies about the micro-scale variability of such properties or related to spatial

variability of root growth are scarcer (Amato, 1991; Tsegaye et al., 1998a,b). Few

64

studies were done, but none of them used robust tools to analyze the results
achieved (Tardieu, 1989; Bruckleret al. 1991b).

Studies of spatial variability demand that sampling procedure be made in
a grid pattern to consider the dependence among sampled regions (lsaaks and
Srivastava, 1989). Since the magnitude of soil properties at one location is
usually related to the magnitude at another location in the same. area, the use of
classical statistics, that assumes spatial independence, is not appropriate (Brown
and Scott, 1984).

A branch of statistics, referred as geostatistics, was developed by
Matheron (1973) to solve the problems concerning the spatial variability.
Geostatistical analysis calculates correlation among observations made at
different locations (Trangmar et al., 1985,1986; lsaaks and Srivastava, 1989).
The observed correlation can then be used in practical applications to estimate
values at unsampled locations by means of kriging and co-kriging techniques
(Journel and Huijbregts, 1978; lsaaks and Srivastava, 1989; Goovaerts, 1992,
1994, 1997).

Considering the natural variability of soil properties and the uneven
distribution of roots on the soil matrix, geostatistical analysis may be a valuable
tool to understand the complex inter-relationships between soil and roots. By
visualizing kriged maps, geostatistics can help us to assess how roots are
distributed in the soil when crops grow under different environmental conditions.

The results obtained in geostatistical analysis could improve water

extraction models. Water extraction could be better predicted and crop yield

65

improved if such water models could take into account spatial variability of root
distribution.

The goal of this study was to apply the principles of geostatistical analysis
in two data sets collected in two different locations, to understand the bypassing

of soil regions and uneven distribution of roots, during periods of water deficit.

Material and Methods

The data set used to describe the spatial variability of soil water content,
roots and soil texture were collected at Michigan State University Research
Farm, East Lansing, MI and Maricopa Agricultural Center, Maricopa, AZ.

All procedures adopted to measure volumetric water content, as well as
the methodology used to analyze root and soil samples were described in
chapter one.

The values for available soil water content in each soil sample were
calculated by the difference between volumetric water content at the end of the
experiment and the respective lower limit for each sample. Values of lower limit
of soil water content for each soil layer, and consequently for each soil sample,
were estimated by using the equations showed in Figure 1.11. In this case, it was
assumed that at upper layers, LL was reached and that this value was equal to
the average of VWC determined by the TDR measurements. By plotting the
VWC found for each sample and the clay content in each sample for the upper
layers, it was possible to establish a simple relationship to estimate the LL for

deep layers based on the values of clay content.

66

Statistical Approach

Statistical analysis were performed on the percent of clay, sand, silt, VWC
(%), AWC (%), and RLD (cm cm?) for all layers in both locations.

Each variable was first subjected to a classical statistical analysis to obtain
the minimum and maximum values, mean, variance and standard deviation
within each soil layer using the GLM procedure in SAS (SAS Institute, 1990). The
degree of spatial variability for each variable, except for sand and silt, was
determined by geostatistical methods (Vieira et al., 1983; Trangmar et al., 1985;
lsaaks and Srivastava ,1989). A semivariogram model, which mathematically
describes the moment of inertia around the diagonal of the associated h-
scatterplot, and thus indicating how points are spatially correlated in a scatterplot
(Trangmar, 1985, Isaak and Srivastava, 1989; Goovearts, 1997), was fitted to
each variable to find the parameters necessary to utilize a kriging procedure.

A semivariogram consists of four basic parts; the nugget effect, range, sill
and the structural variance (Figure 2.1). The nugget effect (Co) represents the
unexplained or random variance, which is caused by sampling error, or variability
of the variable which cannot be detected by sampling scale used (Trangmar et
al., 1985). The Range represents the distance over which the variables exhibit no
spatial dependence, and is defined by the value at which the curve reaches the
sill. Sill approximates the sample variance, and is the region of relatively constant
semivariance. It is the point where semivariance is at its maximum value. The

structural variance (C) is the portion of the total variation correlated with distance.

67

The semivariogram model gives us the values of sill, range, Co, and C,
which are used in the kriging procedure. Kriging is a technique of making optimal,
unbiased estimates of variables studied at unsampled locations using the values
obtained in the semivariogram and the initial data set (Trangmar et al.,1985).
Semivariogram models for the variables in this study were calculated by using
GSPIus+ (Robertson, 1998). Each property was kriged and mapped based on
the values obtained in the semivariogram model, with Surfer V. 6.0 (Golden

Software, 1997).

Results and Discussion

The study of spatial variability by using the geostatistical approach
sometimes requires the same results produced by classical statistics, e.g.,
means, standard deviation, variance, histograms (Isaac and Srivastava, 1989).
This is only the initial step to characterize the spatial variability of the variables
being studied.

The results of the statistical analysis performed for soil texture, VWC and
AWC, RLD are depicted on Tables 2.1, 2.2 and 2.3 for Michigan and Arizona. In
general, clay content increased with depth for both places. In the greenhouse
experiment this increase in clay content was observed below 75 cm in C1 and
below 90 cm in C2. ln C1 the average of clay content changed from 5% at 60 cm
to 13% at 120cm depth. However, maximum values were found at 75 cm, 26% of
clay, and 90 cm, 27% of clay. Variance and standard deviation were also high

for these layers. Yet, the variation of clay content in CZ was about 5 to 13%,

68

almost the same range observed for C1. At MAC, the average of clay content
also increased below 75 cm depth going from 4% to 12%. Contrary to the results
of variance and standard deviation observed at Michigan, at MAC their values
increased with depth. Values for standard deviation varied from 1.1 to 3.2 and
variance ranged from 1.2 to 10.2.

RLD (cm cm'3) decreased with depth in both places and the highest
concentration was found at the upper layers; 0-30 cm and 30-45 cm at Michigan
and 30-45 cm at MAC. The amount of roots at bottom layers was low for both
layers, with complete absence of roots at 105-125 cm at MAC. For this
experiment the high concentration of calcium carbonate at 45-60 cm may have
impeded the root development in deep layers. Subsoil with naturally high
mechanical impedance due to factors such as duripans, fragipans, petrocalcic
layers, or placic layers have a great impact in root growth in some soil layers
(Bennie, 1996). At 45-60 cm maximum values of RLD were around 3.7 and 1.7
cm cm'3, compared to values of 0.85 (60-75cm), 0.43 (75—90 cm) and 0.0 cm
cm'3 (1250m). For the experiment at Michigan, this variation was around 16.3 (0-
30 cm) to 0.79 cm cm'3 (120 cm) with a slight increase in RLD at 75-90 and 90-
105 cm.

An increase in VWC and AWC followed the increase in clay content in
both places and this is probably a consequence of the capacity of water retention
by clay minerals. Nevertheless, the increase in available water content could also
be a consequence of roots bypassing some soil regions leaving water behind

during root growth or connected to the clay content. It was reported by Dwyer et

69

al. (1988) that root depth and distribution were influenced by AWC and soil
physical characteristics, and that the maximum root depth was inversely
proportional to AWC.

