ATLSS: ACROSS-TROPHIC-LEVEL SYSTEM SIMULATION for the Freshwater Wetlands of the Everglades and Big Cypress Swamp Draft Report By D. M. Fleming Everglades National Park Field Station National Biological Survey D. L. DeAngelis Oak Ridge National Laboratory Dept. of Mathematics and Graduate Program in Ecology Univ. of Tennessee, Knoxville L. J. Gross Dept. of Mathematics and Graduate Program in Ecology University of Tennessee, Knoxville R. E. Ulanowicz Chesapeake Biological Laboratory University of Maryland W. F. Wolff Forschungszentrum Julich, Germany W. F. Loftus Everglades National Park Field Station National Biological Survey M. A. Huston Oak Ridge National Laboratory Dept. of Mathematics and Graduate Program in Ecology Univ. of Tennessee, Knoxville PARTICIPATING AGENCIES/ORGANIZATIONS: South Florida/Caribbean Field Laboratory, NBS South Florida Natural Resources Center, NPS Institute for Environmental Modeling, UT/ORNL Chesapeake Biological Laboratory, UM Forschungszentrum Julich, Germany ATLSS Workshop Homestead, Florida June 21-22, 1994 Table of Contents I. EXECUTIVE SUMMARY 1 II. INTRODUCTION AND RATIONALE 3 III. OBJECTIVES 6 IV. OVERVIEW OF METHODS 7 V. LANDSCAPE DESCRIPTION 8 VI. MODELING APPROACHES 10 VII. AQUATIC-BASED (DETRITUS/PERIPHYTON) FOOD WEBS 11 Lower Trophic Level Models 11 Periphyton and Macrophytes 11 Detritus 12 Mesoinvertebrates 12 Macroinvertebrates 12 Intermediate Trophic Level Models 14 Selected macroinvertebrates and fish guilds 14 Higher Consumer Models 17 Colonial wading birds 17 American Alligator 26 VIII. TERRESTRIAL-BASED (MACROPHYTES AND SEEDS) FOOD WEBS 30 Landscape/Vegetation (Vascular Plants) Model 30 White-tailed Deer Model 32 Florida Panther Model 35 IX. EMPIRICAL STUDIES IN SUPPORT OF ATLSS MODEL DEVELOPMENT 38 X. FUTURE PLANS 41 XI. APPLICATION OF ATLSS TO OTHER SYSTEMS 44 XII. ACKNOWLEDGEMENTS 45 XIII. LITERATURE CITED 46 Table of Contents, continued XIV. LIST OF TABLES 53 XV. LIST OF FIGURES 60 EXECUTIVE SUMMARY The Everglades and the Big Cypress Swamp of South Florida are characterized by complex patterns of spatial heterogeneity and temporal variability. Hydrologic patterns which result from the distribution, volume, and timing of water flow are a major driving force controlling the trophic dynamics of these systems. After many decades of intense, adverse water management impacts, hydrologic restoration alternatives are now being developed and proposed. Thus there is a need to predict and to compare the future effects of alternative hydrologic restoration scenarios on the biotic components of these systems. Future projections of the effects of restoration alternatives can be realistically accomplished only by modeling. Trophic interactions are fundamentally important processes which are manifested at variable scales. Primary producers and lower trophic levels are directly driven by biochemical processes while population dynamics and individual behavior tend to dominate the dynamics of upper trophic levels. Thus the use of a single modeling method is not appropriate. ATLSS is a set of integrated models that simulate the hierarchy of whole-system responses across all trophic levels, and across spatial and temporal scales at which interactions actually occur within an ecosystem. ATLSS uses different modeling approaches for each trophic level: (1)(1)process models for lower trophic levels (periphyton and macrophytes, detritus, micro-, meso- and macroinvertebrates); (2) structured population models for intermediate trophic levels (for five functional groups of macroinvertebrates and fishes); (3) individual-based models for top-level consumers (American Alligator, colonial wading birds, Snail Kite, Cape Sable Seaside Sparrow, White-tailed deer, and Florida panther). These are integrated across a spatially explicit (grid-based) landscape model of the Everglades and Big Cypress Swamp. This landscape structure is interfaced with a landscape/vegetation model that simulates within-year and between-year variations in vegetation properties in response to changes in hydrologic conditions and photoperiod, (e.g. rates of vegetation growth and regrowth, caloric and protein content of the vegetation, etc.). Intermediate and long-term vegetation dynamics - regrowth and succession following disturbance and feedbacks between vegetation and geomorphology - will be provided by plant-community succession models currently under development. Spatial resolution at this level is as small as 100m, with the capability to vary this resolution based upon the scale of available input data. ATLSS will be coupled to hydrology models and used to assess the effects of alternative proposed hydrologic restoration scenarios on trophic structure. Simulated or projected changes in species (or taxa) population levels and geographical patterns in response to changes in hydrology and vegetation are able to be statistically analyzed, as well as formatted for display by geographical information systems (GIS). The ATLSS modeling approach will enhance the South Florida Restoration Initiative Task Force's technical capability to address scientific and management questions affecting these systems. It also provides a framework within which past and ongoing ecological research may be synthesized, further enhancing these predictive and diagnostic models that address ecosystem-level questions concerning the Everglades and Big Cypress wetlands. Future work will include refinement of existing models, development of additional models for several remaining key species and/or functional groups, model performance evaluation, integration of ATLSS model components into a common framework, and refinement of a user-friendly interface. ATLSS is being developed through a cooperative effort by the following institutions: South Florida/Caribbean Field Laboratory (at Florida International University and University of Miami), National Biological Survey; Institute for Environmental Modeling, University of Tennessee; Environmental Sciences Division, Oak Ridge National Laboratory; Chesapeake Biological Laboratory, University of Maryland System; and Institut fur Biotechnologie 3, Forschungszentrum Julich, Germany. INTRODUCTION AND RATIONALE The Everglades and the Big Cypress Swamp comprise a mosaic of terrains and drainages which traverse several physiographic sub-regions in southern Florida. This expansive landscape includes freshwater and estuarine wetlands and adjacent nearshore habitats of Florida Bay and the Gulf of Mexico. Knowledge is limited about the structure and function of these subtropical, palustrine and estuarine habitats prior to anthropogenic disturbances. Changes in environmental conditions are hypothesized to have caused the widespread declines of top-level carnivores during the past half century. General downward trends in animal assemblages may be related to alterations in the hydrologic and salinity regimes, and the resultant loss or changes in suitable habitat. Related changes in fire regimes, the invasion of non-indigenous species, and human enrichment and pollution have also been implicated in system decline. Animal issues of concern include the overall decline in historical numbers of native and endemic taxa, changes in historical patterns of the spatial and temporal distribution of species, changes in patterns of community structure, and changes in the pattern and magnitude of energy and material flows. Most animal populations in the region are now characterized by reduced densities and biomass, increased mortality, reduced reproductive potential and juvenile survivorship, and changes in persistence and resilience. These characteristics are usually associated with fragmented, natural landscapes in which populations are frequently affected by environmental stochasticity, demographic stochasticity, social dysfunction, or genetic deterioration (Noss, 1987). The alterations in fundamental landscape characteristics and related trophic disruptions described above may be viewed as chronic stresses to these ecosystems, termed "press disturbances" in the ecological literature (see Bender et al., 1984). By contrast, major perturbations, such as Hurricane Andrew, constitute "pulse type" disturbances, an important dichotomy. The regional biota are adapted to recover from pulse disturbances, even those as severe as Hurricane Andrew. However, the existence of chronic stresses may render these natural systems more vulnerable to permanent changes from a pulse disturbance. An integrative modeling approach is essential for understanding the coupling between the chronic stresses on these systems and the possible long-term consequences of hurricanes. Prior to the current effort, no rigorous, integrative approach for testing hypotheses about chronic stresses and their interactions with major pulse disturbances had been attempted. Such research is required to formulate a system-wide restoration plan within the context of regional water and nutrient budgets. Past research and restoration efforts have focused on individual species or habitats, usually within limited spatial or temporal scales. A lack of integrated understanding of the whole system's response to natural or anthropogenic perturbations severely restricts ongoing restoration and management possibilities. Several critical questions concerning the ecosystem's productivity and resilience must be resolved to produce a scientific basis for restoration and management. Since the 1960's, attempts have been made to quantitatively model entire ecosystems through sets of coupled differential equations; they have failed to provide reliable predictions about system responses to perturbations (Ulanowicz, 1986). Process-oriented models based on such equations have produced reliable predictions for lower trophic-level organisms (e.g., phytoplankton, bacteria, etc.), which are directly driven by biochemical processes. However, extensions of these types of models to populations of intermediate and higher trophic-level organisms (molluscs, crustaceans, fishes, reptiles, birds, mammals) have proven unreliable. Such models fail to incorporate the dependency of the biological processes of higher trophic levels on spatial scale, the complexity of decision-making (behavior) by higher vertebrates in response to environmental variability, and detailed population age and size structure data. Two other modeling approaches, spatially explicit (grid-based), structured-population models and individual-oriented models (which simultaneously simulate the life cycles of many individuals) have since been developed to provide more reliable predictions of higher organism patterns of response to perturbations (Huston et al., 1988; DeAngelis and Gross, 1992). The need for those models becomes even more acute when dealing with populations of organisms that occur at low densities (e.g. endangered or threatened species). The probability of survival for such organisms depends on spatial scale, environmental variability (spatial and temporal), and demographic stochasticity (DeAngelis and Waterhouse, 1987). Because probability of survival of rare species is also dependent on the food webs in which they occur (Pimm, 1991), reliable predictions of changes at lower levels of those food webs are also required. Therefore, it is vital that the process-oriented models of lower trophic levels, and the structured-population and individual-oriented models for the intermediate and higher consumers, respectively, be properly integrated to model the freshwater wetlands of south Florida. To accomplish this coupling, we are developing the Across-Trophic-Level System Simulation (ATLSS) array of simulation models that includes these three types of models. The simulation area includes the historical spatial extent of freshwater wetlands of the Everglades and Big Cypress swamp (Figure 1). The major groups within each trophic level for which simulation model development will be completed during the study period are listed in Table 1. The major trophic pathways of energy flow among these species or functional group of species are depicted in Figure 2. System components to be modeled include: (1) abundant species or primary ecological stocks that make up a significant proportion (5% or greater) of the system's biomass or energy flows, e.g. periphyton, zooplankton, aquatic and terrestrial insects, macroinvertebrates, and fishes; (2) the keystone species that may "regulate" the flows described above, e.g., alligators, colonial wading birds; (3) the rare trophic elements (threatened or endangered species) because preserving rare species is important to maintaining biodiversity, and because system-level changes would also be expected to impact these rare species first (e.g. Florida Panther, Snail Kite, Cape Sable Seaside Sparrow, Wood Stork); and (4) the invaders, non-indigenous species with the potential to disrupt system processess and trophic structure (e.g. pike killifish, Mayan Cichlid, feral hog). The importance of a functional group or species as a food resource to selected higher-order species, and the availability of data on the life histories or cycles of organisms were other factors considered in the selection process. Because we have hypothesized that landscape spatial extent and heterogeneity are critical to higher-order vertebrates, we selected species dependent on large home ranges (Florida panther), patchy (colonial wading birds, Cape Sable Seaside Sparrow, white-tailed deer) or stable resources (American alligator), and ecological specialists (Snail Kite). Models of these species, coupled to appropriate lower trophic models through a spatially explicit (grid-based) landscape structure, will be used to test hypotheses concerning the importance of a landscape of adequate size, heterogeneity, connectivity and hydrologic periodicity to the persistence of native animal assemblages. Although an initial selection of species or functional groups of species has been made for model development within this integrative ATLSS approach, another modeling approach called trophic flow-network modeling (Ulanowicz, 1986) for these freshwater ecosystems is also planned. This system-level approach will identify in a more quantitative manner, the primary ecological stocks or functional types of species comprising each system, and the exchanges of energy and nutrients among them. Completion of this analysis will insure that all major functional groups of species (comprising the major pathways of energy flow and nutrient cycles) are identified for inclusion into later versions of the freshwater ATLSS model. We define an integrated modeling system as models that will be constructed to allow coupling or linkage of the various trophic levels across the landscape. Prior cooperative efforts among the South Florida/Caribbean Field Laboratory of the National Biological Survey (NBS), the South Florida Natural Resources Center (NPS), the Institute for Environmental Modeling (UT/ORNL), and Chesapeake Biological Laboratory (CBL) provide a basic foundation for a system-level modeling approach. Initial versions of most trophic models have been completed. The models will be further refined, then coupled through a spatially explicit (grid-based) landscape structure which incorporates accurate remote-sensing information on elevation, vegetation, and simulation model outputs from hydrologic and plant community succession models. A landscape/vegetation model, under development, will simulate intra-annual variation in vegetation properties, e.g. estimations of the physiological status of current vegetation on a weekly or monthly basis in response to changes in hydrologic conditions and photoperiod. Specific variables include: rates of vegetation growth and regrowth, caloric and protein content of the vegetation, etc. Inputs on intermediate and long-term vegetation dynamics - regrowth and succession following disturbance, and feedbacks between vegetation and geomorphology - will be provided by plant community successional models under development (Sklar, pers. comm.; Huston, 1991). Simulated or projected changes in species population levels and geographical pattern, in response to changes in hydrology and related vegetation response, will be in a form that can be statistically analyzed and formatted for display by geographical information systems (GIS). There are many levels of research and management questions to be addressed in complex systems like the Everglades and Big Cypress Swamp. The ATLSS modeling approach will enhance the South Florida Task Force's technical capability for addressing scientific and management questions concerning these systems. The study requires a multi-year cooperative effort among NBS researchers, university staff, several national research laboratories, and scientists from the Army Corps of Engineers (COE) and public land-management agencies in South Florida. It will provide a framework within which past and ongoing ecological research will be synthesized, thus contributing to predictive and diagnostic models useful to addressing ecosystem-level questions concerning the Everglades and Big Cypress wetlands. OBJECTIVES The Everglades and Big Cypress swamp are characterized by complex patterns of spatial heterogeneity and temporal variability. ATLSS is intended to address fundamental cause-and effect factors associated with these patterns that operate on ecological components of various trophic levels. Such factors affect the population dynamics of key taxa and their vulnerabilities to disturbance. A series of these fundamental factors, arranged in ascending order in the trophic structure of these systems, are stated below as questions: Question 1: How is primary production divided between algal and macrophyte communities? How is production related to hydroperiod? Question 2: What are the relative flows of carbon from detrital and grazing pathways to the higher trophic levels? Question 3: Are there differences in fish and macroinvertebrate biomasses between marshes that are, respectively, detritus-based or grazing-based? Question 4: How do landscape extent, heterogeneity, configuration, and connectivity influence the distribution, abundance and population stability of higher consumer populations ? A major driving force for energy and material transfers among ecological components within and across the trophic levels of these systems is the distribution, volume, and timing of water flow. The relative importance of resulting spatial and temporal hydrologic patterns relates to their influence in structuring trophic patterns and processes. Identification of threshold responses of the biota to spatial and temporal changes in these patterns of water flow is a critical requirement for the design, evaluation, and selection of hydrologic restoration scenarios. Investigating the above-listed questions will be accomplished by comparing trophic responses to pre- and post-drainage patterns of water flow simulated with the same time series of rainfall data. Such comparisons would provide (1) an increased understanding of the causes of trophic disruptions observed, and (2) enable the identification of minimum hydrologic threshold requirements. These requirements include minimum areal extent of inundation, hydrologic periodicity in interior wetlands and sloughs, and minimum freshwater flows and the timing of these flows into downstream estuaries for restoring ecological functions (trophic patterns and processes) that have been disrupted by post-drainage conditions. Knowledge of the minimum thresholds would assist the efficient designing of hydrologic restoration scenarios for restoring a more functional landscape mosaic. We are defining a functional landscape mosaic as a landscape of sufficient areal extent, heterogeneity, configuration, connectivity and hydrologic periodicity to meet the minimum spatial and temporal requirements of top-level carnivores and native animal assemblages in these systems (Fleming et al., in press d). OVERVIEW OF METHODS This project employes a multi-disciplinary approach over several years, requiring collaboration of NBS and other scientists through cooperative and interagency agreements. The modeling domain will initially include only the freshwater wetland types of the historical Everglades and Big Cypress landscapes. Pending additional funding, the domain of the ATLSS model will be extended to the mangrove estuaries to simulate trophic dynamics across the entire historical landscape of the Everglades and Big Cypress Swamp. Construction of process-oriented models follows a state-variable approach (Swartzman and Kaluzny, 1987), using coupled, ordinary differential equations. The spatially explicit, structured-population