Movement heterogeneity of dugongs, Dugong dugon (Müller...

gy and Ecology 334 (2006) 64–83www.elsevier.com/locate/jembe

Journal of Experimental Marine Biolo

Movement heterogeneity of dugongs, Dugong dugon (Müller),over large spatial scales

James K. Sheppard a,b, Anthony R. Preen a,1, Helene Marsh a,b,⁎, Ivan R. Lawler a,b,Scott D. Whiting a,2, Rhondda E. Jones c

a School of Tropical Environment Studies and Geography, James Cook University, Townsville 4811, Australiab CRC Reef Research Centre, P.O. Box 772, Townsville 4810, Australia

c School of Tropical Biology, James Cook University, Townsville 4811, Australia

Received 23 August 2005; received in revised form 3 January 2006; accepted 9 January 2006

Abstract

Seventy dugongs were fitted with satellite PTTs and/or GPS transmitters in sub-tropical and tropical waters of Queensland andthe Northern Territory, Australia. Twenty-eight of the 70 dugongs were also fitted with time-depth recorders. The dugongs weretracked for periods ranging from 15 to 551 days and exhibited a large range of individualistic movement behaviours; 26 individualswere relatively sedentary (moving <15 km) while 44 made large-scale movements (>15 km) of up to 560 km from their capturesites. Male and female animals, including cows with calves, exhibited large-scale movements (LSM; >15 km). Body length oftravelling dugongs ranged from 1.9 to 3 m. At least some of the movements were return movements to the capture location,suggesting that such movements were ranging rather than dispersal movements. LSMs included macro-scale regional movements(>100 km) and meso-scale inter-patch local movements (15 to <100 km) and were qualitatively different from tidally-drivenmicro-scale commuting movements between and within seagrass beds (<15 km). The mean±S.E. macro-scale movement distanceper individual was 243.8±35.4 km (N=14 individuals that travelled >100 km), with a mean±S.E. travel time of 179.8±29.0 h.The mean±S.E. meso-scale movement distance per individual was 49.7±3.3 km (N=28 individuals that made movements of 15–100 km), with a mean±S.E. travel time of 52.3±7.1 h. LSMs were rapid and apparently directed (mean±S.E. travel speeds forGPS tagged animals; meso-scale movements=1.3±0.11 km/h, min=0.3, max=3.0; macro-scale movements=1.6±0.16 km/h,min=0.8, max=1.3). Tracked dugongs rarely travelled far from the coast (mean±S.E. max distance=12.8±1.3 km). Dive profilesfrom the time-depth recorders suggest that dugongs make repeated deep dives while travelling rather than remaining at the surface,increasing their likelihood of capture in bottom set gill nets. Some animals caught in the high latitude limits of the dugongs' rangeon the Australian east coast in winter apparently undertook long distance movements in response to low water temperatures, similarto migrational movements by Florida manatees. Our findings that dugongs frequently undertake macro-scale movements haveimplications for management at a range of scales, and strengthen the aerial survey and genetic evidence for management andmonitoring at ecological scales that cross jurisdictions.© 2006 Published by Elsevier B.V.

Keywords: Dugong; Herbivore; Large mammal; Large-scale movement; Migration; Satellite telemetry; Spatial ecology

⁎
Corresponding author. Graduate Research School, James Cook University, Townsville 4811, Australia. Tel.: +61 7 4781 5575; fax: +61 7 47816126.
E-mail address: [email protected] (H. Marsh).1 Present address: ‘Scott's Plain’, Rollands Plains, NSW 2441, Australia.2 Present address: Biomarine International PO Box 376U Charles Darwin University, Darwin NT, 0801, Australia.

0022-0981/$ - see front matter © 2006 Published by Elsevier B.V.doi:10.1016/j.jembe.2006.01.011

mailto:[email protected]

http://dx.doi.org/10.1016/j.jembe.2006.01.011

65J.K. Sheppard et al. / Journal of Experimental Marine Biology and Ecology 334 (2006) 64–83

1. Introduction

Many large terrestrial mammals perform some kindof large-scale movement, e.g. the seasonal massmigrations of ungulates such as wildebeest and caribou(Sinclair, 1983; Owen-Smith, 2002). Large-scale move-ments of up to thousands of kilometres are also a criticalfeature of the life histories of many species of marinemammals (e.g., Stewart and DeLong, 1995; Kenney etal., 2001; Deutsch et al., 2003a,b). The reasons for suchmovements vary in different species and populations butmay include spatial variability in the availability ofessential resources such as mates, food and warm water,avoidance of predation and ‘evolutionary holdover’–individuals migrating because their ancestors did–despite changes in the bathymetry of ocean basinsover evolutionary time (Stern, 2002).

Members of the order Sirenia (seacows) are the onlyherbivorous marine mammals. They feed on bottomdwelling, emergent and riparian aquatic plants in thetropics and subtropics (Husar, 1978; Atkinson andSmith, 1983; Bengtson, 1983; Ledder, 1986; Hurst andBeck, 1988; Jones, 1994; Preen, 1995b; Mignucci-Giannoni and Beck, 1998; Marshall et al., 2000; Marshet al., 2002). Deutsch et al. (2003b) review the reasonsfor large-scale movements in Sirenians and postulatethat such movements have evolved in response toseasonal changes in environmental variables such astemperature, water levels, salinity and variability offorage.

West Indian manatees, Trichechus manatus, use bothfreshwater and marine environments (Weigle et al.,2001; de Thoisy et al., 2003; Deutsch et al., 2003b) incontrast to the dugong, Dugong dugon (Müller), whichis a seagrass specialist restricted to marine habitats(Husar, 1978; Preen, 1995b; Marsh et al., 2002). Thedugong's range extends through the coastal and islandwaters of over 40 tropical and sub-tropical countries inthe Indo-West Pacific (Marsh et al., 2002, unpublished).Australian waters contain the world's largest dugongpopulation (Marsh et al., 2002) and are recognised as thedugong's stronghold.

Since the 1970s, there has been a significant researcheffort to quantify the distribution and abundance ofdugongs to facilitate management of the species inAustralia. The results of several time series of aerialsurveys flown over very large-spatial scales (>30,000 km2)since the mid 1980s have been used as the basis formanagement planning (Marsh and Saalfeld, 1988, 1989;Marsh et al., 1990, 1991, 1994, 1999, 2002, 2004;Marsh and Lawler, 2001; Gales et al., 2004). Repeatedsurveys of the same region suggest that dugongs

undertake large-scale movements (e.g. see Preen andMarsh, 1995; Marsh et al., 1997, 2004; Marsh andLawler, 2001; Gales et al., 2004). However, there islimited direct evidence for such movements. Anecdotalhistorical evidence comes from a 1890s account byWelsby (1967); “… in skinning one of the animals … wefound in the flesh the head of an aboriginal woodenharpoon, made in the manner in which the nativesaround Port Bowen and further north make theseweapons … I know for a fact that there are no suchharpoons used (in Moreton Bay)…, as all those used inWide Bay and up along the coast as far as Gladstone arein the hands of white men, and are of course made ofiron.” For a dugong to make a long distance movefrom Bowen to Moreton Bay, it would have to travel∼1000 km along the east coast of Queensland.Anecdotal evidence also suggests that dugongs crossdeep ocean trenches. For example, dugongs werereported at Aldabra Atoll, 425 km from Madagascar in2001, for the first time in many years (Marsh et al.,2002). In June 2002, a confirmed report of a solitarymale dugong was made on Cocos (Keeling) Islands inthe Indian Ocean. The atoll is located 1000 km from thenearest mainland coastal habitat (Java, Indonesia) acrosswater 2–4 km deep (Whiting et al., 2005).

Early satellite tracking of dugongs in the Australiantropics confirmed that at least some individualsundertake large-scale movements (LSM) (Marsh andRathbun, 1990; Preen, 2001b). However, these earlystudies provided limited information on the nature ofsuch movements, because of the limitations of thePlatform Transmitter Terminal (PTT) instrumentation.In this paper, we report the unpublished results of earliersatellite PTT tracking combined with recent PTT/GPStracking of 70 dugongs in sub-tropical and tropicalwaters of Queensland and the Northern Territory,Australia. Twenty-two of the recently tracked dugongswere fitted with both PTT/GPS transmitters and time-depth recorders providing new insights into the nature ofthe large-scale movements of dugongs.

2. Materials and methods

Since 1987, 70 dugongs have been caught and fittedwith VHF and satellite telemetry equipment at sixlocations along the east coast of Queensland and theGulf of Carpentaria, Australia (Table 1), using mod-ifications of the capture method described by Marsh andRathbun (1990). Each dugong was herded into shallowwater using a fast boat and captured ‘rodeo-style’ by ajumper before being secured to the side of the vessel.The animal was supported and closely monitored while

Table 1Capture location, transmitter and incidences of large-scale movements (LSMs) for the 70 dugongs satellite tracked in seven sites in Queensland(QLD) and the Northern Territory (NT), Australia between 1986 and 2004

Capturelocation

Lat./Long. Date ofdeployment

No.dugongstracked

Sex Transmittertype

TDR Daystracked

No.dugongsmade LSM

MaxindividualLSM range (km)

Reference

Moreton Bay(QLD)

−27.33°S;153.41°E

June 1988 6 1M, 5F PTT – 32–70 6 15 to 20 Preen (1993)October1988

4 2M, 2F PTT – 20–88 0 – Preen (1993)

January1989

3 1M, 2F PTT – 59–66 0 – Preen (1993)

Hervey Bay(QLD)

−25.20°S;152.65°E

April 2002 1 M PTT/GPS Mk7 28 1 53.5July 2003 7 5M, 2F PTT/GPS Mk7 33–70 7 44.7 to 284.0November2003

3 3M PTT/GPS Mk7and Mk9

20–36 2 23.4 to 63.8

July 2004 9 7M, 2F PTT/GPS Mk7and Mk9

25–70 7 27.5 to 209.5

ShoalwaterBay (QLD)

−22.37°S;150.22°E

March–September1996

10 3M, 7F a PTT Mk4 14–166 3 191.8 to 404.5 Preen (1999)

June 2002 2 2M PTT/GPS Mk7 44–46 0 –ClevelandBay (QLD)

−19.28°S;146.93°E

October1986

2 2M PTT/VHFPTT

– 63–483 1 140.0 Marsh andRathbun(1990)

March/April1998

3 3M – 111–142 3 57.5 to 153.5 Preen (2001a)

HinchinbrookIs (QLD)

−18.25°S;146.19°E

May/October1997

10 2M, 8F b PTT – 24–550 9 47.0 to 560.0 Preen (2001a)

Starcke River(QLD)

−14.50°S;144.80°E

November1987

4 4M PTT – 32–94 0 – Marsh andRathbun(1990)

Borroloola(NT)

−15.58°S;136.31°E

May 1994 5 3M, 2F PTT Mk4 114–199 4 37.4 to 362.5October1994

1 F PTT Mk4 150 1 40.4

a Three females each had an attendant calf each.b Eight females each had an attendant calf each.

