Observations of Shelling Behavior and Potential Evidence of Vertical Social Learning by Bottlenose Dolphins in the Great Sandy Marine Park, Queensland Australia

Abstract

Animal tool use has long been of interest to biologists, anthropologists, and psychologists due to the potentially important ecological, cognitive, cultural, and evolutionary consequences (Mann and Patterson 2013; Lancaster 1968; Gibson and Ingold 1995; Matsuzawa 2001). Among dolphins, tool use has been well established in Shark Bay (Malgana country), Western Australia where two different foraging tactics have been classified as tool use. These include: (i) sponge foraging (hereafter sponging), where a subset of Indo-Pacific bottlenose dolphins (Tursiops aduncus) use marine basket sponges to protect their rostra during foraging on benthic prey (Smolker et al. 1997; Patterson and Mann 2011; Krützen et al. 2005), and (ii) shell foraging (hereafter shelling), where some individuals feed on fish hiding inside large (dead) marine gastropod shells by lifting the shells into the air and subsequently draining the prey (Wild et al. 2020; Mann and Patterson 2013; Allen et al. 2011).

Shelling is rare and thought to occur on an opportunistic basis due to utilizing large dead gastropod shells on the sea floor (Wild et al. 2020; Allen et al. 2011). In Shark Bay, it has been observed in < 2% of the population, however, is believed to be under reported due to the short nature of the foraging strategy (Wild et al. 2020). Despite the potential opportunistic nature of shelling, the behavior requires specialized skills to properly handle the shell as a tool (Allen et al. 2011), potentially reducing intraspecific competition by exploiting a situational novel niche, broadening foraging opportunities and reducing overlap in resources. It is also suspected that the same level of specialization is not needed for shelling as would be for a life-long specialized strategy like sponging. This is due to the situation-dependent nature of the tactic and shelling being primarily socially transmitted non-vertically outside of the mother-calf bond (considered a more flexible learning strategy) (Wild et al. 2020). Yet, evidence for asocial learning (i.e., independent innovation) of shelling is also plausible (Wild et al. 2020; Allen et al. 2011).

Only one study has assessed the transmission pathways of shelling and no observations of transmission have occurred in the wild to date. Wild et al. (2020) ran model simulations and found that diffusion through the social network was most supported, suggesting that shelling was 10 to > 300 times more likely to occur through social transmission than asocial learning or genetic pathways. Nevertheless, shelling has only been observed in individuals from 2 (out of 8) unique haplotypes and learning in these haplotypes was estimated to be ~50 times faster than those of other haplotypes, suggesting potential for genetic/maternal transmission to be important (Wild et al. 2020).

To date, shelling has only been observed with published literature documenting its occurrence in Shark Bay, Western Australia (Allen et al. 2011; Wild et al. 2020). Despite this, there is anecdotal evidence of shelling occurring in southeast Queensland, Australia (Allen et al. 2011). However, there are no images, publications, or information to accompany this anecdote. As such, we present the first published records of shelling observed outside Shark Bay, Western Australia. Two separate sightings were recorded in 2025 within the Great Sandy Marine Park (GSMP), Queensland, Australia during routine cetacean surveys conducted as part of a larger research project in the area. The GSMP (25.35^oS, 152.95^oE) (Butchulla country) includes Hervey Bay, the Great Sandy Straits (Korrawinga), and Tin Can Inlet. Hervey Bay, where the observations occurred (Figure 1), is a large coastal embayment protected and constrained by mainland Australia to the west and the sand island K'gari to the east and includes shallow coral reefs, intertidal wetlands, mudflats, seagrass beds, shifting sandbars, and numerous estuaries and rivers flowing into the system (Gräwe et al. 2009).

Opportunistic boat-based surveys in the GSMP have been consistently undertaken since 2022. Upon sighting a dolphin, the research vessel closely approached the individual(s) to minimize behavioral disturbance but maximize photo identification possibilities. Dorsal fins were photographed using either a Nikon D7500 or Canon 90D DSLR camera with a 100–400 mm telephoto lens. During a survey, dolphin species, individual ID, sex, location (via GPS), water depth, substrate (if known), sea state, water temperature, group size, behavioral state, and life history information (i.e., age-class, reproductive status) were recorded. Sex was determined by views of the genital area or by consistent association with a calf. A group was defined using the 10 m chain rule (Smolker et al. 1992).

