Hydrological implications of woody encroachment in the semi-arid savannas of South Africa.

Abstract

Many grasslands and savannas across the globe are undergoing large-scale degradation in the form of woody encroachment, a phenomenon where native woody plants are gradually increasing in abundance at the expense of grasses and other herbaceous vegetation. Given the notion that increases in woody cover lead to increases in evapotranspiration (ET), and that some of the most severe cases of woody encroachment have been reported in drylands, concern has been expressed that woody encroachment may be placing further strain on already limited water supplies. Given this concern, a considerable number of hydrological studies pertaining to woody encroachment have been undertaken across the globe, mostly in the southern Great Plains and southwest regions of the United States. These studies have demonstrated that the hydrological response to woody encroachment is highly variable and is largely dependent on the local climate. The studies carried out in more mesic climates have mostly corroborated the belief that woody encroachment increases ET, resulting in reductions in streamflow and/or groundwater recharge. Fewer studies have been carried out in more arid climates, and their findings have varied from little influence to a decrease or increase in streamflow and groundwater recharge, with soil properties and geology also having a significant influence on the hydrological impact. The Greater Kruger National Park is South Africa’s largest conservation area and is heavily encroached by indigenous woody species. There has been significant concern about the impact on water resources in the park’s northern part, which experiences the driest conditions, and future climate projections indicate that these conditions will only worsen. The northern Greater Kruger National Park region also has a high cover of Colophospermum mopane (Mopane). Mopane is a protected species of high socio-economic importance, but it has become a dominant encroacher species growing in monospecific stands, and climatic models project that it will extend its current distribution range as a result of climate change. Mopane is thought to be an aggressive competitor for water, tapping into soil water stores that are critical for groundwater recharge. The hydrological implications of woody encroachment in South Africa's savannas have received little research attention to date. The ground-based research equipment required to fully explore the impact of woody encroachment on ET is capital and human resource demanding. However, developing countries such as South Africa have limited financial resources and technical expertise to pursue such research. A further challenge has been that the high spatio-temporal variability in ET necessitates data collection and analysis at large spatial scales as well as over several years to be able to draw meaningful conclusions for land and water management planning purposes. However, advances in ground-based observation methods, as well as access to remote sensing technologies, have allowed for many of the limitations to be overcome. Therefore, the aim of this study was to advance our understanding of the impact of woody encroachment on ET and other components of the water balance in a semi-arid savanna in South Africa by employing cheaper, simpler, yet robust in situ measurements, and to further evaluate these findings across various spatio-temporal scales using freely available remote sensing resources. A combined systematic review and meta-analysis approach was undertaken first to explore trends in previous research on the water use of different vegetation types located in various climates across South Africa, to ultimately evaluate the likelihood of woody encroachment increasing ET losses in South Africa’s semi-arid savannas. The results revealed mixed support for this idea. On the one hand, the fact that woody encroacher species replace grasses and form dense thickets with extensive rooting systems indicated that there is a high potential for woody encroachment to increase ET. On the other hand, rainfall appeared to be a primary factor limiting ET in semi-arid climates, indicating little potential for woody encroachment to have any effect on ET, unless there is an above-average rainfall year, or the vegetation has access to an additional water source. It was noted that there was relatively little ground-based research available on woody encroachment, especially in the South African context, and that further research was warranted in dryland and susceptible areas. In situ ET measurements were therefore carried out in a semi-arid South African savanna to determine the effect of woody encroachment. Over three hydrological years, ET was measured at an experimental woody plant clearing trial using surface renewal, a simpler, more affordable alternative to the well-established eddy covariance method. Two surface renewal approaches, surface renewal 1 (SR1) and surface renewal dissipation theory (SRDT), were tested against eddy covariance in order to assess their potential for sensible heat flux (H) measurement. The SR1 approach best agreed with eddy covariance, and therefore, ET measurements derived from SR1 were compared at two adjacent plots differing in woody plant density; one encroached plot and one plot thinned of Mopane trees. For the two drier years of the study, thinning had little effect on ET. However, for the wettest year of the study, thinning decreased ET by 12%, supporting the theory that woody encroachment can increase ET, at least during years of above-average rainfall. It was also important to evaluate the influence of woody plant thinning on soil hydrological processes because any changes to the movement and distribution of water in the soil can have a direct impact on the production of surface runoff and groundwater recharge. Therefore, at the same site, a field-scale paired-plot experiment was conducted, with soil water content, soil temperature and ET measured in three paired plots over a two-and-a-half-year period. Surface infiltration tests were also carried out. Thinning had minor effects on soil water in the soil profile and soil temperature, and no pronounced effect on daily ET. Only one set of infiltration tests indicated a significant increase in infiltration following thinning. Finally, freely available and relatively user-friendly resources provided by Google Earth Engine were leveraged to assess the relationship between woody cover and ET across various spatiotemporal scales in a woody-encroached, semi-arid savanna catchment. Woody cover in the study catchment was mapped using Sentinel-2 imagery and Gradient Tree Boost algorithms within the Google Earth Engine platform, while remotely sensed ET estimates were obtained from Earth Engine Evapotranspiration Flux (EEFlux) validated by in situ surface renewal measurements. It was found that while woody encroachment can increase ET in semi-arid savannas, this increase occurs gradually over time and typically only during wet seasons and wet years. Moreover, this increase only becomes evident at larger scales. The key conclusions from the research were as follows: • Woody encroachment in semi-arid savannas can increase ET losses, however, this is a gradual process that is only distinguishable during wet seasons or extended wet periods; • The increase in ET in semi-arid savannas caused by woody encroachment is more evident at larger scales; • The SR1 approach with eddy covariance calibration produced reliable ET estimates with reasonable accuracy over semi-arid savanna-type vegetation; • ET in semi-arid savannas is highly seasonal, rising during the wet season in response to increased soil water availability and higher net radiation (Rn), and dropping significantly during the dry season due to low soil water availability and the deciduous nature of the vegetation; • Woody plant thinning had little effect on infiltration and soil water in the soil profile; • Sentinel-2 imagery and supervised pixel-based classification algorithms within Google Earth Engine accurately mapped woody cover in densely wooded areas, but in sparsely wooded areas, woody cover was either over- or under-classified; and • EEFlux was able to estimate daily ET in semi-arid savanna vegetation with a reasonable level of accuracy, despite an underestimation of ET. While there is still a need to continue ET monitoring, further improve remote sensing-based ET models for use in natural environments, and better understand how woody encroachment affects deep drainage and groundwater recharge, this research added to our understanding of the hydrological implications of the woody encroachment phenomenon in drylands. It further provided invaluable insight for developing land and water management strategies to increase resilience under a changing climate and increasing socio-ecological pressures.