Return ticket: spatial use patterns of white sharks (Carcharodon carcharias) from a shark-bite mitigation program in eastern Australia

The inherent susceptibility of white sharks (Carcharodon carcharias), coupled with reported declines in abundance, has led to their global listing as “Vulnerable”, prompting substantial conservation efforts. However, white sharks' propensity to use coastal areas overlaps with human activities, causing safety concerns. In New South Wales (NSW), Australia, they are caught as part of the world's longest-running meshing bather protection program. After an unprecedented spate of shark bites in 2015, Shark-Management-Alert-in-Real-Time (SMART) drumline trials started in NSW as a novel catch-and-release program aimed at non-lethal bather protection. Using seven years of data, this study analyzed SMART drumline recapture events to describe white shark habitat use along the NSW coastline. In total, 890 individuals were tagged with a recapture rate of 24.5% (n = 218). Smaller individuals (fork length < 225 cm) and females were more commonly recaptured, suggesting size and sex-specific habitat preferences. Although recaptures occurred year-round and throughout the entire NSW coast, most occurred during the Austral winter and spring and along the northern NSW coast. There were few consecutive recaptures recorded in the same location, suggesting that white sharks move throughout relatively large ranges within coastal areas. Linear distances between recaptures were on average ~220 (SE ± 25) km, mostly representing northward travels. Despite some variation, most recaptures (~77%) were recorded within a year and ~45% within 90 days. These results are consistent with the described movement ecology for the species along NSW, corroborating the seasonal importance of NSW northern coast for juvenile white sharks, and the potential effect that the Eastern Australian Current has on their distribution. This study provides important insights into immature white sharks' inshore ecology, highlighting the value of capture-mark-recapture data collected from SMART drumlines to forecast shark movement for enhanced bather safety and species management.

White sharks (Carcharodon carcharias) are globally distributed pelagic elasmobranchs with a complex ecological role as predators (1). There are still major knowledge gaps related to the species' life history and reproduction (1). Regardless, there have been substantial advances in white shark research and conservation efforts over the last decade (1), prompted by the species “Vulnerable” conservation status (2) and its mediatic involvement in human-shark conflicts (3). In areas where conservation efforts are well enforced, populations appear to be recovering [e.g., Eastern Pacific (4), and Northwest Atlantic (5)]. However, white sharks' propensity to occupy coastal waters places them in direct conflict with human coastal habitat use, hindering conservation efforts (6–8). White, bull (Carcharhinus leucas), and tiger (Galeocerdo cuvier) sharks are responsible for most unprovoked shark bites and are often referred to as “target” species for bather protection programs. Although unprovoked bites are rare (~70 events and six fatalities per year-Florida Museum of Natural History, International Shark Attack File, https://www.floridamuseum.ufl.edu/shark-attacks, accessed in 11/2024), they often fuel public fears of “shark attacks” (9–11); the media also tends to further sensationalize these events (12). Evidence of a direct relationship between white shark abundance and the number of shark bites is sparse (13, 14). Nevertheless, the recovery of local populations may lead to an increased number of sightings, media events, and overall human interactions (7, 15). Studies on white shark abundance suggest small populations with imperceptible growth (16, 17), but human coastal utilization is increasing rapidly, bringing more people in contact with sharks (9–11, 18). Therefore, there is a need to proactively reconcile public safety with white shark conservation, requiring a better understanding of the species' use of coastal habitats for effective management (1, 3, 19).

Human-shark interactions in Australia have escalated over the last decade, sparking debate around white shark conservation and bather protection (19). Although the species is fully protected in Australian coastal waters, there are exemptions for bather protection measures in the states of New South Wales (NSW) and Queensland (Qld), with the euthanasia of sharks also being authorized, under specific circumstances, in Western Australia (WA) (20). This stems from the fact that Australia has the second-highest annual frequency of unprovoked shark bites, which have increased from an average of nine events between 1990–2000 to 22 events between 2010–2020 (21). Most white shark bites have been recorded along Australia's eastern and southern coastline, particularly within the state of NSW (21), where immature individuals are found in relatively high abundances, moving through the northern areas of NSW, particularly during austral winter and spring (22–26). Both genetic (17, 28) and tagging data (22, 23, 25) support a single population in the areas surrounding eastern Australia and New Zealand, often referred to as the “Eastern Australasian population”. This population is estimated to be around 1,400–6,750 individuals (20, 28) and is thought to have experienced a major (>90%) decline during the last century, a consequence of targeted fishing and lethal bather protection programs (29). Currently, the number of breeding adults appears to be stable but low, and the prospects of recovery to last century population sizes are uncertain (30).

In NSW, mesh nets have been used since 1937 to reduce shark-human interactions, with 51 beaches between Newcastle to Wollongong currently netted from 1 September to 30 April (20, 31). Across 2024–2025, a total of 223 marine animals were captured, of which ~40% were released alive, and only ~10% represented target shark species (31). The high bycatch and mortality rates of mesh nets have long been criticized (32, 33), prompting the development of non-lethal alternatives (34, 35). Compared with mesh nets, traditional drumlines have higher survival rates (36), representing an effective bather protection strategy with reduced impacts (37). Recently, traditional drumlines have been modified into non-lethal Shark-Management-Alert-in-Real-Time (SMART) drumlines, aiming to further reduce mortality and bycatch (38). After an unprecedented 14 unprovoked shark bites (one of which fatal) (21), recorded in NSW in 2015, SMART drumlines were trialed as part of the NSW Shark Management Strategy (now known as the Shark Management Program — SMP) in an experimental framework designed to maximize their efficiency. Results based on these trials showed SMART drumlines to be effective in capturing target sharks, with bycatch of ~30% and an overall mortality rate of 1%−3%; white sharks accounted for >50% of captures (39, 40). Following the success of these trials, beginning in 2022, an operational bather protection program was implemented using 305 SMART drumlines deployed daily across 20 distinct locations along NSW, presenting an unprecedented opportunity to collect data on white sharks.