and individual-oriented models under construction follow the principles and mathematical techniques described in Huston et al. (1988) and DeAngelis and Gross (1992). The modeling system will be designed to run on UNIX workstations or compatible computer hardware. The computer code for the models will be in a standard language with an object-oriented design to allow easy modifications to incorporate both future research and the need for possible changes in algorithms. Major work elements and their schedule for completion are presented in Table 2. The questions previously listed will be investigated using the ATLSS approach. Trophic responses to simulated water flows over the historical Everglades and Big Cypress landscapes will be based upon hydrologic outputs from a natural system model (Fennema et al., 1994) and analysed as a scientific baseline. The biotic responses will then be compared to responses to hydrologic simulations of contemporary conditions, using the South Florida Water Management Model (MacVicar et al., 1984). Pre- versus post-drainage changes in landscape characteristics to be evaluated are listed in Table 3. Such comparative analyses will identify critical natural-system characteristics, and the threshold responses of the biota to changes in those characteristics. Identification of thresholds will also allow the evaluation of proposed restoration scenarios to guide the further development of alternative scenarios, if necessary. No single modeling approach, however, can address all the questions of interest in understanding the particular processes operating in an ecosystem. A complementary diagnostic method for studying the Everglades and Big Cypress systems is trophic flow-network analysis, which allows complicated webs of trophic interactions to be identified and investigated. The underlying trophic structure can then be examined to assess the contributions of any individual species toward each successive level of the overall trophic pyramid. The exchanges of materials among ecological components within and across trophic levels of a system can also be quantified in units of carbon, phosphorus, or nitrogen (the major nutrient cycles) to describe nutrient budgets. The coupling of ATLSS models with trophic flow-network models would allow a more thorough examination of the impacts of altered hydrology on the major nutrient cycles of these systems. Predicted changes in standing stocks of ecological components through forecasting or hindcasting by the ATLSS approach can be converted to equivalent units of carbon, nitrogen, or phosphorus. These units can then be input into flow-network models to examine any resulting effects which altered hydrology may have on major nutrient cycles. Network analysis of these systems is scheduled to begin in FY96 as a complementary approach to ATLSS in addressing hydrologic restoration issues. LANDSCAPE DESCRIPTION Background Considerations. The Everglades/Big Cypress landscape is the template upon which all of its complex biological interactions occur. In the ATLSS approach, interactions among organisms, and between organisms and their environment, occur at relatively small spatial scales appropriate to the sizes and activities of the individual organisms. These spatial scales are generally much smaller than the diffuse, large-scale processes that drive the physical dynamics of the system, such as hydrologic sheet flow, rainstorms, hurricanes, and fire. A major challenge for ATLSS is to link these large-scale processes to the small scales at which organisms operate. The dramatic heterogeneity of the vegetation of South Florida, visible on the earliest maps of the region, as well as on recent satellite imagery, is superimposed on a landscape with very little topographic relief and extremely low hydrologic gradients. On this landscape very small differences in topography have large effects on plant and animal life. Unfortunately, the topography of this region has been very poorly characterized at scales relevant to individual organisms. The best available topographic information, which serves as the basis for the hydrologic models of the region (the Natural Systems Model and the Water Management Model) is of very coarse resolution (2 x 2 mile cells) and is based on approximations that may not correspond well with actual measurements in some areas (R. Fennema, pers. comm.). Landscape Structure. Given this discrepancy between available data and the information needs of a model of individual organisms, the approach taken by the ATLSS modeling group has been to generate pseudo-topography using vegetation maps. This approach is based on the strong relationship between vegetation and local elevation, as indicated by the characteristic hydroperiods of the major vegetation types. Briefly, we use the output of the SFWMD Hydrologic Model to determine the average and extreme hydroperiods for each 2 x 2 mile cell in the area covered by the model. Then, the average hydroperiod for each cell is compared to the expected hydroperiod range for each vegetation type within that cell, as determined by satellite-based vegetation maps. For each vegetation type for which the cell's average hydroperiod does not fall within the expected range, the elevation of that vegetation type is adjusted up or down from the elevation of the 2 x 2 mile cell to produce an appropriate hydroperiod. This approach assumes a level water surface across the entire 2 x 2 mile cell, not unreasonable given the low hydraulic gradient. The vegetation-derived topography interacts with the predicted water levels from the hydrologic simulation models to produce variation in water depth and hydroperiod at a spatial scale relevant to the organisms of interest (deer, panthers, wading birds) across the entire Everglades landscape. The water-mass balance of the hydrologic model is maintained by partitioning the water volume into sub-surface and above-ground components. The landscape model component of ATLSS is thus derived from the larger-scale physical process models (e.g., the Natural System and the Water Management hydrologic models, and the SFWMD landscape model), and will be refined and modified as these other models continue to develop. The approach of generating expected topography from vegetation patterns will also be refined as better vegetation maps and more accurate topographic data become available. To develop the model structure we are currently using the South Florida portion of the Florida Department of Transportation vegetation map of Florida, with 100 x 100 m pixels (an aggregate of 30 x 30 m pixels of the original satellite image) and plan to adopt higher resolution and more accurate vegetation classifications as they become available. Although modeling efforts to date have not addressed the effects of major disturbances which influence South Florida vegetation, the topography-based approach will also allow modeling of the spatial variability in the effects of disturbances such as fires, freezes, and hurricanes. Subtle variation in elevation and consequent effects on the dryness of vegetation will contribute to the patchiness created by fire. Likewise, small-scale variations in water depth can have major effects on the mortality caused by periodic freezes. The fine-scaled landscape model currently being developed will provide the basis for future modeling of successional dynamics and disturbance effects. MODELING APPROACHES The basic premise of the ATLSS approach is the use of an integrated set of models (Figure 3) to describe the response of the top-level carnivores and native animal assemblages of the Everglades and Big Cypress ecosystems to environmental dynamics. For the lower trophic levels we are primarily interested in how much available biomass is created, and how that biomass is affected by abiotic factors. For the higher trophic levels we are interested in the maintenance of populations of the native species. This suggests that models of different types should be used for the lower and higher trophic levels. What is the rationale for choosing appropriate models within the ATLSS approach? In general, there are no "cookbook" recipes for deciding what types of model to use. The choice of model type depends on: (1) the type of question being asked. One usually wants to use the simplest model that will adequately answer the question; and (2) practical considerations, such as the information available to build the model, time availability, and other constraints. Process-oriented models. These are models that describe particular processes at the ecosystem level, such as primary production, secondary production, and decomposition. Their aim is to predict how environmental conditions (temperature, water availability, nutrient loading, etc.) affect these processes. One can usually find fairly simple relationships between ecosystem processes, such as primary production and abiotic parameters (e.g., annual evapotranspiration), without taking into account the complexity of species interactions. Age- and size-structure models. These are models in which populations or functional groups are described to the level of age and/or size classes. Functional groups are sets of populations that play the same role in an ecosystem. Examples of age- and size-structure models include Leslie matrix models and McKendrick-von Foerster partial differential equation models. The dependent variables in these models are numbers of individuals in various age and size classes. In ATLSS, such models are applied to fishes and aquatic macroinvertebrates. The dynamics of these functional groups and their ability to provide prey biomass for wading birds may depend on the age and size structure dynamics of the groups (e.g., predator-prey interactions between larger and smaller size classes, dependence of fecundity on age and size). Individual-based models. These are models of the dynamics of a population, or part of a population, in which each organism and its interactions with other organisms are simulated individually. The rationale for their use is that the maintenance of a population of higher consumers is not usually related in a simple way to the biomass production of lower trophic levels. Instead it depends on details of energy allocation, foraging on a complex, heterogeneous landscape, and social interactions. Also, populations of higher consumers are often so small that demographic stochasticity is an important factor in their probability of survival. Finally, the predictions desired for these large consumers are often not merely population numbers, but also of individual age, size, physiological and/or genetic condition, and spatial location. Individual-based models lend themselves easily to simulation of populations and communities on complex landscapes, and are thus often referred to as "spatially explicit" models (Pulliam et al., 1992; Turner and Gardner, 1991). AQUATIC-BASED FOOD WEBS LOWER TROPHIC-LEVEL MODELS Background Considerations. The ATLSS project integrates, in a common modeling framework, ecological elements of varying complexity and scale. The simplest populations of the smallest organisms in the freshwater wetlands of South Florida are found at the lower trophic levels, where populations may be simulated by aggregated variables belonging to a set of first-order, ordinary differential equations. Such equations can be integrated to portray the behavior of these components on a very fine time scale. This output will provide information on the resources available to the higher, more structured populations at the longer intervals over which the structured-population models are run. Furthermore, the consumption of these resources by higher-level consumers can be fed back into the lower trophic models, thereby providing a two-way, dynamical bridge between the lower trophic levels and the rest of the ecosystem. It follows that output from the lower trophic-level models need not be as precise or as detailed as those issuing from the other modules of ATLSS. In particular, one is interested mainly in the response of these freshwater ecosystems to variation in water levels. The behavior of the ecosystem will, in overwhelming measure, be modulated by the areal amounts of marshland that are left dry during various seasons of the year. Previous work with the individual-based Wood Stork model (Fleming et al., in press c) showed that one could make informed ecological assessments with very crude estimates of resource densities, so long as one has accurate estimates of the areal extent of inundation at any time. That is, the output from the lower trophic- level models need not be highly precise. It merely must be reasonable for the given time and location. Periphyton and Macrophyte Models. Following on these premises, we have created process (differential equation) models of the macrophytes (submerged aquatic plants), periphyton (algae attached to plant and bottom surfaces), and detritus (dead particulate organic material) that constitute the resource base for these freshwater wetlands (Figure 4). During a simulation run of ATLSS, these process models will be applied within each spatial cell of the landscape grid. Our paramount criterion is that these cellular models at all times yield reasonable values for these resources. That is, under all conditions to which the models will be subjected, they must remain within the range of credible population levels based on historical observations. We guarantee that our models will remain within reasonable bounds by choosing dynamics for the two plant compartments that are inherently stable. The rate of growth of the macrophytes and the periphyton has been assumed proportional to the respective stock of these plants, whilst their mortality was assumed to vary as the square of their densities. This means that below some prescribed density of plants, the stocks will grow, whereas at high levels the mortality term will dominate and stocks will decrease. The parameters for each term were chosen so that the population always converges upon some mean of the observed densities. If the growth and death parameters were considered constant, the resulting behavior of the models would be quite uninteresting. Any perturbation of plant stocks would recover toward the mean in monotonic fashion. The models would be insensitive to the seasonal conditions prevailing during recovery. In order to impart a bit more reality to our models, we let the growth and death parameters vary sinusoidally over the year. The parameters for mean stock size and for those regulating the sinusoidal amplitude and phasing were determined by singular-value decomposition (a best-fit technique) using data on periphyton, macrophyte and detritus stocks taken during each season by Browder (1982). This allows the system to respond to a perturbation occurring during different times of the year in a way that is characteristic of the particular time of year during which the perturbation occurs. Detritus Model. The principal source of detritus in these aquatic systems is the death of the macrophytes and associated periphyton. Hence, the generation of detritus is taken as proportional to the death terms in the equations just described governing plant stocks. The disappearance of detritus is assumed to be in proportion to the current stock of detritus. The seasonal behavior of the decay rate is determined from the data of Browder (1982). After a period of dry-down (see below), the stock of detritus will converge to a new, somewhat lower level. This means that the recent history of the system will be written into the current level of detritus. In a crude way this mimics the actual system in that repeated dry-downs leave the system with little available detritus. Mesoinvertebrate and Macroinvertebrate Models. The populations of invertebrates that feed upon the plants and associated microfauna are parsed into two groups according to size. The mesoinvertebrates range in size from about 60 microns up to about a millimeter and include such things as oligochaetes, copepods, amphipods, water fleas, etc. The macroinvertebrates are larger than about one millimeter and consist mainly of snails and various insect larvae. We exclude prawns, crayfish, and Apple Snails which are described by separate age/size-structured models, explained in the following section. The growth of both invertebrate groups is assumed to vary jointly as the product of their own populations and the plant stocks upon which they graze. As with the plant models, death is considered to vary as the square of their respective stocks, thereby keeping their populations from ever becoming unrealistically high. Data are available for both compartments on a monthly basis from an unpublished study by Conrow and Loftus of Everglades National Park. As with plants, seasonal variation of invertebrate growth and death rates are represented by annual sine waves, for which the governing parameters were determined from the data using singular value decomposition. Other Considerations. The major perturbation which ATLSS will be used to investigate is the effect of the distribution, volume, and timing of water flows on the trophic structure and dynamics of these systems. In the current models for the lower trophic levels, all ecosystem dynamics are considered to be independent of actual water level down to a depth of 7 cm. Below that depth, the growth of all species ceases and dieoff is simulated by transferring biomass from the living compartments directly into detritus in proportion to the volume deficit below the 7cm level. Dieoff continues until populations reach 5% of their stocks at the beginning of dieoff. Beyond that point, the 5% reserve is maintained as "seed" stock to initiate regrowth as soon as the cell begins to reflood. Browder (1981) reports that the ecosystem dynamics of the short- and long-hydroperiod gramminoid marshes differ significantly. The periphyton in the long-hydroperiod marshes is characterized by greater representation of diatoms and green algae. By contrast, periphyton in the short-hydroperiod marshes is dominated by harder-to-assimilate, calcareous blue-green algae. To account for the substantial differences in these two communities, separate models were derived to describe each. Data were available from both the Browder and Conrow and Loftus studies to calibrate separate models for each biotype. We discovered significant differences in the short-and long-hydroperiod gramminoid wetlands in terms of the densities of invertebrates. One notes, for example, that the insect larvae densities found in the continuously flooded wetlands are about ten-fold higher than areas with frequent drydown. Model Constraints. These major differences in stocks bear upon the question of how well these empirical models will be able to predict ecological conditions under flow regimens not encountered in the course of data acquisition and model calibration. We believe they should give rather reliable estimates for the gramminoid wetlands. The reason is that the productivities and densities of such marshes that have been flooded without interruption for many years are not likely to differ that much from those we used to calibrate the long-hydroperiod model. The major system-level changes will be occasioned by shifts in the relative areas occupied by long- and short-hydroperiod biotypes. As long as the transition of any particular spatial cell from short- to long-hydroperiod gramminoid type can be accurately judged, the models should give reasonable estimates of productivities for these biotypes under enhanced flow regimens. The threshold for a shift (10 months) of a spatial cell from a short-hydroperiod to a long-hydroperiod gramminoid biotype is calculated over a 5- year running average. A 3-year running average, however, is used to simulate a shift in the opposite direction to reflect the fact that organic matter is lost more rapidly than it is accretated. This threshold, and the related time periods over which it is calculated, reflect our best judgement, pending the completion of on-going analyses to define this. However, there are several biotypes (cypress strands, pinelands, mixed-cypress and pines) for which outputs of these models may be inaccurate. The detritus model will probably underestimate detrital loadings (due to the increased rate of litter fall) for these forested wetlands, while the periphyton models probably overestimate periphyton production (due to decreased light penetration). As a result, invertebrate model predictions may be in error for those forested wetlands by as much as an order of magnitude. The data gaps point to the need for empirical