The study area comprises ~190 km of the 2,137 km NSW coastline (Figure 1) that encompass three major coastal cities and a total of 51 beaches with a single shark net set off each beach. Shark nets are 150 m long and 6 m in height and set in the bottom half of the water column on the sandy seabed approximately 500 m offshore in 10–12 m depth water (Krogh and Reid, 1996). Catch data used for analyses was provided directly by the NSW government to the research team and spanned 22 years from 1998 to 2019. Throughout this time, nets were only in the water for 8 months of the year (01 September to 30 April) as they are removed over winter to reduce potential whale entanglements during historically low months of beach use by swimmers, albeit wetsuits enable surfers to surf year-round. During the first 11 years of sampling, i.e., 1998–2008 (hereafter Period 1), nets were deployed for a minimum of 9 weekdays and every weekend per month. During this period, the maximum time that nets could fish without being checked was 4 days (96 h). From January 2009 onwards (hereafter referred as Period 2) nets were continuously in the water, also from September to April, but checked every 72 h as an effort to reduce mortality of captured individuals (Reid et al., 2011). Detailed descriptions of the SMP methodology can be found in Reid et al. (2011). Live sharks were sexed, measured to fork length and released. Maturity of sharks was determined using published sex-specific maturation sizes (Kohler et al., 1996; Lucifora et al., 2002; Cruz-Martínez et al., 2005; Holmes et al., 2015).

Among the 51 beaches meshed within the SMP several are in close proximity. To avoid bias in catch data, we grouped nets that were within 1 km of each other into a single netted area, producing 41 independent netted areas for our analyses (Figure 1). This grouping resulted in uneven effort, e.g., areas with two or more nets vs. single net areas (Supplementary Table 1). Therefore, fishing effort was standardised to metres of net per fishing day (i.e., 4-day for Period 1 and 3-day for Period 2).

Total catch data comprised only the days in which a net was checked and at least one individual from any of the four shark species captured, so we conducted a second standardisation for analysing fine-scale species-specific catch rates similar to Lee et al. (2018). As nets were checked on Monday/Tuesday and Friday prior to 2009, fine-scale effort (i.e., actual days with fishing effort) could not be accurately determined for Period 1. From 2009, cheques occurred every Monday, Wednesday, and Friday, enabling subdivision of Period 2 into three weekly time intervals (Monday and Tuesday; Wednesday and Thursday; Friday to Sunday) of fishing effort. Catch data were matched using reported dates of capture. All catches of the four shark species are included in the analysis as mortality rates of target sharks in the nets are 95% for Period 1 and 86% for Period 2 and, although live sharks are tagged and released, post-release survivorship is unknown (Dalton and Peddemors, 2018). The impacts of the SMP were only assessed for the two threatened shark species captured, i.e., grey nurse and white sharks.

Seasonality in SMP catches was related to species-specific migratory patterns along the east coast of Australia using the 8 months of fishing effort. The influence of physical characteristics of netted areas influencing shark captures in the nets were assessed using: (i) proportion of each beach that is covered by a net (net:beach) (%); (ii) percentage of rocky reef cover within a 1 km² quadrat centred on net location, downloaded from the NSW Government database (NSW Government, 2019); and (iii) distance to nearest known grey nurse shark aggregation site (Figure 1) to assess whether the species would be more vulnerable to capture in the nets deployed near to where these aggregations occur (Barker and Williamson, 2010). The environmental influence of sea surface temperature (SST; °C) was also tested (Wintner and Kerwath, 2018). Daily values of SST with 0.02° spatial resolution were downloaded from the Australian Ocean Data Network Portal (AODN, 2019) and used to calculate mean values per set during Period 2. The nearest quadrats of data available from the corresponding netted areas were used. To standardise habitat of netted areas with multiple net-sets, a 1 km² quadrat was created around the nets for each area. The percentage of reef area covered in this polygon and distance from its midpoint to the nearest grey nurse shark aggregation site, were then included in the analysis. All physical variables were calculated using QGIS software (version 3.4.0).