Conventional tagging of sharks has long been recognized as a valuable means for studying their life history, movements, and population structure (41). Despite advances in acoustic and satellite tagging, capture-mark-recapture methods are still the most cost-effective and widely used methodology [e.g., (42, 43)]. We analyzed white shark physical recaptures on SMART drumlines deployed as part of the SMP, under the overarching themes of species conservation and human interaction mitigation. Using capture-mark-recapture data collected over a 7-year period, this study aims to quantify white shark movement/behavior ecology with respect to SMART drumlines by: (i) describing factors related to individuals' characteristics affecting recapture probability; (ii) assessing the spatiotemporal variability of recaptures; and (iii) describing the time-at-liberty, linear distance and latitudinal shift between recaptures.

Data collection

Data was collected as part of the ongoing NSW SMP, between December 2015 and March 2023, covering over 7 years of mark-capture-recapture data. SMART drumline trials for the SMP started in December 2015, with drumlines being tested either opportunistically or following more structured deployment schemes in discrete locations along the NSW coastline for various lengths of time (Supplementary Table S1.1). Since May 2022, SMART drumlines have been deployed daily (weather permitting) in 20 locations, from Kingscliff (28.3°S, 153.6°E) to Merimbula (36.9°S, 149.9°E; Figure 1). Within each loccation, SMART drumlines are deployed at 15 permanent sites (20 at Ballina) across 6 to 17 km of coastline. The drumlines are set 500 m offshore in water depths of 6 to 23 m. Each line is deployed between first light and 08:00 a.m. and retrieved at least 2 h before sunset on the same day. This ensures adequate handling time if an animal is caught and time to travel across dangerous river entrances at some locations to return before sunset. Gear retrievals may be made earlier if weather conditions are deemed unsafe to navigate or if rough seas preclude safe handling of animals.

Each SMART drumline consists of a conventional drumline to which a “Catch-A-Live” ® system is added (38). This system utilizes the MLI-s buoy (Marine Instruments™), which emits a GPS signal through the Iridium satellite network, triggered when a magnetic pin is dislocated through an animal pulling on the baited hook. The bait used comprises a single sea mullet (Mugil cephalus). The GPS signal is transmitted to operators who rapidly assess the catch by vessel (within 30 min), aiming to tag target sharks and release them 1 km from shore, or in-situ for non-target animals (44).

After capture, animals were secured to the side of the vessel, identified to species level, measured (total-length-TL, fork-length-FL, and pre-caudal length-PCL) to the nearest cm, and their gender determined by the presence or absence of claspers. For each catch, the time and location of both capture and release were recorded. All animals were checked for tags and tag details, thereby enabling identification of recapture events. If not previously tagged, prior to release, all animals were tagged with conventional external identification tags (yellow plastic, nylon-head dart-tag Hallprint Pty, South Australia); white, tiger, and bull sharks were also fitted with an external acoustic transmitter (V16-6L, Innovasea). Sea surface temperature (SST) was recorded when target species were captured. The collection of other environmental variables was not standardized during the surveyed period, and therefore only SST was considered to describe white shark's recaptures. This is also the most used environmental variable to describe white shark movement (e.g., 6, 26).

Data analysis

Individual characterization

Individuals caught on SMART Drumlines were characterized by sex, size and age class. For sizes, FL measurements were considered most reliable, and the age classes were described using the definitions proposed by Bruce & Bradford (22) [conversions from TL were made following Hunt et al. (45)]. Sharks were separated into: (i) young-of-the-year (YOY; ≤ 152 cm FL), (ii) juveniles (>152–262 cm FL), (iii) sub-adults (>262–422 cm FL for females, and >262–315 cm FL for males) and (iv) adults (females >422 cm FL, and males >315 cm FL). Initially, a general characterization was performed for all individuals caught on SMART drumlines, considering the sizes recorded at the initial capture. After which, aiming to understand which individuals were recaptured, differences between the sex ratio and the FL of individuals recaptured vs. individuals not recaptured were tested. The sex ratio and FL of individuals with multiple recaptures (i.e., with two or more events) were also compared with those with a single recapture. Only recaptures recorded from SMART drumlines were considered, meaning that if an individual was tagged after a capture on a SMART drumline, but recaptured by a fisherman or in a mesh net, it was regarded as “not recaptured”. This study also only considered physical recaptures on the SMART drumlines and did not include electronic detections on the network of tagged shark listening stations in NSW or other receiver networks.

Spatiotemporal variability

Due to the trial phase of the SMP, not all locations were fished for the same period. For this reason, the spatiotemporal analysis had two distinct dimensions: (i) a general description of the complete dataset (from December 2015 to April 2022), and (ii) a detailed analysis of the 12-month period (between April 2022–March 2023) with structured and continuous data collection along all 20 locations being currently fished. The number of recaptures recorded within each of the four Austral seasons was described, considering: Austral summer (December–February), Austral autumn (March–May), Austral winter (June–August), and Austral spring (September–November). The total number of recaptures per surveyed location was also assessed. Specifically for recaptures recorded between April 2022–March 2023 (considering those from both individuals initially caught before and after April 2022), a Generalized Additive Model (GAM) was fitted with restricted maximum likelihood (REML), with a negative binomial error distribution and a log link function, accounting for overdispersion. Seasonal variation was modeled using a cyclic smooth of month, ensuring continuity between December and January. The potential impact of surveyed location was modeled as a smooth random effect. To test whether seasonal patterns differed among locations, a factor smooth interaction between month and location, specified as deviations from the global seasonal pattern, was also included in the GAM. Model assumptions were checked using convergence diagnostics, checks of basis, and diagnostic plots (namely, quantile-quantile plot, histogram of residuals, residuals vs. fitted plot, and response vs. fitted plot). In addition to the GAM, the correlation between the spatial and temporal variability of recaptures recorded between April 2022–March 2023 was further examined by comparing the median latitude of events across the four Austral seasons. This allowed testing for potential seasonal latitudinal shifts. The range of SSTs recorded was also described, and to investigate possible patterns of greater site fidelity to any specific location, the number of individuals consecutively recorded in the same location was assessed during the 12-month period of daily deployments (i.e., between April 2022–March 2023). The proportion of individuals with consecutive events in the same location was compared between the locations with the largest number of recaptures recorded in this 12-month period. To test if individual characteristics influenced the probability of consecutive events in the same location, the sex ratio and FL of individuals with consecutive recaptures were compared with that of individuals with no consecutive recaptures in the same location.

