**4. Conclusions**

Time series analyses methods were employed in this study to investigate the basic structure variability and trend of the HiP NDVI and its response to the variability of climatic conditions. The results of this study indicate that drought stress reaction patterns of vegetation within HiP provide temporal responses to climate variability, suggesting a strong causal influence. Both the NDVI and EVI values, averaged over the study area, decreased suddenly during 2014–2016 to their greatest minima of approximately 0.28 and 0.11, respectively, in 2015. The linear relationship between climatic indices and NDVI indicated that precipitation soil temperature, soil moisture at root-zone (NDII), ET and to some extent ENSO play a significant role in the variability of vegetation health. The Pearson correlation *<sup>r</sup>* and MLR *<sup>p</sup>*-value for precipitation and ENSO were found to be 0.45 and 2.0 × <sup>10</sup>−7, and 0.27 and 8.4 × <sup>10</sup>−4, respectively. While some studies [17] reported temperature as the main meteorological parameter that influences vegetation, in this study, we conclude that the influence of precipitation on vegetation was more significant. Different areas of the HiP are affected differently by the strong El Niño signal because of the special variation of land cover. The southern part of the HiP was affected the most because it is dominated by savanna. On the other hand, the northern part of the HiP seems to not be affected presumably because land cover in this area is dominated by forests which are composed of trees which are drought resistant. Moreover, terrain appears to have additional influence on the state of vegetation in the reserve. For example, the lower NDVI values corresponded with the 2014–2016

drought period, particularly in the south-western (flat) part of the reserve, whereas the northern parts (hilly) seem to have benefited from orographic precipitation which promoted vegetation growth. Terrain is also assumed to restrict wildlife grazing in hilly parts of the reserve where stable NDVI are noticeable, placing more burden in flat areas that are accessible to most grazers.

The Mann–Kendall trend significance test and the sequential version of the Mann–Kendall test statistic revealed a significant decreasing pattern of NDVI during the extreme drought periods of 2003 and 2014–2016, with unprecedented lowest minimum values of NDVI detected in 2015. This study has also demonstrated how the wavelet coherence signal processing technique can serve in identifying periodicities in NDVI time series and can also help demonstrate the temporal response of vegetation status to environmental disturbances. The wavelet coherence power spectra indicate a strong influence of precipitation, soil temperature, soil moisture, and ET on the viability of NDVI. This was revealed by a dominant in-phase relationship between the climatic variables and NDVI, which suggests a positive correlation.

While the El Niño of 2014–2016 was both extended and strong, it is possible that its influence in the study area was also supported by a corresponding positive DMI peak which took place at the same time with the with the 2014–2016 El Niño period. It is, therefore, desirable to use the wavelet coherence technique and other methods to investigate the phase relationship between ENSO and DMI for determining the corresponding influence of rainfall in the north-eastern part of South Africa.

Finally, we conclude that the recent intense drought of 2014–2016 influenced the spatiotemporal pattern of the vegetation condition in the HiP. This holds more implications for the tourism potential of the HiP with attractive grazers such as white rhinos and buffalos that were reportedly affected by this event [11]. The results portend that the freely GEE-archived satellite data is a capable tool for monitoring droughts with a high temporal resolution across game reserves located in drought-prone areas of South Africa and other parts of the world.

**Author Contributions:** N.M. and S.X. designed the research. N.M. performed data analyses, visualization, and interpretation of results. N.M. and S.X. wrote the paper.

**Funding:** This research is funded by the National Research Foundation (NRF) of South Africa and the University of Zululand.

**Acknowledgments:** The authors would like to thank all personnel involved in the development of the Google Earth Engine system and climate engine. We also thank the providers of the important public data set in the Google Earth Engine, in particular, NASA, USGS, NOAA, EC/ESA, and MERRA-2 model developers.

**Conflicts of Interest:** The authors declare no conflict of interest.
