| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Wageningen Univ., Farm Technology Group, P.O. Box 17, 6700 AA Wageningen, The Netherlands
b Wageningen Univ., Dep. of Soil Quality, P.O. Box 47, 6700 AA Wageningen, The Netherlands
* Corresponding author (ana.lima{at}wur.nl; anacrlima{at}hotmail.com).
Received for publication July 19, 2006.
ABSTRACT
Soil quality, as a measure of the soil's capacity to function, can be assessed by indicators based on physical, chemical, and biological properties. Here we report on the assessment of soil quality in 21 rice (Oryza sativa) fields under three rice production systems (semi-direct, pre-germinated, and conventional) on four soil textural classes in the Camaquã region of Rio Grande do Sul, Brazil. The objectives of our study were: (i) to identify soil quality indicators that discriminate both management systems and soil textural classes, (ii) to establish a minimum data set of soil quality indicators and (iii) to test whether this minimum data set is correlated with yield. Twenty-nine soil biological, chemical, and physical properties were evaluated to characterize regional soil quality. Soil quality assessment was based on factor and discriminant analysis. Bulk density, available water, and micronutrients (Cu, Zn, and Mn) were the most powerful soil properties in distinguishing among different soil textural classes. Organic matter, earthworms, micronutrients (Cu and Mn), and mean weight diameter were the most powerful soil properties in assessing differences in soil quality among the rice management systems. Manganese was the property most strongly correlated with yield (adjusted r2 = 0.365, P = 0.001). The merits of sub-dividing samples according to texture and the linkage between soil quality indicators, soil functioning, plant performance, and soil management options are discussed in particular.
IN the state of Rio Grande do Sul, Brazil, rice (Oryza sativa) production is one of the most important regional activities. Rice production is located mainly in the southern lowlands where approximately 5.5 million tons of rice per year are produced, equivalent to 52% of total Brazilian rice production (Azambuja et al., 2004). The inherent fertility of the region has led to the expansion of rice cropping and an increase in land use intensity, mainly in the Camaquã region, over the last half century (Westphal, 1998; Cunha et al., 2001). There is a growing concern with farmers that the land use practices in the Camaquã region may not be sustainable because of their detrimental effects on soil quality.
The most commonly used definition of soil quality is: "the capacity of a specific kind of soil to function, within natural or managed ecosystem boundaries, to sustain plant and animal productivity, maintain or enhance water and air quality, and support human health and habitation" (Karlen et al., 1997). This capacity of the soil to function can be assessed by physical, chemical, and/or biological properties, which in this context are known as soil quality indicators (Wander and Bollero, 1999). Perceptions of what constitutes a good soil vary. They depend on individual priorities with respect to soil function, intended land use, and interest of the observer (Doran and Parkin, 1994; Shukla et al., 2006). Soil quality changes with time can indicate whether the soil condition is sustainable or not (Arshad and Martin, 2002; Doran, 2002). Maintaining soil quality at a desirable level is a very complex issue due to climatic, soil, plant, and human factors and their interactions and it is especially challenging in lowland rice cropping systems because of puddling practices in soil preparation (Chaudhury et al., 2005). Although the concept of soil quality is often advocated (Karlen et al., 2001), it is also criticized in the literature because of "its premature acceptance and institutionalization of an incompletely formulated and largely untested paradigm" (Sojka and Upchurch, 1999).
Minimum data sets of soil quality indicators have been proposed for plot and field scales (Doran and Parkin, 1996), regional scales (Brejda et al., 2000a, 2000b), and for national scales (Sparling and Schipper, 2002, 2004; Sparling et al., 2004). However, there is currently no consensus on a definitive set of soil properties for soil quality monitoring, nor consensus on how the indicators should be interpreted (Schipper and Sparling, 2000). This lack of consensus is partly due to the fact that soil quality is a complex concept and that different site-specific soil conditions may be desirable, depending on the purpose of land use.