Data spanning the entire study period was investigated using a partial redundancy analysis (RDA). Each species was divided into sex-specific groups and maturation status prior to analysis. To account for the differences in net checking regimes between Period 1 and Period 2, a new categorical variable named SMP-period divided shark catches into before and after 2009. The RDA full model also included Area as a geographic component, given SMP distribution is effectively latitudinal (Figure 1), and the corresponding physical parameters of netted areas.

Zero-inflated generalised additive models (ZIGAM) were used to model shark fine-scale occurrence patterns. Models were calculated individually for each species, and the logarithm of fishing effort included as an offset covariate. Groups of continuous candidate predictor variables included: (i) spatial-temporal: netted area and month and (ii) environmental: SST, proportion of beach netted, proportion of reef area cover, and distance to nearest grey nurse shark aggregation site. Interactions between variables were tested to investigate whether complex relationships would better explain shark catch in the SMP. Multicollinearity was assessed with Pearson's correlations. Total number of sharks caught per net deployment were included as response variables using zero-inflated Poisson families of error distribution. Starting from the null models, new variables were gradually included to identify lower AIC values (Wagenmakers and Farrell, 2004). Final models were inspected for normal residual distribution.

To determine whether alternatives to shark nets could assist in shark conservation whilst still affording protection from being bitten by a shark, differences in captures of each species were investigated considering potential scenarios of removing particular netted areas from the SMP. For this purpose, the significant variables found to influence catches of the threatened species (i.e., grey nurse and white sharks) or modifying the netted areas with the highest captures of grey nurse sharks, were used. The values of each covariate responsible for a higher likelihood of capture were used to estimate reductions in catch rates for each species. Each scenario is hereafter referred to as potential changes, and the amount of catch reduction per species compared using ANOVA with post-hoc Tukey tests.

The effects of each potential change upon vulnerability to capture for each shark species were compared based on productivity and susceptibility scores, which are continuous indices ranging from zero to three. This approach is similar to what is used in fisheries research to assess stock vulnerability to becoming overfished (Stobutzki et al., 2001; Patrick et al., 2009; Gallagher et al., 2014). The productivity score has a biological meaning and relates to the capacity of a shark species to produce maximum sustainable yields and to recover if its population is being depleted (Patrick et al., 2009), whereas here the susceptibility score relates to the potential of a species to be impacted by their captures in the SMP. Species-specific productivity scores were obtained from Patrick et al. (2009). Susceptibility scores were calculated based on the average of the productivity score and the following attributes: (i) seasonal migrations: number of months present within the SMP region in relation to total meshing period, i.e., months captured against months with SMP effort; (ii) spatial overlap: overall area used when present in relation to the total 41 meshed areas, i.e., overlap between northernmost and southernmost locations for each species from telemetry studies and the SMP net deployments; (iii) mortality: number of alive vs. dead sharks from each species in the SMP since 2009 (i.e., the last modification of the program); (iv) desirability: 0 = non-targeted and 3 = targeted; (v) management strategy: 0 = no recovery plan in place in Australian waters and 3 = recovery plan in place in Australian waters; (vi) SMP impact: 0 = SMP is not a Key Threatening Process and 3 = SMP is a Key Threatening Process; and (vii) population impact: maximum number of individuals caught per year in the SMP in relation to the East Australian population of the species (Supplementary Table 2). The attributes were first calculated as percentages from either the SMP or published data (Lincoln-Smith and Roberts, 2010; Otway and Ellis, 2011; Tillett et al., 2012; Holmes et al., 2017; Hillary et al., 2018; Bruce et al., 2019; Lee et al., 2019), and then standardised to the same 0–3 scale used in Patrick et al. (2009) for comparison purposes. The catch reduction rates for each species from potential changes to the SMP were similarly converted to this scale in order to be also included as attributes for the final vulnerability scores, calculated using the formula:

In which X is the susceptibility score for the species in the SMP, Y is the scaled catch reduction rate, and Z is the percentage fishing effort reduction (i.e., number of nets removed from the current total 51 nets in the SMP) for the corresponding potential change (Supplementary Table 2). The significance level of the vulnerability score variations as a function of the potential changes were assessed using ANOVA.