The range (R), sill (Co+C), nugget effect (Co), and nugget to sill ratio
(Co/Co+C) for AWC and RLD in all soil depths in both places are depicted on
Table 2.4. The nugget to sill ratio, expressed as percentage, can be regarded as
a criterion to classify the spatial dependence of the variables observed in this
study. If this ratio is less than 25%, the variable has strong spatial dependence; if
the ratio is 25-75%, the variable has moderate spatial dependence; otherwise,
the variable has weak spatial dependence (Chien, et al.,1997).

These results show that AWC had a strong to weak level of spatial
dependence for C1, with values ranging from 20% to 99%. A weak spatial
dependence in CZ was also observed. For the field experiment, the results for
AWC were similar to the results observed in C1 in the greenhouse experiment.
This degree of the spatial dependence varied from 0.9% to 83%. Root length
density also showed moderate to weak spatial dependence both in C1 and in
field experiment. A weak dependence was observed in C2 and at 105-125 cm for
RLD, probably because the low amount of roots found. Usually, the spatial
variability of soil properties is related to intrinsic and extrinsic factors. Intrinsic
factors (soil formation factors, parent material) will determine strong spatial
dependence; on the other hand, extrinsic factors (soil management practices) will

determine weak spatial dependence. Nevertheless, in soil properties it is normal

70

to observe a moderate spatial dependence (Cambarella, 1994; Chien, et al.,
1997)

The range for AWC and RLD varied from 5 cm to about 45 cm in the
greenhouse experiment and from 3 cm to 9 cm in the field experiment (Table
2.4). Usually, the Range observed for some soil properties varies from few
meters to hundreds of meters. For instance, there are reports where sand range
varied from 5 m up to 1000 m, clay range was 723 m, silt 1290 m, and soil water
content with range about 16m (Warrick, 1986; Chien et al., 1997). But,
surprisingly, in this study this variation was about centimeters, and this could be
due to the scale adopted. Depending on the objectives and scale of the
observation needed, these results could be considered as part of the nugget
variation in other studies of spatial variability of soil properties. However, in this
study the spatial variability could be detected at this small scale. Above the
observed R there was no spatial relationship. When linear models were used, R
showed high values, meaning that the spatial dependence could not be detected
in small scale. For example, RLD for soil layers at 30 and 45 cm depth in C1, the
difference between the minimum and maximum values and the standard
deviation observed were high (Table 2.1). This could be caused by some
samples presenting high values of RLD when compared to the average of all
samples.

The values for sill, range and nugget for AWC and RLD were obtained by
using the GSPIus+ software (Robertson, 1998) and then used as input in Surfer

to obtain the krigged surface maps (Figures 2.2, 2.3 and 2.4). These values were

71

obtained after semivariogram models were fitted to pairs of data observed at lag
distances for each variable at each soil layer in both places. For the greenhouse
experiment, 98% of the models fitted were spherical, and 2% were linear. For the
field experiment all models were spherical. These models were fitted
automatically by the GSPIus+ (Robertson, 1998). The user can, in most
situations, change the values obtained in the automatic fit; however, this would
be time consuming and sometimes worthless considering that the model
presented automatically by the program was the one with the best fitted value. To
determine how well these models were fitted to the semivariogram, an indicative
of goodness fit (lGF)(without units) was used, where values close to zero
indicated better fit of the model.

To make the relationship between AWC and RLD more evident, cut-off
values were attributed to RLD and AWC. RLD received a value equal to 1cm
cm’3 and the cut-off value for AWC was 10%. This means that values above the
cut-off values for both variables were made equal to the cut—off values
established. This was done because some samples presented results extremely
high and others extremely low, which made the comparison among surface maps
extremely difficult.

In a broad sense these maps show that AWC was higher at deep layers
than that observed at upper layers. In the greenhouse experiment, AWC below
75 cm depth for C1 was higher than 10%. There are some patches with values
below 10% but these patches follow some patterns in their distribution for all

layers. Exception made at 120 cm where AWC is higher than 10% in almost the

72

entire layer analyzed. This pattern was probably caused by two reasons. One
reason is that AWC was considered to be the amount of water above the LL. As
it was said before, the LL was calculated by using the clay content for each
individual sample. Analyses of correlation performed in the previous chapter
show that a strong correlation between clay content and LL occurred, with R2
values around 96%. This value means a positive correlation, or an increase in
clay content means an increase in LL. In some way, this irregular distribution of
AWC is connected to the amount of clay. The same analogy was adopted when
the analysis for C2 and the field experiment.

The second reason, is a function of the amount of and root distribution in
these layers. Soil layers presenting high values of RLD had less AWC. It means
that roots extracted all water they could independent of their distribution on a
specific layer. This can be seen in Figure 2.2, for layers 30 and 45 cm, where
values for AWC are around 0 to 4% and RLD is as higher as the cut-off value.
However, below 45 cm the pattern of AWC distribution is function of root
distribution. Here we can see that the roots bypassed soil layers leaving water
behind, without being extracted. This characteristic is evident in soil layers at 60,
75, 90 and 105 cm. At 60 and 75 cm depth, spots with high values of AWC were
found outside the zones of high concentration of roots. Arrows in these surface
maps indicate the spots where high concentrations of roots are opposed to low
values of AWC. At 90 and 105 cm the evidence of bypassing is stronger than that
observed in the previous layers. The surface map at 90 cm shows some spots,

indicated by arrows, where the bypassing is occurring. This map clearly shows

73

that in some soil regions roots are clumped and that the water extracted was high
in these regions. However, at short distances far from these clumps, AWC
reaches values around 8-9%. The same thing is observed in the surface map at
105 cm. The explanation for this root clumping may be related to the clay
content. In the greenhouse experiment, for C1, clay content was different at each
soil layer (Table 2.1). The range observed between the minimum and maximum
values of clay content, for instance at 900m depth, varied from 3 to 27%. Such
variation is also observed at other layers. If these samples were taken in a grid
pattern at each soil layer, one can say that clay content presented spatial
variability. Thus, this variability could have affected root growth throughout soil
matrix.

The relationship between AWC and RLD was presented in Figure1.10. In
this figure the R2 is low, meaning a poor correlation; but it was not possible to
give a good interpretation without looking at the surface map of these variables.
Now, it is evident why the correlation was poor; roots were clumped and there
was a spatial variability for AWC.

An interesting feature is observed in Figure 2.2 for soil depths from 60 to
105 cm. The classical statistical analysis showed that the amount of roots
decreased within soil depths (Table 2.1). One can observe that in these layers
the variation in the amount of roots occurred in one specific zone of the soil and
that this zone is located at least at the same position in all soil depths.
Considering the variation observed in RLD in these layers we can say that roots

grew clumped and bypassed the entire soil layer located at the depth of 75 cm;

74

thus, not absorbing water. The same kind of analysis adopted for C1, was used
in CZ and similar results were observed.

For the field experiment, the same evidences of root clumping and soil
bypassing were observed for soil layers at 45 and 60 cm. The RLD and AWC
were spatially distributed and zones of high water extraction were located at
same position where the high values of RLD were found (Figure 2.3). Below 60
cm, the amount of roots decreased and the AWC was more disperse, without
following a specific pattern.

