The Impact of Inundation and Nitrogen on Common Saltmarsh Species Using Marsh Organ Experiments in Mississippi

Abstract

Sea level rise is an escalating threat to saltmarsh ecosystems as increased inundation can lead to decreased biomass, lowered productivity, and plant death. Another potential stressor is elevated nitrogen often brought into coastal regions via freshwater diversions. Nitrogen has a controversial impact on belowground biomass, potentially affecting saltmarsh stability. In this study, we examined the effects of inundation and nitrogen on common saltmarsh plants (Spartina alterniflora and Spartina patens) placed within two marsh organs (a collection of PVC pipes at different levels, the varied elevation levels expose the plants to different inundation amounts) located in the Pascagoula River, Mississippi, USA, with six rows and eight replicates in each row. We randomly fertilized four replicates in each row with 25 g/m² of NH4⁺-N every two-three weeks during the growing season in 2021 and 2022. We concurrently collected vegetative traits such as plant height and leaf count to better understand strategies saltmarshes utilize to maximize survival or growth. We harvested half of the vegetation in Year 1 and the remaining in Year 2 to evaluate the impact of inundation and nitrogen on above- and belowground biomass at different temporal scales. We developed Bayesian models that show inundation had a largely positive impact on S. alterniflora and a mostly negative impact S. patens, suggesting that S. alterniflora will adapt better to increasing inundation than S. patens. Additionally, fertilized plants from both species had higher aboveground biomass than non-fertilized plants for both years, with nitrogen addition only showing impact on belowground biomass in the long term. Our results highlight the importance of long-term study to facilitate more-informed restoration and conservation efforts in coastal wetlands while accounting for climate change and sea level rise.

1. Introduction

Coastal wetlands provide a myriad of resources and services for humans and wildlife [1,2,3]. In these ecosystems, plants are engineers—they shape landscapes through vertical accretion of organic and inorganic materials to form wetland platforms, prevent erosion through their root systems, affect community structure, and provide habitat for various species [1,4,5].

With these factors detailing how coastal wetland plants affect their ecosystems, any alterations or threats to plant growth may affect the system stability, particularly regarding how coastal wetlands tolerate sea level rise and the stress from additional inundation [6,7]. This highlights the importance of understanding how changes in inundation affect plant growth, and subsequently, how biological processes combat or adapt to these changes, alter zonation and competition, and modify their physical environment [4,8,9]. Previous studies show that a small amount of increased flooding may promote biomass growth and productivity but a continued increase in inundation results in decreased biomass and productivity in dominant salt marsh species [4,6,10,11,12]. Other studies show the above- and belowground biomass of Spartina alterniflora and Spartina patens was the highest when inundation was at its minimum, also noting that the negative productivity response to inundation was more pronounced in the S. patens [5]. A marsh organ experiment showed that optimum inundation levels existed for the above- and belowground biomass of Sagittaria lancifolia, and that the biomass decreased quickly beyond the optimum inundation levels [13].

The negative impact on plant productivity from excessive inundation further limits sediment trapping by aboveground biomass and reduces organic matter accumulation by belowground biomass, and therefore decreases vertical accretion, accelerates submergence, and becomes a self-accelerating cycle [5,10,11,12,14,15,16,17,18]. Along with the vulnerability coastal wetlands face due to sea level rise, freshwater diversions from upper rivers can also result in prolonged inundation and higher water levels, as well as impacting salinity dynamics and nutrient cycles of wetland systems [12,19,20,21].

In addition to inundation, nitrogen plays an important role in salt marsh ecosystems. These ecosystems serve as large nitrogen sinks [22,23,24], acting as a stimulant for microbial organic decomposition and positively affecting plant growth and morphological attributes [6,13,22,25]. Studies suggest that exposure to additional nitrogen can be beneficial to marsh systems as the increased nitrogen promotes plant productivity, especially S. alterniflora and S. patens, temporarily alleviates flooding stressors, and broadens its vertical range [6,26,27]. More so, a meta-analysis of nutrient-enriched saltmarsh plants found that after a certain sea-level-rise threshold is passed, the benefits from nutrient enrichment may be overwhelmed [27]. Additionally, in terms of the physical stability of the ecosystem, research suggests that nitrogen additions may lead to a decline in belowground plant productivity and a weaker root system, resulting in potential destabilization of marsh platforms especially in areas that suffer from limited sediment availability [28,29].

Despite its importance, it is yet unclear how salt marsh vegetation responds to the interactive factors of inundation and nitrogen availability on the Mississippi Gulf Coast. Depending on morphological characteristics measured for resource allocation, different species show different phenotypic responses and potential plasticity to inundation and nitrogen. This understanding will facilitate better predictions of coastal resilience at the landscape scale [13,30].

In this research, we aim to: (1) Study the temporal patterns of a variety of morphological traits of two common salt marsh species (S. alterniflora and S. patens) impacted by inundation and nitrogen. (2) Evaluate short- (4-month) and long-term (16-month) impact of inundation and nitrogen on the above- and belowground biomass of the two species. (3) Assess if there is a cumulative effect of nitrogen on above- and belowground biomass over time, showing larger effect over the long term vs. the short term. (4) Gauge alleviatory impact from nitrogen (increased productivity) in regard to the negative effect of inundation on both species, more so for Spartina patens (a high-marsh plant) than Spartina alterniflora (a low-marsh plant). We hypothesize that the fertilized individuals will outperform the non-fertilized individuals, and that the S. alterniflora (low-marsh) plants will adapt better to the marsh organ induction levels than the S. patens (high-marsh) plants.

