Elsevier

Geoderma

Volume 116, Issues 1–2, September 2003, Pages 61-76
Geoderma

Effect of soil organic carbon on soil water retention

https://doi.org/10.1016/S0016-7061(03)00094-6Get rights and content

Abstract

Reports about the relationship between soil water retention and organic carbon content are contradictory. We hypothesized that this relationship is affected by both proportions of textural components and amount of organic carbon. To test the hypothesis, we used the U.S. National Soil Characterization database and the database from pilot studies on soil quality as affected by long-term management. Regression trees and group method of data handling (GMDH) revealed a complex joint effect of texture and taxonomic order on water retention at −33 kPa. Adding information on taxonomic order and on taxonomic order and organic carbon content to the textural class brought 10% and 20% improvement in water retention estimation, respectively, as compared with estimation from the textural class alone. Using total clay, sand and silt along with organic carbon content and taxonomic order resulted in 25% improvement in accuracy over using textural classes. Similar but lower trends in accuracy were found for water retention at −1500 kPa and the slope of the water retention curve. At low organic carbon contents, the sensitivity of the water retention to changes in organic matter content was highest in sandy soils. Increase in organic matter content led to increase of water retention in sandy soils, and to a decrease in fine-textured soils. At high organic carbon values, all soils showed an increase in water retention. The largest increase was in sandy and silty soils. Results are expressed as equations that can be used to evaluate effect of the carbon sequestration and management practices on soil hydraulic properties.

Introduction

Soil water retention is a major soil hydraulic property that governs soil functioning in ecosystems and greatly affects soil management. Data on soil water retention are used in research and applications in hydrology, agronomy, meteorology, ecology, environmental protection, and many other soil-related fields. Soil water retention is measured in some soil survey programs. However, these measurements are impractical at the design stage of some projects, as well as in large-scale applications, and water retention needs to be estimated from other soil properties available from soil survey. Regression equations for such estimation are often called pedotransfer functions (PTF). Extensive research has shown that water retention is a complex function of soil structure and composition Rawls et al., 1991, Wösten et al., 2001. Soil organic matter content and composition affect both soil structure and adsorption properties; therefore, water retention may be affected by changes in soil organic matter that occur because of both climate change and modifications of management practices. Thus, effects of organic matter on soil water retention should be understood and quantified.

Reports on the effect of changes in soil organic matter on soil water retention are contradictory. Table 1 summarizes findings of different authors. Rawls and Brakensiek (1982) and Rawls et al. (1983) found useful to include the organic carbon content in the list of PTF inputs for both −33 and −1500 kPa. Bell and van Keulen (1995) saw the need to use both organic carbon content and pH in estimating water content at wilting point. Beke and McCormick (1985) and Petersen et al. (1968) found it useful to employ data on organic matter content to estimate water content at −1500 kPa, but not at −33 kPa. In contrast, the use of organic matter content improved PTFs at −33 kPa, but not at −1500 kPa, in the work of Calhoun et al. (1973). Viville et al. (1986) indicated that the differences in water retention within soil profiles correlated with profiles of the organic matter content. Hollis et al. (1977) found the organic matter content to be the most influential soil variable to estimate water content at −5 kPa. Lal (1979) and Danalatos et al. (1994) did not find any effect of organic matter content on water retention; the latter attributed that to the generally low organic matter content in their samples. Similarly, Puckett et al. (1985) did not use organic matter in PTFs because of its low level in samples. Bauer and Black (1981) found that the effect of organic carbon on water retention in disturbed samples was substantial in sandy soil and marginal in medium- and fine-textured soils. De Jong (1983) experimented with disturbed soil samples and found that the increase in organic matter content meant higher water content at all suctions. Similar observations were made by several other authors Jamison and Kroth, 1958, Petersen et al., 1968, Riley, 1979, Ambroise et al., 1992, Kern, 1995. Salter and Haworth (1961) argued that organic matter might not be an important predictor to estimate water content at specific suctions, but it is an important factor if water contents at field capacity and wilting point are measured directly. McBride and MacIntosh (1984) found that organic matter content affected water retention at −1500 kPa only when this content was larger than 5%. Most of the cited authors worked with small number of soils from a specific region. Kay et al. (1997) compared relative effects of organic matter on water retention using PTFs developed in different regions and found large regional differences.

The review of the existing studies of the effect of organic carbon on soil water retention begets a hypothesis that this effect may depend on proportions of textural components and amount of organic carbon. The objective of this work was to test the hypothesis using the massive U.S. National Soil Characterization. Database and the database from pilot studies on soil quality as affected by long-term management.

Section snippets

Soil data sets

A subset of about 12,000 samples was extracted from the National Soil Characterization database (Soil Survey Staff, 1995). The samples had data on soil texture, organic matter content, water retention at −33 kPa and −1500 kPa, bulk density at −33 kPa, and a taxonomic characterization. An overview of properties of the subset is presented in Fig. 1. Sandy loams, loams, and silt loams were represented best and together constituted more than 60% of all samples. Silts, sands, sandy clay loams, and

Regression trees

The regression tree for the soil water content at −33 kPa is shown in Fig. 2. Soil textural class and organic carbon content are predictor variables. The first split divides soils by their textural class. Sands, loamy sands, and sandy loams form one large group, and soils with finer texture form another one. The organic carbon content (Corg) is the most important splitting variable in coarse-textured soils, the critical value is 2.1%. Finer soils are further split in the group with fine texture

Discussion

Organic carbon content appeared to be an important soil property to improve estimation of soil water retention from soil texture. Including Corg in regressions improved accuracy of regression tree (Table 1) and GMDH predictions. The 15% and 10% decrease in RMSE were achieved at −33 and −1500 kPa, respectively. This may be related to the fact that the structure-forming effect of organic matter is affecting the water retention at water content close to field capacity to larger extent than water

Conclusion

(1) Relationship of soil water retention to organic carbon content is affected by proportions of textural components.

(2) Soil water retention at −33 kPa is affected more strongly by the organic carbon than water retention at −1500 kPa.

(3) Water retention of soils with coarse texture is substantially more sensitive to the amount of organic carbon as compared with fine-textured soils.

(4) The effect of changes in organic carbon content on soil water retention depends on the proportion of textural

References (41)

  • F.G. Calhoun et al.

    Influence of particle size and organic matter on water retention in selected Florida soils

    Soil Crop Sci. Fla. Proc.

    (1973)
  • L.A. Clark et al.

    Tree-based models

  • B.T. Cristensen

    Carbon in primary and secondary organomineral complexes

  • R. De Jong

    Soil water desorption curves estimated from limited data

    Can. J. Soil Sci.

    (1983)
  • S.J. Farlow

    The GMDH algorithm

  • D. Gimènez et al.

    Prediction of a pore distribution factor from soil textural and mechanical parameters

    Soil Sci.

    (2001)
  • P.I. Good

    Resampling Methods: a Practical Guide to Data Analysis

    (1999)
  • S.C. Gupta et al.

    Hydraulic and thermal properties of a sandy soil as influenced by incorporation of sewage sludge

    Soil Sci. Soc. Am. J.

    (1977)
  • R. Hecht-Nielsen

    Neurocomputing

    (1990)
  • V.C. Jamison et al.

    Available moisture storage capacity in relation to texture composition and organic matter content of several Missouri soils

    Soil Sci. Soc. Am. Proc.

    (1958)
  • Cited by (780)

    View all citing articles on Scopus
    View full text