All surface maps discussed here, for both places, indicated that there is
strong evidence to explain why roots were clumped and bypassed some soil
regions. In this study, by sampling the soil in a grid pattern, it was possible to see
that the same variable differed by several units at same soil depth. Usually it is
assumed that macropores, fissures or other natural openings in the soil cause
non-uniform distribution of roots. Based on the results presented here, clay and
water content had an important role in the root clumping and distribution at each
soil depth. The non-uniform distribution of clay observed in this case affected root
distribution by directing root growth toward soil zones where water content was
lower than that observed in regions with high clay content. Generally, clay can
hold high amount of water; however, if clay content is so high that can cause
impediment to root growth, roots will not penetrate that soil region and will
bypass it leaving water without being extracted. For root to elongate, the
mechanical impedance of the soil against the cross section of the root must be

less than the pressure exerted by the root itself (Bennie, 1996). If root growth

75

pressure is higher than the pressure or resistance that soil and cell wall is
exerting, then root will grow and penetrate that region; otherwise they will stop or
branch in another direction. Depending on the impedance offered by the soil in
determined regions, roots will grow together and will be clumped in regions were
the soil penetration and even water extraction is easier. In this way, a natural
clumping will occur in that region with several consequences, such as, by drying
the soil region on the vicinity of roots as we saw in all surface maps. This could
increase the soil resistance to water movement from other parts of the soil
towards the roots. In this case, there is water in the soil around roots, but its
hydraulic conductivity is low to allow water movement and consequently no water
extraction can occur.

Amato (1991) showed that roots could grow around soil peds but not
penetrate them. In her work, it was seen that the interior of these peds held a
considerable amount of water inside, but this water was not available for roots to
absorb it, resulting in an increase in water stress. Thus, a considerable amount of
water was left, resulting in a localized region of soil water gradients. Thus, it is
true that some regions in the soil can hold more or less water, which affects the
volume of water extracted by roots. To survive in this situation, plants need to
produce more roots at deep layers or to increase water absorption in soil regions
were water still available. Nevertheless, new roots at deep layers or at different
soil region do not mean a guaranty of water supply. The amount of roots
produced can be insufficient to absorb water and supply the total demand of

water by plant as a function of transpiration (Passioura, 1983). Also, water could

76

be held tightly in the soil by clay particles at potentials higher than that plant

could overcome.

Summary and Conclusions

In this study, the analysis of the spatial variability of AWC and RLD for the
greenhouse and field experiments, was the determining factor to demonstrate
that roots can bypass soil regions and leave water without being extracted. As a
consequence of the bypassing, plants will be affected by water deficit even when
a supply of water is still available in the soil. Similar results may be achieved in
other circumstances and for different soils; however, studies like this one in other
soils need to be repeated to verify whether or not this phenomenon can be

manifested again.

77

References

Ahuja, LR, and DR. Nielsen. 1990. Field Soil-water relations. p. 143-190. In
B.A. Steward and DR. Nielsen (ed.) Irrigation of Agricultural crops. Agron.
Monogr. 30. ASA, CSSA and SSSA, Madison, WI.

Amato, M. 1991. Spatial distribution and water uptake of roots in structured
growth media. Ph.D. diss. Michigan State Univ., East Lansing.

Bennie, ATP. 1996. Growth and mechanical impedance. p. 453-470. In Yoav
Waisel et al. (ed.) 2nd ed. Plant Roots: The Hidden Half. Marcel Dekker,
New York, NY.

Brown, D., and HD. Scott, 1984. Dependence of crop growth and yield on root
development and activity. p.101-136. In Barber et al. (ed.) Roots, Nutrient
and Water Influx, and Plant Growth. ASA Spec. Publ. 49. SSSA, Madison,
WI.

Cambarella, C.A,T.B.Moorman, J.M. Novak, T.B. Parkin, K.L. Karlen, R.F. Turco,
and AB Konopka. 1994. Field-scale variability of soil properties in Central
Iowa soils. Soil Sci. Soc. Am. J. 58:1501-1511.

Chien, Y., D.Lee, H. Guo, and K.Houng. 1997. Geostatistical analysis of soil
properties of mid-west Taiwan soils. Soil Sci. 162(4):291-298.

Dwyer, L.M., D.W. Stewart, and D. Balchin. 1988. Rooting characteristics of corn,
soybeans and barley as a function of available water and soil physical
characteristics. Can. J. Soil Sci. 68:121-132.

Golden Software, 1997. Surfer V6.0. for Wrndows. User’s Guide. Golden
Software Inc., CA.

Goovaerts, P. 1994. Study of spatial relationships between two sets of variables
using multivariate geostatistics. Geoderma 62:93-107.

Goovearts, P. 1992. Factorial kriging analysis: a useful tool for exploring the
structure of multivariate spatial soil information. J. Soil Sci. 43:597-619.

Goovearts, P. 1997. Geostatistics for natural resources evaluation. Oxford
University Press, New York, NY.

lsaaks, E.H., and RM. Srivastava. 1989. An Introduction to applied geostatistics.
Oxford University Press, New York, NY.

78

Jones, C.A., W.L. Bland, J.T. Ritchie, and JR. Williams. 1991. Simulation of root
growth. p.91-120. In John Hanks and J.T. Ritchie (ed.) Modeling Plant and
Soil Systems. Agron. Monogr. 31. ASA, CSSA and SSSA, Madison, WI.

Journel, AG, and C. Huijbregts. 1978. Mining Geostatistics. Academic Press,
New York, NY.

Kramer, P.J., and J.S. Boyer. 1995. Water relation of plants and soils. Academic
Press. San Diego, CA.

Kuiper, P.J.C., D. Kuiper, and J Schuit. 1988. Root functioning under stress
conditions: An Introduction. Plant Soil 111(249-253).

Matheron,G. 1973. The intrinsic random functions and their
applications.Adv.Appl.Prob.5:239-465.

Passioura, J.B. 1991. Soil structure and plant growth. Aust. J. Soil Res. 29:717-
728.

Robertson, G.P. 1998. GS+: Geostatistics for the Environmental Sciences.
Gamma Design Software, Plainwell, Michigan, USA.

Russel, RS. 1977. Plant root systems: Their function and interaction with the
soil. McGraw-Hill. England.

Saddiq, M.H., P.J. Wrerenga, J.M. Hendrickx, and MY. Hussain. 1985. Spatial
variability of soil water tension in an irrigated soil. Soil Sci. 140(2):126-132.

SAS Institute 1990. SAS/STAT user’s guide. Version 6. SAS Inst.,Cary, NC.

Tardieu, F. 1989. Root system response to soil structural properties: micro and
macro scale. p. 153-172. In W.E. Larson et al. (ed.) Mechanics and
related processes in structured agricultural soils. Kluwier. London,
England.

Trangmar, B.B., R.S. Yost, and G. Uehara. 1985. Application of geostatistics to
spatial studies of soil properties. Adv. Agron. 38:45-94.

Trangmar, B.B., R.S. Yost, and G. Uehara. 1986. Spatial dependence and
interpolation of soil properties in West Sumatra, Indonesia; l.Anisotropic
variation. Soil Sci. Soc. Am. J. 50:1391-1395.

Tsegaye, T. and R.L.Hill. 1998a. Intensive tillage effects on spatial variability of
soil physical properties. Soil Sci.163(20):143-154.

