Lessons learned from applying eDNA surveying to diadromous fish detection across the Northeast Atlantic region

Abstract

Regular monitoring of diadromous fishes is critical to inform their management and conservation. Yet, the in situ data collection for these species is challenging due to their complex life cycle and low abundance. Focusing on the sea lamprey (Petromyzon marinus, Petromyzontidae) and the European shads (allis shad Alosa alosa and twaite shad Alosa fallax, Clupeidae), emblematic diadromous fishes in the Northeast Atlantic region, this study leverages the use of aquatic environmental DNA (eDNA) samples to monitor their distribution range. For that aim, we developed quantitative PCR (qPCR) and digital PCR (dPCR) assays and applied them to detect sea lamprey and European shad DNA in a network of 44 river basins across Spain, France, Ireland and the UK. We found that qPCR efficiently detected presence/absence of European shads, whereas the higher sensitivity of dPCR was essential for detecting sea lamprey, which is likely to occur at lower abundance. Moreover, sea lamprey showed significantly lower eDNA copies per litre of water compared to European shads, probably due to their larvae spending several years burrowed within soft sediments, reducing eDNA shedding into the water column. The integration of previously recorded data as reference datasets with this wide‐ranging snapshot study enhances our understanding of the distribution of sea lamprey and European shads in Northeast Atlantic rivers. Importantly, the lessons learned within this international collaboration are critical towards achieving a prevailing framework for conservation of migratory fishes, highlighting the need of well‐designed sampling strategies coupled with species‐specific assays applied to eDNA samples to boost long‐term monitoring efforts of diadromous species.

1. INTRODUCTION

Anthropogenic activities, together with climate change, can significantly modify natural habitats and ecosystems (Chu et al., 2005; Dodds et al., 2013). This is particularly more evident for diadromous species, whose complex life cycles, involving a variety of habitats between rivers and open ocean, make them more vulnerable to any environmental alterations (Chaparro‐Pedraza & de Roos, 2019; Limburg & Waldman, 2009; Tamario et al., 2019). Diadromous fishes are emblematic species with crucial ecological roles, providing numerous ecosystem services (Almeida et al., 2023; Ashley et al., 2023; Naiman et al., 2002). They can either be catadromous (migrating to sea to spawn) or anadromous (migrating to rivers to spawn). The sea lamprey (Petromyzon marinus) and European shads (allis shad Alosa alosa and twaite shad Alosa fallax) are anadromous species present along the eastern Atlantic coast with important ecological and economic roles (Almeida & Rochard, 2015; Wilson & Veneranta, 2019). Sea lamprey larvae act as ecosystem engineers through sediment bioturbation and nutrient processing in riverine habitats, whereas adults influence food webs as both predators and preys (Hogg et al., 2014). Economically, sae lamprey was historically exploited in Portugal, Spain and France, and remains culturally significant despite current conservation restrictions (Braga et al., 2019; Hanel et al., 2022). On the contrary, European shads transfer marine‐derived nutrients upstream during spawning migrations and serve as prey for piscivorous fish, birds and mammals (Doherty et al., 2004; Poulet et al., 2022). Their reasonal runs historically supported major commercial and artisanal fisheries along European rivers, particularly for allis shad, which was highly valued for its meat and roe (Sabatié et al., 2009). Thus, the decline of these species therefore compromises ecosystem services and threatens biodiversity as well as cultural heritage.

Their distributions and abundances have decreased markedly in the last century, especially since the 1980s, due to the synergistic effect of anthropogenic impacts, including climate change and habitat fragmentation. For sea lamprey, populations have declined in many regions, with current populations mainly restricted to southwestern European rivers (Almeida et al., 2021). By their side, shad populations have contracted their distributions, with even local extinctions in northern regions, with their core distribution now centred on southwestern Europe (Lassalle et al., 2008; Nachón et al., 2020). Each species faces conservation challenges reflected by their IUCN status: allis shad is categorized by the International Union for Conservation of Nature (IUCN) as ‘Critically Endangered’ globally, whereas twaite shad is categorized as ‘Least Concern’ globally and ‘Near Threatened’ at the European level (Ford, 2024a, 2024b, 2024c, 2024d). The sea lamprey, although decreasing across its global ranges, is listed as ‘Least Concern’ globally and at a European level (Ford, 2024e; Ford, 2024f), but at a regional level is classified as ‘Endangered’ in France and ‘Near Threatened’ in Ireland (King et al., 2011; Nunn et al., 2023; UICN Comité français, 2019). This high level of conservation concern at national levels and transboundary distribution of the species call for concerted inter‐basin conservation and management efforts (Guo et al., 2016; ICES, 2003; Kritzer et al., 2022; Ouellet et al., 2022).

