Spatial conservation of large mobile elasmobranchs requires an understanding of spatio-temporal seascape utilization

Abstract

The positioning of habitats interacts with variability in abiotic factors (e.g. seasonal changes in temperature and extreme weather events) to change how animals use a land or seascape. Marine reserves can regulate how human activities alter fish communities and increase the abundance of targeted species, but the combined influence of reserves and seascape context on species habitat use remains uncertain in many ecosystems. Further, marine reserve effectiveness might be low for mobile species if the size of the reserve is less than a species usual range, reducing the overall time a individual may be protected. In this study, we tracked 19 giant shovelnose rays (Glaucostegus typus), an IUCN listed vulnerable species within the Moreton Bay Marine Park in eastern Australia. We used an array of 28 acoustic receivers within a complex mosaic of seagrass patches, bare sand, mangrove forests and deep-water channels and used regression tree analyses to determine which spatial, temporal and protection factors contributed most to G. typus habitat use. Overall, 50% of the total detections in the study occurred inside marine reserves containing large seagrass beds (>7.09 m²) and in close proximity to mangroves (<7.47 km). During winter (<20.2 °C), G. typus centre of activity increased significantly (p < 0.001), and greater than 50% of the detections occurred in reserves in winter. Conversely, during the rest of the year (water temperature > 20.2 °C), the habitat use of individuals is contained in smaller centres of activity compared with winter, however, protection effects varied. Our results show that seascape context and marine reserves combine to provide the optimal areas for G. typus habitat selection. Limited food resources likely caused larger centres of activity during winter. Identifying priority habitats for vulnerable species is critical for ongoing protection and maintaining effective conservation initiatives. We have shown here that incorporating spatial features into the design of marine reserves can improve conservation outcomes for mobile benthic predators such as G. typus and other species that use such seascapes.

marine reserves, seagrass meadow, tracking

Introduction

Habitat heterogeneity affects species habitat utilization, and the movement of organisms and energy across land or seascapes (Loreau et al., 2003; Hyndes et al., 2014). Further, movement patterns can be influenced by variable prey availability, predator avoidance, temperature changes, and reproductive movements at different times of the year (Bond et al., 2012, Jewell et al., 2013, Heupel and Simpfendorfer, 2014). Understanding patterns in habitat utilization by different species across both space and time is a fundamental goal of ecologists, as these movements can significantly affect the composition and function of ecosystems, and therefore how they are managed (Speed et al., 2010; Andrews and Harvey, 2013). In order to further understand where and when individuals are moving, it is critical to incorporate spatial and temporal factors that will drive the movement of animals (Nathan et al., 2008). Therefore, determining the main external factors driving the distribution and movement of different species, particularly species of conservation (i.e. listed species) or economic concern (i.e. harvested species), across multiple spatio-temporal scales is important for managers (Nathan et al., 2008; White et al., 2014).

No-take marine reserves seek to maintain biodiversity, improve resilience and help re-establish species that have been overharvested (Babcock et al., 2010; Edgar et al., 2014). The optimal spatial arrangement of reserves in heterogeneous seascape and their size must, however, be optimized for individual areas and objectives (Olds et al., 2016). For example, large reserves are most effective in protecting fish over time (Halpern, 2003), but this comes at a cost in terms of reducing fishing areas (Stewart and Possingham, 2005). Although it is well established that the seascape context of an area (e.g. depth, proximity to other habitats, habitat size) can influence the effectiveness of marine reserves (Olds et al., 2016), robust metrics for incorporating connectivity into reserve design are difficult to establish (Kearney et al., 2013; Olds et al., 2016). Notionally, sedentary species should benefit more from marine reserve implementation than more wide-ranging species (Zeller, 1997, Palumbi, 2004, Bryars et al., 2012). Therefore the protection value of reserves for individual species might be reduced if the species range beyond the boundaries of the reserve at different times of the year when they may be susceptible to harvest (Chapman and Kramer, 2000; Claudet et al., 2010; Gaines et al., 2010). Some studies have, however, shown that marine reserves can be beneficial for species that have a large home range or activity space, possibly due to increased habitat quality inside reserves resulting in less movement outside of reserves (Moffitt et al., 2009; Claudet et al., 2010). Further, most studies assessing such an effects are conducted on coral reefs (Olds et al., 2012c; Pittman and Olds, 2015), meaning that the effect of seascape context on such metrics remains untested for many important habitats (Olds et al., 2016). Consequently, it is vital that studies continue to assess such effects to enable marine reserve placement to be optimized in different seascape arrangements and for different habitats.

