SSSAJ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Stutter, M. I.
Right arrow Articles by Billett, M. F.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Stutter, M. I.
Right arrow Articles by Billett, M. F.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Stutter, M. I.
Right arrow Articles by Billett, M. F.
Related Collections
Right arrow Spatial Variability
Right arrow Soil Analysis
Right arrow Soil Chemistry
Published in Soil Sci. Soc. Am. J. 68:1304-1314 (2004).
© 2004 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA

DIVISION S-5—PEDOLOGY

Spatial Variability in Soil Ion Exchange Chemistry in a Granitic Upland Catchment

M. I. Stuttera,e,*, L. K. Deeksb,d and M. F. Billettc,e

a The Macaulay Institute, Craigiebuckler, Aberdeen. AB15 8QH, UK
b National Soil Resources Institute, Cranfield University, North Wyke, Devon. EX20 2SB, UK
c Centre for Ecology and Hydrology, Bush Estate, Penicuik, Midlothian. EH26 0QB, UK
d Soil-Plant Dynamics Unit, Scottish Crop Research Institute, Dundee. DD2 5DA, UK
e School of Biological Sciences, University of Aberdeen. AB24 3UU, UK

* Corresponding Author (m.stutter{at}macaulay.ac.uk).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Advances in quantifying the spatial variability of soil properties made for agricultural soils are not being mirrored for naturally structured upland soils. The objectives of this study were to determine the degree of spatial variability and variance structure of cation exchange chemistry in a granitic, heather moorland site (Northeast Scotland). Two 20 by 20 m soil plots, a Typic Placaquod and a Typic Humaquept, were sampled at O and B horizon depths at 104 locations in a regular grid overlayed with cluster (Placaquod) and transects (Humaquept) patterns. Soils were analyzed for pH and exchangeable cation, physical, and hydraulic properties. Results showed strongly significant vertical differences in exchange chemistry between surface organic and mineral horizons at either site for all chemical properties. Strong lateral variability was also apparent within the plots. Coefficients of variation (CV) were 18 to 52% for O horizon chemical properties, with similar variability between sites. In B horizons CV values were 26 to 119%, the highest associated with Humaquept chemical properties. Compared with the Placaquod the Humaquept had lower mean pH, but higher mean concentrations of exchangeable Ca and Mg in both horizons. Geostatistical analyses highlighted a generally strong degree of spatial dependence to Placaquod properties, particularly in the organic horizon where correlation ranges were greater. By contrast, the majority of properties for the Humaquept showed random, pure nugget variance indicating no spatial correlations at this scale of observation. It is postulated that seasonal water logging of the Humaquept may explain some differences in the exchange chemistry between sites. These differences in chemistry and in the spatial patterns of variability have implications not only for modeling the role of such soils in controlling the hydrochemical environment of the uplands, but also for the design of field soil sampling strategies for accurately quantifying soil properties.

Abbreviations: CEC, cation-exchange capapcity • CV, coefficients of variation


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
SOIL PROPERTIES EXHIBIT a complex degree of variability in both space and time. This variability is both continuous and scale-dependant. The understanding and incorporation of this variability, or the associated uncertainty, into modeling approaches relies on the availability of detailed sampling information and this places a limitation on the development of spatially distributed models concerning a range of environmental processes (Burrough, 1993; Goovaerts, 2001; Park and Vlek, 2002). In contrast some catchment hydrochemical models lump soil variability to reduce model inputs, assuming a constant composition over space, or time (Christophersen, 1982; Cosby et al., 1985), but are open to criticism as they do not adequately consider true variability (Neal, 1996).

Traditional soil science has been concerned with the assessment of how many observations in the field are necessary to quantify certain properties and where the observations should be made. Conventional statistical analyses of such data have assumed a random variation of independent properties to which a mean and confidence interval could be accurately ascribed. However, it is recognized that samples cannot be recorded everywhere and hence the drive to predict properties in the unsampled neighborhood of measured values has necessitated an appreciation of the spatial dependence of the variables of interest. As a result of this requirement geostatistical methods have been developed to allow the spatial dependence between observations to be expressed in terms of a variogram (e.g., Webster, 1985). This analysis describes the average similarity between pairs of points within given separation classes and forms the basis for techniques of kriging, by which properties of unsampled regions are derived (e.g., Burgess and Webster, 1980a,b; Yost et al., 1982).