In this context of general decline, accurate records of diadromous species presence are essential for pinpointing when and where they feed, breed and migrate. Continuous change, especially alterations in species home river ranges due to various pressures, emphasizes the importance of understanding their movements and habitats (Beaulaton et al., 2008; ICES, 2003). Yet, characterizing the distribution of fish spatially and temporally is a challenging task, especially when considering factors like the scale of distribution, the complex life history of diadromous species and resource constraints (Ciannelli et al., 2008). Conventional monitoring methods for diadromous fishes vary. For shads, netting, mark‐recapture, electrofishing, egg kick‐sampling and tagging have been routinely used, whereas for sea lampreys, both adults and larvae are targeted using single‐pass larval electrofishing, sediment excavation or adult trapping (Cantera et al., 2019; Piggott et al., 2021). However, these intrusive, expensive and labour‐intensive methods present several challenges and may be inefficient due to low‐abundant populations and accessibility to some habitats, such as deep or fast‐flowing waters, requiring substantial effort, funding and expertise (Lapointe et al., 2006; Pont et al., 2021). Therefore, using a cost‐effective, scalable and non‐invasive sampling method that increases the probability of detection is key for monitoring rare aquatic species with such a restricted temporal and wide spatial distribution as anadromous fishes (Westhoff et al., 2022). The analysis of environmental DNA (eDNA) is a promising and rapidly evolving approach for aquatic species distribution monitoring (Bhendarkar & Rodriguez‐Ezpeleta, 2024; Nagarajan et al., 2022; Pawlowski et al., 2021; Rodríguez‐Ezpeleta et al., 2021), with potential adaptability in a changing world (Thomsen et al., 2024). Traces of organisms in the form of gametes, faeces, skin cells, saliva, blood and other bodily substances contain DNA (Bohmann et al., 2014) that can be utilized to determine the presence of a species within a specific area (Abbott et al., 2021; Davison et al., 2019). This methodology has been instrumental in examining spatio‐temporal patterns around spawning habitats in rivers for sea lamprey (Bracken et al., 2019; Moser et al., 2021) and European shads (Antognazza et al., 2019; Antognazza et al., 2021), outperforming traditional methods in detecting these species. Moreover, eDNA‐based monitoring has made it possible to identify prospective nursery grounds and distribution of sea lamprey (Baltazar‐Soares et al., 2022). Conversely, the eDNA‐based monitoring is also used to assess surveillance and control measures for invasive populations of sea lamprey in the Great Lakes of North America where the species is considered as a pest (Gingera et al., 2016; Schloesser, 2018; Tkachuk & Dunn, 2020). Therefore, the use of eDNA represents a non‐invasive complementary approach with advantages in adaptability, sensitivity and cost‐effectiveness for shad and lamprey monitoring.

In this study, we conducted a broad‐scale survey using eDNA to monitor the geographical distribution of sea lamprey and European shads in their known range of the Northeast Atlantic region. By using eDNA‐based monitoring, the goal was to provide a non‐invasive, efficient and scalable approach to monitor these species across multiple river basins in Spain, France, Ireland and the UK. We aimed to address four objectives: (i) to assess the regional distribution of sea lamprey and European shads; (ii) to evaluate the efficacy of eDNA analysis as a monitoring tool within this context; (iii) to compare the variability of eDNA detections between quantitative PCR (qPCR) and digital PCR (dPCR) methods; and (iv) to determine freshwater migration limits of sea lamprey and European shads within river‐estuary systems during their upstream migration. We met the first two objectives by comparing qPCR‐based eDNA detections with existing occurrence records from 44 river basins in Spain, France, Ireland and the UK; the third and fourth objectives were addressed using dPCR‐based eDNA detections along river–estuary gradients within Spain. As methodology continues to evolve for fish monitoring and assessment, the integration of eDNA‐based monitoring is likely to become increasingly essential in the assessment of diadromous species. This work improves our understanding of the practicality and efficiency of eDNA analysis at a river basin scale, providing a robust framework for future research and highlighting the importance of a multi‐approach to eDNA analysis interpretation.

2. MATERIALS AND METHODS

2.1. Sampling location and evidence‐based reference data set

Water samples were collected from a network of 44 river basins across Spain (20 rivers; 168 samples), France (2 rivers; 9 samples), the UK (15 rivers; 53 samples) and Ireland (7 rivers; 68 samples, including 3 at sea close to the river mouth) (Figure 1). In the Basque region (Spain), additional water samples were taken from four estuaries: Bidasoa, Oiartzun, Urola and Deba (15 samples) to investigate the abundance and distribution of eDNA along the stretch of river leading to the upstream sites through dPCR. The sampling occurred over different years (2019–2021) in different regions, resulting in a total of 313 samples (Table S1). The selection of sampling sites was based on the known historic distribution of sea lamprey and European shads and was also shaped by opportunistic access through local collaborators.

FIGURE 1.

Area of study. (a) Geographical location of river basins sampled in the Northeast Atlantic region. (b) Close‐up view of the sampling area with primary river channels represented by blue lines and their corresponding numbers. (c) Zoomed‐in map of sampled rivers in the Basque region, where collected samples were analysed using both qPCR and dPCR. (d) Circular bar plot summarizing the number of samples (each segment represents one sampling point) for each river included in this study, analysed by qPCR. Darker segments indicate sea samples taken near the river mouth (Slaney and Suir rivers), and lighter sections indicate sampling points located in tributaries of the main river basin (Oria river). See Tables S1 and S2 for additional details.

This study involved a transnational collaboration among nine institutions in four countries, each adhering to established protocols and facing specific logistical constraints. Consequently, variation in sampling volumes, filters and timing across countries reflects differences in institutional practices and field conditions. Further details on transboundary eDNA sampling and analytical methods can be found in the DIADES Project manual (https://diades.eu/wp‐content/uploads/2020/04/WP6_1_Manual_Final_Version‐compress%C3%A9.pdf).

