Investigation of local movement and regional

2006/809

July 2010

Investigation of local movement and regional migration behaviour of broadbill swordfish targeted by the Eastern Tuna and Billfish Fishery

Karen Evans www.afma.gov.au

Protecting our fishing future

Box 7051, Canberra Business Centre, ACT 2610 Tel (02) 6272 5029 Fax (02) 6272 5175

AFMA Direct 1300 723 621

Copyright and Disclaimer Published by CSIRO Marine and Atmospheric Research © 2010 Commonwealth Scientific and Industrial Research Organisation (CSIRO) and the Australian Fisheries Management Authority. To the extent permitted by law, all rights are reserved and no part of this publication covered by copyright may be reproduced or copied in any form or by any means except with the written permission of the copyright owners. The information contained in this publication comprises general statements based on scientific research. The reader is advised and needs to be aware that such information may be incomplete or unable to be used in any specific situation. No reliance or actions must therefore be made on that information without seeking prior expert professional, scientific and technical advice. To the extent permitted by law, CSIRO (including its employees and consultants) excludes all liability to any person for any consequences, including but not limited to all losses, damages, costs, expenses and any other compensation, arising directly or indirectly from using this publication (in part or in whole) and any information or material contained in it. The use of this report is subject to the terms on which it was prepared by CSIRO. In particular the report may only be used for the following purposes. This report may be copied for distribution within the Client’s organisation; The information in this report may be used by the entity for which it was prepared (the ‘Client’), or by the Client’s contractors or agents, for the Clients internal business operations (but not licensing to third parties); Extracts of the report distributed for these purposes must clearly note that the extract is part of a larger report prepared by CSIRO for the Client. The report must not be used as a means of endorsement without the prior written consent of CSIRO. The name, trademark or logo of CSIRO must not be used without the prior written consent of CSIRO.

National Library of Australia Cataloguing-in-Publication entry: Evans, Karen. Investigation of local movement and regional migration behaviour of broadbill swordfish targeted by the eastern tuna and billfish fishery / Karen Evans ISBN: 9781921605840 (pbk.) Swordfish--Monitoring--Australia. Swordfish--Migration—Australia. Fish communities--Monitoring--Australia. Australian Fisheries Management Authority. CSIRO. Marine and Atmospheric Research. 597.780994

Front cover design: Louise Bell, CSIRO Marine & Atmospheric Research.

TABLE OF CONTENTS

1.

NON-TECHNICAL SUMMARY.................................................................. 1

2.

ACKNOWLEDGEMENTS ......................................................................... 3

3.

BACKGROUND......................................................................................... 4

4.

NEED......................................................................................................... 7

5.

OBJECTIVES ............................................................................................ 9

6.

METHODS............................................................................................... 10 6.1. Tags and tagging operations.................................................................... 10 6.2. Data and analyses ................................................................................... 12 6.2.1.

6.2.2. 6.2.3.

6.2.4.

7.

Data collection ......................................................................................... 12 6.2.1.1. PAT4 and Mk10 PSATs......................................................... 12 6.2.1.2. Mk10 PAT-Fs......................................................................... 13 Potential mortalities ................................................................................. 13 Location estimates .................................................................................. 14 6.2.3.1. PAT4 and Mk10 PSATs......................................................... 14 6.2.3.2. Mk10 PAT-Fs......................................................................... 14 Habitat preferences and behaviour ......................................................... 14

RESULTS ................................................................................................ 15 7.1. Tag attachment........................................................................................ 15 7.2. Potential mortalities.................................................................................. 21 7.3. Location estimation .................................................................................. 21 7.3.2.

Mk10 PAT-Fs .......................................................................................... 21

7.4. Migration and movement patterns............................................................ 23 7.5. Habitat and behaviour .............................................................................. 28

8.

DISCUSSION .......................................................................................... 29 8.1. Tag attachment........................................................................................ 29 8.2. Location estimation .................................................................................. 31 8.3. Migration and movement patterns............................................................ 36 8.4. Habitat ..................................................................................................... 41

9.

BENEFITS/MANAGEMENT OUTCOMES .............................................. 45

10. CONCLUSIONS ...................................................................................... 46 11. REFERENCES ........................................................................................ 47 12. APPENDICES ......................................................................................... 56

i

Movement of swordfish targeted in the ETBF

1.

NON-TECHNICAL SUMMARY

Pop-up Satellite Archival Tags (PSATs) were deployed on 47 large (46 – 240 kg estimated weight) broadbill swordfish (Xiphias gladius) in the western Tasman Sea across 2006 – 2009. An additional 10 prototype towed body PSATs with FastlocTM GPS capabilities were also deployed on swordfish throughout the same region. Attachment durations of PSATs on swordfish ranged 2 – 364 days and data were received from 48 of the 51 PSATs deployed. Tag attachment durations increased throughout the period of the study reflecting the increasing experience of the program. The ability to determine movements and habitat preferences of swordfish across substantive time scales using PSATs is restricted by the ability to achieve long term attachments and the ability to obtain substantive light data from which geo-positions can be determined. The strict diurnal diving behaviour of this species results in little light data being collected by tags and, as a consequence, calculated positions are few and far between or in some cases not able to be determined at all. This reduces the scale at which movement and habitat interaction can be inferred for this species. The pop-up archival tag incorporating Fastloc

