Cracking of shrink–swell soils influences landscape hydrology. Watershed models that address soil crackingphenomena generally use a relationship between shrinkage and current soil water content to estimate theextent of cracking. Although antecedent soil moisture prior to soil shrinkage is found to affect the shrinkingof expansive soils in laboratory measurements, field observations are limited. In a previous study, a series ofin situ surface crack measurements over 10 years indicated the effect of soil moisture just prior to the start ofcracking (antecedent soil moisture) on cracking extent, but this relationship was not specifically analyzed.The objectives of this study were (i) to estimate the antecedent soil water content prior to cracking, (ii) toanalyze the effect of antecedent moisture on crack area density in microhighs and microlows, and (iii) toassess the temporal distribution of antecedent soil moisture in relation to an estimated water availabilityindex. Soil cracking was measured on a 10-m×10-m plot of Laewest clay (fine, smectitic, hyperthermic TypicHapludert) covered with native tallgrass vegetation on 42 dates during 1989–1998. Gravimetric soil watercontent was measured on 50 dates; 18 dates corresponded to crack measurements. Gilgai microtopograhywas mapped, and surface crack area density was calculated. For days when soil water content was notmeasured, it was estimated from precipitation and evapotranspiration. Antecedent soil water content priorto cracking was estimated for depth at 10 cm using daily estimates of soil water content and field notes oncracking. Results indicated that the temporal variation in surface crack area density of the study area during10 years was related to dynamics of current and antecedent soil water content on microhighs and microlows(R2=0.68 and 0.59, respectively). Prediction accuracies improved with classifying drying–wettingconditions during cracking. Dynamic temporal changes in the surface crack area density exhibiteddependence on a long-term (multi-year) cycle of antecedent soil water content superimposed by short-term(within a year) cycles of current soil water content.
Soil cracking, driven by drying of shrink–swell soils, altersinfiltration, runoff, evapotranspiration and redistribution of waterand chemicals. This phenomenon contributes to complex spatial andtemporal variability of water redistribution in the landscape, andcreates challenges to modeling of surface hydrology. Present modelsthat simulate water flow in shrink–swell soils use shrinkagecharacteristics based on current soil water content (e.g. Hendrickxand Flury, 2001; Greco, 2002; Deliberty and Legates, 2003; Arnoldet al., 2005; Cao et al., 2006; Bradley et al., 2007; Bedient et al., 2008;Lepore et al., 2009). Laboratory characterization of soils reinforces theidea that soil shrinkage, and therefore crack formation, is a function ofcurrent soil water content. Commonly used laboratory measurementsare the coefficient of linear extensibility (COLE), which relates soilshrinkage potential to soil water loss (Grossman et al., 1968; Reeve
et al., 1980; Yule and Ritchie, 1980a; Bronswijk, 1990b), and the soilshrinkage characteristic curve (e.g. Haines, 1923; McGarry andDaniells, 1987; Bronswijk, 1988, 1991; Mitchell, 1992; Coulombeet al., 1996; Olsen and Haugen, 1998; Braudeau et al., 1999; Boivinet al., 2006; Cornelis et al., 2006).
In most shrink–swell studies, soil water loss is calculated betweenthe current soil water content and a constant, such as the soil watercontent at field capacity (associated with a matric water potential of−33 kPa or −10 kPa) or at saturation. However, a variable initial soilwater content caused by soil water content history may affect themagnitude of shrinking and swelling of soils containing smectitic clayminerals (Yule and Ritchie, 1980a; Parker et al., 1982; Wilding andTessier, 1988; Tessier, 1990, Santamarina et al., 2001; Wells et al.,2003, 2007; Saiyouri et al., 2004). In this paper, we use the termsantecedent and current soil water content to identify soil moistureprior to surface desiccation cracking, when the soil surface is stillclosed, and at the time of crack measurement, respectively. The terminitial soil water content specifies any soil moisture condition at thebeginning of a measurement, with or without the presence ofdesiccation cracks.
Fig. 2. Crack area density measured on the surface of microhigh and microlowmicrotopography of Laewest clay on 42 dates over the 10-yr study period. Notedifferent scales on each y-axis.
110 A.S. Kishné et al. / Geoderma 157 (2010) 109–117
The complex processes of shrinking and swelling, and associatedcrack opening and closing in Vertisols and vertic intergrades depend onthe microstructure and intra- and interparticle porosity of soils(Wilding and Tessier, 1988; Tessier, 1990; Quirk, 1994; Coulombeet al., 1996). Available water is one of the external factors driving thebalance of attractive and repulsive interparticle forces. These attractive(capillary suction, London–van denWaals and ion–ion correlation) andrepulsive (structural water and osmotic) forces depend internally onmineralogy, cations, electrolyte concentration, and organic matter, butalso externally on varying climate, parent material, topography, landuse, vegetation, and stress history (Quirk, 1994). The effect of stresshistory on soil cracking is most noticeable from extreme wetting anddrying or compaction (Santamarina et al., 2001; Saiyouri et al., 2004).Considering the effect of the history of drying–wetting on soil shrink–swell, it was observed in electron microscopic and X-ray scatteringstudies that initial soil water content played a significant role in shrink–swell processes of smectitic clay samples (Tessier, 1984 reviewed byWilding and Tessier, 1988; Tessier, 1990). On core samples taken fromVertisols, initial moisture prior to desiccation has been also shown toaffect the final magnitude of shrinkage (Yule and Ritchie, 1980a).
In addition to high initial moisture, repeated dry–wet cycles haveenhanced swelling in small soil samples (Parker et al., 1982; Penget al., 2007). Additionally, in a large repacked sample of a smectiticVertisol (76.5-cm×80-cm×30-cm), a sequence of simulated wettingand drying increased swelling and vertical crack depth duringsubsequent wetting and drying cycles (Wells et al., 2003, 2007;Römkens and Prasad, 2006). It was postulated that the spatial andtemporal variability of antecedent soil moisture caused the increase inswelling and cracking during the repeated wetting–drying cycles(Wells, personal communication, 2008).
