Drop it all: extraction-free detection of targeted marine species through optimized direct droplet digital PCR

In-vitro optimization trials for free-floating environmental DNA detection in seawater

Individual Nalgene™ square polycarbonate bottles (Thermo Fisher Scientific, Waltham, MA, USA) were filled with 250 mL of one of the following: tap water, purified water (Milli-Q®; Millipore Sigma™, Burlington, MA, USA), locally collected seawater or artificial seawater with varying salinities (Red Sea Salt-Copepod salt; Red Sea, Germany). New Zealand marine NIS organisms used for testing included Sabella spallanzanii (Gmelin, 1791)—a large Mediterranean fanworm, Styela clava (Herdman, 1881)—a leathery club tunicate and Bugula neritina (Linnaeus, 1758)—a bush-like, calcified bryozoan, all of which were preserved in 99% ethanol and placed individually in separate Nalgene™ bottles. To simulate eDNA release, the organisms were vigorously shaken within the bottles for several minutes. Subsequently, water aliquots were collected from each bottle and directly added to species-specific ddPCR reactions.

The direct-ddPCR assays were optimized to minimize salt inhibition, and various additives and assay manipulations were tested. This included evaluating PCR additives such as glycerol, dimethyl sulfoxide (DMSO), Tween-20, and bovine serum albumin (BSA), as well as the assessment of sample volume, pH adjustment buffers and bases (1M potassium hydroxide (KOH), 1X Tris-acetate-EDTA (TAE), 1M tris aminomethane (Tris) pH 8.0, and 1M 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 7.2), PCR annealing temperature gradient (ranging from 54 °C to 62 °C), and the adjustment of primer and probe concentrations. Additionally, different ddPCR master mixes, including ddPCR™ Multiplex Supermix (No dUTP) (BioRad, Hercules, CA, USA) and ddPCR™ Supermix for Residual DNA Quantification (BioRad, Hercules, CA, USA), were compared. Detailed information regarding the testing and optimization procedures can be found in the Supplemental Material (File S2).

Experimental setup and organism collection

An aquarium-based experiment was conducted between January and February 2023 at a Physical Containment Level 2 (PC2) facility. All work was done in a controlled laboratory setting, and in accordance with biosecurity regulations, i.e., permission to handle organisms under sections 52 and 53 of the Biosecurity Act 1993 (Notice No. MPI 111).

Seven transparent polycarbonate 26 L tanks with lids and individual aeration pumps were enclosed within a temperature-controlled room (21.3 °C). Experimental tanks were maintained without water exchange to eliminate the effects of inflow on eDNA concentrations and water mixing between tanks. Each tank was filled with UV-treated seawater with a pH between 8–8.1, salinity = 35 ppt, and water temperature = 20 °C. Prior to the start of the experiment, the tanks and the system underwent a cleaning process according to facility procedures: scrubbing the tanks and recirculating the system with a 200-ppm bleach solution for two days. Subsequently, the system was rinsed with sodium thiosulfate to neutralize any remaining chlorine and thoroughly rinsed with distilled water.

The aquarium experiment was set up with three morphologically distinct marine invasive species: S. spallanzanii, S. clava and B. neritina. Sabella spallanzanii individuals were collected from a pontoon in Marsden Cove Marina (Marsden Bay, New Zealand: 35.84°S, 174.47°E; 16 January 2023) with local seawater and were shipped in containers on ice to the laboratory. Both S. clava and B. neritina used in the aquarium experiment were sourced from the side of pontoons from the local Nelson Marina (Nelson, New Zealand: 41.26°S, 173.28°E; 17 January 2023) and immediately transported to the laboratory.

The collected specimens were left to acclimate at the laboratory in tanks with seawater for 5 days without feeding. At the beginning of the experiment (designated as time point zero), the live specimens were distributed among three experimental tanks, with an additional fourth tank left empty as a negative control. The allocation of specimens to each tank was determined based on their size and the number of specimens (this factor is referred to as “biomass or biomass treatment” hereafter, Table 1). The total species weight per tank for each species was determined by weighing the organisms after their removal. Note that the assigned biomass treatments (high, medium, and low) were based on the relative weight within each species, rather than between species. To ensure that S. spallanzanii and B. neritina stood upright, organisms were attached to a sterilized stainless-steel bolt using plastic ties (Fig. S1). Styela clava had a string attached to its stem to hang vertically down, to mime their usual orientation in the environment (Fig. S1).