TM

GPS capabilities developed as part of this

project served to increase the number of locations able to be estimated from each tag deployment and also provided for the addition of high resolution location estimates to those derived from light data. Integration of FastlocTM GPS location data with locations derived from light data in a movement model will allow for the incorporation of a number of ‘corrections’ along the light-based estimated track. This will ultimately result in a more realistic estimation of movement and a reduction in the error associated with position estimates. This will have flow-on effects for the determination of habitats of importance, migratory corridors and the responses of individuals to spatial environmental variability as they will be able to be estimated at finer scales than previously possible for this species. Although a number of swordfish undertook extensive movements of up to 1,905.39km, all remained within the Coral/Tasman Sea region, exhibiting restricted east-west movement. The degree of residency demonstrated by swordfish in the Coral/Tasman Seas appears to be unique to this region. There was little synchrony in the direction of movement undertaken by individuals with both movements north and south occurring throughout the winter months. Fish undertaking the largest movements were slightly

1


smaller than those demonstrating restricted or ‘resident’ movement. This may be associated with differences in the degree of movement between gender groups and/or age classes tagged and could also be associated with the degree of spatial separation of foraging grounds to spawning grounds. If regions close to areas where spawning occurs are productive and likely to sustain a population of swordfish, there may be no need for swordfish to undertake large movements away from those areas. Rather than undertaking extensive migrations between foraging and spawning areas, the Tasman Sea and particularly those waters associated with the Tasman Front and seamount regions may provide sufficient forage for swordfish in the Australian region. This would thereby reduce the requirement for individuals to undertake latitudinal movements on similar spatial scales to those observed elsewhere. Individuals demonstrated distinct diurnal diving behaviour and were largely associated with waters which had sea surface temperature of greater than 20°C. Individuals spent the majority of time during the day at depths of 600 – 800m and in waters of 7 – 15°C, ascending at night to depths of less than 200m and 20 – 25°C. Vertical excursions occurred over relatively short time scales (3 – 4 hours) and resulted in water temperature changes of 15 – 20°C. Such behaviour ma y be associated with the vertical migration of prey either directly or via swordfish tracking particular light levels most suitable for ambushing or detecting prey. Swordfish in the Tasman Sea clearly demonstrate an ability to exploit a wide range of habitats reflecting a physiological plasticity which allows for a maximising of resource exploitation in a patchy environment. Seasonal shifts in the depth and water temperature preferences of swordfish were evident and appeared to be closely related to seasonal shifts in the thermal structure of their oceanic environment. Comparisons with those habitat and behaviour data collected from swordfish in other parts of the Pacific Ocean suggest some broad-scale regional differences, possibly associated with regional differences in the thermal structure and the distribution dissolved oxygen concentrations of the oceanic environment. Despite the limitations of the technology used to collect these data, determining such aspects of the life history of this species independent of the fishery would be difficult without the use of PSATs. The data presented in this study represent a major step towards reducing uncertainty about the spatial dynamics of swordfish in the western Pacific region and has already been used to revise the spatial structure of the swordfish stock assessment for the Western Central Pacific Fisheries Commission

2


(WCPFC) region. On the basis of the data collected under this study, collaborations have been established with the Ministry of Fisheries in New Zealand, the Instituto Español de Oceanografica, the Cook Islands Ministry of Fisheries and the Service de la Pêche in French Polynesia. These collaborations have facilitated the deployment of a substantive number of pop-up satellite archival tags on swordfish throughout the WCPFC region and once all data is on hand will allow a regional assessment of swordfish movement and connectivity throughout the WCPFC management area. This will provide further inputs into the stock assessment for this swordfish throughout this area. The evidence of high levels of retainment of swordfish in the Tasman/Coral Sea region supports fishery data for the region and observed serialised depletion in localised stocks not only in the Eastern Tuna and Billfish Fishery but in other areas of the western/central Pacific Ocean. The growing evidence for restricted movement of Pacific Ocean tunas and billfish, particularly in sub-equatorial regions, raises the issue for the Western and Central Pacific Fisheries Commission of how best to manage across the broad range of these fisheries for the sustainability of regional/localised populations. Localised depletions have the potential to significant affect the viability of island and coastal state fisheries, yet fishery-wide controls of effort or catch will be unlikely to prevent localised depletions of these stocks.

KEYWORDS: broadbill swordfish, Xiphias gladius, pop-up satellite archival tag, spatial dynamics, habitat preferences.

2.

ACKNOWLEDGEMENTS

This work was supported by funding from the Australian Fisheries Management Authority. The authors would like to acknowledge the support of Aaron Mattner and Teuira Monikura and the crew of Fortuna II, Dean Reyland and the crew of Seeker, Pavo Walker and the crew of Assassin, Anthony Wilkinson and the crew of Annandale and Martin Wright and the crew of Kendon B. Gary Heilmann and Mike Madden of Mooloolaba Fisheries Investments Pty Ltd and Steve Hall of AFMA provided

3


considerable logistical support in organising tagging operations. Nathan Bicknell (AFMA), Chris Blood (CSIRO), Thor Carter (CSIRO), Scott Cooper (CSIRO), Sam Graham-Taylor (AFMA), Matthew Horsham (CSIRO), David Kube (CSIRO), Mark Rayner (CSIRO), David Webb (AFMA), Grant West (CSIRO) and particularly Matt Lansdell (CSIRO) are thanked for their assistance on the project both in the laboratory and at sea. Heather Bauer, Melinda Braun, Ed Bryant and Ted Rupley and are thanked for their contributions to tag development and data interpretation. Jason Hartog and Toby Patterson are thanked for their contributions to data management, movement modeling and habitat analyses. Considerable support for this project has been provided by Steve Auld, Brad Milic and Wez Norris from the Australian Fisheries Management Authority. Tagging operations were carried out under DPIW AEC permit 08/2007-08 and under scientific permits 901083, 901116, 901186, 901198, 1000075, 1001072 and 1001144, issued by the Australian Fisheries Management Authority.

