Research articlesEnergetic connectivity of diverse elasmobranch populations – implications for ecological resilience
Abstract
Understanding the factors shaping patterns of ecological resilience is critical for mitigating the loss of global biodiversity. Throughout aquatic environments, highly mobile predators are thought to serve as important vectors of energy between ecosystems thereby promoting stability and resilience. However, the role these predators play in connecting food webs and promoting energy flow remains poorly understood in most contexts. Using carbon and nitrogen isotopes, we quantified the use of several prey resource pools (small oceanic forage, large oceanics, coral reef, and seagrass) by 17 species of elasmobranch fishes (n = 351 individuals) in The Bahamas to determine their functional diversity and roles as ecosystem links. We observed remarkable functional diversity across species and identified four major groups responsible for connecting discrete regions of the seascape. Elasmobranchs were responsible for promoting energetic connectivity between neritic, oceanic and deep-sea ecosystems. Our findings illustrate how mobile predators promote ecosystem connectivity, underscoring their functional significance and role in supporting ecological resilience. More broadly, strong predator conservation efforts in developing island nations, such as The Bahamas, are likely to yield ecological benefits that enhance the resilience of marine ecosystems to combat imminent threats such as habitat degradation and climate change.
1. Introduction
How ecosystems respond to human-induced stressors and disturbance regimes has the potential to alter their ecological functions [1]. Conserving the processes that promote ecological resilience in the face of these changes is, therefore, key [2]. Ecological resilience can be defined, in part, by the propensity of ecological systems to rebound following disturbances [3], and is strongly mediated by levels of biodiversity, functional complexity [4] and the diversity of energetic resource pools supporting food-webs [5]. A critical mechanism proposed to support high levels of ecological resilience is the maintenance of strong energy and nutrient flow across habitats [5,6]. This can be facilitated by passive processes, such as current- and wind-based transport of nutrients [7], or more active ecological processes, such as the movements and foraging interactions of large mobile consumers [8–10]. Predators that feed across resource channels help link energy pathways, and aid in the stability and resilience of food webs [5]. For example, Rooney et al. [5] illustrated how food-web stability is strongly mediated by predatory multi-channel feeding, whereby individuals may feed across a distinct energetic pathway with different rates of productivity and turnover.
Animals like whales and sharks that feed across multiple food webs, migrate long distances or occupy diverse habitats, connect disparate regions of the ocean [11]. Yet these widespread functional roles often make them vulnerable to individual- and population-level threats, including human impacts. While some species of marine megafauna are highly protected and are recovering (e.g. some whale and sea turtle species, [12,13]), many sharks continue to experience high levels of overexploitation resulting in population declines [14]. Given that sharks display a variety of functional roles, including top-down control of food webs and connecting trophic levels through their foraging behaviour [15], their ongoing removal is likely to have long-term effects on the structure and function of marine ecosystems [16]. This therefore remains a critical area of research.
Quantifying how mobile marine predators serve as energetic vectors is inherently challenging, owing to their cryptic lifestyles combined with a lack of techniques that can reliably resolve patterns of energy flow. However, the application of biochemical tracers, such as naturally occurring stable isotopes, can provide a non-lethal means for discerning patterns of predator-mediated energy flow among ecosystems [17]. This is because carbon isotope values of animal tissues reflect primary production pathways supporting the base of food-webs, which are isotopically distinct across major producer groups (e.g. C3 versus C4 photosynthetic versus chemoautotrophic production). Nitrogen isotopes complement carbon isotopes because of predictable, stepwise 15N fractionation between predator and prey that reflects a consumer's trophic position within food webs [17]. When combined with isotopic end members (prey species that represent a single energetic resource channel, [18]), an animal's or population's reliance on multiple energetic pathways can be discerned, provided they are isotopically distinct [17,19]. This offers a means for determining the extent to which consumers connect ecosystems themselves [9].
Among target areas for research on trophic coupling, marine seascapes within subtropical and tropical latitudes offer a valuable opportunity to understand patterns of predator-mediated ecosystem connectivity, due to the diversity of species and distinct ecosystems that exist across very small spatial scales [20]. In The Bahamas, acoustic telemetry and biologging revealed that tiger sharks connect coral reef, carbonate banks and seagrass meadows throughout their home range [21], whereby the intermediary seagrass habitats appear to serve as strategic navigation routes and connective pathways between ecosystems [22,23]. At Palmyra Atoll, blacktip reef sharks (Carcharhinus melanopterus), grey reef sharks (C. amblyrhynchos) and red snapper (Lutjanus fulvus) were shown to derive energy from lagoon, forereef and oceanic environments, with grey reef sharks connecting forereef and oceanic components and potentially linking the epipelagic and deep-water environment [9]. More broadly, similar patterns of ecosystem connectivity have been observed across other marine ecosystems such as estuaries (fishes and dolphins [24]), open ocean (whales [25]) and the deep sea (dolphins and whales [26]). However, a critical evaluation of predator-mediated connectivity has yet to be undertaken at any great scale, with most studies focusing on a relatively low number of species. This precludes a comprehensive understanding of predator-mediated ecosystem connectivity and the implications this may have for patterns of ecological resilience.
