Evolution of Coastal Landforms: Investigating Sediment Dynamics,
Hydrodynamics, and Vegetation Dynamics
by
Fateme Yousefi Lalimi
Earth and Ocean Sciences Duke University
Date:______Approved:
______Marco Marani, Supervisor
______Sonia Silvestri
______Allen Murray
______James Heffernan
______Laura Moore
Dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Earth and Ocean Sciences in the Graduate School of Duke University
2018
ABSTRACT
Evolution of Coastal Landforms: Investigating Sediment Dynamics,
Hydrodynamics, and Vegetation Dynamics
by
Fateme Yousefi Lalimi
Earth and Ocean Sciences Duke University
Date:______Approved:
______Marco Marani, Supervisor
______Sonia Silvestri
______Allen Murray
______James Heffernan
______Laura Moore
Dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Earth and Ocean Sciences in the Graduate School of Duke University
2018
Copyright by Fateme Yousefi Lalimi 2018
Abstract
Coastal ecosystems provide a wide range of services including protecting the mainland from the destructive effects of storms, nutrient cycling, water filtration, nurseries for fish and crustaceans, and carbon sequestration. These zones are threatened by human impacts and climate change through more frequent intense storms and sea level rise with a projected increase of up to 16 mm/yr for the last two decades of the 21st century. However, it is not fully understood what mechanisms control the formation and degradation of many of these landforms, and determine their resilience to environmental change. In this work, I use remote sensing, field observations, and mathematical modeling to better understand the interplay of sediment and vegetation dynamics, waves, and tidal flows in shaping key coastal landforms.
First, I develop and use remote sensing analyses to quantitatively characterize coastal dune eco-topographic patterns by simultaneously identifying the spatial distribution of topographic elevation and vegetation biomass in order to understand the coupled dynamics of vegetation and coastal dune morphology. Lidar-derived leaf area index and hyperspectral-derived normalized difference vegetation index patterns yield vegetation distributions at the whole-system scale which are in agreement with each other and with field observations. Lidar-derived concurrent quantifications of biomass and topography show that plants more favorably develop on the landward side of the foredune crest and that the foredune crestline marks the position of an ecotone, which is interpreted as the result of a sheltering effect sharply changing local environmental
iv
conditions. The findings reveal that the position of the foredune crestline is a chief ecomorphodynamic feature resulting from the two-way interaction between vegetation and topography.
In the second part of my work, I focus on the vertical depositional dynamics of salt marshes in response to sea level rise. I investigate the hypothesis that competing effects between biomass production and aeration/decomposition processes determine an approximately spatially constant contribution of soil organic matter (SOM) to total accretion. I use concurrent observations of SOM and decomposition rates from marshes in North Carolina. The results are coherent with the notion that SOM does not significantly vary in space and suggest that this may be the result of an at least partial compensation of opposing trends in biomass productivity and decomposed organic matter. The analyses show that deeper soil layers are characterized by lower decomposition rates and higher stabilization factors than shallower layers, likely because of differences in inundation duration. However, overall, decomposition processes are sufficiently rapid that the labile material in the fresh biomass is completely decomposed before it can be buried and stabilized. The finding point to the importance of the fraction of initially refractory material and of stabilization processes in determining the final distribution of SOM within the soil column.
Finally, in the last part of my work, I focus on estuarine scale modelling and develop a process-based model to evaluate the relative role of watershed, estuarine, and oceanic controls on salt marsh depositional/erosional dynamics and define how these
v
factors interact to determine salt marsh resilience to environmental change at the estuary scale. The results show that under some circumstances, vertical depositional dynamics can lead to transitions between salt marsh and tidal flat equilibrium states that occur much more rapidly than marsh/tidal flat boundary erosion or accretion could.
Additionally, the analyses reveal that river inputs affect the existence and extent of marsh/tidal flat equilibria by both modulating exchanges with the ocean and by providing suspended sediment.
vi
Dedication
To Bijan
vii
Contents
Abstract ...... iv
List of Tables ...... x
List of Figures ...... xi
Acknowledgements ...... xiii
1. Introduction ...... 1
2. Coupled Topographic and Vegetation Patterns in Coastal Dunes ...... 3
2.1 Introduction ...... 3
2.2 Methods ...... 6
2.2.1 Study Area...... 6
2.2.2 In-situ Observations...... 8
2.2.3 Hyperspectral and Lidar Data...... 9
2.2.4 Computation of the NDVI ...... 10
2.2.5 Construction of the Digital Terrain Model ...... 11
2.2.6 Shoreline and Dune Crestline Identification ...... 13
2.2.7 Estimation of the LAI...... 15
2.3 Results and Discussion ...... 16
2.4 Conclusions ...... 25
3. The Spatial Variability of Organic Matter and Decomposition Processes at the Marsh Scale ...... 27
3.1 Introduction ...... 27
3.2 Methods ...... 30
3.2.1 Study Area...... 30
3.2.2 Core Collection and Topographic Survey ...... 31 viii
3.2.3 SOM Analysis ...... 32
3.2.4 Vegetation Cover Quantification ...... 33
3.2.5 Decomposition Rate ...... 35
3.2.6 Flooding Duration ...... 37
3.2.7 Data Analysis ...... 37
3.3 Results ...... 38
3.4 Discussion...... 44
3.5 Conclusions ...... 50
4. Watershed and Estuarine Controls on Salt Marsh Distributions and Resilience ...... 53
4.1 Introduction ...... 53
4.2 Model Structure ...... 55
4.3 Results ...... 61
4.4 Discussion and Conclusions ...... 77
5. Conclusions ...... 83
Appendix A ...... 85
Appendix B ...... 87
Appendix C ...... 88
References ...... 96
Biography ...... 111
ix
List of Tables
Table C.1: Model Parameters ...... 91
x
List of Figures
Figure 2.1: Hog Island (VA) and targeted study sites ...... 7
Figure 2.2: Seaward side of the foredune (dune line closest to the shoreline) on Hog Island ...... 8
Figure 2.3: Comparison between ground elevation values from field GPS measurements and Lidar-derived DTM elevations...... 13
Figure 2.4: DTM (a, d), LAI (b, e) and NDVI (c, f) for site 1 (a, b, c) and site 2 (d, e, f) ..... 17
Figure 2.5: LAI vs. NDVI for the entire site 3 ...... 19
Figure 2.6: LAI as a function of distance from the first crestline for site 1 (a), site 2 (b), and site 3 (c) ...... 20
Figure 2.7: Heat maps of LAI for site 1 (a) and site 2 (b)...... 22
Figure 2.8: a) LAI values on the landward (triangles) and seaward side (circles) located within 15 m of the crestline ...... 23
Figure 2.9: Vegetation percent cover along 3 transects from the shoreline to the backdune area ...... 24
Figure 3.1: Study sites in Wrightsville Beach and Masonboro Island in North Carolina . 31
Figure 3.2: View of the marsh were sampling were conducted in study site 1 along transect T1 ...... 34
Figure 3.3: Marsh surface elevation relative to mean sea level (a, b, and c); soil dry bulk density, BD, (d, e, and f); fractional soil organic matter, SOM, (g, h, and i) ...... 40
Figure 3.4: Soil dry bulk density, BD, (a, b and c), fractional soil organic matter, SOM, (d, e and f), soil organic matter density, OMD, (g, h, and i) ...... 41
Figure 3.5: Organic matter density, OMD, vs. flooding duration along transects T1 ...... 42
Figure 3.6: Decomposition rate (a, b, and c) and stabilization factor (d, e, and f) values as a function of surface elevation ...... 43
Figure 3.7: Decomposition rate (a, b, and c) and stabilization factor (d, e, and f) with respect to flooding duration ...... 44
xi
Figure 3.8: The relation between decomposition rate and stabilization factor along transects T1 ...... 50
Figure 4.1: Scheme of a simplified symmetric estuary ...... 56
Figure 4.2: The relationships between fetch length and river sediment concentration (a), river discharge (b), ocean sediment concertation (c) and tidal range (d) ...... 64
Figure 4.3: The relationships between fetch length and river sediment concentration (a), river discharge (b), ocean sediment concertation (c) and tidal range (d) ...... 67
Figure 4.4: The relationships between fetch length and river sediment concentration (a), river discharge (b), ocean sediment concertation (c) and tidal range (d) ...... 70
Figure 4.5: The relationships between critical fetch length associated with both vertical and lateral dynamics and river sediment concentration (a), river discharge (b), ocean sediment concertation (c) and tidal range (d) ...... 72
Figure 4.6: The relationships between fetch length with basin width (a and c) and estuary length (b and d) ...... 75
Figure 4.7: Sensitivity measures of main effects S (green), total effects T (orange), and effects due to interactions only, T-S (grey) ...... 76
Figure 4.8: Heat map of the interactions between pairs of variables ...... 77
Figure A.1: NDVI as a function of distance from the foredune crestline………………… 85
Figure A.2: Heat maps of NDVI ...... 86
Figure B.1: Soil organic carbon density, OCD, vs. marsh surface elevation relative to mean sea level………………………………………………………………………………….. 87
Figure C.1: The relationships between critical fetch length associated with both vertical and lateral dynamics…………………………………………………………………………... 95
xii
Acknowledgements
This work would not have been possible without the support of my advisor,
Marco Marani, who taught me how to approach complex scientific questions and solve them with the best tools in hand. I would also like to thank my committee members,
Sonia Silvestri, Brad Murray, Laura Moore and Jim Heffernan for their valuable guidance that was a fundamental part of this work.
I am deeply grateful to my friends, colleagues and mentors from Nicholas
School. Thanks to Bijan Seyednasrollah for his great contribution to my field experiments. I owe thanks to the many of my friends who have helped me throughout my PhD, including Anna Braswell, Rebecca Lauzon, Marcella Roner and my lab mate
Enrico Zorzetto. I would also like to thank Dr. Andrea D’Alpaos for his valuable insights on my marsh projects, and his great sense of humor.
I would like to especially thank my parents, Behnaz and Mohamadreza, my sisters, Samaneh, Maryam and Leila, and my nieces, Nila and Neli, who supported me from the other side of the planet. Finally, and most importantly, I would like to express my gratitude and appreciation to the other half of my soul, Bijan, who patiently stood behind me during all these years.
xiii 1. Introduction
Coastal environments provide a variety of ecosystem services including long- term storage of greenhouse gases, improving water quality, outdoor activities and tourism, and habitat for wildlife. They also play an important role in protecting the mainland by mitigating ocean swells and reducing the destructive effects of wind and waves during storms. However, the stability of these landforms is threatened by these forcings and also by environmental change. For example, sea level rise and changing storm patterns have deep and destructive effects on coastal zones. In addition, human activities such as dam construction and flow diversion can also greatly impact the stability of these zones.
Studies show that vegetation plays a central role in stabilizing coastal dunes and barrier islands by reducing erosion and trapping sands, and thus exerts a first-order control on barrier island dynamics [e.g., Short and Hesp, 1982; Hesp, 1988; Hesp, 2002;
Psuty, 2008; Duran and Moore, 2013; Walker et al., 2013]. In addition, coastal marshes are resilient to environmental change due to a positive feedback between vegetation and deposition dynamics, maintaining their elevations and preventing drowning
[Fagherazzi et al., 2012; French and Stoddart, 1992; Kirwan and Murray, 2007; Marani et al., 2007]. However, environmental changes can lead to marsh collapse, retreat or migration under some circumstances such as elevated rates of sea level rise and sudden reductions in sediment supply [D'Alpaos, 2011; Kirwan et al., 2016; Marani et al., 2010;
Morris et al., 2002]. Therefore, it is important to understand the dynamics that shape
1
coastal landforms, to carefully predict their response to environmental change. Here, I use different tools to study some of these dynamics. In chapter two, I focus on barrier island's dunes and use remote sensing to study the coupled vegetation and sediment transport dynamics. In chapter three, I focus on salt marshes behind barrier islands to understand what processes contribute to marsh accretion dynamics as a response to sea level rise. Finally, in chapter four, I develop a mathematical model to predict the evolution of tidal platforms under different scenarios of sea level rise rate and limitations in sediment supply within estuaries.
2 2. Coupled Topographic and Vegetation Patterns in Coastal Dunes
2.1 Introduction
Barrier islands play a key role in the stability of lagoons and estuaries, and in protecting the mainland from the destructive effects of storms. In turn, the stability of barrier islands is governed by coupled biotic and abiotic processes responding to environmental change and anthropogenic interference. Among other factors, sea level rise and more frequent intense storms are projected to have important effects on the geomorphological and ecological processes regulating barrier island dynamics [e.g.,
Moore et al., 2010; Zinnert et al., 2011; Duran Vinent and Moore, 2015]. In particular, vegetation plays a central role in engineering dune topography by trapping and stabilizing the sand, and thus exerts a first-order control on barrier island dynamics [e.g.,
Short and Hesp, 1982; Hesp, 1988; Hesp, 2002; Psuty, 2008; Duran and Moore, 2013;
Walker et al., 2013]. Burial-tolerant species, such as Ammophila breviligulata, are known to increase sediment accumulation in dune areas [Godfrey, 1977; Disraeli, 1984], whereas species less tolerant to burial, such as Spartina patens, are considered to be important in stabilizing low-elevation areas, and in promoting dune recovery after overwash events
[Godfrey, 1977; Ritchie and Penland, 1988; Wolner et al., 2013].
The influence of topography-mediated factors on vegetation development has also been established and quite extensively studied. Elevation, slope, and distance from the shoreline have been identified as main habitat determinants for different dune vegetation species [e.g., Young et al., 2011; Brantley et al., 2014; Tobias, 2015]. It is thus 3
understood that, on one hand, vegetation distribution and biomass depend on a number of different factors, such as salinity, wind exposure, flooding [e.g., see Hayden et al.,
1995; Parker and Bendix, 1996; Molina et al., 2003], and that, on the other hand, vegetation characteristics exert a major control on the development of the topographic features mediating these same abiotic governing factors. Hence, much like what is observed in other landscape systems [Corenblit et al., 2011; Marani et al., 2013], vegetation and geomorphological processes jointly determine the evolution of barrier landforms [e.g., Duran Vinent and Moore, 2015; Goldstein and Moore, 2016], a concept now quite generally known in land-surface science as ecomorphodynamics [Murray et al., 2008; Marani et al., 2010; Corenblit et al., 2011; Pelletier et al., 2015]. However, only relatively recent work has begun addressing vegetation dynamics and dune dynamics as mutually interacting processes [e.g., Stallins and Parker, 2003; Hacker et al., 2012;
Zarnetske et al., 2012; Duran and Moore, 2013; Seabloom et al., 2013]. The concurrent characterization of vegetation distribution and dune topography at the whole-dune or larger scales, is currently tackled using time-intensive longitudinal observations that require decades to acquire [Vandermaarel et al., 1985; Frederiksen et al., 2006]. More recently yet, remote sensing approaches are being explored to expand the simultaneous study of dune vegetation and dune dynamics and to provide rapid, cost-effective, and repeatable data collection across a wide range of temporal and spatial scales
[Kempeneers et al., 2009; Malavasi et al., 2013]. However, vegetation characterization
(e.g., species identification and biomass density) can be challenging in the presence of
4
short and sparse canopies or of highly mixed populations [Silvestri et al., 2003], as is usually the case in coastal dunes.
In this part of my work, I develop and apply new remote sensing analytical techniques to quantitatively characterize barrier island eco-topographic patterns from the plot to the barrier island scale. To this end, I first compare vegetation/topographic characterizations obtained from satellite optical remote sensing (such as hyperspectral- derived Normalized Difference Vegetation Index, NDVI), airborne laser altimetry (such as Lidar-derived Leaf Area Index, LAI), and field observations. Although the simultaneous computation of NDVI and LAI, and their comparison, have been explored in other ecosystems [e.g., Bernasconi et al., 2003; Jin and Eklundh, 2014], this comparison had not yet been performed for short vegetation species typical of coastal systems. Next,
I use the resulting spatial distributions of ecological and topographic properties to identify and study the outcome of coupled ecological and sediment transport processes.
The foredune crestline, i.e. the dune ridge closest to the shoreline, is a significant geomorphic feature, and its elevation is known to exert an important control on dune dynamics [e.g. Long et al., 2014]. Here I characterize its ecomorphodynamic significance.
On one hand, it marks the transition between two zones where environmental conditions are very different (in terms of wind velocity, temperature, salinity, exposure to storm surges), on the other, the crestline position is the result of the mutual interaction between vegetation and sand transport. Here I use remote sensing observations to test the hypothesis that the distribution of vegetation biomass is closely
5
related to the position of the crestline, and that this is as strong a control of vegetation biomass as the distance from the shoreline and ground elevation, commonly assumed to be the sole determinants of vegetation habitat suitability in coastal dunes [Young et al,
2011].
2.2 Methods
2.2.1 Study Area
The study area examined here is located in Hog Island, Virginia (Figure 2.1), part of a chain of barrier islands located along the Eastern Shore of Virginia, USA. Hog Island is a rotational island with a history of alternating erosion and accretion at opposite ends, resulting in changes in the position of shoreline [Hayden et al., 1991]. This island is characterized by multiple continuous dune ridges, with crest elevation between 2 – 4 m, where the most abundant vegetation species are Ammophila breviligulata, Spartina patens, and Panicum amarum (Figure 2.2).
6
Figure 2.1: Hog Island (VA) and targeted study sites. Red boxes mark the location of study sites 1 and 2. Blue dotted lines represent the location of 3 transects where field observations of vegetation distribution were collected. The yellow box shows the location of site 3 which includes site 1, site 2, and the transects where field observations were taken. Sites 1 and 2 were identified to explore within-site variability whereas site 3 encompasses a large study area that would be statistically significant and yet computationally manageable. Green circles mark the locations of GPS survey.
7
Figure 2.2: Seaward side of the foredune (dune line closest to the shoreline) on Hog Island. Vegetation cover is sparser on the seaward side of the dune and increases toward the dune crest (a). The photos on the right illustrate the range of vegetation cover encountered in the area (b and c).
2.2.2 In-situ Observations
Reference topographic data across the study area (used to validate and correct ground elevation from Lidar) were available from a high-resolution GPS survey (using
Trimble R7/8 GNSS; stationary base receiver at a known location broadcasting corrections to a mobile rover receiver) with a cm-scale accuracy performed in August
2010 on 100 plots (0.5 m × 0.5 m) along two sets of transects, respectively perpendicular and parallel to the dune lines [Wolner and Moore, 2010; Wolner, 2011; Wolner et al.,
2013]. I performed a field campaign in July 2015 (at near-maximum seasonal development of dune vegetation biomass), to document the spatial distribution of vegetation cover along 3 shore-perpendicular transects extending from the high tide water mark to the back of the dunes (blue transects in Figure 2.1). To quantify vegetation cover,
8
I analyzed photographs taken with a nadir-looking camera and covering a field of view of about 0.5 m × 0.5 m, with a spacing of 5 m along each transect. A binary map of “green” vs. “non-green” pixels was then created for each photo based on the HSV color space
(cylindrical-coordinate representation of the Red-Green-Blue color model with 3 components: Hue, Saturation and Value). Green-pixel mapping was based on the identification of photo-specific thresholds for H, V, and S values that maximized the correspondence between pixels classified as green (based on the specified thresholds) and those visually classified as such. Adapting the threshold values to each acquired photo allowed to properly adjust for differences in illumination and shadowing (itself a function of the amount of biomass). The vegetation cover fraction attributed to each photo was then computed as the fraction of green pixels with respect to the total number of pixels in the photo.
