Seed Production, Pollinator Dependence, and Plant-Pollinator Networks in Southeastern Native Wildflowers
Ecological Context and Pollinator Dependency Dynamics
Global Angiosperm Pollination Dependencies:
Approximately of all angiosperm species rely directly on animal-mediated pollination for successful reproduction.
Remaining angiosperm species depend on abiotic vectors (e.g., wind) or operate predominantly through self-pollination breeding systems.
Plant species diversity and their corresponding animal pollinator communities are experiencing significant global declines driven by anthropogenic stressors, including habitat loss, pesticide exposure, and agricultural intensification.
Knowledge Gaps in Native Wildflower Biology:
Management plans for crop systems and rare or threatened plant species frequently incorporate detailed evaluations of pollination biology.
Corresponding baseline ecological data—such as specific pollinator reliance, seed set dynamics, and visitor assemblages—remains largely unquantified for most native wildflower species.
Although government and conservation programs actively promote planting native wildflowers to support pollinator populations in human-impacted landscapes, the reciprocal reproductive benefits to wildflower populations are rarely evaluated.
Functional Importance of Floral Traits and Pollination Services:
Key plant traits—including floral morphology, bloom phenology, and reward provision (nectar and pollen)—dictate the composition of visiting insect assemblages.
Pollinator-dependent wildflower species that lack adequate insect visitor service risk reproductive failure and local extirpation, subsequently diminishing future floral resource availability for insect communities.
Direct benefits of insect visitation include increased pollen grain deposition on stigmas, enhancing fertilization rates and total seed quantity per inflorescence.
Indirect benefits stem from cross-pollination and genetic outcrossing, which enhance seed quality and seed weight.
Heavier seeds generally display superior ecological fitness, higher germination rates, and improved seedling establishment compared to lower-weight seeds produced under pollinator-excluded conditions.
Strategic Restoration Applications:
Degraded Habitats: Wildflower species capable of self-pollination or partial self-compatibility can establish and reproduce in landscapes suffering from depleted insect pollinator populations, ensuring continuous floral resource provision.
Restored or Intact Habitats: Highly pollinator-dependent species reinforce mutualistic interactions, increase ecological redundancy, and maximize community-wide plant-pollinator stability.
Seed Mix Architecture: Optimal conservation seed mixes combine both self-compatible and strongly pollinator-dependent native species to accommodate generalist and specialist pollinators across varying degradation gradients.
Southeastern Native Wildflower Species and Conservation Values
Nine native southeastern United States wildflower species across four plant families were evaluated for reproductive reliance and insect visitor relationships:
Apiaceae:
Eryngium yuccifolium Michx. (Rattlesnake master):
Fruit Type: Schizocarp.
Conservation Recommendations: Promoted by the USDA Natural Resources Conservation Service (USDA-NRCS) Conservation Stewardship Program (CSP) for supporting pollinators and beneficial insects, as well as mitigating soil erosion.
Ecological Associations: Host plant for the specialized stem-boring moth Papaipema eryngii Bird (Lepidoptera: Noctuidae).
Breeding System Context: Presumed to be xenogamous with partial self-compatibility.
Asteraceae:
Bidens laevis (L.) Britton, Sterns, & Poggenb. (Smooth beggar-ticks):
Fruit Type: Capsela.
Conservation Recommendations: Designated by the Xerces Society as possessing high value for native pollinators; utilized for erosion control and habitat restoration.
Echinacea purpurea (L.) Moench (Eastern purple coneflower):
Fruit Type: Capsela.
Conservation Recommendations: Recommended by both the USDA-NRCS CSP and the Xerces Society to enhance native pollinator habitat.
Gaillardia pulchella Foug. (Indian blanket):
Fruit Type: Capsela.
Conservation Recommendations: Promoted by the USDA-NRCS CSP for upland wildlife plantings, highway right-of-way revegetation, energy exploration reclamation, and range seed mixes.
Verbesina alternifolia (L.) Britton ex Kearney (Wingstem):
Fruit Type: Capsela.
Conservation Recommendations: Recommended by both the USDA-NRCS CSP and the Xerces Society.
Verbesina virginica L. (Frostweed):
Fruit Type: Capsela.
Conservation Recommendations: Recommended by both the USDA-NRCS CSP and the Xerces Society.
Fabaceae:
Desmanthus illinoensis (Michx.) MacMill. ex B.L. Rob. & Fernald (Illinois bundleflower):
Fruit Type: Schizocarp.