The assessment of soil quality can be viewed as a primary indicator of the sustainability of land management (Doran, 2002). Basically, two types of approaches are employed for evaluating the sustainability of a management system: (i) comparative assessment and (ii) dynamic assessment (Larson and Pierce, 1994). The comparative approach has frequently been used and has shown that multivariate statistical analyses are useful tools to identify indicators and to interpret correlations of indicators (e.g., Wander and Bollero, 1999; Brejda et al., 2000a, 2000b; Govaerts et al., 2006; Shukla et al., 2006). These studies are based on different land uses in a single dominant soil group (Brejda et al., 2000b) or in different soil great groups (Brejda et al., 2000a; Schipper and Sparling, 2000; Sparling and Schipper, 2002) or based on the same land use under different management systems (Chaudhury et al., 2005; Govaerts et al., 2006). Our study relates to one land use (rice production) under three different management systems, on two soil types, suggesting that multivariate analyses of sample data would be appropriate in this case.
To make the results of our study useful for dynamic assessment of soil quality, we used a novel approach. Farmers in the study area mainly evaluate the production potential of the soils according to texture. We sub-divided the samples into four textural classes according to clay and therefore focus the study on those soil quality indicators that are interpretable by farmers. Our approach establishes a minimum data set of soil quality indicators which is able to show not only human effects (as a result of management systems applied by farmers), but also differences due to inherent soil characteristics (soil texture classes). We also analyzed a larger pool of soil properties than is commonly the case.
The study was conducted with the following objectives: (i) to identify soil quality indicators that discriminate both management systems and soil textural classes, (ii) to establish a minimum data set of soil quality indicators, and (iii) to test whether this minimum data set is correlated with yield.
Material and Methods
Area Description and Soil Sampling
Camaquã is located in the south of Brazil, in the Rio Grande do Sul state, between latitude 30°48' and 31°32' S, and longitude 51°47' and 52°19' W (Fig. 1
). Mean annual rainfall is 1213 mm and the average temperature is 18.8°C (Cunha et al., 2001).
|
|
Soil Analysis
Samples collected for microbiological analysis were placed in a cooler with ice packs for transport to the laboratory. The samples were analyzed for microbial biomass (MB), soil respiration (SR), potentially mineralizable N (PMN), β-glucosidase (ßG), acid phosphatase (ACP), and alkaline phosphatase (ALP). Microbial biomass was determined using microwave irradiation of soil (Islam and Weil, 1998). Flow-through respirometry was used to measure SR, using an automatic respirometer with a CO2 analyzer (Sable System, Las Vegas, NE, USA). Potentially mineralizable N was measured using the ammonium production during waterlogged incubation (Bundy and Meisinger, 1994). Enzyme activity was measured using a spectrophotometer according to Tabatabai (1994). Earthworm number (EN) was assessed using the standard method of the Tropical Soil Biology and Fertility Program (TSBF) (Anderson and Ingram, 1993).
Chemical analysis was done using the methodology described by Tedesco et al. (1995). Samples were analyzed for organic matter (OM) using the Walkley–Black method, total N (TN) using the Kjeldahl method, pH (1:1, soil/H2O), and Al saturation (Al sat = 100* (exchangeable Al)/(exchangeable Al + Sum of bases)). Exchangeable Ca, Mg, and Al were extracted by 1 mol L–1 KCl. Ca and Mg were determined by atomic absorption spectrometry and Al by titration with NaOH. Extractable P and exchangeable K were extracted by a Mehlich 1 solution (0.05 mol L–1 H2SO4 + 0.05 mol L–1 HCl). Phosphorus was determined by UV-visible spectrophotometry and K by flame photometry. Potential acidity (PA) was estimated by SMP buffer solution and cation exchange capacity (CEC) as sum of bases + (PA). Micronutrients (Fe, Zn, Cu, Mn) were determined by atomic absorption spectrometry. Manganese was extracted by 1 mol L–1 KCl. Zinc and Cu by 0.1 mol L–1 HCl and Fe by 0.2 mol L–1 ammonium oxalate at pH 3.0.
Bulk density (BD), texture, water stable aggregates (WSA), microporosity (MiP), and soil water retention on pressure plates at –340 and –1500 kPa were the physical analyses measured according to methods described by Klute (1986). The results from textural analysis (hydrometer method) were used to divide the soils into four soil textural classes according to clay content (<20%, 20–40%, 40–60%, >60%), following the standardized division of textural classes used to diagnose soil fertility in the state of Rio Grande do Sul (Sociedade Brasileira de Ciência do Solo, 2004).
