WATER RESOURCES RESEARCH, VOL. 37, NO. 12, PAGES 2863-2872, DECEMBER,
2001
The ecohydrologicalrole of soil texture
in a water-limited ecosystem
Coral Pilar Fernandez-Illescas,
Amilcare Porporato,• FranscescoLaio,•
and Ignacio Rodriguez-Iturbe
Department of Civil and EnvironmentalEngineeringand PrincetonEnvironmentalInstitute
PrincetonUniversity,Princeton,New Jersey,USA
Abstract. Soil textureis a key variablein the coupledrelationshipbetweenclimate,soil,
and vegetation.This couplingand its dependenceon soil textureare studiedin this paper
usinganalyticaldescriptionsfor soil moisturedynamicsand the corresponding
vegetation
water stress.Resultsconfirmthe importanceof soil texturein partitioningrainfall into the
mean valuesof the water balancelosscomponents,namely,evapotranspiration,
leakage,
and runoff, and in determiningvegetationwater stressfor the vegetationand climatic
regimesof the La Copita ResearchArea in Texas.Two separatemechanisms
by which
this sensitivityto soil texture can impact the ecologicalstructureat La Copita are
illustrated.First, dynamicssimilarto the inversetextureeffect,wherebythe optimal soil
texture for a givenvegetationtype changeswith rainfall amount,are demonstratedfor the
two mostcommonspeciesat this site. Second,it is shownthat soil textureplaysa major
role in the modulationof the impactthat interannualrainfall fluctuationshave on the
fitnessand coexistenceof trees and grasses.
between climate, soil, and vegetationwith explicit consideration of the stochasticnature of the rainfall input. The duraIn water-limited ecosystems,
soil moisture and vegetation tion, frequency,and intensityof water stressperiodsas funchave a coupledrelationshipthat is basicto ecosystem
dynam- tionsof climate,soil, and vegetationtype are probabilistically
ics.Figure 1 illustrateshow soil moisturehas a centralrole in characterized and constitute the basic tools to assess the relathe dynamicinteractionbetweenclimate,soil, and vegetation tive suitabilityof environmentalconditionsfor functionally
and makesexplicitthe coupleddependenceof the water bal- different typesof vegetationin water-controlledecosystems.
ance and water stressprocesseson soil moisture.Through Using this framework,preliminarywork by Laio et al. [200lb]
transpiration,plantshave an activerole in soil water use that hints toward the crucial role of soil texture in the coupled
heavily conditionsthe water balance.The water balance, in relationshipbetweenvegetationand soil moistureand in deturn, impactsplant growth, reproduction,and germination terminingfavorableconditionsto the vegetationaccordingto
through the onset of water stress.
the previouslydescribedinversetextureeffect.
Past studieshave recognizedthe importanceof soil texture
This studywill expanduponthe resultsof Laio et al. [200lb]
on the abovecoupledinteractionsand havedemonstratedhow by consideringa wider rangeof climaticandvegetativescenarit influencesvegetationpatternsthrough its impact on soil ios and all of the texturalclasseswithin the U.S.Departmentof
water availability[e.g.,Knoopand Walker,1985].Soil charac- Agriculture(USDA) [1951]soiltexturaltriangleat the La Coteristics affect the distribution and duration of water stored in
pita Research Area in the Rio Grande Plains of southern
the soil, and these,in turn, affectplant distributionand vege- Texas. In addition, the ability of soil texture to encourage
tation structure.Somesoil characteristics
can imposedrought grass/treecoexistenceat this savannasite is examinedunder
on plants even under favorable climatic conditions,as de- different rainfall amounts.The analysiswill focuson the two
scribed,for example,by Newman[1967].Moreover,the same most common vegetation types coexistingin La Copita,
type of vegetation may occur preferentially on coarse soils namely,the herbaceousC4Paspaleumsetaceumand the woody
under low-rainfallconditionsand on fine soilsunder higher- Prosopisglandulosa (honey mesquite). Throughout wellrainfall conditions;a situation referred to as the inversetexture documentedinterannualclimaticfluctuationsthese two specieshaveexperiencedboth dry and relativelywet climaticconeffect [Noy-Meir,1973].
Recentlydevelopedanalyticalschemesof the probabilistic ditions.Thus the importanceof soil textureon the soil water
structureof the soilwater balanceand the relatedplant water dynamics,water stress,and coexistencepatternswill be asdifferent specieswith different
stress[Laio et al., 2001a; Potpotatoet al., 2001] provide a sessedfor two physiologically
rooting
depths
under
quite
different
rainfall conditions.
frameworkfor a comprehensive
descriptionof theseprocesses.
1.
Introduction
The approachallowsfor an analyticaltreatment of the links
2.
•NowatDipartimento
di Idraulica
Transporti
e Infrastructure
Civili,
Soil Texture
Paper number2000WR000121.
Soil texture is a physicalindicator of the size rangesof
particleswithin a soil.All soil particle sizescan be sortedinto
three convenientsizerangescalled separatesor texturalfractions,namely,sand,silt, and clay. Conventionally,the overall
0043-1397/01/2000WR 000121$09.00
textural
Politecnicodi Torino, Torino, Italy.
Copyright2001 by the AmericanGeophysicalUnion.