In a previous study conducted by Kishné et al. (2009) based on a10-yr crack monitoring of Vertisols, findings indicated the effect ofantecedent soil moisture on cracking to cause variation in shrinkageand retained soil water content due to hysteresis. Soil cracks weremapped in field conditions (Fig. 1) on a 100-m2 site of smectiticLaewest clay with gilgai in native grassland in the Texas Gulf CoastPrairie on 42 dates during 1989–1998 (Miller et al., 2010). Thegreatest extent of cracking was measured in 1997 and 1998 (Fig. 2)preceded by months of above normal precipitation. These dates in1997 and 1998 had about 10-fold greater crack area density thandates with similar soil water content but below normal precipitationin the prior months, i.e. 1995 and 1996. Temporal trends in the extentof cracking, expressed as crack area density, were similar on microlowand microhigh gilgai categories (Fig. 2). However, surface crackingoccurred with a much smaller extent in microlows despite of greatershrinkage potential measured as COLE to 1 m depth (Kishné et al.,2009). Current gravimetric soil water content proved to have anoverall weak, negative relationship with surface crack area density.The weak relationship between measured gravimetric soil watercontent and crack area density was somewhat improved byseparating microhighs andmicrolows and by grouping data according
Fig. 1. Field photos demonstrating a) the measurement grid with 10-cm×10-cm cells used
to drying, uniform and wet soil moisture conditions within 10–25 cm.A probable influence of antecedent soil moisture was hypothesized,but not tested because antecedent soil water content prior to crackingwas not the objective of that study.
In the current investigation, we study the temporal dynamics insurface cracking at the previously investigated Vertisol site (i.e.Kishné et al., 2009; Miller et al., 2010). Particularly, we focus analysison the relationship of antecedent soil moisture prior to cracking andlong-term weather variations with cracking densities measured over10 years, in this 10-m by 10-m area of a Vertisol. The specific
to map cracks and b) surface cracking with a 30-cm long ruler on September 3, 1993.
111A.S. Kishné et al. / Geoderma 157 (2010) 109–117
objectives of this study were (i) to estimate the antecedent soil watercontent prior to cracking, (ii) to analyze the antecedent moistureeffect on crack area density on microhighs and microlows, and (iii) toassess the temporal distribution of antecedent soil moisture inrelation to an estimated water availability index.
2. Materials and methods
2.1. Field measurements
A detailed description of the site and data collection was presentedin Kishné et al. 2009 and Miller et al., 2010; nevertheless, we give abrief summary of data regarding this study specifically. A 10-m by 10-m study plot in Victoria County, Texas (28° 39′ 46″ N, 96° 46′ 20″ W),was investigated in 1989–1998. The soil was Laewest clay (fine,smectitic, hyperthermic Typic Hapludert, Soil Survey Staff, 1999) withnative tallgrass prairie vegetation, and circular gilgai microtopogra-phy (Miller et al., 2010). The study area contained distinct microhighmounds and ridges (38%), microlows (19%), and the rest (43%) wasconsidered microslopes (Kishné et al., 2009). The elevation rangedfrom 12.20 to 12.45 mwith amean andmedian of 12.29 m.Microlowswere 2–3 m across and about 6 to 7 m apart. Wilding et al. (1990)reported on the original scope of the study, how the site was selected,characterization of the region, and the botanical survey of the nativeprairie vegetation. Soil morphology was described in detail for themicrohigh and the microlow to depths of 5.9 m and 6.1 m,respectively, close to the study site (Soil Survey Staff, 1990). Thedepth of the soil solum is 3.80 m in microhighs and 3.55 m inmicrolows with wavy boundaries. A permanent groundwater tablewas apparent at 5.1 m in 1988 (Soil Survey Staff, 1990) and wasmostly below 3 m, as indicated by piezometer measurements.However, episaturation, a perched zone of free water, was observedoccasionally in piezometers installed at depths of 0.25, 0.5, 1 and 2 min a field adjacent to the crack study site.
On 42 occasions from June 1989 to July 1998, length and width ofall soil cracks weremeasured with 5 and 0.5 cm accuracy, respectively(Fig. 1, Table 1). The smallest and largest width of measured cracksegments on both microhighs and microlows were less than 0.5 cmand 7 cm, respectively (Kishné et al., 2009). Crack locations wereplotted on engineer graphing paper at a 0.0254 to 1.0-m scale, andsurface crack width data was categorized with limits of 0.5, 1, 2, 5, and7 cm. The crack diagrams were scanned with a 157 pixel cm−1
Table 1Summary statistics for daily mean of crack widths, daily sum of crack lengths, and crackarea density (AD) along with gravimetric soil water content (W) measured on the soilsurface and at a depth of 10 cm, respectively. The total number of measurement dayswas 42 for cracks and 50 for soil water. A subset (n=18) for dates with crackmeasurements, with simultaneously measured soil water (Wc) and estimatedantecedent soil water (Wa) are shown.
resolution, rectified, digitized in ArcView 3.0, and analyzed in ArcGIS9.0 (Environmental Systems Research Institute, 2005). Polylines,defined as crack segments between junctions or end points, and/orchanging crack widths, were clipped according to microhigh, micro-slope, andmicrolow categories. The data are available online at http://soilcrop.tamu.edu/research/pedology/crack.html.
Crack area was estimated for each crack segment by multiplyingthe length of digitized crack segment by the mid-value of widthcategory measured in the field. To assess surface crack area density(surface crack extent), total crack area obtained for a gilgai elementwas normalized by the area it occupied in the study site. Soil watercontent was measured gravimetrically in 2–3 replicates with coresample thickness of 3–5 cm centered at 10, 25, 50, 75 and 100 cmdepths on microhighs and microlows adjacent to the study site on 50occasions (Table 1). On 18 and 15 out of 50 dates, soil moisturesamples were collected on microhighs and microlows, respectively,within 1–2 days of crack measurements with no or only trace amountof precipitation (Table 1). Field observations on cracking-relatedconditions were recorded about every 2 weeks on 181 days over the10-year period.
2.2. Water availability index
Daily minimum and maximum temperature, wind speed, relativehumidity, and precipitation along with hourly solar radiation weremeasured at a weather station at Victoria Regional Airport, Texas,16 km north of the site. Four months (January 1, 1996–April 30, 1996)of missing precipitation data were filled-in using daily precipitationdata collected in Port Lavaca, Texas, about 20 km south of themeasurement site. Cumulative precipitation measurements werecollected at the site about every two weeks. This cumulative raindata was compared to the daily data collected at the weather stationsto check for local anomalies in amounts of rainfall. If the differencebetween precipitation measured at the study site and the weatherstation was greater than 2 mm day−1, then the daily precipitationwas adjusted to match the site precipitation amount. The precipita-tion adjustment was distributed proportionally on the days when rainwas indicated at the weather station.
