rev. 5/22/2008
REPORT
of the
ECOSYSTEM MODELING WORKSHOP #3
Tampa, Florida
May 6-7, 2008
Table of Contents
Abbreviations Used in This Document 2
Introduction 2
Alternative Ecosystem Models 2
EDOM – Carl Walters 2
Using an IBM to Evaluate the Predictions from Foraging Arena Theory – Ken Rose 3
Spatial Ecosystem Model to Assess Fishery and Habitat Risks for Fish and Fisheries in the Gulf of Mexico – Jerry Ault 3
Nutrient Loading in Tampa Bay: Mediation Effects and Bottom-Up Processes in Ecosim – David Chigaris 4
Fresh water – linkage between upland forest and coastal environment – Wei Wu 5
Examination of MPA Scenarios 5
Potential for Application of Marine Reserves in the Recreational Snapper Fishery – Jim Bohnsack 5
Islands in the Stream: A Concept for the Gulf of Mexico – Brian Keller 6
Where Do We Go From Here? SSC Recomendations 8
List of Presentations 9
Workshop Agenda 9
Workshop Participants 11
List of Preparers 12
Abbreviations Used in This Document
AEAM Adaptive environmental assessment and management
CONANP Comison Nacional de Areas Naturales Protegidas (Mexican commission on protected
areas)
EDOM quilibrium Delay-Difference Optimization Model
FGB Flower Garden Banks
FTP File transfer protocol
FWRI Florida Fish and Wildlife Research Institute
IBM Individual-based model
LUMCON Louisiana Universities Marine Consortium
MBRS Mesoamerican Barrier Reef System
MMS Minerals Management Service
MPA Marine protected area
SEFSC Southeast Fisheries Science Center
SSC Scientific and statistical Committee
Introduction
This workshop was a continuation of a series of workshop held on May 8-10, 2007 and September 24-26, 2007 to evaluate whether existing ecosystem models can provide useful advice to the Council about key ecosystem management questions. As with the previous workshop, this workshop was led by Dr. Carl Walters. One focus of this workshop was to examine the capability of ecosystem models to evaluate the effect of MPAs. Participants included the Ecosystem SSC, Ad Hoc Marine Reserves SSC, invited experts on ecosystem modeling, and other interested persons. A complete list of participants appears at the end of this summary.
Alternative Ecosystem Models
The previous workshops focused on developing and evaluating an Ecopath with Ecosim model of the Gulf of Mexico developed by Carl Walters and Behzad Mahmoudi. In this workshop, other models or model applications were looked at that were being developed by workshop participants. Brief descriptions of each of the models discussed follows.
EDOM – Carl Walters
Carl Walters gave a presentation on an Equilibrium Delay-difference Optimization Model (EDOM) that he co-developed to assist California in evaluating MPA policies for a network of marine reserves. The presentation applied the model to the Gulf of Mexico. In practice, the model takes a raster-based image of the Gulf of Mexico and aggregates data into spatial strips for population dynamics predictions. Each strip is treated as a home range for the subpopulation of larval fish that settle in the strips nursery areas (then typically move offshore as they grow). A habitat suitability matrix for juveniles of multiple species is incorporated into the model, as well as the population dynamics of each species included in the model. Displaced fishing effort from an MPA strip is assumed to move into cells that are open to fishing and are more profitable, based on prices, catch rates and costs per unit effort. The model can be optimized to project MPA closure patterns that would result in maximum total profit or sustained fishing effort. The main uses of EDOM to date have been in policy gaming, where MPA options are suggested and quickly evaluated for impact on sustainable fishery profits and fishing efforts. One such example is shown on the cover of this report. The raster image of the Gulf of Mexico is turned on its side, and EDOM was rum to optimize both abundance and catch of red snapper. The optimized result was a network of 16 cross-shelf MPAs covering 24% of the available habitat. The model can currently evaluate only cross-shelf reserves, and is limited by a lack of hard bottom habitat mapping.