66 J.K. Sheppard et al. / Journal of Experimental Marine Biology and Ecology 334 (2006) 64–83

the transmitter was attached, a DNA sample was taken,and length and sex were measured. Locating dugongs indeep and/or turbid waters often necessitated the use of aspotter plane.

2.1. Telemetry housing

Dugongs spend most of their time feeding on theseafloor in shallow water; 72% of recorded dives from15 of the 70 tracked dugongs were within 3 m of thesurface; see Chilvers et al. (2004). Saltwater blocks thetransmission of radio signals, so it is necessary for theantenna of the transmitter to be at the surface to transmita signal. Accordingly, all transmitters were mounted in afloating housing that was tethered to the dugong via aflexible 3 m long nylon rod. The tether was attached tothe dugong via a padded strap around the tailstock. Toreduce battery consumption, a saltwater switch wasincorporated into each unit to shut down attemptedtransmissions when the antenna was submerged.

Because of the potential for this long tether to becomeentangled or for the transmitter housing to be attackedby a shark, a weak link was built into each strap toensure that the dugong could break free if required.The transmitter was recovered at the end of eachtracking period via a remote release system. A signalfrom a handheld transmitter triggered an acceleratedgalvanic reaction across the weak link, causing thetransmitter to detach from the dugong. The unit wasthen located and recovered via short-range VHFsignal. The transmitters usually became detachedwell before the end of the intended deployment period(typically 3–6 months). Thus, actual deployment timesvaried from 15 to 551 days. The longest deploymentsoccurred when dugongs were recaptured and a newtransmitter was attached.

One dugong was fitted with a VHF transmitter, 47dugongs were fitted with satellite PTT tags (ST-14Telonics, Mesa, AZ, USA) linked to Argos/CLS (http://www.cls.fr) and 22 dugongs were fitted with the PTT/

http://www.cls.fr

http://www.cls.fr


GPS satellite tags (Telonics Inc, Mesa, Ariz, USA) (seeTable 1). Argos is a satellite-based location and datacollection system which collects environmental datafrom autonomous platforms. Argos can detect anyplatform with a suitable transmitter anywhere in theworld via Doppler Effect receivers on board NOAAseries satellites (Service Argos, 1996). Six of the PTTtagged dugongs were also fitted with time-depthrecorders (TDR Mk4 Wildlife Computers) (see Chil-vers et al., 2004). All of the PTT/GPS tagged animalswere fitted with Mk7 and Mk9 TDRs (WildlifeComputers, Woodinville, WA, USA). TDRs were fittedto the satellite tag harness at the base of the dugong'stail. Mk4 TDRs can store up to 128,000 readings witha depth resolution of 40 cm. Mk7 TDRs can store up to2 million readings with a depth resolution of 25 cm.The new Mk9 model is smaller and lighter and canstore up to 16 million readings with the same depthresolution. All TDR models have an accuracy of ±1%.The Mk4 TDRs were programmed to sample dive dataevery 1 day in 10 at 1-s intervals. The samplinginterval for the Mk7 and Mk9 TDRs was daily at 1-and 2-s intervals.

2.2. Accuracy and frequency of fixes

The animal fitted with the VHF transmitter wastracked manually by triangulation from three elevatedlocations around Cleveland Bay with a resolution errorof up to 4 km (Marsh and Rathbun, 1990). The PTTsatellite tracking telemetry equipment transmitted at 60-sintervals and operated on 7.5- to 15-h per day dutycycles. Argos predicts that 68% of PTT location class(LC) 3 fixes fall within 150m of the true location. For LC2 the radius is 350 m, and for LC 1 it is 1000 m (ServiceArgos, 1996). Independent field tests suggest these radiiare 361 m, 903 m and 1188 m for LQ 3, 2 and 1,respectively, and that latitudinal error is less thanlongitudinal error (Keating et al., 1991; Vincent et al.,2002). Only PTT LC 1 fixes or greater were used in theanalysis. Outliers were filtered by using a similarapproach to McConnell et al. (1992), which removedfixes if they required improbably high rates of travel toachieve (>3 m/s) or if they were >1000 m inland. ThePTT tags provided location fixes an average of only 2.7(N=50, S.E.=0.2, max=6) times in a 24-h period(Marsh and Rathbun, 1990). In contrast, the PTT/GPStechnology provided regular (max = 72 per day,mean=22.4, S.E.=0.6) and extremely precise (<10 m)spatiotemporal location data. Accordingly, we calculatedminimum possible travelling speeds from the PTT/GPSfixes only. Although the PTT/GPS equipped units

provided both PTT and GPS location fixes, only GPSdata that were downloaded directly from the recoveredunits were used in the analyses.

2.3. Data processing and analysis

We established the criteria which defined a large-scale movement (LSM) based on three discontinuities inthe pattern of dugong movements: (1) the quantitativedifference between the location fix sampling frequencyfor local and long distance movements; (2) themovement distance relative to the diameter of themean dugong kernel home range size, and (3) thetransition of dugong movement structure between smalland large domains of spatial scale using correlatedrandom walks (CRW) and fractal analysis.

Dugongs often clearly made use of multiple, distinct,core habitats separated by areas apparently only used fortransiting between them. In these cases, calculating ahome range using the complete location data set over-inflated the home range boundary, tending to pull theedge of the kernel home range in each core habitattowards the direction of transit in a way not reflective ofthe true use of each area. In order to objectively reducethis distortion of home ranges within core areas, weadopted the following approach: Dugong activitycentres were initially identified using the probabilitycontours of harmonic mean grids (Dixon and Chapman,1980). The harmonic mean was calculated from eachanimal's location fixes using the Animal MovementAnalyst Extension (AMAE) (Hooge and Eichenlaub,1997) for the ArcView v3.3 GIS program (Environ-mental Systems Research Institute, Inc., Redlands, CA).Location fixes within grid cells containing the largestharmonic mean values (proportion: >0.035 GPS fixes,>0.15 PTT fixes) were classified as outliers andexcluded. Fixed kernel home ranges for each dugongwere calculated from the remaining fixes within eachactivity centre using the AMAE (mean±S.E. fixes=382±41.5). All kernels were calculated using a least-squarescross-validation smoothing function (LSCV)—an itera-tive process that estimates the least-biased smoothingfactor (Worton, 1989; Seaman and Powell, 1996;Seaman et al., 1999; Gitzen and Millspaugh, 2003).This approach enabled us to create 95% and 50%utilization distributions (probability contours) with a10 m2 cell size for the areas most frequently used andfor the core habitat areas for each dugong during itstracking period. The generally circular 95% kernel homeranges that were generated were measured across theirmaximum width to provide a measurement of homerange size.


We tested whether the independent individual pathsof dugongs that undertook LSMs differed significantlyfrom those described by the Correlated Random Walk(CRW) model of Skellam (1973), as modified byKareiva and Shigesada (1983). We used the model as anull hypothesis as suggested by Turchin (1996). In themodel, animal movement is random over discrete timeintervals, but has a net directional bias. The equation usesstep lengths, turning angles, and total number of steps tocalculate an expected mean square displacement dis-tance (Rn

2). In this study we used the extension of theequation outlined by Nams (2004) to estimate the overalldeviation between the observed trajectory and expectedRn2 from a bootstrapped simulation of 10,000 iterations,

with 95% confidence intervals determined by thepercentile method (Turchin, 1998). An animal with anobserved trajectory below the expected Rn

2 possessed alower movement displacement than the model (over-predicted) and was deemed sedentary. An animal with anobserved trajectory that fitted the model was classified asa correlated random walker. An animal with an observedtrajectory above the expected Rn

2 possessed a greatermovement displacement than was predicted by themodel and was considered to have made an LSM.

The hierarchical and patchy nature of an animal'shabitat causes movement path structure to vary withspatial scale (Nams, 2004, 2005). The CRW model maybe too simplistic to describe accurately the movementbehaviour of a species that uses complex movementrules across different scales (Morales and Ellner, 2002;Doerr and Doerr, 2004; Nams, 2005). Therefore, asappropriate for our data, we also used fractal analysis tomeasure the tortuosity (Milne, 1991) of dugong paths inorder to detect transitions of movement structure acrossdomains of scale (Wiens, 1989), and to identify theboundaries of dugong spatial habitat use. Fractalanalysis is a geostatistical technique that measures thetortuosity of lines (Mandelbrot, 1967; Milne, 1991). Itincorporates the effects of spatial heterogeneity andprovides a fractal dimension (D) for movement pathsthat can take values between 1, when the line iscompletely linear, and 2, when the line is so tortuous asto completely cover a 2-dimensional plane. It is believedthat a change inDwith spatial scale signifies a transitionbetween domains–different regions in the range ofspatial scales in an animals' environment–and that anincrease in D occurs when an animal is foraging moreintensively (Wiens, 1989; With, 1994; Nams, 1996,2004, 2005; Doerr and Doerr, 2004).

The Fractal Mean estimator, a modification of thedivider method, was used to calculate the D for paths ofdugongs that undertook an LSM, using the VFractal

program of Nams (1996). The correlation of successivecosines estimator was also used to detect use of ahierarchical patchy structure, in which there are patcheswithin patches in the animals' habitat. The estimatormeasures correlation in tortuosity of successive pathsegments at various scales. If one path segment is insideof a patch at scales smaller than the patch size, then theconsecutive one is also likely to be inside the patch andsimilarly for segments lying outside of patches (Nams,1996, 2004, 2005).

Each individual dugong was treated as a samplingunit to avoid pseudoreplication. Because fewer locationfixes were acquired from a dugong undertaking an LSMthan an animal resident in a core habitat area, LSMlocation fixes were spread out in space and time (anaverage of one per 3.4 days for PTT tags, 2.9 days forGPS). Lack of independence among observations canincrease the probability of a type I error by inflating thedegrees of freedom (Legendre, 1993). However, as deSolla et al. (1999) point out, eliminating autocorrelatedfixes from a data set not only reduces the sample size,but may also limit the biological significance of theanalysis if autocorrelation is intrinsic to the movementpatterns of the animal. We assumed dugong locationfixes were sequentially independent and relatively freefrom autocorrelation errors on the basis of: (1) the LSMfixes were spread out in space and time; (2) the largenumber of fixes used to construct kernel home rangespossessed a relatively constant time interval betweensuccessive locations which reduces independence biases(Marzluff et al., 2004); and (3) a large proportion of eachanimals' fixes occurred over intertidal seagrass mea-dows that were available to a foraging dugong onlytwice per day during high tide. This tidal periodicitydrives dugong movement within core habitats andspaces sequential location fixes.

To identify the salient characteristics of the divingbehaviour of dugongs undertaking LSMs, the dive datafor animals tracked via PTT/GPS collected using Mk7and Mk9 TDRs were downloaded and processed usingthe software program Multitrace (Jensen SoftwareSystems, Laboe, Germany) following Chilvers et al.(2004). This procedure enabled a dive profile to bematched to each LSM. Such matching was not possiblefor the animals fitted with the PTT tags only because ofthe low number of location fixes.