Sample collection

Sterile dual filter T.I.P.S® PCR 50–1,000 µL Tip (76 mm) (Eppendorf, Hamburg, Germany) on a compatible micropipette were used to collect water samples (1 mL) from each of the four tanks, 5–7 cm below the surface (Fig. 1). Samples were collected immediately after adding the organisms and then at 4, 8, 24, 48, 72, 96, and 192 (8 days) h. At each sampling occasion, six replicates were taken randomly, targeting different locations in the tank (i.e., back left, back right, front left, front right, and two samples from the middle). Samples were collected in microcentrifuge tubes and kept on ice until further processing (c. 1 h).

After removing the organisms, samples were collected immediately and then at 4, 8, 24, 48, and 72 h as described above. Removed organisms were weighed and photographed (Fig. S2).

Direct droplet digital polymerase chain reaction

Direct droplet digital PCR (direct-ddPCR) was conducted in an automated droplet generator (QX200™ Droplet Digital PCR System; BioRad, Hercules, CA, USA). Copy numbers (per µL) of the Cytochrome c oxidase subunit 1 (COI) gene were measured in all samples using primers and probes specific to S. spallanzanii (Wood et al., 2017) and S. clava (Gillum, 2014) and primers specific to B. neritina (Kim et al., 2018) (Table 2). To compare the results between the two ddPCR chemistries, both the hydrolysis probe (TaqMan, Carlsbad, CA, USA) and DNA binding dye (EvaGreen®, Taoyuan City, Taiwan) assays were utilized in this study. The S. spallanzanii and S. clava direct-ddPCR assays were performed in duplex, as the hydrolysis Sab3-qPCR-Probe was dual-labelled with a 5′ 6-carboxyfluorescein (6-FAM) fluorescent tag and a 3′ Black Hole Quencher. In contrast, the hydrolysis SC1-Probe was designed with a 5′ hexachlorofluorescein (HEX) fluorescent tag and a 3′ non-fluorescent quencher (NFQ). The duplex direct-ddPCR reaction included 10 μL of 2X ddPCR™ Supermix for Probes (No dUTP) (BioRad, Hercules, CA, USA), 1 μL of each primer and probe at 10 pmol, 1 μL of the collected water sample and 4 μL of sterile water for a total volume of 21 μL. For the B. neritina direct-ddPCR assay, each direct-ddPCR reaction included 10 μL of 2X QX200™ ddPCR™ EvaGreen Supermix (BioRad, Hercules, CA, USA), 0.5 μL of each primer at 10 pmol, 1 μL of the collected water sample and 9 μL of sterile water for a total volume of 21 μL. The BioRad QX200™ droplet generator partitioned each reaction mixture into nanodroplets by combining 20 µL of the reaction mixture with 70 µL of BioRad droplet oil, either for probes (Automated Droplet Generation Oil for Probes; BioRad, Hercules, CA, USA) or for EvaGreen (Automated Droplet Generation Oil for EvaGreen; BioRad, Hercules, CA, USA).

The duplex, S. spallanzanii and S. clava direct-ddPCR assay used the following cycle conditions: hold at 95 °C for 10 min, 40 cycles of 95 °C for 30 s, 57 °C 1 min, and a final enzyme deactivation step at 98 °C for 10 min. The B. neritina direct-ddPCR assay used the following cycle conditions: hold at 95 °C for 10 min, 40 cycles of 95 °C for 30 s, 57 °C 1 min, and a final signal stabilization and enzyme deactivation step at 4 °C for 5 min and 90 °C for 5 min. The plates were then analyzed on the QX200™ instrument, including at least one negative control (RNA/DNA-free water Life Technologies) and one positive control (genomic DNA extracted from S. clava and S. spallanzanii or B. neritina). Based on our experience and observation of ddPCR noise (e.g., proportions of fluorescing droplets in water blanks), the detection for all assays was set above the maximum value of the negative controls in the experiment, i.e., 0.08 copies/µL for the B. neritina and S. clava direct-ddPCR assays and 0.130 copies/µL for the S. spallanzanii direct-ddPCR assay.