A wicked problem matrix was developed for our case study of the SMP, providing a series of hypothetical potential changes that could be implemented to the program. These ranged from leaving the status quo, through to selective removal of nets potentially combined with alternative mitigation options, to completely removing all shark nets without replacement with any other mitigative measures. Expected consequences of potential actions for some key stakeholder groups involved in the SMP context (Fraser-Baxter and Medvecky, 2018; Simmons and Mehmet, 2018; SharkSmart NSW, 2021) and for the shark species were graded from low to high resistance and risk.

Five main stakeholder groups were identified: (i) Non-Government Organisations (NGO): supporting conservation through non-lethal measures, (ii) beach goers: people directly affected by any changes to the SMP, (iii) lifesaving associations (volunteer and professional): tasked with protecting beach goers and inevitably, the first responders in the case of a shark attack, (iv) local councils: potential beneficiaries of shark bite mitigation measures, and (v) NSW government: through the Department of Primary Industries Fisheries, is the government agency that administers the SMP. Five groups of primary actions were identified: (i) status quo: leaving the program as is, (ii) selective removal: removing nets as identified in Table 1, (iii) selective replacement: replacing some nets with alternatives, (iv) total replacement: fully removing nets and replacing with alternatives, and (v) completely removing any shark bite mitigation measures (Table 2). Alternatives were identified through stakeholder questionnaires as being the most strongly supported alternative measures as described in the current literature (Crossley et al., 2014; Gray and Gray, 2017; Simmons and Mehmet, 2018). These include Shark Management Alert in Real-time (SMART) drumlines (Guyomard et al., 2019; Tate et al., 2019), traditional drumlines (Sumpton et al., 2011; McPhee et al., 2021), helicopters (Crossley et al., 2014), drones (Butcher et al., 2019), listening stations (VR4Gs) for detecting tagged sharks and informing the public through smartphone apps in real-time (Simmons and Mehmet, 2018), and different combinations of these methods (Table 2). The alternatives are not an exhaustive list but examples of approaches that could be adopted based on current use in Australia and elsewhere (McPhee et al., 2021).

For the proposed changes, resistance levels for stakeholder groups were graded according to a traffic light scale based on results from open and targeted online questionnaires and forums, recently developed as part of the community engagement and social research by the NSW Department of Primary Industries, to better understand the opinions and attitudes of people toward coexisting with sharks (SharkSmart NSW, 2021; Simmons et al., 2021). Since the vulnerability scores calculated for each species are also returned using a traffic light scale, and directly included the corresponding scaled catch reduction rates of each proposed change, these were employed for the grading system of risk for sharks. Finally, the risk level of alternative mitigation strategies to each shark species were based on the literature as to whether they could lead to higher catch rates (i.e., high risk), lower catch rates (i.e., low risk). When information was not available and could not be fully determined it was scored as a medium risk. The significance levels were set at 0.05 and all statistical analyses were performed in R (version 4.0.5).

A total of 106 grey nurse, 217 white, 48 tiger, and 22 bull sharks were confirmed as caught during the study. Bull sharks were historically included in a “whaler shark” category and have only been identified to species level since 2010 (Lee et al., 2019), partly explaining the lower numbers caught in comparison to the other species, therefore bull shark species-specific data were only available for Period 2. More sharks were caught during Period 2 (72.5%). The total number of sharks found alive when nets were checked, and subsequently released, increased from 12.0% during Period 1–40.3% in Period 2. The total bull shark catch was low but equally distributed along the SMP region (Figure 1). Captures for the other species were widespread throughout the SMP region with localised peaks (Figure 1). Captures of grey nurse sharks were highest at Stockton Beach (Area-1; 0.7 sharks/year), followed by Bondi (Area-30; 0.5 sharks/year). Terrigal Beach (Area-13) was the only location near a known grey nurse shark aggregation site (0.7 km) with slightly higher captures of the species (Figure 1) but it still only had an average catch rate of 0.3 sharks/year. White shark catch was also highest at Stockton Beach (1.6 sharks/year), followed by Wattamolla (Area-35; 0.8 sharks/year) (Figure 1). Tiger sharks were mostly caught at Wattamolla (0.4 sharks/year) and Garie (Area-36; 0.5 sharks/year) beaches (Figure 1).

