346
the chosen nondestructive approach. For largescale experiments that will include drought periods,
tensiometers are not recommended as they are typically only functional over a limited range of soil
matric potentials (-0.00 to -0.08 MPa). Neutron
probe approaches have also fallen out of favor recently because they demand site-specific calibration and additional radiation protection procedures.
Soil resistance blocks, soil psychrometers, and
other electronically based sensors are the best
choice for a study in which detailed temporal resolutions on the order of hours or days are required,
but they would be expensive to replicate over large
spatial scale. Autologging of any instrument becomes a problem when replicated over large spatial
scales due to signal losses over long wire lengths
and the disturbance associated with wire installation. Perhaps the best approach for measuring soil
water status over large spatial scales (in the absence
of proven remote sensing methods, see Chapter 13)
is time domain reflectometry (Topp and Davis
1985).
Time domain reflectometry (TDR) has proven to
be a cost effective and durable approach for making
soil water content measurements in large-scale water manipulation studies (Hanson et al. 1995). However, since TDR is only a measurement of soil
moisture content, complementary information on
the relationship between soil water content and soil
water potential for the experimental soils (i.e., a soil
moisture release curve) is required to translate water content data to soil water or matric potential.
Measured TOR yields water content for the bulk
soil including coarse fraction (i.e., soil particles >2
mm), but soil moisture release curves are typically
conducted on soils from which the coarse fraction
has been removed. Therefore, bulk or raw TDR
data (TORr) must be corrected for coarse fraction
content before it is used in a soil moisture release
curve to obtain soil water or matric potential information. Alternatively, soil moisture release curves
including particle size classes greater than 2 mm
should be generated. Paruelo et al. (1987) took the
latter approach and showed that significant soil water could be held by the 2- to 11-mm coarse fraction
of a Patagonian arid steppe soil. They also showed
that large coarse fraction components (i.e., rocks
> 11 mm) decreased soil water holding capacity
nearly proportionate to their volume.
Paul J. Hanson
To obtain soil water potential data from TDRr for
soils having large coarse fraction components (i.e.,
rocks assumed to contain zero water) Hanson et al.
(1998) used the following calculation to correct
TDR values (TORe) for application to their soil
moisture release curves from sieved soil:
TORe = TDR/(100 - Cf ) * 100 (23.1)
where Cf is the appropriate mean coarse fraction
for the soil. For their soil with a mean coarse fraction of 14.2% in the surface 35 cm, TORe values
were typically 2 to 3% higher than observed TDRr.
What is the appropriate depth for making soil
water or matric potential measurements? Unfortunately, there is no simple answer to this question.
Important criteria depend on the questions being
addressed in a particular study and the nature of the
soils being measured. If plant responses are a key
component, one should consider measuring that
portion of the soil profile that includes the "effective" rooting depth of the plants of interest. Multiple soil depths should be considered if time and
money allow. A recent study contrasting water use
by herbaceous and woody plant life-forms in a
shortgrass steppe community (Dodd et al. 1998)
showed that each uses water from different layers
of the soil profile. In such a system it would be
critical to collect water content data from multiple
soil depths corresponding to the water use characteristics of the species of interest. Data on depth of
water resources for plant function are available for
a number of other ecosystems (Ehleringer et al.
1991; Flanagan et al. 1992; Gordon et al. 1989;
White et al. 1985).
Dealing with Spatial Variation
Spatial variation in soil characteristics, species
composition, vegetation cover, slope, and aspect all
interact and lead to potentially significant spatial
differences across sites used for large-scale manipulations. This pretreatment variation must be characterized and understood to ensure that observed
differences in soil water status imposed by the treatment infrastructure are true treatments and not simply inherent patterns driven by variable site characteristics. Pretreatment data collections for soil
water content and/or water potential should be
available for at least one full year to characterize
the soil water patterns across the experimental area
of interest, and the temporal resolution of the pre-
the chosen nondestructive approach. For largescale experiments that will include drought periods,
tensiometers are not recommended as they are typically only functional over a limited range of soil
matric potentials (-0.00 to -0.08 MPa). Neutron
probe approaches have also fallen out of favor recently because they demand site-specific calibration and additional radiation protection procedures.