2.2.3 Hyperspectral and Lidar Data
Hyperspectral data were acquired by the Long-Term Ecological Research –
Virginia Coast Reserve (LTER – VCR) program [Brantley et al., 2011]. The data were collected by an airborne ProSpecTIR VIS hyperspectral imaging spectrometer (SpectIR
Corp., Reno, NV, USA) in August 2008, when vegetation biomass was near its maximum. The data span the 400 nm - 1000 nm spectral range with 128 bands, all 5 nm wide, and a 1 m spatial resolution. The hyperspectral data were subsequently atmospherically corrected to obtain the at-surface reflectance using the MODTRAN4 radiative transfer model [Berk et al., 1999]. The Lidar data (near-infrared light) were acquired in October 2011 using an airborne Riegl LMS-Q680i laser scanner with scan
9
angle range of ±30 degrees [USACE-TEC and JALBTCX, 2014] included a single return for each emitted beam. The Lidar return density is between 20 and 30 returns/m2 and the elevation estimates have a nominal error standard deviation of 20 mm.
2.2.4 Computation of the NDVI
I use the hyperspectral data to characterize the spatial distribution of vegetation biomass through the computation of the NDVI, defined as the difference between two reflectance values in the infrared and red wavelengths (a function of chlorophyll content), normalized by their sum: (near infrared – red) / (near infrared + red). NDVI has been widely shown to be a representative proxy for green biomass [Tucker, 1979; Manso et al., 1998; Lo and Quattrochi, 2003; Bermudez and Retuerto, 2013]. In a large number of applications to different biomes, NDVI has also been successfully related to the LAI, the total one-sided leaf area within the canopy per unit of horizontally-projected ground area, a parameter that mediates photosynthetic, gas exchange, and rain interception processes [e.g., see Soegaard, 1999; Duchemin et al., 2006]. Some studies have used
NDVI to infer dune vegetation biomass using satellite imagery [Alphan, 2011; Castanho et al., 2015], but they have been limited by the coarse resolution of the data available (30 m in the case of Landsat data), particularly in consideration of the small-scale variability and sparseness of vegetation typical of dune areas. Here I use 1 m resolution hyperspectral data and, in order to obtain NDVI values which can be compared to widely studied Landsat-derived values, I average reflectances over appropriate sets of hyperspectral bands to obtain broadband reflectances corresponding to the red (630-690
10
nm) and infrared (770-900 nm) bands of the ETM+ sensor on Landsat 7. I obtained the averages by applying appropriate weights to each hyperspectral band to account for the known ETM+ response function [Chander et al., 2009].
2.2.5 Construction of the Digital Terrain Model
I first processed the Lidar data by removing outliers, defined as unrealistically low values (negative values, which would lie well below local sea level) and unrealistically high values (greater than a value L = M + 5 m, where M is the median elevation calculated for the entire data set, and assuming that the characteristic height of the highest shrubs present in the area is lower than 5 m). To generate the Digital Terrain
Model (DTM), i.e., the digital representation of the topographic surface, the Lidar returns reflected by canopy elements must first be identified and removed. I do this by dividing the area into square cells with side 0.5 m and by identifying the lowest elevation value in each cell as most likely being a ground return (i.e., not reflected by a canopy element). These “lowest points” are then used to construct a Delaunay
Triangulation to define a continuous soil surface throughout the domain. Notice that this construction does not yield a regularly spaced DTM, as the positions of the irregularly spaced “lowest points” induce a Triangular Irregular Network (TIN) representation of the soil surface, which is more flexible in describing local slope and elevation than a grid DTM (where, by definition, elevation is constant within each grid cell). I have extensively tested the effect of the choice of the grid cell size on the resulting
DTM. Larger cells contain a greater number of Lidar returns, such that the probability
11
that a Lidar beam can penetrate the canopy all the way to the ground increases. On the other hand, using a large cell size may cause genuine ground returns, generated by relatively high local elevation maxima, to be interpreted as canopy elements and induces a “smoothing” of small-scale topographic features. I thus compared the DTMs obtained using 6 different grid sizes, 0.25 m, 0. 5 m, 0.75 m, 1 m, 1.25 m, and 1.5 m with the available reference GPS observations. The DTM obtained using a 0.5 m grid size generated the smallest root-mean-square deviation from GPS-determined elevation values and was thus, selected as the optimal grid cell size. The DTM derived through this procedure agrees well with ground elevation measurements (R2 = 0.95, see Figure
2.3). The regression of GPS observations on Lidar estimates indicates the presence of a slight underestimation. This may be due to a mismatch between the datums used in the
GPS and Lidar acquisitions, and to the fact that the minimum elevation return from a cell is likely to be smaller than a GPS observation collected randomly within the same cell. These two sources of error (the bias offset between the datums and the use of the minimum elevation within a neighborhood of fixed size for the estimation of ground elevation) cannot be separated and therefore need to be corrected using ancillary field surveys if absolute elevation with respect to a specified datum is required. The GPS vs.
Lidar regression, however, makes it possible to correct for this limited bias, and the final
DTM elevation is thus obtained by applying the regression/correction relation ZGPS= 0.96
Zlidar + 0.20 m.
12
Figure 2.3: Comparison between ground elevation values from field GPS measurements and Lidar-derived DTM elevations.
2.2.6 Shoreline and Dune Crestline Identification
I use a combination of Lidar and hyperspectral data to define the position of the shoreline. Water is highly absorptive in the infrared part of the electromagnetic spectrum. I thus set a threshold reflectance in the hyperspectral band centered around the 801 nm wavelength. The threshold is based on the maximum value of visually identified water and wet areas along the coast. A “water mask” is defined by all the pixels in the study area with reflectance smaller than the threshold. The instantaneous boundary between land and water areas (which is subject to spatially heterogeneous
13
factors related to nearshore circulation and waves) does not define an equal-elevation shoreline. Thus, from the Lidar data, I determine the maximum elevation of this boundary and use this elevation value to define a reference, equal-elevation shoreline position (similar to extraction of a mean high water shoreline but using an elevation derived from my data set, for internal consistency, rather than a tide gauge). Note that because the relative position from the shoreline, rather than the absolute distance from it, is the relevant factor in the analysis, other definitions of shoreline could be used with no consequences for the results. This is consistent with conventional definitions of the shoreline, the actual shoreline being variable over time due to tidal fluctuations (mean tidal range 1.19 m), waves, and storm surges. I experimented with different choices of the water elevation used to define the shoreline and found that the corresponding modest, and rather spatially uniform, shifts in shoreline position do not appreciably affect my subsequent analyses.
As previously discussed, the dune crestline likely represents an ecotone separating environments where stressors such as sediment and salt deposition rates are very different. I use the DTM to identify the position of the foredune crestline by tracking the value of the local topographic gradient systematically along transects connecting the ocean and the backdune area. The crestline is identified as the set of points where the topographic slope changes sign in an ocean-backdune direction. To find the foredune crestline, I first determine the elevation of the center of each 1 m pixel where the NDVI was previously computed, and subsequently identify the position of
14
local topographic maxima where the derivative of ground elevation changes sign along transects in the ocean-land direction. I manually removed randomly occurring local maxima related to micro-topographic features not belonging to the continuous foredune crestline
2.2.7 Estimation of the LAI
The cloud of Lidar returns has been shown to convey information on the canopy structure, and on the LAI in particular, in a variety of environments [Kempeneers et al.,
2009; Wang et al., 2009; Fieber et al., 2014; Luo et al., 2015]. I use here an approach originally developed for application to forest environments, which is based on the definition of gap probability, i.e., the probability that a laser beam will penetrate the canopy to reach the ground [Houldcroft et al., 2005; Richardson et al., 2009]:
−퐺(휃) 퐿퐴퐼 푃(휃) = exp ( ) (1.1) cos 휃 where 휃 is the view zenith angle, 퐺(휃) describes leaf clumping and the mean leaf projection in the direction 휃 and varies between 0 and 1. I assume here, in line with much of the current literature, a random foliage orientation, which implies 퐺(휃) = 0.5
[Campbell, 1986; Houldcroft et al., 2005]. I also performed sensitivity analyses to the variability in the view angle, and found that the variation in 푃(휃) is negligible when 휃 is varied within ±30˚ (the Lidar’s field of view). Hence, in the following, I consider 휃 = 0 for all returns. To estimate LAI from equation (1.1), I compute 푃(휃), for each grid cell, as the fraction of Lidar returns reflected by the ground with respect to the total number of returns in that cell. The identification of the Lidar returns reflected by the ground is here
15
obtained by first considering the elevation difference between each individual Lidar return and the elevation of the DTM previously obtained by Delaunay Triangulation
(and subsequent bias correction). This difference expresses the height of the target that has caused the return with respect to the best estimate of the local ground elevation.
Lidar returns are here considered to be generated by ground reflections if the height, with respect to the local ground elevation, at which this reflection occurred is smaller than a fixed threshold. This threshold represents the typical minimum height at which canopy elements are expected to occur. I have evaluated the sensitivity of the results to threshold height values between 5 cm and 15 cm. I have found little sensitivity of the properties of the spatial distribution of LAI values inferred using different thresholds in this range. Furthermore, the visual inspection of the frequency distributions of return elevations at several sample sites reveals that a threshold of 10 cm is consistently lower than the bulk of the returns coming from the canopy. As a result, I adopted a 10 cm threshold for all the pixels in the domain. Finally, equation (1.1) was inverted to obtain the estimate of LAI throughout the study sites.
2.3 Results and Discussion
Inspection of the DTMs (Figure 2.4 a and d) indicates the presence of two lines of dunes roughly parallel to the shoreline. At site 1 only the first dune line (i.e., closest to the shoreline) is uninterrupted, whereas the second dune line is highly discontinuous.
Both dune lines are continuous at site 2. The spacing between the two dune lines is about 50 m and 20 m in site 1 and site 2 respectively. The distributions of NDVI and LAI
16
show quite consistent vegetation patterns (Figure 2.4 b, c, e, and f), such as the shapes of the high-density vegetation clusters, or the elongated vegetation patterns aligned with the general crestline/shoreline direction in the proximity of the first crestline.
Figure 2.4: DTM (a, d), LAI (b, e) and NDVI (c, f) for site 1 (a, b, c) and site 2 (d, e, f). The first crestline (closest to the shoreline) and the shoreline are shown in each map.
Since NDVI and LAI are different and independently-derived proxies for vegetation amount, the coherence of the patterns in the two distributions is highly supportive of their significance. The differences visible in the detailed NDVI and LAI patterns (Figure 2.4) can likely be attributed to two main causes. First, NDVI and LAI represent different vegetation properties and in different ways: The NDVI is an optical 17
property whose relation to biomass is empirical and species-specific, while Lidar- derived LAI comes from probing the canopy with narrow laser beams. Second, and possibly more important, co-registration uncertainties between NDVI and LAI can be quite significant, because of the absence of identifiable reference targets. The direct comparison of LAI and NDVI values in the whole study area (site 3) shows an increasing pattern for LAI as NDVI increases despite the uncertainties in the data as described above (Figure 2.5). Therefore, while the general correspondence of NDVI and
LAI spatial features supports the idea that dune biomass distribution can be mapped by remote sensing, I note that LAI estimates are by definition spatially coherent with DTM information, having been obtained from the same Lidar data. Here given the importance of establishing firm relations between topographic and biomass distributions, I focus the following analyses on LAI information.
18
Figure 2.5: LAI vs. NDVI for the entire site 3. The lighter yellow colors represent higher data density whereas the darker blue colors show lower data density. Red circles represent the median of LAI values within NDVI intervals of 0.01.
The foredune crestline at site 1 is both lower and closer to the shoreline than at site 2 (Figure 2.4 a and d). This is consistent with the notion that lower dunes arise when dune-building vegetation becomes established closer to the shoreline and thus in association with a narrower beach providing a lower sediment supply. This sediment supply effect arises from the upward steering of wind by dune topography, which is greater for wider beaches because the effect of the dune on the wind occurs over a broader distance compared to narrow beaches [Duran and Moore, 2013]. The locally high vegetation density near the first crestline (Figure 2.4 b, c, e, and f) is highly suggestive of a tight relation between density and topography. This finding is best appreciated if one plots LAI values against the distance from the most seaward crestline point (Figure 2.6, also see Figure A.1 in appendix A for the analogous analysis involving NDVI). Both distributions indicate that vegetation is sparser in the seaward side of the foredune area and that it is densest near the crestline and in the landward side of the dune. At site 2
(Figure 2.6 b), LAI seems to exhibit some symmetry about the crestline. However, when
LAI distribution is averaged over the whole study area (site 3, which is about 8 times larger than site 2) symmetry disappears, suggesting it is not a characteristic feature of the biomass distribution. In fact, vegetation density is quite high in a whole area near the crest, whereas it decreases very quickly away from the crestline in the direction of the shoreline. This finding, which quantifies the important ecomorphodynamic role of the
19
foredune crestline, is in agreement with previous qualitative observations in other parts of the world [Keijsers et al., 2015].
Figure 2.6: LAI as a function of distance from the first crestline for site 1 (a), site 2 (b), and site 3 (c). Each point corresponds to a grid cell in the considered site. The lighter yellow colors represent a higher number of data points/unit of graph area whereas the darker blue colors show lower data density. Red circles represent the median of LAI values within distance intervals of 3.5 m. In the figure, the shoreline is
20
located to the right of the foredune crestline, such that the seaward and landward sides of the foredune are identified by positive and negative distances to the foredune crestline, respectively. The vertical line shows the location of the foredune crestline.
The analysis of the relation between vegetation density, elevation, and distance from the shoreline (Figure 2.7, see Figure A.2 in appendix A for analogous results involving NDVI) shows the important role of the crestline: Biomass density takes on very different values at equal elevations on the seaward and landward sides of the foredune crestline. I hypothesize this to be related to a sheltering effect that reduces the intensity of environmental stressors (such as salt and sediment deposition and wind) on the landward side of the crestline. In this hypothesis, the observed ecomorphodynamic patterns represent the signature of a two-way interaction between vegetation and the landscape: Vegetation contributes to build the dune through sand trapping and thus engineers an environment where it can develop more abundantly. Figure 2.6 and Figure
2.7 also show that vegetation density remains extremely low (LAI less than about 0.5) on the seaward side of the first crestline, suggesting that little or no vegetation can develop at distances from the shoreline shorter than a critical value. This value is considered as one of the chief factors constraining dune building as it determines the position where vegetation can first colonize the beach, i.e. where stressors are low enough for vegetation to develop, and is analogous to threshold parameters used in ecomorphodynamic modelling [e.g., see Duran and Moore, 2013]. I quantify here this critical distance to be between 20 m and 35 m. Interestingly, the minimum distance for vegetation development also corresponds to the distance from the shoreline where the first sharp
21
increase in topographic slope is observed, about 20-30 m in Figure 2.7. This coincidence of geomorphic and vegetation thresholds is a highly suggestive signature of the feedbacks between sediment transport and vegetation biomass.
Figure 2.7: Heat maps of LAI for site 1 (a) and site 2 (b). Colors represent the median values of LAI for groups of pixels with similar ground elevation and distance to the shoreline. The vertical solid and dashed lines represent the average distance to
22
the shoreline of the first and second crestlines, respectively. The negative distance values denote points landward of the shoreline.
I also analyzed the relation between LAI and both crestline elevation and local elevation on the two sides of the foredune (Figure 2.8). LAI increases with crestline elevation on the seaward side of the dune (Figure 2.8 a), whereas it is mostly constant on the landward side. In addition, on both sides of the crestline, LAI increases with elevation (Figure 2.8 b). However, for a set topographic elevation, LAI values are consistently larger on the landward side of the foredune than on the seaward side. The reversed trend in biomass with respect to distance from the shoreline (increasing with distance on the seaward side, decreasing with distance on the landward side, Figure
2.6), and the different dependencies of biomass on elevation on the two sides of the crestline (Figure 2.8 b), clearly support my hypothesis about the central role of the crestline in defining and marking two widely different environments.
Figure 2.8: a) LAI values on the landward (triangles) and seaward side (circles) located within 15 m of the crestline are associated with the elevation of the closest crestline point. The median of LAI values falling within elevation intervals of 11 cm are subsequently computed and plotted on the y-axis. b) LAI values are associated
23
with the local elevation and medians are computed over local elevation intervals of 7 cm and plotted on the y-axis.
The ecomorphodynamic patterns inferred from biomass distributions are consistent with vegetation density distributions measured in the field (Figure 2.9). This general correspondence holds even though field observations and remote sensing acquisitions were performed in different years and at nearby but different locations.
Consistent with the pattern observed in the remotely-sensed biomass distribution (see
Figure 2.6), vegetation cover tends to be maximum at or near the crestline, and to decay to zero on the seaward side of the foredune, about 20-25 m away from the crestline.
Figure 2.9: Vegetation percent cover along 3 transects from the shoreline to the backdune area. Transects are spaced by 100 m. The shoreline is located to the right (~ 20 – 30 m from the dune crestline), and the landward side of the foredune area is
24
characterized by negative values of the distance to the first crestline. The red solid and dotted lines show the location of the first and second crestlines respectively.
2.4 Conclusions
The comparison of remotely-sensed NDVI and LAI distributions from Hog
Island dunes (VA) shows coherent spatial vegetation patterns. The general features of biomass distribution along transects perpendicular to dune lines are also confirmed by in-situ observations of vegetation cover. This coherence among optical-derived, Lidar- derived, and field-derived vegetation biomass distributions lends support to dune vegetation biomass mapping from airborne remote sensing at the whole-system scale. In particular, I have shown that Lidar remote sensing allows the quantitative evaluation of
LAI for sparse dune vegetation. Importantly, the Lidar-based analyses and processing methods developed here eliminate the difficulties and uncertainties related to co- registering different information sources and the need to perform cumbersome field observations of biomass (e.g., using a large number of small-scale photographs and their digital analysis, or destructive biomass sampling) and of topography (e.g., through necessarily sparse and cumbersome to update GPS measurements). Multiple-return and full waveform Lidar data have been thus far applied to the characterization of tall vegetation only (height ~10 m) [Armston et al., 2013], and currently do not allow the discrimination of the parts of the beam that have been reflected by the ground and by the canopy when vegetation is very short and sparse, such as in the case of coastal dune vegetation. The approach proposed here successfully performs this separation and can
25
be applied to a large set of existing single-return Lidar data opening the door to the full exploration of the space-time evolution of coastal eco-geomorphic patterns.
The spatially-explicit ecomorphodynamic patterns inferred here highlight the signatures of geomorphic controls on the spatial vegetation distribution and the role of the foredune crestline as a chief determinant of vegetation development. Vegetation cover gradients are radically different on the two sides of the crestline, consistent with a strong discontinuity between the conditions in the seaward side of the foredune area, exposed to flooding and high salinity, and in the landward side of the dune area, significantly more sheltered from environmental stressors. However, vegetation does not just passively adapt to topographic conditions, but significantly contributes to dune building, from the seaward position where a dune begins to the location of the crestline.
