Document - Colorado State Forest Service

Transcription

Variability in Rangeland
Water Erosion Processes
COLORBOO STBTE UNIVERSITY
Related Society Publications
Advances in Measurement of Soil Physical Properties: Bringing Theory into
Practice Methods of Soil Analysis
Minirhizotron Observation Tubes: Methods and Applications for Measuring
Rhizosphere Dynamics
Soil Testing and Plant Analysis, Third Edition
Soil Testing: Sampling, Correlation, and Interpretation *
For information on these titles, please contact the ASA, CSSA. SSSA
Headquarters Office; Attn: Marketing; 677 South Segoe Road; Madison, WI
53711-1086. Phone: (608) 273-8080. Fax: (608) 273-2021.
Variability in Rangeland Water
Erosion Processes
Proceedings of a symposium sponsored by Divisions S-l, S-6, and S-7 of the Soil
Science Society of America in Minneapolis, Minnesota, 1-6 Nov. 1992.
Editors
Wilbert H. Blackburn. Frederick B. Pierson, Jr.,
Gerald E. Schuman. and R. Zartman
Organizing Committee
Wilbert H. Blackburn
Gerald E. Schuman
Frederick B. Pierson, Jr.
Editor-in Chief SSSA
Larry P. Wilding
Managing Editor
David M. Krai
Associate Editor
Marian K, Viney
SSSA Special Publication Number 38
Soil Science Socielj of America. Inc.
Madison. Wisconsin, USA
1994
,'RADOST.A
Cover Design: Patricia Scullion
The illustration depicts elevations at Reynolds Creek experimental
watershed. This is Fig. 6-6 in Chapter 6 by M.S. Seyfried and G.N.
Rerchinger.
Copyright © 1994 by the Soil Science Society of America. Inc.
ALL RIGHTS RESERVED UNDER THE U.S. COPYRIGHT LAW
OF 1976 (PL 94-533)
Any and all uses beyond the limitations of (he "fair use" provision
of the law require written permission from the publishers) and or the
authors); not applicable to contributions prepared by officers or employees of the U.S. Government as part of (heir official duties.
Soil Science Society of America, Inc.
677 South Segoe Road. Madison. WI 53711, USA
Library of Congress Cataloging-in-Publication Data
Variability in range I and water erosion processes : proceedings of a
symposium, "variability in range I and water erosion processes,*'
sponsored by Divisions S-1, S-6. and S-7 of the Soil Science Society
of America and the Society of Range Management, held in
Minneapolis. Minnesota, 1-6 November 1992 organizing committee. Wilbert H Blackburn
. [el al J
p.
cm — {SSSA special publication : no. 38)
Includes bibliographical references.
ISBN 0-89118-812*6
1. Soil erosion prediction—Congresses. 2. Rangelands—
Congresses 3. Soil erosion—Research—United States—Congresses.
I Blackburn. W. H. (Wilbert Howard) II. Soil Science Society of
America. Division S-l. III. Title: Rangeland water erosion processes.
IV. Series.
S622.2.V37 1994
94-26299
631.4'S—dc20
OP
Printed in the United States of America
CONTENTS
Page
Foreword
Preface
Contributors
Converson Factors for SI and non-Si units
1 Sources of Variation in Interril] Erosion on Rangelands
Wilbert H. Blackburn and Frederick B. Pierson, Jr.
vii
ix
xi
xiii
1
2 The WEPP Model and Its Applicability for Predicting Erosion on
Rangelands
John M. Laflen, Dennis C. Flanagan, Mark A. Weltz,
Jeffry J. Stone, and James C. Ascough II
11
3 Incorporating Small Scale Spatial Variability into Predictions of
Hydrologic Response on Sagebrush Rangelands
Frederick B. Pierson, Jr., WUbert H. Blackburn,
S. S. Van Vactor, and James C. Wood
23
4 Spatial Pattern Analysis of Sagebrush Vegetation and Potential
Influences on Hydrology and Erosion
K. E. Spaeth, Mark A. Weltz, H. Dale Fox, and
35
5 Temporal Variability in Rangeland Erosion Processes
J. R. Simanton and William E. Emmerich ..
51
6 Influence of Frozen Soil on Rangeland Erosion
M. S. Seyfried and G. N. Flerchinger
67
ri
CONTENTS
7 In Situ Estimation of Effective Hydraulic Conductivity to Improve
Erosion Modeling for Rangeland Conditions
James P. Dobrowolski .
8 Variations in Plants, Soils. Water, and Erosion in a Pinyon Pine and
Juniper Dominated Range Site
M. Karl Wood, David Hereford, Charles E. Sounders,
and Alison Hill . .
83
93
FOREWORD
Land degradation of rangelands continues to be a national and international
research priority, especially in developing countries. Soil erosion is dynamic,
highly variable, and difficult to measure with precision. This publication identities climatic, soil, and biological parameters that control water erosion processes. Range hydrologists have recognized spatial and temporal effects of soil and
vegetation on water erosion for a long time; this publication summarizes the
current state of research knowledge in this area. Hydrologic modeling of rangeland erosion processes has made significant advances in the past five years with
the development of the Water Erosion Prediction Project (WEPP) model.
Research to refine hydrological models will continue because of the extensiveness and variability of rangelands on a global scale. The ability to predict with
accuracy disposition of eroded sediments within a watershed and at off-site locations continues to be a challenge. Likewise, scaling erosion from relatively small
rainfall simulators to natural watershed basins is a difficult, if not impossible
task. This publication, however, serves as a valuable starting point to elucidate
the variables and needed hydrological research to model soil erosion under
rangeland conditions. The organizers of the symposium and authors are to be
commended for bringing this important text to fruition at a time when the public mandates increased attention to sustainability of the biosphere.
LARRY P. WILDING, president
Soil Science Society of America
PREFACE
Historically, erosion research on rangeland has emphasized processes CODtrolling erosion for on-site management with little emphasis on off-site impacts.
This research led to the development of improved vegetation management practices for erosion control with little erosion predictive capabilities. In addition,
existing rangeland erosion predictive technology is heavily influenced by theory
developed for croplands. Current societal interest in environmentally sustainable
rangeland management practices has gone beyond on-site concerns to include the
impact of soil erosion off-site. This, combined with Society's tendency to acquire
its desires through regulation, requires improved erosion prediction technology
for rangelands.
The Water Erosion Prediction Project (WEPP) offers the opportunity to significantly improve erosion predictive technology for rangelands. The WEPP technology has improved rangeland predictive technology through greater model
complexity; however, much of the improvement in simulation accuracy is lost in
techniques used to estimate model parameters. This is particularly true on arid
and semiarid rangelands where the estimation of model parameters is hampered
by significant spatial and temporal variations in erosion processes. Rangeland
interrill erosion is spatially and temporally distributed because of the influence of
different vegetation growth forms, spatially distributed biomass, and the variable
effect climate has on surface soil properties. The WEPP offers a framework to
address the variability found on rangelands and to significantly improve rangeland erosion predictive technology.
This publication is the result of a symposium held during the 1992 annual
meeting of the Soil Science Society of America. The symposium was co-sponsored by the Soil Science Society of America and the Society for Range
Management to address the state-of-the science on variability in rangeland water
erosion processes with emphasis on vegetation induced variability.
WILBERT H. BlJ\CKBURN, coeditor
USDA-ARS
Fort Collins, Colorado
G. E. SCHUMAN, coeditor
USDA-ARS
Cheyenne, Wyoming
CONTRIBUTORS
Junes C. Ascough II
Agricultural Engineer, USDA-ARS. National Soil Erosion Research
Laboratory. 11% Soil Bldg.. Purdue University. West Lafayette, IN
47907-1196
Wilton H. Blackburn
Associate Area Director, USDA-ARS, 1201 Oakridge Dr., Suile 150,
Fort Collins, CO 80525
Jmmes P. Dobmrolskj
Associate Professor of Watershed Science. Department of Range
Science and Watershed Science Unit. Utah State University, Logan, UT
84322-5230. Presently on leave as a Fulbrighi Senior Scholar,
Department d'Ecologie cl Pasloralisme. Instttut d'Agronomique et
Veterinaire Hassan [I. Rabat-Instituls B.P. 6457, Rabat, Morocco
William E. Emmerich
Soil Scientist. USDA-ARS. Southwest Watershed Research Center,
2000 E. Allen Rd-. Tucson. AZ 85719
Denis C. FhuugaB
Agricultural Engineer, USDA-ARS, National Soil Erosion Research
Laboratory. 11% Soil Bldg., Purdue University, West Lafayette, IN
47907-11%
G. N. Flercbinger
Research Hydraulic Engineer. USDA-ARS, 800 Park Blvd., Suite 105,
Boise. ID 83712
H. Dale Fox
Range Scientist. USDA-SCS. Southwest Watershed Research Center,
2000 E. Allen Rd. Tucson. AZ 85719
Dnid Hereford
Technical Director. New Mexico Slate University, 484 Lupton Place,
Las Cruces. NM 88001
AlisooH.II
Research EcofogisL U.S. Army Construction Engineering Laboratories,
USACERL (ENG) P.O Box 9005. Champaign, IL 61826-9005
Jou M. L>fk>
Research Leader. USDA-ARS, National Soil Erosion Research
Laboratory. 1196 Soil Bldg. Purdue University. West Lafayette. IN
47907-1196
Frederick B. P*r»n. Jr
Hydrologist, USDA-ARS. Northwest Watershed Research Center. 800
Park Blvd.. Plaza IV. Bode, ID 83712
M.S.Se> fried
Soil Scientist, USDA-ARS. 800 Park Blvd.. Booe. ID 83712
J. R. StauUn
Research Hydrologist. LSDA-ARS. Southwest Watershed Research
Center. 2000 E. Allen Rd.. Tucson. AZ 85719
Soil Scientist'EcologisI, Gill National Forest Service. 3005 E Camio
Del Bosque. Silver City, MN 88061
CONTRIBUTORS
K. E. Spaeth
Rangcland Hydrologist, USDA Soil Conservation Service, Northwest
Watershed Research Center, 800 Park Blvd., Plaza IV, Suite 105, Boise,
ID 83712
Jeffry J. Stone
USDA-ARS. Southwest Watershed Research Center, 2000 E. Allen
Road, Tucson, AZ 85719
S. S. Van Vaclor
Hydrologist, USDA-ARS, Northwest Watershed Research Center. 800
Park Blvd.. Plaza IV. Suite 105. Boise, ID 83712-7716
Mark.A. Weliz
Research Hydrologist, USDA-ARS, Southwest Watershed Research
Center, Tucson, AZ 85719
James C. Wood
Water Quality Specialist, USDA-SCS, 3244 Elder St.. Boise. ID 83705
M. Karl Wood
Professor of Watershed Management. Department of Animal and Range
Sciences, Box 3-1, New Mexico Stale University. Las Cruces. NM
88003
Conversion Factors for SI and non-Si Units
To convert Column 1
imo Column 2.
multiply by
Column 1 SI Unit
Column 2 non-Si unit
To convert Column 2
into Column 1.
multiply by
1 ..null.
0.621
1094
328
1.0
3.94 x 19'
10
kilometer, km I 111' ml
meter, m
meter, m
micrometer, fan (HI * m)
millimeter, mm (10 ' m)
nanometer, nm(lO-'m)
mile, mi
yard, yd
foot, ft
micron, ft
inch, in
Angstrom, A
1.609
0.914
0.304
1.0
•en
Mi
•quire mile, mi'
•ere
square foot, ft'
square inch, in1
0.405
4.05 x 10-'
2.590
4.05 x III'
9.29 x 10-'
645
25.4
0.1
Ana
2.47
247
0386
1
2.47 K 1010.76
1 .55x10-'
hecure. ha
•quire kilometer, km' (HI' m)'
•quire kilometer, km' (10' m)'
squirt meter, m1
square meter, m'
iquire millimeter, mm1 (10 ' m)1
Volume
9.7J
35.3
6.10
2.84
» 10'
• IP
> 111 '
1.057
3 Si x 10-'
0.265
33.71
2,11
cubic meter, m '
cubic melcr. m '
cubic meter, m '
liter. L (10-' m')
liter, l . ( K l ' m ' )
liler. l.dO-'pV)
liter, L(10 'm')
liter, L(lfr'm')
liter, 1. (Ill' m'l
acre-inch
cubic fool, ft'
cubic inch, in'
bushel, bu
quart (liquid), ql
cubic fool, ft'
gallon
ounce (fluid), oz
pint dim. II, pt
102.8
x 10^
x ID-'
35.24
0.946
2.83
1.64
28.3
3.78
2.96
x 10 '
0.473
::n. HI'
3.52 * 10'
2.205
0.01
1.10*10-'
1.102
1.102
gram.g(10-'kg)
gram, g (Iff' kg)
kilogram, kg
kilogram, kg
kilognm, kg
M" I M J ' l . l t l l
pound, Ib
ounce (avdp), oz
pound, Ib
quintal (metric), i)
ton (2000 Ib), ton
ton (U.S.), ton
(on (U.S.), ton
M,' 1 !
tonne. 1
454
28.4
0.454
too
907
0.907
0.907
Yield and Rule
0.893
7 77 « l(t J
1 4'». l i t 1
VJ,
III
0.107
893
893
0.446
2.24
kilognm per hectare, kg ha '
kilogrun per cubic tncici, kg m '
kilogram per hectare, kg ha '
liter per hectare. L ha~'
tonnes per hectare, 1 ha '
mcgagram per hectare. Mg ha '
megagram per hectare, Mg ha '
meter per second, m s '
pound per acre. Ib acre '
pound per bushel, bu"'
bushel per acre. 60 Ib
bushel per acre, 56 Ib
bushel per acre, 48 Ib
gallon per acre
pound per acre. It) acre
pound per acre. Ib acre'1
(on (2000 Ib) per acre, ton acre'1
mile per hour
1.12
1247
67.19
62.71
53.75
9.35
1.12
x 10 '
1.12» 10-J
2.24
0.447
Specific Surface
10
square meter per kilogram. mj kg '
ujuare meter per kilogram, m' kg '
1000
Kquiirc centimeter per gram, cm1 g '
squurc millimeter per gram, mm' g'1
0.1
0.001
Preisurt
9JO
in
KM)
1.45
x 10"*
megapkacal,MPa(10APa)
megapucal. MPa(IO*Pa)
megagram per cubic meter. Mg m '
pucil, Pa
pascal. Pa
atmosphere
bat
gram per cubic centimeter, g cm'1
pound per square foot, Ib ft '
pound per square inch, Ib m •'
(CMturned on next page)
0.101
01
1.00
47.9
6.90
x 10"
To convert Column 1
into Column 2,
multiply by
Column 1 SI Unil
Column 2 non-Si Unit
Ti > conven Column 2
into Column 1,
multiply by
Temperature
1.00(K-273)
(9/5 'Q + 32
Kelvin. K
Celsius, 'C
x 1(H
0,239
10'
0.733
2.387 x 10-'
joule, J
joule. J
joule, J
joule, J
joule per square meter, J m
Celsius, *C
Fahrenheil, °F
l.OO(°C + 273)
5/9CF-32)
Energ) , Work Quantity of Heat
9.52
1
111
1,43
IK V. 1011.
x 10" '
British thermal unil, Btu
calorie, cal
erg
foot-pound
calorie per square centimeter (langley)
dyne
caloric per square centimeter
minute (irradiancc), cal cm ' nun '
N
wall per square meter, W m '
\J05xW
4.19
10"'
1.36
4.19
Q
2
x 10*
10"'
698
C
0?
o
s
I(j
Transpiration and Photosynthesis
3.60
x iaj
5.56x10-'
milligram per square mclcr second,
mg m-1 B-'
milligram 1 1 ! < » per square mcicr
tccond, mg m ' s '
milligram per square mclcr second,
milligram per square melcr second,
mg m! -. '
gram per square decimeter hour,
g dm J h'1
micromole (H;O) per square centimeter second, fimo\ cm'1 s'1
milligram per square centimeter
second, mg cnrj s*1
milligram per square decimeter hour,
mg dm ' h"1
Plane AnRk
degrees (angle), *
27.8
180
2.78 x 10-'
s
31
Ekclr-kll < •imilnrllvliy. Kkrtrirlly, and Mignetlim
III
Of
liemcn per meter, S m '
lesll. T
millimho per cenlimctcr, mmho cm"1
ii
0.1
10-
Water M»Burtmcnl
'. n, HI'
<»KI » 111'
4.40
8.11
97.28
8.1 « 10'
cubic meter, m'
iurik- melcr per houi, m' h '
cubic meter per hour, m1 h'1
hcL-laic meter-, hi-m
hecurc-melen, ha-m
hectare-centimeter-, ha cm
•cre-incbes, »crc-in
cubic feel per second, fl* r*
U.S. gallons per minute, gal min '
acre-feei. acre-fl
acre-inches, actc-m
acre-reel, acic-ft
102.8
101.9
0.227
0.123
1.03 x
12.33
10-'
('onctnlratloni
1
0.1
1
mole per kilogram, cmol kg
(ion exchange capacity)
gram per kilogram, g kg '
milligram per kilogram, mg kg*1
milliequivalenis per 100 grams, meq
100 g-1
percent, %
parts per million, pprn
1
10
1
lUdkwctlvlty
2.7 « III"
2 7 t III'
100
100
t-cu-ucrel, Bq
becquerel per kilogram. Hq kg '
gray, Gy {tibtorfaetl dote)
tievcrt, Sv (equivalent doae)
2.29
1.20
1.39
1.66
Mnwir.il'
t
K
Ct
M(
curie, Ci
pitiH-unc per gram, pCI g '
rad, rd
rein (rocntgcn equivalent man)
3.7 x 10"
37
0.01
0.01
Plum Nutrient Coavertion
Oxide
'
do
MjO
0.437
0.830
0.715
0.602
Sources of Variation in Interrill
Erosion on Rangelands
U5DA-ARS Northern Plains Area Office
Fort Collins. Colorado
Frederick B. Pitrson. Jr.
USDA-AKS Northwest Watershed Research Center
Boise, Idaho
ABSTRACT
Current rangelaod erosion modeling efforts do not adequately account for vegetation
induced variability in interrill erosion processes. Research examples are presented that
support the premise that range land interrill erosion is spatially and temporally distributed
because of the influence of different vegetation growth forms, spatially distributed biomass, and a variable climate on surface soil properties. The surface soil parameters of
shrub dominated landscapes display greater spatial than temporal variability, but bunchgrass and sodgrass dominated landscapes exhibited greater temporal variability than spatial Improved model parameter estimating techniques are needed to account for the interrill erosion variability found on rangelands.
Cropland and rangeland erosion research have traditionally taken different
approaches. Historically erosion research on cropland has been directed toward
prediction technology (ZJngg. 1940; Lancet al., 1992). Although the early research
emphasis was on-site predictive technology, the approach was consistent with later
needs to predict off-site impacts. This has been a successful research and technology development effort resulting in the Universal Soil Loss Equation, USLE
(Wischmeier & Smith. 1965, 1978), Modified Universal Soil Loss Equation,
MUSLE (Williams, 1975), Revised Universal Soil Loss Equation, RUSLE (Renard
et al., 1991X and the Water Erosion Prediction Project WEPP (Laden et al.. 1991).
Erosion research on rangeland emphasized factors controlling rill and interrill
erosion for on-site management. Early research identified vegetation as a dominate
controlling factor with erosion control research and technology development
emphasizing vegetation management practices. This research led to the development
of improved on-site vegetation management practices for erosion control with little
predictive capabilities (Blackburn, 1975,1983, 1984; Blackburn etai., ! 986; Gifford
Copyright O 1994 Soil Science Society of America. 677 S Scgoc Rd. Madison. WI 53711, USA,
2
BLACKBURN & PIERSON
& Hawkins. 1978; Branson et al., 1981; Rauzi et al., 1968; Packer, 1953; Meeuwig,
1970).
Changing societal demands for environmentally sustainable management
practices and the growing trend to obtain these demands through increased regulation require improved erosion prediction technology for rangelands. In order to
respond to growing societal demands during the 1970s and 1980s, federal and state
land management and regulatory agencies adopted the USLE for use on rangeland
before the background research or validation work to adapt the equation to rangeland was completed (Foster, 1982; Foster etal., 1981; Johnson & Gebhardt, 1982;
Johnson & Gordon, 1988; Johnson et al., 1984, 1985). The result was general misuse of USLE on rangelands (Trieste & Gifford, 1980; Wischmeier, 1976), which
led to the Society for Range Management (SRM) passing a position statement
opposing the use of USLE on rangelands as a determinant of land treatment needs,
land treatment effectiveness, program funding, livestock or wildlife stocking rates,
or any other land management or regulatory decision (SRM Position Statement
approved by board of directors at the 1985 summer meeting in Amarillo, TX).
The Water Erosion Prediction Project was initiated in 1985 to develop a
state-of-the-art model for the prediction of erosion from croplands and rangelands. Through this effort technology for modeling erosion processes on rangelands has improved through greater model complexity; however, much of the
improvement in simulation accuracy is lost in techniques used to estimate model
parameters (Loague & Freeze, 1985; DeCoursey, 1988, 1992; Ferreira & Smith,
1988; National Research Council, 1991). This is particularly true on arid and
semiarid rangelands where little erosion data is available that emphasizes erosion
prediction technology. The estimation of model parameters for rangelands is also
hampered by significant spatial and temporal variations in erosion processes
(Blackburn et al.. 1990; Wilcox et al., 1992). Improvements in model parameter
estimation techniques and in our understanding of vegetation and soil induced
variability are needed to increase erosion predictive capabilities for rangelands
(Blackburn et al., 1992). The guiding premise for the research reviewed in this
chapter is that rangeland intemll erosion is spatially and temporally distributed
because of the influence of different vegetation growth forms, spatially distributed biomass, and a variable climate on surface soil properties. The objective of
this chapter is to present an overview of the influence of vegetation induced variability on intertill erosion processes. Examples illustrating the influence of vegetation on the spatial and temporal variability of interril! erosion are presented.
VEGETATION INFLUENCES—SPATIAL VARIABILITY
Vegetation performs an important role in controlling the spatial variability of
surface soil properties that influence intemll erosion. Canopy cover and standing
biomass are negetatively related to intemll erosion (Rauzi et al., 1968; Devaurs
& Gifford, 1984; Blackburn, 1975; Blackburn et al., 1986; Thurow et al., 1988a)
and are generally regarded as the primary factors influencing intemll erosion
(Lane et al.. 1992). Results from three studies (Johnson & Blackburn. 1989;
Johnson & Gordon, 1988; Thurow et al., 1988a) will be used to help illustrate the
influence of vegetation on intemll erosion.
SOURCES OF VARIATION IN INTERRILL EROSION
3
Results from the study conducted in southwestern Idaho by Johnson and
Blackburn (1989) show the influence of plant canopy cover on interrill erosion
from sagebrush dominated rangelauds. They report the results of rainfall simulation research that contributed to the National Water Erosion Prediction Project
effort to develop improved erosion prediction technology (Laflen et al., 1991).
Research was conducted on three Wyoming big sagebrush sites (Coyote Butte,
Nancy, and Summit), each located on a different soil series. The soils of the
three sites were (i) Coyote Butte site, Power series (fine-silty, mixed, mesic
Xerollic Haplargid), (ii) Nancy site, Gariper series (fine, montmorillonitic,
mesic, Xerollic Paleargid), and (iii) Summit site, Saralegui series (coarse-loamy,
mixed, mesic Xerollic Haplargid). The three-treatments replicated two times at
each study site were undisturbed, clipped (standing vegetation harvested to
ground level by clipping), and bare (standing vegetation harvested to ground
level by clipping and all litter, cryptograms, grass root crowns, and surface rock
harvested by hand). Rainfall simulation plots were 10.7 m long by 3.05 m wide
and were arranged to accommodate a rotating-boom rainfall simulator
(Swanson, 1965). Simulated rainfall was applied at a design rate of 63.5 mm rr1
for a duration of 1 h for the dry soil run and a duration of 30 min for the wet soil
run, 24 h later. Only the results from the wet run are reported here, for detailed
site description, methods, procedures, and results see Johnson and Blackburn,
1989.
Runoff and soil loss were greater on the bare treatments than on either the
undistributed or clipped treatments. Similar runoff and soil loss were measured
from the undisturbed (shrub 17.1% and grass-fort) 5.1% canopy cover) and
clipped (all standing shrub and grass-forb cover removed) treatments (Table
1-1). These results illustrate that factors other than canopy cover or standing biomass are controlling runoff and erosion from sagebrush dominated rangelands.
The indirect effects of shrub canopy cover is much greater than any direct effects
canopy cover might have on interrill erosion. Shrubs and bunchgrasses influence
the site by modifying the microenvironment through addition of litter and organic matter to the soil surface, capturing wind and water bom soil particles, and
enhancing the micro-flora and micro-fauna.
The spatial distribution of amount and type of vegetation has a spatial influence on surface soil characteristics, infiltration, and interrill erosion rates on
rangelands (Blackburn, 1975; Blackburn & Wood, 1990; Johnson & Gordon,
1988; Blackburn et al., 1990; Blackburn et al., 1992; Wood & Blackburn, 1984;
Thurow et al., 1988a). The results of research reported by Johnson and Gordon
(1988) and Thurow et al. (1988a) are provided to illustrate this principle.
TaMc 1-1. Mean treatment ground cover, simulated rainfall (design rate 63.5 mm Ir'), runoff and soil
Ion, data avenged across the Coyote Butie, Nancy, and Summit Study Sites, southwestern Idaho,
Wet Run (Johnson & Blackburn, 1989).
Treaimem
Undisturbed
Clipped
Bin
Ground cover
Simulated rainfall
Runoff
%
68
72
33
30.2
31.1
30.4
4.9
4.6
12^
kjta"
110
60
2000
4
BLACKBURN & PIERSON
The Johnson and Gordon study was conducted on the Reynolds Creek
Watershed in southwestern Idaho on the Nancy site described in the previous
study by Johnson and Blackburn (1989). Circular plots, 0.91 to 0.95 m:, were
installed around shrub coppice areas and in interspace areas between coppices. A
rainfall simulator (Neff, 1982) was used to apply water at a design rate of 127
mm h~ l . Runoff, sediment, vegetation cover, surface topography, and soils data
were measured for each plot. Surface soil organic C, sand content and infiltration
rates were greatest for shrub coppice areas (Table 1-2), where bulk density, silt
content and interrill erosion were greatest for interspace areas. Interrill erosion
was nine times greater from the interspace areas than from coppice areas.
Sagebrush interspace areas are characterized by significant surface soil spatial
variability (Eckert et al., 1986). The study reported in this special publication by
Pierson et al. (1994) (see Chap. 3, this publication) discusses in more detail sagebrush interspace spatial variability and its relationship to interrill erosion.
Research conducted at the Texas Agricultural Experiment Station located on
the Edwards Plateau, Texas, is presented to help illustrate the influence of vegetation on surface soil and interriU erosion (Thurow et al., 1988a). The study site
is an oak-grassland characterized by clusters of live oak (Quercus virginiana
Mill.) and grass dominated interspaces. The most common bunchgrasses are
side-oats grama (Bouteloua curtipendula (Michaux) Torrey) and Texas wintergrass (Stipa leucotricha Trin. & Rupr.) with curly mesquite (Hilaria belangeri
(Steudel) Nash) being the dominant sodgrass. Infiltration rates, interrill erosion,
ground cover characteristic, vegetation standing crop, mulch accumulation, and
surface soil physical properties of grass interspaces were sampled bimonthly
from March 1978 through March 1984. The live oak clusters were sampled once
at the start and once at the end of the study. A drip-type rainfall simulator
(Blackburn et al., 1974) was used to apply water at a design rate of 203 mm Ir1
for 30 min to 0.45 m2 plots.
Surface soil organic C, aggregate stability, and infiltration capacity were
greater and interrill erosion lower in live oak clusters than in the grass interspaces
(Table 1-3). Interrill erosion was 134 times greater in the bunchgrass areas and
316 times greater in the sodgrass areas than in live oak clusters. Surface soil differences between the bunchgrass and sodgrass sites were much less than between
live oak clusters and grass interspace. Organic C, aggregate stability, and infiltration capacity, however, were greater and interrill erosion was less than one-half
for bunchgrass sites than for sodgrass sites. These examples support the premise
Tabte 1-2. Surface soil (0 to 25 mm) organic C bulk density, silt and clay content, infiltration capacity, and inierrill erosion of the shrub coppice and interspace microsites, Nancy Study Site,
Reynolds Creek Watershed in southwestern Idaho, Wei Run (Johnson & Gordon. 1968).
Uuiabte
Shrub coppice
Interspace
OrtnacC.%
Bulk densily, Mg m>
Silt*
day. %
Infiltration opacity, mm IT'
Imemll Erauxt. kg tu
4.0
1.2
38
12
41
45
Z2
1-5
45
13
22
418
5
Table 1-3. Surface soil (0 to 50 mm) organic C, aggregate siabilily. bulk density, sill and clay conlem. infiltration capacity, and intertill erosion of live oak clusters, bunchgrass. and sodgrass interspace siles, Edwards Plateau, Texas, Wet Run (Blackburn et al., 1986; Knight et al.. 1984; Thurow
e t a l - 1986, I988a, b).
Variable
Organic C, %
Aggregate stability, %
Bulk density. Mg m '
Silt.%
Clay, %
Infiltration capacity, mm h
Interrill Erosion, kg ha~
Live oak cluster
7.1
80
0.8
37
43
200
5
Bunchgrass
3.2
75
0.82
41
186
672
Sodgrass
2.9
70
0.89
44
39
170
1579
that the amount and type of vegetation on semiarid rangelands spatially influence
the microenvironment, surface soil characteristics, and interrill erosion. Shrubs or
live oak vegetation exhibit a stronger influence on factors controlling erosion
than grasses, thereby creating spatially distributed microsites that function as run
on and sediment deposition areas with extremely low erosion rates. For all but the
most extreme rainfall events these shrub or tree influenced microsites are erosion
safe sites and the majority of the interrill erosion occurs in the spatially distributed interspace areas.
VEGETATION INFLUENCES—TEMPORAL VARIABILITY
The temporal response of surface soil factors and intem'll erosion rates are spatially different for shrub coppice and shrub interspace areas. Results from studies
(Blackburn et al., 1990, 1992) conducted at the Quonset site on the Reynolds Creek
Watershed in southwestern Idaho are presented to illustrate the nature of temporal
variability on sagebrush dominated rangelands. The vegetation of the site is characterized by Wyoming big sagebrush (Artemisia tridentata Beetle & Young
wyomingensis), Sandberg bluegrass (Poa secunda J.S. Presl), and cryptograms. A
drip-type rainfall simulator (Blackburn et al., 1974) was used to apply water at a
design rate of 99.2 mm h"1 for 30 min to 254 mnr plots placed in shrub coppice
and shrub interspace areas. Rainfall was simulated six times from February through
June on soil that was continuously frozen, diumaliy frozen, or unfrozen. Ground
cover, surface soil bulk density, antecedent soil water, texture, aggregate stability,
and soil frost were measured for each plot. Interrill erosion amount and variability
from the coppice areas was low for all sample periods (Fig. 1-1); however, erosion
from the shrub interspace areas was most susceptible to erosion immediately following a prolonged period of diurnal freeze-thaw cycles that left the surface soil in
a super-saturated state. The frequency of occurrence and length, of this susceptible
period is dependent on local climate and soil water conditions, for this study interspace soils were highly susceptible to interrill erosion for a 19-d period. If an
intense storm or rapid snow-melt event were to occur during this period of high erosional susceptibility an extreme erosion event would occur.
The second example examines the spatial and temporal variability of surface soil aggregate stability, a factor that has been demonstrated to have a strong
BLACkBl RN 4 MEKSON
coppice
2
—
Interspace
4000
-m
i
3000
E S 2000
CO
1000
.
n —
2-\5 222
3/1
3/15 4/13 620
Date (1989)
Fig. 1-1. Cumulative sediment yield during a 30-min interval by sample date for shrub coppice and
interspace areas, Reynolds Creek Experimental Watershed, Idaho (Blackburn et al., 1990)
relationship with interrill erosion. This research was conducted at Reynolds
Creek Watershed in southwestern Idaho and at the Texas Agricultural
Experiment Station on the Edwards Plateau, Texas (Blackburn et al., 1992). The
Idaho study site is the same one described in the previous example and the Texas
