(Held January 13-14, 1998)

Mike Durako
Jay Zieman

PMC Representative
Mike Robblee

December 10, 1998

The overall Florida Bay Program strategy for restoration of the Bay is to develop linked models simulating the dynamics of the Bay and its response to upstream water management. Linked models include circulation, with connection to upstream hydrology, water quality (algal bloom, turbidity, nutrients), seagrass/macrophyte community, and a series of consumer models representative of secondary production within the Bay. Models will be used conceptually to guide research efforts and to identify critical monitoring and research needs. Ultimately, the models will be used to predict the effects of various water management scenarios on the function of Florida Bay. Predictions, effectively the hypothesized response of the Bay to water management operations, will be evaluated (tested) with short- and long-term monitoring. In an iterative fashion water management operations will be modified based on observations and periodic evaluation. This scientific information and modeling base is an essential components for the restoration of Florida Bay "to a naturally functioning ecosystem."

The Strategic Plan for the Interagency Florida Bay Science Program (March 1997) is built around five central questions. Central Question #4:

explicitly deals with the seagrass communities within Florida Bay, although seagrass communities play a role in all of the central questions. The widespread die-off of seagrasses, initiating in 1987, and the resulting changes in the Florida Bay ecosystem are the fundamental reasons behind most of the present research initiatives in the Bay.

The approach being taken to accomplish the goals of the Strategic Plan has been to establish a Florida Bay Scientific Oversight Panel (FBSOP) comprised of senior scientists with significant experience in major estuarine restoration programs, but without involvement in Florida Bay Projects. The Panel has arranged for ad hoc advisory panels of experts in specialized subject areas organized around the five central questions. The advisory panels (or teams) have been charged with the organization of a series of workshops where critical research issues and questions are addressed (e.g., circulation modeling, nutrients, water quality modeling, etc.). The workshop summarized herein was convened to initiate discussions and develop a set of research priorities toward the successful development of a seagrass community model for the Florida Bay ecosystem. The workshop was held over January 13-14 (see attached agenda) and included an advisory panel comprised of Dr. Susan Williams (San Diego State University, Pacific Estuarine Research Lab), a member of the FBSOP, Dr. Kenneth Dunton (University of Texas, Marine Science Institute), Dr. Kenneth Heck (Marine Environmental Sciences Consortium, Dauphin Island Sea Lab), and Dr. Michael Kemp (University of Maryland, Horn Point Laboratory). This panel's responsibility was to make recommendations to the Program Management Committee (PMC) and the FBSOP regarding the best strategies for accomplishing the development and verification of a seagrass community model for Florida Bay.

The members of the Florida Bay seagrass research team responsible for organizing the workshop were: PMC representative Dr. Michael Robblee (USGS/BRD), Team Leaders Dr. Michael Durako (University of North Carolina at Wilmington, Center for Marine Science Research) and Dr. Joseph Zieman (University of Virginia, Environmental Sciences Department), and Team Members Dr. Tom Armentano (Everglades National Park), Dr. Penny Hall (Florida Department of Environmental Protection, Florida Marine Research Institute), and Dr. Christopher Madden (South Florida Water Management District).

Historically seagrasses were found on over 95% of the bottom of Florida Bay. Following the loss of several thousand hectares of seagrasses during the initial and continuing seagrass dieoff episodes, the resulting turbidity plumes and algal blooms showed the stabilizing influence of the previous, nearly continuous, seagrass cover. This was recognized by the Williams Ad Hoc Review panel, whose first summary recommendation was that:

The topics of discussion of the workshop centered on the why and how issues which the workshop organizing committee hoped to address in the workshop. The following considerations were central to how the modeling efforts were to be organized.

• Seagrass communities are the dominant marine habitat in Florida Bay and the FKNMS.

• Seagrasses influence water circulation and quality, sediment characteristics and stability, and higher tropic functions throughout the shallow marine waters of south Florida.

• Loss of seagrasses, especially the dominant species Thalassia testudinum (turtle grass), would be catastrophic to natural communities (e.g. coral reefs) and the economy of south Florida.

• Because of their importance to the South Florida Ecosystem and to the economy of the region we need to have predictive capabilities to evaluate management strategies as they affect seagrass communities and their ecologic function.