3.

BACKGROUND

Broadbill swordfish (Xiphias gladius; hereafter named swordfish) have a widespread geographical distribution throughout the world’s temperate, subtropical and tropical regions. The distribution of individuals has been observed to vary seasonal scales and has been associated with the seasonal extension and retraction of warmer waters into higher latitudes and variability in prey distributions (Palko et al. 1981). Latitudinal distribution appears to be reflective of the sexual dimorphism demonstrated by this species and the size of individuals with fewer, smaller males occurring in colder, higher latitudes than larger females (Palko et al. 1981). Swordfish regularly make vertical movements between the ocean surface and depths of up to 1000m on a diurnal basis, following the distribution of the deep sound scattering layer presumably in search of food (Carey 1992) which predominantly consists of a mix of cephalopods and fish (Scott and Tibo 1986; Hernandez-Garcia 1995; Chancollon et al. 2006; Young et al. 2006). Although, unlike tunas, swordfish are unable to maintain muscle temperatures above ambient water temperatures, a specialised tissue and associated heat exchanger located near the eyes allows swordfish to spend considerable amounts of time at these depths. The tissue and heat exchanger heats the brain and eyes, allowing central nervous system activity to be maintained at low temperatures and

4


increasing temporal resolution in eyesight in the low light levels experienced at depth (Carey 1982; Fritsches et al. 2005). Spawning in general is largely restricted to tropical waters and is associated with surface water temperatures of greater than 24°C in the eastern Australian region (Young et al. 2003). In waters close to the equator, spawning occurs throughout the year, becoming more seasonally defined at higher latitudes (Palko et al. 1981; Young et al. 2003). In the east Australian region swordfish at latitudes lower than 25°S demonstrate a protracted spawning season with spawning occurring between September and March with a peak in December to March (Young et al. 2003). Spawning across the spring and summer months has been similarly been observed in swordfish in other temperate regions, although spawning periods appear to vary to some extent spatially with spawning occurring earlier in the season in some regions and later in others (De Martini et al. 2000; Arocha 2007; Poisson and Fauvel 2009). Males reach sexual maturity at smaller sizes than females, with the L50 of males reported to occur at ~85cm orbit to fork length (OFL; western Pacific Ocean, Young et al. 2003), ~105cm OFL (eastern Pacific Ocean; De Martini et al. 2000) and 113cm OFL (north-west Atlantic Ocean; Arocha and Lee 1996). The L50 of females is reported to occur at ~145cm OFL (eastern Pacific Ocean; De Martini et al. 2000), 159cm OFL (north-west Atlantic Ocean; Arocha and Lee 1996) and ~195cm OFL (western Pacific Ocean, Young et al. 2003). Although there appears to be some spatial structure to the size at which sexual maturity is attained, this is confounded by varying methodologies and definitions of sexual maturity (Young and Drake 2002). Estimates of growth rate suggest that males reach their maximum size at nine years, while females reach their maximum size at 15 years. Individuals are thought to rarely attain ages of 25 years (Ward et al. 2000). A recent study of swordfish in the western Pacific region observed a maximum age of 18 years in females and 15 years in males (Young and Drake 2004). Investigations of catch data and genetic structure suggest there is some population structure to swordfish stocks across the Pacific, Indian and Atlantic Oceans (Reeb et al. 2000; Alvarado-Bremer et al. 2005). In the Pacific Ocean, gene flow appears to have a ⊃-shaped pattern, suggesting movement of animals east-west in the northern and southern hemispheres, with connections across the equatorial zone along the west coast of the Americas. This pattern of gene flow coincides with known spawning locations based on Japanese larval surveys (Nishikawa et al. 1985) and is in line with

5


the hypothesis that there are separate stocks for the north Pacific, eastern Pacific and south-west Pacific proposed by Sakagawa and Bell (1980) based on catch and larval distribution data. These studies all suggest that isolated foraging areas are connected via shared spawning locations, and that foraging areas may be sites of admixture between populations which don’t share spawning areas, similar to that observed in the Atlantic (Alvarado-Bremer et al. 2005). More than half the world’s catches of swordfish are derived from bycatch by distantwater longline vessels targeting tuna (Ward et al. 2000). In the eastern Australian region, swordfish have comprised a component of longline catches since the 1950s when they were initially caught incidentally by Japanese longline vessels targeting tuna. Swordfish continued to be caught incidentally by longline vessels as the domestic fishery for tuna in the area (the Eastern Tuna and Billfish Fishery or ETBF) grew and access by Japanese vessels to the fishery became progressively more restricted. In the mid 1990s improved access to markets in the US resulted in many domestic longline vessels relocating to south-east Queensland to target swordfish specifically. Landings of swordfish grew from less than 50 tonnes per year to 2,373 tonnes in 1998 (Ward et al. 2000), with nominal catch per unit effort peaking in 1997 (Sands et al. 2009). Catches within the ETBF have been largely associated with three oceanographic/geographic regions: (i) inshore activities along the continental shelf associated with oceanographic fronts and eddies of the East Australian Current (EAC); (ii) inshore activities associated with seamounts and (iii) offshore activities associated with oceanographic fronts and seamounts (Ward et al 2000). Catches of swordfish in the ETBF have declined progressively since the late 1990s, with these declines estimated to be in the order of 50 to 70% over the last decade (Campbell 2000). Analyses of swordfish catches throughout the ETBF demonstrate a serial pattern of depletions, with catch declines progressively moving from near shore fishing grounds to those further offshore (Campbell and Hobday 2003). While far offshore areas currently have higher catch rates than those nearer shore, catches offshore are demonstrating a similar trajectory of depletion.