79

Tsegaye, T., and R.L. Hill. 1998b. Intensive tillage effects on spatial variability of
soil test, plant growth, and nutrient uptake measurements. Soil Sci.
163(2):155-165.

Vieira, S.R., J.L.Hatfield, D.R. Nielsen, and J.W. Biggar. 1983. Geostatistical
theory and application to variability of some agronomical properties.
Hilgardia. 51(3):1-41.

Warrick, A..,W D. E. Myers, and D. R. Nielsen 1986. Geostatistical methods
applied to soil science. p. 53- 82. In Klute(ed. )?. Methods of soil analysis.
Part1. 2Ind ed. Agron. Monogr..9 ASA and SSSA, Madison, WI.

Yost, R..,S G. Uehara, and R..L Fox. 1982. Geostatistical analysis of soil

chemical properties of large land areas. I. Semi-variograms. Soil Sci. Soc.
Am. J. 46: 1028- 1032.

80

Table 2.1. Descriptive statistics for root length density (RLD), volumetric water
content (VWC), available water content (AWC), clay, sand, silt and lower limit (LL)
within soil depths. Greenhouse experiment (C1) at Michigan State University.

 

 

 

 

 

 

 

 

 

 

 

Depth Min. Max. Mean Variance SD

(cm)

RLD (cm cmj)

0-30 0.67 16.30 5.60 a 12.35 3.51
30-45 0.00 12.67 3.02 b 7.60 2.76
45-60 0.00 2.14 0.35 c 0.22 0.47
60-75 0.00 0.86 0.06 c 0.02 0.14
75—90 0.00 1.74 0.20 c 0.15 0.38

90-105 0.00 2.20 0.15 c 0.14 0.37
105-120 0.00 0.79 0.03 c 0.02 0.13
VWC (%)

0-30 1.87 19.26 8.62 f 7.76 2.79
30-45 2.41 20.09 9.56 e 6.03 2.45
45-60 4.46 14.40 8.43 f 2.70 1.64
60-75 4.09 22.52 10.35 d 20.85 4.57
75-90 4.16 19.91 13.56 c 14.88 3.86

90-105 8.82 23.77 15.16 b 7.40 2.72
105-120 10.98 26.32 19.14 a 5.98 2.45
AWC (%)

0-30 0.00 3.69 0.54 f 0.81 0.90
30-45 0.00 6.77 1.16 f 1.81 1.34
45-60 0.00 1.65 0.20 f 0.17 0.42
60-75 0.00 13.15 2.51 d 14.30 3.78
75-90 0.00 10.56 4.60 c 10.27 3.20

90—105 0.00 14.56 5.99 b 6.88 2.62
105-120 3.66 17.00 10.12 a 4.92 2.21
Clay (%)

0-30 1.22 10.76 6.55 c 3.26 1.81
30-45 1.66 8.11 5.17 d 1.66 1.29
45-60 2.46 13.52 5.60 c 5.02 2.24
60-75 0.41 26.96 9.33 b 52.47 7.24
75-90 3.34 27.36 13.70 a 31.76 5.64

90-105 3.34 19.97 10.49 b 10.65 3.26
105-120 9.08 18.57 13.03 a 3.05 1.75
Sand (%)

0-30 68.21 77.28 73.89 a 3.68 1.92
30-45 72.82 77.40 74.59 a 0.82 0.91
45-60 70.54 82.12 75.05 a 1.98 1.41
60-75 47.98 80.08 67.63 b 94.15 9.70
75-90 53.07 82.22 64.56 c 57.19 7.56

90-105 50.99 75.90 63.35 c 34.87 5.91
105-120 44.55 74.28 50.60 d 12.55 3.54

 

81

Table 2.1. (cont’d)

 

 

 

 

Silt (%)

030 15.61 24.94 19.56 d 3.74 1.93
3045 16.99 23.42 20.24 d 1.82 1.35
4550 10.06 23.23 19.35 d 4.68 2.16
60-75 15.51 35.66 22.88 c 17.55 4.19
75-90 8.66 33.66 21.73 c 33.30 5.77
90-105 17.82 41.93 26.17 b 39.20 6.26

105120 13.19 42.82 36.37 a 14.04 3.75
LL(%)

0-30 8.84 9.10 8.98 c 0.0029 0.05
3045 8.85 9.03 8.95 d 0.0012 0.03
45-60 8.87 9.17 8.96 d 0.0031 0.06
60-75 8.82 9.54 9.07 b 0.0385 0.20
75-90 8.90 9.55 9.18 a 0.0233 0.15
90105 8.90 9.35 9.09 b 0.0078 0.09

105-120 8.81 9.31 9.16 a 0.033 0.06

 

82

Within columns. means followed by the same letter are not significantly different
Duncan at 0.01 probability.

Table 2.2. Descriptive statistics for root length density (RLD), volumetric
water content (VWC), available water content (AWC), clay, sand, silt, and
lower limit (LL) within soil depths. Greenhouse experiment (CZ) at
Michigan State University, Ml.

 

Depth Min. Max. Mean Variance SD
(cm)

 

RLD (cm cm 5)

 

60 0.00 3.51 0.07 a 0.12 0.34
75 0.00 5.22 0.12 a 0.28 0.53
90 0.00 3.60 0.07 a 0.13 0.37

 

 

 

vwc (%)

60 2.99 13.90 8.26 c 2.74 1.65

75 4.54 17.89 9.91 b 11.81 3.43

90 3.12 25.91 12.93 a 23.88 4.88
ch (%)

 

60 0.00 3.51 0.55 c 0.85 0.92
75 0.00 5.22 2.33 b 0.28 2.49
90 0.00 17.56 5.10 a 17.97 4.23

 

 

 

. Clay (%)
60 0.84 9.60 5.71 c 3.37 1.83
75 0.82 20.09 8.36 b 25.67 5.06
90 2.82 25.51 12.58 a 35.08 5.92
Sand (%)

 

60 69.14 77.89 73.81 a 2.45 1.56
75 52.39 87.26 71.64 ab 104.17 12.00
90 48.64 88.58 70.53 b 112.94 10.62

 

Silt (%)

 

60 16.16 24.69 20.46 a 2.90 1.70
75 10.11 95.65 19.99 a 87.96 9.37
90 5.48 32.74 16.88 b 35.25 5.93

 

LL(%)

 

60 8.29 8.41 8.36 c 0.0006 0.02
75 8.29 8.55 8.39 b 0.0044 0.06
90 8.32 8.62 8.45 a 0.006 0.07

 

Within columns, means follcwed by the same letter are not significantly different
Duncan at 0.01 probability.

'83

Table 2.3. Descriptive statistics for volumetric water content (VWC),
available water content (AWC), root length density (RLD), clay, sand,
silt and lower limit (LL) within soil layers. Field Experiment at
Maricopa Agric.Center, AZ.