To place eDNA‐based monitoring in the context of expected presence or absence, an evidence‐based reference dataset for sea lamprey and European shads for each eDNA sampling site was created by combining presence/absence data from electrofishing, traps and net operations, as well as presence of downstream barriers. Data sources included peer‐reviewed scientific literature, official agency reports and unpublished monitoring datasets, representing the most up‐to‐date information available and reaching up to 2021 (Table S2). For each species, each eDNA sampling site was categorized as ‘Expected’ where there was evidence of past presence and the site was physically accessible by the target species; ‘Not expected’ where there was evidence of absence due to no detection during surveys or the site was physically inaccessible due to an impassable downstream barrier; and ‘Unknown’ where the site was accessible but there was no evidence to indicate the presence or absence of the species because no survey data were available.

2.2. Water sample collection and DNA extraction

The volume of surface water sample filtered ranged from 1 to 30 L (Table S1). Spanish and French samples were filtered through 0.45 μm pore size Sterivex filters and VigiDNA cross‐flow filtration capsules, respectively. Irish samples were filtered through 1.50 μm pore size Whatman filters, whereas for the UK samples 0.22 μm pore size Sterivex filters were utilized. Samples collected in the Basque region, Ireland and the UK included at least one field negative control per year. All filters were kept frozen at −80°C until further processing. For DNA extraction, the Qiagen QIAamp DNA kit was used for Spanish and Irish samples, whereas the French samples were processed using an in‐house procedure (Pont et al., 2018), and UK samples were extracted using the DNeasy Power Water Sterivex kit (Qiagen) following the manufacturer's instructions. One extraction blank was included for the Basque samples per sample batch. Sample handling and pipetting were meticulously performed within a specialized hood to prevent contamination. Each filter was extracted and analysed individually, ensuring the integrity of the DNA extracts from each separate water samples. The extracted DNA was then transported in insulating containers with sufficient dry ice (~−80°C) to AZTI in Spain for further analysis. The concentration of the extracted DNA was quantified using UV spectrometry (Thermo Scientific NanoDrop ND‐1000) and Qubit fluorometer (Life Technologies), and its quality was assessed through agarose gel electrophoresis.

2.3. Species‐specific detection assay development

To detect both species of European shads, we designed genus‐specific primers and a probe for European shads targeting a 94 bp fragment of the mitochondrial cytochrome b (cytb) gene. All allis and twaite shad cytb available sequences were retrieved from GenBank (https://www.ncbi.nlm.nih.gov). These sequences were aligned and a suitable region in which to develop the specific assay was identified using the BioEdit software (Kirmani, 2015) (Figure S1). To ensure the specificity of the primer and probe sets, an in silico analysis was performed using BLAST (Altschul et al., 1990), and no potential amplification of non‐target DNA was detected (Table S4). In vitro specificity testing was not performed with tissue from non‐target species due to the absence of a reference collection of estuarine and riverine samples. However, in silico analyses of primer and probe sequences indicate no risk of cross‐amplification with non‐target taxa. For the detection of sea lamprey, we utilized the assay developed by Gustavson et al. (2015). The sequences of the primers and probes used in this study are detailed in Table 1.

TABLE 1.

Primer and probe sequences used for specific amplification and detection of European shads and sea lamprey.

Primer name	Nucleotide sequence 5′–3′	Target and reference
Alosa_For	ACATTTCAGTTTGATGAAACTTCGG	European shads
Alosa_Rev	GAAGTGTAGTGTATAGCCAGGAA	This study
Alosa_probe	AGGAATGTGTTTAGCGGCAC
Pma_For	TTGGAGGCTTTGGCAACTG	Sea lamprey
Pma_Rev	TGTTTATACGAGGGAAGGCCATA	Gustavson et al. (2015)
Pma_Probe	CTAATACTTGGTGCTCCTG

2.4. Quantitative PCR analysis

Quantitative PCR (qPCR) analyses were performed by combining 5 μL of TaqMan Fast Advanced Master Mix (Applied Biosystems), 0.2 μL of each primer (forward and reverse) and probe (10 μM), 2 μL extracted DNA (10 ng) and 2.4 μL Milli‐Q water to adjust the total reaction volume to 10 μL. The procedure was carried out independently to detect the European shads and sea lamprey under the following conditions: for European shads, initial 50°C for 2 min and 95°C for 10 min, followed by 50 cycles alternating between 95°C for 3 s and 60°C for 20 s; and for sea lamprey, 50 cycles alternating between 95°C for 3 s and 55°C for 20 s. A total of five PCR replicates were analysed for each sample, along with positive and negative amplification controls. Genomic DNA (gDNA) extracted from tissue samples of specimens collected in 2019 from the river basins in the Basque Country by EKOLUR Environmental Consulting (Spain) was used as a positive control, with a 10‐fold standard dilution series included on each qPCR plate. A sample was considered positive for the species if at least one qPCR replicate yielded a cycle threshold (Ct‐value, indicating the number of cycles in qPCR needed to detect the DNA signal) value less than 40 (Takahara et al., 2020; Westhoff et al., 2022). To validate that the PCR was not inhibited, an additional analysis adding a known amount of DNA (0.1 ng) of a targeted species (Internal Positive Control‐IPC) to each sample was performed. The composition of the PCR mixture as well as the thermocycling conditions were the same as mentioned earlier, but the reaction included 2 μL of IPC, with an adjustment to the volume of water. A sample was identified as inhibited if the Ct value was delayed or failed to amplify in this PCR. All field, extraction and amplification controls included in this study tested negative.

