Aerial Surveys Two types of aerial surveys are flown over the Ten Thousand Island study area to collect information on the distribution and abundance of manatees. Parallel strip transects (green lines below) are flown 8 times in the summer following protocols by Miller et al. (1998). Two observers in the plane independently record the location, activity and number of manatees observed in each group. Distribution surveys (yellow line below) are flown monthly following an irregular predefined path across the mainland waterways, the mangrove islands, and offshore bays. [larger version] [larger version]

Satellite Telemetry Satellite telemetry is used to document the movements of individual manatees over time. A total of 26 manatees (8 males, 18 females) were captured and radio-tagged at Port of the Islands, Collier County, Florida through December 2003. Most tagged manatees were fitted with Argos transmitters, providing approximately four location fixes per 24-hour period. GPS tags were used periodically to provide much more accurate locations than the Argos data (approx. 30 m vs. >150 m) every 15-30 minutes, but the battery life was much shorter (8 weeks vs. 6 months). A total of 17,753 high quality Argos fixes over 6,157 tracking days were recorded from 26 tagged manatees (see map below). [larger version] [larger version]

Seasonal Movements Manatees captured in the study area showed variation in migratory behavior. Most tagged animals (16 of 26) overwintered near their capture site. Four of 26 animals moved south to Whitewater Bay during cold events (see below). Four of 26 animals moved north to power plants in Fort Myers or Tampa. Anna:   06Jan02 - 25Jan02   20Dec02 - 08Feb03 Mitzie:   20Dec02 - 08Jan03 Tipper:   11Dec02 - 08Feb03 Surfer:   26Nov00 - 09Feb01 [larger version]

Interzone Movements All tagged manatees make regular movements between offshore seagrass beds and inland sites that provide access to freshwater. [larger version] Surfer - August 2000 [larger version]

Transition Probabilities To model interzone movements, telemetry data is formatted as mark-recapture data and coded for different ecological zones. The example below shows 8 weeks of "resightings" for one tagged manatee: Offshore (S), Travel (T), Inland (I), or no data (0). TSSSSSTSSSSSTT000000000000TTTTTTST0TTT0TSSSSSTTTSSII0T0S 1; SSS0SSSSS0SSSST000STSS0SSSTSSSSSS00I0SSSSSSSTS0I0SSS00SS 1; S0TTSTS0I000SSSTSTTSSSST0000SSSTSSTSS00I0S0SSSSSSSSSS0II 1; I0SSSSSS0S0SS0II00BBSTSSSTT0S00SS0T0SSSSS0I0BTSSSSSSS000 1; Multistate modeling (Lebreton & Pradel 2002) is used to analyze these data and provide movement probabilities between zones. This approach also provides a framework for testing hypotheses as competing models, and for analyzing covariates that may influence movement.

Home Ranges as Networks Manatees have large home ranges, but use a limited proportion of the habitat available to them. They avoid water deeper than 2-3 meters except when traveling. They also avoid shallow water that is not near deeper water, perhaps as a means of avoiding stranding and predation. Network data structures efficiently describe manatee movement pathways and home ranges (see map below). GPS data provides information on movement rates (see graph below) that is directly incorporated into the model. [larger version] Surfer - August 2000 [larger version]

Reinforcement Learning Simulated manatees learn which pathways provide reliable assess to freshwater by positive or negative reinforcement as they choose among alternative pathways. The Rescorla-Wagner model is a simple, field tested method of modeling reinforcement. Vn = C (VMax - Vi) Where V is the strength of association (between a node of the network and the availability of freshwater ),and C is a constant for the individual's rate of conditioning (learning).

Movement as a Monte Carlo Markov Chain process Transition probabilities obtained from the Multistate model are used to simulate the movement of manatees among ecological zones as a markov chain process.

Hydrology component Water quality data is available from a number of monitoring stations in the study area (see graph below). Salinity and water temperature can be modeled along the network used by the manatees. Ideally, an estuarine hydrology model (e.g. TIME) would be linked to the manatee model. As a surrogate for the hydrology model, we are using simple empirical models along the network to simulate changes in hydrology. [larger version] [larger version]

Model Validation Comparison of aerial survey distribution data with snap-shots of simulated population (see graph below from aerial distributional survey). Comparison of real and simulated coarse-scale seasonal home range estimates. Comparison of real and simulated fine-scale foraging movements. Face validation by biologists viewing real-time map displays. Comparison of population trend analysis based on transect surveys and demographic modeling. [larger version]