Modeling at MEERC

The following provides an overview of ecological modeling activities done in conjunction with MEERC research. We briefly describe how this modeling research was coordinated within MEERC, and how MEERC modeling has interacted with other research projects. We also mention how these models may have influenced resource management activities, and we provide descriptions of how two selected MEERC models have contributed directly to Chesapeake Bay management programs. In Table 1, we list 14 distinct modeling projects that were initiated by MEERC investigators to address a range of scientific and applied questions. These models are organized into three groups, which we refer to as: 1) process simulations, 2) spatial models, and 3) fish models.

Process simulations are numerical models of ecological interactions that emphasize dynamic variations in flows of matter in food-webs and biogeochemical processes (e.g., Lauenroth et al. 1998, Kremer et al. 2000). Unless otherwise specified, these models generally represent spatially averaged conditions in a volume of water or sediment surface or both. These particular models tend to focus more on ecosystem-level processes rather than population- or community-level ecological interactions. The pelagic predation model (#6, PPM) is an exception in that it simulates copepod (Acartia tonsa) population dynamics in a highly simplified stage-structured model in the context of interactions with food (algae) and predators (anchovy) and nutrient cycling. The six process models range in scope from simulations of trophic interactions and nutrient cycling in pelagic-benthic mesocosms (#1, SEE), to two models of submersed aquatic vegetation (SAV) production and nutrient cycling (#2, SEM; #3, CSE), to models of intertidal marsh production in relation to biogeochemistry (#4, MEM) and in response to disturbance by fire (#5, MFM). The structure of SEE, which was derived from an earlier model (Bartleson and Kemp 1991, Kemp and Bartleson 1991), has been used to help develop and modify the current water quality model used for nutrient management in Chesapeake Bay (Cerco 2000). Use of SEE as a tool for scaling mesocosms and extrapolating experimental results to nature is described and discussed in a forthcoming paper (Chen et al. 2001). The two SAV/seagrass models developed in association with MEERC (Madden and Kemp 1996, Kemp et al. 1995a, b, Murray et al. 2000, Bartleson et al. 2001) have also contributed to an expanded version of the water quality management model to include nutrient and turbidity effects on SAV (Cerco 2000). Furthermore, these models have provided an initial structure for current modeling of Florida Bay seagrasses in relationship to climatological events and hydrological management (Madden et al. 2001). The marsh models have been effectively used to aid in design of mesocosm experiments (e.g., Zelenke and Madden 1996) and to investigate effects of fire disturbance on ecosystem function (e.g., Schmitz 2000).

Spatial models and related tools have been developed and applied in MEERC research to address questions of spatial scaling for pelagic and SAV-dominated habitats. These models typically use simple versions of numerical structures developed for process simulations and place them in a two- or three-dimensional spatial array to compute how ecosystem processes vary in time and space. These models have been applied in MEERC studies to examine how ecosystem dynamics depend on spatial scales in nature and to address questions of extrapolating from the small scales of experimental systems to the larger scales of natural ecosystems. An initial pilot project produced a generic platform for constructing spatial models and analyzing their scale-dependent behavior (#7, SMS) provided a "proof of concept" that set the stage for two current modeling studies. The first of these (#8, PLM) is examining how spatial heterogeneity is produced from physical and biological processes in pelagic systems. This model will also use individual-based methods to simulate spatial movements of fish schools through the water column (e.g., Kemp et al. 2001a). This model will be used to investigate (among other things) how spatially homogeneous versus heterogeneous ecosystems respond differently to perturbations (simulating mesocosms), and how ecological effects differ from pulse versus press predation by schooling fish. Spatial distributions of plankton and water movement in this model can also be driven by and compared with those derived from statistical models of field data on plankton frequency distributions (#9, SAP). A spatial SAV model (#10, SSM) is being used to consider how plant beds modify their local water circulation and nutrient concentrations depending on their size and patchiness, and how these effects modulate plant susceptibility to eutrophication stress.