 

 

 

 

 

 

 

 

 

 

 

Depth(cm) Min. Max. Mean Variance SD
VWC(%)
30-45 6.38 12.88 9.03 c* 2.35 1.53
45-60 5.23 16.90 12.53 b 4.63 2.15
60-75 6.09 16.72 12.80 b 3.73 1.93
75-90 9.32 19.05 13.10 b 2.83 1.68
90—105 7.37 23.36 14.71 a 8.38 2.89
105-125 10.25 23.06 14.96 a 7.23 2.69
AWC(%)
30-45 0.00 3.29 0.55 c 0.80 0.89
45-60 0.00 10.00 5.14 a 4.88 2.21
60-75 0.00 9.37 5.71 a 3.79 0.47
75-90 0.73 11.24 5.16 a 3.29 1.81
90-105 0.00 15.33 5.55 a 8.31 2.88
105-125 0.00 9.59 4.37 b 5.36 2.31
RLD(cm cm 3)
30-45 0.00 3.72 0.92 a 0.48 0.70
45-60 0.00 1.74 0.23 b 0.17 0.41
60-75 0.00 0.85 0.03 c 0.02 0.13
75-90 0.00 0.43 0.01 c 0.01 0.04
90-105 0.00 0.89 0.01 c 0.01 0.09
105-125 - - - - - -
CLAY (%)
30-45 6.21 11.70 8.82 b 1.62 1.27
45-60 2.53 7.95 5.07 d 1.26 1.12
60-75 1.68 7.33 4.30 e 1.24 1.11
75-90 3.35 8.97 6.26 c 1.67 1.29
90-105 3.75 12.47 9.17 b 3.43 1.85
105-125 7.29 19.89 12.49 a 10.25 3.20
SAND (%)
30-45 74.19 80.48 77.54 c 2.09 1.45
45-60 76.38 85.28 80.21 a 3.20 1.79
60-75 49.14 94.94 80.42 a 20.02 4.47
75-90 69.79 88.32 80.22 a 7.27 2.70
90-105 59.94 83.32 78.75 b 9.52 3.09
105-125 62.79 82.18 74.81 d 30.29 5.50

 

84

Table 2.3. (cont’d)

 

 

 

 

SILT (%)
30.45 10.57 17.27 13.64 b 2.23 1.49
45-60 10.13 19.81 14.72 a 3.34 1.83
60-75 0.43 46.19 15.28 a 21.28 4.61
75-90 6.77 22.54 13.52 b 6.01 2.45
90105 6.11 29.09 12.07 c 9.55 3.09
105125 6.44 23.24 12.70 c 9.52 3.08
LL(%)

3045 7.91 10.25 9.02 b 0.29 0.54
45-60 6.34 8.65 7.42 d 0.23 0.48
60-75 5.98 8.39 7.10 e 0.22 0.47
75—90 6.69 9.10 7.93 c 0.30 0.55
90105 6.86 10.58 9.17 b 0.62 0.79
105125 8.37 13.74 10.59 a 1.86 1.36

 

* within columns, means followed by the same letter are not

significantly different
Duncan at 0.01 probability

85

Table 2.4 Results for geostatistical analysis in the greenhouse experiment
at Michigan State University and field experiment at Arizona. Available

water content (AWC) and root length density (RLD) within soil layers.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Depth Model Range Co Co+C Co/(Co+C) IGF
(cm) (cm) (%)
AWC (%)
30 SPH 12.73 0.068 1.074 93.00 3.54E-01
45 SPH 3.65 0.001 0.967 99.00 2.81 E-01
60 SPH 4.11 0.164 1.033 84.00 7.28E-02
75 SPH 29.76 1.380 5.244 20.83 1.02E-03
90 SPH 66.45 0.114 1.820 93.80 6.32E-03
Michigan 105 SPH 44.68 0.483 1.188 59.30 6.78E-03
C1 120 SPH 10.85 3.397 1.927 63.81 5.18E-03
RLD (cm cm 3)
30 LN 38.8 0.45 1.278 64.80 4.55E-01
45 LN 39.21 1.036 1.483 30.10 ' 2.06E-01
60 SPH 11.59 0.153 1.045 85.00 1.53E-01
75 SPH 5.11 0.151 0.996 84.80 2.19E-05
90 SPH 8.22 0.140 1.007 86.10 1.92E-01
105 SPH 9.65 0.105 1.090 90.40 8.16E-01
120 SPH 5.44 0.001 0.950 99.00 3.57E-01
AWC (%)
60 SPH 5.86 0.171 1.043 83.30 1.22E-01
Michigan 75 SPH 56.10 0.069 1.591 95.60 2.69E-01
C2 90 SPH 38.61 0.175 1.276 86.00 3.39E+00
RLD (cm cm 3)
60 SPH 12.91 0.001 0.954 99.00 8.93E-01
75 SPH 10.77 0.005 0.953 99.00 5.93E-01
90 SPH 13.08 0.001 0.963 99.00 6.45E-01
AWC (%)
45 SPH 4.52 0.198 1.044 81.00 . 2.58E-02
60 SPH 4.11 0.199 1.004 80.20 6.12E-02
75 SPH 3.43 0.245 1.004 75.60 6.02E-06
90 SPH 2.77 0.165 0.996 83.40 1.31 E-01
Arizona 105 SPH 3.02 0.128 0.985 0.87 8.32E-01
RLD (cm cm 3)
45 SPH 5.05 0.069 0.984 93.00 1.02E-01
60 SPH 5.72 0.086 1.025 91.06 1.16E-01
75 SPH 3.53 0.174 1.005 82.70 8.25E—02
90 SPH 9.27 0.368 2.27 83.80 5.47E-01
105 SPH 3.71 0.001 0.993 37.00 4.03E-01

 

LN - linear model, SPH -spherical model, Co-nugget effect, Co+C - sill, C-
structural variance and IGF- Indicative Goodness of Fit.

86

Range (R)

 

 

 

 

 

 

 

d + I

+ .. A
Structural
Variance (C)

Sill
(Co+C)
1
il l L l 1 1 Nugget (Co)
Lag (h)

Figure 2.1. Hypothetical Semivariogram.

87

(cm cm3)

 

 

10 20 30 40 10 20 30 40

Figure 2.2. Root length density (RLD) (cm cm3) and available water content (AWC) within soil
depths. (C1). Greenhouse experiment at Michigan State University, MI.

88

 

120

3%

20—

1%

 

‘_T

 

Figure 2.2. (Cont'd)

89

(cm cm3)

 

 

10 20 30 40

 

Figure 2.3. Root length density (RLD)(cm cm3) and available water content (AWC)
within soil depths. Greenhouse experiment (CZ) at Michigan State University, MI.

90

 

 

 

Figure 2.4. Root length density (RLD) (cm cm3)and available water content (AWC)
within soil depths. Field experiment at Maricopa, AZ.

91

RLD
25—
90

 

25—
1 05
20~

5-4

 

 

 

Figure 2.4. (Cont'd)

l
20

92

ch

 

Chapter 3

Root water extraction by com plants under conditions of limited supply of
water.
Introduction

Water is one of the most important elements for plant growth and
development in conditions of water deficit. The capability of plants to survive in
limited conditions of water supply will largely depend on the uniformity and depth
of its root system (Kozlowski, 1968; Passioura, 1983; Kramer and Boyer, 1995).
Data in literature show that plants can leave an appreciable amount of water in
the soil without being extracted even in conditions of water deficit (Passioura,
1983; Robertson et al., 1993a; Hamblin and Tenant, 1987; Bland and Dugas,
1989). This means that plants will become dormant or die even when a supply of
water still available in the soil. The reasons for this phenomenon were not fully
understood, and several researchers pointed out that the causes would be
related to: (i) root clumping (Passioura, 1983; Tardieu, 1988), (ii) uneven root
distribution of roots in the soil profile (Robertson et al., 1993b), (iii) onset of the
crop maturity before roots make use of the deep water in the soil (Passioura,
1983; Robertson et al, 1993a), or (iv) low root density at deep layers (Passioura,
1983). It is argued that if roots are not present in some regions of the soil, these
regions will not be adequately explored and an incomplete extraction of water
may occur (Ritchie, 1986).

