Repatriating leopards into novel landscapes of a South African province

1. Leopards are often translocated away from where they are caught as non-lethal human-wildlife conflict mitigation. It is alleged that leopards fail to settle where they are translocated to, owing to territoriality. We address the need to publish more accounts of successful repatriation of leopards, but also include novel applications aimed at orphans and confiscated leopards. 2. We satellite collared 16 leopards which included a mixture of relocated and translocated leopards, of which the latter included conventional damage causing animals (DCAs, viz ‘problem animals’), orphans and confiscations. We determined standard home-range metrics and assessed home-range stabilization as a means of determin-ing site fidelity. Premature mortality and site infidelity, that is homing back to origins, were considered failures. We looked at range stabilization by examining successive monthly ranges against that of the preceding month, that is utilization distribution overlap indices (UDOIs). 3. Relocations turned out to be residents ( ˜ 3 km, n = 3), while they were immune to intervention, while translocations resulted in 50% success ( n = 12), which were invariably confiscated adults of unknown origin, and simulations of natal dispersals of orphans ( ˜ 25 km, n ecological and genetic considerations, and not exceeding ˜ 400 km. Other non-lethal mitigation should however be considered before translocation of leopards is contemplated. These findings can be applicable to solitary felids with a similar social organiza-tion.


INTRODUCTION
Leopards Panthera pardus come into conflict with humans over their livestock or game that they keep (Cobb, 1981;Grimbeek, 1992;Mizutani, 1993;Swanepoel, 2008), and thus landowners seek to either destroy them on their properties, or have them removed (Inskip & Zimmermann, 2009). Since attitudes to carnivores have changed in the past decade, landowners may resort to non-lethal approaches (McManus et al., 2014). One of the non-lethal options available is that of translocation, which entails the deliberate movement of an animal from one location to another (Athreya et al., 2007;Fontúrbel & Simonetti, 2011;Linnell et al., 1997), which ostensibly mitigates potential human wildlife conflict at the site (Cobb, 1981;Hamilton, 1981;Linnell et al., 1997;Weise, 2016).
For many years, it has been accepted that translocations of leopard into protected areas (PAs) is futile (Cobb, 1981;Mills, 1991), simply owing to the fact that many of these translocations result in animals returning to an origin or simply continue being a nuisance elsewhere (Hamilton, 1981). Hamilton (1981), however, acknowledged that the failures were simply due to saturated populations in PAs, where it is sometimes forgotten that some of the leopards that were translocated by Hamilton (1981) did in fact remain on some of the PAs (admittedly only two of seven males), though the technology at the time made for inconclusive outcomes. There has thus been a fixation to purport on failed translocations of leopard (Cobb, 1981;Hamilton, 1981;Mills, 1991), while there have been notable strides in improving translocation success of leopards in subsequent years (Briers-Louw et al., 2019;Hayward et al., 2006;Weise et al., 2015).
There has though been negative sentiment on leopard translocations, especially in human-dominated areas (Athreya et al., 2007), where conflicts have even lead to human fatalities (Athreya, 2006), leading to such policies being critically questioned (Athreya et al., 2011). There, satellite-collared individuals have never been directly implicated in conflict, but given high human population densities, the potential is there (Odden et al., 2014), though in Africa, at least, releases of the species are never contemplated in high human use areas.
This study emerged as a need to repatriate leopards which had been confiscated in a law enforcement operation ('Operation Dewclaw'; see Table 1), where we were tasked to examine whether homing to the origin would indeed take place. At the time, there were suspicions that leopard were being illegally caught to be laundered into the trophy hunting industry.
The project was expanded to include other routine translocations of putative problem animals too and to assess whether we can repatriate leopards to novel environments.