2863
classification
of a soil is determined
as a function
of
2864
FERNANDEZ-ILLESCAS
Climate
Soil
ET AL.: ROLE
OF SOIL TEXTURE
Vegetation
Q Transpiration•
Figure 1. Schematicrepresentationof the vegetation-soil
moisturerelationshipin water-limitedecosystems.
the massratios of thesethree textural separates.Within the
USDA [1951] soil textural classification,soilswith different
percentagesof sand,silt, and clay are assignedto 12 different
classes(Figure 2).
Soil texture is also used as a descriptorof soil physical
propertiessuchas porosity,n; saturatedhydraulicconductivity, Ks; soil matric potential at saturation,•s; and pore size
distributionindex, b [Cosbyet al., 1984]. Although other depercentage
by weightsand
scriptorssuchashorizonand structuralsizecertainlyinfluence
the hydraulicparametersof soils,Cosbyet al. [1984]performa Figure 2. The U.S. Departmentof Agriculture(USDA) soil
two-wayanalysisof variance of nine descriptorsto conclude textural triangle.
that soil texture alone can account for most of the discernible
patternsin Ks, •s, b, and n. Under givenclimaticand vegetation conditionsthe above soil-texture-dependent
physical obtainedusingthe regressions
by Cosbyet al. [1984]for sandy
properties,through their influence on soil water movement loam (the observedsoil texture at the La Capita site) are
and the energystateof the water in the soil column,determine comparableto thosereportedat this site [USDA, 1979].
the soil wetness values which in turn establish the water con-
dition of the plant.
The studyof the probabilisticbehaviorof the water balance
and the water deficitfor the vegetationand climatetypicalof
La Capita was focusedon the 12 textural classeswithin the
USDA soil textureclassification.
The soilphysicalparameters
controllingsuch dynamics,i.e., n, Ks, •s, and b, were obtained using the univariate regressionequationsdefined by
Table 5 of Cosbyet al. [1984],whichrelate thesequantitiesto
soil texture.These univariateregressionsare sufficientto describemost of the variability in hydraulicparametersover
texturalclasses[Cosbyet al., 1984].The regressions
were extrapolatedin our studyto includesiltysoils.Typicalvaluesof
the aboveparametersfor the middle point of each of the 12
textural classes within the USDA
soil textural classification
are
3.
Site Description
The La CapitaResearchArea (27ø40'N,98ø12'W)in southern Texas offersvaluabledata on two functional,coexisting
vegetationtypes,namely,C4 grasses(mainlyPaspaleumsetaceum)andwoodyplants,primarilyProsopis
glandulosa
(honey
mesquite).Table 2 containsthe site vegetationparameters
requiredfor the modelingof the soilwater dynamicsof the two
speciesfollowing the analyticalrepresentationof the water
balancepresentedby Laio et al. [2001a].Theseparameters
includethe activesoil depth or root depth,Zr; the maximum
evapatranspiratian
rate, Emax;the soilmatricpotentialat wilt-
ing,ß .... ; the soilmatricpotentialat the levelof incipient
stomatalclosure,•s,s*; the interceptiondepth,A; and the
given in Table 1. The valuesof the soil hydraulicproperties evaporationrate at wilting,E w [Laio et al., 200lb]. The impact
Table 1. ParametersCharacterizing
the 12 Soil TexturalClassesWithin the USDA Soil TexturalTriangle•
Class
Sand
Silt,%
Sand,%
Clay,%
n, %
logI•, inchesh-1
log•,
cm
b
5
92
3
37.3
0.524
0.675
3.387
12
32
82
58
6
10
38.6
41.6
0.371
0.003
0.806
1.12
3.864
4.5
Loam
39
43
18
43.5
- 0.226
1.317
5.772
Silty loam
Sandyclay loam
Clay loam
Silty clay loam
Sandyclay
Silty clay
Clay
70
15
34
56
6
47
20
17
58
32
10
52
6
22
13
27
34
34
42
47
58
46.8
41.6
44.9
47.6
42.3
48.1
46.1
-0.624
0.003
-0.394
-0.731
-0.088
-0.792
-0.547
1.657
1.12
1.461
1.749
1.199
1.801
1.592
4.977
7.203
8.316
8.316
9.588
10.383
12.132
Silt
88
7
5
48
-0.777
1.788
3.705
Loamy sand
Sandyloam
an, porosity;b, pore sizedistributionindex;K•, saturatedhydraulicconductivity;
and •, soil matricpotentialat saturation.Valuesare
obtainedthroughthe univariateregression
equations
of Cosby
et al. [1984].Note that 1 inchh-1 = 0.30cmh-1.
FERNANDEZ-ILLESCAS
ET AL.: ROLE
OF SOIL
TEXTURE
2865
Table 2. ParametersCharacterizingVegetation at La Copita ResearchArea, Texasa
Species
Z r,
cm
E ....
cm/d-1
•I/ ,
MPa
•IJ's
,s*,
MPa
A,
cm
E w,
cm/d-1
Paspalumsetaceum
Prosopis
glandulosa
40
100
0.476
0.442
-4.5
- 3.2
-0.09
- 0.12
0.1
0.2
0.013
0.02
agr,activesoilor rootdepth;Emax,
maximum
evapotranspiration
rate;ß .... , soilmatricpotential
at wilting;•s,s*, soilmatricpotential
at
incipientstomatalclosure;A, interceptiondepth;Ew, evaporationrate at wilting.
