Data Package Metadata   View Summary

Data was obtained using a full factorial combination of nutrient treatments (control or fertilized) and vertebrate consumer density (control or fenced) for a total of four treatments. Replicates were obtained from three blocks in Kentucky and Kansas, five in Minnesota, and six in Iowa. The nutrient-addition treatment was an annual addition of 10 g N x m-2 x yr-1 as time-release urea, 10 g P x m-2 x yr-1 as triple super phosphate, 10 g K x m-2 x yr-1 as potassium sulfate, and 100 g x m-2 of a micronutrient mix (6% Ca, 3% Mg, 12% S, 0.1% B, 1% Cu, 17% Fe, 2.5% Mn, 0.05% Mo, and 1% Zn). N, P, and K were applied annually, and the micronutrient mix was applied once at the start of the study to avoid toxicity of largely immobile micronutrients. The fences were 2.1 m tall and excluded aboveground, nonclimbing mammalian herbivores. The lower 0.9 m was composed of 1-cm woven wire mesh with a 0.3 m outward-facing flange stapled to the ground to exclude digging animals. The top 1.2 m was composed of five equally spaced rows of woven wire to prohibit large animals from entering the plots.

At peak biomass, the cover of all plant species was estimated visually in a 1 x 1 m permanent quadrat located randomly in each plot. Cover estimates included our focal host, A. gerardii. At the same time (i.e., at peak biomass), all aboveground plant biomass (live and dead) was collected in two 0.1 x 1 m strips, sorted to separate live (current year's growth) and dead (growth from past years), dried to a constant mass, and weighed to the nearest 0.01 g. A single mature leaf from three or four individual A. gerardii plants was collected in 2–4 blocks at each site in August of 2014, for a total of 170 samples (KY = 44, KS = 45, IA = 32, MN = 49) with 8–15 plants sampled per treatment at each site. The collected leaf was the second leaf above the uppermost senescent leaf. Leaves were stored in a cooler (about 4degC) until they were surface-sterilized in the lab by immersing them for 1 min each in water, 75% ethanol, 0.4125% sodium hypochlorite (bleach solution), 75% ethanol, and sterile distilled water. Following surface sterilization, samples were stored at -80degC.

Leaves were ground in liquid nitrogen with a mortar and pestle, total genomic DNA was extracted using the Qiagen Plant Mini Extraction Kit (Qiagen N.V., Venlo, the Netherlands), and standardized to 20 ng/lL. Fungal genomic libraries were made by amplifying the ITS region as in Nguyen et al. (2015). Each sample was barcoded with unique 7 base-pair (bp) sequences and the ITS1 region was amplified with the standard primers ITS1f (50- AATGATACGGCACCACCGAGATCTA-CAC-GG-CTTGGTCATTTAGAGGAAGTAA-30) and ITS2 (50-CAAGCAGAAGACGGCATACGAGAT-barcode-CG-GCTGCGTTCTTCATCGATGC-30). The ITS1 primer includes an Illumina Nextera adaptor, linker sequence, and a bar code. Polymerase chain reaction (PCR) was done in triplicate using Roche FastStart High Fidelity Taq (Roche, Indianapolis, Indiana, USA) with annealing temperatures at 51deg, 53deg and 55degC. PCRs with no visible bands were repeated once. Two negative controls (distilled water instead of DNA) were included in every PCR set. Amplicons from the triplicate PCR reactions were pooled for each sample, purified using the QIAQuick Purification Kit (Qiagen N.V., Venlo, the Netherlands), and quantified using the Quant-iT dsDNA HS Assay kit in a Qubit Fluorometer (ThermoFisher, Waltham, Massachusetts, USA). Equal amounts of these purified libraries (25 ng) were pooled and sequenced in Illumina MiSeq at the University of Minnesota Genomics Center (UMGC). Sequence data from the MiSeq runs were combined and analyzed using an analytical pipeline adapted from Nguyen et al. (2015). Adapter and distal priming sites were removed using cutadapt v1.7.1 (Martin 2011) and Trimmomatic v 0.32 (Bolger et al. 2014). Short sequences, homopolymers, and sequences containing ambiguous bases were removed using Mothur v.1.34.4 (Schloss et al. 2009). The clean sequences were then dereplicated and clustered into operational taxonomic units (OTUs) at a 97% cutoff, with chimera sequences removed using USEARCH (Edgar 2010) 5 followed by additional re-clustering using UCLUST (Edgar 2010) implemented in Qiime (Caporaso et al. 2010). Nguyen et al. (2015) found that combining clustering algorithms (i.e., chain-picking) was more stringent and more accurately recovered the fungal taxa in a constructed mock community than using a single clustering method. After removing singleton OTUs, a representative sequence for each OTU was selected and assigned a taxonomy using BLAST (Basic Local Alignment Search Tool) against the UNITE fungal database v 7.0 with BLAST v 2.2.28+ (Camacho et al. 2009). Only annotations greater than 80% identity and greater than 80% alignment length were retained. The pooled OTU counts from the distilled water negative controls were subtracted from each sample. The 748 reads from the distilled water controls represented a fraction of the total number of reads (less than 0.0001) and were clustered into 27 OTUs with a median abundance of 4.5 reads per sample. Seventy-two percent of the negative control reads were classified as a single OTU matching Phoma calidophila (Ascomycota, Dothideomycetes, Pleosporales). Four samples with less than 1,000 reads were removed from the analyses. OTUs that could not be assigned matches against the UNITE database were excluded from the taxonomy report, but included in statistical analyses.