2.5. Digital PCR analysis

Digital PCR (dPCR), utilizing the QIAcuity One, 5plex (Qiagen), was used to validate and compare the detection sensitivities between qPCR and dPCR on samples collected from Spanish rivers. Additionally, we analysed a new set of samples that were collected along transects extending from the estuaries of rivers Bidasoa, Oiartzun, Urola and Deba. The primer and probe sets utilized in the dPCR assays were consistent with those used in the qPCR analyses, with ROX‐labelled probes designated for European shads and Cy5‐labelled probes for sea lamprey detection. Each dPCR reaction contained 10 μL of QIAcuity Probe PCR Kit, 3.2 μL of probe mix, 2 μL of extracted DNA and 24.8 μL of Milli‐Q water, resulting in a total reaction volume of 40 μL. This 40 μL reaction volume was then transferred into microfluidic dPCR nanoplates (QIAcuity) capable of accommodating 24 samples, with each well divided into up to 26,000 partitions. Each of these partitions enables individual PCR reactions to occur within them. The nanoplate was then loaded onto the dPCR instrument (QIAcuity One, 5plex) and subjected to an automated workflow after quantifying the cycling protocol. The thermal cycle was executed under the following conditions: an initial step at 95°C for 2 min, followed by 40 cycles alternating between 95°C for 15 s and 60°C for 60 s. Each plate included positive (standard dilution series with tissue samples) and negative amplification controls (no template controls, NTC). For detection, the ROX fluorescence channel was engaged to monitor European shad amplification, whereas the Cy5 fluorescence channel was used for sea lamprey amplification. Fluorescent images from all PCR wells were acquired and analysed; partitions indicating the presence of the target molecule were discerned by their elevated fluorescence intensity.

The common threshold value of fluorescent intensity (RFU) was set manually, in accordance with Qiagen handbook recommendations (Qiagen, 2022), with both negative and positive controls in each reaction (Passera et al., 2023). This choice allowed us to clearly distinguish between positive and negative partition clouds. The negative control consistently showed negative results for DNA copies. The eDNA copy number per reaction volume was calculated by QIAcuity Software Suite using Poisson statistics on the ratio of positive and negative partitions (Qiagen, 2022).

2.6. Quantification of dPCR and qPCR analysis

To determine the concentration of sea lamprey and European shads eDNA in water samples, in copies per litre (copies/L), we started by assessing the concentration of the DNA template within the reaction volume, which was measured in copies per microlitre (copies/μL). Subsequently, we quantified the concentration of the DNA in the extraction eluate, which was based on a standard volume of 100 μL. For each individual water sample, we used the following formula to compute the overall eDNA concentration (C):

$$\mathrm{C}\left(\text{copies}/\mathrm{L}\right)=\frac{\left[\text{dPCR result}\left(\text{copies}/\mathrm{\mu L}\right)\times \mathrm{\Pi }\mathrm{DF}\_\mathrm{i}\times \text{Volume of Extract}\left(\mathrm{\mu L}\right)\right]}{\text{Volume of original sample}\left(\mathrm{L}\right)}$$

where Π DF_i is the dilution factor in sample, calculated as the volume of each dilution step divided by the corresponding elution volume:

$$\mathrm{\Pi }\mathrm{DF}\_\mathrm{i}=\frac{\text{Volume of first dilution}\left(\mathrm{\mu L}\right)\times \text{Volume of second dilution}\left(\mathrm{\mu L}\right)}{\text{Volume of first elution}\left(\mathrm{\mu L}\right)\times \text{Volume of second elution}\left(\mathrm{\mu L}\right)}$$

where the volume of first dilution refers to the volume of the aliquot created from the DNA extraction eluate, the volume of the first elution is the volume of the DNA extraction eluate used to create the aliquot, the volume of the second dilution is the volume used in the dPCR reaction (25 μL) and the volume of the second elution is the volume of the aliquot used to create the dPCR reaction mix (2 μL). For example, if a 100 μL DNA extraction is obtained, from which 20 μL are taken to create an aliquot of 50 μL (20 μL extraction +30 μL water), and later 2 μL are taken from the aliquot to create the 25 μL dPCR mix, the dilution factor would be:

$$\mathrm{\Pi }{\mathrm{DF}}_{\mathrm{i}}=\frac{50\left(\mathrm{\mu L}\right)\times 25\left(\mathrm{\mu L}\right)}{20\left(\mathrm{\mu L}\right)\times 2\left(\mathrm{\mu L}\right)}=31.25$$

We conducted a linear regression analysis, using R (R Core Team, 2021), to examine the relationship between the eDNA concentrations obtained from dPCR assays (expressed in copies/L) and the CT values from qPCR.