The purpose of the seagrass modeling effort is to simulate the effects of changing physical and biogeochemical conditions on the growth and survivorship of seagrasses in a tropical/subtropical carbonate-based system. Specifically, the resulting models will hopefully simulate seagrass community succession and will be used as a tool to explore short- and long-term hypotheses on seagrass die-off in Florida Bay. A seagrass unit model will be incorporated within a landscape model framework and linked to process-level models of higher trophic levels. The effort to develop the seagrass models will include empirical studies needed to develop information for the models, evaluation of the assumptions of the models, and the calibration and verification of model outputs. The models will be used in the restoration program for predicting the effects of water management within a landscape model framework.

Because the seagrass modeling effort is linked to, and part of, the overall Scientific Program for the Restoration of Florida Bay, and this Program is organized around five central research and modeling areas. The workshop organizing committee discussed the importance of model linkages and produced the following tabular organization of how the models being developed should be integrated. This table was also used as a basis for organizing the workshop and in developing a list of participants and issues.

A sediment model is mentioned specifically here because the effects it incorporates are essential although they may actually be a part of one or the other of the above models.

With the above background the workshop focused on the following three goals:

• Determine the required elements for seagrass models for Florida Bay.

• Produce a prioritized listing of experimental research, field research, and monitoring needs.

• Develop a detailed timeline for accomplishing these tasks

The workshop agenda and presentations were developed to achieve the goals stated above using the collective knowledge of invited panel members, topic speakers, and audience participation (see attached agenda). Because of the short time period for the workshop, the extended background presentations and discussions on the first day, and suggestions from the panel and the participants, the workshop adopted an "Adaptive Agenda" format. The result was an expansion of the time allotted for discussion of specific strategies, data needs, and priorities for the development of a Florida Bay Seagrass Model.

The workshop consensus was that three model types are required for this effort:

1. Species-specific Unit Models
2. Landscape Model
3. Demographic Models

Initially two species-specific Unit Models should be developed describing Thalassia testudinum (turtle grass) and Halodule wrightii (shoal grass). Subsequent models should then be developed for Syringodium filiforme (manatee grass), Ruppia maritima (widgeon grass), and macrophytic algae (i.e., drift red macroalgae, Batophora/Dasycladus, calcareous green algae, etc.). The unit models should include individual species physiological (salinity, temperature, light tolerances), architectural (shoot size, above:below-ground biomass allocation, branching patterns, canopy, etc., demographic (sexual versus vegetative reproduction, mortality, recruitment potential), and growth (especially areal coverage) characteristics, as well as other considerations such as competitive relationships. The models may essentially be viewed as carbon (C) balance models, and will consider both above and below-ground production and turnover. While initially focusing on carbon, the unit and landscape models they will be required to deal with phosphorus and nitrogen dynamics, as nutrient supply and limitation varies widely across Florida Bay. The scaling for these models will address individual short shoot to 1m² scales.

The Landscape Model(s) will incorporate physical and environmental conditions and variations across the Bay and how they influence seagrass distribution and abundance changes by integrating the results of the unit models at a larger scale. The unit models will run in the cells of the landscape model which should address scales of from 10 m²- to-hectare, and then up to basin-to-bay scale. This concept will result in predictions of the spatial response, over time, of each seagrass and, collectively, a representation of the seagrass community to physical and environmental conditions across the seascape of Florida Bay. The link to water management will be made by comparing the predicted response of seagrasses to physical and environmental conditions in Florida Bay as modified by the predicted change in freshwater flow into the Bay (quantity, quality, timing, and distribution) associated with various water management alternatives.

The strategy discussed by the workshop organizing committee was to develop these models on parallel tracks. The Unit Models will initially focus on Thalassia and Halodule. These modeling efforts will be followed, as time and funding allow, by the development of additional unit models. The research team also felt that the Landscape Model should initially be run in representative areas in Florida Bay (e.g., Rabbit Key or Johnson Basin for the west, Rankin Lake or Whipray Basin for the central core, and Eagle Key Basin or in the vicinity of Duck Key in eastern Florida Bay) rather than being developed solely at a Bay-wide scale. Additionally, the team felt that a control area is needed outside the influence of water management and continental water inflow influence (e.g., the 'lakes' area west of Key West or some site on the Atlantic side of Key Largo or Tavernier). To promote program integration and to assist in model validation/calibration, the reference basins and control site should also serve as areas in which research and monitoring efforts will be concentrated.