Seagrasses play an integral role in coastal seascapes, providing nursery areas for numerous harvested fish species that play critical roles in the functions of these habitats in the wider seascape (Bell et al., 1988; Nagelkerken et al., 2001). Seagrasses are highly productive, providing foraging areas for many species, meaning the size of seagrass meadows is a key determinant for the supply of these resources (Bowden et al., 2001; Boström et al., 2006). Seagrasses are of high ecological and economic significance for threatened and harvested species, but are also being reduced in extent globally due to coastal development, sedimentation and eutrophication (Heck et al., 2003; Orth et al., 2006; Waycott et al., 2009; Nagelkerken et al., 2015). Seagrass beds are often positioned within heterogeneous seascapes, making them a suitable habitat to assess the influence of seascape context in driving the movement of individuals within an ecosystem (Connolly and Hindell, 2006; Boström et al., 2011). Early studies in seascape ecology used seagrass habitats to focus on the implications of habitat size, edge effects and distance to the nearest seagrass patch on fishes and invertebrates in seagrass meadows (Robbins and Bell, 1994; Irlandi and Crawford, 1997; Micheli and Peterson, 1999). However, despite the well established importance of highly inter-connected seagrass meadows (Unsworth et al., 2015), how the size of seagrass meadows and connectivity with other habitats influence the movement of individuals within a seascape and the implications of this for marine reserve effectiveness remains under-researched (Connolly and Hindell, 2006). This is particularly important for species that use these habitats, whether it is to forage in or move across from one area to another. It therefore is critical to understand how such species use these areas and provide improved conservation initiatives for them.

The giant shovelnose ray, Glaucostegus typus, is a shark-like batoid that inhabits coastal tropical and sub-tropical waters of eastern Australia and Southeast Asia. G. typus is a benthic predator that feeds on invertebrates and small fish on seagrass meadows, mangrove forest fringes and sandy bottoms (Vaudo and Heithaus, 2011; Bessey and Heithaus, 2013; White et al., 2014). Previous studies into the movement of this species have found that G. typus generally stay in inshore areas for extended periods of time, before moving further away from the coast (White et al., 2014). However, this study was located well within the latitudinal range of the species, suggesting that their movement patterns may alter relative to where in the species range a population is located (Last and Stevens, 2009). Reproductive movements can also alter how this species uses a habitat with G. typus expected to pup from September to November, however this wont occur until individuals are above 200 cm in length (White et al., 2014). The ecological characteristics of this species, its reliance on multiple, diverse habitats, and its status as a threatened species on the IUCN Red List (White and McAuley, 2003; White et al., 2014) make it an ideal species to assess for the effectiveness of spatial conservation measures. In this study, we use acoustic telemetry to determine which spatial (e.g. habitat positioning), temporal (e.g. changes in temperature and rainfall throughout the year) and protection (i.e. within a marine reserve or a fished area) factors contribute most to the habitat use of G. typus within the Moreton Bay Marine Park in central eastern Australia. While the Moreton Bay Marine Park was specifically designed to be representative of all different habitats within the bay, there were also important considerations to protect species of conservation concern and to take an adaptive management approach, allowing assessment to determine the effectiveness of zoning for a range of species and habitats. Approximately 16% of the marine park is located within marine reserves, where no extraction of resources of any form is allowed. Moreton Bay therefore offers a suitable location to test these questions, as (1) there is an extensive acoustic array situated in across both fished and reserve sites; (2) G. typus are under threat as a result of habitat degradation (Orth et al., 2006; Maxwell et al., 2015), being targeted by fishers (Webley et al., 2015) and being by-catch in broader fisheries (Pogonoski et al., 2002); and (3) it offers a heterogeneous seascape comprised of shallow seagrass habitats interspersed with deep channels, coral reefs and mangrove forests at varying distances to the open ocean. Acoustic telemetry offers a unique way of determining how individuals within a population use a range of habitats and whether they may be spending time within reserves (Kramer and Chapman, 1999; Speed et al., 2010; Bond et al., 2012).

