Morphological and chemical attributes of diaspores in myrmecochorous plants have been shown to affect seed dispersal by ants, but the relative importance of these attributes in determining seed attractiveness and dispersal success is poorly understood. We explored whether differences in diaspore morphology, elaiosome fatty acids, or elaiosome phytochemical profiles explain the differential attractiveness of five species in the genus Trillium to eastern North American forest ants. Species were ranked from least- to most-attractive based on empirically-derived seed dispersal probabilities in our study system, and we compared diaspore traits to test our hypotheses that more attractive species will have larger diaspores, greater concentrations of elaiosome fatty acids, and distinct elaiosome phytochemistry compared to the less attractive species. Diaspore length, width, mass, and elaiosome length were significantly greater in the more attractive species. Using gas-chromatography mass spectrometry, we found significantly higher concentrations of oleic, linoleic, hexadecenoic, stearic, palmitoleic, and total fatty acids in elaiosomes of the more attractive species. Multivariate assessments revealed that elaiosome phytochemical profiles, identified through liquid-chromatography mass spectrometry, were more homogeneous for the more attractive species. Random forest classification models (RFCM) identified several elaiosome phytochemicals that differed significantly among species. Random forest regression models revealed that some of the compounds identified by RFCM, including methylhistidine (α -amino acid) and d-glucarate (carbohydrate), were positively related to seed dispersal probabilities, while others, including salicylate (salicylic acid) and citrulline (L-α -amino acid), were negatively related. These results supported our hypotheses that the more attractive species of Trillium—which are geographically widespread compared to their less-attractive, endemic congeners—are characterized by larger diaspores, greater concentrations of fatty acids, and distinct elaiosome phytochemistry. Further advances in our understanding of seed dispersal effectiveness in myrmecochorous systems will benefit from a portrayal of dispersal-unit chemical and physical traits, and their combined responses to selection pressures.

This data file is broken down into three datasets, each collected with a different laboratory method. All data were collected on a set of seeds from five southeastern US species of Trillium (T. catesbaei, T. cuneaum, T. lancifolium, T. discolor, and T. decumbens). Approximately 100 seeds were collected from mature fruits of each species just prior to natural dehiscence from a total of eight study sites across the southern Appalachian region in the summer of 2018. Upon return to the laboratory, we stored seeds at -20°C for 1 month prior to diaspore morphological analyses and 4 months prior to chemical analyses.

The first data set is Trillium diaspore (seed + elaiosome) morphology data. To address our first study objective, which was to quantify interspecific differences in diaspore morphology, we measured diaspore and elaiosome length, width, and mass for 26 diaspores of each species, representing multiple individuals from each study site (n = 130). We took the fresh mass of entire diaspores (g) and then measured the length and width of diaspores (mm) using digital calipers (Mitutoyo Digimatic Caliper, 0.01 mm resolution, Sakado, Japan). We then removed the elaiosome from each seed using a straight razor and repeated the above measurements for the elaiosome. Elaiosome-seed mass ratios were calculated by dividing the mass of the elaiosome by the mass of the entire diaspore.

