The distribution, structure and function of mesic savanna grasslands are strongly driven by fire regimes, grazing by large herbivores, and their interactions. This research addresses a general question about our understanding of savanna grasslands globally: Is our knowledge of fire and grazing sufficiently general to enable us to make accurate predictions of how these ecosystems will respond to changes in these drivers over time? Some evidence suggests that fire and grazing influence savanna grassland structure and function differently in South Africa (SA) compared to North America (NA). These differences have been attributed to the contingent factors of greater biome age, longer evolutionary history with fire and grazing, reduced soil fertility, and greater diversity of plants and large herbivores in SA. An alternative hypothesis is that differences in methods and approaches used to study these systems have led to differing perspectives on the role of these drivers. If the impacts of shared ecosystem drivers truly differ between NA and SA, this calls into question the generality of our understanding of these ecosystems and our ability to forecast how changes in key drivers will affect savanna grasslands globally. Since 2006, an explicitly comparative research program has been conducted to determine the degree of convergence in ecosystem (productivity, N and C cycling) and plant community (composition, diversity, dynamics) responses to fire and grazing in SA and NA.
Thus far, initial support has been found for convergence at the ecosystem level and divergence at the community level in response to alterations in both fire regimes and grazing. However, there have also been two unexpected findings (1) the ways in which fire and grazing interact differed between NA and SA, and (2) the rate of change in communities when grazers were removed was much greater in NA than in SA. These unexpected findings raise a number of important new questions: (Q1) Will exclusion of grazing eventually affect community structure and composition across all fire regimes in SA? (Q2) Will these effects differ from those observed in NA? (Q3) What are the determinants of the different rates of community change? (Q4) How will these determinants influence future trajectories of change? (Q5) Will the different rates and trajectories of community change be mirrored by responses in ecosystem function over time? This project is based on a large herbivore exclusion study established within the context of long-term (25-50+ yr) experimental manipulations of fire frequency at the Konza Prairie Biological Station (KPBS) in NA and the Kruger National Park (KNP) in SA. The suite of core studies and measurements include plant community composition, ANPP, and herbivore abundance and distribution at both study sites to answer these research questions.
This data set was added to the Sevilleta LTER archive at the request of the SEV Principal Investigator Scott Collins.
Data are collected twice each year at each site. Sample periods are equivalent to spring and late summer at each study site (December/January and March/April in South Africa, May and September in North America.
Where the Data were Collected:
Ukulinga Research Farm, Pietermaritzburg, South Africa; Satara Region of Kruger National Park, South Africa; Konza Prairie Biological Station, North America
Additional Geographic Metadata:
Ukulinga Research Farm (URF), South Africa. The URF of the University of KwaZulu-Natal is located in Pietermaritzburg, in southeastern South Africa (30o 24’ S, 29o 24’ E). The site is dominated by native perennial C4 grasses, such as Themeda triandra and Heteropogon contortus, that account for much of the herbaceous aboveground net primary production (ANPP). Mean annual precipitation is 790 mm, coming mostly as convective storms during summer (Oct-Apr). Summers are warm with a mean monthly maximum of 26.4oC in February, and winters are mild with occasional frost. Soils are fine-textured and derived from shales. There has been no grazing at this site for >60 years. Long-term experimental plots were established at URF in 1950 with the objective of determining the optimal fire and/or summer cutting regime to maximize hay production. The experiment is a randomized block (three replicates) split-plot design with four whole-plot haying treatments and 11 subplot fire or mowing treatments. Subplot sizes are 13.7 x 18.3 m.
