Risch et al. 2020 - Raw data for publication...
we assessed how the removal of mammalian herbivores (Fence) and fertilization with growth-limiting nutrients (N, P, K, plus nine essential macro- and micronutrients; NPK) individually, and in combination (NPK+Fence), affected potential and realized soil net Nmin across 22 natural and semi-natural grasslands on five continents. Our sites spanned a comprehensive range of climatic and edaphic conditions found across the grassland biome. We focused on grasslands, because they cover 40-50% of the ice-free land surface and provide vital ecosystem functions and services. They are particularly important for forage production and C sequestration. Worldwide, grasslands store approximately 20-30% of the Earth’s terrestrial C, most of it in the soil (Schimel, 1995; White et al., 2000).
Study sites and experimental design The 22 sites contributing to this project are part of the Nutrient Network Global Research Cooperative (NutNet, https://nutnet.umn.edu/). Mean annual temperature across our 22 sites ranged from -4 to 22°C, mean annual precipitation from 252 to 1,592 mm, and elevations from 6 to 4,261 m above sea level (Fig. 1, Supplementary Table S1). Soil organic C varied from 0.8 to 7.8%, soil total N from 0.1 to 0.6%, and the soil C:N ratio from 9.1 to 21.5. Soil clay content spanned from 3.0 to 35%, and soil pH from 3.4 to 7.6 (Supplementary Table S2). Thus, the sites covered a wide range of environments in which grasslands occur (Fig. 1, Supplementary Table S1 & S2). At each site, the effects of nutrient addition and herbivore removal were tested via a randomized-block design (Borer et al., 2014; Supplementary Fig. S1a). Three replicate blocks with 10 treatment plots each were established at each site, with the exception of the site at bldr.us, where only two blocks were established (Supplementary Fig. S1a). The 10 plots were randomly assigned to a nutrient or fencing treatment, but only a subset of four plots was used in the current study, each with a different treatment (see below; Supplementary Fig. S1a). All plots were 5 x 5 m and divided into four 2.5 x 2.5 m subplots (Supplementary Fig. S1b). Each subplot was further divided into four 1 x 1 m square sampling plots, one of which was set aside for soil sampling (Borer et al., 2014; Supplementary Fig. S1b). Plots were separated by at least 1 m wide walkways. In this study, we collected data from the following four treatments: (i) untreated control plots (Control), (ii) herbivore removal plots (Fence), (iii) plots fertilized with N, P, K, plus nine essential macro and micronutrients (NPK), and (iv) plots with simultaneous fertilizer addition and herbivore removal (NPK+Fence; Supplementary Fig. S1a). The number of years of treatment differed among sites (2 – 9 years since start of treatment; Supplementary Table S1). For the nutrient additions, all sites applied 10 g N m-2 yr-1 as time-release urea; 10 g P m-2 yr-1 as triple-super phosphate; 10 g K m-2 yr-1 as potassium sulfate. A micro-nutrient mix (Fe, S, Mg, Mn, Cu, Zn, B, Mo, Ca) was applied at 100 g m-2 together with K in the first year of treatments but not thereafter. The vertebrate herbivore removal treatment (Fence) was established by fencing two plots, one control and one NPK plot, within each of the blocks (Supplementary Fig. S1a). We designed the fences so that they would effectively exclude aboveground mammalian herbivores with a body mass of over 50 g (Borer et al., 2014). At the majority of sites, the height of the fences was 180 cm, and the fence design included wire mesh (1 cm holes) on the first 90 cm along with a 30 cm outward-facing flange stapled to the ground to exclude burrowing animals; climbing and subterranean animals may potentially still access these plots (Borer et al., 2014). For slight modifications in fence design at a few sites see Supplementary Table S3. While most sites only had native herbivores, a few sites (4) were also grazed by domestic animals (Supplementary Table S1). Potential and realized soil net N mineralization, ammonification, nitrification and other soil properties Each site participating in the study received a package containing identical material from the Swiss Federal Institute for Forest, Snow and Landscape Research (WSL) to be used for sampling and on-site N incubations. For the field incubation, we followed the protocol by Risch et al. (2015, 2019). Briefly, we drove a 5 x 15 cm (diameter x depth) steel cylinder 13.5 cm deep into the soil after clipping the vegetation at randomized locations in each plot. The top 1.5 cm of the cylinder remained empty to