estimation of manatee (trichechus manatus latirostris...

1415

Ecological Applications, 15(4), 2005, pp. 1415–1426q 2005 by the Ecological Society of America

ESTIMATION OF MANATEE (TRICHECHUS MANATUS LATIROSTRIS)PLACES AND MOVEMENT CORRIDORS USING TELEMETRY DATA

RICHARD O. FLAMM, BRAD L. WEIGLE,1 I. ELIZABETH WRIGHT,2 MONICA ROSS,3 AND SHERRY AGLIETTI4

Fish and Wildlife Research Institute, Florida Fish and Wildlife Conservation Commission, 100 8th Avenue, S.E.,St. Petersburg, Florida 33701 USA

Abstract. Effective stewardship of Florida’s coast requires, in part, detailed charac-terizations of ecological components of the marine system. Characterization of one com-ponent, manatees (Trichechus manatus latirostris), involves mapping its distribution andabundance and identifying features of the landscape that are intimately associated with itslife history. In this study, we developed a raster-based spatial model that transformed radio-telemetry point data to surfaces illustrating the areas manatees visited frequently for rel-atively long periods, called ‘‘places,’’ and areas visited frequently for short periods, called‘‘movement corridors.’’ This work involved (1) simulating manatee movement paths be-tween sequential telemetry points using a cost surface based on manatee bathymetric pref-erences that were derived from empirical associations between telemetry locations andwater depth, (2) distributing movement time to cells that were crossed by the simulatedmovement path, and (3) extracting cells from the movement-path map that qualified asplaces and those that qualified as corridors. Movement characteristics of wild and reha-bilitated animals were similar. Movement rates of males were significantly greater thanthose of either females with calves or females without calves. Mean number of visits percell and mean time per visit for the three adult classes were not significantly different.Males had the smallest mean patch size for places, and females without calves had thelargest but fewest places. Based on qualitative evaluations by field biologists, we concludedthat the model performed well in estimating places. The locations of movement corridorswere less certain, although reasonable, given that manatee spatial cognition was consideredsufficient to permit directed movement between places.

Key words: corridors; geographic information system; manatees; marine mammals; movementpatterns; places; raster model; spatial cognition; telemetry.

INTRODUCTION

The Florida manatee, Trichechus manatus latirostris,is an endangered marine mammal that inhabits fresh-water and estuarine ecosystems of Florida (Hartman1974, Powell and Rathbun 1984, O’Shea 1988, Rath-bun et al. 1990, Lefebvre et al. 2001). These near-shoreareas are also heavily used by humans, resulting inencounters between people and manatees. Coexistencewith humans has some adverse consequences, but whatconcerns scientists and managers most are manateemorbidity and mortality caused by collisions with wa-tercraft (Hartman 1974, Campbell and Powell 1976,O’Shea et al. 1985, Ackerman et al. 1995, Wright etal. 1995) and habitat loss (Campbell and Powell 1976,Smith 1993). Attempts at reducing these effects include

Manuscript received 13 July 2004; revised 9 November 2004;accepted 7 December 2004; final version received 3 January2005. Corresponding Editor: P. K. Dayton.

1 Present address: Photo Science, 9800 4th Street North,402, St Petersburg, Florida 33702 USA.

2 Present address: 4730 Toepfer Road, Middleton, Wis-consin 53562 USA.

3 Present address: 13013 Jesup Woods Court, Orlando,Florida 32824 USA.

4 Present address: 103 Springfield Drive, Bainbridge,Georgia 39819 USA.

the development of comprehensive manatee protectionplans by state and local governments, establishment ofslow-speed zones for vessels, and review of permitsfor coastal development (U.S. Fish and Wildlife Ser-vice 2001).

A significant component of protection planning andpermitting involves knowing where manatees aggre-gate (U.S. Fish and Wildlife Service 2001). Most ef-forts at selecting appropriate protection measures foran area include applying mapped manatee aerial surveydata to assess visually where manatees are observed inthe highest concentrations (Ackerman 1995). Becausethis method of interpretation is highly subjective, meth-ods were developed that increased consistency in theinterpretation of aerial survey data (Flamm et al. 2001).Telemetry data are also used by managers in decisionmaking as a tool for determining presence or absenceof manatees when delineating areas of possible manateeconcentration based on clusters of points. These ap-plications, however, did not exploit the full value oftelemetry data. These data also can be used to estimatelocations of movement corridors, movement extentsand rates, residence times, and differences in the move-ment behavior of males, females with calves, and fe-males without calves (Bengston 1981, Powell andRathbun 1984, Deutsch et al. 2003).

1416 RICHARD O. FLAMM ET AL. Ecological ApplicationsVol. 15, No. 4

Also of great interest to scientists and managers ishow manatee distribution and abundance will shift withchanges in the landscape. This shift will manifest itselfas a new spatial arrangement of places and corridors.Of most immediate concern is the future status ofwarmwater refuges. Manatees have long been knownto seek refuge from cold-water temperatures at warm-water discharge sites (Shane 1984) because they aresusceptible to hypothermia (O’Shea et al. 1985). Eco-nomic growth in Florida during the 20th century ledto the construction of electric-power-generating plantsthat discharged warm water along the coast. Manateesdiscovered these discharges and began to aggregate atthem when seeking refuge from the cold (Moore 1956,Hartman 1979, Irvine 1983, Reynolds and Wilcox1994). To some extent, manatees became dependent onthem and remained at the discharge sites during inter-mittent shutdowns (Reynolds and Wilcox 1986, Reyn-olds 2000), and they often returned to the same indus-trial warmwater discharges each year. The expansionof the power production industry in Florida is consid-ered one reason that the winter range of manatees hasextended northward from its more traditional SouthFlorida range (Moore 1951). Telemetry data is expectedto contribute to our assessment of the risk that a chang-ing landscape poses to the population of manatees.

When using UHF telemetry data, several of its at-tributes must be considered. First, locations estimatedby the ARGOS system have varying degrees of ac-curacy depending on the satellite’s angle over the ho-rizon and how many transmissions were received fromthe tag. Second, satellite availability limits data re-cording to about eight times per day in Florida, ofwhich we receive 3–4 good locations per day. Third,because transmission to the satellite cannot occur whenthe transmitter is submerged, biases are introduced. Forexample, we are less likely to receive accurate locationswhen manatees are moving than during feeding or rest-ing (Deutsch et al. 1998). Consequently, areas usedprimarily for movement will be underrepresented bytelemetry points and, therefore, will be identified withless confidence than foraging and resting areas, wherethe tag typically floats at the surface. Fourth, althoughtelemetry is effective in mapping locations, it tells uslittle about activities and paths taken between recordedlocations.

