Using eDNA sampling for species-specific fish detection in tropical oceanic samples: limitations and recommendations for future use

Introduction

Many commercially important fish species aggregate at specific times and places for the purpose of reproduction. The number of individuals present at the timing of aggregation formation is significantly higher (100s–1,000’s depending on the species) than those abundances found during the non-reproductive periods (Nemeth, 2009; Sadovy de Mitcheson & Colin, 2011). The occurrence of these events depends on the availability of suitable areas for feeding and breeding (Claydon, 2004), as well as other factors such as the lunar cycles, diurnal variations, and temperature and current alterations (Sadovy de Mitcheson, 1996; Samoilys, 1997; Nemeth, 2009). Species such as groupers and snappers are commonly known to travel great distances to these aggregation sites, which can be shared by more than one species and sometimes even overlap in timing and duration (Heyman & Kjerfve, 2008). The predictability of these natural events makes them an easy target for fisheries and overexploitation (Sadovy de Mitcheson, 2016; Sadovy & Domeier, 2005). The Caribbean is an unfortunate example highlighting the exploitation of fish spawning aggregations. Some of these species, such as the Nassau grouper (Epinephelus striatus), have been listed as critically endangered by the International Union for Conservation of Nature (IUCN)(Sadovy de Mitcheson, Aguilar-Perera & Sosa-Cordero, 2018). Nassau grouper spawning aggregations have declined by 60% in the Caribbean (Sadovy de Mitcheson et al., 2008), and “fishing down the food web” has led to the decline of other groupers, such as red hind (Sadovy de Mitcheson, 1999; Beets & Friedlander, 1992) along with commercially important species like mutton snapper (Lutjanus analis) and yellowfin grouper (Mycteroperca venenosa) (Luckhurst, 2003; Nemeth et al., 2006). Many exploited fish families such as Haemulidae (grunts), Lutjanidae (snappers), and Serranidae (groupers) have been associated with maintaining a healthy reef ecosystem (Grober-Dunsmore et al., 2007). Considering their ecological and economical importance, the protection and conservation of these aggregation-forming fish species are imperative for the overall resilience of marine food webs.

Fish spawning aggregations (FSAs) have been documented and monitored using multiple field- and personnel-intensive approaches. Some of these methods include fish counting via underwater visual census, which oftentimes is impaired by the sheer number of individuals present. Tens of thousands of fish in constant movement can be influenced by currents and the presence of divers, which can also influence how fish interact and behave during spawning (Kobara & Heyman, 2008). Other methods to first locate FSAs include hydroacoustic surveys such as echosounders (Binder et al., 2021) and underwater microphones (hydrophones) (Appeldoorn et al., 2013; Chérubin et al., 2020). Some of these methods can also be used to monitor FSAs (Schärer et al., 2012), among other methods such as digital imaging using baited or non-baited video cameras (Cappo, Harvey & Shortis, 2006), acoustic tagging (Tuohy et al., 2015), and remote sensing (Kobara & Heyman, 2007). However, all these methods can also be affected by visibility and ocean conditions, and can be costly or require skilled personnel that may not always be available year after year for aggregation monitoring. As a result, obtaining accurate and precise data is not only difficult but can also be costly (Pfrender et al., 2010), and these are just some of the challenges to monitoring known aggregations that are accessible to relocate annually.

Although these methods have been successfully used to monitor FSAs, there remains a need to find a method that detects spawning aggregations in a cost-effective, rapid manner. One promising approach is using environmental DNA (eDNA). Ikeda et al. (2019) proposed eDNA as an efficient and highly sensitive method to assess the distribution of aquatic organisms. The eDNA sampling approach has also been suggested to have the potential to revolutionize marine biomonitoring (Andruszkiewicz, Sassoubre & Boehm, 2017). Environmental DNA is a relatively young field in molecular ecology which uses DNA obtained from the environment (Bohmann et al., 2014) with the goal of identifying a specific organism or group of organisms present in a certain time and space without the need to isolate or capture the organism (Stewart, 2019; Wang et al., 2021). The eDNA sampling technique has also been used successfully to locate and identify a variety of freshwater and temperate latitude marine macro-organisms including those associated with fish spawning aggregations. For example, Ratcliffe et al. (2021), focusing on egg and larval stages, used eDNA sampling while simultaneously collecting ichthyoplankton to compare taxa detection for monitoring purposes. Tsuji & Shibata (2021) identified fish spawning events by detecting a spike in eDNA concentration in a laboratory setting. In field studies, both Tillotson et al. (2018), and Hayer et al. (2020) used eDNA sampling to detect spawning events in constrained freshwater locations. The use of eDNA sampling to detect FSAs in open marine environments has not been studied, but these studies suggest eDNA sampling has the potential to detect and assess FSAs of threatened and commercially important species.

