Data Package Metadata   View Summary

# Experimental design Rainfall manipulation treatments commenced in multiple years and spanned a range of experimental durations at the time of field sampling, from 4 years to 14 years. Plot size depended on the experiment, either 2.5 x 2.5 m or 2.5 x 5.0 m. Most plots contained patches of B. eriopoda, except drought treatments in the older experiments where herbaceous patches exhibited mortality. Most plots also contained a central Prosopis shrub except one experiment (\"LTREB2\"), which excluded Prosopis from the original design. Treatment levels depended on the experiment; all experimental plots had control plots that received ambient rainfall with no rainout shelter or irrigation system. Water treatments were achieved using rainout shelters that decreased incoming precipitation by 50% or 80% and automated irrigation systems that simultaneously applied 50% or 80% of incoming precipitation. During precipitation events, shelters intercepted and redirected incoming rainfall to a PVC irrigation system connected to sprinklers surrounding +50% or +80% treatment plots by means of a solar-powered pump. Manipulation intensities were based on extremes of historical precipitation data for the region. Monthly precipitation sums were obtained from Jornada Basin LTER meteorological stations, available on the Environmental Data Initiative, nearest to the experimental plots (Thatcher and Bestelmeyer 2021), summed over the growing season (June-September), and then adjusted according to the rainfall manipulation treatment. The range of growing season precipitation amount achieved experimentally was 11-206 mm over the experimental years. We present precipitation amount as a continuous variable for our results due to the multi-year, multi-treatment nature of our study design. Some years were considerably drier than others were, and the rainfall amount received is more biologically meaningful to consider as a variable than categorical treatments. Refer to the included \"precip_treatments.csv\" table for a information about the effective precipitation amounts in all experiments and treatments included in this study.  # Field sampling and stable isotope analyses Field collection of leaves and soils for isotopic analyses took place during peak biomass months following the summer monsoon: August 2011, September 2012, September 2018, and September 2020. Leaves were collected from three Bouteloua patches (when possible; some drought plots had zero Bouteloua cover) and the central Prosopis shrub (if present). Specifically, the leaves were collected from the four cardinal directions and the center (54). Surface soil samples (0-10 cm) were collected using 2.54 cm diameter soil corer and composited from five sub-plot samples that were representative of the general ground cover of the plot, ranging from bare ground to underneath dominant plant patches. Soil samples were then passed through a 2 mm sieve. Foliar and soil samples were subsequently dried at 70°C for 48 h and ground into a fine powder using a Desktop High Energy Vibratory Ball Mill (VQ-N ball mill Thomas Scientific) for foliar samples and a mortar and pestle for soil samples. The stainless grinding tools were carefully cleaned with ethanol between each sample. Foliar samples were encapsulated in 4 x 6 mm tins while soil samples were encapsulated in 5 x 9 mm tins. All 2011 and 2012 samples were run on a GVI IsoPrime and an Elementar Cube elemental analyzer at the Boston University Stable Isotope Laboratory. QA/QC procedures for the BU Stable Isotope Lab may be found here: https://www.bu.edu/sil/quality/. One analytical replicate was run per 10 samples, and any anomalous results were rerun. In-house standards of peptone and glycine calibrated to USGS 40 and 41 were alternately analyzed after every 15 unknown samples. All 2018 and 2020 samples were run in analytical triplicates and flash combusted with a coupled continuous-flow elemental analyzer-isotope ratio mass spectrometer system consisting of a Costech EA interfaced to a Delta Advantage peripheral at the METAL Core Laboratory of Arizona State University. Calibration curves for 2018 and 2020 plant samples were built using tomato leaves (NIST 1573a) and for 2018 and 2020 soils using low-nitrogen Montana soil (NIST 2711). In-house glycine standards calibrated to USGS 40 and 41 were analyzed after every third unknown sample and at the beginning and end of each run. Standards and unknowns were corrected for linearity, and unknowns were normalized to isotopic values of standard reference materials using a two-point calibration curve of in-house standards calibrated to USGS 40 and 41 standard reference materials. Stable isotope nitrogen ratios are standardized to atmospheric air and expressed in permil (‰) as: $$\delta = \frac{R_{sa}}{R_{std} - 1}$$ Where _Rsa_ is the molar 15N/14N isotopic ratio of the sample and _Rstd_ is the molar isotopic ratio of atmospheric air (0.0036765). _Please note: This dataset does not include a thorough analysis of any carbon data, and all soil carbon data should be interpreted with caution as carbonates were NOT removed before analysis._ All standards and unknown samples presented in this manuscript underwent quality assessment and quality control. Acceptable accuracy of tomato leaf standards and in-house glycine or peptone standards was defined as having a residual error of ≤ 0.2‰. Acceptable accuracy of the low-nitrogen Montana soil standard was defined as having a residual error of ≤ 0.3‰. Acceptable precision for all standards was defined as having a standard deviation of ≤ 0.2‰. A blank (empty tin cup) was included at the beginning of each analytical run for all 2011, 2012, and 2018 plants and soils, and after every 5-8 unknown samples for 2020 soils. Unknown samples adhered to QA/QC when samples fell within the standard calibration range and exhibited a standard deviation among analytical replicates of ≤ 0.2‰. If unknowns had a standard deviation > 0.2‰, an attempt to meet QA/QC requirements was first conducted by removing one outlier replicate, reducing the number of analytical replicates to 2. If this did not resolve the precision measurement, the sample was flagged, re-ground, and re-run on the IRMS. Any data that remained flagged after re-running were discarded from the final data set. ## Provenance This dataset contains provenance metadata. See below for data sources.