What makes telemetry data special is that it providesmany locations over time for individual animals ratherthan single locations of many animals at a single instantin time. Telemetry data have been used to quantifysurvivorship (Pollock et al. 1989a, b, Riley et al. 1994),habitat use (Bulova 1994, Kieffer and Kynard 1996),home ranges (Thompson 1994, Carroll et al. 1995,Quinn 1995), behavior (Enderson et al. 1995), move-ment differences by sex (Klenner 1987, Bulova 1994),and seasonal movement patterns (Folkow et al. 1996,Deutsch et al. 2003). These analyses typically involved

describing distributions of points over time or mark–resight analysis.

Tracking data can also contribute significantly to ourknowledge about how animals perceive the surround-ing landscape. Environmental cues guiding manateesmight include freshwater and warmwater gradients(Hartman 1979), areas unperturbed by boats (Smethurstand Nietschmann 1998), and visual landmarks such asbottom or shoreline features. How these cues interactto provide the manatee with a geographic picture isunknown. Furthermore, how environmental informa-tion is communicated among manatees during socialinteractions is unknown.

In this paper, telemetry data were used to develop amodel that generates movement paths of manatees inTampa Bay, Florida, USA. These movement paths wereused to map manatee places and movement corridors.A ‘‘place’’ is defined as an area frequented by manateesfor extended periods of time. Places include areas forfeeding, resting, and refuge. Corridors are areas reg-ularly visited by manatees for brief periods as theymove from place to place. Slow movement rates, i.e.,long residence times per grid cell, suggested a manateeplace, whereas rapid movement rates identified a move-ment corridor. Results are interpreted in terms of man-atee spatial cognition and their application to manateeprotection decision-making and landscape-level as-sessments.

METHODS

Study area

The study area was restricted to Tampa Bay and thesurrounding area (Fig. 1) even though movements oftagged manatees ranged from the Florida Panhandle tothe Florida Keys. Tampa Bay is a fairly shallow estuary,has abundant seagrass, and has several freshwater andwarmwater sources (Tampa Bay Estuary Program1996). Between 50 and 300 manatees are found in Tam-pa Bay all year round, depending on water temperatures(Wright et al. 2002).

Field methods

Manatees were captured by means of three methods:land-based captures at a power plant discharge canal,modified land-based captures, and open-water captures(Weigle et al. 2001). Captured manatees longer than230 cm were fitted with a belt that was tethered to abuoyant housing containing a VHF radio transmitterand a platform transmitter terminal (PTT) (Rathbun etal. 1987, Reid et al. 1995, Argos 1996, Deutsch et al.1998, 2003). Staff also tracked tagged manatees 1–2times per week to collect visual observation data (Wei-gle et al. 2001).

Six rehabilitated animals were included in this study.These included a female that was released in November1991 in Charlotte County and tagged at the TampaElectric Company (TECO) power plant discharge ca-

August 2005 1417MANATEE PLACES AND CORRIDORS

FIG. 1. Map of the study area, Tampa Bay, Florida, USA. Key: T, location of Tampa Electric Company Big Bend powerplant warmwater discharge; F, location of Florida Power Corporation Bartow power plant warmwater discharge.

nal; a female tagged and released in July 1996 in LemonBay, Sarasota County; a female released in Hillsbor-ough County and tagged at the Florida Power Corpo-ration (FPC) Bartow power plant; a female released in1993 and tagged in the Braden River, Manatee County;a male tagged and released in Pansy Bayou in SarasotaBay; and a male released in 1990 in HillsboroughCounty and tagged at TECO. None of these animalswere calves when captured and rehabilitated.

Analytical methods

Data from Service ARGOS were processed to gen-erate a digital map of points in the geographic infor-mation system (GIS) ArcInfo (Environmental ResearchSystems Institute, Redlands, California, USA). We ex-amined the GIS map and tabular data and eliminated

duplicate records, obvious locational outliers, and lo-cations having accuracy classes other than 1, 2, or 3(Argos 1996). If two points occurred within 90 min ofeach other, then the one with the lower accuracy classwas removed from the database. Locations for visualobservations collected during tracking were enteredinto the GIS by generating a point coverage using thelatitude and longitude coordinates and attaching its at-tribute information. The PTT and visual databases wereconverted to Universal Transverse Mercator projectionusing the North American Datum 1927 and then joinedinto a single map of point locations.

The goal of our study was to estimate the locationsof manatee places and movement corridors. Mappingmanatee places and movement corridors involved fourprincipal steps. First, we prepared base maps of the


TABLE 1. Weights used to transform the Tampa Bay, Flor-ida, USA, bathymetry map to a cost surface.

Bathymetry class Weight

0–1.8 m 10.1.8–3.6 m 15.3.6–5.5 m 24.5.5–9.1 m 33.9.1 m 65Exposed at low tide 13Dredging spoils 16Inland waters 22Channels 24Areas of frequent change 27

Notes: The greater the value, the more ‘‘expensive’’ or lesslikely a manatee will cross the cell. Original bathymetry dataconsisted of a detailed point survey of depths in feet, hereconverted to meters. The sounding data were contoured andpublished in NOAA nautical charts and denote mean lowwater.

Florida shoreline and bathymetry so that every cellmapped as water in the shoreline map correspondedwith a bathymetry value. Second, we transformed thebathymetry map to a cost surface (Table 1) (Weigle etal. 2001). Third, we evaluated the telemetry data toverify that the locations were positioned in water.Fourth, we delineated movement paths connecting con-secutive telemetry locations and then calculated man-atee residence times (minutes/cell) in cells crossed byeach path. Movement paths were delineated using a‘‘least cost path’’ approach in which bathymetric pref-erences were represented in GIS maps as costs. Fifth,we transformed the raster maps of the movement pathsfor each tagged manatee into maps of number of visitsper cell and mean time per visit per cell. Maps of visitsper cell and mean time per visit were combined andreclassed into manatee places and movement corridorsfor each animal. All maps and analyses were conductedusing raster maps having 25 3 25 m cell size. This cellsize was necessary to help maintain shoreline featuresfor running the movement path algorithm. Steps 1–4are described in the Appendix.

Derivation of manatee places and movement corri-dors.—Maps representing places and movement cor-ridors for each animal were derived from movementpaths that satisfied three criteria that were based onremoving excessive outliers and excluding error intro-duced by a malfunctioning satellite tag. First, the timeinterval between two sequential telemetry points didnot exceed 3 d. Second, the start point of a movementpath was not associated with a malfunction in the tag.Third, the estimated travel rate between two sequentialtelemetry points did not exceed 3000 m/h. Line cov-erages of movement paths for each animal were trans-formed to raster maps illustrating the number of visitsby a manatee in each pixel, called VISITS, and theaccumulated time per visit, called TIME. Summary sta-tistics were then calculated for each map, including thenumber of unique cells visited, total cells visited, meanand standard deviation of visits per cell, and the mean

and standard deviation of time per visit. Summary sta-tistics were calculated by including only those cellsvisited by the manatee.