Limit of detection and limit of quantification assay

To determine the limit of detection (LOD) and quantification (LOQ) for the direct-ddPCR of two assays, a serial dilution of the positive control (genomic DNA extracted from either S. clava and S. spallanzanii or B. neritina (20 ng/µL)) was performed and analyzed. Genomic DNA was initially diluted to 200 pg (1/100 dilution). Subsequently, a 10-fold series of 2X dilutions (e.g., from 1/100 to 1/102,400) was performed, resulting in a final concentration of 0.195 pg for both assays. Six replicates of each dilution and negative control were included in both series, and all dilutions were performed with fresh seawater from the aquarium experiment.

Data analysis

All statistical analyses and visualizations were conducted in R version 4.2.1 software (R Core Team, 2023). A Kruskal-Wallis test and Wilcoxon rank sum test were performed to determine whether there was a significant difference in free-eDNA signal for each species between tanks and between species. To analyze the detection of free-eDNA following the removal of organisms, an exponential decay model was fitted using the ‘easynls’ package in the R software (Kaps & Lamberson, 2009; R Core Team, 2023; Wood et al., 2020). Calculations for limit of detection (LOD), for each direct-ddPCR assay, were based on 95% confidence limit where the lowest level of detection was greater than the maximum value of negative controls (Baker et al., 2018). To infer the limit of quantification (LOQ) of the direct-ddPCR assays, the coefficient of variation (CV) was calculated for each standard, and the LOQ was defined as lowest standard concentration with a CV value below 35% (Klymus et al., 2020). Evaluation of the linearity of quantitative measurement (quantitative linearity) for the direct-ddPCR assays was assessed by the log10-transformed copy concentration measured by direct-ddPCR plotted against the log10-transformed input ng of DNA and fitting it with a linear regression (Zhao et al., 2016). To determine relationships between biomass (weight in g) and free-eDNA concentrations determined by direct-ddPCR, the “lme4” package in R was used to fit a generalized linear mixed model (GLMM). The mixed model was used to consider factors such as species, tank, and weight as predictors of free-eDNA concentration and assess their effects on the variability in free-eDNA concentrations.

Pilot testing and assay optimization for free-floating environmental DNA detection in seawater

Despite trying several methods, none of the additives, buffers, or pH adjustments used could completely eliminate the inhibition of saltwater (File S2). To mitigate the inhibition in observed gene copy numbers due to salt, sample dilution was found to be necessary to detect free-eDNA (Fig. S3). The best volume proportions for the direct-ddPCR assays were determined to be 20 µL of ddPCR master mix (Supermix, primers/probes and water) and 1 µL of the seawater sample (Fig. S3). Additionally, the primer concentration for the B. neritina assay was modified to 250 nM instead of 450 nM. An annealing temperature of 57 °C was established as optimal for all assays following the temperature gradient analysis.

Detection of free-floating environmental DNA from aquaria with organisms present

There was no amplification in samples collected from the negative control tank or the no-template direct-ddPCR controls throughout the entire experiment.

The amplitude of detection from free-eDNA varied and depended on the biomass treatment and/or targeted species. For S. clava, low to no detectable free-eDNA concentrations were observed at most time points when organisms were present, regardless of the tank (Fig. 2A). Conversely, B. neritina consistently exhibited free-eDNA detection throughout all time points, but only in the high biomass treatment (Fig. 2B). On the other hand, S. spallanzanii consistently showed free-eDNA detection at all time points and in all tanks (Fig. 2C). The overall highest free-eDNA copy numbers (13.3 copies/µL) were observed in the S. spallanzanii tank with the highest fanworm biomass (three organisms and a total weight of 32.7 g), at the 192 h sampling time point (Fig. 2 and Table S1).