2.7. Confusion matrix analysis

A confusion matrix is a popular measure in machine learning that sums up the performance of a classification model into a table (Sokolova & Lapalme, 2009). It works for binary and multi‐class classification. We constructed two separate confusion matrices. The first matrix was based on the detected/not detected data obtained from the qPCR‐based eDNA assays and compared to the expected/not expected data derived from reference dataset (Table S2). This allowed us to assess the accuracy of eDNA method for detecting sea lamprey and European shads compared to their known expectations at sample‐level. This matrix serves to quantify how well our eDNA analysis matches the expected presence or absence of the species. Samples categorized as ‘Unknown’ were excluded from the analysis. The second confusion matrix compared the detection rates obtained from qPCR assays and dPCR assays, allowing a direct comparison of these two molecular techniques in their performance in detecting sea lamprey and European shads. The resulting confusion matrix was used to calculate different performance metrics such as accuracy (the proportion of correct predictions of both true positives and true negatives, out of all predictions), sensitivity (correctly identified true positives) and specificity (correctly identified true negatives). These metrics are calculated as:

$$\text{Accuracy}=\frac{\text{True Positive}+\text{True Negative}}{\text{Total Prediction}}$$

$$\text{Sensitivity}=\frac{\text{True Positive}}{\text{True Positive}+\text{False Negative}}$$

$$\text{Specificity}=\frac{\text{True Negative}}{\text{True Negative}+\text{False Positive}}$$

The confusion matrix analysis was performed using R (R Core Team, 2021).

3. RESULTS

3.1. Comparative performance of eDNA analysis

The comparison between the evidence‐based past records of the existence of the species and the current eDNA detection outcomes revealed that among the 58 sampling sites where the presence of sea lamprey was expected, eDNA detection occurred only at 17 sites. No sea lamprey eDNA was detected in the 37 unknown and 38 unexpected sites (Figure 2a; Table S3). For European shads, eDNA detection was successful at 14 out of the 35 sites where the species was previously anticipated. In addition, positive detections were recorded at 4 of the 31 sites that were previously ‘unknown’ and at 3 of the 67 sites where the species was not expected based on the reference dataset (Figure 2b; Table S3).

FIGURE 2.

Detections of (a) sea lamprey and (b) European shads in rivers using qPCR. Each dot represents one sampling point in the river basin, from downstream (left) to upstream (right). Straight lines indicate unpassable barriers and dashed lines, barriers with medium permeability. The left side of dots indicates if species presence (green) or absence (red) is expected or is unknown (orange) based on reference data (Table S1), whereas the right side of dots indicate detection (green) or non‐detection (red) of the species using eDNA. Additionally, the Sankey diagrams on the right illustrate overall flow between expected species presence/absence (left) and their corresponding eDNA detection outcomes (right).

The confusion matrices (Figure 3), excluding the ‘unknown’ category, offer a quantitative overview of the eDNA detection (qPCR method) against anticipated presence/absence derived from reference data sources. For sea lamprey, eDNA detection exhibited an overall accuracy of 57.3% and a sensitivity of 29.3%, indicating that eDNA analyses using qPCR successfully detected this species in only a small fraction of the sites where they were expected. The specificity of 100% shows that in all the sites where reference data predicted sea lamprey absence, the qPCR analyses concurred (Figure 3a). For European shads, eDNA detection using qPCR showcased an accuracy of 76.4% and a sensitivity of 40.0%, emphasizing its proficiency in confirming the presence of European shads in locations where it was anticipated according to reference data. Meanwhile, a specificity of 95.5% indicates that in places where reference data predicted the absence of European shads, eDNA detection often supported this absence (Figure 3b).

FIGURE 3.

Integrative analysis of expected species presence/absence and eDNA detection in water samples, in the form of a confusion matrix for (a) sea lamprey and (b) European shads, where the number of sites for each combination of reference data versus eDNA‐based result (excluding sites for which the species presence/absence was unknown) is indicated.

3.2. River‐specific detection rate

Sea lamprey presence was confirmed exclusively in the Bidasoa, Eo, Tambre, Ulla and Umia rivers in Spain, and the Welsh Dee and Tamar rivers in the UK (Figure 2a). Despite evidence from other survey methods of sea lamprey presence in French, Irish and other UK rivers, the species was not detected with eDNA. In all sites where their presence was unexpected, the qPCR assay provides support for their absence. Conversely, the presence of European shads was confirmed in the Garonne River in France, the Bidasoa River, downstream of the Oria, Deba and Kadagua rivers in Spain, the Slaney, Suir, Blackwater, Ilen and Inny rivers in Ireland, and the Tamar River in the UK (Figure 2b). Notably, upstream sites along the rivers Inny, Ilen in Ireland and Kadagua in the Basque country, where the presence of the species was not expected, resulted in positive detection. Yet, it should be noted that only one replication (out of five PCR replicates) at each of these rivers resulted in a positive detection with a high CT value (CT > 36). Additionally, positive detections were recorded along the rivers Garonne in France, and Urola and Oka in the Basque Country, where European shads’ presence had been classified as ‘unknown’. In the Basque country, specifically Bidasoa, Oiartzun, Urumea, Oria, Urola and Deba, where we conducted eDNA sampling for two consecutive years (2019 and 2020), results showed a consistent detection pattern for sea lamprey and European shads over the course of both years, confirming the utility and the reliability of our assays in monitoring these species.