2. Materials and Methods

We used an in situ mesocosm experiment called a marsh organ to understand mechanistically how different levels of percent of time inundated and nutrients (nitrogen) affected the vegetation growth of Spartina alterniflora and Spartina patens. Marsh organs are applied to evaluate inundation-productivity relationships while also allowing an insight of interactions between nutrient competition and other abiotic stressors [5,6,10,13,31]. Two marsh organs were built in the western Pascagoula River, a tidally influenced brackish river, for the facilitation of varying inundation. Every two to three weeks during the growing season in 2021 and 2022, specifically from the end of July when the marsh organ was set up to the middle of November in 2021 (end of growing season in Year 1) and from April (1 month into growing season in Year 2) to early November in 2022 (end of growing season in Year 2), we measured vegetative traits including plant height, leaf count, leaf length, leaf width, and stem count to evaluate their temporal change and potential tradeoff of different traits. We also randomly selected half of the replicates at each inundation level to add a nitrogen additive to simulate additional nutrients. We harvested the replicates at the end of the growing season for Year 1 and Year 2 to examine short- (4-month) and long-term (16-month) effects of inundation and nitrogen on the above- and belowground biomass.

2.1. Study Species

The Spartina genus is comprised of intertidal C4 grass saltmarsh plants, commonly found along the Atlantic, Pacific, and the Gulf Coast of Mexico. This genus of saltmarsh grows in the summer, reproduces in the fall, and dies back during the winter months, providing habitat for a variety of species through its biomass and sedimentation. While S. alterniflora and S. patens are in the same genus, and often the same region, they occupy different niches. S. alterniflora is a low-marsh species and is found throughout the marsh platform, growing tall along shorelines where it is frequently flooded, typically at every tide and only exposed during low tide [32]. S. patens is a high-marsh species that grows in the upper, more expansive salt marsh and less frequently flooded habitat, as this species has a limited ability to uptake oxygen in anoxic soils [32]. This Spartina zonation is determined by abiotic and biotic factors—S. patens prefers drier, less inundated, and more oxygenated soil in the high marsh and while S. alterniflora can persist in either low or high marsh, it is restricted to the low marsh due to competitive displacement from S. patens [32,33]. When reviewing studies investigating inundation-productivity relationships, some suggest that S. alterniflora may exhibit a quadratic or parabolic shape, indicating an optimal intermediate amount of inundation for maximizing productivity [1,4,10,13]. Alternately, another study showed that both S. alterniflora and S. patens reacted negatively as inundation increased in the Gulf of Mexico [5,13].

2.2. Study Site

The marsh organs were situated in the western channel of the lower Pascagoula River (Figure 1). The Pascagoula River is the largest undammed river in the continental United States by volume and contains roughly 35% of the coastal wetlands of the Mississippi Gulf Coast [1,13,34]. This western channel is minimally anthropogenically impacted compared to the eastern channel and flows southward into the Mississippi Sound, with the watershed receiving abundant rain annually, bringing a large source of freshwater into the Gulf [13,35,36]. Due to this influx of fresh water mixing with the saltwater from the Gulf of Mexico, the marsh organ field site has fluctuating brackish water, with salinity ranging from 0 to 15 ppt (mean of 4.09 ppt and standard deviation of 5.3 ppt), measured with a refractometer from April to November 2022. The study site at (30.393° N and 88.608° W) is located north of Highway 90 and approximately 1200 m north of a US Geological Survey (USGS) water gauge which was used to help design the marsh organs and determine percent of time inundated during the experiment (Figure 1, USGS station ID: 02480285 West Pascagoula River at Highway 90 at Gautier, Mississippi, pulled from the USGS Water Data Resources, https://waterdata.usgs.gov/nwis/inventory?agency_code=USGS&site_no=02480285, last visited January 2023). The saltmarsh in this area is largely comprised of Spartina, Schoenoplectus, and Sagittaria species [13].

2.3. Marsh Organ Design

The marsh organ was constructed of PVC pipes organized by six rows that differ in height and, therefore, differing inundation depth and duration, with eight replicates in each row (Figure 2). The inundation durations were designed to be 90%, 70%, 50%, 30%, 10%, and 0% from bottom to top, determined using the nearby USGS tidal gauge (station ID: 02480285) water level data (Figure 2) and the elevation of the different levels of the marsh organ measured by real-time kinematic (RTK) positioning GPS. The south marsh organ (MO1) contained 48 pipes of S. alterniflora (Figure 3a) and the north marsh organ (MO2) had 48 pipes of S. patens (Figure 3b).

The two marsh organs were constructed in the summer of 2021, placed approximately 10 m apart, facing southward, and situated perpendicular to the marsh–waterline edge. The PVC pipes had a 15 cm diameter and a standardized length of 61 cm. The pipes were screwed and secured together, pushed into the sediment a certain amount depending on row, and then attached to the wooden frame implemented for added stability. Pipes were then packed with local sediment to ensure plant growth at the top of the pipe and to prevent sinking. A nylon mesh was placed at the bottom of the higher pipes, not pushed into the sediment, which still allowed natural lateral water flow but prevented sediment or plant loss.