Here, we compile a large carbon and nitrogen stable isotope dataset to quantify functional diversity displayed across seventeen species of elasmobranch. Specifically, we establish (1) how species use and therefore connect major energetic resource pools and (2) key functional groups defined by interspecific resource use patterns. We use The Bahamas as a case study system, where commercial longlining was banned in 1993 and the exclusive economic zone designated a shark sanctuary in 2011. Bahamian ecosystems, therefore, provide a unique opportunity to understand the functional role of elasmobranchs in the absence of significant fishing pressure. We discuss our results within the context of ecosystem function and patterns of ecological resilience.
2. Methods
(a) Declarations
All sampling was conducted between 2009 and 2020 under research permits issued by the Bahamas Ministry of Agriculture and Marine Resources, Department of Marine Resources.
(b) Compilation of elasmobranch stable isotope values
We compiled elasmobranch carbon and nitrogen stable isotope values from several published studies [27–29], in addition to several unpublished datasets. Animals were captured using a variety of techniques including scientific longlines [30–32], drum lines [21,22,33], gillnets [34], handlines [35,36], hand nets [37] and benthic traps [28]. This included 351 individuals from 17 species sampled from around the Bahamian islands of New Providence, Cat Island, Southern Exuma, South Eleuthera, Andros and Bimini (figure 1). Deep-sea species were exclusively sampled in northeast Exuma Sound and deep waters west of North Bimini (figure 1). Specific information on animal capture, sampling and stable isotope analyses for unpublished datasets can be found in the electronic supplementary material (electronic supplementary material, appendix S1).
Figure 1. General sampling locations (white hashed lines) of elasmobranchs included in this study. Species were sampled from a variety of neritic and oceanic habitats from Bimini, New Providence, Andros, Southern Exumas, South Eleuthera and Cat Island.
Species were grouped based on their major habitat association [38]. Neritic species (i.e. those commonly associated with shallow ocean overlaying the continental shelf) included blacknose sharks (Carcharhinus acronotus), Caribbean reef sharks (C. perezi), lemon sharks (Negaprion brevirostris), nurse sharks (Ginglymostoma cirratum), bull sharks (C. leucas), southern stingrays (Hypanus americanus) and Atlantic chupare (Styracura schmardae). Oceanic/semi-oceanic species included silky sharks (C. facliformis), oceanic whitetip sharks (C. longimanus) and tiger sharks (Galeocerdo cuvier). Deep-sea associated species included Cuban dogfish (Squalus cubensis), smooth dogfish (Mustelus canis insularis), Mead's catsharks (Scyliorhinus meadi), gulper sharks (Centrophorus spp.), Atlantic sixgill sharks (Hexanchus vitulus), bluntnose sixgill sharks (H. griseus) and sharpnose sevengill sharks (Heptranchias perlo). Given similar habitat characteristics between Bahamian islands and similar prevailing biological oceanography (e.g. water temperature, primary production pathways), it was assumed that stable isotope values of elasmobranchs and primary producers would not differ significantly between islands (see [29]).
The goal of this study was to address overall, long-term patterns of resource use, therefore we only included data from tissues with longer isotopic incorporation rates that are relatively insensitive to short-term (i.e. seasonal) fluctuations in energetic resource use. These included red blood cells (herein ‘RBCs’, T95 = 41–582 days, [39,40]), white muscle (T95 = 173–1014 days, [39,41–43]) and fin tissue (T95 = 30–726 days). Because there were several species for which data were generated from multiple tissues, all isotope values were corrected for trophic discrimination prior to statistical analyses (see electronic supplementary material, appendix S1, electronic supplementary material, table S1). Given the lack of trophic discrimination factor values for elasmobranchs, we first used weighted averages from all published studies for each tissue (electronic supplementary material, appendix S1; electronic supplementary material, table S2). However, the weighted averages for muscle tissue resulted in many species falling beyond the isotopic mixing space, indicating that this value was an inappropriate estimate for many species in this study. As a result, we assigned discrimination values for muscle tissue generated for adult lemon sharks, which resulted in most species falling well within the isotopic mixing space [18,44]. We selected this species due to the close taxonomic relatedness to many species included in this study. For RBCs, muscle, and fin clips we assigned trophic discrimination factors of Δ13C = 1.88‰; Δ15N = 2.44‰, Δ13C = 1.3‰; Δ15N = 3.2‰ and Δ13C = 3.6‰; Δ15N = 0.4‰, respectively (electronic supplementary material, appendix S1, electronic supplementary material, table S1).