Methods

Study site and acoustic receiver array

Glaucostegus typus were tracked in the Moreton Bay Marine Park from January 2015 to March 2016. G. typus are a zoobenthivore, feeding predominantly on invertebrates located in the sediment, however, they have also been found to have a small amount of fish in their diet (Froese and Pauly, 2000; Elliott et al., 2007). This study therefore started in the austral summer (December, January, and February) and went through the austral winter (June, July, and August) and end after a second austral summer. Moreton Bay is a sub-tropical embayment in southeast Queensland, Australia (27°S, 153°E; 1582 km²). Moreton Bay is bordered by three large sand islands to the east and to the west by Brisbane, the third largest city in Australia (Figure 1). Oceanic water is exchanged between the eastern sand islands, and three large rivers discharge into the western regions of the bay (Gibbes et al., 2014). The study was conducted in the central eastern sector of the bay; a heterogeneous seascape close to the open ocean that is dominated by shallow seagrass habitats and interspersed with deep channels and mangrove forests (Figure 1).

Figure 1.

The acoustic array in the Eastern Banks region of Moreton Bay, illustrating the main habitat types and the 5 m isobath. Crosshatched areas represent marine reserves.

The acoustic receiver array in Moreton Bay comprised 28 VR2W monitoring receivers (VEMCO ltd, Halifax, Canada) positioned on shallow seagrass beds (between 1.5 and 5 m depth) or in adjacent channels (in 5–25 m water depth). Five receivers were positioned as a gate at the nearest large opening to the open ocean to record individuals entering and leaving the area. The array covers a total area (minimum convex polygon) of approximately 180 km², but detection range is limited to a radius of 500 m around each receiver, resulting in a total detection area of approximately 44 km² (Zeh et al., 2015). The array also covers portions of five separate no-take marine reserves, which make up approximately 31% of the total array area (∼59 km²). Of the 28 receivers in the array, 11 (39%) are positioned either within marine reserves or on the edge. This array was not set up for this project, but the receiver locations provided adequate coverage to evaluate habitat use of G. typus in the study area.

Animal capture, handling, and tracking

G. typus were caught on 25 m bottom-set setlines using ten 10/0 circle hooks on a 2 m wire tracer baited with mullet (Mugil spp.). Once individuals were caught, they were manoeuvred to the side of the boat, where they were turned onto their back and into a state of tonic immobility before being sexed and measured (stretch total length). Individuals in the study ranged from 97 cm to 175 cm, with all individuals fitting into the sub-adult life stage (Last and Stevens, 2009). While this life stage was not specifically targeted, it does suggest that sub-adults may dominate individuals within Moreton Bay. Transmitters were surgically inserted into the abdominal cavity through a small incision on the underside of the individual to reduce tag loss and biofouling (White et al., 2014). VEMCO V13 (13 x 36 mm) acoustic transmitters on a 60–120 s random interval at 69 kHz, which minimized simultaneous detections on the receiver array. Therefore, a single detection refers to any recorded detection on a receiver, by any transmitter this transmits on the 60–120 s random interval. Receiver detections were downloaded on three separate occasions; May 2015, October 2015, and April 2016. The final receiver download coincided with the removal of the array, ending the project after 1 year and 3 months.

Environmental attributes

We used GIS to extract a range of spatial attributes associated with individual receivers (and therefore individual detections). These attributes were; total area of the seagrass meadow in which the receiver was located; the proximity of the receiver to the open ocean and mangrove forests; the deepest water depth within a 1 km radius of the receiver; and whether the receiver was positioned within a no-take marine reserve or in a fished area. Average monthly rainfall [taken from daily recordings at the nearest weather station (BOM, 2016)]; and average monthly water temperature [recorded by the Healthy Waterways Monitoring Program (HWMP, 2015)] were assigned to each detection depending on date of detection. A description, justification, and underlying ecological hypothesis for each variable are given in Table 1.

Table 1.

Explanatory variables to be used in the regression tree analysis.