Daily reference evapotranspiration (ET0) was estimated using theFAO version of the Penman–Montieth equation (Allen et al., 1998),following calculations summarized by Ham (2005). To evaluate therelationship between surface cracking and trends in available water, asimple water availability index (WAI) was calculated,
WAIi = WAIi−1 + nPCPTi−nET0i ð1Þ
where i is a daily index; nPCPT is normalized precipitation; and nET0 isnormalized reference evapotranspiration (Heinsch et al., 2004).Normalization was achieved by dividing the daily totals of precipita-tion and reference evapotranspiration by their 13-year average ofannual totals. The daily WAI was initialized on January 1, 1986. Thefirst crack measurement began on June 7, 1989.
2.3. Estimation of daily soil water content, antecedent soil water, anddata analysis
To analyze the temporal change in soil cracking in relation to priorweather conditions, first, current and antecedent soil water wereestimated for days with no soil water data; second, multivariateanalysis was applied including both current and antecedent soilwater; and finally, the antecedent soil water was investigated incomparison to temporal fluctuation of available water.
Precipitation and evapotranspiration were used to predict soilwater content because they are the main driving factors of soilmoisture dynamics (Delworth and Manabe, 1988; Deliberty andLegates, 2003). Additionally, Doris et al. (2008) reported that time
Fig. 3. Soil water content estimated using cumulative precipitation and referenceevapotranspiration plotted against measured soil water content (n=50) onmicrohighsand microlows.
112 A.S. Kishné et al. / Geoderma 157 (2010) 109–117
history of precipitation and temperature up to 8 weeks wassignificantly related to soil moisture controlling vertical shrink–swell of Vertisols under a concrete block that simulated overburdenpressure in Texas and in Australia. In our study, we developed twomultiple regression equations to estimate daily gravimetric soil watercontent using cumulative precipitation and evapotranspiration and 50measurements of gravimetric soil water content at a depth of 10 cm.Two regression equations were calculated because soil moisture inmicrohighs and microlows were different on a given day. The soilwater content at 10-cm depth was used because the 10-cm depthappeared to be closely related to surface cracking (Kishné et al., 2009)and was strongly affected by weather variability.
To develop these empirical soil moisture equations, variouscumulative time intervals of precipitation and evaporation weretested, and the optimal intervals were chosen based on the best p-values, R2 values, and root mean square deviation (RMSD). Modelresiduals were checked for temporal trends by plotting the residualswith time. Direct effect of runoff and runon on soil water was ignored,particularly because we wanted the soil water estimate to beindependent of the soil cracking data. A possible influence ofoccasional episaturation during prolonged wet periods was assumedto be intrinsically included in the empirical equations through soilmoisture and precipitation measurements.
Antecedent soil water content was estimated for all crackmeasurements using the empirical soil water content model, exceptfor one date when soil moisture wasmeasured. Choosing the value forantecedent soil water content was started by selecting the soil watercontent at the local soil moisture maximum prior to a date of a crackmeasurement. This local maximum was found by using a 4-weekmoving window with daily time steps. Then, the selected soilmoisture was compared to soil moisture values closer, in time, tothe crack measurement until the lowest peak soil water value wasidentified to correspond with observations of closed cracks at the soilsurface.
In this simple approach for estimating antecedent water content,we did not consider that cracks might have been below the soilsurface due to swelling of the surface upon rewetting, nor the unevenmoisture distribution related to preferential wetting or drying aroundmacropores, although these conditions can contribute significantly tobypass flow and uneven distribution of soil water content (Bouma andLoveday, 1988). Multiple linear regression, simple correlation, andtemporal autocorrelation analyses were performed using S-Plus 7.0(Insightful Corp., 2005).
To consider the effect of soil moisture hysteresis on the surfacecracking as a function of soil moisture, we assessed drying–wettingconditions based on measured moisture difference between depths of10 and 25 cm (Kishné et al., 2009). The moisture differences werecompared to what we considered natural measurement variation. Thevariation criterion was estimated as the sum of standard errors ofreplicate soil water content averaged for 50 dates in the two depths.When the moisture difference between 10 and 25 cm exceeded thecriterion, positively or negatively, then a wetting or drying conditionoccurred, respectively. Otherwise, soil moisture was considereduniform in the layer of 10–25 cm.
3. Results and discussion
3.1. Antecedent soil water content
Soil water conditions just before opening of cracks were assessedbased on estimates of daily soil water content and the field notesindicating probably closed surface crack conditions. Table 1 presentsthe summary statistics of measured and estimated soil watercontents. In general, the microlows were wetter than microhighs.
In this study, a 16-week history of rainfall and evapotranspirationwas found sufficient to predict daily soil water content. Based on 50
near-surface measurements of soil water content, the best regres-sion models for microhighs (MH) and microlows (ML) were asfollows:
WMH; i = 0:001PCPTw1 + 0:0008PCPTw2 + 0:0002PCPTw3−16
− 0:0009ETw1−7− 0:0005ETw8−16 + 0:4168;
ð2Þ
WML; i = 0:0012PCPTw1 + 0:0008PCPTw2 + 0:0003PCPTw3−16
−0:0011ETw1−7−0:0004ETw8−16 + 0:4321;
ð3Þ
where W is the gravimetric soil water content (kg kg−1) at a depthof 10 cm, PCPTw1, PCPTw2 and PCPTw3–16, are cumulative precipi-tation (mm) of the first, second, and 3rd to 16th week intervals, andETw1–7 and ETw8–16 are the reference evapotranspiration (mm)from the first seven, and 8th to 16th week periods prior to day i,respectively.
Overall, the prediction models of daily soil water contentexplained 79% of the variation in soil water content in microhighs,and 73% in microlows (Fig. 3). Model residuals were normal,homoscedastic, and not autocorrelated. The model error(RMSD=0.04 kg kg−1) exceeded estimated measurement variabilityof soil moisture by 0.015 kg kg−1.
The estimated antecedent soil water content ranged from 0.26 to0.42 kg kg−1 in microhighs, and from 0.29 to 0.48 kg kg−1 inmicrolows (Table 1). During the longest drying period in 1998,7 months separated the antecedent moisture and a crack measure-ment. It is interesting to note that the antecedent soil water contentprior to crack opening was not a fixed value, but it varied by ±25%from the average over time. Moreover, the greatest antecedent soilwater contents exceeded field capacity values of 0.33 kg kg−1 inmicrohighs and 0.39 kg kg−1 in microlows. The field capacity valuesweremeasured at soil matric water potential of−33 kPa, on soil clodsin laboratory (Soil Survey Staff, 1990). The range in antecedentmoisture supports the hypothesis presented by Kutilek and Germann(2009) that the threshold between macropores and micropores istime-variant, and has a range of transitions rather than a well-definedboundary that may depend on the history of variations in watercontent.