The particular applicability of geostatistics to the earth sciences accounted for its extensive early use in the mining industry (e.g., David, 1977; Verly et al., 1984). Applications in soil science have since been concerned with the estimation of soil textural and hydraulic properties (e.g., Ovalles and Collins, 1988; Lascano and Hatfield, 1992; Stolt et al., 1993), chemical properties (Cambardella et al., 1994; Mallarino, 1996; White et al., 1997; Anderson-Cook et al., 1999), biological and ecological properties (Bonmati et al., 1991; Rochette et al., 1991; Rossi et al., 1992), with some having an emphasis on the optimization of sampling strategies (e.g., Ameyan, 1986; Di et al., 1989; Chien et al., 1997).

Much of the published information on spatial variability, including many of the geostatistical studies above, has concentrated on the properties of agriculturally managed soils. Some of these studies have addressed patterns of soil exchange chemistry in relation to nutrient budgets (Cambardella et al., 1994; Mallarino, 1996; Chien et al., 1997; Anderson-Cook et al., 1999). However, spatial patterns of soil properties in such agricultural soils may be partly attributed to the manipulations of intensive management such as fertilizer application (Mallarino, 1996; Paz-González et al., 2000). However, for undisturbed upland systems, where variability is the product of natural pedogenic rather than anthropogenic factors, heterogeneity is less quantified. Some studies of soil solution compositions have been reported for upland regions in relation to catchment acidification issues in northern Europe (e.g., Neal, 1992; Taugbøl and Neal, 1994), but these were simply to address uncertainty in lumped modeling approaches and no appraisals of the spatial structure of the variability were made. In another study exchangeable cations were evaluated at Birkenes, Norway over a 100 by 100 m hillslope, comprising spodosols and histosols (Mulder et al., 1991). In this case conventional (non-parametric) statistical analyses showed differences between component soil types of the slope in basic and acid cations, but not in cation-exchange capacity (CEC).

The need to accurately model sensitive upland systems and predict their future behavior in relation to environmental change suggests that a greater understanding of their underlying spatial variability is required. Soil exchange properties express important characteristics of how soils interact with their hydrochemical environment, influencing the dynamics of solute transport, buffering capacity and the quality of runoff from catchments (e.g., Billett and Cresser, 1996). In the uplands exchange properties may vary considerably between horizons, for example between organic surface layers and underlying mineral soils (Reid et al., 1981), with characteristic changes in drainage water chemistry accompanying changes in flow contributions through horizons (Billett and Cresser, 1996).

This study comprised an intensive fine-scale investigation of the soil exchange chemistry of two soil units, 600 m apart, within an unmanaged nutrient-deficient upland catchment comprising natural soils and moorland vegetation. The objectives of the study were to compare the within-plot chemical variability and the spatial structure of the variance between the two soil plots. These comparisons were made both between the surface organic and mineral subsoil horizons at each plot and for similar horizons between the two plots.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Study Area
The soils lay within a gently sloping (6–10°) headwater catchment of the River Dee, in the foothills of the Cairngorm Mountains, Northeast Scotland (Fig. 1). Plant communities are boreal heather moorland (Vaccinium–Erica cinerea/Calluna vulgaris) to bog heather moorland (Narthecium–Erica tetralix/Sphagnum spp.), with some remnants of native pinewood (Pinus sylvestris). Management, as a grouse moor, includes periodic heather burning in the catchment, although no evidence of recent burning exists for the two plots. Mean annual precipitation and temperature is 1070 mm and 7°C, respectively. Geologically the catchment is uniformly underlain by coarse pink granite (Fungle Group, Mount Battock series) rich in K feldspar. The minerals soils are dominated by Spodosols on the slopes and skeletal soils to the hilltops, both within the Scottish Countesswells Association (MISR, 1982). These have formed over shallow stoney, gritty, sandy loam drift material overlying deeply weathered granite, which increases soil fine gravel contents. Histosols are found in the low-lying regions of the catchment, up to 1.4 m in thickness.



View larger version (46K):
[in this window]
[in a new window]
  		Fig. 1. Site location and detail of the two soil positions in relation to surface topography and drainage.

