The coupled biophysical interactions between submerged aquatic vegetation (SAV), hydrodynamics (currents and waves), sediment dynamics, and nutrient cycling have long been of interest in estuarine environments. Recent observational studies have addressed feedbacks between SAV meadows and their role in modifying current velocity, sedimentation, and nutrient cycling. To represent these dynamic processes in a numerical model, the presence of SAV and its effect on hydrodynamics (currents and waves) and sediment dynamics was incorporated into the open-source Coupled Ocean–Atmosphere–Wave–Sediment Transport (COAWST) model. In this study, we extend the COAWST modeling framework to account for dynamic changes of SAV and associated epiphyte biomass. Modeled SAV biomass is represented as a function of temperature, light, and nutrient availability. The modeled SAV community exchanges nutrients, detritus, dissolved inorganic carbon, and dissolved oxygen with the water-column biogeochemistry model. The dynamic simulation of SAV biomass allows the plants to both respond to and cause changes in the water column and sediment bed properties, hydrodynamics, and sediment transport (i.e., a two-way feedback). We demonstrate the behavior of these modeled processes through application to an idealized domain and then apply the model to a eutrophic harbor where SAV dieback is a result of anthropogenic nitrate loading and eutrophication. These cases demonstrate an advance in the deterministic modeling of coupled biophysical processes and will further our understanding of future ecosystem change.
Submerged aquatic vegetation (SAV), or seagrass, includes rooted vascular plants that inhabit sediments of estuaries and coastal waters, with a wide global distribution. SAV involves important primary producers in shallow environments and provides a habitat for a number of aquatic organisms; it can slow water velocities and dampen wave energy to trap particulate material (Carr et al., 2010), as well as alter biogeochemical cycles through oxygenation of sediments (Larkum et al., 2006). The positive impact of ecosystem services provided by SAV presence has been well studied (Hemminga and Duarte, 2000; Nixon et al., 2001; Terrados and Borum, 2004; McGlathery et al., 2007; Hayn et al., 2014). The growth of SAV is dependent upon light availability at the leaf surface, which is a function of light attenuation in the water column and the biomass of epiphytic algae growing on SAV stems. During the last several decades, the loss of SAV has accelerated due to anthropogenic pressures (Kennish et al., 2016) or natural causes such as storms (Hamberg et al., 2017). One of the dominant factors of SAV loss is eutrophication through nutrient loading, exemplified by increased phytoplankton growth and epiphytic growth on vegetation. This results in a reduction of light availability (Burkholder et al., 2007), causing a loss of SAV habitat (Cabello-Pasini et al., 2003; Short and Neckles, 1999).
The complex interactions between light availability, nutrient loading, SAV dynamics, hydrodynamics, and sediment transport can be investigated using numerical modeling tools. Few modeling efforts have attempted to couple the effects of hydrodynamics and light availability to model the growth of SAV. Everett et al. (2007) and Hipsey and Hamilton (2008) coupled the effects of chlorophyll and water to account for SAV variability, while Bissett et al. (1999a, b) used spectral underwater irradiance to model the light availability required for SAV growth. Carr et al. (2012a, b) developed a one-dimensional coupled hydrodynamics, sediment, and vegetation growth dynamics model. The model solved for vertical 1-D dynamics of SAV growth through a change in biomass that depended on water temperature, irradiance, and seagrass properties. Ganju et al. (2012) used a three-dimensional circulation model (Regional Ocean Modeling System; ROMS) coupled to a nutrient–phytoplankton–zooplankton–detritus (NPZD) eutrophication (water-column biogeochemistry model) developed by Fennel et al. (2006) and integrated the spectral light attenuation formulation (bio-optical model) provided by Gallegos et al. (2011). These models were linked to a benthic seagrass model to calculate seagrass distribution (Zimmerman, 2003) and applied on the temperate estuary of West Falmouth Harbor (del Barrio et al., 2014). While the model was able to capture the loss of SAV due to insufficient light, it did not include interactions with epiphytes or exchanges with water-column nutrients and gas pools. The hydrodynamic feedbacks (changes in currents and waves) and morphodynamic changes (sediment distribution) due to presence of SAV were also ignored. While these dynamic processes have significant implications for coastal ecosystem resilience, numerical models that allow for the two-way feedbacks between hydrodynamics, sediment transport, and SAV growth and nutrient cycling have generally been lacking.