Time-at-liberty, linear distance, and latitudinal shift between recaptures

Times-at-liberty, linear distances, and latitudinal shifts were assessed to infer general movement patterns between recapture events. The time-at-liberty was measured as days, while the linear distance between events (measured as a straight line) was calculated using the built-in function “length”, available in the QGIS Geographic Information System software v.3.22.12 (46). Due to the general north-south alignment of the eastern Australian coast, linear distances were considered a good approximation of the distances between events along the NSW coastline. Linear distances were only estimated for recaptures of individuals initially caught after April 2022. To infer the general direction of shark travels between recaptures, for individuals initially caught after April 2022, the latitudinal shift was also plotted using the QGIS Geographic Information System software v.3.22.12 (46). Once again, due to the general north-south alignment of the eastern Australian coast, only latitudinal shifts were assessed. The potential effect of individuals' characteristics (sex and size) on the time-at-liberty, linear distance, and latitudinal shift between recaptures was also assessed.

Hypothesis testing

For testing hypotheses related to categorical variables (e.g., sex), either Pearson's chi-square tests (47) or Fisher's Exact tests (48) were used appropriately. For significant correlations, the parameter Cramer's V (49) was estimated, along with the corresponding confidence intervals, computed through bootstrapping.

For comparisons involving numeric variables, both continuous (e.g., individual's size) and discrete (e.g., time-at-liberty), a similar statistical framework to Patil (50) was followed. The normality of the data was checked using density plots or histograms and the Shapiro–Wilk Normality Test (51). If significant deviations from normality were found, for positive skewness, square root (x 1/2), logarithmic (log(x)), and reciprocal (1/x) transformations were applied. For negative skewness, the same transformations were used, after the data was reflected [max(x + 1)−1]. Before performing the hypothesis tests, the homogeneity of variance between groups was checked using violin plots and the Bartlett's Test (52) (when the data was normal) or the Levene's Test (53) (when the normality assumption was not met). When the data was normal and homogeneous, a Student's T -test (54) was performed. If the normality assumption was met, but the data showed significant heteroscedasticity, a Welch's T -test was preferred (55). If the normality assumption was violated (even after applying the beforementioned transformations), the non-parametric Wilcoxon test (56) was computed instead. To compare the median latitude of recaptures between the four Austral seasons, the Kruskall–Wallis's test (57) was used, with post-hoc pairwise comparisons being performed using the Dunn's Test (58), combined with the Holm's adjustment (59). The effect sizes following significant tests were also computed. For both the Student's T -test and the Welch's T -test, the Hedges' g (60) was computed using the non-pooled standard deviation. For the Wilcoxon test, the effect size was computed using the rank-biserial correlation (r rb) (61), and for the Kruskall–Wallis's test, the rank epsilon squared (E r 2) was estimated (62). For all effective size estimations, the confidence intervals were computed, as described by Patil (50). For the correlations between individuals' FL and the time-at-liberty or linear distance between events, scatter plots with added Locally Estimated Scatterplot Smoothing (LOESS) trend lines were constructed. If the LOESS indicated a relevant trend, the appropriate regression (i.e., with the closest shape to the trend estimated) was applied to the correlation.

All statistical analysis was performed using the R statistical software v.4.2.2 (63), and the following packages: “car” v.3.1-2 (64), “rcompanion” v.2.4.26 (65), “rstatix” v.0.7.2 (66), and “ggstatsplot” v.0.11.1 (50). The GAM fitted to explore the spatiotemporal variability of recaptures recorded between April 2022–March 2023 and the diagnostic statistics and plots used to check model assumptions were computed using the package “mgcv” v. 1.9-3 (67). Graphics were plotted using the package “ggplot2” v.3.5.0 (68) or for graphics showcasing comparisons, the package “ggstatsplot” v.011.1 (50).

The results from the normality and homogeneity tests performed prior to the hypothesis testing described in this section are available in Supplementary Tables S2.1–S2.3.

Individual characterization

Between December 2015 and March 2023, there were a total of 1,200 events where white sharks were captured on SMART drumlines. However, in 27 events, individuals were not tagged because they died (n = 4) or freed themselves before being tagged (n = 23). In 11 recapture instances, it was not possible to recover tag information due to it being damaged. Discarding these 38 events, a total of 890 white sharks were included in the analysis, of which 218 were recaptured (recapture rate = 24.5%). Multiple recaptures, either two (n = 41), three (n = 3), four (n = 1), or five (n = 1) times, brought the total number of recapture events to 272 (Table 1).

The sex ratio (F:M) of all individuals caught on SMART drumlines was 1:0.74 (Table 1). When comparing the sex ratio of the 218 individuals recaptured (1:0.49) with that of the 672 individuals not recaptured (1:0.85), there was a significantly higher proportion of females recaptured [Table 1; X 2 (1) = 10.52, p = 0.001, Cramer V = 0.11 (0.05, 0.17)]. For individuals with multiple recaptures, their sex ratio (1:0.24) was also significantly more skewed toward females, when compared with sharks with a single event [sex ratio = 1:0.58; Table 1; X 2 (1) = 4.78, p = 0.029; Cramer V = 0.15 (0.03, 0.26)]. Furthermore, no males were recaptured more than twice (Table 1).

