The increasing frequency and severity of disturbances to coastal dune ecosystems necessitates the development and implementation of restoration strategies that rapidly accelerate reestablishment of vegetation, enhance dune accretion, and ultimately preserve dune ecosystem services. To assess how to rapidly re-establish vegetation to counter ecosystem losses, we conducted a manipulative field experiment on a created dune in Northeast Florida, USA to determine what combinations of planting density, outplant species composition, and nutrient addition maximize dune revegetation rate. After three months, we found that increased planting densities combined with nutrient addition elevated aboveground plant biomass by 868-2,961%, while growth in sparsely planted and unfertilized treatments was negligible or negative. These thickly revegetated, high density + nutrient addition plots accreted 1-5 cm more sediment, demonstrating that this planting method can rapidly kickstart dune-building processes. Additionally, densely planted, fertilized plots containing bitter panicum (Panicum amarum) produced 1.7 times more biomass and accreted 1.5 times more sediment than plots containing sea oats (Uniola paniculata), suggesting that this species and planting scheme may be most effective for rapid dune building. These findings reveal that coupling nutrient addition with dense planting can trigger self-sustaining, reinforcing plant growth and dune building feedbacks within months, likely warranting the cost of additional transplants by enhancing the long-term success of dune restoration investments.

Our field experiment was conducted on recently restored dunes located within the Guana Tolomato Matanzas National Estuarine Research Reserve (GTM-NERR) along the Florida A1A corridor (30° 1’ 20.8164” N, 81° 19’ 21.6732” W). The dune was constructed from May-June 2022 in a project led by St. Johns County using dredge sand sourced from nearshore deposits near the site (Figure 1). In late June 2022, we established 96 2.25 m² plots along an elevation isocline on the crest of initially unvegetated restored foredunes. Plots were spaced 2 m apart to ensure independence and marked at the four corners with bamboo stakes inserted ~0.5 m deep into the sand.

We used a 2 × 2 × 3 factorial design with two levels of nutrient addition (ambient or fertilized), two levels of plant density (36 vs. 144 propagules per 2.25 m²; hereafter referred to as low vs. high), and three levels of plant diversity (monotypic sea oats, monotypic bitter panicum, 50/50 mixed bitter panicum and sea oats) (12 total treatments, n = 8 replicates per treatment). The lower planting density treatment was consistent with standard planting densities used in conventional dune restoration designs, while we halved the plant spacing in the high-density treatment, resulting in densities four times the conventional amount (Rogers & Nash 2003). Sea oats and bitter panicum transplants were purchased from Green Seasons Nursery (Parish, Florida). Transplants were grown in 96 cell count planting trays containing potting soil inoculated with vesicular-arbuscular mycorrhizal (VAM) fungi (Glomus sp.). All transplants were initially of approximately the same height (45 cm) and had a 6.35 cm long root mass. From June 22-July 1^st, transplants were added to treatments. These planting dates fell within the optimal seasonal planting range for dune grasses in this region (April-October) (Williams 2007). Transplants were planted in 15 cm deep hand-dug holes, according to supplier guidelines, and nutrients were added to appropriate treatments in the form of slow-release fertilizer pellets during planting (Osmocote® 15-9-12, 5 g per propagule), consistent with Southeastern US dune restoration manual guidelines (Rogers & Nash 2003; Williams 2007, Miller et al. 2018; FDEP 2021). All transplants were planted in a regularly spaced grid. In treatments with mixed assemblages of bitter panicum and sea oats, the two species were planted in an alternating pattern. Plots were watered for five minutes each immediately following planting (~225 L per plot) and received water from ambient rainfall thereafter. Immediately following the establishment of our experimental plots, the area surrounding all plots was planted with sea oats in a ~40 × 40 cm grid pattern by a contractor overseen by St. Johns County staff. These plants were planted using power augers, a water-retention gel, and slow-release fertilizer pellets.

Aboveground biomass

Plots were monitored immediately following planting and each month thereafter for the 15-month duration of the experiment. During monitoring, we enumerated and measured the heights of all sea oat leaves and bitter panicum stems within a 0.16 m² quadrat placed at the center of each plot. We also determined the percent cover of each plant species. Plot aboveground biomass was estimated from plant height measurements using a regression of stem/leaf height vs. stem/leaf biomass determined from naturally established plants taken from around the experimental area.

