We sampled the epaxial muscle adjacent to the dorsal fin using a hand-held 1 cm diameter stainless steel biopsy and stored on ice before transferring to a -18°C freezer. We collected blood via caudal venipuncture and placed into a lithium-heparin plasma separator vial (BD Vacutainer), centrifuged at 1534 ×g for 3 min and pipetted the plasma into 2 mL vials. We stored the plasma samples temporarily at -18°C in the field and later transferred them to a freezer at -80°C.

Lipids were extracted from samples for subsequent fatty acid analysis to ensure lipid content did not bias carbon isotopes (Newsome et al., 2010; Carlisle et al., 2017). We lyophilised muscle (0.1 ± SD 0.01 g) and plasma (1.2 ± SD 0.78 g) for 48 h (Alpha LD14 plus freeze dryer) and homogenised the samples in a TissueLyser LT (Qiagen) for 2 min at 50 Hz to facilitate optimal lipid extraction. Using a modified Bligh and Dyer (1959) method, lipids were extracted from the lyophilised tissue using 3.8 mL of dichloromethane:methanol:Milli-Q (1:2:0.8 mL) solution for approximately 18 h. We centrifuged the solution at 1534 ×g for 2 min and transferred the supernatant into a glass tube, where we separated the polar and non-polar phases using 1 mL each of dichloromethane and 0.9% NaCl saline solution and centrifuged at 1534 ×g for 2 min. The lower non-polar layer was transferred to a 2 mL vial and dried it under N₂ to a constant weight. We resuspended the total lipid extract in 1.5 mL dichloromethane and stored it at -80°C for subsequent fatty acid analysis.

We air-dried the lipid-extracted tissue overnight in a fume hood to remove residual solvents. After lipid extraction, we removed urea and TMAO using three rounds of sonication in deionised water, following Kim and Koch (2012). We then oven-dried samples at 60°C for 48 h, ground them to a fine powder for 3 min at 50 Hz using a TissueLyser LT (Qiagen) and prepared them for stable isotope analysis.

We analysed fatty acid profiles of 102 muscle (60 female; 42 male) and 39 plasma (26 female; 13 male) tissues following a modified Bligh and Dyer trans-methylation procedure (Bligh and Dyer, 1959; described in Meyer et al., 2017). An aliquot of the total lipid extract was trans-methylated using 3 mL of dichloromethane:methanol:hydrochloric acid (10:10:1 v/v/v) for 2 h at 80°C, then allowed to cool to room temperature before adding 1 mL of Milli-Q. We extracted the resulting fatty acid methyl esters (FAME) from the methylating solvent using three washes of 1.8 ml hexane:dichloromethane (4:1 v/v), centrifuged at 1534 ×g for 5 min, collected the upper solvent layer containing the FAMES in a 2 mL glass vial, dried under N₂ gas, and suspended it in 1 ml of dichloromethane. We identified and quantified individual fatty acids through gas chromatography (GC) and GC-mass spectrometry (GC-MS). We achieved peak separation using an Agilent Technologies 7890B GC (Palo Alto, California USA) with an Equity-1 fused silica capillary column (15 mm × 0.1 mm internal diameter and 0.1 mm film thickness), a flame ionisation detector, a splitless injector, and an Agilent Technologies 7683B Series auto-sampler. Samples were injected in splitless mode and carried by helium gas at an oven temperature of 120°C. The temperature was raised to 270°C at a rate of 10°C per min and then to 310°C at a rate of 5°C per min. We quantified fatty acid peaks using Agilent Technologies ChemStation software (Palo Alto, California, USA) and confirmed identities using a Finnigan Thermoquest DSQ GC-MS system. Fatty acids were converted from chromatogram peak area to percentage of total area.

We weighed 102 dried muscle (60 female; 42 male) and 38 plasma (25 female; 13 male) samples from white sharks (10–20 mg) into tin capsules using an automated microbalance described by Carvalho (2021) to ensure maximum accuracy. We analysed samples for δ¹³C and δ¹⁵N using an isotope ratio mass spectrometer (Thermo Delta V Plus) coupled to an elemental analyser (Thermo Fisher Flash EA) via an interface (Thermo Fisher Conflo IV). Isotopic ratios are expressed in delta (δ) values as the deviations from conventional standards in parts per thousand (‰) using the following formula: δ¹³C or δ¹⁵N = [(R_sample/R_standard – 1)] × 1000 (‰) where R_sample is the ratio of heavy to light isotope and R_standard is the ratio of heavy to light isotope in the reference standard. We measured R_sample against internal working standards (glycine: δ¹³C = -41.8, δ¹⁵N = 2.0; glucose: δ¹³C = -10.5; collagen: δ¹³C = -21.5, δ¹⁵N = 4.8), which were calibrated against international reference materials [(USGS64: δ¹³C = -40.8, δ¹⁵N = 1.8; USGS65: δ¹³C = -20.3, δ¹⁵N = 20.7; USGS64: δ¹³C = -0.7, δ¹⁵N = 40.8 (Schimmelmann et al., 2016)]. We reported δ¹³C and δ¹⁵N values relative to international reference materials V-PDB (Vienna-Pee Dee Belemnite) and atmospheric nitrogen (N₂) with a precision of 0.15 ‰ (δ¹³C) and 0.3 ‰ (δ¹⁵N).

