Methods

This study occurred at six different sites along the Mississippi and Alabama coastlines (Figure 1). Sites were sampled in the summer and winter of 2020. Each of the six sites had a hardened (HS), living (LS), and natural shoreline (NS) segment. This means that in total there were 18 different shoreline segments to compare to each other for a wide suite of ecologically relevant metrics:

• Fetch, energy, erosion rates and slope

• Water turbidity, salinity, temperature, dissolved oxygen

• Sediment grain size, bulk density, organic content

• Vegetation diversity and coverage

Figure 1. Map of six study site sites in Mississippi and Alabama – each site has three shoreline types (natural, living, and a hardened shoreline).

Figure 2. The natural shoreline (blue arrow), living shoreline (yellow arrow) and hardened shoreline (red arrow) for the six study sites. The high wave energy sites are Hancock County Marsh (HC), Swift Tract (ST) and Alonzo Landing (AL). The low wave energy sites are Camp Wilkes (CW), Ocean Springs Inner Harbor (OS) and Grand Bay NERR (GB).

At each of the six sites and three shoreline types, we characterized: (1) hydrographic features, (2) geomorphic features, (3) vegetation abundance, and (4) erosion rates and slope profiles.

The hydrographic factors we measured include wave pressure gauges to collect average wave power recorded at 1Hz Frequency (Temple et al. 2019), and pre-calibrated YSI 6600 series sondes with turbidity, temperature, conductivity (salinity), and dissolved oxygen recorded at 15 min intervals during 5- to 10-day unattended logger deployments.

The geomorphic factors we measured included landscape derived attributes including relative exposure using the methods from LaPeyre et al. (2014), and sediment bulk density, organic content, and grain size distribution derived from shallow (30cm) sediment cores.

Vegetation abundance was measured in ten replicate 1 m2 quadrats spaced equidistant along a 50 m transect running parallel to shore and located three to five meters upslope from the vegetated marsh plant edge of the marsh. The percent coverage for the full quadrat and each of the plant species within was estimated by visualization. Unknown plants were collected and brought back to the lab where they were then identified using taxonomic guides for the northern Gulf of Mexico.

Long term (multi-annual) erosion rates were measured for all three shoreline types at each of the six sites. Google Earth Pro’s timeline feature was used to trace the shorelines in multiple years (2019, 2011, 2005, and 1992). Shoreline retreat over time was calculated following the USGS Digital Shoreline Analysis System – DSAS approach. When possible, images in the summer month were chosen, as almost all years had a clear summer image to use and vegetation was more readily apparent than during the winter months.

Slope was measured using two elevation survey transects, where height above the water was recorded at every 1 m interval for each of the 18 shorelines. These were used to create elevation profiles up to 10 m inland and 10 m offshore (when possible) from the water line. In the field, the ends of the transect were marked by PVC posts with a 20m transect tape extending between the two end posts. Using a ruled H-frame and level, the elevation relative to the reference (set as 0m) midpoint was recorded every 1m along the transect line. The H-frame used consisted of two vertical poles set 1m apart with a level attached to the horizontal pole connecting them. One leg of the H-frame was fixed while the other leg was marked at 1cm intervals with the ability to move up and down within the frame. The fixed leg was first placed on top of the sediment, at the reference mid-point. The ruled leg was then set on the ± 1m point of the transect. Next, the connecter pole was adjusted until the level, attached to the horizontal pole, was level. The difference between the two legs position’s, indicated by the ruler on the sliding leg, was then recorded. The vegetation and sediment found at the spot were also recorded.

Turbidity data was analyzed by two-way ANOVAs using the factors: site & season and average wave power group & season. Data collected from the wave gauges were analyzed in MATLAB to calculate the parameter average wave power (kW/m). The average wave power was compared using the Kruskal Wallis H Test.

Sediment data analysis included the calculations for bulk density, organic matter content, and sediment grain size composition. For each of these parameters a two-way ANOVA was performed using the factors: site & shoreline and average wave power group (high and low) & shoreline. Factors with a significant response (alpha ≤ 0.05) were followed by a Tukey’s post hoc test to find groupings of statistically similar means.

Vegetation data analysis included two-way ANOVAs for species richness, average percent cover, and alpha diversity calculated using the Shannon H and Simpson D indices, and the Bray Curtis dissimilarity index for beta diversity. The three indices were calculated using the R package Vegan.

Multivariate non-parametric ordination (nMDS) as well as a principal component analysis (PCA) were used for further data exploration. The data for both these tests were centered to mean zero and standardized to unit variance. Hydrographic data included were the average wave power and turbidity. Geomorphic data included were relative exposure, erosion rate, average slope, percent sand, and OM. Vegetation factors that were included in the nMDS and PCA were species richness, percent cover and percent cover of dominant species. Not all response variables measured were included due to high correlation among some pairs of metrics.

All data analysis was done in R using RStudio (ver. 1.9, Boston, MA).