Environmental DNA as a complementary tool for biodiversity monitoring: A multi-technique and multi-trophic approach to investigate cetacean distribution and feeding ecology

RESEARCH ARTICLE Environmental DNA as a complementary tool for biodiversity monitoring: A multi-technique and multi-trophic approach to investigate cetacean distribution and feeding ecology Luı´s AfonsoID1,2, Joana CostaID1, Ana Mafalda Correia1,3, Raul Valente1,3, Eva Lopes1,3, Maria Paola Tomasino1, A´ gatha Gil1,4,5, Cla´ udia Oliveira-Rodrigues1,3, Isabel Sousa Pino1,3, Alfredo Lo´ pez2,6,7, Paula Suarez-BreguaID8, Catarina Magalhães1,3* 1 CIIMAR–Interdisciplinary Centre of Marine and Environmental Research, University of Porto, Matosinhos, Portugal, 2 Department of Biology, UA–University of Aveiro, Campus Universita´rio de Santiago, Aveiro, Portugal, 3 Department of Biology, FCUP–Faculty of Sciences of the University of Porto, Porto, Portugal, 4 Department of Biology and Environment, CITAB–Centre for the Research and Technology of Agro- Environmental and Biological Sciences, University of Tra´ s-os-Montes and Alto Douro (UTAD), Vila Real, Portugal, 5 IIM-CSIC–Institute of Marine Research of the Spanish National Research Council, Vigo, Pontevedra, Spain, 6 CESAM–Centre for Environmental and Marine Studies, University of Aveiro, Aveiro, Portugal, 7 CEMMA–Coordinator for the Study of Marine Mammals, Nigra´n, Spain, 8 IEO-CSIC–Spanish Institute of Oceanography of the Spanish National Research Council, Vigo, Pontevedra, Spain * cmagalhaes@ciimar.up.pt Abstract The use of environmental DNA (eDNA) to assess the presence of biological communities has emerged as a promising monitoring tool in the marine conservation landscape. More-

over, advances in Next-Generation Sequencing techniques, such as DNA metabarcoding, enable multi-species detection in mixed samples, allowing the study of complex ecosystems such as oceanic ones. We aimed at using these molecular-based techniques to characterize cetacean communities, as well as potential prey on the northern coast of Mainland Portugal. During four seasonal campaigns (summer 2021 to winter 2022/2023), seawater samples were collected along with visual records of cetacean occurrence. The eDNA isolated from 64 environmental samples was sequenced in an Illumina platform, with universal primers targeting marine vertebrates. Five cetacean species were identified by molecular detection: common dolphin (Delphinus delphis), bottlenose dolphin (Tursiops truncatus), Risso’s dol- phin (Grampus griseus), harbor porpoise (Phocoena phocoena) and fin whale (Balaenop- tera physalus). Overall, except for the latter (not sighted during the campaigns), this cetacean community composition was similar to that obtained through visual monitoring, and the complementary results suggest their presence in the region all year round. In addi- tion, the positive molecular detections of Balaenoptera physalus are of special relevance since there are no records of this species reported on scientific bibliography in the area. The detection of multiple known prey of the identified dolphins indicates an overlap between predator and prey in the study area, which suggests that these animals may use this coastal

area for feeding purposes. While this methodological approach remains in a development PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 1 / 20 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Afonso L, Costa J, Correia AM, Valente R, Lopes E, Tomasino MP, et al. (2024) Environmental DNA as a complementary tool for biodiversity monitoring: A multi-technique and multi-trophic approach to investigate cetacean distribution and feeding ecology. PLoS ONE 19(10): e0300992. https://doi.org/10.1371/journal. pone.0300992 Editor: Petr Heneberg, Charles University, CZECHIA Received: March 8, 2024 Accepted: July 29, 2024 Published: October 16, 2024 Copyright: © 2024 Afonso et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: This work is a result of the project ATLANTIDA (ref. NORTE-01-0145-FEDER-000040) supported by the Norte Portugal Regional Operational Programme (NORTE2020), under the PORTUGAL 2020 Partnership Agreement and through the European Regional Development Fund (ERDF). The funders had no role in study design,

stage, the present work highlights the benefits of using eDNA to study marine communities, with specific applications for research on cetacean distribution and feeding ecology. Introduction Environmental DNA (eDNA) is an emerging tool for biodiversity monitoring that has been gaining prominence in scientific research during the second half of the 21st century, with increasing numbers of scientific outputs being published every year [1]. For marine ecosys- tems, the application of eDNA detection methodologies is still in its infancy, although it has steadily evolved in the last decades given the interest in its potential [2]. Among the multiple applications, these allow verifying environmental health by studying microbial communities, characterizing and quantifying stocks of marine vertebrates, such as teleost fish which repre- sents vital knowledge for good management of the fisheries sector, and determining the pres- ence and abundance of elusive species, such as cetaceans, allowing greater effectiveness of the monitoring work [3–5]. Additionally, technological advances in Next-Generation Sequencing (NGS) techniques, via DNA metabarcoding, allow for simultaneous multi-species detections in environmental samples, permitting the study of multiple trophic levels within the same sam- ples [6–8]. Cetaceans are widely dispersed mammals that inhabit most marine environments, from coastal habitats to neritic waters [9], playing a key ecological role in maintaining the balance of these ecosystems [10]. As keystone species, cetacean conservation is often a top priority in

