EDGE is located at six grassland sites that encompass a range of ecosystems in the Central US - from desert grasslands to short-, mixed-, and tallgrass prairie. We envision EDGE as a research platform that will not only advance our understanding of patterns and mechanisms of ecosystem sensitivity to climate change, but also will benefit the broader scientific community. Identical infrastructure for manipulating growing season precipitation will be deployed at all sites. Within the relatively large treatment plots (36 m2), we will measure with comparable methods, a broad spectrum of ecological responses particularly related to the interaction between carbon fluxes (NPP, soil respiration) and species response traits, as well as environmental parameters that are critical for the integrated experiment-modeling framework, as well as for site-based analyses. By designing EDGE as a research platform open to the broader scientific community, with subplots in all replicates (n = 180 plots) set-aside for additional studies, and by making data available to the broader ecological community EDGE will have value beyond what we envision here.
The six sites were selected to capture the key environmental and ecological gradients of Central US grasslands and represent the major grassland ecosystem types (desert, shortgrass, mixedgrass, and tallgrass) of the region. Site selection criteria included: site characteristics (mean annual precipitation and temperature, dominant vegetation), access and site security, permission to build experimental infrastructure, participation in an existing or future network (e.g., LTER, NEON), and available site support and supporting data (e.g., LTER, USFWS or ARS).
Experimental Treatments and Plots
Our approach will be to impose a significant reduction in growing season precipitation (-66 % of ambient) over a 4-yr period. This is the equivalent of a ca. 50% reduction in annual precipitation because at all sites about 60-75% of annual precipitation falls in the growing season. We will impose this long-term drought either by reducing the size of each rainfall event (event size reduction, E) or by reducing the number of events (delayed rainfall treatment, D).
The control (C) treatment is included for comparison. At each site, the ambient (C) rainfall pattern will be reduced in two ways to impose a severe drought over a 4-yr period.
For the event size reduction treatment (E), each rainfall event will be passively reduced by a fixed proportion. Note that rain event number and the average number of days between events does not differ from ambient treatment.
For the reduced event number (D) treatment, shelters roofs will be removable to permit periods of complete rain exclusion alternating with periods of ambient rainfall inputs. Here, a + 10 mm rule is used to determine when roofs are on or off. When the cumulative precipitation amount in this D treatment falls 10 mm below the E treatment, the roofs are removed until the cumulative precipitation total is 10 mm greater than the E treatment. In this way, total precipitation amounts will be similar at the end of the growing season, but event number will be reduced and the average number of days between events increased, with no change in event size compared to the C treatment.
At each site, we will establish replicate 6 x 6 m experimental plots (n = 10 per treatment, including the control treatment) in a relatively homogeneous area (similar soils, vegetation, etc.) that is representative of the overall site. Plots will be arrayed such that each treatment will be co-located in a single block (n=10 blocks per site), with each block located at least 5 m apart.
The blocking will help control for environmental gradients if present. For each site, all plots within a block (including the control) will be located at least 2 m apart and trenched to 1-1.5 m and surrounded by a 6 mil plastic barrier to hydrologically isolate them from the adjacent soil, and each plot will be covered by the rainfall manipulation infrastructure. The 6 x 6 m plot size includes a 0.5 m external buffer to allow access to the plots and minimize edge effects associated with the infrastructure. The resulting 5 x 5 m area will be divided into 4 2.5 x 2.5 m subplots. One subplot will be designated for plant species composition sampling, two will for destructive sampling (ANPP, belowground productivity, soil sampling, etc.), and the fourth set aside for opportunistic studies.
Rainfall Manipulation Infrastructure
We will passively alter rainfall reaching the plots by using a version of a rainfall reduction shelter (Fig. 6) designed by Yahdjian and Sala (2002). Versions of these shelters (ranging from ~2 to 100 m2 ) are being used by the co-PIs at the Sevilleta, Konza Prairie and Shortgrass Steppe LTERs, as well as by many other ecologists, and thus, they are proven technology. The most significant environmental artifacts of these shelters are a 5- 10% reduction in light due to the acrylic Vshaped shingles and a ~ 20 cm edge effect (Yahdjian and Sala 2002). Shelters will consist of a steel frame that supports a roof. To cover the 36 m2 plots, the shelters will be constructed as modular 3 x 3 m units, with four units per plot. The roof of each modular unit will be slanted at 15° toward the edge of the plot, creating a 6 m long peak along the mid-line of the plot, with two lower 6 m long edges with gutters to move rainwater away from the plots. The peaked roof will facilitate run-off of rainfall and access to the plot, and the lower edge will be oriented to the prevailing wind direction to minimize blow-in. Average leaf canopy height varies among the desert/short-, midand tallgrass prairie sites (~0.2 to 0.6 m), and to maintain a consistent roof-to-canopy distance, peak height of the shelters will be 1.3, 1.55 and 1.8 m, with lower edges of the shelters at 0.5, 0.75 and 1.0 m, respectively, for the four grassland types. Construction of the shelters will begin in Yr 1 (after pretreatment measurements are taken) and treatments will be operational by the early spring of YR 2. For the ESR treatment, the roof will consist of clear acrylic (high light transmission, low yellowness index, UV transparent) v-shaped shingles arrayed at a density to passively reducing each rainfall event by ~66% (Fig. 6). For the REN treatment, the roof will consist of clear, corrugated polycarbonate (high light transmission, low yellowness index, UV transparent) to completely exclude rainfall. For both treatments, the roofs will be constructed to facilitate easy removal via a clamping system. The REN treatment roofs will then be manually deployed and removed at intermittent intervals (see Fig. 6 for more detail). Ambient plots will have a deer netting roof to achieve an average reduction in light similar to the rainfall reduction roofs.