The second data set is Trillium elaiosome fatty acid data. To address our second study objective, which was to quantify interspecific differences in concentrations of elaiosome fatty acids among species of Trillium, we assessed triglyceride, diglyceride, and free fatty acid forms using gas-chromatography mass spectrometry (GC-MS). We removed elaiosomes from frozen seeds as described above and placed them in 15-mL glass scintillation vials. Twelve samples were generated, representing the five study species, each from at least two field sites. Each sample contained a minimum of 30 mg of elaiosomes from one individual plant. The number of elaiosomes used for each sample depended on the species but ranged from 14 to 35. To each sample, we added 2.0 mL of the extraction solvent (2 : 1 dichloromethane : methanol). Samples were centrifuged for 30 m (170 rpm), and supernatants were siphoned off and transferred to new glass scintillation vials. We repeated this treatment three times, until 6 mL of supernatants were generated for each sample. The vials containing all of the combined extraction supernatants were placed in a nitrogen drying apparatus until all the extraction solvent had been evaporated. The dry residue in each vial was suspended in 1 mL isooctane : ethyl acetate (10 : 1) and was then applied to an alumina column (4-cm Pasteur pipette filled with flash alumina pre-equilibrated with isooctane : ethyl acetate 10 : 1), as in Boulay et al. (2006). The column was eluted with 4 ml isooctane : ethyl acetate (10 : 1) to yield triglyceride fractions for each sample. Elution of the diglyceride fraction was realized with 5 mL of isooctane : ethyl acetate (3 : 1). The free fatty acids were captured after elution with 6 mL of isooctane : ethyl acetate : acetic acid (75 : 25 : 2). Fractions were collected in standard glass test tubes. Resulting fractions of triglycerides, diglycerides, and free fatty acids were dried under nitrogen for 3 h. Residue was dissolved in 1 mL of the extraction solvent, and then 400 μL of this volume was transferred to a 2.0 mL vial containing 20 μg of octadecane in methanol as an internal standard. Solutions were dried again under nitrogen and subsequent steps taken to convert fatty acids to their methyl esters for GC-MS analysis. Transesterification for tri- and diglyceride fractions was accomplished by adding 100 µL of methanolic KOH (0.5 M) to the dry triglyceride and diglyceride fractions. After 30 m, the reaction was stopped with 100 µL of HCl (1 N). Then 20 µL of hexane was added to the mixture. Methylation of the free fatty acid fraction was carried out by treating the solution with 200 µL of toluene, 1.5 mL of methanol, and 300 µL of an 8% (w/v) solution of HCL in methanol/water (85 : 15, w/v), sequentially, to the fraction, as in Ichihara and Fukubayashi (2009). The fraction was then heated to 100 C for 1 h. All the obtained fractions were dried one final time and dissolved in 100 μL of isooctane, of which 1 µL was injected into an Agilent 7820 gas chromatograph coupled with a 5977 mass spectrometer set to 70eV electron ionization and equipped with an HP5-MS column (30 m x 0.25 mm i.d., 0.25 μm film thickness; Agilent Technologies, Santa Clara, CA, USA). Ultra-pure helium was used as a carrier gas at a flow rate of 1 mL min -1, a split flow ratio of 100:1, and a front inlet temperature of 275 °C. The following oven conditions were employed: initial temperature 120 °C, hold time 2 min; ramp 1: 15°C min^-1 to 250 °C, hold time 0 min; ramp 2: 5 °C min^-1 to 300°C, hold time 10 min; total run time of 30.67 min. Data were recorded as TIC chromatograms. We produced standard curves by running five concentrations (10, 1, 0.5, 0.1, 0.05, and 0.025 mg/mL) of a fatty acid methyl ester standard mixture of methyl linoleate (20 wt %), methyl linolenate (20 wt %), methyl oleate (20 wt %), methyl palmitate (20 wt %), and methyl stearate (20 wt %) (CRM1891, Millipore Sigma, Merck KGaA, Darmstadt, Germany). Linearity across the relevant range of concentrations was evaluated visually and by calculating the slope of the line and R² values for each compound. GC-MS produced TIC chromatograms for each elaiosome fraction (free, di-, and triglycerides). Peaks were identified based on comparisons of retention times and spectra with authentic standards. Compounds for which we did not have standards were tentatively identified using the NIST 14 library (Scientific Instrument Services, 1027 Old York Rd., Ringoes, New Jersey, U.S.A). Peak areas were determined using the Integrator Trapezoid feature in OpenChrom (Wenig & Odermatt, 2010) and concentrations were determined for known compounds based on response factors of standards relative to the internal standard. Concentrations for tentatively identified compounds were estimated as internal standard equivalents. All concentrations were converted to a percentage of elaiosome fresh weight for final reporting and statistical analyses.