international agreements, especially considering the anthropogenic threats they are currently facing [11]. Addressing the impacting issues in cetacean ecology is crucial, however, obtaining data that provides a detailed understanding of these animals is rather complex. Cetaceans are elusive, spending the vast majority of their time underwater. Also, their distribution range is often very extensive, including areas where access for sampling is difficult due to logistical and financial limitations inherent to the marine wildlife monitoring work or even legal constraints [10]. Therefore, the development of new non-invasive methodologies, such as the metabarcod- ing analysis of eDNA samples, especially in complement to visual monitoring is a promising advance in the optimization of monitoring effectiveness towards a better understanding of these species [10]. Regarding the use of eDNA techniques for cetacean monitoring, few published studies have been focusing on marine mammals (see review on eDNA application for cetacean monitoring under [12]). Nevertheless, there are already successful case studies where it has been possible to identify a variety of cetacean species through environmental samples using both species-spe- cific [5, 13–16] and universal primers [4, 17–23]. The possibility of detecting multiple species within the same environmental samples enabled through metabarcoding, allows for a multi- trophic analysis that widens the utility of the samples for the monitoring of several taxa and

application to various fields of research, including the study of trophic chains and species feed- ing ecology [23, 24]. In the present study, a molecular-based detection methodology was developed, using eDNA samples as a tool for biodiversity monitoring, especially applied to cetacean species in a coastal region of the North of Portugal, located in the Eastern North Atlantic (ENA). The ENA region is an area of great interest regarding the diversity and abundance of cetaceans, with tens of different species of dolphins and whales being recurrently reported over the years [25–36]. Here, we sought to obtain additional and concrete data on the occurrence of cetaceans in this PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 2 / 20 data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist.

area and infer the ecological reasoning behind it by recurring to a universal approach to per- form a multi-trophic analysis. Furthermore, we compared the eDNA results on cetacean spe- cies detection with the data obtained by traditional visual monitoring techniques, in order to assess the true potential of eDNA as a complementary tool across the panoply of methods employed for cetacean monitoring. Methods Study area Surveys to collect eDNA samples were performed on the north coast of Continental Portugal. This subregion, located in the northwest of the Iberia Peninsula, is of particular ecological interest due to the upwelling phenomenon strongly present along the coastline, thus enhanc- ing primary production and providing great conditions for the development of complex and rich trophic chains [37–39]. In addition, the area is part of a particularly dynamic coastal region with several estuaries of rivers that flow into it [40]. Topographically, the study area is entirely placed on the continental shelf, with the presence of the Porto Canyon in the vicinity offshore, a relevant structure as submarine canyons are known to be impactful in hydrody- namic processes of coastal habitats, such as enhancing upwelling and subsequently organic productivity [41] (Fig 1). To survey the area, four seasonal monitoring campaigns were carried out between the sum- mer of 2021 and the winter of 2022/2023. Dates for all carried campaigns are described in Table 1.

Fig 1. Study area with surveyed transect of at-sea monitoring campaigns, conducted between the summer of 2021 and the winter of 2022/2023, on the north coast of Continental Portugal. Isobaths with bathymetry in meters. https://doi.org/10.1371/journal.pone.0300992.g001 PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 3 / 20

Each campaign consisted of a survey transect with eight equidistant parallels, perpendicular to the coastline, spaced by approximately 6 nm, and covering distances of about 10 nm (approx- imately between 2 to 12 nm from the coastline), as shown in Fig 1. Permit for field access was given by the national maritime authority through the captaincy of the Leixões Port and of the Port of Douro. was For the collection of visual monitoring data, a previously established proto- col [33] was followed to sample the occurrence of cetacean species sighted along the established transect (Fig 1). The transect was designed to have an equitable range of observation capacity at all its points, guaranteeing correct and complete visual monitoring of the study area. The col- lected visual monitoring data was imported into ArcGIS Pro for spatial analysis. Water collection and filtration The environmental samples were sampled at pre-defined stations at the vertices of the cam- paign transect (Fig 1 and S1 Table), using a bucket and a rope to collect 5 liters of water. Before water sampling, the samples were poured into 5L containers. All the materials used for sample collection were previously cleaned with 10% bleach, rinsed with MilliQ water, and washed with local seawater (i.e., seawater at the sampling station) just before sample collection. After collection, volumes ranging from 1 to 3 liters were filtered immediately on board through Ster- ivex units (0.22 μm) using a peristaltic pump. After filtration, the samples were kept in liquid

nitrogen, and stored at -80˚C upon arrival at the laboratory. In total, 64 samples were collected, 4 at each station (two summers and two winters), with few exceptions: (1) two additional col- lections were made at S09 and S10 during the first summer campaign (2021); (2) during the first winter campaign, it was not possible to collect samples at stations S07 and S08, due to adverse weather conditions (S2 Table). DNA extraction Total eDNA was isolated using the DNeasy1 PowerWater1 Sterivex™ Kit (QIAGEN), follow- ing the manufacturer’s instructions, with some adjustments to increase the DNA yield, namely: increased vortex times (10 minutes at all vortex steps) and 5 minutes rest time before the last centrifugation step. After extraction, DNA concentration for all samples was quantified using Qubit™ dsDNA High Sensitivity (HS) assay kit (Invitrogen™). Environmental DNA extraction was performed in a specifically dedicated laboratory for the extraction of genetic material, with the bench being always cleaned with ethanol before its use. All materials used were also sterilized in UV light before the start of the extraction process. Library preparation, sequencing and bioinformatic analysis All samples were sequenced in high throughput sequencing in an Illumina MiSeq300 platform, using MarVer3(A) primers [8] and the Superfi II Polymerase (Invitrogen™) (following the Table 1. Seasonal monitoring campaign dates, conducted on the north coast of Mainland Portugal, between 2021 and 2023. Monitoring campaign Dates S01 to S08 S09 to S16