Catches of grey nurse sharks were dominated by females (female:male = 5.93; Figure 2A). More even sex ratios were observed for white (female:male = 1.14; Figure 2B), tiger (female:male = 1.09; Figure 2C), and bull (female:male = 1.71, Figure 2D) sharks. While most grey nurse, tiger, and bull sharks caught were mature, white sharks were almost entirely juveniles with only 5 (2.3%) mature sharks caught during the entire study period (Figure 2). According to Dudley and Cliff (2010), bull sharks larger than 175.5 cm fork length are capable of severe tissue removal and should be considered as potentially hazardous to humans. Most white, tiger, and bull sharks caught in the SMP were indeed larger than 175.5 cm fork length (Figure 2).

The final RDA model revealed that the variables Maturation Status (F = 272.8; p = 0.005), month (F = 13.9; p = 0.005), SMP-period (F = 9.7; p = 0.005) and Area (F = 2.5; p = 0.041) significantly influenced the total shark catches in the SMP (variance explained = 46.8%). While most groups showed similar trends, the patterns of female grey nurse, and male white and tiger sharks varied significantly (Figure 3). Female grey nurse sharks were predominantly adult and occurred more frequently during Period 2 (Figure 3). Similarly, male juvenile white sharks were more frequent during Period 2, but also during the first months of netting and in the northern area of the SMP (Figure 3). By contrast, male tiger sharks were caught toward the end of the netting season, especially during Period 1 (Figure 3).

The capture of grey nurse sharks in the nets was influenced by interactions between netted area and month (deviance explained = 48.0%) and SST and proportion of reef substrate (84.6%) (Supplementary Tables 3, 4). Captures of grey nurse sharks had a northerly peak between September and January and a southerly peak in September (Figure 4A). The southern peak occurred between the aggregation sites of Long Reef and Magic Point, yet over 20 years only seven grey nurse sharks were caught at Area-13 (Terrigal Beach, 1.4 km from Foggies aggregation site) and three at Area-33 (Maroubra Beach, 1.3 km from Magic Point). Significantly lower grey nurse shark captures occurred between the aggregation sites of Foggies and Long Reef, particularly from January to April (Figure 4A). Overall, distance to nearest aggregation site did not significantly influence captures of grey nurse sharks in the SMP (Supplementary Tables 3, 4). However, captures of the species were related to cooler SSTs (<20°C) and exhibited a bi-modal topographic pattern, either more than 40% reef cover or when the seabed comprised <10% reef cover (i.e., sandy substrate) (Figure 4B).

White shark occurrence was influenced by the interaction between netted area and month (46.1%) and the environmental variables SST and proportion of beach netted (84.9%) (Supplementary Tables 3, 5). White shark catches exhibited a northerly peak between September and December and a southerly peak between January and April (Figure 4C). White sharks were more frequent when temperatures were cooler than 20°C (Figure 4D) particularly in areas with proportion of beach netted <10% and >30% (Figure 4E). Tiger shark catches were influenced by netted area (35.7%) (Supplementary Tables 3, 6), being mostly caught south of netted area 20 (Figure 4F).

