Frequent grazing can establish high forage value grazing lawns supporting high grazer densities, but can also produce overgrazed grass communities with unpalatable or low grass basal cover, supporting few grazers. Attempts to create grazing lawns via concentrated grazing, with a goal to increase grazer numbers, are thus risky without knowing how environmental conditions influence the likelihood of each outcome.

We collected grass species and trait data from 33 frequently grazed grass communities across eastern South Africa (28 sites) and the Serengeti National Park, Tanzania (5 sites), covering wide rainfall (336–987 mm.yr-1) and soil (e.g. 44–93% sand) gradients. We identified four grass growth forms using hierarchical clustering on principal components analyses of trait data, and assessed trait-environment and growth form-environment relationships using fourth corner and principal components analyses.

We distinguished two palatable grass growth forms that both attract yet resist grazers, and comprise grazing lawns: 1) ‘lateral attractors’ that spread vegetatively via stolons and rhizomes, and 2) ‘tufted attractors’ that form isolated tufts, and may have alternate tall growth forms. By contrast, 3) tough, upright, tufted ‘resisters’, and 4) ‘avoiders’ with sparse architectures or that grow appressed to the soil surface, are of little forage value and avoided by grazers.

Grazing lawns occurred across a wide range of conditions, typically comprising lateral attractor grasses in drier, sandy environments, and tufted attractor grasses in wetter, low-sand environments. Resisters occurred on clay-rich soils in mesic areas, while avoiders were widespread but scarce.

While grazing lawns can be established under most conditions, monitoring their composition and cover is important, as the potential for overgrazing seems as widely relevant. Tufted attractor-dominated lawns appear somewhat more vulnerable to degradation than lateral attractor-dominated lawns. Increased avoider or resister abundance both reduce forage value, although resisters may provide better soil protection.

Sampling procedures

Grass communities were sampled using 0.25 m² quadrats distributed evenly through the frequently grazed habitat. Most sites had 30 quadrats but the minimum was 15 at one site where sampling was restricted by time. Full details of sampling areas and plot layout are provided in the "SiteInformation" worksheet. Overall, the average distance between quadrats was ~12 m (range: 8 m to 15 m).

All grass species occurring within a quadrat were identified in the field and verified at the National Herbarium in Pretoria, South Africa. For each grass species within a quadrat we recorded percentage aerial cover, median leaf table height (mm), culm orientation (lateral, decumbent, geniculate or upright), stolons (present or absent) and rhizomes (absent, short or long). Leaf table height was assessed visually as the approximate 80th quantile of leaf biomass, with the main bulk of the leaf canopy occurring below this height. We classified short rhizomes as those that incrementally allowed an individual to expand the size of its base, forming a tuft, and long rhizomes as those facilitating the establishment of new ramets with spatially separate aboveground biomass. Percentage bare ground in each quadrat was recorded, and whether grazer dung was present or not. All data were collected by the same observer throughout the study (Gareth Hempson).

Environmental data  

Soil samples were collected at the four corners of each site, and analysed for texture (percent sand, silt and clay), cations (K, Ca, Mg and Na), exchangeable acidity and pH. Cation exchange capacity (CEC) was calculated for each soil sample. South African soil samples were analysed at the Agricultural Research Council Institute for Soil, Climate and Water, in Pretoria, South Africa, and Serengeti soils were analysed at the Sokoine University of Agriculture, Morogoro, Tanzania. Daily rainfall data were extracted from the Climate Prediction Center (CPC) Africa Rainfall Climatology Version 2.0 (ARC2) dataset and used to calculate mean annual rainfall for each site for the 30-year period prior to the sampling date. Access to these data was obtained via the Columbia University International Research Institute for Climate and Society website (iri.columbia.edu). 

Trait indices 

Grass species at our sites persist under frequent grazing, and we sought to characterise the key life history attributes that enable this (i.e. avoidance-attractance, resistance or tolerance). Four trait indices were derived from field measurements: 1) culm orientation index, 2) lateral index, 3) tuft index, and 4) grazer use index. Culm orientation index was calculated as the species mean value after converting each species × quadrat culm orientation record to a numerical value as follows: lateral = 1, geniculate-lateral = 2, geniculate or decumbent = 3, geniculate-upright = 4, and upright = 5. The lateral index was calculated as the proportion of quadrat-level records where a species had stolons or long rhizomes. Similarly, the tuft index was calculated as the proportion of quadrat-level records where a species had a tufted base. To indicate the relative site-level grazing preference of a species, we derived a grazer use index based on the assumption that maximum grazer use is experienced by species with heights close to the median site-level leaf table height, via: 1) taking the ratio of the leaf table height for each species × quadrat to the overall site-level median leaf table height, 2) for values > 1 (i.e. where a species × quadrat is taller than the overall site median), taking the reciprocal of this value, and 3) calculating the overall species mean value across all quadrats × sites.

Data are provided as a .xlsx file created in Microsoft Excel.