Summer 2021 25th June - 1st July 2021 * 24th - 25th September 2021 Winter 2021/2022 2nd - 3rd February 2022 25th - 29th March 2022 Summer 2022 9th - 10th August 2022 30th - 31st August 2022 Winter 2022/2023 20th - 21st March 2023 27th - 28th March 2023 *In Summer 2021, S09 and S10 transects were also conducted between the 25th of June and 1st of July 2021. https://doi.org/10.1371/journal.pone.0300992.t001 PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 4 / 20

protocol from [42, 43]). For marine vertebrate library preparation, a fragment of the vertebrate mitochondrial 16S rRNA gene was amplified and reamplified. In the first amplification step, PCRs were carried out in triplicate in a final volume of 10 microliters (μL), containing 2 μL of template DNA, 0,5 μM of the primers, 1X Platinum SuperFi II DNA Polymerase (Invitrogen), 0,8 mM dNTPs, 1X SuperFi II Buffer, 1X CES [44], and ultrapure water up to 10 μL. The PCR protocol was the following: an initial denaturation step at 98˚C for 30 seconds, followed by 35 cycles of 98˚C for 10 seconds, 49˚C for 10 seconds, 72˚C for 30 seconds, and a final extension step at 72˚C for 5 minutes. Triplicate PCR products were pooled together. The oligonucleotide indices that are required for multiplexing different libraries in the same sequencing pool were attached in a second amplification step with identical conditions but only 5 cycles and with an annealing temperature of 60˚C. Libraries were then purified using the Mag-Bind RXNPure Plus magnetic beads (Omega Bio-tek), following the instructions provided by the manufac- turer. Lastly, the pool was sequenced in a fraction (3/8) of a MiSeq PE300 flow cell (Illumina). DNA metabarcoding library preparation and sequencing were carried out by AllGenetics & Biology SL (www.allgenetics.eu). Following that, sequencing adapters were removed using Cutadapt v3.5 (Martin, 2011) and the originated reads went through a DADA2 [45] tool pipeline to remove other non-biological

DNA sequences (e.g., primers), filter the reads according to their quality, denoise, dereplicate, cluster the resulting sequences into Amplicon Sequence Variants (ASVs), merge correspond- ing forward and reverse reads, and remove chimeric sequences. The taxonomic assignment of each ASV (with a minimum of 5 reads) was performed by querying its representative sequence against a local instance of the NCBI’s Nucleotide database (last updated on 18/09/2023), using the algorithm BLASTn v2.13.0+ [46] with the following parameters: percent identity of 99%, e- value of 1e-05, and a minimum hit coverage of 80%. If, based on these parameters, a matched ASV represented multiple species, the one with 100% identity was considered. For this part of the work, the ASVs corresponding to all cetacean species and Actinopteri superclass species with known occurrence in Continental Portugal (as of the Ocean Biodiversity Information System [OBIS] up to 05/09/2023) were selected. For the construction of the databases, all avail- able nucleotide sequences (containing the 16S rRNA gene) from selected taxa were retrieved from the NCBI nucleotide database. Then, using the Geneious1 software (v.7.0.6), primer sequences (forward and reverse) were annotated (using the ‘Add Primers to the Sequence’ tool), sequences were cut (using the ‘Extract PCR product’ tool) so they only contain the frag- ments spanning (and including) the PCR primers, and all nucleotide sequences not containing the entire fragment of interest were excluded from the database. Positively matched ASVs of

each sample were then grouped into the respective sampling stations and seasonal campaigns for a clear visual representation of the results. Representative stacked bar plots of identified taxa relative abundance and Actinopteri ASVs heatmaps for winter and summer monitoring campaigns were produced using the ggplot2 (v.3.4.3) package [47], in R. Sampling statistical tests To better understand whether there are significant variations in the concentrations obtained from samples taken in the different seasonal campaigns, Kruskal-Wallis chi-squared tests were carried out, and if significant differences were found (p-value < 0,05), we followed with a pair- wise Wilcoxon test. The Wilcoxon test was also applied to compare the concentrations obtained from stations located at different distances from the coast (2 nm versus 12 nm). For this comparative analysis of DNA quantification results, samples that had DNA concentration below Qubit’s HS kit detection limit (0,005 ng/μL) were assigned a value of 0,001 ng/μL. A sample log of all collected and extracted samples can be found in S2 Table. PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 5 / 20

Statistical tests (Kruskall-Wallis test and pairwise Wilcoxon tests, with the level of signifi- cance set to 0,05) were carried out to understand the variation in the total number of ASVs obtained in the samples at the different seasonal campaigns and the different distances from the coast (2 nm versus 12 nm). Results Visual surveys A total of 71 sightings of 4 different cetacean species were recorded during monitoring cam- paigns (S1A Fig and S3 Table). The majority of these records were from common dolphin (Delphinus delphis) with 45 sightings across the entire surveyed area, representing approxi- mately 63.4% of all records. As for the other 3 species—bottlenose dolphin (Tursiops trunca- tus), Risso’s dolphin (Grampus griseus) and harbor porpoise (Phocoena phocoena)–these were sighted on 3 different occasions each. Besides the identified records to the species level, there were also 13 sightings of unidentified Delphinidae (18,3% of total records) and, on 4 occasions, it was only possible to identify cetacean occurrence at the superorder level (Cetacea). All sight- ings recorded with the respective date and geographical coordinates can be found in S3 Table. eDNA sampling Among the campaigns conducted, the summer of 2021 showed much higher DNA concentra- tions when compared to the other seasonal campaigns, with an average quantification of 9.46 ± 9.05 ng/μL (Median = 9,22 ng/μL). This was followed by the winter of 2021/22 with 5.10 ± 6.11 (Median = 3.02 ng/μL), the winter of 2022/23 with 2.26 ± 3.50 ng/μL (Median = 0.8

