Development of an environmental DNA assay for detecting multiple shark species involved in human– shark conflicts in Australia

The number of human– shark interactions has increased

increasing globally (McPhee, 2014). The most problematic shark species include the white shark (Carcharodon carcharias Linnaeus 1758), tiger shark (Galeocerdo cuvier Péron & Lesueur 1822), and bull shark (Carcharhinus leucas Müller & Henle 1839), which are responsible for ~56% of all bites and ~98% of all fatalities over the last three decades globally (McPhee, 2014). These three species co-occur in some temperate and subtropical waters around the world, including the eastern seaboard of Australia, a region recognized as one of the world's shark attack hotspots (Chapman & McPhee, 2016).
While risks to human safety are generally overstated, the increasing frequency in shark attacks has led to public demand for programs aimed at suppressing risks to beachgoers (Colefax et al., 2020;Fraser-Baxter & Medvecky, 2018;West, 2011). These include targeted culling programs, which compromise conservation efforts (Cliff & Dudley, 2011;Gibbs et al., 2020), and non-destructive monitoring programs, which allow for the monitoring of shark movements and collection of critical demographic data (Colefax et al., 2020;Spaet et al., 2020aSpaet et al., , 2020bTate et al., 2019). The most successful non-lethal monitoring program to date has been the Shark

Management Strategy (SMS), which currently operates in eastern
Australia and is one of the shark mitigation programs run by the New South Wales Government. Initiated in 2015, the SMS program has adopted a combination of SMART (Shark-Management-Alert-in-Real-Time) drumlines (Tate et al., 2019), acoustic and satellite tagging (Spaet et al., 2020a(Spaet et al., , 2020b, and drones Colefax et al., 2020) to monitor the movement patterns of individual sharks and communicate reports of near-shore shark visitations to the general public via the purposely built SharkSmart app (www.shark smart. com.au). As of November 2020, 544 C. carcharias, 131 G. cuvier, and 92 C. leucas have been tagged and tracked, and the biological data from individual sharks continue to provide valuable insights into the demographic structure of coastal populations Colefax et al., 2020;Spaet et al., 2020a). However, the program has already come at significant cost to the State Government with over $16 million invested in a 5-year period and an additional $8 million for 2020/21. While the size of resident C. leucas and G. cuvier populations remain uncertain, it has been estimated that approximately 2500-6750 C. carcharias persist on the Australian east coast (Hillary et al., 2018). These figures suggest the number of SMS program tagged sharks represent only a fraction of the total population, highlighting the need for more innovative and cost-effective methods for the detection and monitoring of problem shark species in eastern Australia.
Environmental DNA (eDNA) technologies are revolutionizing the field of wildlife monitoring, providing unprecedented sensitivity for characterizing species presence through the detection of genetic material that organisms shed or excrete into their surrounding environment. The uptake of this technology has increased dramatically in recent years and has been applied widely to surveys of rare or invasive species in freshwater environments (Bohmann et al., 2014;Deiner et al., 2016;Valentini et al., 2016), as well as marine environments, including large pelagic species, providing new insights into species ecology (Berger et al., 2020;Foote et al., 2012;Sigsgaard et al., 2017;Thomsen et al., 2016). Several studies have demonstrated the utility of eDNA technologies for detecting shark species in both shallow and deep-water environments (Bakker et al., 2017;Boussarie et al., 2018;Truelove et al., 2019). Recent examples of eDNA technologies being used for detecting C. carcharias (Lafferty et al., 2018) and C. leucas (Schweiss et al., 2020) have recently emerged. However, in regions where there is sympatric occurrence of multiple target shark species, careful consideration needs to be given to the specificity of eDNA assays in order to reliably discriminate between co-occurring species. This is particularly pertinent in eastern Australia where there is significant overlap in seasonality and habitat use of C. carcharias, G. cuvier, and C. leucas (Espinoza et al., 2016;Holmes et al., 2014;Lee et al., 2019;Lipscombe et al., 2020;Spaet et al., 2020aSpaet et al., , 2020b.
Here, we report the development of a multispecies eDNA assay for simultaneously detecting target shark species in one of the world's major shark attack hotspots. We demonstrate the utility of our assay for detecting and discriminating between C. carcharias, C. leucas, and G. cuvier in eastern Australia using a combination of in silico, laboratory, and field trials. We validate the specificity of the assay by testing for cross-amplification across a range of non-target but co-occurring shark species native to eastern Australia and test the sensitivity of the assay on water samples collected around shark capture events. We then apply the eDNA assay in blind surveys at a known shark visitation hotspot at a seasonal timepoint when all three target species are known to occur, demonstrating the capability of simultaneously detecting these species. This new eDNA multispecies assay has the potential to dramatically enhance shark detection and alerting capabilities in shark attack hotspots such as eastern Australia, providing a rapid, cost-effective, and non-invasive alternative to traditional survey methods. We expect the application of the assay could help reduce risks of shark attack and allow further insights into patterns of species movement, near-shore visitation, and habitat use for informing future conservation management. Kit protocol with the following minor adjustments: 500 μl AL buffer, 500 μl ethanol, and final elution step of 100 μl AE buffer for each sample. Extracted DNA samples were stored at −20°C until required for genetic analysis.