Lipids and urea have known isotopic effects that can alter ecological interpretation (reviewed by [17,45]). For legacy datasets, however, it is often unattainable to ensure all samples are treated in a uniform manner. However, we ensured that all datasets accounted for potential effects of lipids and/or urea in some manner, either through chemical treatment or mathematical normalization (for lipids only) where C:N was high (following DI water rinses to remove urea; [46]). Chemical treatment was not conducted on muscle tissue of a single bluntnose sixgill shark, however the C:N ratio and bulk isotope values fell within the range of lipid-extracted samples for closely related deep-water species [47,48].
(c) Elasmobranch-mediated energy flow
We used stable isotope mixing models to evaluate how elasmobranchs connect ecosystems through energetic resource use (per [9]). Thus, we adopted a well-established isotopic ‘end-member’ approach [9,18,49] to evaluate the reliance of elasmobranchs on four major resource pools – oceanic forage, large oceanic predators (herein ‘large oceanics’), coral reefs, and seagrass. The isotopic mixing space was constructed from resampling published means and s.d.s (n = 100) of key prey species [50] sampled from The Bahamas that are known to derive most energy from a single resource pool. Prey isotope values spanned several geographical areas including Eleuthera [51–53], Cat Island [27], Bimini [54], Inagua and Long Island [55]. Oceanic forage species (δ13C = 18.0‰ ± 1.2‰, δ15N = 5.3‰ ± 0.7‰) included squid (unidentified spp.) and flying fish (family Exocoetidae). Large oceanics (δ13C = −16.7‰ ± 0.6‰, δ15N = 8.6‰ ± 1.2‰) were composed of offshore piscivorous fishes including wahoo (Acanthocybium solandri), dolphinfish (Coryphaena hippurus), blackfin tuna (Thunnus atlanticus), skipjack tuna (Katsuwonus pelamis) and yellowfin tuna (Thunnus albacares). Coral reef endmembers (δ13C = −12.5‰ ± 1.7‰, δ15N = 8.3‰ ± 0.4‰) included mostly piscivorous reef fishes that exhibit high fidelity to large coral reefs, including bar jack (Caranx ruber), graysby grouper (Cephalopholis cruentata), Nassau grouper (Epinephelus striatus), schoolmaster snapper (Lutjanus apodus), yellowtail snapper (Ocyurus chrysurus) and lane snapper (Lutjanus synagris). Seagrass endmembers (δ13C = −8.4‰ ± 1.9‰, δ15N = 5.2‰ ± 3.0‰) included several populations of green sea turtles (Chelonia mydas) and juvenile and adult bonefish (Albula vulpes). Mean δ13C and δ15N (±s.d.) for all endmember species can be found in the electronic supplementary material (electronic supplementary material, appendix 1; electronic supplementary material, table S2).
Bayesian mixing models were implemented in the R package MixSIAR [56], which offers greater flexibility than traditional mixing models by allowing for the inclusion of model covariates. Here, we ran species-specific mixing models which included a random effect of individual to quantify intraspecific variation in resource use. These were compared to a null model (i.e. no covariates) that assumes a uniform resource use distribution across all individuals using leave one out cross validation and Akaike weights [56]. Models were run for 100 000 iterations across three Markov chains with a burn-in of 50 000 and thinning interval of 50. Model convergence was assessed via Gelman-Rubin diagnostics and Geweke statistics [56]. We also assessed multiplicative error terms to ensure carbon and nitrogen isotope data were sufficiently explained by model covariates [56]. For sharpnose sevengill sharks, two energy channels were removed from the mixing model to improve model convergence. This was justified given that this species fell extremely close to a single endmember (large oceanics) [18].
To further evaluate the degree to which different species connected multiple resource pools we calculated Pielou's evenness [57]. This was calculated using the proportional contributions of each resource pool to species diet and provides values between 0 and 1, where = 0 low evenness (i.e. a species relies solely upon a single resource channel), and 1 = high evenness (i.e. a species relies similarly on all available resource pools). Species exhibiting high evenness values are assumed to be greater ecosystem connectors.
(d) Functional diversity of elasmobranchs
We adopted hierarchical K-means clustering using the R package NbClust [58] to determine the diversity of functional roles displayed by elasmobranch species. Median estimates of energetic resource use were calculated from all individuals, for each species and normalized to Euclidean distances prior to clustering. The most appropriate number of clusters was explored using mean within sum-of-squares (WSS) and average silhouette width [58]. The final cluster designation was based on the results of pre-processing algorithms combined with visual scrutiny of the resulting dendrogram. We removed species with low sample sizes (n = 1 or 2) from clustering, which included the bluntnose sixgill shark, sharpnose sevengill shark and Mead's catshark.
