Premium
Temporal changes in spatial patterns of soil moisture following disturbance: an experimental approach
Author(s) -
Guo Dali,
Mou Pu,
Jones Robert H.,
Mitchell Robert J.
Publication year - 2002
Publication title -
journal of ecology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.452
H-Index - 181
eISSN - 1365-2745
pISSN - 0022-0477
DOI - 10.1046/j.1365-2745.2001.00667.x
Subject(s) - replicate , geostatistics , spatial heterogeneity , spatial ecology , environmental science , disturbance (geology) , spatial variability , variogram , soil science , common spatial pattern , sampling (signal processing) , scale (ratio) , spatial dependence , ecological succession , atmospheric sciences , ecology , kriging , biology , mathematics , geography , geology , statistics , cartography , paleontology , filter (signal processing) , computer science , computer vision
Summary 1 We quantified changes in spatial heterogeneity of soil moisture over 2.5 years in a Pinus elliottii Engelm. forest, following disturbance and succession. We harvested or girdled upper canopy trees and measured three components of heterogeneity – global (non‐spatial) variability, spatial dependence and temporal persistence – in replicate plots, using sample points arrayed at a fine scale (0.5–6 m) nested within a coarser scale (5–60 m). 2 Global variability increased after disturbance and then declined, eventually returning to the level recorded in an undisturbed plot. Harvesting resulted in greater, more rapid and more prolonged changes in global variability than girdling. 3 Geostatistical parameters for measuring spatial dependence were largely unaffected by disturbance. Spatial dependence was, however, quite variable across replicate plots and was stronger at the finer sampling scale. 4 Spearman rank correlations showed that the spatial pattern of soil moisture had greater long‐term persistence in the undisturbed and girdled plots than in the harvested plots. 5 Some elements of spatial heterogeneity appear to vary over time in a predictable manner. Detection of temporal trends may be improved if multiple components of heterogeneity are quantified, more than one scale of observation is used, replicate plots are employed and sole reliance on geostatistics is avoided.