study site was described earlier. In the Idaho study, aggregate stability was measured on five microsites, shrub coppice, moss-grass, moss-interspace, bare soil,
and vesicular soil ranging in size from 0.3 to 2 m: were sampled eight times
from October 1989 through July 1990. Aggregate stability showed a significant
spatial and temporal response between and within each microsite (Fig, 1-2).
Aggregate stability was generally lowest for the vesicular microsites and highest for the coppice microsites. The other microsites showed mixed responses
over time with values for each soil falling somewhere between the coppice and
vesicular microsites. All microsites showed the general trend of decreasing
"
O
N
O
J
F
M
A
M
J
J
Sampling Date (1989-1990)
ed, Wafco (Bbddwrn ef aL, 1992).
1 97B
1 979
1 980
1 98 1
1 982
1 983
1 984
Time (Years)
Fig. 1-3. Aggregate stability by sample date for bunchgrass and sodgrass microsi
siles, Edwards
Plateau. Texas (Blackburn et al., 1992).
aggregate stability during approximately the first three sample dates when the
soils were wetting and beginning to freeze and thaw. The soil in the interspa< es
is more susceptible lo temporal variations in soil microclimatic conditions th.ni
soil under shrub canopies.
In the Texas study aggregate stability was measured bimonthly from March
1978 through March 1984 from bunchgrass and sodgrass dominated microsites.
On the Edwards Plateau, soil aggregates are primarily broken down by slaking
and raindrop impact and are not exposed to extreme freeze-thaw processes.
Significant seasonal trends in aggregate stability occurred under both bunchgrass
and sodgrass vegetation growth forms (Fig. 1-3) and temporal changes were
much greater than the small spatial difference that occurred in surface soils
between the two grass growth forms.
These examples support the premise that vegetation induced microsite variability differs spatially and temporally in their erodability. The surface soil parameters of sagebrush dominated landscapes display greater spatial than temporal
variability, but the bunchgrass-sodgrass dominated landscapes exhibited greater
temporal variability than spatial.
CONCLUSIONS
Interrill erosion processes on semiarid rangelands are characterized by significant vegetation induced spatial and temporal variation. The spatial distribution
of the amount and type of vegetation is an important factor controlling surface soil
characteristics that are known to influence interrill erosion. The research findings
presented in this paper support the premise that spatial and temporal variations in
native vegetation and climate found on semiarid rangelands lead to surface soil
properties that are spatially and temporally distributed. Vegetation growth form
and normal variations in climate are the primary factors influencing the spatial and
temporal variability of surface soil processes controlling interrill erosion.
8
BLACKBURN & PIERSON
Current rangeland erosion modelling efforts do not adequately account for
vegetation induced variability in interrill erosion processes. Methods for estimating parameters need to be improved to better account for the variability found on
range lands.
References
Blackburn, W.H. 1975. Factors influencing infiltration and sediment production of semi-arid rangelands in Nevada. Water Resour. Res. 11:929-937.
Blackburn, W.H. 1983. Influence of brush control on hydrologic characteristics of range watersheds,
p. 73-88. In K.C. McDaniei (ed.) Proc. of Brush Management. Symp., Albuquerque, NM. 16
Feb. Texas Tech Univ. Press, Lubbock,
Blackburn, W.H. 1984. Impacts of grazing intensity and specialized grazing systems on watershed
characteristics and responses, p. 927-983. In Developing strategies for rangeland management.
Nat. Res. Council-Nat Academy of Sci. Westview Press. Boulder, CO.
Blackburn, W.H.. R.O. Meeuwig, and CM. Skau. 1974. A mobile infiltrometer for use on rangeland.
J. Range Manage. 27:322-323.
Blackburn. W.H., F.B. Pierson, C.L Hanson, T.L. Thurow, and A.L Hanson. 1992. The spatial and
temporal influence of vegetation on surface soil factors in scmiarid rangelands. Trans. ASAE
35:479-486.
Blackburn, W.H., F.B. Pierson. and M.S. Seyfried 1990. Spatial and temporal influence of soil frost
on infiltration and erosion of sagebrush rangelands. Water Res. Bull. 26:991-997.
Blackburn, W.H., T.H. Thurow, and C.A, Taylor, Jr. 1986. Soil erosion on rangeland. p. 31-39. In
Proc. of Symp. on Cover. Soils and Weather Data on Rangeland Monitoring. Orlando, FL. 12
Feb. Soc. for Range Manage, Denver CO.
Blackburn, W.H.. and M.K. Wood. 1990. Influence of soil frost on infiltration of shrub coppice dune
and dune interspace soils in southeastern Nevada, Great Basin NaL 50:41—46.
Branson, F.A.. G.F. Gifford, K G Renard. and R.F. Hedley. 1981. Rangeland hydrology. Range Sci.
Ser. No. 1. Kendall/Hum Publ. Co., Dubuque, IA.
DeCoursey, D.G. 1988. A critical assessment of hydrologic modeling, p. 478-493. In Proc. of
Modeling Agricultural, Forest, and Rangeland Hydrology. ASAE Int. Symp. Chicago, IL 12-13
Dec. ASAE Pub. 07-88. ASAE, St. Joseph, MI.
DeCoursey. D.G. 1992. Developing models with mote detail: do more algorithms give more truth?
Weed Tech. 6:709-715.
Dcvaurs, M.A., and G.F. Gifford. 1984. Variability of infiltration within large runoff plots on rangelands. J Range Manage. 37:523-528.
Ecfcert, R.E.. Jr., F.F. Peterson, and J.T. Belton. 1986. Relation between ecological range condition
and proportion of soil surface types. J. Range Manage. 39:409-414.
Ferreira, V.A.. and R.E. Smith. 1988. The limited physical basis of physically based hydrologic models, p. 10-18. In Proc of Modelling Agricultural, Forest, and Rangeland Hydrology. ASAE ITJL
Symp. Chicago. IL 12-13 Dec., ASAE Pub. 07-88. ASAE, St. Joseph, ML
Foster. G.R. 1982. Special problems in the application of the t'SLE to rangelands. p. 96-100. In Proc
of Estimating Erosion and Sediment Yield on Rangelands Workshop. Tucson. AZ_ 7-9 Mar.
1981. L'SDA-ARS Western Region Pub. ARM W-26 USDA-ARS Oakland, CA.
Foster. G.R.. J.R. Situation, K.G. Renard, LJ. Lane, and H.B. Osbom. 1981. Discussion of application of the Universal Soil Loss Equation to rangelands on a per-storro basis. J. Range Manage.
34:161-164.
Gifford. G.F., and R.H. Hawkins. 1978. Hydrologic impact of grazing on infiltration: A critical
review. Water Resour. Res. 14:305-313.
Johnson, C W. and W.H. Blackburn. 1989. Factors contributing to sagebrush rangeland soil loss.
Trans. ASAE 32:155-160.
Johnson. C.W.. and K.A Gcbhardt. 1982 Predicting sediment yields from sagebrush rangelands.
p. 145-156 In Proc. of Estimating Erosion and Sediment Yield on Rangelands Workshop.
Tucson. AZ 7-9. Mar.. 1981, USDA-ARS Western Region Publ ARM-W-26 USDA-ARS.
Oakland, CA.
Johnson. C.W., and N.D. Gordon. 1988. Runoff and erosion from rainfall simulator pkMs on sagebrush rangeland Trans. ASAE 31:421-427.
Johnson, C.W., N.D. Gordon, and C.L. Hanson 1985. Northwest rangeland sediment yield analysts
by the MUSLE Trans. ASAE 28:1889-1895.
SOURCES OF VARIATION IN INTERR1LL EROSION
9
Johnson, C.W., MR. Savabi. and S.A. Loumis. I9S4. Rangeland erosion measurements lor the
USLE. Trans. ASAE 27:1313-1320
Knight. R.W, WH. Blackburn, and L.B. Merrill. 1984. Characteristics of oak motto. Eduards
Plateau. Texas. J. Range Manage. 37:534-537.
Laflen. J.M., L.J. Lane, and O R . Foster. IW1 WEPP. A ne* generation of erosion prediction technology, J Soil Water Conserv. 46:34-38.
Lane. LJ.. K i. Renard, G.R. Foster, and J.M. Laden. 1992. Development and application of modern soil erosion prediction technology—the USDA experience. Ausi. J. Soil Res. 30:893-912.
League. K.M . and R.A. Freeze. 1985. A comparison of rainfall-runoff modeling techniques on small
upland catchments. Water Resour. Res. 21:229-248.
Mceuwig. R.O. 1970. Infiltration and soil erosion as influenced by vegetation and soil in northern
Utah. j. Range Manage. 23:185-188.
National Research Council. 1991. Opportunities in the hydroloeic sciences. Nat. Academy Press.
Washington. DC.
Ncff. E.L 1982. Performance characteristics and field operation of two rainfall simulator;. BLMARS Intcragency Agreement YA-515-IA6-3 Rep. USDA-ARS. Washington. DC.
Packer, P.E. 1953. Soil stability requirements for the Gallatin elk winter range. J. Wildl. Manage.
27:401-110.
Rauzi. F, C.I. Fly. and EJ. Dyksierhuis 1%8. Water intake on midcomincntal rangelands as influenced by soil and plant cover. USDA Tech. Bull. 1390. USDA, Washington DC.
Renard, K.G., G.R. Foster, G.A- Weesies, and J.P. Porter. 1991. RUSLE, Revised universal soil loss
equation. J. Soil Waler Conserv 46:30-33.
Swanson, N.P. 1965- Routing-boom rainfall simulator. Trans, ASAE fri.~
Thurow. T.L. W.H. Blackburn, and C.A. Taylor, Jr. 1986. Hydrologic characteristics of vegetation
types as affected by livestock grazing systems. Edwards Plateau. Texas. J. Range Manage.
39--50S-509.
Thurow, T.L. W.H. Blackburn, and C.A. Taylor. Jr.. 1988a. Infiltration and mtemll erosion responses to selected livestock grazing strategies. Edwards Plateau. Texas. J. Range Manage.
41:296-302.
Thurow. T.L, W.H. Blackburn, and C.A. Taylor. Jr.. I988b. Some vegetalkm responses to selected
livestock grazing strategies. Edwards Plateau. Texas, J. Range Manage. 41:108-114.
Trieste, DJ.. and G.F. Gifford. 1980. Application of the universal soil loss equation to rangelands on
a per-storm basts. J. Range Manage. 33:66-70.
Wilcox. B.P.. M. Sbaa. WH. Blackburn, and J.H. Milligan. 1992. Runoff prediction from sagebrush
rangelands using water erosion prediction project (WEPP) technology. J. Range Manage.
45:470-475.
Williams. J.R. 1975- Sediment-yield prediction with universal equation using runoff energy factor,
p. 244-53. In Present and prospective technology for predicting sediment yields and sources.
USDA-ARS. Washington, DC.
Wtschmeier, W.H. 1976. Use and misuse of the universal soil loss equation. J. Soil Water Conserv.
31:5-7.
Wischrneier, W.H.. and D.D. Smith. 1965. Predicting rainfall-erosion tosses from cropland east of the
Rocky Mountains. USDA Agric. Handb. 282. USDA, Washington. DC.
Wischmeier. W.H.. and D.D. Smith. 1978. Predicting rainfall-erosion losses. USDA Agric. Handb.
537. USDA and Sci. and Educ. Admin.. Washington, DC.
Wood, M.K.. and W.H. Blackburn 1984. Vegetation and soil responses to cattle grazing systems in the
Texas Rolling Plains. J. Range Manage. 37.303-3OR
Zjngg. R.W. 1940. Degree and length of land slope as it affects soil loss in runoff. Agric. Eng.
21:59-64.
The WEPP Model and Its
Applicability for Predicting
Erosion on Rangelands
John M. Laflen. Dennis C. Flanagan, and James C. Ascough, II
VSDA-ARS National Soil Erosion Research Laboratory
West Lafayette, Indiana
Mark A. Weltz and Jeffry J. Stone
USDA-ARS Southwest Watershed Research Center
Tucson, Arizona
ABSTRACT
The Water Erosion Prediction Project (WEPP) model is intended to replace the
Universal Soil Loss Equation for predicting soil erosion. The WEPP is a fundamental
process-based model that operates on a daily time step to estimate land, soil and vegetation conditions when a rainfall event occurs, and then uses this information to predict the
hydrology and erosion of single events. The WEPP is used in conjunction with an input
climate data file, long term estimates are based on the accumulated erosion occurring during the period of record covered by the input climate file. This chapter describes the representation of rangelands for making estimates of the land, soil, and vegetation conditions,
and their effect on soil erosion estimates. Additionally, shortcomings and advantages of
WEPP for erosion prediction on rangelands is discussed.
The WEPP brings to the natural resource manager a lool for not only the evaluation of the impacts of management on soil erosion, but also for the evaluation of
offsite impacts related to management decisions.
The USLE (Universal Soil Loss Equation; Wischmeier & Smith, 1965,
1978) and its revision RUSLE (Revised USLE; Renard et al., 1991) is an erosion
prediction technology that has served mankind well. Because of its empirical
nature, however, it has proven to be difficult to apply in some cases, particularly
to offsite problems. Additionally, the empirical database to support its application
to rangelands and to many other situations is very small.
In 1969, Meyer and Wischmeier presented a model of the water erosion
process thai was more basic in nature. The CREAMS model (Chemicals, Runoff,
and Erosion from Agricultural Management Systems; U.S. Department of
Agriculture, 1980) included the more fundamental processes of water erosion
and sediment transport. A more recent effort was initiated to replace the USLE
Copyright O 1994 Soil Science Society of America, 677 S. Scgoe Rd, Madison. Wl 53711, USA.
Variabtluy of Rangtlond Water Erosion Processes, SSSA Special Publication 38.
12
LAFLENETAL.
with fundamental erosion process technologies in a broad based project titled
WEPP (Water Erosion Prediction Project; Foster & Lane, 1987; Nearing & Lane,
1989).
The WEPP is expected to be ready for use at the field level in 1995.
Validation, testing, receding, development of interfaces and parameterization are
underway. Prior to 1995, considerable work is required by action agencies to prepare for implementation. These efforts include training, selection of equipment,
development of input data sets, the development of guidelines and procedures for
use of WEPP These are major tasks and require considerable time and effort.
This chapter is not intended to be a general critique of the WEPP model, but
rather an examination of model components where representing rangeland conditions or parameterizing and modeling the processes may be difficult. These
components are related to hydrology, plant growth, erosion, and soil.
DESCRIPTION OF WEPP
The WEPP is a daily simulation model that computes the conditions of ihe
soil and plant system that are important in runoff and soil erosion. If rainfall
occurs, WEPP computes surface runoff. If surface runoff occurs, WEPP computes the soil that is detached and deposited down a hillslope and tt.e amour'
delivered to a channel at the foot of a slope. These are all computed in the hillslope version of WEPP. Two additional versions (watershed or grid) are used to
compute the erosion, deposition, and delivery of sediment through the channel
system on the field or farm.
The WEPP represents the area where sheet and rill erosion occurs as a series
of overland flow elements (OFE) beginning at the top of the slope and ending Ji
a field boundary or a channel at the bottom of a slope. Each OFE is homogeneous
with regard to the ecosystem, soil, and management.
Within an OFE, sediment detachment and transport occurs on rill and interrill areas. On interrill areas, the detachment is caused by raindrop impact, and
transport is in very shallow flows that are impacted by raindrops. The detached
and transported soil on an interrill area is delivered to a rill. Sediment detachment
in a rill is caused by the hydraulic shear of the flow carried by the rill and is not
affected by raindrops on the water surface. Sediment transport in a rill is also not
affected by rainfall. Sediment deposition may occur in a rill if sediment load
exceeds the transport capacity of the flow.
Plant Growth
The status of plants and plant residue when an erosion event occurs is vital
to accurate estimation of soil detachment and transport. The status of below and
aboveground biomass must be accurately estimated to evaluate the effect of various management alternatives on soil erosion. The WEPP calculates on a daily
basis plant growth and the decomposition and accumulation of residue and litter.
Important plant growth characteristics include canopy cover and height,
mass of live and dead below and aboveground biomass, leaf area index and basal
area, residue, and litter cover. Information about management are input to the
THE WEPP MODEL AND ITS APPLICABILITY
13
model. Many annual and perennial crops, management systems and operations
that may occur on cropland, rangeland, forestland. pastures, vineyards, and gardens have been parameterized. Major efforts are underway to develop an expert
system for selection of parameters to use in WEPP (Deer-Ascough et al., 1993).
While this work is presently for cropland parameters, it is expected that parameters for rangelands will eventually be included.
Representation of the complex plant ecosystem on rangelands have proven
difficult. On croplands, there is generally only one crop grown at a time.
Rangelands are a complex system where numerous species coexist simultaneously, each using different amounts of water each day. and each having different
above and below ground biomass accumulation rates. Additionally, they withdraw water from different soil depths. A question not yet fully answered is can
we represent this complex system with a dominant plant, a few plant species, or
a representative plant community? This question must be answered and necessary parameterization accomplished if we are to have an erosion prediction system on rangelands fully capable of representing existing and potential ecosystems and the varied management schemes practiced and proposed.
[Decomposition is important in estimating residue and litter cover and soil
erosion on rangelands and croplands. Coefficients for use in estimating litter and
residue decomposition have been determined for many crops, but there has been
little work on estimation of decomposition rates of surface litter found on rangelands. Furthermore, the location of litter is also highly variable. There may be in
some ecosystems an accumulation of litter under shrubs, but this may not be the
case for other litter that is more accessible and vulnerable to animal traffic. Both
of these are areas of research needed to apply WEPP to rangelands.
Hydrology
The hydrologic cycle must be well represented if erosion and sediment delivery are to be accurately predicted. The WEPP uses several climate variables,
including storm rainfall amount and duration, ratio of peak rainfall intensity to
average rainfall intensity, time to peak intensity, daily maximum and minimum
temperature, daily miles of wind by station and its direction, and solar radiation.
These variables are required in components related to plant growth and surface
litter decomposition, water balance, and in estimating runoff volume, duration,
and peak rate.
The hydrologic component of the WEPP hillslope profile model is derived
from the research Infiltration and Runoff Simulator (IRS) model (Stone et al.,
1992). The IRS model is an event-based model that uses the Green-Ampt MeinLarson (GAML) infiltration equation as modified by Chu (1978). and the kinematic wave equations as presented by Lane et al. (1988).
Several modifications have been incorporated into the IRS model to address
the implementation constraints of simplicity and speed of execution. Rainfall disaggregation (Nicks & Lane. 1989) of daily precipitation was added to reduce the
amount of data needed to describe rainfall intensity needed by both the GAML
model and the interrill erosion model. An approximate method for computing the
peak discharge at the bottom of a hillslope profile (Hernandez et al.. 1989) was
added to reduce model run time. Parameters for the hydrologk: component can
14
LAFLENETAL.
be identified through calibration, if observed data are available or estimated by
the model from measurable physical properties of the soil and vegetation (Rawls
et al., 1983; Weltz et al., 1992). In continuous simulation mode, baseline hydrologic parameters are adjusted in response to changes in canopy cover and litter
caused by vegetation growth and decomposition, herbicide application, burning,
and grazing by animals.
Preliminary testing of the WEPP model on rangelands has been started using
data from the semiarid rangeland Walnut Gulch Experimental Watershed.
Tiscareoo et al. (1992) found that the hydrologic response of the hillslope model
is most sensitive to rainfall amount, duration, and GAML baseline saturated
hydraulic conductivity. For a given runoff producing rainfall event, the response
is most sensitive to GAML baseline saturated hydraulic conductivity, soil moisture, and aboveground biomass. The parameter estimation techniques within the
model and the procedure used to disaggregate rainfall events have been identified
(van der Zweep et al., 1991) as critical components of the model requiring additional research. Improvements in estimation of the GAML baseline saturated
hydraulic conductivity parameter and in adjusting its baseline value to account
for the influence of changes in canopy cover and surface Utter may greatly
improve model accuracy.
Erosion
The WEPP models erosion on a rangeland hillslope by dividing the soil surface into two regions: rill (concentrated flow paths) and interrill. Rills are flow
paths that form as water flow concentrates. Detachment in these channels is largely a function of flow shear stress (force exerted by water flow on the bed and
banks). In many landscapes, these flow paths form at fairly regular intervals.
The area between rill channels is called the interrill area. Water flow on interrill areas is shallow, and most of the soil detachment here is due to raindrops
impacting the soil surface. The raindrops also act to enhance the transport of previously detached sediment from the interrill area to the rill channels. Rills are the
major sediment transport pathway for all sediment detached—both that from the
rills and that supplied to the rills from the interrill areas.
The basic equation used in the WEPP erosion component is a steady state
sediment continuity equation:
dG/djc = D +Z>
[1]
l
where G is sediment load in the flow down a hillslope (kg s~' m~ ), x is distance
downslope (m), Z) is the interrill sediment delivery rate to the rills (kg s*1 iir2) and
Dr is the rill detachment or deposition rate (kg s~' nr1) (Nearing et al., 1989;
Foster el al., 1989). For erosion computations for each individual storm, the time
period used is the effective duration of runoff computed in the hydrology component of the model. Estimates of dG/dx are made at a minimum of 100 points
down a profile, and a running total of the sum of all detachment and deposition
at each point from each storm is used to obtain monthly, annual, and average
annual values for the simulation.
THE WEPP MODEL AND ITS ATPUCABILITY
15
The interrill component of WEPP is currently a fairly simple sediment delivery function:
Z> = K. It2 Gf Ct S,
[2]
where D is delivery of detached sediment to the rill (kg nr2, K is the interrill
credibility (kg s~' nr4), It is the effective rainfall intensity (ms'1) occurring during
the period of rainfall excess, Gt is a canopy cover effect adjustment factor, C is
a canopy cover effect adjustment factor, and 5f is a slope adjustment factor. The
/c is computed through a procedure that examines the time period over which
rainfall excess is occurring. The effective duration of rainfall excess is passed to
the erosion component from the hydrology component. Equation [2] lumps
together the processes of detachment, transport, and deposition on the interrill
areas.
The Ct is a function of the fraction of the soil surface area covered by canopy
and the height of the canopy. The Gt is a function of the fraction of the interrill
area covered by surface litter, residue, and rocks. The Sf is a function of the interrill slope:
Sf =1.05-0.85 e<-"-S>
[3]
where B is the interrill slope angle. These functions are based on reasonable fits
to data reported by Meyer (1981), M*yer and Harmon (1984, 1989), and Watson
and Laflenf 1986).
An improvement to the WEPP erosion component might be the modeling of
detachment, sediment transport, and sediment deposition as separate processes on
the interrill regions to arrive at a better value for D. Since interrill processes may
be more dominant than rill processes on consolidated rangeland soils, this
improvement to the interrill component might improve erosion estimates for
rangeland situations.
Concentrated flow paths are the major pathway for sediment movement
down most hillslopes. Water flowing in such rills has the ability to both transport
sediment and detach additional soil. When the rill flow becomes laden with sediment from either sediment supplied from the interrill areas or from sediment
detached in the rill channel itself, the rill flow loses some of its ability to detach
soil and transport sediment. If too much sediment is supplied and the flow system is overloaded, then no rill detachment can take place, and sediment deposition occurs. One of the strengths of WEPP is its ability to estimate both rill
detachment and deposition, allowing comprehensive evaluation of both on-site
and off-site effects of erosion.
The WEPP uses separate equations to simulate rill detachment and deposition. Rill detachment is predicted to occur when the flow shear stress exerted on
the soil exceeds a critical threshold value, and sediment transport capacity is
greater than the sediment load:
D = Kt (TAU -TAU^ (1 - C/T)
(4]
16
LAFLENETAL.
where Ds is the rill detachment rate (kg s~' nr2), Kr is the adjusted rill credibility
parameter (s rrr1), TAU is the flow shear stress (PJ, TAU( is the critical flow shear
stress (Pa), G is sediment load (Kg s'1 nr1), and 7"c is the flow sediment transport
capacity (kg s~' nr1). One can see from this equation that as the flow fills with sediment (G approaches Tc) that the rill detachment rate will be predicted to decrease.
Sediment transport capacity in the WEPP model is predicted using the equation:
Te = Jfc( TAU1-'
J 2
[5]
0
where ^ is a transport coefficient [m° s (kg ^)] calibrated and obtained by applying the YaJin (1963) equation at the end of the slope profile (Finkner et al., 1989).
When the sediment load exceeds the sediment transport capacity, the equation used by WEPP to predict deposition is:
D = ((BETA x VJq) (7e - G)
[6]
3
where DT is the rill deposition rate (kg s~' nr ), BETA is a rainfall-induced turbulence
factor (currently set to 0.5), Vctt is an effective particle fall velocity
(m s~'), and q is flow discharge per unit width (nr s~l). An area of concern with the
current deposition equation is the estimation of the Vtf term based upon the particlesize distribution. An evaluation of the procedure that uses the smallest size classes is
underway to determine how well the method and the deposition equation perform.
Other areas for future improvement in the prediction of deposition would be to: (i)
compute the BETA coefficient as a function of rainfall intensity and flow depth,
instead of assigning it a constant value; and, (ii) alter the sediment transport equation used so that it includes a rainfall-enhancement term.
Rill characteristics such as spacing, width, and shape are important in estimating soil erosion. For raogelands, rill spacing is estimated as the average
spacing of vegetation but spacing is never <0.5 m or >5 m. Estimation of rill
width is based on flow and topographic characteristics, while rill shape is
always assumed to be rectangular. These assumptions are being evaluated and
are subject to change as additional information becomes available. Sensitivity
analyses to date have indicated that rill characteristics are not as significant as
several other characteristics in determining erosion and sediment delivery from
rangelands.
SoU
The soil component deals with temporal changes in soil properties important
in the erosion process, and in estimation of surface runoff rates and volumes.
These include random roughness, ridge height, saturated hydraulic conductivity,
soil credibilities, and bulk density. The effects of tillage, weathering, consolidation, and rainfall are considered in estimating the status of soil properties.
Baseline interrill and rill credibility, and critical hydraulic shear for a freshly tilled condition, are adjusted to other conditions based on time since tillage for
cropland soils. For rangeland soils, the baseline condition is thai of a long-term
undisturbed soil under rangeland conditions with surface residue removed. For
17
both range and cropland soils, adjustments to intemll credibility are based on live
and dead roots in the upper 150 mm of the soil and to rill credibility because of
incorporated residue in the upper 150 mm of the soil.
Past efforts to model erosion processes have used USLE relationships for estimating soil credibility. A major WEPP effort has been extensive field studies (Elliot
et al., 1989; Simanton et al., 1987) to develop the technology to predict credibility
values for cropland and rangeland soils from soil properties. A major effort continues for both rangelands and croplands to expand the data bases that support WEPP.
EROSION PREDICTIONS
The use of WEPP to evaluate different management is illustrated by applying the watershed version of WEPP to two common rangeland management scenarios; cattle grazing when the vegetation is brush, and cattle grazing the same
area when it is in grass, perhaps after brush is Controlled by herbicide application
and the grass is established. The watershed is a hillslope on the Lucky Hills 103
watershed near Tombstone, AZ (van der Zweepet al., 1991), and data from these
simulations are presented in Tables 2-1 and 2-2. The WEPP hillslope version
Table 2-1. Information on the management practices simulated using WEPP on the Lucky Hills 103
watershed. Walnut Gulch Experimental Watershed.
Vegetation
Management
practice
AMUt
(ha cow')
Brush
Brush
Grass
Grass
Grass
No grazing
Moderate
grazing
No grazing
Moderate
grazing
Heavy
grazing
Utilization^
Herbicide
Seeding
%
0
18
0
18
none
none
none
none
0
12
0
20
once
once
once
once
2
85
every 4
every 4
tAUM is animal unit month.
Ûtilization is percentage of total standing bi>
y
ed by grazing livestock.
Table 2-2. Average annual watershed runoff volume, 2-yr return period watershed peak discharge.
and hilislope and watershed sediment yield for five management practices for Lucky Hills 103
watershed. Walnut Gulch Experimental Watershed.
2 yr
tegctukn
Brush
Brush
Grass
Grass
Grass
Management
practice
Moderate
grazing
No grazing
Moderate
grazing
Heavy
grazing
Watershed
runoff volume
watershed
peak discharge
nun
mm h '
Hillslope
20
27
33
0.92
134
127
184
15
17
24
26
0.06
a 10
1.70
1.74
18
38
0.16
2.13
Sediment Yield
LAFLENETAL
18
93-0 was also applied to grazing intensity effects on soil erosion, runoff, and sediment concentration for the Edwards Plateau in Texas (Fig. 2-1, 2-2f and 2-3).
The examples shown are for WEPP simulations, the WEPP models are under
development and we MM a mpletel) verified, vali lated, and parameterized Wha
WEPP is fully verified, validated, and parameterized, exact quantitative results
will probably be somewhat different. We do expect the present WEPP model with
our present parameterization to represent trends that would occur in nature.
Table 2-1 lists the characteristics of the management practices for the Lucky
