Comparing the efficiency of DNA extraction protocols across a multinational environmental DNA initiative

10 11 The comparability of methods applied to environmental DNA (eDNA) samples across 12 laboratories remains a significant challenge for international biodiversity monitoring 13 projects. Inconsistently performing practices can jeopardize the reliability of data that is 14 essential for effective conservation strategies across geographic regions and focal 15 species. To address potential discrepancies among four international partner 16 laboratories who are part of a collaborative eDNA initiative, a ring test was conducted to 17 compare extraction efficiencies based on 39 eDNA samples. Each laboratory 18 contributed eight to eleven eDNA samples collected from five locations throughout the 19 North-East Atlantic and Mediterranean Sea near marine megafauna (whales, dolphins, 20 and sharks). After lysis, aliquots were exchanged between laboratories then 21 independently extracted using each facility’s preferred method. Extracts were then 22 returned to their respective laboratories of origin for measurements of total DNA 23 concentration as well as quantitative PCRs using species-specific assays designed for 24 each associated target species. Our findings revealed similar concentrations of total 25 DNA, yet a significant deviation in extraction performance for targeted qPCR reactions 26 by one laboratory. Overall, detection success differed based on the target taxa with 27 sharks being less often detected (and at lower concentrations) than whales and 28 dolphins. Significant interaction effects were found between combinations of 29 laboratories and species, suggesting that particular extraction protocols may be most 30 efficient for specific environmental conditions. Our study serves as a foundational step 31 towards establishing rigorous, reproducible eDNA practices that are crucial for the 32 success of multinational environmental monitoring projects to enable the direct 33 comparison of results. 34 35 Key Words 36 37 assay, cetaceans, DNA extraction, eDNA, Limit of Detection, optimization, ring test38 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447

1 2 Advances in molecular technologies have revolutionized the collective perception and 3 capabilities of assessing biodiversity. Over the past fifteen years, environmental DNA 4 (eDNA) has burgeoned as a noteworthy tool for monitoring the diversity of a system 5 (Beng and Corlett, 2020; Rourke et al., 2022). Through the collection, extraction, and 6 analysis of trace amounts of genetic material shed by organisms into their environment, 7 researchers can now detect the presence of species in environmental samples such as 8 sediment, water, snow, or air (Ficetola et al., 2008; Lynggaard et al., 2022; Miya, 2022). 9 Notably, the inherent attributes of eDNA-based approaches make them particularly 10 suitable for the investigation of rare and/or protected species, due to their noninvasive 11 nature which removes the necessity for direct animal contact (Foote et al. 2012; Baker, 12 Scott 2015; (Juhel et al., 2021; Rojahn et al., 2021). Although eDNA methods are 13 increasingly used by ecologists, especially for such studies on elusive species, their 14 integration into large-scale routine management and decision-making processes 15 remains limited. A prominent obstacle is the need for rigorous international standards 16 and optimized protocols, which could make applications of eDNA monitoring more 17 reliable and comparable across initiatives. 18 19 Considering the highly sensitive nature of methodological choices, efforts to 20 optimize and standardize sampling and analysis methods, especially within the 21 framework of large international projects, are crucial. The multifarious nature of 22 biological systems, coupled with the rapid evolution of technology, present significant 23 challenges to standardization efforts (Thomsen and Willerslev, 2015; Goldberg et al., 24 2016; Bruce et al., 2021; Buxton et al., 2021; Thalinger et al., 2021). Variability in the 25 techniques for sample collection and processing can lead to discrepancies in data 26 interpretation and conclusions, undermining the reproducibility of research findings 27 (Katano et al., 2017; Bruce et al., 2021; Buxton et al., 2021). This issue is further 28 magnified in international projects in which variations in technical expertise, resources 29 (e.g., field or laboratory equipment), and regulatory environments across participating 30 laboratories can exacerbate inter-institutional inconsistencies. To address these 31 challenges, several guidelines have been published, making a first attempt to 32 summarize best practices in eDNA research from preliminary sampling to post hoc 33 bioinformatic processing (Loeza-Quintana et al., 2020; Minamoto et al., 2021; Morisette 34 et al., 2021; Blancher et al., 2022, Bruce et al. 2021). Furthermore, working groups 35 consisting of eDNA specialists are being established internationally to monitor and 36 assess current methods and applications of eDNA research. Examples include a 37 subgroup of the European Committee for Standardization (CEN/TC 230/WG 28; 38 https://www.cencenelec.eu/), the USA Government eDNA Working Group (GeDWG; 39 usgs.gov), and the international eDNA Society (ednasociety.org). 40 41 Among processing steps such as field sampling and target DNA amplification, 42 eDNA-based data is subject to the protocol with which the genetic material was 43 extracted from the environmental sample. DNA extraction encompasses a series of 44 intricate steps, including cellular lysis, DNA isolation, protein and contaminant washing, 45 and final elution of high-quality DNA (Knebelsberger and Stöger, 2012; Barbosa et al., 46 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 3 2016). The widespread adoption of commercial kits, replete with the necessary reagents 47 for extracting DNA from a variety of mediums (e.g., tissue, water, soil), have become 48 common practice throughout recent years as they provide easily accessible, 49 streamlined, and reproducible protocol for retrieving genetic material from a sample 50 (Lear et al., 2018; Pearman et al., 2020). Some of the most widely used and 51 recommended commercially available extraction kits are Qiagen’s (Venlo, The 52 Netherlands) DNeasy kit, which is recommended by the official manual for eDNA 53 research published by the eDNA Society (Ficetola et al., 2008; Lear et al., 2018; Tsuji et 54 al., 2019; Minamoto et al., 2021), and the PowerWater DNA Isolation Kit (Mobio, Hilden, 55 Germany). According to a review of eDNA extraction approaches by Kumar et al., 2019, 56 a distinguishing feature of some kits (such as the PowerWater kit) lies in its built-in PCR 57 inhibitor removal step, which can also be conducted after extraction is carried out (e.g., 58 with Zymo OneStep PCR Inhibitor Removal Kit). This is highly relevant for environments 59 with high levels of suspended particulate matter or poor water quality, which are likely 60 the source of PCR inhibitors such as humic acids, fulvic acids, and polysaccharides 61 (Kuhn et al., 2017; Lear et al., 2018). However, inhibitor removal also introduces the 62 potential of losing target DNA due to increased agitation of the lysate (McKee, Spear 63 and Pierson, 2015; Goldberg et al., 2016) and the inclusion of this step does not 64 guarantee superior extracts. Consequently, the efficacy of the chosen extraction and 65 potential inhibitor-removal approach is contingent upon its compatibility with the specific 66 taxonomic, geographical, and environmental attributes of the study, warranting 67 meticulous consideration. 68 69 Following extraction, total DNA concentration in an extract can be measured via 70 spectrophotometry or fluorometry (Brunker, 2020; García-Alegría et al., 2020). 71 Meanwhile, targeted approaches, such as quantitative PCR (qPCR) and droplet digital 72 PCR (ddPCR) can be used to ascertain the presence and abundance of a specific 73 species within the sampled environment by discerning particular genetic traces of 74 interest amidst a heterogeneous sample, in which the quantity of target DNA is likely 75 present at very low concentrations (Goldberg et al., 2016; Hunter et al., 2017). qPCR 76

(with assays either using an intercalating dye or a fluorescently labeled probe 77 for quantification) are the most widely used technique for attaining species-specific 78 detections (Thalinger et al. 2021). However, its success depends upon precise assay 79 design entailing selectivity that precludes the amplification of nontarget taxa co-existing 80 with the focal species. Moreover, a rigorous validation regimen spanning in silico, in 81 vitro, and in situ evaluations is needed to forestall spurious reactions (primer dimers, 82 hairpins, etc.) and enhance the applicability to eDNA samples collected from the field. 83 These testing protocols have recently been presented by Thalinger et al. (2021) as a 5-84 level validation scale, beginning at Level 1 with simple in silico and nontarget tissue 85 testing to Level 5 with statistical testing of an assay’s detection probability as well as 86 modeling with ecological and physical factors which may influence the rate of perception 87 of a sample’s DNA (Garafutdinov, Galimova and Sakhabutdinova, 2020; Klymus et al., 88 2020). 