The distance and bearing of each LSMwas calculatedusing the AMAE program (Hooge and Eichenlaub,1997). Student's t-tests were used to test for relationshipsbetween dugong gender and LSM factors, and toexamine differences between data collected by PTTversus PTT/GPS telemetry. Correlation matrices were


constructed to test for potential relationships between allfactors. To avoid psuedoreplication, only the longestLSM for each individual animal was used for theanalysis. All statistical analysis was conducted using theStatistica program version-6 (StatSoft, Inc., 2001).Results are presented as means±standard errors. Rawdata from the 13 dugongs PTT-tracked in Moreton Bayby Preen (1993) and the six dugongs PTT-tracked inCleveland Bay by Marsh and Rathbun (1990) wereunavailable for establishing LSM criteria.

All statistics are reported ±standard errors unlessotherwise indicated.

3. Results

3.1. Defining large-scale movements (LSMs)

When a dugong was undertaking an LSM, itssatellite housing was pulled horizontally through thewater. The antenna was underwater and unable tocommunicate with the satellites while the dugong wasswimming. In contrast, when the animal was movingslowly around its core area(s), the cylindrical housingwas vertical and the antenna was often above thesurface and transmitted more frequently. This combi-nation of qualitative behavioural differences andtechnological constraints explains why we obtainedfar more location fixes from foraging dugongs (24.6±0.2 PTT/GPS fixes per day) than travelling dugongs(2.9±0.5 PTT/GPS fixes per day). Between one andseven location fixes (mean=3.4, S.E.=0.6) from eachPTT-tracked dugong were deleted using the locationerror filter.

When a dugong embarked on an LSM, we usuallyobtained a location fix at each end of the journey andfew, or no, fixes along the way. This phenomenon canbe seen in the frequency histograms of distancebetween consecutive locations, which fall off sharplyaround 15 km, suggesting that longer moves arerelatively rare (see Fig. 1c for a representative example).This difference provided a convenient spatial thresholdfor defining a dugong LSM as a movement greater than15 km, a figure comparable to but slightly greater thanthe mean maximum 95% kernel home range diameter of11.0 km±1.5 for the 37 dugongs that undertook anLSM. Thus, a movement of >11 km constituted a stepoutside the usual core foraging area.

Predicted squared displacements calculated using theCRW model were compared with observed dugongmovements using a paired-sample t-test conducted onobserved and predicted slopes. The paths of 21 dugongsexhibited classic CRWs (see Fig. 1d for a representative

example) and 15 were slightly overpredicted by themodel (t=−4.91, df=35, p≤0.001). In the latter case,the observed trajectories lay below the expected Rn

2 (orcovered less distance for a given number of moves) thanpredicted by the CRW model. The mean differencebetween the expected Rn

2 and a random walk for allanimals was −0.41±0.06 (N=36), which implied thatthe observed dugong movements exhibited preferencefor a region. While the tracked dugongs made intensiveuse of core habitats at local scales (<100 km2), they alsomade LSMs at larger scales (>5000 km2) which werenot adequately described by the CRWmodel. Therefore,we used fractal analysis to define transitions of scale indugong movement paths.

The plots of the fractal dimension (D) and correlationof cosines for PTT and PTT/GPS tagged animals allexhibited a discontinuity against spatial scale atapproximately 8–10 km (mean fractal D=1.22±0.02,mean cos=0.06±0.04, N=36) (see Fig. 1d and e forrepresentative examples). Micro-scale movement pat-terns below 8–10 km were tortuous (with a concomitantrise in D and a positive correlation) as the animalforaged around its core habitat areas (as predicted by theCRW model). Above this threshold, the movementpaths were more linear as the animals embarked onLSMs between foraging sites; the D decreased sharplyand the correlation returned to zero, suggesting atransition of movement behaviour between these twodomains of scale.

Considering all these factors together, we conserva-tively defined an LSM as any “displacement betweentwo consecutive stopping points” (Turchin, 1998)greater than 15 km during the entire spatiotemporalrecord of the tracked animal (path). Dugong LSM pathswere multidirectional and of varying lengths, unlike thesustained, north–south seasonal migrations of manatees(Deutsch et al., 2003a,b). Therefore, a stopping pointwas designated when an animal paused at a core habitatpreviously identified by the harmonic mean and kernelhome range analysis or remained stationary for >5 h.Biologists studying moose, another long rangingmammalian herbivore, have suggested a similar defini-tion for a long-distance movement as a one-waymovement that exceeds 10–12 km (Fuller and Keith,1981; Sandgren and Sweanor, 1988). We furtherarbitrarily delineated a dugong LSM into meso-scalemovements (moves 15–100 km) and macro-scalemovements (moves >100 km). Micro-scale movements(<15 km) occurred as tidally-driven foraging andcommuting movements between and within seagrassbeds, e.g. around the Burrum Heads seagrass meadowsof Hervey Bay, Queensland. Meso-scale movements


occurred as local-scale trips between core habitathotspots (home ranges) within bays, e.g. BurrumHeads to Kauri Creek in Hervey Bay, Queensland

0 5 10 15 20 25 30 35Distance between GPS fixes (km)

1

5

50

500

5000

No

of

ob

s(L

OG

)

N = 1851, Me an = 482. 8, StdD v = 1906. 0, Ma x = 33700, Min = 2

Dis

tan

ce2 (k

m)

1

2

3

4

Spatial scale (m) LOG100 1000 10000

Fra

ctal

-D

1.0

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

A

C D

E

(Fig. 1a). Macro-scale movements occurred as regional-scale trips between bays, e.g. Hervey Bay to ShoalwaterBay (Fig. 2).

No. of steps LOG10 100 1000

0

e+6

e+6

e+6

e+6

10 100 1000 10000

Co

rrel

atio

n c

on

seq

uti

ve c

osi

nes

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Spatial scale (m) LOG

B

F

Bul'la location fixes

50% kernel

95% kernel

Coast


3.2. Frequency of large-scale movements

Forty-four dugongs out of the 70 tracked made anLSM as defined above. Of these, 28 made a meso-scaleLSM and 14 made a macro-scale LSM (Table 1 and Fig.2). More than half of the 70 dugongs (52.5%) made atleast one LSM within 40 days of tracking. Fifteen of the70 dugongs (21.4%) made more than two LSMs withina 2-month tracking period.

The maximum recorded distance moved by a dugongfrom its capture site was not correlated with duration oftracking (r=0.06, df=36, p=0.72) and some animalswere sedentary even though they were tracked for longperiods. For example, the immature-sized male that wasVHF tracked for 483 days remained within 10 km of hiscapture (and recapture) site in Cleveland Bay (only 68locations were recorded from this animal, so it ispossible that short-term movements out of ClevelandBay were missed).

3.3. Nature of large-scale movements

The timing, distance, duration, speed and direction ofLSMs were peculiar to each dugong. The maximumdistances moved by a dugong in a single LSM(measured as the shortest distance by water betweenthe two most widely separated beginning and endlocation fixes) were extremely variable. Seven dugongs(18.9% of the total which made LSMs) moved >200 km,while 14 (38.9%) moved >100 km and more than 50%moved >50 km. The mean single LSM distance coveredby the 44 dugongs was 84.6±8.6 km (min=15,max=560, median=54), with a mean time travelling of68.3±6.1 h (min=8.3, max=400.7, median=44.2).Time taken to travel each dugong's maximum LSMwas significantly correlated with move distance for PTTlocation fixes only (r2 =0.44, df=19, p=0.001), PTT/GPS location fixes only (r2 =0.78, df=16, p=0.000) andall location fixes (r2 =0.54, df=36, p≤0.001) (Fig. 3a,Table 2). The total number of location fixes received foreach individual dugong was significantly correlated with

Fig. 1. (A) Location fixes, meso-scale movement patterns and harmonic meanduring July–August 2004. Note the clusters of location fixes in the lower meafixes identified in part of the core habitat circled in (A). (C) Histogram of dis(note log-scale on y-axis). (D) Plot of Correlated RandomWalk (CRW) test fodugong's movements are random. Dashed line represents predicted values, dconfidence intervals. (E) FractalMean D of Bul'la's movement path. (F) Plot ocompares correlation of tortuosities of adjacent path segments versus scale.indicated by a vertical line. On a small scale (<8 km), this dugong moved waround core habitat areas, giving an increasing Distance2 and Fractal D. Odecreasing D (plot E) and a correlation of cosines tending back towards zero (discontinuity of 8 km, where the animal moved out of its core foraging area

total number of days tracked for PTT only (r2 =0.24,df=19, p=0.027) and PTT/GPS tracked animals(r2 =0.35, df=16, p=0.013). Considerably more fixeswere acquired in a shorter period of time for PTT/GPS tracked animals than those tracked via PTT only(Fig. 3b).

LSMs were rapid and apparently directed. Averageminimum travelling speeds recorded for PTT/GPStagged animals (mean±S.E. meso-scale movements=1.3±0.11 km/h, min=0.3, max=3.0; macro-scalemovements=1.1±0.09 km/h, min=0.8, max=1.3)were high in comparison with the speeds recorded bydugongs making micro-scale moves (<15 km) whileforaging around core habitat hotspots (0.4±0.004 km/h,min=0.0006, max=3.0). Minimum mean daily distancetravelled during a meso-scale move was 31.8±2.67 km,min=6.2, max=71.7. Minimum mean daily distancetravelled during a macro-scale move was 25.9±2.23km, min=18.8, max=30.7. LSM travelling speeds werenot correlated with either move distance (r2 =0.02,df=16, p=0.57) or move time (r2 =0.04, df=16,p=0.24).

As explained above, most of an LSM typicallyoccurred between two successive satellite fixes. Weoften failed to receive signals from a transmitter forseveral days and recorded the next location fix >100 kmaway from the last fix (particularly for PTT tags). Anadult female tagged at Hinchinbrook Island was notlocated for 9 days, after which she was recordedapproximately 520 km from the tagging site in PrincessCharlotte Bay on Cape York Peninsula (Fig. 2b). Therewere some exceptions to the low average number oflocation fixes per LSM. For example, we obtained 11location fixes for Bul'la, a 2.9-m-long adult male, duringhis meso-scale LSM down the Great Sandy Strait nearHervey Bay and 37 during the return trip following his20 day residency at Kauri Creek (Fig. 1a). His tortuous18.7 and 55.9 h 81-km journeys necessitated navigatingthrough numerous small islands, shoals and shallowintertidal seagrass meadows. The satellite tag waspresumably often at the surface and able to transmit

grid for Bul'la, a 2.9-m-long male tracked in Hervey Bay, Queenslandn proportion cells. (B) Kernel home range contours calculated using thetance between consecutive location fixes for Bul'la, tracked using GPSr Bul'la's movement path showing CRW behaviour, implying that thisotted line observed values. Upper and lower solid lines represent 95%f correlations of successive cosines for Bul'la's movement path, whichNote discrete disjunctions in plots E and F around the 8000 m markith the tortuous correlated random walk shown in plot D as it foragedn a larger scale (>8 km), it travelled with a straighter path, giving aplot F). Hence, a transition between two domains of scale occurs at theduring a long distance move.