We identified 12 scenarios where modifying the SMP could reduce captures of threatened sharks. These were divided into four significant groups according to the Tukey testing (Table 1). Removing all nets would reduce catches of grey nurse by 5.3 (±6.2) and white sharks by 9.8 (±7.3) sharks/year. It would, however, also reduce catches of tiger sharks by 2.5 (±1.3) and bull sharks by 2.4 (±1.8) sharks/year. On the other hand, removing nets from just Stockton, Bondi or both beaches (effort reductions between 2 and 4%; Tukey B) would lead to similar reductions to that achieved by reducing effort by 12–16% through removal of up to eight nets (Table 1). Stockton (3.5 ± 2.2 sharks/year) and Bondi (1.7 ± 0.8 sharks/year) beaches accounted for ~30% of all grey nurse shark catches. Removing nets solely from Stockton would result in higher reductions in grey nurse shark catch and also lower the capture of white sharks, while removing Bondi alone would have a much less pronounced effect on white sharks (Table 1). Removing more nets (22 to 33% effort reduction; Tukey C) could reduce grey nurse shark captures by a maximum ~50% (Table 1), but it would also greatly reduce captures of potentially dangerous sharks (Table 1), thereby feasibly increasing risk of shark bite if no alternative risk reduction measures are implemented.

The potential changes would effectively reduce vulnerability scores for all species (F = 37.91; p < 0.001) (Figure 5). The status quo would maintain current impacts. At the other extreme, totally removing mitigation measures would completely alleviate impacts from the SMP (Table 2; Figure 5). However, this might increase bather vulnerability, inviting a backlash if serious injuries or fatalities followed. Intermediate options include switching from nets to using more selective, or less lethal, fishing methods, but how this might influence catch rates is unknown as comparable data is not available.

Conventional drumlines that are operated in a manner that generally kills sharks are unpopular with the public (Table 2) (Gibbs and Warren, 2014). SMART drumlines significantly reduce mortality by providing a means of promptly responding to a catch (Guyomard et al., 2019). Selective replacement with SMART drumlines at specific beaches was the most acceptable modification to the SMP across all stakeholder groups (Table 2). However, many beaches are located further than the ~15 km required to reach SMART drumlines and successfully release animals alive (Tate et al., 2019). Removal and replacement of the northern-most beach, Stockton, provides the most conservation outcome for the least reduction in effort, but simultaneously reduces the catch of what is arguably the primary target species of the SMP (i.e., white sharks), thus emphasising the wicked nature of this problem (Table 1; Figure 4). Stockton is adjacent to a port and outside of flight restriction zones, hence alternative mitigation measures could include SMART drumlines and drones (Simmons and Mehmet, 2018). Similarly, replacement of the Bondi net alone would provide substantial benefits to grey nurse sharks (Table 1).

Climate change is increasing the frequency and severity of human-wildlife conflicts by forcing animals and people to share increasingly scarce resources (Abrahms, 2021). Most of the shark catches in the SMP occurred in more recent years (i.e., Period 2). While this could be associated with a higher fishing effort (i.e., nets spending more time in the water during Period 2 compared to Period 1), it may also be related to changes in distributions of potentially dangerous sharks along the East coast of Australia in response to climate change (Niella et al., 2020).

The nearshore occurrence of both threatened shark species increases during periods of water temperatures below 20°C (Lee et al., 2018), particularly in rocky reef and sandy bottoms areas. Our findings corroborate catches of grey nurse sharks in the South African bather protection program (Wintner and Kerwath, 2018), with increased catch apparently driven by cold water upwellings (Roughan and Middleton, 2002; Oke and Griffin, 2011), particularly in the north of the SMP area (Everett et al., 2014). Juvenile white sharks occupy this region as part of their nursery (Bruce and Bradford, 2008, 2012) suggesting increased catch may be related to their exploration of their nursery area. Grey nurse sharks migrate southwards during the spring-summer period (Otway and Ellis, 2011).

White sharks were more frequent in meshed areas with higher proportions of beach covered by the SMP nets (proportion of beach netted > 40%). This, combined with higher occurrence during productive cold-water periods, suggests that when following their prey into the shallows these sharks become more vulnerable to capture. Even though proportion of beach netted was not a significant variable for grey nurse sharks, the southern increase in captures occurred at beaches with a high proportion of the beach covered by the net (Figure 1) and removing these predicted a reduction of catch of ~40% for the species (Table 1). Because SMP nets are set close to the beach, these sharks might have a higher likelihood of being caught at these shorter beaches as the reef is in proximity on either side of the net. Instead of being removed, these nets could be moved to deeper waters in an attempt to reduce captures of grey nurse sharks.

