Swimming against the flow—Environmental DNA can detect bull sharks (Carcharhinus leucas) across a dynamic deltaic interface

Abstract Human activities in coastal areas are accelerating ecosystem changes at an unprecedented pace, resulting in habitat loss, hydrological modifications, and predatory species declines. Understanding how these changes potentially cascade across marine and freshwater ecosystems requires knowing how mobile euryhaline species link these seemingly disparate systems. As upper trophic level predators, bull sharks (Carcharhinus leucas) play a crucial role in marine and freshwater ecosystem health. Telemetry studies in Mobile Bay, Alabama, suggest that bull sharks extensively use the northern portions of the bay, an estuarine–freshwater interface known as the Mobile‐Tensaw Delta. To assess whether bull sharks use freshwater habitats in this region, environmental DNA surveys were conducted during the dry summer and wet winter seasons in 2018. In each season, 5 × 1 L water samples were collected at each of 21 sites: five sites in Mobile Bay, six sites in the Mobile‐Tensaw Delta, and ten sites throughout the Mobile‐Tombigbee and Tensaw‐Alabama Rivers. Water samples were vacuum‐filtered, DNA extractions were performed on the particulate, and DNA extracts were analyzed with Droplet Digital™ Polymerase Chain Reaction using species‐specific primers and an internal probe to amplify a 237‐base pair fragment of the mitochondrial NADH dehydrogenase subunit 2 gene in bull sharks. One water sample collected during the summer in the Alabama River met the criteria for a positive detection, thereby confirming the presence of bull shark DNA. While preliminary, this finding suggests that bull sharks use less‐urbanized, riverine habitats up to 120 km upriver during Alabama's dry summer season.


| INTRODUC TI ON
Human alterations to the global landscape are accelerating shifts in ecosystem structure, function, and service at an unprecedented pace (Halpern et al., 2019). These trends are particularly evident in coastal areas marked by reductions in predatory species and losses of critical spawning and nursery habitats (Lotze et al., 2006).
Increased urbanization of these coastal areas further contributes to changes in habitat by modifying hydrological processes and nutrient dynamics (Lee et al., 2006). Understanding how these anthropogenic activities cascade across ecosystems requires an understanding of how mobile species might act to link adjacent, but otherwise disparate, habitats (Lundberg & Moberg, 2003).
Mobile Bay, Alabama, is a dynamic, shallow, human-impacted coastal ecosystem located in the north-central Gulf of Mexico.
Mobile Bay receives the fourth largest estuarine discharge in the continental United States (Dzwonkowski et al., 2011), 95% of which is accounted for by the Alabama and Tombigbee Rivers (Schroeder, 1978). The extensive discharge from these two rivers is also highly variable; average discharge during the wet season (late winter, early spring) is more than three times greater than average dry season discharge (late summer, early fall) (Webb & Marr, 2016).
Bull sharks are euryhaline generalists that often use freshwater environments as nursery areas (Grant et al., 2019) and thus may act as mobile links connecting the estuarine portions of Mobile Bay and freshwater reaches of the Alabama and Tombigbee Rivers. Acoustic tracking of YOY bull sharks demonstrates extensive use of the northern portion of Mobile Bay, an estuarine-freshwater interface known as the Mobile-Tensaw Delta (MTD). Freshwater from the MTD enters Mobile Bay via two river systems. The Mobile-Tombigbee river system discharges into the northwestern portion of Mobile Bay, along the industrial shores of the Port of Mobile. In contrast, the Tensaw-Alabama river system discharges into the northeastern portion of Mobile Bay, an area with considerably less development (Ellis et al., 2011). Previous telemetry work suggests small-scale habitat selection across these two adjacent areas. Acoustically tagged bull sharks were more frequently detected along the Tensaw-Alabama portion of the MTD compared to the Mobile-Tombigbee system (Drymon et al., 2014). This pattern suggests that YOY bull sharks may be linking freshwater and estuarine habitats in the MTD, but not equally across these two river systems. Determining how YOY bull sharks connect these habitats is critical given the role of mobile links in ecosystem resilience (Lundberg & Moberg, 2003), yet cost-prohibitive when using traditional techniques such as fisheries-independent monitoring or passive acoustic telemetry. Therefore, the objective of the current study was to use a newly developed bull shark environmental DNA (eDNA) assay (see Schweiss et al., 2020) to examine the potential for eDNA approaches to detect bull sharks across this highly dynamic deltaic interface.