Reliable FL measurements were possible for 889 individuals (Table 1). When initially captured, most individuals were juveniles (82.0%, n = 729), followed by sub-adults (14.1%, n = 130), and YOY (3.8%, n = 34; Figure 2). A single adult male was captured, accounting for only ~0.2% of all captures (Figure 2). The mean size of individuals recaptured [208.6 (SD ± 33.1) cm FL] was significantly smaller than that of individuals not recaptured [221.7 (SD ± 43.3) cm FL; Figure 3; Welch's t -test for the log-transformed FL between the two groups: t Welch (460.22) = 4.21, p < 0.001, Hedges' g = 0.31 (0.16, 0.45)]. Individuals with a single recapture averaged 211.6 (SD ± 33.9) cm FL when initially caught. By grouping individuals with two or more recaptures together, the average size recorded in the initial capture for individuals with multiple events was 197.0 (SD ± 27.6) cm FL. Therefore, considering the sizes at initial capture, individuals with multiple recaptures were significantly smaller than individuals with a single recapture [Figures 4A, B; Student's T -test for the log-transformed FL (at initial capture) between individuals with a single recapture vs. those with multiple recaptures: t Student (216) = −2.72, p = 0.007; Hedges's g = −0.46 (−0.78, −0.12)]. Individuals with a single event averaged 228.4 (SD ± 33.5) cm FL after their only recapture. For individuals with multiple events, the mean size measured in their final recapture was 224.3 (SD ± 24.9) cm FL. There was no significant difference between the FL of individuals with a single recapture and that of individuals with multiple recaptures, considering the sizes at their respective last recorded events [Figures 4C, D; Welch's T -test for the log-transformed FL (after the final recapture) between individuals with a single recapture vs. those with multiple recaptures: t Welch (90.82) = −0.71, p = 0.48].

Spatiotemporal variability

The majority (86.0%) of the 272 recaptures were recorded during the Austral winter and spring, with only 8.1% (n = 22) recorded in the Austral summer and 5.9% (n = 16) in the Austral autumn. Forster (n = 60), Evans Head (n = 59), and Ballina (n = 57) accounted for 64.7% of the total recaptures recorded between December 2015 and March 2023.

There were 139 recaptures recorded during the 12-month period of daily deployments, between April 2022–March 2023. Of these, 64 events came from individuals initially caught after April 2022, with the remaining 75 events coming from individuals tagged previously. The 75 recaptures from individuals initially caught prior to April 2022 came mostly from individuals caught during 2021, with 46.7% (n = 35) of these events representing a secondary recapture (i.e., individuals had been previously recaptured at least once); most of these individuals were initially caught in the locations with the highest fishing effort during the earlier trials (i.e., Ballina, Evans Head, and Forster). For individuals tagged after April 2022 (responsible for the remaining 64 events recorded between April 2022–March 2023), the locations of initial capture resembled the overall spatiotemporal variability of recaptures recorded between April 2022–March 2023. Of the 139 recaptures recorded between April 2022–March 2023, 83.4% (n = 116) were recorded between June–October, reflecting a higher abundance of events during the Austral winter and spring. There were very few recaptures recorded between February–May, with a single event per month. The majority of the 139 recaptures was also recorded along the northern coast of the NSW coast, between 29.0–32.5°S. Forster (n = 29), Port Macquarie (n = 24), Coffs Harbour (n = 21), Yamba (n = 17), and Evans Head (n = 15) accounted for 76.3% of all recaptures recorded between April 2022–March 2023. Only 13.0% (n = 18) of recaptures during this period were recorded south of 33°S.

The GAM fitted to the recaptures recorded between April 2022–March 2023 explained 75.8% of the variance, with an adjusted R 2 = 0.68. The model converged fully, and the basis checking results revealed an adequate model fit, corroborated by visual inspection of the diagnostic plots (Supplementary Material 3). Recapture rates exhibited a significant temporal pattern (cyclic smooth of month: X 2 = 45.78, p < 0.001), indicating pronounced monthly variation (Figure 5A). The mean recapture rates also differed significantly among surveyed locations (random effect smooth of location: X 2 =17.96, p < 0.001), indicating spatial heterogeneity along the coast of NSW (Figure 5B). After accounting for the overall temporal pattern, there was no evidence that the trend of monthly variation differed among locations (factor smooth interaction: X 2 =195.24, p = 0.99). Location-specific temporal deviations were strongly penalized toward zero, suggesting that locations shared a common seasonal trajectory. Despite this, a significant correlation between the four austral seasons and the median latitude of recaptures was observed (X Kruskal - wallis 2 (3) = 10.07, p = 0.02; E R 2 = 0.07 [0.03, 1.00]), evidencing a latitudinal shift along the year (Figure 6). Post-hoc pairwise comparisons between the four seasons revealed that recaptures during the Austral summer were recorded in significantly higher latitudes compared with those in the Austral spring (Supplementary Table S4.1).

The mean SST recorded for the 139 recaptures between April 2022–March 2023 was 19.1 (SD ± 1.8) °C, and despite the large range (between 14.3–24.6 °C), 81.3% (n = 113) of recaptures were recorded between 17–21 °C. (Supplementary Figure S4.1).

Of the 139 recaptures between April 2022–March 2023, only 17.3% (n = 24) were recorded at the initial capture location (Table 2). These were from 21 individuals, as two sharks (both females: ID591–recaptured twice and ID 613–recaptured three times) were recaptured multiple times at Forster during this period. The remaining 95 individuals recaptured between April 2022–March 2023 had no consecutive events recorded in the same location. Although 42.9% (n = 9) of the individuals with consecutive events in the same location were recaptured in Forster, there was no significant correlation between the location and the proportion of individuals with consecutive events (Table 2; Fisher's exact test: p = 0.09), considering the locations with the highest number of recaptures recorded between April 2022–March 2023. The sex ratio of individuals recaptured in a different location was 1:0.48 (F:M), while individuals recaptured in the same location had a sex ratio of 1:0.50. No significant correlation was observed between the individual's sex and if they were or not recaptured in the same location of the previous event [ X 2 (1) < 0.01, p = 1.00]. The mean size recorded in the initial capture for individuals recaptured in different locations was 204.4 (SD ± 26.6) cm FL, while individuals consecutively recaptured in the same location averaged 211.8 (SD ± 30.0) cm FL. No significant correlation was found between the FL and individuals being recaptured or not in the same location of the previous event [Student's t -test for the log transformed FL of individuals recaptured in a different location vs. individuals consecutively recaptured in the same location: t Student (116) = −1.09, p = 0.27].

