The formation of aggregations and groups can be found occurring in taxa throughout the animal kingdom with considerable intra- and interspecific variation (Elgar, 1989; Whitehead, 1997). The two types of formations however, differ. The first type is driven by non-mutualistic forces, whereas the latter by forces which are typically mutualistic, in that some benefits are derived from group membership (Whitehead, 2008a). Aggregations can form for a multitude of underlying non-social reasons ranging from attraction to a food resource or specific habitat requirements (Johnson et al., 2002) to synchronized patterns of daily or seasonal activity (Guttal and Couzin, 2010). In contrast, animal groups, often termed “social groups” arise by way of underlying social drivers, with individuals actively preferring to associate with each other, therefore displaying social preferences. Within animal social networks, group members are more highly connected with other members in the group than with other individuals within the broader network (Croft et al., 2008). Social groups have been empirically investigated in a wide range of taxonomic groups including insects, fish, birds, and mammals (Chepko-Sade et al., 1989; Elena et al., 1999; Fewell, 2003; Lusseau and Newman, 2004; Cross et al., 2005; Wittemyer et al., 2005; Croft et al., 2006; Naug, 2008; McDonald, 2009; Vital and Martins, 2011).

Historically, sharks have been perceived as solitary predators, however, according to recent studies, some species may exhibit both aggregation and social grouping (Clua et al., 2013; Bass et al., 2016). Moreover aggregation may well lay the important groundwork for the development of social groups (Sims et al., 2000). Currently, it is understood that sharks have the potential to form complex social structures (Mourier et al., 2019; Papastamatiou et al., 2020). For example, blacktip reef sharks, (Carcharhinus melanopterus), were shown to form stable social groups over multiple years through the use of social network analysis (Mourier et al., 2012, 2017). This pattern of social group formation is likely characteristic of reef shark species which have relatively small core home ranges (consisting of a single reef or multiple reefs in close proximity) where these sharks aggregate during the day and range further at night (Barnett et al., 2012). In another reef shark species, gray reef sharks (Carcharhinus amblyrhynchos), formed large groups during the day which dispersed into smaller groups and individuals at night, most likely for foraging (Papastamatiou et al., 2020). Aggregation and dispersion behavior such as these can be explained by two theoretical frameworks: fission-fusion dynamics and central place foraging (CPF) (Kirkwood and Arnould, 2011; Sueur et al., 2011; Jones et al., 2020). In fission-fusion dynamics, “fission” is where individuals split from a larger group (e.g., reef sharks dispersing at night), and “fusion” is where individuals rejoin the larger group (e.g., reef sharks returning to core area the following day) (Papastamatiou et al., 2020). Conceptually similar, yet slightly nuanced, CPF theory describes behavior where animals periodically move in and out of a central place, normally associated with resting and foraging (Orians and Pearson, 1979; Papastamatiou et al., 2018). The two frameworks are non-mutually exclusive as central place foragers can display fission-fusion dynamics over short temporal periods (Papastamatiou et al., 2020).

Wildlife tourism can facilitate the formation of aggregations through provisioning activities such as using bait to attract and/or feed different shark species. Provisioning activities may be seasonal, short-term or even long-term, for example in locations such as Fiji or South Australia where direct and incidental shark feeding (Meyer et al., in press) have occurred regularly over many years or even decades (Brunnschweiler et al., 2014; Meyer et al., 2019). As such, long-term shark tourism sites offer unique platforms to collect baseline data, test specific hypotheses and more generally observe individual and group behavior and interactions in these artificial aggregations. For example, in Mourier et al. (2012) feeding was shown to promote sociality among blacktip reef sharks by attracting more potential social partners. Indeed, shark feeding may drive fission-fusion and CPF as Mourier et al. (2012) appears to indicate, however, the drivers are reversed, the central place is now a feeding location as opposed to a location used for resting.