The variability of soil moisture just before the start of crackformation has been reported in the literature. No or slight changes in
Fig. 4. Crack area density plotted against measured and estimated current soil watercontents at a depth of 10 cm. Drying, uniform, and wetting conditions were determinedfrom measured water content profile data.
113A.S. Kishné et al. / Geoderma 157 (2010) 109–117
soil volume have been observed with changes in soil water contentduring the structural shrinkage phase (e.g. Stirk, 1954; Yule andRitchie, 1980a; Bronswijk, 1991; Coulombe et al., 1996; Coquet et al.,1998; Olsen and Haugen, 1998; Cornelis et al., 2006). In the basicshrinkage phase, where the change of soil volume is proportional tothe volume of water change (Mitchell, 1992), changes in soil volumewith water loss were observed but soil crack formation was notconsistently observed. These volume changes have been quantified bymeasuring one-dimensional (vertical) shrink–swell (Favre et al.,1997) and by measuring bulk or wet density of soils (Fox, 1964;Grossman and Reinsch, 2002; Bernard et al., 2006; Dudoignon et al.,2007). But no one-dimensional decrease in soil volume has beenfound predominantly as a function of water loss by Aitchison andHolmes (1953), Hallaire (1984) and Bronswijk (1990a). Our study,however, did not address the question of whether soils in the fieldhave a one-dimensional (vertical) shrinkage before starting three-dimensional shrinkage (cracking), because no vertical soil movementwas measured.
3.2. Relationship of surface crack extent and soil water content
Fig. 4 presents the distribution of crack area density versus currentsoil water content at a depth of 10 cm on microhighs and microlows.According to analysis of soil water conditions, there was only onecrack measurement taken on a distinct wetting event on October 20,1992. On that day, crack area density was small on the microhighs(4.8×10−4 m2 m−2), which is expected with high soil water content,and there was no cracking in the microlows. Since this date is the onlycrackmeasurement under a clear wetting process, the further analysisreflects trends based only on dates with measurements underdesiccating conditions, 41 dates on microhighs and 36 dates onmicrolows. Out of 18 dates with measured cracking and soil watercontent, there were three dates when cracks were only in microhighs.
Fig. 4 shows the well-known trend of decreasing cracking withincreasing soil water content. However, in the dry soil water range,between 0.16 and 0.23 kg kg−1, the spread of crack area density waslarge, up to 100 fold. Linear regression models of log-transformedcrack area density as a function of current moisture content poorlydescribed the process for both microhighs and microlows (Models 1aand 1b in Table 2). Stratifying crack area density according to dryingand uniform soil water conditions improved themodels for the dryingcategory, but less or not for uniform conditions (Models 2a, 2b and 3a,3b in Table 2).
An increasing trend of the crack area density with increasingantecedent soil water content was observed in microhighs and
Table 2Regression model results for log-transformed crack area density (m2 m−2) as a function of cuat a depth of 10 cm. For uniform microlows, no correlation was found.
Model Data set N Coefficients for
Intercept
MicrohighsModel 1a* Whole 41 −3.51**
Model 2a** Drying† 10 5.32*
Model 3a Uniform† 7 6.59Model 4a*** Whole 41 −9.62***
† Grouping was based only on measured data which had stratified soil water measurements;
microlows (Fig. 5). In other words, desiccation starting from wetterantecedent soil water content resulted in greater surface cracking.Including antecedent soil water content into the models of log-transformed crack area density resulted in stronger models with R2
values improved to 0.68 and 0.59 for microhighs and microlows,
rrent soil water content (Wc, kg kg−1) and antecedent soil water content (Wa, kg kg−1)
***, **, and * represent significance at p-values less than 0.001, 0.01, and 0.1, respectively.
Fig. 5. Log-transformed crack area density versus estimated antecedent soil watercontent on microhighs and microlows. Crack area densities with same antecedentconditions, but different current soil water contents were averaged. Standarddeviations associated with averaging are expressed with error bars.
Fig. 7. Crack area density versus estimated soil water loss grouped according toantecedent soil water content on microhighs and microlows. Note different scales oneach y-axis.
114 A.S. Kishné et al. / Geoderma 157 (2010) 109–117
respectively, and more significant p-values for model coefficients(Models 4a and 4b in Table 2). Dividing the crack measurements intodrying and uniform moisture conditions improved the multivariatemodels even more based on R2 and RMSD values; however, the p-values were smaller compared to the whole models (Models 5a, 5band 6a, 6b in Table 2).
As a graphical illustration, Fig. 6 demonstrates the multivariatemodels for the whole data set of microhighs and microlows (Models4a and 4b in Table 2). Soil water loss calculated between varyingantecedent and corresponding current soil water content correlatedlinearly with natural log-transformed crack area density, thusexponentially with crack area density. Fig. 7 shows crack datapartitioned into three and four groups of antecedent soil watercontent for microhighs and microlows, respectively. Groupings werechosen based on breaks in the available antecedent water contentdata. In the first three groups, the group limits are identical for
Fig. 6. Log-transformed crack area density versus estimated soil water loss (differencebetween antecedent and current soil water contents) on microhighs and microlows.
microhighs and microlows, but there is an additional group with evengreater antecedent soil water content for microlows. Within eachgroup, the fitted lines represent only trends in surface crack extentand water loss to compare the rates of shrinkage (slope of fitted line).Overall, the rate of shrinkage and the range of water loss increasedwith increasing antecedent moisture. Shrinkage rates were verysimilar in the low antecedent moisture groups compared withinmicrotopography categories. The largest shrinkage rate associatedwith the largest antecedent water content group was outstanding inboth microhighs and microlows.