Time-at-liberty, linear distance, and latitudinal shift between recaptures

For all 278 recapture events recorded between December 2015 and March 2023, the average time-at-liberty was 256.2 (SE ± 17.1) days, ranging between 5–1,384 days (Supplementary Figure S4.2). Despite the large range, 76.5% (n = 208) were recorded within a year, and 44.7% (n = 122) occurred within a season (i.e., 90 days). For the entire dataset, the time-at-liberty between initial capture and first recapture was on average 271.1 (SE ± 19.8) days. For individuals with multiple events, the second recapture occurred after 187.0 (SE ± 27.9) days, and the third recapture 373.4 (SE ± 282.8) days later. ID 553 and ID 613 were recaptured for the fourth time after 66 days and 43 days (respectively), taking an additional 10 days for the fifth and final recapture of ID 613. For individuals with multiple recaptures, the last recapture was on average 473.6 (SE ± 53.0) days (range: 24–1,410 days) after the initial capture, with 41.3% (n = 19) of these individuals having all their recaptures recorded within a single year. There were however five females (ID 347, ID 348, ID 369, ID 395, and ID 448) and single male (ID 214) with multiple recaptures recorded over a period greater than 1,000 days (all these individuals were recaptured twice, except for ID 348, which had three recaptures in total). Two of the individuals with multiple recaptures more than 1,000 days apart (ID 347 and ID 395) were YOY when initially caught, and the remaining four (ID 214, ID 248, ID 369, and ID 448) were small (< 185 cm FL) juveniles. They were also all still immature after their last recapture, and half of them (ID 214, ID 347, and ID 395) were smaller than 225 cm FL when recaptured for the last time.

Of the 64 recaptures recorded from individuals initially caught after April 2022 (i.e. since the implementation of the daily deployments along all 20 locations), these were on average 60 (SE ± 6) days apart (range: 5–273 days; Figures 7A, B), with linear distances between events averaging 219.5 (SE ± 25.1) km (range: 0–967.0 km; Figures 7C, D). Despite these large ranges, 82.8% (n = 53) of recaptures were recorded within 90 days, and 87.5% (n = 56) were within less than 400 km. Regarding the latitudinal shift between recaptures, 64% (n = 41) of sharks traveled northward between recaptures (Figure 8), which is a significant deviation from a 1:1 ratio of northward vs. southward travels [ X 2 (1) = 5.06, p = 0.02].

For individuals initially captured after April 2022, there was no significant correlation between the sex and time-at-liberty until first recapture [females mean = 67 (SE ± 10) days; males mean = 46 (SE ± 6) days; Student's t -test for the log-transformed time-at-liberty until first recapture for females vs. males: t Student (52) = 1.03, p = 0.31]. The distance between initial capture and first recapture was however significantly larger in females [mean = 225.3 (SE ± 29.0 km)] compared to that of males [mean = 141.3 (SE ± 34.2 km); Student's t -test for the square-rooted time-at-liberty until first recapture for females vs. males: t Student (52) = 1.99, p = 0.05; Hedges' g = 0.56 (0.00, 1.12)]. No relationship was evident when either the time-at-liberty or linear distances until first recapture were plotted against the FL of individuals (Supplementary Figure S4.3). The sex ratio (F:M) associated with northward travels (1:0.41) was also not significantly different from that of southward dispersals [1: 0.64; X 2 (1) = 0.28, p = 0.60], and there was no significant difference between the size of individuals (considering sizes recorded in the initial capture) that moved northwards [mean = 207.6 (SD ± 30.1 cm)] compared with those that moved southwards [mean = 210.3 (SD ± 30.9 cm); Student's t -test for the reciprocal transformed size of individuals showing a northward vs. southward travels: t Student (62) = 0.32, p = 0.75].

The two individuals with the overall highest number of recaptures, females ID 553 and ID 613 were recaptured over a period of 659 and 390 days, respectively (Figure 9). ID 553 (recaptured four times) was never recaptured in the same location, being initially caught in Ballina, and then in Coffs Harbour, Forster, Newcastle, and in Lake Macquarie. The recaptures of this individual were recorded along >500 km of coastline. ID 613, after also being initially caught in Ballina, had all its recaptures in Forster, ~380 km from the initial capture. This female was recaptured consecutively five times, twice in November 2021 (i.e., during the Austral spring), only 21 days apart (and both times at the same site), and another three times between June and August 2022 (i.e., during the Austral winter), with these last events being recorded over a period of 53 days.

This study demonstrates the value of using capture-mark-recapture data collected from SMART drumlines to describe the movement ecology of a shark species. It provides one of the first data-rich opportunities to assess white shark physical recaptures from a non-lethal bather protection program. Our results can help to improve white shark management, in the context of bather safety and species conservation. Although capture-mark-recapture studies based on photo-ID on white sharks can have higher recapture rates [e.g., 58% (16)], the total number of physically tagged individuals and the high recapture rate set this study apart from previous capture-mark-recapture studies on white sharks [e.g., (69–71)], as physical recapture rates above 10% are rare in elasmobranchs (42, 43). Furthermore, over 20% of individuals recaptured had multiple recaptures, highlighting the value of extensive capture-mark-recapture data from physical captures and the importance of the NSW coastal areas for the species. A clear effect of size and sex on recapture probability is identified in the present study, as smaller individuals (FL < 225 cm) and females were recaptured more frequently, suggesting sex and size-based differences in habitat preference and foraging behavior. The observed spatiotemporal variability of recaptures on SMART drumlines, along with the times-at-liberty, linear distances, and latitudinal shifts between recaptures, further corroborate the significance of the northern coastal area of NSW (i.e., between 29.0–32.5°S) for juvenile white sharks, particularly during the Austral winter and spring. The spatiotemporal variability of white shark recaptures along NSW coast is also evidence of the potential effect that the strengthening and shifting Eastern Australian Current (EAC) may have on the species distribution.