The mode and rate of growth of seagrasses can vary widely across Florida Bay. Some areas are known to produce fruits and seeds freely, others areas only produce seeds sporadically, and some areas have never been demonstrated to seed. As a result the relative role of vegetative growth and spreading as opposed to areas increase by propagules is poorly known in Florida Bay. Carbon balance models do not readily address issues of population ecology, and so a set of Demographic Models is necessary.

A chronic weakness in all discussions regarding model development and data needs was the (unanswered) question - What is the restoration goal for Florida Bay? Not having a defined answer to this most fundamental question severely undermined the ability to focus the discussions on what was needed from the seagrass model? From the viewpoint of seagrass ecology, targets can be set to reproduce the known seagrass communities in Florida Bay at previous times. The Florida Bay seagrasses were first systematically mapped in the early 1980's, and sufficient data exists to provide targets for the modeling effort during this period of clear water and stable conditions. Similarly there is an abundance of data from the period 1988 to 1998, including detailed water quality. Thus, until a management goal related to seagrasses is developed and articulated, the initial goal of the Seagrass Modeling Program will be to produce models that describe the abundance and dynamics of seagrasses, at selected locations in Florida Bay, in the period of 1980 to 1998.

At this time the most developed and tested seagrass unit models are those developed for Zostera marina for the Chesapeake Bay Program. These models, developed primarily by Richard Wetzel and colleagues at the Virginia Institute of Marine Sciences, are both mechanistic and modular. While carbon based models, they also incorporate nutrient dynamics and epiphytism of the seagrass leaves, both important characteristics in Florida Bay also. While the parameterizations for Zostera from the Chesapeake would need reworking, the basic model structure and modules could serve as a base for development of a south Florida Thalassia model.

Currently, another seagrass model is being developed in south Texas by Adrian Burd and Kenneth Dunton. This seagrass model was developed for USACoE's Waterways Experiment Station for Laguna Madre, Texas, and incorporates unit models for Thalassia and Halodule in a subtropical setting. While these models target the south Florida species, and are carbon balance models, they are driven primarily by physical factors (PAR, temperature, salinity), and do not respond to limiting nutrients. Thus, while the Chesapeake model would have to be modified to reflect the P limitation known to occur in south Florida as opposed to the primarily N limitation further north, the Laguna Madre model would require the addition of both nutrient and epiphyte limitation. The Texas model would be faster to implement in the short term, and clearly could be used initially by the landscape model. Modification of the more physiologically explicit Chesapeake model might to be a more efficient path for the full development of the unit model, however.

Discussions on modeling strategies and data needs were initiated following the comments and recommendations from Rich Batiuk based on his experience and perspectives gained from his participation in the Chesapeake Bay Program. His comments were primarily directed toward the administrative elements of successfully developing the model. He emphasized the importance of directed funding based on program goals and he encouraged the use of RFP's to force integration and a team approach to the directed research questions. The integrated teams should be comprised of participants from both universities and government agencies because both groups bring different perspectives to the research programs. Principal investigators should be accountable for producing the outputs that are specified in proposal scopes of work. The most important administrative element for ensuring the success of a large research program is the development of a realistic (i.e., adequately funded) overall work plan for achieving the milestones within the program.

Florida Bay is a vast mosaic of basins and banks, with over 40 identifiable basin complexes. While there is considerable variation in physical, chemical, and ecologic factors across the bay, there are groups of bank-basin complexes that are representative of large sections of the bay. Further there are several of these complexes that have been studied since 1983, with intense data collection from 1989-90 to the present. There was a general consensus regarding the focusing of data collection, experimental, and modeling efforts in "research" basins in five representative regions of the Bay: Western Bay (Rabbit Key or Johnson Key), Central Bay (Rankin Lake or Whipray), Eastern Bay (Eagle or Nest), southern Bay (Barnes and Arseniker Keys), and Transitional Bays (Little Madeira or Joe). In addition there needs to be a site outside of Florida Bay on the seaside of the keys (Tavernier Key region) outside of Florida Bay influence. By concentrating the limited resources and within the time constraints of the available funding, restricting our efforts to these areas should provide for more rapid progression toward the verification process.