 
The relation of the two sampling plots to surface topography is given in Fig. 1. Table 1 shows representative profiles from each area. Soils at the western site comprised Spodosols, with diagnostic spodic horizons within 50 cm of the surface. The soil at the study plot, occupying a mid-slope position, was a freely draining Typic Placaquod, with a deep organic enriched Bhs1 horizon, overlying a thin, relatively continuous placic horizon with slightly indurated subsoil. At the easterly site the soils were best classified as Inceptisols, with the soil at the second study plot described as a Typic Humaquept. Despite no significant mottling, this soil was considered to remain saturated for a dominant period by virtue of observations of soil wetness, topographic position at the base of a concave slope and the vegetation characteristic of wet locations (Table 1). Some characteristics of leaching, including bright coloring of translocated Fe within the lower B horizon, suggested that leaching occurred during drier weather. Organic horizons of both plots were generally dark, amorphous and well decomposed with thin litter layers. Thin, weakly elluviated A horizons were common to both soils, noticeably more humic for the Humaquept. The two soils are hereby referred to as Placaquod and Humaquept. However, due to a lack of sufficiently detailed chemical analyses of the soils these taxonomic descriptions are given as best estimates of general properties, which are naturally expected to vary considerably over the scale of the grid plots.


View this table:
[in this window]
[in a new window]
  		Table 1. Comparisons of representative soil profiles from the two sites.

 
Sampling
Soils were sampled from the plots (measuring 20 by 20 m at both sites) in July of consecutive years (the Placaquod in 1999 and Humaquept in 2000) following similar periods of dry weather. Sampling design at the Placaquod plot (Fig. 2a) comprised a regular grid of 4 m spacing (36 points) overlain by a centrally biased random cluster (68 points). At the Humaquept plot this design was changed to a regular grid with random transects (Fig. 2b) of the same number of points and similar sample spacing. Individual grid locations were sampled using soil cores of 6 cm length inserted vertically. A central core of undisturbed soil (6 cm diameter) was recovered for hydraulic properties. An outer core (10 cm diameter) around the inner core was used to reproducibly sample the bulk soil bordering each central core and this portion for chemical analyses was transferred to bags. The sampling strategy was designed as a compromise to allow simultaneous appraisal of chemical and hydraulic properties from the same grid locations. Although unconventional, the ‘O’ shaped sample for chemical analyses provides an average of soil properties at the grid location with little of the overall volume being absent from removal of the middle core. Using this method, samples were taken from two depths corresponding to both organic O and mineral B horizon levels. The O horizon cores were inserted with their upper edge 5 cm below the O horizon upper surface and those for the B horizon 10 cm below the upper boundary of the B horizon at the given XY location. Given the thinner accumulation of organic horizon in the Placaquod care was taken to not to include any of the underlying mineral material in the O horizon samples. Total samples numbered 104 from the O horizon and 104 from the B horizon at each soil plot. Due to the thin irregular nature of the A horizons these were not sampled. Samples from each plot were analyzed immediately after collection on each of the 2 yr.



View larger version (14K):
[in this window]
[in a new window]
  		Fig. 2. Sample designs of (a) Typic Placaquod and (b) Typic Humaquept soil plots.

 
Analytical Methods
Bulk samples for chemical analyses were individually air-dried (30°C) and hand sieved to <2 mm. This may have changed the sample support since some variation would be expected in stone contents between samples, although other examples of geostatistical interpretation of properties of <2 mm sieved material have been presented (e.g., Cambardella et al., 1994; Mallarino, 1996). The subsequent sieved material was used for all chemical analyses so the support remained common between determinants. Measurement of pH (in 0.01 mol L–1 CaCl2) used 1:4 and 1:2.5 w/v pastes for organic and mineral soils, respectively. Exchangeable base cations were extracted by shake and filter techniques using neutral 1 mol L–1 ammonium acetate (Thomas, 1982). Concentrations of Na and K in the extractant were determined by photometry and of Ca and Mg by atomic absorption spectroscopy, employing suitable matrix matching. Exchangeable acidity was determined by titration of neutral 1 mol L–1 barium acetate extracts against standardized barium hydroxide (Parker, 1929). Method blanks were employed for all determinations on each batch of samples. The total CEC was taken as the sum of exchangeable bases plus exchangeable acidity. Results were expressed on an oven-dried soil basis. Organic matter content was determined by loss on ignition (30 min at 375°C then 860°C for 6.5 h) after Ball (1964).