Size and sex influence recaptures

The higher frequency of juvenile recaptures on SMART drumlines might be explained by the importance of inshore NSW coastal waters for immature white sharks (22–25). On the east coast of Australia, juveniles and sub-adults are mostly found in continental shelf waters, with the region between Central Coast (32.2° S, 152.5° W), and South West Rocks (30.9° S 153.0° W) comprising one of the two described nurseries for immature (but non-natal) white sharks (24, 25). Adults are observed in eastern Australian coastal waters, inhabiting more offshore areas (72). Bruce et al. (23) hypothesized that individuals start to move offshore after reaching ~250 cm TL (or ~220 cm FL). This is consistent with the size after the final recapture of most individuals with either a single or multiple events in this study being ~225 cm FL, suggesting that individuals are less likely to be recaptured on SMART drumlines after reaching this size. However, most unprovoked bites in NSW are associated with immature individuals (21). Juvenile white sharks are commonly observed in shallow inshore temperate waters globally [e.g., South Africa (73); Northwest Atlantic (74); Northeast Pacific (75)]. Furthermore, the association between immature individuals and shallow coastal areas is not unique to this species, being observed across several large pelagic sharks, such as blue sharks (Prionace glauca) (76), smooth hammerheads (Sphyrna zygaena) (77), and whale sharks (Rhincodon typus) (78). The juvenile preference toward coastal areas might be explained by higher abundances of smaller prey items (e.g., pelagic-neritic and reef-dwelling teleost, or benthonic species) (6, 24, 27). White sharks are generalist predators that undergo an ontogenetic increase in trophic level and dietary inclusion of marine mammals, coupled with increased exploration of offshore food webs among adults [e.g., (79–81)]. Recent studies on the diet of juvenile white sharks are consistent with the use of NSW inshore habitats as important foraging grounds (82, 83). As seen in other shark species (84), adding to the abundance of easy-to-catch prey, the juvenile preference toward inshore habitats might also be influenced by predator avoidance (6, 73, 85). Furthermore, behavioral differences between individuals of different sizes (even if present in the same area) may also contribute to the correlation found between size and recapture probability. It has been shown that juvenile white sharks display inquisitive behaviors toward various potential food and non-food items when moving through NSW coastal areas (7, 86). Data from aggregation sites around mammal colonies also demonstrates how smaller sharks (and therefore less experienced) are generally less successful and selective while foraging [e.g., (87–89)]. This could therefore explain why smaller individuals seemingly interact more often with SMART drumlines. This possible behavioral difference is also consistent with the results from SMART drumline trials in Western Australia, where larger white sharks seem to be less susceptible to capture (90). Lastly, it is important to consider the influence of the bait used (i.e. a single sea mullet) on the size of individuals caught on SMART drumlines. The previously mentioned ontogenetic increase of mammalian prey, coupled with a decrease in teleost consumption [e.g., (79–81)], means that a single sea mullet might not be attractive to larger white sharks, contributing to the virtual absence of adult individuals captured in the present study. SMART drumline trials conducted in Western Australia also used teleosts as bait, and the authors considered that this potentially hindered the likelihood of capturing adult white sharks (90). Despite this potential limitation, the overall dominance of juvenile recaptures on SMART drumlines deployed in the coastal areas of NSW is on par with the ecology of white sharks described using methods independent from SMART drumlines (e.g., 23), while also being consistent with the overall pattern of inshore use by immature individuals from other pelagic shark species [e.g., (76–78)].

The significant effect of sex on recaptures on SMART drumlines is consistent with females being caught in greater numbers along NSW coastal waters, as shown in both acoustic and satellite tagging data (23, 26) and, albeit to a lesser extent, in long-term analysis of individuals caught in mesh nets (19). As is the case for the higher utilization of inshore areas by immature individuals, sexual segregation is also commonly observed in several sharks (91), including in the other target species of the SMP, namely tiger (92) and bull sharks (93), as both exhibit female bias in inshore areas. For white sharks, this pattern has also been reported in other regions, namely in South Africa (85, 93, 94) and in California (96, 97). Given that this female bias has been observed even within immature individuals, some studies based on acoustic data [e.g., (23, 98)] have hypothesized that sexual segregation in white sharks might occur before individuals reach maturity. The significantly higher proportion of female recaptures on SMART drumlines deployed in NSW is consistent with this hypothesis. The underlying factors promoting sexual segregation in white sharks are not fully understood (1), but some evidence suggests differences in foraging ecology between males and females [e.g., (83, 95, 98)]. Male juvenile white sharks seemingly show a greater shift in diet composition with increased size, starting to explore offshore food sources sooner than females (81). The “forage-selection” hypothesis states that differences in nutritional requirements between sexes may generate different foraging strategies, in turn promoting the occupancy of different habitats (99). According to this hypothesis, female white sharks [which mature at a larger size and tend to grow faster (100)] might seek out different prey, or consume higher quantities of it (85, 99). This can therefore explain why they seemingly use NSW inshore areas for longer, having therefore a higher chance of being recaptured on SMART drumlines.

Spatiotemporal variability of recaptures & white shark distribution in NSW

The full year of relatively constant fishing effort across the 20 SMART drumline locations (i.e., the period between April 2022–March 2023) revealed a peak in recaptures during the Austral winter and spring, along with a strong site fidelity between 29.0–32.5°S (i.e., from Evans Head to Forster), further supported by the results from the fitted GAM. This region covers five SMART drumline locations and 75 SMART drumlines deployed daily. Despite there being another 13 SMART drumline locations containing 195 lines to the south of Forster, very few animals were recaptured south of this location. This is consistent with the general description of the species' habitat use in eastern Australia (22–27) and with the proposed northward expansion of the described nursery area in the region (24, 25).

As previously stated, inshore areas with higher productivity seem to be preferred by juvenile white sharks [e.g., (6, 22, 24)], which can also explain the site fidelity to the area between Evans Head and Forster, promoting the higher number of recaptures recorded in this area. Lee et al. (26) inferred a positive correlation between areas with stronger upwelling events and white shark presence in NSW. Upwellings promote the dominance of small pelagic fish, which are vital to transfer energy to higher trophic levels (101), and these areas have already been associated with higher abundances of juveniles from other large pelagic sharks, including shortfin mako (Isurus oxyrinchus) (102). Along NSW, upwelling regimes are strongly influenced by the EAC. Where this outward, western boundary current runs parallel to the coastal shelf of NSW (i.e., north of 31°s), upwellings are common, particularly during the Austral winter (103, 104). It is, however, around the bifurcation point (~31–33°S), where the EAC splits into the stronger eastward-flowing Tasman front and the weaker southward extension (105), that upwellings are the strongest, with conditions favoring these events occurring through Austral spring (103, 104). Only by late Austral spring / early summer are conditions favoring upwellings observed in the southern locations of NSW (>33°S), and even then, these events are generally less frequent and weaker (103, 104). The spatiotemporal variability of the recaptures recorded in SMART drumlines (during April 2022–March 2023) seems to match the overall spatiotemporal variability of upwellings. The highest number of recaptures was recorded during the period when stronger upwelling events are more common (Austral winter and spring), and along the locations (namely Port Macquarie and Forster) where the EAC splits (31–33°S). The northward trend in the median latitude of recaptures recorded in this period (from Austral summer to spring) is also consistent with the spatiotemporal variability described for upwelling favorable conditions.

