Spectra were collected using either an internal pulsed-xenon light source (Jaz, Ocean Optics, Dunedin, FL USA) or a deuterium-halogen light source (DH-2000-BAL (Ocean Optics, Dunedin, FL USA) with a Spectralon white standard (Labsphere, North Sutton, NH) and dark correction to measure percent reflectance from 300-700 nanometers, which is the general range of color perception by many flower-visiting insects, including bees and flies. Floral tissue was illuminated with a collimated beam oriented normal to the floral surface, and spectra were collected by a probe positioned at a 45-degree azimuth, composed of a collimating lens and optical fiber (fiber diameter = 400 microns) connected to the spectrophotometer. We utilized SpectraSuite version 2.0.162 software for capturing spectral data (Ocean Optics, Dunedin, FL USA). Spectra were collected with an integration time ranging from 50-250 milliseconds, a boxcar smoothing width ranging 3-25 nanometers, and with a range of 10-30 average spectral scans (species-specific details of these parameters are included in datasheet). In collection of floral spectra, at least one single petal of the floral unit was measured for each flowering species, or in the instance that a single petal was too small to cover the entire sampling area, multiple petals were overlaid to provide enough surface for the spectrometer to collect a reflectance reading. Within each floral unit, if there was a noticeable change in coloration in the human vision color spectrum or morphological component (e.g.: petal v. labellum), reflectance readings were obtained from various portions across the floral unit. We also searched for any change in ultraviolet reflectance range across the floral display area by viewing live spectrometer reflectance output while moving across the floral tissue surface. Any noted differences within a floral unit for each species is included in datasheet.

Functional traits, particularly those that impact fitness, can shape the ecological and evolutionary relationships among coexisting species of the same trophic level. Thus, examining these traits and properties of their distributions (underdispersion, overdispersion) within communities can provide insights into key ecological interactions (e.g., competition, facilitation) involved in community assembly. For instance, the distribution of floral colors in a community may reflect pollinator-mediated interactions between sympatric plant species, and the phylogenetic distribution of color can inform how evolutionary contingencies can continue to shape extant community assemblages. Additionally, the abundance and species richness of the local habitat may influence the type or strength of ecological interactions among co-occurring species. To evaluate the impact of community size and species richness on mechanisms shaping the distribution of ecologically relevant traits, we examined how floral color (defined by pollinator color vision models) is distributed within co-flowering assemblages. We modeled floral reflectance spectra of 55 co-flowering species using honeybee (Apis mellifera) and syrphid fly (Eristalis tenax) visual systems to assess the distributions of flower color across 14 serpentine seep communities in California. We found that phylogenetic relatedness had little impact on the observed color assemblages. However, smaller seep communities with lower species richness were more overdispersed for flower color than larger, more species-rich communities. Results support that competitive exclusion could be a dominant process shaping the species richness of flower color in smaller-sized communities with lower species richness, but this is less detectable or overwhelmed by other processes at larger, more speciose communities.

Legend (also included in datasheet header) -- 'filename': code for specific spectral file; 'species': plant species; 'Internal_contrast_estimated': whether internal contrast was estimated for species; 'Internal_contrast_detail': detailed location of spectra collection on floral unit; 'spectrometer': spectrometer model (USB2000+ , USB4000, Jaz, Ocean Optics, Dunedin, FL USA); 'integration_time_microseconds': given in microseconds (range of 50-250 milliseconds, which is 50,000-250,000 microseconds); 'number_spectra_averaged': ranges from 10-30 scans; 'boxcar_smoothing_width': ranges 3-25 nanometers; 'number_pixels_processed': number of pixels processed per spectral measurement; 'columns at 300-700': binned reflectance values as percentages (%), binned to 1-nm intervals;