The third data set is Trillium elaiosome phytochemistry data. To address our third study objective, which was to assess interspecific differences in elaiosome phytochemistry, we characterized and compared profiles of elaiosome metabolites using liquid-chromatography mass spectrometry (LC-MS). Each of the five study species was represented by six sample replicates (N = 30). Samples were taken from 2 – 6 individuals from the representative study sites. Using sterile techniques, we removed elaiosomes from frozen diaspores of each species using a straight razor. We recorded the fresh mass of single elaiosomes (g) and placed each elaiosome in a 2.0 mL centrifuge tube. Sample contents were suspended in 1.3 mL of extraction solvent (40 : 40 : 20 HPLC grade methanol, acetonitrile, water with 0.1% formic acid), and were kept at 4°C. Extraction proceeded for 20 min at -20°C before samples were centrifuged for 5 min (16.1 relative centrifugal force [rcf]) at 4°C. Supernatants were transferred to new vials. The remaining elaiosome contents were resuspended in 200 μL of cold (4°C) extraction solvent. The extraction again proceeded for 20 min at -20°C before being centrifuged for 5 min (16.1 rcf) at 4°C. Once more, we transferred supernatants to the vials, and added another 200 μL of extraction solvent to the pelleted elaiosomes for a final wash by repeating the extraction once more. The vials containing all of the combined extraction supernatants were then placed in a nitrogen drying apparatus until all the extraction solvent had been evaporated. We resuspended the residue in 300 μL of sterile water to isolate the polar fraction of phytochemicals and transferred this to 300 μL autosampler vials. Samples were immediately placed in a 4°C autosampler for LC-MS analysis. A 10 μL injection of each sample was separated through a Synergi 2.5 micron Hydro-RP 100 Å, 100 mm × 2.00 mm LC column (Phenomenex, Torrance, CA, USA) maintained at 25°C. The mass spectrometer and chromatographic separation were performed similar to the method outlined in Lu et al., 2010. The eluent was introduced into the mass spectrometer via an electrospray ionization source in negative mode before entering an Exactive Plus orbitrap mass spectrometer (Thermo Scientific, Waltham, MA, USA) through a 0.1-mm internal diameter fused silica capillary tube. Samples were run with a spray voltage of 3 kV, a nitrogen sheath gas flow rate of 10 units, a capillary temperature set at 320°C, and an AGC target set to 3e6. Samples were analyzed in full scan mode with a resolution of 140,000 and a scan window of 85 to 800 m/z for from 0 to 9 min and 110 to 1000 m/z from 9 to 25 min. Solvent A consisted of 97:3 HPLC grade water : methanol, 10 mM tributylamine, and 15 mM acetic acid. Solvent B was HPLC grade methanol. The mobile phase gradient from 0 to 5 min was 0% B, from 5 to 13 min was 20% B, from 13 to 15.5 min was 55% B, from 15.5 to 19 min is 95% B, and from 19 to 25 min was 0% B while maintaining a constant flow rate of of 200 μL/min. Raw files obtained from Xcalibur MS software (Thermo Electron Corp., Waltham, MA) were converted into the mzML format using ProteoWizard (Chambers et al., 2012). The converted files were imported into MAVEN (Metabolomic Analysis and Visualization Engine for LC–MS Data), a software package (Clasquin et al. 2012). Peaks for the known metabolites were picked in MAVEN, which automatically performs non-linear retention time correction and calculates peak areas across samples, using a preliminary mass error of ± 20 ppm and retention time window of 5 min. The University of Tennessee, Knoxville Biological and Small Molecule Mass Spectrometry Core (BSMMSC), through which this analysis took place, has replicated and expanded the method of Rabinowitz and coworkers (Lu et al., 2010). Final metabolite annotations were made using a library of 263 retention time-accurate m/z pairs taken from MS1 spectra. The annotation parameters have been verified previously with pure standards in the course of establishing the method. For a metabolite to be annotated as a known compound, the eluted peak had to be found within 2 min of the expected retention time, and the metabolite mass had to be within ± 5 ppm of the expected value. Metabolite identities were confirmed using the MAVEN software package (Clasquin et al., 2012), and peak areas for each compound were integrated using the Quan Browser function of the Xcalibur MS Software. LC-MS produced relative concentrations of phytochemical compounds across samples, which were compared using z-transformed peak areas from extracted ion chromatograms.

Trillium diaspore morphology data set: This dataset is contained in a single excel file entitled "Trillium_diaspore_morphology," and consists of a data matrix (11 columns, 131 rows). ID refers to the replicate for each species (1 - 26). Data are stacked. Units are included in parentheses after the column header. Replicates include information about the study sites (including the latitude/longitude coordinates for each site) from which they were collected. 