Bulk density and hydraulic measurements were made on the intact core samples. Soil water release characteristics determined on undisturbed core samples using tension tables and pressure plates were used to calculate the fractions of the pore volumes that comprised transmission pores (termed macropores, >50 µm) or storage pores (termed micropores, <50 µm). Saturated hydraulic conductivity was measured on each core using the constant head method (Klute, 1965).

Statistical Methods
The main descriptive statistics were analyzed on raw data to examine means and distributions (data in Tables 2 and 3). Exchangeable base cations were also calculated on the basis of their proportion of the total exchange capacity (expressed as percentages) since this weighting allows simpler comparisons between soils of contrasting CEC (e.g., Billett and Cresser, 1996). Data were tested for normality using Anderson-Darling normality testing at p ≥ 0.05 (e.g., Stephens, 1974). To satisfy the requirements of normality for parametric and geostatistical analyses non-normal data were transformed by Box–Cox techniques (Box and Cox, 1964) and significant differences between populations then determined using unpaired t tests. For Ca and Na in mineral horizons these transformations were performed on n' = n + 0.001 to remove zero values. For chemical data (Table 4), the above detailed transformations were necessary on all mineral soil layer data, while in the case of organic soil layers transformations were only necessary on exchangeable acidity and CEC. For soil physical data (Table 5) transformations were required on all data sets with the exception of transmission pore values for the organic horizons.


View this table:
[in this window]
[in a new window]
  		Table 2. Comparisons of summary statistics (on raw data) for organic horizons of Placaquod and Humaquept plots (n = 104 at each site). Percentage base cations and total base saturation is weighted to a fraction of the total exchange capacity.

 

View this table:
[in this window]
[in a new window]
  		Table 3. Comparisons of summary statistics (on raw data) for mineral horizons of Placaquod and Humaquept plots (n = 104 at each site). Percentage base cations and total base saturation is weighted to a fraction of the total exchange capacity.

 

View this table:
[in this window]
[in a new window]
  		Table 4. Two-sample t tests of soil layer differences between Placaquod and Humaquept plots (n = 104). Details of transformations used are given in the methods.

 

View this table:
[in this window]
[in a new window]
  		Table 5. Comparison of physical and hydraulic soil properties between the Placaquod and Humaquept, given as means (±1 s), with two-sample t tests of soil layer differences between plots.

 
Variograms were constructed to investigate the average rate of change in properties with two-dimensional distance using the software GS+ (Gamma Design Software, Plainwell, MI). The numbers of paired comparisons used at each of the nine lag distance classes (Table 6) are comparable between the two sampling designs. Geostatistical models were chosen on the basis of best fit to the data set using r2 values. The effective range, defining the distance over which samples have spatial dependency, was determined where either a spherical or exponential model best described the data. Although exponential models do not reach a fixed sill variance and finite range, parameters may be defined on the point by which, for practical purposes, the variance stops increasing (Webster, 1985). The spatial dependency of the data was assessed by expressing the nugget variance or random variance (the variance at a lag of zero, see e.g., Webster, 1985) as a percentage of the total variance (e.g., Cambardella et al., 1994). Examples of geostatistical analyses performed on percentage, pH, and transformed data can be found in the literature (e.g., Cambardella et al., 1994; Robertson et al., 1993; Saldãna et al., 1998; Goderya, 1998). Anisotropy was not investigated due to limitations of sample numbers. Geostatistical interpretations of the hydraulic and physical properties of the plots are beyond the scope of this present study.


View this table:
[in this window]
[in a new window]
  		Table 6. Numbers of paired comparisons at each mean lag distance in the geostatistical analyses of Placaquod and Humaquept soil plots.

 

   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Exchange Chemistry Comparisons between Sites and Horizons
Summary statistics of soil pH and exchange data in organic horizons (Table 2) and mineral horizons (Table 3) allow comparisons between different soil horizons and between plots. Organic and mineral horizon soils differed significantly (t tests, p ≤ 0.001) in all chemical variables at both plots. At both sites mean basic and acid cation concentrations and base saturation were considerably higher in the organic horizons. Mean soil pH values were lowest in organic horizons, and had a smaller range than for the mineral horizon. Mean total base saturation for the organic and mineral horizon soils were approximately 9 and 2% respectively; these low values reflecting the acid, nutrient deficient nature of the parent material.