Finally, we evaluated isotopic niche area for each functional group (as inferred through hierarchical clustering) as a proxy for trophic niche variation [59]. We calculated estimates of niche area and proportional pairwise overlap using 95% kernel utilization distributions (KUDs) in the R package rKIN [59,60].
3. Results
Carbon and nitrogen stable isotope values were compiled for 17 elasmobranch species (n = 351 individuals) that associate with neritic, oceanic/semi-oceanic and deep-water ecosystems (table 1). Across sampled elasmobranchs, raw δ13C and δ15N values (i.e. prior to TDF correction) ranged from −17.1‰ (silky shark) to −5.4‰ (nurse shark) and 2.5‰ (Atlantic chupare) to 14.4‰ (bull shark), respectively (figure 2). The greatest ranges of δ13C and δ15N values (where n > 3) were observed for nurse sharks (δ13C = −14.8‰ to −5.4‰) and bull sharks (δ15N = 10.0‰ to 14.4‰), and the lowest ranges were observed for gulper sharks (δ13C = −16.4‰ to −15.8‰; δ15N = 10.3‰ to 11.0‰) (figure 2).
Figure 2. Mean δ13C and δ15N values (± 1 s.d.) adjusted for trophic discrimination for 17 species of elasmobranch spanning neritic (triangles), oceanic/semi-oceanic (squares), and deep-water (circles) associated habitats. Values are shown relative to four major prey resource pools: large oceanics, oceanic forage, coral reefs, and seagrass/banks.
| scientific name | common name | habitat | total length (cm) | n | δ13C | δ13CTDF | δ15N | δ15NTDF | reference |
|---|---|---|---|---|---|---|---|---|---|
| Carcharhinus acronotus | blacknose shark | neritic | 103–119 | 9 | −10.7 | −12.2 (0.8) | 9.6 | 6.7 (0.5) | this study |
| Carcharhinus leucas | bull shark | neritic | 227–278 | 7 | −13.8 | −15.2 (1.3) | 13.1 | 10.0 (1.3) | this study |
| Carcharhinus perezi | Caribbean reef shark | neritic | 81–190 | 69 | −11.6 | −13.5 (1.0) | 9.0 | 6.5 (0.7) | this study |
| Ginglymostoma cirratum | nurse shark | neritic | 113–246 | 53 | −9.9 | −11.4 (2.6) | 8.9 | 6.1 (0.9) | [29]; this study |
| Styracura schmardae | Atlantic chupare | neritic | 25–120 | 20 | −9.6 | −13.3 (1.3) | 4.8 | 4.4 (1.1) | [29] |
| Hypanus americanus | southern stingray | neritic | 45–102 | 20 | −8.1 | −11.8 (0.7) | 7.0 | 6.6 (1.1) | [29] |
| Negaprion brevirostris | lemon shark | neritic | 65–269 | 46 | −12.4 | −14.2 (1.5) | 8.8 | 6.3 (0.7) | this study |
| Carcharhinus falciformis | silky shark | oceanic/semi-oceanic | 106–212 | 13 | −16.0 | −17.3 (0.9) | 9.7 | 6.5 (0.7) | this study |
| Carcharhinus longimanus | oceanic whitetip shark | oceanic/semi-oceanic | 194–307 | 38 | −15.5 | −17.2 (0.2) | 9.4 | 6.5 (0.7) | [27]; this study |
| Galeocerdo cuvier | tiger shark | oceanic/semi-oceanic | 155–304 | 26 | −12.7 | −14.5 (1.6) | 10.1 | 7.5 (1.2) | this study |
| Centrophorus spp. | gulper shark | deep sea | 70–104 | 9 | −16.1 | −17.4 (0.2) | 10.6 | 7.5 (0.2) | [28] |
| Heptranchias perlo | sharpnose sevengill shark | deep sea | 63–87 | 2 | −16.5 | −17.8 (0.5) | 12.8 | 9.6 (0.6) | [28] |
| Hexanchus vitulus | Atlantic sixgill shark | deep sea | 120–151 | 4 | −13.8 | −15.1 (0.6) | 11.2 | 8.1 (0.0) | [28] |
| Hexanchus griseus | bluntnose sixgill shark | deep sea | > 300 | 1 | −12.1 | −14.0 | 8.9 | 6.5 | this study |
| Mustelus canis insularis | smooth dogfish | deep sea | 78–107 | 11 | −11.7 | −13.9 (1.5) | 9.1 | 7.0 (1.2) | [28]; this study |
| Scyliorhinus meadi | Mead's catshark | deep sea | 50 | 1 | −14.1 | −15.4 | 8.0 | 4.9 | [28] |
| Squalus cubensis | Cuban dogfish | deep sea | 39–77 | 22 | −14.1 | −15.4 (0.7) | 10.0 | 6.8 (0.7) | [28] |
(a) Elasmobranch-mediated energy flow
Bayesian isotope mixing models revealed that elasmobranch species were generally reliant upon a mixture of energetic resource channels (figure 3). High resource use diversity was observed for many neritic species including blacknose, nurse and Caribbean reef sharks (figure 3). Most oceanic species were heavily reliant upon large oceanic and oceanic forage species, with little use of neritic resource channels (e.g. coral reefs and seagrass). Several deep-water species, such as the Atlantic sixgill and smooth dogfish exhibited use of both neritic and oceanic prey resources (figures 2 and 3). We observed the highest intraspecific variation in resource use by lemon sharks, nurse sharks, and Cuban dogfish (figure 3). Uniform resource use (i.e. consistent reliance upon the same resource pools across individuals) was observed across many neritic, oceanic and deep-water species, particularly oceanic whitetip sharks, silky sharks and tiger sharks (figure 3). Across the fourteen species included in hierarchical cluster analysis, Pielou's evenness values were generally high and ranged from 0.55 (gulper shark) to 0.99 (smooth dogfish) (figure 4).
