Sympatric ray species show different temporal patterns for accessing provisioned food

Introduction

Nature-based tourism has increased significantly since the 1990s (Silva et al. 2023). Elasmobranch (shark and ray) tourism is a rapidly growing sector (Gallagher and Hammerschlag 2011; Gallagher et al. 2015) that often uses food provisioning to facilitate encounters for paying customers (Brena et al. 2015; Richards et al. 2015; Trave et al. 2017; Patroni et al. 2018). However, it has been shown that feeding wildlife can cause changes in species abundance, behaviour, movement patterns and community structure (reviewed in Trave et al. (2017), Brena et al. (2015) and Patroni et al. (2018)). However, there are some important social and economic benefits (Burgin and Hardiman 2015). For elasmobranchs, which are globally threatened (Dulvy et al. 2014; Pacoureau et al. 2021), such tourism has helped conservation by improving attitudes and highlighting the non-consumptive value of these animals (Brunnschweiler 2010; Clua et al. 2011; Gallagher and Hammerschlag 2011; Vianna et al. 2012). However, educational and financial benefits are typically only maximised if feeding is part of a regulated or managed activity (Dubois and Fraser 2013; Healy et al. 2020).

Wildlife feeding often occurs outside controlled contexts through unregulated tourist attractions, whereby wildlife may be attracted because of incidental provisioning activities, such as dumps or fishing discards (Oro et al. 2013), after which the public takes notice and begin feeding the animals to lure them closer. In these contexts, there are considerable risks to animal and human welfare because of a lack of codes of conduct guiding appropriate human behaviour or education around the risks (Healy et al. 2020). These risks are particularly concerning for elasmobranch tourism where target species can be potentially dangerous to people (e.g. stingray barbs, shark bites). However, the extent to which unmanaged provisioning of wildlife occurs is hard to determine (Dubois and Fraser 2013) and a lack of awareness of such activities leads to a deficiency in data to support monitoring and management and reduce long-term impacts. Research interest does not usually begin until an activity becomes commercialised and attracts public attention, such as when there is growing social and economic pressure to ensure the welfare of the target animals, safeguard the longevity of the activity, or to address emerging concerns such as injuries to or from animals. A prime example is the stingray-feeding tourism attraction at Stingray City Sandbar in the Cayman Islands (Vaudo et al. 2018). Southern stingrays (Hypanus americanus, formerly Dasyatis americana) were originally attracted to the site by fish waste discarded from fishing boats in the 1930s, and in the 1980s SCUBA divers began hand-feeding the stingrays (Shackley 1998). The activity was then commercialised and today is worth US$50 million and the site is visited by over one million tourists annually (Vaudo et al. 2018). However, the first research into potential impacts on the southern stingrays at the site was conducted only once tourist numbers reached ~100,000 annually and the number of stingrays had increased 10-fold (Shackley 1998), more than 50 years after the stingrays were first attracted to the site by commercial fishing activity.

Wildlife tourism often focuses on a single target species and there is a growing body of research into potential impacts on them (e.g. southern stingrays at Stingray City Sandbar, (Vaudo et al. 2018); white sharks (Carcharodon carcharias) in the Neptune Islands, South Australia, (Bruce and Bradford 2013); tiger sharks (Galeocerdo cuvier) at Tiger Beach in the Bahamas, (Hammerschlag et al. 2012); reviews: Brena et al. (2015) and Patroni et al. (2018)). However, provisioning rarely affects any one species in isolation. For example, eight species of sharks are provisioned for diving tourism at Shark Reef Marine Reserve, Fiji, and species abundance and group composition has changed over time owing to competitive exclusion and changes in provisioning activities (Brunnschweiler et al. 2014). At the Neptune Islands, South Australia, various pelagic fishes, reef fishes and rays forage on baits used for white shark cage diving in, but the consumption of baits differs among species (Meyer et al. 2020). These studies have highlighted that a broader, assemblage-level approach to assessing the impacts of wildlife provisioning is needed to aid in the development of more effective management strategies.