89 90 The Limit of Detection (LOD) and Limit of Quantification (LOQ) are two metrics 91 which describe the sensitivity and quantitative precision of DNA assays. The LOD 92 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 4 delineates the lowest amount of DNA that can be consistently detected, while the LOQ 93 specifies the minimum level at which DNA can not only be detected, but also quantified 94 with acceptable accuracy and precision (Hunter et al., 2017; Klymus et al. 2020, 95 Thalinger et al., 2021). However, caution must be exercised when accounting for these 96 metrics during data analysis, particularly when interpreting positive amplifications which 97 fall below an assay’s Limit of Detection, as highlighted by Klymus et al. (2020). Taking a 98 conservative approach by excluding data below LOD may result in the loss of actual 99 detections of the target species. Conversely, the LOD can serve as a comparative 100 benchmark for inter-laboratory processing of the same sample. 101 102 Profoundly influencing an assay’s LOD and LOQ is the design of qPCR primers 103 (and often a fluorescently labeled probe), a process which is now often supplemented 104 by advanced machine learning or automation (Kronenberger et al., 2022; Allison et al., 105 2023). The careful design process ensures the sensitivity and specificity of qPCR, which 106 is vital for distinguishing low abundance targets (Wilcox et al., 2013; Rees et al., 2014). 107 Furthermore, the choice of PCR chemistry and cycling conditions is equally critical for 108 optimizing assay performance (Klymus et al., 2020; Langlois et al., 2021). In summary, 109 the analytical workflow of eDNA extraction and analysis involves intricate considerations 110 ranging from the means of obtaining DNA from a sample (e.g., the selected extraction 111 kit) to the development and refinement of a species-specific assay. Although this is only 112 part of the eDNA workflow, the manifold options available for DNA extraction and target 113 DNA amplification already make comparative tests a requirement before direct data 114 comparisons and applications. 115 116 In the context of international efforts to enhance the detection rates of marine 117 megafauna DNA from environmental samples, a key focus has consistently relied on 118 refining of both field sampling and laboratory protocols. This endeavor led to the 119 initiation of a comparative study (i.e., a ring test) involving four laboratories from 120 different countries who are all working together in an international research project titled 121 eWHALE, which aims to study marine megafauna across the North-East Atlantic and 122 Mediterranean Sea using eDNA-based methods. The four laboratories are: University of 123 Innsbruck (UIBK; Austria), the National Research Institute for Agriculture, Food, and the 124 Environment (INRAE; France), University College Cork (UCC; Ireland) and the Institute 125 for Marine Research (IMR; Norway), each relying upon specialized molecular 126 techniques, particularly eDNA extraction methods. Our aim was to compare the 127 efficiency of extraction protocols for a variety of eDNA samples collected around various 128 marine megafauna species. Additionally, three single-species qPCR assays, which can 129 be utilized by other eDNA specialists in future studies, were developed to specifically 130 amplify sperm whale (Physeter macrocephalus), porbeagle shark (Lamna nasus) and 131 basking shark (Cetorhinus maximus) DNA from environmental samples. We aimed to 132 evaluate the efficacy of laboratory-specific extraction techniques by comparing both 133 total DNA yield and target species DNA yield. This evaluation is crucial for filling the 134 existing gap in the standardization of eDNA monitoring methodologies across various 135 institutions, namely for the purpose of assessing mobile species with spatial ranges 136 beyond country borders. 137 138 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 5

139 140 Field Sampling 141 142 In summer 2023, water samples were collected from different regions throughout the 143 North-East Atlantic and Mediterranean Sea by researchers, students, and partners who 144 were trained in eDNA sampling (Fig. 1; Supplementary File 1). Samples were filtered 145 from the surface of the water column through different commercially available 146 environmental DNA filters: Smith-Root (Vancouver, USA), Sylphium (Sylphium 147 molecular ecology, Groningen, The Netherlands), and Sterivex™ (Millipore®; Merck 148 Chemicals and Life Science GesmbH, Darmstadt, Germany; Table 1). At the end of 149 filtration, all filters were dried by running the pump for an additional 30 s to 1 min outside 150 of the water or pushing air through the filter with a syringe. Storage buffer consisting of 151 TES buffer (0.1 M TRIS, 10 mM EDTA, 2% sodium dodecyl sulfate; pH 8) and 152 proteinase K (20 mg/mL) in a ratio of 190:1 was added to each filter (1.5-3 mL 153 depending on the filter type, see below) except from Smith-Root filters. Between 154 samples, the tubing was rinsed three times with household bleach and three times with 155 tap water (marine species-DNA-free) to prevent cross-contamination. 156 157 In the Mediterranean Sea, samples (n=68) were collected in volumes of either 2, 158 5, or 10 L with a bucket, with 17 samples collected in close proximity to sperm whales 159 (Physeter macrocephalus; Fig. 1). Nine water samples were immediately filtered 160 through self-preserving Smith-Root filter capsules (1.2 µm filter pore size) using a 161 peristaltic pump (Solinst; Model 410; Thomas et al., 2019). Filter capsules were stored 162 at 4 °C on board then in a facility in the harbor of San Remo (Italy) following the cruise 163 until a subset (n=6) was shipped to UIBK in October 2023 for subsequent analysis. 164 165 In the North-Eastern Atlantic Ocean waters around the Azores islands of Faial 166 and Pico (Fig. 1), researchers aboard CW Azores whale watching cruises 167 (cwazores.com) used a bucket to collect 10 L of water (n=42 samples) from sperm 168 whale flukeprints, which were immediately filtered through Sylphium filter capsules (0.8 169 µm filter pore size; ID: SYL010-08-20) using a peristaltic pump (ID: 12.34.SB; 170 Eijkelkamp, Giesbeek, The Netherlands). All filters were filled with 1.5 mL of storage 171 buffer which included an Internal Positive Control (IPC), an artificial fragment of DNA 172 used for quality control, from Sinsoma GmbH (https://www.sinsoma.com/en/). eDNA 173 filters were stored at the University of the Azores in a -20 °C freezer until being 174 transported to UIBK in July 2023 for subsequent analysis (n=5 used in this study). 175 176 In the Shannon Estuary, 9 water samples were collected with a 12 L bucket from 177 the fluke prints of bottlenose dolphins (Tursiops truncatus). One short-beaked common 178 dolphin (Delphinus delphis) sample was collected in the same manner off the South-179 West Coast of Ireland (near Baltimore, Cape Clear Island; Fig. 1). From these water 180 samples, between 1.5 and 2 L were filtered through Sterivex-HV filter capsules (0.45 µm 181 pore size; Merck Millipore ID: SVHV010RS) using 50 mL disposable syringes. 182 Afterwards, 1.5 mL of storage buffer were added. The filter capsules were stored in a 183 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 6 cooler on ice before being transferred to a -20 °C freezer upon return to the laboratories 184 at UCC’s North Mall Campus. 185 186 In the French National Nature Reserve of the Seven Islands in Brittany (Fig. 1) a 187 total of 10 water samples targeting porbeagle sharks (Lamna nasus) eDNA were 188 collected between June and September 2023. Five water samples were collected in 189 5.75 L containers and fully filtered through Sylphium capsules (0.8 µm filter pore size; 190 ID: SYL010-08-20) using a suction pump. The other five water samples were directly 191 filtered from the water using the same type of capsules and the same pump for 5 192 minutes. Once the filters were pumped dry, 3 mL of storage buffer were added. The 193 filters were then stored at -20 °C until analysis at INRAE. 194 195 In the Norwegian Sea by the Lofoten Islands (Fig. 1), 8 surface water samples 196 were collected targeting basking sharks (Cetorhinus maximus). Each 5 L sample was 197 filtered through Sterivex-HV filter capsules (0.45 µm pore size) using a peristaltic pump. 198 A 50 mL syringe was used to push air through the filters before 1.5 mL storage buffer 199 was added. Filters were stored at -20 °C until further analysis at IMR. 200 201 202 Figure 1. Locations in which eDNA samples analyzed for this ring test were collected. 203 Points are colored according to the target species. Cartography was created using 204 QGIS v 3.34.3 using ESRI basemap services (Esri, DeLorme, HERE, MapmyIndia). 205 206 Sample lysis and extraction 207 208 All filters were incubated for 3 h at 56 °C. Prior to incubation, each Smith-Root filter was 209 removed from its housing (using DNA-free forceps) then soaked with 400 µL of storage 210 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 7 buffer. After incubation, each Smith-Root filter was transferred into a plastic inlet placed 211 inside the original reaction tube and centrifuged at 18,626 g for 10 min to separate the 212 lysate from the filter. For Sylphium filter capsules, lysis buffer was removed using a 3 213 mL or 6 mL syringe, resulting in 1 to 1.4 mL lysate per sample for UIBK and 2.25 to 6 214 mL for INRAE. Lysis buffer was removed from Sterivex filters using a 2 mL syringe 215 resulting in 0.5 to 2.0 mL of lysate per sample. 216 217 Each laboratory received a 250 µL aliquot of lysate per eDNA sample. Samples 218 with less than 1 mL lysate were diluted with TES buffer to 1 mL total volume before 219 aliquoting. At UIBK, an extraction IPC (IPC-L: approximately 5,000 copies per sample; 220 Sinsoma GmbH) was added to each aliquot. Generally, lysates were stored at -80 °C 221 prior to shipping (packaged with ice in styrofoam containers) between project partners in 222 fall 2023. We opted for overnight shipping whenever possible, but in some cases, 223 lysates took 1-2 days to arrive at their final destination. Once eDNA lysates arrived at 224 their destination, they were stored at -80°C or -20°C prior to further analysis. 225 226 Table 1. Overview of eDNA filters, assays, and analysis techniques per participating 227 laboratory. 228 Parameter UIBK INRAE UCC IMR Filtration technique Peristaltic pump Suction pump Syringes Peristaltic pump eDNA filter Smith-Root (n=6), Sylphium (n=5) Sylphium (n=10) Sterivex (n=10) Sterivex (n=8) Filter pore size 1.2 μm 0.8 µm 0.8 µm 0.45 µm 0.45 µm Filter