Recently, Beudin et al. (2017) implemented the physical effects of SAV in a vertically varying water column through momentum extraction and vertical mixing as well as accounting for wave dissipation due to vegetation. These processes were implemented within the open-source Coupled Ocean–Atmosphere–Wave–Sediment Transport (COAWST) modeling system that couples the hydrodynamic model (ROMS), the wave model (Simulating WAves Nearshore; SWAN), and the Community Sediment-Transport Modeling System (CSTMS) (Warner et al., 2010). Through this effort, the COAWST framework accounted for the coupled seagrass–hydrodynamics interactions. The model reproduced the turbulent shear stress across the canopy interface and peaked at the top of the canopy similar to the observations of Ghisalberti and Nepf (2004, 2006). The presence of a seagrass patch led to a reduced shear-scale turbulence within the canopy and enhanced wake-scale generated turbulence. For more details on the impact of seagrass on hydrodynamics, readers are referred to Beudin et al. (2017). The inclusion of the physical effects of SAV on flow and sediment dynamics (Beudin et al., 2017) in COAWST allows us to develop a framework that results in dynamic growth of SAV using the temperature, nutrient loading, and light availability in the water column. Therefore, in this work, we implement a SAV growth model that dynamically changes the SAV properties (stem density and height). The growth of SAV is modeled as biomass which includes the aboveground biomass (stems and shoots), belowground (roots and rhizomes) biomass, and epiphyte biomass. Individual biomass equations described in the implementation of SAV growth model (Sect. 2.2) are based on previous SAV biomass models (primarily Madden and Kemp, 1996), some of which have been previously implemented to simulate growth conditions for SAV in three-dimensional numerical model simulations (e.g., Cerco and Moore, 2001). The change in biomass leads to a spatial and temporal variation of SAV density and height. With the inclusion of the SAV growth model, SAV can experience growth or dieback while contributing and sequestering nutrients from the water column (modifying the biological environment) and subsequently affect the hydrodynamics and sediment transport (modifying the physical environment). Conversely, a change in the physical environment, for instance, the amount of sediment in the water column, can decrease light availability and cause SAV dieback, leading to reduced wave attenuation, increased sediment resuspension, and a further decrease of light availability.
We demonstrate the two-way biophysical coupling framework as follows: the SAV growth model and integration into COAWST are discussed in Sect. 2; in Sect. 3, the model setup for the idealized domain and a realistic simulation of West Falmouth Harbor, Massachusetts, are described; in Sect. 4, we present the results from the two model configurations along with a discussion of limitations of the current modeling work; and in Sect. 5, we summarize our work and outline areas of future research.
Beudin et al. (2017) implemented the parameterizations that accounted for the presence of SAV within a coupled hydrodynamic and wave model within the open-source COAWST numerical modeling system (Warner et al., 2008). The COAWST framework utilizes ROMS for hydrodynamics with a wave model (SWAN) via the Model Coupling Toolkit (MCT), generating a single executable program (Warner et al., 2008). ROMS is a three-dimensional, free surface, finite-difference, terrain-following model that solves the Reynolds-averaged Navier–Stokes (RANS) equations using the hydrostatic and Boussinesq assumptions (Haidvogel et al., 2008). The transport of turbulent kinetic energy and generic length scale is computed with a generic (GLS) two-equation turbulence model. SWAN is a third-generation spectral wave model based on the action balance equation (Booij et al., 1999). In ROMS, the presence of SAV extracts momentum, adds wave-induced streaming, and generates turbulence dissipation. Similarly, the wave dissipation due to vegetation modifies the source term of the action balance equation in SWAN. All of these subgrid-scale parameterizations account for changes due to vegetation in the water column extending from the bottom layer to the height of the vegetation. SWAN only accounts for wave dissipation due to vegetation at the bottom layer. The coupling between the two models occurs with an exchange of water level and depth-averaged velocities from ROMS to SWAN and wave fields from SWAN to ROMS after a desired number of time steps. The vegetation properties are separately input in the two models at the beginning of the simulations. Through these changes, the SAV can affect the bottom stress calculations that determine the resuspension and transport of sediment, providing a feedback loop between SAV–sediment dynamics–hydrodynamics and wave dynamics. To account for sediment dynamics, CSTMS (Warner et al., 2010) is used to track the transport of suspended-sediment and bed load transport under the action of current and wave–current forcing. The model can represent an unlimited number of user-defined sediment classes.
The SAV growth model is primarily based upon a previous growth model developed and implemented in Chesapeake Bay by Madden and Kemp (1996). The model simulates the temporal dynamics of aboveground biomass (AGB) that consists of stems or shoots, and the belowground biomass (BGB) that consists of roots or rhizomes. In addition to AGB and BGB, epiphytic algal biomass (EPB) is simulated to account for reductions in light availability to plant leaves due to shading of SAV leaves by epiphytes under high nutrient loading conditions. AGB, BGB, and EPB are simulated as total biomass per unit area, with nitrogen as the currency for biomass. Changes in AGB and BGB pools are simulated as a function of primary production and respiration, mortality (e.g., grazing), and nitrogen exchange through the seasonal translocation of nitrogen between roots and shoots. EPB is modeled as a function of primary production, respiration, and mortality.