studies of primary and detrital production in those wetland types. There are no terms in the lower trophic-level model to represent losses of these resources to higher trophic-level components. Rather, such consumption will be calculated by the calling routines that model these consumers. The amounts consumed will be subtracted in the calling routine from the resource biomasses before that program makes its next call to the resource subroutine. In this manner we will simulate the "top-down" influence of predators upon the lower trophic levels. Also, the current models pertain only to changes in water depths. Data are slowly accruing to show how the productivities of biotypes might respond to different levels of nutrients in the supply water. These data should eventually allow us to expand the process models so that they will apply as well to combined scenarios of alterations in water levels and available nutrients. Finally, we should point out that several functional groups at the lowest trophic levels, the bacteria/fungi and the microinvertebrates (protozoans, ciliates and other fauna less than 60 microns), do not appear in our models. To our best knowledge, no data on these components exist for the Everglades and Big Cypress Swamp. We are able to work around this gap because of the empirical nature of the models we are using. (That is, the influence of such microbiota is implicit in the values and seasonal variations we obtain for the feeding parameters of the two "herbivore" compartments.) As ATLSS develops, however, and more mechanistic detail is included into the models for lower trophic level resources, it will become imperative that field and laboratory effort be expended to elucidate and quantify this hitherto much-neglected, but important portion of the Everglades and Big Cypress ecosystems. INTERMEDIATE TROPHIC-LEVEL MODELS Background Considerations. Freshwater fishes and macroinvertebrates are important components in the aquatic food web of the Everglades and Big Cypress Swamp, acting as primary consumers of vegetation and detritus, at intermediate levels as predators on aquatic insects and crustacea, and as top carnivores. The fishes, in turn, are prey for a myriad of predators and scavengers (Gunderson and Loftus, 1993). The rapid life cycles of small Everglades fishes (Haake and Dean, 1983) enable them to respond rapidly to changes in hydrologic conditions. Loftus and Kushlan (1987) described the fish communities of the Everglades and adjacent wetlands. An assemblage of 30 species occur in the spikerush (Eleocharis Spp.) or wet prairie habitat. Small species of killifishes (cyprinodontids) and livebearers (poeciliids), and juvenile sunfishes (centrarchids) are the common inhabitants of wet prairie and sawgrass marshes. The killifishes and livebearers are short-lived, rapidly growing species (Haake and Dean, 1983), which respond to favorable hydrologic conditions with rapid population growth. The deeper, open-water alligator holes are used by larger fishes such as Florida gar (Lepisosteus platyrhincus), yellow bullhead (Ameiurus natalis), and adult sunfishes, although smaller species, including mosquitofish (Gambusia holbrooki) and sailfin molly (Poecilia latipinna) are also common. The marshes and ponds also provide habitat for the crayfish (Procambarus alleni), the prawn (Palaemonetes paludosus), and the Apple Snail (Pomacea paludosa). The hydrological changes of recent decades in South Florida, particularly increased dry down frequencies in interior wetlands and sloughs, and severe overdrainage of peripheral wetlands, have adverse impacts on these fish communities (Loftus et al., 1990; Loftus and Eklund, 1994). These changes also have affected the top carnivores of these wetland systems. Piscivorous wading birds feed at lower densities in some short-hydroperiod marshes than would be predicted by the area of marsh available (Fleming et al., in press a), possibly because of insufficient prey densities. The number of alligator holes per square kilometer, and presumably the density of alligators, is often but a fraction of that found in long-hydroperiod marshes (Fleming, in review). The low density of alligator ponds may affect fishes by reducing the availability of dry-season refugia and the critical habitat for the large-fish assemblage, thus limiting the number of fishes surviving to recolonize in the next wet season. The areal reduction of peripheral marsh habitats by drainage and development may be critical to the capacity of the system to support large numbers of predators that feed upon marsh fishes (Fleming et al., in press a, b ,c, d). Loftus and Eklund (1994) have shown that the biomass of fishes per unit area is low, even in long-hydroperiod marshes, but the total fish biomass produced across the vast area of the original Everglades was probably many times higher than today. The focus of our present modeling efforts at intermediate trophic levels of these systems has been to predict fish community dynamics and production for selected macroinvertebrates and key fish functional groups in the major biotypes that comprise the Everglades and Big Cypress landscapes under a variety of hydrologic conditions. This predictive capability will assist in the identification of hydrologic thresholds, and the subsequent design and evaluation of hydrologic regimes, for restoring aquatic productivity to degraded wwetlands of South Florida. Model Structure. The spatial models operate within each cell of the ATLSS landscape grid, with the capability to vary cell size: 1 x 1 km, 1/4 x 1/4 km or 100 x 100 m, whatever size is most compatible with the environmental information available. Heterogeneity within the cell is of vital importance to the macroinvertebrate/fish dynamics (Loftus et al., 1990; Loftus and Eklund, 1994). Heterogeneity in the model includes: (a) the possibility of one or more ponds/solution holes and shallow surface depressions of various sizes (diameters); and (2) the possibility of such holes and shallow surface depressions for a spectrum of depths. The hydrologic conditions of each cell on a day-by-day basis are determined from the output of a hydrologic model (e.g. Natural System Model, SFWMD model). In each cell there is a carrying capacity for macroinvertebrates and fish to be determined by the previously described models of lower trophic levels (periphyton, detritus, benthic insects, etc.) that are resources of the fish and macroinvertebrates. Outputs from these models of prey density can change from day to day as the abiotic conditions of the cell change, and as prey are removed by the fish and macro-invertebrate populations in the cell. Several key fish and macroinvertebrate functional groups are modeled as 5-day age classes (Figure 5). The time steps used in the model are also five days, so that on each time step all individuals advance by one age-class. Each age-class, i, of a particular functional group is assigned a size (length, L) calculated from a von Bertalanffy equation: Li = L (1-e^k(i-to)) where L , K, and to are constants. Mature individuals of each functional group produce a number of viable offspring during their reproductive season. Baseline, age-dependent mortality is assigned to each functional group. Both reproduction and mortality are estimated from empirical data. The environmental driving variable is the hydrologic cycle. If and when the average water level falls below the mean elevation of the cell, individuals present in the cell at the time are allocated to whatever ponds/solution holes and shallow surface depressions exist in the cell. Within the ponds/solution holes or shallow surface depressions, these individuals are assumed to undergo the baseline mortality. However, if these areas dry out, individuals that found refuge in them are lost from the population. Those surviving in deeper ponds/solution holes emerge in the cell when the water level rises again. Smaller fish in the ponds/solution holes are exposed to predation from the larger fish. Predation mortality is a function of the ratio of predator biomass to prey biomass (Figure 6). The cellular model simulates the community dynamics and production of the fish and macroinvertebrate functional groups over periods of several years to examine the effects on the macroinvertebrate/fish populations of various hydrologic regimes. Several results from the simulations are stored as output. These include the following: (1) total numbers of fishes or macroinvertebrates per square meter (summed over age classes) for each functional group within a cell on 5-day time steps (calculated separately for both the marsh and pond/solution hole areas); (2) total biomass of each functional group within a cell on 5-day time steps (calculated separately for both the marsh and pond/solution hole areas); and (3) age-class histograms of all functional groups within a cell on selected months and years. Model Constraints. The potential limitations of the model are related to the lack of information on the energetic requirements and expenditures of the fishes and invertebrates. In addition, empirical data are not available for some biotypes, e.g. forested wetlands. There are also gaps in the age-structure data for many species. These data deficiencies, however, are not a serious hindrance to modeling the dynamics of these species in relation to varying hydrologic conditions, but may limit the testing of these models. HIGHER CONSUMER MODELS Colonial Wading Birds. Background Considerations. Wading birds are important components in the aquatic food web as top-level carnivores. As highly mobile animals, wading birds can influence the structure and dynamics of both freshwater and estuarine prey communities (Kushlan 1976, 1978). They also transport nutrients from interior, freshwater wetlands to downstream coastal estuaries (Kahl, 1964; Kushlan, 1978; Frederick and Powell, 1993). Wading birds depend on patchy resources. Since water management regulation, however, ponding of overland flows in northern reaches of the Everglades catchment area and severe overdrainage of the southern reaches downstream of these impoundments has occurred. Combined, these have altered the locations or spatial arrangement, and reduced the areal extent and heterogeneity, of seasonal foraging habitats for wading birds (Fleming et al., in press a, b, c). Declines in wading bird populations have occurred in all feeding guilds concurrent with these landscape changes (Fleming et al., in press a). Decreasing numbers also attempt to nest each year, particularly at traditional colony sites within downstream reaches of the Everglades basin (Fleming et al., in press b). The focus of present modeling efforts for wading birds in ATLSS, therefore, is to develop the simulation capability for investigating nesting colony dynamics in relation to landscape changes caused by altered hydrology. The species selected for modeling in ATLSS represent approximately 85% of the regional wading bird population (Fleming et al., in press a) and include both short and long-legged, tactile and visual-feeding species. Model Structure. Wading bird nesting colony model template. The individual-based wading bird nesting colony model used in ATLSS simulates the activities of potential nesting adults for a period of time immediately preceding the formation of a nesting colony and then the whole of the nesting season. In this model each of the adult nesting birds, as well as each of the offspring of these adults, are modeled as individuals. From this general model template, we have also developed calibrated versions for selected species. A description of this general model is given below, followed by descriptions of different model versions for selected tactile (Wood Stork, White Ibis) and visual (Great Egret, Great Blue Heron) feeding species. The model consists of four basic parts (Fig. 7): (1) a sub-model of the spatially heterogeneous landscape, on which changing water depth is the key variable, (2) a sub-model for the prey population, which varies spatially and temporally across the landscape, (3) models for the behavior and energetics of each potentially nesting adult, and (4) models for the energetics and growth of each nestling, until the nestling is fledged or dies. Landscape sub-model. The importance of spatial heterogeneity of the environment surrounding a colony site requires that heterogeneity be taken into account. This is done through division of the potential foraging area into contiguous, square 25,600 cells, each 1/4 km by 1/4 km. Each cell has its own mean elevation, so that the whole set of cells describes the topography of the wetlands surrounding a selected colony site. Water depths in the spatial cells are based on inputs from hydrologic simulation models and changes on daily time steps. Prey-base sub-model. The prey for the wading birds consist of small fishes and macroinvertebrates. Prey densities are based on the macroinvertebrate and fish guild models previously described and read in at appropriate time steps. A simulation starts near the end of a wet season, with prey being assigned densities across the landscape. The prey in a given cell are assumed to be available to the wading birds in a given cell only when the average water level of the cell is within a certain range (based on a selected species' leg length). The wading birds can feed on the fish and/or macroinvertebrates in that cell and, in so doing, reduce the prey biomass of the cell. Prey densities change due to wading bird foraging on 15-minute time steps. The foraging efficiency of wading birds in a given spatial cell is proportional to the number of prey in the cell, so as prey biomass decreases, the feeding success of the birds in the cell also decreases. The birds also tend to stay for longer periods of time in cells with high prey density. Behavior and energetics sub-models of potential nesting adults. The nesting adult birds are described by a set of species-specific rules that govern their behaviors from one time interval to the next. The time unit chosen to describe the behaviors of the adult birds is 15 minutes. This time unit was chosen because many discrete activities of birds, such as carrying food back to the nest, take time intervals as short as 15 minutes. Notice that this characteristic time differs from the daily time-step changes of cell water depth and prey biomass in a cell. The first choices that must be made by adult pairs during the pre-nesting season are if and when to start nesting. The general rule that is followed is that nesting will begin if the female is able to obtain 20% more than its food needs for three consecutive days during the time period after the end of the rainy season. Eggs are produced asynchronously per pair (the maximum number being dependent on the species being simulated) and hatching takes place over several weeks (the time period also depends upon the species being simulated). After the start of nesting, the decisions made by the adults are guided by various constraints. Each adult must meet a maintenance energy demand each day. To meet this demand the adult uses the first portion of the food it collects in a given day for itself. The time-period each day during which individual birds usually forage is also specified, with each bird deciding when to start and whether to follow others or to go by itself. The location chosen by an individual for foraging is based on its partial information concerning the system. It is assumed that each wading bird has some knowledge, perhaps obtained from visual cues when flying or soaring, concerning the water depth of various locations (cells in the model) in its foraging area. But the wading birds do not know the prey density in a given cell until it has been sampled for some time (at least 15 minutes). The wading bird can select a cell in the appropriate water-depth range, but randomly otherwise, and search for food in that cell. Otherwise, it can decide to join one of several flocks of birds from the colony that may already be feeding. Though each capture of prey by a bird is a stochastic event, the rate of prey captures depends on the current fish density of the cell. If the wading bird has poor success during a 15-minute interval, it moves to another cell, either one nearby or at some distance, although there are greater costs in flying to a more distant foraging site. Again, it may choose to feed solitarily or join a flock. Probabilities for such choices can be assigned in the model. It should be noted that the presence of a flock will usually indicate high food density, though it may have been depleted already by the birds foraging there. Energetics and growth sub-models of nestlings. After obtaining sufficient food for itself, the wading bird will continue to forage for its offspring. It will decide at some point to bring food back to its nestlings, either all the food that it can carry or all that it was able to gather during the time it foraged. The adult regurgitates its food inside the nest where there is competition among the nestlings for the food. According to the rule used in the model, a greater proportion of the food is taken by the largest nestling, the next greatest amount by the second largest, and so forth, until all food is distributed. The daily growth of the nestlings is computed on the basis of the energetic value of the daily ration of food brought by their parents. Depending on their age in days since hatching, there is some maximum amount of food that an individual nestling can consume. This maximum increases linearly at first, then levels off and finally declines as the nestling approaches fledgling. From experimental measurements, it is known that a nestling must attain some threshold level of accumulated food (specified for each species) in order to fledge. If the nestling does not receive this amount of food before the rainy season begins, and adult foraging capability decreases, the model nestling will usually die, because the parents will no longer be able to provide food at a rate sufficient for its survival. If, over any five-day time period during the nesting season, a nestling receives less than a certain percentage of its cumulative food needs, it will die. If the parents cannot find enough food to meet their own energetic demands, they will abandon their nest and the nestlings will perish. The above-description is a very brief outline of the nesting colony model template. A more complete description is presented in Wolff (1994), in which each of the rules mentioned above is discussed in detail, quantitatively. Colony Model Versions for Tactile Feeders (Wood Stork, White Ibis). Wood Stork Colony Model. The wading bird colony model template is specialized to simulate a colony of Wood Storks, (Mycteria americana), as follows. Prey-base sub-model. For convenience, in the current version of this model, all prey which are consumed by the Wood Storks are assumed to be in the form of cyprinodontoid and centrarchid fishes, with an average weight of 1.73 grams (Ogden et al., 1980). The prey in a given cell are assumed to be available to Wood Storks when the water depth does not exceed 40 cm. The sub-model is being modified to accept inputs from the macroinvertebrate and fish guild models previously described. Behavior and energetics sub-models of potential nesting adults. Usually three or four eggs are produced asynchronously by each pair of Wood Storks, and hatching takes place in about three weeks (Kahl, 1964). The individual wood storks usually forage between 1000 h and 1600 h each day (Kahl, 1964), with each bird deciding when to start and whether to follow others or to feed by itself (Figure 8). The location chosen by an individual for foraging is based on its partial information concerning the system. For a tactile forager, such as the Wood Stork, the foraging efficiency should be roughly proportional to the (mean) density of prey within a cell. Energetics and growth sub-models of nestlings. Depending on their age in days since hatching, there is some maximum that an individual Wood Stork nestling can consume (Kahl, 1964). This maximum increases linearly at first, then levels off and finally declines as the nestling approaches fledgling. From experimental measurements, it is known that a nestling must attain some threshold level of accumulated food (taken here as 15 kg) in order to fledge. If the nestling does not receive this amount of food before the rainy season begins, the model nestling will usually die because the parents will no longer be able to provide food sufficient for its survival.. White Ibis Colony Model. The following describes the parameters and rules used in the