Figure 3. Relative use of large oceanic (LO), oceanic forage (OF), coral reef (CR), and seagrass/banks (SB) energetic resource pools by 17 species of elasmobranchs from The Bahamas (n = 351). Colours represent primary habitat associations where green = neritic, blue = oceanic/semi-oceanic, and grey = deep-sea. Boxplots comprise median estimates from each individual, for each species, derived from 100 000 posterior draws. Figure 4. Pielou's evenness for fourteen species of elasmobranchs sampled from The Bahamas. High evenness values indicate even use of available resource pools, whereas low values indicate dominant use of a single resource pool. Colours and symbols indicate functional groups derived from Bayesian mixing models and hierarchical clustering: neritic ecosystem connectors (green circles), neritic–oceanic connectors (purple triangles), oceanic omnivores (blue crosses) and higher oceanic predators (black squares).
(b) Functional diversity of elasmobranchs
Hierarchical cluster analysis revealed at least four distinct species clusters based on the average use of available prey resource channels (figure 5; electronic supplementary material, appendix S1 and figure S1). The first cluster was defined as neritic ecosystem connectors (herein, NECs) and comprised neritic species, including blacknose sharks, nurse sharks, southern stingrays and Atlantic chupares, which derived most energy from seagrass and coral reef prey (figure 5). The second cluster, neritic–oceanic connectors (herein, NOCs), included neritic, oceanic/semi-oceanic and deep-sea associated species that use all available resource channels. NOCs included Cuban dogfish, smooth dogfish, Caribbean reef sharks and lemon sharks. A third cluster, oceanic omnivores (i.e. trophic omnivores; herein, OOs), included oceanic whitetip sharks, silky sharks and gulper sharks. Oceanic omnivores displayed a mixed reliance on oceanic forage and large oceanic resource channels, thereby connecting lower and upper components of oceanic food webs. The final group was defined as higher oceanic predators (herein, HOPs) and comprised species displaying a strong reliance on the large oceanic resource channel. This group comprised tiger sharks, Atlantic sixgill sharks and bull sharks.
Figure 5. Left: Functional groupings of 14 elasmobranch species based on their relative use of energetic resource pools. Four major functional clusters were identified including neritic ecosystem connectors (green), neritic–oceanic connectors (purple), oceanic omnivores (dark blue) and higher oceanic predators (dark grey). Note: bluntnose sixgill sharks, sharpnose sevengill sharks and Mead's catsharks were not included due to low sample sizes (n = 1 or 2). Right: Isotopic niche width of functional groupings were calculated using 95% kernel utilization distributions (KUDs). Isotope values are corrected for tissue-specific trophic discrimination.
The largest niche area was observed for NECs (41.8‰2), followed by HOPs (41.3‰2), NOCs (19.4‰2) and OOs (4.9‰2) (table 2, figure 5). Pairwise isotopic overlap was greatest between NOCs and NECs (greater than 50%), OOs and HOPs (greater than 70%) (table 3), and NECs and HOPs (greater than 78%). Moderate niche overlap was observed between NECs and NOCs, NECs and HOPs, OOs and NOCs, HOPs and NECs, and HOPs and NOCs (27–34%). Low niche overlap was observed between NECs and OOs (1% – 7%), OOs and NECs, and HOPs and OOs (less than 10% (table 3, figure 5).