6


4.

NEED

There are a number of fundamental uncertainties in current stock assessment models utilised on swordfish in the western Pacific region: 1) swordfish are potentially highly migratory, widespread across the south-west Pacific with little known about their migration timing or patterns; 2) the serial decline of in-shore catch rates off eastern Australia suggests swordfish have site affinities or variable residence times in different habitat features, and that these aggregations are important in swordfish dynamics; 3) swordfish are long-lived with age-dependent behavioural characteristics, but there is limited data on behaviour by age, and 4) swordfish exhibit numerous sex-specific characteristics, but there is little sex-specific data collected. In the absence of empirical data these uncertainties are being investigated using simulated data – which at least allows for the quantification of bias and loss of precision that would result from the possible alternatives for each of the factors. However, if management advice is to be robust, it is essential that these uncertainties be resolved to the extent feasible. This will be particularly important if the emphasis in the ETBF moves to considering spatial management and/or if there is an attempt to integrate management at the regional scale as the Western and Central Pacific Fisheries Commission (WCPFC) becomes operationally involved in regional fisheries management. Analyses of the decline in catch rates and anecdotal evidence from fishermen suggest that swordfish are in higher abundance near seamounts than in other parts of the fishery. It is thought that at least a portion of the population follows an annual migratory pattern, moving from spawning grounds in the Coral Sea, south through the fishing grounds off south east Queensland, and then south and east toward New Zealand, eventually returning to the spawning grounds off Australia’s northern east coast. In the context of this scenario, higher densities of swordfish near eastern Australian seamounts may be the result of directed movement or attraction of individuals to the seamounts or a retention of individuals at seamounts once encountered. If swordfish aggregate, a fixed catch strategy, even if instituted via effort controls, could result in an increasing fraction of the total stock being taken annually – thus accelerating the decline in the stock. Spatial closures may provide a solution to this problem, particularly if they are placed in areas where swordfish aggregate. Assuming swordfish are uniformly distributed

7


throughout the fishery, a spatial closure would ensure that a constant fraction of the stock would be protected from fishing. This effectively sets a proportional harvest strategy in place, which in general has been shown to have less risk of causing collapses and/or extinctions, than fixed quotas (Milner-Gulland 2001). If swordfish aggregate in identifiable areas, a management strategy of spatial closures in those areas may be even more conservative, depending on the factors driving the aggregative behaviour. If the aggregation behaviour is driven by longer retention times at the seamounts, closures of seamounts would protect a larger fixed fraction of the population for the same closure area. However, if interactions, such as interference competition during feeding, result in spacing behaviour, then it would be expected that an increasing proportion of the total population would be present at a seamount as the total population size declined. In this case spatial closures would provide a stronger protection for the stock, particularly if unforeseen declines, for example due to fishing pressure outside the Australian Fishing Zone (AFZ) or climatic shifts occur. Local movement dynamics and migration patterns will play a fundamental role in determining the effectiveness of management strategies, whether they utilise catch limits or spatial closures. Spatial closures are potentially attractive as a management tool; however, they require significant knowledge about the movement behaviour of the managed species. The size and location of the closures needs to be considered in the context of the migration dynamics (i.e. if migration rates are very high and unpredictable, small spatial closures could be of limited use, as few individuals would be rendered unavailable to the fishery for much of the time). Several fundamental questions must be answered to determine their effectiveness relative to other options: 1. Does the species aggregate in particular parts of the fishery? 2. Are aggregation areas stable spatially or do they move over time? 3. Is there spacing behaviour involved in the aggregations, or are aggregations driven by higher local residence times relative to other areas? These questions form a set of testable hypotheses that can be investigated in a relatively straightforward manner using a planned tagging study with electronic tags. In addition to the direct benefits to management of understanding movement behaviour, there are significant flow-on benefits. For instance, one of the primary weaknesses in the current stock assessment model for the swordfish is the lack of information on the

8


spatial structure of the stock and movement dynamics of the fish. Because of this shortcoming, it is difficult to determine the appropriate level of subdivision to include in the model, which can result in seriously erroneous estimates of population dynamics, extinction risk, and other important metrics (Cooper and Mangel 1999).

5.

