Dataset DOI: 10.5061/dryad.6wwpzgn9g

Description of the data and file structure

Data associated with a survey of nectar-alcohol concentrations in a Mediterranean climate common garden. An enzymatic assay was used to estimate ethanol concentrations of 147 samples of nectar from 29 species of flowering plants. The five supplementary tables here provide (1) ethanol calibration standard data used to estimate unknown concentrations, (2) a summary of nectar ethanol data at the level of species, (3) raw data associated with individual samples, as well as two tables with summary statistics for a series of phylogenetic generalized linear mixed models used to describe the observed variation in (4) ethanol concentrations and (5) sugar concentrations.

Files and variables

File: Supplementary_Figures_S1-S3.pdf

Titles: Fig. S1 Sugar concentrations in floral nectar; Fig. S2 Calibration regressions of ethanol standards; Fig. S3 Calibration regressions of ethanol standards through intercept

Captions: Fig S1: Sugar concentrations in floral nectar. Average sugar concentrations of floral nectar for 29 angiosperm species arranged according to phylogenetic relatedness. Green branches on the phylogeny refer to monocot species, and orange branches refer to eudicot species. Groups of species are classified into clades as indicated by gray alternating with white shading, as labeled at the bottom of the chart. Boxplots are color-coded by the expected pollinator for that plant species, with corresponding animal silhouettes (from PhyloPic) at the phylogeny tips: purple = hummingbird-pollinated (trochilophily), green = sunbird-pollinated (nectariniophily), salmon = insect-pollinated (entomophily), and cyan = “other” (including chiropterophily, ornithophily, and general zoophily). A black dashed line demarcates the average sugar concentration of 23.4% among all species. Boxplots represent the 25th to 75th percentiles of the interquartile range (IRQ), a thick line color-coded line represents the median, and whiskers represent the minimum and maximum. The median, boxplot and whisker calculations exclude outliers, which are shown as points and represent ethanol values greater than the IRQ multiplied by 1.5, as added to the 75th percentile. Sample size (n) is indicated above each boxplot. Fig S2: Calibration regressions of ethanol standards for four separate assays (plates 1-4), comparing a power regression (blue) and a linear regression (orange). The dashed red square represents the actual data range of assayed nectar samples using the power regression to facilitate visual comparison of the regression fit relative to low concentration calibration standards. Points are colored to correspond to the concentration of sugars added to some standards. Fig S3: Calibration regressions of ethanol standards for four separate assays (plates 1-4), comparing a power regression (blue) and a linear regression forced through the intercept (orange). The dashed red square represents the actual data range of assayed nectar samples using the power regression to facilitate visual comparison of the regression fit relative to low concentration calibration standards. Points are colored to correspond to the concentration of sugars added to some standards.

File: Table_S1_calibrations.csv

Title: Raw data associated with ethanol calibration standards used to create linear models to predict unknown concentrations of nectar ethanol.

Caption: The column plate refers to one of four assay plates used on four independent occasions to analyze the calibration standards alongside the nectar samples. The column absorbance indicates the unitless colorimetric absorbance value at 580 nm. The column etoh cal. (% w/w) indicates the known ethanol concentration of the standard. The column sug. conc. (% w/w) indicates the known sugar concentration of the standard.

File: Table_S2_species_data_with_photos.xlsx & Table_S2_species_data.csv

Title: Summary data for all species in the study.

Caption: The .xlsx version of this table contains the additional columns photo1 and photo2 which contain supplemental photographs of sampled flowers taken in the field that are not reproduced in the .csv version. The column N samples indicates how many individual flowers were averaged for the species, and the column n etoh-positive samples indicates only the value of the averages ethanol-positive samples. The column % etoh-positive samples (n/N×100) was used to calculate the estimated prevalence of yeasts across flower species. The column ethanol (% w/w) is an averaged value of the individual ethanol concentrations for that species. The column ethanol % of positive samples indicates the ethanol concentrations averaged among ethanol-positive samples only. The column max ethanol indicates the maximum ethanol concentration recorded for that species. The column sug. conc. (°Brix) is an averaged value of sugar concentrations, measured in degrees Brix. The column pollinator indicates which group of pollinators each species is associated with in the literature, grouped as either hummingbirds, sunbirds, insects, or other (a pooled category including bat- and general bird-pollinated flowers, indicated in parenthesis). The column pollinator refs contains in-line citations for literature references for the pollinator column, with full references available below the table. The number of references per species is first indicated with a number in parenthesis. Each entry is followed by a description of where in the reference the relevant information can be found. The column clade refers to one of five larger taxonomic groups of angiosperms that each flower species was assigned to, namely asterids, malvids, and the family Cactaceae among the eudicots, and the order Asparagales and the family Bromeliaceae among the monocots. The column Fig1. sequence indicates the sequence of species that correspond to figure 1 in the main text.