| eDNA assays development
Environmental DNA assays have previously been developed for C. carcharias in California (Lafferty et al., 2018) and C. leucas in the Gulf of Mexico (Schweiss et al., 2020). We initially tested these assays to determine their specificity to C. carcharias, C. leucas, and non-target shark species in our study region of northern New South Wales, Australia. Initial in silico analyses of each assay suggested potential compatibility (matching primer and probe sequences) with various non-target shark species native to Australia. Quantitative polymerase chain reaction (qPCR) subsequently confirmed nontarget cross-amplification in both assays (see results). We therefore downloaded the complete mitochondrial genome sequences from GenBank (www.ncbi.nlm.nih.gov) for our three target species and seven non-target shark species known from our study region to design species-specific assays. Unique regions were first identified by aligning all genomes in Geneious (vers. 10.2.5; https://www.genei ous.com). Once target regions were identified, assays were designed using the custom TaqMan® Assay Design tool (https://www.therm ofish er.com/order/ custo m-genom ic-produ cts/tools/ cadt/) and primer and probe specificity was checked in silico using primerblast

| Primary efficiency, limit of detection, and quantification
Primer efficiency, limit of detection (LOD), and limit of quantification (LOQ) were assessed following the protocol and curve fitting method described in Klymus et al. (2020). For the standard curve, serial dilutions of gDNA derived from tissue extractions were prepared in elution buffer AE, Qiagen. The 10-fold dilution series spanned over five orders of magnitude, ranging from 1000 to 0.

| Specificity testing
We tested the specificity of each assay on 10 picograms of gDNA extracted from tissue samples from two individuals for each of 11 shark species native to eastern Australia (captured and provided by the SMS team).

| Field eDNA samples
We validated our eDNA assay with 30 x 1 L water samples collected from sites off the coast of northern NSW (Figure 1). We sampled water adjacent to five captured G. cuvier and three C. carcharias that were being processed next to a vessel ~500 m from the shore as part tect tagged individuals within a ~500 m radius (Spaet et al., 2020b).
Within 5 min of water sample collection, each 1 L water bottle was processed by drawing water into a Hapool 60-ml sterilized disposable syringe (Shandong Hapool Medical Technology) and then pushing the water through an attachable (leur lock) Sterivex® 0.22 μm filter unit (Merck). This process was repeated until 1 L of water had been passed through the filter. Filter units were stored on ice in a dark storage container following collection and shipped to the laboratory for processing. Once at the laboratory, samples were stored at −20°C until DNA extraction.

| Negative field control samples
We took several negative field controls to help assess the multispecies assays. Firstly, we assessed the potential for false positives occurring during eDNA sampling on the boat. A bottle containing sterile water was opened after a shark capture, and 1 L of water was passed through a Sterivex® filter repeating the process that had been undertaken for eDNA sampling (by the same person who took all eDNA samples). This was conducted on three shark capture occasions (samples labeled C31-33).
Secondly, we sampled twice daily (~8.00 AM and ~5.00 PM) for five consecutive days at a VR4G listening station at Evans Head Products were then sequenced using Sanger sequencing (ABI 3730xl, Macrogen Korea) in dual directions using M13 primers and compared to reference sequences to confirm species haplotype authenticity and overall assay specificity.