On 6 July and 1 October 2025 researchers observed two separate shelling events involving three different bottlenose dolphins (“buthu” in Butchulla language). The first event involved a single adult female (Stevie Nicks) who upon initial sighting (0749 h) was foraging (tail-out-peduncle-dive foraging, (Sargeant and Mann 2009)) without evidence of a shell. As the research vessel approached, at 0750 h this individual was using her rostrum to position a bailer shell (Melo amphora) at the water's surface. She proceeded to do three surfacings in multiple directions maneuvering the shell at the surface and lifting the entire shell out of the water (Figure 2). On her next surfacing she had dropped the shell. No other dolphins were with this individual, with the next closest sighting occurring 0.51 km away. This event occurred 3.9 km north of Urangan (−25.2934°S, 152.8938°E) (Figure 1) at a depth of 6.1 m, a sea surface temperature of 16.2°C, a calm sea state (Beaufort 0), and overcast/raining conditions. Unfortunately, due to the rain we could not continue to follow this individual, resulting in a 9-min observation without confirmation of picking up the shell again nor the outcome of whether a fish was drained from the shell (i.e., if the shelling event was successful).

The second observation, on 1 October 2025, involved a mother and calf pair (Maserati and Bentley respectively). The observation occurred at 0829 h, 4.15 km north of Urangan in the same area as our first observation (Figure 1) at a depth of 5.4 m, a sea surface temperature of 23.7°C, a calm sea state (Beaufort 0), and overcast conditions. Upon initial sighting, the adult female was foraging but no shell was observed. Similar to the first observation, no other dolphins were within the vicinity of the mom-calf pair (next closest observation was 0.57 km away). The adult female surfaced with a bailer shell at 0901 h while an unmanned aerial vehicle (UAV) was flying overhead. Video footage (Video S1) showed the female surfacing multiple times with the shell. A fish (species ID cannot be determined from the video) can be seen escaping from the shell immediately followed by the adult female dropping the bailer shell and pursuing the fish. It is unconfirmed whether the fish was caught, as the dolphin submerged while chasing the fish and water visibility was poor. The adult female surfaced multiple times again, never regaining the shell. Three minutes after the mother dropped the shell, the calf was observed surfacing with the shell (0904 h) in the exact same location. The calf can be seen holding the shell in the same way as its mother (rostrum in the bailer shell opening, Figure 3a, Video S2) and was observed slapping the shell against the water's surface (Video S2). UAV footage did not capture prey exiting the shell at any point, but camera footage does show the body/tail of a suspected lizardfish (Synodus sp.) extending from the aperture of the bailer shell while raised above the water's surface (Figure 3b). After a few surfacings with the shell, the calf dropped the bailer (potentially to chase the fish or due to inexperience or lack of efficiency with the required skillset) and resurfaced without the shell. Approximately 30 s later (0905 h) the calf resurfaced with the bailer shell again but dropped it within 7 s and began swimming away (Video S2). Similar to the first event, it is unconfirmed if either the mother or calf were successful at foraging during these events. Throughout the second shelling event the mom and calf were often within or just over 10 m of one another, but the calf did not move into infant/baby position until all shelling finalized. This second observation of the two individuals lasted 29 min.

In addition to the research observations in 2025, two observations (Figure 4) have occurred over the last 12 years on the ecotour vessel Blue Dolphin of Blue Dolphin Marine Tours (personal communication P. Lynch and C. Smith). During these ecotour trips a dedicated photographer has been onboard and captured photographic evidence of each additional sighting (Figure 4). Although specific GPS data are lacking from these observations, the two additional observations occurred within the same area as the events from 2025 and appeared to be foraging in nature (personal communication P. Lynch and C. Smith).

Our work provides the first evidence of shelling (i.e., tool use) outside of Shark Bay, Western Australia contributing evidence of independent innovation of this behavior across the West and East coast of Australia. These findings support the highly intelligent, cognitively complex, and innovative nature of dolphins that has been well studied to date (Marino et al. 2007; Meulman et al. 2013). It is also likely that shelling is underreported in the GSMP and occurs more frequently but has been overlooked due to limited research efforts in the region and the short nature of this behavior (typically occurring in bouts lasting only a few seconds).