The joint ecomorphodynamic patterns identified and analyzed here through remote sensing are, in fact, the result of a two-way interaction between ecological and morphological dynamics: while the foredune crestline defines the space where the maximum biomass production occurs, vegetation growth controls sand deposition, builds the dune, and localizes the crestline. In this respect, the ecomorphodynamic patterns identified by my analyses bear significant implications for modelling. This is where the spatially-explicit characterizations of vegetation and topographic distributions inferred from Lidar remote sensing can contribute, e.g., by quantifying vegetation-airflow-sediment transport interactions [Raupach et al., 1993; Okin, 2008], to pave the way for physically-based dune ecomorphodynamic modelling.
26 3. The Spatial Variability of Organic Matter and Decomposition Processes at the Marsh Scale
3.1 Introduction
Salt marshes play a key role in preventing coastal erosion, creating and maintaining morphological and biological diversity, and providing refuge and nesting for wildlife [Barbier et al., 2011]. Salt marshes sequester carbon as they respond to the local Rate of Relative Sea Level Rise (RRSLR) through inorganic soil deposition and organic soil production [D'Alpaos and Marani, 2016; Kirwan and Mudd, 2012; Marani et al., 2007; 2010; Mitsch and Gosselink, 2007; Morris et al., 2002; Mudd et al., 2009]. Some estimates indicate that the rate of carbon burial per unit area in salt marshes is the highest of any biome, and up to 30 to 50 times larger than forest biomes [Mcleod et al.,
2011]. For example, Roner et al. [2016] estimated an average carbon accumulation rate of
132 g m−2 yr−1 for Venice marshes in Italy, and DeLaune and Pezeshki [2003] found a value of 183 g m−2 yr−1 for coastal marshes in the Gulf of Mexico. Chmura et al. [2003] examined data collected from different marshes along coasts of Atlantic and Pacific oceans, Indian Ocean, Mediterranean Ocean, and Gulf of Mexico, and determined an average carbon burial rate of 210 g m−2 yr−1 at the global scale. Later, Duarte et al. [2005] revised the estimate of the global average carbon burial rate to 151 g m−2 yr−1, by combining observations in Woodwell et al. [1973] and Chmura et al. [2003].
Salt marshes are vulnerable to the effects of climate change, including sea level rise and more frequent intense storms [Howarth and Hobbie, 1981], as well as to anthropogenic pressures, including the reduction of sediment supply to coastal 27
landscapes [Day et al., 2000; Syvitski et al., 2009]. Hence, marsh global extent may significantly decline in the future, and predicting the associated changes in the global carbon cycle requires a better understanding of the processes regulating carbon burial at the marsh scale [D'Alpaos and Marani, 2016; Kirwan et al., 2010; Marani et al., 2007;
Ratliff et al., 2015].
Two main processes contribute to a marsh accretion toward a possible equilibrium with mean RRSLR: inorganic deposition of exogenous sediments and organic soil production by plants. Suspended sediment from rivers or the ocean is deposited over the marsh platform by tidal flooding and leads to inorganic soil accretion, also mediated by vegetation that captures suspended particles [D'Alpaos et al., 2007; Da Lio et al., 2013; Danielsen et al., 2005; Feagin et al., 2009; Marani et al., 2013;
Mudd et al., 2010; Mudd et al., 2004]. Aboveground and belowground biomass as well as the organic matter imported by tidal currents are important sources of soil organic matter (SOM), however, belowground biomass accounts for a significant contribution to
SOM production and overall marsh vertical accretion [Blum, 1993; McCaffrey and
Thomson, 1980; Nyman et al., 2006; Turner et al., 2006]. The importance of the SOM contribution to accretion is mediated by the interplay of belowground biomass production and decomposition processes, the latter determining the fraction of the biomass produced that effectively contributes to soil formation and carbon burial
[Kirwan and Mudd, 2012; Mudd et al., 2009; Nyman et al., 1993; Rybczyk et al., 1998;
Turner et al., 2000]. Field observations and modelling results suggest biomass
28
production and decomposition rates to vary widely across marshes and across different sites. Soil aeration, fractional time of flooding (hydroperiod) and soil drainage are influenced by local topography and elevation, and are first-order controls of both biomass production and SOM decomposition rates [Armstrong et al., 1985; Chen et al.,
2016; Chmura and Hung, 2004; Hazelden and Boorman, 1999; Lemay, 2007; Morris et al.,
2016; Morris et al., 2002; Nyman and Delaune, 1991; Nyman et al., 1995]. In addition to elevation, salt marsh halophytic plants are also capable of modulating soil oxidation by producing a persistently aerated layer below the flooded surface which can alter decomposition processes [e.g., Boaga et al., 2014; Li et al., 2005; Marani et al., 2006;
Ursino et al., 2004; Xin et al., 2013]. Observations from the Venice Lagoon (Italy) show that, even though plant productivity decreases in the lower areas of a marsh located farther away from channel edges, the relative contribution of organic soil production to the overall vertical soil accretion tends to remain constant as the distance from the channel increases [Roner et al., 2016]. This observation suggests that the competing effects between biomass production and aeration/decomposition, modulated by soil elevation, approximately compensate each other in the lower portions of these marshes.
Testing this idea, and its possible more general applicability, requires simultaneous observations of decomposition rates and SOM content at the marsh scale. Such observations will also likely yield more, realistic quantitative representations of SOM dynamics, to improve currently simplistic model parameterizations [e.g., D'Alpaos et al.,
2012; Kirwan and Murray, 2007; Marani et al., 2013; Morris et al., 2016; Schile et al.,
29
2014]. Towards these goals, here I perform simultaneous observations of biomass productivity, decomposition, and SOM in coastal marshes located in North Carolina
(USA).
3.2 Methods
3.2.1 Study Area
The study sites are located in Wrightsville Beach and Masonboro Island (North
Carolina, USA, Figure 3.1). Marsh soils are typically composed of fine sand and silt in approximately even proportions [Croft et al., 2006] and by SOM in the 4-10% range.
Marsh halophytic vegetation species include Spartina alterniflora, Limonium carolinianum and Salicornia perennis. The area is microtidal, with an average tidal range of about 1.2 m, and is subject to an RRSLR of about 2 mm yr-1 [Zervas, 2004], with which marshes are assumed to be in equilibrium. I selected three transects, each with a length of 40 m, at sites 1 and 2 (Figure 3.1). Transect T1 at site 1 is perpendicular to a small creek, with likely modest or negligible transport of inorganic suspended sediment. Transects T2 and
T3 at site 2 (~20 m apart) are perpendicular to a larger channel, likely providing significant sediment input.
30
Figure 3.1: Study sites in Wrightsville Beach and Masonboro Island in North Carolina (a). The segments in b and c show the location of the three transects where soil sampling was performed. The yellow dot marks the position of the NOAA tide gauge station.
3.2.2 Core Collection and Topographic Survey
Using putty knives to avoid compaction, I collected top soil cores (8 cm deep) along three transects in July 2015 [Yousefi Lalimi and Marani, 2016]. Starting from the channel, five cores were collected with a 2.5 m spacing for the first 10 m. Sample spacing
31
was 5 m for distances between 10 m and 20 m, and 10 m between 20 m and 40 m from the channel. In total, I collected nine cores within each transect. The cores were stored in a cooler/fridge until they were analyzed, to minimize decomposition (freezing was avoided to prevent possible volume/porosity changes). I determined the elevation of coring sites using a theodolite with an accuracy of ±3 mm. To express the elevations relative to local mean sea level (MSL, with respect to the National Tidal Datum Epoch defined by NOAA), I recorded the maximum elevation reached by water at each study site at high water slack on the same day. I assumed that the attenuation of the tidal wave crest is negligible as it is a very long wave travelling through well-incised channels over a short distance from the ocean to the study sites. Hence, the maximum elevation reached by water levels, not simultaneous at different sites due to finite propagation celerity, is approximately the same across study sites, and at the nearest NOAA tide gauge station in Wrightsville Beach (station ID: 8658163, latitude: 34° 12.8' N, and longitude: 77° 47.2' W). Knowledge of this common elevation value allows one to link the local datums that had been used for topographic surveys and to express all elevations with respect to local MSL.
3.2.3 SOM Analysis
I applied the loss on ignition (LOI) method to quantify SOM in all the cores. To this end, I first cut the soil samples extracted from the cores in shape of a cuboid (with 8 cm depth from the marsh surface) to later easily measure the dimensions and compute the soil volume. After homogenizing the samples, I dried them at 60°C for 36 hours,
32
such that all the pore water was removed. After salt mass correction, assuming a nominal salinity of 35 ppm (see Dadey et al. [1992]), I determined dry bulk density (BD) by dividing the total mass of the dried sample by its known, undisturbed, volume. I then put samples in a muffle furnace where temperature was increased by 5 °C min-1 until
375°C, which was then kept constant for 16 hours [Frangipane et al., 2009; Roner et al.,
2016]. After this treatment, I again determined the weight of the samples, which represents the inorganic soil mass after combustion of the SOM. Hence, SOM fractional content was determined by subtracting the final sample mass from the mass before combustion and by dividing by the initial dry mass. Soil organic matter density (OMD, weight of organic matter divided by the sample volume) is then computed by multiplying SOM by BD. Using SOM fractional content, I could also estimate organic carbon (OC) in estuarine marsh soils, according to the empirical relation proposed by
Craft et al. [1991]:
OC (%) = 0.40 SOM + 0.0025 SOM2 (2.1)
Organic carbon density (OCD) was later computed by multiplying OC by BD.
3.2.4 Vegetation Cover Quantification
Vegetation coverage varied across the study sites and along transects (Figure
3.2). Transect T1 was mainly colonized by Spartina alterniflora, whereas in inner marsh areas Limonium carolinianum and Salicornia perennis were also present. Spartina alterniflora was the only species present along transects T2 and T3. To determine the percent cover of vegetation along the transects, a proxy for biomass productivity noting that only one
33
species was present in the majority of the locations studied, I took a photo of an area
(about 50 cm × 50 cm) around each core location using a nadir-looking camera. I then processed the photos to derive a binary map of “green” vs. “non-green” pixels based on
Hue-Saturation-Value (HSV) color representation, following the method described in detail in Yousefi Lalimi et al. [2017]. I then determined the percentage of “green” pixels in each photo as a measure of vegetation cover percentage. This information is used as a proxy for aboveground biomass, and, by assuming a constant partitioning between above- and belowground compartments, as a measure of belowground biomass. Note that this method is not used here to infer the absolute value of belowground biomass production, but just its spatial variation, by comparing vegetation cover at different locations within the study transects.
Figure 3.2: View of the marsh were sampling were conducted in study site 1 along transect T1 (a), and examples of digital photos (b and d) with respective vegetation/soil binary maps (c and e) used to determine vegetation cover. Panels b and d correspond to the locations with distances of 20 m and 30 m from the channel 34
respectively. Panels c and e estimate vegetation densities of about 30% and 10% respectively.
3.2.5 Decomposition Rate
I used the approach proposed by Keuskamp et al. [2013] to estimate ‘nominal’ decomposition rates and organic matter stabilization that can be compared across different sites along the transects. The use of a standard organic material makes the derived ‘nominal’ decomposition rates independent of the type of organic material used, such that they may be considered to only be a function of local edaphic and environmental conditions. In the method used, standard green tea bags (EAN 87 22700
05552 5) and rooibos bags (EAN 87 22700 18843 8) are buried in the soil together, and are retrieved after 90 days. The fractional weight, 푊(푡), of a bag content is assumed to change as a result of decomposition processes according to [Keuskamp et al., 2013]:
푊(푡) = 푎푒−푘푡 + (1 − 푎) (2.2) where a is the labile fraction, k is the decomposition rate (in 1/days), and t is the time (in days). The use of two different tea types of known properties makes it possible to estimate a and k without recourse to long data time series [Keuskamp et al., 2013].
According to Keuskamp et al. [2013], stabilization by which a labile material becomes recalcitrant [Prescott, 2010], is quantified through a stabilization factor, S:
푎 푆 = 1 − (2.3) 퐻 where H is the hydrolysable fraction, a known parameter for the standard tea materials.
The authors found a value of 0.842 ± 0.023 for hydrolysable fraction of green tea, Hg, and 35
a value of 0.552 ± 0.050 for hydrolysable fraction of rooibos tea, Hr, by experimenting in the laboratory. Keuskamp et al. [2013] also found that after 90 days, the labile fraction in green tea, mainly composed of leaves, is almost completely decomposed. Hence, the labile fraction ag of green tea can be found and consequently, using equation (2.3), the stabilization factor Sg for green tea can be determined. The authors argued that because stabilization is controlled by environmental conditions and is independent of the hydrolysable fraction, the value of Sg can be assumed to be equal to the stabilization factor Sr for rooibos (i.e., Sg = Sr). Finally, ar and kr for rooibos tea can be estimated using equations (2.3) and (2.2), respectively. In my experiments, I buried two replicas of each tea type at two depths: 3 cm and 8 cm below the marsh surface at all coring positions along the transects. I placed the bags along T1 and T2 in July 2015 and retrieved them after 90 days. In August 2016, I repeated the experiments along T2 and T3 and retrieved the bags after 118 days. To eliminate the effects of small amounts of inorganic sediments that could have entered the bags during the experiments, after drying the contents of retrieved bags and recording the weights, I burned the tea bag contents at 550°C for 3 hours as suggested in Keuskamp et al. [2013]. The material remaining after combustion includes inorganic sediments and salt that entered the bags after burial, as well as tea material that is not easily burned. This latter term was estimated by performing an LOI analysis on the contents of 10 green tea and 10 rooibos unused bags. The average weight of uncombusted material was subtracted from the weights of the material remaining after combustion of the retrieved bags, to estimate the weight of inorganic sediment/salt
36
that had entered each bag after burial. The weight of the extraneous mass was finally used to correct the post-LOI results and obtain an improved SOM content estimate.
3.2.6 Flooding Duration
I determined the fractional flooding duration along the transects for the 3 cm and 8 cm depths where tea bags were buried. Using water level data (available from NOAA: https://tidesandcurrents.noaa.gov/products.html) over the course of the decomposition rate experiments (90 days in 2015 and 118 days in 2016), I determined the amount of time in which the water level was higher than each of the two burial depths. Dividing the flooding time by the total duration of the experiments yields the value of the hydroperiod. This approach neglects the tidal level variations from the NOAA gauge due to the delay in flow and ebb currents through marsh vegetated surface, the delay required for the water table to adjust to changes in the tidal level forcing it, and possible effects related with plant evapotranspiration [Boaga et al., 2014; Ursino et al., 2004].
While this is a simplifying assumption, it is the most accurate estimate of the hydroperiod using the available information.
3.2.7 Data Analysis
The data analysis consists of three main parts, including the relationships based on
1) distance to the channel, 2) surface elevation, and 3) flooding duration as the main factors. I compared the relationships between different variables (such as SOM, OMD and vegetation cover, decomposition rate and stabilization factor) and the main factors
37
across all transects. The comparisons are based on the assumption of linear relationships between variables and the main factors by calculating R2 values for each relationship.
3.3 Results
The characteristics of core samples analyzed represented varied behaviors along three transects (Figure 3.3). Soil elevation increases with distance from the channel along transect T1 (Figure 3.3 a), whereas an opposite trend is found along transects T2 and T3
(Figure 3.3 b and c), where elevation is highest near the channel. Soil dry bulk density
(BD) tends to increase with distance from the channel in T1, however, changes of BD in space do not exhibit marked trends in T2 and T3 (Figure 3.3 d, e, and f). BD values in transect T1 (in the range of 0.6-1.2 g/cm3) are also significantly larger than those in transects T2 and T3 (in the range of 0.2-0.7 g/cm3). Fractional soil organic matter (SOM) shows different behaviors at different sites: it decreases away from the channel along T1, whereas it does not display well-defined trends along T2 and T3 (though fractional SOM minima appear to occur in areas close to the channel and maxima are found away from it, Figure 3.3 g, h and i). The SOM values found were used to derive the organic carbon density (OCD) which represented an average value of 0.023, 0.016, 0.011 g/cm3 for T1, T2 and T3 respectively, with a standard deviation of 0.004 g/cm3 (see Figure B.1 in appendix
B for more details on OCD values). Except for single isolated values, organic matter density (OMD) is greatest near the channel along T1 and smallest near the channel along
T2 and T3 (Figure 3.3 j, k, and l). The areas farther away from the creek in T1 and the areas close to the channel in T2 and T3, which have a higher elevation compared to the
38
other parts of the transects, exhibit smaller amounts of soil OMD. Additionally, vegetation cover exhibits the highest values on the levees in T2 and T3, where it decreases rapidly with distance by a factor of 6, to become approximately constant in the inner part of the marsh (Figure 3.3 n and o). T1, however, exhibits a different behavior as vegetation cover is roughly constant along the transect, even though a slightly lower density may be observed in higher and less frequently flooded areas away from the creek where other species are also present at the distances of 30 m and 40 m (Figure 3.3 m).
Soil dry BD increases with elevation in T1 but shows a mild decrease in T2 and
T3 (Figure 3.4 a, b, and c). As elevation rises, fractional SOM decreases along transect T1
(Figure 3.4 d), whereas a less significant decreasing trend is visible for SOM along T2
(coefficient of determination of 0.3 vs. 0.7) and no notable trend is detectable along T3
(Figure 3.4 e and f). Soil OMD represents clearer trends which decrease with elevation in all transects (Figure 3.4 g, h and i). Meanwhile, vegetation cover in T1 slightly decreases with elevation, though the trend is not significant (Figure 3.4 j). The opposite happens for T2 and T3, where vegetation cover does not exhibit significant trends at lower elevations, but shows sharp jumps at the highest locations (Figure 3.4 k and l).
Additionally, analyzing OMD as a function of flooding duration demonstrates that increasing flooding duration results in a steep and significant increase of OMD for all transects and locations (Figure 3.5).
39
Figure 3.3: Marsh surface elevation relative to mean sea level (a, b, and c); soil dry bulk density, BD, (d, e, and f); fractional soil organic matter, SOM, (g, h, and i); soil organic matter density, OMD, (j, k, and l); and vegetation cover (m, n, and o) for
40
transects T1, T2, and T3 as a function of distance from the channel edge. The marsh edge is located at the 0 m distance.
Figure 3.4: Soil dry bulk density, BD, (a, b and c), fractional soil organic matter, SOM, (d, e and f), soil organic matter density, OMD, (g, h, and i), and vegetation cover
41
(j, k, and l) vs. marsh surface elevation relative to MSL. Dashed lines represent the regression lines.
Figure 3.5: Organic matter density, OMD, vs. flooding duration along transects T1 (open circles), T2 (circle plus symbols) and T3 (closed circles). Solid lines represent the regression lines for each transect.
My observations represent variable relations for decomposition rate and stabilization factor at different sites (Figure 3.6). Along T1, decomposition rate shows increasing trends with elevation for both depths (Figure 3.6 a), whereas it does not exhibit systematic trends with elevation (Figure 3.6 b and c). Similarly, the stabilization factor increases with elevation at both depths in transect T1 (Figure 3.6 d), whereas no meaningful trend is exhibited for T2 or T3 (Figure 3.6 e and f). Interestingly, however, data show that deeper soil layers, experiencing longer flooding durations, tend to show
42
lower decomposition rates and higher stabilization factors at all transects (Figure 3.6 and
Figure 3.7).