3.3. Insight from dPCR and qPCR assessment

Although qPCR was used for all samples from Spain, UK, Ireland and France, a more sensitive dPCR assay to allow more detailed interpretation was applied to the Spanish samples, from which enough DNA was available for additional analyses. Comparisons between dPCR and qPCR results showed similar detection rates in most sampling sites, except in Basque rivers where sea lamprey was detected by dPCR but not by qPCR (Figure 4a; Table S3). However, negative qPCR results for sea lamprey in most samples posed a challenge for holistic comparisons. It is important to note that qPCR analysis involved five replicates for each sample, and the CT value represents the average of the positive values, whereas dPCR analyses were conducted with single replicate, but whose result is the compilation of thousands of partitions (i.e., 26,000 PCR‐independent reactions) within the reaction. The confusion matrix also reflects that dPCR was able to detect 23 sites where qPCR failed, demonstrating an accuracy of 80.4%, a sensitivity of 50.0% and a specificity of 83.8% (Figure 4b). These metrics indicate that dPCR has a higher detection capability than qPCR, particularly for sea lamprey. For European shads, the linear regression between mean qPCR CT values and log‐transformed copies/L determined by dPCR underscores a positive correlation of 0.88 (Figure 4a). This suggests that an increase in eDNA copies corresponds to an enhanced detection rate through qPCR CT values and replications. Most sampling sites displayed a consistent dPCR detection rate compared to qPCR, indicating uniformity in the findings across different locations. In all Galician rivers, European shads remained undetected with dPCR, consistent with the qPCR results. In four sites dPCR did not yield positive whereas qPCR did, although with high CT values. The confusion matrix for European shads showed that the accuracy of dPCR was 96.2%, the sensitivity 92.5% and specificity 98.1% demonstrating a high level of consistency with the qPCR detection rate (Figure 4b).

FIGURE 4.

Comparison of qPCR and dPCR detectability outcomes for sea lamprey (top) and European shads (bottom). (a) Correlation between CT (cycle threshold) values from qPCR and copies/L from dPCR. The bubble sizes are proportional to number of positive qPCR detections among replicates (n = 5), and colour represents the number of DNA copies per litre (in logarithmic scale). Each data point is based on five replications. ND in the x‐axis indicates no detection in qPCR. (b) Confusion matrices for the sensitivity of the qPCR and dPCR approaches.

3.4. eDNA abundance patterns along river stretches

We analysed the distribution patterns of sea lamprey and European shads eDNA at multiple points along 20 rivers, including four estuaries, using dPCR assays. Replicate samples from each sampling site were pooled to ensure a stable analysis. Statistical analysis revealed no significant differences between replicates for either sea lamprey (p = 0.9428) or European shads (p = 0.3887), confirming the consistency of our sampling methodology. Overall, eDNA concentrations of sea lamprey were significantly lower than those of European shads (Figure 5). In the Basque Country, sea lamprey eDNA was mainly detected in the Bidasoa, Oiartzun, Urumea, Oria, Barbadun and Kadagua rivers, whereas in Galicia positive detections were recorded in the Eo, Anllóns, Umia, Ulla and Tambre rivers (Figure 5a). On the contrary, European shads displayed a more restricted distribution, with detections confined to the downstream sites of the Bidasoa and Deba rivers in both years, whereas no detections were recorded in the estuary of any of the basins in 2019 (Figure 5b). A consistent pattern of eDNA detection was observed, with higher concentrations found at downstream sites and progressively lower concentrations upstream, regardless of the sampling period.

FIGURE 5.

Spatial abundance patterns of eDNA detections using dPCR across Spanish sampling sites, where the dot size represents the eDNA concentration (copies/L) for (a) sea lamprey and (b) European shads in each sampling point and red crosses indicate samples where no DNA was detected (ND). For Bidasoa river in 2020, the dots represent the average concentration (mean) of the five sampling dates within the year in which water was collected. Accordingly, the connected scatterplots show a breakdown of the results for the sampling dates within the year. Only results for those dates for which three sampling points were sampled and had at least one of them positive are shown.

4. DISCUSSION

In this study, we built on existing knowledge of sea lamprey and European shads’ distribution through a research initiative spanning 44 river systems across Europe. This effort was made possible through international collaboration among researchers, enabling a large‐scale application of the eDNA‐based monitoring to assess the distribution of diadromous fishes of significant conservation concern in the Northeast Atlantic region. It is important to note that the assay developed for shads is genus‐specific, and therefore species‐specific signals are indistinguishable. Although species‐level resolution could be valuable in the future given their sympatry, this study focuses on the genus level, as both species share similar biological traits and ecology and face comparable conservation challenges. Here we share key practical lessons from this ambitious endeavour, with the goal of guiding similar applications of eDNA for monitoring diadromous fish in similar and other contexts.

4.1. Detection discrepancies among species

Here, we concurrently monitored the presence of both sea lamprey and European shads using the same eDNA sample. Both species are anadromous and exhibit similar habitat preferences during their spawning migrations. During spring, adult European shads and sea lamprey migrate to rivers to spawn, after which most die (Silva et al., 2013). After hatching, juvenile European shads then migrate to the sea in late summer and autumn (Aprahamian et al., 2003; Lochet et al., 2009), whereas sea lamprey larvae (ammocoetes) remain buried in the riverbed for years before becoming parasitic juveniles (OSPAR Commission, 2009). This buried larval stage (Almeida et al., 2023; Limburg & Waldman, 2009), and the fact that we conducted surface water eDNA sampling, might result in low eDNA detection rates for sea lamprey. If ammocoetes are the sea life stage to be detected, collecting DNA from sediment is more appropriate (Baltazar‐Soares et al., 2022). Conversely, European shads, with a primarily pelagic habitat and high mobility during spawning, are more likely to release eDNA into the water column during migrations. According to these predictions, in our study, eDNA detection rates between the species differed: European shads was detected in 40% of the sites where it was expected whereas sea lamprey was detected in 29%. These results align with initial interpretations, highlighting the importance of incorporating species‐specific behaviour patterns and habitat preferences into eDNA sampling design and result analysis (Baltazar‐Soares et al., 2022; Clemens et al., 2022).

