| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Brandenburg University of Technology, Cottbus, Chair of Soil Protection and Recultivation, P.O. Box 101344, D-03013 Cottbus Germany
* Corresponding author (buczko{at}tu-cottbus.de)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: CA, contact angle • HI, heterogeneity index • SDM, sessile drop method • WDPT, water drop penetration time
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Variability of microbial abundance and microbial activity (Grundmann and Debouzie, 2000) will probably influence the pattern of hydrophobicity within the soil, because soil water repellency is, in most cases, induced by organic matter. Therefore, heterogeneity of water repellency can be expected to occur in many soils.
Investigations of water repellency and its variability, especially in field soils under natural conditions, are important, because it can induce unstable wetting fronts and preferential flow (Hendrickx et al., 1993; Ritsema et al., 1998), and erratic water content distributions within the topsoil layers (Ritsema and Dekker, 1998; Dekker et al., 1999). Actual water repellency data, measured on field moist samples or in situ, are often not available. Therefore, in many studies, oven-dried samples are utilized, although the correlation between potential and actual water repellency may be rather poor (Dekker et al., 1998). We hypothesize that heterogeneity of water repellency can be observed and assessed also using disturbed and oven-dried samples, if appropriate indices are used and the results are interpreted with regard to the experimental limitations.
Flow phenomena induced by water repellency are compounded, however, by the influence of other factors, which can also promote preferential flow and heterogeneous water content distributions, for instance, spatial variability of hydraulic parameters, distribution of roots, surface microtopography, and macropores. These factors often concur with water repellency. Thus, quantification of water repellency and its heterogeneity is highly relevant for assessment of the affinity of soils for preferential flow and unstable wetting fronts.
The wettability of granular materials like soils cannot usually be determined directly. Therefore, a variety of indirect methods are used (Letey et al., 2000; Bachmann et al., 2003). Many of them require relatively large soil volumes, and thus, assessment of intrasample variability is not possible. In spite of the wealth of methods to determine and assess the level of hydrophobicity in soils, rigorous interpretation of results and, what is more important, intercomparison between different methods is problematical. This is due to different measurement volumes of the methods and differences in physical significance (for instance, persistence of repellency, as determined by the WDPT test, or relationship between particle surface forces as measured with the critical surface tension test). All published intercomparisons between different methods seem to yield ambiguous correlations with large scatter (e.g., King, 1981; Doerr et al., 1998; Bachmann et al., 2000b, 2003).
An assessment of intrasample variability of water repellency is in principle possible with methods applying single small water droplets at the soil surface, that is, the WDPT test (Van't Woudt, 1959), which consists of measuring the time needed for water absorption, and the SDM (Bachmann et al., 2000a, 2000b), by which CAs at a smoothed soil surface are determined. Quantification of spatial variability of water repellency has been done using variogram analysis (Hallett et al., 2004), which yields information about spatial correlation lengths. It is, however, feasible only for measurements done according to a spatial pattern, that is, in situ or on undisturbed samples. More qualitative descriptions encompass graphical contour plots (Dekker et al., 2001), transects (Doerr et al., 1998; Gerke et al., 2001), or description in terms of statistical parameters (Hallett et al., 2004). For disturbed samples, only heterogeneity indices in terms of bulk statistical parameters are applicable.