| Assay specificity
We developed three species-specific TaqMan® assays that amplified  assay gave the same results even though we changed to PrimeTime assays and different fluorophores for C. leucas and G. cuvier.
We also tested the specificity of the C. carcharias assay from Lafferty et al. (2018) and the C. leucas assay from Schweiss et al.
(2020) on the same 11 shark species found in our study area. These assays were found to amplify their respective target species, but also non-target species; the C. carcharias assay amplified the DNA of I. oxyrinchus, while the C. leucas assay amplified DNA from both C. falciformis and C brachyurus (Table 2).

| Assay efficiency and sensitivity
The standard curves produced from the serial dilution of each target species DNA showed a linear relationship between Cq value and the log of the starting DNA concentration (R 2 > 0.99 for all three assays).

| Field eDNA samples
The multispecies assay detected G. cuvier DNA strongly in 13 field samples (T1-10, B21-22, and B30) with all technical replicate qPCRs being positive (Table 3). Weaker detections were also found in another five samples, where only one or two of the technical replicate qPCRs were positive. The highest concentrations of DNA were found in samples taken from within 2 m of the G. cuvier captures (indicated by the high DNA concentrations in Table 3). C. carcharias DNA was strongly detected in 13 field samples (T5-6, W11-20, and B22) with all technical replicate qPCRs being positive (Table 3). This All detections were above the minimum LOD for each assay.
The three negative field control samples did not detect white shark DNA despite each being taken at the same time as white sharks were captured, indicating that our sampling method is robust to contamination. Similarly, all samples taken over five consecutive days at the Evans Head VR4G listening station (positioned 500 m offshore) did not detect white, tiger, or bull shark DNA, indicating no activity of these species, which is consistent with no captures or acoustic detections in the area for a period of 7 days prior and during the sampling period.
Sanger sequencing of qPCR amplicons confirmed species haplotype authenticity in seven G. cuvier DNA samples (T1-2, T4-5, B22, B25, B30) and six C. carcharias DNA samples (T5, W14-16, B22, and B25). Detections of both G. cuvier and C. carcharias DNA in the same samples were confirmed in three samples (T5, B22, and B25). We were unable to confirm product authenticity of C. leucas qPCR detections via Sanger sequencing due to low DNA concentrations in the two samples where the species was detected.