4.2. When and how to sample

The opportunistic nature of this international monitoring survey of diadromous fish led to distinct sampling approaches by each institution, which may have influenced detection rates. These variations included sampling times, locations, collected water volumes and filter pore sizes and types. Efforts were made to synchronize most of the eDNA sampling with the migration period of both species, assuming higher eDNA shedding rates during this time (Thalinger et al., 2019). However, this synchronization was not fully achieved for sea lamprey (Kelly & King, 2001; McIntyre et al., 2007; Silva et al., 2013; Silva et al., 2019; Taverny & Elie, 2009). In the Basque Country only, sampling was performed during upstream and downstream migration, whereas all other regions were sampled only during downstream migration. Both European shads and sea lamprey eDNA were detected during upstream migration at expected locations, but sea lamprey was largely undetected on downstream migration, likely due to mismatches in timing with peak downstream migration and lower eDNA shedding rates in surface water. High eDNA detection rates that matched with visually observed spawning events of sea lamprey have been reported, highlighting the importance of aligning sampling with activity peaks (Bracken et al., 2019). In contrast, the successful detection of sea lamprey was reported in Galicia as well as in the Tamar and Dee rivers, likely due to the high population density in these areas, which increased the likelihood of eDNA presence in the water samples.

Another aspect of debate is how methodological decision affects our ability to detect species through DNA. Regarding filter pore sizes, small pores retain bacteria and free DNA whereas large pores mainly capture bigger particles such as mitochondria (Power et al., 2023), meaning that bigger pore sizes increase the total DNA versus fish DNA ratio. Although all pore sizes used in this work capture fish DNA, no significant differences in fish DNA detection between 0.22 μm and 0.45 μm pores have been reported (Liu et al., 2024; Stica et al., 2025), whereas increased efficiency has been observed with pores above 0.8 μm (Allan et al., 2025). This suggests that Irish samples, which used the biggest pores, may have achieved a better target‐to‐total DNA ratio and potentially higher detection probability, but the detection rates did not appear to be significantly influenced by this because there is no clear increase of detection rates in those samples. Importantly, our results suggest that single‐species PCR detection is more dependent on DNA concentration than on filtration parameters, given that there is no competition for amplification with non‐target DNA (Eichmiller et al., 2016). Despite using the same pore size (0.45 μm), French samples, which used 30 L of water volume, resulted in lower eDNA detection rates at expected sites than Spanish samples, which used 1–2 L of water volume. Although increasing the filtered water volume can improve detectability, French rivers such as the Garonne and Dordogne are considerably larger and have higher discharge than Spanish rivers, which may lead to greater dilution and dispersion of eDNA in the water column and could explain the lower detection rates in these rivers. Altogether our findings indicate if the concentration of target eDNA in the sample exceeded the assay's detection threshold, it can be detected regardless of the sampled water volume filtered or pore size used. However, given the low abundance of diadromous species in certain rivers, particularly beyond the upstream migration period, and in estuaries, where their concentration is even lower, using the largest water volume possible and a pore size larger than 0.45 μm is recommended to maximize eDNA‐based detectability (Capo et al., 2020).

4.3. Methodological differences

The comparison between results of qPCR and dPCR revealed inter‐specific differences. For European shads, results of both approaches are similar with 96% of the samples resulting in concordant results. In this case, only 6 out of 158 samples exhibited inconsistent results. The four qPCR positives with negative detections based on dPCR could be considered false positives, as they were characterized by only one or two positive replicates and high cycle threshold (Ct) values. Conversely, the two dPCR positive results that were undetected in qPCR could be considered false qPCR negatives due to low DNA quantity, non‐detectable by qPCR. For sea lamprey, the concordance rate dropped to 80%, with 31 of 158 samples exhibiting differing results. Unexpectedly, eight samples were found positive by qPCR but negative by dPCR. Of these, six samples had only one or two positive qPCR replicates; however, two samples had three and four positive replicates with lower Ct values, indicating potential dPCR false negatives. dPCR utilizes extensive sample dilutions divided into thousands of independent partitions assessed through Poisson statistics (Qiagen, 2024). Significant dilution can produce partitions that do not contain the target DNA, especially at low initial concentrations (Whale et al., 2013), which might explain the observed false negatives. Additionally, 23 sea lamprey samples that were positive in dPCR but negative in qPCR highlight potential limitations in the sensitivity of the qPCR method for detecting low concentrations of DNA. Overall, dPCR demonstrates superior efficacy compared to qPCR for detecting species occurring at low abundance. The dPCR is also advantageous as it provides precise quantification of number of eDNA copies, which allows absolute abundance comparisons between samples. Yet, to ensure reliable dPCR results, sample concentrations must be maintained within the dynamic range to reduce partitioning errors, particularly for samples of unknown concentration (Qiagen, 2024). Systematic optimization of dilution factors improves the confidence of dPCR in identifying low target DNA concentrations, thereby reducing false negative rates. This optimization process involves adjusting the dilution levels to achieve the desired sensitivity and precision, which is particularly important when dealing with low‐abundance targets (Jiang et al., 2022). Although in general dPCR excels in sensitivity and accuracy, some applications might prefer using qPCR due to its lower cost, important particularly for large‐scale monitoring (Zhang et al., 2024). This trade‐off between cost and accuracy highlights the importance of selecting the appropriate method based on specific research objectives and available resources.