Variable	Description	Method/Source	Underlying ecological hypothesis
Distance to mangroves	The distance from each receiver to the nearest mangrove forest	(Herbarium, 2015)	Mangroves are important for prey species (White et al., 2014)
Distance to ocean	The distance from each receiver to the ocean	GIS	Proximity to the ocean is important for prey availability and reproductive movements
Seagrass patch area	The area of the seagrass patch that each receiver is present in	(HWMP, 2015)	Larger seagrass meadows provide more resources than small meadows
Protection	Whether a receiver is located in a marine reserve or fished zone	(Queensland Government, 2007)	Higher abundance of prey items (Kramer and Chapman, 1999; Marshell et al., 2011).
Depth contour	The deepest water within a 1 km radius of a receiver	(HWMP, 2015)	Nearby deep water is important for feeding, movement and as a predator refuge
Water temperature	The monthly water temperature	(HWMP, 2015)	Temperature changes will alter prey availability
Rainfall	The monthly rainfall	(BOM, 2016)	Inundation of freshwater and increased sediment may cause individuals to shift movement

Each variable used is listed, along with a description, method attained, and the underlying ecological hypothesis.

Data analyses

We used conditional inference regression tree analyses in the party package of R (Hothorn et al., 2006) to determine which environmental attributes correlated most with habitat use by G. typus. Branches in the tree were permitted only for significant (p < 0.05) splits. Regression trees were conducted on the presence/absence of detections for each week of the study for each receiver. Due to this a binomial distribution was applied to the regression tree analysis. If an individual was found to be present at a receiver during a week, it was marked as present; those not detected at that receiver were marked absent.

Centres of activity were plotted as minimum convex polygons (MCPs) for individuals in GIS (ArcGIS v10.0 ESRI, Redlands, CA, USA) to visualize the area used over different time periods. Kernel densities were not used in this study, as many individuals were not recorded on enough receivers to be sufficient. MCPs represent the extent of an individual’s movement within the acoustic array (Marshell et al., 2011). A residency index (% days detected from first to last detection) was calculated for each individual to show the amount of time individuals are spending inside the area of the array.

As some individuals were only detected on two receivers in some seasons, MCPs could not be constructed for all individuals. Therefore, when MCPs were unable to be used we used the number of receiver which each individual was detected on instead, as a proxy for home range size (Garla et al., 2006). Differences in centres of activity between different times of the year were determined using paired t-tests. Similarly, marine reserve use across seasons was compared using a paired t-test on the number of receivers each individual is detected on within and outside of reserves for the seasons of interest.

Results

Spatial and temporal drivers of habitat use

During this study, detections were recorded for 19 out of 20 tagged individuals, with 22 of 28 receivers detecting individuals. Three individuals were recorded on 7 days or less and so were excluded from analyses. The following analyses are therefore based on 16 individuals. Detections on individuals was highly variable with some individuals being detected <50 times, while others were detected >3500 times (Supplementary Table S1). Individuals were detected on average across a 280-day period (the time between the first and last detection) (Supplementary Table S1). Detections only occurred on ∼15% of the total number of tagged days (number of days all individuals were tagged, 4202 tagged days), with ∼58% of days detected having a detection inside a reserve. Of these, individuals were detected across an average of 280 days (min. 121, max. 441) (Supplementary Table S1) and residency index (% days detected from first to last detection) averaged 7.56 +/- 1.8 SE (minimum 0, maximum 28.73; Supplementary Table S1). The area that each individual used varied greatly, with an average minimum convex polygon size of 24.7 km² +/- 5.6 (minimum 2 km², maximum 76 km²; Supplementary Figure S1). Size (t = 1.165, p = 0.262) or gender (f = 0.003, p = 0.956) of individuals did not result in a significant difference in the size of minimum convex polygons.

The regression tree analysing the presence or absence of individuals at individual receivers showed that monthly average water temperature was the most influential factor, causing the first split in the tree (Figure 2). When water temperature was below 20.2°C (June–August), the seascape factors distance to mangrove, seagrass patch area and distance to ocean all caused further splits in the tree. Therefore, during winter, individuals were most likely to be found at sites in larger seagrass beds (> 25.84 km²) close to mangroves (< 7.47 km). Conversely, when temperature was above 20.2°C (April–May and September–April), individuals were most often detected on receivers surrounded by shallow water and on smaller seagrass patches below 25.84 km².

Figure 2.

A conditional inference tree on (a) the presence and absence of individuals at different receivers throughout the study period. Interpolated maps of the acoustic array highlight when water temperature was (b) above 20.2 °C and (c) below 20.2 °C. The colour scale of purple-white-orange (black-white-grey) indicates a high-medium-low presence of individuals.