Analyses included comparing movement character-istics among males (M), females without calves (FN),and females with calves (FW). A few females weremonitored during periods with and without a calf. Onlythe paths for the sex class, FN or FW, that composedthe majority of their tagging period were used. Move-ment was categorized as characteristics of individualanimals (range), movement paths (movement rate,RATE; mean frequency of visits per cell, MVISIT; andmean time per visit per cell, MTIME), and attributesof movement-behavior patches (places and movementcorridors). RATE was calculated per animal as the totallength of travel paths divided by the total hours tracked.MVISIT is the mean number of times cells were visitedby an individual manatee. Only those cells with at leastone visit were included in this mean. MTIME was cal-culated by first dividing the duration of a travel pathby the number of cells crossed. This provided the cellvalue TIME, time per visit. After TIME was calculatedfor all cells and summed for all travel paths, then theTIME map was divided by the VISIT map to arrive atthe mean time per visit estimate for each cell. MTIMEcan be interpreted as the mean manatee residence timefor each cell. First, we tested the hypothesis that cap-tivity and release of manatees did not bias the responsevariables. If the results suggested no differences, re-habilitated animals were pooled with the wild-caughtmanatees. If the data satisfied the criteria for parametriccomparisons, we conducted a parametric analysis ofvariance to compare the mean number of visits per celland the mean time per visit for the three sex classes.Otherwise, we conducted a Dunn’s nonparametric mul-ticomparison test (Dunn 1964).

Manatee places and corridors were classified basedon mean residence times and the number of visits andwere represented as four maps for each animal. Mapsof only those cells with one or two visits and a meanof at least 5 min/visit were called secondary places.Maps of cells with more than two visits and a mean ofat least 5 min/visit were called primary places. Simi-larly, maps with cells of one or two visits and a meanof less than 5 min/visit were called secondary corridors,whereas maps with cells of greater than two visits wereconsidered primary corridors. The resulting four mapswere then processed with a 3 3 3 cell spatial filter, inwhich the value of the focal cell, i.e., the center cell,in the output map was set to the sum of the cell valuespositioned within the 3 3 3 filter. Cell values in theoutput map of five or greater, i.e., five or more cells inthe neighborhood of and including the focal cell wererepresentative of the theme of interest, were reclassifiedto a value of one, and the other cells were set to zero.The final map revealed a set of patches showing areasof comparatively greater activity for each place andcorridor class for each animal.


Landscape indices for the movement-behavior mapswere calculated using the spatial statistics softwareFragstats (McGarigal and Marks 1995). Patches weredefined using the eight-neighbor rule. Fragstats outputfor each animal’s maps of places and movement cor-ridors were imported into SAS (SAS Institute 1990)where area (AREA), largest patch index (LPI), numberof patches (NP), and mean patch size (MPS) of thedifferent movement-behavior patch classes were com-pared among the three manatee sex classes.

Composite maps for each sex class were also com-pared. For each sex class, maps of the primary placeswere merged across animals, as were the maps of pri-mary movement corridors. Using the map of primaryplaces as an example, the result of a merge was a binarymap in which a cell value of ‘‘1’’ represented at leastone incidence of a primary place among the maps beingmerged. The same process was used to merge the pri-mary corridor maps.

RESULTS

Characteristics of individual animals

Forty-one manatees, 19 males and 22 females, hadmovement paths mapped in the study area and weremonitored for at least 30 consecutive days. Five of thefemales moved with their dependent calves for the en-tire time that they were tagged, and 12 did not havecalves during the monitoring period. The tagging pe-riod of six females was split between times with andwithout calves. Six of the animals were rehabilitatedanimals. Details of characteristics of tagged animalsused in this study and their movement paths are pre-sented in Weigle et al. (2001).

Predominant movement patterns included aggregat-ing at warmwater discharge sites during cold weather,moving along the shoreline, and sometimes crossingTampa Bay. Almost all coastal areas of Tampa Baywere visited by manatees, although some areas werevisited much more frequently than others. Movementpatterns varied considerably among individual animals;some remained in Tampa Bay during the tagged periodwhile others moved well outside the study area (Weigleet al. 2001). Fig. 2 illustrates movement paths for asingle animal, one that ranged over a small area.

Characteristics of movement paths

Although MVISIT, MTIME, and RATE for M, FN,and FW usually satisfied the assumptions necessaryfor conducting parametric statistics, deviations weresufficient to warrant running nonparametric compari-sons between rehabilitated and wild animals andamongst sex classes (Table 2). Movement-path char-acteristics MVISIT (P . x2 5 0.4386), MTIME (P .x2 5 0.8827), and RATE (P . x2 5 1.000) were notsignificantly different between captive-released andwild-caught manatees based on results of a Kruskal-Wallis test. Therefore, rehabilitated and wild animalswere pooled in subsequent analyses.

Mean movement rates were 417, 323, and 307 m/hfor M, FN, and FW, respectively (Table 2). Movementrates were significantly different (PROB . x2 50.0118), with M significantly greater than FN and FW.MTIME, which is related to the inverse of movementrate and reflects cell residence time, was not stronglysignificantly different at the 0.05 level (MTIME, P .x2 5 0.0550). The mean number of visits per cell wasnot significantly different among sex classes (MVISIT,P . x2 5 0.1824).

MTIME was not related to the number of days ananimal was tagged (Table 3). MVISIT was positivelycorrelated with tagging duration, resulting, in part,from seasonal variation in sites visited. This result issupported by the decreasing ratio of cells visited onceto the number of cells visited at least once with in-creasing total movement-path length (Fig. 3).

Characteristics of manatee placesand movement corridors

Fig. 4 illustrates the places and movement corridorsof all tagged manatees combined. Places were concen-trated in West Tampa, the TECO power plant dischargecanal and south to the mouth of the Little ManateeRiver, Terra Ceia Bay, and the Gandy Bridge area, in-cluding the FPC Weedon Island power plant. Move-ment corridors were found primarily along the coast-line, although some extended across Tampa Bay. Themost prominent corridor ran from the TECO powerplant past the southern edge of the Tampa peninsulaand across Old Tampa Bay to the Bartow power plantdischarge.