The residency of eDNA in biological systems is influenced specifically by four factors: origin, state, transport, and fate of the DNA (Turner, Uy & Everhart, 2015). In aquatic environments, feces have been identified as one of the major sources of fish DNA, but these can easily become diluted in water, so samples must be collected as close to the source as possible (Caldwell, Payment & Villemur, 2011). Specifically for marine systems, the transport of DNA could refer to the movement through the water column, currents, or settling down in the benthos. Biotic factors such as the density of the organism or group of organisms present in the environment (Dejean et al., 2011), the organism’s size and life history stage, immunological state, reproductive status, metabolic rate, or stress to which the organism is being subjected (Sassoubre et al., 2016; Stewart, 2019) influence the condition of DNA in the system. Feeding, diet, and migration also influence DNA concentration, as does the shedding rate (Rourke et al., 2022). Abiotic factors such as salinity, pH, temperature, UV radiation, water currents and dilution processes in the water column (Sassoubre et al., 2016; Ficetola et al., 2008) can influence the residency and decay of DNA (Stewart, 2019). Additionally, water current speed and direction, as well as tidal fluctuations, can influence the direction and transport DNA downstream from the source, which could result in false-positive detections (Jeunen et al., 2019). For this reason, it is imperative to understand the dynamics of the study system in question and consider all the possible factors of the environment in which the samples will be collected in order to facilitate the greatest chance for successful DNA detection.

Choosing the right genetic marker can be difficult, specifically for eDNA studies which require smaller barcodes since DNA is mostly found fragmented or degraded when sampled directly from the environment. Thus, using or designing primers should be limited to roughly 200–250 bp to increase PCR success rate (Freeland, 2017; Bylemans et al., 2018; Zhang, Zhao & Yao, 2020). Mitochondrial genes are often used in eDNA metabarcoding (or, in general, DNA barcoding) studies because of their high efficiency for detection among degraded DNA and their availability in reference databases. These genes have been used to assess the biodiversity of fish communities (Shu, Ludwig & Peng, 2020), the phylogenetic diversity of bony fish and elasmobranchs (Marques et al., 2021), and fish population connectivity (Okumuş & Çiftci, 2003). Two of the most popular genetic markers for these studies include Cytochrome c Oxidase Subunit 1 (CO1) and 12S rRNA (12S). The amplified fraction of the CO1 gene is usually about 650 base pairs, depending on the primers, and is highly efficient as a genetic marker commonly used for universal DNA barcoding for metazoans (Rach et al., 2017). The CO1 gene has been previously used for the detection and identification of aquatic macro-organisms (Zemlak et al., 2009; James, Bolick & Suzumoto, 2010; Deiner et al., 2013). The 12S marker, a portion of the small mitochondrial rRNA gene, is commonly used to detect a wide variety of aquatic organisms, including fish (Polanco et al., 2021; Gold et al., 2021). Both markers have been previously used in eDNA biodiversity studies to estimate species richness (Shu, Ludwig & Peng, 2020), compare, and detect fish communities (Hallam et al., 2021) and for monitoring fish communities in spawning areas (Ratcliffe et al., 2021). The CO1 marker has also been used for detecting invasive marine species (LeBlanc et al., 2020), for species-specific studies (Wang et al., 2020) and for detecting spawning distributions of European anadromous shads (Antognazza et al., 2019) but there is no evidence to support CO1 or 12S as the most effective marker for species-specific detection in tropical marine environmental samples.