ng/μL) and, finally, the summer of 2022 with 1.08 ± 2.38 ng/μL (Median = 0.67 ng/μL). Statisti- cal tests showed significant differences between the quantified concentrations of the different campaigns (p-value = 9.63e-06), these being explained by differences between summer 2021 and summer 2022, summer 2021 and winter 2022/23, and also between summer 2022 and win- ter 2022/23 (S4A Table). At the seasonal level, the summer shows higher concentrations, with 5.52 ± 8.18 ng/μL (Median = 3.51 ng/μL) compared to the 3.59 ± 5.02 ng/μL (Median = 1.49 ng/μL) obtained for the winter. However, this difference was not statistically significant (p- value = 0.79). Taking distance into account, the sampling stations near the coast (2 nm) had an average DNA concentration of 5.48 ± 8.18 ng/μL (Median = 3.2 ng/μL), while the DNA obtained in stations at 12 nm was quantified as 3.75 ± 5.21 ng/μL (Median = 1.26 ng/μL) on average. These differences were not statistically significant (p-value = 0.45). eDNA monitoring: Sequencing output and taxonomic identification A total of 5 876 226 reads were generated by the Illumina platform with an approximate aver- age of 90 403.48 ± 42 311.20. After quality-filtering steps, the final output was 3 735 272 total reads with an average of approximately 57 465.72 ± 40 520.33 (representing 65.6% of the input). There were 5 outlier samples that produced less than 50 reads. A total of 173 156 repre- sentative ASVs (of previously determined marine vertebrate species: cetaceans and selected

value = 0.009). Pairwise tests highlight differences between the two summer campaigns and both winter campaigns with the summer of 2022 (S4B Table). With regard to distance from the shore, samples collected closer (2 nm) resulted in an average number of ASVs of 988.13 ± 2087.06 (Median = 0), while those further away (12 nm) resulted in 4,423.31 ± 17,126.03 ASVs (Median = 52.5). However, differences on obtained ASVs in rela- tion to the distance to the coast were not significant (p-value = 0.64). Of the 64 eDNA samples sequenced through metabarcoding, 33 (51.6%) had positive detec- tions of representative ASVs, from which 10 of the samples (15.6%) had positive detections of cetacean ASVs. By grouping the results by sampling station, we noticed that all sampling stations had positive detections for either Cetacea or Actinopteri. Relative abundances of all cetacean spe- cies identified, and Actinopteri taxa detected in 4 or more stations, are illustrated in Fig 2. Focusing on cetaceans, 5 different species were detected, with at least one of them identified in 7 out of the 16 stations (Fig 2 and S4 Table). The common dolphin (Delphinus delphis) was the most detected species, being identified in 6/16 stations, followed by the harbor porpoise (Phocoena phocoena) and the fin whale (Balaenoptera physalus), occurring in 2/16 sampling stations. The Risso’s dolphin (Grampus griseus) and the bottlenose dolphin (Tursiops trunca- tus) were detected in 1/16 stations. Regarding Actinopteri species, 19 species were identified, with at least one of them detected

in all sampling stations (Figs 2 and 3). Of all these detections, the most notable result was the Fig 2. Taxonomic assignment from environmental DNA samples collected in 16 sampling stations, in the north coast of Mainland Portugal, between 2021 and 2023. The plot represents the relative abundance (%) of sequences matched for each marine vertebrate species, considering pre-determined Cetacea, and Actinopteri species with positive detection in 4 or more stations (see methods). “Others” represents Actinopteri superclass species identified in 3 or fewer stations. https://doi.org/10.1371/journal.pone.0300992.g002 PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 7 / 20

widespread presence of the sardine (Sardina pilchardus)—in 14/16 sampling stations; the European seabass (Dicentrarchus labrax)—in 9/16 stations; and the European anchovy (Engraulis encrasicolus)—in 6/16 of the sampled sites (Figs 2 and 3). All the Actinopteri species identified, as well as the respective number of ASVs detected at each station, can be found in S4 Table. Additionally to the frequency of the detections across sampling stations, there is also a higher abundance of captured ASVs of the species S. pilchardus and D. labrax (Fig 3). In 5/16 sampling stations, 100% of the matched ASVs belonged to either S. pilchardus or D. labrax (Fig 2). The highest number of ASVs was obtained for S. pilchardus in Station S15 (off north of Viana do Castelo) (Fig 3). Station S01 (southwest of Porto, at ~2nm from the coast) had posi- tive detections for a higher number of selected Actinopteri species in both seasons (Fig 3). The Station S08 (near Po´voa de Varzim) only had one minor detection (9 BLAST hits for its respec- tive ASV, see S5 Table) for S. pilchardus across all monitoring campaigns. Cross-checking eDNA with visual monitoring data Overall, by cross-referencing the data obtained by the two sampling methodologies applied in this study, we can verify the occurrence of the following four species in both datasets (visual Fig 3. Number of Amplicon Sequence Variants (ASVs) detected for each species of the Actinopteri superclass molecularly identified in environmental

