Faunal engineering stimulates landscape-scale accretion in southeastern US salt marshes

Abstract

The fate of coastal ecosystems depends on their ability to keep pace with sea-level rise-yet projections of accretion widely ignore effects of engineering fauna. Here, we quantify effects of the mussel, Geukensia demissa, on southeastern US saltmarsh accretion. Multi-season and -tidal stage surveys, in combination with field experiments, reveal that deposition is 2.8-10.7-times greater on mussel aggregations than any other marsh location. Our Delft-3D-BIVALVES model further predicts that mussels drive substantial changes to both the magnitude (±<0.1 cm·yr^-1) and spatial patterning of accretion at marsh domain scales. We explore the validity of model predictions with a multi-year creekshed mussel manipulation of >200,000 mussels and find that this faunal engineer drives far greater changes to relative marsh accretion rates than predicted (±>0.4 cm·yr^-1). Thus, we highlight an urgent need for empirical, experimental, and modeling work to resolve the importance of faunal engineers in directly and indirectly modifying the persistence of coastal ecosystems globally.

Fig. 1. Conceptual figure outlining the spatial and temporal scale of all field components and associated hypotheses.

1a–c Landscape assays of sediment deposition (i.e., 9 cm filter papers; ) were distributed across 13 area types over four 24 h tidal deployments. We hypothesized that sediment deposition atop mussel aggregations would be as high as deposition on levee crests. (2a, b) Experiment 1 involved tracking the fate of fluorescently tagged previously-settled and newly ejected biodeposits from mussel aggregations (24 h deployments). We hypothesized that sediment would be rapidly redistributed across marsh platforms from these local hot spots of deposition. 3a, b Experiment 2 involved the deployment of seven treatments containing a range of mussel and cordgrass biomass. Treatments were deployed at the creekhead and on the marsh platform in sediment catchment devices, designed to capture all sediment deposited throughout the 1-month deployment. We hypothesized that mussel biomass would drive sediment deposition at this intermediate temporal and spatial scale. (4) Experiment three involved the removal of mussels from one tidal creekhead and the transplantation of these mussels to another proximate creekhead. We hypothesized that the removal of mussels inhibits accretion at the landscape scale, while addition increases it relative to an unmanipulated control. Locations of each experiment are highlighted in the panels at left. Numbers and colors correspond to the experiment of relevance.

Fig. 2. Creekhead density and mussel areal coverage at sites distributed across the South Atlantic Bight.

a Regional surveys quantified the percent area of tidal creekheads occupied by mussels (values on the left of Panel a with sites denoted in open circles; n = 6 sites) and the density of tidal creekheads within larger creekshed areas (# 1 km⁻²; values on the right of Panel A with sites denoted in black circles; n = 7 sites). b Sapelo Island sites were similarly assessed for creekhead mussel area (open circles; n = 6 sites) and creekhead density (black circles; n = 3 sites). c There were no statistically significant differences in creekhead density (light grey bars) or creekhead mussel area (dark gray bars) across northern sites (n_{Creekhead Density} = 4; n_{Creekhead Mussel Area} = 3), Sapelo Island sites (n_{Creekhead Density} = 3; n_{Creekhead Mussel Area} = 6), and southern sites (n_{Creekhead Density} = 3; n_{Creekhead Mussel Area} = 3; mean + SE). Finally, assuming a creekhead area of 2500 m², areal coverage of creekheads and of mussels range from 25–32% and 2.5–3.0%, respectively across the region (mean values reported to right of error bars).

Fig. 3. Filter paper results.

a Summer neap, b summer spring, c winter neap, and d winter spring tide results are presented across 13 marsh location types (n = 15 filters/location/tide). Mean sediment deposition is presented as gray bars (mean ± SE) on each marsh location, with letters denoting statistically significant differences among treatments (Season*Tide*Location: F _{12, 825} = 4.7; p < 0.0001). For ease of interpretation, Tukey HSD post hoc analyses are conducted separately for each season using a corrected p-value of p < 0.001. Results collected from atop mussel aggregations are presented as gray bars with a black border (at 0 m, 10 m, and 20 m from the tidal creekhead). Results from locations not associated with mussel aggregations (i.e., all other marsh location types) are presented as gray bars with no border.

Fig. 4. Results from experiments 1 & 2.

Box plots of sediment redistribution (in cm traveled per tide) is shown of previously settled biodeposits (a) and newly ejected material (b) in both the creekhead (light green) and marsh platform (dark green) zones (n = 6 mounds/zone). Box plots present minimum, 25th percentile, median, 75th percentile, and maximum as lines. Outliers are presented as solid circles. c The relationship between live mussel biomass (x axis) and sediment biodeposition (y axis) is presented in both the creekhead and on the marsh platform. Model summaries are presented inset with adjusted R² values. For reference to no-mussel, no-cordgrass control treatments (0 M, 0 C), we present the mean value of this treatment on the y axis. Month-long dry sediment deposition is presented in units of kg/m²/day.

Fig. 5. Delft3D-BIVALVES Model Results.

a Model domain and area types are delineated and color coated to depict locations of different vegetation and mussel patterning. b Mean local scale accretion (1 m²) of the five marsh area types in the baseline scenario (defined as vegetation present, mussel coverage 10%) are shown (mean ± SD; n > 150 per area type, depending on number of cells occupied by area type). For each marsh location type and scenario, we then calculate percent change in local deposition (m⁻²) compared to the baseline scenario in (c). Change in landscape scale deposition (d) and accretion (e) from baseline scenarios are presented for both mussel removal (0%) and mussel addition (20%) scenarios in the creekhead (top panels) as well as in the entire domain (bottom panels).

Fig. 6. Creekshed Mussel Manipulation.

Prior to deployment in 2017, (a) we first delineated the creekhead area (2500 m²) for the mussel removal (highlighted in red, dashed), mussel addition (blue, dashed), and control creeks (white, dashed). Aerial imagery from 2020 (b) shows the same experimental areas and visual changes to the condition of the marsh platform (e.g., noticeable grid of mussel aggregations and increased primary productivity on addition creekhead outlined in blue, and loss of those features on the removal creekhead outlined in red). The 2020 DEM (c) depicts marsh elevation across the entire creekshed (red colors indicate higher elevations, shifting down to orange, yellow, and green at lower elevations). We then compare differences in elevation between the control and each of the experimental treatment plots from two methods in 2017 and from the DEM creekshed area in 2020 (d). Results highlight that, relative to the control trajectory, the mussel removal creek lost elevation (−1.7 cm yr⁻¹) while the mussel addition creek gained elevation (+0.4 cm yr⁻¹).

Fig. 7. Fauna engineering effects on ecogeomorphology of vegetated coastal ecosystems.

a Conceptual model (adapted after Fagherazzi et al.) depicting the mechanisms through which fauna engineers alter ecogeomorphology, with direct and indirect mussel effects highlighted (blue and green boxes, inset). b Mussel engineering effects on marsh ecogeomorphology are illustrated and further described. Other fauna engineers likely alter vegetated coastal ecosystem accretion processes, such as through c bioturbating effects of lugworms (Arenicola marina) and d above and belowground grazing effects of the omnivorous marsh crab, Sesarma reticulatum.