This work is important in the context of the oft purported failure of leopard translocations (Hamilton, 1981), which in many cases was related to outdated technology, that is VHF radio-telemetry. In the interim, the use of more advanced satellite technology has allowed better clarification on translocation success and shown mixed success, with certain documented failures (Odden et al., 2014;Weilenmann et al., 2011), while there have been notable successes too (Briers-Louw et al., 2019;McManus, 2009;Weise et al., 2015), where , for example, 66.7% of translocated leopards successfully established home-ranges (HR) in Namibia (Weise et al., 2015).
Defining translocation success has been debated and various criteria have been proposed ( Fischer & Lindenmayer, 2000;Fontúrbel & Simonetti, 2011;Linnell et al., 1997), culminating in established standards (IUCN/SSC, 2013). Leopard-specific studies share similar criteria, that is alleviating conflict at source, refraining from conflict at release site, site fidelity, no homing or exploratory behaviour, and to contribute to the gene pool (Briers-Louw et al., 2019;Weilenmann et al., 2011;Weise et al., 2015), importantly in this, is HR stabilization, reproduction and survival past a year (Briers-Louw et al., 2019).
Of published literature for reintroductions of radio-tracked leopards per study using older radio-telemetry there are sample sizes of one (Cristescu et al., 2013;Hayward et al., 2006;Weise et al., 2015), two (Hamilton, 1976;Mondal et al., 2013), and 10 leopards (Hamilton, 1981) from which to examine this. However, in the era of satellite technology, the sample sizes per study have not improved much, and these include reintroductions of one in Botswana (Houser et al., 2011); two into an Eastern Cape PA of South Africa (McManus, 2009); then four into a Botswana PA (Weilenmann et al., 2011); five in India (Odden et al., 2014); and six into PAs of Malawi (Briers-Louw et al., 2019) and Namibia (Weise et al., 2015), respectively.
The premise for deciding upon suitable release sites for this species in South Africa's North West Province (NWP) was the paucity of leopard occurrence records. Since a province wide mammal-based inventory (2010-2013) used a leopard-specific survey design (Power et al., 2019) proven to be without their mother, who has ostensibly been removed or killed. Natal origin is the location where they were first captured when pre-weaned juveniles (<3 months) Evidence of orphaning is determined from camera trap footage placed at the individual's location for a period of time that exceeds that which a mother should tend to it. Leopards are released when old enough (>2 years) with a collar Rehabilitating Known/Unknown Leopard obtained/captured, which has endured an injury and undergoes treatment and rehabilitation in temporary captivity, before later release These include snared individuals, and those which have damaged their teeth, and they are released and monitored camera-trapping effort of at least 500 trap nights is done (Ngoprasert et al., 2012), and some of the vegetation types of NWP had no detections despite >˜1600 trap nights ( Figure 1; Power et al., 2019), thus these areas were earmarked as ideal release areas. It was further acknowledged that the species was likely not absent entirely, but thought to be either unsaturated or functionally extinct, as is a wellestablished hypothesis outside PAs (Balme et al., 2010;Marker & Dickman, 2005;Rosenblatt et al., 2016).
Notwithstanding anthropogenic pressures outside PAs, and the need to mitigate conflict, it was assumed there would be only nominal human disturbance on the edges of smaller PAs (<300 km 2 ), and these areas would have territorial vacancies (Balme et al., 2010;Hamilton, 1981), such that site fidelity would likely occur, and sufficient conspecifics would ensure site attraction (Hayward, Adendorff et al., 2007a;Smith & Peacock, 1990).
We aimed to assess successful establishment of leopards in release areas, that is site fidelity. Our objectives were to (a) assess whether leopard remained in the proximity of the release area and ascertain that HR stabilization took place.
Given this, we appraised the success of leopard reintroductions by examining (a) reproduction, whether by males or females, which would be a proxy for territoriality, and (b) survival after the first year of monitoring. From this we would address the appropriateness of translocations.