White Ibis nesting colony model. Only rules different from the corresponding ones given in the general colony model are stated explicitly; if the rules are the same (or very similar) only the new parameter values are given. Prey-base sub-model. White Ibises are tactile foraging birds that feed mostly on small prey items of approximately 2 cm. long. The food of a White Ibis in southern Florida consists mainly of crayfish (Kushlan, 1977b). The average energy content of the diet is 4.05 kcal/gram-wet weight (16.945 kJ/gram-wet weight) with a ratio of 4:1 wet weight:dry weight (Kushlan, 1977b). This prey-base sub-model is being coupled to the macroinvertebrate and fish models for inputs on White Ibis prey. Behavior and energetics sub-models of potential nesting adults. The American White Ibis (Eudocimus ruber albus) is a medium-sized ibis, in which adults have white plumage and juveniles are darker. Males and females have identical plumages, but the male is substantially larger (Kushlan, 1977a). A male weighs about 975 grams, a female about 775 grams (Hancock et al, 1992). The energy requirement of White Ibises has been measured by Kushlan (1977c). For birds with an average weight of 948 grams, Kushlan estimated the following energy requirements: basal metabolic rate (BMR) 85 kcal/day (355.640 kJ/day) existence metabolism (EM) 113.8 kcal/day (476.139 kJ/day) aviary existence metabolism (AEM) 138.7 kcal/day (580.32l kJ/day) assimilation efficiency (AE) 79.7% Using the mean assimilation efficiency of 79.7% and an average wet weight to dry weight of prey of 4:1, an adult White Ibis thus requires 203 grams food per day or 21.4% of its body weight. A similar number (21%) was obtained for Wood Storks (Kahl, 1964). The initiation of nesting is governed by rules similar to the Wood Stork model. In particular, the female must obtain enough food to meet the additional energy requirements of ovary and egg production, i.e. 250 kcal (1046 kJ) during 7 consecutive days. This represents an energetic requirement of more than 25% of its daily energy needs. This rule probably underestimates the actual energy required. For nonpasserine birds, the ovaries comprise about 6% of the female's body weight, i.e. 45 grams for a female weighing 750 grams. Average energy content of the ovaries is about 8 kJ/gram. Total egg weight is also about 6% of the female's body weight, with an average energy content of 5 kJ/gram (i.e. 225 kJ/egg). For a clutch of two eggs, the female thus needs 810 kJ (194 kcal), for a clutch of three eggs 1035 kJ (247 kcal). Using a conversion efficiency of 75% (ratio of food consumed to energy utilized), a female laying two eggs needs 258 kcal (1079 kJ), and a female laying three eggs needs 330 kcal (1381 kJ). Clutch sizes in White Ibis nests range from two to five eggs, averaging 2.45 in the Everglades (Kushlan, 1977c). The first egg is laid 5-6 days after copulation, and additional eggs at 1-2 day intervals, the latter being usual. Incubation starts as soon as the first egg has been laid and lasts for about 21 days (Hancock et al., 1992). The rules for foraging flights have also been changed. In the general wading bird colony model, each bird chooses its foraging site independent of the choices of other birds. White Ibis, however, frequently fly in cohesive flocks between feeding and roosting sites (Kushlan, 1978). If an adult leaves the nest for a foraging trip, it may either join one of the flocks leaving the colony, or it may choose a new foraging site independent of other birds also leaving the colony (Figure 9). At every given time interval, all birds leaving for a foraging trip are thus divided into one or several sets, i.e. flocks. The birds in each flock fly to a common foraging site. Each bird in the flock requires the same time to arrive at the site. The energy expenditure, however, is calculated separately for each bird according to its weight, using the formulas in Pennycuick and DeSanto (1989). Although males are larger than females, they do not feed more successfully (Bildstein, 1987). The foraging success rate is thus assumed independent of a bird's weight. As no data are available on the distribution of flock sizes leaving a colony, the model allows for two possibilities: (1) the bird chooses among the flocks leaving the colony with equal probability or, (2) with a probability proportional to the size of the flock. In the first case, flocks leaving the colony at any given time are of equal size, whereas in the second case there will be one large and one or several smaller flocks. The rule for selecting a feeding site is the same rule as the corresponding rule in the wood stork model, i.e. a flock leaving a colony site is treated as a single bird; and, flocks form at a feeding site from local enhancement (Kushlan, 1976, 1978). The model White Ibises discriminate among feeding sites according to the distance from their colony site. Whenever a White Ibis has to choose a new feeding site, sites within 10 km from the bird's colony location are preferred to more distant sites (based on Fig. 25.5 in Bancroft et al., 1994). In the model, 90% of sites less than 10 km away are chosen over more distant sites. This rule is only used when a bird, feeding at such nearby sites, can obtain enough food to meet its energy requirements during the previous three days. If a bird is not able to satisfy its energy requirements by foraging at sites nearby, it will prefer more distant sites, i.e. the probabilities in the last-stated rule are reversed. Energetics and growth sub-models of nestlings. Nestlings are modeled in the same manner as the adult birds. Simulating nestlings, however, is much simpler because their main activity consists of eating the food brought back by their parents, and sleeping. Because our main interest is in determining whether or not a nestling receives enough food to fledge successfully, it suffices to keep track only of their food intake. To model the growth of the nestling, their relative sizes are assumed to be solely determined by their individual total cumulative food intakes. Total food intake, therefore, determines whether the nestlings have grown to a size enabling them to leave the nest and forage on their own. Nestlings are fed by regurgitation. Adults begin feeding their young soon after hatching. After the last chick of a nest is born, the oldest, largest chick is usually fed first. As a result of synchronous hatching and differential feeding, siblings differ in size and vigor. As a consequence, brood reduction is severe (Kushlan, 1977c). Young leave the colony when about 50 days old (Hancock et al., 1992). The total energy requirement for the growth of one young to fledgling is 8629 kcal of food (36066kJ) (Kushlan, 1977b,c). The pattern of nestling food uptake is similar to that of Wood Storks (Kushlan, 1977b; Kahl, 1964). Therefore, nestling subsistence and growth rules are modeled using the same rules as used for Wood Storks, but with an appropriate change of parameters. (1) Each nest initially contains three eggs which are laid two days apart. (2) Each egg is incubated for 21 days before hatching. (3) The amount of food each nestling can maximally ingest on a single day increases linearly to 250 grams during the first three weeks after hatching. It remains at this value for the following three weeks and then decreases linearly to about 150 grams per day during the final three weeks. (4) The amount of food a nestling receives can fall below these values for some days, but the accumulated actual food intake for five days in a row must not fall below 50 percent of the accumulated maximal values; if it does, the nestling will be assumed to have starved. (5) Each nestling must be provided with a total of 7.8 kg of food over the whole nesting period to be successfully fledged. This value corresponds to an average daily food intake of about 200 grams during the second stage of highest food demand. (6) Nestlings can fledge after 50 days and can stay up to 55 days in the nest. (7) The relative size of the chicks in a nest is determined by their total cumulative food intake. (8) Larger nestlings obtain more food than their smaller mates. Colony Model Versions for Visual Feeders (Great Egret, Great Blue Heron). In the following we summarize how the general colony model was modified to simulate visually feeding species, in particular Great Egrets (Casmerodius albus) and Great Blue Herons (Ardea herodias). Prey-base sub-model. Visually feeding, long-legged waders exhibit a much broader range of sizes of consumed prey than tactile feeding species (Kushlan, 1978). To reflect this difference, the basic unit of prey is taken to be 20 g. In this way the adult bird could simultaneously consume prey of 20, 40, 60 g, representing usual weights of prey consumed by visual foragers in the prey sub-model. In the present versions of the individual wading bird models the prey base is estimated from empirical data. However in the integrated model, prey availability will be supplied by the fish and the aquatic macroinvertebrate models. Behavior and energetics sub-models of potential nesting adults. The actual mechanisms of visual feeding (locating the prey, stalking, and handling) (Kushlan, 1978) are more subtle than tacto-locating (groping), as is done by tactile feeders such as the Wood Stork and White Ibis. With visual feeders, a number of environmental factors affect their feeding success, i.e. intensity and direction of sunlight, wind, water depth, water temperature, refraction between the prey's apparent and real position in the water column, etc. (Bovino, 1979; Frederick and Loftus, 1993; Katzir et al., 1989). Foraging success rates also differ by age of the bird (Quinney and Smith, 1979). No data are available on the effects of sunlight intensity and direction, or wind, on the foraging success of visual feeders. The foraging behaviors in the general wading bird colony model, therefore, were only modified to account for visual feeding behavior characteristics with respect to: 1) water depth, in that herons and egrets can forage more successfully in deeper water areas than tactile feeders (Kushlan, 1978). (2) the actual prey resources of the cell, i.e. the handling time of larger prey items by visual foragers, increases exponentially relative to smaller prey sizes (Kushlan, 1978). (3) the time of day, i.e. visual foragers feed more efficiently during afternoon hours, when the sun is at its peak (Bovino, 1979), and reduce foraging effort (not success) during the early morning and late afternoon hours (Kushlan, 1978; Palmer, 1962) and, (4) whether the bird is an adult or fledgling, i.e. increased foraging success rates for adults (Quinney and Smith, 1979). The model assumes linear functional forms for foraging success rates in relation to (1) and (3) in order to adjust the actual prey resources of a cell to that available to a visually feeding bird. The behavioral rules for selection of foraging sites, in relation to the location of other birds, were also modified. Two authors (Krebs, 1974; Kushlan, 1976) showed that feeding aggregations of Great Blue Herons form only when the prey density is high. Birds which forage in such groups gain more food simply because flocks form when food is abundant. When resources are not so abundant, herons often forage solitarily and establish and defend feeding territories. Such solitary visual feeders spend more time searching neighboring sites, walking slowly to search for prey when it is more scarce. The cell size in the landscape sub-model was changed, therefore, to 50m x 50 m (i.e. a finer resolution) to account for the increased time spent by solitary visual feeders searching nearby sites for prey. However, if resources are abundant, this territoriality breaks down and additional birds are allowed to join. This is modeled by the following rule: An adult heron samples a cell. If, during the first day, it has satisfied all of its energetic needs while in the cell, it will return the next day. If this happens for more than one day, other birds searching for prey are also allowed to feed in the same cell. Great Egret Colony Model The specific details of the Great Egret (Casmerodius albus) model are as follow: Prey-base sub-model. Great Egrets are visual feeders that forage both in shallow and deep water areas (Kushlan, 1978). The maximum depth for wading is set at 28 cm (Palmer, 1962). They are diurnal feeders (Palmer, 1962). The prey for Great Egrets consist mainly of fish and macroinvertebrates (Kushlan, 1978). Behavior and energetics sub-models for potential nesting adults. The adult Great Egret weighs 1.0-1.5 kg, usually with the male about 10% larger than the female (Palmer, 1962). The energy requirements are as follows: Basal metabolic rate: 78 Free living metabolic rate: 145 Assimilation efficiency: 77% The initiation of nesting is governed by rules similar to those in the previous models. A threshold of 280 kcal for a week is the required energy that a female must get in order to start a nest. Their primary foraging mode is by standing and wading slowly (Kushlan, 1978), so sampling of adjacent cells is usually part of their foraging strategy. They can choose to forage in flocks, following the rules explained in the Wood Stork model. Energetics and growth sub-models of nestlings. Every nesting pair lays 3-4 eggs, asynchronously, with a period of two days between eggs (Palmer, 1962). Each egg is incubated for 24 days before hatching (Palmer, 1962). The amount of food that a nestling can maximally ingest grows linearly at a constant rate of 25 g/day reaching a maximum of 300 g, within the first two weeks (Custer and Peterson, 1991), remains at the maximum for the next two weeks, finally decreasing linearly to 100 g at the final period in the nest. Each nestling fledges successfully after 6 weeks from hatching, if it has accumulated a total of 12 kg of food throughout that period, assuming it has not starved (less than 50% of the accumulated values for a period of 5 consecutive days). Intraspecific competition in the nest has been taken into account (i.e. older chicks are assumed to have an advantage in feeding (Mock et al., 1987)). Great Blue Heron Colony Model. The specific details of the Great Blue Heron (Ardea herodias) model are as follows: Prey-base sub-model. Great Blue herons are long-legged, visual feeders foraging both in shallow and deep-water areas (Palmer, 1962). Their foraging mode is to usually stand and wait (Kushlan, 1978). The maximum depth for wading is set to be 39 cm (Palmer, 1962). They are diurnal and nocturnal feeders (Krebs, 1974). The prey for Great Blue Herons consists mainly of fish (Frederick and Collopy, 1988). Behavior and Energetics Sub-models for potential nesting adults. The adult Great Blue Heron weighs 3.0-3.5 kg, with the male usually about 10% larger than the female (Palmer, 1962). The energy requirements are as follows: Basal metabolic rate: 62 Free living metabolic rate: 166 Assimilation efficiency: 75% The initiation of the nest is governed by rules similar to those in the previous models. A threshold of 600 kcal for a week is the required energy that a female must obtain in order to start a nest. Energetics and growth sub-models of nestlings. Each nesting pair lays 3 eggs, asynchronously, with a period of two days between eggs. Each egg is incubated for 28 days before hatching (Palmer, 1962). A nestling fed at the maximal rate can grow at the rate of 40 g/day reaching a weight of 400 g within 10 days, and a maximal weight in ~ 3 1/2 weeks (Owen, 1960), remains at the maximum for the next two weeks, finally decreasing linearly its daily demands to 150 g at the final period in the nest. Each nestling fledges successfully after 60 days from hatching, if it has accumulated a total of 15 kg of food throughout that period, assuming it has not starved (less than 50% of the accumulated values for a period of 5 consecutive days). Intraspecific competition in the nest has been taken into consideration. Depending on the size of the prey, contest or scramble competition among nestlings has been simulated. Wading Bird Nesting Colony Model Constraints. These models have been specifically constructed to evaluate various hypotheses concerning nesting colony success. Due to this focus, the time span in the model is one breeding season, and therefore a variety of assumptions must be made to evaluate colony success over several years. In particular, there is no model component dealing with survival outside of the nesting season. The model does not include any adult or nestling mortality factors from disease or predation, and currently does not include any mixed-species flocking. Additionally, coupling of the models to specific hydrology and landscape data is very preliminary, as is the coupling to the fish and macroinvertebrates models. Despite the above constraints, many aspects of the model outputs may be directly compared to observational data on wading bird foraging behavior and nesting success. American Alligator Model. Background considerations. The American alligator (Alligator mississippiensis) is a keystone species in the Everglades and Big Cypress Swamp, as defined by its role as a top-level carnivore and architect within these systems and its influence on the structure, distribution and abundance of native plant and animal communities (Craighead, 1968; Kushlan and Hunt, 1979; Loftus and Eklund, 1994). Although the alligator is a large, mobile carnivore that represents a versatile and selectively opportunistic predator, it depends upon stable resources within its local environment, e.g. presence of surface water and related prey resources (Spotila et al., 1972). During a long life span of nearly continuous growth, an individual feeds on a variety of prey species and sizes that vary by an alligator's size. As alligator populations consist of overlapping size classes, they are consumers at all trophic levels (Murphy, 1981). The alligator also modifies its environment through construction and maintenance of 'alligator holes' to regulate its body temperature (Craighead, 1968; Spotila et al., 1972). These holes also serve as critical dry-season refugia for a variety of other aquatic animals (Neil, 1971; Craighead, 1968; Loftus et al., 1990; Loftus and Eklund, 1994) upon which wading birds and other predators feed (Craighead, 1968; Kushlan, 1976). Because loss of such a keystone species can lead to drastic re-ordering and simplification of total ecosystem functioning (O'Neil et al, 1986), recent severe declines in the abundance of the American alligator in South Florida are of concern. These declines are attributed to alterations in the natural hydrologic regime of the region (Craighead, 1968; Hines et al., 1968; Hines, 1969; Kushlan and Jacobsen, 1984, 1990; Jacobsen and Kushlan, 1984, 1989; Fleming, in review). General Structure of the Model. A model is presently under construction to simulate alligator responses to varying hydrologic regimes in a variety of freshwater, local environments. These local environments include short- and long-hydroperiod wetlands (Figures 10 and 11). The general structure of the model is shown in Figure 12. The model consists of two parts ('modules'). The first module simulates the life stages of individual adult and subadult alligators and, in particular, nesting female alligators and their reproductive performance. The module produces daily data on adult and subadult locations and numbers within these wetland types, and tracks the state of each alligator, which includes: age, sex, weight, and various factors related to reproduction. The alligator data change daily, based upon the behavior and physiological responses of adult and subadult alligators to changes in air and water temperature, water depth, season, food intake, and other factors. The second module simulates the life stages of hatchlings and juveniles (modeled as cohorts) in typical nest areas. The module tracks their daily growth and survival based upon a set of environmental factors and stochastic survival probabilities. Adult and Sub-Adults (Individual-Oriented Module). Adults and sub-adults are modeled as individuals. The basic time step to model individual alligators is a day (or less, if necessary; see below). Food intake during each day is monitored, as well as the amount of energy spent for activities during the day (and the night). An (average) assimilation coefficient from feeding experiments is used to determine the daily energy budget. These budgets are accumulated over a longer time period (e.g. a month) and are then converted into growth or regression, depending on whether the accumulated energy budget is positive