Trillium elaiosome fatty acid data set: This dataset is contained in an excel file entitled "GC_MS_Trillium," containing 4 sheets. The data were generated using GC-MS as specified in the Methods above. The sheets include: IRF calcs, Sample Areas, Sample Quant mg_ml, and Sample Quant % fw. 

IRF calcs: This sheet contains four data tables. In the first table ("Concentrations"), the concentration of the internal standard, octadecane, and the concentrations of the five methyl esters in the standard mixture are recorded (mg/mL). In the second table ("Raw Data Standard Areas"), the standard areas from the 9 GC runs for the internal standard and the five methyl esters in the standard mixture are recorded (these are area under the curve values without a unit). The third table ("RRFs") records the relative response factors for each of the five methyl esters in the standard mixture (relative to the internal standard) for each of the 9 GC runs.  The fourth table ("Avg") simply records the averaged RRFs for each of the five methyl esters in the standard mixture. 

Sample Areas: This sheet consists of a data matrix with 37 columns and 19 rows. The data contained in the matrix are the raw sample areas (areas under the curve, no units) for each of 13 compounds identified in the elaiosome samples. Column names specify the fatty acid form (di = diglyceride, tri = triglyceride, fa = free fatty acids) followed by an underscore and a two letter abbreviation for the species specific epithet ("cu" = cuneatum, "ca" = catesbaei, "de" = decumbens, "di" = discolor, and "la" = lancifolium) and a two letter abbreviation for the study site ("ca" = cave, "jg" = jocassee gorges, "wf" = whitewater falls, "bh" = blue hole, "om" = old mine, "pb" = pocket branch, "tb" = tilton bridge). The third row of the data matrix specifies the exact concentration of the internal standard (mg/mL), and the fourth row of the matrix specifies the fresh weight of each elaiosome sample (mg). The numbers contained in the parentheses after the names of each compound are the average retention times. 

Sample Quant mg_ml: This sheet is similar to Sample Areas; it is a data matrix with 37 columns and 15 rows. The data contained in the matrix are the concentration of the compounds per sample (mg/mL); these are generated from a formula that multiples the internal standard concentration for each sample by the raw area under the curve for each sample; that product is then divided by the raw area under the curve of the internal standard for each sample. 

Sample Quant % fw: The final sheet is again similar to both Sample Areas and Sample Quant mg_ml; it contains a data matrix with 37 columns and 15 rows, corresponding to the 36 elaiosome samples and the 13 compounds (plus some column headers). The data contained in this matrix are the sample concentrations as percentage fresh weight; these are generated from a formula that multiplies the sample concentration (mg/mL) by the total solvent volume in which the extract was suspended for analysis (0.1 mL), divides that product by the fraction of the total extract that was in the aliquot (0.4 mL), and then divides that value again by the sample fresh weight (mg). Finally, the value is multiplied by 100 to obtain a percentage. The last row in this matrix is a summation of all 13 compounds, to generate a total percentage of fresh weight that is represented by fatty acids. A second table in this sheet lists the fresh weight of each sample (mg) and the number of elaiosomes used to generate the sample, and then calculates the average free oleic and free linoleic acid per elaiosome per species. 

Trillium elaiosome phytochemistry data set: This data set consists of two .csv files. The first is entitled "LC_MS_Trillium_Knowns.csv", and it contains a data matrix consisting of 124 columns and 31 rows. The first four columns are metadata columns, and contain information about the geographic distribution of each species (endemic v. widespread), the species specific epithet, the average probability of seed removal in the field, and two-letter abbreviation for the study site from which the sample was collected ("TB" = Tilton Bridge, "PB" = Pocket Branch, "OM" = Old Mine, "CA" = Cave, "WF" = WhiteWater Falls, "BR" = Boat Ramp, and "JG" = Jocassee Gorges). The remaining columns are the names of chemical compounds identified using Liquid-chromatography mass spectrometry. Each cell contains the raw area under the curve for each compound in each sample. 

The second file is entitled "LC_MS_Trillium_Unknowns.csv", and it is identical to the Knowns data matrix, except that it contains > 7,500 columns, each corresponding to an unknown chemical compound detected by LC-MS. These are simply denoted by numbers.