For the MO1, S. alterniflora was transplanted into each PVC pipe that originated from a mixture of east and west channels sites in the Pascagoula River (plants from each channel were haphazardly assigned to each pipe). However, for the MO2, S. patens was collected only from the eastern channel, as no large patches of this species were found in the western channel. The plants were also cleansed of their original site sediment. During transplanting, we aimed to plant five individuals of S. alterniflora and ten individuals of S. patens into each PVC pipe. These densities were based on previous literature [37], which suggests 3–4 stems per PVC pipe, plus the extra we added to account for potential die-off related to the stress of relocation. The pre-conditions of the vegetative morphology (leaf count, height, etc.) were recorded as they likely affected the vegetation growth. More local sediment was packed into the pipes after plant transplantation to account for the potential of gradual compaction and reduce the risk of the plants floating out of the pipes when inundated [13].

2.4. Trait Monitoring and Biomass Processing

Starting in July 2021, we haphazardly selected two individuals from each PVC pipe replicate to measure their heights (base of stem to the highest part of a plant), leaf counts, leaf lengths (the second leaf from the top), and petiole width (the second leaf from the top) (

Table 1. Summary of vegetative trait measurements on Spartina alterniflora (Sa.) and Spartina patens (Sp.).

Species	Traits Measured	Visual Representation
Sa. and Sp.	Height from base of stem to highest part of the plant
Sa. and Sp.	Leaf count
Sa. and Sp.	Length of the second leaf from the top
Sa. and Sp.	Width of the petiole of the second leaf from the top
Sa.	Width of the second leaf from the top
Sp.	Stem count

). Additionally, for the Spartina alterniflora individuals, we measured the width of the second leaf from the top of each individual and for the Spartina patens individuals, we measured total stem count in each pipe to observe plant density (Table 1). These measurements were performed every 2 to 3 weeks during the growing season from July to November 2021 for year one and April to November 2022 for year two, to capture short-term morphological changes over a long-term period. Altogether, we conducted the measurements six times in 2021 (09/03, 09/24, 10/08, 10/23, 11/05, and 11/18) and twelve times in 2022 (04/08, 04/22, 05/09, 05/27, 06/17, 07/08, 07/21, 08/05, 08/29, 09/14, 10/14, and 11/02). At each visit starting on 09/10/21, we applied 25 g per m² of NH4⁺-N with a syringe into the soil (15 cm below surface) of half of the replicates, randomly selected to simulate the scenario of added reactive nitrogen in the environment [6]. Consequently, at the end of the growing season of Year 1 in November 2021 (short-term impact), we harvested half of replicates and harvested the other half at the end of the growing season of Year 2 in November 2022 (long-term impact), as above- and belowground biomass samples can elucidate the integrated effect of inundation and nitrogen. In each harvest, we randomly selected half of the replicates with nitrogen additions and half without in each row. Aboveground biomass was bagged and ready for immediate processing and belowground biomass was allocated into bags based on depth: 0–5, 6–10, 11–15, 16–20, 21–25, and 26–30 cm, which were then stored in laboratory refrigerators 4 °C for subsequent processing.

Aboveground biomass samples were processed within two weeks of harvest, with live and dead parts of the biomass separated into pre-weighed, oven safe aluminum trays that were then oven-dried at 75 °C until a constant weight was reached, around 3–5 days [1]. The demarcation between live and dead aboveground biomass was based on color. Live biomass had green stems and leaves while dead biomass ranged from yellow to brown [38]. Pre-dried and post-dried weights were collected from these samples.

To process belowground biomass, we first washed sediment and mud off the belowground biomass using a 1 mm mesh sieve and removed extraneous objects such as sticks and snails. Once washed, live and dead biomass was separated based on the buoyancy by submerging it in water first, combined with color and high turgidity (Figure 4). Floating biomass with light colors and turgidity was classified as live biomass, while dark matter that sunk to the bottom of the container, and felt and looked flaccid, was classified as dead biomass [38]. The live and dead biomass were separated into the pre-weighed aluminum trays, then weighed again, and dried in an oven for several 3–5 days to remove water content until a constant weight was reached. After removal from the oven, the sample trays were weighed to collect the dry weight. See Figure A1, Figure A2, Figure A3 and Figure A4 in Appendix A and Tables S1–S4 in Supplementary Materials, respectively, for displayed measured data of the biomass and morphological characteristics for both years.

2.5. Statistical Analyses

Using the collected morphological attributes and biomass data, we applied multi-level Bayesian models to the morphological attributes and single-level Bayesian models to the biomass data to evaluate the impact of inundation and nutrients on vegetation traits over time and vegetation productivity (Figure 5). Bayesian statistics is a form of statistical inference involving the Bayes theorem. Hierarchical Bayesian models decompose high-complexity problems into simpler conditional distributions in a fully consistent framework [39,40,41]. Using Hierarchical Bayesian models allows data assimilation while accounting for various uncertainties and provides inference based on posterior distributions [39,40]. We developed models for each trait measured during sampling visits, and above- and belowground biomass. We accounted for senescence (days since installation of marsh organs in Year 1), and seasonality (temperature since onset of growing season in Year 2), pre-condition, and channel when evaluating the impact of inundation and nutrient addition (Figure 5; Equation (1)). As biomass was measured at the end of growing seasons, their models differed from morphological characteristics in that they do not have time as a covariate.