Fig. 2. Map of the Queensland and Northern Territory coast of Australia, showing tagging sites (circled) and examples of macro-scale moves fromthree dugongs: (A) “Doreen”, a female who moved from her capture site at Borroloola 335 km east to Forsyth Island in 1994. (B) “MT”, a female whomoved from the capture site around Hinchinbrook Island 560 km north to Princess Charlotte Bay in 1997. (C) “Wunai”, a 1.9 m male who movedfrom the capture site in Hervey Bay 513 km north to become resident around Clairview in 2003.


location fixes acquired during each LSM (even thoughthe animal was travelling rapidly; 3.0 km/h). A likelyexplanation is that the animal frequently paused in itstravels (perhaps to feed, orient or rest) long enough forGPS fixes to be taken. In comparison, no location fixeswere obtained when this dugong travelled through deepwater on its way from Burrum Heads to Big WoodyIsland, at the mouth of the Sandy Strait.

Dugongs apparently do not stop to feed or rest forsignificant periods while on these journeys, even thoughthey often pass by areas where significant seagrassresources are available. For example, a 1.9-m-long maleanimal (Wunai) was PTT/GPS tracked in Hervey Bay inJuly 2003. Two days following tagging, the animal leftthe bay and was not located again for 8 days, after which

he was found 350 km north at Port Clinton. Eight dayslater he left Port Clinton and was relocated a further135 km north at Clairview, where he was resident forthe remainder of the tracking period (51 days, Fig. 2c).

The likelihood of a dugong engaging in an LSM andthe scale of that movement appeared to be independentof the demographic category of the dugong. Somemales, females and females with calves travelled macro-scale distances of over 200 km, while other individualswithin each demographic group moved in micro-scalesless than 15 km within the tracking period. Body lengthhad no apparent influence on frequency or character-istics of LSMs.

The mean time of the first LSM post-capture for allanimals was variable (mean=33.4±10.3 days, min=0

0 100 200 300 400 500 600

Move distance (km)

0

2

4

6

8

10

12

14

16

18

Mo

ve t

ime

(day

s)

GPS PTT

0 100 200 300 400 500 600

Total days tracked

-200

0

200

400

600

800

1000

1200

1400

1600

1800

2000

2200

To

tal N

o. l

oca

tio

n f

ixes

GPS PTT

A B

Fig. 3. (A) Plot of maximummove time for each individual dugong against maximummove distance grouped according to telemetry type. (B) Plot oftotal number of location fixes received for each individual dugong against total number of days tracked grouped according to telemetry type.


days, max=271.6 days), suggesting that most animalsdid not undertake an LSM in response to being captured.Mean times varied according to capture location, withdugongs captured off Burrum Heads leaving for theirfirst LSM much earlier (mean=4.1±2.0 days, N=17)than those captured around Hinchinbrook Island(mean=102.5±31.1 days, N=11).

3.4. Relationship between movement and seatemperature

Dugongs rarely travelled far from the coast (meanmax distance=12.8±1.3 km), however, there wereexceptions. The meso-scale movements of at least 13dugongs captured close to the high latitude portion ofthe dugongs' range on the Australian east coast in winterseemed to be prompted by water temperature. DuringJuly and early August of 2003 and 2004, six dugongsmade an 80 km direct LSM across the middle of HerveyBay (−25.22°S, 152.69°E, Fig. 4a and d). The animalstravelled from their capture location at Burrum Heads tothe deep water at the northernmost tip of Sandy Cape,Fraser Island; the one-way trip took 2 to 6 days tocomplete. Once off Sandy Cape, they spent an averageof 10 days moving slowly about a small area (<40 km2),sometimes less than 100 m off the shoreline, beforeeither returning to Burrum Heads or travelling downGreat Sandy Strait. A preliminary seagrass surveyconducted off Sandy Cape with a grab sampler indicateda general absence of seagrass food resources, mostlikely a consequence of the unstable benthic sediments.

Analysis of sea surface temperatures (SST) for theregion acquired via a NOAA thermal imaging satellite(http://www.aims.gov.au/pages/remote-sensing.html)

show that during July, Hervey Bay effectively acts like agiant ‘cup’ of cold water; isolating water cooled by thecoastal winter air from the warmer southerly EastAustralian current (Fig. 4a and d). Mean daily SSTs inthe inshore region of Burrum Heads (where dugongswere tagged) in the beginning of July 2003 and 2004(when dugongs began making meso-scale LSMs to theouter bay) averaged 19 °C; benthic sea temperaturesrecorded off Burrum Heads by scuba divers conductingseagrass surveys at the same time averaged 16 °C(Sheppard, personal observation). These temperaturesare lower than the SSTs of 21 °C recorded for watersaround Sandy Cape. Temperature loggers incorporatedinto the GPS tags and the TDR units recorded congruenttemperature variances (Fig. 4c and f). Once SSTs in theBay began to warm from mid-August, the trackeddugongs no longer made LSMs to Sandy Cape.Dugongs in Hervey Bay and Moreton Bay in Queens-land, as well as Shark Bay in Western Australia (Marshet al., 1994), seem to respond to a water temperaturethreshold of 17–18 °C, below which they undertakemeso-scale thermoregulatory movements.

In Moreton Bay, Queensland, at the high latitudeextreme of the dugong's area of occupancy on the eastcoast of Australia, dugongs were able to make micro-scale movements of about 10 km to escape the coldwinter temperatures in the bay. During winter 1988–1989, all six tracked dugongs left the bay at some stage.They moved through South Passage to where the watersof the East Australian current were >5 °C warmer thanthe 17–18 °C water in the bay. The dugongs regularlyrode the currents in and out of the bay to move betweencool water feeding areas and warm water resting areas(although none made meso- or macro-scale

http://www.aims.gov.au/pages/remote-ensing.html

Table 2Correlation matrix of potential behavioural interactions characteristic of dugong large-scale moves (LSMs) (significant correlations in bold)

Total numberdays tracked

Total numberlocation fixes

Distance oflargest move

Move time Mean min.speed

Size core hotspot95% Kernel

Patch residencytime beforelargest LSM

Proportiontime diving

Total numberlocation fixes

(GPS: r2=0.35p=0.013)(PTT: r2=0.24p=0.027)

– – – – – – –

Distance oflargest move

(r2=<0.01p=0.69) a

(r2=0.01p=0.62) a

– – – – – –

Move time (r2=<0.01p=0.69) a

(r2=0.04p=0.23) a

(GPS: r2=0.78p=0.000)(PTT: r2=0.44p=0.001)

– – – – –

Mean min. speed (r2=0.01p=0.77) b

(r2=0.14p=0.14) b

(r2<0.02p=0.57) b

(r2=0.04p=0.24) b

– – – –

Size core hotspot95% Kernel

(r2=0.06p=0.16) a

(r2=0.09p=0.07) a

(PTT: r2=0.55p=0.000)(GPS: r2=0.02p=0.58)

(r2=<0.01p=0.93) b

(r2=0.05p=0.39) b

– – –

Patch residencytime beforelargest LDM

NA NA (r2=0.01p=0.62) a

(r2=0.03p=0.30) b

(r2=<0.01p=0.95) b

(r2=0.08p=0.27) b

– –

Proportion timediving

(r2=0.31p=0.12) b

(r2=0.01p=0.85) b

(r2=<0.01p=0.87) b

(r2=0.03p=0.66) b

(r2=<0.01p=0.96) b

(r2=0.02p=0.75) b

(r2=0.02p=0.71) b

–

Mean max. divedepth

(r2=0.14p=0.32) b

(r2=0.02p=0.71) b

(r2=0.01p=0.83) b

(r2=<0.01p=0.87) b

(r2=0.04p=0.61) b

(r2=0.33p=0.11) b

(r2=0.19p=0.25) b

(r2=0.36p=0.09) b

a Calculated using GPS and PTT location fixes.b Calculated using GPS location fixes only.

74J.K

.Sheppard

etal.

/Journal

ofExperim

entalMarine

Biology

andEcology

334(2006)

64–83

Fig. 4. (A) Burando, a 2.7 m male tagged in Hervey Bay during July 2003, made this 79.1 km, 4.2-day meso-large-scale movement (LSM) from thecapture site at Burrum Heads to Sandy Cape. The move is superimposed over Sea Surface Temperature (SST) measurements for Hervey Bay from aNOAA thermal imaging satellite at the start of the move (note temperature gradient across the Bay and the lack of location fixes between start and endpoints). (B) Projected bathymetric profile of the move if the animal followed a straight-line trajectory from start to end fix across the Bay. (C)Maximum dive depth and temperature profile of the LSM as recorded by the TDR. (D, E and F) The LSM, SST and dive characteristics of the returntrip back to Burrum Heads. Note the position of 15 fixes obtained during the return move marked as A and B in (D) and (F).


Table 3Dive profile characteristics of 22 large-scale moves recorded from nine individual dugongs

DugongID

Move time(days)

Distance(km)

No. ofdives

Timewithin1.5 msurface

Total timediving(>1.5 m)

Proportiontime within1.5 m surface(%)

Proportiontimediving (%)

Dive time Max.divetime

Max.divedepth (m)

Dive depth(m)

Mean ±S.E. Mean ±S.E.

Bur'ando 4.7 79.1 3206 13:27:46 99:46:24 11.9 88.1 01:52 :02 09:20 33 5.8 0.1Bur'ando 6.6 72.7 4081 25:15:49 133:40:16 15.9 84.1 01:58 :02 08:14 33 7.2 0.1Wunai 16.7 513.0 10,298 74:33:36 326:05:24 18.6 81.4 01:54 :01 10:30 24 4.3 0.0Gul'awa 0.6 34.9 444 2:25:48 11:32:10 17.4 82.6 01:26 :04 06:40 16 4.4 0.2Gul'awa 0.6 44.7 449 1:38:33 13:19:30 11.0 89.0 01:47 :04 07:32 20 5.3 0.2Gul'awa 0.9 38.6 549 4:54:01 17:04:54 22.3 77.7 01:52 :04 09:56 18 4.0 0.1Gul'awa 0.8 42.4 466 4:17:26 15:29:06 21.7 78.3 02:00 :05 08:28 13 4.1 0.1Kal'ba 2.4 71.4 1644 9:20:40 49:22:22 15.9 84.1 01:48 :03 08:48 27 4.9 0.1Kal'ba 2.7 84.6 1619 10:06:17 53:51:26 15.8 84.2 02:00 :03 11:04 29 4.0 0.1Yuan'gan 1.0 43.0 875 6:25:44 17:56:50 26.4 73.6 01:14 :02 06:00 12 3.6 0.1Yuan'gan 1.7 43.3 1271 14:55:44 25:43:00 36.7 63.3 01:13 :02 04:46 16 3.7 0.1Yuan'gan 1.2 34.4 921 9:02:04 20:39:23 30.4 69.6 01:21 :03 05:16 24 3.4 0.1Bunda 0.5 18.7 365 1:56:28 10:04:22 16.2 83.9 01:40 :06 06:50 19 5.0 0.3Bunda 2.0 18.3 1212 6:07:28 42:08:54 12.7 87.3 02:05 :04 07:16 21 6.6 0.2Bunda 2.2 27.5 1877 5:41:48 46:23:41 10.9 89.1 01:39 :02 07:26 19 4.7 0.1Bunda 0.9 23.1 575 2:09:10 19:08:54 10.1 89.9 02:00 :05 06:38 13 4.7 0.1Bunda 0.4 18.1 246 0:59:44 8:21:11 10.6 89.4 02:02 :07 06:16 17 6.6 0.3Bumkaman 15.1 284.0 10,797 22:02:46 340:39:45 6.1 93.9 01:48 :01 07:50 27 4.4 0.0Bumkaman 12.0 273.0 7884 32:16:06 255:04:28 11.2 88.8 01:54 :01 07:46 35 5.7 0.1Gandhau 2.0 77.6 2566 4:36:42 42:45:06 9.7 90.3 01:00 :01 07:16 24 4.5 0.1Gandhau 3.7 110.9 3775 5:25:14 84:16:06 6.0 94.0 01:19 :01 07:58 27 3.7 0.1Wooinbum 1.1 23.4 675 3:46:46 21:45:32 14.8 85.2 02:03 :04 09:10 14 3.7 0.1

Date & Time

12:

45

01:

15

01:

30

01:

45

02:

15

02:

30

02:

45

03:

15

03:

30

03:

45

04:

15

04:

30

04:

45

05:

15

05:

30

21/0

7 01

:00

21/0

7 02

:00

21/0

7 03

:00

21/0

7 04

:00

21/0

7 05

:00

Div

e d

epth

(m

)

-24

-22

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

2

Fig. 5. Sample of dugong Bumkaman's dive profile (4.5 h) during his284 km large-scale movement from Burrum Heads to Great KeppelIsland. Note repeated 5-min dives to >21 m.


movements). No tracked animals were recorded leavingthe bay during spring or summer (Preen, 1993).