The southern region caught more tiger and white sharks, with white sharks mostly caught during the latter half of the season (January to April) coincident with known broadscale movements (Bruce et al., 2019), supporting South African research suggesting white shark catch was related to longshore movements (Cliff et al., 1989). The increased catch of both white sharks and grey nurse sharks in the nets on either extremity of the SMP region may therefore be due to the “boundary effect” first described by Cliff et al. (1989). The boundary effect proposes that sharks are more likely to be captured when they first encounter nets on the edge of these programs. Removing nets from extreme locations would not necessarily reduce shark catches but simply move the boundary. A more effective approach to reduce mortality of threatened species may rather be to change practises, for example deploying alternative mitigation strategies.

Management of human-wildlife conflicts requires that both the human and ecological dimensions of the problem are properly identified and accounted for (Carter et al., 2014; Teixeira et al., 2021). Recognising local attitudes toward the animals involved is often among the main strategies to tackle these situations (Madden, 2008; Goodale et al., 2015; Dillon et al., 2016). This includes tolerance of wildlife presence (Bruskotter and Wilson, 2014; Slagle and Bruskotter, 2019), which in turn is affected by personal experiences such as being involved in direct contacts or attacks by these animals (Treves and Bruskotter, 2014; Dorresteijn et al., 2016).

At one end of our matrix, maintaining the status quo of the SMP would result in ongoing catches of threatened grey nurse and white sharks, and zero impact on reducing this Key Threatening Process to their population viability. In Australia, there are strong pressures toward adapting bather protection programs, particularly in Western Australia and New South Wales (Fraser-Baxter and Medvecky, 2018), suggesting that this option would not be supported by NGOs nor beach goers. Simultaneously, this option has limited value to the government organisation charged with enhancing the conservation status of these species. However, despite public and NGOs being supportive of taking nets out of the water, complete removal of the SMP may exacerbate the wicked problem due to the possibility of increased risks to human safety. Therefore, we focus on the evidence supporting the replacement of nets with alternatives.

It is unlikely that all nets could be replaced by new methods as logistically replacing nets with SMART drumlines off beaches further than 15 km from a port is unfeasible if sharks are to survive (Tate et al., 2019). However, traditional drumlines could be installed at remote beaches as they do not require immediate response following a capture and incur lower mortality than nets (Sumpton et al., 2011). Survivorship in traditional drumlines off Queensland varies considerably (grey nurse = 50%, white = 47.5%, tiger = 31%, bull = 25.9%; Sumpton et al., 2011). They may represent a compromise by reducing mortalities of threatened species while still providing some mitigation (Cliff and Dudley, 2011), however they are generally not supported by the majority of the community (Gibbs and Warren, 2014).

The release of potentially dangerous sharks did not directly correlate with higher incidence of shark bites in South Africa (Cliff and Dudley, 2011), nor has it in NSW since it was formally implemented in 2009, supporting the use of non-lethal measures for mitigating shark hazard. Although approximately half of grey nurse sharks are released following capture in the SMP (Dalton and Peddemors, 2018), there is little information on post-release survivorship. Grey nurse sharks are prone to stress-induced lactic acid build-up and subsequent mortality following capture (Smale et al., 2012; Otway, 2015). Apparently positive post-release survival from the SMP (Dalton and Peddemors, 2018) and the two trials of shark nets on the NSW North Coast (NSW DPI, 2017, 2018) suggests that SMART drumlines will lead to even higher post-release survivorship for this species. SMART drumlines do not appear to attract potentially dangerous sharks into nearshore areas, however, neither are they or any other existing mitigation measure capable of completely stopping sharks from using a particular area (Guyomard et al., 2020). Combining SMART drumlines with aerial surveillance (Simmons and Mehmet, 2018) could provide a potential alternative to shark nets.