We collected muscle tissue samples from 18 potential white shark prey items to calculate prey contributions to white shark diet ( Supplementary Table S1 ). Prey items were selected based on recent studies of immature white sharks in the same region (Grainger et al., 2020; Clark et al., 2023). Muscle samples were lipid-extracted using the above protocol, and urea was removed from elasmobranch prey samples using the outlined deionised water treatment. We could not obtain samples from cownose rays (Rhinoptera neglecta), so we used values from Raoult et al. (2019), which were not lipid extracted. To account for the effects of lipids on δ¹³C, these values were mathematically lipid-corrected before analyses using the below equation, described in Post et al. (2007).

Of the 57 fatty acids detected, we used those with averages > 0.1% (14 muscle; 17 plasma; Table 1 ). We used PRIMER 7 + PERMANOVA (Permutational multivariate analysis of variance; Plymouth Routines in Multivariate Ecological Research, Anderson, 2017) to run multivariate statistical analyses, and R v4.3.1 (R Core Team, 2023) for linear and Bayesian mixing models.

PERMANOVA analysis (Type I, based on 9999 permutations) was used to investigate differences in white shark fatty acid profiles using season and sex as fixed factors, location as a random factor, and total length as a continuous covariate. PERMANOVA analyses were completed using a Euclidean distance matrix on normalised untransformed data. Pairwise comparisons were performed if the main PERMANOVA tests showed significant differences. Principal Coordinate Ordination analysis (PCO) was used to visualise the clustering of individuals and correlation between fatty acids.

If the PERMANOVA showed significant differences in fatty acid profiles, we further explored dietary patterns by fitting univariate linear models to a subset of fatty acids selected for their ecological relevance and high correlation on the PCO. We were unable to include location as a factor in the models because of the low sample size across some locations ( Supplementary Table S4 ). The effect of season (4 levels; fixed), sex (2 levels; fixed), and TL (continuous) on individual fatty acids (16:0, 18:0, 18:1ω9, ARA, EPA and DHA) was assessed by analysis of variance derived from each linear model. We used the same linear model to investigate changes in isotopic signatures in muscle and plasma:

Where Y is the vector of fatty acid or stable isotope values observed on each shark and X is a matrix of indicator variables with columns representing season of capture, sex and TL. The vector β contains coefficients estimated by maximum likelihood. All models were tested for normality and homogeneity in the residuals (e). The models were also used to estimate the mean response to season with 95% confidence intervals.

We used Bayesian stable isotope mixing models of the muscle samples using the simmr package (Govan et al., 2023) to calculate the overall contributions of prey sources to the diet of white sharks. We excluded plasma stable isotope values from the analysis, as the C:N after lipid extraction was greater than the threshold (3.5), introducing potential bias to the mixing model results (Post et al., 2007; Hussey et al., 2012a). Before running the mixing model, we visually assessed the isotopic signatures of consumers, ensuring they were generally within the isospace of sampled prey ( Supplementary Table S1 ). After the application of the trophic enrichment factor, the following prey did not fall within the isospace of the consumers and were removed from the model ( Supplementary Figures S2 , S3 ): cownose ray, pilchards (Sardinops sagax), tiger shark (Galeocerdo cuvier), and bronze whaler (Carcharhinus obscurus).

We used K-means clustering; assigning observations to the group with nearest centroid so that within group variance is minimised, to aggregate the remaining prey items (n = 13; Table 2 ) into three groups. This was recommended by Phillips et al. (2014), whereby model accuracy is greatest when the number of sources is equal to the number of tracers + 1. Stable isotope values of prey groups were adjusted for trophic enrichment using values for lipid-extracted sand tiger shark (Carcharias taurus) muscle from Hussey et al. (2011; δ¹³C = 0.9 ± 0.3‰, δ¹⁵N = 2.29 ± 0.2‰). We ran the mixing model with 3600 iterations over 4 MCMC chains and deemed model convergence satisfactory through a Gelman diagnostic value of 1.