We created and ran these models in R applying Markov Chain Monte Carlo Simulation (MCMC) with the “rjags”, “MCMCvis”, and “coda” packages (https://cran.r-project.org/web/packages/available_packages_by_name.html, accessed 1 March 2023). We compared the models using Deviance Information Criterion (DIC) and predictive posterior loss (PPL)—selecting the best model based on the lowest DIC or PPL [40,42]. DIC was used as the main model selection criterion, but we also considered PPL during the parameter selection in addition to DIC when the models differed in their hierarchies (see Tables S5–S13 in the Supplementary Materials for more information on the model comparisons). Once selected, we summarized medians and quantiles of 95% and 50% credible intervals (CIs) for the parameters of the covariates based on the posterior outputs [40,43]. The 95% CIs represent the range from the 2.5% to 97.5% quantiles, while the 50% CIs indicate the range from the first to third quartile. These 95% or 50% CIs indicate a 95% or 50% probability that the covariate’s coefficient lies within the intervals. If the CI does not overlap zero, there exists evidence for the covariate to have a strong or moderate positive or negative effect on the dependent variable [40]. For full Bayesian posteriors, see Figures S1–S14 in the Supplementary Materials.

The conceptual model illustrates the hierarchical structure with complexity decomposed into stages of data, process and parameters (vertical direction) and the association of different spatial scales (horizontal direction). The symbols TP denotes temperature (for Year 2 models, Year 1 models have days since transplant as the covariate at the time scale), IN is the proportion of time inundated, N denotes nitrogen treatments, PC stands for the pre-condition of the marsh plants transplanted, and CH refers to channel. The symbols of αs, βs, λs, and τs are the parameters in the model denoting the coefficients for the covariates, with τ_1, τ_2, and τ_3 denoting precision at the time, row, and PVC pipe scales. (Adapted from [39,40])

$$\begin{gathered} [{\beta }_0, {\beta }_1, {\alpha }_1, {\alpha }_2, {\lambda }_1, {\lambda }_2, {\lambda }_3, {𝜏}_1, {𝜏}_2, {𝜏}_3| {\mathrm{y}}_{\mathrm{j}\mathrm{i}\mathrm{t}}] \\ \propto \prod _{t=1}^{12}\prod _{i=1}^6\prod _{j=1}^8 \\ dnorm({\mathit{y}}_{\mathit{j}\mathit{i}\mathit{t}} | {\mathit{\lambda }}_{0it, }{\mathit{\lambda }}_1, {\mathit{\lambda }}_2, {\mathit{\lambda }}_3, {\mathit{𝜏}}_3) \\ dnorm({\mathit{\lambda }}_{0it }| {\mathit{\alpha }}_{0t}, {\mathit{\alpha }}_1, {\mathit{\alpha }}_2, {\mathit{𝜏}}_2) \\ dnorm({\mathit{\alpha }}_{0t}| {\mathit{\beta }}_0, {\mathit{\beta }}_1, {\mathit{𝜏}}_1) \\ dnorm({\mathit{\beta }}_0| 0, 0.000000001) dnorm({\mathit{\beta }}_1|0, 0.000000001) \\ dnorm({\mathit{\alpha }}_1| 0, 0.000000001) dnorm({\mathit{\alpha }}_2|0, 0.000000001) \\ dnorm({\mathit{\lambda }}_1| 0, 0.000000001) dnorm({\mathit{\lambda }}_2|0, 0.000000001) \\ dnorm({\mathit{\lambda }}_3| 0, 0.000000001) dgamma({\mathit{𝜏}}_1|0.1, 0.1) \\ dgamma({\mathit{𝜏}}_2| 0.01, 0.01) dgamma({\mathit{𝜏}}_3|0.001, 0.001) \end{gathered}$$

(1)

Equation (1) Bayesian posterior for the multi-level models to predict vegetation morphology in Year 2 (2022). See Figure 5 for denotations of the symbols. Models for biomass are similar, simpler without the time level.

3. Results

3.1. Inundation Duration

In Year 1 (2021), inundation durations at different rows of the Spartina alterniflora marsh organ were similar to what we designed (Figure 6a). However, due to extreme droughts in 2022, the inundation durations decreased in Year 2, with the maximum duration reaching only 60% to 70% of time (Figure 6a). Inundation durations of the Spartina patens marsh organ were lower than those of the Spartina alterniflora marsh organ as they were situated at a higher platform (Figure 6b). Just like the Spartina alterniflora marsh organ, the inundation durations were lower in Year 2 than in Year 1.

3.2. Response of Aboveground Biomass

In Year 1, the aboveground biomass of Spartina alterniflora (Sa. thereafter) and Spartina patens (Sp. thereafter) exhibited moderate and strong parabolic relationships with inundation time, respectively (“moderate” defined as the 50% CIs do not overlap 0, while “strong” defined as 95% CIs do not overlap 0) (

Table 2. Summary of signs of 50% and 95% credible intervals (CIs) based on the posteriors of Bayesian models for the biomass of Spartina alterniflora and Spartina patens. Positive impact is denoted by (+) and negative impact is marked by (-). Also see 50% and 95% CIs in Figures S1–S4 in the Supplementary Materials (Sa. denotes Spartina alterniflora, marked in red, Sp. denotes Spartina patens, marked in black) (If no signs were provided for a particular CI, it means the CI intercepted 0).