MATERIALS AND METHODS
The NWP is covered by the savanna biome in the northern parts ( Figure 1), and it experiences a subtropical to semi-arid climate (Mucina & Rutherford, 2006), and agriculture and mining are prominent economic activities in NWP.
Leopards were either caught using standard walk-in cage traps, or they were confiscated from perpetrators illegally keeping them.
Damage causing animals (DCAs) were mostly captured and then collared (Table 1).
Leopards were immobilized using standard chemical immobilization for the species (McKenzie, 1993). A CO 2 powered Dan-inject ® dart gun pressure set for˜5 m was used to immobilize leopards with Zoletil (Zoletil 100 ® , Virbac RSA, Halfway House), with dosages of 5 mg/kg (Bertram & King, 1976), administered by a qualified veterinarian or one of the authors experienced in this (in a veterinarians presence).
The locations of all animals registered four times per day. The times 00:00, 06:00, 12:00, and 18:00 were selected (after Swanepoel, 2008), while the Sirtrack collar registered hourly intervals, and the Tractgroup collar registered multiple points when active, and these were reduced to 6-h intervals for comparability.
Leopards were transported in ventilated transport crates, and if not released immediately, leopards were kept in temporary captivity (McKenzie, 1993), and the South African National Standards (SANS 1884-3:2008) adhered to as policy. If young (<2 years), we awaited for maturity (> 2 years), for release, or the outcomes of court orders for release. Temporary keeping facilities were registered with the province and adhered to the keeping specifications, while the same could not be said of facilities where they were initially kept illegally, where in one situation, the dimensions were the same as that of the animal.
Study animals were of overlapping categories (cf. Weise et al., 2015; Table 1), and categories could be compared against one another.
Sixteen leopards were obtained for collaring, release and monitoring (Table 2) We calculated HRs in ArcGIS (ESRI, 2019), and where a dispersal or translocation took place discrete areas were examined. We compared across the different categories and grouped confiscated and rehabilitated animals as translocated animals for comparability (cf. Weise et al., 2015).
The nature of translocations followed established protocols (after IUCN/SSC, 2013), while a relocation was defined as a translocation less than the maximal diameter of the specific gender's HR, based on ecological or geographic benchmarks.
Success was gauged when (a) HR stabilization took place or that (b) the leopards remained for at least a year on the release area, that is property or reserve.
For relocated or translocated individuals, successful HR establishment was deemed to have occurred when HRs stabilized and were within at most one HR diameter away from the release sites.
We worked out monthly Home-Range Overlap Indices (HROIs) and Utilization Distribution Overlap Indices (UDOIs) using the approach of Fattebert et al. (2016), but we used shorter temporal periods, that is 30 days and worked out HR overlap based on the previous month's usage. We benchmarked translocated leopards' HRs against that of residents to determine HR stabilization.
We looked at proximity of the release sites (Norton & Lawson, 1985;Weise et al., 2015), and whether the release area was within the 50% and 95% kernel HRs or not. Scaled HR diameters were measured at distances progressively away from an origin (after Fattebert et al., 2015) to a release site. If within this area, this would be seen as a successful establishment, while any scaled HR diameter away from this, progressively less so, until the extreme of returning to the known origin, which would be a certain scaled distance away from the release site.
Success was also evaluated in terms of whether there was sur- Decollaring was planned for the non-releasable collars, a year after deployment, and longer if there was sufficient battery life on the collar, and GPS clusters of kills were located (Pitman et al., 2013;Swanepoel, 2008), and cage traps were placed for recapture and collar removal, with the same sedation procedure employed.

RESULTS
Two leopards registered no HR data, as one died prematurely, and the other lost its collar. HR sizes were calculated for all remaining leopards (SuppInfo Figures 2-4) and presented (Figures 2-4). All leopards had their collars removed (https://www.youtube.com/watch?v= qlWj9ugBMUQ), except two females which evaded capture even after repeated trap conditioning.
Residents which were relocated ( Figure 5(b)), and DCAs having returned to their origins ( Figure 5(d)), had minimum UDOIs and HROIs of 0.15 ( Figure 5(b, d)) and 0.4 (SuppInfo Figure 5), respectively, and so this was used as the benchmark to determine HR stabilization in translocated leopards (Table 3). Translocated leopards of unknown origin had wide variation in HROIs, despite initial stabilizing ( Figure 5(a)), which may be due to large HR sizes ( Figure 2).
A cost breakdown was determined for each individual (SuppInfo Table 4) and summarized (Table 4).
As for the success of translocations, where stabilized HR is formed within one HR diameter of the release areas, or animals were present on release areas, this was the case for three of five females (n = 5, see Tables 3 and 4), while for males, this was three out of seven (n = 7). Altogether across all leopards this would be a 50% success (cf. Weise et al., 2015), when excluding relocation of residents (Table 3). When looking at survival, this tracked HR stabilization in males, but in the case of females this became two out of five, resulting in survival after a year across all leopards to be 45.4% (n = 11, Table 4).
Leopards kept in temporary captivity >100 days were for the most part successful in establishing HRs (Tables 2 and 3), while also more costly ( as they either homed back to where they came from, or died on the release sites owing to intraspecific competition or from wire snares (Table 3). The least expensive intervention was to relocate leopards, being half that of translocations (Table 4).

