Herbivory Weakens the Stability of a Floral Polymorphism Mia Gaughan1-2,7-8, Vincent S. Pan2-4,7-8, Sylvie Martin-Eberhardt2-3,5, William C. Wetzel6, Kadeem J. Gilbert2-5. 1:Department of Biology, Washington University in St. Louis, St. Louis, MO, USA 2: W. K. Kellogg Biological Station, Michigan State University, Hickory Corners, MI, USA 3: Department of Integrative Biology, Michigan State University, East Lansing, MI, USA 4: Ecology, Evolution, and Behavior Program, Michigan State University, Easting Lansing, MI, USA 5: Department of Plant Biology, Michigan State University, East Lansing, MI, USA 6: Land Resources and Environmental Sciences, Montana State University, Bozeman, Montana, USA 7: Corresponding authors: vsbpan [at] gmail [dot] com, gmia [at] wustl [dot] edu 8: Co-first authors Dataset DOI: 10.5281/zenodo.18807895 General descriptions 480 Hesperis matronalis were subjected to an induction (clipping) treatment across 40 sites. We measured a bunch of fitness surrogates to look at whether herbivory affects frequency dependent selection. We performed a choice/no-choice assay to confirm what was going on with the herbivores. See the associated manuscript for more details on the methods. Manuscript abstract Herbivores have profound impacts on the maintenance of plant diversity. While simple optimal foraging models predict that herbivores should selectively attack common plant phenotypes, models that allow for differential plant resource acquisition can predict a preference towards rarer phenotypes, leading to the destabilization of intraspecific polymorphism. We show that herbivores may play such a destabilizing role in the dimorphic Hesperis matronalis system by weakening negative frequency-dependence. Across 40 patches of varying morph ratios, locally rarer H. matronalis morphs grew larger in the early season but received disproportionately more herbivory in the mid- and late season, leading to lower female fitness and survival. We rescued the advantage of being rare by experimentally reducing later herbivory through a plant immunity induction treatment. Feeding assays with a mustard-feeding caterpillar Trichopulsia ni confirmed that herbivores had higher survival on locally rare morphs and preferred to feed on rarer morphs. Our results are consistent with the hypothesis that herbivores may have evolved to use rarity as a cue to track resources, potentially weakening the advantage of being rare, a critical component of stable coexistence. We argue that this phenomenon of insect herbivores weakening the maintenance of plant diversity is extremely common in stable polymorphic systems. Data descriptions We use 'NA' to denote missing values in general. base_map.jpg This is a screen shot of google maps displaying the sites studied. DRFDS_manuscript_cleaned_field_data_03_06_2025.csv This is the field data + no-choice experiment. Each row is a plant or caterpillar that ate the specific plant. This is the field data + no-choice experiment. Each row is a plant or caterpillar that ate the specific plant. siteID: The unique identifier of the 40 sites plantID: Unique plant identifier trt: the clipping treatment of the plant (c = control, a = aggregated clipping, d = dispersed clipping) leaf_number_pre: number of leaves counted before the leaves were clipped largest_leaf_len_pre: largest leaf length in cm two days after the leaves were clipped height: the height of the plant in cm roughly a month later clipping when the plant is flowering herb_intensity: herbivory severity at the time of flower survey (n = no damage, l = low damage, m = medium damage, h = high damage) flower_count: the number of flowers counted date_flower_survey: moth-day-year of flower survey date_trt: the date of clipping treatment (month-day-year) lat: the latitude in degree decimals of the plant lon: the longitude in degree decimals of the plant notes: general notes date_pollen_processed: the date pollen is scored pollen_alive: the number of pollen grains counted that were stained red with tetrazolium (0.5% TTC for >24h) pollen_dead: the number of pollen grains counted that did not stain with TTC anther_prop_red: the proportion surface area of anther stained red in the TTC staining solution anther_prop_red_logit: logit transformed anther_prop_red. num_stalks: the number of stalks (either flowering or not flowering) of the same plant leaf_feeding_start: the year-month-day H:M:S of when the no-choice leaf feeding assay with T. ni started leaf_feeding_end: the year-month-day H:M:S of when the no-choice leaf feeding assay with T. ni ended cat_wt: the milligram of the T. ni caterpillar post weight (after leaf feeding trial) cat_dead: is the caterpillar dead after the no-choice feeding trial (1 = yes, 0 = no) herbivory_assay_notes: notes specific to the no-choice herbivory assay seed_collection_date: year-month-day of when the seeds were collected. late_herb_prop: Proportion of herbivory on the plant in the late season survey late_herb_survey_date: year-month-day of when late season herbivory was surveyed late_herb_sruvey_notes: Notes specific to late season herbivory survey silique_count: Number of siliques counted silique_weight: the weight in grams of the siliques collected. PC1: the first principal component of the RGB values of the scanned flower color hex: the hex code of the scanned flower color red: the red intensity (0-1) of the flower color green: the green intensity (0-1) of the flower color blue: the blue intensity (0-1) of the flower color clipped: which clipping treatment did the plant get? (clipped: clipped; control: not clipped) site_PC: The average PC1 value of all plants measured at a site flower_color: a rough classification of flower color as either purple or white. alive_next_year: whether the plant survived until next year (1: yes, 0: no) next_year_survey_date: year-month-day of when the plant survival was surveyed next_year_notes: notes specific to survival survey DRFDS_manuscript_cleaned_choice_experiment_10_22_2023.csv This is the choice experiment. Each row is a caterpillar. containerID: The experimental arena ID cat_sourceID: colony ID of where the caterpillar comes from cat_pre_wt: Caterpillar pre-weight in milligrams leaf_trt: Whether the leaf disk comes from a purple (p) or white (w) morph. Each leaf disk is ordered in a clockwise fashion, separated by a dash. percent_herb: Percent herbivory on each leaf disk in the same order is leaf_trt cat_post_wt: Caterpillar post weight in milligrams date_experiment_setup: year-month-day of when of the experiment was setup date_experiment_takedown: year-month-day of when the experiment ended notes: general notes dead: Did the caterpillar die by the end of the experiment? (1: yes, 0: no) Code DRFDS_analysis.html The knitted output of the manuscript analyses DRFDS_analysis.Rmd The code for manuscript analyses DRFDS_analysis_utils.R Some functions used in DRFDS_analysis.Rmd for analysis. Dependencies A list of code dependencies, including their version numbers, can be found in the file dependencies.txt Usage Do whatever you want with the data. For questions, email Vincent Pan (vsbpan [at] gmail [dot] com).