Hills watershed near Tombstone, AZ, The two bmsh scenarios consist of no grazing or moderate grazing with no herbicide application or reseeding of grass. The
three grass scenarios consist of an initial herbicide treatment to remove the brush,
reseeding with grass, and three grazing intensities. The heavy grazing management practice necessitates reapplication of the herbicide and reseeding every 5 yrs,
The climate (precipitation, ternperature,and solar radiation) used for the simulation of each management practice was a 15*yr sequence generated by the
CLIGEN model (Nicks & Lane, 1989), Initialization of infiltration parameters
was taken from van der Zweep et aL (1991). Soil credibility parameters were
taken from Laflen et al. (1991).
As grazing intensity increased, water and sediment yields also were predicted to increase, while conversion from brush to grass was predicted to have the
opposite effect (Table 2-2). Increases in vegetation density and amount of residue
on the soil surface because of brush to grass conversion or because of a lower
grazing intensity increases infiltration, decreases runoff, and protects the soil surface from detachment by raindrop impact The most significant impact was on
hillslope sediment yield where conversion from brush to grass with no grazing
was predicted to decrease hillslope sediment yield 91%.
15
230 kg eafces
10
Q
i
o
0
0.2
0.4
06
-
Animate ha'
Fig. 2-1
Effect of grazing intensity on annual soil erosion for the Edwards Plateau region of Texas.
19
200
230 kg calves
150
o
c
100 h
92 kg sheep
50
0.2
0.4
Animals ha
Fig. 2-2,
0.8
0.6
T
Effect of grazing intensity on surface nrnoff fof the Edwards Plateau region of Texas,
10
8 230 kg calves
8
c
0?
92 kg sheep
0,2
Fig. 2-3
0.4
0.6
Animals ha'
Effect of grazing intensity on sediment
0.8
for ike Edwards Ptaleaa regkm of
LAFLENETAL
20
The example shown in Fig. 2-1, 2-2, and 2-3 is for the Edwards Plateau
region of Texas- As indicated earlier, information presented in Fig. 2-1, 2-2, and
2-3 are based on simulations using the WEPP model that is still under development. As development continues, predicted quantities and relationships will probably change because of model improvements, improved data, and improved parameter estimation. The information presented here is to demonstrate the power of
the WEPP model, and to demonstrate potential use, not give exact quantitative
results.
In this example, the WEPP hiilslope version was run for a 20-yr weather period. The climate was again generated using CLIGEN (Nicke & Lane, 1989).
Average annual generated rainfall was 625 mm. For this simulation, 92 kg sheep
(Ovis aries) were contrasted with 230 kg stocker calves (Bos taunts) to demonstrate
the sensitivity of the rangeland component to different stocking rates of livestock.
Grazing periods simulated were from about 15 March to 31 October of each of the
years of simulation. Some grazing rates were probably in excess of feasible rates.
As shown in Fig. 2-1, WEPP demonstrated a sensitivity to grazing intensity.
Soil erosion predicted in this case was sediment delivered from a 100 m long 9%
slope. Soil erosion rates were quite high when stocking rates were high for the
230 kg calves, but until stocking rates exceeded 030 animals per ha for the 7.5mo grazing period for this size animal, there was little impact of stocking rates
on soil erosion rates. The model demonstrated as forage consumption increases,
the risk of soil erosion increases once a critical threshold of canopy and ground
cover has been passed- The WEPP model estimates daily biomass growth and
daily biomass use and loss. This information is then used to estimate canopy and
liner cover For a given climate, the WEPP model would predict that increased
stocking rates would increase daily forage consumption, which would decrease
canopy cover and increase soil erosion and runoff. The daily forage consumption
per animal is based on the work by Brody (1945) as expressed in Eq. [7].
F = 0-l(BwftTV/>)
[7]
where F is daily forage consumed (kg) per animal, B^ is the body weight of the
animal (kg), and D is digestibility (a fraction between 0 and 1) of the forage.
Similarly, as shown in Fig. 2-2, simulated average annual runoff would be
predicted to increase as grazing intensity increased, but not as dramatically as did
soil erosion. For this simulation, intensive grazing was predicted to reduce
ground cover, both litter and canopy, which was predicted to increase surface
runoff and to increase soil erosion.
Estimated average annual sediment concentrations, based on estimated soil
erosion and runoff amounts, were low until stocking rates increased above a
threshold level (Fig. 2-3). Sediment concentration is an important parameter to
those interested in offsite effects of management, but it is not a parameter that can
be computed using Universal Soil Loss Equation prediction technology. This
illustrates one of the new uses for which the WEPP technology can be applied,
on both rangeland and cropland. Additional available information includes
enrichment ratios based on specific surface area of eroded sediment delivered
from hillslopes, and for sediment delivered from small watersheds.
THE WEPP MODEL AND ITS.APPLICABILITY
21
Research on the influence of grazing by livestock on erosion has demonstrated that light grazing can not be detected from no grazing (Thurow et al.,
1986; McGinty el al., 1979; Blackburn et al.? 1982). In many cases research has
demonstrated that moderate grazing is similar to no grazing in respect to soil erosion and runoff volume (Weltz & Wood, 1986a. b; Johnson et aL, 1980; Wood
et al., 1986). The WEPP model reflects this fact by not indicating an acceleration
in soil erosion until the stocking rate of 30 230-kg animals per square kilometer
(0.3 animals ha~') has been reached. If we use the soil tolerance concept of 21 ha"1
as excessive erosion then the maximum stocking rate would be between 40 and
2
2
50 230 kg animal knr or nearly 80 to 92 kg animals km" . This example demonstrates how the WEPP model may be able to assist ranchers and conservationists
in setting stocking rates that avoid accelerated erosion on western rangelands.
SUMMARY AND CONCLUSIONS
The WEPP model for soil erosion prediction is being developed to work for
all land situations in the USA. Its major limitations on rangelands are accurate
representation and parameterization of rangeland soils, surfaces, and ecosystems.
Major efforts are underway to overcome these limitations.
As shown here, WEPP can be used to evaluate alternative rangeland management
for specific sites. In the past, it has been difficult to easily evaluate the effectiveness of
a specific practice across a wide range of conditions- The WEPP's ability to simulate
the wide range of climates, topographies, ecosystems, and soils should make such
evaluations routine when WEPP is completely parameterized and validated
The WEPP brings to the managers* tool kit a new tool that provides new
information of importance not only for protection of the grazing resources, but
for evaluation of off site impacts of rangeland management and conservation practices- As the demands of the twenty-first century increase our reliance on a dwindling natural resource base, WEPP and other natural resource models will assume
greater roles in management of these resources.
REFERENCES
Bbckburn. W H., R W Knight and M K Wood. I98Z Impact of grazing on watersheds: A state of
knowledge- Texas Agric Exp. MP1496. College Station.
Brody, S, 1945. Bioencigciics and Growth. ReinhoW Publ Corp., New YorkCbu, ST. 1978. Infiltration during an unsteady ram. Water Resour Res. 14:461^466.
Deer-Ascougfa, LA.. G-A- Weesics, J.C Ascough II, D E_ SloU and J.M Laflen. 1993- Plant database development and linkage with a knowledge-based syslem lor use in erosion prediction models, p, 40^-417. In Proc, Application of Advanced Information Technologies for Management
of Natural Resources, Spokane. WA. 18-19 June 1993, ASAE, St. Joseph, ML
Elliot, WJ.,A-M. LeibcnowJ.M. Laffeiuand 1C D. Kohl. 1989. A compendium of soil credibility data
from WEPP cropland soil field erodibility experiments 1987 and 1988. NSERL Rep. 3. USDAARS Nad. Soil Erosion Res, Lab. Purdue Univ.. West Lafayette, IN.
Finkncr S C . M A Neanng, G.R. Foster, and J.E. Gilley, 1989 A simplified equation for modeling
ledimem transport capacity. Tram. ASAE. 32:1545-1550.
Foster, G R.. and LJ Lane. 1987. User Requirements: USDA-Uter Erosion Prediction Project (UtPP).
NSERL Rep. 1. USDA-ARS Mritt Soil Erosion Res. Lab.. Purdue Univ. West Labycae, IN,
Foster, G.R.. LJ. Lane, MA. Nearing, S.C Finkner. and D.C Hanagan, 1969. Emston ComponcnL
p. IO.I-10.il Im LJ- Lane and M.A. Nearing (ed.) USOA-Waler erouon prediction project:
Hillslopc profile model documenuiwr NSERL Rep. 1 USDA-ARS NaU. Sot! Erosion Res.
Lab,, Purdue Univ. West Lafayette.
22
LAFLENETAL
Hernandez, M-, LJ. Lane, and JJ. Stone. 1989. Surface Runoff, p. 5.1-5.18. In LJ, Lane and M.A.
Ncanng (ed) USDA-Waler erosion prediction project: Hillslopc profile model documentation.
NSERL Rep- 2. USDA-ARS Nail. Soil Erosion Res, Lab. Purdue Univ. W-sl Lafayette. IN.
Johnson, C.W., G.A. Schumaker. and IP. Smith. 1980. Effects of grazing and sagebrush control on
potential erosion. J. Range Manage. 33:451-454.
Laflen, J.M.. W.J. Elliot, J.R. Simanton, C.S, Holzhey, and K,D. Kohl. 1991. WEPP: Soil credibility
experiments for rangeland and cropland soils- J. Soil Water Cons. 46:39-44.
Lane. LJ , ED. Shirley, and V.P Singh. 1988. Modeling erosion on hiilslopes. p. 287-308. In M.G.
Anderson (ed.) Modeling geomorphic systems. John Wiley SL Sons. New York.
McCimy. W.A., F.E, Smein*. and LB. Merrill 1979, Influence of soil, vegetation, and grazing man*
agemcni on infiltration rate and sediment production of Edwards Plateau rangeland, J. Range
Manage. 32:33-37.
Meyer. L.D. 1981. How rain intensity affects intern 11 erosion. Trans. ASAE 24:1472-1475.
Meyer, L.D.. and W.C Harmon. 1984. Susceptibility of agricultural soils to interrill erosion. Soil Sci.
Soc, Am, J. 48:1152-1157.
Meyer, UD., and W.C. Harmon- I9H9. How row-sideslope length and steepness affect sideslope erosion. Trans. ASAE 31639-644.
Meyer, L.D., and W.H. Wischmeier. !969. Mathematical simulation of the process of soil erosion by
waicr Trans. ASAE I2:754~76Z
Nearing, M.A-, G.R. Foster. LJ Lane, and S.C. Finkner. 1989. A process-based soil erosion model
for USDA-Waler erosion prediction project technology. Trans. ASAE 32:1587-1593.
Nearing, M.A., and LJ. Lane. 1989. USDA-Water erosion prediction project: Hillslope profile model
documentation. NSERL Rep. 2. USDA-ARS Nad. Soil Erosion Res. Lab., Purdue Univ., West
Lafayette, IN,
Nicks, A.D., and LJ. Lane. 1989. Weather Generator, p. 2.1-2.19. In LJ. Lane and M.A. Nearing
(ed.) USDA-Water erosion prediction project: Hillslope profile model documentation. NSERL
Rep. 2- USDA-ARS Nail, Soil Erosion Res. Latx, Purdue Univ., West Ufayettc. IN.
Rawts, W.J., D.L Brafcensiek. and V Miller, 1983, Green-Amp* infiltration parameters from soils
data. J. Hydraul. Eng. 109:62-70.
RenanL K.G-, G.R- Foster, G.A. Wcesies, and J.P. Porter 1991. RUSLE: Revised universal soil loss
equation. J. Soil Wfclcr Cons. 46:30-33.
Simanton. J.R,, M A . Weliz, LT. West, and G.D. Wingate. 1987. Rangeland experiments for water
erosion prediction project. Pap, 87-2545. Am. Soc Agric. Eng., St. Joseph, Ml.
Stone, JJ., LJ. Lane, and E.D. Shirley, 1992. Infiltration and runoff simulation on a plane. Trans.
ASAE 35:161-170.
Thurow, T.L, W,H_ Blackburn, and CA. Taylor 1986. Hydrologk characteristics of vegetation types
as affected by livestock grazing systems, Edwards Plateau, Texas, J. Range Manage. 39305-509.
Tiscareno, M . V.L Lopes, and J J. Stone. 1992. Sensiitvity of the WEPP Watershed model for rangeland conditions. Pap- 92*2018. Am. Soc. Agric. Eng., St. Joseph, MI.
U.S. Depanmeni of Agriculture. 1980. CREAMS-A field scale model for chemicals, runoff, and erosion from agricultural management systems. Conservation Research Rep. 26. USDA. Sci. and
Educ. Admin. U.S. Gov. Print Office, Washington, DC
van der Zweep. R.. V.L Lopes, and JJ. Stone. 1991. Vdidatioo of the WEPP watershed model. Pap.
91-2552. Am. Soc. Ayic. Eng-t St Joseph,ML
Watson, D.A-.and J.M. Laflen, 1986. Soil strength, slope, and rainfall intensity effects on imerrill erosion. Trans. ASAE 29:98-102.
Weltz, M A . A.B. Arslan. and LJ. Lane. 1992. Hydraulic roughness coefficients for native rangelands. J. Img. Drain. DTV Am. Soc. Civ. Eng. 118(S):776-79a
Wdtz. M.A .and M K . Wood 1986a. Short duration grazing in Central New Mexico: Effects on infiltratioft rates. J. Range Manage. 39:365-368.
Weltz, MX, and M.K. Wood. 1986k Short duratkxi grazing in Central New Mexico: Effect on sediment production. J. Soil Water Cons. 41:262-266,
Wodmeier, W.H., and D.D. Smith. 1965, Predicting rainfalUrasion losses from cropland east of the
Rock} Mountains—Guide for selection of practices for soil and water conservatioa USDA
Agric. Handb. 282- US Gov, Print. Office, Washington. DC.
Wischmeier, W.H., and D.D Smith. 1978. Pred*aing rainfall enxion losses- USDA Agric. Handb.
537- Sci. and Educ. Admin.. U.S. Gov. PriaL Office, Washmgton, DC
Wood, M.K,, C.B Donan. and MA- Wieliz. 1966. Comparairve infiltration rates and sediment production oo fertilized and grazed blue grama nagrlial J, Range Manage. 39:371-374.
Yalin, VS. 1963. An exprtssiott for bed-load transportation, J. Hydraul. Div. Am. Soc Civil Eng. 89
(HY3):221-250-
Incorporating Small Scale Spatial
Variability into Predictions of
Hydrologic Response on
Sagebrush Rangelands
Frederick B. Pierson, Jr. and S. S. Van \ actor
USDA-ARS Northwest Watershed Research Center
Boise, Idaho
USDA-ARS Northern Plains Area Office
Fort Collins, Colorado
m
James C Wood
USDA-SCS Idaho State Office
Boise, Idaho
ABSTRACT
A rainfall simulation study was conducted on sagebrush rangeland to quantify the
small scale spatial variability in soil, plant, and hydrologic characteristics between four
different surface soil-vegetation-microeopographic microsites (coppice, moss-grass,
bare, and vesicular crust). The impact that this small scale spatial variability in hydrologic characteristics has on predictions of runoff and erosion from sagebrush rangeland was
also investigated. The coppice and moss-grass microsites had significantly lower runoff
and intenill erosion rates than the bare and vesicular crust microsites. TVo averaging techniques (arithmetic mean and area weighted mean) were used to estimate the runoff and
erosion response from a larger integrated area using measurements of runoff and erosion
from the four surface microsites. The area-weighted average approach provided significantly better integrated estimates of infiltration, runoff, and intertill erosion than ibe arithmetic mean approach. Both averaging approaches produced poor integrated estimates of
intenill erosion. These results have a significant impact on how hydrologk: and erosion
processes are modeled on rangelands. The commonly used assumption of viewing a hillslope as a uniform plane that can be modeled using a single set of parameters would
appear to be adequate for modeling infiltration and runoff on sagebrush rangciand, but not
for modeling intern!! erosion. Data presented in this chapter indicate thai only a small portion of the soil panicles that are detached in the intenill erosion process are actually delivered to the bottom of the hillslope. This suggests that the erosion process is transport limited and not detachment limited as often assumed.
CopyrighE O 1994 Soil Science Society of America. 677 S. Scgoe Rd. Madison. W1 53711, USA.
of Rangcbwi Wairr Ervsio* Processes, SSSA Special PuM*caHoci 38-
24
PIERSONETAL,
Sagebrush intermixed with shrubs, grasses, and forbs make up plant communities that extend across -39 million ha throughout Nevada, Utah, Idaho, Oregon,
Montana, Wyoming, and Colorado (Tisdale et at, 1969). Runoff and erosion
from these communities can significantly influence floods and sediment yields,
particularly if they have been disturbed by harsh management treatments. Several
studies have been conducted on the natural variability of hydrologic and erosion
processes within different sagebrush plant communities. In Nevada, Blackburn
(1975) studied several sagebrush communities and found that the highest infiltration and lowest sediment production occurred on coppice dune sites with wellaggregated surface soils and no vesicular crusting. The interspace soils had well
developed vesicular soil crusts which reduced infiltration and increased runoff
and erosion. Johnson and Gordon (1988) found very similar results for a
Wyoming big sagebrush (Artemisia tridentata wyomingensis Beetle & Young)
site on the Reynolds Creek Experimental Watershed in southwest Idaho.
Additional studies have shown that the type of spatial variability in hydrologic and erosion processes on sagebrush rangelands exhibited in the above studies is also temporally dependent. Blackburn et al. (1990) found that soil frost significantly influenced infiltration, runoff, and intertill erosion on both coppice and
interspace soils within a Wyoming big sagebrush plant community. They found
that the presence of soil frost significantly reduced infiltration on both soil types,
but that the impact was greatest on the interspace soils. Erosion was consistently
higher for the interspace soils compared with the coppice soils with the greatest
erosion occurring during periods of repeated diurnal freeze-thaw cycles.
Freezing and thawing of the soil helps maintain a saturated surface soil condition
in which the soil aggregates slake and breakdown, thus greatly increasing soil
credibility. Blackburn and Wood (1990) found similar results for a blackbnish
(Coleogyne ramosissimum Torr.) site near Crystal Springs, Nevada.
Most studies conducted on the variability of hydrologic and erosion processes on sagebrush rangelands have focused on quantifying the variability and not
explaining the causative factors or the impact the observed variability may have
on our ability to predict runoff and erosion over large portions of the landscape.
A study conducted by Blackburn et al. (1992) presented evidence that vegetation
is the primary factor influencing the spatial and temporal variability of surface
soil properties, which in turn, control the infiltration and credibility on semiarid
rangelands. In particular, on sagebrush rangelands it is the vegetation growth
form (e.g., bunchgrass, sodgra&s, and shrub) that is the primary factor influencing the surface soil properties related to runoff and erosion- Using this premise,
Seyfried (1991) studied the influence of macropore flow on the infiltration
process for both coppice and interspace soil types within a Wyoming big sagebrush plant community. He found that macropore flow was important in infiltration on both soils but had a greater impact on coppice soils. He hypothesized that
surface soils should be considered separately when trying to predict infiltration
for large areas. This could be accomplished by taking an area-weighted average
of predicted infiltration rates for both soil types as indicated in the following
equation:
[i]
INCORPORATING SMALL SCALE SPATIAL VARIABILITY
25
where RO is the total steady state runoff,/ and/^ are the fractions of the total area
covered by coppice and interspace soils, respectively,-4 is the rainfall application
rate, and ic and / are the infiltration rates for the coppice and interspace soils,
respectively.
"lie objectives of this chapter are to: (i) quantify the small scale spatial variability in soil, plant, and hydrologic characteristics within a sagebrush plant community; and (ii) determine if dividing an area into different microsites when estimating hydrologic characteristics will improve predictions of runoff and erosion
from sagebrush rangeland. The second objective is an expansion of the hypothesis set forth by Seyfried (1991) as depicted in Eq. [1].
MATERIALS AND METHODS
Site Characteristics
This study was conducted at the Nancy Gulch site on the Reynolds Creek
Experimental Watershed in southwest Idaho. The annual precipitation is ~32 cm,
falling primarily during the winter as both rain and snow. Summers are usually
dry with only occasional thunderstorm activity. The elevation of the site is 1400
m and the average slope is 7%. The vegetation composition is representative of
many of the regional sagebrush-dominated rangelands. The overslory consists
mostly of Wyoming big sagebrush, the understory is Sandberg bluegrass (Poa
secunda J.S. Presl) and bottlebrush squirreltail (Sitanion hystrix Natt.), and the
ground cover is liner, rocks and moss (Tortula ssp.). The soil is a Gariper silt
loam classified as a fine, montmorillonitic, mesic Xerollic Paleargid. The argillic
B horizon begins at -10 to 20 cm below the soil surface.
Surface Soil Classification
The study area was subdivided using the criteria presented by Eckert et al.
(1986) and in more detail by Eckert et al. (1989). They broke the surface soils
within the sagebrush coppice-interspace complex into four types based on
detailed study of microtopography and soil physical characteristics of many similar soils. The authors labeled the four soil types as simply Soil I through IV We
preferred to provide names for each soil type that referenced their relationship to
soil and vegetation characteristics and microtopographic position. We also chose
to refer to a soil type as a microsite. Therefore, throughout this chapter the soils
are referred to as Type I, which is a coppice microsite. Type 11 is a moss-grass
microsile, Type III is a bare microsite, and Type IV is a vesicular crust microsite.
The four microsites with their associated microtopographic positions and soil and
vegetation characteristics are shown in Fig. 3-1.
Experimental Procedures
Small Plots
Ten replicates of each of the four surface microsites were randomly selected
within a uniform 0.25-ha area so that the surface within each plot was entirely of
PIERSON ET AL.
Shrub
Coppice
11
G'«sand
Bar£
MOSS
g .,
Pnrî—i—|—|"T"Ts T
Vesicular
CrusI
Surface
Zone
Shrub
Cover
Ground
Cover
Soil
Crusting
Coppice
High
High
No
Moss/Grass
None
High
No
Bare
None
Low
Yes
Vesicular
None
None
Yes
Fig. 3-1. Micro topographic positions of surface microsiles and [heir associated cover and crusting
characteristics within a typical sagebrush coppice-interspace complex.
one microsite. The area of each irregularly shaped plot was measured by tracing
the outline of the plot on a piece of paper then using a planimeter to determine
the area within the outline. Plot areas ranged from 0.10 to 0.15 m2. For the coppice
plots, the shrubs were cut at ground level to reduce rainfall interception losses*
Rainfall simulation was conducted between 17 July and 27 July 1989, and
was applied using a small, portable, oscillating-arm rainfall simulator similar to
that described by Meyer and Harmon (1979). The oscillating-arm was 3 m above
the plot surface and a H1/2USS Veejet 80100 spray nozzle (Spraying Systems
2
Co-, Wheaton, IL)' was used at a pressure of 0.7 kg cm" measured at the oscillating arm. The oscillating arm passed over the plots -102 times every minute
resulting in intermittent simulated rainfall similar to naturally occurring rainfall
(Meyer & Harmon, 1979). Simulated rainfall was applied to each plot at 66.8 mm
h~l for 60 min starting at antecedent soil moisture conditions (dry run). The plots
were covered with a plastic sheet and allowed to drain for 24 h before simulated
rainfall was again applied at the same rale for 30 min (wet run).
Runoff from each plot was pumped continuously into a collection reservoir
using a peristaltic pump. The depth of runoff in the reservoir was continuously
measured using a bubbler gage. Rainfall rate was not measured, however, a calibration of rainfall rate based on number of times the oscillating arm passed over
the plot indicated that the calculated rainfall rate did not vary from the design
application rate by more than ±1%. For this reason the design application rate
was used in all calculations. Infiltration rate was calculated as the difference
between rainfall rate and runoff rate. Final infiltration rate was taken as the value
at the time when steady state was reached in most cases the last measured value.
A final sediment concentration was determined by taking a single, well-mixed,
homogenous 1 -L sub&ample from the runoff reservoir at the end of the rainfall
simulation. Suspended sediment samples were filtered through a 45-/mi filter.
Mention of manufacturers b for (he convenience of the reader only and implies no c
the pan of the author* or USDA-ARS
27
dried at 105°C for 24 h, weighed, and converted to sediment yield (kg ha"1) using
the measured area and runoff volume of each plot
Percentage cover of cryptograms, forbs, grasses, shrubs, litter, rocks, and bare
soil were ocularly estimated for each plot Following the wet run, all cryptograms,
forbs, grasses, and litter were cut at ground level, dried at 60°C for 48 h, and
weighed to determine aboveground biomass. Before the dry runs three soil samples
were collected adjacent to each plot for determination of particle-size distribution
by the hydrometer method (Bouyoucos, 1962), aggregate stability by the vaporwetting, wet-sieve method (Kemper & Rosenau, 1986) and organic C by the
Walkley-Black method (Walkley & Black, 1934). Soil cores were collected adjacent to each plot before the dry and wet runs for determination of bulk density and
gravimetric water content. Aggregate stability was also measured after the wet run,
Large Plots
Estimates of runoff and erosion were also collected from two 3.05 by 10-7 m
undisturbed plots using methods outlined under the Water Erosion Prediction
Project (WEPP) and described by Johnson and Blackburn (1989), Rainfall simulation was conducted in August 1989, on two natural treatment plots al the Nancy
Gulch site described by Wilcox et al. (1992). The area of each microsite type within the large plots was determined from composite low-level areal photographs of
the large plots. Boundaries of each zone were determined by visual inspection of
plot photographs, then the boundaries were digitized to determine the areal extent
of the zones. The percentage area of each zone was calculated assuming that the
total digitized area of the four soil-vegetation zones summed to 100%.
Preparation of Data for Analysis
For small plots, runoff depth within the collection reservoir was converted lo
a measure of flow rate using a calculated calibration curve during the conversion
of the strip chart data to a breakpoint-digital record- A constant rainfall rate equal
to the design rate was assumed. Runoff and rainfall were converted to an equivalent depth and rate for each breakpoint period- Infiltration depth and rate was calculated as the difference between rainfall and runoff depth and rate, respectively.
Sediment yield was calculated during each breakpoint period by multiplying the
measured runoff during that breakpoint period by the final sediment concentration and dividing by the plot area. Total runoff, infiltration, and sediment yield
were calculated by summing across breakpoint periods. Final rates were calculated at 60 and 30 min for the dry and wet runs, respectively.
Results for the large plots were prepared in a similar manner with two exceptions. The measured runoff did not have to be converted to a flow rate (the data was
recorded as a rate), and sediment concentrations were measured for each breakpoint
period. Runoff values for the large plots were scaled by the percentage of the difference between rainfall rates for the small and large plots so total rainfall amounts
were the same for both the small and large plot rainfall simulation runs.
Statistical Procedures
Analysis of variance for a completely random experimental design and
Student-Newman-Keuls* multiple range lests (Steel & Tome, 1980) were used to
28
P1ERSON ET AL
test for differences between means of measured and derived hydrologic, soil, and
vegetation parameters. The assumptions of normality and homogeneity of van*
ance required by these tests were assessed using Bartiett's test for homogeneity of
variance (Steel & Torrie, 1980) and the Shapiro-Wilk test for univariate normality
(Zar, 1974), Substantial departures from the required assumptions were rectified
using transformations. Where no other transformation yielded satisfactory results,
a rank transformation was used- Rank transformations reduce the power associated with tests of differences so that differences detected tend to be conservative.
Hydrologic results from four small plots were excluded from all analyses due to
mechanical or procedural problems encountered during rainfall simulation.
RESULTS AND DISCUSSION
Soil Characteristics
Average soil characteristics for each surface microsite at the Nancy site are
given in Table 3-1. No differences in particle-size distribution were found
between microsites with one exception. The vesicular crust microsites had significantly lower sand and higher silt contents than the other microsites. Wind and
water work together to transport soil particles from one microtopographic location to another helping to form a vesicular crust in the lower microtopographic
locations. Bulk density before the dry and wet runs, increased in the order coppice, moss-grass, bare and vesicular crust microsites. Bulk densities for the bare
and vesicular oust microsites were not significantly different before the wet run.
The largest increase in bulk density came between the moss-grass and the bare
microsites. This is probably due to the abrupt changes in surface cover and organic C content also observed between the two microsites (Table 3-2 and Table 3-1,
respectively).
Table 3-1. Average soil characteristics for each surface microsile at the Nancy Gulch site on the
Reynolds Creek Experimental Watershed, 1989.
Microsite
Characteristics
Coppice
Moss-Grass
Bare
Vesicular
39.81*
4&9b
13 Ja
37.5 a
48.4b
U.I a
39-la
0.9d
O8c
Me
Mb
IJb
L2a
1.3a
3-5a
3.7»
lib
2Jb
37â
4.7t
32-8a
17b
23.9b
l.lc
21.*
a?d
77.9i
54.4a
64.9b
48.Sa
45.7c
Partide sue (%)
Sand
Silt
Clay
Bulk density (Mg or1)
Before dry run
Beforewctrun
Gravimetric moisture control (%)
Before iky run
Before wtt run
Organic Carboo (%)
Aggregate stability (%)
Before dry nm
After wet nm
'Means followed by same letter within rows are not significantly different (P < 0.05).
29
Table 3-2. Average shrub cover, ground cover and bionrass characteristics for each surface microsjte
at the Nancy Gulch site on the Reynolds Creek Experimental Watershed, 19X9.
Micros ire
Characteristics
Coppice
Cover type, (%)
Shrub
Grass
Fbrb
Litter
Cryptogam
Rock
Bare
Vesicular
0.0
5-5a
0.2a
4.9a
84.6b
0.9b
3.8b
0.0
0.0
l.3b
0.2a
0.6b
0.6c
3JJa
94. la
0.0
0.0
0.8b
O.la
O.Sb
0.2c
4.4a
O.Oc
93.9
43.0b
1.5a
27882Ja
I08.4a
21. 2b
Ua
1447.4a
I.Tb
1239.2a
3.3a
26.9b
21. 4a
51.8c
156. 1b
l.Obc
100.0
4.2a*
0.2a
6.4a
99.9a
O.Oc
O.Oc
0.0
Bare ground
Vesicular crust
Moss-Grass
Total biomass, (kg m 0