Conservation Recommendations: Widely utilized in revegetation projects; host plant for several skipper butterfly species; historically presumed to be self-pollinating.
Desmodium floridanum Chapm. (Florida ticktrefoil):
Fruit Type: Legume (loment containing 3 to 5 deltoid segments).
Conservation Recommendations: Recommended by the USDA-NRCS CSP; host plant for the velvetbean caterpillar (Anticarsia gemmatalis Hübner: Erebidae).
Lamiaceae:
Monarda punctata L. (Spotted beebalm):
Fruit Type: Nutlet.
Conservation Recommendations: Recommended by both the USDA-NRCS CSP and the Xerces Society.
Experimental Methodology and Plot Architecture
Site Locations and Planting Layout:
Eight species were established at the Auburn University Bee Lab in Auburn, Alabama, USA ().
One naturally occurring population of Bidens laevis was evaluated along a pond bank at Auburn University's Agricultural Heritage Park ().
Plot preparation at the Bee Lab involved applying a concentration of glyphosate (Roundup, Monsanto, St. Louis, MO, USA) over a strip, followed by rototilling two weeks later.
Planting occurred on 30 April 2019. Each species was assigned to three contiguous subplots forming a plot strip, separated from adjacent species strips by a unplanted buffer.
Total experimental strip layout spanned in length.

Propagation, Pre-treatments, and Planting Densities:
Seed-derived species (D. illinoensis, D. floridanum, M. punctata, V. alternifolia, V. virginica) were sourced from Roundstone Seed LLC (Upton, KY, USA).
Stratification and scarification treatments:
V. alternifolia and V. virginica: Cold moist stratification in dampened sand at for .
D. illinoensis and D. floridanum: Mechanical scarification via vigorous dry sand shaking (Quikrete Premium Play Sand) for .
M. punctata: No pre-treatment required.
Indoor germination began in February 2019 using 72-cell starter trays with soil (Scotts Miracle-Gro) under 16-hour light cycles ( full-spectrum grow lights). Seedlings were thinned to 1 plant per cell upon developing true leaves.
Hardening off commenced on 16 April 2019 over two weeks under tree shade with daily incremental outdoor exposure increases starting at 08:00.
Transplants sourced as 1-year-old potted plants from Superior Trees Inc. (Lee, FL, USA) to guarantee first-year blooms: E. purpurea, E. yuccifolium, G. pulchella.
Subplot density arrangements to achieve full vegetative cover:
Arrangement A (): Monarda punctata, Eryngium yuccifolium.
Arrangement B (): Gaillardia pulchella, Verbesina alternifolia, Verbesina virginica.
Arrangement C (): Echinacea purpurea, Desmanthus illinoensis, Desmodium floridanum.
Bidens laevis: Designated a () plot; individual plant density was unquantified due to dense growth.
Bloom Phenology Across Species:
Field observations documented distinct, overlapping bloom windows from May through November 2019.

Insect Exclusion Field Protocol:
Treatments were assigned across individual plants to control for plant-level variation.
Bagged Treatment (Insect-Excluded): Unopened flower buds enclosed in fine-mesh () polyester pail strainers (The Cary Company, Addison, IL, USA) to exclude insect visitors while allowing passive wind pollination.
Open Treatment (Open-Pollinated): Unopened flower buds tagged with red flagging tape, allowing free access to foraging insects.
Defined Flowering Units:
Composite head: B. laevis, D. illinoensis, E. purpurea, E. yuccifolium, G. pulchella, V. alternifolia, V. virginica.
Panicle cluster: D. floridanum.
Apical 3 glomerules of a single branch: M. punctata.
Seed collection occurred post-senescence. Heads were dried naturally in paper bags indoors, after which mature seeds were counted and weighed to the nearest thousandth of a milligram () on a digital analytical balance (VWR-124B).
Targeted Sweep Netting Surveys:
Timed sweep netting surveys were performed on active plots containing flowering units between 9 May 2019 and 7 November 2019.
Sampling conditions required warm temperatures (), sunny skies ( cloud cover), and low precipitation probability () across morning, midday, and afternoon intervals.
Captured insects were either identified and released in the field or chilled on ice for laboratory identification using generic and specific taxonomic keys.
Statistical Modeling Framework
Seed Count and Weight Analyses:
Evaluated using Generalized Linear Mixed Models (GLMMs) in R via the
glmmTMBpackage.Models were fitted individually by plant species due to distinct data distribution structures.