Metrics	Species	Inundation Time		Inundation Time Squared		Nutrient		Pre-Condition		Channel (0—East, 1—West)
		95% CI	50% CI	95% CI	50% CI	95% CI	50% CI	95% CI	50% CI	95% CI	50% CI
Aboveground Biomass YR 1	Sa.		-		+	+	+			-	-
	Sp.	-	-	+	+	+	+	-	-
Aboveground Biomass YR 2	Sa.	+	+		-	+	+				-
	Sp.		-			+	+
Belowground Biomass YR 1	Sa.		+		-				+	-	-
	Sp.		-
Belowground Biomass YR 2	Sa.	+	+		-	+	+				-
	Sp.	-	-			+	+

). While uncertainty was large, we found the biomass reached minimum productivity at 41.3% and 51.6% (medians) inundation time for Sa. and Sp., respectively. By Year 2, inundation time showed a moderate negative effect on the aboveground biomass of Sp., while the aboveground biomass of Sa. reached maximum productivity at 50.0% of inundation (median). Nutrient addition had a strong positive impact on aboveground biomass for both species in the short and long term (Table 2). Pre-condition (stem count) only had a strong negative impact on Sp. in the short term, with no impact observed by Year 2 for both species. For Spartina alterniflora, channel had a strong negative impact in Year 1, and moderate negative impact in Year 2, indicating that the east channel plants strongly and then moderately outperformed the west channel plants as time progressed.

3.3. Response of Belowground Biomass

The belowground biomass of Sa. and Sp. responded to inundation very differently in the short and long term (Table 2). However, the relationships were consistent for individual species in both years and demonstrated increased impact by Year 2. In both years, the belowground biomass of Sa. indicated a parabolic relationship with the maximum reaching at 42.6% (median) and 44.3% (median) of inundation time, respectively. The belowground biomass of Sa. responded negatively to inundation. While nutrient addition had little to no impact on belowground biomass in Year 1, it showed strong positive impact on both species in Year 2. Pre-condition had only a moderate positive impact on Sa. in the short term, with little to no impact observed by Year 2 for either species (Table 2). Like aboveground biomass, the vegetation from the east channel outperformed the vegetation from the west channel, with the stronger difference in Year 1.

3.4. The Impact of Inundation on Morphological Traits

In Year 1, inundation time had mixed impact on both species. Inundation had a moderate negative linear impact on Sp. leaf count and height, and a strong negative linear impact on Sa. leaf count; inversely, inundation had a moderate positive linear impact on Sa. and Sp. stem width, Sa. leaf width, and Sp. leaf length, as well as a strong positive impact on Sp. stem count (

Table 3. Summary of signs of 50% and 95% credible intervals (CIs) of linear and quadratic inundation impact based on the posteriors of multi-level Bayesian models for metrics of Spartina alterniflora (marked in red) and Spartina patens (marked in black). Positive impact is denoted by (+) and negative impact is marked by (-). Also see 50% and 95% CIs in Figures S5–S14 in the Supplementary Materials. (If no signs were provided for a particular CI, it means the CI intercepted 0).

		Year 1 (4 months)				Year 2 (16 months)
Metrics	Species	Inundation Time		Inundation Time Squared		Inundation Time		Inundation Time Squared
		95% CI	50% CI	95% CI	50% CI	95% CI	50% CI	95% CI	50% CI
Plant Height	Sa.	+	+		-	+	+	-	-
	Sp.		-			-	-
Leaf Count	Sa.	-	-				+		-
	Sp.		-			-	-
Leaf Length	Sa.		-		+		+		-
	Sp.				+	-	-	+	+
Stem Width	Sa.		+			+	+
	Sp.		+			-	-
Leaf Width	Sa.		+			+	+		-
Stem Count	Sp.	+	+			-	-		+

). Sa. plant height had a moderate parabolic relation with inundation and it reached the maximum at 56% of inundation time (median). Meanwhile, Sa. leaf length had moderate parabolic relations with inundation, and they reached the minimum at 36% of inundation time (median) (Table 3).

By Year 2, inundation time had more consistent impact across individual species. For Sp., inundation exhibited a strong negative impact on plant height, leaf count, and stem width. Leaf length and stem count, on the other hand, showed parabolic relations with inundation and they reached the minimum at the inundation time of 52% and 84% (medians), respectively. For Sa., most of the morphological traits, including plant height, leaf count, leaf length and leaf width, showed strong or moderate parabolic relations with inundation with the maximums reached at 89%, 35%, 39%, and 56% (medians) of inundation time, respectively (Table 3). Inundation had a strong positive linear impact on Sa. stem width (Table 3).

The different response of different vegetation traits to inundation time can help reveal strategies of vegetation in maximizing survival or growth under elevated inundation. In the short time, focusing on linear response, Sa. increased stem width and leaf width at the expense of leaf count, while Sp. increased leaf length, stem width and stem count at the expense of plant height and leaf count. In the long term, Sa. also showed increased stem width with elevated inundation while other traits showed parabolic relations to inundation.