Polyethersulfon e (PES) Polyethersulfone (PES) PVDF PVDF Target species Sperm whale (Physeter macrocephalus) Porbeagle shark (Lamna nasus) Bottlenose dolphin (Tursiops truncatus) and Common dolphin (Delphinus delphis) Basking shark (Cetorhinus maximus) Extraction

Qiagen BioSprint® 96 Workstation using the Biosprint 96 Macherey-Nagel NucleoSpin Tissue Kit** Qiagen DNeasy Blood and Tissue Kit** Qiagen DNeasy Blood and Tissue Kit** Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 8 tissue protocol ** Sanger sequencing Eurofins Genomics Germany GmbH (Ebersberg, Germany) GenoScreen (Lille, France) Eurofins Genomics Germany GmbH (Ebersberg, Germany) University Hospital of North Norway (Tromsø, Norway) ** Modifications made to protocol (see main text for details) 229 230 Extraction protocol per laboratory 231 232 Each project partner employed a DNA extraction protocol that is commonly used in the 233 facility for high numbers of eDNA samples, each including slight modifications from the 234 original manufacturer’s protocols (Table 1). All extractions were performed in dedicated 235 laboratory spaces with proper ventilation and cleaning procedures adhering to the 236 processing of eDNA samples (e.g., surface cleaning with bleach, sterilized DNA-free 237 gloves and protective wear; Thalinger et al., 2021; Hymus, 2016). Additionally, PCR 238 preparation was conducted in separate rooms with appropriate PCR-dedicated 239 workbenches that are disinfected by UV light at least once per working day. 240 241 At UIBK, DNA extraction was performed with the BioSprint 96 instrument 242 (QIAGEN; Venlo, The Netherlands) using the BioSprint 96 DNA blood Kit (ID: 940057; 243 QIAGEN) in accordance with the manufacturer's instructions except for using 100 µL of 244 TE buffer instead of AE buffer for elution (Supplementary Material 1, DOI: 245 dx.doi.org/10.17504/protocols.io.q26g71p83gwz/v1). In total, 39 lysates were extracted in 246 one Biosprint run with one extraction control containing only elution buffer. 247 248 At INRAE, DNA extraction of lysates (n=39 eDNA, 1 control) was performed 249 using the Macherey-Nagel NucleoSpin Tissue kit (ID: 740952.50, Düren, Germany) 250 according to the manufacturer’s recommended protocol with the addition of 25 µL 251 proteinase K at the lysis step. To maximize DNA yield, the Buffer BE was heated at 70 252 °C and elution was repeated twice with the same 100 μL of Buffer BE with 3-minute 253 incubation time (Detailed extraction protocol can be found here: DOI: 254 dx.doi.org/10.17504/protocols.io.4r3l2q2yql1y/v1, Private link for reviewers: 255 https://www.protocols.io/private/7FDA25FB19BE11EFAE230A58A9FEAC02 to be 256 removed before publication. 257 258 At UCC, lysates were incubated at 56 °C for 1 hr prior to extraction. DNA 259 extraction of lysates (n=39 eDNA, 2 controls) was performed using the Qiagen DNeasy 260 Blood and Tissue Kit (ID: 69504; QIAGEN) with 100 µL elution volume in the final step. 261 262 At IMR, DNA extraction of lysates (n=39 eDNA, 4 controls) was performed using 263 the Qiagen DNeasy Blood and Tissue Kit (ID: 69504, QIAGEN). A QiaVAC 24 Plus 264 vacuum system (ID: 19413, QIAGEN) was used instead of centrifugation for spin 265 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 9 column steps, with a final elution of 100 µL in a centrifuge. Detailed extraction protocols 266 for UCC and IMR using the Qiagen DNeasy Blood and Tissue Kit can be found here: 267 DOI: dx.doi.org/10.17504/protocols.io.n92ld8m2ov5b/v1, Private link for reviewers: 268 https://www.protocols.io/private/565B501A19BA11EFAE230A58A9FEAC02 to be 269 removed before publication. 270 271 Per extract, 4 aliquots (25µL each) were generated and sent back to their 272 laboratory of origin for further analysis using the same shipping conditions as before. 273 274 Total DNA quantification 275 276 Each project partner measured the total DNA and the target DNA of extracts from their 277 original lysates (e.g., UIBK measured the extracts generated from the 11 sperm whale 278 eDNA samples for extracts created at all participating laboratories: UIBK, INRAE, UCC 279 and IMR). Total DNA concentrations (ng/µL) per extract were measured via a Qubit™ 280 fluorometer using the Qubit dsDNA High Sensitivity (HS) Assay Kit (Life Technologies, 281 Carlsbad, California, US; ID: Q32851). Qubit standards and DNA sample tubes were 282 prepared using low-bind tubes (ID: Q32856; Thermo Fisher Scientific, Waltham, MA, 283 USA) and 5 µL of extract (protocol: https://dx.doi.org/10.17504/protocols.io.bc6vize6). 284 All tubes were measured in triplicate. 285 286 Species-specific eDNA quantification 287 288 Assay development and validation 289 290 Targeted qPCR TaqMan MGB assays were developed for this study in order to amplify 291 DNA from the species of interest for field samples contributed by UIBK, INRAE, and 292 IMR. Primarily, full mitochondrial sequences from target and nontarget species (i.e., 293 closely related and/or co-occurring species) were obtained from publicly available 294 repositories (GenBank database at the National Center for Biotechnology Information 295 (NCBI), https://www.ncbi.nlm.nih.gov/genbank/). Sequences were aligned with Clustal 296 Omega (Sievers et al., 2011), and preliminary species-specific qPCR assays were 297 selected using assayID, a publicly available software tool 298 (https://github.com/jammc313/assayID). This program scans the inputted mitochondrial 299 sequence alignment file using Primer3 (Koressaar and Remm, 2007; Untergasser et al., 300 2012) to design primer/probe sets for previously defined windows across the full 301 mitogenome. Given a DNA sequence template, Primer3 generates primer/probe sets 302 optimized for various parameters that are critical to assay performance. This includes 303 primer/probe length, melting temperatures (Tm), GC content, and avoidance of 304 secondary structure formations, among others. The software is designed to maximize 305 specificity and efficiency in amplification, minimizing potential issues such as 306 dimerization or hairpin formation that can impair the qPCR assay's accuracy and 307 sensitivity. Sequence diversity and distance metrics are calculated for the regions 308 covered by the designed assays, including measures of target species genetic diversity 309 and distance measures between target and nontarget sequences (e.g., Shannon 310 Entropy, sequence similarity, nucleotide divergence). The assays are then ranked 311 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 10 according to their specificity and sensitivity. An ideal assay will target a region that has a 312 combination of a low genetic diversity for the target species sequences, and high 313 genetic distance to nontarget species sequences. A multivariate statistical method: 314 Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) is finally 315 used to rank designed assays and identify those with optimal specificity and sensitivity. 316 Assays with the highest rankings from this program were selected for further manual 317 development and in silico testing (described below per lab). 318 319 Three unique species-specific TaqMan assays were ultimately optimized for the 320 species of interest from UIBK. INRAE, and IMR (Table 2; Supplementary File 2). At 321 UIBK, a sperm whale qPCR assay was designed targeting the Cytochrome B (CYTB) 322 region of the mitochondrial genome. Binding regions and primer lengths were manually 323 adapted to enhance specificity (i.e., sufficient mismatches with nontarget taxa), 324 adhering to standard recommendations for TaqMan assays (Applied Biosystems Primer 325 Express v3.0.1; Life Technologies, Foster City, CA, USA) and minimizing the 326 occurrence of secondary structures using BioEdit v7 (Hall, 2004), Primer3 (Untergasser 327 et al., 2012), Primer Premiere (PREMIER Biosoft), and Primer Express 3.0.1 (Applied 328 Biosystems). The probe was labeled with 6-FAM and MGB-Q530 quencher (5’ and 3,’ 329 respectively, Table 2). 330 331 At INRAE, the assayID program identified a total of 295 primer/probe 332 combinations for porbeagle sharks. A total of 25 combinations that met the criteria of a 333 window size of 150-180 bp, no hairpin, oligo not ending with G and no “GGGG” string in 334 the oligos were retained. They were BLASTed (https://blast.ncbi.nlm.nih.gov/Blast.cgi) 335 to check for specificity with porbeagle sequences, and results matching with other 336 species or with mismatches with porbeagle sequences were excluded. The best 337 candidate targeted the ND1, and to improve its specificity, the last bp was manually 338 removed from the probe. The probe was labeled with 6-FAM and a BHQ-1 quencher (5’ 339 and 3’, respectively; Table 2). 