or negative (Jacobsen and Kushlan, 1989). Daily energy budget. In the current version, the model simulates an adult alligator under different environmental conditions (i.e. wetland type, ambient air and water temperature, inundation patterns and related prey resources) using time steps shorter than a day (presently 15 minute steps, as in the wading bird models). The model operates under the assumption that an alligator will try to achieve and maintain a preferred, core body temperature (CBT) (Smith, 1979) by making use of local topographical features (Figure 13). During the day an alligator will thus try to warm up (e.g. by basking in the sun) or to cool itself (by moving into shade, if available, or by submerging its body in a pond). In the current implementation, the model uses data from Kushlan (1979c) and Kushlan and Hunt (1979) for air and water temperatures in and near an alligator pond. The biophysical ecology (insulation, metabolic, and evaporative water loss rates, heat conductance to the ground, absorption of incoming direct or indirect thermal radiation) and the heat energy budget have been determined by Spotila et al. (1972) (see also Tracy, 1982). Starting from these data, the model simulates the transport of heat within the alligator's body, in particular how much and how fast heat is transported to the core of the alligator's body. The dominant mechanism involved in heat transport is blood circulation. A simple heat transport model is used with blood circulation parameters as input data (e.g. heart beat frequency, heat exchange rate between blood stream and surrounding tissue, etc.). Average parameters are currently used (Smith, 1979; Smith and Adams, 1979). A refinement of this model will be made at a later date, when the relevant input parameters have been determined. Food intake. The availability of food (primarily fishes, aquatic insects and snails) is simulated using outputs from the models for fish and aquatic macroinvertebrates (currently a dummy table is used). Food intake rates also depend on an alligator's CBT and its current nutritional status. The model assumes that alligators do not eat every day, but capture some large prey items once every several days, and a few smaller prey items in between. The smaller prey items are treated on an average basis, whereas the larger prey captured are treated as separate and distinct food items. The model assumes that these larger prey are 'offered' to the individual alligator, which accepts or rejects them, depending on its state variables, e.g. satisfaction/hunger, CBT. Given the food intake and the composition of the diet, the model then calculates the energy budget of the animal using food conversion rates and maintenance energy requirements determined from feeding experiments (Table 4) (Staton et al., 1990, 1992). Courtship and nesting. A female chooses a nest area and, under normal conditions, stays there. It can, but does not have to, leave its nest area if it becomes flooded or dries out. Nest areas are categorized according to some basic characteristics, such as location, duration of inundation, food availability, suitability for building a nest, etc. This classification is used as a basis for defining more detailed rules specifying female behavior at a nesting site, and reviewed in the following section. The timing of courtship and nesting are correlated with ambient air temperature (Kushlan and Jacobsen, 1990) and water levels (Fleming, in review). These variables are used to trigger nest construction in the simulation. Behavior, energetics, and related reproductive performance of nesting females. Food intake by a potential nesting female alligator is simulated by the food intake rules described previously for adults and subadults. A nesting female alligator's energy budget, however, is determined by the caloric value of its total food intake, basal metabolism and/or growth, and cost of egg production. Reproductive performance or potential hatchling yield of a nesting female depends on the amount and the composition of its diet (Cardeilhac, 1991). Nesting rate is determined by protein content of the diet during egg follicle development (Cardeilhac, 1991) which, in South Florida, occurs during the late dry to early wet season (March through mid-April) (Fleming, in review). Clutch size varies in relation to the fat content in the diet during this same period. Infertility rate varies in relation to the relative proportions of fat and protein content in the diet annually (Cardeilhac, 1991). In the model simulation, the fat and protein content of the food intake are recorded on a daily time step. These values, suitably averaged over these periods prior to nesting, are then used to determine whether the female will nest, and if so, its clutch size and the rate of egg infertility. Because prey availability will differ among wetland types and also under different hydrologic conditions at each of these sites, the amount and composition of the alligator's food intake will change, thereby affecting its reproductive performance. Potential hatchling yield is therefore estimated as a function of diet and represented by a female's nesting rate, clutch size, and the rate of egg infertility: potential hatchling yield per female per year ~ nesting rate * clutch size * (1 - egg infertility rate). Factors other than egg infertility may also affect egg hatching rate. These factors include nest flooding, arrested embryonic development, predation, egg crushing, and desiccation. With the exception of egg loss from nest flooding, egg mortality from the other factors is low and relatively constant from year to year (Fleming, in review), and are therefore ignored in the model's present version. The probability of nest flooding during the incubation period is a function of egg-clutch height above prevailing water levels at the time of egg deposition, and subsequent water-level rise during the incubation period. Clutch elevation is modeled using a regression relationship of egg-clutch height to average water level at the time of nest construction (from Kushlan and Jacobsen, 1990). This regression model, in combination with simulated daily water levels (inputs from a hydrologic model), is used to determine whether a nest becomes flooded or not during the following 60 day incubation period (July-August) of each year. Simulated egg loss in a nest due to flooding is then subtracted from potential hatchling yield to estimate actual hatchling yield per nesting female. In the current implementation, the model assumes the sex ratio of hatchlings to be 1:1. However, this ratio will be changed in future model versions to account for temperature-dependent sex determination of the developing embryo during the incubation period (Ferguson and Joanen, 1982) Since hatchling yield of nesting females in a population is a major determinant of overall population size increases or decreases (Nickols et al., 1976) (all other factor being equal), different hydrologic regimes can be evaluated by their estimated effect on this index of potential recruitment alone. Hatchlings and Juveniles. Alligator hatchlings and juveniles stay near their nest the first two (possibly three) years. Mortality during these years is high (approximately 60%) (Jacobsen and Kushlan, 1984). Hatchlings and juveniles can therefore be simulated as cohorts (with no discrimination between males and females) because mortality is a chance event that randomly eliminates individuals from the various cohorts, i.e. year classes. If we assume that individual properties are distributed by a well-behaved probability distribution in the space of state variables, mortality is effectively equivalent to a random mixing process. We therefore end up with cohorts of hatchlings and juveniles that are random samples from some initial ensemble. Variations between different types of nest areas and the resulting variability in survival of hatchlings and juveniles will, however, be taken into account. The size of hatchlings and juveniles after one and two years also depends on the type of nest area and environmental conditions, such as climate, hydrologic conditions, and food availability. Catastrophic Events (Dispersal and Mortality related to Drought). Alligators rely on the availability of (open) water to regulate their body temperatures. If their ponds dry down completely, female alligators will usually disperse from their nest site. Other alligators in the surrounding marsh will also disperse with drying of the marsh. During their travel, alligators run the risk of desiccation. Therefore, the model assumes a mortality that increases with increasing distance to the closest refugia. The high density of alligators and the lack of food at these sites lead to antagonistic behavior and, frequently, to cannibalism. Smaller-sized animals run the greatest risk of being preyed upon. Therefore, the model assumes that the risk of being eaten decreases with increasing size. If a female alligator chooses not to disperse from its nest site but rather retreated to its underground den, its survival probability will be greater during drought conditions, as long as the ground water level does not fall below the deepest point in the den. This being the case, the den will retain some water and enable the alligator to remain cool. Depending on this below-ground depth and groundwater level, the den may dry; initially evaporation may keep the den and the alligator cool, but evaporation will eventually cease, resulting in the death of the alligator. Model Constraints. The relationships between the amount and composition of the alligator's diet and its growth and reproductive performance have been determined for captive animals in experimental studies. The model is largely based on these relationships and therefore assumes that these relationships are still true for animals in the wild. If so, the model will provide a reasonable estimate of alligator growth rates and reproductive performance for the hydrologic conditions under evaluation. Because the module for adults and sub-adults is based on individuals, it provides a variety of outputs that may be directly compared to the field data on individuals, in addition to comparisons of data on nesting success under differing environmental conditions. The model does not include detailed individual-based components for the hatchling and juvenile stages. Considerable effort must still be made to couple the model to landscape and hydrology data sets, as well as to integrate it with the fish and macroinvertebrate models. TERRESTRIAL-BASED FOOD WEBS Landscape/Vegetation (Vascular Plants) Model Background Considerations. The vegetation of South Florida responds at multiple temporal and spatial scales to a variety of forces, including seasonal variation in water level, interannual variation in water levels which results in unusually wet or dry years, fires, human- and animal-caused disturbances, hurricanes, freezes, etc. The current focus of the ATLSS project is the short-term, intra- and inter-annual variations in water level and hydroperiod which affect vegetation phenology, and the quantity and quality of primary production available to consumers. Longer-term changes such as succession and recovery from hurricanes and fires, peat accumulation, and vegetation-geomorphology feedbacks will be addressed by other modeling efforts, including small-scale individual-based models of plant growth and dynamics (Huston, 1991), and larger-scale biogeochemical and geomorphological models (F. Sklar, pers. comm). The goal of the current landscape/vegetation modeling effort is to reproduce the spatial and temporal variation in forage available to the Everglades and Big Cypress deer herds. Properties of these deer populations that are relevant to their interaction with vegetation include: 1) extremely low densities over much of the area (Fleming et al., in review); and 2) long-term stability in population size (Smith and Bass, 1994). Deer diets and growth energetics have been extensively studied in locations such as Michigan and Texas, but virtually no data on forage availability and energetics are available from South Florida (R. Labisky, pers. comm.). Available data do confirm that, like deer everywhere, deer in South Florida obtain the majority of their food intake from a very small number of plant species (Harlow, 1965). Based on rumen samples collected in the Everglades over the period of November to January, 65% of the volume of deer forage was comprised of four species (pond lily, royal fern, swamp lily, and willow) (Loveless, 1959, cited in Harlow, 1965). Clearly, the component of vegetation relevant to deer population dynamics represents a very small proportion of the total plant community, in terms of both species and biomass. In the absence of any quantitative information about the temporal variation in productivity of deer forage in different vegetation types, our approach to modeling the vegetation has been to use the qualitative information available to create a framework that can accommodate quantitative data as they become available. Model Structure. Primary features of the vegetation model are: 1) emphasis on the small proportion of the total vegetation that provides the high-quality forage selected by deer; and 2) temporal variation in forage quality and quantity driven by hydroperiod dynamics. The vegetation landscape model, like other lower trophic level models in the ATLSS project, is not an individual-based model, but rather simulates the quantity of particular vegetation types present across the landscape. This approach sacrifices details, such as species composition, within the vegetation for the ability to model spatio-temporal vegetation dynamics over a very large area. For each vegetation type, the model tracks the availability of three forage classes based on "forage quality." Forage quality is defined by the amount of energy (kcal) available from the forage, which is a function of the total energy content, and the proportion of the total that can be released by digestion, or "digestibility." Total energy content varies much less than digestibility, so digestibility is the primary determinant of forage quality (Short, 1986). We use two classes of relatively high forage quality: I (2000 kcal/kg) and II (1500 kcal/kg), both of which are able to maintain deer weight or allow growth when they are available. A third forage class (800 kcal/kg) represents low quality forage that is always available, but is not of sufficient energy content to allow deer to maintain body weight. Although individual plant parts change seasonally in their digestibility as they initiate growth, mature, and senesce during the dry season, we have adopted the convention of leaving quality constant within a forage class, and varying the amount of forage in each class to reflect seasonal changes in the vegetation. Thus, a proportion of the forage mass in Class I is transferred into Class II as its quality declines. The amount of high quality forage increases rapidly at the beginning of the growing season, then declines gradually as foliage matures, followed by a rapid decline in quality during the dry season. Growth initiation and senescence are regulated by the hydroperiod, which varies across the landscape as a function of elevation and local hydrology. Because there are currently no quantitative data on the spatio-temporal availability of different forage types in the Everglades and Big Cypress Swamp, we use approximations based on anecdotal information from the Everglades and other parts of central and south Florida. Our basic assumption is that the oligotrophic nature of most biotypes within the Everglades and Big Cypress Swamp results in low availability of forage for deer. The densities for Everglades deer reported by Fleming et al. (in review), are two orders of magnitude lower than the densities supported by optimal habitat in Short's Habitat Suitability models (Short, 1986). Consequently, we expect that the availability of high quality forage in the Everglades is substantially lower than most of the scenarios used in HSI models of Florida deer. In such marginal environments, variation in forage quality should be manifested by regional variation in deer densities and weights, as reported by Fleming et al. (in review), and others, as well as by seasonal variation in deer densities, as deer move to areas with the highest availability of quality forage. Therefore, one of the primary tests of the landscape model of vegetation will be a comparison of the spatial and temporal variation in deer densities and body weights with the spatial and temporal patterns of forage quantity and quality across the landscape. White-Tailed Deer Model Background Considerations. The White-tailed deer (Odocoileus virginianus seminolus) is the only large (native) herbivore in the Everglades and Big Cypress Swamp. In addition, it is the main prey base for the endangered Florida Panther (Smith and Bass, 1994). Deer are selective feeders and depend upon patchy resources (Short, 1986). Consequently, population sizes are closely coupled to both vegetation and water level in the Everglades and adjacent wetlands (Loveless, 1959; Land, 1993; Fleming et al., in review). Water levels have effects on this consumer species both directly, due to restrictions imposed on deer movements by water level (Loveless and Ligas, 1959; Fleming et al., in review), as well as indirectly through effects on the quantity of quality forage (Loveless and Ligas, 1959; Harlow, 1965; Harlow and Jones, 1965; Schortemeyer, 1980). As the primary large ungulate in South Florida wetlands, deer also have the capability to alter the vegetation through the impact of foraging. Model Structure. The deer model component of ATLSS simulates the birth, growth, movement, and death of individual deer as they move across the Everglades and Big Cypress landscapes (Figure 14). Each deer is simulated using an energetically based model that tracks energy input through food intake and energy loss through basal metabolism, movement, lactation, and other activities. Weight losses and gains, as well as size at maturity, are consequences of the amount and quality of forage available to each individual deer and the energy it must expend to obtain that food. Individual deer utilize the landscape at the scale of 100 x 100 m blocks, which is the resolution of the current vegetation model. At this scale, deer make decisions about where to move on the basis of forage availability and water depth, which can restrict their movement under some conditions. Most aspects of deer behavior and energetics are dependent on the size of each individual, while certain behaviors, such as rutting, lactation, seasonal activity, and social group dynamics, are determined by a combination of sex, age, and energetic status. Key background information for various model components is found in Short (1986). During a simulation, each deer begins the day by choosing where to start foraging. This decision is based on the assumption that a deer knows where in its local area the best forage can be found. From a n x n array of blocks centered on its current location (each block is 500 x 500 m, composed of 25 of the 100 x 100 m cells on which vegetation is modeled, and n is dependent on the maximum daily searching distance), the deer chooses the nearest block that has quality Class I forage above some threshold amount. If no Class I forage is available, the deer chooses the nearest block with Class II forage. There is a maximum total distance allowed for movement each day, with different values for fawns, does and bucks. Once a block has been selected for foraging, a deer moves to the 100 x 100 m cell within that block that has the largest amount of quality forage available. The deer moves among the 100 x 100m cells, consuming forage at a rate that is a function of its size, up to a maximum daily amount that is also a function of its size. If a single 500 x 500 m block of cells contains sufficient high quality forage to meet the deer's daily energy requirements, it consumes what it needs and remains in the block until the next day. If the cell does not contain sufficient forage of appropriate quality, and if time and distance limitations have not been reached, the deer moves to the next nearest block with quality forage, and so on. Each day, the deer's energy balance is calculated to determine whether it gains, maintains, or loses weight. Energy losses include a basal metabolic rate that is a function of deer weight (parameters for South Florida deer are not available, so general values derived from the literature are used), and energetic