Coastal wetlands display ecohydrological zonation such that vertical differences of plant zones are driven by varying groundwater levels over tidal cycles. It is unclear how variable levels of tidal drainage directly impact biotic and abiotic factors in coastal wetland ecosystems. To determine the impacts of drainage levels, simulated tides in mesocosms with varying degrees of drainage were created with Spartina alterniflora, the salt marsh coastal ecosystem dominant species on the United States Atlantic Coast, and Salicornia pacifica, the Pacific Coast dominant. We measured biomass production and photosynthesis as indicators of plant health, and we also measured soil and porewater characteristics to help interpret patterns of productivity. These measures included above and belowground biomass, porewater pH, salinity, ammonium concentration, sulfide concentration, soil redox potential, net ecosystem exchange, photosynthesis rate, respiration rate, and methane flux. We found the greatest plant production in soils with intermediate drainage levels, with production values that were 13.7% higher for S. alterniflora and 57.7% higher for S. pacifica in the intermediate flooding levels than found in more inundated and more drained conditions. Understanding how drainage impacts plant species is important for predicting wetland resilience to sea level rise, as increasing water levels alter ecohydrological zonation.

This experiment was conducted using two ecosystem-dominant salt marsh species Spartina alterniflora and Salicornia pacifica.  Mesocosms were used to quantify the effects of the varying drainage treatments on the coastal wetland plant growth and soil conditions. The mesocosm soil was prepared by combining 25% marsh soil that was collected from the Tuckerton Peninsula, New Jersey (39.5388°, -74.3254°), 25% multipurpose sand (Lowes, Philadelphia, PA, USA), and 50% mushroom compost (Primex Garden Center, Glenside, PA, USA). Soils were screened at 2-mm to remove shell hash and coarse organic matter, homogenized with a cement mixer, and added to plant pots. The S. alterniflora plants were contained in pots that were made from cylindrical PVC pipes that were 20-cm long x 10-cm in diameter and were screened at the bottom to allow for water exchange. The S. pacifica plants were contained in 20-cm long square (7.6 x 7.6-cm) treepots, also screened at the bottom. 

We conducted a 2 x 4 factorial experiment where two types of plants were exposed to four different drainage treatments, with a total of 64 experimental units. Pots were placed in 150 L tidal mesocosms constructed from stock tanks and painted white. Simulated tides were created using tubing to pump water between the experimental tank and a reservoir, using timers to create a once-daily three-hour high tide and a 21-hour low tide. Inside four separate tanks, there were four drainage treatments: where the low tide water level was never drained (0-cm), and where the low tide water levels were drained to 5-cm, 10-cm, and 20-cm below the surface of the plant pot, The plants were propagated outdoors receiving natural precipitation. Plants were propagated in freshwater for one week, transitioned to 10 ppt salinity for one week, and finally changed over to 20 ppt salinity for the remainder of the experiment. Saline conditions were created utilizing Instant Ocean and adjusted weekly. Plants were propagated for 12 weeks from 13 May 2021 to 3 August 2021.

To measure the productivity of the wetland species Spartina alterniflora and Salicornia pacifica, the aboveground and belowground biomass of each treatment was measured. Aboveground and belowground biomass was harvested at the end of week 12. Stalks growing from the soil were cut, plants were washed over a 2-mm sieve, dried at 60°C, and weighed. Belowground biomass was separated from the soil by hand, rinsed over a sieve, dried at 60°C, and then weighed. Biomass values were scaled to a square meter through multiplication by a constant.

Measurements of the effects of soil drainage on abiotic parameters included porewater salinity, pH, ammonium, and hydrogen sulfide. Porewater was collected with a push pointsampler on 7 August 2021 (PPX36, MHE, East Tawas, MI, USA) at a depth of 20-cm. Samples for ammonium, pH, and salinity were frozen for later analysis. Subsamples (2 mL) were fixed with 0.22% zinc acetate solution, and frozen, for later analysis of hydrogen sulfide. Samples were analyzed for salinity and pH using a calibrated portable pH/Conductivity meter (Star A325 pH/Conductivity Portable Multiparameter Meter, Thermo Scientific Orion). UV-vis spectrophotometry was used for measurements of porewater ammonium using the indophenol blue method. Hydrogen sulfide was measured spectrophotometrically on the zinc acetate-fixed samples following the methylene blue method. All samples below the detection limit (~0.25 µM) were assigned a sulfide concentration of zero for the purposes of data analysis. Redox potential measurements were taken at depths of 5-cm and 15-cm for the Spartina alterniflora soils, and depths of 3-cm and 9-cm for Spartina pacifica soils to reflect the size of the pots using a bench-top redox meter (Bench-Top pH/mV Meter, Sper Scientific).