of differentplant physiologyand rootingdepthson the water Sh;thewiltingpoint,Sw;the levelof incipientstomatalclosure,
balanceunder variable soil texture thusmay be undertaken. s*; andfieldcapacity,
Sfc,in thelossfunction;
andthemean
Climate at the site is subtropicalwith an average annual soil moisture, (s), in the pdf increase).These shifts occur
so do the adsorptemperatureof 22.4øCand a mean annualrainfall of 70 cm. becauseasthe percentageof clayincreases,
However, significantdeviationsfrom the long-term (1930- tive forceswhich permit higher water retention at any given
1986)mean rainfall over this regionwere observedduringthe suction.It is also clear from Figures3 and 4 that thesesoil1941-1960droughtandthe 1961-1983wetter-than-average
pe- texture-relatedpatternsare robustto vegetationand climatic
riod. Statisticallyhomogeneousgrowingseason(i.e., May- characteristics.
While the pdf'sof soilmoistureshownin Figure4 are quite
September)rainfall statistics,
namely,the meandepthof rainfall events, a, and the rate of storm arrivals, X, were obtained different, all of them havetheir modebetweenthe correspondfrom the daily rainfall data recorded at Alice rain station, ing Sw and s*, indicatingthe high likelihood that the two
The mode, however,is
Texas (Rainfall data are availableat http://www.ncdc.noaa.speciesat this site are water-stressed.
gov.)For the 1950-1960dryperiod,Xdry
is estimated
as0.166 more pronouncedfor sandthan for siltyloam or claybecause
d-1, andadryis estimated
as1.342cm,witha totalgrowingof the higherprobabilityof soilmoisturebeingat or nearSwfor
seasonrainfall of 34.1 cm. For the 1973-1983relativelywet sandysoilsresultingfrom the larger slopeof the lossfunction,
period,/•vet
isestimated
as0.202d-I, andI;•we
t isestimated
as p(s), in the {Sw,s* } range.As observedin Figure4, shallow1.417cm,with a total growingseasonrainfallof 43.8cm.In this
manner, the role of soil texture in the water balancemay also
be assessed for different climatic scenarios.
p(s)(d-1)
a)
0.08
4.
Sensitivityof Soil Water Dynamics to Soil
0.06
In this section,the analyticaltoolsdevelopedby Laio et al.
[2001a]to characterizethe probabilisticsoilmoisturedynamics
are appliedto thevegetationandclimaticregimesof La Copita
to investigatethe role of soil textureon the water balanceat
0.05
and vegetationtype. The reader is referred to Laio et al.
[2001a]for a detaileddescription
of thesetools.Analysiswill
focuson the sensitivityof the lossfunction(e.g., evapotranspiration plus leakage),X(s); the probabilitydensityfunction
(pdf) of soilmoisture,p(s); and the meancomponents
of the
water balanceto variationsin soil texture at a point.
/
SiltyLoam
Texture
this site and the manner in which this role varies with climate
Sand
0.07
/
tIt
Clay
.
0.04
/
0.03
,
/
/
/
/
/
0.02
0.01
[
0.4
0.2
, ...•", ,
p(s)(d-1)
i
0.6
,
.
,
,
0.8
4.1. Soil Moisture Probability Density Function
ativesoilmoisture,
p(s) = X(S)/tlZr,is shown
in Figures
3a
.
,
• S
1
b)
0.16
The dailyaveragedependenceduringthe growingseasonof
the normalizedevapotranspirationand leakagelosseson rel-
,
Sand
0.14
SiltyLoam
l
/
/
0.12
0.1
,'
Clay
.-'
and3b for Prosopis
glandulosaandPaspaleumsetaceumand for
threedifferentsoiltexturalclasses,
namely,sand,siltyloam,
0.08
and clay.Figure 4a showsthe behaviorof the growingseason
steadystatepdf of soil moisturefor Prosopisglandulosaon
thesethree soil textural classesunder the dry climatic conditions observedin the region during the 1950-1960 period,
whileFigure4b showsthe samefor Paspaleum
setaceum
under
the wet climatic conditionsobservedduring the 1973-1983
period.Figure5 showsthe steadystatemeansoilmoisture,(s),
correspondingto the different categoriesof the USDA soil
texturaltrianglefor Prosopis
glandulosa
underthe dry climatic
conditions.Figures3, 4, and 5 indicatethat as the percentage
of clay increases,the pdf and the lossfunctionshift to higher
soilmoisturevalues(i.e., the valuesof the hygroscopic
point,
0.06
tt
//
/
/
/
/
0.04
/
0.02
0.2
0.4
0.6
0.8
Figure 3. Normalized loss function p(s) for (a) Prosopis
glandulosa
and(b) Paspaleum
setaceum
on sand,siltyloam,and
clayat La CopitaResearchArea.Ks,•s,n, andb weredetermined
from soil textureusingthe univariateregression
equationsof
Cosbyet al. [1984].The parameters
characterizing
Prosopis
glandulosaandPaspaleum
setaceum
are givenin Table 2.
2866
FERNANDEZ-ILLESCAS
ET AL.'
ROLE
OF SOIL
TEXTURE
cieslike honeymesquite.As expected,a wetter climateresults
in higher(s) valuesfor both species(not shown).
p(s)
2O
Sand
17.5
\
----
Silty
Loam
4.2.