costs associated with foraging, travelling, and seasonal energy demands such as lactation and rutting. Energy gains come solely from foraging. Deer energetic parameters used in the model are listed in Table 5. Deer are able to gain or maintain weight if they consume forage Classes I and II, but lose weight if only Class III forage is available. Weight gains will vary depending on how far deer have to move to obtain sufficient forage, and other energy losses such as lactation or rutting activities. Each deer's weight is partitioned into lean body mass (muscle) and fat, with deer able to gain or lose differing proportions of fat or lean mass (with different conversion rates of calories to mass) (Torbit et al., 1985). Many physiological and life history processes are scaled as a function of physiological status, including probability of conception and successful birth of fawns, probability of twins, lactation amount, and rutting success. A deer's physiological status and probability of successfully completing various activities decreases with weight loss. Starvation mortality for an individual occurs when weight decreases to 70% of maximum lean body mass previously attained. (Torbit et al., 1985). Deer herd structure is based on multi-generational groups of females. Fawns travel with their mother until 18 months of age, at which time male fawns disperse. Females remain with their mother, foraging as a group if sufficient forage is available, but scattering if forage becomes scarce (Harlow, 1965b). Deer mortality occurs as the result of a number of processess, including starvation, predation, road hazards, catastrophic disturbances, disease, and age-dependent mortality. Food intake is decreased when movement is restricted by water depth (Loveless, 1959). In some parts of the Everglades, deer may become stranded on tree islands during extended periods of inundation (Fleming, in review). Movement across standing water is restricted as a function of body size, so fawns are affected most. Bucks are assigned a higher probability of moving out of stranding situations. Starvation may also occur due to seasonal regional depletions of high and medium quality forage, or (in fawns) due to poor nutrition of the mother. Mortality due to panther predations is modeled explicitly as a function of the encounter rate between panthers and deer. The bobcat is another important predator for deer in parts of the Everglades (Boulay, 1992). Bobcats are not modeled individually, but are included as a stochastic factor along with other sources of mortality. Spatially-explicit mortality risks due to road hazards and hunting are currently being added to the model. Each simulation run of the ATLSS landscape model of hydrology/vegetation/deer/panther interactions is initiated with a specific number of deer and panthers. As the model is run over a number of years (with water levels driven by the Natural Systems Model or the Water Management Model), populations increase or decrease as a result of births and deaths until they approach a steady-state that represents the carrying capacity for the system as it is defined by vegetation parameters, deer and panther parameters, and the hydrologic regime. A typical model run may stabilize with approximately 10,000 individual deer and 30 panthers. Model Constraints. Because the model is based on individual animals, there are many levels of output at which the model can be tested. These levels include: 1) regional variation in deer densities; 2) regional variation in deer size and age distributions; 3) regional, intra-, and inter-annual variation in fecundity, fat content, percentage of twins, weight at weaning, etc.; and 4) intra- and inter-annual variation in deer distributions in response to climate and hydrology. Given the current lack of quantitative data for important deer and vegetation variables in South Florida, we do not expect the model to perform well at the level of an individual deer's physiological status. Model development and parameter refinement must proceed in close collaboration with ongoing research on vegetation, deer, and their predators. At this stage, however, we do expect the model to reproduce the major spatial and temporal variations in deer density in relation to inundation patterns and related forage characteristics. These model predictions can be tested against the quantitative data available (e.g., Boulay, 1992; Sargent, 1992; Zultowsky, 1992; Smith and Bass, 1994; Fleming et al., in review). Florida Panther Model Background Considerations. As the largest terrestrial carnivore in the Everglades and Big Cypress Swamp, and an animal that depends upon large home range sizes, the Florida Panther (Felis concolor coryi) serves as a key species for assessing the health of these ecosystems. A spatially explicit, individual-based model of the Florida panther population has been constructed to be coupled with the White-tailed deer model previously described. The individual-based model for deer in the Everglades and Big Cypress swamp allows the deer population to be simulated as individuals daily, on a landscape divided into 500 meter x 500 meter spatial cells, with varying vegetation and water depth characteristics. Data from a hydrology model determines changing water depths in each cell on a regular basis (this may be daily, weekly, or monthly). The deer model proceeds, as described above, to produce daily data on deer distribution, and tracks the state of each deer (including location, age, sex, weight, maximum weight attained, and various factors related to reproduction). The deer data change daily, based upon the behavioral and physiological responses of the deer to changes in water depth, the quantity of quality forage, season, and other factors. Model Structure. This spatially explicit model of the Everglades and Big Cypress Swamp landscapes has been extended by adding individual panthers. Each panther is assigned a state that includes: age, sex, weight, maximum weight ever attained, location, whether it is at a kill site or not, and number of days since last kill. Additionally, there are sex-specific states. For a female, the state may include an associated set of kittens (linked to the states of these kittens), the number of days pregnant, and a next mating date if she is not pregnant. For males, whether or not a female is within mating distance is a state variable. Each panther is assigned an initial value for all of the above components of its state (Figure 15). Individual panthers utilize the landscape at a scale of 500 m resolution, which is the between-days scale at which the deer model operates (although the deer forage within a day at 100 m resolution). At this scale, panthers move daily based upon specific behavioral rules which allow for (a) short-distance search for prey, (b) remaining at a kill site until the kill is either eaten or spoiled, (c) intermediate scale and long-distance dispersal when local searches for prey are unsuccessful, and (d) long-distance dispersal in search of a mate (for males). These movements are affected by water depth, which can restrict movement under some conditions, as well as by underlying vegetation habitat type. During a simulation, each panther starts the day at the location it ended the previous day. If this was a kill site, and meat remains that is not spoiled (spoilage time depends upon season), it remains at this site. Otherwise it searches neighboring pixels for deer, with a certain probability it detects deer presence (this probability depends linearly on local deer abundance up to a maximum value). If a deer is detected, with a certain probability the panther successfully ambushes the deer. If a deer is not captured, the panther moves to a neighboring pixel, which becomes the center for a new search for deer. This process continues until a deer is captured, or until a panther's maximum daily movement distance has been reached. Water depth limits these movements, with different limitations for kittens, adult males and adult females. If the above searching procedure does not provide a deer capture within 14 days, the panther relocates to a randomly chosen pixel (of one of five habitat types appropriate for panther resting sites) within its maximum dispersal distance. If no deer are captured in a 21-day period, the panther relocates to a randomly chosen pixel (of one of the five habitat types) anywhere on the landscape. The model allows one to calculate home range readily just as one could from field data on daily position from radio-tracking information, as well as compare model results to deer mortality data (e.g. Land, 1986) and data on panther food habits (Dalrymple and Bass, 1994). Each day, the panther's energy balance is calculated to determine whether it gains, maintains, or loses weight. Energy losses include a basal metabolic rate that is an allometric function of panther weight (parameters for Florida panther are not available, so general values derived from the literature on puma are used as listed in Table 6), and energetic costs associated with searching for prey and dispersal. Maximum daily food intake is also an allometric function of weight, and is obtained only from recent deer kills. On days when a panther is unsuccessful in searching for deer, the model assumes that it will successfully meet its daily energy requirements 40% of the time (from other prey species of the panther including lagomorphs and small carnivores). The other 60% of the time, the panther's weight drops according to a uniform distribution. Mortality occurs in the model due to: starvation, aggression, dispersal (primarily assumed to be due to automobiles), and natural mortality (e.g. disease). The model has the capability to include mercury effects and road hazards, but these are not yet implemented. Death due to starvation is assumed to occur when body weight drops to 70% of the maximum attained, and if a mother dies, her kittens younger than 10 months will also die. Aggression leads to death of a panther with a certain probability, when two panthers occupy the same pixel and one is at a kill site. Which panther dies depends upon its sex, age, and whether the panther is pregnant or has kittens. Long-range dispersal leads to panther death with a certain fixed probability. When a female comes into estrus, she is serviced with a probability proportional to the number of males within mating distance to her. If she does not become pregnant one month, then one month later she enters estrus again. Pregnancy lasts a fixed-time period, and litter size follows a distribution with a mean of two. Once a month, males check to determine if they are within mating distance of a female. If not, the male disperses with a certain probability to a random pixel within mating distance of the nearest female. Each simulation run of the panther model takes into account hydrology/vegetation/deer/panther interactions and is initiated with specific numbers, locations, and states for all deer and panther across the landscape. Models are run with hydrology input files from either of the hydrology models for South Florida, and the panther population changes through time due to hydrology effects on vegetation and deer, stochastic demographic factors operating within the panther population, and the current structure of the population. To remove effects of initial conditions, the model is run for an appropriate time period (determined from the simulations). The predicted spatial distribution and abundance of panthers from this initial model run is then used to initialize all further simulations, with a single set used to compare alternative hydrologic scenarios. Since the model contains stochastic components for individual behavior, as well as for mortality and reproductive factors, a number of simulations must be run for each alternative hydrologic scenario. Model Constraints. As with the deer model, since the panther model is also based on individual animals, there are many levels of output at which the model can be tested. These include: (1) detailed comparisons of panther movement and activity area, as calculated from the model to those estimated from field data; (2) regional variation in panther success, as measured by such factors as survival and fecundity in various basins; (3) regional variation in size and age distributions; (4) causes of mortality, as compared to those observed in the field; and (5) seasonal changes in movement patterns, as affected by hydrology. Given the general lack of detailed physiological and behavioral data on Florida panther, it is unreasonable to expect the model to provide accurate assessments of such details. Of necessity, the model includes a number of assumptions about mating success, fecundity, mortality, and movement patterns for which inadequate data exist from the Everglades and Big Cypress areas. A key assumption in constructing the model is that data on panthers from many different locations in the U.S. (not necessarily the Florida panther), i.e. studies on single life-stage components, and observations of various behaviors, may be reasonably combined to provide a model which covers the animal's entire life cycle. However, the model has the capability to be directly compared with data on panther distribution and movement (e.g. Smith and Bass, 1994). The model should be very useful and relatively accurate in predicting how panther movement and population structure vary regionally, and how this interacts with hydrology. Additionally, a proxy measure for individual panther health (current body weight/maximum body weight attained) is also readily obtained from the model as a means to assess the spatially explicit impacts of alternative hydrologic scenarios on individuals in particular basins. Clearly, the model also provides a means to carry out a risk assessment concerning the relative probability under alternative hydrologic scenarios for the panther population to remain above some viable level. Genetic sub-models for each individual can also be included to assess population viability in terms of its genetic variability. EMPIRICAL STUDIES IN SUPPORT OF ATLSS MODEL DEVELOPMENT Research, modeling, and monitoring of biotic responses to hydrologic manipulations are integral components in the effective restoration and management of South Florida wetlands. However, given the present extent of data gaps, time, and funding constraints, an efficient research effort is required to develop a predictive understanding of complex patterns of spatial heterogeneity and temporal variability within these wetlands. Not everything that can be studied and modeled needs be. Goal-directed research in support of modeling provides an efficient framework for identifying critical data gaps, and prioritizing empirical studies (field and laboratory) to address those needs. Such research will, by necessity, involve both short- and long-term studies. Model constraints, described in the preceding sections of this document, have identified several areas of priority research needs at the various trophic levels of the Everglades and Big Cypress swamp. These needs require either preliminary or further study to improve ATLSS capabilities to evaluate whole-system responses to hydrologic restoration efforts. Priority studies to address these research needs are summarized below, in ascending order of the trophic structure of these ecosystems. Landscape: Topography - fine-scale spatial variation in topography (e.g. solution hole depths and spatial autocorrelation) - spatial distribution of ponds and pond sizes - appropriate sampling of topography for the Big Cypress Swamp Hydrology - appropriate spatial sampling of hydrology throughout the year to evaluate validity of hydrology model inputs - spatial details on local variation in water depth at scales finer than 2km x 2km Lower / Intermediate Trophic Levels: Vegetation - baseline data on detritus and primary production of major biotypes in relation to variable water levels and nutrients in the water supply, particularly for forested wetland biotypes - caloric and nutrient composition of vegetation types as they vary seasonally and spatially - assessment of accuracy of current vegetation maps Periphyton - biomass, spatial distribution, and composition for unsampled biotypes - productivity estimates correlated with hydrology for unsampled biotypes Bacteria/fungi - biomass, production, and food web relationships for all biotypes - productivity estimates correlated with hydrology for all biotypes Microinvertebrates - contributions to food webs in all biotypes - biomass, production estimates, and relative abundances in all biotypes Selected fish and macroinvertebrates - life-history information on important fish functional groups, snails, prawns, aquatic and terrestrial insects, and crayfish - production estimates, energy efficiencies, and dispersal distances in all biotypes - species relative abundances, density, and biomass fluctuations in unsampled biotypes Higher-consumer Groups: Colonial wading birds. - species-specific prey capture rates in relation to water depth, vegetation structure, prey abundance, conspecific and intra-specific competition - species-specific nestling subsistence needs (cumulative total food intake) - species-specific maximum transport (amount of food) of nesting adults - frequency, size distribution and species composition of flock-feeding associations American Alligator. - seasonal food habits for all sex and age/size classes - digestive rates (stomach clearance rates) of major prey items identified in above study - chemical analysis of major prey items - growth dynamics, sexual maturity (size), and reproductive performance in relation to diet quality - population numbers (census) - size-specific survival rates in relation to resources - movement, activity patterns and habitat preferences of all sex and age/size classes, particularly for adults, and related energy requirements and expenditures White-tailed Deer. - seasonal food habits for all sex and age classes - temporal variation in the seasonal productivity and quality of deer forage by vegetation types Florida Panther. - energetic requirements and expenditures for individuals of all sex and age classes - mating success, fecundity, and mortality rates - movement patterns and related behavioral characteristics in relation to short - long distance dispersal in search of prey, mates, etc. Cape Sable Seaside Sparrow. - food habits (high intensity sampling during breeding season, low intensity for non-breeding season) - breeding numbers and territory size in relation to resources - natality, mortality, recruitment and productivity rates in relation to resources - dispersal rates of adults and fledglings among core and peripheral satellite populations FUTURE PLANS Planned work on the ATLSS approach to the freshwater wetlands of South Florida over the next three years (FY 95-97) will focus on four different tasks. These are: (1) completing modifications to existing ATLSS model components; (2) conducting model performance evaluations; (3) integration of all ATLSS model components for these wetlands into a common framework (an integrated, modular, hierarchical model); and (4) development of several additional model components for inclusion in ATLSS. Modifications to Existing ATLSS Model Components Lower trophic-level model components. The landscape/vegetation model will be modified to include: (1) high resolution pseudotopography; (2) a more recent and accurate vegetation map being prepared by the University of Georgia; (3) complete parameterization of plant variables; (4) incorporation of plant phenology; (5) incorporation of major types of disturbances into the landscape model: freeze severity and extent, hurricane effects, fire extent, intensity, and pattern; and (6) the capability of reading in simulated successional changes in vegetation across the landscape, as predicted from a series of plant community successional models under development as a separate project. Versions of existing cellular models of periphyton, meso- and macroinvertebrates will be calibrated for forested wetland biotypes, based upon available data. Additionally, nutrient variables will be included in all the process-oriented models developed for lower trophic-level functional groups of species. Mathematical expressions for rates of microbial methylation/demethylation (as provided by other researchers) will be incorporated into the bacteria/fungi process-oriented model scheduled to be developed (see below). Mercury uptake and fate will also be incorporated into the existing invertebrate models. Intermediate trophic-level model components. The