Fbrb
Cryptogam
Litter
Root
.14. 4a
86.4c
109.4b
II. 7c
'Means foUowed by same letter within rows ait not significantly different (P < 0.05),
All microsites were extremely dry prior to rainfall simulation (Table 3-1), Before
the wet run water contents of the coppice and moss-grass microsites, however, were
much higher than for the bare and vesicular crust microsites. This could be due to differences in infiltration as affected by the large differences observed in bulk density
between these sets of microsites. Large differences in organic C content were also
observed between microsites. Percentage organic C significantly decreased in the
order coppice, moss-grass, bare, and vesicular crust microsites (Table 3-1).
Aggregate stability was measured both before the dry run and following the
wet run. Large variations were found within each microsite making it more difficult to show statistical differences between microsires; however, the same general trend was found as for other variables. Coppice and moss-grass microsites
tended to have higher aggregate stability compared with bare and vesicular crust
microsites. This is probably due to the higher organic C contents found in the
coppice and moss-grass microsites,
Vegetation and Surface Cover Characteristics
Average cover and biomass characteristics for each microsite are presented
in Table 3-2, The coppice microsites were selected on the basis of their location
directly under the sagebrush canopy, therefore they had 100% shrub cover, while
all other microsites had 0%. The coverage of grass, forb, litter, cryptogam, rock,
bare ground, and vesicular crust were all examined for each microsite. The largest
differences between microsites were found for cryptogams, bare ground, and
vesicular crust The coppice and moss-grass microsites had thick moss layers,
while the bare areas were dominated by bare ground and the vesicular crust
microsites are —94% vesicular crust
The biomass of grasses, forbs, cryptogams. litter, and roots were also examined within each surface microsite (Table 3-2), The biomass of grass was low for
30
PIERSON ET AL
all microsiies but was highest on the moss-grass microsites. Few forbs were found
and no significant difference in forb biomass was found between any microsites.
Both the coppice and moss-grass microsites were found to have very high biomass estimates of cryptogams compared with the other microsites. The coppice
microsites had nearly 28000 kg m~: of cryptogam biomass. Litter biomass followed a similar trend. Root biomass estimates were highest for the moss-grass
microsite followed by the coppice and bare and vesicular cnist microsites,
Hydrologic and Erosion Response
Average hydrologic and erosion characteristics for each microsite are given
in Table 3-3. Final infiltration rate and cumulative infiltration both decreased by
surface microsite in the order coppice, moss-grass, bare, and vesicular crust
microsites for both the dry and wet runs. There were no significant differences
between the bare and vesicular crust microsites for the wet run. The coppice
microsites had a final infiltration rate nearly six times that of the bare and vesicular crust microsites for both runs. Cumulative infiltration for the coppice
microsites was almost five times greater than the vesicular microsites and three
times greater than the bare microsites after the dry run and approximately six
times greater than both the bare and vesicular microsites following the wet run.
Results for final runoff rate and cumulative runoff followed similar but inverse
trends compared with infiltration,
Some differences were found between the hydrologic responses of the dry
and wet runs (Table 3-3). Final infiltration and runoff rates were generally lower
and higher, respectively, for the wet run compared with the dry run. Little difference was found between runs for the vesicular crust microsites. Results for cumulative infiltration and runoff were not statistically compared because twice the
amount of rainfall was applied during the dry run compared with the wet run.
Cumulative sediment was also found to increase across microsites in the
order coppice, moss-grass, bare, and vesicular crust microsites. No significant
Table 3-3. Hydrologic and erosion characteristics for each surface microsiie by dry and wet run at
the Nancy Gulch site on the Reynolds Creek Experimental Watershed. 1989.
Microsite
BMC
Drv^p run
Coppice
Moss-Grass
t. Final infiltration me (mm h )
Z Cumulative infiltration (mm)
3. Final runoff rate (mm h »
4. Cumulative runoff (mm)
5. Cumulative sediment (kg ha'1)
64. la*
65 la
42-5b
51. lb
24Jc
10.7c
23.4c
15.7c
43.4b
U.lc
228.4b
3807 Ja
53.6a
48S0.6a
60.6a
35. lb
22.6b
3ZOb
5.0c
5.4c
62.1a
7Jc
4.6c
59.8a
10.9b
28-la
1 898.0*
28.9*
1873.3.
V
V
2.7d
1.7d
56-fb
Vesicular
5.0d
13.2d
61.8a
Wet run
V . Final infiltration rate (mm h"1)
2- Cumulative infiltration (mm)
3, Final runoff rale (mm h ' >
4. Cumulative runoff (mm)
5. Cumulative sediment (kg ha i
30.4a
6-5c
3.1c
16.4c
115-flb
'Means followed by same letter * iihm rows are not significantly different (P < 0.05).
31
differences in cumulative sediment were found between bare and vesicular crust
microsites for either the dry or wet mns. The differences in cumulative sediment
between the coppice and moss-grass microsites compared with the bare and
vesicular crust microsites were much higher than can be explained by the differences in cumuUtive runoff found between the same groups of microsites. While
differences in runoff across all microsites were less than one order of magnitude,
differences in cumulative sediment were of two to three orders of magnitude
between coppice microsites and bare and vesicular crust microsites. Thus, the
credibility across microsites dramatically increased in the order coppice,
moss-grass, bare, and vesicular crust, with the bare and vesicular crust microsites
being similar and much higher than the othci microsites.
Estimating Large Plot Hydrologic and Erosion Responses
Two approaches were explored for combining measured hydrologic responses of the different surface microsites for estimating the hydrologic response of a
larger integrated area. First, a simple average of the responses for the four
microsites was multiplied by the area of the large plot to estimate the large plot
response. This assumes equilibrium conditions exist, that each microsite contributes equally to the large plot response and that the total response is a linear
addition of each surface microsite- Second, an area weighted average approach
was used to combine responses of all surface microsites- The hydrologic response
of each microsite was multiplied by the proportion of the area of the large plot
made up of each respective microsite, then added to give a total for the entire
large plot This again assumes equilibrium has been reached and that the total
response is a linear addition of each surface microsite, but does not assume that
each microsite contributes equally to the large plot response.
A comparison of approaches for predicting large plot hydrologic response is
given in Table 3-4. The area weighting approach provided better predictions of
all large plot hydrologic variables than the equal weighting approach. Predictions
for final infiltration rate and cumulative infiltration using the area weighting
method were very close to observed values, while predictions of the same van*
ables using the equal area method were substantially more in error. Predictions of
Table 3-4. Comparison of weighting schemes for aggregating estimates of inflltralion, runoff and
erosion variables from coppice* moss-grass, bare and vesicular surface microsiies 10 predict hydrologk and erosion variables on targe pkKs ai the Nancy Gulch site on the Reynolds Creek
ExpcriroenUJ Waerahed, 1989.
Weighting scheme
Equal weighting
DT
FnaJ infiltration rale (mm h")
Cumulative infiltration (mm)
Final mnoff rale (mm h ')
Cumulative runoff (nun)
Cumulative sediment (kg ha~*)
Area weighting
scheme
Wet
Or?
342
27 JO
15.7
36.2
28.6
2225.1
40.1
17.8
975.9
39.4
47.7
27.4
19.1
9038
30.6
Wet
32.9
20.4
34.1
13.2
45 1 J
Average of
large plots
Dry
40.7
50J
26.1
1&2
260.4
Wet
35.0
21.2
32.0
12J
103.4
32
PIERSON ET AL.
runoff responses on the large plots were very similar to those for infiltration. Thus,
the assumption that all surface microsites contribute equally to the larger scale
infiltration and runoff processes is not valid. The data does suggest, however, that
the assumption of linear additivity of spatially variable point estimates of hydrologic processes for predicting integrated landscape responses may be valid.
These results provide insight into how we should model the hydrologic
processes of integrated landscapes. Often when modeling infiltration and runoff
processes, a hillslope is viewed as a uniform plane that can be modeled using a
single set of model parameters to represent the entire landscape. Little field data
is available to test this assumption on rangeland. Data presented in this study
indicates that this assumption may indeed be valid if parameters are estimated
using many samples that are weighted proportionate to the amount of landscape
area they represent. This is mathematically equivalent to taking a large random
sample across the entire landscape. Parameter estimation procedures based on a
single sample or a composite of a small number of samples are inadequate particularly if the magnitude of small scale spatial variability is highPredictions of large plot erosion response was quite poor for both the equal
weighting and area weighting approaches. Both approaches greatly over estimated the amount of sediment eroded from the large plots. Using an area weighting
scheme did reduce the amount of error by more than one-half compared with the
equal weighting method, however, the quantity of error was still much more than
would be explained by the error in runoff discussed above. This suggests that only
a portion of the sediment that was detached from different points within the large
plots was actually delivered down slope to the outlet. On a small scale a great deal
of erosion and deposition takes place across a landscape that does not result in
significant large-scale movement of soil down slope or influence the sediment
loads delivered to stream channels. Soil particles are eroded then deposited only
a short distance away.
These results suggest that there are elements of the erosion process that are
not well understood on rangelands. The assumptions of hillslope uniformity and
linear additivity discussed above do not seem to be valid with respect to erosion.
Modeling a hillslope as a uniform area with only one set of parameters estimated as a simple average of point measures may not be adequate for describing all
the controlling processes across a landscape. Field observation has shown that
overland flow through some sagebrush plant communities concentrates in the
lower microtopographic positions between shrubs. This means that overland flow
follows a very tortuous path and is concentrated on the more erosive bare and
vesicular crust microsites. More experimentation is needed to determine what
additional processes should be included in our modeling efforts and how they
might be quantitatively represented.
CONCLUSIONS
Significant differences in runoff and erosion were found between four soil
surface microsites within the sagebrush shrub-interspace complex. The coppice
and moss-grass microsites had much lower runoff and interrill erosion rates than
J3
the bare and vesicular crust microsites. Both a simple average and weighted average of measurements of runoff and erosion from these zones were used lo estimate the runoff and erosion response from a larger integrated area. The weighted
average approach provided significantly better integrated estimates of runoff than
using a simple arithmetic average with estimation errors for values of runoff
under antecedent soil moisture conditions of <10%. While the weighted average
approach also produced the best estimate of intenill erosion, both approaches still
over estimated erosion by at least 250%.
These results have significant bearing on how hydrologic and erosion
processes are modeled on rangelands. When modeling infiltration and runoff
processes, the commonly iiscd assumption ^f v urging a hilklopc as a uniform
plane that can be represented by a single set of model parameters would appear
to be adequate as long as the parameters truly represent the entire landscape.
Parameters should be estimated from many samples that are weighted proportionate to the amount of landscape area they represent.
The same assumptions of hillslope uniformity and estimating parameters as
linear additives of point measurements may not be valid with respect to modeling erosion on rangeland The data presented suggest that only a small portion of
the soil particles that are detached from different points across the landscape are
actually delivered to the bottom of the hillslope- Most of the detached soil particles are redeposited only a short distance away. This suggests that the erosion
process is transport limited and not detachment limited as often assumed.
Additional important processes may be mi&sing from our current view of how the
erosion process occurs on shrub-dominated rangeland. One such process is the
concentration of overland flow in the interspaces between shrubs. More experimentation is needed to determine what processes truly control erosion on rangelands and if any additional factors should be included in our erosion modeling
efforts.
REFERENCES
Blackburn, W.R 1975. Factors influencing infiltration and sediment production of semiarid rangeVands in Nevada. Water Resour. Res. 1 ] :929-937.
Bbcfcbum, W.H.. FB. Picnon, CL Hanson. T.L. Thurow, and A.L Hanson. 199Z The spaiial and
temporal influence of vegetation on surface soil factors in semiahd rangelands- Trans. ASAE
35:479-^86.
Blackburn, W.H., FB. Piereon, and M.S Seyfrwd. IW1 Spatial and temporal influence of soil fiat
OB infiltration and erosion of sagctmfa raogdancb. Water Resour Bull 26:991-997.
BlackbunLWH. and M-K. Wood 1990. lrtBCBDBrf^fcM«ltfMiQBaf*0*aDppc*rfMi
and done interspace soils in southeastern Ncvatfa. Great Basin Nal. 50:41-46,
Bouyoucos. GJ 1962. Hydrometer method improved for making particle size analysis of soil. Agron.
J. 54:464-165.
Eckcrt R.E. J r . FF Peterson, and J-T Bdtoo. 1986. Relation between ecological-range coodiuon and
proportion of soil-surface types. J. Range Manage- 39:409—114.
Eckert,R.EJr., FF Peterson. MK, WMKLW.H. Btack&um, and J.L. Stephens. 1989. The rote of soilsurface morphology mine function of semiand rangebnck. Nevada Agnc. Exp. Stn-Tedt Bull
TB-8W)1 University of Nevada Reno.
Johnson, C W , and W.H- Blackburn. 1989, Factors contributing to sagebrush rangeland soil Ion.
Traat ASAE 32:155-160.
CW. and N.D Gordon- 19SS. Runoff and erosion from rainfall simulator plots on sagebnrt range land Trans. ASAE 31-421-427.
34
PIERSONETAL
Kcmpcr, W.D,. and R.C. Rosenau. 1986. Aggregate stability and size distribution, p. 424—442. In A.
Klute fed.) Methods of soil analysis. Part 1. 2nd cd. Agron. Monogr. 9. ASA and SSSA,
Madison, \\l
Meyer LD^ and WXT. Harmon- 1979. Multiple-intensity rainfall simulator for erosion research on
row sideslopes. Trans. ASAE 22:100-103.
Seyfricd, M.S. 1991. Infiltration patterns in simulated rainfall on a semiarid range I and. Soil Sci. Sac
Am, J. 55:1726-1734.
Steel, R.G.D., and J.H. Tonic. 1980. Principles and procedures of statistics. McGraw-Hill Book Co.,
New York.
Tisdale, E.W.. M. HJronaka, and M,A, Fosberg. 1969. The sagebrush region in Idaho—A problem in
range resource management. Idaho Agric. Exp. Stn. Tech. Bull- 512. Univ. of Idaho, Moscow.
Walkley, A., and A,I. Black. 1934. An examination of the Deqtjareff method for determining soil
organic matter and a proposed modification of the chromic acid filiation method. Soil Sci.
37:29-38.
Wilcox, B.P, M, Sbaa, W.H. Blackburn, and J.H. Milligan, 1992. Runoff prediction from sagebrush
rangelands using water erosiun prediction project (WEPP) technology. J. Range Manage.
45:470-474.
Zar, J.H. 1974. Biostatislicai analysis. Prentice-Hall, EngJewood Hiffs, NJ,
Spatial Pattern Analysis of
Sagebrush Vegetation and
Potential Influences on
Hydrology and Erosion
j
K. E. Spaeth
USDA-SCS, Northwest Watershed Research Center
Boise, Idaho
MarkA. Weltz
Tucson, Arizona
Dak Foi
USDA-SCS Southwest Watershed Research Center
Tucson, Arizona
USDA-ARS Northwest Watershed Research Center
Boise, Idaho
ABSTRACT
Climate induced temporal variation, spatial patterns of vegetation and microeovironments. plant growth forms, soils and geology, and topographic factors influence
hydrologk processes in rangeland environments. In Part I of this study, a gradient analysis of 13 environmental variables identified temporal and spatial gradients in sagebrush
coppice and interspace soil surface cover types. Spatial cover types and temporal cyclic
variations were distinct for both soil surface cover types. Part II of the study identified
different spatial patterns for several sagebrush species. Wyoming big sagebrush
(Artemisia tridentaia Beetle & Young nyomi/igensis) and mountain big sagebrush (A- tridentaia Rydh veseyana) were both associated with uniform distribution patterns. Low
sagebrush (A. arbuscula Nutt. arbuscula) exhibited a random pattern. Spatial patterns of
vegetation (random, clumped, and uniform distribution) effect the degree of tortusoity of
flow paths and hydraulic roughness CHI rangelands. Additional refinements to the Chezy
friction coefficient that incorporates estimates of roughness coefficients for rills and
interrill areas should be considered through additional resistance factors such as plant
dispersion coefficients.
Copyright O IW4 SoU Socnc* Society of America, 677 S. Segoc Rd.. Madtfon, Wl 53711. USA
35
SPAETH ETAL.
Spatial heterogeneity and pattern is a universal feature in natural plant communities. Hydrologic and erosion processes are effected by the amount, type, and
spatial distribution of rangeland vegetation- Spatial and temporal changes with
regards to hydrologic and erosion processes have been documented by
Blackburn, 1975; Spaeth, 1990; Blackburn and Wood, 1990; Blackburn et aL,
1990, 1992. If the development of new technologies and modeling of hydrologic
and erosion processes in natural plant ecosystems is to proceed, more ecological
information is needed regarding distribution patterns of major shrub types in the
western USAWhy study pattern or spatial distribution of plants in the context of erosion
prediction modeling efforts? Many hydrologic modeling efforts represent overland flow as areas of broad, uniform sheet flow. Surface flow in shrublands is tortuous, water velocity is reduced, and flow may be impeded by shrub coppices,
which act as surface dams. Flow paths are longer because of this tortuosity, which
increases the surface area of some rills.
In general, current hydrologic modeling efforts approach plant distribution
too simplistically, i.e., measurements of plant density and single species composition. Density is the number of individuals per unit area; however, density without other supporting information is static both from ecologic {Harbour et aL.
1987) and hydrologic perspectives. Plant density does not reveal the dynamic
interactions that affect spatial distribution between members of the same or different species. Different plant patterns may be present on the landscape, which
may be due to a number of factors that vary from site to site. Natural plant com*
munities are not homogeneous, even in seemingly monotonous expanses of
grasslands or prairie where shrubs and trees are virtually absent. Since individual
plant species have characteristic affects on hydrologic processes (Thomas &
Young, 1954; Mazarak & Conrad, 1959; Rauzi & Kuhlman, 1961; Dee et al.f
1966; Gifford, 1985), hydrologic models should consider the effects of shrub
communities that are dominated by one or two species compared with communities that are more diverse.
Gleason (1920) recognized that minor differences in the environment could
disrupt uniformity in vegetation. Since then, the detection and study of patterns
in plant communities has been a subject of interest among plant ecologists (for
reviews see Pielou, 1969; Grieg-Smith, 1979, 1983; Diggle, 1983; Ludwig &
Reynolds, 1988). Three main types of plant distributions are recognized in natural populations: random, uniform (regular), and aggregated (clumped or contagious) (Whittaker, 1975; Grieg-Smith, 1983; Pemberton & Frey, 1984).
The causal factors of pattern in vegetation are complex and multifactorial in
nature. Patterns in vegetation can be attributed to the morphology of the species
(Kershaw, 1959; Grieg-Smith, 1961); dispersal mechanisms from the parent plant
(Laraacraft et aL, 1983; Kershaw & Loooey, 1985; Whitford, 1986); environmental heterogeneity (Gulmon & Mooney, 1977; Beaty, 1984; Shumar &
Anderson, 1986; Ludwig et aL, 1988); sociological pattern involving competition, genetics, and other types of interaction among individuals (Fowler &
Antooovics, 1981; Aarssen & Turkington, 1985a; Fitter, 1987; Szwagrzyk,
1992); demographic characteristics and succession (Aarssen & Turkington,
SPATIAL PATTERN ANALYSIS OF SAGEBRUSH VEGETATION
37
1985b; Symonides & Wierzchowska, 1990); management history, i.e., burning or
grazing (Lament & Fox, 1981; Wright & Bailey, 1982; Matus & Tothmeresz,
1990; Ter Heerdt et a!., 1991); and stochastic pattern resulting from random variation of any of the preceding factors (Hutchinson, 1953).
If progress is to be made on overland flow models for rangeland vegetation
t; pes, rangeland communities must be assessed for spatial heterogeneity, plant
patterns (distribution), and density. The purpose of this chapter is to examine spatial and temporal relationships of soil, vegetation, hydrology, and soil erosion in
a sagebrush community and examine spatial patterns of several sagebrush
species. Out of the 15 basic rangeland types in the USA, the sagebrush grassland
is one of the largest types in the western USA (39 million hectares; Holechek et
ah, 1989). Sagebrush grasslands occur extensively in Oregon. Idaho, Nevada.
Utah, Montana, and Wyoming. In Wyoming, nearly two-thirds of the state is
occupied by one or several of 13 different sagebrush species (Beetle & Johnson,
1982).
This study is organized into two parts: The objective of Part I was to demonstrate spatial and temporal relationships of soil, vegetation, hydrology, and soil
erosion (Blackburn et al., 1990, data set) and examine multivariate relationships.
The objectives of Part II were to: (i) on a preliminary basis, conduct a spatial pattern analysis (SPA) of sagebrush vegetation types with distance based density
estimates; and (ii) relate results that are relevant to hydrology and soil erosion
modeling.
We recognize that there are many questions to be answered regarding spatial
distribution patterns in natural plant communities and causal factors that affect
distribution patterns. We propose several hypotheses that we have evaluated on a
preliminary basis and will be used, with modification as necessary, for subsequent papers involving more sites and other rangeland plant communities. We
hypothesize that within discrete ecological range sites where soil heterogeneity
on a large scale is minimized, Wyoming big sagebrush stands will exhibit a uniform pattern where sagebrush is the dominant shrub species and no interspecific
shrub species exists on the site. Where other brush species are codominants with
sagebrush, especially root sprouters, other patterns should emerge. For example,
where shrub species are root sprouters, a clumped pattern may develop.
Mountain big sagebrush is often associated with other shrub species such as
mountain snowberry (Symphortcarpos oreophilus Gray) and waxleaf ceanothus
(Ceanothus velulinus Dougl.) on the Reynold's Creek Experimental WatershedThese two species can reproduce by root sprouting; therefore, the pattern of the
codominant sagebrush population may tend to deviate from uniformity.
Parti
Site Characteristics, Spatial and Temporal Study
Part I of this study was located at the Quonsel site in the Reynold's Creek
Experimental Watershed in southwest Idaho (*80 km southwest of Boise). The
SPAETH ETAL
38
study site is characterized by an aridic moisture regime with average annual precipitation of 281 mm, 70% from rain and 30% snow. The elevation is 1193 m,
slope 6%, and aspect 344°, The soil is classified as a Larimer series (fine-loamy
over sandy or sandy-skeletal, mixed, mesic of Xerallic Haplargid. These well
drained soils occupy old alluvial terraces and colluvial foot-slopes and are
derived from weathered basalt. Two soil-vegetation surface cover types were
identified: (i) shrub coppice, and (ii) interspace between shrubs. The A horizon of
the shrub coppice is characterized by weakly granular structure, loam texture, and
the surface is dominated by moss (Polychidium spp. and Tortula spp.) and to a
lesser degree, lichens. Wyoming big sagebrush is the dominant shrub (30%
canopy cover), rubber rabbitbrush (Ckrysothamnus nauseousus [Pall] Britt) is a
subdominant shrub (<2% canopy cover). Sandberg bluegrass (Poa secunda J.S.
Presl) is the dominant graminoid species with cheatgrass (Bromus tectorum L,)
and bottiebrusb squirreltail (Sitanian hystrix Nun.) as subdominanls. The A horizon of the interspace area is characterized by a 15 mm thick vesicular crust (platy
structure), loam texture, and the surface layer is sparsely covered by gravel,
graminoid species, mosses, and lichens,
Field Methods and Analysis
Rainfall was simulated with a drop type simulator at a rate of 88.2 mm h~l
for 30 min on two soil surface cover types: sagebrush coppice dune and interspace (see Blackburn et alM 1990, for details). Rainfall was simulated at six different dates (15, 16 February; 22 February; 1, 3 March; 15, 16 March; 13, 14
April; and 20, 21 June) on soil that was continuously frozen and diurnally frozen
or unfrozen. For each date and soil surface cover type, four to six simulations
were made.
Detrended correspondence analysis (DCA; Hill, 1979), an eigenvector ordination technique based on reciprocal averaging (RA; Hill, 1973a) was used to
evaluate two soil surface cover types, six dates, 13 environmental variables, and
identify associated gradients within this data matrix (Table 4-1). Ordination is
Tabte 4-1- DefuiidoRs of symbols and units of measurement vied in (he analyses.
Symbols
Unit of Measurement
SSWC
Surface soil water content (%), 0-5-cm depth
PB
Bulk Density (Mg nr*X 0-5-on depth
AGST
Aggregate stability (%), 0-5-cm depth
OC
SAND
Organic C (%), 0-5-ctn dcpih
Sand (%). 0-5-cm depth
SILT
ROCK
UTTER
CRYPT
BIO
GRASS
INF
CSED
Silt (%), 0-5-cm depth
Suffice rock com (%)
Surface litter cover (%>
Surface Cryptogam cover (%)
Above gfomd biomass (kg m *)
Surface grass cover (%)
infiltration capacity (cm hrr)
Cumulative intern!! sediment (kg ha'1 30 mm)
39
"the arrangement of species and samples in a low-dimensional space such that
71
similar entities are close by and dissimilar entities far apart (Gauch, 1982,
p. 109). Ter Braak (1987, p. 91) defines ordination as "the collective term for
multivariate techniques that arrange sites along axes on the basis of data on
species composition." The objective of ordination is to condense complex data
sets to define emergent relationships. The Blackburn et aL, (1990) study was set
up as a complete block design.
Partn
Study Areas, Pattern Analysis Study
The study area in Part II consists of seven sites (Table 4—2)Field Methods and Analysis
The T-square distance sampling technique (Besag & Gleaves, 1973) was used
on all seven sagebrush sites. Sampling was confined to discrete range sites. T-square
sampling has been found to be a robust technique to detect pattern (especially
clumped and uniform) in vegetation (Diggle, 1983; Lamacraft et al., 1983). The Tsquare sampling procedure requires two distances: (i) x, the distance from a random
point to the nearest individual, and (ii) v, the distance from that individual to its nearest neighbor At the Buffalo, WY, and Blackfoot, ID, sites, one set of 100 random x
(cm) andy (cm) measurements were made. For the five sites at the Reynold's Creek
Experimental Watershed, one set of 50 x and y measurements in centimeters were
made along the contour of the slope and vertically down the slope. From the two
distance measurements, an index of spatial pattern (Q was calculated
ft
I
N
where N is the total number of sample points. When C is =0.5, the pattern of
individuals in a population suggests random distribution. If individual plants are
clumped, C will be significantly >0.5; significantly <0.5 implies uniform pattern. Significance of departure of C from 0.5 was tested with the standard normal deviate (z),
2-'
C-0-5
Canonical discriminant analysis was used to drive canonical variables (linear
combinations of the quantitative variables) that have the highest multiple correlation with the qualitative classes (SAS Institute, 1988). The purpose of this
analysis was to predict group membership of sagebrush species from a set of variables (C x, and y). Groups were Wyoming big sagebrush, mountain big sagebrush, and low sagebrush.
Table 4-2. Site dcsuipiuint, Wyoming anil Iduhn
Snil
Slope
Elcvition
dtg
Buffalo, WY
1372
. loamy,
Mr.in annual
precipitation
mm
90
•u,
Wyoming big sagebrush, western
[F.fymux smithn (Rydh.) Gould], prairie juncgrass
\Ktntleria cri\iutei (L.) Persb], green necdlcgrais
(Stipa viridula Inn J, Hoods phlox I / V J / . F , / Jf »i/;n
Rich.)
350
457
Mountain big aagehru&h, big bluegnu (Pkw juncifolia
Sctibn.), L ^Herman needle grass (Slipa Iciicrtnanii
Vn&ey), prairie juncgruss
250
Wyoming big sagebrush. hoiiLchiush squirrel i.nl,
cheaigrtss
nn M *!. meiii
• .,
U
SummH
Ro4)in (fin*. tiJty,
miKcd, cry ic Ptchlc
PalchuroU)
7.0
Current vegetal ion
Saralegui(coarMliiimy, mucil. inrsu
Xemllic
LUM
Reynold*!
Creek, ID
l-aururr (fine, | n , n n \ . over
»undy tu siamly \kf lcu>h p mixed,
m«tic, Utlolllc
"'-
344
261
Wyoming big sagebrush, rubber rabbiibru&h. Sandhcig
blucgrasa
y Oukh
Reynold'iCrtek. ID
Uariper (fine.
montmorlhoniik. metk
Xerollu Pikirgld)
1414
342
•,i •
Wyoming big tagebnuh, Sandbctg blucgrttt
Lower Sheep
Raytwld'i t'lwk. ID
Ciablca
mixed, frigid Uthic
Ull
Low Mgebniih. Sandberg blucgrusi
Reynold^ Creak, ID
7.0
28.0
Argue rolU)
Reynold1*
Reynold1! Creek. ID
Hullrey (finc-kwmy,
inucd. PiR-hic CVyobototl)
70
Mountain big ugcbrusht mounluin snowberry
Spatial and Temporal Study
A two-axis DCA ordination of spatial and temporal data from the Quonset
site was representative of both spatial and temporal gradients (Fig. 4-1). The first
DCA axis (eigenvalue 0.105, 91% of variability)was interpreted as a spatial gradient, which is clearly defined by the sagebrush coppice dune and the interspace
soil surface cover types. Soil surface cover type, as a categorical dummy variable