| ME THODS
Water samples were collected at a series of estuarine, deltaic, and freshwater sites in Alabama once in the wet winter season (February 19-20, 2018) and once in the dry summer season (August 21-22, 2018). In total, five estuarine sites were sampled from Dauphin Island to the northern extent of Mobile Bay, six sites were sampled within the MTD, and ten freshwater sites were sampled in two river systems: the Mobile-Tombigbee and the Tensaw-Alabama. In each river system, sites spanned ~190 km north of the mouth to just south of the Coffeeville Lock and Dam and Claiborne Lock and Dam in the Tombigbee and Alabama Rivers, respectively ( Figure 1a). All sampling sites were spaced 15-25 km apart (Figure 1a). At each site, 5 × 1 L water samples and environmental data, including temperature (°C), salinity (psu), dissolved oxygen (mg/L), and depth (m), were collected approximately 0.5 m below the surface of the water. While bull sharks are known to use the entire water column, previous studies have identified surface water sampling as the most suitable method for eDNA collection in headwaters (Katano et al., 2017). In Mobile Bay, water samples were collected in proximity to structures (e.g., oil rig and lighthouse), and at all riverine sites, samples were collected across the width of the river.
All eDNA field and laboratory protocols and controls followed Schweiss et al. (2020), and new gloves were used at each sampling site. Water samples were collected in the field using sterile, 1 L high-density polyethylene Nalgene ® bottles and stored on ice in clean marine coolers, or frozen, until water filtration (see Schweiss et al., 2020). In addition to the filtration, DNA extraction, and PCRnegative controls described in Schweiss et al. (2020), negative collection controls were also included. The negative collection controls consisted of autoclaved deionized water, which was taken onto the boat and placed in a clean marine cooler on ice with field samples to test for field contamination (e.g., Jerde et al., 2011). All negative control samples (collection, filtration, DNA extraction, and PCR) were processed and analyzed in replicates of five, according to the protocols of Schweiss et al. (2020), and were defined as negative if they did not meet any of the criteria for positive detections. Water samples were vacuum-filtered in a laboratory using 47-mm-diameter, 0.8-μm nylon filters (Cole Parmer ® ) and preserved in 95% ethanol at room temperature (see Schweiss et al., 2020). Freshwater discharge during the time of sampling was 510 m 3 /s, 1 characteristic of dry season flow conditions. At this site, the water was warm (29.6°C), normoxic (7.5 mg/L), and fresh (0.07 psu) (Table 1). Water samples collected at all other sites during the dry summer season and wet winter season did not meet any of the criteria for positive detections. None of the collection, filtration, DNA extraction, and PCR controls met any of the three analysis criteria for positive detections (see Schweiss et al., 2020); therefore, samples were considered free from contamination by target DNA.

| D ISCUSS I ON
The headwaters that pass through the MTD and feed the Mobile Bay estuary encompass the richest freshwater fauna in North America (Boschung & Mayden, 2004;Lydeard & Mayden, 1995), including many rare and endemic species. Monitoring the populations of such biodiverse fish fauna across such a vast expanse can be challenging.
Our findings provide evidence that bull sharks can occupy freshwater upstream habitat in the Alabama River and further demonstrate the ability of eDNA to identify rare species in Alabama rivers (e.g., Alabama Sturgeon Scaphirhynchus suttkusi, Pfleger et al., 2016). To place our findings into context, a synthesis of 150 years of survey data collected from 3,716 sampling locations throughout Alabama's many rivers and the Mobile Basin noted only two bull sharks, both of which were located in estuarine waters south of the MTD (Mettee et al., 1996). Thus, our findings represent the first scientific evidence of bull shark habitat use in this freshwater riverine system.
Although our data are limited to one survey in each of the dry summer and wet winter seasons, they provide preliminary information on potential spatial and temporal patterns of the occurrence of bull sharks in this region. The positive eDNA sample from site 16 suggests that at least one bull shark was present at that site or further upstream. The Alabama River contributes to one of the largest discharge volumes in the continental United States (Dzwonkowski et al., 2011); therefore, shed eDNA is expected to disperse downriver from the source relatively quickly (e.g., Jane et al., 2014;Wacker et al., 2019). While dispersing, eDNA molecules are subject to biological and physical degradation, which is accelerated in fresh, warm (e.g., >20°C) waters with high levels of microbial activity (Collins et al., 2018;Strickler et al., 2015). The positive bull shark eDNA detection in this study occurred during the dry summer season, when water temperatures were warm and discharge in the Alabama River was relatively low (i.e., less than ~500 m 3 /s, Webb & Marr, 2016).
Given these conditions, the persistence time of the detected bull shark DNA was likely short. Studies of DNA degradation under similar conditions found an eDNA half-life of ~3 hr, with a life span of ~6 hr for bony fish (Tsuji et al., 2017). When combined with a lack of  (Bethea et al., 2015). Similar size-based segregation has been widely demonstrated for bull sharks off the east (Curtis et al., 2013) and west (Simpfendorfer et al., 2005) coasts of Florida and in Texas estuaries (Matich et al., 2020), where YOY individuals preferentially occupy riverine habitats. In Florida, YOY bull sharks move upriver into shallow freshwater habitats during periods of low discharge to take advantage of pulsed resources (Matich & Heithaus, 2014) while residing in a low-mortality environment (Heupel & Simpfendorfer, 2011). Increasing urbanization of coastal regions and hydrological modifications to riverine ecosystems is accentuating the burden placed on species that use these habitats (Grant et al., 2019). This is particularly acute for YOY bull sharks in Alabama's riverine system, an ecosystem referred to as "North America's neglected hotspot" (Lydeard & Mayden, 1995). While preliminary, our findings add to a body of literature documenting the importance of freshwater habitats to bull sharks in both the eastern (Simpfendorfer et al., 2005) and western (Matich et al., 2020) Gulf of Mexico. By functioning as predatory mobile links across marine and freshwater habitats in coastal Alabama, bull sharks play a critical role in this ecosystem through their ability to influence prey abundance and behavior, maintain biodiversity, and buffer against invasive species (see Ferretti et al., 2010;Ritchie et al., 2012). In the north-central Gulf of Mexico, eDNA represents a powerful tool to identify how future changes in freshwater discharge and/or urbanization may impact habitat use by bull sharks, with important implications for the overall health of this system. Sciences of the National Institutes of Health. We thank Matthew Jargowsky, Katya Jagolta, Garrett Dunne, and Sarah Cunningham for assistance in the field, and Ariel Williams and Morghan Ellis for assisting with water filtration. Thanks to Amanda Jefferson for comments that strengthened an early draft of this manuscript.

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data associated with this publication (sample site coordinates, environmental data, number of copies of target DNA/ul for each sample and limit of detection plot) can be accessed on Dryad