structured-population, cellular models developed for five functional groups of macroinvertebrates and fishes will be modified to allow for movement of individuals among cells, as affected by hydrology and vegetation type. Mercury uptake and fate components will also be incorporated into these structured-population models. Higher consumer model components. The simulation capability of the alligator model will also be expanded. The alligator model will be further refined for simulating the complete life cycles of individuals for all sex and age classes. This will include adding drought-related dispersal and mortality (desiccation and cannibalism) which affect individuals of all sex and age classes in a population. Existing individual-based, wading bird nesting colony models will be modified to incorporate vegetation and fish model outputs. Preliminary individual-based, wading bird colony models have also been constructed which include mercury uptake and effects (Henry, 1992; Hallam et al., in review), and serve as a template for extension to other species. The panther model also has a mercury uptake and effects component, not currently implemented, but planned for further development. Pending additional funding above base-project budget level, mercury uptake and effects sub-models will be added to the individual-based alligator, wading bird, and panther models. Other recommended (but not critical) modifications to higher consumer models include: (1) incorporating in the deer/panther model several changes in behavior, landscape and model capabilities; (2) developing and incorporating individual-based feral hog and bobcat models into the deer/panther model; and (3) incorporating genetics into the panther model. These have not been included in the base-project funding estimates, but could be completed in the three-year time period noted above. Performance Evaluation of ATLSS Model Components: This involves three separate activities: (1) sensitivity analysis; (2) uncertainity analysis; and (3) observed versus predicted comparisons. Within the next three years, these evaluations will be carried out on the various model components of ATLSS, not on the integrated ATLSS system. Development of Additional ATLSS Model Components: Several new models are planned for development. Process-oriented models for several key lower-trophic-level functional groups of species not included in the present version of ATLSS are planned (e.g., bacteria/fungi, microinvertebrates, other macroinvertebrates). These modeling efforts, however, will only be done pending the funding of empirical studies and the timely availability of study results in support of this effort. The set of age- and size-structured models of intermediate trophic level groups will be expanded to include several invasive, non-indigenous fish species and selected macroinvertebrates (i.e., Apple Snail). These additional structured-models will also include mercury fate and bioaccumulation components. The set of higher consumer models will be expanded to include individual-based models for the Snail Kite and the Cape Sable Seaside Sparrow. Completion of the Cape Sable Seaside Sparrow model, however, will be dependent on continued funding of an ongoing, empirical study over the next three years. Inclusion of mercury uptake and effects in these additional individual-based models will depend on securing funds above the base-project budget level. In addition, a multi-species, multi-colony simulation capability for wading birds will be developed. Integration of Model Components in a Common Framework: Currently, the models are rather loosely linked across trophic levels. The lower trophic level models are used to drive some components of the intermediate trophic levels, and the intermediate level output is being integrated into the alligator and wading bird models. None of these efforts are closely linked to the landscape vegetation and hydrology models being used in the deer/panther model. Therefore, we will develop a fully integrated system that will allow a unified approach to questions which cross trophic levels, and which will also allow complete analysis of alternative hydrologic scenarios without requiring different model runs. This complete integration will be accomplished during the production coding of all models (FY 96 - FY 97), once an initial research development phase of each model has been completed. The computer code for the models will be in a standard language and have an object-oriented design to allow easy modifications to incorporate future research and the need for possible changes in algorithms. This modeling system will be designed to run on UNIX workstations or compatable computer hardware. A usable first version of an integrated system will be completed within three years. The fully integrated ATLSS will also be capable of including appropriate model outputs from other component systems within the region, aside from the freshwater areas. Additionally, a fully integrated system will facilitate comparisons of model results to complementary modeling approaches, such as those available from a trophic flow-network analysis. ATLSS Application The fully integrated ATLSS system for the freshwater wetlands of the Everglades and Big Cypress Swamp will be ready for application to restoration issues in FY98. The ATLSS approach could also be developed for other sub-regions of Central and South Florida. Work tasks and time frames for accomplishing these modeling activities for other sub-regions are discussed in the following section. APPLICATION OF ATLSS TO OTHER SYSTEMS The modeling domain of interest to the South Florida restoration initiative is the hydrologic basin of central and south Florida. The region encompasses approximately 18,000 square miles and is comprised of linked hydrologic drainages or basins that include the Kissimmee River basin, Lake Okeechobee, Everglades, Caloosahatchee River basin, Big Cypress Swamp, Florida Bay, and the Florida reef tract (Gunderson and Holling, 1991). The ATLSS model under development for the Everglades and Big Cypress basins also provide a useful template for modeling trophic dynamics within the other basins of this region. ATLSS can test the hypotheses that underlie hydrologic restoration scenarios, and may also be used to compare future effects of such scenarios on the biotic components of the system. Identification of threshold responses of the biota to changes in water flow regimes is crucial if hydrologic scenarios are to be efficiently designed and evaluated. Because the ATLSS approach uses appropriate (varying) scales at which trophic interactions occur, analyses of model outputs provide meaningful assessments of threshold responses at each trophic level, and therefore offer more reliable predictions of whole-system responses. Attempts to model trophic dynamics at the regional or macro-scale within the framework of a single model structure would severely restrict or limit the spatial and temporal resolution required to achieve meaningful predictions. The development of ATLSS models for the remaining major areas of environmental concern within the region, the Kissimmee River basin, Lake Okeechobee, Florida Bay, Florida reef tract (Figure 16), and coupled through appropriate model inputs and outputs, however, would provide meaningful assessments using appropriate scales. By coupling these meso-scale or sub-regional models, outputs could still be analyzed within the context of a regional water budget. In this regard, modeling approaches, such as the UFAEA model (Tate, 1990) or habitat-assessment methodologies (Mitchell, 1992) that are regional in scope but fail to incorporate causal mechanisms or linkages at appropriate scales, cannot provide the predictive capability required for reliable model outputs of trophic responses (Fleming et al., 1993). The scope of work required for expanding the ATLSS approach to other sub-regions involves the complementary development of hydrologic, landscape structure/vegetation, and related plant community successional models for these regions. Process-oriented, structured-population, and individual-based models can then be developed and coupled for simulating trophic responses in these basins to hydrologic and related vegetational change. Hydrologic and landscape models for all but one (reef tract) of these areas are already under development or planned for development. Therefore, additional modeling activities should focus on the development of process-oriented and structured-population models for lower and intermediate trophic-level taxa in those systems. Most of the individual-based models already developed or planned for development (alligator, wading bird, white-tailed deer, Snail kite) can be directly incorporated in ATLSS models for the other freshwater systems, where appropriate. These same models can also be used as a template for modeling similar vertebrate species, i.e. crocodiles, Osprey, Bald Eagles, etc., that occur in the estuarine areas (mangrove estuaries, Florida Bay). ATLSS model outputs for each sub-region could then be analyzed with diagnostic models (trophic flow network and multivariate analysis techniques) to evaluate such scenarios within the context of regional water and nutrient budgets for the Central and South Florida study area. ACKNOWLEDGEMENTS The work described in this document has benefitted from contacts with a very large number of biologists and modelers. In particular, Dona Neff and Mindy Nelson assisted in data file summarizations for the development of process-oriented and structured-population models for lower and intermediate trophic level organisms. Nancy Urban and Roxanne Conrow shared data for development of the invertebrate models. We specifically appreciate the expertise of Sonny Bass, Jim Schortemeyer, Dave Maehr, and Daryl Land in support of individual-based model development for White-tailed deer and the Florida panther. Modelers who have greatly contributed to the development of ATLSS include Ethel Jane Comiskey, Hang-Kwang Luh, Yiannis Matsinos, Moris Shorrosh, and Yegang Wu. Financial support from the National Biological Survey, the National Park Service, and the U.S. Army Corps of Engineers has been essential to this project. In addition, the State of Tennessee, the Institute for Environmental Modeling, and the Science Alliance Center of Excellence at the University of Tennessee, Knoxville, have also provided support. LITERATURE CITED Bancroft, G.T., A.M. Strong, R.J. Sawicki, W. Hoffman, and S.T. Jewell. 1994. Relationships among wading bird foraging patterns, colony locations, and hydrology in the Everglades. Pages 615-658 in: S.M. Davis and J.C. Ogden (eds.). Everglades: the ecosystem and its restoration. St. Lucie Press, Delray Beach, Florida. Bender, E.A., T.J. Case, and M.E. Gilpin. 1984. Perturbation experiments in community ecology: theory and practice. Ecology 65: 1-13. Bildstein, K.L. 1984. Age-related differences in the foraging behavior of White Ibises and the question of deferred maturity. Colonial Waterbirds 7:146-148. ------------ 1987. Energetic consequences of size dimorphism in White Ibises (Eudocimus albus). Auk 104:771-775. Boulay, M.C. 1992. Mortality and Recruitment of White-tailed Deer Fawns in the Wet Prairie/Tree Island Habitat of the Everglades. Unpublished Masters Thesis, Department of Wildlife and Range Sciences. University of Florida, Gainesville. 77 pp. Bovino, R.R. and E.H. Burtt. 1979. Weather-dependent foraging of Great Blue Herons (Ardea herodias). Auk 96:628-630. Browder, J.A. 1981. Perspective on the ecological causes and effects of the variable algal composition of Southern Everglades periphyton. South Florida Research Center Report T-643. 110 pp. ----------- 1982. Biomass and primary production of microphytes and macrophytes in periphyton habitats of the Southern Everglades. South Florida Research Center Report T-662. 49 pp. Cardeilhac, P.T. 1991. Diagnosis and treatment of infertility in captive alligators (II). Pp. 501 -555 in Aquaculture Report Series. Vol. I - III. 1987 - 1991. Florida Department of Agriculture and Consumer Services, Tallahassee, Florida. Custer, T.W. and D.W. Peterson. 1991. Growth rates of Great Egret, Snowy Egret, and Black-crowned Night-Heron chicks. Colonial Waterbirds 14(1): 46-50. Craighead, F.C. 1968. The role of the alligator in shaping plant communities and maintaining wildlife in the southern Everglades. Florida Naturalist 41 (and 42): 2-7, 69-74. Dalrymple, G.H. and O.L. Bass. 1994. Food habits of Florida Panther in Everglades National Park. Draft Manuscript. DeAngelis, D.L. and J.C. Waterhouse. 1987. Equilibrium and nonequilibrium concepts in ecological models. Ecol. Monog. 57: 1-21. ------------- and L.J. Gross (eds.). 1992. Individual-based Approaches in Ecology: Concepts and Models. Routledge, Chapman and Hall. 620 pp. Fennema, R.J., C.J. Neidrauer, R.A. Johnson, W.A. Perkins, and T.K. MacVicar. 1994. A computer model to simulate natural south Florida hydrology. Pages 249 - 289 in S.M. Davis and J.C. Ogden (eds.). Everglades: the ecosystem and its restoration. St. Lucie Press, Delray Beach, Florida. Ferguson, M.W.J. and T. Joanen. 1982. Temperature of egg incubation determines sex in Alligator mississippiensis. Nature (Lond.) 296:850-853. Fleming, D.M. In review. American alligator nest distribution, nest abundance, and reproductive performance in relation to landscape characteristics of the southern Everglades. American Midland Naturalist. ------------, W.F. Loftus, and M.B. Robblee. 1993. SFRC critique of the wetland value assessment methodology and community models (WVA) document and an alternative approach. South Florida Research Center, Everglades National Park report to the C & S F Project Comprehensive Review Study, U.S. Army Corps of Engineers, Jacksonville, Florida. 10 pp. -------------, J. Schortemeyer, W. Hoffman, and D.L. DeAngelis. In press a. Colonial wading bird distribution and abundance in the pre- and post-drainage Everglades landscapes. Oecologia. -------------, J. Schortemeyer, W. Hoffman, and D.L. DeAngelis. In press b. Colonial wading bird nesting in the pre- and post-drainage Everglades landscapes. Oecologia. -------------, W.F. Wolff, and D.L. DeAngelis. In press c. The importance of landscape heterogeneity to wood storks in the Everglades. Environmental Management. -------------, D.L. DeAngelis, and W.F. Wolff. In press d. The importance of landscape in ecosystem integrity: the example of Everglades restoration efforts. In: Perspectives on Ecosystem Integrity. Kluwer Academic Publications. -------------, J. Schortemeyer, and J. Ault. In review. White-tailed deer distribution and abundance in the Everglades. American Midland Naturalist. Frederick, P.C., and M.W. Collopy. 1988. Reproductive ecology of wading birds in relation to water conditions in the Florida Everglades. Tech. Rept. No. 30. Florida Cooperative Fish and Wildlife Research Unit, School for Forestry Research and Conservation, University of Florida, Gainesville. ------------ and M.W. Collopy. 1989. Nesting success of five ciconiiform species in relation to water conditions in the Florida Everglades. Auk 106(4):625-634. ______ and W.F. Loftus. 1993. Responses of marsh fishes and breeding birds to low temperatures: a possible behavioral link between predator and prey. Estuaries 16: 216-222. ------------ and G.V.N. Powell. 1994. Nutrient transport by wading birds in the Everglades. Pages 571-584 in S.M. Davis and J.C. Ogden (eds.). Everglades: the ecosystem and its restoration. St. Lucie Press, Delray Beach, Florida. Gunderson, L.H. and C.S. Holling. 1991. Central and southern Florida ecosystem modeling and reconnaissance study: results of July 1991 workshop. Arthur R. Marshall Jr. Laboratory, University of Florida, Gainesville. 34 pp. ----------- and W.F. Loftus. 1993. The Everglades, Pages 199-255 in W.H. Martin, S.G. Boyce, and A.C. Echternacht (Eds.). Biodiversity of the Southeastern United States: Lowland terrestrial communities, Chapter No. 6. John Wiley and Sons, Inc., New York. Haake, W. and J. Dean. 1983. Age and growth of four Everglades fishes using otolith techniques, Technical Report SFRC-83/03, Everglades National Park, Homestead, Florida. Hallam, T.G., T. Trawick and W.F. Wolff. In review. An individual-oriented model of toxicant effects on a wading bird nesting colony with applications to mercury contamination. Ecological Modeling, Special Issue on Resource Management. Hancock, J.A., J.A. Kushlan and M.P. Kahl. 1992. Storks, ibises and spoonbills of the world. Academic Press. London. 385 pp. Harlow, R.F. 1965. An evaluation of white-tailed deer in Florida. Tech. Bull. No. 5, Florida Game and Fresh Water Fish Comm. Tallahassee. 64 pp. ------------- and F.K. Jones. 1965. The white-tailed deer in Florida. Tech. Bull. No. 9, Florida Game and Fresh Water Fish Comm. Tallahassee. 240 pp. Henry, T.L. 1992. Effects of Mercury Contamination in a Wading Bird Nesting Colony. Unpublished Masters Thesis, University of Tennessee, Department of Mathematics. 132pp. Hines, T. 1969. The past and present status of the alligator in Florida. Proceedings of the Southeastern Association of Game and Fish Commissioners. 33:224-232. Hines, T., M.J. Fogarty, and L.C. Chappell. 1968. Alligator research in Florida: a progress report. Proceedings of the Southeastern Association of Game and Fish Commissioners. 22: 166-180. Hoffman, W., G.T. Bancroft, and R.J. Sawicki. 1994. Foraging habitat of wading birds in the water conservation areas of the Everglades. Pp. 585-614 in: S.M. Davis and J.C. Ogden (Eds.). Everglades: the ecosystem and its restoration. St. Lucie Press, Delray Beach, Florida. Huston, M.A. 1991. Use of individual-based forest succession models to link physiological whole-tree models with landscape-scale ecosystem models. Tree Physiology 9:293-306. ______ D.L. DeAngelis and W.M. Post. 1988. New computer models unify ecological theory. Bioscience 38: 682-691. Jacobsen, T. and J.A. Kushlan. 1984. Population status of the American alligator (Alligator mississippiensis) in Everglades National Park. Report to South Florida Research Center, Everglades National Park, Homestead, Florida. 50 pp. ----------- and J.A. Kushlan. 1989. Growth dynamics in the American Alligator, (Alligator mississippiensis) in the southern Florida wetlands. Copeia 1974: 993-996. Kahl, M.P. 1964. Food ecology of the Wood Stork (Mycteria americana) in Florida. Ecological Monographs 34:97-117. Katzir, G., A. Lotem, and N. Intrator. 1989. Stationary underwater prey missed by reef herons, Egretta gularis: head position and light refraction at the moment of strike. Journal of Comparative Physiology A (1989) 165:573-576. Krebs, J.R. 1973. Significance of head tilting in the Great Blue Heron. Nature 242:533-535. ----------- 1974. Colonial nesting and social feeding as strategies for exploiting food resources in the Great Blue Heron (Ardea herodias). Behavior 51: 99-134. Kushlan, J.A. 1976. Wading bird predation in a seasonally fluctuating pond. Auk 93:464-476. ----------- 1977a. Sexual dimorphism in the White Ibis. Wilson Bull. 89:9298. ----------- 1977b. Growth energetics of the White Ibis. Condor 79:3136. ----------- 1977c. Population energetics of the American White Ibis. Auk 94:114-122. ----------- 1977d. Foraging behavior of the White Ibis. Wilson Bulletin 89:342-345 ---------- 1978. Feeding ecology of wading birds. Pages 249-298 in A. Sprunt, J.C. Ogden, and S. Winckler (eds.), Wading Birds. Research Report 7. National Audubon Society. New York, New York. ----------- 1979a. Feeding ecology and prey selection in the White Ibis. Condor 81:376389. ----------- 1979b. Prey choice by tactile-foraging wading birds. Proc. Col. Waterbird Group 3:133- 142. ---------- 1979c. Temperature and oxygen in an Everglades alligator pond. Hydrobiologia, 67:267- 271. ----------- and M.S. Kushlan. 1975. Food of the White Ibis in Southern Florida. Fla. Field Naturalist 3:31-38. ----------- and B. Hurt. 1979. Limnology of an alligator pond in south Florida. Florida Scientist 42: 66-84. ______ and T. Jacobsen. 1984. Management of alligator nest flooding. South Florida Research Center In-house Report, Everglades National Park, Homestead, Florida. 18 pp. ------------ and T. Jacobsen. 1990. Environmental variability and the reproductive success of Everglades alligators. J. Herpetol. 24:176-184. Land, E.D. 1986. Big Cypress deer/panther relationships: deer mortality. Final Report of Study Number 7509. Bureau of Wildlife Research, Florida Game and Freshwater Fish Commission. 29 pp. Loftus, W.F. and J.A. Kushlan. 1987. Freshwater fishes of southern Florida. Bulletin of the Florida State Museum, Biological Sciences 31:147-344. ______ J.D. Chapman, and R. Conrow. 