Figure 3.6: Decomposition rate (a, b, and c) and stabilization factor (d, e, and f) values as a function of surface elevation for the two depths at which bags were buried (3 and 8 cm). The number of data points varies across transects and for different years, because not all buried bags could be retrieved at all sites. The dashed and solid lines represent the regression lines for 3 cm and 8 cm depths respectively. Open (3 cm depth) and closed (8 cm depth) symbols represent data collected in 2015 and 2016
43
(circles and triangles respectively). R2 values corresponding to 3 cm and 8 cm depths are shown in italic grey and black respectively.
Figure 3.7: Decomposition rate (a, b, and c) and stabilization factor (d, e, and f) with respect to flooding duration at two depths (3 and 8 cm) along transects T1, T2 and T3. The dashed and solid lines represent the regression lines for 3 cm and 8 cm depths respectively. Open (3 cm depth) and closed (8 cm depth) circles and triangles represent data collected in 2015 and 2016 respectively. R2 values corresponding to 3 cm and 8 cm depths are shown in italic grey and black respectively.
3.4 Discussion
The difference in elevation profiles observed at site 1 and site 2 can be due to different inorganic sediment amounts received through the channels. The increasing trend for soil elevation along T1 at site 1 can be a result of limited inorganic sediment being transported to the marsh through the creek (Figure 3.3 a). The opposite trend
44
along T2 and T3 at site 2 can be explained by noting that the relatively larger channel feeding the marsh likely carries a greater amount of inorganic sediment, sufficient to induce a levee-like morphology near the channel itself (Figure 3.3 b and c). In other words, the portion of T1 closest to the creek, and the areas in T2 and T3 furthest from the creek, are all characterized by low equilibrium elevations likely due to a lower availability of inorganic sediment in the water column [Marani et al., 2013]. On the contrary, the other ends of transects are characterized by higher elevations where smaller SOM fractional values tend to occur [Roner et al., 2016]. I interpret this reduction in SOM with elevation as the result of a greater decomposition rate caused by more microbial activities due to shorter inundation duration [Hazelden and Boorman, 1999;
Howarth and Hobbie, 1982]. The OCD values estimated are relatively lower than the ranges of OCD observed by Roner et al. [2016] (between 0.03 g/cm3 and 0.06 g/cm3) which have been measured for shallower core samples (3 cm deep vs. 8 cm in this study). This can be due to a decrease in belowground biomass with depth and thus, the shallower soil layers may encounter higher OCD than deeper layers [Keller et al., 2015;
Saunders et al., 2006]. Several other studies have found OCD values similar to my results [e.g., Hinson et al., 2017; Kelleway et al., 2016].
The distribution of fractional SOM (being the ratio of the weight of organic material to the total weight of a soil sample) depends on both the amount of organic material generated yearly in the soil and on the amount of inorganic sediment deposited yearly. Thus, analyzing soil OMD (which I recall to be the weight of organic matter
45
divided by the sample volume) should clarify the distribution of the amount of organic material alone. OMD measures represent smallest values at lower elevations close to the channel along T1 and highest elevations in the inner marsh area along T2 and T3 (Figure
3.3 j, k and l). This behavior is similar to what is found in Venice Lagoon marshes (Italy), where OMD is minimum in more elevated areas close to the channel, but remains approximately constant elsewhere [Roner et al., 2016]. My observations are compatible with the expected dependence resulting from elevation controls on flooding duration, soil aeration, and decomposition rate. The analysis of vegetation cover shows that it does not significantly vary with elevation (R2 = 0.17) at T1 (Figure 3.4 j). Additionally, the levees found close to the channel at T2 and T3 are covered by the highest amounts of vegetation density (Figure 3.4 k and l). This behavior is observed in several tidal environments [Marani et al., 2013] and is due to a reduced water logging stress [Ursino et al., 2004]. Overall, soil OMD shows a significant decrease with elevation, which cannot be explained in terms of observed vegetation cover trends. These observations support the hypothesis that at lower locations, where inundation is more prolonged, and the soil is less oxidized, a greater proportion of the fresh organic matter produced may be buried as SOM and results in a greater OMD [Hansen et al., 2017]. This hypothesis is further supported by the analysis of OMD as a function of flooding duration, a more direct proxy of soil aeration. In other words, the significant increase of OMD with flooding duration (Figure 3.5) is in agreement with the observed decreasing OMD values with elevation (Figure 3.4 g, h and i) and confirms once more the hypothesis that
46
the oxygen availability is the driving factor in the accumulation of organic matter in marsh soils.
The value of hydrolysable fraction of dry mass of Spartina alterniflora, the dominant species in my study sites, containing 10 g leaves and 5–7 g fine roots is 0.604
[Harmon et al., 2009] which is within the ranges of hydrolysable fractions of green and rooibos tea (Hg = 0.842 ± 0.023 and Hr = 0.552 ± 0.050) used for litter decomposition experiments. In fact, the hydrolysable fraction of a dry mixture containing 10 g green tea and 6 g rooibos tea will be around 0.660 which is comparable to the hydrolysable fraction of Spartina. Thus, green tea and rooibos tea can be considered as adequate representatives of marsh vegetation. Here, the increasing trend for decomposition rate values of rooibos tea with elevation along T1 for both depths (Figure 3.6 a), consistent with the hypothesis that elevation, by increasing soil aeration, is a primary control of decomposition rate and a main determinant of SOM. However, such inference is not conclusively corroborated by data from transect T2 and T3 (Figure 3.6 b and c). In these transects, decomposition rates do not represent systematic trends in space or with respect to elevation. Hence, decomposition rates from T2 and T3 do not offer an explanation as to how the higher vegetation cover at higher elevation may be compensated for to yield relatively flat SOM profiles. Various studies suggested that the relation between decomposition rate and elevation can be highly varied [Hackney, 1987;
Halupa and Howes, 1995; Janousek et al., 2017; White and Trapani, 1982]. The findings show that elevation can be positively linked to decomposition rate at site 1 but it does
47
not affect the decomposition rate at site 2. The variations in decomposition rate can be partially due to grain size distribution which can consequently affect organic matter content [Gibeault et al., 2016]. In addition, other factors can alter decomposition rate such as vegetation composition, microbial community composition, and the physical and chemical micro-environment conditions [Janousek et al., 2017]. The stabilization factors also represent large variations at different transects (Figure 3.6 d, e and f). While elevation and flooding duration seem not to offer any reasonable explanation for the variations seen in decomposition rates and stabilization factors across all sites, the deeper soil layers at most of the sample locations along each transects are characterized by smaller decomposition rates and larger stabilization factors rather than the shallower layers (Figure 3.8). This finding reinforces the notion that soil aeration is the main determinant of decompositions processes and, hence, of SOM. Elevation in T2 and T3 is generally lower than in T1, and flooding duration is correspondingly about 10-15% longer. Despite these differences, the observed ranges of decomposition rate and stabilization factor are quite similar. This observation is suggestive of a possible saturation of the effect of increased flooding duration on decomposition processes, such that decomposition rate and stabilization factor, everything else being equal, become stabilized if elevation is sufficiently low.
The average value of the decomposition rate at 8 cm depth across all transects is about 0.010 d-1. This is equivalent to a decay of 95% of the labile organic matter in about
300 days, based on the exponential relation in equation (2.2). On the other hand, a marsh
48
surface accretion in equilibrium with a nominal RRSLR of 2 mm yr-1 requires 40 years before the surface sediment is buried to a depth of 8 cm. Hence, a comparison between the time scale needed for decomposition (order of months) and for sediment burial
(order of tens of years) suggests that the labile fraction of SOM is completely decomposed before it can reach deep soil layers and become stabilized. With variations of accretion rate in space or with time, the age of soil layers can vary spatially which leads to changes in decomposition properties. My analysis supports that, for the purpose of marsh accretion and morphodynamic evolution, the key role in determining SOM content is played by the initial refractory fraction of the fresh organic material, and by the stabilization factor. Stabilization, in fact, reduces the amount of organic matter subject to decomposition processes, thereby potentially increasing the eventual SOM amounts.
The values observed for decomposition rate and stabilization factor for marshes in my study are comparable with data from other environments (Figure 3.8) [Keuskamp et al., 2013]. Differences are also notable. In deserts, for example, where moisture is very low, organic matter decays at about the same rate found for marshes, while the stabilization factor is increased. Overall, a visual inspection of Figure 3.8 suggests that marshes exhibit some of the lowest decomposition rates among all biomes (e.g., unlike fringe mangroves in tropical and subtropical areas where high temperatures favor decomposition processes). The labile fraction of SOM in marshes tends to be less
49
stabilized by local edaphic conditions compared to other ecosystems, including peat
[Currey et al., 2010; Wang et al., 2015].
Figure 3.8: The relation between decomposition rate and stabilization factor along transects T1 (circles), T2 (triangles) and T3 (squares) at two different depths, 3 cm (open symbols) and 8 cm (closed symbols). The symbols represent the average values and the error bars show the 50th percentile of data distribution along x and y axes. The diamond-plus markers represent data for different ecosystems according to Keuskamp et al. [2013].
3.5 Conclusions
The results suggest that soil dry BD, SOM and biomass productivity, here represented through vegetation cover, vary with distance from the marsh edge and
50
elevation at the marsh scale. As soil elevation increases, SOM exhibits large fluctuations beclouding possible trends, but OMD is significantly reduced. This finding may be ascribed to the greater aeration at higher soils, where the inundation time is shorter. I find that plant biomass production does not seem to define the eventual (after burial and decay) SOM content and OMD. Both parameters, in fact, show greater covariation with elevation and hydroperiod than with vegetation cover. The quantification of the spatial dependence of decomposition processes overall corroborates the idea that this covariation is the sign that soil oxidation tends to overwhelm initial difference in biomass productivity and to largely determine the spatial distribution of SOM. The results also suggest that decomposition rate and stabilization of the labile part of belowground biomass vary with position within a marsh and depth below the soil surface. Decomposition rate is greater in the upper soil layers and stabilization is greater deeper away from the surface, highlighting the role of elevation and flooding duration on decomposition rate and stabilization factor within a marsh. Even though decomposition rates vary significantly at the marsh scale, observed values are sufficiently large that the labile component of the standardized root material (rooibos) is fully decomposed over the course of 2-3 years within the root zone. This result suggests that only the refractory part of the belowground biomass contributes to soil accretion and emphasizes the role of the stabilization factor in modulating marsh vertical accretion: a higher stabilization factor increases the effectively refractory fraction, thereby contributing to increase SOM accumulation. Overall, the findings provide the
51
basis for refining representations of the SOM accumulation process to inform the next generation of spatially-explicit bio-morphological models of salt marsh evolution [e.g.,
D'Alpaos and Marani, 2016; Fagherazzi et al., 2012; Kirwan and Murray, 2007; Marani et al., 2013; Mudd et al., 2004]. These new models will be a key tool to evaluate impending rates of marsh loss and the associated carbon release.
52 4. Watershed and Estuarine Controls on Salt Marsh Distributions and Resilience
4.1 Introduction
Estuaries provide a wide range of ecosystem services including protection of coastal areas, nutrient cycling, water filtration, nurseries for fish and crustaceans, and carbon sequestration [Barbier et al., 2011; Mcleod et al., 2011; Zedler and Kercher, 2005].
These coastal wetlands are threatened by climate change through sea level rise (SLR) with a projected increase of up to 16 mm/yr for the last two decades of the 21st century
[Church et al., 2013; Craft et al., 2009; Morris et al., 2002]. Rivers feeding estuaries can also be highly impacted by human activities within watersheds and coastal environments such as dam construction and flow diversion, altering river discharge and restricting the sediment supply delivered to salt marshes [Kennish, 2001; Lotze et al., 2006]. Studies show that salt marshes are resilient to environmental change due to a positive feedback between vegetation and deposition dynamics, maintaining their elevations and preventing drowning [Fagherazzi et al., 2012; French and Stoddart, 1992; Kirwan and
Murray, 2007; Marani et al., 2007]. However, environmental changes can lead to marsh collapse, retreat or migration under some circumstances such as elevated rates of SLR and sudden reductions in sediment supply [D'Alpaos, 2011; Kirwan et al., 2016; Marani et al., 2010; Morris et al., 2002].
Marani et al. [2007] find that three equilibrium states exist for tidal platforms
(subtidal zones, tidal flats and salt marshes) by solving the sediment mass conservation at equilibrium regulated by sediment dynamics, biotic processes and SLR. The study of 53
the coupled evolution of tidal flat and marsh platforms has been recently expanding, suggesting that tidal flats can regulate marsh erosional and depositional dynamics
[Allen, 1990; D'Alpaos et al., 2005; Fagherazzi et al., 2006; Mariotti and Fagherazzi, 2013].
The sediment eroded from tidal flats due to wind wave dynamics can be transported to marshes by tides and deposited on the surface [Carniello et al., 2005; Tambroni and
Seminara, 2006]. In addition, wind waves formed in tidal flats can also govern the migration of the margin between tidal flats and marshes by determining the relative significance of margin erosion compared to margin accretion [Lawson et al., 2007;
Leonardi and Fagherazzi, 2015]. The erosion of tidal flat beds and marsh margins is indeed dictated by the morphology of the estuary through fetch and water depth, which control the wave power needed for erosion [Defina et al., 2007; Marani et al., 2011].
Mariotti and Fagherazzi [2013] suggest that a critical width exists for tidal flats above which irreversible marsh retreat occurs. In addition, the geometric properties of the tidal flat and estuary depend on both river discharge and tidal dynamics [Gisen and Savenije,
2015; Savenije, 2015]. Noting that the morphology of tidal flats is subject to change due to environmental disturbances on the local scale and watershed and estuarine scales
(such as variations in river discharge or sediment concentration, storm surges, and changes in SLR rates), it is essential to understand the full range of mechanisms that control salt marsh evolution [FitzGerald et al., 2008]. Despite numerous field and modeling studies on morphodynamic processes of coastal platforms, only few studies
(and those only qualitatively) discuss the upland and estuarine controls of marsh
54
existence [Allen, 2000; Day et al., 2000; Gunnell et al., 2013; Redfield, 1965; Braswell &
Heffernan, in review]. Consequently, we lack an understanding of how these large-scale controls affect the tidal flat and marsh erosional and depositional dynamics. Especially, it is not clear how the mixture of ocean and riverine sediment inputs can influence marsh distribution within estuaries. Here, I develop a large-scale model of tidal flat and salt marsh evolution and explore how watershed extent and characteristics, mediated by riverine liquid/solid discharge, determine jointly with oceanic inputs and tidal and wave forcings, the relative extent of marsh vs. tidal flat area and possible transitions between competing equilibrium states. By understanding the full range of mechanisms controlling marsh persistence and degradation, the effectiveness of restoration policies can be increased.
4.2 Model Structure
I consider a simplified symmetric geometry which includes a salt marsh platform, a tidal flat, and a main river channel (Figure 4.1). The model incorporates vegetation dynamics, hydrodynamics and sediment dynamics. I assume that the marsh has a fixed landward boundary, but the seaward margin of the marsh can erode or accrete. My approach determines the conditions under which the marsh platform expands (in case of accretion on the margin between marsh and tidal flat exceeding the erosion), or contracts (in case of margin erosion exceeding deposition). These conditions are controlled by the fetch length which regulates the fetch-limited wave energy and the erosional dynamics on the margin.
55
Figure 4.1: Scheme of a simplified symmetric estuary, (a) top view and (b) side view, containing a salt marsh platform, a tidal flat and a main river channel. The red dashed lines mark the control volume considered for the model. Shapes are not drawn to scale.
In my simulations, I assume that the intertidal surfaces consist of fine sediments which can be transported throughout the system resulting in a well-mixed and constant suspended sediment concentration in the water column within the whole system [Allen,
1994; D'Alpaos et al., 2006]. Here, I describe the state of the system by the average suspended sediment concentration, 퐶푠, and three geomorphic variables: tidal flat width,
푏푓, tidal flat depth, 푑푓, and marsh depth, 푑푚. Accordingly, the mass conservation equation incorporating all sediment inputs and outputs to/from the control volume:
휕푀 푠푒푑 = 𝑖푛 + 𝑖푛 + 𝑖푛 − 표푢푡 − 표푢푡 + 푛푒푡 (3.1) 휕푡 푟푖푣푒푟 표푐푒푎푛 푒푟표푠푖표푛 푑푒푝표푖푠푡푖표푛 푒푥푝표푟푡 푚푎푟푔푖푛
The left-hand side of equation (3.1) is the rate of change of sediment mass, 푀푠푒푑 , in the control volume:
푀푠푒푑 = 퐶푆(푏푓푑푓 + 푏푚푑푚)푙푒 (3.2)
56
where 푏푚 is the marsh width and 푙푒 is the length of estuary. The rate of sediment input from the river is computed as:
𝑖푛푟푖푣푒푟 = 퐶푅푄푅 (3.3) where 퐶푅 and 푄푅 are river sediment concentration and river discharge respectively. The rate of sediment input from the ocean is controlled by ocean tidal prism, 푉푂, and ocean sediment concentration, 퐶푂, and is computed as:
푉 𝑖푛 = 퐶 푂 (3.4) 표푐푒푎푛 푂 푇 where 푇 is the tidal period. Due to the presence of a river and its contribution to filling the basin, 푉푂 is limited to the available volume in the system that can be filled with ocean water during one tidal cycle:
푉푂 = (푏푓푑푓 + 푏푚푑푚)푙푒 − 푄푅푇 (3.5)
I assume that erosion processes are controlled only by wind wave action and the influence of tidal currents is negligible [Christiansen et al., 2000; D'Alpaos et al., 2007]. In addition, due to the effective wind wave dissipation by vegetation, marsh bed erosion can be neglected [D'Alpaos et al., 2007; Moller et al., 1999]. Therefore, bed erosion can happen only on the tidal flat platform and is governed by the wave-induced bed shear stress, 휏 (see appendix C). The sediments removed from the tidal flat through erosion processes are added back to the system as a source of sediment input. I estimate the rate of sediment input due to bed erosion, 𝑖푛푒푟표푠푖표푛, by:
휏−휏푐 𝑖푛푒푟표푠푖표푛 = 푡푠퐸0 푏푓푙푒 (3.6) 휏푐
57
where 퐸0 is a bed erosion coefficient and 휏푐 is the critical shear stress (휏푐 = 0.3 Pa, see the complete list of model parameters in Table C.1 in appendix C). 푡푠 accounts for the fraction of time within a time step (set to one tidal period) when the tidal flat is submerged and can be eroded, and is computed as:
1 1 −1 퐻−푑푓 − 푠𝑖푛 ( ) 𝑖푓 푑푓 < 2퐻 푡푠 = { 2 휋 퐻 (3.7) 1 𝑖푓 푑푓 ≥ 2퐻 where 퐻 is the tidal amplitude. Deposition, however, can occur on both tidal flat and marsh surfaces. The rate of deposition on the tidal flat, 퐴푓, depends on the average concentration of the system and is limited to the total available sediment in the water column. Hence, it is computed as:
1 퐴 = min (푡 퐶 푏 휔 푙 , 퐶 푏 푑 푙 ) (3.8) 푓 푠 푠 푓 푠 푒 푇 푠 푓 푓 푒 where 휔푠 is the settling velocity. I assume that due to the presence of vegetation all sediment in the water column above the marsh deposits on the marsh surface. Therefore, the rate of deposition on the marsh, 퐴푚, is:
1 퐴 = 퐶 푏 푑 푙 (3.9) 푚 푇 푠 푚 푚 푒
The sum of these two deposition terms gives the rate of total deposition, 표푢푡푑푒푝표푖푠푡푖표푛:
표푢푡푑푒푝표푖푠푡푖표푛 = 퐴푓 + 퐴푚 (3.10)
The rate of sediment export, 표푢푡푒푥푝표푟푡, from the control volume with outgoing ebb currents depends on the sediment concentration and the size of the tidal flat, and is computed as:
1 표푢푡 = 퐶 푏 min(푑 , 2퐻) 푙 (3.11) 푒푥푝표푟푡 푇 푠 푓 푓 푒
58
The last term in equation (3.1), 푛푒푡 푚푎푟푔푖푛, reflects the net rate of sediment that can be introduced to the system due to the erosion of the margin between the marsh and tidal flat (hereafter referred to as ‘the margin’), or can be removed from the system due to deposition on the margin and marsh expansion. This term is computed as:
̇ 푛푒푡 푚푎푟푔푖푛 = 푏푓(푑푓 − 푑푚)푙푒휌푠 (3.12)
̇ where 푏푓 is the change of margin position in time and 휌푠 is the sediment bulk density. To model the dynamics of the margin, I use the approach proposed by Mariotti and Carr
̇ [2014] and describe 푏푓 based on the competition between margin erosion, 퐵푒, and progradation, 퐵푎:
̇ 푏푓 = 퐵푒 − 퐵푎 (3.13)
Margin erosion is controlled by wave power density at the margin and is computed as:
1 퐵 = 푘 훾푐 퐻 2 (3.14) 푒 푒 16 푔 푤 where 푘푒 is a margin erodibility coefficient, 훾 is the specific weight of water, 푐푔 is the wave group velocity, and 퐻푤 is the wave significant height on the tidal flat (see appendix C). Margin progradation is dictated by sediment accretion on a gently sloping surface and is computed by:
1 퐵푎 = 푘푎휔푠퐶푠 (3.15) 휌푠 where 푘푎 is a margin accretion coefficient which describes the geometry of the marsh boundary [Mariotti and Fagherazzi, 2013]. Mariotti and Carr [2014] discuss that estimations of 푘푎 and 푘푒 cannot be independently performed, however, the selection of
푚 푊 −1 푘 = 2 (equivalent to a slope of 26°) corresponds to 푘 = 0.16 ( ) which provides 푎 푒 푦푟 푚 59
the best fit for margin migration data from marshes along the US Atlantic coast reported in Mariotti and Fagherazzi [2013]. For simplicity, I consider a wind regime blowing perpendicular to the margin with a constant velocity of 6 m/s [Mariotti and Carr, 2014].