Movement corridor and place characteristics werenot significantly different between captive-released andwild animals based on results of a Kruskal-Wallis test(corridor characteristics, AREA, PROB . P 5 0.8537;largest patch index [LPI] PROB . P 5 0.2378; numberof patches [NP] PROB . P 5 0.8537; mean patch size[MPS] PROB . P 5 0.8973; place characteristics,AREA PROB . P 5 0.8248; LPI PROB . P 5 0.3760;NP PROB . P 5 0.9853; MPS PROB . P 5 0.7122).Therefore, rehabilitated and wild animals were pooledin subsequent patch comparisons among sex classes.

Characteristics of movement corridors were similaramong sex classes (Table 4). The primary differencewas that females without calves had significantly fewercorridor patches and total corridor patch area. Placesdid not differ among sex classes for total patch area.Females without calves had significantly fewer butlarger places as reflected by a low NP and high MPSand LPI. Alternatively, males had a much lower MPSand higher NP.

DISCUSSION

The model presented here transformed telemetrypoints into a movement history of manatees from whichwe extracted frequently used places and movement cor-ridors. The results of this analysis are analogous to


FIG. 2. Movement paths derived from the model for two manatees. Key: T, location of Tampa Electric Company BigBend power plant warmwater discharge; F, location of Florida Power Corporation Bartow power plant warmwater discharge.TTB048 and TTB034 are identifiers for the animals. TTB048 means ‘‘telemetry, tagged in Tampa Bay, animal number 48.’’

TABLE 2. Mean visits per cell (MVISIT), mean time per visit (MTIME), and mean movementrate (RATE) for males (M), females without calves (FN), and females with calves (FW),with results of Shapiro-Wilk test for normality.

Sex N Mean SD Range W: Normal P

Mean visits/cell, P . x2 5 0.1824M 19 2.22 0.733 1.42–4.64 0.778060 0.0003FN 13 2.17 0.611 1.29–3.26 0.961117 0.7221FW 9 2.47 0.493 1.86–3.47 0.910703 0.3157

Mean time/visit (min), P . x2 5 0.0550M 19 4.75 0.955 3.01–6.46 0.973779 0.8326FN 13 5.79 1.887 3.50–10.58 0.897652 0.1211FW 9 5.77 1.080 3.92–7.38 0.973479 0.9181

Mean travel rate (m/h), P . x2 5 0.0118M 19 416.9 147.8 275–934 0.735086 0.001FN 13 323.5 90.7 170–487 0.95989 0.7272FW 9 307.1 65.2 221–426 0.930107 0.4330

Notes: The hypothesis of normality is not rejected if the value of the test statistic, W: Normal,is large. P is the probability of a smaller W: Normal.


TABLE 3. Results of a linear regression testing the hypothesis that the slopes of the lines ofmean visits/cell and mean time/visit on total days tagged were not significantly differentfrom zero.

Sex Variable df Parameter estimate SE

t for H0

Parameter 5 0 P

M MVISIT 1 0.109909 0.01991192 5.520 0.0001FN MVISIT 1 0.131697 0.06548507 2.011 0.0695FW MVISIT 1 0.107786 0.05659841 1.904 0.0986M MTIME 1 0.020054 0.04303780 0.466 0.6472FN MTIME 1 20.124823 0.23355084 20.534 0.6037FW MTIME 1 0.166624 0.13830000 1.205 0.2674

Note: Total days tagged was divided by 50. Student’s t is reported for the null hypothesisthat the parameter population mean is 0, along with the probability of a greater absolute valuefor this t value.

FIG. 3. Linear relationship between the ratio of cells vis-ited once to the number of cells visited at least once and thenumber of days the animal was monitored.

those of home-range mapping via telemetry data (Klen-ner 1987, Joshi et al. 1995), but rather than using theclusters of points to delineate a home range, we canuse clusters of manatee locations to help identify aparticular environmental feature such as a warm- orfreshwater discharge, where manatees spend extendedperiods of time or aggregate. Telemetry data collectedon other non-territorial species show similar charac-teristics with respect to aggregation areas (Tiebout andCarey 1987, Franz 1995).

It is important to note that the places and corridorsdisplayed in Fig. 4 were delineated to illustrate outputsof the model. The data were filtered, and the criteriaused to separate places from corridors were not chosenfor the purpose of assisting in any specific managementdecision. Nor are the maps meant to be the definitiverepresentation of manatee places and corridors in Tam-pa Bay. Depending on the management problem andits spatial and temporal scale, the criteria used to iden-tify places and corridors and those used to display themon a map need to be evaluated for their applicabilityfor representing manatee movement appropriate for theproblem.

Characteristics of movement paths

We were interested in comparing movement ratesamong the sex classes. However, the model indicatedthat there was not a clear distinction in movement-pathcharacteristics among the three sex classes. For ex-

ample, although cell residence times and the MVISITdid not differ significantly among classes, movementrate, which is related to the inverse of residence times,was significantly faster for males. This apparent dis-crepancy may simply be due to the subtle differencesin how travel rates and residence times were calculated.However, males of many species typically travel moreextensively than the conspecific females (Goodwin andMarion 1979, Franz 1995). For manatees, the matingseason is long, and males travel extensively in searchof females (Hartman 1979). In this regard, Deutsch etal. (2003) reported greater daily travel for adult malesthan for females during the warm season along theAtlantic coast. This greater travel should translate intohigher movement rates for male manatees.

We explored site fidelity (Deutsch et al. 2003) bycomparing the number of days an animal was taggedto cell mean residence times. Mean residence times didnot change with an increase in tagging duration beyond30 d, suggesting that the general movement behaviorwas revealed early in the monitoring period, that is,commuting between places where extended time isspent. This result, when considered in conjunction withthe proportion of cells visited once to total cells visiteddecreased as the duration of tracking increased, sug-gests a degree of site fidelity in manatees.

The observed variability in movement patternsamong individuals, the locations of their places andcorridors, and the discovery of areas new to individualmanatees is due to choice-motivated behavior, aboutwhich little is known. For manatees, choice is moti-vated, in part, by avoidance of extreme air and watertemperatures, the desire to drink fresh water, mate seek-ing, hunger, raising a calf, and possibly some humanactivities (Hartman 1979, Buckingham et al. 1999).How manatees receive, store, retrieve, and process en-vironmental information is unknown. Some data sug-gest that knowledge is transferred from mother to off-spring (Deutsch et al. 2003), but new knowledge couldalso be acquired through interactions with other man-atees (O’Shea and Kochman 1990) or by chance.

