Structure and Parameterization of an Agent-based
Manatee Model for Southwestern Florida

Brad Stith¹, Dean Easton², Jim Reid², Lynn W. Lefebvre², and Don DeAngelis³

¹AScI Corp. under contract for U.S.G.S.

²U.S. Geological Survey,
Florida Integrated Science Center, Sirenia Project
412 NE 16^th Ave., Room 250, Gainesville, FL 32601

³University of Miami, Miami, Florida

Presented at the Manatee Population Ecology and Management Workshop,
1-4 April 2002, University of Florida, Gainesville, Florida.
 

Abstract

We are developing a spatially explicit, individual-based model of the Florida manatee (Trichechus manatus latirostris) in southwestern Florida. This model is part of the Across Trophic Level System Simulation program (ATLSS) that will be used to predict and assess the potential effects of altered hydrologic regimes on manatees in South Florida. Two "agents" will be simulated in the model; humans who alter the flow of freshwater in the study area, and manatees who respond to the alterations in the freshwater flow. Virtual manatees assess GIS map layers of water depth, sea grass beds, fresh water sources, and water temperature to influence their behavior. Analysis of telemetry data from GPS and ARGOS tags helps calibrate the model with movement parameters (e.g. speed), pathways (provided by network diagrams), home range classes, habitat preferences, time budgets, and behavioral rules.  Our current focus is on manatees using the Ten Thousand Islands and western coastal areas of Everglades National Park.

• Develop a spatially explicit, agent-based manatee model.
• Incorporate detailed GIS layers such as water depth, submerged aquatic vegetation, freshwater, and water temperature.
• Parameterize the model using a combination of detailed (15 min.) GPS telemetry data and coarser (6 hour) Argos data.
• Program manatees as intelligent agents with behavior that simulates real manatees.
• Use the model to predict changes in manatee abundance and distribution associated with different hydrologic restoration scenarios.

• Avoidance of water deeper than 6 meters.
• Movement pathways follow networks (memory maps) of nodes (destination points) and arcs (travel corridors).
• Variable-length, multi-day foraging bouts (3 - 7 days) followed by short-duration movements to freshwater drinking source.
• Foraging concentrated in offshore seagrass beds and inshore bays with healthy SAV.
• Cool season movement into inshore thermal refugia.

• Each individual has a unique network of preferred feeding, resting, drinking, and thermoregulating sites. These sites are connected by travel corridors.
• Maternal transmission of memory maps to offspring.
• Repeat visits to favored foraging areas to feed on new vegetative regrowth.
• Preferential use of select travel corridors.
• Inter-annual site fidelity to inshore thermal refugia during cool season.
• Occasional exploratory forays which add new sites to memory map.

Movement Among Habitat Types

Speed and Directionality of Movements

• Size and position of home range – both offshore and inshore.
• Different memory maps/networks.
• Preference for foraging offshore vs. inshore bays.
• Variable cold tolerance affects length of stay in thermal refugio.
• Frequency of use of freshwater sites.
• Propensity to switch freshwater sites.

Demographic Parameters

• Mortality and fecundity of different age-stage classes will be incorporated into the model to simulate population dynamics.
• Information on vital rates for southwest Florida will be supplemented with estimates from better-studied areas.
• Environmental stochasticity due to prolonged cold spells will affect mortality.
• Catastrophic mortality due to red tides and hurricanes will be considered.

Validation Methods

• Face validation by biologists viewing real-time map displays.
• Comparison of real and simulated coarse-scale seasonal home range estimates.
• Comparison of real and simulated fine-scale foraging movements.
• Comparison of aerial survey distribution data with snap-shot of simulated population.
• Comparison of population trend analysis based on transect surveys and demographic modeling.

Alternative Restoration Scenarios

Simulations will be run to investigate differences in distribution and abundance of manatees under different conditions.

• Restoration with seasonally pulsed water input at a few point sources.
• Restoration with seasonally pulsed water input at multiple point sources.
• Restoration with even sheet flow over broad areas.
• Drought years.
• High rainfall years.

Acknowledgements

Support for this project was provided by U.S.G.S. Place Based Funds. Logistical support and data were provided by Ten Thousand Island National Wildlife Refuge, Everglades National Park, Rookery Bay National Estuarine Research Reserve, and Florida Marine Research Institute.