Although simple representations of fish schools have been added to spatial models of ecosystem dynamics in MEERC, more detailed analyses of fish populations and assemblages have also been examined with a range of other fish models. An individual-based model of fish behavior in enclosures (#11, FBE) was developed to analyze potential artifacts associated with studies of fish in experimental enclosures. Provocative results provide widely applicable lessons for design and interpretation of fish experiments (Heath and Houde 2001). Bioenergetic models developed for bay anchovy (#12, FBM) have been used to design experiments in terms of maximum fish densities, and to interpret experimental results in terms of fish growth and predation pressure on copepods. More detailed analyses of factors regulating fish feeding have been accomplished with the use of foraging models (#13, FFM). Algorithms in all three of these fish models have been useful in the development of a process simulation model to examine top-down and bottom-up interactions in experiments on anchovy predation (#6, PPM). A trophic network model was also developed under the auspices of MEERC (#14, TNM) to consider, among other things, the extent to which rates and cycles of material flows and trophic level designations depend on the conceptual scale (level of detail included) of the model (Abarca 2000).

Versions of many of these models have already been applied by state and federal agencies for use in practical problems of managing wastes and living resources in Chesapeake Bay and elsewhere. As indicated earlier, the equation structures used in our SAV models (e.g., Kemp et al. 1995a, b, Madden and Kemp 1996, Bartleson et al. 2001) have been used in revising the Chesapeake Bay water quality management model to include seagrass interactions with nutrients and turbidity (Cerco 2000). This was accomplished through EPA-sponsored close coordination between scientists and engineers through countless meetings and email exchanges. It has led to a productive relationship that has persisted for a decade or more. Other spin-offs from this modeling effort include developments of seagrass models for management of Florida Bay resources (Madden et al. 2001). We have further applied components of these SAV models into an algorithm that uses data collected in the Chesapeake Bay monitoring program to calculate the suitability of different Bay regions for SAV growth and survival (Batiuk et al. 2000; Kemp et al. 2001). Here, the model helps to compute the light attenuation associated with nutrient-stimulated growth of epiphytic algae on SAV leaves under variable conditions of depth, water exchange rate, and herbivorous grazing (Kemp and Bartleson 2001). These two applications combine to produce a powerful approach to using monitoring data to assess habitat conditions and using a large numerical model to forecast changes in SAV distribution and abundance under different nutrient loading regimes. In addition, we have worked with the same Bay modelers to incorporate some of the algorithms developed in MEERC-related process simulations that improve representation of trophic interactions and nutrient cycling (Cerco 2000). Particular attention has been paid to the problem of correctly simulating the production and biomass of phytoplankton and the partitioning of plankton community respiration. We have also used these simple models to explore how trophic transfer efficiencies change with nutrient enrichment in pelagic ecosystems (Kemp et al. 2001b).

We believe that the MEERC project has been rich with many modeling activities involving a diversity of methods and approaches to address a range of topics involving the theory and application of scaling relationships in coastal ecosystems. There has been abundant interactions among modeling projects and between modeling and empirical research, leading to a strong program. Many of these models have had direct utility in bringing scientific knowledge to bear toward effective resource management in Chesapeake Bay and other coastal ecosystems around the world.

Abarca, L. G. 2000. Comparing networks and food web analyses: Ecosystem patterns, trends and scale considerations. PhD Dissertation, University of Maryland, College Park.

Bartleson, R. D. and Kemp, W. M. 1991. Preliminary ecosystem simulations of estuarine plankton-benthic interactions: The planktonic submodel. Pages 243-252 in J. A. Mihursky and A. Chaney (eds.). New Perspectives in the Chesapeake Bay System. Chesapeake Research Consortium Publication, Solomons, MD.

Batiuk, R. A. and 15 others. 2000. Chesapeake Bay submerged aquatic vegetation water quality and habitat-based requirements and restoration targets: A second technical synthesis. EPA Chesapeake Bay Program Report, Annapolis, MD (available at http://www.chesapeakebay.net/pubs/sav/index.html).

Cerco, C. F. 2000. Chesapeake Bay eutrophication model. Pages 363-404 in J. E. Hobbie (ed.). Estuarine Science. Island Press, Washington, DC.

Chen, C.-C., J. E. Petersen and W. M. Kemp. 2001. Effects of spatial scale on the dynamics of experimental planktonic-benthic ecosystem: A simulation model. Ecol. Model. (in preparation).