The limited research that does exist on multi-species groups has focussed on community composition and dynamics (Clarke et al. 2013; Brunnschweiler et al. 2014) or consumption (Meyer et al. 2020). There is very little research comparing the movement and site-use patterns of multiple species using the same provisioned resource. This lack of research is surprising, given that the majority of research on single provisioned species demonstrates that behavioural changes are widespread (reviewed in Brena et al. (2015), Trave et al. (2017) and Patroni et al. (2018)). However, the behavioural responses of provisioned elasmobranchs documented in the literature have been rather mixed, both among different species at the same site and the same species at different sites. For example, in Hamelin Bay, Western Australia, smooth stingrays (Bathytoshia brevicaudata), black stingrays (B. lata, formerly Dasyatis thetidis) and southern eagle rays (Myliobatis tenuicaudatus) are fed as part of an unregulated tourism attraction (Newsome et al. 2004). Although all species showed similar patterns in use of the site that were linked with tourist numbers, the two larger stingray species showed aggressive behaviour towards the smaller southern eagle rays during feeding (Newsome et al. 2004). In other locations where different populations of the same species are provisioned, behavioural responses have differed. For example, provisioning associated with cage-diving in South Africa has negligible effects on white shark foraging behaviour (Laroche et al. 2007), whereas white sharks in the Neptune Islands, Australia, exhibited considerable behavioural changes in response to increased intensity of cage-diving operations, including increased periods of residency, greater time spent near cage-diving vessels, and altered diel habitat use patterns (Bruce and Bradford 2013). This difference suggests that the impacts of provisioning are species-, site-, and/or context-specific (Hammerschlag et al. 2012), but these drivers, and the interplay among them are poorly understood. Assessing sympatric species exposed to the same provisioning conditions at the same location can help isolate species-specific behavioural responses from those driven by environmental factors, providing insight into the variability observed in responses to provisioning across different contexts, and potentially furthering our understanding of when the impacts of provisioning will be limited or more substantial.

This study aimed to compare the movement patterns of two sympatric ray species, the smooth stingray and southern eagle ray, around a site where they are fed as part of an unregulated but popular tourist attraction in south-eastern Australia. Using passive acoustic telemetry, we compared the presence, duration of visits, and activity spaces of the two ray species and how they relate to patterns in provisioning activity. On the basis of the previous research at Hamelin Bay, Western Australia (Newsome et al. 2004), where the context is similar, we expected both species to show similar patterns in presence and space use, and for these patterns to be strongly linked to when and where provisioning occurred. We also expected the larger smooth stingrays to have stronger attachment (i.e. higher presence, longer visits, and smaller activity spaces) during peak provisioning times (i.e. afternoons and weekends) as a result of competitive exclusion of the smaller eagle ray species.

Materials and methods

Study site

This study was conducted along the open coastline surrounding Boat Harbour Beach, Bendalong, New South Wales (NSW), Australia (35.245°S, 150.539°E; Fig. 1). Boat Harbour Beach is a 300 m long beach with two boat ramps and fish cleaning facilities on the rock platform at the eastern end (Fig. 1b). Smooth stingrays and southern eagle rays have been provisioned in the shallow waters at the eastern end of the beach for many years (Fig. 1b). Provisioning began with recreational anglers discarding fish cleaning waste from the fish cleaning tables at the eastern end, which led to the handfeeding of fish waste and commercially produced baits to the rays by the public. At the time of this study, the feeding of the rays was unregulated, and knowledge of the activity was largely spread by word-of-mouth (e.g. Trip Advisor Australia 2018).

Fig. 1.

(a) Map of the study location around Bendalong, NSW, Australia, with the positions of the 10 VR2W passive acoustic receivers used to track smooth stingrays and southern eagle rays. Inset shows geographical location of the study site in Australia. (b) Aerial photograph of Boat Harbour Beach in Bendalong, NSW, Australia, where smooth stingrays and southern eagle rays are provisioned. Aerial photograph sourced from Google Earth Pro ver. 7.3.3.7786 (imagery date: 24 February 2016).