Choice is based, in part, on the level of knowledge,or memory, that the animal has about its environment


FIG. 4. Places and corridors generated by the model and combined for 41 animals. Places and corridors were identifiedfor individual animals first and then combined. Frequently used places were defined as cells with at least two visits by anindividual manatee and a mean visitation length of 5 min or more per visit. Frequently used corridors were defined as cellswith at least two visits by an individual manatee and a mean visitation length of less than 5 min/visit. For illustration purposes,places and corridors were expanded by five cells in all directions. Patches of fewer than 5000 cells, or 3.125 km2, weredeleted. Only the largest places and corridors were shown. Smaller places and corridors were scattered along the coast ofTampa Bay. Key: T, location of Tampa Electric Company Big Bend power plant warmwater discharge; G, West Gandy Bridgearea including the Florida Power Corporation Bartow power plant warmwater discharge; W, West Tampa; M, Little ManateeRiver; C, Terra Ceia Bay.

TABLE 4. Mean rank scores and results of Kruskal-Wallis tests (x2 approximation) comparingplace and corridor patch characteristics among males (M), females without calves (FN), andfemales with calves (FW).

Variable

Mean rank scores

M FN FW x2 P

PlacesPatch area 21.3 17.6 25.3 2.2247 0.3288Largest patch index 15.0 30.9 19.3 13.861 0.0010Number of patches 25.1 13.2 23.5 8.1232 0.0172Mean patch size 15.1 28.3 23.0 9.7746 0.0075

Movement corridorsPatch area 24.9 14.6 21.9 5.8055 0.0549Largest patch index 19.2 24.6 19.5 1.7370 0.4196Number of patches 25.1 14.0 22.6 6.8438 0.0327Mean patch size 22.3 19.3 20.7 0.48461 0.7848


(Balda et al. 1997, Janson 2000). For example, severalspecies of primates appear to recall locations of re-sources and show preferences for more rewarding sites(Garber 1989, Menzel 1991, Janson 1998). In corvidspecies, spatial memory was concluded to improve themore dependent the individuals were on the resourcesfor survival (Balda and Kamil 1989). Although the spe-cies cited in these examples respond to their own setof environmental cues when exercising their spatialcognitive ability, the basic motivations driving theirchoice behavior, hunger, thirst, safety, reproduction,etc., are the same as for manatees. As such, some levelof environmental knowledge is likely possessed bymanatees because of their high level of fidelity to im-portant resource sites such as high-quality warmwaterdischarge sites during cold weather (Deutsch et al.2003). If no prior environmental knowledge is as-sumed, then manatees would probably follow environ-mental cues until they reached non-navigable areas,undesirable environmental features, or preferred sites.

Behavior of geographically naı̈ve animals suggeststhat manatees do follow environmental cues as theygain the appropriate knowledge base to establish hab-itat usage patterns commonly observed in free-rangingmanatees. One adult manatee, Sweet Pea, was capturedin the Houston, Texas, USA, ship channel and releasedin the Homosassa River, in northwest Florida (Weigleet al. 2001). Mitochondrial DNA suggested that thisanimal lived with the Florida population and had trav-eled west to Texas. Because the animal also had scarpatterns more common to Florida manatees and lesscommon to their counterparts in Mexico it was believedless likely that this manatee had traveled north up fromMexico. Regardless, this animal’s movement behaviorwas atypical. After release, Sweet Pea traveled widely,exploring the coastline up to Apalachicola Bay duringthe summer, then turned south arriving in the FloridaKeys by winter, and then north along the east coast ofFlorida. She returned to Homosassa River while on herway south toward the Florida Keys, displaying possiblerecently learned behavior. However, she did not settleinto a predictable movement pattern by the time thesatellite tag stopped functioning. Released captive-bornanimals are slightly different in that they are not onlygeographically naı̈ve, but also have not developed theknowledge and skills to survive in the wild. Althoughvariation among naı̈ve manatees is common, generallytypical observations of tracked captive-born animalsreveal a period of several weeks to months duringwhich the animal doesn’t move very far from the re-lease site. As their world and knowledge base increaseswith time, they do start exploring their surroundings.If they join local individuals or groups of wild man-atees, the captive-born animals seem to adapt their be-havior to that of the wild animals and their movementbehavior becomes more predictable over time.

We suspect that manatees maintain a mental map ofsome of the less common resource areas, such as fresh-

water and warmwater discharge sites and possibly sev-eral of the prime foraging areas. The existence of fre-quently used travel corridors and relatively few keyaggregation sites, as revealed by the model, supportthis contention (Janson 2000). The extent and com-plexity of their mental map, however, would be difficultto quantify. The gentle slope of the line in Fig. 3 sug-gests that the number of places in the manatee’s mentalmap is extensive and that these animals continuallylearn of and travel to areas not previously visited. Othermammals, such as chimpanzees, can recall up to 16locations (Menzel 1973), and humans perform simi-larly when memorizing a map (Taylor and Tversky1992).

Another characteristic of manatee movement is thevast distances these animals trekked to reach specificsites along the west coast of Florida. Ultimately, move-ment is a cost–benefit balance between energy ex-pended and the reward upon arrival (Steudel 2000).Given that the mass of manatees is supported by water,moving their large size may be relatively economical(Tucker 1975), so long-distance movements may beaccomplished fairly easily. Furthermore, manateestravel along a plane, move slowly most of the time,and have a ballast system that may require little energy.Such efficient movement would allow manatees flexi-bility in their travel patterns, because they might beless constrained by energy expenditures. This efficien-cy and the fact that foraging areas are abundant andwidely distributed may have contributed to the widerange in travel patterns and spatial extents observedamong individual manatees (Weigle et al. 2001). Con-straining this flexibility is the potential heat loss duringtheir winter travels, because manatees have a relativelyhigh thermal conductance (Irvine 1983).

Behavioral characteristics such as maintaining amental map of resource locations and being able totravel long distances have implications for cumulativemovement patterns of manatees. First, locations of lesscommon resources such as freshwater and warmwaterdischarge sites are known by all manatees. Since man-atees can move over long distances (access is limitedby physical obstruction only) these sites can be reachedby manatees from anywhere along the east coast ofFlorida. Furthermore, many of these areas are sur-rounded by feeding areas that are visited frequentlyduring aggregation around such sites, resulting in apattern of central-place foraging. Second, movementand aggregation behaviors are strongly influenced byseasonal temperatures so that foraging behavior, andtherefore travel patterns, change over the course of ayear. As a result, Tampa Bay, primarily near the shore-line, is blanketed by manatee movement paths.

Characteristics of places and corridors

The model did an excellent job of mapping places.All places identified by the model were confirmed byfield biologists as areas where manatees aggregate, for-


age, or simply spend extended periods of time. Manyof the places were warmwater or freshwater dischargesites, most of which were man-made. Reasons for dif-ferences in characteristics of place patches, especiallybetween females without calves and the other two sexclasses, are unknown.