To evaluate whether bitter panicum and/or sea oats gained more aboveground biomass when in monoculture versus mixed assemblages, we compared each species’ aboveground growth rates at the end of the first growing season. We calculated the change in aboveground biomass as the difference between the estimated final and initial aboveground biomass in each plot.

Belowground biomass

Belowground biomass was evaluated near the end of the first growing season on September 25th, 2022, and again at the end of the second growing season on September 26^th, 2023, from four 3 × 15 cm cores taken from the center of each plot. For each sample, macroscopic roots and rhizomes were separated, dried to a constant weight at 70°C, and weighed to the nearest 0.01 g.

Abiotic condition metrics

To explore whether there were self-reinforcing feedbacks between aboveground plant growth and abiotic stress, we quantified abiotic conditions in all treatments. In August 2022, one month following deployment, we quantified water loss from identical small kitchen sponges placed in each plot as a proxy for plot desiccation stress at the sediment surface (Silliman et al. 2011). Four plastic toothpicks were inserted into each sponge to create “legs” that elevated sponges 1 cm above the dune surface so they would not collect any sediment. We conducted this experiment on a clear day with minimal cloud cover, with temperatures ranging from 30.7-31.2°C, and wind speeds between 9.01-10.9 km/h. Immediately prior to deployment, the dry weight of each sponge was determined in the field using a battery-powered scale. Sponges were then wetted with fresh water and weighed again. A wet sponge was deployed to the center of each plot for 2.5 h. After this time, each sponge was collected and immediately weighed again. Percent water loss was determined by: [(initial wet weight−final wet weight)/(initial wet weight−dry weight)]*100.

Following our desiccation stress experiment, we used an infrared thermometer to determine the difference between sediment surface temperature at the center of each plot and the temperature of adjacent unvegetated substrate. Additionally, we quantified the difference in photosynthetic photon flux (µmol m^-2 s^-1) at the sediment surface within plot interiors and on adjacent bare sediment using a quantum meter (Spectrum Technologies model LQS-QM). We calculated the amount of light attenuated by vegetation in each plot as the percent difference in photosynthetic photon flux between plot interiors and adjacent bare substrate. Finally, we evaluated average sediment moisture content (%) at a 10 cm depth (n = 3 readings per plot) using a ΔT SM150 soil moisture kit.

Sediment surface elevation

Near the end of the first growing season, Hurricane Ian made landfall in southwestern Florida on September 28^th, 2022, and subsequently passed to the south of our study site as a tropical storm. We used this event as an opportunity to assess potential differences in the functional capacity of vegetation in experimental plots to accrete sediment and thus initiate the vertical dune-building processes.

We took elevation measurements of the dune plots using Unmanned Aerial System (UAS) Structure from Motion (SfM) almost one month after the hurricane on October 21^st, 2022, using a Parrot Anafi Ai. This dataset exhibited a relative accuracy of less than 2 cm. All experimental plots were georeferenced using Real-Time Kinematic Global Navigation Satellite System (RTK GNSS) and adjusted to the NGS CORS network, ensuring alignment with the established geodetic reference system, with baselines to the base station kept below 2 km. UAS SfM data were processed using Agisoft Metashape and were also aggressively filtered based on confidence values. Subsequently, datasets were further processed using ArcGIS Pro to generate Digital Elevation Models (DEMs).

We compared sediment surface elevation in treatment plots to unmarked locations between treatment plots that had been restored by contractors using conventional methods (hereafter referred to as control plots). The dimensions of these plots were roughly equivalent to those of treatment plots and were truncated by 0.5 m from either the northern or southern edge. Treatment and control plot areas were subdivided using 3 × 3 grids and plot statistics (maximum, minimum, and mean elevations) extracted from each of the resulting nine sub-grids using the Geoprocessing tool Zonal Statistics. We then calculated the difference in mean plot elevation (calculated from the nine plot subdivisions) of each treatment plot and the control plot adject to its southern border.

Following our initial evaluation of plot elevation, the site was impacted by a nor’easter (November 5th 2022) and Hurricane Nicole (November 10th, 2022). We assessed plot relative sediment surface elevation again using identical methods after one year on September 26^th, 2023 to determine if trends observed at the end of the first growing season continued during the second.