We did not quantify when provisioning activity occurred during this study; however, there are specific temporal patterns inherent in recreational fishing and tourist activity at the site that are used as a proxy. First, provisioning occurs throughout the day but is higher in the afternoons, driven by tourists spending the day at the beach and recreational anglers returning to fish cleaning facilities in the afternoon (Pini-Fitzsimmons et al. 2018; Lynch et al. 2020). Second, because of the recreational nature, tourist numbers and fishing effort is higher on weekends than weekdays when people are not at work (Sunger et al. 2012; Finger and Lehmann 2012; Flynn et al. 2018; Kendall et al. 2021).

The study was primarily conducted during spring and summer months, which is a period of moderate to high tourism activity at the site. During this time, tourist numbers and provisioning activities typically increase, particularly during summer and school holidays, before decreasing during cooler months. While different seasons or holiday periods may lead to varying provisioning patterns and associated impacts on the rays, seasonal variation in provisioning was not directly quantified as part of this study.

Results

Acoustic tracking

All seven tagged southern eagle rays and six tagged smooth stingrays were detected on the acoustic tracking array post-tagging (Fig. 2). Following the removal of detections within the first 12 h following tagging, all recorded acoustic detections (n = 141,779) were considered valid and retained in subsequent analyses. Tagged smooth stingrays were tracked for an 232 ± 49.642 days (mean ± s.e.) and tagged southern eagle rays were tracked for 197 ± 49.123 days. The detection index for southern eagle rays was 0.823 ± 0.111, and for smooth stingrays it was 0.643 ± 0.130; however, this difference was not significant (Kruskal–Wallis test: χ² = 2.947, P = 0.086).

Fig. 2.

Detection plots for individual (a) southern eagle rays and (b) smooth stingrays tagged in this study. VR2W passive acoustic receivers on which the rays were detected are ordered from north to south on the y-axis for each plot and date (local time) is given on the x-axis. Points indicate that the individual was detected on that day on the given receiver and lines between them connect consecutive detections. Vertical dashed lines indicate the date of tagging. The receiver closest to the provisioning site at Boat Harbour Beach, Bendalong, NSW, Australia, is highlighted by the horizontal red dashed line.

Overall space use

Both core and total overall activity spaces were larger for smooth stingrays (core = 5.248 km², total = 14.795 km²) than for southern eagle rays (core = 0.859 km², total = 7.397 km²), with southern eagle rays exhibiting more restricted space use centred around the provisioning site (Fig. 3). The extent of overlap for the two species encompassed the provisioning site for both core and total overall activity spaces (Fig. 3). The area of overlap of core overall activity space for the two species was 0.843 km², encompassing 98.2% of the southern eagle ray and 16.1% of the smooth stingray activity spaces (Fig. 3a). The area of overlap of total overall activity space for the two species was 7.389 km², encompassing 99.9% of the southern eagle ray and 49.9% of the smooth stingray activity spaces (Fig. 3b). Smooth stingrays, therefore, travelled further within the acoustic array than did southern eagle rays, which showed higher attachment to the provisioning site.

Fig. 3.

Maps denoting areas of the (a) core (50%) and (b) total (95%) overall activity spaces for smooth stingrays and southern eagle rays, highlighting areas used by each species and areas where they overlapped. Note that for southern eagle rays, 98.2% of the core activity space and 99.9% of the total activity space overlapped with the core and total activity spaces of smooth stingrays.

Presence and duration of visits

At the receiver closest to the provisioning site, 6435 hourly presences and 30,580 hourly absences were recorded for southern eagle rays, and 2469 hourly presences and 26,211 hourly absences for smooth stingrays. In total, 5480 visitation events were recorded on the receiver closest to the provisioning site (smooth stingrays = 1470, southern eagle rays = 4010). These visitation events lasted between 5 min and 7.5 h (mean ± s.e. = 22.5 min ± 44 s) for smooth stingrays and between 5 min and 10.5 h (mean ± s.e. = 37.1 min ± 347 s) for southern eagle rays.