DNA samples, collected at the sampling stations surveyed in the summer (top) and winter (bottom) in the north coast of Mainland Portugal, between 2021 and 2023. The number of ASVs was logarithmised for scaling purposes. https://doi.org/10.1371/journal.pone.0300992.g003 PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 8 / 20

records, eDNA molecular detection): D. delphis, P. phocoena, T. truncatus and G. griseus. In addition to these, positive molecular detections of B. physalus were obtained through eDNA analysis. Regarding the spatial distribution of these records, it is clear that D. delphis was sighted not only in greater abundance but also with a dispersed distribution throughout the region studied, which is also in line with the data obtained by molecular detection. For the other identified species, their DNA abundance was lower and there were fewer visual records obtained as well. Overall, no pattern of spatial or temporal correlation between molecular detections and visual records is clear. An inter-seasonal comparison of the spatial distribution of the records obtained by the visual surveys with the eDNA detections reveals contrasting results (Fig 4). In Fig 4. Seasonality of cetacean records obtained by visual monitoring (A) and molecular analysis of environmental DNA (B), for summer (left) and winter (right) campaigns carried out in the North coast of Mainland Portugal, between 2021 and 2023. https://doi.org/10.1371/journal.pone.0300992.g004 PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 9 / 20

terms of composition of the cetacean community, in the summer campaigns, all the 4 species mentioned above were sighted; while in the winter campaigns, sightings of D. delphis accounted for all records, excepting one sighting of T. truncatus (considering sightings identi- fied at the species level) (Fig 4A). Through eDNA analysis though, all the species were molecu- larly detected in the winter campaigns (including also B. physalus), while in the summer season the only species identification through molecular detection of cetacean species was of D. delphis (Fig 4B). Disregarding seasonality and accounting for spatial distribution only, the location of molecular detections of each species at the sampling stations seems to align better with the visual records at the latitudinal level than in relation to distance to the coastline (i.e., longitudinally). In summary, it is worth highlighting that: i) in both datasets, D. delphis was found to be distributed across the entire area with higher frequency north of Po´voa de Varzim; ii) T. truncatus was distributed in the north of the study area, visually recorded at different dis- tances from the coast but only detected with molecular methods in the most coastal sampling station (S13); iii) P. phocoena was detected in both datasets south of Po´voa de Varzim, and mostly in stations near coast (with only one sighting further from the coast); iv) the molecular detections of B. physalus occurred in the stations closer to the coast, one in the north and another in the south of the study area (Fig 4B). Multi-trophic analysis

Across all campaigns, multiple Actinopteri detections coincided with the molecular presence of cetacean species (Table 2). D. delphis was detected by eDNA simultaneously to S. pilchardus Table 2. Coincident cetacea and Actinopteri species detections by environmental DNA at sampling stations surveyed across seasonal monitoring campaigns carried in the north coast of Mainland Portugal, between 2021 and 2023. Monitoring campaign Sampling station Cetacean Actinopteri Summer 2021 S11 Delphinus delphis Belone belone S13 Delphinus delphis Sardina pilchardus Engraulis encrasicolus Dicentrarchus labrax S15 Delphinus delphis Scomberesox saurus Winter 2021/22 S01 Phocoena phocoena Sardina pilchardus Dicentrarchus labrax Labrus bergylta Alosa alosa Pleuronectidae spp. Dicoglossa cuneata Hyperoplus lanceolatus Conger conger S04 Phocoena phocoena Sardina pilchardus Dicentrarchus labrax Alosa alosa Dicoglossa cuneata Scomber scombrus Sarpa salpa Winter 2022/23 S01 Delphinus delphis Labrus bergylta S02 Grampus griseus Sardina pilchardus S04 Delphinus delphis Sardina pilchardus Balaenoptera physalus https://doi.org/10.1371/journal.pone.0300992.t002 PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 10 / 20

(on two occasions), B. belone, E. encrasicolus, D. labrax, S. saurus and L. bergylta. P. phocoena detections were coincident with a higher richness of Actinopteri species, including S. pilchar- dus, D. labrax, S. scombrus, L. bergylta, Pleuronectidae spp., H. lanceolatus and C. conger (Table 2). G. griseus and B. physalus were also detected at the same time as S. pilchardus (Table 2), while T. truncatus was solely detected without associated potential Actinopteri prey. Discussion Environmental DNA sampling and uptake Conducting the dedicated sampling surveys was highly constrained by weather conditions. This is particularly evident in the summer 2021 campaign, in which sampling was carried out with an interval of approximately 3 months at each end of the season (June and September, Table 1). Although this time interval between samplings from the same summer campaign is suboptimal, it still acts as a valid representation of the season when compared to winter. This is supported by the uniform oceanographic dynamics of the study region during the summer months [39, 48, 49]. The sampling techniques here employed were chosen due to their on-field practicability (collection of seawater and filtration) and lower risk of contamination (no direct handling of filters during DNA extraction). To date, there are no generalized standard proto- cols for the detection and identification of cetacean species through eDNA, with a wide varia- tion of methods currently being implemented [12]. The lack of a standardized, reliable, and

repeatable protocol for sample collection (and subsequent processing and analysis) still consti- tutes a major obstacle to the use of this monitoring tool for marine biodiversity monitoring [50]. Upwelling phenomena on the Iberian Peninsula tend to occur on a higher scale in the sum- mer months, with uniform upwelling-favorable conditions along the north coast [39]. There- fore, it was expected DNA concentrations to be uniformly higher in the samples collected during summer campaigns, which was not the case. However, while the 2021 summer cam- paign presented the highest values of DNA concentration, which could be explained by the mentioned oceanographic phenomena, the exact opposite was observed in 2022. This signifi- cant difference, although not ruling out possible human error in the different sampling steps, may be due to some unknown oceanographic event happening during sampling days or to other environmental variables with known impact on DNA preservation, such as temperature [51, 52], pH [53, 54] and/or UV radiation exposure [55]. The lack of inter-seasonal significant differences in the DNA concentration obtained from eDNA samples, in contrast to the signifi- cant inter-campaign differences observed, may be evidence that the seasonal (i.e., inter-annual cycles such as seasonal upwelling) influence is less relevant than the impact of less cyclic and more ephemeral phenomena on the DNA preservation and retrieval. Regarding the number of target ASVs, significant differences found between both seasons and monitoring campaigns,