340 341 At UCC, 8 primer pairs were selected, 3 from the above-mentioned assayID 342 program, 2 created using IDT PrimerQuest Tool, and 3 from existing literature 343 (Stoeckle, Mishu and Charlop-Powers, 2018; Greiner-Ferris, 2020). The specificity and 344 efficiency of these primers was initially tested in vitro (via conventional PCR and gel 345 electrophoresis) using DNA extracts of bottlenose dolphin (5 ng/µL, 0.5 ng/µL and 0.05 346 ng/µL), short-beaked common dolphin, harbour porpoise (Phocoena phocoena), killer 347 whale (Orcinus orca), long-finned pilot whale (Globicephala melas), sperm whale, fin 348 whale (Balaenoptera physalus) and grey seal (Halichoerus grypus). The primers 349 designed by Greiner-Ferris (2020), targeting the displacement loop (D-loop) region of 350 the mitochondria, were selected because they were the most specific to the target 351 species (Table 2; Supplementary File 2). The last base pair at the 3' end of the reverse 352 primer was removed so that the primers would better amplify the bottlenose dolphin 353 haplotypes found in the study area (Nykänen et al., 2019). A putative probe for TaqMan 354 chemistry was initially designed for this modified version of the primer pair using the IDT 355 PrimerQuest™ Tool. The probe/primer assay was then extensively tested using 356 standard dilutions of tissue-derived bottlenose dolphin DNA, but failed to detect target 357 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 11 DNA beyond 0.01 ng/µL. The primers were tested using SYBRgreen mastermix (without 358 the use of a probe) in which it was possible to detect target DNA beyond 0.001 ng/µL. 359 Thus, it was decided that all subsequent runs would be on SYBRgreen based chemistry 360 using the aforementioned primer pair. 361 362 At IMR, the assayID program resulted in several potential assays for Cetorhinus 363 maximus. Further in silico testing for target species specificity and tendency to form 364 secondary structures using Primer-BLAST (Ye et al., 2012) and Integrated DNA 365 Technologies OligoAnalyzer Tool (Integrated DNA Technologies, 2023), resulted in the 366 selection of the best performing assay targeting part of the ND5 region. The probe was 367 labeled with 6-FAM and NFQ-MGB quencher (5’ and 3,’ respectively, Table 2). 368 369 The specificity for all assays presented herein was verified in silico via Primer-370 BLAST (Ye et al., 2012), with standard settings and the nr database. No amplification of 371 closely related or co-occurring species was found for UIBK, IMR, and INRAE assays. 372 The assay for UCC amplified all Delphinidae species, including the target species 373 bottlenose dolphin and common dolphin. DNA extracts from target and nontarget 374 species were used for in vitro testing of all selected assays’ specificity. Tissue samples 375 used for in vitro testing of species-specific assays were dried in a fume hood and then 376 extracted using either the Qiagen DNeasy Blood and Tissue Kit (ID: 69504, QIAGEN) or 377 the Macherey-Nagel NucleoSpin Tissue Kit (ID: 740952.50, Düren, Germany) following 378 the manufacturer’s instructions. 379 380 Upon optimization of cycling conditions (Supplementary Material 2), the Limit of 381 Detection (LOD) and Limit of Quantification (LOQ) were calculated for each assay 382 following the definitions of Klymus et al., 2020 using measurements from triplicate 383 standard curves (per qPCR plate) of serial dilutions of target DNA from known 384 concentrations (ng/µL; Table 2). Ultimately, there was insufficient statistical power to 385 calculate each assay’s LOQ in accordance with the defined calculation method (Klymus 386 et al., 2020). 387 388 Table 2. Assays used in the current study for the amplification of target species DNA. 389 Institute Target Species Gene Name Forward (5’-3’) Reverse (5’-3’) MGB Probe (5’-3’) Fragment length (bp) Optimal Annealing Temp. Assay LOD** UIBK Sperm whale (Physeter macrocephalu s) CYTB Phy-cat-S939 * Phy-cat-A939 * P030_Phy-cat * CCTACCACACAAT CAAAGACACC GGTTTGATGTGT GTTGGGGTAT TAGTGGATTTGCT GGGGTGTA 144 61 0.0001 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 12 INRAE Porbeagle shark (Lamna nasus) ND1 LnND1-F209 * LnND1-R380 * LnND1-P242 * TCAGCATCTTCCC CTTTCCT ATCCGGAGCCCA AGATAGTG CCCACAATGGCT CTTACACTGGCC CTCCT 172 60 0.0005 26 UCC Bottlenose dolphin (Tursiops truncatus) and Common dolphin (Delphinus delphis) D-loop TtDloopF TtDloopR CACACGTGCATG CTAATATTTAG GAGTGACCATAG GATATAATGGAG 159 60 0.0000 1 IMR Basking shark (Cetorhinus maximus) ND5 CetoMaxND5F _01 * CetoMaxND5R _01 * CetoMaxND5P _01 * AGTTTCCGCCCT ACTCCATT GCTGCGGTAAAG AGGGTAGT AGTCGTTGCCGG CGTCTTCCTGCTA 144 60 0.0001 * Created for this study 390 ** As described by the discrete method presented in Klymus et al., 2021 391 392 qPCR 393 394 Each participating laboratory performed qPCR with triplicates of each eDNA extract and 395 triplicate serial dilutions of known concentration (6-8 points of tenfold dilution starting at 396 either 1 or 0.526 ng/µL) on each plate. 397 398 At UIBK, qPCRs were carried out using a qTOWER3G (Jena, Germany). Primary 399 qPCRs sought to amplify IPC-F and IPC-L to assess the amount of DNA that is lost 1) 400 during transportation and storage of eDNA filters and 2) during extraction. Triplicate 401 dilution series of IPC-F (1:10, 15 copies/µL, 6 points) and IPC-L (1:10, 1,000 copies/µL, 402 6 points) as well as two No Template Controls (NTC; i.e., nontarget extracts) and nine 403 Negative Controls (NC) were included on each plate. The 10 µL reactions for both IPCs 404 contained 5 µL 2x TaqMan Environmental MM (ID: 4396838, EMM, Life Technologies), 405 1.0 µL of Primer mix and 0.4 µL of Probes mix (Sinsoma GmbH), 0.6 µL nuclease-free 406 water, and 3 µL eDNA extract or positive control or NTC. Optimized thermocycling 407 conditions were 1) an enzyme activation step at 95 °C for 10 min, 2) a denaturation step 408 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 13 (40 cycles) at 95 °C for 15 s, and 3) a combined annealing and extension step (40 409 cycles) at 60 °C for 90 s. 410 411 For the sperm whale assay (UIBK), 10 µL reactions contained 5 µL 2x TaqMan 412 Environmental Master Mix, 1 µL primer/probe mix (0.4µM per primer and 0.2µM probe), 413 1 µL nuclease-free water, and 3 µL eDNA extract or NTC. Triplicate dilution series of 414 sperm whale DNA (1:10, 1 ng/µL, 6 points as well as two NTC and nine NC were 415 included on each plate. Optimized thermocycling conditions were 1) an enzyme 416 activation step at 95 °C for 10 min, 2) a denaturation step (40 cycles) at 95 °C for 15s, 417 and 3) a combined annealing and extension step (40 cycles) at 61°C for 90s. 418 At INRAE, qPCRs were carried out using a BioRad CFX96 (Bio-Rad Laboratories, 419 Hercules, CA). Triplicate dilution series of porbeagle shark DNA (1:10, 0.526 ng/µL, 6 420 points) as well as one NTC and one NC were included on each plate. The 20 µL 421 reactions contained 10 µL 2x TaqMan Environmental MM (ID: 4396838, EMM, Life 422 Technologies), 1.0 µL of Primer mix and 1 µL of probe, 4 µL nuclease-free water, and 3 423 µL eDNA extract or positive control or NTC. Optimized thermocycling conditions were 1) 424 an enzyme activation step at 95 °C for 10 min, 2) a denaturation step (49 cycles) at 95 425 °C for 30 s, and 3) a combined annealing and extension step (49 cycles) at 60°C for 426 00:50. 427 428 At UCC, qPCRs were run on Applied Biosystems 7500 Real-Time PCR System 429 (Foster City, CA). Triplicate dilution series of bottlenose dolphin DNA (1:10 starting at 1 430 ng/µL, 6 points) were used as standards on each plate as well as one negative control. 431 An internal PCR positive control, Kavlick IPC was used to test for possible inhibitors in 432 the extracted eDNA samples in a separate run (Kavlick, 2018). 10 µL reactions 433 contained 5 µL 2x SYBR™ Green PCR Master Mix (ID: 4309155, Applied 434 Biosystems™), 0.8 µL of Primer mix (0.4 µL forward and reverse primer, 2.2 µL 435 nuclease-free water, and 2 µL eDNA extract or positive control or NTC. Optimized 436 thermocycling conditions were 1) an enzyme activation step at 95 °C for 10:00, 2) a 437 denaturation step (40 cycles) at 95 °C for 00:15, and 3) a combined annealing and 438 extension step (40 cycles) at 60°C for 1 min. 439 440 At IMR, qPCRs were carried out using an Applied Biosystems QuantStudio Flex 441 6 Real-Time PCR system (Applied Biosystems, Foster City, CA, USA). Triplicate dilution 442 series of basking shark DNA (1:10, 1 ng/µL, 8 points) were used as standards as well 443 as one NC. The 10 µL reactions contained 5 µL 2x PerfeCTa qPCR Master Mix (ID: 444 101419-220, QuantaBio), 0.5 µL 150 nM TaqMan Custom gene expression assay, 1.3 445 µL nuclease-free water, 1 µL ThermoFisher IPC Exo Mix, 0.2 µL ThermoFisher IPC Exo 446 DNA, and 2 µL eDNA extract or positive control or NTC. Optimized thermocycling 447 conditions were 1) an enzyme activation step at 95 °C for 10 min, 2) a denaturation step 448 (55 cycles) at 95 °C for 15 s, and 3) a combined annealing and extension step (55 449 cycles) at 60°C for 1 min. 450 451 Sequencing 452 453 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 14 Following positive amplification of target species DNA using qPCR, amplifications were 454 verified via Sanger sequencing of both the forward and reverse strands. At UIBK, 455 positive qPCR amplifications of target DNA were purified using an enzymatic treatment 456 kit (ExoSAP-IT® Express PCR Product Cleanup Reagent; Affymetrix-USB Corporation, 457 Santa Clara, California, USA) then sent to Eurofins Genomics Germany GmbH 458 (Ebersberg, Germany). At INRAE, the target fragments from eDNA products yielding 459 positive qPCR amplifications were re-amplified using end-point PCR (Supplementary 460