Simulation of water extraction is the extreme importance in water models

to determine the amount of water extracted by plants (Ahuja and Nielsen, 1990).

93

Most of these models are based on the relationship between the amount of roots
present and water extraction by roots (Molz, 1981; Hector, 1993); however, the
importance of using roots as a parameter in such models has being questioned
by several researchers (Hamblin, 1985; Bland and Dugas, 1989; Dardanelli,
1997)

The reason is that these models assume roots as being uniformly
distributed in the soil or decreasing exponentially with root depth in the soil
(Dwyer, 1988; Klepper, 1990). Nevertheless, due to the heterogeneity of the soil
environment (Russel, 1977; Hamblin, 1985; Zobel, 1996), to the intrinsic plant
factors that determines root growth, and to the interaction among soil-root, the
assumption of uniform root distribution is not adequate. The validity of using
roots to explain water depletion from the soil may be inappropriate (McIntyre et
al., 1995; Lacape et al., 1998).

The adequate solution would be the development and use of water
extraction models that are not root based to describe the pattern of water
depletion by plants (McIntyre, 1995). A functional model of water extraction was
proposed by Ritchie et al. (1999). This model assumes that roots reached a
threshold in each soil layer, that water demand is higher than supply and that
lower limit was precisely measured or estimated. Its simulates water depletion in
a soil profile in a daily basis. An empirical constant K, was included in this model,
and represents the fraction of extractable water in a specific soil layer per day.

The objectives of this work were: (i) to estimate from soil water depletion

curves the empirical constant K for corn plants growing in water deficit

94

conditions, (ii) to determine the beginning of the maximum water extraction rate
at each soil layer and, (iii) to simulate water extraction by using the functional

model proposed by Ritchie et al. (1999).

Material and Methods

This experiment was conducted from DOY 336 in 1997 to DOY 81 and 96
in 1998. Two steel cylinders with dimensions of 60cm in diameter, 180 cm in high
and 0.6 cm of wall thickness were pushed into the soil to obtain undisturbed soil
samples. The soil was collected at Michigan State Research Farm, MSU, East
Lansing and is classified as Metea B (Loamy, mixed, mesic Arenic Hapludalfs),
according to USDA, Soil Conservation Service. More details are given in chapter
one of this thesis.

The cylinders with contained soil were removed and placed in greenhouse
at Pesticide Research Center, Michigan State University, East Lansing, and
covered with plastic to maintain the initial moisture content. Three corn seedlings
(variety Cargill 333), with 4 leaves were transplanted on DOY 336 in 1997 in C1
and 20 days later in C2. Plants in each cylinder were fertilized with a NPK
formula (19-19-19) at rate of 500 kg ha'1 at transplanting day and side fertilized
with nitrogen (100 kg ha") 20 days later after transplanting. After transplanting,
plants were watered daily until DOY15 in 1998, and water was withheld
aftenlvards.

Daily minimum and maximum temperatures were obtained from a

minimum data set recorder (Model Ll-1200, Li-cor, Lincoln, NE) located inside

95

the greenhouse. Temperature data from DOY 23 in 1998 to DOY 91 in 1998 are
shown in Figure 1.2.

Soil water content was measured by using neutron scattering technique
(CPN, Model 503 DR) during the plant growth period in limited supply of water.
One access tube (aluminum, 50 mm id.) was seated in the center of each
cylinder. Readings were made using a 32 3 count at 15 cm intervals up to a
depth of 1.20 m, every 48 hours during the drying cycle.

The neutron probe was calibrated by using two steel barrels containing
soil. An equation fitted to the neutron-probe readings in these two barrels, one
with dry soil and the other one with wet soil, was used to calculate the volumetric
water content in this experiment. Although the soil used in these barrels is not the
same soil used in this experiment, it was considered that the slope found in this
equation was constant for the neutron-probe and that only the intercept varied as
function of the soil type where the probe was being used. By keeping this slope
fixed, a new intercept for the calibration equation was found by using the last
neutron-probe readings and the data of volumetric water content obtained from
the TDR readings. This was the methodology used to determine the volumetric
water content considering that there was not a specific neutron probe calibration
where the experiment was performed.

A model that does not take into account root length density (RLD) (cm
cm'3), but generalizations from measured soil water content changes, was used
to simulate the rate of water extraction by com plants. The model proposed by

Ritchie et al. (1999), assumes that roots reached a minimum density for each soil

96

layer to extract water at a minimum rate. Also it assumes that the demand of
water by transpiration is greater than the water supply in the root zone, and that
the lower limit of water content can be precisely measured or estimated. The

equation that produces the daily change in water content is:

96 = 911-1 - (Gd-1 - GLL) * K (1)

where, ad is the volumetric water content in a given day (cm3cm'3); 0.1-1 =
volumetric water content in the previous day (cm3cm‘3); GLL = volumetric water
content at lower limit (cm3cm'3); and K(day’1) = an empirical constant that
represents the maximum fraction of available water extraction within one day. For
most crop situations this K constant is assumed to be approximately 0.1 (Ritchie
et al., 1999). Based on data found in the literature this model is the only one at
present time that does not depend on the amount of roots, that is why it was
used in this experiment.

The measured soil water content obtained with the neutron probe gauge
was used to determine the constant K for all soil layers in both cylinders. This
value was obtained by curve fitting the logarithm of the difference between
volumetric water content and the lower limit for each soil layer.

For the upper layers, LL was considered equal to the lowest value of VWC
obtained with the neutron probe in these layers. A relationship between VWC

and clay content observed at upper layers was established, and the resulting

97

equation from this relationship was used to estimate LL at deep layers in both
cylinders (Figure 1.11).

The condition in the model that water demand be greater than the water
supply In the root zone can be confirmed by the daily evapotranspitration

(mmday’1) values in Figure 1.7.

Results and Discussions

The time course for changes in WVC (cm3cm'3) used to determine K for all
soil layers in both cylinders (C1 and C2) is depicted on Figures 3.1 and 3.2. The
data showed that there was an exponential decay in water change in the upper
layers from 15 to 75 cm in C1; however, below that, this exponential decay was
not so evident. Results also showed that some water extraction was occurring at
deep layers but at low rates. Below 75 cm depth the change in water content was
not so steep. In both cases, a low density of roots at deep layers or a bypassing
of some soil regions may have contributed to the low water extraction observed
below 75 cm.

The log transformation of the difference between AWC and LL for each
soil layer as a function of time for water extraction is presented in Figures 3.3.
Curves with different slopes fit quite well the measured data during rapid
extraction when the time of measurement for each time was biased to the date
when water decay became exponential. As mentioned by Ritchie et al. (1999)
this bias time is related to the time when roots obtain sufficient density and

uniformity to extract water at it is maximum rate during the period of limited

98

 

supply. The slopes, K, determined here (Table 3.1) at the upper layers are very
close to the value of K=0.1 mentioned in the functional model proposed by
Ritchie et al. (1999).