In both soil horizons of the Humaquept mean exchangeable base cations occurred in the order of abundance Mg > Ca > K > Na. Although the same order of abundance was observed for the Placaquod organic horizon, that of the Placaquod mineral subsoil differed (K > Mg > Na > Ca). Previous analysis of bulk soil mineralogy adjacent to the Pacaquod plot showed a dominance of quartz (51%), with 26% K-feldspar, 15% Na-Ca feldspar albite, and 3.8% of dioctahedral phyllosilicates (M. Stutter,unpublished data, 2001), probably comprising kaolinite, mica and illite (Bain et al., 1994). Such mineral compositions are most likely to contribute K and Na through weathering, accounting for the increased prevalence of K with depth in the Placaquod. Unfortunately an analysis of the variation of mineral compositions across and between the plots was beyond the scope of the present study. In nutrient-deficient upland systems however, base cation inputs may be largely derived from atmospheric, rather than weathering sources (Stutter et al., 2003). Atmospheric inputs comprise dominantly marine-derived components (Na+ and Mg2+), although atmospheric sources of Ca2+ have been previously observed to exceed those from weathering in similar soils in this region (Stutter et al., 2003). Distributions of divalent cations down the profile, especially from deposition sources, may reflect the nature of the exchange material since they make more stable associations with organic material than monovalent cations.

Consistent differences were observed in the spread of the data between O and B soil horizons at each site. Surface organic horizon chemical data exhibited distributions close to normality with departures tending to negative skewness. In contrast mineral horizon chemical data showed marked deviations from normality with consistent positive skewness. Organic horizon pH data (Table 2) showed significant differences (p ≤ 0.001) between the Placaquod and Humaquept, with a higher mean, range, and positive skewness in the Placaquod O horizon pH. Exchangeable acidity and CEC were significantly different (p ≤ 0.001) between the sites, with Humaquept values being strongly negatively skewed compared with those of the Placaquod. Table 4 shows that differences in concentrations between O horizons were highly significant for K and Mg, weakly significant for Ca and insignificant for Na. However, the between-site difference in exchange-weighted percentage of Na was highly significant (p ≤ 0.001), while percentage of Ca was nonsignificant. Base cation concentrations were generally higher in the Humaquept O horizon, especially for Mg.

The mean, range, and degree of positive skewness of chemical properties were higher for the Placaquod mineral horizon compared with the Humaquept mineral horizon (Table 3). Differences in the data between sites were generally significant (Table 4), with the exceptions of exchange acidity, CEC, and percentage of total base saturation. The significance of differences in concentrations and exchange-weighted proportions of basic cations showed greater statistical differences between Placaquod and Humaquept mineral horizons than when surface horizons were compared. Of the base cation concentrations, the least significant differences were again observed for Ca. Mean concentrations of Ca and Mg were higher in the Humaquept, while monovalent basic cations had higher mean concentrations in the Placaquod plot. While sample distributions for percentages of Ca and Mg showed a high degree of skewness and kurtosis for the Placaquod plot, conversely the same characteristics of non-normality were observed in the Humaquept data for the concentrations of these cations and not for the exchange-weighted proportions.

Figure 3 summarizes exchangeable base cation compositions in the different horizons of the Humaquept and Placaquod plots and highlights differences in the spread of the data. The cations Na and K are grouped to reflect the difference between divalent and monovalent ionic species. Compared with the Placaquod, the Humaquept shows a greater spread of cation compositions, especially for the organic horizon. The Placaquod B horizon soil chemistry has greater contributions of (Na + K) to the overall cation compositions and plots much closer to the mean cationic composition of precipitation (Fig. 3) for this region (AEA Technology plc., 2002), perhaps suggesting a greater degree of leaching with solutions of similar composition to rain water.



View larger version (46K):
[in this window]
[in a new window]
  		Fig. 3. Triangular diagrams contrasting base cation distributions between plots and horizons of (a) Placaquod and (b) Humaquept soils. Cation proportions are expressed on a molc basis. The precipitation chemical composition is given for comparison ({square}).