SAV growth model variable descriptions.
SAV growth model input parameter descriptions and their values.
The remaining section describes the source terms that calculate the
evolution of AGB, BGB, and EPB, denoted by
The primary
production of
The availability of photosynthetically active radiation (PAR)
represented by mathematical symbol
SAV respiration terms are partitioned into active and
basal respiration, where the active respiration term represents respiration
that is dependent on the photosynthesis rate, and the basal rate represents
maintenance respiration rate. The active respiration term is defined as
The mortality of SAV is computed separately for
aboveground and belowground biomass, where aboveground mortality (
Belowground mortality,
The translocation from roots/rhizomes to leaves (upward translocation) is
modeled as a simple linear function of belowground biomass (denoted by
The EPB is computed similarly to SAV biomass by simulating EPB as a function of primary production, respiration, and mortality (e.g., grazing).
The primary production of
EPB depends on the maximum potential growth rate (
The light availability (
Epiphyte respiration terms are partitioned into active
and basal respiration, where the active respiration term represents
respiration that is dependent on the photosynthesis rate, the basal rate
represents the maintenance respiration rate. The active respiration term is
defined as
The basal respiration term is defined as
The mortality of epiphytes depends on mortality and grazing
of algal cells, as well as losses associated with SAV sloughing (which
effectively removes epiphytes from a cell). The mortality term is given as a
simple linear form:
The AGB computed in the SAV growth model is utilized
to obtain SAV shoot height (meters) and stem density (stems m
The SAV growth model is built to interact dynamically with the water-column
biogeochemistry model (BGCM)
within the COAWST modeling framework. We
utilize one of the existing BGCM models developed by Fennel et al. (2006)
that accounts for nutrients (
Each state variable is calculated based on the tracer transport equation
with tracer concentrations calculated at the grid cell centers as follows:
The changes in water-column variables (dissolved and particulate nitrogen,
dissolved oxygen, dissolved inorganic carbon) due to the SAV growth model
occur locally at the bottom cell through the source terms (
The addition of the SAV growth model leads to the biological evolution of SAV properties based on temperature, light, and nutrient availability. The modeled SAV community exchanges nutrients, detritus, dissolved oxygen, and dissolved inorganic carbon with the water-column BGCM. Changes in SAV biomass and canopy characteristics also impact hydrodynamics, wave dynamics, and sedimentary dynamics (resuspension–transport). By lowering the current speed and attenuation of wave flow, the reduction in bed shear stresses in the vegetation canopy reduces sediment resuspension, thereby altering sediment transport in the model (as described in Sect. 2.1), which uses the feedback to control light availability and, in turn, potential seagrass biomass production. This methodology of including the SAV growth model enables the COAWST framework to have a two-way feedback in hydrodynamic–biological coupling. Figure 1 describes the coupling process between different modules schematically.
Schematic showing the coupling of SAV growth module implementation in the COAWST model.
Empirical relationships between aboveground biomass and SAV shoot
height for
Planform view of the idealized test domain simulation.
The implementation of the SAV growth model within the COAWST framework is
first tested on an idealized domain. The test case consists of an idealized
rectangular domain of 9.2 km width and 9.8 km length with a 1 m deep basin.