Plant species composition, species traits, stem density, and light availability
In the subplot designated for species composition, we will establish a permanent 2 x 2 m sampling plots, which will be divided into four 1 x 1m quadrats in which canopy cover of each species will be visually estimated to the nearest 1%. For each site, these measures will be repeated at least twice during the growing season of each year to sample early and late season species. Maximum cover values of each species will be used to determine richness, diversity and dominance and changes in composition, species turnover, and species associations over time.
Collecting the Data:
Net primary production data is collected twice each year, spring and fall, for both sites. Spring measurements are taken in April or May when shrubs and spring annuals have reached peak biomass. Fall measurements are taken in either September or October when summer annuals have reached peak biomass but prior to killing frosts. Winter measurements are taken in February before the onset of spring growth.
Vegetation data is collected on a palm top computer. A 1-m2 PVC-frame is placed over the fiberglass stakes that mark the diagonal corners of each quadrat. When measuring cover it is important to stay centered over the vegetation in the quadrat to prevent errors caused by angle of view (parallax). Each PVC-frame is divided into 100 squares with nylon string. The dimensions of each square are 10cm x 10cm and represent 1 percent of the total area.
The cover (area) and height of each individual live (green) vegetative unit that falls within the one square meter quadrat is measured. A vegetative unit consists of an individual size class (as defined by a unique cover and height) of a particular species within a quadrat. Cover is quantified by counting the number of 10cm x 10cm squares filled by each vegetative unit.
Niners and plexidecs are additional tools that help accurately determine the cover a vegetative unit. A niner is a small, hand-held PVC frame that can be used to measure canopies. Like the larger PVC frame it is divided into 10cm x 10cm squares, each square representing 1% of the total cover. However, there are only nine squares within the frame, hence the name “niner.” A plexidec can help determine the cover of vegetative units with covers less than 1%. Plexidecs are clear plastic squares that are held above vegetation. Each plexidec represents a cover of 0.5% and has smaller dimensions etched onto the surface that correspond to 0.01%, 0.05%, 0.1%, and 0.25% cover.
It is extremely important that cover and height measurements remain consistent over time to ensure that regressions based on this data remain valid. Field crew members should calibrate with each other to ensure that observer bias does not influence data collection.
Grasses-To determine the cover of a grass clump, envision a perimeter around the central mass or densest portion of the plant, excluding individual long leaves, wispy ends, or more open upper regions of the plant. Live foliage is frequently mixed with dead foliage in grass clumps and this must be kept in mind during measurement as our goal is to measure only plant biomass for the current season. In general, recently dead foliage is yellow and dead foliage is gray. Within reason, try to include only yellow or green portions of the plant in cover measurement while excluding portions of the plant that are gray. This is particularly important for measurements made in the winter when there is little or no green foliage present. In winter, sometimes measurements will be based mainly on yellow foliage. Stoloniferous stems of grasses that are not rooted should be ignored. If a stem is rooted it should be recorded as a separate observation from the parent plant.
Forbs, shrubs and sub-shrubs (non-creosote)-The cover of forbs, shrubs and sub-shrubs is measured as the horizontal area of the plant. If the species is an annual it is acceptable to include the inflorescence in this measurement if it increases cover. If the species is a perennial, do not include the inflorescence as part of the cover measurement. Measure all foliage that was produced during the current season, including any recently dead (yellow) foliage. Avoid measuring gray foliage that died in a previous season.
Cacti-For cacti that consist of a series of pads or jointed stems (Opuntia phaecantha, Opuntia imbricata) measure the length and width of each pad to the nearest cm instead of cover and height. Cacti that occur as a dense ball/clump of stems (Opuntia leptocaulis) are measured using the same protocol as shrubs. Pincushion or hedgehog cacti (Escobaria vivipara, Schlerocactus intertextus, Echinocereus fendleri) that occur as single (or clustered) cylindrical stems are measured as a single cover.
Yuccas-Make separate observations for the leaves and caudex (thick basal stem). Break the observations into sections of leaves that are approximately the same height and record the cover as the perimeter around this group of leaf blades. The caudex is measured as a single cover. The thick leaves of yuccas make it difficult to make a cover measurement by centering yourself over the caudex of the plant. The cover of the caudex may be estimated by holding a niner next to it or using a tape measure to measure to approximate the area.
Height is recorded as a whole number in centimeters. All heights are vertical heights but they are not necessarily perpendicular to the ground if the ground is sloping.
Annual grasses and all forbs-Measure the height from the base of the plant to the top of the inflorescence (if present). Otherwise, measure to the top of the green foliage.
Perennial grasses-Measure the height from the base of the plant to the top of the live green foliage. Do not include the inflorescence in the height measurement. The presence of live green foliage may be difficult to see in the winter. Check carefully at the base of the plant for the presence of green foliage. If none is found it may be necessary to pull the leaf sheaths off of several plants outside the quadrat. From this you may be able to make some observations about where green foliage is likely to occur.
Perennial shrubs and sub-shrubs (non-creosote)-Measure the height from the base of the green foliage to the top of the green foliage, ignoring all bare stems. Do not measure to the ground unless the foliage reaches the ground.
Plants rooted outside but hanging into a quadrat-Do not measure the height from the ground. Measure only the height of the portion of the plant that is within the quadrat.
Additional Information on the personnel associated with the Data Collection / Data Processing
Nathan Gehres 2014-present; Michell Thomey 2012-2014