was highly correlated with Axis 1.
The second DCA axis (eigenvalue 0.010, 8.5% of the variability) was interpreted as a temporal gradient, which appears to be cyclic. February and June sample dates are synchronous for both cover types, February at the top portion of the
ordination diagram (Fig. 4-1) and June at the bottom of the ordination diagram.
Cryptogam cover, organic C, soil moisture, infiltration capacity, aboveground biomass, and aggregate stability were negatively correlated (P s 0.05)
with DCA Axis 1 (Table 4-3), the spatial gradient. The magnitude of these variables became smaller toward the interspace soil surface cover type (Fig. 4—2).
Percentage of rock cover, litter cover, percentage of silt, and cumulative sediment
;
2/22
3/1
Interspace
2/22
2/15
<40
A 3/1
Shrub
Coppice Dune
20
81
Axis 1
Fig- 4-1. Detrended correspondence analysis of 13 environmental variables from two soil cover types
six sampling dales.
SPAETH ETAL,
Table 4-3. Coefficient of correlation (r) of soil and plant variables with two DCA ordination axes
Variable
Axisl
Ax is 2
Surface soil water conieru, %
Bulk density
Aggregate stability
Organic C
Sand
Silt
Cumulative sediment
Infiltration capacity
Rock cover
Litter cover
Above ground biomass
Grass cover
Cryptogam tc cover
Soil surface cover type
-0.71*
0.49
-0.64*
-0.79'
-0.54
0.590.56*
-0.71*
0.93'
0.69'
-0.880,52
-0.94-0.93'
0.36
-0.51
-0.70*
0.06
-0.40
0.26
0.64-
-0.62'
0.59*
0.16
-0.64*
-0.11
-0.62*
-0.59*
•Significant at P s 0.05.
Sediment
2/22
,ow
3/1 A *
Interspace
2/22
A 2/15
c
u
:- -
•x
<
igh
igh
igh
igh
ligh
ligh
Shrub
Coppice Dune
A 4/13
Aggregate Statxh
Infiltration
Biomass
Crypt, Cover
Soil Water
Organic Carbon
Hign Sed»ment
High Silt
High Rock Cover
Axis 1 (sd units)
Fig. 4^2. Dctrcnded correspondence analysis with gradients. Gradients
significantly correlated with Axes I and II loading!
variables that * e re
43
were positively correlated (P s 0.05) with DCA Axis 1. These variables increased
from the sagebrush coppice dune toward the interspace soil surface cover type.
Along the temporal gradient (DCA Axis 2), aggregate stability, biomass,
infiltration capacity, and cryptogamic cover were negatively correlated (P s 0.05;
Table 4—3)- Values tended to be greater during the wanner months (gradient
direction from 20 June to 15 February)- Higher infiltration capacity was related
to later dates, which reflects differences in soil freezing, higher aggregate stability, greater biomass, higher cryptogamic cover, and higher organic C Cumulative
sediment yield and rock cover were positively correlated (P s 0-05) with DCA
Axis 2 (Table 4—3). Sediment yield was highest in the interspace during 15 and
22 February when the upper 10 mm of surface soil was diurnally frozen and the
soil at 50 and 100 mm was continually frozen.
Spatial Pattern Analysis of Sagebrush Types
The results of distance based sampling for seven sites is given in Table 4-4.
There were differences in spatial patterns between sagebrush species. Wyoming
big sagebrush was associated with a uniform pattern on all four sites. Mountain
big sagebrush was also associated with a uniform pattern at both sites. At Lower
Sheep Creek, low sagebrush was randomly distributed horizontally to the slope,
Table 4-4. SpatiaJ pattern analysis data from sagebrush sites in Wyoming and Idaho.
Location
C*
z*
.
X*
Buffalo site, WY
HV*
0.43
-2J5—
25.56
Blackfooc site, ID
HV
0.43
-159-"
34.14
Nancy Gulch, Reynold's Creek Exp. Watershed, ID
tT
0,43
-1.7742J2
V*
039
-2.65*
42.80
HV
0,41
-3.13*
42.46
Lower Sheep Creek, Reynold's Creek Exp. Watershed,
H
034
1.06
33,32
V
0.43
-1.61*
26,00
HV
0.49
-038
29.66
Summit site, Reynold's Creek Exp. Watershed. ID
H
0.42
-ZOO"
3O2S
V
0,47
-0.68
3T82
HV
0,45
-1.89**
34.05
Quonstt site, Reynold's Creek Exp. Watershed. ID
H
(U7
-0^8
44.43
V
0.41
-3.07*
42-69
HV
0,44
-2.74'
43,61
Reynold's Mountain, Reynold s Creek E*p. Watershed.
H
0,44
-134*
33.95
V
0.45
-1.17
37.74
HV
0.44
-1.92**
35.84
Y*
Species
40.79
Wyoming big
55.00
Mountain big sagebrush
7734
78.44
77J9
ID
42-86
44.14
43 JO
Wyoming big sagebrush
Low sagebrush
Low sagebrush
Lo* sagebrush
54.00
57JO
55.75
looming big sagebrush
Wyoming big
71-91
77.72
74.65
ID
53.47
54,12
53JO
Wyoming big
Wyoming big sagcbne*
Wyoming big sagebnah
Mountain big
MtaflUin big
Mniill big
V '• Significant«the 0,05 and 0.01 probability levels, respectively.
t Significant at (he 0,1 probability kvcL
t C = T-square index of spttiaJ pattern, z - standard normal deviate and test of significance of an
departure of C = OJ, X = avenge rfatanrr from random point to nearest individual, Y =
distance from individual to nearest neighbor. H = Horizontal. Bcawtncvts OB the conteM*, 50
> points; V = Vertical 50 x, y points, and HV = horizontal and vertical combined, 100 x,y
44
SPAETH ET AL
somewhat uniform vertically, but random for the horizontal -vertical composite
sample. Low sagebrush differs from both Wyoming big sagebrush and mountain
big sagebrush in that il is a dwarf shrub of irregular form, 40 to 80 cm in diam., and seldom >50 cm tall (Tisdale & Hironaka, 1981)- Low sagebrush also grows on soils that
are drier and more rocky than those supporting Wyoming big sagebrush and mountain
big sagebrush. An edaphic restriction exists on low sagebrush sites (Sabrinski & Knight,
1978) in that soil depth is either <33 cm to an impermeable B horizon, bedrock, or if
deeper, contain 30% more gravel and cobbles in the horizon (Fosberg, 1964). The shallow soils arc a result of periglaaal erosion and are low in moisture holding capacity and
become very dry in summer (Tisdale & Hirooaka, 1981).
Relating to our original hypotheses, there is some indication that mature
stands of Wyoming big sagebrush, within discrete ecological range sites, where
soil characteristics are relatively homogeneous (textures-silts to loams, and rock
outcrops, surface stones, and boulders do not influence distribution), tend toward
uniform distribution.
At the Reynold's Mountain site where mountain snowberry, a root sprouting
species, was also present, the pattern of mountain big sagebrush tended toward
uniformity. We have not evaluated the relative abundance of each shrub in precise quantitative terms to determine if mountain snowberry's presence can be
considered a codbminant thereby possibly affecting the distribution of mountain
big sagebrush- Initial transect estimates of mountain snowbeny cover are -5%.
At the low sagebrush site, distribution tended toward randomness. The
Gabica soil series (loamy-skeletal, mixed, frigic Lithic Argixeroll) contains angular cobblestones 7,5 to 25.4 cm in diam. and gravels over the soil surface. Gravels
can constitute 20 to 50% of the surface area, while cobbles, stones, and exposed
bedrock areas cover 10 to 30% of the surface. Further research will investigate if
the pattern of these cobbles and exposed bedrock areas are correlated with low
sagebrush distribution*
Discriminant Analysis of Spatial Pattern Data
A canonical discriminant analysis was performed using three variables: T-square index of spatial pattern (C), distance from random point to nearest individual
(xX and distance from individual to nearest neighbor (y) as predictors of member*
ship in three sagebrush species groups. The first two discriminant functions accounted for 77 and 23%, respectively, of the between-group variability. In Fig. 4-3, the
first canonical discriminant function discriminates mountain big sagebrush from low
sagebrush, with Wyoming big sagebrush falling between the two groups. Ttie T*
square index was somewhat associated with the first canonical discriminant function
(Table 4-5). IT* second discriminant function was highly oxidated with y. The
average distance of y for the three sagebrush species was: low sagebrush, 43-5 cm;
mountain big sagebrush, 54.1 cm; and Wyoming big sagebrush, 66,6 cm. The average distance between Wyoming big sagebrush plants was greater than low sagebrush
and mountain big sagebrush TTie C index was negatively correlated with the second
canonical discriminant function (Table 4-5). The average distance of the C index for
the three sagebrush species was: low sagebrush, 0.49; mountain big sagebrush, 0.44;
jod Wyoming big sagebrush, 0.43. In Fig. 4-3, C decreases along the second discriminant axis, meaning that Wyoming big sagebrush was more uniform according
45
SPATIAL PATTERN ANALYSIS OF SAGEBRL SH VEGETATION
I
Wyoming big sage
0.5
OJ
O
-0.5
Mountain btg sage
-1
Low sage
-1.5
-2
-1.5
-1
-0.5
1.5
0.5
CAN1
Rg. 4-3. Plots of three group cemrokis on two caaoaial dacrininam functions derived from spatial
variables (T-Square index; z, distance from random point to nearest plant; and y. distance from plant
10 nearest neighbor).
T«Ne 4-5. Result of canonical discriminant analysts of spatial variables.
Correlations of predictor variables
with canonical discriminant functions
Predictor variables
T- Square indc* (C)
Distance of iMdm
1
0.27
-0.15
2
-061
0.62
-0.09
0.90
Pooled wi thin-group
correlatkNB anoog predictors
C
1
X
0.16
1
y
-0.26
O.W"
poirt lo plant (A)
Distance of plan to
nearest neighbor (Y)
1
ta
Significant at the 0.001 probability level.
to the index. The x value, was also correlated with the second discriminant function,
but j by itself is not meaningful from a hydrologic perspective.
Pooled within-group correlations among the three predictors are shown in
Table 4-5, There is a positive relationship between distance -r and to distance y>
with r = 0.90, P ^ 0,0001, The SAS discriminant procedure (SAS Institute, 1988)
was used to classify the 17 horizontal, vertical, and horizontal-vertical data sets.
The analysis classified 100% of the data sets into their specific a priori sagebrush
groupings. The results of this discriminant analysis to hydrologic modeling suggests that between plant distances and indices of dispersion may be useful in
parameterizing spatial characteristics, especially if more supporting information
such as soil characteristics and environmental variables are correlated.
SPAETH ET AL
CONCLUSIONS
If discriminant analysis is to be used appropriately, both a clear idea of the
statistical problem and insight about the ecological data are required (Williams,
1983), This chapter is an exploratory in nature, we are not reporting the results as
statistically confirmatory. More sagebrush sites and replications within sites are
needed; however, it is interesting to note the possible significance of this data in
terms of hydrologic models. Distinct spatial cover types and temporal cyclic van*
ations exist in rangeland plant communities, all which effect or influence hydrologic processes. From a modeling perspective, the spatial distribution of cover
types, whether they are shrub coppices, caespitose grasses, sod forming grasses,
microphytic crusts, or bare ground, are important. The distribution and spacing of
these cover types, i.e., uniform, clumped, or random pattern will also affect overland flow and hydraulics. In the case of Wyoming big sagebrush, if this plant is,
for the most part ubiquitously associated with uniform pattern and predictable
plant distances, the theoretical aspects and effect of shrubs and coppice dunes on
two-dimensional overland flow models would be simplified.
In reality, however, specific plant taxa are probably not associated with any
one pattern. For example, the spatial distribution of creosote bush (Larrea divaricaia Cav.), a ubiquitous shrub of the warm desert region of North America, can
be uniform, aggregated, or random depending upon the environment (Barbour,
1969). Barbour et al., (1977) concluded that in the more arid regions of the
Mojave desert, and mesic stands in the Chihuahuan desert, creosote bush may
show a clumped or random pattern, rather than a uniform pattern. The desert
pavement is not homogeneous and pattern on the local scale can be related to very
small washes, gravel pavement, depth to caliche, rodent burrows, and microtopography. Creosote bush densities have also been attributed to differential seeding survival and creosote bush clumps appear to arise from asexual reproduction
(Barbour, 1969). Creosote bush also appears to be independent of other species
densities and distribution (Barbour et al-, 1977).
Among the work done in creosote bush communities, controversy exists as to
the type of dispersion pattern; however, a large part of the controversy is due to plot
size dependency, the specific site studied, and the mathematical techniques
employed (Barbour et al,t 1977). It is difficult to evaluate the literature regarding
spatial patterns—a consistent methodology is needed. Pielou (1977,1979) makes a
distinction between natural and arbitrary sample units. Natural sampling units, for
example, may be insects found on a given leaf or fruit; whereas, shrubs on rangeland, trees in a forest, and grasses in a prairie occur in continuous habitats. Usually
a plot or quadrat is used—an arbitrary approach, and the detection of pattern will
ultimately depend on the size of the plot or quadrat. Ladwig and Reynolds (1988),
recommend that with continuous or nondiscrete habitats, quadrat variance models
or distance models be used in lieu of frequency distribution models (Poisson and
negative binomial). The use of arbitrary sample units with continuous habitats may
result in incorrect assumptions regarding pattern because of the relationship of size
and shape of plots or quadrats to the type of pattern detected; the type of errors ecologists want to avoid (Ludwig & Reynolds, 1988). This study approached spatial
pattern analysis from a distance model or plotless perspective.
47
Still, the problem remains: How can spatial variability be represented in
hydrologic models? Studies have documented that there are significant hydrologic differences between soil surface cover types. Soil detachment and erosion
can also be significantly different between cover types. Plants exhibit spatial pattern and this is a universal feature in natural plant communities. If plant distribution and pattern are correlated to edaphic. environmental, and ecological factors (competition), this information could be used in modeling spatiality of vegetation. Specific components, concerning plant effects on hydrologic processes,
may center on the degree of tortuosity of flow paths, hydraulic roughness, two*
dimensional overland flow models, coefficients of dispersion, and the use of
qualitative variables to categorize pattern, i.e., randomness, uniformity, or
clumping.
Hydraulic roughness coefficients are used to determine surface runoff, flow
velocity, time of concentration, and routing of runoff hydrographs (Gilley et aL,
1992a). Soil microrelief, standing vegetation, litter cover, surface rocks, soil
crusts, and raindrop impacts effect resistance of surface flow and contribute to
total hydraulic resistance. In the WEPP model, the Chezy friction (C) coefficient
is used to model uniform flow characteristics. The Chezy hydraulic roughness
coefficient can be determined directly from the Darcy-Weisbach hydraulic roughness coefficient (Gilley & Finkner, 1991) using the relationship
cwhere g = acceleration due to gravity, and/= Darcy-Weisbach roughness coefficient (Chow, 1959). The Darcy-Weisbach hydraulic roughness coefficient is
given as:
I "
where S = average slope, V = flow velocity, and R = hydraulic radius (see Gilley
& Finkner, 1991), suggest that field experimentation is needed to determine the
effect of the Reynolds number on roughness coefficients. The Reynolds number,
/?e, is used to express the ratio of inertia! forces to viscous forces- They also suggest that generalized equations should be developed that relate roughness to particular characteristics of rangeland plants. Characteristics could be incorporated
for spatial distribution and pattern of rangeland vegetation. The additive property of roughness coefficients has been successfully demonstrated by Weltz et ah,
(1992) and they represent the total roughness coefficient for rills on rangeland
A -4+4+4+4
where fm = roughness coefficient for rills, f^ - roughness coefficient for grav<
and cobbles, £ = roughness coefficient for litter and organic residue, and f^
SPAETH ETAL.
48
roughness coefficient for plants, Weltz et al. (1992) developed an equation for
estimating the friction coefficient (T) for plants on rangeland areas:
OJJ
M
=39.0C C + 125.91 B*
where Cc and Bt are the fractions of canopy cover and basal plant cover, respectively. Gilley et aL, (1992a,b) also give a similar equation for rangeland interrill areas
In the WEPP model, additional refinements to the Chezy friction coefficient,
which incorporates estimates of roughness coefficients for rills and inlerrill areas,
should be considered through additional resistance factors that are related to plant
distribution patterns.
Until more field data is available on plant distributions and hydrologic effects
of specific plant species, some theoretical approach is needed to model hydrologic processes on rangelands. The current issues to be addressed are: (i) hydrologic
differences of soil surface cover types; (ii) spatial patterns of shrubs and other
plants that exhibit a tufted, caespitose, or pedestalled growth form; (iii) develop
other resistance coefficients pertinent to rangeland settings for more accurate estimates of roughness for rills and interrills; and (iv) initiate an effort to model two
dimensional overland flow where plant coppice dunes are recognized*
Many of the current modeling efforts such as WEPP base erosion prediction
OD a process-based approach that include the fundamentals of infiltration theory,
hydrology, soil physics, plant science, hydraulics, and erosion mechanics. The
WEPP project has been a major scientific effort and the knowledge gained from
this multidisciplinary approach is technically noteworthy. We subscribe to the
point of view, however, that with current technological limitations in experimentation, especially experimental designs and problems that are inherent in rangeland field studies, not to mention the economic limitations, qualitative variables
may offer a means to increase precision of model variables. Since many model
parameters are estimated by means of regression equations, qualitative variables
in combination with quantitative variables (covariance models, see Neter et al,
1989) can improve predictability. If prediction is important, the prediction equation for parameter estimates may need to include an effect due to some category
or classification of variables. Spatial patterns can be expressed in quantitative and
qualitative terms; however, the indexes of dispersion that are currently available
are unitless. Indexes of dispersion are in essence, defined by the mathematics of
the equation and ultimately, are measures of qualitative categorization such as
random, clumped, or uniform.
Terrestrial plant communities are in a constant state of flux, whether disturbed or undisturbed. Spatial patterns and hydrologic processes are cyclical in
time. Pattern and distribution of plants may be self-induced (autogenic) or environmentally determined (allogenic). A pattern that is allogenic over more than a
hundred years may be autogenic more than a thousand (Hill, 1973b). Therefore,
due to the ecological complexity of rangeland ecosystems with respect to pattern
49
and distribution, hydrologic models could be enhanced by at least considering
large scale trends, affinities, or disposition of shrub species distribution patterns.
REFERENCES
Aarssen, LW., and R. Turlcington. I985a. BitXic specializahon between neighboring genotypes in
Lolium pertnne andTnfolium repens from a permanent pasture. J. Ecol. 73:605-614.
Aaresen, LW., and R. Turkington. 1985b. Comparative relations among species from pastures of different ages. Can. J. BOL 63:2319-2325.
Barbour, M.G. 1969. Age and space distribution of the desert shrub Larrca divaricate. Ecology
50:679-685.
Barbour, M.G,, J-H. Burk, and W,D. Pitts. 1987. Terrestrial plant ecology, 2nd cd. The Benjamin
Cum m ings Publ. Co., New York.
Barbour. M.C.. J,A, MacMahon, S.A Brambcrg, and J.A. Ludwig- 1977, The structure and distribution of Lanea communities, p. 227-251. in TJ. Mabry, el al. (ed) Creosote bush: Biology and
chemistry of Larrca in new world deserts- Dowden, Hutchinson & Rose, Stroudsburg, PA.
Beaiy, S.W. 1984. Influence of microtopography and canopy species on spatial patterns of forest
undersiory plants. Ecology 65:1406-1419.
Beetle, A,A., and K.L Johnson. 1982. Sagebrush in Wyoming. Agric. Exp. Stn. Bui, 779. Univ. of
Wyoming. Laramk.
Besag, J.E , and J.T. Cleaves. 1973. On the detection of spatial pattern in plant communities. Bull
Ins*. Staiis. InsL 45:153-158.
Blackburn. W.H. 1975. Factors influencing infiltration and sediment production of semi-arid rangelands in Nevada. WUer Resour. Res. 11,-929-937,
Blackburn, WH.. F B Piereon, and M.S. Seyfricd. 1990. Spatial and temporal influence of soil frost
on infiltration and erosion of sagebrush rangelands. Wiicr Resour Bui, 26:991-997.
Blackburn, W,R, F.B Pieraon. CL Hanson, T-L Thurow. and A.L Hanson. 1992, The spatial and
temporal influence of vegetation on surface soil factors in semiarid rangcUnds, Trans. ASAE
35:479-486.
Blackburn, WH., and MX Wood 1990- Influence of soil frost on infiltration of shrub coppice dune
Md dune interspace soils in southeastern Nevada. Great Basin Nat 50:41-46Chow, V.T. 1959. Open channel hydraulics. McGraw-Hill New York.
Dec, F F. T.W. Box, and E. Robertson. 1966. Influence of grass vegetation on water intake of Pullman
sflty day loam, J. Range Manage. 19:77-79.
Diggle, PJ 1983. Suiisucal analysis of spatial point patterns. Academic Press, New York.
Fitw, A.H 1967, Spatial and temporal separation of activity in plant communities: prerequisite or
consequence of coexistence, p. 119-139. In J.H.R. Gee and PS. Ciller (ed) Organization of
communiues, fast and present. Proc. of the 27th Symp of the British Ecdogkal Soc.
Aberystwyth. 1966, BUckwcll Scientific PuM.t London.
Fo&berg, MA. 1964. Characteristics and genesis of patterned ground in Wisconsin time in a chestnut
soil of southern Idaho. Soil Sci. 99:30-37.
Fowter, N.. and L Antonovks. 1981. Competition and coexistence 10 a North Carolina grassland:
T- Pinerns in undisturbed vegetation. J. Ecol. 69:825-841
Gauch, RG. 198Z MuJtrvariaie analysis in community ecology. Cambridge Univ. Press, New York.
Giffori G.F 1965. Cover allocation in rangeland watershed management A review, p. 23-31. /n B
Jones and T Ward (ed.) \ftiershed management in the eighties. Proc. of a Symp^ Denver, CO.
30Apr.-l May. 1985. ASAE, St. Joseph, Mt
Gilley, J.E., and S.C Finkner. 1991. Hydraulic roughness coefficients as affected by random roughness. Trans. ASAE 34:897-903.
Gilley, J.E.. DC. Flanagan, E.R. Kottwitz, and MJL Weltz. 1992b Darcy-Weisbach roughness coefficients for overland flow. p. 25-52- In AJ. Paraons and AD. Abrahams (ed) Overland flow,
hydrauiics and erocion mechanics UCL Press, Univ. College London. London.
Gilley, J.E-, E.R. Kottwitz, and G A. Wicnun. T992a. Darcy-Wcisbach roughness coefncknts for
jnvel and cobWe surfaces. J Irhg. Drain. Eng. 118:104-112
Gkttoa, HJV. 1920. Some appJicafioos of the qudnt mctfcod Bull, Torrey BoL Ouh. 47:21-33.
Grcig-Smith, P. 1961. Data on pattern within plant commuties: L The analysts of pattern. J, EooL
49:703-708.
Greig-Smith, P 1979. Pattern in •••inn J. Ecol. 67:755-779.
so
SPAETH ETAL.
Greig-Smilh, P. 19X3. Quantitative plant ecology. 3rd ed. Univ. California Press, Berkley.
(.m'rnon. S-L. and RA. Mooney. 1977. Spatial and temporal relationships between two desert shrubs,
Airiple* hymenelytra and Tidestronua oblongtfolia in Death Valley, California. J. EcoL
65:831-838.
Hill, M-tX I973a. Reciprocal averaging: An eigenvector method of ordination. J. EcoL. 61:237-249.
Hill. M.O- 1973t>. The intensity of spatial pattern in plant communities. J. Ecol. 61:225-236.
HilU MX). 1979. DECORANA-A FORTRAN program for detrended correspondence analysis and
reciprocal averaging- Cornell Univ.% Ithaca, NY.
Holecheck, J.L, R.D. Pieper, and C.H HcrtaeL 1989. Range Management, principles and practices.
Prentice-Hall, Englewood GifTs. NJ.
Hutchinsoru E.G. 1953. The concept of pattern in ecology. Proc Acid NaL Sci. Philadelphia
104:1-11.
Kershaw, K.A 1959. The pattern of Dactylis gtomcraia, Lolium perenne and Tnfotium repens:
III. Discussion and conclusions. J, EcoL 47:31-53.
Kershaw, K_A., and J.H. Looney. 1985. Quantitative and dynamic plant ecology. 3rd ed Edward
Arnold, Baltimore, MD,
Lamacraft, R.R., M-H. Friedel, and V.H. Chewings- 1983. Comparison of distance based density estimates for some arid rangeland vegetation. Aust J. Ecol. 8:181-187.
Rauzi, F. and A.R. Kuhlman. 1961. Water intake as affected by soil and vegetation on certain westem South Dakota rangelands. J. Range Manage. 14:267-271.
Sabnnski. D W.. and D.H. Knight. 1978, Variation within the sagebrush vegetation of Grand Teton
National Park, Wyoming. Northwest Sci. 51:195-204.
SAS Institute. 1988, SAS/STAT user's guide. Release 6.03 ed. SAS InsL, Gary, NC
Shumar. M.L., and J.E. Anderson, 1986, Gradient analysis of vegetation dominated by two subspecies
of big sagebrush. J. Range Manage. 39:156-159.
Spaeth, ICE. 1990. Hydrologic and ecological assessments of a discrete range site on the southern
High Plains. PkD. diss. Texas Tech. Univ., Lubboct
Svmonides E-, and U. Wierzchowska. 1990. Changes in the spatial pattern of vegetation structure and
of soil properties in early old-field succession, p. 210-213. In K.F. Agncw andJ.H. WHlems(ed,)
Proc. of Spatial Processes in Plant Communities Workshop, Ublicc. 18-22 ScpL, 1989. SPB
Acad Publ bv. The Haugue
Szwagrxyk, J. 1992, Small scale spatial panems of trees in a mixed Pious sylvestris^Fagus sylvatica
forest For. Ecol. Manage. 51:301-315
Ter Braik, CJ.F 1987. Ordinatkxt p. 91-169. h R.H.G. Jongman, et it (ed.) Data analysis in community and landscape ecology. PUDOC Wageningcn, The Netherlands,
TerHecrduG.NJ., Bakfcer, J P , and J. De Leeuw 1991. Seasonal and spatial variation in Living and
dead plant material in a grazed grassland as related to plant species diversity J. Appi. Ecol.
28:120-127,
Thomas, G.W.. and J_A_ Young, 1954. Relation erf soils, rainfall, and grazing management to vegetation, western Edwards Plaieau of Texas, Texas Agric Exp. Stn. Bull, 786, College Station.
TUdak, E.W , and M. Hiroiuka. 1981. The sagebrush-grass region: A review of the ecological literature. Forest, Wildlife and Range Exp. Stn. Bull. 33. Univ of Idaho, Moscow.
Weltz, M A . A Arslan, and LJ. Lane. 1992, Hydraulic roughness coefficients for natrvt rangelands.
J. Irrig. Drain. Eng, 118:776-790.
Whitford, WG (ed_) 1986. Pattern and process in desen ecosystems. Univ. of New Mexico Press,
Albuquerque.
Whiiuker. R.H. 1975. Communities and ecosystems, 2nd. cd, MacMillan PuW. Co,, New York.
Williams, B.K. 1983. Seme observations on the use of discriminant analysis in ecology. Ecology
64:1283-1291,
Wnght H.A., and A.W. Bailey. 1982. fire ecology. John Wiley & Son*, New Yort
Temporal Variability in
Rangeland Erosion Processes
J. R. Simanton and William E. Emmerich
Tucson, Arizona
ABSTRACT
Rainfall simulation experiments conducted on large plots at various rangeland sites
in southeastern Arizona were used to determine temporal variability in rangeland soil erosion. Measured soil credibility varied monthly, seasonally, and yearly and appeared to
depend on vegetation and soil type. Short term (monthly or seasonally) variability was
greater than year to year variability unless treatment effects were interacting. The RUSLE
K factor, computed within the RUSLE model from an algorithm based on frost-free peri*
od and annual ^-values, cycles differently than the rainfall simulator measured credibility; RUSLE estimates of K were the highest when measured credibilities were the lowest
Time related changes in erosion rates associated with rangeland treatment need to be evaluated during a multiyear period using multiplot studies-
Temporal variability in the soil erosion process is both a rangeland and cropland
phenomena. The most dramatic variability is observed on cropland areas where
mechanical disturbance, growing and harvesting crops, and fallow conditions
cause almost instantaneous changes in soil credibility. Temporal variability of
natural components within a rangeland site seem static year to year, day to day,
and even hour to hour. The seasonal change is perceived to be the greatest
because of variations in vegetation canopy and ground surface cover; however,
the near soil surface and surface factors affecting soil erosion processes are
changing continuously. For example, daily soil temperature fluctuations affect
the soil moisture content and flux, which in turn influence soil infiltration capacity, biotic activity, and soil structural properties (Jaynes, 1990), Temporal van*
ability can be compared with spatial variability in that soil properties, biotic
activity, and surface cover characteristics in a heterogeneous rangeland landscape
can change in very short distances and very short time frames.
Temporal varying factors affecting the rangeland erosion process include:
(0 soil properties of infiltration capacity, porosity, bulk density, organic content, moisture, structure, and ousting; (ii) vegetation canopy cover and growth stage and surface cover, and (iii) surface microtopography. Many of the soil properties are affected temporally by frost action, wetting and drying, and rainfall compaction (Schumm
& Lusby, 1963; GifFoid, 1979; Simanton & Renari 1982; Blackburn et al.. 1990).
Copyriht O 1994 Soil Science Societ of Amend, 677 £ Scgoc R<1, Madison, WI 53711, USA.
Processes, SSSA Special Publication 38.
51
SIMANTON & EMMERICH
52
Temporal variability in rangeland soil erosion may not be as dramatic as for
cropland areas, but is critical because of the limited topsail resource associated
with rangeland ecosystems- This temporal variability is recognized in current soil
erosion prediction models. If prediction of soil erosion variability can be
achieved, then rangeland soil loss can be quantitatively assessed. This chapter
reports temporal variability in rangeland soil erosion found during rainfall simulator studies conducted during the past 10 yr in southeastern Arizona.
Rangeland experiments to quantify temporal variability in soil erosion have
relied heavily on rainfall simulation techniques (Gifford, 1979; Devaurs &
Gifford, 1984; Simanton & Renard, 1986;Lane et aL, 1987; Johnson & Gordon,
1988; Seyfried, 1991; Wilcox et al, 1992). Rainfall simulators have been used
extensively for evaluating the hydrologic and erosional responses of the natural
environment (Neff, 1979). The advantages of rainfall simulation, especially on
arid and semiarid rangelands, are that there is maximum control over where,
when, and how data are collected and there is no need to wait for natural storms,
which are usually very sporadic. Runoff and erosion responses can be compared
both temporally and spatially because similar rainfall sequences, intensities, and