Heath, M. R. and E. D. Houde. 2001. Pages xx-xx in R. H. Gardner, W. M. Kemp, V. S. Kennedy and J. E. Petersen (eds.). Scaling Relations in Experimental Ecology. Columbia University Press, New York. (In press).

Kemp, W. M. and R. D. Bartleson. 1991. Preliminary ecosystem simulations of estuarine plankton-benthic interactions: The benthic submodel. Pages 253-264 in J. A. Mihursky and A. Chaney (eds.). New Perspectives in the Chesapeake Bay System. Chesapeake Research Consortium Publication, Solomons, MD.

Kemp, W. M. and R. Bartleson. 2001. Epiphyte contributions to light attenuation and availability for submersed plants: Model estimates of water quality effects. Aquat. Bot. (in review).

Kemp, W. M., R. Batiuk, and 15 others. 2001. Habitat requirements for submerged aquatic vegetation in Chesapeake Bay: Water quality, light regime, and physical-chemical factors. Estuaries (in preparation).

Kemp, W. M., W. R. Boynton and A. J. Hermann. 1995a. Simulation models of an estuarine macrophyte ecosystem. Pages 262-278 in B. Patten and S. E. Jørgensen (eds.). Complex Ecology. Prentice Hall, Englewood Cliffs, NJ.

Kemp, W. M., W. R. Boynton and A. J. Hermann. 1995b. Ecosystem modeling and energy analysis of submerged aquatic vegetation in Chesapeake Bay. Pages 28-42 in C. A. S. Hall (ed.). Maximum Power. University Press of Colorado, Miwot, CO.

Kemp, W. M., M. T. Brooks and R. R. Hood. 2001b. Nutrient enrichment, habitat variability and trophic transfer efficiency in simple models of pelagic ecosystems. Mar. Ecol. Prog. Ser. (In press).

Kemp, W. M., J. E. Petersen and R. H. Gardner. 2001a. Scale-dependence and the problem of extrapolation: Implications for experimental and natural coastal ecosystems. Pages xx-xx in R. H. Gardner, W. M. Kemp, V. S. Kennedy and J. E. Petersen (eds.). Scaling Relations in Experimental Ecology. Columbia University Press, New York (in press).

Kremer, J. N., W. M. Kemp, A. Giblin, I. Valiela, S. Seitzinger and E. Hofmann. 2000. Linking biogeochemical processes to higher trophic levels. Pages 299-341 in J. E. Hobbie (ed.). Estuarine Science. Island Press, Washington, DC.

Lauenroth, W., C. Canham, A. Kinzig, K. Poiani, W. M. Kemp and S. Running. 1998. Simulation modeling in ecosystem science. Pages 404-415 in M. Pace and P. Groffman (eds.). Successes, Limitations and Frontiers in Ecosystem Science. Springer, New York.

Madden, C. J., D. Gruber and W. M. Kemp. 2001. Growth and survival of the seagrass, Thalassia testudinum, in shallow coastal waters of South Florida: A model of key ecological processes. Ecol. Model. (in preparation).

Madden, C. J. and W. M. Kemp. 1996. Ecosystem model of an estuarine submersed plant community: Calibration and simulation of eutrophication responses. Estuaries 19(2B): 457-474.

Murray, L., R. B. Sturgis, R. Bartleson, W. Severn and W. M Kemp. 2000. Scaling submersed plant community responses to experimental nutrient enrichment. Pages 241-258 in S. Bortone (ed.). Seagrasses: Monitoring, Ecology, Physiology, and Management. CRC Press, Boca Raton FL .

Scheurer et al. 2001. Ph.D. Thesis, University of Maryland, College Park (in preparation).

Schmitz, J. P. 2000. Meso-scale community organization and response to burning in mesocosms and a field salt marsh. M.S. Thesis, University of Maryland, College Park

Zelenke, J. and C. J. Madden. 1996. Simulation model of biogeochemical processes in marsh mesocosms, pp. In: W. M. Kemp et al. (eds.) Ecosystem models of Chesapeake Bay relating nutrient loading, environmental conditions and living resources. Report to U.S. EPA, Annapolis MD.

