Probabilistic description of vegetation ecotones using remote sensing
Date
2018-07
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Elsevier
Abstract
Ecotone transitions between vegetation types are of interest for understanding regional diversity, ecological processes and biogeographical patterns. Ecotones are seldom represented on vector, line-based vegetation maps, which imply an instantaneous change from one vegetation type to another. We use supervised, probabilistic classification of remotely sensed (RS) imagery to investigate the location, width and character of ecotones between acid Sandstone and alkaline Limestone fynbos on the Agulhas plain at the southern tip of Africa, known for rapid speciation of plants and exceptional plant biodiversity at the global scale. The resultant probability map, together with the probability graphs developed for a few transects across the transition, are able to map and describe (1) sharp, narrow ecotones (under five meters); (2) moderate ecotones that have a distinct band of transition (over a few hundred meters); and (3) complex ecotones that include slow transitions, interdigitated boundaries and outliers. The latter class of transitions include portions where vegetation types change sharply over a few meters, but due to the interdigitated boundaries they are mapped over hundreds of meters to a kilometre at a landscape scale. In this study area, our findings suggest that the character of the Agulhas limestone-acid ecotone is probably more complex than often noted. Moderate transitions and broad mosaics are difficult to indicate in a vector vegetation map, whereas RS probabilistic classifications can output images indicating core areas, important for key species and biodiversity pattern, and transitional zones, important for ecosystem processes and perhaps plant evolution, which distinction is important for conservation planning.
Description
CITATION: De Klerk, H. M., Burgess, N. D. & Visser, V. 2018. Probabilistic description of vegetation ecotones using remote sensing. Ecological Informatics, 46:125-132. doi:10.1016/j.ecoinf.2018.06.001.
The original publication is available at https://www.sciencedirect.com
The original publication is available at https://www.sciencedirect.com
Keywords
Probabilistic classification -- Remote sensing, Ecotone -- Remote sensing, Vegetation transition -- Remote sensing
Citation
De Klerk, H. M., Burgess, N. D. & Visser, V. 2018. Probabilistic description of vegetation ecotones using remote sensing. Ecological Informatics, 46:125-132. doi:10.1016/j.ecoinf.2018.06.001.