The free-eDNA concentrations among the three species were significantly different (Kruskal-Wallis chi-squared = 58.152, df = 2, p-value < 0.0001 (p = 2.358e−13)). Further pairwise comparisons revealed significant differences between S. spallanzanii and S. clava (p = 0.0037), as well as between S. spallanzanii and B. neritina (p-value < 0.0001 (p = 9.3e−13)), but there was no significant difference between S. clava and B. neritina (p = 0.8207). It is important to note that S. clava free-eDNA had a much lower positive detection rate (6.25%) compared to B. neritina (24.3%) and S. spallanzanii (74.3%) (Fig. 2 and Table S1).

There were significant differences in S. spallanzanii free-eDNA copies/µL between high and low biomass treatments (p = 0.0015) as well as low and medium biomass treatments (p < 0.0001 (p = 5.4e−09)), but not between high and medium biomass treatments (p = 0.258) (Kruskal-Wallis chi-squared = 29.715, df = 2, p-value < 0.0001 (p = 3.528e−07)) (Fig. 2C and Table S2). Similarly, for B. neritina, significant differences were between high and low biomass treatments (p = 0.043) and medium and high biomass treatments (p = 0.042), but not between low and medium biomass treatments (p = 1.0) (Kruskal-Wallis chi-squared = 8.922, df = 2, p-value = 0.011) (Fig. 2B and Table S2). In contrast, no significant differences were observed among the three biomass treatments for S. clava (Kruskal-Wallis chi-squared = 2.5247, df = 2, p-value = 0.283) (Fig. 2A and Table S2).

Detection of free-floating environmental DNA after removal of organisms

After organisms were removed at the 192 h time point, an exponential decrease in free-eDNA concentrations was observed. For S. clava in the low biomass treatment, the free-eDNA signal was undetectable around 24 h after organism removal (Fig. 3A). However, for B. neritina and S. spallanzanii, some free-eDNA signal was still detectable for up to 72 h, in the medium and low biomass treatments, and the high and medium biomass treatments, respectively (Figs. 2B and 2C).

Notably, for S. clava, no detection of free-eDNA was observed in the high or medium biomass treatments after the removal of organisms (below 0.08 copies/µL after time point zero) (Fig. 2A). In the case of B. neritina, only one out of the six replicates in the high and low biomass treatments showed detectable levels of free-eDNA at 72 h (0.162 and 0.120 copies/µL, respectively). In contrast, all six replicates in the high biomass treatment for S. spallanzanii exhibited detectable levels of free-eDNA at 72 h (average = 0.257 copies/µL), and two replicates in the medium biomass treatment also showed detectable levels (average = 0.160 copies/µL).

For S. spallanzanii, the free-eDNA concentrations remained relatively stable within the first 8 h in the high biomass treatment, averaging 17.5 copies/µL (STDEV = 2.82). However, a decline was observed at the 24 h mark (average = 6.72 copies/µL). The exponential signal decrease continued after 24 h, and at the 72 h mark, the average free-eDNA concentration decreased to an average of 0.257 copies/µL (Figs. 2C and 3B).

The goodness of fit for the exponential models was evaluated using the R-squared values. Sabella spallanzanii free-eDNA showed the best fit with an R-squared value of 0.9735 (Fig. 3B), while S. clava (Fig. 3A) and B. neritina (Fig. 3C) had R-squared values of 0.8225 and 0.8545, respectively (Table S3).

Limits of detection and quantification

In general, the dilution series showed an exponential decrease in DNA concentrations for all three species and direct-ddPCR reaction exhibited good linearity (all R² > 0.737, p-value < 0.0001) (Fig. S4). However, for S. clava, the DNA concentrations exhibited stability across dilutions of 1/1,600 and 1/6,400, showing a consistent mean concentration of 0.232 copies/µL (STDEV = 0.020). The lower 95% confidence limit ranged from 0.075 to 0.128, which falls just outside the mean concentration of the blank (0.071 copies/µL). Based on these findings, we determined the LOD to be >0.234 copies/µL (dilution 1/1,600) (Table 3 and Fig. S5). The LOD for S. spallanzanii was considered to be >0.380 copies/µL (dilution 1/6,400), and for B. neritina LOD was calculated as >0.536 copies/µL (dilution 1/12,800) (Table 3 and Fig. S5).