The first animal tagged by Marsh and Rathbun(1990) in Cleveland Bay in 1986 may also have madeone meso-scale LSM in response to a 2 °C seatemperature drop during an unseasonable cold front.The animal travelled rapidly from Upstart Bay to itsoriginal capture location, even though the temperature inUpstart Bay (>27 °C) was well within the thermaltolerance range of dugongs. The motivations and cuesdriving the LSMs of the other 31 dugongs are unknown.

3.5. Diving behaviour during long distance movements

The mean proportion of travelling time spent on thesurface (defined as <1.5 m depth, see Chilvers et al.,2004) during an LSM for nine dugongs was 15.7±2.3%,suggesting that dugongs travel through the watercolumn rather than at the surface (Table 3). Analysisof the dive profiles of these animals show repeated deepdives are made along the travel route (Fig. 4). A malethat made a return 290 km macro-scale LSM fromBurrum Heads north to Great Keppel Island in July 2003made consecutive 5-min deep dives up to 27 m for the15 days it took to make the initial journey (Fig. 5). Hedived to 33.5 m on the 12-day return trip. The regular U-

shaped profile of these dives suggests that they were tothe seafloor; however, this cannot be substantiatedbecause the paucity of location fixes acquired during theLSM precludes matching the depth of the dive to thebathymetric depth at its location.


Dugongs travelling to and from Sandy Cape madesimilar regular deep dives as they travelled across themiddle of Hervey Bay (e.g., Fig. 4c and f), with oneanimal making a 5-min dive to a maximum of 36.5 m.The dive depths of sedentary dugongs around corehabitat areas (i.e. within the kernel home range) weregenerally shallow (<3 m), because of the shallowbathymetry of intertidal foraging sites (similar toChilvers et al., 2004).

3.6. Home range size

The 95% kernel home range size of individualdugongs varied from 0.5 km2 for a sedentary dugongwith high site fidelity to its core habitat to 733 km2 foran animal that ranged widely (Fig. 6). Home range sizewas correlated with maximum LSM length for PTT-tracked animals (r2 =0.55, df=19, p≤0.001) but notGPS-tracked animals (r2 =0.02, df=16, p=0.58). Homerange size was not a function of the number of days ananimal was tracked (r2 =0.06, df=35, p=0.16) or thenumber of location fixes recorded (r2 =0.09, df=35,p=0.07) (Table 2).

Home range size varied with the geographic locationwhere a dugong was resident (Fig. 6). The mean homerange size for dugongs resident off Burrum Heads wasconsiderably smaller (26.3±12.4 km2) than thoseanimals resident around Hinchinbrook Island (82.6±17.9 km2) and Borroloola (280.3±122.3 km2). Meanhome range size was consistently smaller for PTT/GPS

Home range location

Burru

m H

eads

(N=1

5)

Clairv

iew (N

=1)

Kauri

Ck (n=

1)

Borro

loola

(N=5

)

Hitchin

broo

k Is (

N=7)

Prince

ss C

harlo

tte B

ay (N

=1)

Port C

linto

n (N

=1)

Shoalw

ater

Bay

(N=2

)

Clevela

nd B

ay (N

=4)

Mor

eton

Bay

(N=1

3)Mea

n s

ize

95%

ker

nel

ho

me

ran

ge

(km

2 ) ±

s.e.

0

100

200

300

400

500

600

GPSPTT

Fig. 6. Mean±95% kernel home range sizes for all dugongs organisedaccording to home range location and transmitter type.

(24.0±11.0 km2) tracked dugongs and generally largewith considerable size variation (131.1±27.9 km2) forPTT tracked dugongs. These differences are presumablyan artefact of the coarse-scale precision and resolutionof PTT location fixes compared to PTT/GPS, althoughthe within-PTT kernel size variation also suggests ageographic influence. In other words, kernel size waslargely influenced by geography, but variability at eachsite was affected by the coarse PTT location fixresolution.

4. Discussion

4.1. Heterogeneity of large-scale movements ofdugongs

The 70 individual dugongs we tracked for periodsranging from 15 to 551 days exhibited a large range ofindividualistic movement behaviours; 26 individualswere relatively sedentary while 44 moved distances ofup to 625 km from their capture sites (Table 1 and Fig.2). The animals that exhibited these large-scale move-ments were from all age–sex classes including cowswith calves (Table 1). At least some of the movementswere return movements, suggesting that such move-ments were ranging rather than dispersal movements.Macro-scale (>100 km) and meso-scale (15–100 km)movements were qualitatively different from micro-scale movements (<15 km). Although at least some ofthe large-scale movements of animals caught atlocations close to the high latitude limits of the dugong'srange in eastern Australia were movements to warm-water habitat, animals caught in the tropics alsoundertook large-scale movements without such thermalbenefits.

Minimum mean±S.E. daily travelling rates fordugongs during meso-scale LSMs (31.8±2.67 km, thisstudy) were comparable to the rates±S.D. of travel ofFlorida manatees during their southward autumn (33.5±7.6 km) and northward spring (27.3±10.5 km) migra-tions (Deutsch et al., 2003b). However, the maximumrecorded speed of a PTT/GPS-tracked dugong under-taking a meso-scale LSM (3.0 km/h) was slightly lowerthan that of a Florida manatee (3.6 km/h) (Deutsch et al.,2003b).

4.2. Nature of dugong long distance movementbehaviour

4.2.1. Coastal fidelityWith the exception of the six animals that travelled

across the centre of Hervey Bay to Sandy Cape (e.g.,


Fig. 4a), the large-scale movements documented herewere coastal, the tracked dugongs rarely venturing morethan 20 km off the coast. The dugong's dependence onseagrass for food (Heinsohn and Birch, 1972; Lipkin,1975; Wake, 1975; Marsh et al., 1982; Preen, 1995b)limits them to coastal waters but dugongs are frequentlysighted much further from the coast than 20 km on aerialsurveys (Marsh and Saalfeld, 1988; Marsh et al., 1997;Gales et al., 2004; Marsh and Lawler, 2002). A plausibleexplanation for dugongs following the shoreline onmost long distance movements is that it is generally theshortest distance between origin and destination (as wascrossing the centre of Hervey Bay).

Dugongs may also be using the coastal geomorphol-ogy to navigate using visual, magnetic, chemosensoryor tactile cues. Dugongs possess the most developedsinus hairs of any animal, especially on the muzzle(Kamiya and Yamasaki, 1979; Marshall et al., 2003).This sensitive bristle array may allow travellingdugongs to exploit geomorphic variations in the benthosas an orientation aid. Presumably such assistance wouldnot be available to dugongs crossing deep oceanic water(>4 km deep) where the bottom is well beyond theirknown or likely diving ability (Marsh et al., 2002).

4.2.2. Deep divingThe hypothesis that dugongs find their way around

coastal areas by tracking the bottom is supported by theTDR evidence that travelling dugongs make repeateddeep dives (presumably along the seafloor) (Fig. 5,Table 3). In addition to the hypothesis that such dives arean orientation aid, there are other explanations for adugong undertaking deep dives during a large-scalemovement. The coastline where this behaviour wasobserved is generally bereft of suitable seagrass foodresources (e.g. the exposed and often rocky shoresbetween Bundaberg and Port Clinton) suggesting thatthese were not foraging dives. Such behaviour may bean anti-predator strategy. Sharks, the primary predator ofdugongs, often surprise-attack from below (Heithaus etal., 2002). Migrating green turtles (Hays et al., 2001)and foraging elephant seals (Crocker et al., 1997)exhibit similar subsurface swimming or resting beha-viours, which may reduce the visibility of theirsilhouette against the surface to hunting sharks.

Alternatively, dugongs may travel along the seafloorto save energy, despite the energetic cost of travelling tothe surface every 5 min or so to breathe. Compared withmany other diving mammals (e.g. elephant seals,Hindell et al., 1991 and Weddell seals, Watanabe etal., 2003), dugongs operate in very shallow water, andso diving costs to the benthos may be energetically

negligible. In currents, fluid dynamics exert less drag inthe boundary layer just above the benthos than in thewater column. Travelling dugongs may be counteringthe drag from tidal flows and oceanic currents byswimming in this manner. This energy conservingbehaviour has been observed in migrating elvers inexperimental flume tanks (Barbin and Krueger, 1994),as well as in fish inhabiting natural reefs (Fulton et al.,2001). However, to verify that a dugong is tracking thebottom, the depth of a dugong's dive must be matchedwith the bathymetry to determine whether the animal ison the seafloor rather than suspended in the watercolumn. The animal's location then must be matchedwith the water speed and flow direction to judge whetherbottom-tracking is energetically viable. Unfortunatelyeven using PTT/GPS transmitters, few location fixeswere obtained during LSMs, making it impossible for usto test such hypotheses.

4.3. Stimuli for long distance movement behaviour

4.3.1. Thermal regulationSeasonal movements in response to low water

temperatures are an important feature of the movementsof Florida manatees that live at relatively high latitudes(Layne, 1965; Hartman, 1979; Reynolds and Wilcox,1985, 1994; Weigle et al., 2001; Deutsch et al., 2003a,b). Dugongs are also sensitive to water temperaturesbelow about 18 °C (Anderson, 1986, 1994; Preen, 1989;Marsh et al., 1991; Lanyon et al., 2005). We tracked sixdugongs making repeated return meso-scale 80 kmmovements across Hervey Bay in winter (e.g., Fig. 4aand d) to the warmer oceanic waters off Sandy Cape,despite the apparent lack of suitable seagrass meadowsin the area and the high numbers of sharks reportedduring aerial surveys. During winter, dugongs capturedin Moreton Bay were also found up to 15 km outsidethat bay in a region where water temperature is up to 5 °Chigher. In Shark Bay, dugongs move from shallowinshore summer feeding areas to deeper offshore water inwinter where the temperature remains higher (Anderson,1986; Marsh et al., 1994; Gales et al., 2004).