Two possible scenarios might describe the desiccation paths. Onone hand, shrinkage may have started with a low rate similar toshrinkage rates exhibited in groups 1 and 2 then followed anincreasing shrinkage rate of the line fitted to all measured data inthe group of largest antecedent soil moisture. The phase with smallshrinkage rate at the beginning of desiccation may correspond to thecurvilinear phase between the maximum swelling and the macro-porosity limit in the structural phase identified in well structured soilcores, and was demonstrated by several shrinkage curve studies(Braudeau et al., 2004; Boivin et al., 2004; Boivin, 2007; Milleret et al.,2009). On other hand, shrinkage in 1993, 1997 and 1998 may haveindividual linear shrinkage paths. Although there were no observa-tions to confirm individual shrinkage paths, different paths might besuggested by the large variation in crack extent at large water losses.On microhighs, at water loss of 0.15–0.20 kg kg−1 and 0.20–0.25 kg kg−1, there were about 2- and 3-fold differences in crack
Fig. 8. Temporal distribution of the daily water availability index (WAI) and antecedentsoil water content (Wa). Peak WAI values prior to the crack measurements are markedby triangles.
115A.S. Kishné et al. / Geoderma 157 (2010) 109–117
area density. On microlows, there was a similar difference in crackarea density between water loss of 0.25 and 0.30 kg kg−1.
In the group of lowest antecedent soil water content of 0.25–0.3 kg kg−1, the very small cracking and near zero slope of the fittedline indicated a short basic shrinkage phase, and shrinkage startingpossibly from soil moisture close to the shrinkage limit. The within-group variation of current soil water content and crack area densitywas likely influenced by hysteresis, as suggested by Kishné et al.(2009) and some errors in estimation of soil water content. The effectof antecedent moisture on crack length was more distinguishablethan on crack width. Under conditions of large water loss, the dailysum of crack length was higher in both microhighs and microlows(data not shown).
In situ observations indirectly relating soil cracking to antecedentsoil water content have been reported in the literature (Lin et al.,1998; Das Gupta et al., 2006). In a study on steady-state infiltration ofa Ships clay (very-fine, mixed, active, thermic chromic Hapludert), Linet al. (1998) showed a greater than 10% increase in apparent steady-state infiltration rate after a week-long tropical rainfall eventcompared to prior measurements at the same soil water content.This increase in infiltration may have been related to changes inmacroporosity, including cracks, and hence to antecedent moistureeffects on cracking in the soil surface (Lin, personal communication,2008). In another study conducted also on a site of Ships clay in Texasby Das Gupta et al. (2006), hydraulic conductivity measured at zeromatric potential was eight times greater following a rainy fall andwinter in 2003 and a seasonal drying period than earlier at about thesame initial soil water content.
The effect of antecedent soil water content on the relationship ofsoil water and soil cracking seems to be twofold. The first effect,though potentially minor, is the effect on hysteresis. When drying ofsoil starts from greater antecedent soil water content the soil will havemore retained water at the same matric potential than when a soilstarts drying from smaller antecedent soil water content (Croney andColeman, 1954; Lal and Shukla, 2004, p.437). This hysteretic waterretention is expected to influencemainly the capillary flow, and not ornegligibly the water flow in shrinkage cracks and permanentmacropores (Chertkov and Ravina, 2002; Greco, 2002; Briaud et al.,2003; Gerke, 2006). In microlows, the generally greater antecedentand current soil water contentsmay also be related to greater retainedsoil water due to hysteresis frommore distinct structure in addition tothe lower topographic position of microlows. This hysteresis effectmay be strongest near the soil surface and decreases with depth(Mitchell and Mayer, 1998).
The second effect of antecedent soil water content is on therelationship of soil cracking and water loss. In soil shrinkage curves,water lossmeasured fromwetter initial conditions have been shown andmodeled toproduceproportionally greater shrinkage at the samecurrentsoil water content during basic shrinkage (e.g. Yule and Ritchie, 1980a,b;Wilding and Tessier, 1988; Tessier, 1990; Chertkov and Ravina, 1998. Atmicroscopic scale, this effect is attributed mainly to change of watercontent and energy level in the interparticle pores and in the internalstructure of clay fabric (Wilding and Tessier, 1988; Tessier, 1990;Coulombe et al., 1996; McGarry and Yule, 2006). In drying calcareoussmectitic Vertisols, thewalls in claymicrostructure fold on each other, asan accordion— an analogy used byWilding and Tessier (1988, p. 76). Thedegree of shrinkage depends on the magnitude of previous swellingrelated to the initial moisture condition and the thickness andcomposition of microstructure. Great intensity of drought causesincreased number of layers comprising substacks in themicrostructures,thus the interlayer spacing decreases, and so the overall shrinkageincreases (Tessier, 1990). Thus, the collapse of microstructure andintraparticle porosity of Ca-smectite in the soil depends on initial (orantecedent) soilwater content, aswell as on layer charge, layerflexibilityandextensibility of overlapping layers. Intensive, prolongedprecipitationwith repeated wetting–drying cycles, and consequent high antecedent
soil moisture may promote this process. We speculate that the varyingantecedent soil water content might relate to slowly changing particlegeometry. The microporosity reached at the wet end of basic shrinkagephase (macroporosity limit) and swelling limit may increase or decreasedue to rearrangement of particles forming interparticle pores.
In this field study, the positive correlation of crack area density tothe antecedent soil water content was significant (Figs. 5 and 6). Thecomplex Vertisol cracking conditions observed probably exhibitedboth types of antecedent soil water content effect on soil shrinkage.Furthermore, an even earlier history of wetting–drying may haveimpacted the measured soil shrinking behavior in fieldconditions. Peng et al. (2007) found on small repacked clayey soilsamples that the maximum intensity of previous wetting–dryingcycles influenced significantly the magnitude of pore shrinkage.Besides, a possible memory mechanism of soils (Santamarina et al.,2001, p. 142) related to soil shrinkage phenomena and repeatedwetting–drying cycles cannot be ruled out.
The influence of antecedent soil water content on soil shrinkagemay not be limited to the soil surface. It may also affect shrinking–swelling of deeper soil subhorizons, in addition to current soil watercontent, although overburden confining pressure may limit itscontribution. Consequently, this may be one of the influentialvariables in varying the shrinkage ratio between height change andwater storage change from one soil layer to another (Kirby et al.2003). It may also explain some of the differences in temporal andspatial variation of shrinking–swelling soils on different landscapepositions (Baer and Anderson, 1997).