Using an IBM to Evaluate the Predictions from Foraging Arena Theory – Ken Rose
Ken Rose was unable to make it to the workshop due to airline delays, but his presentation on an application of individual based models (IBM) was presented by Carl Walters. In classical predator-prey relationships, prey is assumed to be continuously available to predators, and the amount of prey eaten rises in direct proportion to the predator density. In Foraging Arena Theory, it is assumed that prey behavior (such as hiding) limits prey availability to predators. Even as the predator density increases, only those prey organism that become available are eaten. Thus, the amount of prey eaten will level off even as predator density increases. To test this theory, an IBM model was constructed of a Louisiana tidal march community consisting of six species, grass shrimp, blue crab, sheepshead minnow, Gulf killifish, inland silverside and bay anchovy. The tidal marsh model was divided into a 100 x 100 grid, within which the model on an hour by hour basis several processes for each individual organism, including feeding, growth, reproduction, and movement. In addition, habitat conditions such as dissolved O2, temperature, tidal stage, prey and predator density, and individual size were also modeled. Individual predator and prey organism movements were modeled based on a process in which cells adjacent to the one currently occupied by the organism were evaluated by a number of factors to produce a fitness potential score, with the organism subsequently moving into the cell with the highest score.
Simulations were run in which the mortality of predators was increased or decreased, and in which the density of benthic organisms was decreased. While the results to date are not yet conclusive, shrimp appear to conform the predictions from Forage Arena Theory, whereas crabs appear to show true depensatory mortality. Additional runs incorporating random movements of organisms will be run to compare to the results where movements are determined by the fitness potential score, in order to determine if the results are strong enough to be conclusive.
Spatial Ecosystem Model to Assess Fishery and Habitat Risks for Fish and Fisheries in the Gulf of Mexico – Jerry Ault
Jerry Ault presented a multispecies spatial ecosystem model to assess risks for fish and fisheries from a geospatial dynamics approach. Dynamics of the model were divided into several layers:
- human impact layers (exploitation, water quality, MPAs)
- predator layers (tarpon, sharks, billfish, snapper)
- prey layers (menhaden, mullet, sardines, shrimp)
- hydrodynamics layers (currents, salinity, temperature, dissolved oxygen)
- habitat layers (bathymetry, benthic substrate)
The model integrates these layers through a series of multispecies conservation equations, and tracks cohorts of prey and predators at age, time and space as the spawn, settle grow and age. Examples of some of the results include:
- The Gulf ecosystem can be describes as a highly dynamic coupled biophysical environment. Shelf waters (less than 30 fathoms) accounts for more than 70% of fish landings.
- The ecosystem is under significant anthropogenic and environmental stress.
- An ecosystem approach should protect habitats that support a range of productivity.
Nutrient Loading in Tampa Bay: Mediation Effects and Bottom-Up Processes in Ecosim – David Chigaris
Often times, a third party organism interferes with predator prey interactions that occur in an ecosystem. This interference is here referred to as mediation and can either facilitate or provide protection from predation. In addition, a mediation effect may occur on primary producers to increase or reduce productivity. These mediations can be demonstrated using an Ecopath with Ecosim (EwE) model of Tampa Bay. The model is linked (using a forcing function) to a hydrographic time series that represents nutrient loading in Tampa Bay over the past 50 years. The general trend of nutrient loading in Tampa Bay indicates a peak in the late 1960s to early 1970s when pollution in the bay was high. Since the early 1980s a clean-up effort has been in place and nutrient loads into the bay have been reduced. The model indicates strong bottom-up processes related with nutrient loading showing that increased nutrient input results in increased phytoplankton biomass, which results in growth of zooplankton biomass, thereby providing more food to zooplanktivorous fishes that are heavily preyed upon by piscivores. However, historical trends of abundance do not appear to follow this pattern. This exercise demonstrated a mechanism for Ecosim to represent indirect effects that ecosystem processes have on foraging behavior and the ability to test hypotheses using this application.
First, a mediation effect was introduced using phytoplankton as the mediating group on seagrass productivity to represent the “shading” of sunlight by phytoplankton (and thereby dissolved organic material). Once this interaction was accounted for, the model provided better fits to historical seagrass abundances. Secondly, to represent the effect that increased phytoplankton may have on water clarity thus affecting the ability of visual predators to locate prey, a mediation effect was applied to mackerel that reduces the vulnerability of their prey when phytoplankton biomass was high. This allowed to model to estimate the low biomass values of the 1970s indicated by historical time series derived from stock assessments. Finally, it is generally known that seagrass provides refuge for many estuarine dependent species. Therefore, a mediation effect was included to represent how increased biomass of seagrass will reduce the vulnerability of spotted seatrout to their predators. This mediation effect caused the model to provide better fits to historical abundance estimates where seatrout and seagrass biomass was low during the 1970s and 1980s.