Centres of activity (shown as MCPs for individuals that appeared on three or more receivers) changed dramatically for some individuals for times when water temperature was below 20.2°C compared with above, with some MCPs being as small as 2.64 km² in summer and changing to 56 km² (Figure 3). After the initial split (temperature < or > 20.2°C) in the regression tree assessing the number of detections, a paired t-test showed that individuals were recorded on more receivers when water temperature was below 20.2°C compared with above (p = 0.006).

Figure 3.

A minimum convex polygon (e.g. grey shaded area) for three individuals during summer (> 20.2°C, top) and winter (< 20.2°C, bottom). Marine reserves overlapping with the minimum convex polygon are highlighted in white. Minimum convex polygons varied through seasons: (a) 4.95 km², (b) 2.35 km², (c) 6.6 km², (d) 14.75 km², (e) 54.99 km², and (f) 63.2 km².

Marine reserve effectiveness

Acoustic receivers within reserves represent 39% of the total array in Moreton Bay, suggesting that when the proportion of detections was above this value, there was greater use of the reserves than outside for individual within the study. Fifty-three percent of the total detections in this study were recorded in reserves during winter, with only 23% in reserves in the remainder of the year. Receivers that were located within protected areas had more detections when water temperature was below 20.2°C. Overall the number of days detected inside marine reserves for all individuals was 58% when compared to number of days detected. With a further 58% (∼34% of the total number of days detected) of the days individuals were detected inside marine reserves occurring when water temperature was below 20.2°C. Individuals were also recorded on a greater number of receivers inside reserves when temperature was below 20.2°C (p = 0.003, Figure 3).

Discussion

Seascape context interacts with marine reserves to alter fish abundance, community composition, and ecological processes globally (Berkström et al., 2012; Pittman and Olds, 2015; Olds et al., 2016). Understanding species movement and habitat use across both space and time is important for management, as these movements can significantly affect the composition and function of ecosystems (Speed et al., 2010). Our results show that the seascape context of habitats is critical in the use of marine reserves by G. typus, with marine reserves located in large seagrass meadows and near deep water being used more by individuals. Our findings concur with other studies that have reported positive seascape effects on marine reserve performance (Huntington et al., 2010; Olds et al., 2012b; Martin et al., 2015), however, these studies focused primarily on fish abundance and fish community composition. Here, we focus on how seascape factors influence habitat use and the success of marine reserves on a highly mobile species. We show that beyond these priority effects of seascape, habitat use and marine reserve use by G. typus varies according to temperature. Approximately 70% of detections occurred within marine reserves during winter, while less than 25% were inside marine reserves during the remaining periods of study. Approximately, 58% of the days individuals were detected inside reserves occurred during winter. Our results suggest that the current placement and amount of no-take marine reserves in Moreton Bay provides periodic protection to the vulnerable G. typus, and that the success of these marine reserves in protecting such species also depends on the seascape context, especially seagrass bed size.

Spatial use by animals in heterogeneous seascapes is governed by the habitat size and positioning relative to other habitats and habitat patches (Dorenbosch et al., 2005; Almany et al., 2009; Berkström et al., 2012). In this study, seascape context played a pivotal role in the presence of individuals and the amount of time spent in different positions in the seascape. Larger seagrass beds provide a more productive ecosystem to meet resource needs of individuals (Prado et al., 2008; Smith et al., 2010; Espinoza et al., 2015), including a more stable food supply (Bowden et al., 2001; Boström et al., 2006). While seagrass bed size played a key role in the amount of detections, the availability of nearby deep water was also critical in the space use of individual, therefore suggesting that the inclusion of this important refuge in spatial conservation plans would be suitable (Jankowski et al., 2015; Papastamatiou et al., 2015). As many minimum convex polygons were generally limited by the western margin of the acoustic array and detections throughout the study were low, it is likely that deep water habitats to the west play a more integral part in the movement of this species than this array allows to determine. Deep-water habitats are critical for many species, they offer a refuge from predation and fish and alternate food resources (Jankowski et al., 2015). Re-designing reserves to include multiple habitats or deep-water refuges, in particular allowing reserve boundaries to not be constrained by habitat boundaries (Knip et al., 2012; Graham et al., 2016), may further increase the effectiveness of reserves in this heterogeneous seascape. This would also likely benefit a range of other species within the Moreton Bay Marine Park, with previous work in seagrass meadows and coral reefs supporting the need for better connected reserve sites that include multiple habitat types (Olds et al., 2012a; Gilby et al., 2016; Henderson et al., 2017).