3.5. The Impact of Nutrients on Morphological Traits

In Year 1, nutrient addition had a strong positive impact on Sp. leaf count, Sa. leaf length, and Sa. leaf width, and Sa. stem width, and a moderate positive impact on Sp. leaf length and Sp. stem width (

Table 4. Summary of signs of 50% and 95% credible intervals (CIs) of nitrogen impact based on the posteriors of multi-level Bayesian models for metrics of Spartina alterniflora (marked in red) and Spartina patens (marked in black). Positive impact is denoted by (+), negative impact is marked by (-), and (*) denotes varying impact. Also see 50% and 95% CIs in Figures S5–S14 in the Supplementary Materials. (If no signs were provided for a particular CI, it means the CI intercepted 0).

		Year 1 (4 months)		Year 2 (16 months)
Metrics	Species	Nutrient		Nutrient
		95% CI	50% CI	95% CI	50% CI
Plant Height	Sa.			+	+
	Sp.			+	+
Leaf Count	Sa.	*	*	+	+
	Sp.	+	+	+	+
Leaf Length	Sa.	+	+	+	+
	Sp.		+	+	+
Stem Width	Sa.	+	+	+	+
	Sp.		+	+	+
Leaf Width	Sa.	+	+	+	+
Stem Count	Sp.			+	+

). Nitrogen had a varying effect on Sa. leaf count at different inundation levels in the short term, lending insight to the interactive effect between nitrogen and inundation. Nitrogen addition increased leaf count in the more inundated vegetation, helping alleviate the stress of inundation to some degree. Starting after the first sampling period when the nutrient addition began, the strong positive effect of nutrient addition on leaf count started to show up in the third sampling event in Rows 1-4, while it did not show up until the fourth sampling event in Row 5, and indicated little to no effect on Row 6, the least inundated row (Table 4 and Figure 7). By Year 2, nutrient addition had a strong positive impact on every metric for both species (Table 4).

3.6. The Impact of Pre-Condition on Morphological Traits

Pre-condition largely had a positive impact on most of the morphological traits in Year 1, with strong positive impact on Sa. and Sp. plant height, Sa. stem width, and Sp. stem count, as well as a moderate positive impact on Sa. leaf width and Sp. stem width (

Table 5. Summary of signs of 50% and 95% credible intervals (CIs) of pre-condition impact based on the posteriors of multi-level Bayesian models for metrics of Spartina alterniflora (marked in red) and Spartina patens (marked in black). Positive impact is denoted by (+) and negative impact is marked by (-). Also see 50% and 95% CIs in Figures S5–S14 in the Supplementary Materials. (If no signs were provided for a particular CI, it means the CI intercepted 0).

		Year 1 (4 months)		Year 2 (16 months)
Metrics	Species	Pre-Condition		Pre-Condition
		95% CI	50% CI	95% CI	50% CI
Plant Height	Sa.	+	+	+	+
	Sp.	+	+	+	+
Leaf Count	Sa.				-
	Sp.			+	+
Leaf Length	Sa.			+	+
	Sp.				+
Stem Width	Sa.	+	+
	Sp.		+
Leaf Width	Sa.		+		-
Stem Count	Sp.	+	+	-	-

). Year 2 exhibited more mixed impact, with strong positive impact on Sa. and Sp. plant height (consistent with Year 1), Sp. leaf count, Sa. leaf length, a moderate positive impact on Sp. leaf length, a now strong negative impact on Sp. stem count, and a moderate negative impact on Sa. leaf count and leaf width (Table 5). The mixed effects can mean the tradeoffs of different morphological traits in response to the vegetative initial condition, shown in the relative long term. The tradeoff showed in the long term. For Sa., increased plant height and leaf length contrasted with decreased leaf count and leaf width with better initial condition. For Sp., increased plant height, leaf count and leaf length contrasted with decreased stem count.

3.7. The Impact of Senescence or Seasonality on Morphological Traits

In Year 1, we focused on the impact of time since the vegetation was planted in the marsh organ. We expected some impact attributable to adaption and senescence. It appears that senescence had strong negative impact on Sa. and Sp. plant heights and moderate negative impact on Sa. stem width, while it allowed individuals of both species to continue producing new leaves as time progressed (

Table 6. Summary of signs of 50% and 95% credible intervals (CIs) of senescence/seasonality impact based on the posteriors of multi-level Bayesian models for metrics of Spartina alterniflora (marked in red) and Spartina patens (marked in black). Positive impact is denoted by (+) and negative impact is marked by (-). Also see 50% and 95% CIs in Figures S5–S14 in the Supplementary Materials. (If no signs were provided for a particular CI, it means the CI intercepted 0).