Models are designed to answer defined questions and test specific hypotheses. During discussions on strategies and data needs, several questions and hypotheses were presented as elements that can be used to focus the model development:

Why are there seagrasses in Florida Bay

What precipitated seagrass die-off in Florida Bay?

What factors are responsible for current seagrass change in Florida Bay (Thalassia ¯ the west, Halodule ­ ¯ )?

A critical element in the development of hypotheses that may drive the type of model which is developed is that they have to be testable. This is where emperical data and manipulative experiments can be used to verify the output and test the assumptions of the models.

Examples of testable hypotheses:

1. The changes that are taking currently place in Florida Bay are no different than historical changes.

This hypothesis could be evaluated by assessing historical data, imagery, and observations.

2. Where light is not limiting, Thalassia testudinum growth is limited by salinity, temperature, nutrients, and/or H₂S.

These hypotheses could be tested by a combination of directed monitoring and manipulative experiments.

3. Stressed Thalassia testudinum is more susceptible to disease.

This hypothesis could also be evaluated by field and laboratory experiments.

The unit models should be spatially explicit. Several approaches to unit models were discussed. They represent three hierachical scales:

a) physiological at the shoot scale: C-balance ® Thalassia, Halodule, later on Syringodium, Ruppia, macroalgae.

These models depend on an understanding of factors controlling photosynthesis versus respiration, photosynthesis versus irradiance, shoot size (LAI), biomass allocation (root:shoot).

b) physiological and architectural at the shoot-to-plant (ramet-to-genet scale): metabolic ® growth.

These models consider carbon fixed, biomass allocation (LAI, root:shoot), and turnover (i.e., carbon lost).

c) demographic at the genet-to-population scale: population dynamics -ss/m²/y, apicals, productivity.

These models consider recruitment versus mortality. They should be data driven (empirical). Existing FHAP data and FIU WQ data may be used in this effort to investigate possible forcing relationships.

Specific existing models that were discussed included the Fong model and the Wetzel (and related Madden/Kemp) model. The new Laguna Madre WES seagrass model may be another very applicable alternative. This model should be available within a couple of months.

Discussion on the required inputs to the unit model included the following:

a) Light - quantity

b) Salinity - mean & variance

c) Temperature - mean & variance

d) Disease - relation to physical environment

e) Nutrients - limitation/excess C:N:P

f) Competition/inhibition species interaction

g) Herbivory

h) Sedimentation

There is, at least, some existing information available on most of these input elements. However, much of the information is limited in scope. There was a consensus that the highest data need priority was for continuous light monitoring data within the four research basins.

Examples of inputs to the unit model about which very little is currently known:

a) k_d = k_water + k_TSS +k_{phytoplankton} + k_CaCO3

This equation refers to how the total light attenuation coefficient (light quantity) is partitioned among contributions from water, total suspended particles, phytoplankton, and resuspended fine carbonate sediments. How much each of these components contributes to the total loss of light in the water column is very important in terms of possible management scenarios. This relates to the second important aspect of the light field in Florida Bay - light quality (PUR). Because phytoplankton contain chlorophyll, they preferentially absorb the wavelengths of light that are also required to support seagrasses. Thus, their effect may be proportionally more important than non-wavelength specific attenuation by particles. Resuspended carbonate muds may actually result in little light reduction due to their high albedo.

b) PAR:PUR

This refers to light quantity versus light quality (photosynthetically active radiation versus photosynthetically usable radiation action spectra). Very little is known about either of these parameters in the Bay. And no one at the workshop was aware of any data on photosynthetic action spectra of any seagrass. This could be important if pigments or Gelbstoff are important contributors to the total k_d.

c) Benthic N & P fluxes

Although some information exists regarding sediment N & P flux dynamics, almost nothing is known regarding species-specific nutrient uptake kinetics or affinities (K_s) in leaves versus roots for the seagrasses in Florida Bay.

d) Factors Affecting Photosynthesis versus Irradiance

There are P versus I data for the seagrasses in the Bay, but little data are available on the effects of salinity, temperature, or H₂S levels on these characteristics. Quantifying these relationships will increase the robustness of the unit models.

e) H₂S toxicity

There are data that show that increases in H₂S can kill Thalassia, but almost nothing is known about the relative susceptibility of the other seagrass and macroalgal species to H₂S toxicity.

f) Disease

The etiology of the disease is not clear. There are data on the effects of Labyrinthula infection on P vs I characteristics in Thalassia. There has been verification of Koch's Postulates for the infections. There exists a high degree of coincidence in the distributions of Labyrinthula and Thalassia losses. However, there has not been clear in situ evidence demonstrating that this organism is responsible for die-off.