 
Burt and Park (1999) previously described variability in exchangeable Ca and Mg across a granitic hillslope (Southwest England). They attributed minimal variation of Ca and Mg in surface soils to similarities in behavior as nutrient cations, with tight cycling in vegetation and uniform atmospheric and litter fall inputs. Greater variability in lower horizons was ascribed to lateral drainage, with lower concentrations in leached soils upslope and enrichment downslope. Data from the present study, however, suggests the contrasting behavior of Ca and Mg, with Mg demonstrating considerably more variation between the sites. Differences in surface horizon CEC, exchangeable acidity, K and Mg, and to a lesser degree Ca between plots, may arise due to differences in the nature of organic material as a result of varying litter decomposition rates and vegetation types and differences in Mg particularly likely related to varying nutrient dynamics.

Overall variability within the plots can also be assessed against values given by Burt and Park (1999) by comparing CV. These authors found that variability of exchangeable base cations, in their case at the whole hillslope scale, decreased in the order Mg > Ca > Na > K (43, 29, 26, and 23%, respectively) with a low variability for CEC (15%), reported as being comparable with values from other studies. Mulder et al. (1991) also demonstrated comparable variability at a 100-m grid scale for a spodosol site in Norway. The present study demonstrates considerably larger variability at a smaller scale. Such high percentage CV values, being especially pronounced for Ca, are comparable with those reported for agricultural situations where high variability is attributed to differences in management practices (e.g., Goderya, 1998). Within-plot variability for all components is much greater for mineral than organic horizons. Cation-exchange capacity was also considerably more variable than the 15% value reported by Burt and Park (1999), with values up to 42% for the Placaquod mineral horizon. Manderscheid and Matzner (1995) also demonstrated the variability in soil solution cations across a 25 by 25 m grid plot on granitic parent material in Germany. In their case the site comprised mature forestry (Picea abies), which likely accounts for the greater variability, with CV values of 41, 132, 57, and 42% for solution concentrations (35 cm depth) of Na, K, Ca, and Mg, respectively. Geostatistical investigation of their data showed purely random variation at their scale of observation (although at n = 59).

Comparisons of Physical and Hydraulic Properties
Soil physical and hydraulic characteristics are given in Table 5, showing statistical testing between properties for common horizons between plots. The mean organic matter content of the B horizon was <10% at either site, with statistically no significant difference between B horizons of the two sites. Mean bulk densities and transmission pore-size abundances showed significant differences between the Placaquod and Humaquept plots in the B horizon only. Highly significant differences were observed between O horizons in terms of organic matter content and saturated hydraulic conductivity, mean values of which were both greater in the Humaquept. Within-site variability in properties was similar between common horizons of the soils with the exception of the residual pore class composition of the mineral horizons, which demonstrated a greater standard deviation for the Humaquept. In both soils micropores dominated the pore composition, especially in organic horizons, indicating potential for long residence times of waters in surface soils at both plots. Lower horizon pore-size distributions show that the Placaquod exhibited a slightly lower mean proportion of micropores (69%) than the Humaquept (74%). Mean total porosities were approximately 80% in organic and 50% in mineral horizons. Despite this, mean hydraulic conductivity was commonly higher in B than O horizons.

Variance Structure of Soil Exchange Properties
Geostatistical information is summarized in Table 7 and selected variograms are depicted in Fig. 4 for Placaquod O and B horizons in (i) and (ii), respectively, and Humaquept O and B horizons in (iii) and (iv), respectively. Percentage of nugget values show marked differences in the degrees of spatial dependence for variables between both sites and horizons. Cambardella et al. (1994) previously defined percentage of nugget values of <25, 25 to 75, and >75 as categories of strong, moderate, and weak spatial dependence, respectively, and this scheme is also used in the discussion below. Strong spatial dependence was exhibited by all variables for the Placaquod organic horizon and generally strong to moderate spatial dependence for the B horizon variables. Fits of spherical and exponential models to the data were generally very good for the Placaquod horizons. By contrast, many variables of the Humaquept were described by linear variograms, with varying degrees of scatter. These exhibited pure nugget variance, indicating those variables to be spatial independent, with purely random variance. Exceptions to this were the concentrations of H ions (derived from the pH) and exchangeable Na and acidity in the O horizon and the exchangeable acidity, CEC, and percentage of K in the B horizon, all of which exhibited moderate spatial dependence as described with exponential models.