The objective of this study was to assess and quantify the potential water repellency and to estimate the magnitude of its intrasample variability (i.e, at a scale smaller than 100 cm3) and its variability with soil depth using disturbed and oven-dried soil samples, both in terms of strength (CAs as determined with the SDM) and persistence (WDPT test), for a sandy soil under a forest transformation chronosequence.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
Soil Sampling and Sample Preparation
For each of the four plots, disturbed soil samples were taken in November and December 2001 according to soil horizon boundaries from profile pits down to a depth of 160 cm. The discerned horizons along with the pertinent depth intervals are compiled in Table 2. For each horizon, samples were taken by pressing 100-cm3 steel cylinders into the undisturbed profile walls at the midpoint depths between the horizon boundaries. Three samples for each horizon were taken in immediate horizontal proximity to each other and from the same soil depth, such that the three soil cores were extracted along a horizontal within a distance of 15 cm. The total number of samples was 72. For the mixed stands (P76/B34 and P114/B57), care was taken that the profile pits were at the midpoint between the nearest pine and beech trees. In the laboratory, the samples were oven dried at 60°C for 3 d, and equilibrated for approximately 24 h with the ambient temperature (approximately 21°C) and relative air humidity (approximately 40%) in the laboratory. The gravel fraction (diameter > 2 mm) was separated by dry hand-sieving from the dried soil material and the fraction < 2 mm was used for the SDM and the WDPT test. Further sieving was not done, the sample surfaces were only gently smoothed by hand for these tests after filling the disturbed soil material into circular dishes of 10-cm diameter.
|
Sessile Drop Method
The SDM applied in this form is described in Bachmann et al. (2000a, 2000b). The soil materials < 2 mm were not separated into different particle size fractions (similar as in Bachmann et al., 2000a), because of the relatively uniform particle size distribution of the sandy soil substrate investigated here. We note that the soil preparation necessary for the SDM (sieving of the fraction > 2 mm, oven drying) precludes an assessment of the impact of aggregates (which were not present in this sandy soil) and moisture content on the CA, as described by Goebel et al. (2004) using the capillary rise method.
Contact angle measurements were done on a thin layer of soil material (such that droplets could not infiltrate). These were fixed on a double-sided adhesive tape (TESA, type 55733, Beiersdorf, Hamburg, Germany), which was glued on a glass slide. The soil material was attached to a 2.5 by 5.5 cm (1375 mm2) area by pressing the slide manually for several seconds on the same smoothed soil surfaces which were used later for the WDPT test. The slide was then shaken carefully to remove nonadhering soil particles. This procedure was repeated three times for each sample. The few remaining particle-free gaps on the tape were filled by sprinkling soil material on the slide. Contact angles were measured within 15 min after sample preparation using a microscope fitted with a goniometer scale (OCA10, DataPhysics, Filderstadt, Germany), and at a room temperature of 21°C and a relative humidity of 40%. Using a pipette, a drop of deionized water (drop volume 20 µL) was placed on the soil sample, and shortly thereafter CAs were measured by adjusting the goniometer as a tangent at the point of the three-phase contact, first at the left, and then at the right side of the droplet. Subsequently, another drop was placed on the sample surface and CAs were measured. This procedure was repeated 10 times for each sample. The reason for measuring CAs immediately after placement of the drop, and only thereafter placing the next drop on the sample, was the observed rapid decrease of CAs on the sample surface. The successive droplets applied on the surface exhibited on average no time-dependency in the measured CAs (data not shown), which might be induced, for instance, by water vapor adsorption from neighboring droplets.
Water Drop Penetration Time Test
For the WDPT test (Van't Woudt, 1959), several small drops (40 µL) of distilled water from a laboratory pipette were placed on the same smoothed soil surface which were used for the preparation of the SDM samples, and the time for a water drop to infiltrate into the soil recorded as WDPT. To assess the intrasample variability of the WDPT, we placed 10 drops of distilled water on the surface of each soil sample, with nearest distances of the droplets to each other of approximately 5 mm. For each drop the WDPT value was measured with a stopwatch. The shortest measurable penetration time was 0.5 s (corresponding to instantaneous infiltration). The measurement procedure was terminated after 3600 s. To water droplets with penetration times in excess of 3600 s, a WDPT value of 3601 s was assigned. This cutting-off at 0.5 and 3601 s when measuring WDPT values produced exceptionally low HI values for low and high mean values of log(WDPT) (see results and discussion section).