3) and bands corresponding to the expected size on a 2% agarose gel were 461 extracted and purified using the NucleoSpin Gel and PCR Clean-up (Macherey-Nagel, 462 Dünel, Germany) before being sequenced by GenoScreen (Lille, France). At UCC, 463 qPCR products yielding amplifications were run on a 2% agarose gel at 100 V for 1 h. 464 Bands which corresponded to the expected length of the target species fragment (200 465 bp) were extracted and purified using the QIAquick Gel Extraction Kit (Qiagen, Venlo, 466 The Netherlands). The products were sent to Eurofins Genomics Germany GmbH 467 (Ebersberg, Germany). At IMR, positive qPCR amplifications of target DNA were 468 purified using ExoSAP-IT®, Sanger sequencing reactions were performed using 469 BigDye™ Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems™ 4404310), and 470 sequencing products were sent to University Hospital of North Norway (Tromsø, 471 Norway). 472 473 Resulting sequences were trimmed to remove low quality ends using Bioedit 7.7 474 (Informer Technologies), AliView 1.28 (Uppsala University), Geneious 7.1.9 (Biomatters, 475 Auckland, New Zealand) and UGENE 50.0 (Unipro). The taxonomic classification of all 476 DNA sequences obtained from sequencing was determined using BLAST (NCBI). 477 478 Statistical analysis 479 480 Total DNA measurements (via Qubit fluorometer) were tested for statistically significant 481 differences between laboratories using a Generalized Linear Mixed Model (GLMM) in 482 SPSS (v 28; IBM Corp, Armonk, NY, USA) with the following formula: 483 484 Yijk= µ + πi + αj + παij + εijk 485 1) 486 487 In which Υijk is the DNA concentration (in ng/µL) of the extract for the k-th observation in 488 the j-th laboratory from the i-th source, µ is the general mean, πi is the random effect of 489 the i-th source in which samples originated from (i.e., the different target species), αj is 490 the fixed effect of the j-th laboratory performing the eDNA extraction, παij is the random 491 effect of the i-th source in which sample extracts originated from the j-th laboratory, and 492 εijk is the error term. 493 494 For qPCR amplifications, statistical tests were carried out on two datasets: one 495 with all positive amplifications of the target species (hereafter “all-inclusive dataset”) and 496 one with a subset of positive amplifications that were at or below an assay’s Limit of 497 Detection following the definition provided in Klymus et al. (2021) in which at least 95% 498 of PCR replicates for a given standard DNA concentration is amplified (hereafter 499 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 15 “conservative dataset”). qPCR amplifications were tested for significant differences 500 between detection probability (based on a binary variable for target species 501 amplification) using Pearson’s Chi-Square tests. Furthermore, Chi-Square was also 502 applied to test for significant deviations in the detection probability by species (i.e., 503 basking shark, dolphin species, porbeagle shark, sperm whale). The detection 504 sensitivity of positive amplifications (defined henceforth by the Ct values, a proxy for the 505 DNA concentration within an extract; measured in triplicate per extract) were analyzed 506 using a Kruskal-Wallis test with a post-hoc Dunn’s test to identify which extraction 507 technique had a positive or negative effect on amplification strength (i.e., high or low Ct 508 values). High Ct values correspond to lower concentrations of target DNA within an 509 extract since it takes a higher cycle number (Ct) for the DNA to be detected by the 510 cycler. Statistical significance was defined for all tests at a p-value < 0.05. 511 512 Figures were created in R (v 4.3.1; R Core Team, 2023) using ggplot2 (v 3.4.4; 513 Wickham et al., 2024), dplyr (v 1.1.3, Wickham et al., 2023) reshape2 (v 1.4.4; 514 Wickham, 2020) and viridis (v 0.6.5; Garnier et al., 2024). All code that was used to 515 generate figures and statistical results can be found at https://github.com/eWHALE-516 DNA. All data used for this publication can be found in the Supplementary Files: 517 Qubit_Data.csv and qPCR_Data.csv. 518 519

520 521 Total DNA quantification 522 523 Total DNA (ng/µL) measurements for all samples did not differ significantly between 524 laboratories (GLMM p > 0.05; Fig. 2). The sample source (i.e., the random factor) and 525 the sample source*laboratory (i.e., target species and laboratory performing the 526 extraction) interaction were both found significant (p < 0.05), with sample source having 527 the largest effect size (Table 3). Initially, the model also considered the effect of sample 528 transportation as a binary variable to assess potential DNA degradation during shipping. 529 However, this factor did not significantly influence the results and led to a higher AICc, 530 indicating a less efficient model fit (Bolker et al., 2009). 531 532 Table 3. Results from GLMM regarding the effects of extraction protocol on samples 533 from each partner. Laboratory x Sample source indicates the interaction term between 534 laboratory-specific extraction protocol and the source from which samples were sent 535 from (i.e., different target species). 536 Model Variable Sum of Squares DF Mean Square F Sig. Intercept Hypothesis 3742.60 1 3742.60 2.38 0.220 Error 4712.87 3 1570.94 Laboratory Hypothesis 128.43 3 42.81 2.03 0.180 Error 190.18 9.01 21.10 Hypothesis 4733.47 3 1577.82 74.50 < 0.05** Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 16 Sample Source Error 190.65 9 21.18 Laboratory x Sample Source Hypothesis 190.70 9 21.19 7.24 < 0.05** Error 1453.83 497 2.93 * Sum of Squares indicates the total variation in the data accounting for the presence of 537 other factors included in the model. Degrees of Freedom (DF) indicates the number of 538 observations of this parameter minus the number of estimated parameters. The Mean 539 Square is the average variation explained by random effects. F is a statistic taking into 540 account the mean square of the effect and the mean square of residuals. Significance 541 value estimates the statistical significance of model parameters to the overall fit of the 542 model. 543 ** Indicates statistical significance 544 545 546 Figure 2. Total DNA concentration per eDNA sample (x-axis) and extract (ng/µL) 547 measured in triplicate with a Qubit fluorometer. Individual points represent replicate 548 measurements (n=3) per extract. Note the variation of y-axis range per plot. For an 549 expanded version of this graph, showing the exact measurements in triplicate per 550 laboratory per sample as their own box plots, see Supplementary Fig. 1. 551 552 Species-specific DNA quantification 553 554 Out of 468 qPCR reactions across all target species (not including standards, NCs or 555 NTCs), 115 successfully amplified their respective target species DNA during qPCR (all-556 inclusive dataset; Fig. 3, Supplementary Material 5). Out of those detections, 79 557 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 17 exhibited DNA concentrations at or below the LOD (conservative dataset; Fig. 3; 558 Supplementary Material 5). 559 560 Across all 39 water samples, 28 yielded at least one positive qPCR reaction (out 561 of triplicate measurements) for the associated target species in the all-inclusive dataset. 562 For UIBK samples, the target species (sperm whale) was detected in 7 out of 11 563 samples: 4 samples with one PCR replicate from UIBK and UCC extracts and 3 564 samples with at least 2 PCR replicates from UIBK, INRAE, UCC, and/or IMR extracts. In 565 total, 43 out of 132 reactions (11 samples, 4 extracts each, measured in triplicate 566 qPCRs) amplified sperm whale DNA. For INRAE samples, the target species 567 (porbeagle shark) was detected in 8 out of 10 samples: 4 samples with one PCR 568 replicate from UIBK, INRAE, and UCC extracts and 4 samples with at least 2 replicates 569 from UIBK and UCC extracts. In total, 13 out of 120 reactions (10 samples, 4 extracts 570 each, measured in triplicate qPCRs) amplified porbeagle shark DNA. For UCC samples, 571 the target species (dolphin species) was detected in all 10 samples: 3 with one PCR 572 replicate from INRAE, UCC, and IMR extracts and 7 with at least 2 PCR replicates from 573 UIBK, INRAE, and/or UCC extracts. In total, 56 out of 120 reactions (10 samples, 4 574 extracts each, measured in triplicate qPCRs) amplified dolphin DNA. For IMR, the target 575 species (basking shark) was detected in 3 out of 8 samples: all detections were only 576 from one PCR replicate from either UIBK or UCC extracts. In total, 3 out of 96 reactions 577 (8 samples, 4 extracts each, measured in triplicate qPCRs) amplified basking shark 578 DNA. There were no amplifications of NC or NTC. 579 580 581 Figure 3. Species-specific detections with standard (i.e., target extract) dilution series of 582 known DNA concentrations (ng/µL) for each lab. The LOD for each assay is denoted as 583 a dashed blue line. Only positive detections below or at the assay’s LOD (i.e., below the 584 blue line) were included in the conservative dataset. 585 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 18 586 Chi-Square Tests of detection probability between extraction methods for all 587 qPCR reactions indicated that IMR detected target species significantly less than other 588 labs for both datasets (all-inclusive dataset: Chi-Square=21.155, df=3, p < 0.05, Fig. 4A; 589 conservative dataset: Chi-Square=14.302, df=3, p < 0.05, Fig. 4B). Detection probability 590 did not differ significantly between UIBK, INRAE, and UCC (Fig. 3; see Supplementary 591