View this table:
[in this window]
[in a new window]
  		Table 7. Geostatistical analysis of variability within the two horizons of Placaquod and Humaquept sites.

 


View larger version (32K):
[in this window]
[in a new window]
  		Fig. 4. Variograms for selected properties of (i) O and (ii) B horizons of the Placaquod and (iii) O and (iv) B horizons of the Humaquept.

 
In the Placaquod B horizon, H ions (pH), base cation concentrations, CEC, and exchangeable acidity generally had lower ranges of spatial dependency than the O horizon. This showed that spatial dependence was confined to shorter separations of samples in the mineral material for these variables. The exception was exchangeable Ca concentration, for which a moderate spatial dependence was indicated up to 6.9 m in the Placaquod B horizon. This contrasted with Placaquod B horizon Mg concentrations, for which a much stronger spatial dependence was indicated over a much smaller range (0.9 m).

Moderate spatial dependence indicated for variables of exchangeable Na and acidity in the O horizon of the Humaquept had small ranges when compared with the same properties of the Placaquod O horizon. However, the fits of the Na and acidity models to the Humaquept O horizon data (Fig. 4; Table 7) were poor (r2 < 0.3); hence, the results must be interpreted with caution. A long range value to Humaquept O horizon H ion concentrations (pH derived) results from the exponential model, which shows that variance increases in line with lag distance to over twice the distance of spatial dependence indicated for the Placaquod O horizon. Fitted models for exchangeable acidity and CEC showed moderate spatial dependence in the Humaquept B horizon data (supported by r2 values of 0.6) and had larger range values than in the Placaquod B horizon. Several of the variograms for the Humaquept that were best described by linear models (e.g., those for Mg concentrations in either horizon) demonstrated higher variance associated with smaller average lag distances and these were assigned a pure nugget linear fit.

Considering differences in the nature of the exchange complex of organic and mineral material, the prevalence of biological cycling of nutrients nearer the soil surface and other hydrological factors it is not surprising that vertical differences in exchange chemistry between O and B horizons exceeded those between common horizons across the two soil units. The two horizons are not unrelated however, due to the nature of eluviation–illuviation processes and it has been noted that vertical anisotropy is often much greater than lateral anisotropy in many soil attributes (Wilding, 1984). Such observations have led to the development of statistical methods aiming to describe soil properties in three-dimensions (Park and Vlek, 2002). Since both soils exhibit morphological characteristics of leaching (Table 1) patterns in spatial variability may potentially be transferred through the profiles depending on hydrological conditions and transport mechanisms. However, relationships between horizons in the spatial structure of variance were not clear for the Placaquod, while in the Humaquept variance was randomly structured for both horizons. The shorter ranges observed to the spatial dependence of the majority of chemical variables of the Placaquod B horizon compared with the O horizon may reflect differences in the scales of preferential flow pathways. Observations at this site showed the development of vertical crack structures in Placaquod organic horizons prone to seasonal drying, which may impart a coarser network of transport pathways to the surface soil than that of the mineral B material. Despite the relatively uniform composition of the granite dominated parent material, dissolution of spatially distributed deposits of calcite that have been observed associated with fractures and veins in similar geology (e.g., Bain et al., 1994) may offer a possible explanation for the different spatial behavior of exchangeable Ca in the Placaquod B horizon compared with other cations. However, the presence of rapidly weathering Ca-bearing minerals is not supported by higher mean exchangeable Ca concentrations in Placaquod B horizon material.

The linear ‘pure nugget’ fit of models to many of the Humaquept properties, implies no spatial correlations at the sampling separations used. This element of pure randomness is very distinct from the stronger spatial dependence of the Placaquod properties and may correspond to interrelated differences in conditions of hydrology and soil formation between sites. Mechanisms impacting on soil formation that may contribute to differences in exchangeable cations are local-scale variability in parent material or vegetation, the effects of which cannot be assessed with the present information. Nutrient cycling differences may explain differences in the variogram structure for cations between the organic horizons of the two soils (especially since the non-nutrient cation Na shows similarities between the sites), it is unlikely that this could explain the random nature of variation in the Humaquept subsoil.