Some indicators were calculated from the measured data set: available water (AW = difference between water content at field capacity and permanent wilting point), macroporosity (MaP = difference between total porosity and MiP), mean weight diameter (MWD = 
(mean diameter x aggregates weight)/sample dry weight)), and microbial quotient (Mq = the ratio of soil microbial biomass carbon to soil total organic carbon). These indicators have been suggested as useful for soil quality monitoring (Doran and Parkin, 1994; Schipper and Sparling, 2000).
Grain was collected (before soil sampling) and manually separated from the straw, weighed, and moisture content was determined. The final yield was calculated based on 13% moisture.
Statistical Analysis
Multivariate statistical analysis of the soil properties was conducted using factor analysis and discriminant analysis. Factor analysis was used to group the 29 soil properties into statistical factors (or principal components) to reduce the entire data set for subsequent discriminant analysis. Principal component analysis was used as the method of factor extraction and factors were subjected to varimax rotation. The general principles of principal component and factor analyses can be found in Webster and Oliver (1990) and guidance for summarizing data is provided by Webster (2001).
Stepwise discriminant analysis was used to select the component(s) that best discriminated among the different management systems and also among the different soil textural classes. Following selection of the best discriminating component(s), soil quality properties that comprised these components were also subjected to stepwise discriminant analysis to select the best set of properties for forming the discriminant functions. Discriminant analysis, therefore, proceeds with the derivation of the discriminant function and the determination whether a statistically significant function can be derived to separate the two or more groups (in our case the three rice management systems and the four soil textural classes).
Holdout method was employed as a technique to validate the results of discriminant analysis. For the methodological rationale related to the factor and discriminant analyses see also Sharma (1996) and Hair et al. (1998). All statistical analyses were conducted with SPSS 11.0 software (SPSS, 1998).
Results
Significant correlation (P < 0.01) was observed between 232 of 406 soil property pairs for the Camaquã samples (Table 2 ). The strongest positive correlations (r > 0.90) were observed for OM with TN, Ca, MiP, and CEC, and for CEC with Ca, PA, and MiP. The strongest negative correlations (r > 0.90) were observed for BD with OM, TN, Ca, MiP, and CEC.
|
90% of variance in BD, TN, CEC, MiP, and Al, and 80% of the variance in PA, ALP, ßG, Mg, MB, Cu, Zn, Al sat, and pH. However, the five components explained <60% of the variance in AW, EN, and MaP (Table 3).
|
0.95) of OM (0.97), TN (0.95), and Ca (0.95), and a high negative loading of BD (–0.95). It also had positive loadings of CEC, MiP, PA, ßG, ACP, ALP, SR, Mg, Fe, PMN, MB, WSA, MWD, and Al (>0.50). We identified this component as the "organic matter component" because all soil properties comprised in this component were significantly correlated (P < 0.01) with OM (Table 2). The second principal component explained 9.68% of the variance (Table 3) and was identified as the "micronutrients component" because it had the highest positive loading for Cu (0.77) and Zn (0.64). This component also had a positive loading for MWD (0.62) and a negative loading for AW (–0.61). Moderate positive loadings were found for K (0.43) and WSA (0.43), resulting from the significant correlation (P < 0.01) between K and Cu, Zn, MWD, AW, and WSA (Table 2). A moderate negative loading was found for EN (–0.45), resulting from the significant correlation (P < 0.05) between EN and Cu.
The third principal component was identified as the "acidity component." It explained 9.57% of the variance (Table 3) and had a high positive loading for pH (0.73) and a high negative loading for Al sat (–0.87) and Al (–0.76). These soil properties were grouped together because all three were significantly correlated (P < 0.01; Table 2).
The fourth principal component explained 9.01% of the variance (Table 3). It had the highest positive loading for Mn (0.74); therefore, it was identified as the "Mn component". This component had a moderate loading for K (0.59) resulting from the largest correlation between this property and Mn (0.60, Table 2). It also had a high and a moderate negative loading for Mq (–0.86) and MB (–0.60), respectively (Table 3).
The fifth principal component was identified as the "P component" because it had the highest positive loading for P (0.78); it also had a moderate positive loading for pH (0.47) and the highest negative loading for MaP (–0.72) (Table 3). These three soil properties were grouped together because the largest correlation of MaP was with P and the largest correlation of P was with pH (Table 2). This component explained 6.14% of variance (Table 3).
