Assessment of soil property spatial variation in an Amazon pasture: basis for selecting an agronomic experimental area
Introduction
Understanding the distribution of soil properties in the field is important in refining agricultural management practices (McBratney and Pringle, 1999) while minimizing environmental damage. Soil property variation within a field often has been described by classical statistical methods assuming a random distribution Goovaerts, 1999, Webster, 2000, Conant and Paustian, 2002. Natural soil spatial variation occurs primarily from pedogenetic factors (Trangmar et al., 1985). In addition, variation can occur as a result of land use and management Paz-González et al., 2000, Stenger et al., 2002. As a consequence, soils usually exhibit marked spatial variation on macro (White et al., 1997) and micro scales (Yang et al., 2001).
In many instances, spatial variation is not random but tends to decrease as distances diminish between points in space Goovaerts, 1998, Webster, 2000. Spatial dependence has been observed for a wide range of soil physical Mapa and Kumaragamage, 1996, Castrignano et al., 2000, chemical Boyer et al., 1996, Bragato and Primavera, 1998, Stenger et al., 2002 and biological properties Robertson et al., 1997, Goovaerts, 1998, Gaston et al., 2001, but typically the size of the studied area is relatively small, commonly ranging from 1 m2 to 1 ha.
Increasingly geostatistical techniques are being used in soil science for spatial variation studies on scales ranging from centimeters to kilometers White et al., 1997, Goovaerts, 1998, Castrignano et al., 2000, Yang et al., 2001. These techniques have provided the means to characterize and quantify spatial variation, have been used to process this information for rational interpolation, and have been applied to estimate the variance of interpolated values Isaaks and Srivastava, 1989, McBratney and Pringle, 1999, Webster, 2000, Gaston et al., 2001, Stenger et al., 2002.
The Amazon has the largest intact tropical forest in the world (Neill et al., 1997), but it has also the world's highest deforestation rate (Fearnside and Barbosa, 1998). One of the principal causes of deforestation in the Amazon forest has been the conversion of natural forest to cattle pastures, the main land use in deforested areas (Graça et al., 1999). However, 5 to 10 years after deforestation, pastures become degraded with low productivity mainly due to overgrazing and invasion of weeds (Fearnside and Barbosa, 1998). Other bad management practices have long been recognized as major causes of the in-site fertility reduction, as well as soil compaction, decrease in water supply, soil erosion and nutrient loss acceleration. Off-site consequences may include the increase of CO2 level with increasing organic matter mineralization as well as a decrease of groundwater recharge, pollution by nutrients, soil deposition in valley bottoms or reservoirs (IGBP, 1995). All these effects may dramatically jeopardize natural ecosystems and agricultural systems durability.
Despite the predominance of degraded pasture areas, little information exists about the spatial variation of soil properties including nutrients and carbon. The first results obtained at a regional scale have shown large variations of C, N, Ca and pH, due to vegetation and soil type Bernoux et al., 1998, Cerri et al., 1999. At the field scale, variations may also occur and have to be better understood. Indeed, it is in that scale that agronomic experiments have been installed and carried on to support strategies for conservation practices and policies.
However, soil property variation has been a familiar problem to agricultural scientists who must constantly deal with cumulative effects of micro and macro variation that can easily mask treatment differences in agronomic experiments Perrier and Wilding, 1986, Goovaerts, 1999. An ideal experimental field is a land area in which the plot size and soil variability have been minimized for a specific plant or soil physical/chemical treatment (Davis, 1986). It should have a minimum point-to-point variability (Trangmar et al., 1985). In addition, proper interpretation of experimental data largely depends on the “best” estimation of experimental error (Webster, 2001).
Therefore, our main objective was to assess the spatial variation of chemical and physical soil properties and then use this information to select an appropriate location to install a pasture rehabilitation experiment in the 63 ha area at Nova Vida ranch (Amazon region, Brazil).
Section snippets
Study area
The study was conducted at Nova Vida Ranch (62°49′27ʺW; 10°10′05ʺS), a 22,000 ha cattle ranch located about 250 km south of Porto Velho in central Rondônia State (Brazil), part of the Amazon basin (Fig. 1).
Mean annual temperature at Nova Vida Ranch is 25.5 °C, with 2200 mm precipitation (Bastos and Diniz, 1982), including a dry season (4 to 5 months) lasting from May to September. Predominant natural vegetation of the ranch has been classified as “open humid tropical forest” (Pires and Prance,
Classical statistical analysis
Descriptive statistics were applied to all eight-soil properties (C, N, δ13C, δ15N, pH in H2O, pH in KCl, clay, and sand) at each depth. We evaluated all data together (Table 1), and afterwards, the modeling set data and the validation set data were separately considered. Analyzing the data using a classical approach, no discrepant values were observed. Data followed the same behavior (Table 1) found in other studies in the same region Moraes et al., 1996, Neill et al., 1997.
At the study area,
Conclusions
We found a large amount of spatial heterogeneity in this 63-ha field, despite the fact of the site appears to be as homogeneous as any pasture field in the region. The site had been cleared of original vegetation and used as a pasture for about 17 years prior to the start of this study, with neither chemical fertilizer added nor mechanized agricultural practices adopted. We thus did not expect to find very large differences in important soil properties across the site.
The present study
Acknowledgements
This work was supported by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP-99/07103-0), Coordenacao de Aperfeicoeamento de Pessoal de Nivel Superior (CAPES-1240/01-3), The Ecosystems Center (Woods Hole, USA), National Science Foundation (IBN-9987996), and Consortium for Agricultural Mitigation of Greenhouse Gases (CASMGS) with support from USDA/CREES. We thank Dr. Brigitte Feigl, Dr. Marisa Piccolo, Dr. Marcelo Zacarias and Dr. Plinio Camargo for helping in the soil analysis. João
References (47)
- et al.
Dynamics of soil organic matter as reflect by natural 13C abundance in particle size fractions of forested and cultivated Oxisols
Soil Biol. Biochem
(1992) - et al.
Study of spatial relationships among soil physico-chemical properties of a field in central Italy using multivariate geostatistics
Geoderma
(2000) - et al.
Spatial variability of soil organic carbon in grassland: implications for detecting change at different scales
Environ. Pollut
(2002) - et al.
Soil carbon changes from conversion of forest to pasture in Brazilian Amazon
For. Ecol. Manag
(1998) Geostatistics in soil science: state-of-the-art and perspectives
Geoderma
(1999)- et al.
Variability of soil properties in a tropical Alfisol used for shifting cultivation
Soil Technol
(1996) - et al.
Three-dimensional GIS cartography applied to the study of the spatial variation of soil horizons in a Swiss floodplain
Geoderma
(2000) - et al.
The effect of cultivation on the spatial variability of selected properties of an umbric horizon
Geoderma
(2000) - et al.
Spatial variation of nitrate-N and related soil properties at the plot-scale
Geoderma
(2002) Is soil variation random?
Geoderma
(2000)