The GLMM modelling the drivers of presence (n = 65,695 presences/absences) explained 20.5% of the observed variance. Southern eagle rays showed significantly higher probability of presence at the provisioning site than did smooth stingrays across time period, with the exception of night (Fig. 4a), and across different day types (Fig. 4b). By time period, southern eagle rays had the highest probability of being present in the morning, followed by the afternoon and lowest probability in the early morning and night (Fig. 4a), but showed no variation across day type (Fig. 4b). In contrast, smooth stingrays had the highest probability of being present in the afternoon, followed by the morning, night, and then early morning (Fig. 4a), and had a higher probability of presence on weekends (Fig. 4b).

Fig. 4.

Estimated marginal means (±s.e., reverse-transformed) for the probability of smooth stingray (red) and southern eagle ray (blue) (a–b) presence and (c–d) duration of visitation events at the receiver closest to the provisioning site (Boat Harbour Beach, Bendalong, NSW, Australia) by (a, c) time period and (b, d) day type on the basis of the respective generalised linear mixed-effects models. Significant differences, as outlined in the post hoc summaries given in Table S2 (presence) and Table S4 (duration of visitation events) respectively, are denoted by brackets: significant between-species comparisons are denoted by black brackets and significant within-species comparisons are given by colour-coded brackets. Note: overlapping error bars do not necessarily indicate a non-significant difference (see Lenth 2024).

The GLMM modelling the duration of visitation events at the provisioning site (n = 5480) explained 8.3% of the variance observed. Similarly to the presence model above, southern eagle rays had longer visitation events than did smooth stingrays at the provisioning site across day type (Fig. 4d) and time period, with the exception of the night-time period (Fig. 4c). The duration of smooth stingray visitation events did not vary across time period (Fig. 4c) or day type (Fig. 4d), but southern eagle rays had significantly longer visitation events when they began in the morning, followed by the early morning, afternoon and night (Fig. 4c).

In conjunction, the results of these two models indicate that southern eagle rays were more likely to be present and remain longer at the provisioning site than did smooth stingrays overall, particularly in the mornings. In contrast, smooth stingrays did not vary in the length of time they stayed at the provisioning site over time, but the frequency of their visits (presence) was higher in the afternoon and on weekends.

Temporal activity spaces

The GLM modelling time period activity spaces (n = 6892) explained 9.9% of the observed variance (R² = 0.099). Activity spaces away from the provisioning site were consistent in size for southern eagle rays across all time periods and day types, whereas smooth stingrays had larger activity spaces at night than in the early morning and morning time periods, but showed no difference between day types (Fig. 5a, c). In contrast, around the provisioning site, smooth stingrays showed reduced activity spaces during daylight hours (morning and afternoon) and southern eagle ray showed reduced activity spaces in the morning (Fig. 5b). For smooth stingrays, activity spaces during the afternoon and night were significantly smaller around the provisioning site than were those away from it for the same time period, whereas early morning activity spaces were larger (Fig. 5a, b). Further, their weekday activity spaces were significantly smaller around the provisioning site than for those away from it (Fig. 5c, d). Southern eagle rays had consistently smaller activity spaces than did smooth stingrays for all time periods and day types both around and away from the provisioning site (Fig. 5). Overall, the model indicates a timing and location effect at the provisioning site that is more pronounced for smooth stingrays (Fig. 5).

Fig. 5.

Effect plots (estimated marginal means ± s.e., reverse-transformed) for the generalised linear model comparing the area of time period activity spaces (km²) of individual smooth stingrays and southern eagle rays by whether (b, d) or not (a, c) they occurred around the provisioning site during the time period (a, b) and day type (c, d) for which the area was calculated. Significant between-species post hoc comparisons are given by black brackets and significant within-species post hoc comparisons are given by colour-coded brackets (bold brackets indicate within-species comparisons between locations). Details are given in Table S6. Note that overlapping error bars do not necessarily indicate a non-significant difference (see Lenth 2024).