OBJECTIVES •

Test alternative hypotheses regarding swordfish movement and retention times with distance offshore and proximity to major benthic features, such as seamounts

•

Develop an understanding of swordfish migratory behaviour in the western Pacific Ocean, Tasman and Coral Seas

•

Create a dataset that will inform spatial management of the Eastern Tuna and Billfish fishery, particular the targeting of closures to remediate depletions specifically associated with the Mooloolaba grounds

•

Provide information that can be used to clarify the appropriate structure for swordfish stock assessment models under development for the WCPFC management region

•

Initiate development of an Australian-based database on swordfish behaviour and movement that can be utilized as a basis for international collaboration on swordfish dynamics and fisheries issues

•

Enable the opportunistic collection of data on the migration patterns and habitat use of other key apex predators within the pelagic ecosystem of the eastern AFZ within a experimental design framework

9


6.

METHODS

6.1. Tags and tagging operations Pop-up satellite archival tags (PAT4: n = 24, Mk10: n = 23, Wildlife Computers, Redmond USA) were deployed on swordfish in the waters of the northern Tasman Sea/southern Coral Sea across the months of November to April in 2006, 2007, 2008 and 2009 (n = 47). Swordfish were caught as part of commercial longline operations with those considered in good condition and of an adequate size and weight (>150cm OFL and >50kg wet weight) lead alongside the vessel to a position near the sea door. The anchor of each tag was inserted into the dorsal musculature of the fish in a position just ventral to the primary dorsal fin using a custom built tagging pole similar to that described in Chaprales et al. (1998). Once tagged, a biopsy sample was also collected from each individual using a custom built biopsy pole similar in design to the tagging pole with a stainless steel biopsy punch screwed into the tip. As soon as each tag was deployed and the associated biopsy sample collected, the fish was either cut from the line or the hook removed and then allowed to swim away from the vessel. A custom made stainless steel floy-type anchor was used as a primary anchor for each tag. The primary monofilament leader on all tags was fitted with a depth release device (RD-1800, Wildlife Computers, Redmond USA), designed to cut the tag off the fish at a depth of 1800m, thereby preventing implosion of the tag at depth. Each tag was printed with an identification number, information about a reward offered and where to return the tag. The deployment position of all tag releases was recorded using the vessels’ onboard GPS system. Initial results from deployments revealed that the diving behaviour of swordfish resulted in the calculation of geo-position from light data being problematic. Diving behaviour was observed to be almost strictly diurnal in nature, with the majority of time during the day spent by individuals at depths of approximately 600m (see Results for further details). Such a large proportion of time spent at depth during the day resulted in little light data being collected by PSATs deployed and as a result, calculated

10


positions were few and far between (see Results for further details). This thereby reduced the scale at which movement and habitat interaction could be inferred. However, swordfish were observed to spend a significant amount of time at the oceans surface at night and in other areas of the Pacific Ocean are known to demonstrate time at the surface during the day known as ‘basking’ behaviour (Ward et al. 2000). The ability to spend extended periods of time at the oceans surface and the possibility of basking behaviour prompted initial discussions with electronic tag manufacturers about the viability of alternative technologies to be utilised to determine movement in swordfish. The availability of Fastloc

TM

global positioning system (GPS) technology in

current generation transmitting archival tags prompted CSIRO and the electronic tag manufacturer Wildlife Computers to initiate a collaboration focused on developing a similar tag able to be externally attached to an animal and with pop-up capabilities. In order to determine the likelihood of complete surfacing of towed body tags on swordfish we modified two Mk10 PSATs (Wildlife Computers, Redmond) so that they transmitted signals to Service Argos satellites opportunistically once clear of the ocean surface. The two tags were programmed for 60 day deployments and deployed on swordfish in the western Tasman Sea in January 2007. One tag prematurely detached after five days, the other detached on time after the full 60 days. During the 60 day deployment, transmission attempts were received by Service Argos on six occasions, demonstrating that the swordfish was swimming close enough to the surface for the tag to clear the ocean surface. In response to the successful trial of these deployments, further tests were conducted on current generation transmitting archival tags incorporating FastlocTM GPS technology (Mk10-AF, Wildlife Computers, Redmond USA) to determine the capability of the GPS antenna to successfully acquire GPS constellation signals under a range of degrees of tilt (0-40°). After determining that tilt throughout the range of 0-40° had negligible effect s on the ability of the GPS antenna to acquire GPS constellation signals, the development of a prototype design for such a tag went ahead. The design essentially took the components of a Mk10-AF and transferred them into a configuration which had the circuitry in front of the GPS antenna (rather than underneath) and a AA battery hung below the GPS antenna. The components were then encased into a float comprised of beaded epoxy and fashioned into a streamlined shape (Figure 1). Ten prototype towed body PSATs with FastlocTM GPS capabilities were deployed in the same region as the PAT4s and Mk10s across the months February to December 2008.

11


Figure 1. Prototype towed body Mk10 PAT-Fs. The first iteration of the prototype is in the foreground with the finalised prototype in the background.