We ran separate Bayesian mixing models on winter, spring, and summer to examine proportional prey contribution differences among seasons. Autumn was excluded from this analysis to avoid unreliable contributions, as only four samples were collected during this season ( Supplementary Table S4 ). Due to prey seasonality, two species, sea mullet (Mugil cephalus) and Australian salmon (Arripis trutta), were removed from the spring and summer models because their abundance is substantially reduced throughout these seasons (Lester et al., 2009; Hughes, 2012).

White shark muscle fatty acid profiles varied among seasons (PERMANOVA-pseudo-F = 4.34, p< 0.01) but were not influenced by sex or location (PERMANOVA- pseudo-F = 1.65, p > 0.05; Supplementary Table S5 ). Pairwise PERMANOVA revealed winter being different to both spring and summer, and autumn being different to both summer and spring (p< 0.05). Muscle samples collected in spring and summer had higher proportions of essential FA 20:4ω6 (ARA), 20:5ω3 (EPA), and 22:6ω3 (DHA) compared to samples collected in winter and autumn, which were dominated by saturated FAs 14:0, 16:0 and 18:0 ( Figures 2 , 3 ). Plasma fatty acid profiles, however, did not vary between season, sex, or location ( Supplementary Table S5 ).

In muscle tissues, all six individual fatty acids were most affected by season, with autumn and winter having high 16:0, 18:0, and 18:1ω9, and low ARA, DHA, and EPA ( Supplementary Tables S6, S7 ; Figure 3 ). Sex also affected muscle EPA and ARA ( Supplementary Figure S8 ), while shark length affected muscle EPA and DHA ( Supplementary Table S6 ; Supplementary Figure S9 ).

In white shark muscle tissue, δ¹³C ranged from -14.6 to -17.5‰ (mean ± standard deviation: -16.2 ± 0.5‰), while δ¹⁵N ranged from 14.6 to 16.4‰ (15.5 ± 0.3‰). Plasma δ¹³C ranged from -14.3 to -17.8‰ (-15.3 ± 0.8‰) and δ¹⁵N ranged from 12.4 to 14.9‰ (13.8 ± 0.5‰). Stable isotope values of muscle were influenced by seasons ( Supplementary Table S10 ; Supplementary Figure S11 ), δ¹³C was lowest in winter and δ¹⁵N lowest in spring ( Figure 4 ). Sex did not affect stable isotope values, but a weak but significant relationship existed in muscle and plasma δ¹⁵N and total length ( Supplementary Figures S12 , S13 ).

Stable isotope values of white shark tissue were within the isoscape of potential prey sources ( Figure 5 ; Supplementary Figure S3 ). Low trophic level δ¹³C enriched prey were the primary resource used by immature white sharks and had an average contribution of 0.49 (± 0.04) to the annual diet of immature white sharks ( Figure 6 ). High trophic level δ¹³C enriched prey) contributed an average of 0.37 (± 0.05), while mid-trophic level δ¹³C depleted prey contributed 0.14 to the diet of immature white sharks (± 0.01; Figure 6 ).

Seasonal mixing models showed substantial variation in prey proportions. Contributions from low trophic level δ¹³C enriched prey) were highest in spring and summer, at 0.57 and 0.43 (± 0.04 and 0.02), respectively and decreased to 0.29 in winter ( Figure 7 ). Mid-trophic level δ¹³C depleted prey (contributed minimally during spring and summer at 0.07 and 0.14 (± 0.04 and 0.09), respectively and increased to 0.40 (± 0.1) in winter ( Figure 7 ). Proportions from high trophic level δ¹³C enriched prey were lowest in winter, 0.31 (± 0.03), increased during spring to 0.36 (± 0.06), and were highest in summer at 0.43 (± 0.04; Figure 7 ).