1990. Hydroperiod effects on Everglades marsh food webs, with relation to marsh restoration efforts. Pp. 1-22 in G. Larson and M. Soukup (Eds.). Proceedings on the Conference on Science in the National Parks, 1986. Vol. 6. Fisheries and Coastal Wetlands Research, Fort Collins, Colorado. 22 pp. ------------ and A.M. Eklund. 1994. Long-term dynamics of an Everglades small-fish assemblage. Pages 461-484 in S.M. Davis and J. Ogden (Eds.). Everglades: the ecosystem and its restoration. St. Lucie Press, Delray Beach, Florida. Loveless, C.M. 1959. The Everglades deer herd, life history and management. Tech. Bull. No. 6. Florida Game and Fresh Water Fish Comm., Tallahassee. 92 pp. ----------- and F.J. Ligas. 1959. Range conditions, life history, and food habits of the Everglades deer herd. Transaction North American Wildlife Conference 24:201-215. MacVicar, T.K., T. Van Lent, and A. Castro. 1984. South Florida Water Management Model. Documentation Report. Technical Publication 84-3. South Florida Water Management District, West Palm Beach. 123 pp. Meyerriecks, A.J. 1960. Comparative breeding behavior of four species of North American Herons. Publ. Nuttal Ornithol. Club No. 2. Miller, K.E. 1993. Habitat Use by White-Tailed Deer in the Everglades: Tree Islands in a Seasonally Flooded Landscape. Unpublished Masters Thesis, Department of Wildlife and Range Sciences, University of Florida. 105 pp. Mitchell, L. 1992. Wetland value assessment methodology and community models. U.S. F.W.S., Lafayette, Louisiana. 72 pp. Mock, D.W., T.C. Lamey, and B.J. Ploger. 1987. Proximate and ultimate roles of food amount in regulating Egret sibling aggression. Ecology 68: 1760-1772. Murphy, T.M. 1981. The population status of the American alligator on the Savannah River Plant, South Carolina. Savannah River National Environmental Research Park Publication, Savannah River Ecology Laboratory, Aiken, South Carolina. 20 pp. Neil, W.T. 1971. The last of the ruling reptiles (alligators, crocodiles, and their kin). Columbia University Press, New York. 486 pp. Nickols, J.D., L. Viehman, R.H. Chabreck, and B. Fenderson. 1976. Simulation of a commercially harvested alligator population in Louisiana. Louisiana Agricultural Experiment Station Bulletin 69: 59 pp. Noss, R.F. 1987. Protecting natural areas in fragmented landscapes. Natural Area Journal 7: 2-13. Ogden, J.C., H.K. Kale, and S.A. Nesbitt. 1980. The influence of annual variation in rainfall and water levels on nesting by Florida populations of wading birds. Transaction Linn. Society, New York, 9:115-126. O'Neil, R.V., D.L. DeAngelis, J.B. Waide, and T.F.H. Allen. 1986. A hierarchial concept of ecosystems. Princeton University Press, Princeton, New Jersey. 253 pp. Owen, D.F. 1960. The nesting success of the Heron Ardea Cinerea in relation to the availability of food. Proceedings Zoological Society London 133: 597-617. Palmer, R.S. 1962. Handbook of North American Birds, Vol. 1. New Haven and London, Connecticut, Yale University Press. Pennycuick, C.J. 1978. Fifteen testable predictions about bird flight. Oikos 30:165-176. -------------. 1989. Bird flight performance: A practical calculation manual. Oxford University Press. Oxford. 153 pp. ------------ and T. DeSanto. 1989. Flight speeds and energy requirements for White Ibises on foraging flights. Auk 106:141-144. Petit, D.R. and K.L. Bildstein. 1987. Effect of group size and location within the group on the foraging behavior of White Ibises. Condor 89:602-609. Pimm, S.L. 1991. The Balance of Nature? Ecological issues in the conservation of species and communities. University of Chicago Press, Chicago, Illinois. 434 pp. Pulliam, H.R., J.B. Dunning, Jr., and J. Liu. 1992. Population dynamics in a complex landscape: a case study. Ecological Applications 2:165-177. Quinney, T.E., and Smith, P.C. 1979. Comparative foraging behavior and efficiency of adult and juvenile great blue herons. Canadian Journal of Zoology 58: 1168-1173. Sargent, R.A. 1992. Movement Ecology of Adult Male White-Tailed Deer in Hunted and non-Hunted Populations in the Wet Prairie of the Everglades. Unpublished Masters Thesis, Department of Wildlife and Range Sciences, University of Florida. 66 pp. Schortemeyer, J.L. 1980. An evaluation of water management practices for optimum wildlife benefits in Conservation Area 3A. Florida Game and Fresh Water Fish Commission, Ft. Lauderdale. 74 pp. Short, H.L. 1986. Habitat Suitability Index Models: White-tailed Deer in the Gulf of Mexico and the South Atlantic Coastal Plain. Biological Report 82 (10.123) of the Division of Wildlife and Contaminant Research, U.S. Fish and Wildlife Service. Smith, E.N. 1979. Behavioral and physiological thermoregulation of crocodilians. American Zoology 19:239-247. Smith, E.N., and S.R. Adams. 1978. Thermoregulation of small American alligators. Herpetologica, 34:406-408. Smith, T. and O.L. Bass. 1994. Factors regulating spatial patterns by the Florida panther and white-tailed deer in the Everglades. Pp. 693-708 in S.M. Davis and J.C. Ogden (Eds.). Everglades: the ecosystem and its restoration. St. Lucie Press, Delray Beach, Florida. Spotila, J.R., O.H. Soule, D.M. Gates. 1972. The biophysical ecology of the alligator: heat energy budgets and climate spaces. Ecology 53: 1094-1102. Staton, M.A., H.M. Edwards Jr., and I.L. Brisbin Jr. 1990. Protein and energy relationships in the diet of the American alligator (Alligator mississippiensis). Journal Nutrition 120:775-785. Staton, M.A., H.M. Edwards Jr., and I.L. Brisbin Jr. 1992. The influence of environmental temperature and dietary factors on utilization of dietary energy and protein in purified diets by alligators, Alligator mississippiensis (Daudin). Aquaculture 107:369. Swartzman, J.L. and S.P. Kaluzny. 1987. Ecological Simulation Primer. MacMillan Co., New York. Tate, D. 1990. Adaptive Environmental Assessment Everglades Simulation Guide. Users Guide. The Arthur R. Marshall, Jr. Laboratory, Dept. of Zoology, University of Florida, Gainesville. 59 pp. Torbit, S.C., L.H. Carpenter, D.W. Swift, and A.W. Alldredge. 1985. Differential loss of fat and protein by mule deer during winter. J. Wildl. Manage. 49(1):80-85. Tracy, C.R. 1982. Biophysical modeling in reptilian physiology and ecology. Pp. 275-324 in C. Gans (Ed.). Biology of the Reptilia. Vol. 12. Academic Press, London. 536 pp. Turner, M.G., and R.H. Gardner (eds.). 1991. Quantitative Methods in Landscape Ecology. Springer-Verlag, New York. Ulanowicz, R.E. 1986. Growth and Development: Ecosystem Phenomenology. Springer-Verlag, New York. 203 pp. Wolff, W.F. 1994. An individual-oriented model of a wading bird nesting colony. Ecol. Modelling 72:75-114. Zultowsky, J.M. 1992. Behavioral and Spatial Ecology of Female White-tailed Deer in the Everglades Ecosystem. Unpublished Masters Thesis, Department of Wildlife and Range Sciences, University of Florida. 81 pp. LIST OF TABLES Table 1. An overview of the major trophic elements or compartments of the Everglades and Big Cypress landscapes for which simulation models are being developed. Table 2. ATLSS project's multi-year work tasks and study schedule. Table 3. Comparison of pre- and post-drainage characteristics of South Florida wetland landscapes. Table 4. Energetics parameters used in the individual-based American alligator model. Table 5. Energetics parameters used in the individual-based white-tailed deer model. Table 6. Energetics parameters used in the individual-based Florida panther model. An overview of the major trophic elements or compartments of the Everglades and Big Cypress landscapes for which simulation models are being developed. Italicized taxa are non-indigenous species. _________________________________________________________________ ________________________________ Trophic Elements or Compartments by Simulation Model Type _________________________________________________________________ ________________________________ Process-oriented Structured-population Individual-oriented _________________________________________________________________ ________________________________ periphyton * selected macroinvertebrates * ** American (prawn, crayfish, Apple snail) Alligator * bacteria/fungi ** American Crocodile microinvertebrates ** amphibians ** colonial wading birds * (protozoans, ciliates, (frogs, salamanders) (herons, egrets, ibises, rotifers, etc.) storks) reptiles ** (snakes, turtles, lizards) mesoinvertebrates * fish * ** Snail Kite ** (oligochaetes, copepods, (killifishes, livebearers, amphipods, etc.) centrarchids, gar, catfish, pike killifish, Mayan Cichlid, marine species) Cape Sable Seaside Sparrow ** macroinvertebrates * ** (aquatic and terrestrial insects, etc.) White-tailed Deer * Florida Panther * _________________________________________________________________ ________________________________ * models under development ** models to be developed Project's multi-year work tasks and study schedule. _________________________________________________________________ _____________________ Work Tasks 1994 1995 1996 1997 1998 1999 _________________________________________________________________ _____________________ I. Scopes of Work, Coop. Agree., etc. _______ II. Model Component Development: Quantitative model development _____________________ _ _ _ _ _ Model performance evaluation ___________________________ III. Model Component Integration: Serial Coded Version ____________ PVM Coded Version ______ IV. ATLSS Model Application to Restoration Issues ____ V. ATLSS Model Documentation, User Interface and Guide ______ V. Reports/Publications _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ________________ _________________________________________________________________ _______________________ A Comparison of Pre- Versus Post-drainage Landscape Characteristics of South Florida Freshwater Wetlands _________________________________________________________________ __________________________ Landscape Characteristics Pre-drainage Post-drainage _________________________________________________________________ __________________________ Wet season sheet flow Uninterrupted Sheet flow fragmented sheet flow by levees into impoundments and/or overdrained marshes Short-hydroperiod marshes Extensive Lost to development or degraded by drainage Seasonal changes in water Attenuated changes Pronounced fluctuation in water depths Major drydowns in sloughs Rare Frequent Dry season freshwater Greater flows Reduced flows flows into the estuaries _________________________________________________________________ ___________________________ Table 4. Energetics parameters used in the individual-based American alligator model. _________________________________________________________________ ___________________________ An alligator - consumes only 3 - 5 % of its total body weight per day - converts 40% of food eaten into body mass Daily maintenance requirement: - 6 - 8 kcal/kg live body weight - 0.5 - 0.9 g protein/kg live body weight Optimal digestible energy: crude protein ratio ~ 9 : 1 kcal/g protein Feeding ceases below 22 C Seeks heat after feeding to process food consumed more quickly: - rate of protein digestion and absorption two times faster at 28 C than at 20 C - rate of protein digestion and absorption three times faster at 30 C than at 15 C (Caiman crocodilus) Avoids heat during periods of fasting to reduce metabolic costs _________________________________________________________________ ___________________________ Table 5. Energetics parameters used in the individual-based white-tailed deer model. _________________________________________________________________ ___________________________ Maximum Forage Intake Rate (kg/hr) = 0.01164 x bodyweight**0.75 (decreased if forage drops below 5 kg/ha) Maximum Forage Amount (kg/day) = bodyweight**0.29 Forage Classes: I = 2000 kcal/kg II = 1500 kcal/kg III = 800 kcal/kg Metabolic Energy Requirements: Basal Metabolic Rate (kcal/day) = 87.1 x bodyweight**0.75 Travel Energy Costs (kcal) = (2.97 x bodyweight**0.66) x distance (km) Lactation = 1.25 x BMR Rutting = 1.12 x BMR Pregnancy = 1.25 x BMR Conversion of Energy to Body Weight: 1 kcal = 0.00021 g muscle 1 kcal = 0.000105 g fat _________________________________________________________________ ___________________________ Table 6. Energetics parameters used in the individual-based Florida panther model. _________________________________________________________________ ___________________________ Daily Food Requirement (kg/day) = bodyweight ** 0.75 Metabolic Energy Requirements: Basal metabolic rate (kcal/day) = 221.95 x bodyweight ** 0.75 Travel energy costs (kcal) = 5.695 x bodyweight ** 0.66 x distance (km) Male mating energy costs (kcal) = 1.30 x BMR Predation Energy Costs = 1.10 x BMR Lactation Energy Costs = 1.05 x BMR Pregnancy Energy Costs = 1.10 x BMR Conversion of Energy to Body Weight: 1 kcal = 0.00021 g muscle 1 kcal = 0.00105 g fat _________________________________________________________________ ___________________________ LIST OF FIGURES Figure1. Map of the south Florida restoration initiative study area and the Everglades and Big Cypress ATLSS simulation subarea (shaded). Figure2. Food web diagrams (from Gunderson and Loftus, 1993) showing known trophic relationships among characteristic Everglades and Big Cypress animals with (a) macrophyte base and (b) detritus/periphyton base. Shaded components of the food web are included in the ATLSS model(s) under development for these freshwater wetlands. Figure3. Schematic diagram of the general structure of the Across-Trophic-Level System Simulation (ATLSS). Figure4. Schematic diagram of the lower trophic-level models: mass-flow diagram. Figure5. Schematic diagram of the general structure of the model of fish and some aquatic macroinvertebrate population dynamics. Figure6. Schematic diagram showing the seasonal movements of fish between marsh and ponds/solution holes as the marsh is alternately flooded or drying. Figure7. Schematic diagram of the general structure of the wading bird nesting colony model Figure8. Schematic decision diagram of the rule used by a model Wood Stork to find a foraging site. Figure9. Schematic decision diagram of the rule used by a model White Ibis to find a foraging site. Figure10. Schematic diagram of typical alligator nesting areas within representative wetland land-facets. Figure11. Schematic profile of an alligator pond and surrounding environment. Figure12. Schematic diagram of the general structure of the alligator model. Figure13. Schematic decision diagram of the rule used by a model alligator to achieve a preferred core body temperature. Figure14. Schematic diagram of the general structure of the white-tailed deer model. Figure15. Schematic diagram of the general structure of the Florida panther model. Figure16. Map of the sub-regions of major environmental concern within the South Florida restoration initiative study area. 1. Kissimmee River, 2. Lake Okeechobee, 3. Big Cypress Swamp, 4. Everglades, 5. Mangrove estuaries of the lower Gulf coast and Florida Bay, 6. Florida Bay, 7. Florida reef tract. Appendix.Estimated Budget for NBS Modeling Activities for the Freshwater Wetlands of the Everglades/Big Cypress Swamp Sub-Regions of the Central and South Florida Comprehensive Review Study Area $ x 1,000 Work TasksFY95FY96FY97FY98TOTAL I. ATLSS Model A. Model Modifications: 1. Landscape/vegetation model 120 95 a. read-in capability for plant succession model outputs b. update vegetation map c. complete plt. var. param. d. incorporate plant phenology e. incorporate disturbance regimes for freeze, fire, hurricane 2. Lower trophic-level models a. calibrate Type I and II models for each functional group for all major forested biotypes b. add nutrient variables into the process-oriented models developed for each functional group c. add mercury fate into the process- oriented models developed for each functional group 3. Intermediate trophic-level models a. incorporate read-in capability for hydrology and vegetation b. incorporate macroinvertebrate and fish movement among cells 4. Higher trophic-level models a. incorporate complete life cycle for individuals of all sex and age classes into the alligator model, including draught-related dispersal and mortality b. single species wading bird models B. Model Performance Evaluations C. Model Integration D. Additional Models 1. Snail Kite 2. Cape Sable Seaside Sparrow 3. Wading bird multi-species, multi-colony simulation capability E. Model Application 1. PI salary support (3) 2. PD salary support (5) 3. hardware support CRITERIA FOR SELECTION (1) Abundant species or primary ecological stocks, e.g. periphyton, zooplankton, aquatic and terrestrial insects, macroinvertebrates, and fishes; (2) The keystone species that may "regulate" the flows described above, e.g., alligators, colonial wading birds; (3) The rare trophic elements, threatened or endangered species, e.g. Florida Panther, Snail Kite, Cape Sable Seaside Sparrow, Wood Stork, American Crocodile; (4) The invaders, i.e. introduced species, e.g. pike killifish, Mayan Cichlid, feral hog; (5) Functional groups or species important as food resources to selected higher consumers; (6) The availability of data on the life histories or cycles of organisms were other factors considered in the selection process; and (7) Species dependent on: large home ranges (Florida panther), patchy resources (colonial wading birds, Cape Sable Seaside Sparrow,white-tailed deer) stable resources (American alligator), and ecological specialists (Snail Kite) OBJECTIVE Develop a landscape-scale, ecosystem level modeling approach to: Investigate causes of trophic disruptions; Investigate minimal hydrologic thresholds (e.g. minimum area of inundation, hydrologic periodicity, etc.) required to maintain native plant and animal communities; and Based upon above, define a more functional landscape mosaic (area, heterogeneity, configuration, connectivity, and hydrologic periodicity) that meets the minimal spatial and temporal requirements of top-level carnivores and native animal assemblages The ATLSS Approach Hierarchical model structure to simulate whole-system responses, i.e. across-trophic-level responses at appropriate (varying) spatial and temporal scales at which trophic interactions occur; Different model types (resolution) at each trophic level; Coupling of these models across a spatially explicit (grid-based) landscape model of the region; this landscape model includes: (1)a landscape/vegetation submodel to simulate within-in year and between-year variations in vegetation properties in response to changes in hydrologic conditions and photoperiod, (e.g. rates of vegetation growth and regrowth, caloric and protein content of the vegetation, etc.); (2sub-models to simulate the effects of important disturbance regimes, (i.e. fire, freeze, and hurricanes), on the major vegetation types, (i.e. regrowth and succession following disturbance) by coupling to plant community succession models under development; ATLSS will be coupled to hydrology models and used to assess the effects of alternative proposed hydrologic restoration scenarios on trophic structure; Because appropriate model types and scales are used to simulate organisms' responses at each trophic level, reliable qualitative predictions are provided of whole-system responses.Animal Issues of Concern Decline in historical numbers of native and endemic taxa Changes in historical patterns of the spatial and temporal distribution of species Changes in patterns of community structure Changes in the pattern and magnitude of energy and material flows Most animal populations characterized by: reduced densities and biomass increased mortality reduced reproductive potential and juvenile survivorship changes in persistence and resilience These characteristics are usually associated with a fragmented natural landscape in which populations are frequently affected by: environmental stochasticity demographic stochasticity social dysfunction genetic deterioration Questions Concerning System Behavior (arranged in ascending order in the trophic structure): Question 1: How is primary production divided between algal and macrophyte communities? How is production related to hydroperiod? Question 2: What are the relative flows of carbon from detrital and grazing pathways to the higher trophic levels? Question 3: Are there differences in fish and macroinvertebrate biomasses between marshes that are, respectively, detritus-based or grazing-based? Question 4: How do landscape extent, heterogeneity, configuration, and connectivity influence the distribution, abundance and population stability of higher consumer populations?