Identifying the interactions of functional, biotic, and abiotic factors that define plant–insect communities has long been a goal of community ecologists. Metabolomics approaches facilitate a broader understanding of how phytochemistry mediates the functional interactions among ecological factors. Ceanothus velutinus communities are a relatively unstudied system for investigating chemically mediated interactions. Ceanothus are nitrogen-fixing, fire-adapted plants that establish early post-fire, and produce antimicrobial cyclic peptides, linear peptides, and flavonoids. This study takes a metabolomic approach to understanding how the diversity and variation of C. velutinus phytochemistry influences associated herbivore and parasitoid communities at multiple spatiotemporal scales. Herbivores and foliar samples were collected over three collection times at two sites on the east slope of the Sierra Nevada Mountain range. Foliar tissue was subjected to LC-MS metabolomic analysis, and several novel statistical analyses were applied to summarize, quantify, and annotate variation in the C. velutinus metabolome. We found that phytochemistry played an important role in plant–insect community structure across an elevational gradient. Flavonoids were found to mediate biotic and abiotic influences on herbivores and associated parasitoids, while foliar oligopeptides played a significant positive role in herbivore abundance, even more than abundance of host plants and leaf abundance. The importance of nutritional and defense chemistry in mediating ecological interactions in C. velutinus plant–herbivore communities was established, justifying larger scale studies of this plant system that incorporate other mediators of phytochemistry such as genetic and metageomic contributions.

4.1. Study Site

Ceanothus (Rhamnaceae) is a species-rich North American genus of woody perennial that includes about 55 species, with 51 species found in Western North America [67] and with the highest diversity (38 species) and endemism in the California Floristic Province [68,69]. Ceanothus have symbiotic relationships with nitrogen-fixing bacteria [70] and play an important role in ecosystem nitrogen availability [67]. Ceanothus velutinus (Tobacco brush) was chosen as the focal species for this study. C. velutinus is typically found at elevations between 1600 m a.s.l. and 2500 m a.s.l. and at times in very dense stands. Eighteen plots were established across two sites on the Eastern slope of the Sierra Nevada, referred to as the Dog Valley (39°31′39.12″ N, 120° 1′53.40″ W) and Mt. Rose locations (39°20′33.55″ N, 119°52′11.33″ W). Both sites are characterized by eastern facing slopes with Dog Valley having a lower minimum elevation (1632 m a.s.l.) and slope (46 m/m) and Mt. Rose having a higher maximum elevation (2550 m a.s.l.) and slope (145 m/m). The plots were 10 m in diameter and set up in triplicate along an elevational gradient between 1600 m a.s.l. and 2600 m a.s.l. at “low”, “medium”, and “high” elevation bands (referred to here as subsite). Elevation for each band was established relative to the elevational range of each site, ±50 m. Each plot was at least 10 m away from any trail or road, spaced at least 100 m apart from any other plot, and contained at least one C. velutinus individual, chosen as the center point of the plot. The plots were established in 2018. Ecological and chemical data were collected on the same day from each plot monthly from May to October of 2018. Due to the extra time involved with locating sites and setting up the plots, we repeated data collections in 2019 to ensure we captured a complete dataset for all sites over the season.

4.2. Ecological Data

For each plot, the total plant species diversity and the estimated leaf abundance of each plant within the plot was determined. Leaf abundance was estimated by counting the leaves on two representative branches from a given plant, then estimating the number of such branches on the plant. Each C. velutinus individual within the plot was numbered and the lepidoptera were collected using a beat sheet. Voucher samples were collected for all C. velutinus individuals and are available from the herbarium at the University of Nevada, Reno Natural History Museum. All collected caterpillars were reared individually in plastic cups in the lab at UNR to either adult or parasitoid emergence. Host plant foliage was replaced every 2 days and all pupae were checked daily. The adult Lepidoptera and parasitoids that emerged from the pupae were allowed to fully harden and then placed in a freezer for storage before pinning and identification. Each caterpillar collected was photographed and assigned a unique voucher code, linking the individual to the plot and plant number in the database. Morphotypes were later condensed and validated using reared adult specimens and larval photographs as references.