amounts can be applied and antecedent conditions controlled.
Rangeland USLE Study (1981*1984)
The Southwest Watershed Research Center, of the USDA-ARS, began
rangeland erosion plot studies in 1981 to develop rangeland soil loss factors for
the Universal Soil Loss Equation (USLE) (Wischmeier & Smith, 1978). The
plots were located in southeastern Arizona on the Walnut Gulch Experimental
Watershed -90 km southeast of Tucson (Fig. 5-1). Average annual precipitation
on the watershed is 320 mm and is bimodally distributed with 60 to 70% occurring during the summer thunderstorm season of July to mid-September (Osborn
et alM 1979). In the winter, snowfall occurs periodically, but rarely remains on the
*
'
WVJlL
- OULJH
Fig, 5-1- Location of the Walnut Gulch, Santa Rita, and Empire Ranch study sites in
TEMPORAL VARIABILITY IN RANGELAND EROSION PROCESSES
53
ground for >2 d; soils may freeze to shallow depths (<20 mm) overnight, but
thaw during the day. Soils are generally well drained, calcareous, gravelly loams
with large percentages (>50%) of rock and gravel on the soil surface and up to
60%, by volume, gravel and cobbles in the surface 100 mm (Gelderman, 1970).
Three sites were selected that had different soil and vegetation complexes.
The soil series and descriptive classification were: Bernardino (fine, mixed, thermic Ustollic Haplargid), Cave (loamy, mixed, thermic, shallow Typic Paleorthid),
and Hathaway (loamy-skeletal, mixed, thermic Aridic Calciustoll). Table 5-1
lists general soil properties of the surface 5 cm of these three soil series.
Vegetation on the Bernardino soil series site was grass dominated; blue grama
[Bouietoua gracilis (H.B,K.) Lag], bkieoals grama [B. curtipcuJula (Michx)
Tort-], Shrubs dominated the Cave soil series site and consisted of creosotebush
[Larrea tridentata (DC,) Coville], white-thorn (Acacia constricta Benth.), tarbush (Flourensia cernua DC.), and burroweed [Aplopappus tenuisectus (Green)
Blake], The Hathaway soil series site contained grasses; blue grama, sideoats
grama, and bush muhly (Muhlenbergia Porteri Scribn.), and shnibs; snakeweed
[Gutierrezia Sarothrae (Pursh) Britt. & Rusby], creosotebush, white-thorn and
burroweed. Canopy cover averaged 65, 30, and 50% for the Bernardino, Cave,
and Hathaway soil series sites respectively.
Procedures used in these studies are described by Simanton and Renard
(1986) and included the use of a rotating boom rainfall simulator (Swanson,
1965) that applied rainfall intensities of 65 or 130 mm h~'- Rainfall simulations
were made in the spring and fall on plots 10.7 m long by 3.05 m wide, with three
treatments under three soil moisture conditions [dry (existing soil moisture condition), wet (field capacity), and very wet (saturated)]. The treatments were: natural cover (control); clipped (vegetation clipped to a 20 mm height and clippings
removed); bare (vegetation clipped at the soil surface and all clipping, surface litter and rock fragments >5 ram removed). Each site was fenced to exclude grazing and plot treatments were made prior to each season's rainfall simulations.
Detailed plot characterizations of vegetation composition, foliar canopy cover
and ground surface cover were measured before and after treatment Surface
cover characteristics included: soil, gravel (5-20 mm), rock (>20 mm), litter, and
basal plant cover. Ten 3.05-m long line transects, perpendicular to the plot and
equally spaced along the plot, were measured lo produce 490 point readings to
describe each plot's surface and vegetation canopy cover.
TaWc 5-1. General soil properties of the Bernardino, Cave, Haihaway. and White House soil series
evaluated *ith the rainfall simulator.
General Soil Properties
.
Series
Sand
Silt
Clay
Organic matter
84
66
10
26
08
1.8
1-5
1.0
1.7
Cave
ftabaway
While House
74
17
68
22
6
8
9
10
Hathaway (Empire)
66
22
12
Bernardino
54
SIMANTON & EMMERICH
Burn Study (1987-1991)
Rainfall simulation studies to determine vegetation burning effects on runoff,
erosion, and nutrient cycling were conducted on the Santa Rita Experimental
Range and Empire Ranch-Cienega Resource Conservation Area in southeastern
Arizona (Fig. 5-1). Average annual precipitation at these areas is 420 mm. Both
areas have similar yearly precipitation and temperature characteristics as those
found at the Walnut Gulch Experimental Watershed. Soils are generally well
drained, calcareous, gravelly loams with up to 25% rock and gravel on the soil
surface and -20% in the surface 100 mm. The soil series at the Santa Rita site
was a White House (fine, mixed, thermic Ustollic Haplargid). The soil series at
the Empire Ranch site was a Hathaway. Table 5-1 lists general soil properties of
the surface 10 cm of these two soil series. Vegetation at the Santa Rita site was
dominated by an introduced grass, Lehmann lovegrass (Eragrostis Lehmanniana
Nees.)- Mean live and dead standing biomass was 4170 kg ha'1 and ground litter
was 1650 kg ha"1. Native grasses dominated the Empire Ranch site and included
black grama (B. eriopoda Torr), hairy grama (B. hirsuta Lag.), and sideoats
grama. Mean live and dead standing biomass was 2310 kg ha~' and ground litter
1
was 420 kg ha" .
The rainfall simulation procedures used are described in detail by Emmerich
and Cox (1992) and were similar to Simanton and Renard (1986), except rainfall
simulations were made only at initial soil moisture conditions. The simulations
were conducted in the fall and spring seasons for 2 yr. Treatments at both sites
included a natural (control) and a first-year burned treatment (all litter and vegetation burned just prior to the rainfall simulation). Amounts of live and dead
standing biomass and ground litter were determined prior to burning and rainfall
simulation; plot surface characterizations were not made.
Erodibilit) Study (1991-1992)
A temporal rangeland soil credibility study began in 1991 at the Walnut
Gulch Experimental Watershed using rainfall simulation techniques described by
Simanton et al. (1991). This study was conducted on the Bernardino soil adjacent
to the 1981-1984 USLE Bernardino soil site. Monthly evaluations of erosion
rates were made on two treatments at the three soil moisture conditions described
for the 1981-1984 study- The treatments included a natural (control) and a
clipped (vegetation clipped to a 20 mm height and clippings removed) treatmenL
Results from all these studies in southeastern Arizona showed that there was
monthly, seasonally, and yearly temporal variability in rangeland soil erosionMonthly Erodibility
The 1991-1992 credibility study showed that the clipped plot erosion rate,
per millimeter of runoff, cycled through the year by a factor of three between the
TEMPORAL VARIABILITY IN RANG ELAND EROSION PROCESSES
—
96 RAT
cc
>;
* i
MONTH OF MEASUREMENT
Erosion rates per millimeter of runoff for the clipped plots at dry and wet soil moisture con
on the 1991 to 1992 and the April and November 1981 Bernardino soil series stf
highest and lowest erosion rates (Fig. 5-2). This same cycle was found for all
three soil moisture conditions with erosion rates increasing with increasing soil
moisture. The dry and wet soil moisture erosion rates of the 1991-1992 April and
November simulations were very near those of the April and November 1981 dry
and wet soil moisture erosion rates. The agreement between the 1991-1992 and
1981 erosion rates implies that the monthly cycle is repeatable and that the within year erosion rates are greater than between years. The monthly erosion rate on
the control plots was slightly less than the clipped plot erosion rate, but followed
a similar cycle. The large decrease in erosion rate from November to January was
attributed to the first freeze-lhaw sequence that usually occurs in December The
freeze-thaw would loosen the soil surface that had been compacted and sealed
during the high intensity summer thunderstorms and the wetting and drying
cycles that start in July. July also marked the change from a relative constant erosion rate to an increasing rate during the summer rainstorm period
Seasonal Credibility
Four years of spring and fall rainfall simulations were made on the
198U1984 USLE Walnut Gulch study site plots and 2 yr on the bum study plots
at the Santa Rita and Empire Ranch sites. Seasonal runoff and erosion differences
were found at all sites. The magnitude of these differences appears to be both
treatment and soil type dependent- Greater runoff and erosion occurred from the
fall simulations on the nonvegetaied (clipped, bare, and burned) plots compared
SIMANTON & EMMERICH
40 '
E
i
D
30
20-
* BERNARDINO
or
• HATHAWAY
< 10
»CAVE
o SANTA RITA
* EMPIRE RANCH
10
20
X
40
SPRING RUNOFF, mm
4000
E
o
^
3000 -
O
O
* BERNARDINO
2000
LU
* HATHAWAY
E
O
ft CAVE
(A
o SANTA RITA
* EMPIRE RANCH
111 1000
1000
2000
3000
4000
SPRING SEDIMENT CONC., mg-1• L
Fig. 5-3. Relation between spring and fall runoff vohmes tod average sediment concentrations on
(bare, clipped, and burned) plots for all study sites.
with the vegetated control plots (Fig. 5-3). Except for the 1981-1984 Bernardino
site, the vegetated (control) plots had more runoff in the fall than the spring (Fig.
5-4). Vegetated plots at all sites had lower sediment concentrations in the fall than
the spring (Fig. 5-4). Similar seasonal runoff differences have been reported for other
rangeland sites (Schumm & Lusby, 1963; Achouri & Gifford, 1984; Simanton et ah,
1986; Blackburn et al., 1990). Inadequate data prevent specific explanations for these
seasonal differences at the Arizona sites; however, the differences could be related to
soil surface aggregate destruction and compaction caused by raindrop impact, wetting and drying cycles with crust formation, and vegetation growth cydes,
Sediment yields from the vegetated plots at the Walnut Gulch sites were lower
in the fall than spring whereas the nonvegetaled plots had higher yields in the fall
;
40
* BERNARDINO
• HATHAWAY
30 -
E
*CAVE
o SANTA RITA
x EMPIRE RANCH
20 •
10 -
u.
10
20
30
40
SPRING RUNOFF, mm
4000
ERNARC'INO
O)
E
-A HAVM
O 3000
« CAVE
*
o
SANTA RJTA
o
EMPIRE RANCH
2000
UJ 1000
2
,
0
1000
2000
3000
4000
SPRING SEDIMENT CONG., rag'1 L
Relation between spring and fall runoff volumes and average sediment concentrations on
(control) plots for all study sites.
than spring (Fig- 5-5). At the Santa Rita site there was no difference between the
spring and fall sediment yields, but the Empire Ranch site had significantly more
sediment yield in the fall than spring (Table 5-2). The vegetated and bum plot data
from the Santa Rita and Empire Ranch sites were pooled for seasonal evaluation as
there was no difference between treatments evaluated immediately after the burn.
Yearly Erodibility
Nonlinear least squares fits of measured erosion rate per unit erosion index
vs. time data for the control and clipped plots from the Walnut Gulch 1981-1984
study are shown in Fig- 5-6. El, the product of rainfall energy and maximum 30
min. intensity and a measure of rainfall erosrvity (Wischmeier & Smith, 1978),
-
58
SIMANTON & EMMERICH
2000
(a)
01 1500
A
1000 -
• BERNARDINO
D HATHAWAY
LU
» CAVE
500 -
500
1000
1500
2000
SPRING SEDIMENT, kg ha'1
2000
1500
U
1000
UJ
500
0
500
1000
1500
2000
SPRING SEDIMENT, kg ha '
Fig. 5-5. Relation between spring and fall sediment yields for the vegetated (a) and
(b) plots for the 1961 to 1984 Walnut Gulch study sites.
Table 5-2, Mean sediment yield per rainfall simulator event from pooled control and burn treatment
data for the Santa Rita and Empire Ranch study sites.
Season
Site
Fail
kg ha
Santa Rita
Empire
26a'(30)
222a(l69)
32a(46)
(97)
'Within rows, data followed b> different letter* are significantly different (P < G.G5). Values in
arc standard deviat
2
'° I
59
BERNARDINO
HATHAWAY
CAVE
1.5
UJ
..---
1.0
I
/
CUPPED
..-
2
z
o
w
o
5
0.5
CONTROL
81
62
83
84
85
YEAR
Fig, 5-6. Nonlinear least squares fit of measured erosion rale vs, time for the control and clipped plots
from the 1961 to 1964 Walnut Gulch study (Simanton A Rcnard 1992).
used to normalize the rainfall energy inputs to the different treatments. Each
soil had a different shaped erosion rale vs. tune curve. Erosion rates of the control plots decreased for the Bernardino and Hathaway soils and increased for the
Cave soil (Fig. 5-6)- The different shapes of the erosion rale curves probably
reflected vegetation differences- The Bernardino control plots were dominated by
perennial grasses, the Cave control plots were shrub and forb dominated, and the
Hathaway control plots had both grass and shrub canopy cover
The decrease in erosion rate on the Bernardino and Hathaway soils was
probably due to increases in soil porosity and organic matter associated with
increases in plant basal and canopy cover that occurred because of release from
grazing and addition of spring and fall moisture from the rainfall simulations.
The initial increase in erosion rate on the Cave soil could be the result of additional rainfall energy added during the rainfall simulations on the shrub dominated site. This would force the relatively unstable mounds found under the
shrubs to reach a new equilibrium with the additional energy inputs of the rainfall simulation.
60
SIMANTON & EMMERICH
The clipped plots response to the loss of vegetation canopy produced three
different trends in erosion rate (Fig- 5-6). The Cave soil (shrub vegetation) erosion rate increased with time after the initial clipping and then leveled out in ~2
yr This type of response indicated a new equilibrium was being established with
the rainfall energy inputs the same way the control plots responded, only to a
greater extent due to the vegetation canopy removal. The Hathaway soil (grass
and shrub vegetation) erosion rate continued to increase throughout the four year
study whereas the Bernardino (grass vegetation) erosion rate curve decreased
with time over the four year period. The decrease in erosion rate of both the control and clipped plots at the Bernardino site could be reflected in a soil response
to grazing release. Prior to exclusion from grazing at the beginning of the study,
the Bernardino site was the most heavily grazed of the three Walnut Gulch sites.
The bare soil plots produced the greatest erosion rate changes with time of
all the treatments (Fig, 5-7). The rates for the Bernardino and Cave soils
increased with time for -2 yr before reaching a new equilibrium with the energy
input- As with the clipped plots, the Hathaway bare soil treatment erosion rale
was still increasing after 4 yr The erosion rate increase for the bare soil treatment
BARE PLOTS
LU
BERNARDINO
HATHAWAY
CAVE
o
55
O
o:
LU
CUPPED PLOTS
YEAR
Fig, 5-7, Nonlinear least squires fit of measured erotkxt rate vs. time for the bare and clipped plots
from ihe 1981 to 1984 Walnut Guich study (Simancon & Rcnard, 1992).
61
Table 5-3. Mean surface runoff and sediment yield for the control and burned plots after treatment
and 1 yr later on pooled data for the Santa Rita and Empire Ranch study sites.
Treatment
Surface runoff
Sediment yield
mm
— k g hi"' —
After burn
Control
Burn
5.8 (7 S?
7.0 (10)
76 (98)
106(145)
1 yr later
Control
Bum
4.4 (6.7)
19031
67 (97)
454(388)
'Values in parenthesis are standard deviations
closely emulated runoff changes that may be attributed to the decrease in root
and residue material in the soil, which
in turn decreased the soil macropore
•
structure (Dixon & Simanton, 1979), Other reasons for the erosion rate
increase could be the formation of better defined concentrated flow paths that
would be more efficient in sediment transport and a decrease in surface roughness associated with vegetation and rock fragment removal. Most likely, the
increase in runoff and erosion rates is a combination of these factors rather
than the result of any single factor. The bare soil plots on the Bernardino soil
had an erosion rate -90 times greater than the control at the end of 4 yr For
the other two soils, the bare soil plots had erosion rates that were *-30 times
greater than the control
If the vegetative canopy cover was a dominate factor controlling erosion
rates, a dramatic increase should have been found immediately in the clipped plot
results. The clipped plot's erosion rate did change with time, but not as drastically as the bare plot (Fig. 5-6 and 5-7). This suggests a small influence on erosion
rale of vegetative canopy cover removal and a more dominate effect of soil surface cover controlling (i.e., rock fragment cover) erosion rates. Similar results
have been reported by Simanton et al. (1991) for nine other rangeland sites
throughout the western USA- The clipped plot response also suggests that the loss
of soil surface cover from the bare plots and increase in concentrated flow paths
were probably dominating the bare plot responseThere were no significant differences in runoff or sediment yield between the
control and burned treatments after the bum treatments were imposed at the Santa
Rita and Empire Ranch sites (Emmerich & Cox, 1992) (Table 5-3). One year
after the burn there was over four times the runoff from the burned plots compared with the control and more than six times the sediment yield These results
agree with the Walnut Gulch data in that increases in erosion will occur with time
ooce the soil surface protection is removed. The Santa Rita and Empire Ranch
sites did not have rock cover to protect the soil surface and within a year increases in erosion rates were observed. The Walnut Gulch clipped plots had soil surface rock fragment cover to protect the soil surface and prevent an erosion rate
increase with time.
62
SIMANTON & EMMERICH
40
HATHAWAY SOIL
30 4V:
1981-1984 STUDY
Of
O 20
0
cr
10 -f
:
^-^_
CONTROL
....
CLIPPED
Fig, 5-8. Change in percentage litter cover of the control and clipped plots from the Hathaway site of
rhe 1981 to 1984 Walnut Gulch study
Plot Characteristics
Of the yearly changes in measured plot surface characteristics, the change in
litter cover was most surprising- As would be expected, the clipped plot litter
cover decreased with time as the canopy cover was continually being clipped.
However, the litter cover of the control plots from all three 1981 to 1984 Walnut
Gulch sites also decreased with time, though at a slower rate (Fig. 5-8). Only
results from the Hathaway site are shown, but the results from the other sites followed a similar trend. Associated with this litter cover decrease was a corresponding increase in plot surface bare soil* Litter being removed by the runoff
was riot evaluated, but this mechanism for litter disappearance is possible- The
reason for the corresponding increase in bare soil, however, is not clear. These
two changes in plot surface characteristics could both be explained by increases
in termite activity. In the Chihuahuan desert, under conditions of high relative
humidity, termites have been shown to remove a large fraction of surface plant
material, while moving large amounts of mineral soil to the surface (Whitford et
ah, 1982; El kins et af., 1986). The spring and fall rainfall simulations were made
during naturally dry periods, thus extending the period of termite activity. The
disappearance of litter, forage for termites; and an increase in surface soil, a byproduct of termite foraging; may be evidence enough to explain the decrease in
litter cover and corresponding increase in surface soil.
6J
~
T
RL
-TO
X
0.40
-J
cn
20
EROSION RAT
I
JAN
I
MAR
I
I
MAY
I I
JUL
MONTH
Fig. 5-9. Monthly measured erosion rale per millimeter of precipitation Tor the wet soil moisture
condition on the clipped plots of the 1991 to 1992 Bernardino study and the RUSLE estimated
monthly K.
RUSLE Soil Erodibility Factor K
The Revised Universal Soil Loss Equation (RUSLE) (Renard et al, 1991) is
the soil loss prediction equation developed (o replace the L'SLE. The equation is:
X = f i x / : x L S x C x P where,
A
R
K
LS
C
P
1
= average annual soil loss (t ha' ),
= average annual erosivity (MJ x mm/ha x h x y).
= soil erodibility (t x ha x tiha x MJ x mm).
= topographic effect,
= cover-management, and
= conservation practice.
The soil erodibility factor (K) of the RUSLE is varied throughout the year and the
variance is described by an algorithm dependent on length of frost-free period and
average annual R. Monthly measured erosion rates, kilogram per hectare per millimeter precipitation, for the Bernardino soil at field capacity (wet soil moisture),
from the 1991 to 1992 Bernardino credibility study are plotted with the RUSLE
estimated K factor for that soil (Fig. 5-9), The measured erosion rates are lowest
between May and July when RUSLE estimates the K factor to be at its highest.
Also, RUSLE estimated K to be lowest in November when the highest measured
erosion rate occurred This period of highest soil erodibility, however, coincides
with the period of lowest rainfall erosivity. This discrepancy in the cycle of soil
64
SIMANTON & EMMERICH
credibility extremes may be due to the lack of freeze-thaw intensity in the
Bernardino soil as compared with the soils from which the RUSLE K algorithm
was developed- The RUSLE K algorithm was developed on cropland soils from
the east and midwestern USA. The Bernardino soil was not in the standard plot
(continuous fallow, up-down slope cultivation) condition (Wischmeier & Smith,
1978) as required for a K factor determination, therefore, the plot erosion rate and
the K factor cannot be compared directly, but yearly trends in the cycle should be
in agreement.
CONCLUSIONS
Soil credibility, measured by rainfall simulation experiments conducted at
various rangeland sites in southeastern Arizona, varied monthly, seasonally, and
yearly and appears to depend on vegetation and soil type. Short-term (monthly or
seasonally) variability is greater than year-to-year variability unless treatment
effects are interacting. The RUSLE K factor cycles differently than measured
credibility and estimates the highest credibilities when, in fact, they have been
measured to be at their lowest. Time related changes in erosion rates associated
with rangeland treatment need to be evaluated during a multiyear period using
multiple! studies. Biotic, both flora and fauna, influences can play a major role in
the temporal variability of the rangeland soil erosion process.
REFERENCES
M.. and G.F Giffard. 1984. Spatial and seasonal variability of field measured infihrauon
on a ""p4**"1 site in Utah, J. Range Manage. 37:451-455.
Blackburn, \v H,. RB. Pierson, and M.S. Seyfried 1990, Spatial and temporal influence of soil frost
on infiltration and erosion of sagebrush rangelands- Water Res Bull. 26:991-997.
Devaura, M-, and G.F Gifford. 1984, \feriabtlity of infiltration within large runoff plots on rangelands,
J. Range Manage. 37:523-528.
Dixon, R.M., and J.R. Siman loo. 1979. Water infiltration processes and air-earth interface concept,
PL 314-330 In HJ. Morei-Stytoux et aJ. (ed_) Surface and subsurface hydrology. Water Res.
Publ,, Ljnletoo, CO.
Elkins, NX, G.V Sabol, TJ. Ward, and WG. Whitford 1986. The influence of subterranean termites
on the hydroiogjcaj characteristics of a Chihuahuan desert ecosystem. Oecotogia 68:521-528.
Emmerich, W.E., and J.R. Cox. 1992- HydroJogic characteristics immediately after seasonal burning
on introduced and native grasslands. J. Range Manage. 45:476-479.
Geiderman. F W, 1970. Soil survey. Walnut Gulch experimental watershed, Arizona Special Report,
USDA-SCS.
Gifford, G.F 1979, Infiltration dynamics under various rvngeland treatments on uniform sandy-loam
sous in southeastern Utah. j. Hydrol. (Amsterdam) 42:179-185.
D.B. 1990. Temperature variations effect on field measured infiltration- Soil Sci. S«- Am, J.
Johnson, C.\V. and N.D. Gordon. 1988. Runoff and erosion from rainfall simulator plots on sagebra* rangeland. Trans. ASAE 31:421-427.
Lane. LJ-. J.R- Simairton, TE. Hafconon, and E_M Romney, 1967. Lirgc-plot infiltration studies in
desert and scmuhd nngriand areas of the Southwestern U-SJL p. 365-367, lit Ent Cocif. oo
infiltration development and application, Matioa, HI. Jan. 1987, Univ. of Hawaii. Mam.
Neff, E-L 1979, Why rainfall simulation? p. 3-7. /« E.L Neff (cd) Proc. Rainfail simulator worksbop, Tucson, AZ- 7-9 Mac 1979. USDA-Sci. and Educ. Admin.. Agric Rev. and ManualsWÎO. Agnc. Res-So and Educ. AdmtiL, Oakland, CA.
Osborn, H B,, K-G- Renanf. and J.R. Simantoa 1979. Dense networks to measure ccnvecthr rainfall
in the sooth*esicra United Stales, Water Resour. Res- 15:1701-1711.
65
Renard, (CO., G.R. Foster, G,A Wcesics. and J.P Porter 1W1 RUSLE Revised Universal SoiJ Loss
Equation. J. Soil Water Conserv. 46:30-33,
Schumm, S.A.. and C.C. Lusby 1963. Seasonal variation or infiltration capacity and runoff on hillslopes in western Colorado. J. Geophys. Res* 68:3655-3666.
Seyfried, M.S. 1991. Infiltration patterns from simulared rainfall on a semiand rangetaod soil. Soil
Sci. Soc. Am- J. 55:1726-1734.
Simanton. J.R,. C.W. Johnson, J \V. Nyhan, and I M. Rommry. 1986. Rainfall simulation on rangeland erosion plots, p. 11-17. In LJ- Lane (ed ) Erosion on rangelands: Emerging technology and
database. Proc of the rainfall simulator workshop, Tucson, AZ, 14-15 Jan. 1985, Soc. for Range
Manage., Denver, CO.
Simamon, J.R.. and K.C. Renard. 1982. Seasonal change in infiltration and erosion from USLE plots
in southeastern Arizona, p. 37-46, In Hydrology and water resources in Arizona and the
Southwest Vol. 12. Office of Arid Land Studies, Univ. of Arizona, Tucson.
Simanton. J.R., and K G Renard 1986. Time related changes in rangeland erosion, p. 18-22. In LJ.
Lane (cd.) Erosion on rangelands Emerging technology and data base. Proc. of the rainfall simulator workshop, Tucson. AZ. 14-15 Jan. I9K5. Soc for Range Manage. Denver. CO.
Simamoru JR.. and K.C. Renard 19Q2. Upland erosion research on raigclaad p. 335-37S. In AJ.
Parsons and A.D. Abrahams (ed.) Hydraulics and erosion mrchaiici of Overland Flow, Univ.
College London. GuiJdford, England
Simanton, J.R., Vf.A. Weltz_and H D. Larsen. 1W1. Rangeland experiments to parameterize Ihe water
erosion prediction project model: Vegetation canopy cover effects. J. Range Manage.
44:276-282SVMOO, N.P. 1965. Rotaiing-boom rainfall simulator. Trans. ASAE 8:71-72.
W'bitfori W.G , Y. Slernberger, and G Eltershank 1982. Contributions of subterranean termites to the
êconomy" of Qiihuahuan Desert Ecosystems. Oecologia 53:298-302.
Wilcox, B.P., M. Sbaa, W.H. Blackburn, and J.H Milligan, 1992. Runoff prediction from sagebrush
using erosion prediction project | WEPP) technology, J- Range Manage, 45:470-474.
r, W.H., and D.D. Smith. 1978, Predicting rainfall erosion losses—a guide 10 cooservaiioB
planning, USDA Agnc. Haadb. 537, U.S. Gov Print Office. Washington, DC
Influence of Frozen Soil
on Rangeland Erosion
M. S. Seyfried and G. N. Flerchinger
USDA-ARS, Northwest Watershed Research Center
Boise, Idaho
ABSTRACT
Surface runoff and erosion from frozen soils are widespread phenomena thai have
been reported in most regions of the world that have significant soil freezing. In some
regions, such as the interior Pacific Northwest, frozen soil is associated with the majority
of flooding and erosion events. The processes involved in frozen soil erosion are somewhat different from those in normal, unfrozen soil erosion. When the soil solution freezes,
some portion of the total water content remains as liquid water. This is critical because the
amount of ice formed largely determines the impact of soil freezing on the soil properties
that affect erosion. Soil with high ice content may be essentially impermeable so that
when the soil surface thaws it becomes highly credible. These conditions can generate
runoff and erosion with little or no rainfall (e.g., if there is snowmeU), Accurate estimation of runoff and erosion from frozen soils requires knowledge of soil freezing occur*
rence and depth, the effect of freezing on infiltrability and surface runoff, and the effect of
freezing on soil credibility. Current models can accurately predict frost occurrence and
depth but not infiltrability or credibility. Field observations of frozen soil runoff have
shown extreme spatial and temporal variability over a range of scales- Accurate description of the effects of soil freezing on surface runoff and erosion on rangelands will ultimately require that models incorporate landscape scale processes.
The action of soil water freezing can greatly increase the potential for soil erosion. Historically, the primary emphasis of erosion research has been on summertime, unfrozen processes. The importance of frozen soil erosion and its coosequences have now been accepted. Wischmeier and Smith (1978), for example,
estimated that 90% of the erosion on the steeply rolling wheatlands of the Pacific
Northwest was related to frozen soil. They further noted that the Universal Soil
Loss Equation (L'SLE) did not account for this kind of erosion.
Frozen soil erosion can potentially occur wherever there is significant seasonal frost. The U.S. Soil Conservation Service depiction of land resource units
affected by significant freeze-thaw action indicate that a substantial portion of
the contiguous USA, much of it rangeland, is susceptible to frozen soil erosion
(Fig. 6-1). Adjacent portions of Canada and similar regions in Europe and Asia
are also affected. In addition to the Pacific Northwest, erosion due to frozen soil
has been noted across southern Canada (Hayhoe et al.t 1992) and in the northCopynghl O I9W Soil Soence Sockty of America, 677 S. Segoe Rd, Madison, WI 53711, USA.
Vfnabifay of Rangthmd Waier Erosion Processes, SSSA Special Publication 38.
68
SEYFRIED AND FLERCHINGER
Soil Freezc-Thaw Boundary
Fig. 6-1. Ponion of continental USA affected by significant soil freezing.
eastern (Storey, 1955), midwestern (Garstka, 1945), and plains regions (Spomer
& Hjelmfelt, 1983) of the USA.
Soil freezing profoundly affects the potential for erosion by altering both
soil bydrologic and hydraulic properties- These properties are unique to frozen
soil and are often not associated with normal unfrozen soil erosion. For this reason, the first part of this chapter will be devoted to a brief description of bow
soil freezing affects soil properties critical to the generation of surface runoff
and the resultant erosion. We will then describe the current status of research
quantifying the important processes involved. In some cases, recently obtained,
previously unpublished work will be presented as examples of the state of
research. Finally, we will use measured data to illustrate how these processes
operate on rangelands.
SOIL FREEZING EFFECTS
Liquid Water Content
Soil is considered to be frozen when the soil solution is frozen. In general,
this begins to occur between -0.1 and -0-3°C but may occur at temperatures as
low as -2 or -3°C (Tsytovich, 1975)- This reduction in the freezing point below
0°C is known as the freezing point depression and is attributed to physical interactions and solute effects (Miller, 1980). As the temperature drops below the