Preen (2004) sighted large tightly aggregated herdsof dugongs in winter in the Arabian Gulf, in contrast totheir dispersed distribution in summer. He suggests thatdugongs may aggregate around thermal springs inwinter, which would entail large-scale movements ofapproximately 400 km from the summer habitats.Dugongs are seen in New South Wales coastal waterin summer to about 27°S but not in winter (Allen et al.,2004), suggesting that these animals move back toQueensland waters when the water temperature drops.


Thus, there seem to be two types of large-scalemovements in response to water temperatures belowabout 18 °C: meso-scale movements to warm-waterhabitats without seagrass (e.g. movements acrossHervey Bay or out of Moreton Bay) or macro-scaleseasonal movements to seagrass beds in warmer waterssuch as those that occur in Shark Bay or along the NSW-Southern Queensland coast. This delineation of move-ment behaviours is analogous to Florida manatees,which make macro-scale migrations in winter tosouthern Florida where warmer waters prevail. Floridamanatees also make meso-scale movements to warm-water refuges (springs, power plants) in proximity toseagrass beds, where they commute between foragingand thermal areas (Deutsch et al., 2003a).

4.3.2. Spatiotemporal heterogeneity of seagrass foodresources

Dugongs are primary consumers and seagrassspecialists (Heinsohn and Birch, 1972; Marsh et al.,1982; Lanyon et al., 1989; Preen, 1995a). Hence,dugong distribution is constrained by the availabilityof suitable seagrass beds. Seagrass meadows varyacross bathymetric, nutritional and seasonal gradients(see Short et al., 2001 for reviews) and are subject tolarge-scale periodic diebacks (e.g. Preen and Marsh,1995; Preen et al., 1995). Grazing herbivores typicallyadjust their densities and vegetation utilization patternsbased on the productivity potential of their foodsources, aggregating where the potential is high anddispersing from areas where that potential is low(McNaughton, 1984). Consequently, it is reasonable toexpect that dugongs have evolved to cope with theunpredictable and patchy nature of seagrass bymoving to alternative areas known to have adequatefood resources in the past.

Knowledge of the spatiotemporal distribution ofpreferred food resources should enable foragingdugongs to better exploit their heterogenous environ-ments. Our tracking data suggest that dugongs maintaina spatial memory of specific habitat hotspots—patchesof quality seagrass food resources, which they visitperiodically. Dugongs may use these visits to monitorthe quantity and quality of food resources, balancing thenet rate of energy gain from the current patch with thepotential energetic costs and benefits of moving toanother meadow. Movements of Florida manatees alsoappear to be influenced by seasonal and regionalfluctuations in biomass and nutritional content of theirprincipal forage plants. Individual Florida manatees alsoexhibit high site fidelity to areas of core habitat (Weigleet al., 2001; Deutsch et al., 2003b).

4.3.3. Extreme climatic degradation of seagrass foodresources

As explained above, seagrasses are vulnerable toextreme episodic weather events such as cyclone andfloods. These events can result in the loss of hundreds ofsquare kilometres of seagrass from wave action, shiftingsubstrates, adverse changes in salinity, and increasedturbidities resulting from increased runoff and riverflooding (Heinsohn and Spain, 1974; Kenyon andPoiner, 1987; Preen, 1995a; Preen et al., 1995; Poinerand Peterken, 1996). For example, in 1992, more than1000 km2 of seagrass were lost from Hervey Bay,Queensland following two floods and a cyclone in quicksuccession, reducing the local dugong population from1700 dugongs to approximately 630 (Preen and Marsh,1995). Some animals appeared to survive by success-fully relocating to other localities such as Moreton Bay(−27°S) 400 km to the south (although the evidence forthis is anecdotal) and there were increased reports ofdugong sightings in New South Wales. Two dugongswere found drowned in shark control nets south ofSydney (−34°S), more than 500 km south of MoretonBay (Preen and Marsh, 1995; Allen et al., 2004). Thedugongs remaining in the Hervey Bay region stoppedbreeding and many died of starvation 6–8 months afterthe flooding event (Preen and Marsh, 1995).

Extreme climatic degradation of seagrass beds in theNingaloo/Exmouth Gulf region of north-western Aus-tralia by tropical cyclone Vance (category 5 intensity) in1999 is the probable cause of the rapid, large-scaleincrease in dugong numbers in Shark Bay, 400 km to thesouth (Gales et al., 2004). Results from the 1987, 1991,1996 and 2001 Torres Strait and 2000 northern GreatBarrier Reef aerial surveys suggest that change indugong population and distribution in these regionscannot be attributed to natural increase in the absence ofimmigration. The reasons for large-scale movementsappear to be associated in part with large-scale episodicdisturbance to habitat by cyclones and floods (Marshand Lawler, 2002; Marsh et al., 2002, 2004).

4.4. Spatial memory of foraging patch distribution

Dugongs undertaking movements of several hundredkilometres often bypass known dugong habitats. Forexample, two dugongs that moved north from HerveyBay to Great Keppel Island and Clairview respectively,passed by Rodds Bay, an area of high seagrass density185 km north of Burrum Heads (Lee Long et al., 1993).Two animals tagged off Hinchinbrook Island made 1-day return ‘visits’ to Cleveland Bay, 150 km to thesouth. These animals did not explore any known


seagrass meadows along their travel route, nor did theyspend any length of time sampling the extensivemeadows at their destination, which preclude explor-atory foraging behaviour despite the high energetic costof these moves. Florida manatees also commonlybypass known habitats during their seasonal migrations(Deutsch et al., 2003b).

Social cues may be driving these behaviours. Forexample, male dugongs may be following oestrousfemales. Intraspecific social dominance hierarchies andcompetitive exclusion may preclude subordinate ani-mals from occupying certain seagrass habitats. Acaptive calf reared into adulthood that was recentlyreleased into Moreton Bay was recaptured 8 monthslater very close to his release point, emaciated andcovered with serious tusk wounds from other dugongs.Hodgson (2004) directly observed dugong herds in clearwater in Moreton Bay over several months using ablimp mounted video and saw no agonistic or territorialbehaviour; there may have been no newcomers duringher observation periods'. Dugongs making short termvisits to established habitats may be maintaininginformation about other meta-herds and seeking out ormonitoring the status of bulls or oestrous females.

Another plausible explanation for dugongs bypassingseemingly high-quality habitat and for the individualvariability in the range of areas used by differentindividuals is that movements are learned behaviours,reflecting the individual's experience as a dependentcalf. Support for this hypothesis comes from Floridamanatees (Deutsch et al., 2003b). Researchers havetracked Florida manatee calves whose mothers weretracked some years earlier (Deutsch et al., 2003b andreferences within). These calves exhibited strong fidelityto the same areas used frequently by their mothers(Weigle et al., 2001; Deutsch et al., 2003b). In contrast,most captive-raised Florida manatees released as sub-adults along the Atlantic coast did not undertake seasonalmigrations, despite migratory behaviour occurring in87.5% of the wild population (Deutsch et al., 2003b).

There is considerable observational data as well assome experimental evidence that spatial memory alsoplays a role in the foraging behaviour of grazingterrestrial herbivores, such as ungulates (Bailey et al.,1989; Gillingham and Bunnell, 1989; Kotler et al.,1994; Edwards et al., 1996; Wallis de Vries, 1996).Many species of large terrestrial herbivores haveaccurate spatial memories and have the ability to usespatial memory to improve foraging efficiency (Saar-enmaa et al., 1988; Folse et al., 1989; Bailey etal., 1996; Edwards et al., 1996; Dumont and Hill,2001).

4.5. Implications for management

Our finding that dugongs frequently undertake LSMshave implications for management at a range of scales,and strengthen the aerial survey and genetic evidence formanagement and monitoring at ecological scales thatcross jurisdictions (Marsh et al., 2002, 2004; Gales etal., 2004; Brenda McDonald, pers. comm). Thegovernment agencies responsible for dugong conserva-tion manage at geo-political scales. Thus the dugongpopulation of a single bay is often managed as a self-contained entity. For example, aerial surveys of thedugong population of Moreton Bay were commissionedin response to the outbreak of the toxic algae Lyngbya,initially without reference to the population of nearbyHervey Bay (Marsh et al., 1990; Marsh and Lawler,2001). Clearly, with movement between bays a commonoccurrence, estimates of population size and trends canonly be meaningfully made at regional scales. Thecapacity of large-scale monitoring programs to detecttrends in dugong numbers at scales of even thousands ofkm2 is confounded by the dugongs' tendency toundertake LSMs (Gales et al., 2004; Marsh et al.,2004). Conservation actions implemented at smallerscales may not be sufficient to ensure the sustainabilityof dugong populations. When developing strategies fordugong conservation, the dugong population of any onebay should be considered not as an isolated entity but asa fluctuating component of a spatially dynamicmetapopulation.

The tendency for dugongs to track the bottom onlarge-scale movements may increase their vulnerabilityto incidental capture in bottom set gill nets, suggestingthat that strategy of protecting dugongs in a string ofseparate Dugong Protection Areas in which nettingpractices are modified or banned may not be as effectiveas previously thought (Marsh, 2000). In addition, ifdugongs transfer their spatial knowledge of the locationof quality food resource patches to their offspring, thenlocal depletions will lead to loss of this knowledge.Areas of high quality seagrass may thus becomeunknown to dugongs. In the absence of grazing pressuresuch areas may become less valuable as dugong habitatif the preferred early seral stage species of seagrassconvert to more fibrous species (Preen, 1995a).

Acknowledgments

Funding and in-kind support for this research wasgenerously provided from the following sources: CRCReef, Australian Research Council, Sea World Researchand Rescue Foundation, Great Barrier Reef Marine Park


Authority, The Ian Potter Foundation, The Society forMarine Mammalogy, The Norman Wettenhall Founda-tion, The Rothwell Wildlife Conservation Fund,Queensland Parks and Wildlife Service, QueenslandEnvironmental Protection Agency, McArthur RiverMine, the Wondunna Aboriginal Corporation and theMabunji Aboriginal Resources Association Inc. Wethank Nick Gales, Col Limpus and Paul Harvey for theirvalued contributions and Galen Rathbun and Jim Reidfor help with the design of the transmitters and harnessassembly. We are grateful to the many field assistantswho helped capture and tag wild dugongs, in particular,Tom Stevens, Jody Kruger, Guido Parra, AmeerAbdulla, Pam Quayle, Amanda Hodgson, and BradChainey from Hervey Bay Air Adventures. We aregrateful for the knowledge and assistance of TomSimon, Graeme Friday, Steve Johnson and ArchieJohnson and for the helpful comments of two anony-mous reviewers who greatly improved the manuscript.[RH]

References

Allen, S., Marsh, H., Hodgson, A., 2004. Occurrence and Conserva-tion of the Dugong (Sirenia: Dugongidae) in New South Wales.Proc. Linn. Soc. N. S. W. 125, 211–216.