4.4. Can eDNA be used as reliable sources of presence/absence for diadromous fish?

The findings from this survey revealed inconsistencies between eDNA detection and expected occurrences of sea lamprey and European shads. Focusing on sites where the species were expected, we identified false negatives, which could be attributed to low abundance of the species, resulting in low eDNA concentrations in the dPCR or qPCR assays. Importantly, this limitation is not unique to environmental genetics, and low abundance also reduces the efficacy of traditional methods, reflecting a general challenge in monitoring these species. Because false negatives can misrepresent species presence and hinder conservation efforts (Desrochers et al., 2010), it is important to reduce them by adapting sampling times to coincide with peak migration, especially during spawning periods, filtering large volumes of water and applying more sensitive methods such as dPCR. In a few cases, detections were observed in sites where the species is not expected, based on historical records and known distribution. These detections were based on a single qPCR replicate out of five, with high Ct values (37–39), and were not confirmed by dPCR. Given this combination of factors, we interpret them as likely false positives, possibly resulting from stochastic amplification or low‐level contamination. As such, we treated these sites as true absences in our final interpretation. Focusing on the sites for which presence or absence in the species was unknown, all cases resulted negative for sea lamprey. However, for European shads, we revealed new sites where the species could be present. Interestingly, in river Deba (Spain), the initial assessment of European shads’ presence was ‘unknown’, but the positive eDNA detection caused the monitoring agency to return to the river where they found European shads (Ekolur, personal communication).

In this study, the availability of a comprehensive reference dataset has enhanced our understanding of the eDNA‐based results, facilitated interpretation and identification of potential causes of false negatives and positives. This contextual information was critical for assigning detection outcomes at each site. We therefore propose that eDNA serves as a valuable complementary tool for diadromous species monitoring, offering enhanced spatial and temporal coverage due to its logistical ease, non‐invasiveness and cost‐effectiveness.

AUTHOR CONTRIBUTIONS

NR‐E, CC and MB contributed to the conceptualization of the study. Methodology was developed by NR‐E, CC, MB, NFF and IM. Data acquisition and field work were carried out by DJN, PID, TB, COL, WKR, IA and M‐LA. Data curation was performed by CC and MB. Formal data analyses were conducted by CC, MB and NR‐E, with CC and MB responsible for data visualization. All authors contributed to interpretation of the results. NR‐E, CC and MB wrote the original draft of the manuscript. All authors contributed to review and editing of the manuscript. NR‐E supervised the work and was responsible for project administration. Equal contribution: CC and MB contributed equally to this work.

FUNDING INFORMATION

This work has been funded by the European Regional Development Fund (ERDF) through the INTERREG Atlantic Area Programme (DIADES project) (EAPA_18/2018), by the Department of Economic Development and Infrastructure of the Basque Government through project ‘GENGES‐Development and application of genetic methods to improve marine ecosystem management’, by the State Research Agency (AEI) of the Ministry of Economy, Industry and Competitiveness of Spain (EDAMAME project) (CTM2017‐89500‐R), by the Game & Wildlife Conservation Trust, Queen Mary University (QMUL) of London and Cefas’ Science Excellence fund. This work was performed under a doctoral grant awarded to Mukesh Bhendarkar by the Indian Council of Agricultural Research (ICAR), Government of India.

CONFLICT OF INTEREST STATEMENT

The authors declare that there are no conflicts of interest.

Supporting information

FIGURE S1. Sequence alignment of the primers and probe specific to shads, non‐target freshwater and diadromous fish occurring in the same rivers as European shads and shad sequences. The full alignment is provided as a separate file in extended supplementary information (10.5281/zenodo.17608649).

FIGURE S2. Amplification plots for serial dilutions of positive controls of (a) sea lamprey and (b) European shads. From left to right, dilutions correspond to 10, 1, 0.1, 0.01 and 0.001 ng DNA/μL. Each dilution was analysed in duplicate.

TABLE S1. Overview of eDNA sampling and processing of collected samples across the different regions indicating the year, period, number of samples and replicates, and other sampling features. Field control corresponds to a negative control, where ultra‐pure water is filtered alongside collected field samples to assess possible contamination leading to false positives. In all cases PCR controls were included.

TABLE S2. Sample sites classified per river basin and region. The number of samples collected per site and their location with respect to sea are indicated, with 1 being the closest point to the sea. The presence of an obstacle downstream and its passability is also indicated. Previous evidence of the presence of the species is indicated by showing the type of evidence.

TABLE S3. eDNA detection outcomes for sea lamprey and shad across the study area by qPCR and dPCR. Note that dPCR was performed only on samples collected in Spain.

TABLE S4. Summary of the in silico analysis of potential non‐target amplification by primers and probes designed for shads. Detailed information about BLAST results can be found in separate files in extended supplementary information (10.5281/zenodo.17608649).