The bull shark (Carcharhinus leucas) is a large, mobile species that undertakes long-range movements (Heupel et al., 2015) and is not commonly known for displaying aggregating behavior associated with CPF or fission-fusion, nor for displaying social associations under natural conditions [but see Daly et al. (2014) for example of aggregating]. Nevertheless, in the only such study for the species to date utilizing social network analysis, Loiseau et al. (2016) documented social interactions and relatively strong paired associations for two pairs of females suggesting some level of sociality among bull sharks. The study, conducted at an aquaculture farm around Reunion Island, although pioneering, was limited both in terms of sampling time (22 days) and number of identifiable individuals (n = 8). These constraints are not applicable to the Shark Reef Marine Reserve (SRMR), a long-term, multi-species shark provisioning site located on the southern coast of Viti Levu, Fiji (Brunnschweiler et al., 2014). At the SRMR, up to ∼80 bull sharks per day form short-term aggregations with individuals displaying different degrees of site fidelity. Visitation patterns vary with some individuals present almost year round while others remain absent for extended periods of time (Brunnschweiler and Barnett, 2013). The general pattern in diel movements for the bull sharks is to use the area around the provisioning site during the morning hours before dispersing over the broader neighboring reef systems at night (Brunnschweiler and Barnett, 2013). Direct shark feeding (Meyer et al., in press) has taken place since 2003, occurring up to five mornings per week, thus providing a unique opportunity for the assessment of long-term association patterns among individual bull sharks in an isolated insular ecosystem. The specific aims of this study were: (i) to investigate whether this artificial aggregation has facilitated social links between individuals, and (ii) if so, whether the associations persist temporally (i.e., are individuals reconnecting repeatedly at the provisioning events).

At the Shark Reef Marine Reserve, the ongoing wildlife tourism activity of shark diving has resulted in the formation of temporary aggregations of bull sharks. This aggregation was artificially contrived, by virtue of a non-social driving factor: that of direct feeding. Our results appear to indicate that the ongoing nature of these feeding activities, and the aggregations which ensue, have served to facilitate the development of social associations, some of which are temporally stable. With numbers reaching nowadays ∼80 individuals on a single dive, the conditions are prime for social behaviors to occur such as agonistic interactions.

Long-term preferred companionships and strong affinity between individuals were observed throughout the study. Overall and before 2009, the temporal stability of the associations was best described by preferred companionships (i.e., associations occurring more often than expected by chance) and casual acquaintances (i.e., associations lasting from a few days to a few years wherein individuals may dissociate and reassociate again), and for the time periods between 2009 and 2010, and from 2011 by two levels of casual acquaintances. This transformation within the social parameters can be explained by the numerical increase of newly identified bull sharks (Brunnschweiler and Baensch, 2011). Apart from the steady increase over time in numbers of bull sharks (whether named or unnamed) visiting the SRMR (Brunnschweiler et al., 2014), an unusually high number of individuals were identified at the beginning of 2009 (Brunnschweiler and Baensch, 2011). We can only speculate about the reasons underlying such an influx of new individuals in a relatively short period of time, but find it noteworthy. There is anecdotal evidence supporting the notion that bull sharks are indeed gregarious to an extent and travel together. Contrary to other large, apex predator species such as tiger (Galeocerdo cuvier) or white (Carcharodon carcharias) sharks, juvenile and adult bull sharks are rarely seen alone but more often observed in pairs or small groups (Brunnschweiler and Compagno, 2008; Loiseau et al., 2016). Therefore it is possible that a group of bull sharks new to the SRMR show up together at the provisioning site, causing a temporary alteration of the existing social structure by reconfiguring the nature of associations.

In natural settings, individuals form groups and associate to benefit from reduced predation risk, improved foraging efficiency or individual fitness (Krause and Ruxton, 2002). At the level of the individual, associating with conspecifics can provide a number of benefits, from increased access to resources and potential mates, social learning, and information dissemination (Krützen et al., 2005; Croft et al., 2006). Drivers of association amongst conspecifics can include overlapping core ranges, relatedness, behavioral phenotype, and familiarity (Krause and Ruxton, 2002; Croft et al., 2009). At the SRMR where bull sharks temporarily aggregate because of repeated direct feeding (Brunnschweiler and Barnett, 2013), the observed network structure and perceived possible sociality may likely be an artifact of spatio-temporal overlap and not attributed to active social interactions or behavioral strategies in grouping patterns. Hence, finding non-random associations does not necessarily mean that individuals actively choose to group with preferred social partners. The SRMR is not a natural setting, namely bull sharks form artificial aggregations solely in response to feeding by the tourism operator. This is contrary to, for example, white sharks which aggregate naturally around seal colonies and are then lured to cage-diving operators through the use of chum (Schilds et al., 2019). Yet despite the artificial driver in this instance, bull sharks aggregating to exploit a food source could be considered natural, identical to ephemeral natural feeding aggregations at whale carcasses or spawning aggregations (Graham and Castellanos, 2012; Lea et al., 2019). In this regard, the SRMR being an artificial aggregation site could have laid the groundwork for natural social interactions between sharks by simply providing regular and consistent opportunities for those interactions to occur (Clua et al., 2010; Pini-Fitzsimmons et al., 2021).