| functional group | species composition | nindividuals | niche area (‰2) |
|---|---|---|---|
| neritic ecosystem connectors (NECs) | blacknose shark, nurse shark, southern stingray, Chupare stingray | 102 | 41.8 |
| neritic–oceanic connectors (NOCs) | lemon shark, Caribbean reef shark, Cuban dogfish, smooth dogfish, | 148 | 19.4 |
| oceanic omnivores (OOs) | oceanic whitetip shark, silky shark, gulper shark | 60 | 4.9 |
| higher oceanic predators (HOPs) | Atlantic sixgill shark, tiger shark, bull shark | 37 | 41.3 |
| functional group | neritic ecosystem connectors | neritic–oceanic connectors | oceanic omnivores | higher oceanic predators |
|---|---|---|---|---|
| neritic ecosystem connectors | — | 0.312 | 0.004 | 0.332 |
| neritic–oceanic connectors | 0.67 | — | 0.070 | 0.786 |
| oceanic omnivores | 0.036 | 0.277 | — | 0.725 |
| higher oceanic predators | 0.336 | 0.369 | 0.086 | — |
4. Discussion
As climate change and human impacts continue to threaten ecosystem stability, management and conservation groups in many regions have been increasingly forced to set priorities, such as the increased expansion of marine protected areas [63]. As such, some habitats, like coral reefs, have been given more attention than others despite the inherent connectivity between ecosystems. Here we provide an evaluation of predator-mediated ecosystem connectivity from The Bahamas archipelago, highlighting the functional connections facilitated by sharks and rays between disparate habitats. We observed functional diversity within and across elasmobranch species which were found to energetically couple neritic and oceanic waters, including the deep-sea. Patterns of evenness suggested that most species are rarely reliant on a single resource pool, and therefore connect at least two different ecosystems in space and time. These findings underscore the value of predator biodiversity in facilitating the conditions required to ensure ecological resilience, specifically energy and nutrient flow across habitats [3,64].
(a) Ecosystem connectivity and functional diversity of Bahamian elasmobranchs
Neritic ecosystem connectors (NECs) comprised species that facilitated energy flow between seagrass and coral reef environments and included blacknose sharks, nurse sharks and two larger-bodied stingray species. These findings align with the known habitat use of these species and their physical mobility between neritic regions of the seascape [37], which comprise a large portion of The Bahamas Exclusive Economic Zone [21,22]. Individuals within this functional group were not homogeneous, with NECs exhibiting the broadest isotopic niche and significant overlap with other functional groups. This suggests considerable variation among individuals—a trend that was particularly evident within nurse sharks. Such variation indicates that not all sharks and rays in this functional group link coral reef and seagrass habitats equally, and individual differences may buffer populations from disturbance events, such as habitat degradation [65], climate change [66] and species invasions [67]. We thus draw two major conclusions from NECs: (1) this functional group is critical for facilitating energetic transfer between seagrass and coral reef environments, and (2) species within this group may be buffered from environmental disturbance due to the use of multiple prey resource pools [5] provided energetic needs can be adequately sustained by remaining undisturbed pools.
The second functional group, neritic–oceanic connectors (NOCs) comprised species that predominantly relied upon coral reef and oceanic-derived energy (figure 5), which is supported by existing knowledge of vertical movement behaviors (reviewed by [68]). Caribbean reef sharks relied upon energy derived from seagrass, coral reef and oceanic forage species, reflecting their movements across ecotones [21–23,29]. Their reliance on oceanic forage species may also, in part, explain the functionality of commonly observed deep-diving behaviours [62,68,69]. Several deep-sea elasmobranchs, namely Cuban dogfish and smooth dogfish were assigned to this group, because both species were reliant upon oceanic and coral reef-derived energy. In regions such as the Exuma Sound, these dogfishes typically occupy depths of 300–900 m along the slope [31,61], which probably receives significant inputs of carbonate matter washed off shallow banks and algal-covered reef walls. Based on vertical habitat preferences [31,61], it is plausible that both species may provide a connection between the neritic extent of the slope (abutting the near-vertical reef wall) and deep benthic environments, which could be particularly important for deep-sea habitats that typically exhibit low levels of productivity. Our findings here suggest that this connectivity may, in part, be mediated by the vertical movements of both large- and smaller-bodied sharks within this functional group.
Oceanic omnivores (OOs) such as oceanic whitetip sharks and silky sharks relied upon a mix of oceanic forage and larger oceanic prey [27]. Despite their large body sizes, these findings suggest that these species may not constitute true apex predators in offshore, oceanic ecosystems of The Bahamas, but rather feed across multiple trophic levels. This is certainly true for silky sharks, for which data were only available for juvenile and subadult individuals. This functional group exhibited the narrowest isotopic niche and lowest overlap with other functional groups, which suggests a uniform functional role among individuals that is largely divergent from other sampled taxa. These species likely serve as connectors of lower and upper trophic levels within oceanic food webs. For oceanic whitetip sharks, which routinely perform mesopelagic excursions [35,36,70], individuals may facilitate energetic connectivity between epi- and mesopelagic zones.