The number of interior domain points is 90 in the
The model is forced by oscillating the water level on the northern boundary
with a tidal amplitude of 0.25 m and a period of 12 h. Northern boundary
conditions include a water temperature variation between 1.5 and
18.5
In their study, del Barrio et al. (2014) used an offline coupling of the COAWST model with a
bio-optical seagrass model (Zimmerman et al., 2003) to study the influence
of nitrate loading and sea-level rise on seagrass presence/absence in West
Falmouth Harbor, Massachusetts, USA. Nitrate concentrations in groundwater
exceeded 200
Planform view of
Planform view of
Time series of
Magnitude of bottom stress
Simulations of the coupled hydrodynamic–biogeochemical–SAV model revealed
the integrated nature of estuarine dynamics in response to submerged
macrophytes. In these simulations, suspended sediment concentration
(SSC) was imposed at the northern open
boundary at concentrations of 0.5
Mean over 22 d of
The temporal evolution of SAV biomass in response to the SSC input at the
northern boundary further emphasizes the self-stimulating role of SAV in the
idealized simulations. A comparison of model simulations at two locations
within the initially described SAV bed of the idealized domain (indicated in
Fig. 5a and corresponding to
As mentioned above, the SSC input on the northern boundary of the idealized
domain causes a region of suboptimal light conditions that lead to the SAV
dieback, while the SAV growth occurs in the remaining bed where favorable
light conditions exist. The effect of change in SAV density and height on
the hydrodynamics and morphodynamics at the end of the simulation can be
demonstrated by using the same idealized domain. To this end, two transects
are chosen that are along the length of the SAV bed and extend from the
northern boundary towards the southern boundary. The transects are chosen
inside at
Time series of
The simulation of the idealized domain demonstrates the capability of the modeling framework to perform two-way feedbacks between hydrodynamics, sediment, and biological dynamics. The SSC input in the northern boundary affects the light attenuation in the domain and causes SAV dieback close to the northern boundary. The SAV grows in the region where favorable light conditions exist. The SAV dieback leads to increased bottom stresses, while the growth of SAV leads to a decrease in bottom stresses, illustrating the fact that the SAV acts as bottom sediment stabilizer by reducing SSC.
The present-day simulation of seagrass dynamics reproduces the patterns of
chlorophyll (via phytoplankton), light attenuation, and near-bottom PAR
simulated by del Barrio et al. (2014). Nitrate loading from shoreline point
sources led to increased phytoplankton growth indicated by increased
chlorophyll and light attenuation in the landward, northeast portion of the
harbor (Fig. 8a, b), while bathymetric controls in the deeper central basin
led to decreased near-bottom PAR (Fig. 8c). Peak AGB exceeds 100 mmol N m
Change in outcomes between impacted and non-impacted scenarios
(nitrate loading scenario – no loading scenario). Difference in mean over
22 d of
Time series of these parameters (Fig. 9) from selected outer and inner harbor locations over the first 22 d demonstrate the diurnal variability, as well as the rapid loss of AGB in the inner harbor due to the local nitrate loading, phytoplankton proliferation, and degraded light climate. The sizable diurnal variability in AGB (Fig. 9d) appears to be an artifact of production/respiration formulations that are based on seasonal responses to environmental forcing, rather than diurnal responses to solar irradiance. Future modifications could attenuate this variability by utilizing daily averaged environmental forcing or modifying the frequency of biomass updating.
The modeling framework developed in this work can be used to create hypothetical scenarios to estimate future environmental responses. For example, we ran the model setup of West Falmouth Harbor described in Sect. 3.2 with no nitrate loading to simulate a hypothetical scenario where the groundwater input has no influence from the wastewater treatment plant (non-impacted past or future scenario). The elimination of nitrate loading results in negligible changes in the outer harbor but greatly reduces chlorophyll and light attenuation in the inner harbor (Fig. 10a, b), while increasing the near-bottom PAR (Fig. 10c). Peak AGB responds to the decreased chlorophyll and increased light attenuation with an increase in the inner harbor (Fig. 10d). This implementation represents an incremental improvement to the prior modeling exercise (Ganju et al., 2012; del Barrio et al., 2014), because the interactions between SAV and the nitrogen pools are explicitly accounted for. For example, this model can now be used to test how changes in seagrass coverage influence nitrogen retention within the estuary or export to the coastal ocean. Further, the introduction of seagrass kinetics will allow for investigation of water-column oxygen budgets with and without seagrass under present and future scenarios.
Modeled
In order to qualitatively evaluate the seagrass growth model, we have
compared the modeled results with observations by del Barrio et al. (2014)
that measured the extent of seagrass coverage in West Falmouth Harbor (red
outline in Fig. 11). The field data are only available for the northern
region of West Falmouth Harbor where the model–data comparisons are performed. The model
results are compared by extracting the peak AGB on
14th day of the simulation and normalized with the initial aboveground
biomass. The ratio of
Direct estimates of aboveground SAV biomass have also been recently made in
West Falmouth Harbor (Hayn et al., unpublished data). Although these
measurements were not made during the same year as our simulations
(measurements in 2006, 2007, 2013; model 2010), the mean aboveground
biomass measured in the outer harbor of 49.5 (21 June–6 July 2006), 45.3
(6–19 June 2007), and 41.5
While this modeling approach represents an advance in modeling coupled
biophysical processes in estuaries, there are limitations that must be
addressed in future work:
The modeling of SAV dieback/growth scenarios may require long-term
simulations on decadal timescales (Carr et al., 2018). However, the short
model time step limits the duration of such simulations. The time step size
is of the order of seconds (typical of 3-D ocean models) and this, combined
with the fact that the presence of SAV in the hydrodynamic model further
limits time step size (due to hydrodynamic stability constraints), overall
limits the applicability of the model to be utilized from monthly to annual
timescales at this juncture. The biomass equations described in Sect. 2.3 are formulated for
seasonal timescales and are being used in the model implementation at every
ocean model time step. This leads to large daily variations in above- and
belowground biomass that likely do not occur in the environment, although
diel variations on SAV growth have been measured in situ (Kemp et al., 1987).