Table 1. Preliminary listing of ecological models developed within the framework of MEERC, including the investigators involved and interactions within and beyond the MEERC Project.

1) Scaling Experimental
Ecosystems (SEE)

Petersen, Chen, Kemp

PPM, PLM

EPA / Chesapeake Bay Program

SEE is a 2-sector (upper, lower layers) model of ecological processes and properties in water column and sediments and on walls of Pelagic-Benthic (PB) mesocosms, including variables for phytoplankton, benthic algae, wall periphyton, O₂, DIN, POC, DOC, bacterioplankton, zooplankton. It has been used to interpret results of nutrient addition studies, and to extrapolate from experiments to conditions without walls.

SEM is a spatially averaged submerged aquatic vegetation (SAV) model which was calibrated using field and mesocosm data and used to interpret results from studies of water residence time, nutrient loading frequency, and top-down control of eutrophication. It has also been linked to a PB model (predecessor of SEE) for analysis of multicosm (multi-habitat mesocosms) experiments.

CSE is a spatially averaged dynamic model of the interactions of nutrients, light, suspended particulates and SAV (P. perfoliatus, Z. marina) growth in mesohaline Chesapeake Bay. It is articulated into a 3-sector model representing 3 depths (corresponding to the EPA restoration framework), and tracking nutrient exchanges between the target shallow littoral zone and mid-channel Bay.

MEM is a 2-sector (high marsh and low marsh) model of the MEERC marsh mesocosms that incorporates information on experimental additions of nitrate via groundwater, and the processing of nitrogen by the marsh mesocosm. Components and processes include variable rates of N addition, groundwater flow, tide schedules, vegetation complexity and harvest rates.

MFM is a simple model of marsh plant biomass and detritus, designed to interpret empirical results on interannual patterns of marsh growth in mesocosms and to explore differences in responses to fire in experimental and natural ecosystems.

PPM is a pelagic ecosystem simulation model designed to interpret and extrapolate experimental results from PB mesocosm studies of top-down control. The model includes variables for phytoplankton, inorganic nutrients, copepods (adults, copepedites, juveniles) and bay anchovy.

SMS is a spatial modeling platform written in C++ and designed to simulate plankton dynamics across a (expandable-collapsible) spatial grid with individual planktivorous fish moving according to behavioral rules in schools across the grid and interacting with the plankton variables.

PLM is a 2-dimensional spatial model (resolution 1m, extent at least 1km) of pelagic ecosystems (NPZD formulation) that characterizes the spatial variation of biota due to interactive effects of nutrient availability and trophic dynamics.

SAP uses spectral analysis and geo-statistics to characterize spatial patterns observed in continuous records of planktonic chlorophyll-a, temperature, salinity, zooplankton and oxygen for perspective on experimental scales and to calibrate spatial numerical models.

SSM is a spatial model of SAV interactions with physical circulation in shallow estuarine environments, focusing on how plant structure alters drag coefficients and other physical parameters. This model will be coupled with SEM to examine how spatial patterns of SAV distribution influence plant-nutrient interactions and plant susceptibility to light and nutrient stress.

Behavioral patterns of planktivorous and piscivorous fish are simulated with FBE to evaluate and understand effects of enclosure in mesocosms on feeding success and fish growth.

Potential food consumption by bay anchovy in all life stages (larvae through adult) is estimated with FBM to assist in design of experiments and interpretation of results.

Interactions between fish predators and plankton prey are simulated with FFM to evaluate effects of fish on plankton communities.

Food-web structures supporting fish production are represented with TNM to compute primary production to support fish growth, and to analyze the web in terms of material cycling and through-put.

-------------
^a See list of models (1-4) for definition of acronyms.
^b Process simulations are numerical models of ecological interactions that emphasize dynamic variations in flows of matter in food-webs and biogeochemical cycles.
^c Spatial models include numerical simulations with finite-difference or finite-element methods that describe variations of ecological entities in time and space.
^d Fish models are systems of equations simulating or computing bioenergetics, behavior, and/or population dynamics of fish or food-webs that support fish production.

MULTISCALE EXPERIMENTAL ECOSYSTEM RESEARCH CENTER

Ecological Modeling in MEERC