Higher oceanic predators (HOPs), the fourth functional group, comprised bull sharks, tiger sharks, and Atlantic sixgill sharks, which exhibited a strong reliance on large oceanic prey. For Atlantic sixgill sharks, it is likely that cluster designations may in part, represent opportunistic reliance upon larger sinking carrion such as whales and larger fishes, which are commonly observed in deep-water environments of the eastern Bahamas [71], rather than active hunting of large pelagic fishes. A surprising assignment was that of bull sharks to this group, which commonly traverse neritic ecosystems [72]. This observation could be explained by their common occurrence in Bahamian marinas where they are highly provisioned with offshore offal (O.N.S. 2023, personal communication), or by known movements across the oceanic environment of the Florida straits during migrations between The Bahamas and Florida [73].
From an energy flow perspective, it remains unclear the extent to which these species connect regions of the oceanic food web, given the lack of empirical data on vertical habitat preference and movement trajectories. Based on the high δ15N values of HOPs, it could be inferred that these species occupy apex positions throughout oceanic food webs, but the moderate body sizes of individuals sampled (TL = 120–304 cm), suggest that this is unlikely. Instead, we predict that bluntnose sixgill sharks, deep-diving great hammerhead sharks (Sphyrna mokarran), and toothed whales may assume this role for the deeper oceanic environment in The Bahamas [31]. However, we did not have sufficient data to support this inference within the confines of this study. Further, these regions also support a variety of large, deep-diving toothed whales [74] that may fill this ecological role. Despite assessing 14 different species, expanding the taxonomic resolution of mobile predators studied in the context of energy flow is clearly required for a more holistic understanding of ecosystem structure and function. For Bahamian elasmobranchs, this may include greater attention on the collection of samples from species such as blacktip sharks (C. limbatus), great hammerhead sharks (Sphyrna mokarran), bonnethead sharks (Sphyrna tiburo) and yellow stingrays (Urobatis jamaicensis).
When considering deep water elasmobranchs in the fourth functional group, physiological processes such as tissue catabolism could be responsible for high δ15N values [75]. Similarly, the nitrogen physiology of deep-water sharks is unique relative to other vertebrates, whereby individuals must balance concentrations of urea and trimethylamine n-oxide to maintain osmotic balance under high hydrostatic pressures [76]. If this requires alteration to overall nitrogen balance [77], such as the mobilization of nitrogen from structural tissues to synthesize urea and/or TMAO, this may leave the residual δ15N values of tissues 15N-enriched. While the δ15N patterns of deep-sea elasmobranchs may be linked to both physiological and ecological processes [48], low δ13C values for species such as Atlantic sixgill and sharpnose sevengill sharks indicate an exclusive reliance upon oceanic resource pools, suggesting these species are unlikely to be critical connectors of ecosystems.
(b) Implications for connectivity and resilience
Although these findings have specific implications for Bahamian seascapes (as outlined above), they are highly relevant when considering the broader role of marine predators in terms of ecosystem connectivity and resilience. Our results highlight how, when faced with diverse energetic resource pools, mobile predators are rarely reliant upon energy derived from a single energetic pathway, as reflected in consistently high Pielou's evenness values (all > 0.5). These observations agree with those reported across a variety of marine, freshwater and terrestrial settings and are strongly tied to theories of ecosystem structural asymmetry and stability [78]. For example, previous work has suggested that the use of phytoplanktonic and detrital energy channels by predators, diversifies ecosystem linkages and can increase overall stability [5,6]. Similarly, the ecological coupling of marine and freshwater food webs by highly mobile predators may increase the resilience of estuarine systems that frequently experience extreme weather events (e.g. [24]). This is due to disparity of productivity and energetic turnover between distinct energetic pathways, which buffers against long-term perturbations. Although the overall rate of energetic turnover has yet to be determined for energy channels supporting many sub-tropical ecosystems, this offers a challenging, albeit critical avenue for further work. Similarly, exploring how predators may shift resource acquisition in response to environmental and anthropogenic cues is required to fully appreciate the mechanisms driving stability [79]. Such information is essential for predicting the long-term, ecosystem-wide effects of chronic stressors, such as climate change.