In this study, we apply the use of eDNA sampling to the tropical Caribbean waters of Puerto Rico. The focus of this study was to test the effectiveness of three different eDNA collection methods and targeted detection of specific fish species using two molecular markers (CO1 and 12S). In addition, a comparison of molecular techniques between traditional polymerase chain reaction (PCR) and quantitative PCR (qPCR) was performed with both genes. Our target organisms were four commercially important fishery species that form well-documented FSAs including red hind (Epinephelus guttatus), Nassau grouper (Epinephelus striatus), yellowfin grouper (Mycteroperca venenosa) and mutton snapper (Lutjanus analis). These species were chosen because of their commercial importance in Puerto Rico and based on long-term monitoring of their spawning aggregations, where abundance estimates are well known during the time of aggregation formation. The results highlight the challenges of working with eDNA in open ocean tropical latitudes and provide species-specific primer sets for future use in eDNA studies.

Materials & Methods

Sample collection sites

The eDNA samples were collected off the west coast of Puerto Rico for the four target species at their respective known spawning sites (Fig. 1) during the aggregation periods from January to May 2019 (Table 1). Sampling for each species aggregation was targeted during the species’ most active (highest fish count) aggregation month corresponding to the lunar cycle. For example, a species that forms an FSA during January to March, with January being the historically highest fish abundance, was only sampled during January. The chosen month was based on long-term monitoring of these FSA sites (Sadovy de Mitcheson, 1999; Ojeda-Serrano, Appeldoorn & Ruź-Valentín, 2007; Schärer, Nemeth & Appeldoorn, 2010; Kobara et al., 2013). In some cases, due to weather conditions, additional sampling may have occurred during the species’ other aggregation months and some sampling methods could not be conducted at the FSA sites due to logistical constraints of sampling at deep depths. All sample collection, equipment deployment, and monitoring were conducted under a local permit (DRNA scientific permit #2018-IC-075).

Table 1:

The number of eDNA samples collected per sampling method and per fish species.

The sampling month denotes that only one target fish species was sampled at each eDNA sampling event during that time frame, corresponding to the FSA peak. N Fish is maximum count of that species observed by divers on the sampling day.

Target species	Year	Month	Diver	VanDoorn	Sediment	Control	Total	N Fish
Red hind	2019	Late January	24	12	30	30	96	32
Nassau grouper		Early March	9	6	12	24	51	58
Yellowfin grouper		Late March–April	20	12	27	30	89	45
Mutton snapper		Late April–May	21	9	24	24	78	0

DOI: 10.7717/peerj.14810/table-1

Results

Discussion

Working in an open ocean tropical system proved to be a difficult task for single-species detection from eDNA sampling. A recent study determined that the concentration of eDNA and the overall DNA yield was higher in semi-closed environments such as estuaries compared to open environments such as offshore and clear water habitats (Kumar et al., 2022). Furthermore, the persistence of DNA can vary between the sampling methods, where water is found to have higher degradation rates compared to sediment (Sakata et al., 2021). Given these limitations, our study made a significant effort to focus sampling at the exact known times and places of FSA formation and using a variety of sampling methods to obtain the highest DNA yield as possible. In addition, specific primers were designed to amplify our target species, which in a priori testing proved to effectively detect only the target species of interest. Despite these controlled factors, both traditional PCR and qPCR proved to be ineffective at detecting target species DNA in our eDNA samples.

Sensitivity to detect a target individual or group of individuals greatly depends on the dispersion of target DNA in the sampling site (Furlan et al., 2016). In our project, the use of the AWAC provided a retrospective look at what occurred during our exact times of sampling. For example, at the yellowfin aggregation, the current velocity was roughly two knots from the surface to the depth of sampling. Thus, water currents could have quickly dispersed and reduced our target DNA concentrations, despite sampling during high abundances (relative) of target fish and at the known peak spawning times where the fish are found to aggregate close together. Additionally, factors such as UV radiation disturb DNA base pair bonds, influencing and reducing eDNA persistence throughout the water column (Pilliod et al., 2014) and warm temperatures, such as those in the tropics, are known to exacerbate eDNA degradation rates (Dejean et al., 2011). Higher water temperatures promote fish mobility and metabolism (Petty et al., 2012; Xu, Letcher & Nislow, 2010) increasing the rate at which an organism sheds DNA. Higher temperatures then could be associated with higher DNA shedding rates but simultaneously reduce DNA concentrations altogether. This could account for the variation in our eDNA sample concentrations, where general DNA concentration varied widely from 0.2 ng/µL to 88.2 ng/µL between samples and sampling methods.