In the upper layers, the K values ranged from 0.07 to 0.09 for C1 and C2.
These values are approximately similar to those values obtained for cotton
(Bland and Dugas, 1989; Lacape et al.,1998), pearl millet (McIntyre et al. 1995),
soybean, sunflower, maize and peanut (Dardanelli, 1997). However, for deep
layers K values were lower than 0.07. K values as low as 0.03 was found in C1,
and K value ranging from 0.008 to 0.025 was found in C2 (Table 3.1). Ritchie et
al. (1999), attributed values of K less than 0.1 to root clumping. Soil constraints,
such as high clay content would cause cracks or fissures where roots could grow
clumped. Once roots are clumped, a considerable bypassing by the root system
in the soil volume can occur which would lead to an incomplete water extraction
and consequently causing K to be low. Another possible reason for low K values
is the low root density at deep layers, when further downward root growth ceased
(Ritchie, et al., 1999).

In the chapter 2 of this thesis, it was observed that roots were clumped
and bypassing soil regions. This bypassing was confirmed here by the low values
of K observed at deep layers.

The maximum water extraction rate in soil layers below 45 cm depth
occurred after DOY 45 in C1 (Table 3.2). This means that when the period of
water deficit began in DOY 45 (Figures 1.4, 1.5 and 1.6) corn plants had already

extracted water at the maximum rate they could at 30 cm but this rate decreased

99

with soil depth. It is most likely that roots were present in these layers before
plants had been affected by water deficit on DOY 45, but the maximum rate of
water extraction began with only few days of difference within each other at these
layers. The time interval when maximum water extraction began in these layers
varied from 1 to 3 days in C1, and about 9 days in C2. Plants could not extract
water at same rate below 45 cm because roots were clumped and also because
the low root densities at deep layers. As a consequence an incomplete water
extraction from deep layers occurred.

In CZ, it was not possible to get information about the amount of roots
present in the first 30cm depth. Nevertheless, considering the K value of 0.09
(Table 3.1), water extraction was at the maximum rate in this layer (Figure 3.3).
This means that this soil layer was probably fully colonized by roots and that
almost all existent water was extracted. Based on the graphs of water depletion
(Figure 3.2), the beginning of water extraction is not clear; but the maximum rate
of extraction began by DOY 50. For soil depth from 45 to 75 cm, the maximum
water extraction rate began around DOY 59 (Figure 3.4), almost the same period
when water deficit began affecting plants in C2 (Figure 1.5 and 1.7). Also, it was
noticed that the beginning of water extraction for these layers began around DOY
27, thus taking almost 30 days for roots colonizing these layers. However, based
on the Table 3.1 and Figure 3.2, this colonization was not complete and K values
for these layers were 0.065. In this situation it is probable that the low root
density caused the incomplete water extraction. For soil depth from 90 to 120

cm, some water extraction was verified after DOY 65; however, K was

100

consistently low, and the probable cause was the low root density. The severe
water deficit effects on these plants impaired root growth and at an appreciable
rate and density to fully extract water from deep layers.

The data presented until now show that the rate of water extraction was at
its maximum for upper layers when root density was high enough to allow that.
For deep layers, K values, were consistently lower than 0.01, as previously
suggested by Ritchie et al. (1999). For both cylinders, the causes of K being low,
seems to be associated to a low root density, root clumping and also to root
system bypassing soil regions.

Using the K values found and the time indicated in Table 3.2 for C1 and
C2, simulations of water extraction were performed for some selected soil layers.
The water extraction simulations performed for soil layers at 30, 90 and 120 cm
depth for both cylinders indicated that the model proposed by Ritchie et al.
(1999) was adequate to describe water extraction by roots (Figure 3.4). In this
study, K value at 30 cm depth was around 0.09 for C1 and C2, but different at
deep layers. At 90 cm depth, K was equal to 0.07 in C1 and equal to 0.025 in CZ.
These differences may have been caused by the low root density observed in CZ
and by the clumped roots found at C1. At 120 cm in both cylinders, K reached
values as low as 0.03 and 0.008 for C1 and C2 respectively, and the reasons for
such low K values may be related to a low root density found in both cylinders.
As proposed by Ritchie et al. (1999), the K value equal to 0.1 can be used to
describe the daily fraction of water extracted by plants; although, to correctly

describe the water extraction in conditions of restricted water supply, K values

101

different than 0.1 were used in the simulations performed here. These different
values are correct, when it is assumed that K varies as function of the amount
and distribution of roots for each soil layer as proposed by Ritchie et al. (1999). In
this study K decreased with soil depth to account for the bypassing and low root

density.

Summary and Conclusions

When measurements of volumetric water content are accurate and
periodically obtained, it is possible to simulate water extraction by plants without
knowing the amount of roots for each soil layer. This means that it is possible to
simulate water extraction based on the model proposed by Ritchie et al. (1999)
and based on K values for each soil layer. lf K is not calculated, a value of 0.1
may be used to predict water extraction. However, in this case, a clear
exponential decay in water content for soil layers along the time must be
observed before assuming K equal to 0.1. In case no clear exponential decay is
observed different values of K must be used for each soil layer to accommodate

the correct water extraction by plants.

102

References

Ahuja, LR, and DR. Nielsen. 1990. Field Soil-water relations. p. 143-190. In B.A.
Steward and DR. Nielsen (ed.) Irrigation of Agricultural crops. Agron.
Monogr. 30. ASA, CSSA and SSSA, Madison, WI.

Bland, W.L., and W. Dugas. 1989. Cotton root growth and soil water extraction.
Soil. Sci. Soc. Am. J. 53:1850-1855.

Dardanelli, J.L., O.A. Bachmeier, R. Sereno, and R. Gil. 1997. Rooting depth and
soil water extraction patterns of different crops in a silty loam haplustoll.
Field Crop Res. 54:29-38.

Dwyer, L.M., D.W. Stewart, and D. Balchin. 1988. Rooting characteristics of corn,
soybeans and barley as a function of available water and soil physical
characteristics. Can. J. Soil Sci. 68:121-132.

Hamblin, AP. 1985. The influence of soil structure on water movement, crop root
growth, and water uptake. Adv. Agron. 38:95-158.

Hamblin, A. and D. Tennant. 1987. Root length density and water uptake in
cereals and grain legumes: How well are they correlated? Aust. J. Agric.
Res. 382513-527.

Hector, D.J., K. Gregson, and M. McGowan. 1993. A computer simulation
describing water extraction by crop root systems. Field Crop Res. 32:287-
304.

Klepper, B. 1990. Root growth and water uptake. p.281-322. In B.A. Steward and
DR. Nielsen (ed.) Irrigation of Agricultural crops. Agron. Monogr. 30. ASA,
CSSA and SSSA, Madison, WI.

Kozlowski,, T.T. 1968,. Water deficits and plant growth. Academic Press, New
York, NY.

Kramer, P.J., and J.S. Boyer. 1995. Water relation of plants and soils. Academic
Press. San Diego, CA.