The morphological evolution of the bed topography of the tidal flat platform is governed by local erosion and deposition processes and varies with SLR rate, 푅:
̇ 1 1 푑푓 퐸0 휏−휏푐 푑푓 = −min (푡푠 퐶푠휔푠, 퐶푠 ) + 푡푠 + 푅 (3.16) 휌푠 휌푠 푇 휌푠 휏푐
̇ where 푑푓 is the rate of change of tidal flat depth. The first and second terms on the right- hand side of equation (3.16) incorporate sediment deposition and erosion processes,
̇ respectively. The rate of change of marsh depth, 푑푚, however, is mainly governed by sediment deposition processes from inorganic and organic sediment, and SLR rate:
̇ 1 푑푚 푑푚 = − 퐶푠 − 푂 + 푅 (3.17) 휌푠 푇 where 푂 is the rate of organic matter production. I estimate 푂 from:
푂 = 푘퐵 퐵 (3.18) where 푘퐵 incorporates vegetation characteristics and the density of the organic soil produced, and 퐵 is the biomass production rate. I model 퐵 using the approach proposed by Marani et al. [2010] which describes the production rate based on the vegetation fractional cover and carrying capacity of the system, 퐵푚푎푥. Biomass is assumed to be in equilibrium with marsh platform elevation and thus the steady state solution for
푚 biomass is estimated by 퐵 (1 − ), where 푚 and 푟 are mortality and reproduction 푚푎푥 푟 rates, respectively. 푚 and 푟 are functions of the elevation of the marsh platform relative to mean sea level (MSL), 푧, and are computed as:
60
푧 푚 = 0.5 (3.19) 퐻
푧 푟 = −0.5 + 1 (3.20) 퐻 assuming that the common halophytic salt marsh vegetation, Spartina alterniflora, is the dominant species colonizing the marsh with its highest density at 푧 = 0 and lowest density at 푧 = 퐻 due to adaptation to hypoxic conditions [Morris et al., 2002]. The assumption of steady state biomass is valid only if 푧 ≥ 0, otherwise, if the marsh surface is below MSL, the vegetation cannot grow (i.e. if 푧 < 0, 퐵 = 0) [Silvestri et al., 2005].
The set of nonlinear dynamical equations describing lateral margin dynamics, the vertical marsh and tidal flat dynamics, and the concentration in the well-mixed water column describe the evolution of the system and are here solved numerically, both in steady and unsteady state (equations 3.1, 3.13, 3.16, and 3.17, see appendix C for more detailed description).
4.3 Results
The numerical investigations of the morphodynamic evolution of marsh and tidal flat show that the equilibrium solution for the position of the tidal flat/marsh margin is unstable; any perturbation leads to either a tidal flat-dominated estuary or to a marsh-dominated one. This confirms the previous finding by Mariotti and Fagherazzi
[2013] that marsh and tidal flat cannot coexist at a stable state of "horizontal" erosional/depositional processes. Even though I consider both riverine and ocean sediment input in this work, no stable solution is found. This is due to a positive feedback between the erosive wave energy and margin erosional/depositional dynamics
61
[Mariotti and Fagherazzi, 2013]. In the case of a small amount of net erosion at the margin, tidal flat width and thus the fetch increase which result in a larger wave energy, leading to more margin erosion until the marsh ultimately disappears. The opposite happens for the case of a small amount of net deposition at the margin, where the erosive energy decreases due to a reduction in fetch, leading to expansion of the marsh, ultimately filling the basin. Estuarine geometry is dictated by large-scale morphology
(e.g., the origin of the area that is now a coastal area, fluvial inputs, tidal regime). This morphology sets constraints to the size of different parts of the estuary, which constitute the initial conditions for the system and thus determine the subsequent dynamics. The sets of initial conditions of the system dynamics that lead to estuaries filled with either marsh, or tidal flat define the “basins of attractions”. Therefore, two regions in the multidimensional space of morphodynamic forcings can exist: basin of attraction of the marsh-dominated estuaries, and basin of attraction of the tidal flat-dominated estuaries.
I consider a reference set of forcings to illustrate the possible dynamical mechanisms controlling the basins of attractions. Different forcings may have different impacts on the critical fetch, dividing the tidal flat- and marsh-dominance (Figure 4.2).
The critical fetch lines correspond to the unstable equilibrium solutions of the dynamical system as a result of lateral erosional and depositional dynamics on the margin. Basins with a fetch smaller than the critical fetch value (lying in the grey area) will be eventually filled with marsh due to marsh margin progradation whereas basins with a larger fetch than critical fetch (lying in the white area) will be eventually filled with tidal
62
flats due to margin erosion (Figure 4.2). Such behavior, for instance, exists for marshes of
Virginia Coast Reserve where the data analyzed from aerial photographs show that marshes with a fetch smaller than 1 km expand whereas they retreat with a larger fetch
[Mariotti and Fagherazzi, 2013]. The basin of attraction of a marsh-dominated system grows with increasing river and ocean concentrations, and hence, the chances that initial conditions lead to estuaries filled with marsh increase (Figure 4.2 a, c). With larger external sediment input to the system and the corresponding increase in accretion at the margin, a larger fetch is needed to produce the greater wave energy necessary to maintain a balance between margin accretion and erosion. Therefore, the unstable equilibrium is marked by a larger critical length for increased sediment concentrations.
A similar behavior exists when river discharge is increased (Figure 4.2 b) because with increasing river discharge (which has a higher sediment concentration than the ocean in the reference conditions: 퐶푅 = 50 mg/l whereas 퐶푂 = 10 mg/l), less ocean water enters the system, resulting in a higher amount of total external sediment input. Here, I highlight the importance of the ocean to river discharge ratio, 푄푟푎푡푖표, which is the ratio of the ocean tidal prism entering the control volume during one tidal period to river discharge:
푉푂 푄푟푎푡푖표 = (3.21) 푇푄푅 on dynamics of the system. The sediment mass import to the system depends on both ocean and river inputs of water and sediment concentrations. 푄푟푎푡푖표, in fact, determines the relative amount of water entering the system from ocean and river and
63
consequently, the relative significance of their concentrations that contribute to the average concentration of the system. Therefore, 푄푟푎푡푖표 can regulate the final yield of sediment import to the system rather than ocean and river inputs alone.
Figure 4.2: The relationships between fetch length and river sediment concentration (a), river discharge (b), ocean sediment concertation (c) and tidal range (d) determining marsh- (grey area) vs. tidal flat- (white area) dominance. The lines bordering the two basins of attraction correspond to the critical fetch length. The
64
reference river concentration is 50 mg/l, river discharge is 50 m3/s, ocean concentration is 10 mg/l, and tidal range is 1.4 m. The sea level rise rate is 6 mm/yr.
Different mechanisms and forcings regulate the migration rate of the margin and lead to different slopes of the critical fetch lines (Figure 4.2 a, b and c). For instance, the fetch equal to the critical fetch for systems with river concentrations larger than 50 mg/l and the associated tidal flat depth are sufficiently large to cause erosion on tidal flat bed
(휏 > 휏푐 under such initial conditions of the system; Figure 4.2 a). Thus, the eroded sediment from the tidal flat can contribute to the concentration of the system, altering the dynamics of the system. However, the shear stress becomes too weak to cause any erosion on the tidal flat for fetch equal to the critical fetch and the associated tidal flat depth corresponding to river concentrations less than 50 mg/l (휏 ≤ 휏푐). Surprisingly, I find that tidal range does not seem to change the erosional and depositional dynamics of the margin (Figure 4.2 d). The behavior of the critical fetch tends to remain constant as the tidal range increases. This behavior can be explained by considering that the equilibrium depths of tidal flat and marsh platforms are almost constant for any tidal range, and depths quickly adjust to the equilibrium values [Mariotti and Fagherazzi,
2013]. Increasing the tidal range only moves the platforms to higher elevations, without changing the margin dynamics, thus, the critical fetch line stays quite constant.
In addition to the lateral erosion and progradation of the margin, the results show that the vertical dynamics on the tidal flat can also affect the transition from tidal flat- to marsh-dominance (Figure 4.3). Vertical dynamics become the controlling process when accretion on the tidal flat surface occurs sufficiently rapidly for the surface to
65
emerge above MSL, making it favorable for vegetation to establish [Silvestri et al., 2005], and consequently, tidal flat conversion to a new marsh occurs. Therefore, the results suggest that the basin of attraction for the marsh-dominated equilibrium (hatched area in Figure 4.3) is larger than that implied by existing theories (grey area in Figure 4.3), which only consider horizontal erosion/deposition dynamics.
66
Figure 4.3: The relationships between fetch length and river sediment concentration (a), river discharge (b), ocean sediment concertation (c) and tidal range (d) determining marsh- (grey and hatched areas) vs. tidal flat- (white area) dominance. The border of the hatched areas corresponds to the critical fetch found including both vertical and lateral dynamics, whereas the border of grey area is linked only to the lateral erosional/depositional dynamics on the marsh/tidal flat margin. The reference river concentration is 50 mg/l, river discharge is 50 m3/s, ocean concentration is 10 mg/l, and tidal range is 1.4 m. The sea level rise rate is 6 mm/yr.
67
The effect of vertical dynamics on the size of the basins of attraction in the tidal range and fetch plane is particularly large (Figure 4.3 d). The reason for the large addition to the marsh-dominance basin across tidal ranges is that with increasing tidal range, the tidal flat platform accretes in order to maintain the equilibrium depth and is more likely to pass MSL. This promotes tidal flat conversion to marsh and significantly increases marsh-dominance. The equilibrium depth of tidal flats lying in the tidal flat- dominated basin is dictated by the fetch length (the tidal flat accretes/erodes until it reaches an elevation where erosion balances deposition) and increases with increasing fetch (Figure 4.3 d). In order to lie in these basins, the initial conditions of the system should be such that the fetch is sufficiently large to create a large bed shear stress. This large shear stress is needed to prevent tidal flat conversion to marsh due to vertical dynamics. However, for tidal ranges greater than a threshold value (about 3.3 m in the setup of the model examined here), the associated depth of tidal flat becomes too large to produce a bed shear stress exceeding 휏푐 and causing erosion. In fact, a fetch larger than what the basin can sustain is needed to compensate for the depth and cause erosion. Therefore, there is a sudden jump of the critical fetch to its highest possible value limited by the basin size (10 km). In other words, no tidal flat-dominated systems can exist for tidal ranges larger than this threshold value, hence, leading to full expansion of the marsh-dominance basin.
The increase in the extent of marsh-dominance due to vertical dynamics can be small or even negligible if the river concentration and discharge are small, due to the
68
lack of sufficient sediment for vertical dynamics to happen (e.g., 퐶푅 = 5 mg/l and 푄푅 = 5 m3/s, Figure 4.4). It is also important to note that when the river concentration is lower than the ocean concentration (here 퐶푅 = 5 mg/l and 퐶푂 = 10 mg/l), an increase in river discharge reduces the basin of attraction of marsh-dominance by changing the 푄푟푎푡푖표 because the tidal prism entering the bay during a tidal cycle decreases, resulting in a reduction in the average sediment concentration of the system (Figure 4.4 b).
69
Figure 4.4: The relationships between fetch length and river sediment concentration (a), river discharge (b), ocean sediment concertation (c) and tidal range (d) determining marsh- (grey and hatched areas) vs. tidal flat- (white area) dominance. The border of the hatched areas corresponds to the critical fetch found including both vertical and lateral dynamics, whereas the border of grey area is linked only to the lateral erosional/depositional dynamics on the marsh/tidal flat margin. The reference river concentration is 5 mg/l, river discharge is 5 m3/s, ocean concentration is 10 mg/l, and tidal range is 1.4 m. The sea level rise rate is 6 mm/yr.
70
Similarly, SLR can also affect the basins of attractions by altering the dynamics of the intertidal platforms (Figure 4.5; for the effects of SLR rate on a system with a smaller river see Figure C.1 in appendix C). The critical fetch value is here associated with both lateral and vertical dynamics. An elevated SLR rate results in a larger reduction in the marsh-dominance basin. This is because with higher SLR rates, more sediment is needed to fill the space created by SLR in order to prevent drowning, which results in a reduction in the average concentration of the system. For example, in the reference
3 conditions when 푄푅 = 50 m /s, 퐶푅 = 50 mg/l, and 퐶푂 = 10 mg/l, the average concentration of the system is halved (퐶푆 = 7 mg/l) if SLR rate increases from 2 mm/yr to
8 mm/yr and tidal platforms deepen. Therefore, with a reduction in 퐶푆, the critical fetch decreases and marsh-dominance basin shrinks until sufficiently large SLR rates (e.g., larger than 10 mm/yr, Figure 4.5 a) can lead to marsh drowning for some system conditions, indicating that the accretion is not sufficient for the marsh to keep up with
SLR.
71
Figure 4.5: The relationships between critical fetch length associated with both vertical and lateral dynamics and river sediment concentration (a), river discharge (b), ocean sediment concertation (c) and tidal range (d) determining marsh- vs. tidal flat- dominance basins. The lines show the critical fetch lines found from both vertical and lateral erosional/depositional dynamics for different rates of SLR. The reference river concentration is 50 mg/l, river discharge is 50 m3/s, ocean concentration is 10 mg/l, and tidal range is 1.4 m.
72
The volume of the system is limited by the tidal range and the size of the basin.
In the reference conditions, if the tidal range is smaller than 1.3 m and the SLR rate is less than 2 mm/yr, a river with a discharge of 50 m3/s can dominate the systems with a fetch equal to the critical fetch and with the associated platform depths, during one tidal period. In other words, under these circumstances the volume of water entering the system from the river is greater than that of ocean (푄푟푎푡푖표 < 1), and hence, there is a net sediment export from the system with increasing tidal range up to 1.3 m (Figure 4.5 d).
Such behavior occurs because the sediment concentration of ocean water is 10 mg/l but the ebb current has a higher concentration due to mixing with river water which has a concentration of 퐶푅 = 50 mg/l. For example, in the reference conditions, if the tidal range is equal to 1 m and 퐶푂 = 10 mg/l, the ebb current has a concertation of 18 mg/l. While increasing the tidal range results in an increase in the amount of sediment mass imported by the ocean, the amount of sediment mass exported from the system (due to the excess of the tidal range) is higher because the ebb has a higher concentration. This net sediment export from the system results in a reduction in critical fetch, which explains the existence of the minimum point for critical fetch at tidal range of 1.3 m
(Figure 4.5 d). This behavior reflects the importance of 푄푟푎푡푖표 in governing the marsh- and tidal flat-dominance basins.
The geometric properties of the estuary also affect the critical fetch length (Figure
4.6), however these properties have much smaller effects on systems with small river
3 input (e.g., 퐶푅 = 5 mg/l and 푄푅 = 5 m /s). In fact, these effects can be almost negligible
73
for estuary lengths larger than about 2 km and basin widths larger than about 5 km (the change in critical fetch per unit length of the size will be smaller than 0.001 m/m; Figure
4.6 c and d). Interestingly, the results show that estuary length and width influence the system differently. Changes in basin width are directly linked to changes in marsh area, which is a sink for sediment (under the assumption that all sediment in the water column above marsh surface deposits on the marsh). However, changes in estuary length alter both marsh and tidal flat area, thus controlling both sediment sink and sediment export (sediment export occurs through ebb currents and depends on the tidal flat size). Hence, changing the width and length of the estuary can have different impacts on the dynamics of the system.
74
Figure 4.6: The relationships between fetch length with basin width (a and c) and estuary length (b and d) in determining marsh- (grey and hatched areas) vs. tidal flat- (white area) dominance. The border of hatched areas corresponds to the critical fetch found based on both vertical and lateral dynamics, whereas the border of grey area is linked only to the lateral erosional/depositional dynamics on the margin. The reference river concentration and river discharge are 50 mg/l and 50 m3/s (a and b), and 5 mg/l and 5 m3/s (c and d), respectively. The other parameters are fixed including ocean concentration (10 mg/l), tidal range (1.4 m) and sea level rise rate (6 mm/yr).
75
Inspection of the sensitivity of the model results to various environmental factors and geometric properties for basins with similar widths reveals that the river concentration has the largest effect (Figure 4.7; see appendix C for more detailed description). Tidal range, wind velocity and river discharge are also important parameters, whereas estuary length and SLR rate are the least important. These physical properties also contribute to model output variations through interactions with each other, and suggest that the interactions between river concentration, river discharge and tidal range are more significant than other interactions (Figure 4.8; see appendix C for more detailed description).