Hence, with the current formulations, the output from the biomass model
needs to be analyzed as a daily averaged quantity. The current implementation of the SAV growth model is limited to only one
SAV species. However, it should be extended to include multiple SAV species
to investigate competition under variable salinity and to make the model
applicable to a wider variety of locations.
The present study adds to the open-source COAWST modeling framework by implementing a SAV growth model. Based on the change in SAV (aboveground, belowground) biomass and epiphyte biomass, SAV density and height evolve in time and space and directly couple to three-dimensional water-column biogeochemical, hydrodynamic, and sediment transport models. SAV biomass is computed from temperature, nutrient loading, and light predictions obtained from coupled hydrodynamics (temperature), biogeochemistry (nutrients), and bio-optical (light) models. In exchange, the growth of SAV sequesters or contributes nutrients from the water column and sediment layers. The presence of SAV modulates current and wave attenuation and consequently affects modeled sediment transport and fate. The resulting modeling framework provides a two-way coupled SAV–biogeochemistry–hydrodynamic and morphodynamic model. This allows for the simulation of the dynamic growth and mortality of SAV in coastal environments in response to changes in light and nutrient availability, including SAV impacts on sediment transport and nutrient, carbon, and oxygen cycling. The implementation of the model is successfully tested in an idealized domain where the introduction of sediment in the water column (SSC) at one end of the domain provides suboptimal light conditions that cause SAV dieback in that region. The model was applied to the temperate estuary of West Falmouth Harbor, where simulations show the coupled effect of enhanced nitrate loading in the inner harbor leading to poor light conditions for the SAV to grow, thus modeling the physical effect of eutrophication leading to the loss of a SAV habitat. Among other applications, in the future, the model will be used assess the effects of sea-level-rise scenarios that limit light availability and potentially cause the loss of SAV habitats.
The implementation of the SAV growth model has been conducted within COAWST v3.4. This particular version is available for download at
The major code development that was done for this project is contained
within the COAWST folder on the following path:
This folder contains several methods of computing water-column biogeochemistry. Other than the I/O component of our implementation, the algorithmic development in this study only modifies two files on this path: “estuarybgc.h” and “sav_biomass.h”. The file “sav_biomass.h” contains all the newly added equations for the growth of SAV based on the nutrient loading in the water column. The forcings to the SAV growth model (temperature, light, nutrient availability, exchanges nutrients, detritus, dissolved inorganic carbon, and dissolved oxygen) are provided through the file “estuarybgc.h” which calls “sav_biomass.h”. The file “estuarybgc.h” solves for the water-column biogeochemistry and was based on existing modeling framework developed by Fennel et al. (2006) (also coded as “fennel.h”).
Other important paths that existed in the framework prior to the current
modeling effort but are being used in the modeling process include the following:
The main kernel of the 3-D non-linear Navier–Stokes equations is contained
within this part and links all the submodels: biological, vegetation, and
sediment models; The kernels that account for seagrass–hydrodynamics interactions can be found here: The kernels that account for sediment transport can be found here:
The model data were released as per the USGS model data release policy, and
separate digital object identifiers were created as part of the release
(
TSK implemented the SAV growth model in the COAWST framework. JMT provided guidance on the mechanistic processes affecting the growth of SAV from biomass parameterizations and developed linkages between the SAV growth model and the water-column biogeochemical model. NKG developed the test case and the realistic domain case. TSK and NKG performed the data analysis from the output of the test cases and were responsible for model data release. The paper was prepared with contributions from all co-authors.
The authors declare that they have no conflict of interest.
Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the US Government.
We thank Joel Carr at the US Geological Survey Patuxent Wildlife
Research Center in Beltsville, Maryland, for providing his feedback to
improve the quality of the paper. We thank the reviewers for their careful
reading of our manuscript and their insightful comments and suggestions. We
thank Alfredo Aretxabaleta for valuable discussions through the length
of the project. We would also like to thank Melanie Hayn, Robert Howarth,
Roxanne Marino, and Karen McGlathery for sharing
This paper was edited by David Ham and reviewed by Jon Hill and David Ham.