Fresh water – linkage between upland forest and coastal environment – Wei Wu
Overfishing and habitat destruction are two of the biggest concerns in fishery management. Large areas of earth’s surface are linked to the ocean by freshwater, and therefore, habitats in marine ecosystem can be damaged by terrestrial inputs through freshwater. A well-known example is the Gulf of Mexico hypoxia zone. During the SSC modeling workshop on May 6-7, Wei Wu from the University of Southern Mississippi presented her work in modeling fresh-water quantity and quality at the upland forest ecosystems under land use / land cover change and climate change, which has important implications for marine habitats. Different statistical modeling approaches were introduced, including generalized linear mixed model and geographically weighted regression, which applies in modeling response variables that belong to exponential family such as occurrence data. Process models linking atmospheric deposition, vegetation, soil and surface water were presented in the application of modeling evapotranspiration and water quality under global changes. In addition, she presented a fully Bayesian hierarchical modeling approach to combine data modeling and process modeling, which can account for uncertainties from observations, parameters and model structures, to derive a full predictive distribution of water quantity under climate change, which is deemed to provide a basis for more informed sustainable water resource management. She gave an example in modeling the habitat and control of invasive species using multi-scale analysis combining landscape modeling and population demographic modeling. She emphasized the need of considering cross-system stressors in selecting marine protection areas.
Examination of MPA Scenarios
Ecosystem model applications to marine reserves included evaluation of the EDOM model (discussed above) and a mathematical model to illustrate the potential effect of marine reserves on age distribution, spawning biomass, and yield from the red snapper fishery with assumptions of low, medium or high migration patterns out of the reserve (discussed below). The SSC also received a presentation describing the Islands in the Stream concept, but sis not attempt to model it. Previous workshops modeled small offshore marine reserves using Ecosim or Ecospace, and concluded that such reserves were too small to have a measurable impact on managed stocks.
Potential for Application of Marine Reserves in the Recreational Snapper Fishery – Jim Bohnsack
Jim Bohnsack and coauthor Brian Linton (NMFS, Miami) gave a Powerpoint presentation examining the feasibility of using marine reserves as a tool to extend the red snapper recreational fishing season over the short-term and increase long-term yield. Marine reserves act to both increase average of capture and reduce total fishing mortality. Advantages of marine reserves include no removals of targeted fishery species and zero bycatch mortality within the protected area (if fully enforced); protection of genetic quality of stocks from size-selective fishing; and increased supply of fish in fishing grounds from a spillover effect of some larger fish that leave reserves and increased supply of fish larvae that disperse from reserves to fishing areas. Compared to a minimum size limits alone, marine reserves would allow the stock to grow over a wide range of ages, whereas a minimum size limit tends to compress to age distribution to a small number of younger age groups.
The presentation demonstrated an MPA model in which 20% of all representative habitat in the Gulf (and initially 20% of the fish) were protected by marine reserves. Movement of fish into or out of the MPA was either low (25% of the fish move), medium (50% move) or high (75% move). Results were projected to the year 2032 and compared to status quo projections based on the 2003 red snapper assessment. Lower movement produced a large increase in the proportion of older (age 6+) snapper and total spawning stock relative to the status quo projection. Under medium or high movement, increases in older fish and spawning stock still occurred, but to a much lesser extent than under the low movement scenario. Long-term benefits of marine reserves were much greater in the western than eastern Gulf of Mexico in terms of proportional differences in size structure, absolute abundance, and fishery yield. These results are based on a relatively simple simulation study and a more complex study should be conducted if marine reserves are adopted as part of the red snapper management strategy.
In support the validity of the low movement assumption, recapture data from tagged red snapper at artificial reefs off Mississippi showed high site fidelity and low movements (maximum a few km) in years without hurricane disturbance. However, dispersal distances were orders of magnitude higher for red snapper tagged before and recaptured after hurricane Opal. Hurricane disturbance appears to cause pulsed dispersal more similar to the medium movement model and leads to the prediction that hurricane disturbance significantly reduces the protection benefits marine reserves for red snapper in the northern Gulf of Mexico.