Discussion

There is a growing body of research assessing the movements of marine species targeted for wildlife tourism that uses food provisioning; however, most studies focus on a single species or contexts that are at least partially regulated or managed. Still, conflicting behavioural responses are commonly documented, including among different species at the same sites (e.g. rays at Hamelin Bay, Western Australia; Newsome et al. 2004). These inconsistencies highlight the need for multi-species studies to better understand when provisioning activities might result in limited or more substantial impacts. We used passive acoustic telemetry to compare the movements of two sympatric ray species, the smooth stingray and the southern eagle ray, at an unregulated ray-feeding tourist attraction in Bendalong, NSW, Australia. Our findings indicated that both species respond to unregulated food provisioning at the study site, but in different ways. Southern eagle rays showed greater philopatry to the site, whereas smooth stingrays showed lower attachment, but their presence peaked during periods of high provisioning activity (i.e. afternoons and weekends). Despite this, the amount of time smooth stingrays spent at the provisioning site did not vary with patterns in provisioning activity, suggesting that smooth stingrays may only make short visits that increase in frequency during high provisioning periods. Further, the two species overlapped in their space use, mostly around the provisioning site, with smooth stingrays showing reduced activity spaces during daylight hours and southern eagle rays in the morning compared with away from the site, indicating a strong association with the location and timing of provisioning activity.

Animals can quickly learn associations with food when it is predictably available in time and space (Reebs 1993; Mulder et al. 2013). Food provisioning for tourism leverages this ability, and elasmobranchs have been shown to learn these associations very quickly (Guttridge et al. 2009; Heinrich et al. 2020, 2021). Previous studies have reported strong site attachment in both smooth stingrays and southern eagle rays at provisioning sites (Newsome et al. 2004; Pini-Fitzsimmons et al. 2018). Thus, it is unsurprising that both species showed attachment in this study. However, like these earlier studies, our study lacked data from non-provisioned control populations, limiting our ability to attribute observed behaviours solely to provisioning. This absence of baseline information on natural space use is an important caveat when assessing site attachment.

Despite the observed attachment by smooth stingrays and southern eagle rays in the present study, the two species exhibited different usage patterns. It is possible they may be avoiding competition over provisioned resources through differential temporal use of the provisioning site (e.g. Every et al. 2019; and Kinney et al. 2011). Our findings suggest that smooth stingrays and southern eagle rays may have adopted different temporal patterns for accessing provisioned food (i.e. time-sharing). Smooth stingrays were generally more absent from the provisioning site, had consistently shorter visits, and larger activity spaces, but their presence at the provisioning site peaked during the day and on weekends when provisioning is generally more intense. Smooth stingrays may therefore be maximising their chances of obtaining provisioned food by focussing on peak provisioning times with frequent visits. Given their larger size (Last et al. 2016), likely higher energy requirements, and provisioned food being a limited resource, it makes sense for them to concentrate their efforts during peak times. In contrast, southern eagle rays showed more consistent presence at the provisioning site, with longer stays and restricted activity spaces, suggesting they may be habituated to the study site, whereby their constant presence allows them to quickly access provisioned food when it is available.