Fatty acid and stable isotope results suggest that white sharks use coastal habitats for foraging. Muscle and plasma ARA proportions were similar to previous studies in New South Wales and South Australia (Pethybridge et al., 2014; Meyer et al., 2017; Gallagher et al., 2019) and are comparable to other coastal sharks and benthic elasmobranchs (Davidson et al., 2011; de Sousa Rangel et al., 2021c; Zhang et al., 2023). ARA derives from benthic and coastal primary producers (Sardenne et al., 2017). Therefore, animals feeding in these areas would have elevated levels of this polyunsaturated fatty acid (Caraveo-Patiño et al., 2009; Hartwich et al., 2013) compared to those feeding in offshore habitats. Over a longer-temporal scale, δ¹³C values support the fatty acid results of high coastal habitat use. However, the δ¹³C values revealed high inter-individual variation, ranging between -14.6 to -17.5‰, which encompasses habitats dominated by coastal macrophytes (-14‰) and pelagic phytoplankton (-18‰; Hobson, 1999). The variation in δ¹³C reported here indicate some individuals have a higher reliance on δ¹³C depleted prey, similar to great hammerhead sharks (Sphyrna mokarran) that forage on coastal prey in the same region (Raoult et al., 2019).

Overall proportions of ARA and DHA indicate feeding on coastal prey. These fatty acids may reflect the contribution of fish and cephalopods to the diet of white sharks (Couturier et al., 2013; Pethybridge et al., 2014), with high ARA proportions representing benthic-feeding fishes and elasmobranchs (Couturier et al., 2013). This was further supported by the stable isotope Bayesian mixing model, showing white shark diet consisted primarily of low trophic level prey from coastal habitats with a substantial contribution from species at higher trophic levels These findings are similar to previous studies reporting that immature white sharks feed on various species from benthic and pelagic ecosystems (Hussey et al., 2012b; Tamburin et al., 2020). Previous stomach content and eDNA metabarcoding analyses showed that Australian salmon (Arripis trutta) and sea mullet (Mugil cephalus) are important prey for eastern Australian immature white sharks (Grainger et al., 2020; Clark et al., 2023). Benthic and benthopelagic elasmobranchs also contribute to the diet of immature white sharks in eastern Australia (Grainger et al., 2020) and in the North-Eastern Pacific Ocean (Tamburin et al., 2020) but were rarely detected in another study using eDNA metabarcoding of cloaca swabs (Clark et al., 2023). The differences in prey composition across these studies reflect the broad range of species consumed by white sharks, likely linked to seasonal variation in prey availability or broad-scale movements exhibited by most sharks, from Queensland to Victoria Spaet et al., (2020b). Additionally, the differences in δ¹³C and δ¹⁵N values seen here, indicate there are substantial dietary variations among individuals which may not be reflected in the contributions from the mixing model.

Stable isotope mixing models from muscle samples showed seasonal shifts in diet, with substantial changes from low and higher trophic level δ¹³C enriched prey in spring and summer to mid-trophic level δ¹³C depleted prey in winter. The fatty acid results further support this dietary shift. Proportions of DHA increased substantially during the spring and summer months and are similar to those found in benthic elasmobranchs (Dunstan et al., 1988; El Kebir et al., 2007) and bigeye tuna (Thunnus obesus; Peng et al., 2013). Fatty acid 18:1ω9 is a strong indicator of mesopelagic fish, such as yellowfin tuna (Thunnus albacares; Sardenne et al., 2016) and cephalopods (Phillips et al., 2001; Meyer et al., 2019), suggesting higher consumption of these prey during autumn and winter when this fatty acid increases. Seasonal variations in environmental conditions can impact the distribution and abundance of prey (Poloczanska et al., 2007; Last et al., 2011). The strengthening EAC during late spring causes persistent upwelling over shelf waters, subsequently increasing productivity (Roughan and Middleton, 2002). This coincides with white shark presence and capture in the region, which peaks during winter and spring, when water temperatures are below 21°C (Spaet et al., 2020a; Lipscombe et al., 2023). Consequently, the seasonal changes detected in our study are likely linked to the changes in prey availability associated with water temperature and season. The East Australian Current strongly influences water temperatures (Malcolm et al., 2011) and prey distribution on the east coast of Australia (Booth et al., 2007). Thus, these factors also influence shark occurrence and capture (Spaet et al., 2020a; Lipscombe et al., 2023), and habitat use by potential prey species (Gillanders et al., 2001; Zischke et al., 2012). Fatty acids are known to be influenced by sea surface temperature (Dalsgaard et al., 2003; Meyer et al., 2019); therefore, by combining these with stable isotopes, we have greater confidence in the inferences of diet. Future studies should focus on how changes in prey distribution and availability in response to the strengthening East Australian Current may impact the trophic ecology of immature white sharks.

