Soil databases and the problem of establishing regional biogeochemical trends

Karen Holmes, D.A. Roberts, S. Sweeney, I. Numata, E. Matricardi, T.W. Biggs, G. Batista, O.A. Chadwick

    Research output: Contribution to journalArticle

    22 Citations (Scopus)

    Abstract

    Regional and global environmental modeling depend on soil data for input layers or parameterization. However, randomly located observations, such as provided by agricultural databases, are not always representative of trends identified in field studies conducted under carefully controlled conditions. Many researchers lament the paucity of soil profile data in Amazonia, and suggest that given more data, regional studies would more closely approximate field research results. We assess the ability of a well-populated regional database collected in the southwestern Brazilian Amazon to reproduce expected biogeochemical trends associated with forest clearing and pasture establishment, and explore the ramifications of relying on independently collected soil data for regional modeling. The Soteron database includes analyses of approximately 3000 soil cores collected for zoning purposes in the state of Rondonia. Pasture ages were determined from a time series of Landsat TM images classified using spectral mixture analysis.Although regional averages showed some of the temporal trends expected based on field study results (e.g. increase in pH following forest clearing), the trends were not statistically significant. Stratification by precipitation and other variables showed pasture age to be important but difficult to separate from other potential controls on soil conditions, mainly because of the reduced number of observations in each stratum. Using multiple regression, which permitted the inclusion of all potential explanatory factors and interactions, pasture age was shown to be a statistically significant predictor of soil conditions. However, the expected temporal sequence of changes documented by field chronosequence studies could not be reproduced. Properties dominated by large-scale environmental gradients - pH, sum of base cations, aluminum saturation, and exchangeable calcium - were moderately well modeled, while those more strongly linked to dynamic spatially heterogeneous processes such as biological cycling and land management, particularly organic carbon and nitrogen, could not be modeled.Management-induced soil changes occur at too fine a scale to be captured by most maps, and the relative changes are small compared with spatial heterogeneity caused by controls on soil development over large regions. Therefore, regardless of whether chronosequence-derived models of biogeochemical response to land-cover change are correct, the results of these models will not lead to spatially explicit maps that can be validated by regional reconnaissance, nor will they facilitate realistic predictions of the regional biogeochemical consequences of land-cover change. The change from local to regional scale entails a change in the relative importance of processes controlling soil property behavior.
    Original languageEnglish
    Pages (from-to)796-814
    JournalGlobal Change Biology
    Volume10
    DOIs
    Publication statusPublished - 2004

    Fingerprint

    pasture
    Soils
    chronosequence
    soil
    land cover
    environmental modeling
    environmental gradient
    organic nitrogen
    Landsat thematic mapper
    land management
    zoning
    multiple regression
    soil profile
    soil property
    parameterization
    stratification
    aluminum
    cation
    calcium
    organic carbon

    Cite this

    Holmes, K., Roberts, D. A., Sweeney, S., Numata, I., Matricardi, E., Biggs, T. W., ... Chadwick, O. A. (2004). Soil databases and the problem of establishing regional biogeochemical trends. Global Change Biology, 10, 796-814. https://doi.org/10.1111/j.1529-8817.2003.00753.x
    Holmes, Karen ; Roberts, D.A. ; Sweeney, S. ; Numata, I. ; Matricardi, E. ; Biggs, T.W. ; Batista, G. ; Chadwick, O.A. / Soil databases and the problem of establishing regional biogeochemical trends. In: Global Change Biology. 2004 ; Vol. 10. pp. 796-814.
    @article{bac23cbd139e48dabc11202421dcae75,
    title = "Soil databases and the problem of establishing regional biogeochemical trends",
    abstract = "Regional and global environmental modeling depend on soil data for input layers or parameterization. However, randomly located observations, such as provided by agricultural databases, are not always representative of trends identified in field studies conducted under carefully controlled conditions. Many researchers lament the paucity of soil profile data in Amazonia, and suggest that given more data, regional studies would more closely approximate field research results. We assess the ability of a well-populated regional database collected in the southwestern Brazilian Amazon to reproduce expected biogeochemical trends associated with forest clearing and pasture establishment, and explore the ramifications of relying on independently collected soil data for regional modeling. The Soteron database includes analyses of approximately 3000 soil cores collected for zoning purposes in the state of Rondonia. Pasture ages were determined from a time series of Landsat TM images classified using spectral mixture analysis.Although regional averages showed some of the temporal trends expected based on field study results (e.g. increase in pH following forest clearing), the trends were not statistically significant. Stratification by precipitation and other variables showed pasture age to be important but difficult to separate from other potential controls on soil conditions, mainly because of the reduced number of observations in each stratum. Using multiple regression, which permitted the inclusion of all potential explanatory factors and interactions, pasture age was shown to be a statistically significant predictor of soil conditions. However, the expected temporal sequence of changes documented by field chronosequence studies could not be reproduced. Properties dominated by large-scale environmental gradients - pH, sum of base cations, aluminum saturation, and exchangeable calcium - were moderately well modeled, while those more strongly linked to dynamic spatially heterogeneous processes such as biological cycling and land management, particularly organic carbon and nitrogen, could not be modeled.Management-induced soil changes occur at too fine a scale to be captured by most maps, and the relative changes are small compared with spatial heterogeneity caused by controls on soil development over large regions. Therefore, regardless of whether chronosequence-derived models of biogeochemical response to land-cover change are correct, the results of these models will not lead to spatially explicit maps that can be validated by regional reconnaissance, nor will they facilitate realistic predictions of the regional biogeochemical consequences of land-cover change. The change from local to regional scale entails a change in the relative importance of processes controlling soil property behavior.",
    author = "Karen Holmes and D.A. Roberts and S. Sweeney and I. Numata and E. Matricardi and T.W. Biggs and G. Batista and O.A. Chadwick",
    year = "2004",
    doi = "10.1111/j.1529-8817.2003.00753.x",
    language = "English",
    volume = "10",
    pages = "796--814",
    journal = "Global Change Biology",
    issn = "1354-1013",
    publisher = "John Wiley & Sons",