6.2. Data and analyses 6.2.1. Data collection 6.2.1.1. PAT4 and Mk10 PSATs PAT4 PSATs were programmed to record pressure (depth), temperature and light at 60 second intervals whilst Mk10 PSATs were programmed to record pressure, temperature and light at 10 second intervals. Both versions of PSATs were programmed to release from the fish after predetermined time periods (60 days: Mk10 n = 2; 90 days: PAT4 n = 9; 180 days: PAT4 n = 5; Mk10 n = 7; 274 days: PAT4 n = 5; Mk10 n = 8; 365 days: PAT4 n = 5 Mk10 n = 6), after which they floated to the ocean surface and transmitted their archived data via the ARGOS satellite service (Service Argos, Toulouse, France). Due to limited transmission bandwidth, data collected by the PSATs were summarised into one (PAT4: n = 9), two (PAT4: n = 5; Mk10: n =7), four (PAT4 n = 5; Mk10 n = 8), six (Mk10 n = 2) and twelve (PAT4 n = 5; Mk10 n = 6) hour

12


time periods prior to transmission. The summary data for each time period consisted of distributions of the proportion of time spent within preset depth and temperature bins and temperature-depth profiles. For those PSATs recovered (n = 7) the full archived data set was downloaded from the tag. Only those data collected from tags at liberty greater than seven days were included in analyses to avoid possible behavioural changes imposed from the process of tagging.

6.2.1.2. Mk10 PAT-Fs Tags were programmed initially to collect GPS position data only and did not incorporate a light sensor in the design (n = 4). Later versions were programmed to collect depth (pressure) and temperature data in addition to GPS position data and incorporated a light sensor in the design (n = 6). All tags were programmed to collect depth and temperature at 10 second intervals with later versions collecting light also at 10 second intervals. All tags were programmed to sample GPS constellation signals at two minute intervals. Data were summarised into 1 h time periods prior to transmission via Service Argos. All tags were programmed to release from tagged fish after 60 days. Similarly to the PAT4s and Mk10 PSATs, the summary data for each time period consisted of distributions of the proportion of time spent within preset depth and temperature bins and temperature-depth profiles. After release, each tag floated to the ocean surface and transmitted their archived data via the ARGOS satellite service (Service Argos, Toulouse, France).

6.2.2. Potential mortalities Wildlife Computers PSATs transmit diagnostic information from which attachment failures and mortalities can be inferred. This information includes the identification of whether the corrodible tether pin has broken, the depth at which the premature release status was activated and the last recorded depth greater than 600m and when this was recorded. In circumstances on the continental shelf, a mortality will be indicated by the tag maintaining a constant shallow bottom depth (as indicated by the time at depth data and the depth at which premature release was activated) over an extended period of up to 48 hours followed by release of the tag, indicating that the tag and animal were on the sea floor. In pelagic waters, mortalities can be identified when there is gradual sinking of the tag (as indicated by the time at depth data) to depths beyond what would be expected to be the maximum dive depth of the species (as indicated by

13


the last recorded depth greater than 600m) and to the point that the depth cut-off device is activated. While mortalities in swordfish over the continental shelf could be identified using the criteria above, given the deep diving capabilities of swordfish in waters off the continental shelf, we could not necessarily associate very deep final depths (greater than 1000m) in close proximity to the transmission of data with mortalities.

6.2.3. Location estimates 6.2.3.1. PAT4 and Mk10 PSATs Estimates of location were produced using the TrackIt package implemented within the R statistical environment (R Foundation for Statistical Computing, 2006) from light data recorded by the tags. This package estimates a most probable track of geographic positions via the state space model described in Nielsen and Sibert (2007) and provides robust estimates of uncertainty associated with the positions produced. The model attempts to estimate two positions for each day, one at dawn and one at dusk directly from the time series of light data recorded by the tag.

6.2.3.2. Mk10 PAT-Fs GPS locations were calculated from acquisitions recorded by the tag utilising proprietary software (Fast-GPS Solver version 1.0.56, Wildlife Computers, Redmond USA). For those Mk10 PAT-Fs that collected complimentary light data, light-based position estimates were estimated as per the PAT4s and Mk10PATs.

6.2.4. Habitat preferences and behaviour Time-At-Depth (TAD) and Time-at-Temperature (TAT) data transmitted consisted of normalized histograms giving the proportion of time within the pre-programmed summary period that the fish spent within a given depth or temperature range. Aggregate time-integrated indices of the temperature and depth preference were calculated by calculating the average TAD and TAT value in each histogram bin and the proportion of time in a given depth or temperature range examined. PAT-style Depth Temperature (PDT) profiles were investigated for spatio-temporal variability in oceanographic conditions encountered by individuals.

14


7.

RESULTS

A total of 47 PSATs and 10 prototype towed body GPS PSATs were deployed on swordfish estimated to weigh 45 – 240 kg gilled and gutted wet weight (mean±SD: 105.35 ± 47.63 kg; Table 1, Figure 2). Of the 57 PSATs deployed, data were received from 48 (84.2%) via Service Argos. A further one tag which failed to transmit was recovered, providing the full archival records from this tag. A total of seven (12.28%) tags were recovered and the full archival record was retrieved from five of these tags. Data derived from deployments longer than seven days were retrieved from 39 PSATs.

7.1.