We considered the LOQ for S. spallanzanii, to be 0.698 copies/µL (dilution 1/3,200) (Table 3 and Fig. S6). Despite the CV of 36.5% observed for the B. neritina standard curve at dilution 1/1,600, we determined the LOQ for B. neritina to be 0.739 copies/µL at dilution 1/3,200. Dilutions below 1/3,200 consistently exhibited CV values exceeding 35% (Table 3 and Fig. S6). The S. clava assay had the most variability and we considered the LOQ for S. clava, to be 24 copies/µL (dilution 1/2.5) (Table 3 and Fig. S6).

Feasibility of free-floating environmental DNA in marine environments

Despite the reduced susceptibility of ddPCR to inhibition, the high concentrations of salts and metal ions such as K⁺, Na⁺, and Mg² in saltwater samples can affect DNA stability and decrease the fidelity and enzymatic activity of DNA polymerase during the PCR reaction (Davalieva & Efremov, 2010; Kuffel, Gray & Daeid, 2021; Lorenz, 2012; Mubarak et al., 2020). Therefore, detecting free-eDNA directly from seawater samples is not a trivial task and requires in vitro optimization of the direct-ddPCR reaction, as was done in the present study.

Mitigating the inhibitory effects of salt was a critical step in achieving accurate detection and reducing the sample volume to 1 µL proved to be an effective approach (Fig. S3). Previous research has demonstrated that dilution can enhance reaction efficiency by reducing the concentration of inhibitors and has been used successfully as a method to detect bacteria in drinking water (Benson, Fode-Vaughan & Collins, 2004; Kokkoris et al., 2021). However, it is important to acknowledge that using a 1 µL volume may not be optimal in highly dynamic marine environments, especially when targeting rare species or new incursions that may require larger water volumes to increase likelihood of detection (Bowers et al., 2021; Diaz-Ferguson & Moyer, 2014). To adapt this technique for field studies and reduce inhibition of salt further optimization of direct-ddPCR reactions may be necessary, such as using low salt ddPCR Supermix, testing alternative DNA polymerase, diluting the sample prior to testing in the reaction, and evaluating alternative polymer-buffer systems (Hedman et al., 2010; Sidstedt et al., 2017) (File S2). One example is the use of the multiplex master mix and buffering the water solution prior to sample collection, allowed for additional sample input. We found that using the ddPCR™ Multiplex Supermix and diluting the sample with buffers yielded a positive detection of free-eDNA during our in-vitro optimization trials. This is likely attributed to the replacement of the reduced volume of master mix with nuclease-free water and overall dilution, enabling an increase in the input sample volume up to 3 µL (File S2). Although even with optimization, the volumes will be lower compared to traditional methods, the benefit of this technology lies in its ability to allow for a massive increase in the number of continuous samples, enabling higher spatial-temporal coverage.

Additionally, we acknowledge that the distribution of free-eDNA within an ecosystem can be unpredictable and patchy (Bowers et al., 2021; Eichmiller, Bajer & Sorensen, 2014; Harper et al., 2018; Itakura et al., 2020). In this study, samples were only collected from the top ~5 cm of the tank, corresponding to the length of a P1000 tip. Consequently, certain free-eDNA molecules may have remained undetected despite aeration-induced mixing. Hence, researchers should tailor the testing of direct-ddPCR assays in the field to their specific environment and target species, considering factors such as detection probabilities and accordingly adjust sample dilution, volume, sampling depth, and sampling design.