Model diagnostic checks (zero-inflation, dispersion, residual fits) were confirmed using the
DHARMapackage; model selection was guided by Akaike's Information Criterion (AIC).Random intercepts accounted for experimental structure: bag number for most species, or bag number nested within plant ID for species with multi-plant replication (M. punctata, E. yuccifolium, G. pulchella).
Zero-inflated negative binomial distributions were specified for overdispersed seed count data (D. illinoensis, G. pulchella, B. laevis, V. virginica).
Linear mixed-effects models with case-resampling bootstrapping ( iterations via
lme4andparameterspackages) were implemented for species violating standard GLMM assumptions (V. alternifolia and E. yuccifolium seed counts; V. alternifolia and E. purpurea seed weights).Seed weight analysis was restricted exclusively to inflorescences that produced mature seed, using Gaussian or Gamma (log-link) distributions.
Insect Visitor Community Analyses:
Visitor abundance and species richness per survey were analyzed using negative binomial GLMMs.
Models incorporated an offset for the total number of survey events conducted per plant species to standardize varying sampling effort across bloom durations.
Low-visitation species (D. floridanum, D. illinoensis, V. virginica) were excluded from statistical GLMM comparisons due to non-convergence from zero-inflation.
Pairwise group comparisons were evaluated using Tukey-adjusted estimated marginal means () via the
emmeanspackage.
Plant-Pollinator Network Structure:
Network interactions were constructed and analyzed using the
bipartitepackage (plotweb(),networklevel(),specieslevel()).Evaluated network topology parameters included connectance, modularity (), overall interaction specialization (), web asymmetry, and topological robustness to node loss.
Quantitative Seed Production and Weight Dynamics
Overview of Evaluated Inflorescences:
A total of insect-excluded (bagged) and open-pollinated (open) inflorescences were processed across all wildflower species.

Seed Production per Inflorescence (Seed Count):
Significant Pollinator Enhancement:
Bidens laevis: Open-pollinated inflorescences produced seeds versus in bagged inflorescences ( increase; CL: ; ).
Desmanthus illinoensis: Open-pollinated inflorescences produced seeds versus in bagged inflorescences ( increase; CL: ; ).
Echinacea purpurea: Open-pollinated inflorescences produced seeds versus in bagged inflorescences ( increase; CL: ; ).
Monarda punctata: Open-pollinated inflorescences produced seeds versus in bagged inflorescences ( increase; CL: ; ).
Verbesina virginica: Open-pollinated inflorescences produced seeds versus in bagged inflorescences ( increase; CL: ; ).
Non-Significant Treatment Differences (Self-Pollination Potential):
Eryngium yuccifolium: Open-pollinated inflorescences produced seeds versus in bagged inflorescences ( ratio; CL: ; ).
Gaillardia pulchella: Open-pollinated inflorescences produced seeds versus in bagged inflorescences ( ratio; CL: ; ).
Verbesina alternifolia: Open-pollinated inflorescences produced seeds versus in bagged inflorescences ( ratio; CL: ; ).
Desmodium floridanum: Insufficient seed yield prevented statistical seed count modeling.
Weight per Seed per Inflorescence (Seed Quality):
Significantly Heavier Seeds in Open Treatments:
Bidens laevis: Open seeds weighed versus when bagged (difference of ; CL: ; ).
Echinacea purpurea: Open seeds weighed versus when bagged (difference of ; CL: ; ).
Eryngium yuccifolium: Open seeds weighed versus when bagged ( proportional increase; CL: ; ).
Gaillardia pulchella: Open seeds weighed versus when bagged ( heavier; CL: ; ).
Verbesina alternifolia: Open seeds weighed versus when bagged (difference of ; CL: ; ).
Verbesina virginica: Open seeds weighed versus when bagged (difference of ; CL: ; ).
Inverse Weight Trend:
Desmanthus illinoensis: Bagged seeds were significantly heavier () than open-pollinated seeds () (difference of ; CL: ; ), reflecting resource allocation trade-offs where lower seed set in bagged plants concentrated maternal resources into fewer seeds.
Non-Significant Weight Difference:
Monarda punctata: Open seeds weighed versus when bagged (difference of ; CL: ; ).