    }

    Holmes, K, Roberts, DA, Sweeney, S, Numata, I, Matricardi, E, Biggs, TW, Batista, G & Chadwick, OA 2004, 'Soil databases and the problem of establishing regional biogeochemical trends' Global Change Biology, vol. 10, pp. 796-814. https://doi.org/10.1111/j.1529-8817.2003.00753.x

    Soil databases and the problem of establishing regional biogeochemical trends. / Holmes, Karen; Roberts, D.A.; Sweeney, S.; Numata, I.; Matricardi, E.; Biggs, T.W.; Batista, G.; Chadwick, O.A.

    In: Global Change Biology, Vol. 10, 2004, p. 796-814.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - Soil databases and the problem of establishing regional biogeochemical trends

    AU - Holmes, Karen

    AU - Roberts, D.A.

    AU - Sweeney, S.

    AU - Numata, I.

    AU - Matricardi, E.

    AU - Biggs, T.W.

    AU - Batista, G.

    AU - Chadwick, O.A.

    PY - 2004

    Y1 - 2004

    N2 - Regional and global environmental modeling depend on soil data for input layers or parameterization. However, randomly located observations, such as provided by agricultural databases, are not always representative of trends identified in field studies conducted under carefully controlled conditions. Many researchers lament the paucity of soil profile data in Amazonia, and suggest that given more data, regional studies would more closely approximate field research results. We assess the ability of a well-populated regional database collected in the southwestern Brazilian Amazon to reproduce expected biogeochemical trends associated with forest clearing and pasture establishment, and explore the ramifications of relying on independently collected soil data for regional modeling. The Soteron database includes analyses of approximately 3000 soil cores collected for zoning purposes in the state of Rondonia. Pasture ages were determined from a time series of Landsat TM images classified using spectral mixture analysis.Although regional averages showed some of the temporal trends expected based on field study results (e.g. increase in pH following forest clearing), the trends were not statistically significant. Stratification by precipitation and other variables showed pasture age to be important but difficult to separate from other potential controls on soil conditions, mainly because of the reduced number of observations in each stratum. Using multiple regression, which permitted the inclusion of all potential explanatory factors and interactions, pasture age was shown to be a statistically significant predictor of soil conditions. However, the expected temporal sequence of changes documented by field chronosequence studies could not be reproduced. Properties dominated by large-scale environmental gradients - pH, sum of base cations, aluminum saturation, and exchangeable calcium - were moderately well modeled, while those more strongly linked to dynamic spatially heterogeneous processes such as biological cycling and land management, particularly organic carbon and nitrogen, could not be modeled.Management-induced soil changes occur at too fine a scale to be captured by most maps, and the relative changes are small compared with spatial heterogeneity caused by controls on soil development over large regions. Therefore, regardless of whether chronosequence-derived models of biogeochemical response to land-cover change are correct, the results of these models will not lead to spatially explicit maps that can be validated by regional reconnaissance, nor will they facilitate realistic predictions of the regional biogeochemical consequences of land-cover change. The change from local to regional scale entails a change in the relative importance of processes controlling soil property behavior.

    AB - Regional and global environmental modeling depend on soil data for input layers or parameterization. However, randomly located observations, such as provided by agricultural databases, are not always representative of trends identified in field studies conducted under carefully controlled conditions. Many researchers lament the paucity of soil profile data in Amazonia, and suggest that given more data, regional studies would more closely approximate field research results. We assess the ability of a well-populated regional database collected in the southwestern Brazilian Amazon to reproduce expected biogeochemical trends associated with forest clearing and pasture establishment, and explore the ramifications of relying on independently collected soil data for regional modeling. The Soteron database includes analyses of approximately 3000 soil cores collected for zoning purposes in the state of Rondonia. Pasture ages were determined from a time series of Landsat TM images classified using spectral mixture analysis.Although regional averages showed some of the temporal trends expected based on field study results (e.g. increase in pH following forest clearing), the trends were not statistically significant. Stratification by precipitation and other variables showed pasture age to be important but difficult to separate from other potential controls on soil conditions, mainly because of the reduced number of observations in each stratum. Using multiple regression, which permitted the inclusion of all potential explanatory factors and interactions, pasture age was shown to be a statistically significant predictor of soil conditions. However, the expected temporal sequence of changes documented by field chronosequence studies could not be reproduced. Properties dominated by large-scale environmental gradients - pH, sum of base cations, aluminum saturation, and exchangeable calcium - were moderately well modeled, while those more strongly linked to dynamic spatially heterogeneous processes such as biological cycling and land management, particularly organic carbon and nitrogen, could not be modeled.Management-induced soil changes occur at too fine a scale to be captured by most maps, and the relative changes are small compared with spatial heterogeneity caused by controls on soil development over large regions. Therefore, regardless of whether chronosequence-derived models of biogeochemical response to land-cover change are correct, the results of these models will not lead to spatially explicit maps that can be validated by regional reconnaissance, nor will they facilitate realistic predictions of the regional biogeochemical consequences of land-cover change. The change from local to regional scale entails a change in the relative importance of processes controlling soil property behavior.

    U2 - 10.1111/j.1529-8817.2003.00753.x

    DO - 10.1111/j.1529-8817.2003.00753.x

    M3 - Article

    VL - 10

    SP - 796

    EP - 814

    JO - Global Change Biology

    JF - Global Change Biology

    SN - 1354-1013

    ER -