Tag attachment

Of the 49 tags from which data were received, 16 achieved the full deployment period set. The number of days that tags remained attached varied considerably both within and between deployment years and in relation to expected deployment length (Table 2). Deployment periods ranged 2 – 364 days and the proportion of the original deployment period achieved ranged 0.01 – 1.00. The proportion of time tags remained attached in relation to the period set steadily increased across the four years of tagging from 0.29 ± 0.28 in 2006 to a high of 0.65 ± 0.41 in 2009, reflecting the increasing experience of the program (Figure 3). The overall proportion of time that tags remained attached across the four years was 0.56 ± 0.41. The proportion of the set period achieved demonstrated a negative trend as the period set increased with tags set at 365 days only remaining attached for an average period of 40.7% of their total set attachment period in comparison to an average of 71.2 % for tags set at 60 days (Figure 3). Archival records from recovered PSATs recorded two tags detaching from swordfish after the depth-cut off device was activated as a consequence of deep diving behaviour in these individuals. Summary depth records suggest that deep diving behaviour set off depth cut off devices prematurely releasing a further ten tags from swordfish.

15


Table 1. Release and pop-up details for PSATs deployed on broadbill swordfish in the Tasman Sea 2006–2009. TAL: time at liberty; G&G: gilled and gutted; *Prototype Mk10-PATF; ^Tag recovered Tag

2006 04P0336 04P0574 04P0443 04P0572 04P0550 04P0575 04P0405 04P0577 04P0578 04P0576 2007 06A0718 06A0719 04P0588 04P0617^ 04P0563^ 04P0474 04P0472 04P0564 04P0462 04P0471 04P0473 04P0429 04P0453 04P0468 2008 04P0338 04P0610

Releases

Pop-up transmissions

Date

Latitude (°S)

Longitude (°E)

Est. wt (G&G in kg)

20 Mar 2006 07 Oct 2006 08 Oct 2006 09 Oct 2006 30 Oct 2006 31 Oct 2006 03 Nov 2006 03 Nov 2006 03 Nov 2006 28 Nov 2006

26.22 25.88 26.11 26.28 27.28 27.20 26.05 28.09 28.07 24.93

159.43 156.96 157.02 157.02 156.90 156.65 156.77 156.84 156.81 156.49

45 50 50 60 60 45 60 100 90 55

Failed to transmit 06 Mar 2007 24.40 06 Aug 2007 29.12 06 Nov 2006 21.21 08 Nov 2006 23.76 02 Nov 2006 26.50 Failed to transmit 17 Mar 2007 29.56 16 Dec 2006 16.87 24 Feb 2007 27.84

28 Jan 2007 29 Jan 2007 30 Jan 2007 31 Jan 2007 1 Feb 2007 7 Feb 2007 4 Mar 2007 4 Mar 2007 10 Mar 2007 28 Nov2007 28 Nov 2007 29 Nov 2007 15 Dec 2007 15 Dec 2007

25.61 25.21 24.75 25.51 25.21 24.84 25.46 25.37 24.52 27.17 27.30 27.50 27.21 27.20

157.22 157.58 157.84 157.68 157.61 156.34 157.30 157.28 156.49 157.29 157.42 157.88 157.90 157.85

80 50 50 70 70 90 70 80 55 120 140 100 200 160

19 Jan 2008 20 Jan 2008

28.37 27.80

157.38 156.07

150 150

Date

Latitude (°S)

Longitude (°E)

TAL ( days)

161.61 166.44 155.78 156.33 156.33

150 302 28 9 2

155.99 153.92 158.87

134 43 88

28 Mar 2007 14.30 03 Feb 2007 23.75 7 Jun 2007 38.25 Failed to transmit 16 Feb 2007 24.53 07 Aug 2007 32.01 01 Jun 2007 22.80 30 Aug 2007 23.42 Failed to transmit 30 Nov 2007 27.41 25 Feb 2008 30.41 Failed to transmit 28 Dec 2007 26.13 11 Jan 2008 24.50

152.07 157.52 170.24

59 5 128

157.77 157.29 156.53 157.06

15 181 89 179

157.87 153.61

2 89

154.62 158.80

13 27

17 Apr 2008 31.42 Failed to transmit

156.13

89

16


Tag

06A1162^ 07A0889* 06A1161 06A1159 06A1164 06A1165 06A1166^ 07A0890* 06A1140 06A1160 07A0894* 06A1130 06A1133 06A1137 06A1157 06A1163^ 06A1151^ 08A0103* 08A0098* 07A0893* 08A0071*^ 08A0101* 08A0096* 08A0100* 06A1132 06A1138 2009 06A1167 06A1156 06A1134 06A1135 06A1139

Releases

Pop-up transmissions

Date 25 Feb 2008 28 Feb 2008 20 Mar 2008 21 Mar 2008 21 Mar 2008 21 Mar 2008 22 Mar 2008 23 Mar 2008 24 Mar 2008 24 Mar 2008 25 Mar 2008 26 Mar 2008 22 Apr 2008 22 Apr 2008 23 Apr 2008 23 Apr 2008 24 Apr 2008 25 May 2008 19 Jun 2008 6 Nov 2008 8 Nov 2008 12 Dec 2008 13 Dec 2008 15 Dec 2008 16 Dec 2008 16 Dec 2008

Latitude (°S) 29.36 29.36 25.68 25.98 25.70 25.95 25.97 25.98 25.87 25.73 25.95 26.02 25.75 25.97 26.07 26.15 26.11 29.07 28.71 24.75 25.55 25.73 25.73 25.89 25.82 25.89

Longitude (°E) 159.60 159.60 156.68 156.95 156.55 156.57 156.88 156.96 156.57 156.28 156.82 156.98 156.19 156.62 156.73 156.85 156.76 154.03 154.04 155.46 155.65 156.09 156.24 155.36 155.54 155.34