5 for further details). 592 593 A) 594 B) 595 Figure 4. Results of qPCRs organized by the laboratory performing the extraction 596 (triplicate reactions; x-axis) and the eDNA sample number and target species (y-axis). 597 Tile color represents the Ct value; reactions without amplification were left blank. Panel 598 A shows the all-inclusive dataset: all 115 detections, whereas panel B shows the 599 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 19 conservative dataset: 79 positive detections that were at or below the respective 600 assay’s LOD. For alternative visualization, see Supplementary Fig. 2. 601 602 Across all positive amplifications, extracts generated by INRAE had significantly 603 lower DNA concentrations (i.e., higher Ct values) than the other partner laboratories for 604 both datasets (all-inclusive dataset: Chi-Square=12.215, df=3, p < 0.05; conservative 605 dataset: Chi-Square=24.322, df=3, p < 0.05; Fig. 3 & 4). Pairwise comparisons of the 606 extracts (Table 4) generated in different laboratories via Dunn’s Multiple Comparison 607 Test revealed a significant difference in DNA concentrations (Ct values) between UCC 608 and UIBK for the all-inclusive dataset (Z-score=2.730, p < 0.05), likely due to the 609 variation in detections of porbeagle shark above the assay’s LOD, resulting in UCC 610 having higher Ct values on average than UIBK (Fig. 4A). Additionally, Dunn’s Multiple 611 Comparison Test identified a significant difference between the pairwise Ct values of 612 UIBK and INRAE for both datasets (Table 4), with UIBK having an average Ct value 613 3.189 lower than INRAE (i.e., higher target DNA concentration) for the 10 samples in 614 which both laboratories had positive amplifications (based on the all-inclusive dataset). 615 616 Table 4. Pairwise comparisons of triplicate Ct values by laboratory-specific extraction 617 method. Columns 2-3 represent the all-inclusive dataset with all 115 detections, 618 whereas columns 4-5 represent the conservative dataset with only the 79 detections 619 that were at or below the respective assay’s LOD. 620 Significance values have been adjusted by the Bonferroni correction for multiple tests 621 ** Indicates significantly different Ct values per sample between extraction methods 622 623 At UIBK, 9 (triplicates from 3 samples) out of 144 reactions (11 water samples 624 extracted by four partners and analyzed in triplicate in PCR) did not detect the 625 extraction IPC (IPC-L), suggesting either inhibition in these samples or human error 626 (e.g., from pipetting errors). Therefore, these extracts were further assessed for 627 inhibition by spiking 0.5 µL into a species-specific qPCR targeting Ichthyosaura 628 alpestris. All amplifications of the spiked reactions showed detections with similar Ct 629 values, indicating a lack of inhibition. At UCC, 5 out of 80 reactions (10 water samples 630 extracted by four partners and analyzed in duplicate in PCR) did not detect the IPC. At 631 Institute-Institute All-inclusive dataset Conservative dataset (Detections at or below LOD) Sample 1-Sample 2 Z-score Adjusted P-value Z-score Adjusted P-value UCC-UIBK 2.730 0.019** 1.796 0.217 UCC-IMR -0.457 1.000 0.267 1.00 UCC-INRAE 0.801 1.000 3.092 0.006** UIBK-IMR 1.439 0.451 1.458 0.434 UIBK-INRAE 3.250 0.003** 4.907 2.77e-06** IMR-INRAE -1.004 0.946 -1.854 0.191 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 20 IMR, 132 out of 138 qPCR reactions amplified the IPC added to the PCR master mix, 632 the remaining 6 reactions contained an IPC Block (Non-Amplification Controls) and did 633 not amplify. 634 635 Sanger Sequencing 636 637 For all positive PCR products sent for sequencing by UIBK, at least 1 qPCR replicate 638 (per DNA extract) was successfully Sanger sequenced and matched to the target 639 species (Physeter macrocephalus) using the Basic Local Alignment Search Tool 640 (BLAST; https://blast.ncbi.nlm.nih.gov/Blast.cgi). There were two instances in which the 641 sequence was also >97% identical to another species (Table 5). From the extracts 642 sequenced after qPCR by INRAE, two sequences were matched to the target species 643 (Lamna nasus) using BLAST. At UCC, out of the 26 PCR products sent for sequencing 644 13 resulted in low-quality reads (less than 50 bp) and the remaining 13 matched (>98% 645 identity) to the target species (Tursiops truncatus) using BLAST. All PCR products with 646 a visible band were successfully matched to the target species, whereas PCR products 647 that had no visible band on agarose gels yielded low-quality reads which could not be 648 positively matched with any species. Of the 4 qPCR products sent for Sanger 649 sequencing from IMR, all returned low-quality reads with no significant hits in BLAST. 650 651 Table 5. Detections of target species DNA via Sanger sequencing of positive PCR 652 products. 653 Detections UIBK INRAE UCC IMR Number of PCR products sent in for Sanger Sequencing 48 7 26 4 Number of PCR products matching to target species (> 97%) 32 2 13 0 Number of PCR products unable to be matched to any species* 15 1 13 4 Number of PCR products matching to a nontarget species (>97%) 2 (Drosophila spp., Scotophilus heathii) NA NA NA * Due to sequencing error or insufficient quantity of DNA for sequencing. 654 655

656 657 We aimed to evaluate the efficacy of laboratory-specific extraction techniques by 658 comparing both total DNA yield and target species DNA yield from extracts generated 659 from the same lysate by different laboratories. Our findings confirm that variations in 660 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 21 DNA extraction methodologies significantly influenced the detection of targeted marine 661 megafauna species in eDNA samples. Despite the general uniformity of total DNA 662 concentrations of extracts generated from the same eDNA sample, this ring test 663 identified significant disparities in the detections of target-specific DNA in extracts 664 generated by IMR using their modified Qiagen DNeasy Blood and Tissue Kit extraction 665 protocol. Three novel species-specific TaqMan assays were designed to amplify marine 666 megafauna species. Out of all positive qPCR replicates utilizing these assays (which 667 were sent in for Sanger sequencing), 55% were successfully sequenced with at least 668 one positive verification for all extracts in which target DNA was detected. Notable 669 discrepancies in detections were observed between combinations of laboratory-specific 670 extraction protocols and target species, suggesting that the effectiveness of a particular 671 extraction technique was highly dependent on the sample type (e.g., the type of filter 672 used). Regardless, there were significantly less detections of shark species than marine 673 mammals overall, indicating species-specific variation in eDNA samples. This study 674 exemplifies the importance of assessing gaps in the reliability of eDNA analysis 675 protocols post field collection across multiple laboratories. 676 677 Lysates extracted by IMR with the modified Qiagen DNeasy Blood and Tissue kit 678 yielded significantly lower detection rates for all target species for both datasets (all-679 inclusive and conservative regarding the assay’s LOD), albeit extraction protocols being 680 almost identical between IMR and UCC, whose DNA extracts in total had the highest 681 number of positives in PCR (41 out of 115 reactions vs IMR 12 out of 115 reactions). 682 Upon reflection, two differences in the DNeasy protocol were found between IMR and 683 UCC. The first occurred prior to extraction in which lysates were incubated for an 684 additional hour before any subsequent work at UCC. This warming potentially improved 685 the binding of DNA to silica membranes (i.e., reduced the chance of clogging) by 686 preventing the precipitation of AL buffer (Lear et al., 2018). The other difference stems 687 from a vacuum being used at IMR during the DNA purification step for the purpose of 688 drawing the sample and binding buffer through the spin column which the DNA binds to 689 the membrane while other cellular components are washed away. The vacuum 690 technique versus the commonly used centrifugation protocol (which is also employed by 691 INRAE’s NucleoSpin protocol) may vary in their extraction performance due to several 692 factors: incomplete binding of DNA resulting from insufficient vacuum pressure, 693 inefficient washing of contaminants by the vacuum, and/or higher saturation of the 694 extraction column resulting in lower DNA yields. The difference between IMR and 695 UCC’s detection probability of target DNA could not be directly attributed to any of these 696 factors, but this finding demonstrates the effect of protocol modifications for downstream 697 analyses. Concerning differences induced by the mechanisms used for DNA binding 698 and separation of lysate components, the extraction robot used at UIBK has been 699 shown to be more robust in attaining the amount of total DNA within samples as the use 700 of paramagnetic beads avoids these steps altogether (Wallinger et al., 2017). 701 Accordingly, UIBK extracts detected the target species across multiple replicates with 702 higher concentrations of target DNA than other labs (i.e., lower Ct values; average Ct 703 across all positive amplifications: UIBK=32.54, INRAE=34.35, UCC=34.82, IMR=33.21). 704 705 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 22 All qPCR assays provided herein met Level 3 of the validation scale presented 706 by Thalinger et al. (2021) in which the target organism was successfully detected from 707 an environmental sample (Supplementary Material 4). In all cases, the specifics of DNA 708 extraction and concentration of eDNA from the environmental sample were reported. 709 The assays almost satisfied Level 4 of the validation scale in which the LOD has to be 710 calculated and in vitro qPCRs on co-occurring nontarget species have to be carried out, 711 yet failed to meet the expectation of extensive field testing as this was a preliminary 712 study on a subset of samples. For qPCR assays targeting a particular species of 713 interest, the assayID program can be used as a preliminary means to identify candidate 714 primer/probes. Further testing with manual in silico techniques and software is 715 recommended for all automated selections prior to ordering a costly hydrolysis probe. 