DISCUSSION
The status quo remains unchanged that translocating (not relocating) problem leopards has limited success -at least in this study, and at a regional level, that is a South African province.
This is not to say that success cannot be garnered elsewhere as there are many agencies that allude to success, while the most convincing documented successes in a province of a similar size are those in the Eastern Cape (Hayward et al., 2006;McManus, 2009 (Weise et al., 2015). It is argued that a carnivore moving out of its recipient area may not necessarily constitute a failure if the animal had little impact on its environment (Weise, 2016). In our case if not exactly in the confines of a PA where the animal was released, we considered relatively nearby to be a good enough proxy using the HR diameter as a yardstick (Table 3)  Carnivore translocations should exceed 100 km in an attempt to prevent homing behaviour (Fontúrbel & Simonetti, 2011;Hamilton, 1981;Lemeris, 2013), and this is indeed confirmed empirically (Weise et al., 2015), where perhaps the relatively short distance transloca- be what is required to prevent homing behaviour, while our shorter translocations were constrained by the extent of the NWP (Figure 1), while we also heeded genetic considerations (Ropiquet et al., 2015) and the possible existence of different ecotypes, that is Kalahari and Bushveld.
Dispersing subadult male leopards can move up to 353 km from their natal sites or 195 km when measured in a straight line from the natal origin (Fattebert et al., 2013), so such long distances (˜200-400 km) can be contemplated for translocating subadult animals, and also for genetic reasons, as has been recently discovered in human-influenced landscapes in South Africa where unde-sirable natal philopatry of males readily occurs (Naude et al., 2020).
In this study, we have demonstrated successful translocations of leopards of both orphaned and unknown origin which were obtained from confiscations, which has also been the case in work in Namibia (Weise et al., 2015) and Botswana (Houser et al., 2011).
It was unclear why confiscated animals were more successfully established in novel HRs than DCA leopards. It could be hypothesized that their longer period of confinement (Table 2) could have been enough to break their homing tendency (Hayward, Adendorff et al., 2007a;Hunter, 1999;Weise et al., 2015), while prior to being seized by authorities these animals may have been in captivity for longer than what we were aware of. Be that as it may, all adults that were obtained in such a manner exhibited success in establishing HRs where they were placed (Figures 2(a), 2(b), 2(d), and 3(a) and Table 3); in one case, the arithmetic mean centre of the HR was˜1 km from the release site ( Figure 2(a)). One theory as to the success of these leopards was that LF02 and LF03 (Figure 2(a) and 2(b)) were coincidentally repatriated to where they were originally obtained from. Though not an impossibility for at least one to be from the release area, this is unlikely, given that a relatedness test was undertaken as part of a forensic examination,

F I G U R E 4
HRs of leopards in the western NWP, with dark (50% kernel), and light grey (95% kernels) for (a) relocated LF07 (Aug-15 to Oct-16) and translocations of (b) LF09 and LM10 (Jun-16 to Oct-16) and (c) LF16 (Sep-19 to Apr-20). Capture and release sites are indicated, as are arithmetic mean centres of HRs and sites of death

F I G U R E 4 (Continued)
and they were found to be unrelated, which would be unusual, given female leopard philopatry Fattebert et al., 2015;Naude et al., 2020).
Of the orphaned leopard, there was success for these animals in general, where at least two, a male and female established a stable HR (Figure 2(c) and 2(d)), and the success of these HR establishments was simply due to applying biological knowledge and simulating a dispersal as closely as possible by repatriating these animals relatively near to where they were born, viz < 50 km (see also Fattebert et al., 2015).
For the first time, we document a negative impact associated with translocation in that a resident has killed translocated individuals (Table 3), and in the one case the vacated range was filled by another, as is known to occur (Bailey, 1993;Balme et al., 2009). In most cases, the translocated leopards simply home back to their origins (Weilenmann et al., 2011). Failures such as this are important to publish (Fischer & Lindenmayer, 2000), as there tends to be a publication bias towards successful translocations (Fontúrbel & Simonetti, 2011).
The presence of an existing population of conspecifics at a site may affect the success of a reintroduction through attraction or avoidance (Hayward, Adendorff et al., 2007a;Smith & Peacock, 1990). Where resident conspecifics occur, translocated carnivores typically undergo extensive exploratory movements (Briers-Louw et al., 2019;Hamilton, 1981;Odden et al., 2014;Weilenmann et al., 2011;Weise et al., 2015) , which is a prelude to their expulsion, which has been evident in the NWP (Figures 2(d) and 3(a)), which is due to the absence of vacancies in the territorial system of stable leopard populations (Bailey, 1993;Balme et al., 2009;Weilenmann et al., 2011). Similarly, with pumas Puma concolor, when translocated they can only establish into the territorial matrix if vacancies are present (Ruth et al., 1998), and in the case of translocated tigers Panthera tigris, they are invariably killed by residents (Seidensticker et al., 1976). We were thus guilty of underestimating local leopard occurrence. Lemeris (2013) has produced a spatial model to determine release site suitability for leopards where various ecological parameters are incorporated, and this appears to be a robust approach to be followed in this.
Furthermore, it is not inconceivable that translocated males may even cause social disruptions at recipient sites when challenging established territory holders (Athreya, 2006;Bailey, 1993;Balme et al., 2009;Hamilton, 1976), and they may elicit infanticide , which could induce downward spiralling source-sink like dynamics (Balme et al., 2009). To date this is speculative and there is more empirical support for the converse (Table 3; Seidensticker et al., 1976).  Vacant leopard territories are re-colonized by male dispersers within 3 months of a predecessor's death (Balme et al., 2009), and in our example with the death of LM08 (Table 3), this took place after about 6 months when we could speculate of a territory take-over, which was possibly a male leopard that had been residing within its territory (or margins) for a while (cf. Naude et al., 2020).