Any area that provides requisites with some minimallevel of reliability can qualify as a potential place formanatees. Most notable are freshwater discharge sitesand warmwater refuges, because of their predictablehigh manatee abundance. Seagrass beds and other feed-ing areas are sufficiently abundant and widely distrib-uted and thus do not attract dense concentrations ofmanatees to any single patch. Depending on weatherconditions, those foraging patches near freshwater andwarmwater discharges may be used more due to theiraccessibility from the discharge sites than those furtheraway (Packard 1984).

Probably more than any other factor, human land-use choices have influenced the current distribution offrequently used places and, consequently, manatee dis-tribution and abundance in Florida. Industrial discharg-es from paper mills and power plants are thought tohave expanded the winter range northwards by provid-ing refuge from cold temperatures (Hartman 1974,Campbell and Irvine 1981). Canal construction, whichoccasionally opens pores in the Florida aquifer, canresult in seeps of fresh water as well as provide moretemperate deep-water refuges (Shane 1984). Seawallsthat concentrate run-off via discharge pipes and sewerlines also contribute to the distribution of freshwaterdischarges, in addition to rivers, streams, and naturalseeps.

Locations of movement corridors were less certain.Corridors, unlike the places, tended not to be associatedwith clusters of telemetry locations unless they over-lapped places such as seagrass patches. Furthermore,because of the ephemeral nature of manatees in thecorridors, field biologists could not confidently identifythe locations of frequently used corridors. One corridordelineated by the model and considered a likely routeis the one connecting the two primary warmwater dis-charge sites that are located on opposite sides of TampaBay (the Florida Power Corporation’s Bartow plant andthe Tampa Electric Company’s plant) (Fig. 4).

Manatee movement corridors are not considered tobe the same as conservation corridors. Conservationcorridors are typically related to physical features ofthe environment (Hurlbert 1969, Healy 1975, Hersheyand Forester 1980, Beier and Noss 1998) and are de-lineated solely by their structure, such as greenbelts(Diamond 1975, Wilson and Willis 1975, Noss 1987).Manatee corridors are defined as being a route selectedby many manatees to travel between places. Their ex-istence suggests a relatively standard cognitive pro-tocol among members of the population. Required en-vironmental attributes for a corridor might be minimal,possibly as simple as sufficient water depth to permit

passage. Manatees may show preferences for certainwater depths and may avoid areas of heavy boat use,but they will cross less preferred areas to reach theirdestination. In this scenario, the primary influence be-hind the location of corridors would be the spatial ar-rangement of places. If a manatee’s spatial cognitionis high, i.e., if many places are maintained in an internalmap, it could reduce travel distances between places.

Some forms of minimizing movement between plac-es are reported in the literature. For example, organismswill move more frequently between nearby patchesthan between distant patches (Reed et al. 1998). Wheth-er or not animals are aware of the most direct routebetween patches is less certain. For example, stripednewts and eastern narrow-mouthed toads did not usenarrow migratory corridors. Rather, their movementwas directed between terrestrial habitats and breedingponds (Dodd and Cade 1997). For manatees, the pro-cess behind choosing routes between places is un-known. The results of the model do suggest that cor-ridors are part of the manatee’s movement behavior.

The mapping of places and corridors clarifies areasof importance for manatees beyond using telemetrypoint data alone. Telemetry data can help delineatesome of the obvious places but is less effective at map-ping movement corridors. Furthermore, examiningpoints for presence/absence lends itself to a broad spec-trum of interpretations among reviewers as to wherethe important manatee areas might be. The results ofour model synthesize elements of the data that are dif-ficult to visualize, such as residence times and mappingplaces and corridors in a consistent, relatively objectivemanner. What must be decided when managers are ap-plying this model is the criteria for defining a place orcorridor, the range of values designating a high-useplace or corridor as opposed to low use, and the spatialextent of a place and width of a corridor.

ACKNOWLEDGMENTS

We thank Chip Deutsch, Elsa Haubold, and three anony-mous reviewers for their comments on the manuscript. Wealso thank Alexander Smith for helping with the figures andRobert Bonde for his input on spatial cognition of geograph-ically naı̈ve animals. This work was supported by the FloridaFish and Wildlife Conservation Commission’s Save the Man-atee Trust Fund. This research was conducted under the au-thority of scientific research permit no. MA773494.

LITERATURE CITED

Ackerman, B. B. 1995. Aerial surveys of manatees: a sum-mary and progress report. Pages 13–33 in T. J. O’Shea, B.B. Ackerman, and H. F. Percival, editors. Population bi-ology of the Florida manatee. National Biological ServiceInformation and Technology Report number 1. UnitedStates Department of Interior, Fish and Wildlife Service,Southeast Region, Atlanta, Georgia, USA.

Ackerman, B. B., S. D. Wright, R. K. Bonde, D. K. Odell,and D. J. Banowetz. 1995. Trends and patterns in mortalityof manatees in Florida, 1974–1992. Pages 223–258 in T.J. O’Shea, B. B. Ackerman, and H. F. Percival, editors.Population biology of the Florida manatee. National Bio-logical Service Information and Technology Report number


1. United States Department of Interior, Fish and WildlifeService, Southeast Region, Atlanta, Georgia, USA.

Argos. 1996. User’s manual. Service Argos, Landover,Maryland, USA.

Balda, R. P., and A. C. Kamil. 1989. A comparative studyof cache recovery by three corvid species. Animal Behavior38:486–495.

Balda, R. P., A. C. Kamil, P. A. Bednekoff, and A. G. Hile.1997. Species differences in spatial memory performanceon a three-dimensional task. Ethology 103:47–55.

Beier, P., and R. F. Noss. 1998. Do habitat corridors provideconnectivity? Conservation Biology 12:1241–1252.

Bengtson, J. L. 1981. Ecology of manatees (Trichechus man-atus) in the St. Johns River, Florida. Dissertation. Univer-sity of Minnesota, Minneapolis, Minnesota, USA.

Buckingham, C. A., L. W. Lefebvre, J. M. Schaefer, and H.I. Kochman. 1999. Manatee response to boating activityin a thermal refuge. Wildlife Society Bulletin 27:514–522.

Bulova, S. J. 1994. Patterns of burrow use by desert tortoises:gender differences and seasonal trends. HerpetologicalMonographs 8:133–143.

Campbell, H. W., and A. B. Irvine. 1981. Manatee mortalityduring the unusually cold winter of 1976–1977. Pages 86–91 in R. L. Brownell, Jr., and K. Ralls, editors. The WestIndian manatee in Florida. Proceedings of a workshop heldin Orlando, Florida, 27–29 March 1978. Florida Depart-ment of Natural Resources, Tallahassee, Florida, USA.