Lacape, M.J., J. Wery, and D.J.M. Annerose. 1998. Relationships between plant
and soil water status in five field—grown cotton (Gossypium hirsutum L.)
cultivars. Field Crop Res. 57:29-43.

McIntyre B.D., S.J. Riha, and DJ. Flower. 1995. Water uptake by pearl millet in a
semiarid environment. Field Crop Res. 43267-76.

103

 

Molz, F .J. 1981. Models of water transport in soil-plant-system: A review. Water
Resourc. Res. 17(5):1245-1260.

Passioura, J.B. 1983. Roots and drought resistance. Agric. Water. Management
7:265—280.

Ritchie, J.T. 1986. Soil water and plant productivity. Transactions of the Xlll.
Congress of the International Society of Soil Science. Hamburg 13—20
August. pp 13-28.

Ritchie, J.T., J.L. Dardanelli, M. Calmon, and J.M. Andriani. 1999. A functional
model for root water uptake. Agron. J. (submitted).

Robertson, M.J., S. Fukay, M.M. Ludlow, and G.L. Hammer. 1993a. Water
extraction by grain sorghum in a sub-humid environment. I. Analysis of the
water extraction pattern. Field Crop Res. 33:81-97.

Robertson, M.J., S. Fukai, MM. Ludlow, and G.L. Hammer. 1993b. Water
extraction by grain sorghum in a sub-humid environment. ll. Extraction in
relation to root growth. Field Crop Res. 33:99—112.

Tardieu, F. 1988. Analysis of the spatial variability of maize root density. ll.
Distance between roots. Plant Soil 107:267-272.

Zobel, R.W. 1996. Genetic control of root systems. p. 21-30. In Yoav Waisel et al.
(ed.) 2"d ed. Plant Roots: The Hidden Half. Marcel Dekker, New York, NY.

104

 

Table 3.1.

experiment at Michigan State University.

K values for different soil layers in C1 and C2. Greenhouse

 

 

 

K
Depth (cm) C1 C2

0-30 0.080 0.090
3045 0.080 0.065
45-60 0.075 0.065
60-75 0.075 0.065
75-90 0.070 0.025
90-105 0.070 0.010
105-120 0.030 0.008

 

Table 3.2. Day of the year (DOY) for water extraction within soil layers for C1 and
C2. Greenhouse experiment at Michigan State University, Ml.

 

 

 

Michigan Depth(cm) DOY
Maximum Interval

C1 0-30 24 -
30-45 45 21

45-60 46 1

60-75 47 1

75-90 50 3

90-105 53 3

1 05-120 55 2

C2 0-30 50 -
45-75 59 9

75-90 63 4

90-1 05 65 0

1 05-120 70 5

 

105

 

0.30
0.25 .
0.20 «
0.15 «
0.10 (

0.05 1

 

 

 

 

0,00

VWC (cm 3cm‘3)

 

0.30
0.25 «
0.20 I
0.15 1
0.10

0.05 1

 

VWC (cm 3cm'3)

0.00 1

 

75

 

 

 

 

0.30
0.25 1
0.20 1
0.15 1
0.10 1

0.05 I

VWC (cm 3cm'3)

 

/, /8 K

 

 

 

0.00

 

0.30

0.25 1

0.20 1

0.15 1

0.10 1

WVC (cm 3cm'3)

0.05 1

 

120

/

 

 

0CD

f

102030405060708090

DOY

102030405060708090

DOY

Figure 3.1. Volumetric water content (VWC)(cm30rn'3) within different soil depths
(cm) as function of the day of year (DOY). Data for C1 in the greenhouse

experiment at Michigan State University, MI.

106

vwc (cm’cm'3) VWC (cm’cm'a) VWC (cm’cm'3)

VWC (cm3cm'3)

0.10 1

0.05

0.30

0.25 1

0.10 1

0.05

0,30

0.25 1

0.20 1

0.10 1

0,05

0.25 1

0.20

0.15

0.05

 

 

 

 

45

 

 

 

 

75

 

 

 

 

 

 

 

 

 

 

107

Figure 3.2. Volumetric water content (VWC)(cm3cm'3) within different soil depths
(cm) as function of the day of year (DOY). Data for C2 in the greenhouse experi-
ment at Michigan State University,Ml.

 

 

 

 

log(AV\C-LL)

 

 

 

 

 

DOY
4O 45 50 55 a) 65 7O 75 m 85 so % 100
-1-5 L #L L 1 1 1 1 A L L 1

105cm
120cm

 

 

 

 

~20 1

-2.5

«30 1

-3.5 1

Iog(Avvc-LL)

~40 1

~45 1

 

 

 

 

-5.0

Fig ure 3.3. Logarithm of the difference between AWC and LL as function of
DOY. The slope of straight lines represents K value for each soil depth. Data for
C1 and C2 at greenhouse experiment at Michigan State University, MI.

108

VWC(%)

WVC(%)

WVC(%)

 

025

 

 

 

 

 

 

° 30cm-Ct
o
0201
015i
010-
0 Measured
Predicted
005 f a
2025303560455055606570750085
DOY
025
90cm-C1

0.20 1

0‘51

 

 

 

 

 

 

 

 

0 10 1
0 Measured
— Predicted
o 05 T ‘7 ‘T V V Y Y I
40 45 50 55 60 85 70 75 80 85
DOY
0.25
120cm-C1
0.20 . ' o
0.15 1
0 10 1
0 Measured
— Predicted
o 05 T T f 1 Y Y
50 55 60 65 70 75 80 85
DOY

 

 

 

005

-—- Predicted

 

 

35

0.25

7 T fi fi I f Y I’ T

556065 70 7580 859095100
DOY

40 4550

 

020‘

0151

0.10 1

 

005

 

90cm-CZ

. Measured
— Pin-diced

 

 

35

T

404550556005707580859095100

DOY

 

0.25

0.20 1

 

  

1 ZOcm-CZ

0 Measured
— Predided

 

 

0.15
60

65 70 75 00 85 90 05

DOY

Figure 3.4. Measured and simulated VWC (%) at different soil depths for
C1 and C2 in the greenhouse experiment at Michigan State University,M|.

109

General Conclusions

The main objective of this dissertation was to demonstrate that plants
growing in limited supply of water could leave an appreciable amount of water in
the soil without being extracted by roots. To confirm the hypothesis that roots
bypassed soil regions in detriment of others, which caused the incomplete water
extraction, this study aimed to verify the spatial variability of root distribution,
water content and soil texture in soil samples collected in a grid pattern in two
different locations.

Based on the krigged maps for both locations, it was confirmed that not
only roots but also soil water content were spatially distributed in soil layers. The
uneven distribution of roots and the spatial variability of soil water content for
each soil layer studied confirmed the hypothesis of root bypassing.

Also, it was verified that classical statistics was not adequate to describe
spatial variability. This kind of statistics does not take into account the distance
among samples. Considering the scale, the spatial dependence among soil
samples and the objectives of this study, geostatistics approach proved to be
more adequate to describe the micro-scale spatial variability.

The functional model used to simulate soil water extraction in this study
closely predicted the water extraction for soil layers observed in both cylinders
(C1 and CZ) in the greenhouse experiment at Michigan State University, MI. The
constant value, K, in the functional model, must be defined for each soil layer if

roots are not fully occupying a specific soil layer.

110