Southern eagle ray presence and visit duration also peaked in the morning, coinciding with smaller activity spaces, possibly indicating anticipation of increased provisioning during the day. Indeed, anticipatory behaviour of provisioning events by rays has been documented previously (Gaspar et al. 2008; Corcoran et al. 2013; Pini-Fitzsimmons et al. 2018). Although, we might also expect similar increased presence in the afternoon because of higher provisioning, their presence actually decreased, and their activity spaces increased. Spending less time at the provisioning site could suggest that they are being competitively excluded from the site by the larger smooth stingrays during this period, when smooth stingray presence is highest, and their activity spaces are smallest at the provisioning site. Competitive exclusion of southern eagle rays by larger stingrays has been observed at Hamelin Bay, Western Australia, where larger smooth stingrays and black stingrays actively displace southern eagle rays in a similar provisioning context (Newsome et al. 2004). Although Newsome et al. (2004) observed active chasing of southern eagle rays, they did not observe a decrease in the total number of eagle rays present with increasing numbers of larger stingrays. In contrast, our tracking data indicated that tagged southern eagle rays reduced their time at the site when smooth stingray presence was highest. However, this difference may reflect methodological differences, whereby Newsome et al. (2004) looked at total numbers of individuals, whereas we tracked the movements of a subset of individual animals. The Newsome et al. (2004) study took place during the summer holiday period when tourist numbers peak and it may be that there was enough provisioning occurring at Hamelin Bay to support both species. Whereas rays for our study were tracked predominantly during spring (September–November; Fig. 2) when tourism is moderate, so the amount of provisioned food would be more limited and thus competition may be greater. Competitive exclusion based on size has also been identified as a potential mechanism for changing abundances of shark species at Shark Reef Marine Reserve, Fiji (Brunnschweiler et al. 2014). It could be argued that if smooth stingrays were competitively excluding southern eagle rays, we might also expect to see southern eagle ray presence and duration of visits decrease on weekends and activity spaces to increase when smooth stingray presence increases; however, this was not observed. Although, weekend provisioning is likely to be more intense than is afternoon provisioning because tourist numbers would be higher for much of the day on weekends, and the biomass of food provisioned may be sufficient to support individuals of both species, as may be the mechanism at Hamelin Bay. Because these patterns are not especially clear from this study, additional monitoring involving direct observations of behaviour over different times of the year (low and high tourist/provisioning activity), coupled with active tracking, would help clarify how smooth stingrays and southern eagle interact over provisioned resources in this system.

Differences in the behavioural ecology of the two species might also explain their different temporal use of the provisioning site. Southern eagle rays may naturally forage in the morning, whereas smooth stingrays may normally forage during the day, resulting in their presence peaking in the morning and across the day respectively. Similarly, being a large species, smooth stingrays may normally have larger activity spaces or move around more continuously than do the smaller southern eagle rays (e.g. Crook et al. 2022), despite eagle rays (Myliobatidae) being considered to have greater swimming efficiencies than do stingrays (Dasyatidae) (Rosenberger 2001). However, not enough is known about the natural space use and ecology of these species to discuss such circumstances with any certainty. Filling these knowledge gaps should therefore be the objective of future research.

Although our data suggest that smooth stingrays and southern eagle rays are responding to unregulated provisioning activities at Boat Harbour Beach, albeit with southern eagle rays showing stronger attachment and potential habituation, it has been argued that if provisioned animals maintain regular movements away from provisioning sites for other purposes (e.g. reproduction), the potential for longer-term impacts owing to provisioning may be limited (e.g. Brunnschweiler and Barnett 2013;Huveneers et al. 2013;Meyer et al. 2009;Maljković and Côté 2011; and Hammerschlag et al. 2012). Research from New Zealand has suggested that smooth stingrays aggregate offshore to breed in summer (Le Port et al. 2012). For the present study, smooth stingrays were detected for 2/3 (~64%) of their days tracked, and preliminary evidence suggests that some made large-scale movements out of the study array in spring–early summer (J. Pini-Fitzsimmons, unpubl. data), including one individual that showed a clear transition from frequent detections to a period of reduced and more sporadic activity before largely disappearing from the array (individual #18131; Fig. 2). This shift in space use may reflect a departure from the provisioning site to fulfil other ecological needs, such as reproduction or seasonal migration. These data support the possibility that unregulated food provisioning at Boat Harbour Beach may not be affecting the broader movements of smooth stingrays. Confirmation of the purpose of the long-distance movements by smooth stingrays is of management interest, and given that only a single reproductive site has been identified for this species (New Zealand; Le Port et al. 2012), and that there is limited gene flow between the Australian and New Zealand populations of smooth stingrays (Le Port and Lavery 2012), identification of breeding sites is of conservation interest.