suggest a higher target DNA uptake in winter months. Nevertheless, this result should be inter- preted with caution, since these differences are mainly justified by the low numbers evidenced in the summer 2022 campaign (for which a very low DNA concentration was obtained). As far as the proximity to the coast, although with some variation, there were no significant differ- ences in terms of DNA concentrations or the number of ASVs obtained. Researchers have reported a faster DNA degradation in inshore areas associated with intense anthropogenic pressures [56]. Nevertheless, coastal areas are also likely to have higher biomass, in comparison with offshore areas, due to the coastal upwelling phenomena [39, 50] and the input of the rivers [57, 58]. The coastal waters of the north coast of Continental Portugal are heavily influenced by the input of various fluvial water masses such as the Douro, Ca´vado, and Minho rivers [58, 59], which could also be interfering with sampling results. The river plume of the Douro River PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 11 / 20

is known to vary substantially interannually, with increasing extension into the ocean in winter months [49]. Overall, we consider that further studies are required in the future to monitor the influence of environmental conditions on eDNA uptake, and therefore evaluate sampling suc- cess in dynamic marine coastal ecosystems. Bioinformatic analysis criteria To compile the BLASTn databases, we established a strict criterion, comprising the amplified fragment of interest and the sequences of the forward and reverse primers for the aforemen- tioned species of interest. However, this type of approach for sequence selection has its pros and cons. On the positive side, it allows for greater certainty that a given sequence belongs to a respective species since smaller fragments can result in greater similarity between species (con- servative approach). However, this method can lead to a loss of information by possibly removing sequences of interest. Moreover, and particularly for assembling the database of sequences corresponding to the Actinopteri species, the approach was only focused on the spe- cies already recorded along the Portuguese coast, substantially reducing the size of the data- base. Such a decision was made since, in this context, the purpose for the detection of Actinopteri species was tied to the ecological role they play in the diet of cetaceans. Given the parameters applied in the BLASTn algorithm used for the taxonomic assignment (identity higher than 99%), there were only two cases in which multiple species matched to a

given ASV: one within Cetacea taxa, and another within Actinopteri. A specific example of multi-species assignment to an ASV was a case where the ASV was assigned to D. delphis with 100% identity, but other taxa were assigned with a percentage identity of 99.14% (the genera Tursiops spp. and Stenella spp.). This was to be expected given the genetic similarity of these species pertaining to the oceanic dolphin family (Delphinidae), as described by McGowen et al. [60]. Especially between D. delphis and the T. truncatus, there are only two Single Nucleo- tide Polymorphisms (SNPs) on the amplicon obtained with the MarVer3 primers. Besides this case, there was a double identification with 100% percent identity for the same ASV—7 sequences for the Atlantic horse mackerel (Trachurus trachurus) and 1 for the megrim (Lepi- dorhombus whiffiagonis) in the NCBI database. In this case, the species T. trachurus was selected as the correct identification. Such a decision was made because the matched sequence of L. whiffiagonis referred to samples collected in the Mediterranean Sea, in which case the authors of the work demonstrated a great variation in the mitochondrial 16S rRNA region between the Mediterranean and the Atlantic Ocean populations [61]. Cetacean detections: eDNA versus visual monitoring As described in the results, the species detected through visual monitoring were also detected by eDNA, with the additional record of the species B. physalus, which was only recorded through molecular methods in the environmental samples. This is a species with non-existent

sightings in scientific literature in this subregion of Continental Portugal, even though it has been recorded in other parts of the country’s coastline [62]. Additionally, in the Galician coastal waters, adjacent to the study area, the species occurs rather frequently [30, 31, 63, 64]. Offshore Galicia, there are relevant topographical structures, seamounts, that may act as hot- spots for pelagic biodiversity [65], such as the Galicia Bank [66]. The obtained detections of B. physalus DNA may be evidence of an occasional occurrence of the species in waters off north- ern Portugal, as it has been previously recorded in SCANS-IV surveys [31]. This showcases a likely case of eDNA displacement from its origin place of origin. As shown by the results, the common dolphin represented the vast majority of the sightings in the study area. The wide distribution and frequent presence of D. delphis along the entire PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 12 / 20

northern Portuguese coast was corroborated by molecular detection on eDNA samples. P. pho- coena was detected via eDNA in two stations located closer to the coastline. That result is in line with the frequent occurrence of the species at the mouth of the Douro River (located in proximity to the sampling sites), as described by Gil et al. [67] with occurrence records obtained through visual monitoring. As for T. truncatus and G. griseus, both species were only sighted on very few occasions. By eDNA, each of these was detected at one station, emphasiz- ing the less frequent occurrence in the north of Continental Portugal, in comparison with D. delphis and P. phocoena. T. truncatus was detected via eDNA in northern latitudes, where the sightings of the species were also registered. As for G. griseus, the DNA was detected in the southern quadrant of the study area, where the species was also recorded through visual moni- toring, where they were also recorded by molecular detection. On the other hand, T. truncatus is a frequent species in the northwest of the Iberian Peninsula [68], therefore detection through visual and molecular methods was anticipated, although more records were to be expected. Overall, the combination of the results from visual monitoring and eDNA suggests the year- round presence of the dolphin and porpoise species on the north coast of Continental Portu- gal. Given the difficult weather conditions for visual monitoring in the ENA for a large part of the year, mostly in winter months, and consequently low monitoring effort in the region, com-