Anderson, P.K., 1986. Dugongs of Shark Bay, Australia—seasonalmigration, water temperature, and forage. Natl. Geogr. Res. 2,473–490.

Anderson, P.K., 1994. Dugong distribution, the seagrass Halophilaspinulosa, and thermal environment in winter in deeper waters ofeastern Shark Bay, Western Australia. Wildl. Res. 21, 381–388.

Atkinson, M.J., Smith, S.V., 1983. C:N:P ratios of benthic marineplants. Limnol. Oceanogr. 28, 568–574.

Bailey, D., Rittenhouse, L., Hart, R.H., Richards, R.W., 1989.Characteristics of spatial memory in cattle. Appl. Anim. Behav.Sci. 23, 331–340.

Bailey, D., Gross, J., Laca, E., Rittenhouse, L., Coughenour, M., Swift,D., Sims, P., 1996. Mechanisms that result in large herbivoregrazing distribution patterns. J. Range Manag. 49, 386–400.

Barbin, G.P., Krueger, W.H., 1994. Behaviour and swimmingperformance of elvers of the American eel, Anguilla rostrata, inan experimental flume. J. Fish Biol. 45, 111–121.

Bengtson, J.L., 1983. Estimating food consumption of free rangingmanatees in Florida. J. Wildl. Manage. 47, 1186–1192.

Chilvers, B.L., Delean, S., Gales, N.J., Holley, D.K., Lawler, I.R.,Marsh, H., Preen, A.R., 2004. Diving behaviour of dugongs,Dugong dugon. J. Exp. Mar. Biol. Ecol. 304, 203–224.

Crocker, D.E., Le Boeuf, B.J., Costa, D.P., 1997. Drift diving in femalenorthern elephant seals: implications for food processing. Can. J.Zool. 75, 27.

de Solla, S.R., Bonduriansky, R., Brooks, R.J., 1999. Eliminatingautocorrelation reduces biological relevance of home rangeestimates. J. Anim. Ecol. 68, 221–334.

de Thoisy, B., Spiegelberger, T., Rousseau, S., Talvy, G., Vogel, I., Vie,J., 2003. Distribution, habitat, and conservation status of the West

Indian manatee Trichechus manatus in French Guiana. Oryx 37,431–436.

Deutsch, C.J., Edwards, H.H., Smith, A.J., 2003a. Argos-linked GPStags as a new tool to investigate manatee behavior: preliminaryresults on winter use of a warm-water refuge. In: Proceedings ofthe Argos Animal Tracking Symposium. March 24–26, 2003.Annapolis, Maryland. 6 pp.

Deutsch, C.J., Reid, J.P., Bonde, R.K., Easton, D.E., Kochman, H.I.,O'Shea, T.J., 2003b. Seasonal movements, migratory behavior,and site fidelity of West Indian manatees along the Atlantic Coastof the United States. Wildl. Monogr. 151, 1–77.

Dixon, K.R., Chapman, J.A., 1980. Harmonic mean measure of animalactivity areas. Ecology 61, 1040–1044.

Doerr, V.A.J., Doerr, E.D., 2004. Fractal analysis can explainindividual variation in dispersal search paths. Ecology 85,1428–1438.

Dumont, B., Hill, D.R.C., 2001. Multi-agent simulation of groupforaging in sheep: effects of spatial memory, conspecific attractionand plot size. Ecol. Model. 141, 201–215.

Edwards, G.R., Newman, J.A., Parsons, A.J., Krebs, J.R., 1996. Theuse of spatial memory by grazing animals to locate food patches inspatially heterogeneous environments: an example with sheep.Appl. Anim. Behav. Sci. 50, 147–160.

Folse, L.J., Packard, J.M., Grant, W.E., 1989. AI modelling of animalmovements in a heterogeneous landscape. Ecol. Model. 46, 57–72.

Fuller, T.K., Keith, L.B., 1981. Woodland caribou populationdynamics in north-eastern Alberta. J. Wildl. Manage. 45, 197–213.

Fulton, C., Bellwood, D., Wainwright, P., 2001. The relationshipbetween swimming ability and habitat use in wrasses (Labridae).Mar. Biol. 139, 25–33.

Gales, N.J., McCauley, R.D., Lanyon, J., Holley, D.K., 2004. Changesin abundance of dugongs in Shark Bay, Ningaloo and ExmouthGulf, Western Australia: evidence for large-scale migration. Wildl.Res. 31, 283–290.

Gillingham, M.P., Bunnell, F.L., 1989. Effects of learning on foodselection and searching behaviour of deer. Can. J. Zool. 67, 24–32.

Gitzen, R.A., Millspaugh, J.J., 2003. Evaluation of least squares crossvalidation bandwidth selection options for kernel estimation.Wildl. Soc. Bull. 31, 823–831.

Hartman, D.S., 1979. Ecology and Behaviour of the Manatee(Trichechus manatus) in Florida. Spec. Publ. Am. Soc.Mammal. 5.

Hays, G.C., Åkesson, S., Broderick, A.C., Glen, F., Godley, B.J.,Luschi, P., Martin, C., Metcalfe, J.D., Papi, F., 2001. The divingbehaviour of green turtles undertaking oceanic migration to andfrom Ascension Island: dive durations, dive profiles and depthdistribution. J. Exp. Biol. 204, 4093–4098.

Heinsohn, G.E., Birch, W.R., 1972. Foods and feeding habits of thedugong, Dugong dugon (Erxleben), in northern Queensland,Australia. Mammalia 36, 414–422.

Heinsohn, G.E., Spain, A.V., 1974. Effects of a tropical cyclone onlittoral and sub-littoral biotic communities and on a population ofdugongs (Dugong dugon (Muller)). Biol. Conserv. 6, 143–152.

Heithaus, M.R., Dill, L.M., Marshall, G.J., Buhleier, L.M., 2002.Habitat use and foraging behavior of tiger sharks (Galeocerdocuvier) in a seagrass ecosystem. Mar. Biol. 140, 237–248.

Hindell, M.A., Slip, D.J., Burton, H.R., 1991. Diving behaviour ofadult male and female southern elephant seals. Aust. J. Zool. 39,595–619.

Hodgson, A.J., 2004. Dugong behaviour and responses to humaninfluences. PhD thesis, James Cook University, Townsville,Queensland, Australia. p. 294.


Hooge, P.N., Eichenlaub, B., 1997. Animal Movement Extension toArcview. ver. 1.1. Alaska Biological Science Center. U.S.Geological Survey, Anchorage, AK, USA.

Hurst, L.A., Beck, C.A., 1988. Microhistological Characteristics ofSelected Aquatic Plants of Florida: With Techniques for the Studyof Manatee Food Habits. Fish and Wildlife Service, USDepartment of the Interior, Washington, DC, USA, p. 145.

Husar, S.L., 1978. Dugong dugon. Mamm. Species 88, 1–7.Jones, W., 1994. Nutrient composition of freshwater plants and its

correlation to manatee food preferences. First InternationalManatee and Dugong Research Conference, Gainsville, Florida.

Kamiya, T., Yamasaki, F., 1979. A morphological note on the sinushair of the dugong. In: Marsh, H. (Ed.), The Dugong: Proceedingsof a Seminar/Workshop Held at James Cook University of NorthQueensland. James Cook University, Department of Zoology,Townsville.

Kareiva, P.M., Shigesada, N., 1983. Analysing insect movement as acorrelated random walk. Oecologia 56, 234–238.

Keating, K.A., Brewster, W.G., Key, C.H., 1991. Satellite telemetry:performance of animal-tracking systems. J. Wildl. Manage. 55,160–171.

Kenney, R.D., Mayo, C.A., Winn, H.E., 2001. Migration and foragingstrategies at varying spatial scales in western North Atlantic rightwhales. J. Cetacean Res. Manag. 2, 251–260 (Special Issue).

Kenyon, R., Poiner, I., 1987. Seagrass and Cyclones in the WesternGulf of Carpentaria. Information Sheet (1). Marine Laboratories,CSIRO Australia.

Kotler, B.P., Gross, J.E., Mitchell, W.A., 1994. Applying patch use toassess aspects of foraging behaviour in Nubian Ibex. J. Wildl.Manage. 58, 299–307.

Lanyon, J.M., Limpus, C., Marsh, H., 1989. Dugongs and turtles:grazers in the seagrass system. In: Larkum, A.W.D., McComb, A.J., Shepherd, S.A. (Eds.), Biology of Australian Seagrasses, AnAustralian Perspective. Elsevier, Amsterdam.

Lanyon, J.M., Johns, T., Sneath, H.L., 2005. Year-round presence ofdugongs in Pumicestone Passage, south-east Queensland, exam-ined in relation to water temperature and seagrass distribution.Wildl. Res. 32, 361–368.

Layne, J.N., 1965. Observations on marine mammals in Floridawaters. Bull. Fla. State Mus. 9, 166–168.

Ledder, D.A., 1986. Food Habits of the West Indian Manatee,Trichechus manatus latirostris, in South Florida. University ofMiami, Coral Gables, FL, p. 114.

Lee Long, W.J., Mellors, J.E., Coles, R.G., 1993. Seagrasses betweenCape York and Hervey Bay, Queensland, Australia. Mar. Freshw.Res. 44 (I), 19–31.

Legendre, P., 1993. Spatial autocorrelation: trouble or new paradigm?Ecology 63, 0548–0562.

Lipkin, Y., 1975. Food of the Red Sea dugong (Mammalia: Sirenia)from Sinai. Isr. J. Zool. 24, 81–98.

Mandelbrot, B., 1967. How long is the coast of Britain? Statistical self-similarity and fractional dimension. Science 156, 636–638.

Marsh, H., 2000. Evaluating management initiatives aimed at reducingthe mortality of dugongs in gill and mesh nets in the Great BarrierReef world heritage area. Mar. Mamm. Sci. 16, 684–694.

Marsh, H., Lawler, I.R., 2001. Dugong Distribution and Abundance inthe Southern Great Barrier Reef Marine Park and Hervey Bay:Results of an Aerial Survey in October–December 1999. GreatBarrier Reef Marine Park Authority, Townsville.

Marsh, H., Lawler, I.R., 2002. Dugong Distribution and Abundance inthe Northern Great Barrier Reef Marine Park. Great Barrier ReefMarine Park Authority/James Cook University, Townsville, p. 46.

Marsh, H., Rathbun, G.B., 1990. Development and application ofconventional and satellite radio-tracking techniques for studyingdugong movements and habitat usage. Aust. Wildl. Res. 17,83–100.

Marsh, H., Saalfeld, W.K., 1988. The distribution and abundance ofthe dugong in the Torres Strait region. Report to The AustralianFisheries Service, The Great Barrier Reef Marine ParkAuthority and the Fisheries Management Branch of theQueensland Department of Primary Industries, Townsville,Queensland. James Cook University, p. 24.

Marsh, H., Saalfeld, W.K., 1989. Distribution and abundance ofdugongs in the northern Great Barrier Reef marine park. Aust.Wildl. Res. 16, 429–440.