3.3. Analysis of antecedent soil water content and weather variation
Strikingly, long, generally drying periods in 1988–1990 shown bythe decreasing trend in the water availability index (Fig. 8) wereassociated with small cracking, and relatively low antecedent soilwater contents (0.29–0.32 kg kg−1). On the other hand, extremelywet periods in 1997 and 1998 preceding extensive droughts wereassociated with large crack area density and relatively largeantecedent soil water contents, 0.39 and 0.37 kg kg−1 on microhighsand 0.46 and 0.48 kg kg−1 on microlows, respectively. The varying
116 A.S. Kishné et al. / Geoderma 157 (2010) 109–117
antecedent soil water content seemed to follow the long-termfluctuation of available water content because the latter determinesthe degree of saturation; thus there is a similar temporal distributionof peaks of water availability index and antecedent soil water content(Fig. 8).
As a consequence of this trend, dynamic temporal changes in thesurface crack area density developed in microhighs and microlowsseemed to follow a long-term cycle (approx. 6 years) of antecedentsoil water content superimposed by short-term (within-year) cyclesof cracking, related to current soil water content. Because surfacecrack area density is related to crack volume (Peng and Horn, 2007),we anticipate similar trends in the effect of antecedent soil moistureon crack volume of Vertisols. Because of lowwater permeability of thesoil matrix, the bulk clayey soils may swell or shrink on a very slowrate under cyclic moisture conditions; therefore, reaching their newmoisture equilibrium conditions may take years (Lal and Shukla,2004). As noted in Section 3.2, this slow hydration–dehydration ofbulk soil matrix probably drives the change in antecedent soil waterconditions while the short-term crack opening and closing is likelytriggered by more frequently changing seasonal wetting–dryingconditions. Under monotonic conditions of available water, such ashomogeneous climate or regular irrigation without long-termfluctuations, the effect of antecedent soil moisture would be minimal.
To overcome the limitations of our study, which were theestimation of antecedent and current soil water content, and onlyone measurement site, further field studies for monitoring long-termvertical and horizontal shrink–swell processes are needed. To developa more coherent understanding of the phenomenon observed, furtherinvestigations of temporal and spatial dynamics of shrink–swellmechanisms related particularly to antecedent soil water content andprior wetting–drying cycles would be desirable in various Vertisolsunder different field and climatic conditions.
4. Conclusions
In this study of surface cracking of a Laewest clay with gilgai, wehave analyzed the temporal variation of field-measured crack areadensity of Laewest clay and estimated antecedent soil water content,along with a calculated water availability index. The main results areas follows:
(i) Antecedent soil water content was assessed using daily soilwater content estimated from cumulative precipitation andreference evapotranspiration covering a total of 16 weeks, andavailable field observations on cracking-related conditions.
(ii) Variation in the surface crack extent in microhighs andmicrolows depended not only on current soil water contentnegatively, but more strongly on antecedent water conditionspositively. Including antecedent soil water content in regres-sion models with current soil water content improvedprediction of crack area density significantly on microhighs(R2=0.68) and microlows (R2=0.59). Accounting for soilwater hysteresis (drying or uniform soil water conditions)improved the multivariate regression models with R-squaredup to 0.95.
(iii) Shrinkage rate in basic shrinkage phase might be a variable ofantecedent soil water content in Laewest clay.
(iv) Antecedent soil water content just prior to cracking inmicrohighs and microlows showed a long-term (multi-year)fluctuation.
More generally, short-term oscillation of daily current soil waterseemed to trigger relatively short desiccation cycles which weresuperimposed on a long-term multi-year cycle of antecedent soilwater. This varying antecedent soil water content may depend onslowly changing particle geometry due to repeated rewetting anddrying conditions in the bulk smectitic soil matrix. Using antecedent
soil water content to account for the long-term variability in availablewater conditions might enhance modeling crack extent in expansivesoils—especially in climates with oscillating wetting and dryingcycles. To achieve this, the mechanism of antecedent soil moistureaffecting soil cracking needs to be further investigated in carefullydesigned field experiments to validate the results, to furtherfundamental understanding and also to explore potential improve-ments for modeling of water and chemical movement in shrinking–swelling soils at pedon and landscape scales.
Acknowledgements
This research was funded by a cooperative agreement betweenTexas AgriLife Research, and the Texas USDA-NRCS. The authors arethankful to Drs. K. McInnes and J. Heilman for comments on thecalculation of water availability index and the manuscript.
References
Aitchison, G.D., Holmes, J.W., 1953. Aspects of swelling in the soil profile. Aust. J. Appl.Sci. 4, 244–259.
Allen, R.G., Pereia, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration: guidelines forcomputing crop water requirements. Irrig. Drain. FAO, Rome, p. 56.
Arnold, J.G., Potter, K.N., King, K.W., Allen, P.M., 2005. Estimation of soil cracking and theeffect on the surface runoff in a Texas Blackland Prairie watershed. Hydrol. Process.19, 589–603.
Baer, J.U., Anderson, S.H., 1997. Landscape effects on desiccation cracking in an Aqualf.Soil Sci. Soc. Am. J. 61, 1497–1502.
Bernard, M., Dudoignon, P., Chevallier, C., Pons, Y., 2006. Shrinkage and subsidence in amarsh soil: measurements and preliminary model. Soil Sci. Soc. Am. J. 70,1834–1842.
Boivin, P., 2007. Anisotropy, cracking, and shrinkage of vertisol samples. Experimentalstudy and shrinkage modeling. Geoderma 138, 25–38.
Boivin, P., Gardner, P., Vauclin, M., 2006. Modeling the soil shrinkage and waterretention curves with the same equations. Soil Sci. Soc. Am. J. 70, 1082–1093.
Boivin, P., Garnier, P., Tessier, D., 2004. Relationship between clay content, clay type,and shrinkage properties of soil samples. Soil Sci. Soc. Am. J. 68, 1145–1153.
Bouma, J., Loveday, J., 1988. Characterizing soil water regimes in swelling clay soils. In:Wilding, L.P., Puentes, R. (Eds.), Vertisols: Their Distribution, Properties, Classifi-cation and Management. Texas A&M Univ. Printing Center, College Station, TX, pp.83–96.
Bradley, C., Mosugu, M., Gerrard, J., 2007. Seasonal dynamics of soil–water pressure in acracking clay soil. Catena 69, 253–263.
Braudeau, E., Constatine, J.M., Bellier, G., Colleuille, H., 1999. New device andmethod forsoil shrinkage curve measurement and characterization. Soil Sci. Soc. Am. J. 63,525–535.
Braudeau, E., Frangi, J.-P., Mohtar, R.H., 2004. Characterizing nonrigid aggregated soil–water medium using its shrinkage curve. Soil Sci. Soc. Am. J. 68, 359–370.