Lack of amplification and/or negative results could also be false negatives. False negatives are known as undetected taxa that are actually present (Ficetola et al., 2015) and can start to occur within 48 h of sampling since 90% of eDNA concentration degrades in the first two days (Forsström & Vasemägi, 2016; Sassoubre et al., 2016). Most of the PCR products were deemed immediately unsuccessful (no amplification, smear bands) or were found to be erroneous after validation through sequencing indicated non-specific amplification despite valid visual representation on the gel (e.g., the size of the amplified band was of the expected size). False negatives occur when low amounts of target DNA are actually present in the eDNA water sample, but the results fall below needed detection levels (Rees et al., 2014). This phenomenon can occur despite the target organism or group of organisms being present during sampling (Ficetola et al., 2008). However, DNA belonging to non-target species can exacerbate the chance of false negatives (Rees et al., 2014) which is an uncontrollable factor in this study’s open ocean system. In our study, the target species was observed at the time of sampling for three out of our four aggregations, and samples were often collected directly under or next to the target species itself. So, it can be implied that DNA shed from these three targeted species was collected. Even so, if the quantity of DNA collected falls below the detection level, no detection would be possible.

Takahara et al. (2012) found that the abundance of a target fish species could be measured by the quantity of eDNA, and conversely Kelly et al. (2014) found that the eDNA sequences are directly related to the relative abundance of a target population. In our study, low individual count—albeit high compared to non-spawning times—could have resulted in low concentrations of target species DNA and overall scarce eDNA yield. These results were confirmed after performing a traditional PCR with Nassau grouper eDNA samples in which these samples were spiked with DNA from Nassau grouper tissue extraction. Since the species-specific primers were able to successfully amplify the target DNA from the eDNA samples, we can conclude that our primer set was valid, and our sampling methodology was not hindering the ability to detect the target fish.

The maximum fish abundance observed on our peak sampling days was 32 red hind, 58 Nassau grouper, 45 yellowfin and no mutton snapper were observed at their FSA. These numbers are low compared to previously documented counts, such as for red hind which contained 282 individuals (Shapiro, Sadovy & McGehee, 1993; abundance calculated by converting their density to our survey area). Other spawning aggregations are known to range from 100s to 1,000s (Nemeth, 2005; Sadovy de Mitcheson et al., 2008; Sadovy de Mitcheson & Colin, 2011), such as the Cayman Islands for Nassau grouper (Whaylen et al., 2004). In other words, compared with past aggregations or those in other neighboring islands, the number of individuals documented during this study was exponentially lower. Similar studies focused on fish spawning aggregations also use zooplankton nets to collect ichthyoplankton as well as eDNA water samples to compare and correlate detection rates to oocytes and spermatozoa present (Bylemans et al., 2017; Hayer et al., 2020). Spermatozoa are small and abundant in the water column during spawning events which makes them a major source of eDNA during spawning. Gamete release is directly related to peaks in eDNA concentration and overall DNA abundance and detection (Tsuji & Shibata, 2021), and spermatozoa specifically remain in the water column (Bylemans et al., 2017). However, in our case, the exact time of fish spawning is unknown mdash and has never been observed at any of these FSAs—but is estimated to occur at the low light hours of the crepuscular evening, when conditions at sea are unfavorable for diving and navigating these aggregation sites.

Conclusions

Environmental DNA has the potential to become a promising tool for the study of marine fish spawning aggregations in the Caribbean. However, open oceans present numerous uncontrolled variables that amplify the degradation rates of DNA. For single species detection, the target species DNA must be in detectable concentrations in these open systems which are largely based on the number of target individuals present. For future eDNA studies, it is imperative to take into consideration the importance of species abundance for eDNA yield and species detection.

Supplemental Information