Quantification of Intrasample Variability of Water Repellency
Intrasample (i.e., at a scale of 100 cm3) variability of water repellency was quantified with the following:
 | [1] |
4). This HI has advantages in comparison with other statistical measures, like the standard deviation and the coefficient of variation: Due to the utilization of the quantiles Q90 and Q10, the index is applicable also for non-normal distributions; moreover, the normalization by the scale factor SF enables a better comparability of variabilities which pertain to values of different magnitude, compared to the coefficient of variation, which attains very high values when calculated for small mean values. The value of HI is a measure of the variability within a given sample volume (in this case, 100 cm3). It is comparable to the nugget effect of a variogram, and by increasing the support volume, information about the increase of variability with increasing spatial extent, similar as in variograms, can be obtained. However, in variogram analysis, the nugget effect is considered as disturbing background noise. However, this background noise consists of subscale variability, and pure measurement error. Measurement error is defined here as the variability caused by the uncertainty of the observation or the measuring device. To assess the proportion of pure measurement error inherent in the estimated total variability, control measurements both of CA and WDPT were done on pure quartz sand with a narrow particle size distribution of 0.1- to 0.4-mm diameter. This sand is sold as building material (QuickMix, Osnabrück, Germany) and originated from a glaciofluvial deposit in northern Germany, but has been processed before being sold: The organic matter contained in the original sediment was removed by elutriation, therefore no residual ash contents were present. The prior content of organic carbon has not been recorded previously, but was presumably below 1%. After elutriation, the sand was dried by hot air at 360 to 430°C. Due to the glaciofluvial origin of the sand, particles are moderately well rounded (roundness factor about 0.8). Due to the removal of organic matter and the intense heating, the sand exhibited no signs of water repellency (CA = 0° and WDPT < 0.5 s). This material was hydrophobized by sprinkling with naphtha (CAS no. 64742–48–9). Contact angles were measured using the pure hydrophobized sand. This hydrophobized sand yielded consistently WDPT values in excess of 5 h (after that time, it was assumed that disappearance of droplets is governed by evaporation, even though sample dishes were covered by lids), making an assessment of the measurement error in WDPT impossible. Therefore the hydrophobized sand was thoroughly mixed with the tenfold amount of untreated sand to test pure measurement error of the WDPT test. |
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
The different repellency levels which were documented previously in those stands (Buczko et al., 2002, 2005), using the WDPT test and the ethanol percentage–critical surface tension test (Roy and McGill, 2002), with relatively high repellency in the mixed stands and lower repellency in the pure stands, have been related to prevailing humus forms and thicknesses of the forest floor layers (Buczko et al., 2002). Analyses of chemical characteristics of different forest humus forms, for instance using 13C nuclear magnetic resonance spectroscopy, often yield for mor or moder humus forms greater contents of hydrophobic aromatic and alkyl compounds than for mull humus forms (e.g., Kögel-Knabner et al., 1988). It is worth noting, that characterization of humus forms at the two mixed stands P76/B34 and P114/B57 is based on transects with about 100 sampling points (Bens et al., 2006), and the prevailing humus form moder rich in fine humus is found mainly in regions which are more or less at equal distances from both pine and beech trees (as the sampling pits for soil water repellency samples used here). Within the influence zone of the pine trees, mor type moder is more prevalent at these stands (Bens et al., 2006). Therefore, soil water repellency at these stands presumably is variable at a larger scale (several meters), in addition to the small-scale variability investigated in the present study. Moreover, it has been shown that the repellency levels at this site are fluctuating according to the seasons of the year (Buczko et al., 2005), and that, especially for samples with field water contents near the wilting point, actual water repellency can be more persistent than potential repellency (Buczko et al., 2002).
For samples taken from the same site and at the same date, Woche et al. (2005) determined advancing CAs using the Wilhelmy plate method which are on average 30° higher than the CAs determined in the present study with the SDM, although the overall correlation is relatively high (r = 0.85). The great differences may be interpreted as the difference between advancing (Wilhelmy plate method) and static (SDM) CAs, and have been described for other soils as well (Bachmann et al., 2003).