716 The optimized assays were shown to be highly effective at detecting target species 717 DNA down to a low concentration as denoted by their LODs (0.0001 ng/µL for sperm 718 whale, porbeagle shark, and basking shark, 0.00001 ng/µL for dolphins; Klymus et al., 719 2020). Therefore, non-detections of target species throughout the course of this study 720 are likely due to the lack of (or extremely low concentrations of) quantifiable target DNA 721 in the extract (Eichmiller, Miller and Sorensen, 2016; Hunter et al., 2017). The LOD of 722 UCC’s SYBRgreen assay was one dilution point higher than the LOD of the TaqMan 723 (hydrolysis probe-dependent) assays. We hypothesize that this is due to the efficiency 724 at which target DNA can hybridize to the primer pair, whereas the primer pair + probe 725 assay for TaqMan-based qPCR is highly specific and may not bind to all target DNA 726 within an extract, which is suggested by other comparative studies (Cao and Shockey, 727 2012; Zhang et al., 2015). 728 729 Detection rates across all qPCRs varied exceptionally depending on the species: 730 sperm whales and dolphins were detected in 43 and 56 qPCR replicates, respectively, 731 whereas shark species were only detected in three (in the case of basking shark) and 732 13 (porbeagle shark) qPCR replicates and were completely absent from the 733 conservative dataset. This is justified from previous environmental DNA work with shark 734 species, which report low concentrations of eDNA from elasmobranch (sharks and rays) 735 taxa (Dunn et al., 2023). In contrast, extracts generated by all partner laboratories 736 amplified sperm whale DNA across almost all replicates for three separate eDNA 737 samples and dolphin DNA in eight samples for more than one qPCR replicate. Marine 738 mammals notably lose sloughed skin and dispel fecal matter while resting at the surface 739 of the water column in between feeding events (Whitehead et al., 1990; Konrad et al., 740 2018). Therefore, the genetic material that was collected during sampling events near 741 surfacing individuals likely provided sufficient quantities of eDNA to be collected, 742 filtered, and extracted. Not only does the behavior of each species affect their ability to 743 be detected by eDNA, but differences in field sampling may have also factored into 744 detection probability. We attribute the low frequency of detections for IMR’s assay 745 (basking shark) in both the all-inclusive and conservative datasets to field sampling at 746 too-great distances (spatially and temporally) from the target species in order to identify 747 significant concentrations of eDNA. Also, water volumes of samples taken near sharks 748 (both species) were 5 liters, whereas sperm whale samples were all 10 liters. Lysate 749 volumes from INRAE Sylphium filters were higher (up to 6 mL in some instances 750 compared to 1.5 mL from UIBK Sylphium filters) due to improper drying of the filter, 751 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 23 which may have consequently diluted the eDNA to a point in which it was no longer 752 detectable. Though only two liters of water were collected near dolphins, their tendency 753 to travel closely in groups at the water’s surface likely enhanced the amount of eDNA 754 present (Shane, Wells and Würsig, 1986; Acevedo-Gutiérrez and Parker, 2000; Gridley 755 et al., 2017). Therefore, while the extraction protocols did impact the capability of 756 partner laboratories to detect different species, there are many other factors to account 757 for during further ecological applications of these detections amongst others. 758 759 There was a significant difference observed between combinations of laboratory-760 specific extraction and the subsequent target species. This implies that the nature of the 761 sample (e.g., different environmental parameters at the sampling site) could have 762 affected the efficiency of DNA extraction following its collection from the field. Such a 763 theory is consistent with the findings of the Lear et al., 2018 review concerning eDNA 764 extraction, storage, amplification, and sequencing methods. Thus, while particular 765 extraction techniques may perform well with certain sample types (e.g., samples 766 including PCR inhibiting substances), they may not be universally applicable across all 767 eDNA samples, introducing significant discrepancies especially in the context of inter-768 laboratory comparisons. For example, INRAE yielded lower total DNA measurements 769 than other laboratories for their own eDNA samples (Sylphium filters, target species: 770 porbeagle shark) as well as UIBK eDNA samples (Sylphium and Smith-Root filters, 771 target species: sperm whale), yet similar measurements to other labs for UCC samples 772 (target species: dolphin species, Sterivex filters) and IMR samples (target species: 773 basking shark, Sterivex filters). This suggests that the NucleoSpin Tissue Kit may be 774 most effective with DNA lysed from Sterivex filters, which is in accordance with Tsuji et 775 al. (2019) who present that each commercial DNA extraction kit has shown dependence 776 on a combination of the eDNA collection method and the condition of water samples 777 (e.g., the degree of inhibition). For samples in which inhibition (compounds which may 778 disrupt PCRs) may be expected (e.g., in turbid marine environments), inhibitor-removal 779 kits and/or additional extraction protocol steps are often utilized to enhance the 780 detection of target DNA (Rees et al., 2014). For the purposes of this study, however, all 781 partner laboratories agreed to avoid inhibitor removal steps. Ultimately, the 782 implementation of various Internal Positive Controls corroborated the lack of inhibition 783 across all generated DNA extracts, confirming our choice to not risk the loss of target 784 DNA via an additional inhibition removal protocol. The use of high-quality PCR 785 chemistry could have also influenced the detection success of each qPCR assay (Beng 786 and Corlett, 2020; Thalinger et al., 2021), but this effect was not specifically tested in 787 the current study. However, external factors, namely the effect of shipment of eDNA 788 lysate and extracts, were included in preliminary analyses but did not show any 789 significant effect on the resulting concentrations of DNA per extract. 790 791 A total of 36 positive amplifications fell below the assays’ LOD and were 792 excluded from one of the two datasets used for statistical testing. Although the results of 793 statistical tests between the full dataset (with all positive amplifications regardless of the 794 assay’s LOD) and the subsetted dataset showed the same result - IMR’s DNA extracts 795 amplified the target species significantly less than the other lab’s extracts - the detection 796 rate of each lab’s DNA extracts decreased by approximately 20% (41% in the case of 797 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 24 IMR) in the subsetted dataset. The analysis of eDNA detections above or below an 798 assay’s LOD is an important factor to consider for future interpretations of eDNA 799 samples in marine ecosystems in which genomic traces are expected to be far lower 800 than the desired confidence of detection (Paul, Jeffrey and DeFlaun, 1987; Collins et al., 801 2018). On that note, almost all positive detections which fell above the assay’s LOD 802 were positively sequenced to match the target species’ DNA, proving that the above-803 LOD amplifications were true positives of the target species. However, for the purpose 804 of a cross-laboratory comparison, the dataset provided in the conservative dataset 805 (amplifications at or below the LOD), is most appropriate as stated by Klymus et al., 806 2020. Ultimately, the DNA was extracted from the same environmental sample and 807 should theoretically be the same across independent laboratories. 808 809

810 811 Overall, this study demonstrates a comprehensive effort to evaluate the consistency 812 and accuracy of DNA extraction, quantification and species detection across four 813 different laboratories. Our findings support the general reproducibility of eDNA analyses 814 across eWHALE partners, despite some disparities in one lab’s extraction performance. 815 This analysis demonstrates both the variability introduced by different target species, 816 sampling protocols and extraction methods as well as the critical importance of 817 interpreting data within the context of a given assay’s Limit of Detection. As a result of 818 this study, adjustments were made to improve IMR’s extraction protocol, thereby 819 enhancing its effectiveness. This research demonstrates the necessity of conducting 820 preliminary validation tests for research projects involving multiple laboratories, ensuring 821 reliable and comparable results. Our work represents a significant step towards the 822 successful implementation of standardized protocols, promoting consistent performance 823 across an international environmental DNA initiative. 824 825

826 827 This research was conducted within the eWHALE project funded by 1) FWF Project no. 828 I 6389 (UIBK), 2) ANR-22-EBIP-0011 (INRAE), 3) EPA Research Programme 2030; 829 Project code “2022-NE-1170 eWHALE”; Project reference “R21568” (UCC), 4) 830 Research Council of Norway via the Sharks on the Move project RCN #326879 (IMR), 831 5) M2.2/eWHALE/001/2023 via the Fundo Regional para a Ciência e Tecnologia – 832 FRCT, Governo Regional dos Açores. Porbeagle shark samples were collected under 833 permit number 708/2023 (delivered 6 June 2023). Basking shark eDNA samples were 834 collected onboard “Rind” provided by the Directorate of Fisheries. 835 836 We thank Dania Tesei, Michael Costello, Emer Rogan, Allen Whitaker, Oriol 837 Giralt Paradell, Jasmine Stavenow, Des Requins et des Hommes, Anne-Laure Besnard, 838 Loïc Baulier, Ingrid Bruvold, and Antonia Klöcker for their dedicated efforts to facilitate 839 sample collection and transportation. We thank Tanja Hanebrekke and Daniela Sint for 840 their diligent work in the laboratory. 841 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 25