Briaud, J.-L., Zhang, X., Moon, S., 2003. Shrink test—water contentmethod for shrink andswell predictions. J. Geotech. Geoenviron. Eng. 129, 590–600.
Bronswijk, J.J.B., 1988. Modeling of water balance, cracking and subsidence of clay soils.J. Hydrol. 97, 199–212.
Bronswijk, J.J.B., 1990a. Shrinkage geometry of a heavy clay soil at various stresses. SoilSci. Soc. Am. J. 54, 1500–1502.
Bronswijk, J.J.B., 1990b. Shrinkage of Dutch clay soil aggregates. Netherlands J. Agric.Sci. 38, 175–194.
Bronswijk, J.J.B., 1991. Relation between vertical soil movements and water-contentchanges in cracking clays. Soil Sci. Soc. Am. J. 55, 1220–1226.
Cao, W., Bowden, W.B., Davie, T., Fenemor, A., 2006. Multi-variable and multi-sitecalibration and validation of SWAT in a large mountainous catchment with highspatial variability. Hydrol. Process. 20, 1057–1073.
Chertkov, V.Y., Ravina, I., 1998. Modeling the crack network of swelling clay soils. SoilSci. Soc. Am. J. 62, 1162–1171.
Chertkov, V.Y., Ravina, I., 2002. Combined effect of interblock and interaggregatecapillary cracks on the hydraulic conductivity of swelling clay soils. Water Resour.Res. 38, 1157–1171.
Coquet, Y., Touma, J., Boivin, P., 1998. Comparison of soil linear shrinkage curve fromextracted cores and in situ. Aust. J. Soil Res. 36, 765–781.
Cornelis, W.M., Corluy, J., Medina, H., Diaz, J.M., Hartmann, R., Van Meirvenne, M., Ruiz,M.E., 2006. Measuring and modeling the soil shrinkage characteristic curve.Geoderma 137, 179–191.
Coulombe, C.E., Dixon, J.B., Wilding, L.P., 1996. Mineralogy and chemistry of Vertisols.In: Ahmad, N., Mermut, A.R. (Eds.), Vertisols and Vertisol Technologies for TheirManagement. Elsevier, Amsterdam, pp. 115–200.
Croney, D., Coleman, J.D., 1954. Soil structure in relation to soil suction (pF). J. Soil Sci. 5,75–84.
Das Gupta, S., Mohanty, B.P., Köhne, J.M., 2006. Soil hydraulic conductivities and theirspatial and temporal variations in a Vertisol. Soil Sci. Soc. Am. J. 70, 1872–1881.
117A.S. Kishné et al. / Geoderma 157 (2010) 109–117
Deliberty, T.L., Legates, D.R., 2003. Interannual and seasonal variability of modeled soilmoisture in Oklahoma. Int. J. Climatol. 23, 1057–1086.
Delworth, T., Manabe, S., 1988. The influence of potential evapotranspiration on thevariability of simulated soil wetness and climate. J. Clim. 5, 523–547.
Doris, J.J., Rizzo, D.M., Dewoolkar, M.M., 2008. Forecasting vertical ground surfacemovement from shrinking/swelling soils with artificial neural networks. Int. J.Num. Anal. Methods in Geomech. 32, 1229–1245.
Dudoignon, P., Causseque, S., Bernard, M., Hallaire, V., Pons, Y., 2007. Vertical porosityprofile of a clay-rich marsh soil. Catena 70, 480–492.
Favre, F., Boivin, P., Wopereis, M.C.S., 1997. Water movements and soil swelling in a dry,cracked Vertisol. Geoderma 78, 113–123.
Fox, W.E., 1964. A study of bulk density and water in a swelling soil. Soil Sci. 98,304–316.
Gerke, H.H., 2006. Preferential flow descriptions for structured soils. J. Plant Nutr. SoilSci. 169, 382–400.
Greco, R., 2002. Preferential flow inmacroporous swelling soil with internal catchment:model development and applications. J. Hydrol. 269, 150–168.
Grossman, R.B., Brasher, B.R., Franzmeier, D.P., Walker, J.L., 1968. Linear extensibility ascalculated from natural-clod bulk density measurements. Soil Sci. Soc. Am. Proc. 32,570–573.
Grossman, R.B., Reinsch, T.G., 2002. Bulk density and linear extensibility. In: Dane, J.H.,Topp, G.C. (Eds.), Methods of Soil Analysis. Part 4, Physical Methods: SSSA BookSeries, no. 5. SSSA, Madison, WI.
Haines, W.B., 1923. The volume-changes associated with variations of water content insoil. J. Agric. Sci. 13, 296–310.
Hallaire, V., 1984. Evolution of crack networks during shrinkage of a clay soil undergrass and winter wheat crops. In: Bouma, J., Raats, P.A.C. (Eds.), Proc. ISSSSymposium onWater and Solute Movement in Heavy Clay Soils, ILRI, Wageningen,The Netherlands.
Ham, J.M., 2005. Useful equations in micrometeorology. In: Hatfield, J.J., Baker, J.M.,Viney, M.K. (Eds.), Micrometeorology in Agricultural Systems: ASA-CSSA-SSSA. No.47, pp. 533–560.
Heinsch, F.A., Heilman, J.L., McInnes, K.J., Cobos, D.R., Zuberer, D.A., Roelke, D.L., 2004.Carbon dioxide exchange in a high marsh on the Texas Gulf Coast: effects offreshwater availability. Agric. Forest Meteorol. 125, 159–172.
Hendrickx, J.M.H., Flury, M., 2001. Uniform and preferential flow mechanisms in thevadose zone. Panel on Conceptual Models of Flow and Transport in the FracturedVadose Zone, U.S. National Committee for Rock Mechanics, Board on Earth Sciencesand Resources, Commission on Geosciences, Environment and Resources, NationalResearch Council, Conceptual models of flow and transport in the fractured vadosezone. National Academic Press, Washington D.C, pp. 149–187.
Insightful, 2005. S-PLUS version 7.1. Insightful Co., Seattle, Washington, USA.Kirby, J.M., Bernardi, A.L., Ringrose-Voase, A.J., Young, R., Rose, H., 2003. Field swelling,
shrinking, and water content change in a heavy clay soil. Aust. J. Soil Res. 41,963–978.
Kishné, A.Sz., Morgan, C.L.S., Miller, W.L., 2009. Vertisol crack extent associated withgilgai and soil moisture in the Texas Gulf Coast Prairie. Soil Sci. Soc. Am. J. 73,1221–1230.