plementary results from eDNA prove to be very useful to increase baseline knowledge of these coastal populations. Besides multi-species detection on eDNA samples, the application of the methodology for single-species detection may be of particular importance. This is the case given the need to monitor resident populations in decline, as shown by Ma et al. [14] where eDNA tools were applied for conservation efforts. In the north coast of Continental Portugal, we believe that this approach is extremely pertinent for species of conservation interest, such as P. phocoena, whose population in the mainland Portuguese coastline is predicted to be extinct in 20 years (Critically Endangered, [69]). Despite the aforementioned promising results, some challenges remain to be addressed for the future consistent application of this novel monitoring tool. In previous eDNA metabarcod- ing studies, it was usually not possible to distinguish between species of the same genus, as observed in Valsecchi et al. [22] for the Tursiops and Stenella genera, and even within the Del- phinidae family [4, 20]. In the present work, this was also a challenge during the assignment of ASVs, even though a conservative selection criterion was established. Therefore, this limitation represents a significant obstacle to meet monitoring objectives. In addition, although the use of universal primers may provide us with relevant information at the multi-trophic level, there is a major drawback when the technique is used to detect a certain taxon (e.g., monitoring cetacean

occurrence). Besides the likelihood that primers bind to non-target DNA in greater abundance at the time of amplification, which can result in false negative detections [70, 71], it is impossible to pre-select the samples with the target DNA before the sequencing step as the detections in the post-PCR electrophoresis may allude to different taxonomic groups. Thus, in the context of the applicability of eDNA for more efficient and accessible (i.e., cost-effective) taxon-specific monitoring, it is advisable to use taxon-specific primers. In the specific case of cetaceans, where, in addition, relative abundance may often be rather low (in comparison to other marine taxa), the development of cetacean-specific primers is one of the next key steps to address in the con- tinued emergence of the application of eDNA tools for cetacean monitoring. Monitoring multiple trophic levels The introduction of universal primers in the study of marine communities allows the collec- tion of data from multiple trophic levels, with posterior inference of possible ecological PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 13 / 20

interactions and prey/predator relationships. Universal primers also provide a deeper under- standing of unusual distribution patterns, as was the case in Zhang et al. [24] where the authors used data obtained by eDNA metabarcoding to assess the food resources of Eden’s whale (Balaenoptera edeni edeni), and in Visser et al. [72] where the cephalopod community compo- sition was studied to better understand G. griseus and Cuvier’s beaked whales (Ziphius caviros- tris) foraging behaviors. In this study, it was possible to detect numerous species from the Actinopteri superclass. In addition to S. pilchardus, highly abundant and dispersed in the study region, and reported as main target for the diet of D. delphis on the Iberian coast [73, 74], other species detected have also been found in its stomach contents in the portuguese continental coast, such as the afore- mentioned E. encrasicolus, the sole (Soleidae spp.), the needlefish (B. belone), the scads (Tra- churus spp.) and the Atlantic mackerel (Scomber spp.) [74]. Another species with abundant and dispersed detections in this study, D. labrax, was also found in the stomach contents of D. delphis in the English Channel [75]. This latter coastal fish species, of high economic interest, is of particular relevance from a cetacean conservation perspective since its feeding preferences overlap with those of the dolphins, with reported implications for their bycatch by promoting interactions with the fisheries sector [76]. S. pilchardus is also extremely relevant to this topic

due to its commercial importance in Portugal, being a main target of purse seine fisheries in the country [77]. In Continental Portugal, interactions between various dolphin species with this fishing method are frequent and can result in the bycatch of these animals [78, 79]. In the case of P. phocoena, many of the species identified simultaneously, such as Scomber scombrus and the ballan wrasse (Labrus bergylta), and others occurring in other stations, such as T. tra- churus, are also known prey for these animals in the ENA [80, 81]. Although not detected in coincidental stations, a wide variety of fish taxa identified in the studied area were previously found in T. truncatus stomach contents in Iberia, namely S. pilchardus, the European conger (Conger conger), E. encrasicolus, Trachurus spp. and Soleidae spp. [82]. Therefore, data obtained from metabarcoding techniques, in addition to occasional observed foraging behavior during dolphin sightings (S3 Table), suggest that this study area may act as a relevantfeeding ground for coastal cetaceans. It also highlights the prey species available to their populations, which may have relevant implications for management and con- servation strategies. Environmental DNA is thus proving to be a promising tool for multi- trophic assessments, not only to study biodiversity occurrence, but also to infer about ecological processes and investigate trophic relationships. Conclusion In summary, in this work we provide a proof-of-concept of the potential of metabarcoding

methods applied to eDNA samples, for biodiversity assessments, with special relevance as a cetacean monitoring technique for the study of cetacean distribution and feeding ecology. However, there are still obstacles and difficulties to overcome. More studies are necessary to better understand this novel sampling method, as there are still knowledge gaps in the applica- tion of the method, from environmental sampling to the analysis of the sequencing results. Positive detections of cetacean species in this work constitute important data, as it was pos- sible to characterize the northern coast of Continental Portugal in terms of cetacean occur- rence, not only reproducing similar results but also complementing the data obtained through visual monitoring. Here, eDNA monitoring allowed us to conclude that dolphin and porpoise species that are less often sighted during visual surveys, even with known occurrence in the study area, have a probable all-year-round presence in this subregion of the ENA. Addition- ally, this study also shows the potential of this molecular-based technique to collect data when PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 14 / 20

the monitoring effort is hampered by the inherent conditions of the sampled region (e.g., weather conditions), especially for less frequent species (possibly without existing visual rec- ords). Furthermore, by expanding the analysis to lower trophic levels, the overlap detection of dolphins and multiple known prey allows us to infer about their use of these coastal area for feeding purposes. In conclusion, this work has demonstrated the potential of an innovative monitoring meth- odology for studying complex marine biological communities, such as cetaceans, even allow- ing for a multi-trophic approach, essential for conservation efforts. Therefore, although the effectiveness of using eDNA as a tool in cetacean monitoring programmes remains under development, this work represents a step forward towards that goal. Supporting information S1 Fig. A—Cetacean occurrence records and spatial distribution registered during visual sur- veys in the North coast of Mainland Portugal, between 2021 and 2023, with reference in-land points. Coloured points represent cetacean sighting locations. B—Spatial representation of sampling stations surveyed in the North coast of Mainland Portugal, between 2021 and 2023. The map represents the positive molecular detection for cetacean species on environmental DNA samples (Phocoena phocoena, Delphinus delphis, Tursiops truncatus, Grampus griseus, Balaenoptera physalus). (TIF) S1 Table. Pre-defined eDNA sampling stations on the vertices of the monitoring transects, with distance from coastline and gegraphical coordinates (DD). (XLSX)