The results from this study are based on the reliable long-term identification of individual bull sharks at a single site, hence our findings come with some caveats. The individual identification of sharks using distinctive markings and coloration has its limitations (Marshall and Pierce, 2012). For example, while all individuals in this study were identified using permanent identifiable features such as scars, wounds, missing and/or damaged fins, and images and video footage was collected throughout the entire study period, it is inevitable that human error in observation, identification and recording occurred. For example, depending on the uniqueness and/or obviousness of natural marks, identification of individuals can be challenging (Brunnschweiler and Barnett, 2013). Misidentification of an individual or the failure to confirm an individual which was actually present but not recorded could also be due to ocean conditions impeding visibility or the level of experience of the trained observer. In addition, we only observed bull sharks at a single feeding site; therefore, the validity of our results is limited when compared to studies that include multiple sites (e.g., an acoustic receiver network; Armansin et al., 2016; Papastamatiou et al., 2020) or those investigating multiple areas and sampling both provisioning and non-feeding sites (Mourier et al., 2012). However, the large numbers of sharks and the long study period provides a robust data set that adequately characterizes the associations of bull sharks aggregating at the SRMR, laying the foundation for further investigating sociality in this species. For example, although kinship appears not to drive associations and affiliations among blacktip reef sharks (Mourier and Planes, 2021), group assortment in bull sharks may be influenced by genetic relatedness. Unlike most other shark species, bull sharks spend the first few years of their lives in estuaries and rivers before moving out to the ocean. Consequently, further work investigating genetic relatedness of individual bull sharks encountered at the SRMR as well as those found in known nurseries (Glaus et al., 2019) is warranted in order to explore the possibility that during those first crucial years of their lives, long lasting relationships are formed within cohorts.

Direct shark feeding at the SRMR appears to drive fission-fusion dynamics, where the feeding event temporarily fuses a large number of individual bull sharks in a central place. In terms of CPF theory, of interest is bull shark behavior following dispersion from the central place (feeding event). If bull sharks indeed choose to group with preferred associates, namely turn up together at the feeding site, specific paired associations or groups of bull sharks observed at the SRMR would be expected to be observed together at other locations as well. This hypothesis could be tested by monitoring other dive sites, feeding and not, in Fiji where bull sharks are encountered. Ward-Paige et al. (2020) reported bull shark groups of variable sizes from six areas in Fiji. Individuals would need to be independently identified at these sites to ensure that they are also visitors to the SRMR and vice versa. Anecdotal reports show that several bull sharks visually identified at the SRMR were also recorded in Kuata, an island of the Yasawa Group approx. 200 km away, where a shark feeding site was established back in 2015. In the absence of direct observation data collection at each dive site, the joint movement of individuals may be monitored using acoustic telemetry. Brunnschweiler and Barnett (2013) found that the overall diel patterns in movement are for acoustically tagged bull sharks to use the area around the feeding site in the morning before spreading out over Shark Reef throughout the day and dispersing over a larger coastal area at night. Trophic information suggests that they continue to forage on natural prey (Abrantes et al., 2018), quite possibly in the Navua estuary at night (Brunnschweiler and Barnett, 2013), nevertheless we are lacking key information including regarding resting behavior (e.g., where, when or if they rest in groups). Unfortunately, our data from acoustically tagged bull sharks are inconclusive with respect to the existence of pairs or groups free ranging together as a result of small numbers of individuals tagged together for longer time periods (Brunnschweiler and Barnett, 2013). To better understand and explore the depth and breadth of sociality within the SRMR bull shark population and determine how group behavior and dynamics align with CPF, a network analysis study including focal observations combined with tracking data should be undertaken in Fiji. A tangential study could be conducted at a location such as Mozambique where adult bull sharks aggregate without being fed (Daly et al., 2014) and the results compared and contrasted. This would provide for a more thorough examination of the associative nature of this species.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethical review was not required for this study; field work was carried out in the Shark Reef Marine Reserve with the knowledge and the approval of the Ministry of Fisheries and the traditional owners of Shark Reef. No animals were caught or handled.

JB designed the study. TB, NL, and JB analyzed the data. All authors contributed to drafting the manuscript and its revisions and read and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.