Est. wt (G&G in kg) 60 70 120 150 90 200 115 135 80 140 240 140 65 130 85 60 140 70 100 60 50 145 165 170 90 130

12 Jan 2009 13 Jan 2009 14 Jan 2009 15 Jan 2009 11 Feb 2009

27.22 26.97 26.68 26.73 23.93

156.18 156.68 156.62 156.62 155.91

150 170 160 175 100

Date Latitude (°S) 23 Feb 2009 28.12 05 Mar 2008 28.44 11 Dec 2008 33.28 25 Mar 2008 25.65 Failed to transmit 14 Dec 2008 23.26 04 Apr 2008 24.65 23 Apr 2008 23.56 22 Dec 2008 36.10 12 Oct 2008 24.70 23 May 2008 36.28 09 Dec 2008 22.86 15 Aug 2008 23.40 20 Jan 2009 29.61 Failed to transmit 25 Apr 2008 26.58 22 Jan 2009 23.99 02 Jun 2008 29.81 17 Aug2008 21.83 04 Jan 2009 15.91 21 Nov 2008 25.42 09 Feb 2009 33.24 10 Feb 2009 32.82 12 Feb 2009 34.77 19 Dec 2008 25.31 18 Dec 2008 25.72

Longitude (°E) 160.42 161.08 157.22 155.73

24 Jan 2009 24.91 14 May 2009 21.84 Failed to transmit 13 July 2009 24.87 14 July 2009 -40.88

TAL ( days) 364 6 266 4

155.67 154.87 161.84 151.27 155.69 151.77 158.25 154.88 158.98

268 13 31 273 202 59 258 115 273

157.29 161.61 153.82 159.06 158.27 157.55 156.45 159.24 153.94 155.97 155.72

2 273 8 59 59 13 59 59 59 3 2

157.57 163.36

12 121

158.24 153.64

179 153

17


Figure 2. Release and pop-up locations of pop-up satellite archival tags released off the east coast of Australia 2006-2009.

18


Table 2. Deployment periods and attachment durations (days) achieved by PSATs deployed on broadbill swordfish in the Tasman Sea 2006–2009 (excludes tags that failed to transmit). Year (n) 2006 (8) 2007 (12) 2008 (25) 2009 (4) Total (49)

Programmed deployment period Mean ± SD Range 314.50 ± 53.99 264–365 122.50 ± 51.90 60–180 192.82 ± 124.93 60–365 180.0 ± 0.0 180 194.47 ± 112.08 60–365

Attachment period Mean ± SD 94.50 ± 100.48 66.75 ± 66.33 112.68 ± 117.63 116.25 ± 73.44 98.76 ± 100.61

Range 2–302 5–181 2–364 12–179 2–364

Proportion of deployment period attached Mean ± SD Range 0.29 ± 0.28 0.01–0.83 0.53 ± 0.44 0.02–1.00 0.62 ± 0.42 0.01–1.00 0.64 ± 041 0.07–1.00 0.56 ± 0.41 0.01–1.00

19


Figure 3. Deployment periods of pop-up satellite archival tags: (a) average deployment period set (black line), average deployment period achieved (red line) and the proportion of the set deployment period achieved (blue line) and (b) average deployment period achieved (black line) and the proportion of the set period achieved (blue line) for each of the set deployment periods.

20


7.2.

Potential mortalities

Of all tags that transmitted data, only one pin breakage was recorded. Only PAT-4s recorded the most recent depth greater than 600m and the date that this occurred; Mk10s instead recorded a maximum depth. Depths greater than 1,000 m were reported from two fish (4.08% of the total tags data were received from) within two days before data transmission suggesting that detachment occurred as a result of the depth cut off device either as a consequence of deep diving or possible mortality in these fish. Summary depth records suggest that a further two individuals demonstrated similar behaviour within two days before data transmission with depth cut off devices detaching tags from swordfish either as a consequence of deep diving or possible mortality in these fish. One further fish demonstrated regular diving for 28 days and then appeared to sink to a depth of 216m where it maintained a constant depth until the premature release mechanism was triggered (48 hours later) and the tag was released from the fish. Of those swordfish considered to be potential mortalities (n = 5), tag release occurred 2–28 days post-deployment.

7.3.

Location estimation

The number of days for which locations could be calculated was on average 22.15 ± 35.69% of total days at liberty. The proportion of days on which locations could be estimated decreased as time at liberty increased (Figure 4). The state-space model used to estimate location from the time series of light data recorded failed to converge for eight tags (time at liberty ranged 12 – 273 days) resulting in a failure to produce a most probable track for these tags.

7.3.1.

Mk10 PAT-Fs

The model used to calculate GPS positions was initially developed for acquisitions collected close in space and time and associated with a high number of satellites (>8). Tag clock drift is assumed to minimal and the deployment location programmed as a start point entered when setting up the tag is assumed to be accurate. Acquisitions collected by tags on swordfish however, were often associated with substantial clock drift, were widely separated in space and time and typically associated with 4 – 6 satellites (mean ± SD: 5.56 ± 1.39; range: 4–9). The nature of the acquisitions

21


collected by the tags deployed on swordfish resulted in a lack of convergence within the model and the production of obvious erroneous position estimates. On further investigation, the time search window of the original model was found to be too large causing large errors associated with locations generated from acquisitions derived from