Southern eagle rays are also suggested to make annual inshore or southward migrations during summer (Potter and Hyndes 1994; Cadwallader 2020), but the purpose is unclear. In New Zealand, southern eagle rays have highest inshore densities in autumn and possible mating in summer (Hartill 1989), indicating that these movements may relate to breeding behaviour. Our data showed large gaps in detections for some individuals and increased variability for others in summer to autumn. Disappearance from the array indicates that despite their strong attachment to the provisioning site, southern eagle rays may still be making regular movements out of the immediate provisioning area, but it is unlikely that they are making large-scale migrations. In general, migratory behaviour and the normal movement patterns of this species in natural contexts are poorly understood. Characterising these movements is necessary to fully understand the influence food provisioning may be having on their broader behaviour and ecology.

Interestingly, the provisioned populations of both species studied here are entirely female and mid- to late-stage gravid individuals are common (J. Pini-Fitzsimmons, pers. obs.). This female bias has been observed in other provisioned elasmobranch populations (e.g. rays: Newsome et al. 2004;Corcoran et al. 2013;Pini-Fitzsimmons et al. 2018; sharks: Maljković and Côté 2011;Hammerschlag et al. 2012;Brunnschweiler et al. 2014). Potential hypotheses include (i) competitive exclusion of smaller males by females for species with such sexual dimorphism (e.g. Corcoran et al. 2013;Newsome et al. 2004), (ii) females having increased energy requirements associated with reproduction (Wearmouth and Sims 2008) and using provisioned resources to supplement their diets (Hammerschlag et al. 2018), (iii) provisioning sites occurring in locations species use for gestation normally (e.g. shallow, inshore areas) (Hammerschlag et al. 2012; Sulikowski et al. 2016), or (iv) sexual segregation whereby females use inshore waters whereas males use deeper waters (Wearmouth and Sims 2008; e.g. McElroy et al. 2006;Knip et al. 2012; Werry and Clua 2013). Given gene flow in smooth stingrays around New Zealand is male-biased, suggesting that males are more transient than females (Roycroft et al. 2019), there is some support for the latter for smooth stingrays studied here. However, additional research is needed to understand this sex-bias, which is an important consideration for the management of provisioning activity and its potential broader effects on reproduction and population viability.

This study has provided novel insights into species-specific responses to food provisioning in a previously unstudied context; however, several limitations should be acknowledged. Many individuals were tracked for short periods (2–6 months), reducing our ability to assess seasonal and long-term responses. Tracking was also concentrated during spring, limiting inference during peak summer tourism. The absence of baseline data on natural space use patterns in non-provisioned environments further constrained our ability to fully attribute observed behaviours to provisioning. Although patterns in space use and visitation duration were observed, the models explained a low proportion of the variance, particularly for the duration of visits and size of temporal activity spaces both around and away from the provisioning site. This suggests that other unmeasured environmental or behavioural factors may also be influencing the observed patterns. Additionally, both smooth stingrays and southern eagle rays showed shifts in the use of the array during the study period, and some evidence of movements out of the array. Given the broader ecological functions of large-scale movements, including possible reproductive migrations, are not yet well understood, further investigation of natural drivers of behaviour in these species is required. These limitations highlight the need for longer-term studies incorporating both telemetry and behavioural observations to fully understand the consequences of unregulated provisioning on ray populations.

Currently, there is no management of the ray-feeding tourist attraction studied here. Left unmanaged, elasmobranch-feeding tourism carries the highest risks to animals and people when compared with other elasmobranch-focussed tourism activities such as cage diving, scuba diving and snorkelling (Healy et al. 2020). Lessons from Hamelin Bay, Western Australia, where management recommendations have been largely unmet (Lewis and Newsome 2003; Newsome et al. 2004; DeLorenzo and Techera 2019), highlight the need for effective oversight and monitoring. Our study has highlighted that while provisioning occurring at Boat Harbour Beach influences the smaller-scale space use of both smooth stingrays and southern eagle rays, broader impacts to the rays may be minimal. However, the growing popularity of ray-feeding at the site warrants continued monitoring of both ray behaviour and tourist numbers to enable rapid adoption of informed management protocols if necessary.

Supplementary material

Supplementary material is available online.