Another factor potentially contributing to the observed spatiotemporal variability of white sharks recaptured on SMART drumlines is water temperature. Despite the species inhabiting a wide range of SSTs (between 5–30 °C), white sharks show a marked preference toward waters between ~18–20 °C across their global range [e.g., (6, 24, 76, 106)]. Water temperature might be especially important for juveniles (6, 22), since, despite the endothermic capabilities of the species (107), smaller individuals (with a larger ratio of surface area to volume) have greater difficulty in maintaining a high body core temperature during prolonged exposure to colder waters (108). Most recaptures were recorded between 17–21 °C, in agreement with the species' preferred SST range. Furthermore, the described seasonal shift in the median latitude of recaptures recorded between April 2022–March 2023 is also consistent with white sharks inhabiting more northern latitudes of NSW during colder months, starting to move south as temperatures increase (23, 25). Similar seasonal latitudinal shifts have been described in other areas (6). Despite SST being commonly used as a predictor for white shark presence (e.g., (6, 26)], the mechanisms driving this association are not fully understood (1). As is the case for other elasmobranchs, water temperature might directly affect the species' metabolic rate and overall physiology, including its growth rate (109–111), which is crucial for juvenile survival and recruitment into the adult population (84). However, SST might also act as a proxy for other critical factors, including prey availability (24, 26). Regardless of the mechanisms driving the relationship between white shark abundance and SST, the shelf waters of NSW are facing a rate of ocean warming 3–4 times the global average, which has been directly associated with the intensification and southward shift of the EAC (112). This warming has not been uniform across the coast of NSW, being particularly prevalent in regions south of 32°S (113). This means that the changes to the EAC are directly influencing both the upwelling and the SST regimes along NSW (114), and by extension, white sharks' recaptures on SMART drumlines, and the overall species' use of NSW coastal areas. The strengthening of this current has been linked to an increased risk of white shark bites along the northern coast of eastern Australia (18), and its effect on the distribution of other target shark species for bather protection programs [bull (115), and tiger sharks (116)] has also been hypothesized.

Despite the overall site fidelity described for the broad area between Forster and Evans Head, during the period between April 2022–March 2023, there were few consecutive recaptures recorded in the same location. This may be a consequence of the “fright and flight” response of white sharks following capture, which is described as a rapid offshore movement post-release (86, 117). It is also consistent with the described broader movement patterns of white sharks throughout coastal areas [e.g., (25, 118, 119)], contradicting the preconception that individuals remain within the same locations over long periods of time. Even though there was no statistically significant correlation between the location and whether individuals were recaptured consecutively, the large portion of individuals consecutively recaptured in Forster (including the only two individuals-ID 591 and ID 613-with more than two consecutive recaptures in the same location) implies how juvenile white sharks may have a stronger site fidelity to this location. Along with being at the bifurcation point where the EAC splits, one of the main features of the wider Forster region is the estuary system of Wallis Lake. This coastal lagoon has been shown to be extremely productive and has higher abundances of potential white shark prey species, even when compared with other estuaries in NSW (120). There is evidence supporting the utilization of areas around productive estuaries by immature white sharks in NSW (27), and other studies have also hypothesized how the Wallis Lake system might promote higher local abundances of this species (121). Forster might therefore present an ideal location for white sharks foraging along NSW, resulting in a stronger site fidelity, and therefore a higher number of consecutive recaptures. The fact that there were also no significant differences between the sex or size of individuals with consecutive events vs. those without consecutive events in the same location also demonstrates that, rather than the individual's characteristics, specific habitat features are more likely to promote higher site fidelity to a particular location.

Time-at-liberty, linear distance, and latitudinal shift between recaptures & white shark movements in NSW

Predicting movements from capture-mark-recapture has very clear inherent limitations (41). This is especially true in the current study, considering the high movement capacity of white sharks, able to cover distances greater than 10,000 km in a single year (25). However, capture-mark-recapture data can still help to infer general movement patterns (41). For the present study, the description of time-at-liberty, linear distance, and latitudinal shift between recaptures further supports the previously described patterns of white sharks' habitat utilization in NSW.

Considering all recaptures recorded since the trial period of the SMP (i.e., 278 events), over 75% were recorded within a year, and for over 40% of individuals with multiple recaptures, all their events were also within a year. This is consistent with the idea that after a certain stage, larger/older white sharks seemingly depart from NSW shelf areas (22, 23) or they may become less susceptible to capture (89). However, the wide range of days between recaptures (5–1,384 days), provides evidence for some level of intraspecific variability, as some sharks (albeit less commonly) were recaptured more than 1,000 days later, demonstrating how some individuals (particularly females) use or return to the inshore areas of NSW across multiple years, possibly associated with their annual movement along eastern Australia (25). The six individuals recaptured multiple times over more than 1,000 days apart were all either YOY or small (< 185 cm FL) juveniles when initially caught. They were also all very close to 225 cm FL when recaptured for the last time. This means that despite being recaptured multiple times over more than 3 years apart, these events are still consistent with larger/older individuals being less commonly recaptured. Furthermore, the fact that of these six individuals, only one was a male, also corroborates the previously described pattern of sexual segregation, where females likely stay or return to NSW for longer periods. However, it is noteworthy to mention another potential limitation of this study, related to tag shedding (122). Since no effort was made to incorporate photo-ID, and the physical recaptures were not cross-referenced with acoustic detections, it is impossible to evaluate tag retention. Most recaptures being recorded within a single year could be an artifact of using single-barbed plastic identification dart tags, since these show high levels of shedding for long-term deployments (>1 year) (122). Even the observed female bias in recaptured individuals might be a consequence of tag shedding, considering that females have thicker skin than males (123). Despite this inherent limitation, the average size of the individuals when recaptured for the last time is extremely similar to the size at which individuals seemingly move away from inshore areas of NSW, as inferred from acoustic and satellite data (23). This means that for most individuals, the tag was likely reattained during the period when individuals are more likely to interact with SMART drumlines. Therefore, there is a strong argument for the validity of comparing our results with the species' ecology inferred using other methodologies. Furthermore, the more in-depth analysis conducted on the spatiotemporal variability (described in the previous subtitle), and for the times-at-liberty, linear distances, and latitudinal shifts (described in the following paragraphs), was conducted for recaptures recorded between April 2022–March 2023. This comprises a relatively shorter deployment period (1 year), somewhat mitigating the potential effect of tag shedding (122).