		Year 1 (4 months)		Year 2 (16 months)
Metrics	Species	Days		Month Temp
		95% CI	50% CI	95% CI	50% CI
Plant Height	Sa.	-	-
	Sp.	-	-
Leaf Count	Sa.		+	+	+
	Sp.		+		+
Leaf Length	Sa.			-	-
	Sp.				+
Stem Width	Sa.		-
	Sp.				+
Leaf Width	Sa.				+
Stem Count	Sp.				+

). By Year 2, when the vegetation should have been acclimated to the new environment as recovery generally takes only a few weeks [40], we focused on the impact of seasonality because the 2022 measurements spanned the spring, summer, and fall seasons. Temperature showed a largely positive impact. The exception is the little to no impact of temperature on both species’ plant height and Sa. stem width, and negative impact on Sa. leaf length. Again, the mixed impact of temperature can suggest tradeoffs of varying morphological traits. In Year 2, increased leaf count and leaf width contrasted with decreased leaf length with higher temperature in Sa, while Sp. responded to temperature by optimizing the production of new leaves and stems, favoring increased width and length rather than vertical growth.

3.8. The Impact of Channel on Morphological Traits

Three traits of Sa. plants, leaf length, stem width, and leaf width, from the eastern channel outperformed individuals transplanted from the western channel in both years (

Table 7. Summary of signs of 50% and 95% credible intervals (CIs) of channel impact based on the posteriors of multi-level Bayesian models for metrics of Spartina alterniflora (marked in red). Positive impact is denoted by (+) and negative impact is marked by (-). Also see 50% and 95% CIs in Figures S5–S14 in the Supplementary Materials. (If no signs were provided for a particular CI, it means the CI intercepted 0).

		Year 1 (4 months)		Year 2 (16 months)
Metrics	Species	Channel		Channel
		95% CI	50% CI	95% CI	50% CI
Plant Height	Sa.			-	-
Leaf Count	Sa.		+	-	-
Leaf Length	Sa.	-	-	-	-
Stem Width	Sa.	-	-
Leaf Width	Sa.		-

). Larger values were consistent in leaf length for the entire duration of this study. The one exception is that vegetation from the west channel exhibited moderately higher leaf counts in Year 1. However, the advantage of the west channel in leaf count disappeared in Year 2 with larger leaf counts from east channel transplants witnessed in Year 2 (Table 7).

4. Discussion

We studied the impact of the diverse environmental factors and initial condition of individuals on biomass along with a wide variety of key morphological traits of two common salt marsh species at two different temporal scales. While three-dimensional biomass (g/cm³) showed the accumulated result for vegetation growth, studying one-dimensional morphological traits (ex. flat leaves) can provide a more in-depth explanation. This approach can provide additional insights into the tradeoffs associated with vegetation growth, helping us understand potential strategies plant individuals or species undertook to maximize survival and growth in a stressful environment. For example, we found that S. alterniflora tended to increase stem width under elevated inundation in both Year 1 and 2.

While our experiment only lasted for two years, the different results in Year 1 and Year 2 emphasize the importance of longer-term study to gain better understanding and make more accurate predictions on vegetation’s response to environmental stressors. Long-term experiments can also cover larger variability of environmental conditions and reveal vegetation’s response to more extreme events. In our study, precipitation for 2022 (Year 2) was almost 20% lower than the precipitation for 2021 (Year 1), and in some months, exhibited up to 70% less precipitation in the winter months (precipitation data gathered from https://www.wunderground.com/history/daily/us/ms/gautier/KPQL, accessed 1 March 2023). This was particularly noticeable over the winter seasons—which may have resulted in even lower inundation levels due to low, seasonal tides. More so, there may have increased plant stress in Year 2, exacerbated by the lower precipitation and higher salinity. It should be noted that the overall inundation percentages for Year 1 reflects the original marsh organ design fairly well. However, Year 2 percentages failed to meet the designed inundation percentages. This could be attributed to many reasons, such as storms, the lower precipitation, and seasonal and tidal water changes. Hurricane Ida occurred shortly (~ two months) after the beginning of the marsh organ experiments, which may have brought additional inundation in Year 1.

The impact of inundation on biomass largely matched their corresponding morphological traits in Year 2, especially for Spartina alterniflora. Optimum inundation levels exist for all the Spartina alterniflora Year 2 traits except stem width; however, different traits reached the maximum at different inundation levels. Focusing on the medians of the predictions, plant height continued to increase until 89% of inundation time, while leaf count and leaf length increased only until 35% and 40% of inundation time, respectively. Leaf width lay in the middle, reaching the maximum when inundation was 57% of time. Stem width, as an exception, increased linearly with inundation. These combined responses result in above- and belowground biomass reaching the maximum at 50% and 44% of inundation time. This suggests that Spartina alterniflora increased plant height and stem width at the expense of leaf count, leaf length, and then leaf width. This strategy may have allowed individuals to grow tall and keep leaves above water with strong stem support for photosynthesis under increased inundation.

Spartina patens, on the other hand, negatively responded to inundation in biomass, plant height, leaf count, and stem width. When inundation surpassed 52% of time, leaf length began to increase but was not able to offset the reduction in other traits, and therefore biomass continued to decrease. The better adaption of Spartina alterniflora to inundation than Spartina patens is consistent, as Spartina patens occupies marsh habitat at higher elevation and less inundation than compared to Spartina alterniflora [5,44,45]. Our findings also align with previous marsh organ experiments [5,10,11,31].