freezing point depression, increasing amounts of soil water freeze. Most of the
soil water freezes between 0 and -5°C (Tice et al., 1976), but some unfrozen (liquid) water may be found at temperatures as low as ^M)°C (Anderson & Tice,
1972). Thus, whenever the soil is frozen the total soil-water content is composed
of both liquid (unfrozen) and solid (frozen) water. The liquid water in frozen soil
contains all the unprecipitated soil solution solutes and is located next to the soil
particlr
INFLUENCE OF FROZEN SOIL ON RA.NGELAND EROSION
0.30
~ 0.25
.§ 0.20
|
o 0.15
O
0.10
" 0.05
0.00
Grov
Unfrozen
Frozen
Fig. 6-2- Frozen and unfrozen liquid water content of a sand at -5°C as a function of total water content. GRAY refers to grav imetrically determined soil-water content. Both the frozen and unfrozen
liquid water contents were measured with lime domain reflectometry (TDR).
The following expression for the soil matric and osmotic potentials has been
derived from the Clausius-Ciapeyron equation given the assumption that the ice
pressure is zero:
0^(7-273X7=*.
where o| is the density of liquid water (Kg/m3), L, is the latent heat of fusion
(J/kg), T is the temperature (K), *m is the soil matric potential (Pa), and <X>o is the
soil solution osmotic potential (Pa)- Equation [1] can be combined with equations
relating matric potential to soil-water content to estimate the liquid water content
(6J across a range of temperature. For example, we used the van Genuchten
equation (van Genuchten, 1980) and measured soil solution concentration to
obtain an expression for 0L. It predicts that 6L is dependent on soil properties and
temperature, but not on the total water content. The approach was tested in sand
using time domain reflectometry (TDR) to measure 6L_ The unfrozen 6 values are
the total water content, the frozen are the measured liquid water contents (Fig.
6-2). These results agree fairly well with calculated value of 3.5% (TDR accuracy is ±2%). Work is continuing to determine if the apparent increase in 6L with
total water content is due to deviations from the theory or experimental artifact
Frozen Soil Properties
One of the most obvious properties of frozen soils is hardness. A high ice
content soil can have a hardness similar to concrete. Hardness has, in fact, been
used to detect the presence or absence of frozen soils. The primary determinate
of hardness is the ice quantity; however, qualitative effects determined by the
freezing conditions and soil properties can also affect the structure and thus
76
SEVFRIED AND FLERCHINGER
strength of frozen soil. At least three different structures of soil ice have been
observed (Haupt, 1967): concrete, porous concrete, and stalactite. Concrete frost
is the hardest and is associated with high ice contents.
Conversion of soil water to ice has important impacts on the soil thermal
regime. This is clearly an important consideration because the soil ice content is
ilependent on temperature. The thermal conductivity of ice is about four times
greater than that of water, while the heat capacity of soil water is about double
that of ice. This means that heat can be transmitted much more rapidly through
frozen than unfrozen soil with the same total water content. Heat transmission
can be greatly moderated by the energy released upon freezing of water, or the
heat of fusion, which buffers soil temperatures near the freezing point.
The affects of freezing are, in some ways, analogous to those of drying.
Liquid soil-water interacts with ice in much the same way that it interacts with
mineral soil. Water is bound more strongly to mineral and ice surfaces as the
water (liquid) content decreases. Thus freezing results in large matric potential
decreases. Koopmans and Miller (1966) demonstrated this quantitatively for
some soils by accounting for the ratio of the surface tension at the air-water interface to that at the ice-water interface in mineral soil. Soil freezing from the surface downward can therefore result in large upward matric potential gradients.
There are numerous observations of substantial upward redistribution of soil
water towards the freezing front by this mechanism (Willis et aK, 1964; Kane &
Stein, 1983; Granger & Gray, 1990; Pikul & Zuzel, 1990). When heat consumption from the surface is balanced by heat release from the freezing of water transported upward to the frozen zone, the downward progress of the freezing front
can be halted and ice lenses formed that cause heaving (Miller, 1980).
As with soil drying, soil freezing results in a dramatic reduction in hydraulic
conductivity. Ice blocks soil pores, effectively reducing the cross sectional area
available for water transport. Measurement of frozen soil hydraulic conductivity
is experimentally challenging and no technique is universally accepted, but one
result seems clear: the saturated hydraulic conductivity drops dramatically as
soon as ice is present (Black & Miller, 1990). Williams and Burl (1974) measured
a drop in saturated hydraulic conductivity of five orders of magnitude from -10"*
1
M
1
m s^ near 0°C to -IQ~ m s' at -0.4°C This means that infiltration into frozen
soil is largely through the air-filled porosity. In order to estimate this quantity, 9L
and the expansion of water upon freezing (9%) must be taken into account
QUANTIFICATION OF FROZEN SOIL RUNOFF AND EROSION
Adequate deterministic description of runoff and erosion from frozen soils
requires the following information: (i) occurrence and depth of frozen soil, (ii)
the effect of frozen soil on infiltration and surface runoff, and (iii) the effect of
soil freezing on soil erodibilitv
Frozen Soil Occurrence and Depth
The occurrence and depth of frozen soil are dependent on the interrelated
processes of heat and water transfer at the soil surface and within the soil profile.
INFLUENCE OF FROZEN SOIL ON RANGELAND EROSION
71
Therefore, factors controlling heat and water transfer will have a direct effect on
soil freezing. The energy balance at the soil surface is defined as:
o
•
L J
where Ro is absorbed solar radiation, Lo is net long-wave radiation exchange, H is
sensible heat transfer, X£ is latent heat vaporization at the surface, and G is heat
transferred to the soil all in units of W or2- Heat and water transfer at the soil surface are governed by the meteorological and environmental conditions at the
soil-atmosphere interface- Meteorological variables governing heat and water
exchange at the surface include solar radiation, wind speed, precipitation, humidity, and, most importantly, air temperature.
Environmental conditions affect occurrence and depth of frozen soil by altering one or more of the modes of heat transfer at the surface. Snow, organic
residue, and vegetative cover insulate the soil surface, alter surface albedo and
change long-wave exchange characteristics. Thorud and Duncan (1972) found the
average depth of frozen soil was 38 cm deeper in plots without snow than in plots
with 13 cm of snow cover. During the same period, plots without litter froze 20
cm deeper than litter-covered plots. Pikul et al. (1986) found that standing stubble reduced frost penetration an average of 35% compared with tilled or baresurface plots. Other important environmental conditions that alter the solar radiation incident on the soil surface include the slope and aspect
Critical soil properties governing soil freezing include soil-water content and
bulk density. Although soil hydraulic properties do not drastically affect frost
depth, their affect on moisture migration to the freezing front, and therefore ice
content and infiltration into the frozen matrix, can be significant (Flerchinger,
1991). High soil-water content prior to freezing can result in slower, shallower,
frost penetration due to increased latent heat of fusion, but may also lead to
impermeable-frozen soil due to higher ice contents (Willis et al., 1961).
With sufficient information of heat and water transfer within the soil, depth of
soil freezing and thawing can be predicted quite accurately. Some models that have
been developed for predicting soil freezing include those of Benoit and Mostaghimi
(1985), Grant (1992), Lundin (1990), and Flerchinger and Saxton (1989).
The Simultaneous Heat and Water transfer (SHAW) model, developed by
Flerchinger and Saxton (1989) was recently used to study the effects of snow and
plant residue on the depth of soil freezing. Required model inputs include hourly
or daily weather data, slope, aspect soil solution concentration, soil physical
properties, vegetation, snow, and residue cover The study sites included mountainous rangeland sites with extreme differences in snow cover (Fig. 6-3X and
agricultural plots in interior Alaska with different tillage and residue cover (Fig.
6-4), The Quonset and Reynolds Mountain rangeland sites (Fig. 6h-3) are on the
Reynolds Creek Watershed, which will be described in more detail later The
Quonset site elevation is 1190 m and has a much wanner, drier climate than the
Reynolds Mountain site located a few miles away at 2020 m. The frozen soil
depth was much deeper but less persistent at the Quonset due largely to the lack
of insulation from snow. Simulated and measured snow and frozen soil depth are
in good agreement for both rangeland sites even though the temperature and snow
72
measured mow depth
frost ti.be
simulated frost ft snow depths
250 280 310 340
5
35
65
95
125 155
Day of Year
Fig. 6-3. Simulated and measured depth of soil freezing and snow accumulation for
(A) the Quonset site, (B) the Reynolds Mountain site (Flerchinger et al., 1990).
winter at
10
10
~ 20
30
o
* 40
c
50
60
70
80
90
285
290
295
300
305
310
315
320
325
Day of Tear
Fig. 6-4. Simulated and measured depth of soil freezing and snow accumulation under different
tillage and residue regimes in Alaska (Flerctunger et al., 1990).
73
regimes were quite different {Fig. 6-3)- The effect of plant residue is shown in
Fig. 6-4, where simulated and measured frost depths are plotted for disked and
no-till plots having 8500 and 2000 kg ha"1 of residue cover, respectively. Again
the agreement between the measured and simulated values is quite good. These
data show that, given sufficient information, soil freezing can be simulated well
under a wide range of conditions.
Frozen Soil Infiltration and Runoff
Unfortunately, simply predicting frozen soil occurrence and depth is not sufficient to accurately predict the susceptibility of frozen soil to surface runoff and
erosion. Soil freezing may result in either a slight increase or a decrease in infiltrability (Haupl, 1967; Blackburn & Wood, 1990). We have measured infiltration
rates of zero in frozen soil. Several factors have been linked to infiltration rates
in frozen soils including soil-water content, ice structure, soil structure, cover,
and soil texture. Probably the most important is the total soil-water content
(Stoeckeler & Weitzman, 1960; Kane, 1980; Granger & Gray, 1990). Factors
such as texture and plant cover strongly influence the soil-water content. Soil ice
structure is also largely determined by soil-water content. The most important
manifestation of soil structure is via macropores, which can have a substantial
impact on infiltration when they penetrate below the frozen zone (Harris, 1972;
Pikuletal., 1992).
Freezing affects infiltration by two basic mechanisms, an increased matric
potential gradient as described above, and pore blockage by ice. These tend to
work in opposite directions, hence the observation that freezing can result in
either increased or decreased infiltration rates. In drier soils with relatively Httle
ice, the increased matric potential gradient may actually increase infiltration
rates. As the ice content increases, however, the amount pore blockage results in
a large decrease in infiltration rate approaching zero. Measurements of infiltration into frozen soil have confirmed that higher water contents do lead to reduced
infiltration rates (Kane, 1980).
We have yet to develop a good quantitative description of water infiltration
into frozen soil This is due partly to conceptual problems and partly to experimeDtal problems. Conceptual problems are related to the unknown effects of ice
formation on the permeability of frozen soil. The experimental difficulties stem
from the basic incompatibility of frozen soil and unfrozen surface water Addition
of water to soils at temperatures below the freezing point depression results either
in the freezing of incoming water the thawing of resident iceT or both. This will
continue until either all the ice is melted or there is no pore space available. This
limitation has been observed in field studies of infiltration (Kane & Stein, 1983;
Blackburn & Wood, 1990)- Thus there is no steady-slate infiltration rate analogous to that in unfrozen soil.
One approach to this problem is to use an alternative fluid that has a freezing point substantially below the temperature of interest. We have used air for this
purpose. Measured air permeabilitiescan be linked to the hydraulic conductivity
provided that viscosity and density differences between the fluids are accounted
for (Gary et al., 1989). Preliminary experimental results show that reductions in
air flow rates in sand due to freezing agree qualitatively with expectations
74
SEYFRIEDAND FLERCHINGER
1500
5%
1200
O
o
a
900
j2
10%
15%
20%
25%
30%
600
Unfrozen
Frozen
Fig, 6-5. Air flow rate in a sand under frozen (-5°C) and unfrozen conditions for different volume!
ric soil-water contents expressed in percent
(Fig. 6-5). For example, we see little change with freezing at low water contents
where all or most of the water remains as liquid during freezing. As the total
water content increases, the amount of ice formed increases so that the pore
blockage increases. Thus the air permeability is seen to decrease more at greater
total water contents; however, we have yet to develop expressions that accurately describe this phenomena.
i Soil Erodibility
The credibility of frozen soil is, in some respects, similar to unfrozen soil.
Factors such as texture and organic matter content are important to both. There
are, however, additional considerations in frozen soil that can have a great impact
on soil erosion.
Erosion from frozen soil is generally from soil that is at least partially
thawed. The presence of liquid water on the soil surface, a prerequisite for runoff
and water erosion, indicates thawing conditions. Except at very low ice contents,
the soil is very unerodible when still frozen (Haupt, 1967). It is when the soil
begins to thaw that conditions change dramatically. At that point, altered aggregate stability and water conditions can lead to highly erosive conditions.
The effects of freezing on soil aggregate stability are quite complex, being
sensitive to aggregate size, soil type, porosity, organic matter, bulk density, and
the soil solution concentration (Benoit & Vborhees, 1990; Perfect et al., 1990;
Lehrsch et at, 1991). It has generally been found that increasing the soil-water
content at the time of freezing leads to reduced aggregate stability (Benoit, 1973;
Bullock et al., 1988; Lehrsch et ah, 1991). It is likely that greater progress in that
field wilt occur when traditional aggregate stability studies incorporate soil physical interactions as suggested by Gary (1992).
Soil-water contents may be quite high near the soil surface due to upward
movement of soil water. This means that a large amount of water is present upon
INFLUENCE OF FROZEN SOIL ON RANG ELAND EROSION
75
thawing. In addition, the underlying frozen soil zone may be virtually impermeable, so thai drainage is prevented. Given an addition of rain water or snowmelt,
saturated soil conditions can exit thus shear strength becomes negligible (Miller,
1980; Formanek et ah, 1984), and the soil becomes extremely erodible. Very high
erosion rates have been measured under these conditions (Edwards & Burney,
1989; Blackburn et al., 1990). Note that high erosion rates can occur with either
low or no (in the case of snowmelt) precipitation input, which is very different
from the normal, unfrozen case,
Another extremely important consideration, which is specific to frozen soils,
is the effect of snow cover The insulating effect of snow has been described previously. This can work to minimize seasonal frost or maintain it during specific
thaw events. It can also have a strongly retarding effect on erosion similar to the
presence of mulch (Haupt, 1967).
FIELD OBSERVATIONS IN RANGELANDS
The discussion thus far has focused on general principles of frozen soil erosion and runoff without particular reference to rangelands. We fett that this was
necessary because frozen soil processes are so different from the more commonly assumed erosion processes and because we wish to provide a basis for the reader to extrapolate beyond the examples we describe here to other, possibly more
relevant, settings.
The field data reported here will originate primarily from the Reynolds
Creek Watershed in southwest Idaho. The watershed measures -250 km2 inextent
most of which is semiarid sagebrush rangeland. The elevation ranges from HOD
to 2100 m with the highest elevations in the south (Fig. 6-6). The average annual precipitation ranges from <250 mm at the lower elevations to >1000 mm at the
higher elevations. This provides a climatic gradient to examine trends in runoff
and erosion production.
There is a network of precipitation gauges and subwatershed weirs within the
larger Reynolds Creek watershed Four subwatersheds will be discussed: Flats,
Nancy's Gulch, Lower Sheep Creek, and Reynolds Mountain. These were selected because they encompass the range of elevational and climatic conditions in the
watershed (Table 6-1)In addition to the dramatic increase in precipitation with elevation there is an
even more dramatic increase in streamflow (Table 6-1). Virtually all slreamflow
from the higher elevations is derived from snowmelt- Surface runoff is rarely
observed. In contrast stream flow from the lower elevations is from surface
runoff, which is almost entirely related to frozen soil. Analysis of the conditions
associated with runoff at Lower Sheep showed that almost all the surface runoff
came under conditions of frozen soil, shallow (<35 cm) snow cover, and rainfall
(Seyfried et al., 1990).
The dominance of frozen soil ranoff from mid and lower elevations results
from a combination of soil and climatic conditions. Climatic factors include the
precipitation distribution and intensity in the region. About 60% of the precipitation falls from November through April at the lower elevations, which brackets
the time that the soil is frozen. Most of these storms are regional frontal systems
SEYFRIEDAND FLERCHINGER
N
a Flats
b Nancy's Gulch
f
c Lover Sheep Creek
d Reynolds Mountain
Fig- 6-6, Elevations at Reynolds Creek experimental watershed with locations of subwalersheds.
Table 6-1. Avenge frozen soil runoff related to total runoff, precipitation and elevation at four subwêidttds at the Reynolds Creek watershed for water years 1972 lo 1984. Water year is from
October 1 to September 30.
Subwatershed
Flats
Nancy's Gulch
Lower Sheep
Reynolds Mountain
'Mean annual precipitation.
Elevation
•
1183
1414
1588
2018
MAP*
Total runoff
Frozen soil runoff
mm
255
317
379
1138
nun
1.6
1J
83
%
91
94
83
600,9
4
INFLUENCE OF FROZEN SOIL ON R VNGELAND HROSION
77
[
characterized by fairly low (<5.0 mm h ) precipitation intensities (Hanson &
Morris, 1982). This compares with measured soil infiltration capacities of >60
mm h - 1 at Lower Sheep (Johnson et aL 1984), and ~4<) mm h" 1 at Nancy's Gulch,
which is very similar to the Flats (Johnson et aL, 1984). Under these conditions,
a reduction in soil infiltrability is required to generate cool season runoff.
As expected, erosion follows these trends. Johnson ei aL (1985) divided sediment producing runoff events into three categories, those resulting from: (i) rainfall alone, (ii) snowmelt alone, and (iii) combined snowmeft and rainfall. They
estimated that 90% of the sediment produced at Reynolds Creek came from combined events, while only 3% resulted from rainfall only. The combined event conditions, which dominate sediment production, are essen?^"y the *amp a* those
thai cause surface runoff,
Large Scale Variability
*
Hydrographs from frozen soil runoff events are characteristically spiked, rising quickly, and falling quickly (Fig. 6-7) relative to the broad seasonal snowmelt
patterns. There is little surface storage at such times, leading to the rapid rise, and
runoff ceases quickly due to either an end of the precipitation, a change of rain to
snow, or soil thawing. This is consistent with other observations (Garstka, 1945).
E
E
o
-
Q-
8
0
10 20 30 40 50 60 70 80 90 100110120
Day of Y«ar
Fig. 6-7. StFeunflow measured if the weir per unit area and precipiuawi tt Lower Sheep Creek
during 1975.
78
Note also in Fig. 6-7 that later season precipitation produced virtually no surface
runoff, which is consistent with the discussion above.
Such distinctive hydrographs are traceable throughout the watershed to other
weirs. This allows the interpolation of runoff between subwatersheds to get a picture of the spatial and temporal distribution of runoff generated from specific
events. An example of this was provided by Johnson and McArthur (1973), who
mapped the measured precipitation and corresponding runoff resulting from two
storms in 1972 (Fig. 6-8). The watershed outlet and low-est elevation (1000 m) is
near the middle of the watershed at the northern extreme. Elevations increase in
all directions from there with the highest points near the southern border Runoff
from the 18 January event was greatest at elevations below 1570 m where the soil
was deeply frozen and there was shallow (10-15 cm deep) snow cover, even
though the total precipitation, which came as rain, was lowest (13 mm) on that
part of the watershed. In contrast, the soil had completely thawed at lower elevations on 2 March, and runoff was greatest near the 1570 m elevations where the
soil was still frozen and there was a shallow snow cover Precipitation during both
storms was mainly snow at the highest elevations, resulting in a maximum runoff
at the intermediate elevations for the 2 March. The pattern of the second event is
the more common (Seyfried et ah, 1990).
Small Scale Variability
Clearly, there is considerable spatial and temporal variability of frozen soil
runoff production. This is because it is highly sensitive to the transient, unstable
conditions in which ice and liquid water coexist. These same factors create heterogeneities at much smaller scales, which makes generalization and extrapolation of subwatershed behavior difficult.
One of the primary causes of heterogeneity is the vegetation. The presence
of perennial shrubs creates a substantially different microenvironment that results
in contrasting soil properties beneath and between shrubs (Hugie & Passey,
1964). ID general, the soils between shrubs (interspace soils) have lower infiltration capacity and are more erodible than those soils under the shrubs (Blackburn,
1975; Johnson & Gordon, 1988; Seyfried, 1991). Similar effects are also present
in frozen soils. Blackburr et al. (1990) showed that, as with unfrozen soils, there
was considerably more runoff and erosion from interspace than shrub soils when
frozen.
Another important impact of vegetation is its effect on snow accumulation
due to interaction with wind- In much of the watershed the maximum height of
snow accumulation is determined by the height of the vegetation. The taller the
vegetation, the more snow is trapped. As mentioned above, frozen soil events are
largely restricted to shallow snow covers. This tends to restrict frozen soil events
to shrub vegetated rangetands at elevations below the tree line. Other important
heterogeneities are related to factors that affect the energy balance and water
movement such as slope, aspect and soil type.
Much of the concern with frozen soil runoff from rangelands is with large
scale events that result in flooding and massive sediment loads. Accurate prediction of these events and potential amelioration of their effects requires that watershed variability be incorporated into large scale models. This may be accom-
INFLUENCE OF FROZEN SOIL ON RANG ELAND EROSION
79
0.5—:
0.0
0.0—
^
1.0
1.5
1.5
1.0
1JO
0.0
Fig, 6-8. Runoff and precipitation dtstribiflioa at Reynolds Creek for two different sunns in 1972.
(a) Precipitation distribution for the 18 Immmy event, (b) Strcamflow distribution (per i
for the 18 January event- (c) Precipitation distribution for die 2 March event and (d)
distribution for the 2 March event
M
plished either by developing large scale measurement techniques (e.g., remote
sensing) that implicitly integrate variability, or extrapolating our process-level
research up in scale. Both approaches have limitations. The large scale measurement techniques have yet to be developed. On the other hand, we still lack coherent understanding of some of the basic processes involved at a small scale.
Research is currently under way on both of these topics.
SUMMARY
We have attempted to briefly describe the important processes associated
with soil freezing and how they affect surface runoff and erosion from rangeland
soils. Description and ultimate prediction of these processes requires knowledge
of frozen soil occurrence and depth, and the effect of soil freezing on soil infiltrability and credibility. Research to date has been reasonably successful in
describing frozen soil depth and occurrence but not so successful for frozen soil
infiltrability and erodibility. Accurate characterization of the relationship
between soil-water content and frozen soil infiltrability is an important first step.
A better understanding of how ice interacts with soil aggregates in conjunction
with changing soil water content and temperature will aid in describing effects of
soil freezing on erodibility. Application of this knowledge to rangelands will
require incorporation of the effects of small and large-scale spatial variability on
soil freezing, infiltrability, and erodibility.
REFERENCES
Anderson. D.M., and A.R. Tke. 1972. Predicting unfrozen water contents m frozen sods from surface
area measurements. High. Res. Rec. 393:12-18.
BenoiL G.R- 1973. Effect of frceze-thaw cycles on aggregate stability and hydraulic conductivity of
three soil aggregate sizes. Soil Sci. Soc. Am, Proc, 37:3-5.
BenoiL G.IL, and S. Mosughimi. 1985. Modeling soil frost depth under three tillage systems. Trans.
ASAE 28:1499-1505.
Bcnoit, G.R., and W.B. Vborhees. 1990. Effect of freeze-thaw activity on water retention, hydraulic
conductivity, density, and surface strength of two soils frozen at high water content, p. 45-53. In
ICR. Cooky (cd.) Frozen soil impacts on agriculture, range, and forest soils. Proc Int.
Symposium, Spokane, WA- 21-22 Mar. 1990- CRREL Special Rep. 90-1. U.S. Army Cold
Regions Re*, and Eng. Lab., Hanover, NH
Black, PR, and RJX Miller. 1990. Hydraulic conductivity and unfrozen water content of air-free
frozen silL Water Resour. Res. 26:323-329.
Blackburn, W.R 1975. Factor* influencing infiltration and sedMcaf production of semiarid rangelands in Nevada. Water Resour. Res, 11:929*437.
BUcktaurn, W.H-. F.B. Ptenon, and M.S. Seyfrkd. 1990. Spatial and temporal influence of soil frost
oa infiltration and erosion of sagebrush rangelands Water Rcsour. Bull. 26:991-997.
Blackburn, W.H.. and M.K. ttfood. 1990, Influence of soil frost on infiltration of shrub coppice dune
and dune interspace soil in southeastern Nevada. Great Basin Nat. 50:41 -46
Bullock, M.S.. W.D, Kemper, and SJ> Ncbon_ 1988, Soil cohesion as affected by freezing, water
content, time and tMl^fti Soil Sci, Soc. Am. J. 52-770-776.
Gary, J.W. 1992- Comments on "Freezing effects on aggregate stability affected by lexture, mineralogy, and ocgank matter- Sod Sci. Soc Am- J. 56:1659.
Gary, J,W., C.S Simmons, and J.F McBnde. 1989. Permeability of air and immiscible organic liquids in porous media. Water Resour. Bull. 25:1205-1216,
Edwards, UM.T and J R. Burney. 1989, The effect of antecedent freeze-thjw frequency on runoff and
soil loss from frozen soil with and without subsoil rnayaction and g\\*m\ cover. Can. j. Soil
Sci 69:799^-811,
81
Flerchinger, G.N. 1991. Sensitivity of soil freezing simulated b> the SHAW model, Trans. ASAE
34:23X1-2389.
Flerchingen G.N,, R.F. Cullum. CL Hanson and K.E. Saxton. 1M9Q. Soil freezing and thawing simulation with the SHAW model, p, 77-£6. In K.R. Cooky (cd.) Frozen soil impacts on agriculture, range, and forest soils, Proc. Int. Symposium. Spokane. WA. 21-22 Mar. 1990. CRREL
Special Rep. 90-L U.S. Army Cold Regions Res, and Eng, Lab., Hanover NH.
Fte:chinger. G.N.. and K.E. Saxton. 1989. Simultaneous heat and *ater model of a freezing snowresidue-soil surface. I. Theory and development. Trans. ASAE 32:565-571.
Formanelc, G.E., D.IC McCool, and R,!. Papendiclc. 1984. Freeze-lhaw and consolidation effects on
strength of a wet silt loam. Trans, ASAE 27:1749-1752.
Formanek, G.E., G.B. Muckel. and W.R. Evans. 1990. Conservation applications impacted by soil
freeze-ihaw. p. 108-112, In K.R, Cooley (ed) Frozen soil impacts on agriculture, range, and forest soils. Proc. from International Symposium. Spokane, WA, 21-22 Mar. 1990. CRREL Special
Rep. 90-1. ILS. Army Cold Regions Res. and Eng. Lab., Hanover, NH.
Garsika, W.U. 1945, Hydrology of small watersheds tinder winter conditions of snow cover and
frozen soil. Trans. Am. Geophys. Union. 25:838-871.
Granger, RJ.T and D.M. Gray. 1990, The impact of frozen soil on prairie hydrology, p. 247-256, /«
K.R. Cooky (ed.) Frozen soil impacts on agriculture, range, and forest soils. Proc. Int.
Symposium' Spokane, WA, 21-22 Mar. 1990. CRREL Special Rep. 90-1. U.S. Army Cold
Regions Res, and Eng. Lab.. Hanover, NH.
Gram. R.F. 1992. Dynamic simulation of phase changes in snowpack and soils. Soil Sci. Soc. Am. J.
56:1051-1062Hanson, C.L, and R.P. Morris. 1982. Precipitation-duration-frequency relationships for a mountainous watershed in southwest Idaho. Trans. ASAE 25:1637-1640.
Harm, A.R. 1972. Infiltration rate as affected by soil freezing under three cover types. Soil Sci. Soc.
AOL Proc, 36:489^191
Haupt. H.F 1967. Infiltration, overland flow, and soil movement on frozen and snow-covered plots.
Water Resour. Res. 3:145-161.
Hayboe, H.N,, R.C. PclleticrtandS- Moggridge, 1992-Analysis of freezc-thaw cycles and rainfall on
frozen soil at seven Canadian locations. Can. Agri. Eng. 34:135-142.
Hugie, V.It, and H.B. Passey. 1964. Soil surface patterns of some semi and soils in northern Utah,
southern Idaho, and northeastern Nevada. Soil Sci. Soc Am. Proc. 28:786-792.
Johnson, C.W.. and N.D. Cordon. 1988. Runoff and erosion from rainfall simulator pkxs on sagebrush nngeUixL Trans, ASAE 31 42l-*2"
Johnson, C.W., N D Gordon, and CL Hanson. 1985. Northwest rangeland sediment yield analysis
by the MUSLE- Trans. ASAE 28:1889-1895.