File: Table_S3_nectar_samples.csv

Title: Raw data associated with individual nectar samples used in the study.

Caption: The column plate refers to one of four assay plates used on four independent occasions to analyze the calibration standards alongside the nectar samples. The column absorbance indicates the unitless colorimetric absorbance value at 580 nm. The column ethanol (% w/w) indicates the individual ethanol concentration of that sample and the column sug. conc. (°Brix) the individual sugar concentration, measured in degrees Brix. The column pollinator indicates which group of pollinators each species is associated with in the literature, grouped as either hummingbirds, sunbirds, insects, or other (a pooled category including bat- and general bird-pollinated flowers, indicated in parenthesis). The column time of day indicates the number of hours after 8 AM that the sample was collected. The columns temperature (°C) and % relative humidity correspond to the time of sample collection.

File: Table_S4_etoh_pglmm.csv

Title: A list of summary statistics of a series of phylogenetic generalized linear mixed models (PGLMMs) associated with ethanol concentrations.

Caption: The column R formula indicates the code used in the ‘pglmm’ function of the ‘phyr’ package in R, with all models composed of ethanol as the dependent variable, followed by a tilde after which are listed any fixed variable(s) used in the model, followed by either (1 + sp__), indicating two random effects (species effects and phylogenetic effects) or by (1 + sp), indicating only species effects as a random effect. Among the fixed variables, sug. conc. indicates sugar concentrations (as measured in degrees Brix), clade indicates membership in one of five phylogenetic groups (asterids, malvids, the family Cactaceae, the order Asparagales and the family Bromeliaceae), cotyledon indicates membership in either monocots and dicots, and pollinator indicates the pollinator associated with the species via published literature (hummingbird, sunbird, insect, or ‘other’). Two models did not contain any fixed effects (denoted with the number ‘1’) were used to inspect the relative weight of random effects on the model. The column AIC indicates Akaike’s Information Criterion, and models with the  lowest AIC values are considered to be the most parsimonious in balancing the explanatory power of the model and the number of explanatory variables in the model. The column R2_lik indicates the coefficient of determination of the model, calculated using the likelihood method. The columns species_random effect size, phylogenetic_random effect size, and residual_random effect size indicate the relative random effect size of species categories, higher phylogenetic relationships, and individual sample effects, respectively. The Intercept column indicates the mean ethanol concentrations when fixed variable(s) are at an x-intercept of zero, and the column pval_intercept indicates the associated p-value. The column fixed_1 indicates a fixed trait, and is followed by effect_1 which specifies the fixed trait’s intercept (mean-centered scaled), and a column titled pval1 indicates the p-value associated with that intercept. These are followed by four more series of three similarly named columns, to accommodate up to five fixed groups.

File: Table_S5_brix_pglmm.csv

Title: A list of summary statistics of a series of phylogenetic generalized linear mixed models (PGLMMs) associated with sugar concentrations

Caption: The column R formula indicates the code used in the ‘pglmm’ function of the ‘phyr’ package in R, with all models composed of sugar concentrations (sug.conc.) as the dependent variable, followed by a tilde after which are listed any fixed variable(s) used in the model, followed by either (1 + sp__), indicating two random effects (species effects and phylogenetic effects) or by (1 + sp), indicating only species effects as a random effect. Among the fixed variables pollinator indicates the pollinator associated with the species via published literature (hummingbird, sunbird, insect, or ‘other’). Two models that did not contain any fixed effects (denoted with the number ‘1’) were used to inspect the relative weight of random effects on the model. The column AIC indicates Akaike’s Information Criterion, and models with  lowest AIC values are considered to be the most parsimonious in balancing the explanatory power of the model and the number of explanatory variables in the model.  The column R2_lik indicates the coefficient of determination of the model, calculated using the likelihood method. The columns species_random effect size, phylogenetic_random effect size, and residual_random effect size indicate the relative random effect size of species categories, higher phylogenetic relationships, and individual sample effects, respectively. Note: one model, ethanol ~ 1 + (1 | sp), does not have a phylogenetic effect component and the corresponding phylogenetic_random effect size column for that model is presented as ‘NA’. The Intercept column indicates the mean ethanol concentrations when fixed variable(s) are at an x-intercept of zero, and the column pval_intercept indicates the associated p-value. The column fixed_1 indicates a fixed trait, and is followed by effect_1 which specifies the intercept (mean-centered scaled) of a single group belonging to a fixed trait, and a column titled pval1 indicates the p-value associated with that intercept. These are followed by two more series of three similarly named columns, to accommodate up to three fixed groups.