Nutrient addition had a strong positive impact on aboveground biomass for both species in the short and long term. However, nutrient addition had little to no impact on Sa. and Sp. belowground biomass in Year 1. It was only in Year 2 that we observed the strong positive impact, suggesting that belowground biomass had a lagged response to the nitrogen addition, indicating that the effect of nitrogen on above- and belowground biomass became more pronounced over time. This lagged response of belowground biomass to the nitrogen addition in Year 1 may be due to the resource allocation of the plants, that the extra nitrogen uptake may be used in the aboveground biomass first to conduct photosynthesis. Another potential reason for this lagged response may be that the stress the plants experienced from transplantation into the marsh organ could have prevented nitrogen’s positive impact on the belowground biomass; thus, by Year 2, this stress was reduced and the positive impact showed up. Previous papers support these findings in that N fertilization increases Spartina productivity over-time (S. patens, S. americanus, and S. alterniflora) [6,26,27]. We did not find the interaction between nutrient and inundation important in the majority of the traits with the exception of Sa. leaf count in Year 1. This interaction shows that nutrient addition increased leaf count, especially in the higher inundation levels, helping alleviate the stress of inundation.

There exist contrasting results describing elevated nutrient impacts on belowground biomass and the further effects on root strength [30]. Hollis and Turner [46] and Turner [47] showed a significant decline in root strength after small increases in nutrient availability. Other studies [48,49,50] did not show decreased belowground production when introduced to high nutrient addition, and one study in particular showed that nitrogen fertilization on S. alterniflora increased belowground biomass, including root biomass [25]. Particularly, greenhouse studies demonstrated that biomass production of Spartina patens increased with nitrogen addition, with a more pronounced increase in lower salinity than in higher salinity [51]. Elsey-Quirk et al. [28] pointed out that larger sediment availability is necessary to offset the negative impact from the larger variability of environmental factors or excessive nutrients driven by freshwater diversions. However, it is not clear how much nitrogen is assimilated versus lost through the denitrification process. Whether our findings on the increased above- and belowground biomass from nutrient additions based on the marsh organ experiments can be transferred to field observations requires further investigation.

The pre-condition of plants when they were first planted exhibited some impact on biomass in the short term, but not in the long term, indicating that the environmental factors played a more important role in productivity than the initial condition of the plants. The negative effect of pre-condition on the aboveground biomass of Sp. might be due to the artifice of the selected pre-condition metric (in this case, stem count). When we studied the individual morphological traits, we found positive to no impact of pre-condition in Year 1, but mixed results in Year 2. The mixed impact might offset each other somehow, partially explaining why there is no impact in the biomass, in addition to the diminishing impact of initial condition compared to environmental conditions.

When testing for differences in response variables between the S. alterniflora sites, the eastern channel plants’ biomass strongly outperformed the western channel plants in the short term. However, this effect lessened to moderate outperformance in the long term, suggesting that environmental conditions played a more important role than the source of vegetation as the time went by likely through the adaptation mechanism. The larger biomass in the vegetation from the eastern channel was also reflected in the majority of the morphological traits. The eastern channel is much more anthropogenically influenced from urban development, industrial factories on the shorelines, and dredged channels when compared to the western channel. The vegetation transplanted from the disturbed sites may have been able to outperform the local individuals given that their environmental constraints were removed. In the field observation, the western channel of the Pascagoula River contained significantly higher belowground biomass than the eastern channel, while vertical distribution of belowground biomass did not strongly vary between channels [13]. While more work could further evaluate the differences between the east and west channel plants, we do not believe that incorporating plants from the two channels strongly impacted vegetation’s response to inundation and nitrogen, especially as the outperformance decreased over time, indicating origin of an individual mattered less with time. Collecting plants from both channels may actually help better link a biomass function from this study to saltmarsh wetlands under a range of anthropogenic stressors.

Studying the impact from seasonality, measured either in days since transplant in Year 1 or monthly temperature in Year 2, the mixed responses witnessed in Year 1 may be due to the plants adapting and senescence acting together. In Year 2, after adapting and covering more seasons, temperature had a mostly positive impact on vegetative traits. For S. alterniflora, the results suggest existence of a tradeoff, where leaf count and leaf width favorably responded to higher temperature at the cost of leaf length.

5. Conclusions

There are a few general trends noticeable from the results of this study. Increased inundation negatively impacted Sp.; however, there existed an optimal inundation level for Sa. in end-of-season biomass. This suggests that the low-marsh plant S. alterniflora will, to some degree, adapt better to increasing inundation while high-marsh plant S. patens will struggle to adapt. Nutrient addition stimulated both below- and aboveground biomass for both species, where this effect was more pronounced in the long term. The incorporation of temporal scales explicitly highlighted the importance of how a longer-term in situ mesocosm study lends insight on the adaptation of transplanted individuals to their new environment conditions, which played a more important role than the initial vegetation’s status as time progressed. Some impacts increased over time, such as nitrogen addition, while others decreased, such as source of vegetation. In this study, we observed some tradeoffs between different morphological traits in response to environmental stressors, indicating various growth strategies to maximize survival and productivity.

This study benefits our understanding of the factors relevant to sea level rise, freshwater diversions, and climate change in their impact on common coastal wetland vegetation in the short and longer term. These findings will therefore potentially facilitate further evaluation of conservation and restoration practices on coastal wetlands while filling in the lack of mechanistic understanding of vegetation’s response and their adaptive strategies in this region.