Water Balance Components
15
/'",
Clay
I/'x
12.5
10
7.5
Figure 6 showsthe mean levels of the componentsof the
water balance, namely stressed evapotranspiration, unstressedevapotranspiration,runoff, and leakage, expressed
as percentageof growingseasonrainfall for Prosopisglandulosa under dry climatic conditions on the 12 textural
classes within
2.5
......;•...•'• .................
."-.
•- s
0.1
0.2
0.3
0.4
0.5
0.6
0.7
p(s)
2O
Sand
17.5
Silty Loam
Clay
15
12.5
10
/\
7.5
5
\
2.5
0.1
0.2
0.3
0.4
0.5
0.6
the USDA
soil textural
classification.
Their
analyticalexpressionsare given by Laio et al. [2001a]. Interceptionis independentof soil texture and is therefore not
shown. Each component shows considerable variability
within the soil texture triangle.
From Figures6a and 6b one observesthat sandysoilshave
the lowest mean levels of stressedevapotranspiration,while
unstressed
evapotranspiration
is highestfor the siltyloam region of the triangle.The mannerin whichaveragepercentages
of stressedand unstressedevapotranspirationdepend on soil
textureis a consequence
of the soil-texture-dependent
patterns
in the probabilityof soilmoisturebeingbelows*, P(s* ) (Figure 7a) andthe probabilityof soilmoisturebeingaboves* (the
complementof Figure 7a), respectively.
Figure6c indicatesthat runoffconstitutes
a negligibleportion
of the water balanceat this site and its variabilitywithin the soil
triangleappearsto followgradientsin mean soil moisture,
(Figure5). Clay soils,with highermeanvaluesof soilmoisture,
are associated
with largeramountsof runoff.Mean leakagelevels
are alsoquite smalland, as shownin Figure6d, increaseas the
ofsbeingbetween
Sicand1 (seeFigure7b)increases.
Figure 4. Soil moistureprobabilitydensityfunctionfor (a) probability
for Prosopisglandulosaunder dry climatic
Prosopisglandulosaunder dry climaticconditions(i.e., mean As a consequence,
claysoilstend to exhibita greatervolumeof leakage
depthof rainfalleventsa = 1.342cmandrateof arrivalX = 0.166 conditions,
d-•) and(b) Paspaleum
setaceum
underwetclimatic
conditionsthansandysoilsdespitehavinga muchlowerKsvalue,althoughin
(i.e., mean rainfalldepth a = 1.417cm and frequencyof the both cases the amounts are minimal.
rainfallevents
X = 0.202d-•) onsand,siltyloam,andclayat La
Changingthe climaticandvegetativecharacteristics
at La CoCopitaResearchArea. SeeTable 2 for otherparameters.
pita resultsin quantitativebut not qualitativechangesin the
relationship
betweenmeanvaluesof the water balancecomponentsand soiltexture(not shown).For dry climaticconditions,
rooted speciessuchas Paspaleumsetaceum,with their faster
stressed
evapotranspiration
isthemostimportantlosscomponent
temporal fluctuationsin soil moisture resultingfrom their
of the water balancein termsof its magnitudefor both Prosopis
smaller active soil depth Zr, tend to show larger standard
glandulosaand Paspaleumsetaceum.This suggests
that under
deviationsin the soil moisturevaluesthan deeper-rootedspesuchclimaticconditionsand for any soil textureconditionsboth
species
will sufferwater deficitmostof the time. Leakageand
runoffare negligiblefor Prosopis
glandulosa
and are onlyslightly
largerfor Paspaleum
setaceum.
As the dimate becomeswetter,
the magnitudeof the unstressed
evapotranspiration
increases
significantlyfor bothspecies,
andtheoccurrence
of leakagebecomes
greaterfor the shallow-rooted
plant.
% Silt
The sensitivityof the soil water dynamicsto soil texture is
evident from the large range of valuesfor mean soil moisture
0.25-0.45
andmeanlevelsof the componentsof the water balancewithin
the USDA soil textural classification.This sensitivityis robust
< 0.25
to vegetationand climate.Differencesin soil water dynamics
due to soil texture are large enoughto suggesta pronounced
lO
9o
influence on vegetationwater stressregimes,which are the
>0.575
:'•..
'••7 0.45_0.575
9O
7O
5O
% Sand
3O
10
Figure 5. Impact of soil textureon mean relativesoil moisture duringthe growingseasonfor Prosopis
glandulosaunder
dryclimaticconditions
(i.e.,meandepthof rainfalleventsa =
1.342cm and rate of arrival• = 0.166d-x) at La Copita
ResearchArea. See Table 2 for other parameters.
focus of section 5.
5. Sensitivity of Soil Moisture Crossing
Properties and Water Stress to Soil Texture
Porporatoet al. [2001] derive analyticalexpressions
for the
expectedfrequencyand duration of soil moisture excursions
FERNANDEZ-ILLESCAS
ET AL.:
Stressed Evapotranspiration
a)
5J:i!•'•
.:-'
.'.....
%Silt
b)
I 77%
70
50
30
TEXTURE
2867
9
0
I
% Silt
% Clay
I 65-74%
l< 65%
5 t.d::•-:•iii:;::..::..?
--'--"----:.
:::'
.F'•,0
..• 10-12.5%
<10%
1
10
o%[OandO
lO
%Sand
Runoff
Leakage
•
>0.01%
___l 0.3%
................
:•'
'*•'•••!