Animal
Where HRs are not asymptotic, and ever increasing in size, this is the case with dispersing subadults (Fattebert et al., 2016;Mizutani & Jewell, 1998), or when animals are engaged in extensive exploratory movements (Briers-Louw et al., 2019;Odden et al., 2014;Weilenmann et al., 2011). Area-observation curves can be analysed by calculating the cumulative monthly HR size change (Briers-Louw et al., 2019), where we found it sufficient to look at monthly-level HR stabilization as to whether asymptotic HR had been attained or not ( Figure 5). Comparing translocated animals to that of residents proved worthwhile in this study (see also Weise et al., 2015), but one needed to subjectively decided upon a particular UDOI threshold, and despite clear territoriality (i.e. reproduction), large HRs exhibited by some translocated leopards did confound the UDOI approach a bit, particularly when short temporal periods were examined.
In our study, leopards settled into HRs relatively quickly (Table 3), not unlike attempts made in Namibia (Weise et al., 2015). Translocated leopards were found to establish HRs as early as 2 weeks in Namibia (Weise et al., 2015), 2 months in India (Mondal et al., 2013) and 4 months in Malawi (Briers-Louw et al., 2019). These results (Table 3) Hamilton's (1981) supposition that female leopards cannot establish HRs when translocated.
Interestingly also, Weise et al. (2015) found no significant differences between the survivorship for leopards which were translocated compared to residents (Weise et al., 2015), while we do report lower survival ( Table 4). The reality is that it is difficult to tease apart the prevailing mortality rates of any given area, whether natural or anthropogenic (Swanepoel et al., 2015), as one has no sure way of ascribing it to translocation. Hence we did not focus on survival by the end of the first year as it has to be placed in context of what the background mortality rate is.
We have found that young adult females (2-3 years) readily established HRs (Table 3; Figures 2(c), 3(b) and 4(a)), which may be due to unsaturated populations recovering from disturbance (Balme et al., 2009;Fattebert et al., 2016), so perhaps such females outside PAs should simply be relocated nearby. However, in general, young adults may still be suitable candidates for translocation (Weise, 2016). This would explain the success of young orphaned leopards (Tables 2-4), and other translocations where leopards were relatively young (Briers-Louw et al., 2019;Houser et al., 2011;Mondal et al., 2013) and settled readily when compared to older candidates, which showed continuous increases in their cumulative HRs (Briers-Louw et al., 2019). With pumas, the best results came when translocated between 12 and 27 months of age (Ruth et al., 1998), which is when they disperse, and are more likely to accept an unfamiliar area compared to an adult who has spent considerable time in a given place and is adamant upon returning (Ruth et al., 1998).
Young leopard males typically disperse from their maternal HRs (Bailey, 1993;Fattebert et al., 2015), sometimes over considerable distances (Fattebert et al., 2013), resulting in exposure to a range of novel environmental stresses which can resemble effects artificially created through translocation (Weise, 2016). In contrast, subadult females usually take over part of their maternal range (Bailey, 1993;Fattebert et al., 2015) and are said to be less suited for translocation (Weise, 2016), but this has not been the case in our study (Table 4), just as some of the published case studies would suggest which demonstrate that young adult females can sometimes successfully cope with translocation, become self-sustaining and contribute to recruitment in a breeding population (Briers-Louw et al., 2019;Houser et al., 2011;Weise, 2016). Normally, leopard females breed for the first time from 3 years , but here, at least two females between 2 and 3 years reproduced (Tables 2 and 3).