(c) Caveats and considerations
While previous research on our study species supports the inferences we draw from our results, we must acknowledge several important assumptions. We acknowledge that for many species, individuals may display ontogenetic shifts in diet, which may explain the relatively high individual-level variability seen within lemon sharks, nurse sharks and Cuban dogfish (figure 2). The goal of this study, however, was to provide a general species-level overview of energetic resource use. For many species (i.e. deep-sea and oceanic/semi-oceanic), the use of specific primary production channels is unlikely to change across ontogeny. Future studies should aim to investigate how the functional role of different species changes across size, to develop conservation strategies for specific life-history stages. Second, we recognize that several shark species did not fall directly within the isotopic mixing space [18,80], including bull sharks and sharpnose sevengill sharks (figure 2). However, isotope values of most individuals were extremely close to a single end member, promoting high confidence that individuals use a specific energetic resource channel. This observation partly reflects challenges associated with the use of a unified mixing space for seventeen different species comprising data generated from multiple tissues and associated TDFs, which remain poorly defined for elasmobranchs. Moreover, this pattern could be driven by variation in isotopic baselines between sampling locations, most likely due to localized nutrient dynamics [81]. If this were the case, baseline δ15N values would be most impacted among islands of varying population densities with notable differences in run-off and sewage treatment. While we are not able to explicitly explore differences in isotopic baselines among islands using the data presented, the relatively low standard deviations of most elasmobranchs and literature-derived prey stable isotope values used to derive the mixing space, indicate these effects are relatively small. Work by Shipley et al. [29] provides the only known compilation of baseline isotope values for the wider Bahamas, which suggested relatively low spatial variation, at least for some primary producers referenced in this study (e.g. seagrass). Further work exploring baseline differences among islands would greatly improve knowledge of local nutrient dynamics and how this impacts isotopic baseline composition, which provides a useful avenue for future work. Finally, we acknowledge the relatively low sample sizes for some species, which pertain largely to species captured from deep-sea environments, such as bluntnose sixgill shark, Atlantic sixgill shark and Mead's catshark. This reflects the logistical challenges associated with sampling deep-water species in remote regions of The Bahamas; this gap will be filled by focused sampling of deep-water species in the future.
5. Conclusion
Across The Bahamas, the assemblage of sub-tropical elasmobranch species occupies a diversity of functional roles that facilitate energy flow across discrete components of the seascape. The observed energetic connectivity between disparate ecosystems, such as connecting neritic waters and the deep-sea are considered to increase overall stability through asynchronous ecosystem effects (see [5]). Marine protected areas, such as The Bahamas shark sanctuary, will have ecological spillover effects extending well beyond habitat and ecosystem boundaries due to the mobile nature of many elasmobranch species [21,22,28,35,36,47,48,61,62,82]. Successful predator conservation efforts in island nations, including the shark sanctuary of The Bahamas, will potentially yield ecological benefits for ecosystem structure, function and resilience, which may catalyze the broader adoption of similar measures across other countries. This is imperative, given many small developing, low-lying nations face continued threats from overfishing [83] combined with the severe effects of global climate change [84], which is likely to intensify throughout much of the Anthropocene. However, despite conservation efforts, many large mobile predators are still being removed from ocean ecosystems in vast quantities [85,86]. Our findings strongly suggest that large-scale removal of predators may decouple intrinsic energetic linkages that are critical for long-term stability.
Ethics
Research was conducted under permits granted from The Bahamas Ministry of Agriculture and Marine Resources (including MAF/FIS/9, MAF/FIS/17 and MAF/FIS/34).
Data accessibility
All data and code supporting this submission can be accessed through the electronic supplementary material.
The data are provided in electronic supplementary material [87].
Authors' contributions
O.N.S.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, software, supervision, validation, visualization, writing—original draft, writing—review and editing; P.M.: conceptualization, data curation, formal analysis, investigation, methodology, writing—original draft, writing—review and editing; N.E.H.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, writing—original draft, writing—review and editing; A.M.L.B.: data curation, funding acquisition, investigation, resources, writing—review and editing; D.C.: data curation, funding acquisition, investigation, methodology, resources, writing—review and editing; M.G.F.: data curation, funding acquisition, investigation, resources, supervision, writing—review and editing; A.E.G.: data curation, funding acquisition, resources, writing—review and editing; T.L.G.: conceptualization, data curation, funding acquisition, investigation, resources, validation, writing—review and editing; L.A.H.: data curation, funding acquisition, investigation, resources, validation, writing—review and editing; S.K.: data curation, investigation, project administration, resources, validation, writing—review and editing; D.J.M.: data curation, funding acquisition, investigation, methodology, resources, validation, writing—review and editing; O.O.: data curation, funding acquisition, resources, validation, writing—review and editing; N.V.P.: data curation, funding acquisition, investigation, resources, validation, writing—review and editing; M.P.: data curation, funding acquisition, investigation, resources, writing—review and editing; M.J.S.: data curation, funding acquisition, investigation, resources, validation, writing—review and editing; E.V.C.S.: data curation, funding acquisition, investigation, resources, validation, writing—review and editing; B.D.S.: data curation, investigation, resources, validation, writing—review and editing; B.S.T.: data curation, formal analysis, investigation, resources, validation, writing—review and editing; M.W.: data curation, investigation, resources, writing—review and editing; E.J.B.: data curation, funding acquisition, project administration, resources; A.J.G.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, validation, writing—original draft, writing—review and editing.
All authors gave final approval for publication and agreed to be held accountable for the work performed therein.
Conflict of interest declaration
We declare we have no competing interests.
Funding
This work was supported by anonymous philanthropic grants.
Acknowledgements
We would like to thank the countless volunteers that supported fieldwork, sample collection and laboratory analyses. Special thanks to G. Henkes, V. Atudorei and L. Burkamper for supporting isotope analyses.