Effect of biomass and species on the detection of free-floating environmental DNA with droplet digital polymerase chain reaction

To explore the influence of species on the detectability of free-eDNA, this study focused on three morphologically and biologically distinct marine invertebrates: S. spallanzanii, S. clava and B. neritina. The results revealed significantly higher positive detection levels and free-eDNA concentrations for S. spallanzanii compared to both B. neritina and S. clava (Tables S1 and S2). These findings are consistent with the observations of Wood et al. (2020), who also reported higher eDNA concentrations of S. spallanzanii compared with S. clava in laboratory-controlled conditions. The variations in detectability were attributed to anatomical differences between organisms, with S. spallanzanii potentially shedding more eDNA due to its fragile feeding tentacles, while S. clava’ s tougher tunic may result in lower eDNA release (Wood et al., 2020). Bugula neritina, with its interconnected zooids enclosed within a calcified exoskeleton (Keough & Chernoff, 1987; Trindade-Silva et al., 2010), may exhibit lower release of free-eDNA compared to S. spallanzanii observed in this study. Given the observed detection differences among organisms in this study, we recommend that end-users tailor their eDNA detection approach based on the target species, acknowledging that this method may not be viable for all species. To overcome interspecies detectability variations, researchers may consider increasing sample size, focusing sampling near the organisms’ habitats, and accounting for the species’ seasonal and life-history traits (Burian et al., 2021; Mauvisseau et al., 2019b; Wood et al., 2020).

The study also aimed to assess the potential impact of species biomass on free-eDNA detection. Previous studies have reported positive relationships between biomass abundance and eDNA concentrations (Bradley et al., 2022; Doi et al., 2015; Everts et al., 2021; Lacoursière-Roussel et al., 2016; Rourke et al., 2022; Tillotson et al., 2018). Although some recent studies have implied that eDNA may not be useful to infer abundance of some species (Rourke et al., 2023). In our analysis, we found significantly positive linear relationships between weight and free-eDNA concentration for B. neritina and a moderate positive linear relationship for S. spallanzanii (Fig. S7). However, S. clava showed a non-significant weak positive correlation, indicating a less pronounced relationship between weight and free-eDNA concentration (Fig. S7). Further analysis using a generalized linear mixed model (Table S4) revealed that weight alone did not have a significant effect on concentration. This suggests that weight alone may not be a reliable predictor of free-eDNA concentrations and that multiple factors, such as tank (intra-species variation) and species, which contribute to the variability in free-DNA direct-ddPCR data, should be considered when interpreting the results.

Variations in eDNA shedding have been observed among individuals, even when exposed to the same environment and exhibiting similar behaviour, with some studies reporting up to a 100-fold variation from the same fish under controlled conditions (Rourke et al., 2022). Factors such as stress and viability can influence interspecific variation in eDNA shedding, as seen in our study where animals, specifically S. spallanzanii, may have experienced stress during transportation and adaptation to the new aquarium conditions. Stress can create an imbalance between eDNA accumulation and decay (Rourke et al., 2023). Furthermore, the death of some organisms during the study could also have implications for eDNA shedding, with some models suggesting that shedding rates may increase after death (Tillotson et al., 2018).

Considerations should be given to the sensitivity and precision of species-specific ddPCR assays, as well as methodological aspects such as filtration and DNA extraction, as they significantly impact detection and recovery rates of eDNA (Capo et al., 2021; Hinlo et al., 2017; Schweiss et al., 2019). Assessing the performance of direct-ddPCR assays involves evaluating the LOD and LOQ, which reflect the assay’s sensitivity and ability to accurately quantify low levels of target sequences (Klymus et al., 2020). In the present study, the LOD was similar across all three assays, but the LOQ for S. clava differed, suggesting the need for further optimization for routine point-of-need (PON) applications (Table 3). Evaluating sensitivity also requires considering the ddPCR chemistry employed, such as the hydrolysis probe (TaqMan, Carlsbad, CA, USA) and DNA binding dye (EvaGreen®, Taoyuan City, Taiwan). Both the EvaGreen assay (for B. neritina) and the duplex probe assay (for S. spallanzanii and S. clava) demonstrated comparable sensitivity in detecting free-eDNA, as indicated by the LOD values (Table 3). These findings align with previous research that reported similar sensitivity for these ddPCR chemistries (Falzone et al., 2020; McDermott et al., 2013).