S2 Table. Sample log of all eDNA samples collected in pre-defined stations (see methods), with respective sampling date, DNA concentration quantified using Qubit™ dsDNA High Sensitivity (HS) assay kit (Invitrogen™) and total number of ASVs generated by Illumina MiSeq300 platform after read quality-filtering. (XLSX) S3 Table. Recorded sighintgs during the visual surveys on the season monitoring cam- paigns between the summer of 2021 and the winter of 2022/2023 in the north coast of con- tinental Portugal, sighing date, identified species and respective geographical coordinates (DD). Observed foraging behaviour was also recorded whenever it was possible to clearly determine it. (XLSX) S4 Table. Pairwise Wilcoxon test results between the collected eDNA samples originated data in the various monitoring campaigns carried out between the summer of 2021 and the winter of 2022/2023 in the north coast of Continental Portugal. A—DNA concentra- tions by monitoring campaign. B—Total number of target taxa ASVs (cetaceans and selected Actinopteri species with known occurrence in Continental Portugal). P-values of less than 0.05 were regarded as statistically significant. (XLSX) S5 Table. Grouped number of total ASVs detected for all identified species in the eDNA samples collected between the summer of 2021 and the winter of 2022/2023 in the north coast of continental Portugal. (XLSX) PLOS ONE Environmental DNA as a complementary tool for biodiversity monitoring PLOS ONE | https://doi.org/10.1371/journal.pone.0300992 October 16, 2024 15 / 20

Acknowledgments Ph.D. fellowships for authors RV (SFRH/BD/144786/2019) and AG (PD/BD/150603/2020) were granted by Fundac¸ão para a Ciência e Tecnologia (FCT, Portugal) under the auspices of Programa Operacional Regional Norte (PORN), supported by the European Social Fund (ESF) and Portuguese funds (MECTES). This work is a result of the project ATLANTIDA (ref. NORTE-01-0145-FEDER-000040) supported by the Norte Portugal Regional Operational Pro- gramme (NORTE2020), under the PORTUGAL 2020 Partnership Agreement and through the European Regional Development Fund (ERDF); and of the project EMPHATIC funded by Biodiversa+, the European Biodiversity Partnership, under the joint call 2022–2023 Biodiv- Mon for research proposals, co-funded by the European Commission and with the following funding organisations: Fundacio´n Biodiversidad (FB, Spain), Fundac¸ão para a Ciência e Tec- nologia (FCT, Portugal), Agence Nationale de la Recherche (ANR, France), and the Ministry of Universities and Research (MUR, Italy). Author Contributions Conceptualization: Luı´s Afonso, Joana Costa, Ana Mafalda Correia, Maria Paola Tomasino, Catarina Magalhães. Formal analysis: Luı´s Afonso, Joana Costa, Ana Mafalda Correia, Raul Valente. Funding acquisition: Isabel Sousa Pino, Catarina Magalhães. Investigation: Luı´s Afonso, Joana Costa, Ana Mafalda Correia. Methodology: Luı´s Afonso, Joana Costa, Ana Mafalda Correia, Raul Valente, Eva Lopes, Cla´u- dia Oliveira-Rodrigues, Paula Suarez-Bregua. Resources: Isabel Sousa Pino, Catarina Magalhães.

Supervision: Joana Costa, Ana Mafalda Correia, Alfredo Lo´pez, Catarina Magalhães. Visualization: Luı´s Afonso, A´ gatha Gil. Writing – original draft: Luı´s Afonso, Ana Mafalda Correia. Writing – review & editing: Luı´s Afonso, Joana Costa, Ana Mafalda Correia, Raul Valente, Eva Lopes, Maria Paola Tomasino, A´ gatha Gil, Cla´udia Oliveira-Rodrigues, Isabel Sousa Pino, Alfredo Lo´pez, Paula Suarez-Bregua, Catarina Magalhães. References 1. Takahashi M, Saccò M, Kestel JH, Nester G, Campbell MA, Van Der Heyde M, et al. Aquatic environ- mental DNA: A review of the macro-organismal biomonitoring revolution. Sci Total Environ. 2023; 873:162322. https://doi.org/10.1016/j.scitotenv.2023.162322 PMID: 36801404 2. Dı´az-Ferguson EE, Moyer GR. History, applications, methodological issues and perspectives for the use of environmental DNA (eDNA) in marine and freshwater environments. Revista de biologia tropical. 2014; 62(4):1273–1284. https://doi.org/10.15517/rbt.v62i4.13231 PMID: 25720166 3. Bohmann K, Evans A, Gilbert MTP, Carvalho GR, Creer S, Knapp M, et al. Environmental DNA for wild- life biology and biodiversity monitoring. Trends in Ecology and Evolution. 2014; 29(6):358–367. https:// doi.org/10.1016/j.tree.2014.04.003 PMID: 24821515 4. Andruszkiewicz EA, Starks HA, Chavez FP, Sassoubre LM, Block BA, Boehm AB. Biomonitoring of marine vertebrates in Monterey Bay using eDNA metabarcoding. PLoS One. 2017; 12(4). https://doi. org/10.1371/journal.pone.0176343 PMID: 28441466 5. Parsons KM, Everett M, Dahlheim M, Park L. Water, water everywhere: Environmental DNA can unlock