' •-'
''O.005-. 01%
3e
I 20%
12.5-20%
,c----•.•4?ji:.....¾---.':•'--•{•
74- 77%
•'• .........................
';:;/::
:•::::•5:.:•>•
o
90
OF SOIL
Unstressed Evapotranspiration
I
%Clay.? **
ROLE
I
I
0
le
*'•:...•i"-•*'•'•
:•0.1-0.3%
0.001-0.005%
<0.001%
o. o3-o. 1%
..
I
<0.03%
.'0
90
70
50
30
10
%Sand
Figure 6. Mean componentsof the water balancefor Prosopisglandulosaunder dry climaticconditionsat La
Copita ResearchArea for the USDA soil textural triangle. The partition of incomingrainfall among (a)
stressed
evapotranspiration,
{E,), (b) unstressed
evapotranspiration,
{E,•,), (c) runoff,{Q), and (d) leakage,
(L), is reportedhere. Canopyinterception(I), not shown,is assumedconstantwith respectto soiltextureat
0.2 cm. See Table 2 and Figure 5 for relevant parameters.
below
the level of soil moisture
associated
with
the onset of
water stress,i.e., s *. Thesestatistics,referred to asthe crossing
propertiesof the soil moistureprocess,are combinedwith the
expectedlevel of water stressexperiencedduring each excursion, i.e., the staticwater stress,to provide a new indicator of
the overall conditionof a plant under given edaphic and climatic characteristics
(seePorporatoet al. [2001]for a detailed
derivation). This new quantity is called the "dynamicwater
stress"of vegetation.This sectionuses the conceptsof soil
moisturecrossingpropertiesand water stress(both staticand
dynamic)to better understandthe role of soil texture on the
water deficitprocessfor the two vegetationtypesand climatic
conditionsat La Copita.
Figure 8 showsthe mean durationof an excursionbelows *,
Ts.; the mean number of crossingof s*, •.; and the mean
staticwater stress,•', for Prosopis
glandulosaunder dry climatic
conditionsfor all the soiltexturalclassesin the USDA triangle.
The maximumvalue of • is normalizedto 1, correspondingto
the case when the stressevent takes place at or below the
wiltingpointof the plant [Porporatoet al., 2001].Its calculation
includesthe parameterq, which is a measureof the nonlinear
impactsof the soil moistureon plant conditions.
Accordingto Figure 8a the Ts. valueswithin the soiltextural
triangle show a minimum for sandy soilswhile they show a
maximumfor the siltyloam soils.In contrast,Figures8b and 8c
indicate that both the •s* and •' values show the opposite
pattern,i.e., a minimumfor the siltyloam soilsand a maximum
for the sandiestsoils. These patterns appear to mimic the
variability of the quantity n(s* - Sw) among soil textures
(Figure 9). This quantityrepresentsthe layer of water per unit
_
_
_
depth neededto rise the relative soil moisturecontentof a soil
columnfrom sw to s*. For soiltextureswith smalln (s * - sw),
wilting and therefore water conservationoccurat soil moisture
levelscloseto the onsetof stresss*. As a result,any amountof
rainfall is likely to bringthe plant out of stress,diminishingthe
meandurationof the stressperiod,•'s*, but increasing
the
mean numberof crossings
of the stresslevel, •s*- Small n(s*
- Sw) valuesalsoresultin a higherprobabilityof excursionsof
the soil moisture trace to soil moisturevalues near Sw, which
areassociated
withhighervaluesof staticstress,
i.e.,higher•'
values.Conversely,soiltextureswith highn(s* - Sw) resultin
large•'s* andsmallhs. and•'.
The abovepatternsof •s*, h•., and•' withrespectto soil
texturedo not qualitativelychangewith vegetationtype and/or
climate in the caseof La Copita. Nevertheless,there are important quantitativechangesin thosemeasuresas the climate
and the vegetationtype change.For both the dry and the wet
climate, shallow-rootedspeciessuch as Paspaleumsetaceum
(not shown)spendon averagelesstime under stressand cross
the stresslevelmore often than do deep-rootedspeciessuchas
Prosopisglandulosa.Also, the grassestend to experience
higher •' valuesthan treesindependentlyof the wetnessof the
_
climate.A wetterclimateresults,
asexpected,
in lower½'valuesand lower•'s* values.Valuesof hs. increasewith the
wetnessof the climate for both species.
Valuesof thedynamical
waterstress,
3, areshownin Figure
10a and 10b for Prosopisglandulosaunder dry and wet climatic
conditionsat La Copita, respectively.They are normalizedto a
maximumof 1, and their calculationincludesthe parameter k,
representingthe indexof plant resistanceto water stress[Por-
2868
FERNANDEZ-ILLESCAS
ET AL.:
ROLE
OF SOIL
TEXTURE
P(s*)
a)
lO
> 0.95
% Silt
% Clay
0.915-.935
< 0.915
lO
9o
70
50
% Sand
10
30
1- P(sfc)
b)
90•.0
...
70.•
.-/:
---'ii
• '• ":"•.30
% Clay
..•i
..•.
% Silt
50 •...:•:•::•.::
......
-
..
..:,..:..
"}•.
30 ,:
:":;..•.:.70
•,.
,..•,,
.:
< 0.0003
-
....
9O
10'
..