Our study focused on assessing the feasibility of detecting free-eDNA in seawater rather than directly comparing it with traditional methods. However, when comparing the ddPCR results from our study with previous research by Wood et al. (2020), it becomes evident that traditional filtration and extraction methods may yield higher starting concentrations of S. spallanzanii and S. clava eDNA compared to the direct detection of free-eDNA. This disparity in methodology could potentially impact the detection rates and should be taken into account when designing a survey (e.g., by adjusting the replication levels).

Persistence of free-floating environmental DNA in seawater

Understanding the fate and persistence of free-eDNA is crucial for optimizing sampling strategies, improving detection accuracy, and interpreting findings (Farrell, Whitmore & Duffy, 2021; Harrison, Sunday & Rogers, 2019; Zaiko et al., 2018).

We observed that the free-eDNA signal for high biomass treatments of S. spallanzanii and B. neritina was detected up to 72 h, while that of S. clava—to only 24 h (Fig. 3). These findings contrast with a previous study by Wood et al. (2020), who reported that the eDNA signal of S. spallanzanii declined below the detection limits of ddPCR within 35 h, while S. clava could still be detected up to 87 h. Although this suggests that free-eDNA may degrade at a slower rate than initially anticipated, these discrepancies also highlight the complexities and variability in eDNA degradation processes, which may be influenced by abiotic conditions, system setup, and inherent variability within the organisms (e.g., stress, viability). It is also important to consider that free-eDNA may contain cellular contents released from damaged cells or lysed cell particles during high-temperature PCR cycles (Shehadul Islam, Aryasomayajula & Selvaganapathy, 2017). Therefore, further research is necessary to accurately classify and characterize free-eDNA using a combination of laboratory and in-field methods capable of capturing different size fractions of free-eDNA (Jo et al., 2019; Moushomi et al., 2019; Turner et al., 2014; Wilcox et al., 2015).

Future directions

This study has demonstrated the feasibility of detecting targeted marine species directly from a water sample using direct-ddPCR, marking the initial step towards automated in-field technologies. While further work is necessary, such as expanding to other target plankton or nekton organisms, in-field experiments and development and validation of mobile ddPCR technology, there have already been promising strides toward in-field automation. Mobile devices capable of detecting targeted eDNA in field settings are actively being developed (Hansen et al., 2020; Marx, 2015; Ponce, Arismendi & Thomas, 2021; Sepulveda et al., 2018; Thomas et al., 2020). However, the current approach involves collecting and manipulating samples with a field-based DNA extraction kit followed by the mobile PCR device, reducing the practicality of such tools and introducing potential biases (Marx, 2015; Nguyen et al., 2018; Sepulveda et al., 2018; Thomas et al., 2020). Our results suggest that coupling direct PCR with ddPCR could potentially lead to the development of automatic in-situ monitoring technology, mitigating some of these biases and enhancing sensitivity.

Along with targeted species analysis for eDNA, this study could help spark new ideas for in-situ assessment and screening of broader communities through the utilization of metabarcoding. However, for metabarcoding analysis, significantly larger volumes of water samples are needed for an accurate assessment of marine biodiversity (Kawakami et al., 2023; McClenaghan et al., 2020). In recent years, technologies such as robotic environmental sample processors (ESP) have been introduced with the ability to automate water sample collection and filtration, preserving the sample for further analysis (Hendricks et al., 2023; Preston et al., 2023; Sepulveda et al., 2020; Truelove et al., 2022), with some having the ability to filter large volumes (Govindarajan et al., 2022). Additionally, some second generations ESP can perform in-situ qPCR (Hansen et al., 2020). With the development of newer generations of ESP, it is easier to envision the collection and filtration of larger volumes and the ability to conduct direct PCR on filters (Stojan et al., 2023). Moreover, the success of mobile sequencing technologies such as Oxford Nanopore on marine eDNA samples (Egeter et al., 2022; Truelove et al., 2019), demonstrates the potential of coupling these technologies with ESP and direct PCR to create in-situ metabarcoding monitoring, although further ground-truthing of the technology is needed.