Campbell, H. W., and J. A. Powell. 1976. Endangered spe-cies: the manatee. Florida Naturalist 49:15–20.

Carroll, J. P., R. D. Crawford, and J. W. Schulz. 1995. Graypartridge winter home range and use of habitat in NorthDakota. Journal of Wildlife Management 59:98–103.

Deutsch, C. J., R. K. Bonde, and J. P. Reid. 1998. Radio-tracking manatees from land and space: tag design, imple-mentation, and lessons learned from long-term study. Ma-rine Technological Society Journal 32:18–29.

Deutsch, C. J., J. P. Reid, R. K. Bonde, D. E. Easton, H. I.Kochman, and T. J. O’Shea. 2003. Seasonal movements,migratory behavior, and site fidelity of West Indian man-atees along the Atlantic coast of the United States. WildlifeMonographs 151:1–77.

Diamond, J. M. 1975. The island dilemma: lessons of modernbiogeographic studies for the design of nature reserves.Biological Conservation 7:129–146.

Dodd, C. K., Jr., and B. S. Cade. 1997. Movement patternsand the conservation of amphibians breeding in small, tem-porary wetlands. Conservation Biology 12:331–339.

Dunn, O. J. 1964. Multiple comparisons using rank sums.Technometrics 6:241–252.

Enderson, J. H., J. Larrabee, Z. Jones, C. Peper, and C. Lep-isto. 1995. Behavior of peregrines in winter in south Texas.Journal of Raptor Research 29:93–98.

Flamm, R. O., L. I. Ward, and B. L. Weigle. 2001. Applyinga variable-shape spatial filter to map relative abundance ofmanatees (Trichechus manatus latirostris). LandscapeEcology 16:279–288.

Folkow, L. P., P. Martensson, and A. S. Blix. 1996. Annualdistribution of hooded seals (Cystophora cristata) in theGreenland and Norwegian Seas. Polar Biology 16:179–189.

Franz, R. 1995. Habitat use, movements and home range intwo species of ratsnakes (genus Elaphe) in north Floridasandhill. Florida Game and Freshwater Fish CommissionNongame Wildlife Program Project Report GFC-86-014,Tallahassee, Florida, USA.

Garber, P. A. 1989. Role of spatial memory in primate for-aging patterns: Saguinus mystax and Saguinus fuscicollis.American Journal of Primatology 19:203–216.

Goodwin, T. M., and W. R. Marion. 1979. Seasonal activityranges and habitat preferences of adult alligators in a north-central Florida lake. Journal of Herpetology 13:157–164.

Hartman, D. S. 1974. Distribution, status, and conservationof the manatee in the United States. U.S. Fish and WildlifeService, National Fish and Wildlife Laboratory. ContractReport Number 14-16-0008-748, Publication NumberPB81-140725. National Technical Information Service,Springfield, Virginia, USA.

Hartman, D. S. 1979. Ecology and behavior of the manatee(Trichechus manatus) in Florida. American Society ofMammalogists 5:1–153.

Healy, W. R. 1975. Breeding and postlarval migrations ofthe red-spotted newt, Notophthalmus viridescens, in Mas-sachusetts. Ecology 56:673–680.

Hershey, J. L., and D. C. Forester. 1980. Sensory orientationin Notophthalmus viridescens (Amphibia: Salamandridae).Canadian Journal of Zoology 58:266–276.

Hurlbert, S. H. 1969. The breeding migrations and interhab-itat wandering of the vermillion-spotted newt Notophthal-mus viridescens (Rafinesque). Ecological Monographs 39:465–488.

Irvine, A. B. 1983. Manatee metabolism and its influence ondistribution in Florida. Biological Conservation 25:315–334.

Janson, C. H. 1998. Experimental evidence for spatial mem-ory in foraging wild capuchin monkeys Cebus apella. An-imal Behavior 55:1129–1143.

Janson, C. H. 2000. Spatial movement strategies: theory,evidence, and challenges. Pages 165–203 in S. Boinski andP. A. Garber, editors. On the move: how and why animalstravel in groups. The University of Chicago Press, Chicago,Illinois, USA.

Joshi, A. R., D. L. Garshelis, and J. L. D. Smith. 1995. Homeranges of sloth bears in Nepal: implications for conser-vation. Journal of Wildlife Management 59:204–214.

Kieffer, M. C., and B. Kynard. 1996. Spawning of the short-nose sturgeon in the Merrimack River, Massachusetts.Transactions of the American Fisheries Society 125:179–186.

Klenner, W. 1987. Seasonal movements and home range uti-lization patterns of the black bear, Ursus americanus, inWestern Manitoba. Canadian Field Naturalist 101:558–568.

Lefebvre, L. W., J. P. Reid, G. B. Rathbun, and D. P. Domning.2001. Status and biogeography of the West Indian manatee.Pages 425–474 in C. A. Woods and F. E. Sergile, editors.Biogeography of the West Indies: new patterns and per-spectives. CRC Press, Boca Raton, Florida, USA.

McGarigal, K., and B. J. Marks. 1995. Fragstats: spatial pat-tern analysis program for quantifying landscape structure.General Technical Report PNW-GTR-351. U.S. Depart-ment of Agriculture, Forest Service, Pacific Northwest Re-search Station, Portland, Oregon, USA.

Menzel, C. R. 1991. Cognitive aspects of foraging Japanesemonkeys. Animal Behavior 41:397–402.

Menzel, W. E., Jr. 1973. Chimpanzee spatial memory orga-nization. Science 182:943–945.

Moore, J. C. 1951. The range of the Florida manatee. Quar-terly Journal of the Florida Academy of Science 14:1–19.

Moore, J. C. 1956. Observations of manatees in aggregations.American Museum Novitiates 1811:1–24.

Noss, R. F. 1987. Protecting natural areas in fragmented land-scapes. Natural Areas Journal 7:2–13.

O’Shea, T. J. 1988. The past, present, and future of manateesin the southeastern United States: realities, misunderstand-ings, and enigmas. Pages 184–204 in R. R. Odom, K. A.Riddleberger, and J. C. Ozier, editors. Proceedings of theThird Southeastern Non-game and Endangered WildlifeSymposium. Georgia Department of Natural Resources,Game and Fish Division, Social Circle, Georgia, USA.

O’Shea, T. J., C. A. Beck, R. K. Bonde, H. I. Kochman, andD. K. Odell. 1985. An analysis of manatee mortality pat-


terns in Florida, 1976–81. Journal of Wildlife Management49:1–11.