Selecting Soil Quality Indicators
Soil Textural Classes
A stepwise discriminant analysis based on the five principal components obtained from the factor analysis showed that only the first discriminant function was significant and explained 99.1% of the total variance. The components "Organic matter," "Micronutrients," and "Mn" were the most powerful discriminators of the four soil textural classes (Table 4
).
|
|
|
|
Yield
To investigate whether the soil quality indicators in the minimum data set are correlated with rice yield, a regression was performed with rice yield as dependent variable and the eight soil properties as independent variables (adjusted r2 = 0.365). The results (Table 6
) show that only Mn significantly predicts yield (P < 0.001).
|
We found the same set of components "Organic matter," "Micronutrients," and "Mn" to be the most important principal components discriminating among both management systems and textural classes. However, different sets of soil properties came out as indicators of soil quality related to management practices than those related to soil textural classes.
Compared to other studies (Wander and Bollero, 1999; Brejda et al., 2000a, 2000b; Shukla et al., 2006), we did not detect just one soil property with the greatest potential for monitoring regional soil quality. Instead, our minimum data set includes eight soil properties: three physical (AW, BD, and MWD), four chemical (OM, Zn, Cu, and Mn) and one biological (EN). However, if fewer soil properties were to be used to monitor soil quality in rice fields, as related to both management systems and textural classes, micronutrients, specifically Cu and Mn, appear to offer the greatest potential, also for improving management systems.
Our approach selected OM, Cu, EN, Mn, and MWD as the soil quality indicators to distinguish the soil management systems. Available water, BD, and micronutrients (Cu, Zn, Mn) were the soil quality indicators to distinguish the soil textural classes. Using principal component analysis, Chaudhury et al. (2005) also reported that MWD was an important physical property to be retained in their minimum data set, while Mn was not considered an effective indicator of soil quality for different rice growing practices in India.
Mangenese was the only soil property correlated with yield, which suggests that the Mn concentration of soils has a strong influence on rice production. This micronutrient also contributed significantly to differentiating the management systems and soil textural classes. The speciation of Mn in soil is extremely complex and involves both chemical and microbial interactions (Fageria, 2001) and studies about micronutrients in lowland soil under inundation are scarce and contradictory (Assis et al., 2000). No Mn fertilization recommendations are supplied to the local farmers as it is assumed that the majority of the regional soils contain an adequate supply of micronutrients, including Mn (Sociedade Brasileira de Ciência do Solo, 2004). Consequently, the soil laboratories do not perform any micronutrient analysis on a routine basis. The results of this study suggest that Mn may have to be considered as a factor determining the sustainability of rice production in the region.
Farmers in the study area evaluate the production potential of the soils based on texture. Bulk density, AW, Zn, Cu, and Mn are the soil quality indicators which were identified based on the textural classes. As Mn is also correlated with yield, it is probably sound to base farming practices on textural classes. None of the seven other soil indicators identified in the discriminant analysis for both soil textural classes and management systems could significantly predict yield. These other seven indicators provide information on the most basic soil functions: water infiltration, storage and supply (AW, BD, MWD, EN, OM), nutrient storage, supply and cycling (OM, micronutrients, EN, AW), and sustaining biological activity (OM, EN). As a corroboration, OM and EN are widely recognized as useful indicators of all these soil functions together. Beyond that, OM is the simplest, least expensive to measure and EN is an indicator that farmers can observe themselves.
The statistical approach used here avoided arbitrary choices in selecting indicators within the significant principal components and discriminant analysis. Without performing separate analyses based on textural class and management system, we would not have been able to decide whether the soil indicators would have been the consequence of inherent soil properties (i.e. texture) or management. The minimum data set, therefore, may provide an early warning tool to evaluate land management options, such as growing alternative crops and where to farm and buy land, or, in other words, to evaluate the sustainability of land use.
Further research is needed to validate our approach to arrive at the minimum data set in different regions, under different management and land use.
ACKNOWLEDGMENTS
This work was financially supported by CAPES (Fundação Coordenação de Aperfeiçoamento de Pessoal de Nível Superior- Federal Agency for Post-Graduate Education, Brazil). The authors thank Dr. Ron de Goede (Wageningen University) for his suggestions on an earlier version of this manuscript.
NOTES
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | ||||
| Soil Science Society of America Journal | Journal of Plant Registrations | The Plant Genome | |||