90
70
50
% Sand
30
10
Figure 7. Prosopis
glandulosaunder dry climaticconditionsat La Copita ResearchArea. (a) Probabilityof
soilmoisture
tobebelows*, theonsetofwaterstress,
and(b) probability
of soilmoisture
tobeabove
Sfcfield
capacity.
SeeTable2 andFigure5 for relevantparameters.
poratoet al., 2001]. Under the dry climaticscenariothis vege-
significantloss mechanismin more arid climates, favoring
tationtypehasa betteroverall
condition
(i.e.,lower• values
) sandysoils that are capableof quickly draining water away
on a sandiersoilthan on a finer texturesoilsuchasa siltyloam
(Figure10a).Paspaleum
setaceum
alsoprefersa coarsersoilto
a finer-texturedsoil when the climateis dry (not shown).An
from the surfaceand avoidinghigh levelsof water lossdue to
direct soil evaporation.
Our vertically integrated model of soil moisture does not
increase in the wetness of the climate results in a shift of the
differentiatenear-surfacewater availablefor direct soil evaplow• regionto thefinertexturesoiltypesfor bothvegetation oration from water available for transpiration within Z rtypes
(ascanbeobserve
d in Figure10bforProsopis
glandu- Therefore it doesnot resolveall of the vertical processesusulosa), corrob0ratingthe inversetexture effect as describedby ally associatedwith the inverse texture effect. Instead, the
Noy-Meir[1973,p. 37]: "The samevegetationtypecanoccurat previouslyobservedeffect is, within our modelingframework,
lower rainfall on coarse soils than it does on fine ones."
drivenby the sensitivityof the overallconditionof the plant to
Previousstudies[Noy-Meir,1973]have attributedthe occur- the total amount of water that it needs to make the transition
fence of the inversetextureeffect to the impactof climateand from being at wilting to being abovestressand the manner in
soil textureon plant water availabilityand the manner in which which this sensitivitychangeswith climate. Dry climatesfavor
this impact changeswith the degreeof climate aridity. Water soil textural typeswhere a minimum amount of water is relossin humid climatesis greatlyinfluencedby leakagelosses, quired to raise soil moisture levels above s*, and thus even
favoringplant growth in soilswith high water-holdingcapaci- weak storm events are capable of taking vegetation out of
ties suchas silty loams. In contrast,surfaceevaporationis a water stress.Consequently,vegetationwill experiencelower
FERNANDEZ-ILLESCAS
ET AL.:
ROLE
OF SOIL
TEXTURE
2869
determinethe importanceof the mean duration,intensity,and
frequencyof water stressto the overall conditionof a plant as
Ts, (s*) (d)
expressed
bythedynamic
waterstress
•.
_____J
>100
"½:' "'i:'•
/'
% Glay
•,.•:.•..;::
75-100
"••••i•4
•½'--'•'•:'"'•'••••'
' '
% Silt
6.
50-75
< 50
,•
ß.........
':':!:•'"'"'""•,•..._•'..:•4•,•..,.....
..... .•.
90
70
......
50
30
% Sand
•0
Interaction
to Determine
of Climate
Coexistence
and Soil
Patterns
This section studies the role of soil texture in the coexistence
patterns of the two principal vegetationtypes present in La
Copita, under the rainfall variabilitycharacteristicof the area.
Figure 11 showsthe relationship between the total growing
seasonrainfall, t9, and the differencein dynamicalwater stress
between Paspaleum setaceum and Prosopis glandulosa,
AO•_•a, for six soil texturalclasses.
As before,the correspondingK•, xp•, b, and n have been estimatedthrough applicationof the univariateregressions
by Cosbyet al. [1984]to
the midpoint of the appropriatesoil textural regionwithin the
USDA textural triangle. To more accuratelydescribeconditions at this site, the soil hydraulic properties reported by
USDA [1979] were used to characterizethe predominantsoil
type at La Copita (sandyloam). The six textural classeswere
•(s*)
%Clay
• ....'"5:i:•::.:.
%Silt
5!?:"::
• 0
chosenbecausetheir calculatedAO•_•a valuesspanthose
< 1.5
90
70
50
30
% Sand
10
observedfor all 12 soil textural classesin the USDA triangle.
The total growing seasonrainfall is computedby linearly increasing,in a joint manner,the mean depth of rainfall events,
a, and the mean frequencyof arrival of storms,X, from their
dry to their wet period values.
Figure 11 showsa decreasingsensitivityof the differencein
theoverall
condition
of thetwospecies,
A•_•a, to total
rainfall as the climate gets wetter. Figure 11 also shows a
transition
inthesignof A•'•-•a withincreasing
growing
season rainfall for all six soil textures.For dry periods, grasses
C)
tendtodominate
(values
ofA•,•_•g < 0), whiletheopposite
is trueforwetperiods
(values
of A•_•g > 0). Thusthe
9!&10
I •>0.575
..:..;.•-:.•
-'•'•.'•..'....x.•.•.
4-->•.
-.. •:-•:•--•--•:
0
t .x9½•;
"-"•"'•½-•:•:•:•u
••
•;{:?•g.•½ ..*:•*"':
..........
0
O.