The rapid decrease of repellency with depth indicates, that the substances inducing hydrophobicity are derived largely from forest litter, and only to a minor degree from tree roots. The decrease of water repellency with increasing soil depth is characteristic for soils under forest (Crockford et al., 1991; Huffman et al., 2001). The depth distribution of CAs and WDPT values found in this study seem qualitatively similar to each other (compare Fig. 1 and 2). In a previous study at the same site (Buczko et al., 2002), ethanol-percentage (EP) values were found to decrease more rapidly with depth, compared to WDPT values, a finding which is in agreement with results reported from other sites (Dekker and Ritsema, 1994). The physical "message" of EP values and CAs is similar, because both quantities depend on the strength of soil water repellency, that is, the free energy of the particle surfaces. In theory, for CAs < 90°, water droplets should infiltrate instantaneously into the soil, corresponding to a wettable soil with WDPT values < 5 s (Carrillo et al., 1999). Accordingly, all samples with CA < 90° would be wettable, and the strength of water repellency as expressed by CA would decrease more rapid through depth (Fig. 1) in comparison with WDPT values (Fig. 2), in line with the depth decrease of EP values reported previously for the same site (Buczko et al., 2002).
In contrast to WDPT values, there is no generally applied classification for water repellency as determined by CA measurements with the SDM. King (1981) proposed a classification of CAs between "not significant" (CA < 75°) and "very severe" (CA > 98°), which was based on CA determined using the capillary rise method.
Relation between Contact Angle and Water Drop Penetration Time Test
A plot of measured CAs vs. log(WDPT) values (Fig. 3
) reveals a roughly linear relationship for CAs between 50 and 100°, although with a broad scatter. The horizontal plateaus for WDPT values of 0.5 and 3601 s are probably artifacts caused by the lower and upper boundaries inherent in the WDPT recording procedure. The significance of the scatter is enhanced, considering that the WDPT values in Fig. 3 are logarithmically transformed. The scatter in WDPT values is especially large around CAs of 90°, which is consistent with previous findings reported in the literature (Bachmann et al., 2000b). For samples displaying CAs of about 90° (i.e., between 85 and 95°), the corresponding WDPT values range between less than 100 s and 1 h. A possible explanation for this variability is that the WDPT test is sensitive only in the narrow zone at CAs of about 90° (Bachmann et al., 2000b). On the other hand, the physical significance of the WDPT test and CAs is not identical. Whereas CAs yield a measure of the boundary forces of the soil particle surfaces, WDPT values indicate the persistence of water repellency of those particles. Theoretically, an applied water drop will infiltrate into the soil only when the CA becomes lower than 90° (provided that the applied drop is significantly larger than the pore size). This means that samples with WDPT values < 5 s should exhibit SDM CAs < 90°, and vice versa. On the other hand, several samples exhibit CAs lower than 90°, but WDPT values distinctly higher than 5 s. This phenomenon cannot be explained by measurement errors of CA and WDPT values alone, because the data points are systematically shifted into the region with WDPT > 5 s and CA < 90°. Possibly, also the "subcritical" water repellency (cf., Hallett et al., 2001) has an impact on infiltration, which can be measured by the WDPT test. This is compounded, however, by two factors:
|
Another factor is the roughness of the prepared soil surface. Although the soil material was fairly homogeneous, differences between the maximum and the minimum particle size class do inevitably occur. Consequently, it is doubtful where the plane of reference for the CAs actually lies. In the case of extreme roughness, a great part of the particle surfaces would be perpendicular to the glass slide. This would mean that actual CAs for those surfaces were 90° less than the values measured with the glass plane as a reference (Roy and McGill, 2002). For the prepared surfaces employed here, this effect is presumably less significant. To obtain prepared surfaces as smooth as possible, it would have been necessary to utilize narrowly sieved particle size fractions and to apply droplets with diameters 20-fold larger than the surface roughnesses (Bachmann et al., 2000b). Sieving into different particle size classes was not done here, because the SDM should be utilized for assessment of intrasample heterogeneity as well. On the other hand, even for narrowly sieved soil fractions, WDPT values > 5 s and CA < 90° were observed by Bachmann et al. (2000b), similar as in our study.