Figure 4.7: Sensitivity measures of main effects S (green), total effects T (orange), and effects due to interactions only, T-S (grey) for ocean concentration (CO),
76
river concentration (CR), river discharge (QR), estuary length (le), tidal range (a), sea level rise rate (R), and wind velocity (vw).
Figure 4.8: Heat map of the interactions between pairs of variables: ocean concentration (CO), river concentration (CR), river discharge (QR), estuary length (le), tidal range (a), sea level rise rate (R), and wind velocity (vw). The dendrogram shows the hierarchical binary cluster tree based on the comparison between the pair effect values.
4.4 Discussion and Conclusions
The findings reveal that vertical depositional processes on tidal flats in addition to the horizontal dynamics of margin migration can lead to transitions from tidal flat- to marsh-dominance. It is important to note that these vertical dynamics can influence the basins of attractions leading to marsh- or tidal flat-dominance in two ways: 1) by adding
77
to the marsh-dominance basin and 2) by influencing the rate at which the transitions from one basin of attraction to another can happen. The latter is supported by the finding by Mariotti and Fagherazzi [2013] that the depositional vertical dynamics happen at faster timescales than horizontal ones. Therefore, transitions from tidal flat to marsh can happen faster if vertical dynamics are involved in addition to the horizontal dynamics. Moreover, the relatively slow horizontal dynamics, asymptotically leading to marsh- or tidal flat-dominance, suggest that estuaries are likely to be found, at any given time, in a transition state between marsh- and tidal flat-dominated estuaries, dictated by temporal variations in environmental forcings. Despite their importance in determining marsh- vs. tidal flat-dominance, however, previous studies only consider the horizontal depositional/erosional processes that result in margin migration, and neglect the situations under which tidal flats can emerge and convert to marsh regulating transitions between tidal flat- and marsh-dominance [e.g., see D'Alpaos and Marani,
2015; D'Alpaos et al., 2007; Kirwan and Murray, 2007; Kirwan et al., 2016; Mariotti and
Fagherazzi, 2013; Mariotti and Carr, 2014; Mudd et al., 2009]. These vertical depositional processes are especially critical when addressing system response to environmental change in different locations with varying tidal ranges. For instance, the vertical processes can largely alter the dynamics of the system and significantly add to the extent of the marsh-dominance basin (e.g., Figure 4.3 d and Figure 4.4 d). In addition, while considering only the lateral margin dynamics suggests that marshes in a macrotidal environment with a tidal range of 5 m can be vulnerable to a SLR rate of 6 mm/yr,
78
incorporating the vertical depositional dynamics shows that marshes are indeed resilient to SLR and can survive (Figure 4.3 d).
Moreover, the model experiments reveal that river inputs of water and suspended sediment influence the stability and extent of wetland and embayed estuary equilibria, though they do not alter the basic dynamics that produce these alternative stable states. Increasing river discharge promotes marsh expansion when the river concentration is greater than the ocean concentration and leads to marsh retreat when the river concentration is smaller, having a similar effect to SLR in terms of shrinking marsh-dominance basin. Therefore, in a system with varying ocean and river concentrations, the ocean to river discharge ratio is one of the key controlling factors of the system behavior as it determines the final yield of sediment import to the system.
In addition to the river characteristics, other environmental and topographic parameters govern the morphodynamics of tidal flat/marsh platforms. For instance, increasing the ocean sediment concentration leads to marsh-dominance expansion and stability of marshes in response to changes of SLR rates. The effect of increasing tidal range is similar: marshes are more resilient in a macrotidal environment compared to mesotidal and microtidal systems [D'Alpaos et al., 2011; D'Alpaos et al., 2012; Kirwan and Guntenspergen, 2010]. In contrast, SLR and wind velocity favor tidal flat expansion and small marsh-dominance basins.
Estuary size, when comparing estuaries with similar river characteristics and ocean sediment inputs, is another key determinant of marsh vs. tidal flat extent. Indeed,
79
the size of the basin shapes the behavior of the system by controlling the water volume and determining the relative importance of ocean to river discharge. Moreover, the results suggest that estuary length and basin width influence the system through different mechanisms. Estuary length influences both tidal flat and marsh whereas the basin width mainly controls marsh area. In general, with higher ratios of tidal flat area to marsh area, the system tends to lose sediment through export via ebb currents, whereas lower ratios favor sediment absorption by the system through effective sedimentation on the marsh surface. It is worth noting that the results of my model are quantitatively limited to the different system configurations considered in this work (since the marsh vs. tidal flat boundaries can shift in case of system settings due to variation in the average concentration of the system and depth), however, they are qualitatively robust in explaining the relationships between different factors in determining tidal flat vs. marsh extent.
A sensitivity analysis suggests that among all parameters, basin width and river concentration have the most influence on the model output variations, whereas SLR rate is one of the least influential parameters. In addition, when basin width is fixed, the influence of both river concentration and SLR rate increases for basins with smaller widths. This finding confirms that riverine sediment inputs can especially help marshes in narrower basins, where they are highly threatened by elevated SLR rates, prevent drowning in response to SLR.
80
My model addresses the relevant mechanisms for morphodynamic evolution of tidal flat and marsh within an estuary, however, it has some limitations as some processes have been simplified. For instance, my model considers a constant effective sediment concentration and falls short of addressing varying concentration during a tidal cycle as a result of erosion and deposition, and with changing water depth. Due to these limitations and the limitations discussed earlier about the model framework, the model should only be used to understand the role of different dynamical mechanisms, rather than to produce detailed predictions.
In summary, the model developed is the first estuarine scale model which includes the full range of relevant geomorphic processes including local processes at the scale of an individual marsh/tidal flat and larger scale processes at the scale of watershed and estuary. The findings reveal that different mechanisms regulate the transitions from tidal flat- to marsh-dominance: lateral erosion and accretion dynamics at the margin, and vertical accretion dynamics on tidal flat surface that can happen much more rapidly than lateral dynamics. I suggest that considering both lateral and vertical dynamics is critical and otherwise it is not possible to address the full range of mechanisms regulating marsh extent. In addition, the results show that rivers can affect the extent of marshes in two ways: 1) by fluvial sediment delivery to tidal platforms, and
2) by regulating exchange with the ocean through riverine discharge. Among various environmental forcings and geometric properties controlling tidal flat- vs. marsh- dominance, river concentration is found to be a key factor influencing the marsh vs. tidal
81
flat extent for basins with similar widths. Considering that different parts of the estuary may have different sizes and orientations, the same wind regime can represent a very different forcing in different parts of the system (due to the potential variations in the fetch length and the angle between direction of wind and marsh margin in different locations). Additionally, sediment availability may be quite variable across an estuary.
All these situations may explain the simultaneous occurrence of marsh- and tidal flat- dominance in different parts of a single system.
82 5. Conclusions
The approach proposed here as a simultaneous use of lidar and hyperspectral data successfully characterizes the vegetation patterns in relation to topographic features in coastal dunes and opens the door to the full exploration of the space-time evolution of coastal eco-geomorphic patterns. In addition, I found that the spatially-explicit ecomorphodynamic patterns highlight the signatures of geomorphic controls on the spatial vegetation distribution and the role of the foredune crestline as a chief determinant of vegetation development. The joint ecomorphodynamic patterns identified and analyzed through remote sensing are, in fact, the result of a two-way interaction between ecological and morphological dynamics: while the foredune crestline defines the space where the maximum biomass production occurs, vegetation growth controls sand deposition, builds the dune, and localizes the crestline. In this respect, the ecomorphodynamic patterns identified by my analyses bear significant implications and pave the way for physically-based dune ecomorphodynamic modelling single system.
In addition, I find that plant biomass production does not seem to define the eventual SOM content and OMD. However, both parameters, in fact, show greater covariation with elevation and hydroperiod than with vegetation cover. My results also suggest that only the refractory part of the belowground biomass contributes to soil accretion and emphasizes the role of the stabilization factor in modulating marsh vertical accretion: a higher stabilization factor increases the effectively refractory
83
fraction, thereby contributing to increase SOM accumulation. Overall, these findings provide the basis for refining representations of the SOM accumulation process to inform the next generation of spatially-explicit bio-morphological models of salt marsh evolution. These new models will be a key tool to evaluate impending rates of marsh loss and the associated carbon release.
The findings of marsh modelling study reveal that different mechanisms regulate the transitions from tidal flat- to marsh-dominance: lateral erosion and accretion dynamics at the margin, and vertical accretion dynamics on tidal flat surface that can happen much more rapidly than lateral dynamics. I suggest that considering both lateral and vertical dynamics is critical and otherwise it is not possible to address the full range of mechanisms regulating marsh extent. In addition, the results show that rivers can affect the extent of marshes in two ways: 1) by fluvial sediment delivery to tidal platforms, and 2) by regulating exchange with the ocean through riverine discharge.
Among various environmental forcings and geometric properties controlling tidal flat- vs. marsh-dominance, river concentration is found to be a key factor influencing the marsh vs. tidal flat extent for basins with similar widths. Considering that different parts of the estuary may have different sizes and orientations, the same wind regime can represent a very different forcing in different parts of the system (due to the potential variations in the fetch length and the angle between direction of wind and marsh margin in different locations).
84 Appendix A
Figure A.1: NDVI as a function of distance from the foredune crestline for site 1 (a), site 2 (b), and site 3 (c). The lighter yellow colors represent higher data density whereas the darker blue colors show lower data density. Red circles represent the median of NDVI values within distance intervals of 3.5 m. In the figure, the shoreline is located right of the foredune crestline, such that the seaward and landward sides of
85
the foredune are identified by positive and negative distances to the foredune crestline, respectively. The vertical line shows the location of the foredune crestline.
Figure A.2: Heat maps of NDVI for site 1 (a) and site 2 (b). Colors represent the median values of NDVI for groups of pixels with the same ground elevation and the same distance to the shoreline. The vertical solid and dashed lines represent the average distance to the shoreline of the first and second crestlines, respectively.
86
Appendix B
Figure B.1: Soil organic carbon density, OCD, vs. marsh surface elevation relative to mean sea level along transects T1 (open circles), T2 (circle plus symbols) and T3 (closed circles). Solid lines represent the regression lines for each transect.
87
Appendix C
Model Construction
The erosion processes of the tidal flat bed depend on the wave-induced bed shear stress, 휏. I consider a depth-averaged shear stress which is computed as:
1 휏 = 휌 푓 푢 2 (C.1) 2 푤 푤 푤 where 휌푤 is water density, 푓푤 is friction factor and 푢푤 is the maximum horizontal orbital velocity. 푓푤 is set equal to:
3 − 퐻푤 4 푓푤 = 0.4 ( ) (C.2) 푘0 푠푖푛ℎ 푘푤ℎ where 퐻푤 is the significant wave height, 푘0 is roughness, 푘푤 is the wave number computed from the dispersion relationship for monochromatic waves [Marani et al.,
1 2010], and h is equal to (푑 + 푚푎푥(0, 푑 − 2퐻)) as the tidally averaged reference water 2 푓 푓 depth where 퐻 is the tidal amplitude. The maximum horizontal orbital velocity, 푢푤, is equal to:
휋퐻푤 푢푤 = (C.3) 푇푤 푠푖푛ℎ 푘푤ℎ where 푇푤 is the peak wave period. The wave characteristics, 퐻푤 and 푇푤 are computed using the approach proposed by Young and Verhagen [1996] using 2푏푓 as the reference fetch, ℎ as the reference water depth, and an isotropic wind regime. In this method, the significant wave height 퐻푤 and the peak wave period 푇푤 on the tidal flats are computed with semi-empirical equations:
0.87 푔퐻푤 퐵1 2 = 0.2413 (tanh 퐴1 tanh ) (C.4) 푣푤 tanh 퐴1
88
and
0.37 푔푇푤 퐵2 = 7.518 (tanh 퐴2 tanh ) (C.5) 푣푤 tanh 퐴2 with
푔ℎ 0.75 퐴1 = 0.493 ( 2 ) (C.6) 푣푤
0.57 −3 푔휒 퐵1 = 3.13 ∗ 10 ( 2 ) (C.7) 푣푤
푔ℎ 1.01 퐴2 = 0.331 ( 2 ) (C.8) 푣푤
0.73 −4 푔휒 퐵2 = 5.215 ∗ 10 ( 2 ) (C.9) 푣푤 where 푔 is the gravitational acceleration, 푣푤 is wind velocity, and 휒 is the fetch length. 푐푔 is wave group velocity and is computed as:
2π 1 2kwh 푐푔 = (1 + ) (C.10) kwTw 2 sinh 2kwh
Simulations Setup
The model experiments begin with uniform marsh and tidal flat platforms with depths set to ¼ and ¾ of the tidal range, respectively. I used two different approaches to run the model simulations under different conditions of a system containing marsh and tidal flat platforms. First, I set the initial value of tidal flat width equal to 100 m and numerically solved the system of equations (including equations 3.1, 3.13, 3.16, and 3.17) at steady state to find the equilibrium solutions. To do this, I used a Matlab® solver that implements the Trust-Region Dogleg method, an advanced Newton method for nonlinear problems (fsolve). In this approach, I disregarded the elevation at which the 89
dominant marsh vegetation, Spartina patens, starts to grow. Therefore, this approach does not account for tidal flat conversion to marsh. Instead, it only considers the conditions under which the horizontal erosion and deposition dynamics on the marsh/tidal flat margin lead to transitions between tidal flat- and salt marsh- basins of attraction.
In the second approach, I used a Matlab® ordinary differential equation solver that implements the Numerical Differentiation Formulas (ode15s). I simulated the evolution of the system for 1000 years using time steps equal to one tidal period. Starting with 1 m as the initial tidal flat width, if at the end of 1000 years the system was filled with marsh or the marsh was still expanding, the system was considered to lie on a marsh-dominance basin, whereas if the marsh was fully eroded or was still contracting, the system was considered to lie on a tidal flat-dominance basin. I kept repeating this procedure with larger initial values of tidal flat width in 1 m increments up to the size of basin width. The initial tidal flat widths lying on the border between marsh- and tidal flat-dominance basins were considered the solutions at equilibrium. Here, I allowed tidal flat conversion to marsh by setting the elevation at which vegetation starts to grow to mean sea level, and thus, both vertical and horizontal dynamics can lead to transitions between basins of attraction.
In both approaches, the reference values for variables were set to: tidal range =
1.4 m, river concentration = 5 and 50 mg/l, river discharge = 5 and 50 m3/s, ocean concentration = 10 mg/l, basin width = 5 km, estuary length = 5 km, wind velocity = 6
90
m/s, and sea level rise rate = 6 mm/yr. Table C.1 lists the parameter values considered for my simulations.
Table C.1: Model Parameters
Symbol Parameter Value Source 푘푔 퐸 Bed erosion coefficient 10−4 Marani et al. [2007] 0 푚2푠
휏푐 Critical shear stress 0.3 푃퐴 푚푚 휔 Settling velocity 0.5 Mariotti and Carr [2014] 푠 푠 푘푔 휌푠 Sediment bulk density 1000 Mariotti and Carr [2014] 푚3
푘0 Roughness 1 푚푚 Mariotti and Carr [2014] Margin erodibility 푚 푊 −1 푘푒 0.16 ( ) Mariotti and Carr [2014] coefficient 푦푟 푚 Margin accretion Mariotti and Fagherazzi 푘푎 2 coefficient [2013] 2 2 푘퐵 Vegetation coefficient 푚 푚 Marani et al. [2010] × 10−3 푦푟 푘푔 Maximum biomass 푘푔 퐵푚푎푥 1 Marani et al. [2010] density 푚2 푁 훾 Specific weight of water 9800 푚3 푘푔 휌푤 Water density 1000 푚3
Sensitivity Analysis
I evaluated the relative importance of different environmental parameters and geometric properties on the variability of the model outputs. I applied a variance-based method which considers the contribution of variables to the variance of model outputs as a measure of sensitivity, known to be effective for environmental studies [Pianosi et
91
al., 2016]. The general form of the model is given by 푦 = 푓(푥1, 푥2, … , 푥푛), where 푦 is the model output and 푥푖 is the input variable for 𝑖 = 1, 2, … , 푛. The “main effect”, 푆푖, is a measure for the direct contribution of the 𝑖th variable in the total variation of model outputs and is defined by [Saltelli et al., 2010]:
푉푥 [퐸푥 (푦|푥푖)] 푆 = 푖 ~푖 (C.11) 푖 푉(푦) where 푉 denotes the variance, 퐸 denotes the expected value, and 푥~푖 refers to all variables except for 푥푖. 푆푖 corresponds to the variation in the outputs that can be explained by 푥푖. In other words, it accounts for the reduction of output variance if 푥푖 is fixed. To measure the overall contribution of the 𝑖th variable in the model results, the
“total effect”, 푇푖, is then defined by [Homma and Saltelli, 1996]:
퐸푥 [푉푥 (푦|푥~푖)] 푇 = ~푖 푖 (C.12) 푖 푉(푦)
푇푖 describes all the influences of 푥푖 on the outputs including both the main effects and joint effects due to possible interactions with other variables. I also analyzed the joint influence of different variables due to the possible physical interactions between them
[Saltelli et al., 2010]:
퐸푥 [푉푥 푥 (푦|푥~푖푗)] 푇 = ~푖푗 푖 푗 (C.13) 푖푗 푉(푦)
Here, 푇푖푗 accounts for the influence that the interaction between the pair of 푥푖 and 푥푗 can have on the model results. I considered eight variable inputs including river concentration (ranging between 10 and 5000 mg/l), river discharge (ranging between 10 and 5000 m3/s), ocean concentration (ranging between 5 and 65 mg/l), tidal range
92
(ranging between 0.5 and 16.5 m), wind velocity (ranging between 2 and 10 m/s), sea level rise rate (ranging between 2 and 18 mm/yr), basin width (ranging between 1 and 17 km) and estuary length (ranging between 1 and 17 km). I considered five different values for each variable producing 85=390625 different settings for an estuary system.
The results suggest that the influence of basin width on the critical fetch length is the highest among all the parameters (with 푆 = 0.67 and 푇 = 0.69 which are more than three times larger than those of other variables). In addition to the indirect controls of basin width on critical fetch (e.g., determining the ocean to river discharge ratio and controlling the amount of sediment import/export and accretion/erosion processes), basin width also directly affects the critical fetch length by setting its boundary limits.
This direct control overshadows the influence of other parameters. To resolve this issue,
I fixed the basin width and computed the main and total effects of other parameters. I repeated this for different values of basin width. The results show that the effects of river concentration are the largest (Figure 4.7). Tidal range, wind velocity and river discharge are the other influential parameters. Despite the physical interactions between river concentration and discharge, the effect of river discharge is less than half that of river concentration. This could be because of a possible saturation of the effect of increased discharge on the behavior of the system, such that river discharge fills the system and the excess water does not contribute to the system dynamics. The results also show that the influences of SLR rate and ocean concentration when considered alone are the smallest, however, they contribute to output variation through interactions
93
with other parameters. The difference between S and T describes the portion of the total effect that is only due to interactions with other parameters, and is largest for river concentration and tidal range. This suggests that the interactions between these parameters are probably greater than with other parameters. Analyzing the joint influences of parameters verifies this point (Figure 4.8). It is also important to note that influence of river discharge is the largest when interacting with river concentration, determining the sediment import to the system through the river.