Marsh, H., Channels, P.W., Heinsohn, G.E., Morrissey, J., 1982.Analysis of stomach contents of dugongs from Queensland. Aust.Wildl. Res. 9, 55–67.

Marsh, H., Saafeld, W.K., Preen, A.R., 1990. The Distribution andAbundance of Dugongs in Southern Queensland Waters: Implica-tions for Management, Zoology Department. James CookUniversity. Queensland Department of Primary Industries,Townsville.

Marsh, H., Prince, R.I.T., Saalfeld, W.K., 1991. The Distribution andAbundance of the Dugong in Shark Bay and Exmouth Gulf-Ningaloo. Western Australian Department of Conservation andLand Management, pp. 1–16.

Marsh, H., Prince, R.I.T., Saalfeld, W.K., Shepherd, R., 1994. Thedistribution and abundance of the dugong in Shark Bay, WesternAustralia. Wildl. Res. 21, 149–161.

Marsh, H., Harris, A.N.M., Lawler, I.R., 1997. The sustainability ofthe indigenous dugong fishery in Torres Strait, Australia, PapuaNew Guinea. Conserv. Biol. 11, 1375–1386.

Marsh, H., Eros, C., Corkeron, P., Breen, B., 1999. A conservationstrategy for dugongs: implications of Australian research. Mar.Freshw. Res. 50, 979–990.

Marsh, H., Penrose, H., Eros, C., Hugues, J., 2002. Dugong: statusreport and action plans for countries and territories. In: Group, I.S.S.S. (Ed.), UNEP: Early Warning and Assessment Report Series,Townsville, p. 162.

Marsh, H., Lawler, I.R., Kwan, D., Delean, S., Pollock, K., Alldredge,M., 2004. Aerial surveys and the potential biological removaltechniques indicate that the Torres Strait dugong fishery isunsustainable. Anim. Conserv. 7, 435–443.

Marshall, C.D., Kubilis, P.S., Huth, G.D., Edmonds, V.M., Halin, D.L.,Reep, R.L., 2000. Food-handling ability and feeding-cycle lengthof manatees feeding on several species of aquatic plants. J.Mammal. 81, 649–658.

Marshall, C.D., Maeda, H., Iwata, M., Furuta, M., Asano, S., Rosas, F.,Reep, R.L., 2003. Orofacial morphology and feeding behaviour ofthe dugong, Amazonian, West African and Antillean manatees(Mammalia: Sirenia): functional morphology of the muscular-vibrissal complex. J. Zool. 259, 245–260.

Marzluff, J.M., Millspaugh, J.J., Hurvitz, P., Handcock, M.S., 2004.Relating resources to a probabilistic measure of space use: forestfragments and steller's jays. Ecology 85, 1411–1427.

McConnell, B.J., Chambers, C., Fedak, M.A., 1992. Foraging ecologyof southern elephant seals in relation to the bathymetry andproductivity of the Southern Ocean. Antarctic Sci. 4, 393–398.

McNaughton, S.J., 1984. Grazing lawns: animals in herds, plant form,and coevolution. Am. Nat. 124, 863–886.

Mignucci-Giannoni, A.A., Beck, C.A., 1998. The diet of the manatee(Trichechus manatus) in Puerto Rico. Mar. Mamm. Sci. 14,394–397.


Milne, B.T., 1991. Lessons from applying fractal models to landscapepatterns. In: Bissonette, J.A. (Ed.), Wildlife and LandscapeEcology. Effects of Pattern and Scale. Springer-Verlag, NewYork, pp. 32–69.

Morales, J.M., Ellner, S.P., 2002. Scaling up animal movements inheterogeneous landscapes: the importance of behaviour. Ecology83, 2240–2247.

Nams, V.O., 1996. The VFractal: a new estimator for fractal dimensionof animal movement paths. Landsc. Ecol. 11, 289–297.

Nams, V.O., 2004. Fractal analysis measures habitat use at differentspatial scales: an example with American marten. Can. J. Zool. 82,1738–1747.

Nams, V.O., 2005. Using animal movement paths to measure responseto spatial scale. Oecologia 143, 179–188.

Owen-Smith, R.N., 2002. Adaptive Herbivore Ecology: FromResources to Populations in Variable Environments. CambridgeUniversity Press, Cambridge. 374 pp.

Poiner, I.R., Peterken, C., 1996. Seagrasses. In: Zann, L.P., Kailola, P.(Eds.), The State of the Marine Environment Report for Australia.Technical Annex: 1. Great Barrier Reef Marine Park Authority,Townsville, pp. 40–45.

Preen, A.R., 1989. Dugongs: Volume 1. The Status and Conservationof Dugongs in the Arabian Region. Meteorological and Environ-mental Protection Administration.

Preen, A.R., 1993. Interactions between dugongs and seagrasses in asubtropical environment. PhD thesis, James Cook University,Townsville, Queensland.

Preen, A.R., 1995a. Impacts of dugong foraging on seagrass habitats:observational and experimental evidence for cultivation grazing.Mar. Ecol. Prog. Ser. 124, 201–213.

Preen, A.R., 1995b. Diet of dugongs—are they omnivores? J.Mammal. 76, 163–172.

Preen, A.R., 1999. Dugongs in the Shoalwater Bay area. Report to theRoyal Australian Navy, Townsville.

Preen, A.R., 2001a. Dugongs, Boats, Dolphins and Turtles in theTownsville-Cardwell Region and Recommendations for a BoatTraffic Management Plan for the Hinchinbrook Dugong ProtectedArea: vii 88. Great Barrier Reef Marine Park Authority,Townsville, Queensland, Australia.

Preen, A.R., 2001b. Dugongs, Boats, Dolphins and Turtles in theTownsville-Cardwell Region and Recommendations for a BoatTraffic Management Plan for the Hinchinbrook Dugong ProtectionArea, Department of TESAG. JCU. Great Barrier Reef MarinePark Authority, Townsville, Qld., pp. vii, 88.

Preen, A.R., 2004. Distribution, abundance and conservation status ofdugongs and dolphins in the southern and western Arabian Gulf.Biol. Conserv. 118, 205–218.

Preen, A.R., Marsh, H.D., 1995. Response of dugongs to large-scaleloss of seagrass from Hervey Bay, Queensland, Australia. Wildl.Res. 22, 507–519.

Preen, A.R., Lee Long, W.J., Coles, R.G., 1995. Flood and cyclonerelated loss, and partial recovery, of more than 1000 km2 ofseagrass in Hervey Bay, Queensland, Australia. Aquat. Bot. 52,3–17.

Reynolds, J.E.I., Wilcox, J.R., 1985. Abundance of West IndianManatees (Trichechus manatus) around selected Florida powerplants following winter cold fronts, 1982–1983. Bull. Mar. Sci. 36,413–422.

Reynolds, J.E.I., Wilcox, J.R., 1994. Observations of Florida manatees(Trichechus-manatus-latirostris) around selected power plants inwinter. Mar. Mamm. Sci. 10, 163–177.

Saarenmaa, H., Stone, N.D., Folse, L.J., Packard, J.M., Grant, W.E.,Makela, M.E., Coulson, R.N., 1988. An artificial intelligencemodelling approach to simulating animal/habitat interactions.Ecol. Model. 44, 125–141.

Sandgren, F., Sweanor, P.Y., 1988. Migration distance of moosepopulations in relation to river drainage length. Alces 24, 112–117.

Seaman, D.E., Powell, R.A., 1996. An evaluation of the accuracy ofkernel density estimators for home range analysis. Ecology 77,2075–2085.

Seaman, D.E., Millspaugh, J.J., Kernohan, B.J., Brundige, G.C.,Raedeke, K.J., Gitzen, R.A., 1999. Effects of sample size on kernelhome range estimates. J. Wildl. Manage. 63, 739–747.

Service Argos, 1996. User's Manual. CLS Argos.Short, F.T., Coles, R.G., Pergent-Martini, C., 2001. Global seagrass

distribution. In: Short, F.T., Coles, R.G. (Eds.), Global SeagrassResearch Methods. Elsevier Science B.V., Amsterdam, pp. 5–30.

Sinclair, A.R.E., 1983. The function of distance movements invertebrates. In: Swingland, I.R., Greenwood, P.J. (Eds.), TheEcology of Animal Movement. Clarendon Press, Oxford,pp. 240–258.

Skellam, J.G., 1973. The formulation and interpretation of mathemat-ical models of diffusionary processes in population biology. In:Bartlett, M.S.a.H., R.W. (Eds.), The Mathematical Theory of theDynamics of Biological Populations. Academic Press.

Stern, S.J., 2002. Migration and movement patterns. In: Perrin, W.F.,Wursing, B., Thewissen, J.G.M. (Eds.), Encyclopaedia of MarineMammals. Academic Press, San Diego, pp. 742–748.

Stewart, B.S., DeLong, R.L., 1995. Double migrations of the northernelephant seal. J. Mammal. 76, 196–205.

Turchin, P., 1996. Fractal analysis of animal movement: a critique.Ecology 77, 2086–2090.

Turchin, P., 1998. Quantitative Analysis of Movement. Sinauer Assoc,Inc, Sunderland, p. 396.

Vincent, C., McConnell, B.J., Ridoux, V., Fedak, M.A., 2002.Assessment of Argos location accuracy from satellite tagsdeployed on captive gray seals. Mar. Mamm. Sci. 18, 156–166.

Wake, J.A., 1975. A study of the habitat requirements and feedingbiology of the dugong Dugong dugon (Muller). B.Sc. Hons. thesis,Biological Sciences. James Cook University, Townsville, p. 119.

Wallis de Vries, M.F., 1996. Nutritional limitations of free-rangingcattle: the importance of habitat quality. J. Appl. Ecol. 33,688–702.

Watanabe, Y., Mitani, Y., Sato, K., Cameron, M.F., Naito, Y., 2003.Dive depths of Weddell seals in relation to vertical preydistribution as estimated by image data. Mar. Ecol. Prog. Ser.252, 283–288.

Weigle, B., Wright, I.E., Ross, M., Flamm, R., 2001. Movements ofRadio-Tagged Manatees in Tampa Bay and Along Florida's WestCoast 1991–1996. Florida Marine Research Institute TechnicalReports TR-7. St. Petersburg, Florida.

Welsby, T., 1967. Schnappering. In: Thomson, A.K. (Ed.), TheCollected Works of Thomas Welsby. Jacaranda Press, Brisbane,pp. 104–105.

Whiting, S.D., Thorn, R., Murray, W., 2005. A lone dugong on Cocos(Keeling) Islands, Indian Ocean. Sirenews 44, 5–6.

Wiens, J.A., 1989. Spatial scaling in ecology. Funct. Ecol. 3, 385–397.With, K.A., 1994. Using fractal analysis to assess how species perceive

landscape structure. Landsc. Ecol. 9, 25–36.Worton, B.J., 1989. Kernel methods for estimating the utilization

distribution in home-range studies. Ecology 70, 164–168.

Movement heterogeneity of dugongs, Dugong dugon (Müller...

Documents

Transcript of Movement heterogeneity of dugongs, Dugong dugon (Müller...