Intrasample Heterogeneity of Water Repellency
Pure measurement error was estimated to contribute a HI of 6 to 10% for CA measurements (for CA between 90 and 110°, corresponding to standard deviations of 3 to 5°; for comparison, Bachmann et al. [2000b] give an accuracy of about 1–2° for CA measurements on smooth surfaces.) and 5 to 6% for the WDPT test (for WDPT values between 1200 and 3000 s). Thus, the measurement error for WDPT seems to be slightly lower than for CAs, but overall within the same range of magnitude. These results correspond for the assigned intervals to a proportion of measurement error of the observed heterogeneity of 55% for CA measurements and 39% for WDPT values. In the following, however, uncorrected HI values are given, because pure measurement error probably changes with magnitude of the measured values, but could not be assessed here for nonrepellent (CA 0° and WDPT < 0.5 s) and extremely repellent conditions (CA > 110° and WDPT > 3600 s).
The intrasample heterogeneity of CAs is in general highest at shallow soil depths at about 5 to 10 cm, whereas directly within the topsoil layer at 0- to 5-cm depth, heterogeneity is somewhat smaller (Fig. 1, Table 2). An exception is the soil depth 53 to 86 cm at plot P114/B57 with an HI of 27.7% which is higher than the HI for the topsoil horizons at this stand. Compared with the HI values, coefficients of variation show for each stand maximum values at greater soil depths (Table 2). This is because the coefficient of variation is sensitive to the magnitude of the mean value, which for both the CA and log(WDPT) in general decreases with soil depth.
For log(WDPT), maximum values of HI are observed for all stands in the topsoil layers at about 5- to 15-cm depth (Fig. 2, Table 2). There, the values of HI for log(WDPT) are distinctly higher than the respective values of HI for the CAs. At stand P114/B57 for the soil depth 53 to 86 cm, the HI for log(WDPT) is 0 (all measured WDPT values were 0.5 s), in contrast to the respective HI for CAs.
Heterogeneity index values depend in a distinct manner from the sample volume. In Table 2 and Fig. 4 , HI values for each of the three 100 cm3 samples are compiled along with the values calculated for all three 100-cm3 samples of a given soil depth. In general, HI values for all three 100-cm3 samples are higher than those for single 100-cm3 samples, in a few cases even to a considerable degree. The single samples at each soil depth were extracted from immediately adjoining locations, and therefore the increase in HI values probably reflects a scale effect.
|
When comparing these results with variabilities of CA measurements given in the literature, Bachmann et al. (2000b) report for plain surfaces an accuracy of CA measurements of 1 to 2°, but for prepared soil surfaces a standard deviation of 5 to 15° (i.e., slightly less than determined in the present study, Table 2); but even the standard deviation in Bachmann et al. (2000b) could represent both measurement error and heterogeneity. Bachmann et al. (2000a) ascribe standard deviations of up to 20° for a sandy soil material to measurement error.
For the WDPT test, an assessment of measurement error or reproducibility is seldom reported in the literature. Whereas for most soil physical measurements standard deviations or measurement errors are routinely given (water content, bulk density, etc.), for WDPT measurements, these data conspicuously are mostly lacking. On the other hand, in almost all studies, appreciable heterogeneity of WDPT values has been observed, but this is usually not quantified and is attributed to an inherent heterogeneity of the soil. An assessment of "pure" measurement error for the WDPT test is problematical, because it is not possible to apply two or more water droplets on an identical spot of the soil surface. A spot where a water droplet had been applied in the course of WDPT measurement is affected by this measurement process and does not attain its a priori level of hydrophobicity, because the hydrophobic properties are altered (Doerr, 1998).