For recaptures of individuals initially caught after April 2022, the time-at-liberty and linear distances between events further corroborate the patterns of spatiotemporal variability described previously (22–27). Most recaptures were recorded within a season, and close to 90% of events were recorded within 400 km. Considering that SMART drumlines are deployed across almost the entire NSW coastline (~1,300 km), this further supports the fidelity to the ~350 km of coastline between Forster and Evans Head, particularly during the Austral winter and spring. The significantly higher portion of northward travels between recaptures is also consistent with the higher utilization of the northern areas of NSW. It might also be a consequence of the observed northward shift in the median latitudes of recaptures recorded between April 2022–March 2023. As previously stated, this shift is most likely influenced by upwelling and SST regimes along NSW, both influenced by the EAC (103, 104). These animals are potentially migrating further offshore as they come around the south-eastern point of Australia and moving up the NSW coast toward the warmer and more productive northern areas of NSW. This area forms the southern limits of the described white shark nursery (24, 25). Similar dispersal and presence patterns have also been reported for tagged white sharks detected on acoustic stations throughout NSW (24).

Sex was only significantly correlated with the linear distance until the first recapture. There were no significant differences between the time-at-liberty until the first recapture or the proportion of northward vs. southward travels between recaptures between females and males. Females in coastal bays in South Africa were observed to cover larger ranges when foraging, compared with males (95). It is possible that the same holds true for females in NSW, explaining their longer linear distances between recapture events on SMART drumlines. There was also no significant correlation observed between the FL of individuals and the times-at-liberty, linear distances, or direction of the latitudinal shift between recaptures. The general lack of a significant correlation between the estimated travels between recaptures and the sex or size of individuals can be interpreted as further evidence for the similarities between the movement ecology of smaller immature individuals within inshore areas of NSW, prior to the described sexual segregation (24, 25, 27).

Finally, the times-at-liberty and linear distances associated with ID 553 and ID 613 (i.e., the two individuals with the overall highest number of recaptures), while showcasing two contrasting scenarios, are consistent with the inferences made about white shark behavior and movement ecology in relation to SMART drumlines along the east coast of Australia. Both individuals were females <250 cm FL after their last recapture. Also, despite the large number of recaptures, both individuals were recorded over a relatively short period, with ID 613 having all five recaptures within 390 days and ID 553 having its four recaptures within 659 days. Individual ID 553, was recorded at some of the locations with the highest overall number of recaptures, but had no consecutive events in the same location. However, ID 613 had five consecutive recaptures in Forster, which further demonstrates the possible stronger local site fidelity to this specific location. The recaptures of ID 613 were also recorded during the Austral winter and spring of two consecutive years, which is consistent with the described seasonal use of the NSW coast.

Implications for human-wildlife conflicts and future directions

Our results corroborate the importance of NSW coastal area for white sharks. In doing so, we provide further evidence of the need to continue to manage this species, which involves a fine balance between conservation and ensuring bather protection (3, 19). Despite some inherent limitations, the inferred effect of size and sex on white shark recaptures on SMART drumlines is most likely a consequence of the species ecology, rather than an artifact of the methodology used. Any current or future management have therefore the potential to affect distinct age classes and sexes differently, given the higher utilization of NSW inshore areas by smaller individuals (< 225 FL cm) and by females. These results are consistent with the already described (non-natal) nursery for the Australasian white shark population [see (24) and (25)], which is also a strong argument for the continuous use of non-lethal bather protection measures, such as SMART drumlines (3, 19). Along with corroborating the description of white sharks' ecology in NSW, emphasizing the importance of appropriate management for the species, this study also highlights new avenues for white shark research. As with other studies (e.g., 22, 27, 85), while the patterns of size and sexual segregation are clearly observable, the mechanisms driving these patterns are still not fully understood. This is also true for the possible correlation between white shark abundance and the EAC [see also (26)]. Given the ongoing changes of the EAC flow (114), modeling its potential impact on the distribution of white sharks is prudent for the overall management of the species, especially for bather protection measures in NSW. Furthermore, the potential stronger local site fidelity to the Forster region provides evidence of how even relatively small regional features [such as the Wallis Lake system (121)] may influence the habitat utilization of this large vagile predator, which presents an interesting question for future studies.

Conclusions

This study aimed to quantify white shark movement/behavior ecology using capture-mark-recapture data from SMART drumlines in NSW. Despite possible limitations (namely related to the bait used and with tag shedding), we successfully identified that smaller individuals (FL < 225 cm) and females are being more frequently recaptured. Our results describe a strong spatiotemporal variability for recaptures along the northern coast of NSW, during the Austral winter and spring. We also detailed times-at-liberty, linear distances, and latitudinal shifts between recaptures, matching the previously described ecology for the species. By helping to identify key areas for the species, and infer the biotic and abiotic mechanisms driving their distribution, all framed within the broader context of species conservation and bather protection, this study addressed key research priorities for white sharks. It therefore demonstrates the extensive value of capture-mark-recapture data collected from SMART drumlines to inform shark management.