Kruger National Park (KNP), South Africa. The KNP is a 2 million ha protected area of savanna grassland that includes many of the large herbivores common to southern Africa (22º 25' to 25º 2 32' S, 30º 50' to 32º 2' E). The extant abundance and grazing intensity of herbivores in KNP is considered moderate for regional savanna grasslands. In the south-central region of KNP where our research takes place, average rainfall is 537 mm with most falling during the growing season (Oct-Apr). The dormant season is mild, dry and frost free, and summers are warm with mean monthly maximum air temperature of 28.9oC in January. Because of the importance of fire in savanna grassland ecosystems, the Experimental Burn Plot (EBP) experiment was initiated in 1954 to examine the effects of fire frequency (control-no fire, 1-, 2-, 3-, 4- and 6-yr return interval) and season [early spring (Aug), spring (Oct), mid-summer (Dec), late summer (Feb), and fall (Apr)] on vegetation communities in the park. Four blocks of 12 plots (two were later split for the 4- and 6-yr trts), each ~7 ha (370 x 180 m) in size, were established in four primary vegetation types covering the two major soil types (granites and basalts) and spanning the precipitation gradient in the park. Each plot has 50+ years of known fire history, and native herbivores have had unrestricted access, thus fire and grazing effects are combined. This research focuses on the EBPs located near Satara where precipitation, soil type, and the mix of herbaceous and woody plants are similar to KPBS. Vegetation on the blocks is co-dominated by C4 grasses, such as Bothriochloa radicans, Panicum coloratum and Digiteria eriantha, and woody plants, such as Acacia nigrescens and Sclerocarya birrea. Soils are fine-textured and derived from basalts. Adjacent to one of the Satara blocks is the Cape buffalo enclosure, erected in 2000 for veterinary purposes. The 200 ha permanent enclosure contains 65-80 animals and is divided into 4 blocks burned on a rotational basis. The grazing intensity inside is comparable to the moderate levels imposed in the park and at KPBS. Two blocks are burned annually while others are burned infrequently (approximately once every 4-yr).
Konza Prairie Biological Station (KPBS), North America. The KPBS is a 3,487 ha savanna grassland in northeastern Kansas, USA (39o 05’ N, 96o 35’ W) dominated by native perennial C4 grasses such as Andropogon gerardii and Sorghastrum nutans that account for the majority of ANPP. Scattered shrub and tree species include Cornus drummondii, Gleditsia triacanthos, and Prunus spp. Numerous sub-dominant grasses and forbs contribute to the floristic diversity of the site. The climate is continental, with mean July air temperature of 27°C. Annual precipitation is ca. 820 mm/year, with 75% falling as rain during the Apr-Oct growing season. Soils are fine textured, silty clay loams derived from limestone and shales. KPBS includes fully replicated watershed-level fire and fire/grazing treatments, in place since 1977 and 1987, respectively. Replicate watersheds (mean size ~60ha) are burned at 1-, 2-, 4-, 10- and 20-yr intervals, mainly in April, to encompass a range of likely natural fire frequencies and management practices. A subset of watersheds has not been grazed for more than 30 years. To address the role of native grazers and fire/grazing interactions, bison (~260 individuals) were reintroduced to KPBS in a 1000-ha fenced area that includes replicate watersheds burned in the spring at 1-, 2-, 4- and 20-year intervals. The overall grazing intensity is considered moderate.
Study Area 1:
Study Area Name: Ukulinga Research Farm
Study Area Location: Near Pietermaritzburg, South Africa
Elevation: 840 m above sea level
Landform: Colluvium fan
Geology: Marine shales and dolerite colluvium
Soils: Dystric leptosols, Chromic luvisols, Haplic plinthisols
Vegetation: Native grassland
Climate: Mean annual precipitation is 844 mm, Mean annual temperature 17.6C
Site history: Ungrazed since 1950
Single Point: 29o 40’ S / 30o 20’ E
Study Area 2: Kruger National Park, South Africa
Study Area Name: Satara Experimental Burn Plots and Cape Buffalo Exclosure
Study Area Location: Near Satara rest camp
Elevation: 240-320 meters above sea level
Landform: Level Upland
Soils: Rhodic nitisols, Haplic luvisols, Leptic phaeozems
Climate: Mean annual precipitation 544 mm; mean annual temperature 21.2–23.3C
Site history: Grazed by native herbivores
Single Point: 23–25o S /30-31o E
Study Area 3: Konza Prairie Biological Station
Study Area Name: Konza Prairie
Study Area Location: Watersheds N20B, N4D, N1B, N4B; 1D, 4F, 20B
Elevation: 320-444 meters above sea level
Landform: Alluvial terrace
Geology: Cherty limestone and shale
Soils: Udic argiustolls
Climate: Mean annual precipitation 835 mm; mean annual temperature 12.7C
Site history: Ungrazed watersheds (since 1971), watersheds grazed by native herbivores (since 1987)
Single Point: 39o 05.48’ N / 96o 34.12’ W
We used comparable experimental designs and sampling procedures at both URF and KPBS. At URF we used three replicate plots (not hayed or mowed) that have been burned every 1 and 3 years in the spring, and those left unburned (N=9 plots). At KPBS, we established replicate plots in experimental watersheds burned every 1 and 4 years in the spring, and those left unburned (N=9 plots). Thus, the only difference in design between NA and SA was the intermediate burn frequency. In 2005 at both sites we established four 2x2m areas in each replicate of the 1-yr, 3-4 yr burned, and unburned plots (N=36 subplots). We then randomly selected two of the subplots for the fertilization treatment and the other two subplots served as controls (Fig. 1). Starting in 2006 at KPBS and 2007 at URF, we began adding 10 gN/m2/yr as NH4+NO3- to assess the interactive effects of fire frequency and nitrogen limitation on plant community composition, structure and dynamics.