Seasonal variations in diet have the potential to impact the nutritional condition of white sharks. The higher proportions of saturated fatty acids and lower proportions of essential dietary fatty acids in autumn and winter suggest that prey consumed in these months is of lower nutritional quality or certain prey species are less abundant. Saturated fatty acids are ubiquitous in all animals and, combined with monounsaturated fatty acids, are the primary fatty acids catabolised for energy (Tocher, 2003). In contrast, polyunsaturated fatty acids are conserved for critical biological processes (Tocher, 2010). Lipid metabolism is relatively understudied in elasmobranchs, yet unlike teleost fishes, lipid storage occurs in the large liver (Zammit and Newsholme, 1979). The proportions of saturated fatty acids in muscle may suggest a metabolic adjustment to compensate for the lack of high-quality prey. Conversely, substantial increases in the proportions of the essential fatty acids, particularly DHA, in spring and summer imply a shift in diet to prey that supports cellular functions and physiological processes, such as growth, during these periods. Although the seasonal change in diet is evident, links to nutritional condition and the implications of poor nutrition in sharks are speculative and should be examined in more detail in future studies. Additionally, assessing responses and metabolic adaptations to dynamic coastal ecosystems may be invaluable for forecasting distribution changes associated with anthropogenic impacts.

Ontogenetic diet shifts are well documented in white sharks from the Pacific (Estrada et al., 2006; Kim et al., 2012b). Due to limitations in sample availability, few (n = 8) samples were collected from sharks > 3 m; therefore, only a small increase in δ¹⁵N was detected with shark length. Furthermore, there was high variability among individuals, with a smaller shark (2.5 m) having the highest δ¹⁵N value and larger sharks exhibiting values below 15.5‰, lower than those recorded for immature sharks in Mexico and the north-east Pacific (Carlisle et al., 2012; Tamburin et al., 2020).

An animal’s energy and nutritional requirements typically increase with size (Gallagher et al., 2014). Although the relationships between fatty acid proportions and length are yet to be described in white sharks, the negative relationships seen here are vastly different from those of other shark species (de Sousa Rangel et al., 2021a; de Sousa Rangel et al., 2021c). We expected fatty acid proportions to increase with size as larger and higher quality prey were consumed. However, we observed decreasing proportions of DHA and EPA in muscle. Increases in DHA with total length were found in the plasma of tiger sharks (Galeocerdo cuvier) in the Atlantic (de Sousa Rangel et al., 2021a), where larger mature sharks exhibited higher values compared to immature sharks, which was linked to nutritional adjustments in preparation for reproduction. Similar results have been observed for mature male blacktip sharks in EPA, DHA and ARA (de Sousa Rangel et al., 2021b). Based on these results, it is plausible that our results reflect the immature life stage of white sharks, where energetic investments for reproduction are not yet required. More extensive sampling of mature white sharks across a broader spatial scale may assist in clarifying the relationships between total length and specific fatty acids when sharks are closer to maturity.

Consideration must be given to the limitations of using bulk tissue stable isotopes in trophic ecology studies. Specifically, mixing models are sensitive to trophic enrichment factors and isotopic routing, which can lead to inaccurate results (Phillips et al., Bond and Diamond, 2011). Species- and tissue-specific trophic enrichment factors are limited for large sharks, as controlled feeding experiments are impractical. Currently, no trophic enrichment factor is available for white sharks, and as a large-bodied predator occupying a high trophic level, there are several physiological (e.g., growth rate, isotope routing) and environmental (e.g., temperature) factors that may impact the discrimination of stable isotopes (Caut et al., 2009; Hussey et al., 2010; Phillips et al., 2014). Subsequently, the value provided by Hussey et al. (2010) from multiple species of sharks was deemed the best alternative. Isotopic routing remains unclear in shark tissues, yet it is well documented to cause significant issues when interpreting mixing models in teleosts, resulting in over or underestimating the contribution of sources to diet (Kelly and Martínez del Rio, 2010). We also acknowledge that while we aimed to include a range of prey items of white sharks, δ¹³C values indicate many of these sharks are feeding on coastal prey with values around -15‰, that were not included in the mixing model. There may also be another prey group between the low trophic and high trophic level δ¹³C enriched groups that these sharks are feeding on that we have been unable to sample in this study. One of the many limitations of mixing models is their inability to detect missing prey or differentiate if consumers values are the average of two separate prey groups with distinct isotopic signatures (Phillips et al., 2014; Stock et al., 2018). Additionally, the more prey included in the mixing model, the less precise the estimates of contribution (Phillips et al., 2014). Furthermore, interpretation of stable isotopes requires caution, as spatial and temporal variation of isotopes exists in marine environments and are influenced by environmental and anthropogenic factors (Pethybridge et al., 2018; Matich et al., 2021).