Please use this identifier to cite or link to this item:
Title: Spatio–temporal variation of vegetation heterogeneity in groundwater dependent ecosystems within arid environments.
Authors: Mpakairi, Kudzai Shaun.
Dube, Timothy.
Dondofema, Farai.
Dalu, Tatenda.
University of the Western Cape
University of Western Cape
University of Venda
School of Biology and Environmental Sciences
Keywords: Khakea–bray.;Rao’s Q.;Random forest.;Vegetation heterogeneity.;Spectral variation hypothesis (SVH).;Shannon–Weiner.
Issue Date: 2022
Publisher: Elsevier
Abstract: Climate change, land cover change and the over–abstraction of groundwater threaten the existence of Groundwater-Dependent Ecosystems (GDE), despite these environments being regarded as biodiversity hotspots. The vegetation heterogeneity in GDEs requires routine monitoring in order to conserve and preserve the ecosystem services in these environments. However, in–situ monitoring of vegetation heterogeneity in extensive, or transboundary, groundwater resources remain a challenge. Inherently, the Spectral Variation Hypothesis (SVH) and remotely-sensed data provide a unique way to monitor the response of GDEs to seasonal or intra –annual environmental stressors, which is the key for achieving the national and regional biodiversity targets. This study presents the first attempt at monitoring the intra–annual, spatio–temporal variations in vegetation heterogeneity in the Khakea–Bray Transboundary Aquifer, which is located between Botswana and South Africa, by using the coefficient of variation derived from the Landsat 8 OLI Operational Land Imager (OLI). The coef ficient of variation was used to measure spectral heterogeneity, which is a function of environmental hetero geneity. Heterogenous environments are more diverse, compared to homogenous environments, and the vegetation heterogeneity can be inferred from the heterogeneity of a landscape. The coefficient of variation was used to calculate the α- and β measures of vegetation heterogeneity (the Shannon–Weiner Index and the Rao’s Q, respectively), whilst the monotonic trends in the spatio–temporal variation (January–December) of vegetation heterogeneity were derived by using the Mann–Kendall non–parametric test. Lastly, to explain the spa tio–temporal variations of vegetation heterogeneity, a set of environmental variables were used, along with a machine-learning algorithm (random forest). The vegetation heterogeneity was observed to be relatively high during the wet season and low during the dry season, and these changes were mainly driven by landcover- and climate–related variables. More specifically, significant changes in vegetation heterogeneity were observed around natural water pans, along roads and rivers, as well as in cropping areas. Furthermore, these changes were better predicted by the Rao’s Q (MAE = 5.81, RMSE = 6.63 and %RMSE = 42.41), than by the Shannon–Weiner Index (MAE = 30.37, RMSE = 33.25 and %RMSE = 63.94). These observations on the drivers and changes in vegetation heterogeneity provide new insights into the possible effects of future landcover changes and climate variability on GDEs. This information is imperative, considering that these environments are biodiversity hot spots that are capable of supporting many livelihoods. More importantly, this work provides a spatially explicit framework on how GDEs can be monitored to achieve Sustainable Development Goal (SDG) Number 15.
DOI: 10.1016/j.ecoinf.2022.101667
Appears in Collections:Journal articles

Show full item record

Google ScholarTM



Items in UMP Scholarship are protected by copyright, with all rights reserved, unless otherwise indicated.