O’Shea, T. J., and H. I. Kochman. 1990. Florida manatees:distribution, geographically referenced data sets, and eco-logical and behavioral aspects of habitat use. Pages 11–22in J. E. Reynolds III and K. D. Haddad, editors. Report ofthe workshop on geographic information systems as an aidto managing habitat for West Indian manatees in Floridaand Georgia. Florida Marine Research Publications 49.Florida Marine Research Institute, St. Petersburg, Florida,USA.

Packard, J. M. 1984. Impact of manatees Trichechus manatuson seagrass communities in eastern Florida. Acta ZoologicaFennica 172:21–22.

Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. S.Curtis. 1989a. Survival analysis in telemetry studies: thestaggered entry design. Journal of Wildlife Management53:7–15.

Pollock, K. H., S. R. Winterstein, and M. J. Conroy. 1989b.Estimation and analysis of survival distributions for radio-tagged animals. Biometrics 45:99–109.

Powell, J. A., and G. B. Rathbun. 1984. Distribution andabundance of manatees along the northern coast of the Gulfof Mexico. Northeast Gulf Science 7:1–28.

Quinn, T. 1995. Using public sighting information to inves-tigate coyote use of urban habitat. Journal of Wildlife Man-agement 59:238–245.

Rathbun, G. B., J. P. Reid, and J. B. Bourassa. 1987. Designand construction of a tethered, floating radio-tag assemblyfor manatees. Document PB 87-161345/AS. National Tech-nical Information Service, Springfield, Virginia, USA.

Rathbun, G. B., J. P. Reid, and G. Carowan. 1990. Distri-bution and movement patterns of manatees (Trichechusmanatus) in northwestern peninsular Florida. Florida Ma-rine Research Intitute Publications 48:1–33.

Reed, J. M., M. D. Silbernagle, K. Evans, A. Engilis, Jr., andL. W. Oring. 1998. Subadult movement patterns of theendangered Hawaiian stilt (Himantopus mexicanus knud-seni). Auk 115:791–797.

Reid, J. P., R. K. Bonde, and T. J. O’Shea. 1995. Reproductionand mortality of radio-tagged and recognizable manateeson the Atlantic Coast of Florida. Pages 171–191 in T. J.O’Shea, B. B. Ackerman, and H. F. Percival, editors. Pop-ulation biology of the Florida manatee. National BiologicalService Information and Technology Report number 1.United States Department of Interior, Fish and WildlifeService, Southeast Region, Atlanta, Georgia, USA.

Reynolds, J. E., III. 2000. Distribution and abundance ofFlorida manatees (Trichechus manatus latirostris) aroundselected power plants following winter cold fronts: 1999–2000. Annual Report to Florida Power and Light Company,Juno Beach, Florida, USA.

Reynolds, J. E., III, and J. R. Wilcox. 1986. Distribution andabundance of the West Indian manatee Trichechus manatusaround selected Florida power plants following winter coldfronts: 1984–1985. Biological Conservation 38:103–113.

Reynolds, J. E., III, and J. R. Wilcox. 1994. Observations ofFlorida manatees (Trichechus manatus latirostris) aroundselected power plants in winter. Marine Mammal Science10:163–177.

Riley, T. Z., J. B. Wooley, Jr., and W. B. Rybarczyk. 1994.Survival of ring-necked pheasants in Iowa. Prairie Natu-ralist 26:143–148.

SAS Institute. 1990. SAS user’s guide. Version 6. SAS In-stitute, Cary, North Carolina, USA.

Shane, S. H. 1984. Manatee use of power plant effluents inBrevard County, Florida. Florida Scientist 47:180–188.

Smethurst, D., and B. Nietschmann. 1998. The distributionof manatees (Trichechus manatus) in the coastal waterwaysof Tortuguero, Costa Rica. Biological Conservation 89:267–274.

Smith, K. N. 1993. Manatee habitat and human-relatedthreats to seagrass in Florida: a review. Report preparedfor the Florida Department of Environmental Protection,Division of Marine Resources, Office of Protected SpeciesManagement, Tallahassee, Florida, USA.

Steudel, K. 2000. The physiology and energetics of move-ment: effects on individuals and groups. Pages 9–23 in S.Boinski and P. A. Garber, editors. On the move: how andwhy animals travel in groups. University of Chicago Press,Chicago, Illinois, USA.

Tampa Bay Estuary Program. 1996. Charting the course: thecomprehensive conservation and management plan forTampa Bay. Tampa Bay Estuary Program, St. Petersburg,Florida, USA.

Taylor, H. A., and B. Tversky. 1992. Descriptions and de-pictions of environments. Memory and Cognition 20:483–496.

Thompson, F. R., III. 1994. Temporal and spatial patterns ofbreeding brown-headed cowbirds in the mid-western Unit-ed States. Auk 111:979–990.

Tiebout, H. M., III, and J. R. Carey. 1987. Dynamic spatialecology of the watersnake, Nerodia sipedon. Copeia 1987:1–18.

Tucker, V. A. 1975. The energetic cost of moving about.American Scientist 63:413–419.

U.S. Fish and Wildlife Service. 2001. Florida manatee re-covery plan. Third revision. U.S. Fish and Wildlife Service,Atlanta, Georgia, USA.

Weigle, B. L., I. E. Wright, M. Ross, and R. O. Flamm. 2001.Movements of radio-tagged manatees in Tampa Bay andalong Florida’s West coast 1991–1996. Florida Marine Re-search Institute Technical Reports TR-7. Florida Fish andWildlife Conservation Commission, St. Petersburg, Flori-da, USA.

Wilson, E. O., and E. O. Willis. 1975. Applied biogeography.Pages 522–534 in M. L. Cody and J. M. Diamond, editors.Ecology and evolution of communities. Belknap, Cam-bridge, Massachusetts, USA.

Wright, I. E., J. E. Reynolds III, B. B. Ackerman, L. I. Ward,B. L. Weigle, and W. A. Szelistowski. 2002. Trends inmanatee (Trichechus manatus latirostris) counts and habitatuse in Tampa Bay, 1987–1994: implications for conser-vation. Marine Mammal Science 18:259–274.

Wright, S. D., B. B. Ackerman, R. K. Bonde, C. A. Beck,and D. J. Banowetz. 1995. Analysis of watercraft-relatedmortality of manatees in Florida 1979–1991. Pages 259–268 in T. J. O’Shea, B. B. Ackerman, and H. F. Percival,editors. Population biology of the Florida manatee. Na-tional Biological Service Information and Technology Re-port number 1. United States Department of Interior, Fishand Wildlife Service, Southeast Region, Atlanta, Georgia,USA.

APPENDIX

Details of the method used to generate the manatee movement paths are available in ESA’s Electronic Data Archive:Ecological Archives A015-039-A1.

estimation of manatee (trichechus manatus latirostris...

Documents