Plenary: Advances in Methodology

The Albany Thicket (AT) has been over-grazed during recent times, with resultant transformation of large areas. Restoration is linked to its ability to store C efficiently. To determine whether AT is a viable sink for atmospheric C, we established an eddy covariance system on the Ezulu Game Reserve. A Campbell Scientific eddy covariance system, comprising a CSAT sonic anemometer and an IRGASON infra-red gas analyzer were installed in October 2015. Associated instruments provide 30 min values for radiation, soil heat flux, soil moisture and soil temperature. An automatic weather station provided radiation, temperature, relative humidity and rainfall. The system provided instantaneous fluxes for C and H2O, which were corrected for site-specific parameters using EddyPro. The site has been free of domestic livestock for >30 years, and there are small populations of wild herbivores. There is visual evidence of the recovery of Portulacaria afra, where the mean clump height >1m, and clump density >2000 clumps ha-1. The daily CO2 fluxes of the site are comparable to those being experienced by other semi-arid regions in southern Africa, with maximum midday rates of 25-30 umols m-2 s-1. These rates are sustained during a long growing season from October to May. Nighttime respiration rates are low and subject to high reporting error due to low wind conditions at night. Appropriate gap-filling algorithms were used to generate these night-time respiration values. Results for two years of continuous recording indicate that the AT has been a net C sink, accumulating 0,52 (in 2016-17) and 0,59 (in 2015-2016) gC m-2 day-1. Based on current daily accumulation, the annual C gain for this site will be 0.54-0.62 kg C m-2 year-1 (i.e. 5.4-6.2 tons C ha-1 year-1). This still needs to be partitioned between above- and below-ground biomass using the ratio appropriate for this vegetation type. These results are compared with the MODIS net primary production product PsnNet( MOD17A2). MODIS PsnNet predicted 0,52 kg C m-2 year-1 in 2016-2017. The MODIS PsnNet data for 2015-16 had several periods of missing data due to satellite malfunction, but using gap-filling, we estimate that during 2015-2016, 0.69 kg C m-2 year-1. The data provided in this summary reflect two relatively dry years (October 2015 - October 2017), when 238  mm and 252mm of rain were measured per 365 day period. Evapotranspiration (ET) was higher than precipitation in both years, being 278 mm and 324 mm respectively. The higher ET for the site is most likely attributable to a link between larger trees using groundwater, but this needs to be confirmed using isotope analysis. These results represent the first data-rich approximation of the sequestration rates in the AT using eddy covariance, and are the first conclusive evidence of the long-term sequestration ability of the AT. Our results concur largely with the findings of Mills and Cowling (0.42 kg C m-2 year-1 for healthy intact thicket at Krompoort), and the predictions of the MODIS 17A product.

Considering the nature and scale of agriculture in general, remote sensing techniques offers the ideal tool for monitoring management inputs and response over time. As a result, current advancements in data collection and monitoring are frequently based on combining in situ measurements, airborne sensors and satellite observations. Simultaneously, remote sensing technologies are evolving at a rapid rate and together with increased number of earth observing satellites, high resolution remotely sensed imagery are more accessible than ever. Remotely sensed data are used to describe both small and large-scale processes, but each system has its unique spatiotemporal constraints. Ground based assessments are time consuming, labour intensive and very expensive. The main impediment to using remotely sensed data from satellites are the unavailability of cloud free, high spatial (0.3 m or better) and spectral resolution (multiple narrow bands) data. In recent times, the affordability and accessibility of Remotely Piloted Aircrafts (RPA’s) opened new opportunities as transporters of remote sensing equipment. Currently, RPA-mounted sensors are in common use for bridging the gap between ground-based surveys and satellite-derived data. As a result, RPA’s are commonly used to monitor an array of agricultural applications including, vegetation cover, plant health, crop condition, nutrient status, yield, productivity and many more. While most RPA’s provide an accurate, geo-referenced platform which allows for instant results at regular intervals, it has its limitations with regard to flight time, adverse weather conditions and often it’s sensing capacity. In order to correctly interpret the imagery derived in this manner, it is important to take cognisance of the capacity or capability of the particular sensor in use, as well as the reflection pattern which is derived from light that was absorbed and reflected from the electro-magnetic spectrum (EMS). The reflectance of vegetation differs throughout the EMS and is generally found to be very low in the blue and red regions of the EMS, slightly higher in the green region and high in the near infra-red region. This information can be used to determine stress levels in plants linked nitrogen status and correlated with chlorophyll concentration. This case study provides a first-hand account on the application of high resolution remote sensing imagery and illustrating the potential for this approach in Agricultural Management. 

Rangelands in communal lands, particularly in the former Transkei in the Eastern Cape, have been utilized for crop production and livestock farming for many years. Even though livestock farming has been practised for a long time, there have not been many studies addressing the distribution patterns of livestock in these rural areas. Geographic Information Systems (GIS) and remote sensing are tools that have been used to map land use and to monitor animal movement. In the South African context, the use of GIS combined with Global Positioning Systems (GPS) in domestic livestock needs to be given more attention. In this study we develop a predictive spatial model for domestic livestock using the ArcGIS Predictive Analysis Tool.  We used a combination of a GPS collar dataset from tracked livestock during the dry and wet season and landscape variables that are expected to influence livestock distribution to predict potential distribution across the landscape. The selected landscape variables were (1) altitude (height above sea level), (2) slope (steepness) and (3) aspect (solar radiation intercepted by that surface), which were all generated from a Digital Elevation Model (DEM), (4) Normalized Difference Vegetation Index (NDVI) which is used to represent the presence of vegetation and lastly, (5) water sources. We addressed three research questions: (1) where do livestock spend time? (2) Can we predict potential livestock distribution in other areas? (3) What management strategies can be implemented to overcome the under- and overutilization of resources available? The study was conducted in a village near Cala, Eastern Cape. Preliminary results show that core livestock distribution is located on grazing resources near homesteads and water sources, suggesting that these areas on the landscape are important in prediction models. The results also demonstrate that without adequate fencing and livestock management, livestock may over-utilize local grazing resources. With the decline in the use of herders within these communities, livestock are concentrated around human settlements. The research will benefit the local community members by emphasising the need for strategic herding and labour through the identification of areas that are potentially underused and overused. The research could also provide baseline information needed for rangeland strategies such as rotational grazing.