It should be recognized that the Humaquept site might have been an accumulation area for colluvium from upslope. The present structure of the soil may be impacted by mixing of material during transport, resulting with deposits comprising the present day subsoil being more evenly laterally redistributed over the area of the Humaquept plot. Depending on the processes of colluviation this may also have imparted a degree of vertical fractionation according in terms of particle size. A possible further explanation for random structured variance may relate to soil water conditions at the Humaquept site. Although the Humaquept B horizon shows relatively bright oxidized colors to B horizon material, the site was much wetter through most of the year, even during the July period of sampling. While both sites probably experience periods of downward leaching, it is likely that the Humaquept site becomes waterlogged during winter-spring by virtue of its slope foot position. The site may then become saturated by events of upwelling ground water as a result of hydraulic pressure from upslope. The saturated pore-spaces then would provide a means of connectivity by which exchange sites could equilibrate and, with the high hydraulic conductivity, distribute cations more uniformly throughout the exchange complex. However, localized upwelling of ground waters enriched in base cations is not supported at the Humaquept plot by elevated subsoil base cations, or base saturation, which are similar between plots (Table 3).

Geostatistical approaches are highly sensitive to the scale of the experimental observations (e.g., Webster, 1985); hence, the Humaquept plot of pure nugget variances indicate no spatial dependency only at the sampling scale we have used here. Consideration of properties at larger sampling scales may have allowed links to be made between chemical variables and factors such as vegetation, terrain parameters, soil depth and texture, and parent materials.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Both conventional statistical and geostatistical approaches show clear differences in soil exchange properties between the Placaquod and Humaquept. This highlights the importance of addressing differences in broader soil units when assessing variability in exchange chemistry at finer scales. Comparisons between the soil types showed similarities in acid and base saturation. Differences were however observed in the composition of constituent base cations between plots, with the Humaquept having greater concentrations and proportions of divalent cations, in particular Mg. These differences were most pronounced in mineral horizons, where Ca2+ and Mg2+ were observed to have different characteristics of variability. Placaquod mineral layers showed a greater dominance of Na+ and K+, most likely resulting from more active current leaching. Many of the geostatistical analyses of the Humaquept showed pure nugget variance, suggesting that no spatial correlations were found at the sampling separations used. In contrast, the strong spatial dependence for variables at the Placaquod plot suggests a ‘hot-spot’ distribution pattern for many important chemical properties. The differences in spatial structure of the variance between horizons of the Placaquod are important considerations for deterministic catchment models, many of which are driven by vertical changes in soil chemistry, with horizonated soil inputs.

The data provided is not sufficient to allow all factors of soil formation affecting cation exchange compositions between the sites to be addressed and conclude what processes control the differences in variance structure between these soils. However, important differences have been shown that have implications for soil sampling strategies aiming to characterize soil heterogeneity. Descriptions of soil variability are important to improve the accuracy of distributed models and reduce the time and effort of data gathering in future studies. Our results show that, at the current scale of observations, the characterization of the variability in exchange chemistry for these particular soils required a considerable number of samples in both cases. However, consideration of the spatial orientation of the strategy would only yield extra information about the variance in the Placaquod. Increasing the scale of observations of such properties for these soils would address the implications of how these seemingly distinct soil units fit into soil chemistry variation at the catchment level. Future assessment of whether the observations of exchange chemistry are translated into the chemistry of soil solutions would be highly beneficial to stream-flow generation and water quality modeling.


   ACKNOWLEDGMENTS
 
The authors acknowledge the financial support of the Scottish Executive Environment and Rural Affairs Department for the funding of this consortium research project undertaken between the partners Aberdeen University, Macaulay Institute and the Scottish Crop Research Institute.

Received for publication September 3, 2002.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Stutter, M. I.
Right arrow Articles by Billett, M. F.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Stutter, M. I.
Right arrow Articles by Billett, M. F.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Stutter, M. I.
Right arrow Articles by Billett, M. F.
Related Collections
Right arrow Spatial Variability
Right arrow Soil Analysis
Right arrow Soil Chemistry

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