94
Figure C.1: The relationships between critical fetch length associated with both vertical and lateral dynamics with river sediment concentration (a), river discharge (b), ocean sediment concertation (c) and tidal range (d) in determining marsh- vs. tidal flat-dominance basins. The lines show the critical fetch lines found from both vertical and lateral erosional/depositional dynamics for different rates of SLR. The reference river concentration is 5 mg/l, river discharge is 5 m3/s, ocean concentration is 10 mg/l, and tidal range is 1.4 m.
95
References
Allen, J. R. L. (1990), Salt-Marsh Growth and Stratification - a Numerical-Model with Special Reference to the Severn Estuary, Southwest Britain, Mar Geol, 95(2), 77- 96.
Allen, J. R. L. (1994), A Continuity-Based Sedimentological Model for Temperate-Zone Tidal Salt Marshes, J Geol Soc London, 151, 41-49.
Allen, J. R. L. (2000), Morphodynamics of Holocene salt marshes: a review sketch from the Atlantic and Southern North Sea coasts of Europe, Quaternary Sci Rev, 19(12), 1155-1231.
Alphan, H. (2011), Comparing the utility of image algebra operations for characterizing landscape changes: The case of the Mediterranean coast, J Environ Manage, 92(11), 2961-2971.
Armston, J., M. Disney, P. Lewis, P. Scarth, S. Phinn, R. Lucas, P. Bunting, and N. Goodwin (2013), Direct retrieval of canopy gap probability using airborne waveform lidar, Remote Sens Environ, 134, 24-38.
Armstrong, W., E. J. Wright, S. Lythe, and T. J. Gaynard (1985), Plant Zonation and the Effects of the Spring-Neap Tidal Cycle on Soil Aeration in a Humber Salt-Marsh, J Ecol, 73(1), 323-339.
Barbier, E. B., S. D. Hacker, C. Kennedy, E. W. Koch, A. C. Stier, and B. R. Silliman (2011), The value of estuarine and coastal ecosystem services, Ecological Monographs, 81(2), 169-193.
Berk, A., et al. (1999), MODTRAN4 radiative transfer modeling for atmospheric correction, P Soc Photo-Opt Ins, 3756, 348-353.
Bermudez, R., and R. Retuerto (2013), Living the difference: alternative functional designs in five perennial herbs coexisting in a coastal dune environment, Funct Plant Biol, 40(11), 1187-1198.
Bernasconi, L., I. Pippi, and S. Raddi (2003), Comparisons among normalized vegetation indices for the determination of LAI, Remote Sensing for Agriculture, Ecosystems, and Hydrology Iv, 4879, 145-153.
Blum, L. K. (1993), Spartina-Alterniflora Root Dynamics in a Virginia Marsh, Mar Ecol Prog Ser, 102(1-2), 169-178.
96
Boaga, J., A. D'Alpaos, G. Cassiani, M. Marani, and M. Putti (2014), Plant-soil interactions in salt marsh environments: Experimental evidence from electrical resistivity tomography in the Venice Lagoon, Geophys Res Lett, 41(17), 6160- 6166.
Brantley, S. T., J. C. Zinnert, and D. R. Young (2011), Application of hyperspectral vegetation indices to detect variations in high leaf area index temperate shrub thicket canopies, Remote Sens Environ, 115(2), 514-523.
Brantley, S. T., S. N. Bissett, D. R. Young, C. W. V. Wolner, and L. J. Moore (2014), Barrier Island Morphology and Sediment Characteristics Affect the Recovery of Dune Building Grasses following Storm-Induced Overwash, Plos One, 9(8).
Campbell, G. S. (1986), Extinction Coefficients for Radiation in Plant Canopies Calculated Using an Ellipsoidal Inclination Angle Distribution, Agr Forest Meteorol, 36(4), 317-321.
Carniello, L., A. Defina, S. Fagherazzi, and L. D'Alpaos (2005), A combined wind wave- tidal model for the Venice lagoon, Italy, J Geophys Res-Earth, 110(F4).
Castanho, C. D., C. J. Lortie, B. Zaitchik, and P. I. Prado (2015), A meta-analysis of plant facilitation in coastal dune systems: responses, regions, and research gaps, Peerj, 3.
Chander, G., B. L. Markham, and D. L. Helder (2009), Summary of current radiometric calibration coefficients for Landsat MSS, TM, ETM+, and EO-1 ALI sensors, Remote Sens Environ, 113(5), 893-903.
Chen, S., R. Torres, and M. A. Goni (2016), The Role of Salt Marsh Structure in the Distribution of Surface Sedimentary Organic Matter, Estuaries and Coasts, 39(1), 108-122.
Chmura, G. L., and G. A. Hung (2004), Controls on salt marsh accretion: A test in salt marshes of Eastern Canada, Estuaries, 27(1), 70-81.
Chmura, G. L., S. C. Anisfeld, D. R. Cahoon, and J. C. Lynch (2003), Global carbon sequestration in tidal, saline wetland soils, Global Biogeochem Cy, 17(4).
Christiansen, T., P. L. Wiberg, and T. G. Milligan (2000), Flow and sediment transport on a tidal salt marsh surface, Estuar Coast Shelf S, 50(3), 315-331.
Church, J. A., et al. (2013), Sea Level Change, in Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, edited by T. F. Stocker, D.
97
Qin, G.-K. Plattner, M. Tignor, S. K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P. M. Midgley, pp. 1137–1216, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.
Corenblit, D., A. C. W. Baas, G. Bornette, J. Darrozes, S. Delmotte, R. A. Francis, A. M. Gurnell, F. Julien, R. J. Naiman, and J. Steiger (2011), Feedbacks between geomorphology and biota controlling Earth surface processes and landforms: A review of foundation concepts and current understandings, Earth-Sci Rev, 106(3- 4), 307-331.
Craft, C., J. Clough, J. Ehman, S. Joye, R. Park, S. Pennings, H. Guo, and M. Machmuller (2009), Forecasting the effects of accelerated sea‐level rise on tidal marsh ecosystem services, Front Ecol Environ, 7(2), 73-78.
Craft, C. B., E. D. Seneca, and S. W. Broome (1991), Loss on Ignition and Kjeldahl Digestion for Estimating Organic-Carbon and Total Nitrogen in Estuarine Marsh Soils - Calibration with Dry Combustion, Estuaries, 14(2), 175-179.
Croft, A. L., L. A. Leonard, T. D. Alphin, L. B. Cahoon, and M. H. Posey (2006), The effects of thin layer sand renourishment on tidal marsh processes: Masonboro Island, North Carolina, Estuaries and Coasts, 29(5), 737-750.
Currey, P. M., D. Johnson, L. J. Sheppard, I. D. Leith, H. Toberman, R. van der Wal, L. A. Dawson, and R. R. E. Artz (2010), Turnover of labile and recalcitrant soil carbon differ in response to nitrate and ammonium deposition in an ombrotrophic peatland, Global Change Biol, 16(8), 2307-2321.
D'Alpaos, A. (2011), The mutual influence of biotic and abiotic components on the long- term ecomorphodynamic evolution of salt-marsh ecosystems, Geomorphology, 126(3-4), 269-278.
D'Alpaos, A., and M. Marani (2015), Reading the signatures of biologic–geomorphic feedbacks in salt-marsh landscapes, Advances in Water Resources.
D'Alpaos, A., and M. Marani (2016), Reading the signatures of biologic-geomorphic feedbacks in salt-marsh landscapes, Advances in Water Resources, 93, 265-275.
D'Alpaos, A., S. M. Mudd, and L. Carniello (2011), Dynamic response of marshes to perturbations in suspended sediment concentrations and rates of relative sea level rise, J Geophys Res-Earth, 116.
D'Alpaos, A., C. Da Lio, and M. Marani (2012), Biogeomorphology of tidal landforms: physical and biological processes shaping the tidal landscape, Ecohydrology, 5(5), 550-562.
98
D'Alpaos, A., S. Lanzoni, S. M. Mudd, and S. Fagherazzi (2006), Modeling the influence of hydroperiod and vegetation on the cross-sectional formation of tidal channels, Estuar Coast Shelf S, 69(3-4), 311-324.
D'Alpaos, A., S. Lanzoni, M. Marani, and A. Rinaldo (2007), Landscape evolution in tidal embayments: Modeling the interplay of erosion, sedimentation, and vegetation dynamics, J Geophys Res-Earth, 112(F1).
D'Alpaos, A., S. Lanzoni, M. Marani, S. Fagherazzi, and A. Rinaldo (2005), Tidal network ontogeny: Channel initiation and early development, J Geophys Res-Earth, 110(F2).
Da Lio, C., A. D'Alpaos, and M. Marani (2013), The secret gardener: vegetation and the emergence of biogeomorphic patterns in tidal environments, Philos T R Soc A, 371(2004).
Dadey, K. A., T. Janecek, and A. Klaus (1992), Dry-Bulk Density: Its Use and Determination.
Danielsen, F., et al. (2005), The Asian tsunami: A protective role for coastal vegetation, Science, 310(5748), 643-643.
Day, J. W., G. P. Shaffer, L. D. Britsch, D. J. Reed, S. R. Hawes, and D. Cahoon (2000), Pattern and process of land loss in the Mississippi Delta: A spatial and temporal analysis of wetland habitat change, Estuaries, 23(4), 425-438.
Day, J. W., L. D. Britsch, S. R. Hawes, G. P. Shaffer, D. J. Reed, and D. Cahoon (2000), Pattern and process of land loss in the Mississippi Delta: A Spatial and temporal analysis of wetland habitat change, Estuaries, 23(4), 425-438.
Defina, A., L. Carniello, S. Fagherazzi, and L. D'Alpaos (2007), Self‐organization of shallow basins in tidal flats and salt marshes, Journal of Geophysical Research: Earth Surface, 112(F3).
DeLaune, R. D., and S. R. Pezeshki (2003), The Role of Soil Organic Carbon in Maintaining Surface Elevation in Rapidly Subsiding U.S. Gulf of Mexico Coastal Marshes, Water, Air and Soil Pollution: Focus, 3(1), 167-179.
Disraeli, D. J. (1984), The Effect of Sand Deposits on the Growth and Morphology of Ammophila-Breviligulata, J Ecol, 72(1), 145-154.
Duarte, C. M., J. J. Middelburg, and N. Caraco (2005), Major role of marine vegetation on the oceanic carbon cycle, Biogeosciences, 2(1), 1-8.
99
Duchemin, B., et al. (2006), Monitoring wheat phenology and irrigation in Central Morocco: On the use of relationships between evapotranspiration, crops coefficients, leaf area index and remotely-sensed vegetation indices, Agr Water Manage, 79(1), 1-27.
Duran, O., and L. J. Moore (2013), Vegetation controls on the maximum size of coastal dunes, P Natl Acad Sci USA, 110(43), 17217-17222.
Duran Vinent, O., and L. J. Moore (2015), Barrier island bistability induced by biophysical interactions, Nat Clim Change, 5(2), 158-162.
Fagherazzi, S., L. Carniello, L. D'Alpaos, and A. Defina (2006), Critical bifurcation of shallow microtidal landforms in tidal flats and salt marshes, Proceedings of the National Academy of Sciences, 103(22), 8337-8341.
Fagherazzi, S., M. L. Kirwan, S. M. Mudd, G. R. Guntenspergen, S. Temmerman, A. D'Alpaos, J. Koppel, J. M. Rybczyk, E. Reyes, and C. Craft (2012a), Numerical models of salt marsh evolution: Ecological, geomorphic, and climatic factors, Reviews of Geophysics, 50(1).
Fagherazzi, S., et al. (2012b), Numerical models of salt marsh evolution: Ecological, geomorphic, and climatic factors, Reviews of Geophysics, 50(1), n/a-n/a.
Feagin, R. A., S. M. Lozada-Bernard, T. M. Ravens, I. Moller, K. M. Yeager, and A. H. Baird (2009), Does vegetation prevent wave erosion of salt marsh edges?, P Natl Acad Sci USA, 106(25), 10109-10113.
Fieber, K. D., I. J. Davenport, M. A. Tanase, J. M. Ferryman, R. J. Gurney, J. P. Walker, and J. M. Hacker (2014), Effective LAI and CHP of a Single Tree From Small- Footprint Full-Waveform LiDAR, Ieee Geosci Remote S, 11(9), 1634-1638.
FitzGerald, D. M., M. S. Fenster, B. A. Argow, and I. V. Buynevich (2008), Coastal impacts due to sea-level rise, Annu Rev Earth Pl Sc, 36, 601-647.
Frangipane, G., M. Pistolato, E. Molinaroli, S. Guerzoni, and D. Tagliapietra (2009), Comparison of loss on ignition and thermal analysis stepwise methods for determination of sedimentary organic matter, Aquat Conserv, 19(1), 24-33.
Frederiksen, L., J. Kollmann, P. Vestergaard, and H. H. Bruun (2006), A multivariate approach to plant community distribution in the coastal dune zonation of NW Denmark, Phytocoenologia, 36(3), 321-342.
100
French, J. R., and D. R. Stoddart (1992), Hydrodynamics of Salt-Marsh Creek Systems - Implications for Marsh Morphological Development and Material Exchange, Earth Surf Processes, 17(3), 235-252.
Gibeault, C., U. Neumeier, and P. Bernatchez (2016), Spatial and Temporal Sediment Dynamics in a Subarctic Salt Marsh (Gulf of St. Lawrence, Canada), J Coastal Res, 32(6), 1344-1361.
Gisen, J. I. A., and H. H. Savenije (2015), Estimating bankfull discharge and depth in ungauged estuaries, Water Resour Res, 51(4), 2298-2316.
Godfrey, P. J. (1977), Climate, Plant Response and Development of Dunes on Barrier Beaches Along United-States East Coast, Int J Biometeorol, 21(3), 203-215.
Goldstein, E. B., and L. J. Moore (2016), Stability and bistability in a one-dimen- sional model of coastal foredune height, J. Geophys. Res. Earth Surf., 121.
Gunnell, J. R., A. B. Rodriguez, and B. A. McKee (2013), How a marsh is built from the bottom up, Geology, 41(8), 859-862.
Hacker, S. D., P. Zarnetske, E. Seabloom, P. Ruggiero, J. Mull, S. Gerrity, and C. Jones (2012), Subtle differences in two non-native congeneric beach grasses significantly affect their colonization, spread, and impact, Oikos, 121(1), 138-148.
Hackney, C. T. (1987), Factors Affecting Accumulation or Loss of Macroorganic Matter in Salt-Marsh Sediments, Ecology, 68(4), 1109-1113.
Halupa, P. J., and B. L. Howes (1995), Effects of Tidally Mediated Litter Moisture- Content on Decomposition of Spartina-Alterniflora and S-Patens, Mar Biol, 123(2), 379-391.
Hansen, K., C. Butzeck, A. Eschenbach, A. Grongroft, K. Jensen, and E. M. Pfeiffer (2017), Factors influencing the organic carbon pools in tidal marsh soils of the Elbe estuary (Germany), J Soil Sediment, 17(1), 47-60.
Harmon, M. E., W. L. Silver, B. Fasth, H. Chen, I. C. Burke, W. J. Parton, S. C. Hart, W. S. Currie, and Lidet (2009), Long-term patterns of mass loss during the decomposition of leaf and fine root litter: an intersite comparison, Global Change Biol, 15(5), 1320-1338.
Hayden, B. P., R. D. Dueser, J. T. Callahan, and H. H. Shugart (1991), Long-Term Research at the Virginia Coast Reserve, Bioscience, 41(5), 310-318.
101
Hayden, B. P., M. C. F. V. Santos, G. F. Shao, and R. C. Kochel (1995), Geomorphological Controls on Coastal Vegetation at the Virginia-Coast-Reserve, Geomorphology, 13(1-4), 283-300.
Hazelden, J., and L. A. Boorman (1999), The role of soil and vegetation processes in the control of organic and mineral fluxes in some western European salt marshes, J Coastal Res, 15(1), 15-31.
Hesp, P. (1988), Surfzone, beach, and foredune interactions on the Australian South East Coast, J Coast Res(Special issue 3), 15–23.
Hesp, P. (2002), Foredunes and blowouts: initiation, geomorphology and dynamics, Geomorphology, 48(1-3), 245-268.
Hinson, A. L., R. A. Feagin, M. Eriksson, R. G. Najjar, M. Herrmann, T. S. Bianchi, M. Kemp, J. A. Hutchings, S. Crooks, and T. Boutton (2017), The spatial distribution of soil organic carbon in tidal wetland soils of the continental United States, Global Change Biol, 23(12), 5468-5480.
Homma, T., and A. Saltelli (1996), Importance measures in global sensitivity analysis of nonlinear models, Reliab Eng Syst Safe, 52(1), 1-17.
Houldcroft, C. J., C. L. Campbell, I. J. Davenport, R. J. Gurney, and N. Holden (2005), Measurement of canopy geometry characteristics using LiDAR laser altimetry: A feasibility study, Ieee T Geosci Remote, 43(10), 2270-2282.
Howarth, R. W., and J. E. Hobbie (1981), The Regulation of Decomposition and Heterotrophic Microbial Activity in Salt-Marsh Soils, Estuaries, 4(3), 289-289.
Janousek, C. N., K. J. Buffington, G. R. Guntenspergen, K. M. Thorne, B. D. Dugger, and J. Y. Takekawa (2017), Inundation, Vegetation, and Sediment Effects on Litter Decomposition in Pacific Coast Tidal Marshes, Ecosystems, 20(7), 1296-1310.
Jin, H. X., and L. Eklundh (2014), A physically based vegetation index for improved monitoring of plant phenology, Remote Sens Environ, 152, 512-525.
Keijsers, J. G. S., A. V. De Groot, and M. J. P. M. Riksen (2015), Vegetation and sedimentation on coastal foredunes, Geomorphology, 228, 723-734.
Keller, J. K., T. Anthony, D. Clark, K. Gabriel, D. Gamalath, R. Kabala, J. King, L. Medina, and M. Nguyen (2015), Soil Organic Carbon and Nitrogen Storage in Two Southern California Salt Marshes: The Role of Pre-Restoration Vegetation, Bulletin, Southern California Academy of Sciences, 114(1), 22-32.
102
Kelleway, J. J., N. Saintilan, P. I. Macreadie, and P. J. Ralph (2016), Sedimentary Factors are Key Predictors of Carbon Storage in SE Australian Saltmarshes, Ecosystems, 19(5), 865-880.
Kempeneers, P., B. Deronde, S. Provoost, and R. Houthuys (2009), Synergy of Airborne Digital Camera and Lidar Data to Map Coastal Dune Vegetation, J Coastal Res, 25(6), 73-82.
Kennish, M. J. (2001), Coastal salt marsh systems in the US: A review of anthropogenic impacts, J Coastal Res, 17(3), 731-748.
Keuskamp, J. A., B. J. J. Dingemans, T. Lehtinen, J. M. Sarneel, and M. M. Hefting (2013), Tea Bag Index: a novel approach to collect uniform decomposition data across ecosystems, Methods Ecol Evol, 4(11), 1070-1075.