The scaling of landscape models should range from 1m²-to-hectare-to-basin-to-bay. Because of the great amount of time spent discussing unit model considerations, there was relatively little discussion on data needs and priorities for the development of seagrass landscape models other than to generate the following list of important inputs.

a) Physical Forcing: Light, Salinity, and Temperature

These are similar to the unit model requirements, but the question of grid scaling becomes important in the model development.

b) Substrate Characteristics and Depth

This is especially important in Florida Bay. Sediment depths may limit community development in the northeast part of the Bay. The trend of increasing sediment depth from northeast to southwest is paralleled by increasing seagrass community development. The resuspendability of the sediments is also an important consideration because of the shallow water depths and expansive mud-bank system.

c) Basin Topography

Depth distribution is important where turbidity is high, although the depth range is small in the Bay there are dramatic differences in seagrass community structure in basin versus basin edge versus bank. There is some evidence that die-off initiated along the basin edges. Loss of seagrass cover on some of the western banks has contributed to the chronic turbidity in this region. One of the most important current management issues is how/if we can re-establish seagrasses along the western banks.

d) Basin Hydrology

Basins in the northeast part of the Bay are generally large and shallow with narrow mud banks. The size of the basin affects it's fetch and the effect of wind events on resuspension, depth is also important in the ability of wind waves to resuspend sediments. Western basins and banks are subjected to significant tides, whereas central and eastern regions have negligible tides. The basin-mud bank physiography of Florida Bay greatly restricts circulation and results in variations in residence times of water masses within basins and the effects of evaporation/rainfall on salinity characteristics. Current velocity can be high in channels through mud banks, especially in the west. Generally, there is an increasing influence of wind-driven currents from west-to-east.

e) Species Specific Factors

This is where the development of realistic unit models is critical. At the landscape levels, conversion of the carbon balances to actual growth rates and the resulting areal extent of the different macrophytes increases the scale at which the model can operate.

f) Architectural Constraints

The species-specific variations in allocation of new carbon to above versus belowground biomass greatly affects their distribution in the submerged landscape of the Bay. The different seagrasses exhibit very different growth patterns with Thalassia growing monopodially and Halodule exhibiting sympodial growth. This affects the production and density of apical meristems, which are the source of all new biomass. Little information exists on environmental affects on branching patterns or the species-specific rates and patterns of horizontal spread. Another fundamental architectural difference between Thalassia and Halodule is the size of their rhizomes and how the rhizomes function in accumulating storage reserves which can be used when environmental conditions are stressful. There are data to suggest that Thalassia is able to tolerate longer periods of low light than Halodule. Short-shoot characteristics are also quite different between these two species and the differences in short-shoot biomass, height, density and demography (longevity, recruitment and mortality) will greatly affect their landscape-scale responses to environmental changes.

g) landscape features

At what spatial and temporal scales are seagrass communities organized? These questions are important in determining the scale of the landscape model grids and the scale of sampling. We know little about the relative importance of acute versus chronic factors in controlling seagrass community dynamics.

Model structure and required inputs for particular models are largely driven by what are the desired outputs. The discussion on this issue was again constrained by what is the restoration goal for the Bay. However, the following list of outputs represents a consensus of generally desirable outputs for a Florida Bay seagrass model. The most basic model output should be some indication of the seagrass community structure. Characteristics such as short-shoot density, standing crop (or leaf area index), and above:below-ground biomass are examples of structural outputs for a unit model. The landscape model should be able to provide outputs on habitat type, such as monospecific versus mixed species beds, homogeneous versus patchy distributions, and habitat characteristics for secondary producers (benthic infauna, fish, shrimp, etc.). The seagrass models should be linked to other models to provide realistic outputs in response to interactions with sediments - how resuspension affects seagrasses and how seagrass cover affects resuspension; interactions with water quality - particularly with respect to P concentrations and availability. Because of the shallow nature of Florida Bay and the dominance of seagrasses in the landscape, the models should be able to provide outputs on the interaction of seagrasses with circulation within the Bay. There are data that demonstrate the importance of seagrass cover on holding water on vegetated banks.