| DISCUSS ION
The increasing frequency of human-shark interactions on the east coast of Australia has prompted investment into intensive shark monitoring programs in recent years (Colefax et al., 2020;Spaet et al., 2020b;Tate et al., 2019). In this study, we developed a highly specific multispecies eDNA assay that will provide added capabilities for detecting and discriminating between the three shark spe- autumn-winter months (Bruce et al., 2019;Spaet et al., 2020a), with near-shore visitation highest between July and December when annual sea-surface temperatures are at their lowest (Spaet et al., 2020a(Spaet et al., , 2020b. Conversely, G. cuvier tend to be found in warmer northern waters and typically move southward, with nearshore visitation peaking in summer and autumn, although they are found in NSW waters year-round (Holmes et al., 2014;Lipscombe et al., 2020). There is considerable overlap in near-shore visitation during autumn in Australia for both species, so it is not surprising that we were able to detect both species from water samples collected at this time of year. This is an important finding given the risks both species pose to beachgoer safety at similar times of the year and suggests that eDNA approaches could be used to gain a better appreciation of spatial and temporal patterns of near-shore habitat use by each species.
The eDNA detections of C. leucas from two field samples were relatively weak, compared with some of the C. carcharias and G. cuvier detections. We did not capture any C. leucas during the sampling period and therefore could not directly sample water in an area with a confirmed presence. While present all year round within the study area, C. leucas are generally in lower densities and less active during daylight hours than either C. carcharias or G. cuvier and tend to occur in adjacent river systems (Werry et al., 2011). It is also worth noting that the C. leucas qPCR assay targets a larger amplicon (228 bp) compared to the C. carcharias (128 bp) and G. cuvier (92 bp) assays, which may affect field detectability due to the propensity for larger amplicons to degrade more quickly. Unfortunately, discrimination between C. leucas mtDNA sequences and non-target shark species was difficult and was only possible with a larger amplicon (the assay in Schweiss et al. (2020) is also of a similar amplicon size). Further testing is therefore recommended to assess whether amplicon size affects detectability of C. leucas from water samples. Ideally, this would be undertaken with known presence samples (e.g., water samples adjacent to C. leucas captures).
The multispecies assay developed here appears to be more specific than other eDNA-based assays developed for C. carcharias (Lafferty et al., 2018) and C. leucas (Schweiss et al., 2020), at least in our study region. Both of these previously developed assays detected non-target shark species occurring in eastern Australia, which compromises the use of these particular assays in this region. We expect our multispecies assay to be specific to C. carcharias, C. leucas, and G. cuvier populations in other regions of the world, as our primers and probes for each species were designed based on all available data on sharks in GenBank (Table S1). However, further specificity testing in regions where shark community assemblages are notably different is required to exclude the potential for cross-amplification of local non-target shark species.
The multispecies eDNA assay developed in this study will add to the shark detection capabilities of bather protection programs and provide an alternative tool for monitoring shark populations, that currently includes drumlines for capture (Tate et al., 2019), acoustic and satellite tagging and detection (Spaet et al., 2020a(Spaet et al., , 2020b, and drones Colefax et al., 2020), all paired with the SharkSmart app to provide real-time updates to the general public. Current automated acoustic monitoring techniques are limited to the detection of tagged individuals, which are likely to be only a small fraction of the total shark population present in eastern Australia (Davenport et al., 2021). Our eDNA assay has the potential to monitor a greater proportion of the shark population, particularly along coastal regions in hotspot areas at a fraction of the cost of current traditional monitoring approaches. Recent technological advances also allow in-field detection capabilities using mobile qPCR machines (Thomas et al., 2020), offering the potential for this multispecies eDNA assay to act as an early warning system to beachgoers for shark activity in the general area. Currently, the minimum time for extracting DNA and undertaking the multispecies assay is 6-24 hr from receipt of samples in a laboratory. An infield detection system could cut this time down to 1-2 hr, making it possible for patrolled beaches to be tested routinely and results reported into shark management programs such as the SharkSmart app. Converting this assay for use with recombinase polymerase amplification (RPA) combined with a lateral flow strip detection method could further reduce time, costs and increase useability (Rohrman & Richards-Kortum, 2012). However, further research is needed to understand the spatial and temporal limitations of eDNA detections, particularly in the context of the real-time nature of a detection.
At present the primary driver for the development of innovative and cost-effective shark monitoring tools in Australia is around increasing public safety at beaches. However, strategic application of our multispecies eDNA assay has the potential to address critical knowledge gaps associated with the biology, ecology, and population dynamics of C. carcharias, C. leucas, and G. cuvier in Australia and potentially abroad. Other eDNA sampling and detection approaches (e.g., eDNA metabarcoding (Stat et al., 2017)) in combination with this multispecies assay could also be used to help understand the relationship between marine fauna and seasonal shark activity at hotspots. Similarly, targeted eDNA sampling across the species distribution (both in Australia and abroad) could greatly assist in improving our knowledge of the seasonality of shark movements and habitat use in both coastal and offshore environments.

ACK N OWLED G EM ENTS
Project funding and support were provided by the New South

AUTH O R CO NTR I B UTI O N S
All authors were involved in the design of the study and contributed to the writing of the manuscript. ARW and ADM led manuscript preparation, AvR undertook assay design, specificity testing and eDNA sample analysis, and PAB and ZC undertook fieldwork.

DATA AVA I L A B I L I T Y S TAT E M E N T
All primers/probe sequences and results are available in the main text.