capture incoming N from run-off or deposition with a polyester mesh bag (mesh-size 250 µm) filled with 13.2 ± 0.9 g of acidic and alkaline exchanger resin (1:1 mixture; ion-exchanger I KA/ion-exchanger III AA, Merck AG, Darmstadt). The bag was fixed in place with a metal Seeger ring (Bruetsch-Rüegger Holding, Urdorf, Switzerland). We then removed 1.5 cm soil at the bottom of the cylinder and placed another resin bag to capture N leached from the soil column. The exchange resin was saturated with H+ and Cl- prior to filling the bags by stirring the mixture in 1.2M HCl for 1 h and then rinsing it with demineralized water until the electrical conductivity of the water reached 5 μS/cm. The cylinders were then re-inserted into the cored hole, level with the soil surface, and incubated for an average of 42 days (range 40 to 57days). The individual site coordinators chose the timing of incubation to start approximately six weeks prior to peak plant biomass production. All incubations were completed between February 2015 and January 2016 accounting for differences in growing season between northern and southern hemispheres. At the end of the incubation, the cylinders were re-collected and immediately shipped to WSL in an insulated box together with cold packs to halt further mineralization. Gloves were worn at all times to avoid contamination of the samples. Upon arrival at WSL, we extracted the resin bags and a 20 g subsample of sieved soil (4 mm) separately in a 100 ml PE-bottle with 80 ml 1 M KCl for 1.5 h on an end-over-end shaker and filtered through ashless folded filter paper (DF 5895 150, ALBET LabScience). We measured NO3- (colorimetrically; Norman & Stucki, 1981) and NH4+concentrations (flow injection analysis; FIAS 300, Perkin Elmer) on these filtrates. At the start of the field incubation, we additionally collected two soil cores of 5 x 12 cm (diameter x depth) in each sampling plot and composited them to measure potential soil net Nmin, soil chemical and biological properties (see below). We also collected an additional sample (5 x 12 cm) to assess soil physical properties, which remained within the steel cylinder. Both ends were tightly closed with plastic caps. Cores were carefully packed to avoid further disturbance, and together with the composited soil samples, were shipped to the laboratory at WSL within a few days after collection. From the composited samples, we extracted an equivalent of 20 g dry soil with KCl, as described above, and measured NO3- and NH4+concentrations. Realized soil net Nmin was then calculated as the difference between the inorganic N content of samples collected at the end of the incubation (plus N extracted from the bottom resin bag) and the N content at the beginning of the incubation; values were scaled to represent daily mineralization rates (mg N kg-1soil d-1; Risch et al., 2015). Realized soil net Nmin values represent an average period of 42 days prior to peak biomass, typically the highest period of biological activity, and not the entire year (Risch et al., 2019). A second subsample of the composited sample was used to determine potential soil net Nmin in the laboratory (Risch et al., 2019). Briefly, we weighed duplicate samples (8 g dry soil) into 50-ml Falcon tubes. Soil moisture was brought to 60% of the field capacity of each plot, the Falcon tubes tightly closed and then incubated at 20°C for 42 days in a dark room. Every week the Falcon tubes were opened and ventilated. At the end of the incubation, the soil samples were extracted the same way as described above and NO3- and NH4+ was determined. Potential soil net Nmin was calculated as the difference between the N content before and after the incubation and scaled to represent daily values (mg N kg-1soil d-1). Using our NO3- and NH4+ measures we also calculated potential and realized soil net nitrification and soil net ammonification to be able to better understand the drivers of fertilization and herbivore removal effects on potential and realized soil net Nmin. A third subsample of the composite soil sample was sieved (2 mm mesh) and microbial biomass (μg Cmic g-1 soil dry weight) was estimated by measuring the maximal respiratory response to the addition of glucose solution (4 mg glucose per g soil dry weight dissolved in distilled water; substrate-induced respiration method) on approximately 5.5 g of soil (Anderson & Domsch, 1978). The rest of the composited sample was dried at 65°C for 48 h, ground and sieved (2 mm mesh) to assess a series of soil chemical properties (Risch et al., 2019). We measured the percentage of clay as an indicator of