The following list represents a "clearinghouse" of research questions that were put forth by the workshop participants in the context of what was need in the development of workable seagrass models for Florida Bay. Some of the topics below were discussed above in terms of required inputs into the seagrass models.

- Light ® k + Temperature & Salinity

What is the relative importance and interaction of these three physical variables on the growth and survival of seagrasses in the Bay? What are the salinity x temperature interactions on productivity?

- Monitoring network

This was previously discussed above. Essentially we need to focus our data collection, experimentations and modeling efforts to common, representative areas. By focusing our efforts we should be able to collect accurate and detailed spatial data within the test sites (research areas). We need to come up with rationale, defensible reasons for choosing the test sites and the sites chosen for seagrass research should also serve as reference areas for correlary research and modeling efforts of the other research groups.

- Partition k

This was discussed above. Determining the relative contributions of various fractions is important in affecting possible future management decisions.

- Evaporation/heat budget

This research question is very important in supporting mass balance approaches to modeling salinity patterns within the Bay. Salinity may be a primary forcing function on the seagrass unit models and it may control disease distribution (i.e., low salinity refugia).

- Nutrient limitation of growth

Is it P or N? Where and when are these limiting? How does nutrient limitation compare with light, salinity, temperature, disease in determining species distributions?

- Action spectra of photosynthesis

PAR versus PUR. As stated above, almost nothing known about this in Florida Bay.

- Whole plant Photosynthesis versus Irradiance characteristics

The data that are available for Florida Bay seagrasses has been collected in lab incubations using tissue pieces or individual short-shoots. There is a need for in situ chamber-based data collection to support and calibrate the unit models

- C-limitation

Because of the unusual geochemistry of this carbonate dominated, high pH system there is a very real possibility that CO₂ availability may be limiting to seagrass productivity in some regions of the Bay.

- C:N:P within Clones

How much clonal integration exists for distributing nutrients of highly variable spatial availability among ramets of a genet? What are the transport rates, transport distances and controls on these integration processes? Short-shootD Rhizome, Short-shootD Short-shoot?

- In situ - infection healthy versus "stressed" (ss/m², T and S)

It is accepted that Labyrinthula is somewhat ubiquitous. If it is the organism responsible for die-off, why now and what are the factors that make Thalassia susceptible to the disease. This is very important in separating the relative significance of disease versus light limitation in current losses of seagrasses in the western and central parts of the Bay.

- Top Down Control of distribution

The apparent lack of epiphytes in this now nutrient-enriched (seagrass necromass from die-off) environment is puzzling. Are epiphyte controlled by grazers?

- Sedimentation ® horizontal versus vertical growth

There are data available on the effects of sedimentation on the growth characteristics of several seagrasses, but little or no data for the species in Florida Bay or in a biogenic carbonate sediment environment. This is basically a question of how carbon is allocated from the physiological scale to the ramet/genet scale. This question could be investigated empirically or with manipulative experiments.

The discussions summarized above dominated the last day of the workshop. There was little time to collectively discuss specific timelines for the data collection, research, and modeling efforts. The existence of several similar conceptual models (Zieman, Carlson/Durako, Rudnick, etc.) provides some justification for designing monitoring and experimental projects to test specific elements of these conceptual models as a short-term goal.

In addition, there exists extensive empirical data on spatial and temporal changes in seagrass and macroalgal distribution (FHAP, DERM/SFWMD) and water quality (FUI, SFWMD) data that have not been examined to determine if any strong correlations exist. This could be done in the short-term. The expected near-future availability of the Texas seagrass model may offer an opportunity to adapt and test this model within the context of the USACoE Water Quality modeling effort and independently in the context of its use as a seagrass community model. In parallel, longer-term efforts should be initiated to develop unit and landscape models specific to the unique environment of Florida Bay.