Johnson, C.W., and R.P. MeArthur. 1973. Winter storm and flood analysis. Northwest interior- p
35^-369. la Proc. of the Hydraulic Division Specialty Conf.. Bozeman. MT. 15-17 Aug. ASCE,
New York.
Johnson, C. W . M.R. Savabi. and S.A. Loomis. 1984. Rangeland erosion measurements for the USLETrans. ASAE 27:1313-1320,
Kane, D.L 1980. Snowmeit infiltration into seasonally frozen soils. Cold Reg, Sci. TechnoL
3:153-161.
Kane, D.L, and J. Stein. 1983. Water movement into seasonally frozen soils. Water Resour. Res.
19:1547-1557.
Kooproaas, R. W.R.. and R. D Miller. 1966. Soil freezing and soil-water characteristic curves. Soil Sci
Soc Am. Proc 30:680-683.
Lchrsch, G.A., R E Sojka, D.L. Gaiter, and P.M. Joiley. 1991. Effects of freezing on aggregate stability affected by texture, mineralogy, and organic matter. Soil Sci- Soc. Am, J. 55:1401-1406.
Loodux LC 1990- Hydraulic properties in an operational model of frozen soil. J. Hydro!.
(Amsterdam) 118:289-310.
Miller, RJX 1980. Freezing phenomena in soils, p. 254-299. lit D Hilfel (ed.) Applications of soU
physjct. Academic Press, New Yoct
ET W K.P van Loon, B D Kiy. and PH. Groenevelt. 1990. Influence of ice scgraytion and
on soil structural sttbility/Can J, Soil Set 7O371-58I.
Not, J.L, and J.F. Zuzel. 1990, Heal and *aier flux in a diumally freezing aad thawing soil. p.
113-119. fn K_R. Cooley (ed) Frozen soil impacts on agriculture, range, and forest soils. Proc,
lni_ Symposium. Spokane, WA. 21-22 Mar 1990. CRREL Special Rep. 9Q-I. U.S. Army Cold
Regions Res. and Eng. Lad, Hanover. N H
Hbil* J-U ir-. J F Zuzel. and R.N. GrecrmaiL 1986. Formation of soil frost *s influenced by tillage
and residue manwBcmeoL J. Soil Water ConserN 41.196-199.
82
Pikul, J.L, J.F. Zuzd, and D,E. Wilkins. 1992. Infiltration into frozen soil as affected by ripping.
Trans, ASAE 35:83-90,
Seyfried. M.S. 1991. Infiltration patterns from simulated rainfall on a semiarid range land soil. Soil
Sci. Soc. Am, J. 55:1726-1734.
Seyfried, M.S., B.P. Wilcox, and K.R. Cooky- 1990, Environmental conditions and processes associated with runoff from frozen soil at Reynolds Creek Watershed, P. 125-134. In KLR. Cooky (ed.)
Frozen soil impacts on agriculture, range, and fores* soils, Proc. Int. Symposium, Spokane, WA.
21-22 Mar. 1990. CRREL Special Rep. 90-1. U.S. Anny Cold Regions Res. and Eng. Lab.,
Hanover, NH.
Spomer, R.G,T and S.T, Hjelmfell, Jr. 1983. Snowmeli runoff and erosion on Iowa loess soils. Trans.
ASAE 26:1109-1116.
Sioeckekr J.hL and S Wcitzman. I960. Infiltration rales in frozen soils in Northern Minnesota. Soil
Sci. Soc-Am. Proc. 24:137-139.
Storey, H.C 1955, Frozen soil and spring and winter floods, p, 179-184. In Water. USDA, Yearbook
of Agriculture. U.S, Gov, Print Office, Washington DC
Thorud. D B.. and D.P. Duncan. 1972. Effects of snow removal, litter removal and soil compaction on
soil freezing and thawing in a Minnesota oak stand. Soil Sci. Soc. of Am. Proc, 36:153-157.
Tice, A.R., D.M. Anderson, and A. Banin. 1976. The prediction of unfrozen water contents in frozen
soils from liquid limit determinations- CRREL Rep- 76-8. Cold Regions Res, and Eng, Lab..
Hanover, NH.
Tsytovich. N,A. 1975. The mechanics of frozen ground. McGraw-Hill, San Francisco.
van Gcnochten, M.Th. 1980. A closed-form equation for predicting the hydraulic conductivity of
unsaturated soils. Soil Set. Soc. Am. J, 44:892-898.
Williams, PJ., and TP. Burt, 1974. Measurement rf hydraulic conductivity of frozen soils. Can.
GeocedLJ. 11:647-650.
Willis, W.O., C.W, Carlson. J. Alessi, and HJ. Haas. 1961- Depth of freezing and spring run-off as
to fall soil-moisture level. Can. J. Soil Sci. 41:115-123.
Willis, WO , H.L Parkinson, C.W. Carlson, and HJ. Haas. 1964. Water table changes in soil moisture loss under frozen conditions. Soil Sci, 98:244-248.
Wischmeier, W.H., and D.D. Smith. 1978. Predicting rainfall erosion losses, p. 58. In USDA-SEA,
Agric, Handbook 537. United Engineering Center. New York.
Variations in Plants, Soils, Water,
and Erosion in a Pinyon Pine and
Juniper Dominated Range Site
M. Karl Wood and David Hereford
New Mexico Slate University
Las Cruces, New Mexico
Charles Souders
J
Gtia National Forest
Silver City, New Mexico
%
Alison Hill
U.S. Army Corps of Engineers
Champaign, Illinois
ABSTRACT
Surface erosion on pinyoo pine (Pinus edulis Engelm.)-juniper (Juniperus spp.)
dominated rangelands varied spatially and temporally due to the confounding effects of
erratic climate, topographic changes, incongruities of soil and geologic substrate, and various other perturbations. Measurement of spatial variation was affected by plot size. Small
plots (1 m2) were influenced by differences in soils, geologic substrate, and plant community structure. Therefore, many plots were needed to stratify vegetation, soil, and geologic differences. Runon and runoff processes between coppice dune and dune interspaces
could not be measured. Large watershed size plots (a few hectares) were highly influenced
by topographic features such as watershed slope, aspect, and shape. Plots that were 4 by
25 m could be located to minimize topographic changes, yet were large enough to include
;cs in soil, geologic substrate, and plant community structure.
Pinion pine and juniper dominated woodlands extend across >19 million hectares
in the western USA. Much of this vast area of land was once grass and shrubland,
but now is dominated by pinyon pine-juniper as a consequence of human activities, panicuarly overgrazing and fire suppression. Some lands were invaded by
pinyon pine and juniper, while these trees were present, but relatively scarce in
other lands. Pinyon pine-juniper woodland has its own place in the ecosystem
where it is climax on hillslopes that are too rocky to have enough understory to
carry a fire. Pinyon pine^juniper woodlands are valuable because they provide
benefits to humans and their animals. These include fuelwood, fence posts, poles,
CopyngbcC 1994 Soil Science Society of Amenca, 6T7S. Seg« Rd_. Madoocu WI 53711, USA.
Water Emm Proc&tt
WOOD ET AL.
pinyon nuts, Christmas trees, outdoor recreation, and cover for big game species
such as deer, turkeys, and elk. Predominated areas also provide opportunities for
recreation, site stability, livestock grazing, and habitat for wildlife.
Because pinyon pine and juniper have invaded and dominated large areas,
control methods have been used for several decades. Fire was not commonly used
because sparse understory made burning nearly impossible as the fires did not
carry well from tree to tree due to low densities of the woodland due to limited
nutrients and intolerance to shade. In the 1950s and 1960s, mechanical methods,
especially with bulldozers and anchor chains were mostly used. Because of rising costs of energy and public perceptions of environmental disaster, this method
was abandoned by the early 1970s. By the early 1980s, several herbicides were
developed and useful. Because of public perceptions of environmental disaster
associated with pesticides and agency fear of legal ramifications, all herbicides
were banned from U.S. Forest Service lands by the late 1980s, Some attempts
have been made to aid burning of pinyon pine and juniper woodlands with highly flammable chemicals, which replace the energy lost by not being able to bum
an associated understory. Today, most pinyon pine-juniper woodland control is
conducted by allowing the public access to the woodland for fuelwood harvest
Obviously, their impact on the total acreages occupied by pinyon pines and
junipers are very low.
The purpose of this research project was to quantify over time and space the
effects of fuelwood harvesting and slash disposal on runoff, sediment concentration, sediment production, bedioad, soil water content, and understory vegetal
responses. From the literature (Evans, 1988; Pieper, 1991), it was suspected and
confirmed in the field that temporal and spatial variation would be high. Methods
to measure and minimize the variation were employed. The purpose of this chapter is to describe the variations under undisturbed natural conditions and how
they were dealt with in order to determine fuelwood harvest and slash disposal
treatment effects.
METHODS
Study Site Location
A study site was sought that would represent the pinyon pine-juniper woodland
of southern New Mexico. A site was sought where pinyon pine and juniper had
invaded or increased enough to dominate a former grassland. The most desirable site
would have uniform slope, aspect, geology, soil, mature vegetation, climate, and
:ment
Plot Size
Many plot sizes were considered from very small (1 m7) to watersheds of
many hectares. It was recognized that every size has advantage* and disadvantages relative to spatial variability and environmental constraints.
Climate
A weighing-iy pe recording rain gauge was placed on each ridge. Daily rain
gauges were placed on the eastern and western ridges, while a weekly rain gauge
VARIATIONS IN PLANTS, SOILS, WATER, AND EROSION
95
i
was placed on the center ridge. A hygrothemiograph was placed on the eastern
ridge. All three ridges were located adjacent to each other and the entire study
area was -20 ha, which would insure a similar climate.
Soils
A soil pit was excavated in the middle of the south side of each plot between
the runoff and southern plot borders. Soils were dug deep enough to classify or
to bedrock. Information taken for each pit included soil horizons, depth, color,
texture, control section, and any other property that is necessary in defining specific soil families.
Vegetation
Plant species composition was determined by examining and identifying all
plants in all plots. Cover was determined by ocular estimation using five 0.5 m2
quadrats in each plot. Plant mass was determined by clipping plants to 1.5 cm
stubble height in four 0.5 m2 quadrats in each plot
Tree density was determined by counting every tree by species in every plot
Tree diameter was measured with a diameter tape at 30 cm above the ground.
Tree height was determined with measuring poles on every tree in every plot.
Fuel wood from each plot was stacked and measured after each plot was
clearcut
Runoff and Erosion
Each 12 by 25 m plot (experimental unit) was marked with permanent steel
stakes at each corner. The centrally located runoff plots were contained by sheet
metal that was 30 cm high and buried 15 cm in the soil. On the downslope end,
a drop box and H-flume of 25.4-cm depth were installed, which led to a
Coshocton-type runoff sampler (Brakensiek et al., 1979). The flume contained
a stilling well and stage recorder. The Coshocton wheel was connected by 2.54
cm-diam. pipe to a runoff collection tank where a 1 L subsample could be taken
to measure sediment. Bedload was collected with screens placed in the drop
boxes.
RESULTS
Study Site Location
The study site was located on a mesa top that has been invaded by pinyon
pine and alligator juniper (7. deppeana SteudJ. which now dominate the site. The
study site is on Spring Mesa adjacent to Corduroy Canyon in the Black Range
Ranger District of the Gila National Forest or more specifically in Section 36 of
T9S, R12W in Catron County, New Mexico. The elevation is 2245 m. The study
site extends across -20 hectares on the western exposure of three north to south
ridges that join on the north end with Spring Mesi Slopes were 12 to 16%.
[n 1986, 20 experimental plots were established in an area where commercial fuelwood harvesters remove pinyon pine and juniper stems. Five plots were
located in each of four blocks. TTie most western ridge contained one block of
five plots, the middle ridge contained two blocks of five plots each, the most east-
WOODETAL,
ern ridge contained one block of five plots. These (plots) experimental units were
15 by 25 m and contained centrally located runoff plots, which were 3.66 by
21.13 m or 0.0081 ha. This size is the same length and twice the width as plots
used to develop the Universal Soil Loss Equation. The plots were arranged in a
random ized-complete block design. Prior to assigning slash disposal treatments,
these plots were used for collection of preliminary data on climate, vegetation,
soils, runoff, sediment yield, and soil moisture during the growing seasons of
1987 and 1988.
Plot Size
2
Measurement of spatial variation was affected by plot size. Small plots (1 m )
were influenced by differences in soils, geologic substrate, and plant community
structure. They were found to be useful for characterizing individual components
of the understory (Ward & Bolton, 1991). Many plots were needed, however, to
characterize the understory after stratifying vegetal, edaphic, and geologic differences, and run-on and run-off processes between coppice dune and dune interspaces could not be measured.
Large watershed size plots (a few hectares) were highly influenced by geomorphic features. Several watersheds empty into Cordoroy Canyon, but they are
quite different because of unique combinations of watershed shape, slope, aspect,
drainage pattern, drainage density, and stream order
Plots that were 12 by 25 m could be located to minimize topographic
changes, yet were large enough to include changes in soils geologic substrate,
and plant community structure.
CUmatc
The climate of the study site is dry temperate. Mean maximum and mini*
mum temperatures are 19°C and -L68°C. Mean annual precipitation for the last
20yr is 319 mm with —111 frost-free days from 5 June through 24 September.
Precipitation in the form of snow is experienced during December and
January. Substantial amounts of precipitation are received as short-lived summer rains that are of mild to moderate intensity. Differences in amounts of precipitation received each week varied greatly throughout the monsoon season
(Table 8-1). There were also some large differences between gauges (14 May
1987 and 27 August 1987). Although these differences could be from observer
errors (11 Nov. 1987 gauge no. 2), most are attributed to being located -30 m
apart, and the storms being very localized,
Sods
Soils were in the ton ti-Poley-Rough Broken Land Association (fine, mixed,
mesic Ustollic Haplargid). Soils in this association generally have a thin surface
layer of reddish-brown noncalcareous gravelly loam over reddish-brown gravelly clay or heavy gravelly clay loam to a depth of 30 to 90 cm.
Most of the pits at the study site had an extremely gravelly sandy loam texture (18 of 20) (Table 8-2). There was a gravel layer at the surface that may be
97
Table 8-1. TocaJ precipitation (mm depth) for each week as measured by two precipitation gauges.
Date
Gauge no. I
Gauge no. 2
MINI
14 May 1987
17
9
18 May 1987
5
0
6
3
0
5
26 May 1987
1 June 1987
9June 1987
16 June 1987
23 June 1987
30 June 1987
7 July 1987
14 July 1987
21 July 1987
29 July 1987
4 Aug.
11 Aug.
18 Aug.
27 Aug.
1987
1987
1987
1987
12 Sept 1987
25 Sept. 1987
2OO-1987
HOcL 1987
16 Oct. 1987
4
0
0
2
4
"2
6
3
2
.. 38
5
2
0
8
1
0
6
3
0
3
0
0
3
4
3
7
4
1
18
2
3
2
mtssiag
3
due to erosion. There is evidence of soil loss on the plots such as pedestalled grass
plants, and soil being deposited on the uphill side of logs and trees*
The parent material in the study area is Gila [coarse-loamy, mixed (calcareous), thermal Typic Torriflavent] conglomerate. Often the exact boundary of the
bedrock was difficult to find. The bedrock becomes soft due to weathering. The
C horizon material grades into the weathered bedrock- Because of this, soils associated with several pits were marginal toward Typic. Most of the soils are considered Lithic because most of the roots stop as they move into the harder C horizon material*
At the subgroup level, the soils are fairly uniform (Table 8-3). Ninety percent of the soils were Lithic Haplustaife. At the family level, the variability
increases, especially in the percentage of clay and rock fragments.
These soils are not the original top portions of the A horizon that formed
under grasses. Erosion of 15 to 30 cm of topsoil since establishment of pinyon
pine and juniper trees has resulted in most of the topsoil that formed under the
grass to be washed from the site. Present soils are classified as Uthic Haplustalfs
which is indicative of shallow bedrock contact with minimum horizon belonging
to the order Alfisol and having an ustic moisture regime.
Vegetation Composition
The vegetation of the study site consists of a moderately low tree density
dominated by two-needle pinyon and alligator juniper with a scattering of gray
oak (Quercus grisca (LJebm.) and an occasional ponderosa pine (Pinus ponderosa Laws) (Hill, 1990).
WOOD FT AL
Table 8-2. SoiJ characteristics for each plot.
Texture'
Plot
DO.
1
2
3
4
5
Mean
Slope
Soil depth
%
cm
15
14
15
16
14
14.8
61
46
41
48
38
46.8
5%
17%
Surface
Subsoil1
GRX-SL
GRX-SL
GRX-SL
GRX-SL
GRX-SL
GRV-SL
GRV-SCL
GRV-SCL
GRV-C
GRV-C
GRV-SL
GRX-SL
GRX-SL
GRX-SL
GRX-SL
GRV-SCL
GRV-SCL
GR-C
GR-C
GR-C
GRX-SL
GRX-SL
GRX-SL
GRX-SI
GRX-SL
GRV-SCL
GR-C
GR-C
GR-C
GR-C
GRX-SL
GRX-SL
GRV-SL
GRX-SL
GRX-SL
GR-C
GR-C
CR-C
GR-C
GRV-SCL
Cocfficienl
of variation
6
15
16
14.2
46
48
41
48
48
46.2
of variation
11%
6%
11
15
12
13
16
7
8
9
10
Mean
12
13
15
Coefficienl
14
15
Mean
Coefficient
of variation
16
17
18
19
20
Mem
Coefficient
of variation
Overall mean
43
43
14
38
61
33
14.8
43.6
5%
22%
14
16
16
16
16
15.6
33
48
48
48
48
45.0
5%
17%
14.8
45.4
7%
14%
15
14
Coefficient
of variation
*GRV = very gravelly (35-59% local rock fngniaMs with most gravel size), GRX = extremely gravelly (6Q+% tool rock fragments with mosl gravel size). SL = sandy loam. SCL = sandy day loam,
and C = day lectures.
*Subsoil may mean the control section (25-100 cm) or the argillic horizon (horizon of day
Most are the argillic horizon textures and rock fragment
' ~
99
Table 8-3. Additional soil characteristics for each plot
-
Plot
t» !
Clay
Rock
15
32
30
45
50
34.4
40
55
45
loamy-skeletal, mixed, mesic Typic Ustochrepc
loamy -skeletal, mixed, mesic Lithic Haplustalf
loamy -skeletal, mixed, mesic Lithic Haplustalf
40
clayey -skeletal, mixed, mesk Lithic Haplustalf
clayey -skeletal, mixed, mesk Lithic Haplustalf
^f
Soil family
no.
1
2
3
4
5
Mean
35
43.0
Coefficient
of variation
36%
16%
^
6
7
8
9
10
Mean
Coefficient
of variation
11
12
13
14
30
30
55
50
55
35
40
44.0
20
30
20
29.0
26%
28%
30
40
50
SO
45
45
50
50
15
Mean
40
40
40.0
Coefficient
of variation
16%
5%
45
45
30
SO
45
40
45
SO
Mean
40.0
46.0
of variation
14%
8%
Overall mean
39.6
41.5
Coefficient of
variation
25%
23%
40
loamy -skeletal, mixed, mesk Ljthic Haplustalf
loamy-skeletal, mixed, mesk Lithic Haplustalf
clayey, mixed, mesk Lithic Haplustalf
clayey, mixed, mcsic Uthic Haplustalf
clayey, mixed, mesk LJihic Haplustalf
loamy -skeletal, mixed, mesk
clayey-skeletal, mixed, mesk
clayey -skeletal, mixed, mesk
clayey-skeletal, mixed, mesk
dayey-skeletaj. mixed, mesk
Lithic Haplusialf
Uthic Haplustalf
Uthic Haplusialf
Uthic Haplustalf
EJthk HaplustaJf
dayey-skeletaJ, mixed, mesk
clayey -skektaJ. mixed, mesk
dayey-skeletai mixed, mesk
clayey -skeletal, mixed, mcsic
Uthic Haplusulf
Uihic HaplustaJf
Uthk Haptustalf
Lithic HaplustaJf
48.0
16
17
18
19
20
40
*
Coefficient
loMiy-skcknl* mixed, mesk Uthk Haplusulf
WOODETAL
100
The herbaceous growth comprises a variety of grass and forb species- The
common grass species include:
poverty threeawn
Aristida divaricata Humb. & Bonpi. ex WUld.
pine dropseed
Blepharoneuron trichloepis (Torr.) Nash
blue grama
June grass
wolftail
iongtongue
rouhly
mountain muhly
pinyon
ricegrass
little biuestem
nash squirreltail
Bouteloua gracilis (Willd ex Kunth) Lagasoa ex Griffiths
Koeleria cristata (L.) Pers.
Lycurus phleoides H.B.K.
Muhlenbergia longiligula Hitchc.
Muhlenbergia montana (Nutt.) Hitchc
Piptochaetium fimbriatum (H.B.K.) Hitchc.
Schizachyrium scoparium (Michaux) Nash
Sitanion hystrix (Nutt) J.G. Smith
The common forb species include:
milkvetch
sedge
dayflower
Astragalus spp.
Carer spp.
Commelina dianthifotia Delile
indigo bush
Dalea spp*
fleabane
Wright buckwheat
slender goldenweed
pepperweed
mustard
golden eye
Erigeron spp.
Eriogonum Wrightii Torr. ex Benth.
Haplopappus gracilis (Nutt) A- Gray
Lepidium spp.
Sisymbrium linearifolium (Gray) Pay son
\lquiera dentata
Vegetation distribution appeared to be homogeneous based on aerial photographs and visual observations made on the ground. Most trees appeared to be
mature with a few seedlings, saplings, and immature trees.
Vegetation Cover and Mass
Life forms with abundant cover such as trees in the overstory; grasses; grass,
forb, and needle litter; and bareground (Table 8-4) had coefficients of variation
Table 8-4, Mean overstory tree, undersiorv plant lioer, rock, and bareground cover (%) for all
Blots with coefficients of variation.
.
Miniinuni
value
Vanabk
Mean
value
^n
Mjft
•
J
• P^ • ••^Pfe n • fl^^K
^^
^^^^^^^^^K V
—1
86.90
ZO3
14.18
0.15
j-u-
*
^r- -M-^B—» ^« —
_
37
212
49
145
436
181
36
Uwkntory tree
Grass
Fofb
MOM
Lktea
Grass, forb, Md
0-28
41-83
8135
18J5
0.00
2J3
0.00
0.00
0.00
19.18
litter
Tree branch »ad
2.46
6.78
0.55
73
1.15
47.94
3.45
69.38
0.00
14.63
78
f"^^— . I n . i ift—^^
VJVUUUij
UCC
aemltOer
Rock
43-11
0.25
5-94
0.03
0.02
OJ3
1.73
' ^H
•
VARIATIONS IN PLANTS. SOILS, WATER, AND EROSION
Table 8-5. Mean grass, forb. and lotal plant mass (kg ha~') for each block and overall with coefficients of variation.
Block
Maximum
value
Mean
Minimum
value
Coefficiem
of variation
82
142
76
kg ha
Grass production
1
2
3
4
Total
Forb production
1
2
3
4
Total
193
166
106
114
145
318
193
151
155
318
1(0
76
40
12
27
19
49
90
78
79
43
73
. H3 .
108
97
80
113
61
57
55
20
20
19
24
17
51
35
196
216
166
126
126
26
10
10
26
31
.
Total plant production
1
2
3
4
Total
282
245
185
157
217
_
404
276
206
236
404
from 28 to 49%. They also had great differences between the maximum and minimum values such as 68.55% for tree overstory. Life forms with very small percentages of cover such as trees in the underctory; foibs; mosses; lichens; and tree
branch and stem litter had coefficients of variation that ranged from 73 to 436%.
Mean grass, forb, and total plant mass varied by 10 to 51% for various
blocks (Table 8-5). All three variables varied from 31 to 40% across all plots.
Grass values were always about two times higher than forb values. Combining
grass and forb values resulted in the lowest coefficients of variation within blocks
and overall. Cover and mass were not as homogeneous as they visually appeared.
IVw Density
The number of trees per ploc or tree density was greatest for pinyon pine followed by alligator juniper and gray oak (Table 8-6). Only one ponderosa pine
minimim
Tifcfeg-d Mo*
Specks
Meanptor1
Maximum value pior'
Minimum value p*or'
Mea« bcctvc ]
Maximum value hectare'*
Minimum value hectare'1
Coefficient of variation
23
37
10
767
1.233
333
32%
" ^^^^"^^^
ABigaior
juniper
Gray
6
14
3
200
467
100
45%
3
12
0
100
Total
400
0
89%
0.1
1
0
3
33
0
—
32
46
19
1,070
1433
633
23%
102
WOOD ET AU
was found on one plot. The coefficient of variation was lowest when all species
were combined, but higher within species. The 23% of coefficient of variation for
all species may explain why the tree ridges visually appeared to be homogeneous
from observations made both on the ground and in the air. Lower mean values
had higher coefficients of variation. The ranges were also quite high with differences between the maximum and minimum values being 26 trees per plot for
pinyon pine and 27 trees per plot for all combined species. Table 8-6 also shows
the number of trees per hectare based on each plot being 0.03 or 3% of a hectare.
On a hectare basis, a wide range is found for total number of trees.
Tree Stem Diameter
Mean tree stem diameter in each plot was greatest for alligator juniper followed by pinyon and ponderosa pines (Table 8-7). The coefficient of variation
was lowest for pinyon pine and highest for alligator juniper. Higher mean values
had higher coefficients of variation. The ranges were also quite high with differences between the maximum and minimum values being 0-77 ra for alligator
juniperTree Height
Mean height of pinyon pine and juniper trees were 4.52 and 4.23 m tall,
respectively. Gray oak is a browse species that has a mean height of 2,72 m,
which is out of reach of many
f wildlife.
Mean tree height in each plot was greatest for ponderosa pine followed by
pinyon pine and alligator juniper and gray oak (Table 8-8). The coefficient of variation was lowest for pinyon pine and highest for gray oak. Lower mean values had
TaMe 8-7. Mean, coefficient of variation, maximum, and minimum tree stem diameter (m) for each
and all species.
Species
»^P^fc»îP^îr •
Parameter
Mean plor
Maximum value ploT1
Minimum value ploc
Pinyon
pine
Alligator
juniper
033
0.76
1.05
028
19%
0.40
0.20
18%
^^—*^
Gray
Ponderosa
pine
Weighted
total
—
0.25
0.25
-
0.25
0.40
051
0.22
28%
oak
—
—
—
TaMe 8-8. Mean, coefficient of variaiioa. maximum, and minimum tree height (m) for cacti and all
Species
Parameter
Mean ptar1
Maximum value plor'
Minimum vaJuc ptar'
Pinyoa
•
pine
Alligator
•
•
jumper
4.52
42?
5.52
5.47
2.45
3.65
9%
19%
Gray
oak
2.72
4^2
135
35%
Ponderosa
V
536
5.56
536
^
^•h
TotaJ
3.93
4.88
332
12%
103
•
3
Table 8-9. Wood volume (m ) for each plot with mean and standard deviation with conversions to
volume per hectare (m3 ha'1) and acre (cords acre'1)
Pl«
~
Volume
•_^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^H
no.
j
m ptor
1
2.40
2
3
4
5
7.70
l
—
3.07
4.18
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^•i
K
_
A
cords acre-'
64.59
207.21
7.21
23.13
—
8159
112.49
9.22
12-56
—
_
6
7
8
9
10
11
12
13
14
15
1*
17
18
19
2O
^
6.01
21.74
5.46
133 JO
53.82
194.78
146.93
5.12
137.78
15.38
—
170
Jl
156.89
111.68
19.04
17.52
4.96
2.00
7.24
—
- 6.34.
5.83
4.15
—
14.90
16.40
—
12.47
^_
^
11964
273 .52
145.64
13J6
30.54
28.60
1626
5.50
147.99
16-52
Maximum value
10.16
273-52
30.54
Minimum value
2.00
53.82
6.01
Mean
Cocfficiem of variation
4.45
10.16
9.52
5.41
41%
256.19
41%
41%
higher coefficients of variation. The ranges were also quite high with differences between the maximum and minimum values being 3.02 m for alligator
juniper
Wood Volume
Wood volume for each plot are shown in Table 8-9. The plots without values were control plots and were not measured, because wood volume wasmeasured from stacked piles after cutting. Differences in wood volume are probably
due more to differences in density than diameter and height although these may
also be contributors to wood volume differences.
Erosion
Large differences in sediment concentration were found between plots and
between blocks of plots (Table 8-10). These values represent an accumulation
over an entire sampling period, which was from the middle of May through the
middle of October. This period is the growing season and when most high inien-
WOOD ET AL,
104
Table A-10. Mean runoff sediment concentration (g L'1) for each block and overall with coefficients
of variation.
Sediment concentration
Coefficient of
Variation
Block
Pkx
no.
no.
1
1
2
3
4
5
1.54
0.59
0.63
0.29
2.02
1.01
66%
2
6
7
8
9
10
0.23
0.40
0.44
0.84
0.43
0.47
43%
3
11
12
13
14
15
0.21
0.47
0.09
0.31
44%
16
17
18
19
20
134
2.03
1.07
54%
4
Mean
Plot
Block
gL-'
OJ9
OJ7
0.80
OJ2
0.86
0.72
79%
sity thunderstorms occur which result in runoff and erosion. All blocks had val1
1
ues that were close to 1 g L' and close to 0 g L" .
Like sediment concentration, large differences in bedload were found
between plots and between blocks of plots (Table 8-11). These values also represent an accumulator! over an entire sampling period. Values ranged from 349 to
3728 g plor1.
Although these erosion values are quite variable, they seem quite low,
Before invasion or increases of pinyon pine and juniper trees, historical
records show this area was dominated by grasses with a lesser component of
shrubs and forbs. Adjacent areas without pinyon pines and junipers have a
prominent A horizon, which suggests the study site also had a prominent A
horizon before invasion. Pedestalling of relic tree stumps shows a loss of from
15 to 30 cm of A horizon. The less erosive B horizon is now exposed.
Therefore, the site has experienced three periods of erosion: low erosion under
grasslands, high erosion under pinyon pines and junipers, and low erosion
under pinyon pines and junipers after the A horizon is gone in the second
period.
VARIATIONS IN PLANTS, SOILS, WATER. AND EROSION
Table 8-11. Mean bed load (g plot"1) for each block and overall wilh coefficients of variation.
Bedioad
Coefficient of
Block
Plot
no.
no.
1
I
2
3
4
5
2
3
Van at ion
ptol
877
46%
6
7
8
9
10
1432
622
1187 '
2329
791
1272
47%
11
12
13
1497
2691
1110
608
784
1338
55%
1750
68%
19
1842
436
2166
577
20
3728
16
17
18
Mean
Block
1272
1221
349
438
1106
14
15
4
Plot
1309
65%
CONCLU
The blocks were originally located based on geomorphic positions. Now the
question arises: How should the plots be put into homogeneous blocks based on
soils, vegetation, and erosion? Correlation analysts was performed between all
the variables, i.e., tree height, soil depth, and bedload. No significant and highly correlated relationships were found- The plot size (12 by 25 m) was large
enough to include variations in the soil, but obviously too small to include the
large differences in vegetation. The plots may have been too large to account for
differences in erosion. Larger plots would have included large topographic differences, which would have influenced soils, vegetation and erosion. Therefore,
the 12 by 25 m plots were probably the best choice, but not ideal. The last question then is: given the large pretreatment differences, should treatments be
applied to these plots? Which is another way of asking, would a siudy with large
pretreatmeni differences be better than no study? The slash disposal treatments
were so different and made such a big change on the land, this is considered a
valid and valuable study.
WOOD ET AL.
REFERENCES
Brakensiek, D.L.. H.B. Osbom, and WJ. Rawls. 1979. Field manual for research in agricultural
hydrology. USDA Agric. Handb. 224. U.S. Gov. Print. Office. Washington. DC
Evans, R.A. 1988. Management of pinyon-juniper woodlands. USDA-FS Gen. Tech. Rep. IN-249.
Intermountain Res. Sin*, Qgden, UT.
Hill. A. 1990- Ecology and classification of the pinyon-juniper woodlands in wesiern New Mexico.
Ph-D. dissertaiion. New Mexico Stale Univ., Las Cruces (Diss, Abstr. DA 9111293).
Pieper, R.D. 1991. Above-ground Ecology/Bio logy, p, 6-9. In Proc, Pinon Conference, Santa Fc. NM.
23 Apr. 1991. New Mexico Agric, Exp, Sui,T New Mexico Univ., Las Cruces.
Ward. TJ.. and S,M. Ballon. 1991. Hydrologic parameters for selected soils in Arizona and New
Mexico as determined by rainfall simulation. Tech. Completion Rep. 259. New Mexico Water
Resour Insu New Mexico Univ., Las Cruces,

Document - Colorado State Forest Service

Transcription

Similar documents

Purple Diamond® Semi

EROSION CONTROL is for planting

Enkamat® General Brochure

Classic Caladiums

Calla Lilies

“Mark Ricciuto”

Spring - Walter Andersen Nursery

Monthly Newsletter - MU Extension

Phycology / Microbiology

The Great Garden Dilemma