5-0.575interannual rainfall variability present at the site drives the
systemtoward different stateswhere soil texture plays a key
role in determiningthe dominantvegetationtype. Notice that
a sandysoil in this siteleadsto muchsmallerdifferencesin the
overall condition of the two speciesunder stronginterannual
..•_------%---..-..••½••.••••;•::...
rainfall variabilitythan thoseoccurring,for example,within a
silty clay loam. Moreover, it is clear from Figure 11 that most
90
70
50
30
10
of the impact of the interannualrainfall fluctuationstends to
% Sand
favor the C4 Paspaleumsetaceum.Wet periodsare more conFigure 8. Crossing properties and static water stress for ducive to Prosopisglandulosa,but the differencesin fitness
Prosopis
glandulosaunder d• climaticconditionsat La Copita between the two speciesare significantlysmaller than those
ResearchArea for the USDA soiltexturaltriangle.(a) Mean occurringin dry periodsin favor of the grasses.This suggests
durationof an excursionbelows*. (b) Mean numberof down- that the long-termrelativeincreaseof woodyplantsin this area
crossings
of s*. (c) Mean staticwater stressusingq = 3. The
durationof the growingseasonis T .... = 153 days.See Table cannotbe solelyattributedto pronouncedinterannualrainfall
variabilityexistingin the region.As important as thesefluctu2 and Figure 5 for other parameters.
ations are in driving the systemtoward different vegetation
states,a sizablelong-term increasein woody specieswill also
dependon the impact of grazingand fire control.
To further study the coexistencepatterns at La Copita, a
levels of dynamicwater stressfor coarse-texturedsoils with growingseasonrainfall equal to the 1950-1985 mean value of
small n(s* - Sw). In contrast,wetter climatesfavor fine- 40 cm was selected.The mean depth per rainfall event, a, and
texturedsoilswith largen(s* - Sw) so that if s* is reached the meanfrequencyof arrivalof storms,X, were then assigned
during a dryingperiod, the soil moisturetrace doesnot fall to from the previouslydescribedlinearlyvaryingvaluesto match
levelsassociatedwith large staticstressvalues.Thus the finer- the meanrainfall(i.e., a = 1.387cm and X = 0.188d-•,
textured soilshave the lowestdynamicwater stressvaluesun- respectively).
The corresponding
A•_•a valuesfor the
der wet climatic conditions.
USDA soil texture triangle are shown in Figure 12, which
In terms of the inverse texture effect, soil texture and climate showsthat the transition between grassand tree domination
• Clay
,:•;•--.,½,•
:::::::::::::::::::::
• S•lt
• ' ff:;;'•;'•½'•;:.:':"•"'
'
0
0.475-0.5
2870
FERNANDEZ-ILLESCAS
ET AL.: ROLE OF SOIL TEXTURE
I
% Silt
% Clay
I >0.08
!g•ii,.•--,
•---•-•,
,,-.:,•••
,•.,
ß
..........
•'•i'""•-•J'•;•;•;
O.07-0.08
0.05-0.07
< 0.05
10
9O
90
70
50
% Sand
30
Figure 9. Layer of water per unit depth neededto rise the relative soil moisturecontentof a soil column
fromSwto s* in the caseof Prosopis
glandulosa.
SeeTable2 andFigure5 for otherrelevantparameters.
can occurat this site at almostany soiltextureand evenmore
sowithinthe sandyloamregion,whichis the predominantsoil
moisturehavebeenstudiedin detailbyD'Odoricoetal. [2000].
The persistence
of prolongedperiodsof abnormally
wet and
in the area.
dryperiodscombinedwith the impactof grazingandfire conThe climaticvariabilityin thisregionand its impacton soil trol drivesthe systemtowardstatesof treeor grassdomination.
a)
90 •,,'::":
'.•..10
>0.7
70 '
'"':•::•'
% Silt
•
ß
..?:•?i":•"
.....
'"'•"'
'"'•...
50
% Clay
.:;½'-'::•:,•,•
.,...
0.625-0.7
0.55-0.625
3O
< 0.55
....
90
70
50
•
30
1¸
Sand
b)
*"&•'
90/?"-510
> 0.44
70
30
% Clay
% Silt
50
50
•
0.43-0.44
0.425-0.43
30
70
10
90
.,•
70
50
% Sand
30
<0.425
90
10
Figure10. Dynamic
waterstress
3forProsopis
glandulosa
ontheUSDAsoiltextural
triangle
under(a)dry
climaticconditions
and(b) wet climaticconditions.
T.... = 153days,q = 3, andk = 0.5. SeeTable2 and
Figure 5 for other parameters.
FERNANDEZ-ILLESCAS
ET AL.'
ROLE
OF SOIL
TEXTURE
2871
0.05
0.00
-0.05
-0.10
I
/
/
[]
-0.15
360.0
loam
Optimal
coexistence
-
-0.20
340.0
[]
380.0
400.0
420.0
440.0
O(mm)
Figure 11. Differencein the dynamicwater stressbetweenPaspaleum
setaceum
andProsopis
glandulosa,
AOes-ea,asa functionof thetotalincoming
rainfallduringthegrowingseason,
©. The parameters
a anda
vary linearlywith © increasingfrom their dry to wet values.T .... = 153 days,q = 3, and k = 0.5. See Table
2 and Figure 5 for other parameters.
Thesechangesin vegetationin the regionare well documented nentsof the water balanceshowhigh sensitivityto soil texture.
by Archeret al. [1988] and VanAuken [2000].
Soil texture alsohas a controllingeffect on the mean duration,
frequency,and intensityof water deficit periodsof the herbaceousand woody speciesat this site under dry and wet condi7.