Carbon dioxide and methane fluxes were measured to calculate the photosynthesis, respiration, and net ecosystem exchange (NEE) rates of the plants. The fluxes were measured in the field on 7 August 2021 from 10 a.m. to 4 p.m. (Eastern Standard Time), using an LGR ultra-portable greenhouse gas analyzer via cavity ring-down spectroscopy (ABB-Los Gatos Research, San Jose, CA, USA). Positive fluxes were defined as those in which gas concentrations increased over time within the chamber, and negative fluxes as those in which concentrations decreased (Nakanoa et al., 2004). Photosynthesis rates were obtained by measuring the carbon dioxide concentration as a function of time in the light and dark conditions, and then subtracting the dark measurement from the light measurement. Respiration rates were obtained by measuring the carbon dioxide concentration as a function of time in the dark conditions. 

All statistical analyses were performed in R 4.2.3 (R Core Team, 2023). Bar graphs and scatter plots were made using the package ggplot2. The responses of the study plants Spartina and Salicornia to drainage were analyzed using permutational multivariate analysis of variance (perMANOVAs; Anderson, 2001), utilizing the R package vegan. The dependent variables included above and belowground biomass, porewater pH, salinity, ammonium concentration, sulfide concentration, soil redox potential, NEE, photosynthesis rate, respiration rate, and methane flux. We partitioned the variability and obtained F-statistics on matrices of Euclidean distances calculated from the original raw data. P-values were calculated using 10,000 random permutations of the appropriate exchangeable units. Where values of certain variables were missing, the means of a treatment group factor were added to the matrix to ensure each variable had the same number of values. Redox was rank-transformed prior to analysis to remove negative values. Because the interaction term was significant, we conducted contrast tests using the same function (adonis2), with a Bonferroni correction applied for the number of contrasts (0.05/16 = 0.0031).

Because the results of the contrasts did not reveal strong groupings, we conducted follow-up two-way ANOVAs to identify the factors contributing to significant differences between the species and drainage levels. ANOVAs were performed on rank-transformed data to account for the non-normal error structure. ANOVAs were run in base R using the aov command. Follow-up least-square means comparisons tests were completed on main effects and interactions, where appropriate, using package lsmeans using the Tukey or Sidak methods for p-value adjustments as appropriate.

A principal component analysis (PCA) was used for dimension reduction of the multivariate measurements to examine variability between treatments and plant species, as well as to determine species and treatment grouping patterns with respect to the dependent variables. The eigenvalues provide a measure of the amount of variance explained by each principal component. The data matrix was constructed using the 11 dependent variables which included above and belowground biomass, porewater pH, salinity, ammonium concentration, sulfide concentration, soil redox potential, NEE, photosynthesis rate, respiration rate, and methane flux.  The packages used for the PCA include cor, ggcorrplot, factoextra. To perform the PCA, the data was first normalized, and a correlation matrix was computed and visualized. Visualization of the PCA was constructed as a biplot of the first vs. second principal components. A Kaiser-Meyer-Olkin (KMO) test was used to determine how suited the data is for factor analysis, utilizing the EFAtools package. A Bartlett's test of sphericity was used to test for significant correlations among at least some of the variables, utilizing the psych package.

As many of the measured variables varied in response to the drainage treatments in a multi-dimensional way, we constructed loess curves (span=0.40) to describe the relationship between treatment and the response variables of above and belowground biomass, porewater salinity, porewater pH, porewater ammonium and sulfide concentration, NEE, and redox potential. A loess curve is a nonparametric method for smoothing a series of data and was used because of the non-linear structure of the relationship between drainage and response variables.