Although it ais difficult to assess reproduction in leopards which are very elusive (Hayward, Kerley et al., 2007b;Swanepoel, 2008), as with any felid population when at depressed densities, the opportunity for hastened reproduction arises as a result of a low population density allowing normally subordinate individuals to breed earlier than in established populations (Fattebert et al., 2016;Hunter, 1999).
Like our study (https://www.youtube.com/watch?v=NVWZba96l-c; Table 3), reproduction in translocated female leopards has also been confirmed (Briers-Louw et al., 2019;Weise et al., 2015), and suspected (Houser et al., 2011), while males have been observed copulating (Weise et al., 2015), suggesting a genetic contribution to a population. Reproduction is the ultimate sign of success, particularly where there is the birth of a wild-born generation (Hayward, Kerley et al., 2007b). In our study, a female conceived as early as 3 months after release (Table 3; , while other studies purport this to be at 8 months (Weise, 2016) to over a year (Briers-Louw et al., 2019). Confirmed breeding events have important implications as they may eventually compensate for initial mortalities, demonstrating that translocations can locally supplement and support free-ranging gene pools (Table 3;  Monetary compensation for livestock losses has been considered in lieu of translocation (Athreya et al., 2011;Fontúrbel & Simonetti, 2011), but the costs of leopard translocation may sometimes be less costly (Weise et al., 2014). In the NWP, given an average estimate of ZAR 9000 per heifer (see www.vleissentraal.co.za) this would suggest a break-even point at the loss of five such livestock, or two animals, if no collar is involved with the perpetrating animal (Table 4). There is often no significance in whether translocation or compensation is opted for (Weise et al., 2014). Ultimately, practitioners should decide for themselves what to consider given their available resources. As for nonconflict related leopards, these are generally even more costly (Table 4), but such costs, that is veterinary treatment, satellite collars, temporary keeping, can be recovered from public interest groups (see Weise et al., 2014).
Why some of the translocations of DCA animals failed in the NWP may be due to relatively high human population densities (and anthropogenic mortality), which is said to be the reason for failures in many Indian situations (Athreya et al., 2011), while successes seem to prevail in sparsely inhabited parts of Africa (Briers-Louw et al., 2019;Weise et al., 2015).
The results of this study have been sufficient to dictate a preferred policy of relocation over and above translocation as defined. However, practitioners charged with this on the ground are more inclined to consider translocation, for fear of recurring problems of returning individuals. This is also because local farmers have borne witness to practitioners releasing animals adjacent to capture origins and this has not endeared local farmers to authorities.
Given the largest HR sizes (SuppInfo Figures 2-4), maximal distances for relocation in the NWP would amount to 28 km in the NWP, which is well below genetic threshold distances (Ropiquet et al., 2015).
Outside PAs in South Africa, leopards face threats such as illegal hunting, trapping and snaring (Swanepoel et al., 2015), while they may also be subject to the ills of haphazard translocations (Swanepoel et al., 2016). Mammalogists had to be observed (Gannon & Sikes, 2007).
We thank the following organizations who made this work possible: The veterinarians who assisted in immobilizations: PB, Drs Scoustra, Caldwell, Steenkamp and Venter. We also thank the helicopter pilots who assisted us with searching for collared animals (Messrs Botes, Molteno and Botha) as well as those who flew fixed-wing aircrafts for us (Messrs Sutton and Holland-Ramsay).

CONFLICT OF INTEREST
The authors declare no conflict of interest.

AUTHORS' CONTRIBUTIONS
RJP and M-VB initiated the work, conceived the ideas and methodology, with LV and PB joining in on successively later occasions. RJP led the writing, and all authors contributed critically to the drafts and gave final approval for publication.

DATA AVAILABILITY STATEMENT
The datasets analysed during the current study are available in the Movebank Data Repository (Power & Venter, 2020).