Summary of Sample Sizes and Reproductive Output Statistics:
Species Name | Inflorescences (Count: Bagged / Open) | Seed Count Effect | Inflorescences (Weight: Bagged / Open) | Seed Weight Effect | Seed Failure Count (Bagged / Open) |
|---|---|---|---|---|---|
Eryngium yuccifolium | 44 / 34 | No diff () | 44 / 32 | Open higher () | 0 / 2 |
Bidens laevis | 18 / 25 | Open higher () | 15 / 25 | Open higher () | 3 / 0 |
Echinacea purpurea | 23 / 19 | Open higher () | 23 / 19 | Open higher () | 0 / 0 |
Gaillardia pulchella | 40 / 32 | No diff () | 38 / 31 | Open higher () | 3 / 1 |
Verbesina alternifolia | 203 / 253 | No diff () | 177 / 252 | Open higher () | 26 / 1 |
Verbesina virginica | 24 / 37 | Open higher () | 22 / 37 | Open higher () | 2 / 0 |
Desmanthus illinoensis | 32 / 40 | Open higher () | 10 / 40 | Bagged higher () | 22 / 0 |
Desmodium floridanum | NA / NA | Insufficient | NA / NA | Insufficient | NA / NA |
Monarda punctata | 22 / 24 | Open higher () | 21 / 24 | No diff () | 1 / 0 |
Insect Floral Visitor Abundance, Diversity, and Network Topology
Survey Effort and Visitor Composition:
A total of timed sweep netting surveys yielded individual insect floral visitors across genera/species, families, and insect orders.
Survey counts by plant species: B. laevis ( events over ), E. purpurea ( events over ), V. alternifolia ( events over ), G. pulchella ( events over ), M. punctata ( events over ), E. yuccifolium ( events over ), V. virginica ( events over ), D. illinoensis ( events over ), and D. floridanum ( events over ).
Order-Level Visitor Breakdown:
Bees (Hymenoptera): ( individuals).
Wasps (Hymenoptera): ( individuals).
Butterflies and Moths (Lepidoptera): ( individuals).
Flies (Diptera): ( individuals).
Beetles (Coleoptera): ( individuals).

Visitor Abundance and Richness Comparisons:
Insect Abundance per Survey (Estimated Marginal Means):
Gaillardia pulchella: visitors/survey ( CL: ; Tukey group a).
Eryngium yuccifolium: visitors/survey ( CL: ; Tukey group ab).
Monarda punctata: visitors/survey ( CL: ; Tukey group b).
Bidens laevis: visitors/survey ( CL: ; Tukey group bc).
Echinacea purpurea: visitors/survey ( CL: ; Tukey group bc).
Verbesina alternifolia: visitors/survey ( CL: ; Tukey group c).
Insect Richness per Survey (Estimated Marginal Means):
Highest visitor richness observed on E. yuccifolium (), M. punctata (), and G. pulchella () (all Tukey group a).
Statistically lower richness observed on B. laevis (; group ab), E. purpurea (; group ab), and V. alternifolia (; group b).
Plant-Pollinator Interaction Network Architecture:

Network Topology Metrics:
Connectance: (proportion of realized interaction links relative to all possible links).
Modularity (): (high modularity reflecting compartmentalized sub-communities).
Specialization (): (moderate specialization across the network).
Web Asymmetry: (higher richness of insect visitors relative to plant species).
Robustness: Pollinators demonstrated higher resilience to plant species loss () than plants to pollinator loss ().
Dominant Insect Floral Visitor Accounts:
Bees:
Halictus poeyi/ligatus (Halictidae): Total individuals visiting 7 plant species; of visits () concentrated on G. pulchella.
Apis mellifera (Apidae): Total individuals visiting 8 plant species; of visits () on E. purpurea and () on B. laevis.
Bombus impatiens (Apidae): Total individuals visiting 3 species; of visits () concentrated on late-blooming B. laevis.
Xylocopa virginica (Apidae): Total individuals visiting 7 species, most frequently M. punctata ( visits) and B. laevis ( visits).
Wasps:
Scolia nobilitata (Scoliidae): Total individuals visiting 5 species; of visits occurred on E. yuccifolium () and V. virginica ().
Scolia dubia (Scoliidae): Total individuals observed exclusively on V. alternifolia.
Sphex ichneumoneus (Sphecidae): Total individuals visiting M. punctata (), V. alternifolia (), and E. yuccifolium ().
Non-Hymenopteran Taxa:
Mordella sp. (Coleoptera: Mordellidae): Total individuals; ( visits) recorded on G. pulchella.
Hover flies (Diptera: Syrphidae): of all syrphid visits targeted fall-blooming B. laevis. Dominant species included Eristalis dimidiata ( visits), Eristalis transversa ( visits), and Helophilus fasciatus ( visits)—all restricted to B. laevis.
Hylephila phyleus (Lepidoptera: Hesperiidae, Fiery skipper): Total individuals; ( visits) recorded on E. purpurea.