In the present study, slightly lower pure measurement error of the WDPT test (equivalent to HI of 5–6%) compared with the SDM (equivalent to HI of 6–10%) were found. Consequently, when the effects of measurement error are accounted for, the resultant average heterogeneity at the Kahlenberg site is higher for log(WDPT) values than for CAs.
In the literature, there is no information about "pure" measurement error in connection with the WDPT test. A few studies, however, are concerned with reproducibility of the WDPT test. For instance, Doerr et al. (2002) report for 12 different soils from different sites all over the world a mean intrasample standard deviation of 0.178 classes, when categorizing the measured WDPT values between <5 and >18000 s into 15 classes and applying three droplets per sample. They regard this as good reproducibility. In fact, when the WDPT data of Kahlenberg are categorized into the same classes as in Doerr et al. (2002), the mean standard deviation would be 0.553 classes. Possible reasons for this discrepancy are that in Doerr et al. (2002) the samples were sieved and thoroughly homogenized, whereas the samples taken here from the Kahlenberg site were neither sieved nor homogenized. Moreover, in Doerr et al. (2002), three droplets were applied per sample, whereas in the present study 10 droplets per sample were applied. Using similar classes of WDPT values as Doerr et al. (2002), Doerr (1998) reports for laboratory testing of sieved and thoroughly mixed sandy loam soil (Humic Cambisols and Umbric Leptosols) samples a mean standard deviation of 0.186 categories, when applying 15 droplets per sample. In Doerr (1998) the samples had been merely hand sieved to remove the skeletal fraction (>2 mm), but not thoroughly homogenized. In contrast to Doerr (1998) and Doerr et al. (2002), the intrasample variability of the Kahlenberg soil seems to be much higher. This could be due to some extent to the sample treatment before WDPT measurements (no homogenizing, only hand sieving to remove the gravel fraction), but also to a higher intrinsic variability of WDPT values at the Kahlenberg site. Similar to our results, Crockford et al. (1991) and Huffman et al. (2001) report poor reproducibility (although not quantified) of WDPT measurements using unsieved samples.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The intrasample variability, as quantified by HI, based on the difference between the 90 and 10% quantiles of measured data, normalized by the range of possible data values, tends to be higher in the upper soil layers and declines with depth. When the proportion of estimated "pure" measurement error is subtracted from HI values, the average heterogeneity is higher for log(WDPT) (mean HI = 8.9%) than for CAs (mean HI = 6.7%). The mean HI for CAs is essentially unchanged, when calculation of HI is based on cos(CA). Values of HI show mostly a clear scale dependence, with higher values for larger soil volumes (in this case 3[100 cm3]) compared to single samples of 100 cm3.
It should be noted, that soil water repellency of oven-dried samples, and presumably also its variability, depend also from several other factors that were not considered in this study: drying temperature (Dekker et al., 1998), air humidity (Doerr et al., 2002), and water content before oven drying (Buczko et al., unpublished data, 2005). Moreover, a possible seasonal variability should be taken into account (Buczko et al., 2005).
Summarizing, the preferential flow phenomena which were observed at this site, despite the ostensible textural homogeneity, may be explained, at least to some extent, by the high degree of heterogeneity of water repellency, although other factors may be important as well (e.g., funnel flow, root channels). For soils with extremely repellent surface horizons and uniform WDPT values in excess of 3600 s, as described for instance from the same site in the summer months (Buczko et al., 2005), preferential flow may be induced even in a homogeneously repellent upper soil horizon. In such cases, rainwater may flow laterally over the soil surface and concentrate in small surface depressions and infiltrate preferentially from there, or it may utilize root and worm channels for infiltration.
Future studies should focus on the following aspects:
|
ACKNOWLEDGMENTS |
|---|
Received for publication June 10, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  U. Buczko, O. Bens, and R. F. Huttl Tillage Effects on Hydraulic Properties and Macroporosity in Silty and Sandy Soils Soil Sci. Soc. Am. J., October 27, 2006; 70(6): 1998 - 2007. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education |
Vadose Zone Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