Conclusions
Previous
tions.
studies have hinted
toward
the crucial role of soil
texture in the soil water balance and water stressprocesses
[Laio et al., 2001b].Our resultsfurthercorroboratethesefindingsby extendingtheir analysisto a broader range of vegetation, climate, and soil textures. For the two vegetation and
climaticconditionsat the La Copita site the coupledrelationship between soil moisture and vegetation is found to be
stronglydependent on soil texture. The probability density
function of soil moisture and the mean levels of the compo-
A relationship between soil texture and rainfall in accordance with the inverse texture effect [Noy-Meir, 1973] was
identifiedby Laio et al. [200lb] for a singlespeciesand here
proves to hold for the two coexistingspeciesat La Copita,
namely, the woody Prosopisglandulosaand the herbaceous
Paspaleumsetaceum.Resultssuggestthat the inversetexture
effect is driven by the sensitivityof the overall conditionof the
plant to the total amount of water it needsto make the transition from being at wilting to being above stressand the
10
•
71
% Clay
>0.01
% Silt
0-0.01
50
:•? -0.025-0
ß
< -0.025
.
9O
7O
.
5O
% Sand
3O
10
Figure 12. Differencein the dynamicwaterstressbetweenPaspaleum
setaceum
andProsopis
glandulosa,
AOps_pa
, foranaverage
climate,
a = 1.387
cmandX = 0.188d-], fortheUSDAsoiltextural
triangle.
Tseas
--153 days,q -- 3 and k -- 0.5. See Table 2 and Figure 5 for other parameters.
2872
FERNANDEZ-ILLESCAS
ET AL.: ROLE
OF SOIL
TEXTURE
water-controlledecosystems:
Active role in hydrologicalprocesses
manner in which this sensitivitychangeswith climate. The
andresponseto waterstress,II, Probabilisticsoilmoisturedynamics,
inversetextureeffect is observedwithoutneedingto explicitly
Adv. WaterResour.,24(7), 707-723, 2001a.
resolveall of the vertical processes
typicallyinvokedto de- Laio, F., A. Porporato, C. P. Fernandez-Illescas,and I. Rodriguezscribe it.
Iturbe, Plantsin water-controlledecosystems:
Active role in hydro-
logicprocesses
and responseto water stress,IV, Discussion
of real
It hasalsobeen shownthat soiltextureplaysa major role in
cases,Adv. WaterResour.,24(7), 745-762, 200lb.
the impact that interannualrainfall fluctuationshave on the
relative fitnessof woody and herbaceousspeciesin southern Newman, E. I., Responseof Aira precox to weather conditions,I,
Responseto droughtin spring,J. Ecol., 55, 539-556, 1967.
Texas,with the differencein fitnessbeing larger in favor of Noy-Meir, I., Desert ecosystems:
Environmentand producers,Annu.
grasses
duringdry periodsthanit is in favorof treesduringwet
Rev.Ecol. Syst.,4, 25-44, 1973.
years. For typical values of rainfall occurrenceand storm Porporato,A., F. Laio, L. Ridolfi, and I. Rodriguez-Iturbe,Plantsin
water-controlledecosystems:
Active role in hydrologicalprocesses
depths,Prosopisglandulosaand Paspaleumsetaceummay coandresponse
to waterstress,III, Vegetationwaterstress,
Adv. Water
existin a rangeof soil texturesat La Copita.
Resour.,24(7), 725-744, 2001.
References
Archer, S., C. Scifres,C. R. Bassham,and R. Maggio, Autogenic
successionin a subtropicalsavanna:Conversionof grasslandto
thorn woodland,Ecol. Monogr.,58, 90-102, 1988.
Cosby,B. J., G. M. Hornberger, R. B. Clapp, and T. R. Ginn, A
statisticalexplorationof the relationshipsof soil moisturecharacteristicsto the physicalpropertiesof soils,WaterResour.Res.,20(6),
682-690, 1984.
D'Odorico, P., A. Porporato, and I. Rodriguez-Iturbe,Preferential
statesof seasonalsoil moisture,WaterResour.Res.,36(8), 22092219, 2000.
Knoop, W. T., and B. H. Walker, Interactionsof woody and herbaceousvegetationin a southernAfrican savanna,J. Ecol., 73, 235253, 1985.
Laio, F., A. Porporato,L. Ridolfi, and I. Rodriguez-Iturbe,Plantsin
U.S. Department of Agriculture (USDA), Soil surveymanual, U.S.
Dep. Agric.Agric.Handb., 18, 503 pp., 1951.
U.S. Departmentof Agriculture(USDA), Soil surveyof Jim Wells
County, Texas, report, 111 pp., Soil Conserv.Serv., Washington,
D.C., 1979.
Van Auken, O. W., Shrub invasions of North American semiarid
grasslands,
Annu. Rev. Ecol. Syst.,31, 197-215, 2000.
C. P. Fernandez-Illescas
and I. Rodriguez-Iturbe,Department of
Civil andEnvironmentalEngineering,PrincetonUniversity,Princeton,
NJ 08544,USA. ([email protected];
[email protected])
F. Laio and A. Porporato, Dipartimento di Idraulica Trasporti e
InfrastruttureCivili, Politecnicodi Torino, CorsoDuca degliAbruzzi,
24, 1-10129Torino, Italy. ([email protected];[email protected])
(ReceivedNovember30, 2000; revisedJuly 19, 2001;
acceptedJuly23, 2001.)