Hill slope seep wetlands are ecologically and economically important ecosystems as they supply a variety of ecosystem services to society. In South Africa, livestock grazing is recognised as one of the most important disturbance factors changing the structure and function of hill slope seep wetlands. This study sought to investigate the potential effect of livestock grazing on the resilience and vulnerability of hillslope seep wetland vegetation cover using a trait based approach (TBA). Changes in vegetation cover were used as a surrogate for indicating grazing intensity. The degree of human disturbances was assessed using the Anthropogenic Activity Index (AAI). A TBA was developed using seven plant traits, resolved into 27 trait attributes. Based on the developed approach, plant species were grouped into vulnerable and resilient groups in relation to grazing pressure. It was then predicted that species belonging to the vulnerable group would be less dominant at the highly disturbed sites, as well as in the winter season when grazing pressure is at its peak. The approach developed enabled accurate predictions of the responses of hillslope plant species to grazing pressure seasonally, but spatially, only for the summer season. The predicted responses during the winter season across sites did not match the observed results, which could be attributed to the difficulty in species identification and accurate estimation of vegetation cover during winter. Overall, the approach developed here provides a general framework for applying the TBA and can thus be tested and applied elsewhere.

Developing a trait-based approach (TBA) for assessing the vulnerability vegetation cover to livestock grazing in hillslope seep wetlands

The approach followed in this study was largely adapted from Odume et al. (2018) and follows five steps in classifying plant species into vulnerability groups:

1. Reviewing the literature for reported grazing modes of stress on plant species (more details on the manuscript).
2. On the basis of the reported grazing modes of stress, select and measure traits that are mechanistically linked to the modes of stress (The selected plant traits were those that were deemed mechanistically linked to livestock grazing disturbances. Seven plant traits resolved into 27 trait attributes were selected. The seven plant traits were measured according to the standardised world-wide protocol described by Cornelissen et al. (2003) and included plant height, specific leaf area, palatability, leaf size, leaf dry matter content, longevity (years), and resprouting potential (more details on the manuscript).

3. Identifying plant traits deemed mechanistically linked to disturbance factors, such as livestock grazing (more details on the manuscript)
4. Identifying vulnerable trait attributes from the selected trait categories (more details on the manuscript).
5. Grouping plant species into three vulnerability groups, based on the combination of plant traits (more details on the manuscript).

Species, vegetation cover and AAI

Vegetation cover was used as a surrogate measure for grazing intensity within each hillslope seep wetland. Direct grazing measurement could not be taken but, given that the hillslope seeps were situated in a rural catchment where no other major activity impact vegetation cover apart from grazing, this measure was deemed appropriate. A 100 m transect was established at the centre of each seep wetland in order to avoid the possibility of sampling terrestrial plant species. This is particular important because of the very small sizes of hillslope seep wetlands, and thus care was taken not sample terrestrial plant species. However, while care was taken to avoid sampling terrestrial plant species, the sampling strategy deployed may also have led to under sampling and estimation of edged species. Nevertheless, similar sampling strategies have been deployed by (Wardrop et al., 2004). Each transect was marked with small steel pegs so that they could be accurately located in the next sampling season. The vegetation in each site was sampled in two ways. First, the cover was determined following a non-destructive method of Flombaum and Sala (2007). Five (0.2 x 1 m) quadrats were placed along each transect at intervals of 20, 40, 60, 80, and 100 m. In each quadrat, species relative cover and total percentage vegetation cover was visually estimated based on the quadrat area covered by grass using categories shown in Table 4. Second, all the vascular species were identified and recorded along the transect to determine species composition using the step-point method (Evans and Love 1957). Species used for traits were collected using this approach. To reduce potential bias from using cover classes, mid-point of each class was used to estimate vegetation cover.

The anthropogenic Activity Index (AAI) was used to quantify general anthropogenic disturbance in the studied hillslope seep wetlands. The index consists of five metric, i) surrounding land use intensity, ii) soil disturbance, iii) hydrology and habitat alteration, and iv) vegetation community. The five metrics were assessed at each wetland, scored and then aggregated to provide an AAI value per site. AAI value range between 1 and 15, where scores 1-5 indicate least disturbance, 6-10 moderate disturbance and > 10 high disturbance (Ervin et al, 2006).

Data process

Species abundance data was captured in excel per site.

The seven plant traits were measured according to the standardised world-wide protocol described by Cornelissen et al. (2003), out of those all leaf traits and included plant height, were measured specific leaf area, while longevity (years), and resprouting potential were derived from literature and expert knowledge. These were all typed in excel with their values. Vulnerability scores for each species were calculated using percentile distributions.

Vegetation cover was used as a surrogate measure for grazing intensity- the cover collected from each quadrat was then summed to give the overall for each site and mid-point of each class was used to estimate vegetation cover.

AAI data-The five metrics were assessed at each wetland, scored and then aggregated to provide an AAI value per site. AAI value range between 1 and 15, where scores 1-5 indicate least disturbance, 6-10 moderate disturbance and > 10 high disturbance (Ervin et al, 2006). All scored values of the five metrics were summed to obtain the degree of disturbance per site per season and then transferred to excel.