Fig. 1. Experimental design and sampling for the proposed studies: A) the role of long-term fire regimes (without megaherbivores), B) the importance of grazing and grazing/fire interactions, and C) the role of megaherbivore diversity. Moveable exclosures (3/plot) will be used to estimate ANPP in the grazed plots. N addition subplots (2 x 2 m) will be divided into 4 1 x 1 plots, with two designated for plant species composition sampling and the other two for destructive sampling. Soil samples will be collected from areas not designated for ANPP or plant composition sampling. Note that the same annually and infrequently burned plots at Kruger and Konza will be used in (B) and (C). In addition, similar plots will be established minus the N addition subplots in the 1-yr and 4-yr burned blocks of the Buffalo enclosure for (C).
Each of the 2x2m subplots was divided into four 1x1m quadrats. Annually since 2005 (prior to nitrogen addition) canopy cover of each species rooted in each quadrat was visually estimated twice during the growing season to sample early and late season species. As a surrogate for aboveground production, we measured light availability at the end of the growing season above the canopy at the ground surface in each quadrat (N=4 per subplot) using a Decagon ceptometer.
Net primary production measurements: Prior to the 2005 growing season we established plots (13.7 m by 18.3 m) in ungrazed areas burned annually, at 3–4-year intervals, and unburned (n = 3 per fire treatment) at both KBPS and URF. Areas with trees or large shrubs were avoided as our main goal was to evaluate responses in the herbaceous plant community. ANPP was estimated from end-of-season harvests starting in 2005 (September for KBPS, April for URF). In 10, 0.1-m2 (20 cm by 50 cm) quadrats randomly located in each plot (n = 30/treatment/site), we harvested the vegetation at ground level and separated it into grass, forb, and previous year’s dead biomass. Samples were dried at 60C to a constant weight. For annually burned plots, total biomass harvested represents ANPP. For the intermediate and unburned sites, we calculated ANPP by summing all but the previous year’s dead component.
To assess the impacts of fire on ANPP in grazed areas, we established herbivore exclusion treatments in KBPS in North America and KNP in South Africa. Herbivore exclosures in grazed areas in KPBS and KNP were erected prior to the 2006 growing season. The exclosures were 7 m in diameter, 2 m tall, and constructed of diamond mesh (5-cm diameter). Seven exclosures were established in each of three blocks of the three fire treatments— annually burned, intermediate burn (3- years for KNP or 4-years for KPBS), and unburned (n = 21 exclosures/treatment/site). As our focus was on ANPP responses of the herbaceous layer, exclosures were not located beneath trees or where dense shrub patches were present. Additionally, in the Satara region of KNP is a 900-ha permanent enclosure containing 80–90 adult African buffalo (S. caffer). This enclosure was erected in 2000 and was divided into six areas (100–200 ha each), with these burned on a rotational basis including plots burned annually and plots that were unburned. We used the unburned and annually burned areas in the buffalo enclosure to provide a direct comparison for determining the effects of a single-species large grazer in KNP and KPBS, and to assess the effects of large herbivore diversity at adjacent sites in KNP. Similar exclosures were built in the African buffalo enclosure at KNP. We placed 7 exclosures in the three blocks of each fire treatment (annually burned and unburned) resulting in 21 exclosures/treatment. We sampled ANPP by harvesting plant biomass from three 0.1 m2 quadrats per herbivore exclosure at the end of the growing season starting in 2006.
See all Sevilleta Publications