soil texture (Gee & Bauder, 1986; Risch et al., 2019). Statistical analyses Potential and realized soil net Nmin were square root transformed to account for a highly skewed data distribution (yt = sign(y)*sqrt|y|; negative values in the data set impeded log transformation). To assess treatment effects on potential and realized soil net Nmin, we used linear mixed effects models (LMMs) fitted by maximum likelihood using the lme function from the nlme package (version 3.131.1; Pinheiro et al., 2016), R version 3.6.1; R Foundation for Statistical Computing. Treatment (Control, Fence, NPK, NPK+Fence) was a fixed factor, with site and block as random factors, where block was nested within site. We also tested for effects of time since start of treatments in preliminary analyses by adding total treatment years as an additional fixed factor. We did not find a significant effect of years of treatment, and thus dropped this variable from the models. The LMMs were corrected using varIdent if the homogeneity of variance criterion was not met (Pinheiro et al., 2016). To visualize our results, we calculated treatment effects using Cohens’ d statistic (Cohen, 1977; Koricheva et al., 2013). Note that calculating response ratios (or log response ratios) was not possible with our data, as we have both negative and positive values. We also fitted LMMs for potential and realized soil net ammonification and nitrification to gain more insight into how global change affects the processes underpinning potential and realized soil net Nmin. We also sqrt-transformed (see above) these dependent variables. Treatment was included as a fixed factor with random factors as described above. In addition, we assessed how potential and realized soil net Nmin were related to potential and realized soil net ammonification and nitrification, respectively. For this, we calculated site averages for each treatment separately. We then ran LMMs, with potential and realized soil net Nmin as the dependent variable, potential and realized soil net ammonification/nitrification as the independent ones. Site was included as a random factor. Based on our previous work (Risch et al., 2019) and the existing literature (Schimel & Bennett, 2004; Liu et al., 2017), we developed a priori causal conceptual models of relationships among treatments, environmental drivers, and potential and realized soil net Nmin (Supplementary Fig. S2) to test with structural equation modelling (SEM) using a d-sep approach (Shipley, 2009; Lefcheck, 2016). The variables included in the model were long-term climatic conditions, specifically, site-level mean annual precipitation (MAP) and temperature of the wettest quarter (T.q.wet), plot-level soil texture (clay content) and soil microbial biomass. Mean annual precipitation and T.q.wet were obtained from WorldClim (Hijmans et al., 2005) (http://www.worldclim.org/) and together with the experimental treatments were predicted to directly affect soil properties and soil net Nmin (Supplementary Fig. S2). Soil clay content was, in turn, predicted to affect microbial biomass and soil net Nmin. Because we determined microbial biomass prior to incubating the samples in the laboratory or field, we assumed that the abundance of these microbes would be responsible for N process rates and not vice versa (Supplementary Fig. S2). We tested our conceptual model (Supplementary Fig. S2) using the piecewiseSEM package (version 2.0.2; Lefcheck, 2016) in R 3.4.0, in which a structured set of linear models are fitted individually. This approach allowed us to account for the nested experimental design, and overcome some of the limitations of standard structural equation models, such as small sample sizes (Shipley, 2009; Lefcheck, 2016). We used the lme function of the nlme package to model response variables, including site as a random factor. Good fit of the SEM was assumed when Fisher’s C values were non-significant (p > 0.05). For all significant interactions between covariates and experimental treatments detected in the SEMs, we calculated treatment effect sizes, i.e. the differences in potential or realized soil net Nmin between Control and treatments (Fence, NPK, NPK+Fence) and plotted these values against the climate or soil covariates. Finally, we fitted LMMs for the soil variables included in our SEMs, with treatment as the fixed factor, and with site and block as random factors, where block was nested within site.
|Last updated||August 13, 2020|
|Created||August 13, 2020|
|License||ODbL with Database Contents License (DbCL)|
|Access Restriction||Restricted info only available to editors|