Soil resistance blocks, soil psychrometers, and
other electronically based sensors are the best
choice for a study in which detailed temporal resolutions on the order of hours or days are required,
but they would be expensive to replicate over large
spatial scale. Autologging of any instrument becomes a problem when replicated over large spatial
scales due to signal losses over long wire lengths
and the disturbance associated with wire installation. Perhaps the best approach for measuring soil
water status over large spatial scales (in the absence
of proven remote sensing methods, see Chapter 13)
is time domain reflectometry (Topp and Davis
1985).
Time domain reflectometry (TDR) has proven to
be a cost effective and durable approach for making
soil water content measurements in large-scale water manipulation studies (Hanson et al. 1995). However, since TDR is only a measurement of soil
moisture content, complementary information on
the relationship between soil water content and soil
water potential for the experimental soils (i.e., a soil
moisture release curve) is required to translate water content data to soil water or matric potential.
Measured TOR yields water content for the bulk
soil including coarse fraction (i.e., soil particles >2
mm), but soil moisture release curves are typically
conducted on soils from which the coarse fraction
has been removed. Therefore, bulk or raw TDR
data (TORr) must be corrected for coarse fraction
content before it is used in a soil moisture release
curve to obtain soil water or matric potential information. Alternatively, soil moisture release curves
including particle size classes greater than 2 mm
should be generated. Paruelo et al. (1987) took the
latter approach and showed that significant soil water could be held by the 2- to 11-mm coarse fraction
of a Patagonian arid steppe soil. They also showed
that large coarse fraction components (i.e., rocks
> 11 mm) decreased soil water holding capacity
nearly proportionate to their volume.
Paul J. Hanson
To obtain soil water potential data from TDRr for
soils having large coarse fraction components (i.e.,
rocks assumed to contain zero water) Hanson et al.
(1998) used the following calculation to correct
TDR values (TORe) for application to their soil
moisture release curves from sieved soil:
TORe = TDR/(100 - Cf ) * 100 (23.1)
where Cf is the appropriate mean coarse fraction
for the soil. For their soil with a mean coarse fraction of 14.2% in the surface 35 cm, TORe values
were typically 2 to 3% higher than observed TDRr.
What is the appropriate depth for making soil
water or matric potential measurements? Unfortunately, there is no simple answer to this question.
Important criteria depend on the questions being
addressed in a particular study and the nature of the
soils being measured. If plant responses are a key
component, one should consider measuring that
portion of the soil profile that includes the "effective" rooting depth of the plants of interest. Multiple soil depths should be considered if time and
money allow. A recent study contrasting water use
by herbaceous and woody plant life-forms in a
shortgrass steppe community (Dodd et al. 1998)
showed that each uses water from different layers
of the soil profile. In such a system it would be
critical to collect water content data from multiple
soil depths corresponding to the water use characteristics of the species of interest. Data on depth of
water resources for plant function are available for
a number of other ecosystems (Ehleringer et al.
1991; Flanagan et al. 1992; Gordon et al. 1989;
White et al. 1985).
Dealing with Spatial Variation
Spatial variation in soil characteristics, species
composition, vegetation cover, slope, and aspect all
interact and lead to potentially significant spatial
differences across sites used for large-scale manipulations. This pretreatment variation must be characterized and understood to ensure that observed
differences in soil water status imposed by the treatment infrastructure are true treatments and not simply inherent patterns driven by variable site characteristics. Pretreatment data collections for soil
water content and/or water potential should be
available for at least one full year to characterize
the soil water patterns across the experimental area
of interest, and the temporal resolution of the pre-