The observations presented here are suggestive of vertical (maternal) social transmission between mother and offspring through observational learning, during which a calf has picked up and attempted to shell immediately following the mother performing this behavior. Despite our limited sample size, this suggests an element of vertical (mother to calf) social transmission/social learning that has not previously been documented. Although Wild et al. (2020) found support for non-vertical social transmission outside of the mother-calf bond (i.e., horizontally from peer to peer or obliquely from non-maternal elders to younger individuals), our second observation is suggestive of a “do-as-mother-does” strategy previously reported in other specialized foraging behaviors (Mann and Sargeant 2003). Prolonged dependency and maternal care in dolphins (Mann et al. 2000; Mann 2019) facilitate ample opportunities for vertical social transmission observed here and in the transmission of other behaviors in dolphins (Mann and Sargeant 2003; Rendell et al. 2019). Additionally, vertical social transmission of foraging skills can be advantageous in heterogenous marine environments with patchy prey that is indefensible, ultimately reducing competition on an intra-population level, benefiting the offspring and modifying the inclusive fitness landscape to competition mainly among kin (Krützen et al. 2014). As such, it is not surprising that maternal (vertical) transmission would be a common pathway for the transmission of shelling. It is possible multiple pathways are responsible for the transmission of shelling, and some evidence of this can be found in Wild et al. (2020) (i.e., evidence for social and asocial learning as well as the propensity of dolphins with specific haplotypes to exhibit shelling). Additionally, due to limited observations within this study and Shark Bay, as well as the likelihood of undetected events/individuals, it is also plausible that we have yet to properly tease out the true mechanisms of transmission for this behavior. Ongoing observations culminating in larger datasets over time would benefit further analyses.

Within our study, shelling was observed exclusively among adult females and one of their offspring (unknown sex). This potential sex bias observed among adults (albeit from a very small sample size) is interesting as this differed from Wild et al. (2020), where a lack of sex bias was observed. Nonetheless, highly specialized foraging techniques (such as sponging) tend to be female biased (Mann and Sargeant 2003) due to sex-specific reproductive strategies. Specifically, males invest their time in consorting females and in some locations form and maintain alliances to improve these efforts (Connor and Krützen 2015; Gerber et al. 2020). These behaviors are time-consuming, rendering them less compatible with highly specialized and spatially restricting behaviors such as specialized foraging techniques (Mann et al. 2008; Patterson et al. 2016). Conversely, females, due to the energetic demands attributed to reproduction, exhibit high foraging investment and therefore have more capacity for time-intensive or specialized foraging strategies (Galezo et al. 2020, 2018; Krützen et al. 2014; Mann et al. 2008).

Based on these shelling observations and the findings of Wild et al. (2020), it is possible shelling is more specialized than previously thought. For example, shelling appears to meet several criteria consistent with a foraging specialization: (1) shelling is not ubiquitous within the population, but instead is restricted to a small subset of individuals, (2) among these individuals, shelling is performed repeatedly across time, indicating behavioral consistency rather than only isolated, opportunistic events (Wild et al. 2020), and (3) shelling likely requires a distinct behavioral repertoire and specialized knowledge, including locating suitable empty gastropod shells, effectively manipulating shells to retain prey, and exploiting habitats in which shells occur, as the behavior has been observed to be spatially structured rather than uniformly distributed across the study area (Wild et al. 2020). Collectively, the restricted occurrence, repeatability, inferred skill requirements, and socially or maternally learned nature of shelling seen here and in the Wild et al. (2020) study supports the notion of shelling as a specialized foraging strategy and may further facilitate ecological niche differentiation. Ongoing studies focused on the various foraging strategies within Hervey Bay would help elucidate the novelty and potential specialization of shelling within this population.

We provide additional evidence (the first from Queensland, Australia) in support of shelling behavior as a rare foraging tactic indicative of tool use highlighting that this behavior occurs across the East and West coasts of Australia. Additionally, we provide the first observations consistent with this behavior possibly being socially transmitted vertically among bottlenose dolphins. However, the small sample size warrants that our findings be taken with caution and not conclusive of vertical transmission. Further research is needed across both Hervey Bay and Shark Bay to fully unpack the nuances of shelling behavior in these populations and provide more clarity on transmission pathways, specializations, and links to ecology, cognition, culture, and evolution.

Alexis L. Levengood: conceptualization, methodology, data curation, investigation, formal analysis, funding acquisition, visualization, project administration, resources, writing – original draft, writing – review and editing. Georgina V. Hume: data curation, investigation, project administration, writing – review and editing. Jessica M. Buckman: data curation, investigation, methodology, writing – review and editing. Isla Steel: investigation, writing – review and editing. Bridget E. Toohey: writing – review and editing, investigation. Peter Lynch: writing – review and editing, funding acquisition, data curation. Cassandra Smith: writing – review and editing, data curation.

This work was supported by the Winifred Violet Scott Charitable Trust and CID Foundation.

The authors declare no conflicts of interest.

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.