Kutilek, M., Germann, P.F., 2009. Converging hydrostatic and hydromechanic conceptsof preferential flow definitions. J. Contam. Hydrol. 104, 61–66.
Lal, R., Shukla, M.K., 2004. Principles of Soil Physics. Dekker, New York.Lepore, B.J., Morgan, C.L.S., Norman, J.M., Molling, C.C., 2009. A mesopore and matrix
infiltration model based on soil structure. Geoderma 152, 301–313.Lin, H.S., McInnes, K.J., Wilding, L.P., Hallmark, C.T., 1998. Macroporosity and initial
moisture effects on infiltration rates in Vertisols and Vertic intergrades. Soil Sci.163, 2–8.
McGarry, D., Daniells, I.G., 1987. Shrinkage curve indices to quantify cultivation effectson soil structure of a Vertisol. Soil Sci. Soc. Am. J. 51, 290–297.
McGarry, D., Yule, D.F., 2006. Soil shrinkage. In: Lal, R. (Ed.), Encyclopedia of SoilScience. 2nd. Taylor and Francis, Boca Raton, pp. 1633–1636.
Miller, W., Kishné, A.Sz. and Morgan, C.L.S., 2010. Vertisol morphology and seasonalcracking patterns in the Texas Gulf Coast Prairie. Soil Survey Horizons, Spring Issue,In print.
Milleret, R., Le Bayon, R.-C., Lamy, F., Gobat, J.-M., Boivin, P., 2009. Impact of roots,mycorrhizas and earthworms on soil physical properties as assessed by shrinkageanalysis. J. Hydrol. 373, 499–507.
Mitchell, R.J., Mayer, A.S., 1998. The significance of hysteresis in modeling solutetransport in unsaturated porous media. Soil Sci. Soc. Am. J. 62, 1506–1512.
Olsen, P.A., Haugen, L.E., 1998. A new model of the shrinkage characteristic applied tosome Norwegian soils. Geoderma 83, 67–81.
Parker, J.C., Amos, D.F., Zelazny, L.W., 1982. Water adsorption and swelling of clayminerals in soil systems. Soil Sci. Soc. Am. J. 46, 450–456.
Peng, X., Horn, R., 2007. Anisotropic shrinkage and swelling of some organic andinorganic soils. Eur. J. Soil Sci. 58, 98–107.
Peng, X., Horn, R., Smucker, A., 2007. Pore shrinkage dependency of inorganic andorganic soils on wetting and drying cycles. Soil Sci. Soc. Am. J. 71, 1095–1104.
Quirk, J.P., 1994. Interparticle forces: a basis for the interpretation of soil physicalbehavior. Adv. Agron. 53, 121–183.
Reeve, M.J., Hall, D.G.M., Bullock, P., 1980. The effect of soil composition andenvironmental factors on the shrinkage of some clayey British soils. J Soil Sci. 31,429–442.
Römkens, M.J.M., Prasad, S.N., 2006. Rain infiltration into swelling/shrinking/crackingsoils. Agri. Water Manag. 86, 196–205.
Saiyouri, N., Tessier, D., Hicher, P.Y., 2004. Experimental study of swelling inunsaturated compacted clays. Clay Miner. 39, 469–479.
Santamarina, J.C., Klein, K.A., Fam, M.A., 2001. Soils and waves. Particulate MaterialsBehavior, Characterization and Process Monitoring. Wiley, Chichester, UK.
Soil Survey Staff, 1990. Classification and Management of Wet Soils. Guide book forTexas. VIII Int. Soil Correlation Meeting, Baton Rouge, LA, and College Station, TX.Dept. of Soil and Crop Sciences, TAMU and USDA Soil Conservation Service, USDANational Soil Survey Center, Lincoln. Data is available online for Lab Pedon Number:89P0032 and 89P0033 at http://ssldata.nrcs.usda.gov/querypage.asp.
Soil Survey Staff, 1999. Soil Taxonomy: A Basic System of Soil Classification for Makingand Interpreting Soil Surveys, 2nd ed. Agric. Handb., 436. US. Gov. Print. Off,Washington.
Stirk, G.B., 1954. Some aspects of soil shrinkage and the effect of cracking upon waterentry into the soil. Aust. J. Agric. Res. 5, 279–290.
Tessier, D., 1984. Etude de l'organisation des materiaux argilleux. Ph.D. Thesis, Univ. ofParis.
Tessier, D., 1990. Behaviour and microstructure of clay minerals. In: De Boodt, M.,Hayes, M., Herbillon, A. (Eds.), Soil Colloids and Their Associations in Aggregates.Plenum Publishing Co., New York, NY, pp. 387–414.
Wells, R.R., DiCarlo, D.A., Steenhuis, T.S., Parlange, J.-Y., Römkens, M.J.M., Prasad, S.N.,2003. Infiltration and surface geometry features of a swelling soil followingsuccessive simulation rainstorms. Soil Sci. Soc. Am. J. 67, 1344–1351.
Wells, R.R., Römkens, M.J.M., Parlange, J.-Y., DiCarlo, D.A., Steenhuis, T.S., Prasad, S.N.,2007. A simple technique for measuring wetting front depths for selected soils. SoilSci. Soc. Am. J. 71, 669–673.
Wilding, L.P., Tessier, D., 1988. Genesis of Vertisols: shrink–swell phenomena. In:Wilding, L.P., Puentes, R. (Eds.), Vertisols: Their Distribution, Properties, Classifi-cation and Management. USDA Soil Management Support Services, Tech. Monogr.18. Texas A&M Univ. Printing Center, College Station, pp. 55–81.
Wilding, L.P., Williams, D., Miller, W.L., Cook, T., Eswaran, H., 1990. Close interval spatialvariability of Vertisols: a case study in Texas. In: Kimble, J.M. (Ed.), Proc. 6th Int. SoilCorrelationMeeting (ISCOM), Characterization and Utilization of Cold Aridisols andVertisols. USDA Natural Resources Conservation Service, National Soil SurveyCenter, Lincoln, NE, pp. 232–247.
Yule, D.F., Ritchie, J.T., 1980a. Soil shrinkage relationships of Texas Vertisols: I Smallcores. Soil Sci. Soc. Am. J. 44, 1285–1291.
Yule, D.F., Ritchie, J.T., 1980b. Soil shrinkage relationships of Texas Vertisols: II. Largecores. Soil Sci. Soc. Am. J. 44, 1291–1295.