4.3. Sample Preparation

Immediately after caterpillars were collected using a beat sheet, leaf samples were collected from each C. velutinus individual for phytochemical profiling. Young leaves were cut from the terminal end of the branches, placed in paper bags, and stored in a cooler with dry ice until the end of the collection day, and subsequently stored at −80 °C. Leaves were then transferred to plastic centrifuge tubes with a tungsten steel bead, lyophilized, and ground to a fine powder at 30 Hz for two minutes using a tissue lyser (Qiagen Tissuelyser II; Hilden, Germany). The ground plant material (~20 mg) was transferred to screw-cap scintillation vials with 1 mL of 70% aq solution of HPLC-grade, denatured ethanol (Fisher, Pittsburgh, PA, USA) in 18 MΩ water. The samples were vortexed, sonicated for 10 min and incubated overnight on a shaker at room temperature and randomized before filtration through a 96 well filter plate with 1.0 mL capacity and 1.0 μM glass fiber filters (Acroprep 96, Pall Corporation, Port Washington, NY, USA) into a 96 well plate with 1 mL glass inserts, sealed with a silicone cap mat, and stored at −20 ℃ for less than one week until analysis.

4.4. LC-TOF analysis of foliar plant tissue

Chromatography was performed on an Agilent 1200 analytical HPLC equipped with a binary pump, autosampler, column compartment and diode array UV detector, coupled to an Agilent 6230 Time-of-Flight mass spectrometer via an electrospray ionization source (ESI-TOF; gas temperature: 350 °C, flow: 8 L/m; nebulizer pressure: 35 psig; VCap: 3500 V; fragmentor: 175 V; skimmer: 65 V; octopole: 750 V). Internal standards were selected to represent the major metabolite classes found in Ceanothus: flavonoid glycosides (narignin), flavonoid aglycones (naringenin), and peptides (cholecystokinin fragment), but were not found in plants. Multiple internal standards were used to account for differences in ionization efficiency among phytochemical classes. Extracts (1.00 μL) were co-injected with 1.00 μL of internal standard mix (naringenin, 100. μM; naringin, 101 μM, Cholecystokinin Fragment 30–33 Amide, (52 μM) Sigma-Aldrich) and eluted at 0.400 mL/min through a Kinetex EVO C18 column (Phenomenex, 2.1 × 100 mm, 2.6 μ, 100 Å) at 40 °C. The linear binary gradient was comprised of buffers A (water containing 10 mM ammonium acetate) and B (acetonitrile) changing over 25 min accordingly: 0 min 5% B, ramp to 40% B at 8 min, ramp to 100% B at 14 min, 14–20 min 100% B ramping to 0.70 mL/min before re-equilibrating the column from 21–25 min at 5% B, 0.4 mL/min.

The agilent-formatted raw data were converted to mzML format in ProteoWizard MSConvert 3.0 [71] before processing using the Bioconductor package XCMS [72] in R. Peaks and were extracted using the centwave function before retention time correction using obiwarp, peak density correspondence analysis, and gap-filling to yield 1308 aligned retention time and m/z bins (chromatographic features). Features were visually inspected to remove 260 peaks representing areas of baseline noise above peak threshold. CAMERA [73], an XCMS wrapper, was then used to identify groups of features (pseudospectra) having similar retention time and were correlated across samples (r > 0.7), with similar (r > 0.5) peak shape and by isotopic pattern. The CAMERA command annotate was used to identify and remove features resulting from ammonium adducts that might lead to the misidentification of nitrogenous compounds. The feature within a pseudospectrum group with the highest abundance was then chosen to represent the peak area of each of the 300 resulting pseudospectra. To minimize between-run batch effects, each peak bin was subjected to ANOVA analysis grouped by batch and 133 peaks exhibiting significant batch effects (p < 0.10) were excised.

These data include the LC-TOF peak abundances of m/z-rt peak bins generated from analysis of Ceanothus velutinus leaf extracts and the ecological data with which they are associated. ReadMe files will be uploaded with keys describing column names in tables.