In order for marine reserves to provide adequate protection, a significant proportion of the life cycle of an animal needs to be protected (Hooker et al., 2011; Graham et al., 2016). Marine reserves within the Moreton Bay Marine Park were important for the protection of G. typus; however, the strength of the effect was increased in winter, was dependent on depth, seagrass patch size and proximity to mangrove forests. Therefore, reserves within Moreton Bay only periodically protect these individuals, which is crucial at these early life stages (Heupel et al., 2007). While it is critical that these early life stages are protected, management also needs to focus on incorporating all life stages into management plans (Kinney and Simpfendorfer, 2009). Most current marine reserves within MBMP, however, only cover small shallow seagrass meadows or sand flats that do not provide optimal habitat for feeding for G. typus (White and McAuley, 2003; Queensland Government, 2007). Most of the reserves in the network are defined by the edge of these habitats, and may be missing crucial parts of the species range, especially deep water or connected mangroves (White et al., 2014; Dance and Rooker, 2015).

Temporal variation in the movement of species may be because of physiological temperature limits (Reid et al., 1991; Heupel and Simpfendorfer, 2014), spawning/breeding migrations (Costa et al., 2012; Taylor and Mills, 2013) or resource availability (Kramer and Chapman, 1999; Marshell et al., 2011). However, due to the presence of individuals year round, the latter two are most likely the factors causing the largest changes in movement patterns in this region. In this study, individuals extended their centre of activity to react to what could be a likely reduction in available resources, as has been seen in many marine (Green et al., 2015) and terrestrial ecosystems (Moorcroft et al., 2006). Given that the majority of detections during this study are inside a marine reserve, we expect that marine reserves, when designed to incorporate important seascape properties, may be capable of managing fluctuations in resource availability throughout a year (Dell et al., 2015). Alternatively, this species has been shown to increase its centre of activity or migrate during different parts of the year (White et al., 2014), however this species was present year round in Moreton Bay, only increasing detections during winter. In other locations, this species and other shovelnose ray like species use inshore embayment’s for pupping and mating resulting in highly variable home ranges and core activity areas throughout different seasons (Talent, 1985). Whether the changes in habitat use by G. typus are influenced predominantly by resource availability or a need to migrate for spawning or breeding, marine reserves still need to be designed to provide protection for a species throughout the entire year. Here, we have seen that while these factors are highly influential, incorporating seascape properties further into the design of marine reserves can provide protection across these temporal scales. This can also benefit a range of other species in Moreton Bay; with connectivity between habitats being well known to improve reserves, it would be beneficial to improve reserves in the MBMP by crossing habitat boundaries, increasing the size of reserves and having better connected reserves (Olds et al., 2012a; Henderson et al., 2017).

In this study, we show that the protection effects of marine reserves for G. typus are greater when the reserves are positioned in large seagrass beds, near other important seascape features. Further, this study provides proof-of-concept that acoustic telemetry and seascape context factors can be combined to determine the effectiveness of conservation measures and find what factors can be used to improve reserves. The use of acoustic telemetry in marine reserve assessment is suitable as it provides long-term data on the use of an area across multiple spatial and temporal scales (Meyer et al., 2007; Marshell et al., 2011). Incorporating the knowledge that is gained from studies such as this one and others like it, managers can assess how re-designing marine reserves to include a range of seascape properties to improve reserve performance for this species and others (Knip et al., 2012). Consequentially, understanding the movement and habitat use of harvested species, such as G. typus, a species that is representative of a large proportion of the community, is vital to the adequate design of successful marine reserves (Schofield et al., 2013). While this study has only been conducted on a single species in a location at the southern end of its range. The results are still critical for providing further understanding of this species, but this also serves as an important case study on determining important seascape factors that drive species movement and habitat use and can be beneficial in reserve design. Therefore, reserve effectiveness across multiple scales needs to be investigated further and results incorporated into the improved design and monitoring of marine reserves.

Acknowledgements

Funding for this project was provided by the Margaret Middleton Fund as part of the Australian Academy of Science. We thank the staff of the Moreton Bay Research Station and K. Finlayson, C. March, B. Hodgson, S. Brodie, N. Harcla-Goody and V. Thomson for field assistance. Animal ethics protocols were followed as per Griffith University Ethics approval ENV/07/13/AEC.

Supplementary data

Supplementary material is available at the ICESJMS online version of the manuscript.