Is the grass always greener on the other side? Comparing vegetation dynamics of a protected area and adjacent communal rangeland during growing seasons with extreme climatic events

Szewczuk, Alekzandra Mari
Journal Title
Journal ISSN
Volume Title
Recent changes in global climate have led to an increase in the severity and frequency of climatic events with current climate models predicting that global temperatures and regional rainfall patterns will continue changing in the future. Annually, the initiation of any phenological event is triggered by seasonal changes in temperature, photoperiod and rainfall. Enhanced Vegetation Index (EVI), a measure of vegetation greenness, obtained from remotely sensed MOD13Q1 16-day composite imagery (250 m resolution), was used to extract phenological time series data and the associated phenometrics of growing seasons. Six consecutive growing seasons (2011-2017) were assessed, during which two “extreme” events occurred. These were a flood in 2013 (annual rainfall = 757.8 mm) and a drought from 2014-2016, in 2014-15 (annual rainfall = 343.7 mm) and 2015-16 (annual rainfall = 307.5 mm). To test if land use and extreme climatic events had an influence on the phenometrics of a mopane dominated region, phenological responses of vegetation to these extreme events was assessed. This was conducted at multiple sites, within a communal rangeland and a protected area- on either side of the north-western fence line of the Greater Kruger National Park, near Pafuri, with a total area of 1250 ha on either side of the fence. The phenological dynamics were compared to rainfall, temperature and photoperiod. Lastly, land cover was determined using three different remote sensing platforms (MODIS MCD12Q1, 500 m resolution; Landsat 8 OLI and TIRS, 30 m resolution; Sentinel-2A, 10 m resolution) for both sides of the fence and the change across the growing seasons was assessed. Results showed that differences in phenology and the associated phenometrics measured within the communal land and protected area were driven by land use, rather than by vegetation type. The start of greenup and the length of the growing season were significantly earlier (six days; p<0.001) and longer (five days; p<0.001) in the communal land. However, the end times did not differ notably for the two sites (p = 0.410). Although the communal land had slightly greater productivity (1%, p<0.05), the protected area had slightly more seasonally active vegetation (1%, p<0.01). EVI was not as strongly influenced by rainfall (protected area: R2 = 0.180; communal land: R2 = 0.190) yet, the phenometrics derived from EVI were more susceptible to changes in seasonal rainfall. The flood year had the longest growing season (protected area = 229 days; communal land = 244 days; p<0.05) and greatest vegetation productivity (flood year to first drought year: protected area = 13% decrease; communal land = 13% decrease, p<0.05; flood year to second drought year- protected area = 9% decrease; communal land = 8% decrease, p<0.05). All the phenometrics were significantly influenced by change in austral years. The end time of the growing season was the only phenometric to not be significantly influenced by land use (p = 0.405). Land use had a significant effect on the response of the StartT, PeakT and MidT to conditions associated with the flood year and first drought year. Dense shrub was the most dominant land cover class for both sides of the fence as detected by the finer scales of Landsat 8 (communal land = 50%, protected area = 42%) and Sentinel (communal land = 43%, protected area = 28%). The MODIS assessment, however, showed significant differences across the fence line for the two common land cover classes between the two sides of the fence, with grass being the most dominant cover class (communal land = 83%, protected area = 95%, p<0.01) followed by deciduous broadleaf trees (communal land = 12%, protected area = 95%, p<0.05). All three datasets showed change in land cover across most of the austral seasons. More importantly, Landsat 8 and MODIS showed land cover to change considerably after the flood year, and Landsat 8 and Sentinel-2 showed change after the drought years.This study highlights the impacts of extreme climatic events and the response of vegetation under different land management strategies to these events. As well as providing important insight for land managers and conservation practitioners, given the prediction of more frequent extreme climatic events across South Africa in the future. Keywords: Drought, Enhanced Vegetation Index (EVI), flood, Greater Kruger National Park, land use, land cover change, phenology, remote sensing, semi-arid savanna, vegetation response.
Master of Science Dissertation: School of Animal, Plant and Environmental Sciences, University of the Witwatersrand, Johannesburg May 2019