Drones are a cost-effective substitute for manned aircraft, however, their effectiveness still depends on the ability of the pilot to identify a shark in variable environmental conditions (Butcher et al., 2019). The development of artificial intelligence is likely to reduce reliance on pilots in future, but there will also still be issues related to proximity with airports and other critical infrastructure. Sole reliance on aerial surveillance may, therefore, not be practical in many coastal regions. Drones have been shown to be capable of acting as aerial support for shark detection in nearshore waters (Butcher et al., 2019; Gorkin et al., 2020), however, until flight automation becomes more advanced it is unlikely to be considered viable at understaffed beaches for beach authorities. They could, however, be applied in combination with SMART drumlines off more popular beaches, e.g., Bondi Beach where high grey nurse shark and low white shark captures, coupled with the popularity of this Australian beach, point to a high likelihood of success.

Recognising the ecological patterns underlying a conservation issue while including knowledge of human behaviour is an effective strategy for solving wicked problems, but the science must be evidence-based (Game et al., 2014; Mason et al., 2018). Grey nurse sharks aggregate at particular locations along the east coast of Australia (Bansemer and Bennett, 2008) leading to calls to remove nets near aggregation sites (Barker and Williamson, 2010). However, this and previous studies (Green et al., 2009) found no evidence that proximity to aggregation sites had any influence on capture probability (Supplementary Table 4), implying that removal of these nets would be an ineffective strategy to reduce the impacts of the SMP, possibly misdirecting efforts. This argues strongly for evidence-based conservation decision making, particularly when sensitive to public scrutiny. It is important to emphasise that the nets are not a barrier that prevents sharks from interacting with beach goers as they do not cover the entire beach, rather they are a fishing agent. Research trials of shark nets on the NSW north coast found that only 6 out of 145 animals (4%) were caught on the shoreward side of the nets (NSW DPI, 2018). By contrast in South Africa, 35% of the sharks caught were done so on the shoreward side of the nets, i.e., when moving out to sea (Dudley, 1997). While removing nets might be considered controversial for risk-averse government agencies, it is evident from our findings that management efforts focused at only two beaches (i.e., Stockton and Bondi) could substantially improve shark conservation with little projected risk for people in this region, provided alternative mitigation measures are implemented in their place.

In human-wildlife interactions people are not only part of the problem but the essential component to finding effective solutions toward coexistence (Frank and Glikman, 2019). Many human-wildlife conflicts involve humans venturing into wildlife habitat, or exploiting areas in new ways, whether becoming exposed to sharks in the sea or large dangerous wildlife on land. In the case of shark bite risk, while management agencies are tasked with developing public safety strategies the public can also contribute to lowering their own risk, for example, by wearing personal electronic deterrents (Bradshaw et al., 2021). Balancing human risk with increasingly disturbed environments and ever diminishing populations of wildlife (Bar-On et al., 2018) requires robust approaches. Effectively measuring stakeholder attitudes toward potential alternatives for human-wildlife conflict can be challenging, as they may have divergent perspectives toward wildlife (Ceauşu et al., 2019), and these might further vary with time depending on people's personal experiences, culture and changes in beliefs. It is also important to weight the impacts and trade-offs involved in the risk of conflicts during recreational activities (Kubo and Shoji, 2014), including their decision making process around the threat of shark bites. Under the context of the SMP, no single definite solution could be found to replace current methods as some stakeholders, including members of the public, were found to agree with the use of shark nets (Gray and Gray, 2017). Any changes to the program are likely to be location-specific, depending on feasibility of use for each of the alternatives proposed in this study, whilst simultaneously accounting for the attitudes of the people involved. While we acknowledge that our findings might not be definitive, we hope they will provide decision makers with sufficient understanding to move toward a more holistic approach to managing marine ecosystems in which humans and sharks can coexist. The framework adopted in our study incorporates (i) identification of species-specific occurrence patterns, (ii) determining potential actions toward reducing impacts for wildlife while keeping people safe, (iii) evaluation of stakeholder attitudes toward proposed alternatives. Our results indicate that integrating this framework into human-wildlife conflict resolution strategies is not only applicable to other programs of mitigating shark bite risk, but also for managing human-wildlife conflicts in other situations where empirical data on species occurrence and ecology are available.