Kirwan, M. L., and A. B. Murray (2007), A coupled geomorphic and ecological model of tidal marsh evolution, P Natl Acad Sci USA, 104(15), 6118-6122.
Kirwan, M. L., and G. R. Guntenspergen (2010), Influence of tidal range on the stability of coastal marshland, J Geophys Res-Earth, 115.
Kirwan, M. L., and S. M. Mudd (2012), Response of salt-marsh carbon accumulation to climate change, Nature, 489(7417), 550-+.
Kirwan, M. L., D. C. Walters, W. G. Reay, and J. A. Carr (2016), Sea level driven marsh expansion in a coupled model of marsh erosion and migration, Geophys Res Lett, 43(9), 4366-4373.
Kirwan, M. L., G. R. Guntenspergen, A. D'Alpaos, J. T. Morris, S. M. Mudd, and S. Temmerman (2010), Limits on the adaptability of coastal marshes to rising sea level, Geophys Res Lett, 37.
Lawson, S., P. Wiberg, K. McGlathery, and D. Fugate (2007), Wind-driven sediment suspension controls light availability in a shallow coastal lagoon, Estuaries and Coasts, 30(1), 102-112.
Lemay, L. E. (2007), The impact of drainage ditches on salt marsh flow patterns, sedimentation and morphology: Rowley River, Massachusetts.
Leonardi, N., and S. Fagherazzi (2015), Effect of local variability in erosional resistance on large-scale morphodynamic response of salt marshes to wind waves and extreme events, Geophys Res Lett, 42(14), 5872-5879.
103
Li, H., L. Li, and D. Lockington (2005), Aeration for plant root respiration in a tidal marsh, Water Resour Res, 41(6), n/a-n/a.
Lo, C. P., and D. A. Quattrochi (2003), Land-use and land-cover change, urban heat island phenomenon, and health implications: A remote sensing approach, Photogramm Eng Rem S, 69(9), 1053-1063.
Long, J. W., A. T. M. de Bakker, and N. G. Plant (2014), Scaling coastal dune elevation changes across storm-impact regimes, Geophys Res Lett, 41(8), 2899-2906.
Lotze, H. K., H. S. Lenihan, B. J. Bourque, R. H. Bradbury, R. G. Cooke, M. C. Kay, S. M. Kidwell, M. X. Kirby, C. H. Peterson, and J. B. Jackson (2006), Depletion, degradation, and recovery potential of estuaries and coastal seas, Science, 312(5781), 1806-1809.
Luo, S. Z., C. Wang, F. F. Pan, X. H. Xi, G. C. Li, S. Nie, and S. B. Xia (2015), Estimation of wetland vegetation height and leaf area index using airborne laser scanning data, Ecol Indic, 48, 550-559.
Malavasi, M., R. Santoro, M. Cutini, A. T. R. Acosta, and M. L. Carranza (2013), What has happened to coastal dunes in the last half century? A multitemporal coastal landscape analysis in Central Italy, Landscape Urban Plan, 119, 54-63.
Manso, A. F., P. Illera, J. A. Delgado, and A. F. Unzueta (1998), Climatic interpretation of the NDVI: applications for vegetation monitoring in Castilla y Leon (Spain), Remote Sensing for Agriculture, Ecosystems, and Hydrology, 3499, 372-383.
Marani, M., C. Da Lio, and A. D'Alpaos (2013), Vegetation engineers marsh morphology through multiple competing stable states, P Natl Acad Sci USA, 110(9), 3259- 3263.
Marani, M., A. D'Alpaos, S. Lanzoni, L. Carniello, and A. Rinaldo (2007), Biologically- controlled multiple equilibria of tidal landforms and the fate of the Venice lagoon, Geophys Res Lett, 34(11).
Marani, M., A. D'Alpaos, S. Lanzoni, L. Carniello, and A. Rinaldo (2010), The importance of being coupled: Stable states and catastrophic shifts in tidal biomorphodynamics, J Geophys Res-Earth, 115.
Marani, M., S. Silvestri, E. Belluco, N. Ursino, A. Comerlati, O. Tosatto, and M. Putti (2006), Spatial organization and ecohydrological interactions in oxygen-limited vegetation ecosystems, Water Resour Res, 42(6).
104
Mariotti, G., and S. Fagherazzi (2013), Critical width of tidal flats triggers marsh collapse in the absence of sea-level rise, P Natl Acad Sci USA, 110(14), 5353-5356.
Mariotti, G., and J. Carr (2014), Dual role of salt marsh retreat: Long- term loss and short- term resilience, Water Resour Res, 50(4), 2963-2974.
McCaffrey, R. J., and J. Thomson (1980), A Record of the Accumulation of Sediment and Trace Metals in A Connecticut Salt Marsh, in Advances in Geophysics, edited by B. Saltzman, pp. 165-236, Elsevier.
Mcleod, E., G. L. Chmura, S. Bouillon, R. Salm, M. Bjork, C. M. Duarte, C. E. Lovelock, W. H. Schlesinger, and B. R. Silliman (2011), A blueprint for blue carbon: toward an improved understanding of the role of vegetated coastal habitats in sequestering CO2, Front Ecol Environ, 9(10), 552-560.
Mitsch, W. J., and J. G. Gosselink (2007), Wetlands, 4 ed., Wiley, Hoboken, NJ.
Molina, J. A., M. A. Casermeiro, and P. S. Moreno (2003), Vegetation composition and soil salinity in a Spanish Mediterranean coastal ecosystem, Phytocoenologia, 33(2-3), 475-494.
Moller, I., T. Spencer, J. R. French, D. J. Leggett, and M. Dixon (1999), Wave transformation over salt marshes: A field and numerical modelling study from north Norfolk, England, Estuar Coast Shelf S, 49(3), 411-426.
Moore, L. J., J. H. List, S. J. Williams, and D. Stolper (2010), Complexities in barrier island response to sea level rise: Insights from numerical model experiments, North Carolina Outer Banks, J Geophys Res-Earth, 115.
Morris, J. T., P. V. Sundareshwar, C. T. Nietch, B. Kjerfve, and D. R. Cahoon (2002), Responses of coastal wetlands to rising sea level, Ecology, 83(10), 2869-2877.
Morris, J. T., et al. (2016), Contributions of organic and inorganic matter to sediment volume and accretion in tidal wetlands at steady state, Earths Future, 4(4), 110- 121.
Mudd, S. M., S. M. Howell, and J. T. Morris (2009), Impact of dynamic feedbacks between sedimentation, sea-level rise, and biomass production on near-surface marsh stratigraphy and carbon accumulation, Estuar Coast Shelf S, 82(3), 377- 389.
Mudd, S. M., A. D'Alpaos, and J. T. Morris (2010), How does vegetation affect sedimentation on tidal marshes? Investigating particle capture and
105
hydrodynamic controls on biologically mediated sedimentation, J Geophys Res- Earth, 115.
Mudd, S. M., S. Fagherazzi, J. T. Morris, and D. J. Furbish (2013), Flow, Sedimentation, and Biomass Production on a Vegetated Salt Marsh in South Carolina: Toward a Predictive Model of Marsh Morphologic and Ecologic Evolution, in The Ecogeomorphology of Tidal Marshes, edited, pp. 165-188, American Geophysical Union.
Murray, A. B., M. A. F. Knaapen, M. Tal, and M. L. Kirwan (2008), Biomorphodynamics: Physical-biological feedbacks that shape landscapes, Water Resour Res, 44(11).
Nyman, J. A., R. D. Delaune, H. H. Roberts, and W. H. Patrick (1993), Relationship between Vegetation and Soil Formation in a Rapidly Submerging Coastal Marsh, Mar Ecol Prog Ser, 96(3), 269-279.
Nyman, J. A., R. D. Delaune, S. R. Pezeshki, and W. H. Patrick (1995), Organic-Matter Fluxes and Marsh Stability in a Rapidly Submerging Estuarine Marsh, Estuaries, 18(1b), 207-218.
Nyman, J. A., R. J. Walters, R. D. Delaune, and W. H. Patrick (2006), Marsh vertical accretion via vegetative growth, Estuar Coast Shelf S, 69(3-4), 370-380.
Okin, G. S. (2008), A new model of wind erosion in the presence of vegetation, J Geophys Res-Earth, 113(F2).
Parker, K. C., and J. Bendix (1996), Landscape-scale geomorphic influences on vegetation patterns in four environments, Phys Geogr, 17(2), 113-141.
Pelletier, J. D., et al. (2015), Forecasting the response of Earth's surface to future climatic and land use changes: A review of methods and research needs, Earths Future, 3(7), 220-251.
Pianosi, F., K. Beven, J. Freer, J. W. Hall, J. Rougier, D. B. Stephenson, and T. Wagener (2016), Sensitivity analysis of environmental models: A systematic review with practical workflow, Environ Modell Softw, 79, 214-232.
Prescott, C. E. (2010), Litter decomposition: what controls it and how can we alter it to sequester more carbon in forest soils?, Biogeochemistry, 101(1-3), 133-149.
Psuty, N. (2008), The coastal foredune: A morphological basis for regional coastal dune development, Coastal Dunes, eds Martínez ML, Psuty NP (Springer, Berlin), 171, 11–27.
106
Ratliff, K. M., A. E. Braswell, and M. Marani (2015), Spatial response of coastal marshes to increased atmospheric CO2, P Natl Acad Sci USA, 112(51), 15580-15584.
Raupach, M. R., D. A. Gillette, and J. F. Leys (1993), The Effect of Roughness Elements on Wind Erosion Threshold, J Geophys Res-Atmos, 98(D2), 3023-3029.
Redfield, A. C. (1965), Ontogeny of a Salt Marsh Estuary, Science, 147(3653), 50-&.
Richardson, J. J., L. M. Moskal, and S. H. Kim (2009), Modeling approaches to estimate effective leaf area index from aerial discrete-return LIDAR, Agr Forest Meteorol, 149(6-7), 1152-1160.
Ritchie, W., and S. Penland (1988), Rapid Dune Changes Associated with Overwash Processes on the Deltaic Coast of South Louisiana, Mar Geol, 81(1-4), 97-122.
Roner, M., A. D’Alpaos, M. Ghinassi, M. Marani, S. Silvestri, E. Franceschinis, and N. Realdon (2016), Spatial variation of salt-marsh organic and inorganic deposition and organic carbon accumulation: Inferences from the Venice lagoon, Italy, Advances in Water Resources.
Rybczyk, J. M., J. C. Callaway, and J. W. Day Jr (1998), A relative elevation model for a subsiding coastal forested wetland receiving wastewater effluent, Ecological Modelling, 112(1), 23-44.
Saltelli, A., P. Annoni, I. Azzini, F. Campolongo, M. Ratto, and S. Tarantola (2010), Variance based sensitivity analysis of model output. Design and estimator for the total sensitivity index, Comput Phys Commun, 181(2), 259-270.
Saunders, C. J., J. P. Megonigal, and J. F. Reynolds (2006), Comparison of belowground biomass in C-3- and C-4-dominated mixed communities in a Chesapeake Bay brackish marsh, Plant Soil, 280(1-2), 305-322.
Savenije, H. H. (2015), Prediction in ungauged estuaries: An integrated theory, Water Resour Res, 51(4), 2464-2476.
Schile, L. M., J. C. Callaway, J. T. Morris, D. Stralberg, V. T. Parker, and M. Kelly (2014), Modeling Tidal Marsh Distribution with Sea-Level Rise: Evaluating the Role of Vegetation, Sediment, and Upland Habitat in Marsh Resiliency, Plos One, 9(2).
Seabloom, E. W., P. Ruggiero, S. D. Hacker, J. Mull, and P. Zarnetske (2013), Invasive grasses, climate change, and exposure to storm-wave overtopping in coastal dune ecosystems, Global Change Biol, 19(3), 824-832.
107
Short, A. D., and P. A. Hesp (1982), Wave, Beach and Dune Interactions in Southeastern Australia, Mar Geol, 48(3-4), 259-284.
Silvestri, S., M. Marani, and A. Marani (2003), Hyperspectral remote sensing of salt marsh vegetation, morphology and soil topography, Phys Chem Earth, 28(1-3), 15-25.
Silvestri, S., A. Defina, and M. Marani (2005), Tidal regime, salinity and salt marsh plant zonation, Estuar Coast Shelf S, 62(1-2), 119-130.
Soegaard, H. (1999), Fluxes of carbon dioxide, water vapour and sensible heat in a boreal agricultural area of Sweden - scaled from canopy to landscape level, Agr Forest Meteorol, 98-9, 463-478.
Stallins, J. A., and A. J. Parker (2003), The influence of complex systems interactions on barrier island dune vegetation pattern and process, Ann Assoc Am Geogr, 93(1), 13-29.
Syvitski, J. P. M., et al. (2009), Sinking deltas due to human activities, Nat Geosci, 2(10), 681-686.
Tambroni, N., and G. Seminara (2006), Are inlets responsible for the morphological degradation of Venice Lagoon?, J Geophys Res-Earth, 111(F3).
Tobias, M. M. (2015), California foredune plant biogeomorphology, Phys Geogr, 36(1), 19-33.
Tucker, C. J. (1979), Red and Photographic Infrared Linear Combinations for Monitoring Vegetation, Remote Sens Environ, 8(2), 127-150.
Turner, R. E., E. M. Swenson, and C. S. Milan (2000), Organic and inorganic contributions to vertical accretion in salt marsh sediments, Concepts and Controversies in Tidal Marsh Ecology, 583-595.
Turner, R. E., C. S. Milan, and E. M. Swenson (2006), Recent volumetric changes in salt marsh soils, Estuar Coast Shelf S, 69(3-4), 352-359.
Ursino, N., S. Silvestri, and M. Marani (2004), Subsurface flow and vegetation patterns in tidal environments, Water Resour Res, 40(5).
USACE-TEC, and JALBTCX (2014), High resolution LiDAR Data for Hog Island, VA, 2011. Virginia Coast Reserve Long-Term Ecological Research Project Data Publication knb-lter-vcr.232.3 (http://www.vcrlter.virginia.edu/knb/metacat/knb- lter-vcr.232.3/lter ).
108
Vandermaarel, E., R. Boot, D. Vandorp, and J. Rijntjes (1985), Vegetation Succession on the Dunes near Oostvoorne, the Netherlands - a Comparison of the Vegetation in 1959 and 1980, Vegetatio, 58(3), 137-187.
Walker, I. J., J. B. R. Earner, and I. B. Darke (2013), Assessing significant geomorphic changes and effectiveness of dynamic restoration in a coastal dune ecosystem, Geomorphology, 199, 192-204.
Wang, C., M. Menenti, M. P. Stoll, A. Feola, E. Belluco, and M. Marani (2009), Separation of Ground and Low Vegetation Signatures in LiDAR Measurements of Salt- Marsh Environments, Ieee T Geosci Remote, 47(7), 2014-2023.
Wang, H. J., C. J. Richardson, and M. C. Ho (2015), Dual controls on carbon loss during drought in peatlands, Nat Clim Change, 5(6), 584-U145.
White, D. A., and J. M. Trapani (1982), Factors Influencing Disappearance of Spartina Alterniflora from-Litterbags, Ecology, 63(1), 242-245.
Wolner, C. W. V., and L. J. Moore (2010), Topographic GPS profiles of Hog and Metompkin Island Cross-shore Transects of the Virginia Coast Reserve 2010. Virginia Coast Reserve Long-Term Ecological Research Project Data Publication knb-lter-vcr.198.15 (http://www.vcrlter.virginia.edu/knb/metacat/knb-lter- vcr.198.15/lter ).
Wolner, C. W. V., L. J. Moore, D. R. Young, S. T. Brantley, S. N. Bissett, and R. A. McBride (2013), Ecomorphodynamic feedbacks and barrier island response to disturbance: Insights from the Virginia Barrier Islands, Mid-Atlantic Bight, USA, Geomorphology, 199, 115-128.
Woodwell, G. A., P. H. Rich, and C. S. A. Mall (1973), Carbon in Estuaries.
Xin, P., J. Kong, L. Li, and D. A. Barry (2013), Modelling of groundwater–vegetation interactions in a tidal marsh, Advances in Water Resources, 57, 52-68.
Young, D. R., S. T. Brantley, J. C. Zinnert, and J. K. Vick (2011), Landscape position and habitat polygons in a dynamic coastal environment, Ecosphere, 2(6).
Young, I. R., and L. A. Verhagen (1996), The growth of fetch limited waves in water of finite depth .1. Total energy and peak frequency, Coast Eng, 29(1-2), 47-78.
Yousefi Lalimi, F., and M. Marani (2016), Organic Matter and Decomposition Rate Observational Data from Salt Marshes of North Carolina.
109
Yousefi Lalimi, F., S. Silvestri, L. J. Moore, and M. Marani (2017), Coupled topographic and vegetation patterns in coastal dunes: Remote sensing observations and ecomorphodynamic implications, Journal of Geophysical Research: Biogeosciences, 122(1), 119-130.
Zarnetske, P. L., S. D. Hacker, E. W. Seabloom, P. Ruggiero, J. R. Killian, T. B. Maddux, and D. Cox (2012), Biophysical feedback mediates effects of invasive grasses on coastal dune shape, Ecology, 93(6), 1439-1450.
Zedler, J. B., and S. Kercher (2005), Wetland resources: Status, trends, ecosystem services, and restorability, Annu Rev Env Resour, 30, 39-74.
Zervas, C. (2004), North Carolina bathymetry/topography sea level rise project: determination of sea level trends, NOAA Technical Report NOS CO-OPS 041.
Zinnert, J. C., S. A. Shiflett, J. K. Vick, and D. R. Young (2011), Woody vegetative cover dynamics in response to recent climate change on an Atlantic coast barrier island: a remote sensing approach, Geocarto Int, 26(8), 595-612.
110
Biography
Fateme Yousefi Lalimi was born in 1984, Sari, Iran. Prior to starting her PhD at
Duke, she earned her Master of Science degree in applied mathematics from Sharif
University of Technology in 2010 with a focus on optimization theory, and her Bachelor of Science degree in applied mathematics from University of Semnan in 2008.
At Duke University, Fateme was supported by the Duke Graduate Student
Training Enhancement Grant in 2016 to spend a semester in Padova, Italy, in order to advance her research on modeling salt marsh evolution. She was also awarded the Duke
Wetland Center Graduate Student Research Grant in 2015 to design and implement a field research in salt marshes of North Carolina.
Fateme has published her research on ecomorphology of coastal dunes in a peer- reviewed journal: Yousefi Lalimi, F., Silvestri, S., Moore, L.J., and Marani, M. (2017),
Coupled Topographic and Vegetation Patterns in Coastal Dunes: Remote Sensing
Observations and Ecomorphodynamic Implications, Journal of Geophysical Research-
Biogeosciences, 122, 119–130, doi:10.1002/2016JG003540. She also has submitted her research on variability of organic matter in salt marshes to the same journal that is currently under review: Yousefi Lalimi, F., Silvestri, S., D’Alpaos, A., Roner, M., and
Marani, M., The Spatial Variability of Organic Matter and Decomposition Processes at the Marsh Scale, Journal of Geophysical Research-Biogeosciences.
111