842 Acevedo-Gutiérrez A, Parker N (2000) Surface Behavior of Bottlenose Dolphins Is 843 Related to Spatial Arrangement of Prey. Marine Mammal Science 16: 287–298. 844 https://doi.org/10.1111/j.1748-7692.2000.tb00925.x 845 846 Allison MJ, Warren RL, Lopez ML, Acharya-Patel N, Imbery JJ, Coombe L, Yang CL, 847 Birol I, Helbing CC (2023) Enabling robust environmental DNA assay design with 848 “unikseq” for the identification of taxon-specific regions within whole mitochondrial 849 genomes. Environmental DNA 5: 1032–1047. https://doi.org/10.1002/edn3.438 850 851 Barbosa C, Nogueira S, Gadanho M, Chaves S (2016) Chapter 7 - DNA extraction: 852 finding the most suitable method. In: Cook N, D’Agostino M, Thompson KC (Eds), 853 Molecular Microbial Diagnostic Methods. Academic Press, San Diego, 135–154. 854 https://doi.org/10.1016/B978-0-12-416999-9.00007-1 855 856 Beng KC, Corlett RT (2020) Applications of environmental DNA (eDNA) in ecology and 857 conservation: opportunities, challenges and prospects. Biodiversity and Conservation 858 29: 2089–2121. https://doi.org/10.1007/s10531-020-01980-0 859 860 Blancher P, Lefrançois E, Rimet F, Vasselon V, Argillier C, Arle J, Beja P, Boets P, 861 Boughaba J, Chauvin C, Deacon M, Duncan W, Ejdung G, Erba S, Ferrari B, Fischer H, 862 Hänfling B, Haldin M, Hering D, Hette-Tronquart N, Hiley A, Järvinen M, Jeannot B, 863 Kahlert M, Kelly M, Kleinteich J, Koyuncuoğlu S, Krenek S, Langhein-Winther S, Leese 864 F, Mann D, Marcel R, Marcheggiani S, Meissner K, Mergen P, Monnier O, Narendja F, 865 Neu D, Pinto VO, Pawlowska A, Pawlowski J, Petersen M, Poikane S, Pont D, Renevier 866 M-S, Sandoy S, Svensson J, Trobajo R, Zagyva AT, Tziortzis I, Hoorn B van der, 867 Vasquez MI, Walsh K, Weigand A, Bouchez A (2022) A strategy for successful 868 integration of DNA-based methods in aquatic monitoring. Metabarcoding and 869 Metagenomics 6: e85652. https://doi.org/10.3897/mbmg.6.85652 870 871 Bolker BM, Brooks ME, Clark CJ, Geange SW, Poulsen JR, Stevens MHH, White J-SS 872 (2009) Generalized linear mixed models: a practical guide for ecology and evolution. 873 Trends in Ecology & Evolution 24: 127–135. https://doi.org/10.1016/j.tree.2008.10.008 874 875 Bruce K, Blackman R, Bourlat SJ, Hellström AM, Bakker J, Bista I, Bohmann K, 876 Bouchez A, Brys R, Clark K, Elbrecht V, Fazi S, Fonseca V, Hänfling B, Leese F, 877 Mächler E, Mahon AR, Meissner K, Panksep K, Pawlowski J, Yáñez PS, Seymour M, 878 Thalinger B, Valentini A, Woodcock P, Traugott M, Vasselon V, Deiner K (2021) A 879 practical guide to DNA-based methods for biodiversity assessment. Advanced Books 1: 880 e68634. https://doi.org/10.3897/ab.e68634 881 882 Brunker K (2020) DNA quantification using the Qubit fluorometer. Available from: 883 https://www.protocols.io/view/dna-quantification-using-the-qubit-fluorometer-bc6vize6 884 (February 8, 2024). 885 886 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 26 Buxton A, Matechou E, Griffin J, Diana A, Griffiths RA (2021) Optimising sampling and 887 analysis protocols in environmental DNA studies. Scientific Reports 11: 11637. 888 https://doi.org/10.1038/s41598-021-91166-7 889 890 Cao H, Shockey JM (2012) Comparison of TaqMan and SYBR Green qPCR Methods 891 for Quantitative Gene Expression in Tung Tree Tissues. Journal of Agricultural and 892 Food Chemistry 60: 12296–12303. https://doi.org/10.1021/jf304690e 893 894 Collins RA, Wangensteen OS, O’Gorman EJ, Mariani S, Sims DW, Genner MJ (2018) 895 Persistence of environmental DNA in marine systems. Communications Biology 1: 1–896 11. https://doi.org/10.1038/s42003-018-0192-6 897 898 Cruaud P, Vigneron A, Fradette M-S, Charette SJ, Rodriguez MJ, Dorea CC, Culley AI 899 (2017) Open the SterivexTM casing: An easy and effective way to improve DNA 900 extraction yields. Limnology and Oceanography: Methods 15: 1015–1020. 901 https://doi.org/10.1002/lom3.10221 902 903 Dunn N, Curnick DJ, Carbone C, Carlisle AB, Chapple TK, Dowell R, Ferretti F, Jacoby 904 DMP, Schallert RJ, Steyaert M, Tickler DM, Williamson MJ, Block BA, Savolainen V 905 (2023) Environmental DNA helps reveal reef shark distribution across a remote 906 archipelago. Ecological Indicators 154: 110718. 907 https://doi.org/10.1016/j.ecolind.2023.110718 908 909 Eichmiller JJ, Miller LM, Sorensen PW (2016) Optimizing techniques to capture and 910 extract environmental DNA for detection and quantification of fish. Molecular Ecology 911 Resources 16: 56–68. https://doi.org/10.1111/1755-0998.12421 912 913 Ficetola GF, Miaud C, Pompanon F, Taberlet P (2008) Species detection using 914 environmental DNA from water samples. Biology Letters 4: 423–425. 915 https://doi.org/10.1098/rsbl.2008.0118 916 917 Garafutdinov RR, Galimova AA, Sakhabutdinova AR (2020) The influence of quality of 918 primers on the formation of primer dimers in PCR. Nucleosides, Nucleotides & Nucleic 919 Acids 39: 1251–1269. https://doi.org/10.1080/15257770.2020.1803354 920 921 García-Alegría AM, Anduro-Corona I, Pérez-Martínez CJ, Guadalupe Corella-Madueño 922 MA, Rascón-Durán ML, Astiazaran-Garcia H (2020) Quantification of DNA through the 923 NanoDrop Spectrophotometer: Methodological Validation Using Standard Reference 924

and Sprague Dawley Rat and Human DNA. International Journal of Analytical 925 Chemistry 2020: 8896738. https://doi.org/10.1155/2020/8896738 926 927 Garnier S, Ross N, Rudis B, Sciaini M, Camargo AP, Scherer C (2024) viridis: 928 Colorblind-Friendly Color Maps for R. Available from: https://cran.r-929 project.org/web/packages/viridis/index.html (March 4, 2024). 930 931 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 27 Goldberg CS, Turner CR, Deiner K, Klymus KE, Thomsen PF, Murphy MA, Spear SF, 932 McKee A, Oyler-McCance SJ, Cornman RS, Laramie MB, Mahon AR, Lance RF, Pilliod 933 DS, Strickler KM, Waits LP, Fremier AK, Takahara T, Herder JE, Taberlet P (2016) 934 Critical considerations for the application of environmental DNA methods to detect 935 aquatic species. Methods in Ecology and Evolution 7: 1299–1307. 936 https://doi.org/10.1111/2041-210X.12595 937 938 Greiner-Ferris K (2020) Testing the Potential of Environmental DNA: Genetic Monitoring 939 of Estuarine and Coastal Bottlenose Dolphin (Tursiops truncatus) Stocks in South 940 Carolina. Master’s Thesis, Coastal Carolina University, South Carolina, USA. 941 942 Gridley T, Elwen SH, Rashley G, Badenas Krakauer A, Heiler J (2017) Bottlenose 943 dolphins change their whistling characteristics in relation to vessel presence, surface 944 behavior and group composition. Proceedings of Meetings on Acoustics 27: 010030. 945 https://doi.org/10.1121/2.0000312 946 947 Alzohairy, A (2011) BioEdit: An important software for molecular biology 60-61-2. GERF 948 Bulletin of Biosciences. Available from: 949 https://www.researchgate.net/publication/258565830_BioEdit_An_important_software_f950 or_molecular_biology 951 952 Hunter ME, Dorazio RM, Butterfield JSS, Meigs-Friend G, Nico LG, Ferrante JA (2017) 953 Detection limits of quantitative and digital PCR assays and their influence in presence–954 absence surveys of environmental DNA. Molecular Ecology Resources 17: 221–229. 955 https://doi.org/10.1111/1755-0998.12619 956 957 Hymus C (2016) The Presence of Environmental DNA Within a Newly Established 958 Research Laboratory: Comparing ‘Dirty’ and ‘Clean’ Areas. Master’s Thesis, Murdoch 959 University, Perth, Australia. 960 961 Wood J (2017) Methods for the extraction, storage, amplification and sequencing of 962 DNA from environmental samples. NZES. Available from: 963 https://newzealandecology.org/nzje/3323 (March 4, 2024). 964 965 Juhel J, Marques V, Polanco Fernández A, Borrero‐Pérez GH, Mutis Martinezguerra M, 966 Valentini A, Dejean T, Manel S, Loiseau N, Velez L, Hocdé R, Letessier TB, Richards E, 967 Hadjadj F, Bessudo S, Ladino F, Albouy C, Mouillot D, Pellissier L (2021) Detection of 968 the elusive Dwarf sperm whale (Kogia sima) using environmental DNA at Malpelo island 969 (Eastern Pacific, Colombia). Ecology and Evolution 11: 2956–2962. 970 https://doi.org/10.1002/ece3.7057 971 972 Katano I, Harada K, Doi H, Souma R, Minamoto T (2017) Environmental DNA method 973 for estimating salamander distribution in headwater streams, and a comparison of water 974 sampling methods. PLOS ONE 12: e0176541. 975 https://doi.org/10.1371/journal.pone.0176541 976 977 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 28 Kavlick MF (2018) Development of a universal internal positive control. BioTechniques 978 65: 275–280. https://doi.org/10.2144/btn-2018-0034 979 980 Klymus KE, Merkes CM, Allison MJ, et al. Reporting the limits of detection and 981 quantification for environmental DNA assays. Environmental DNA. 2020; 2: 271–282. 982 https://doi.org/10.1002/edn3.29 983 984 Klymus K, Ruiz-Ramos D, Thompson N, Richter C (2021) Development and Testing of 985 Species-specific Quantitative PCR Assays for Environmental DNA Applications. Journal 986 of Visualized Experiments. https://doi.org/10.3791/61825 987 988 Knebelsberger T, Stöger I (2012) DNA Extraction, Preservation, and Amplification. In: 989 Kress WJ, Erickson DL (Eds), DNA Barcodes. Methods in Molecular Biology. Humana 990 Press, Totowa, NJ, 311–338. https://doi.org/10.1007/978-1-61779-591-6_14 991 992 Konrad CM, Gero S, Frasier T, Whitehead H (2018) Kinship influences sperm whale 993 social organization within, but generally not among, social units. Royal Society Open 994 Science 5: 180914. https://doi.org/10.1098/rsos.180914 995 996 Koressaar T, Remm M (2007) Enhancements and modifications of primer design 997 program Primer3. Bioinformatics (Oxford, England) 23: 1289–1291. 998 https://doi.org/10.1093/bioinformatics/btm091 999 1000 Kronenberger JA, Wilcox TM, Mason DH, Franklin TW, McKelvey KS, Young MK, 1001 Schwartz MK (2022) eDNAssay: A machine learning tool that accurately predicts qPCR 1002 cross-amplification. Molecular Ecology Resources 22: 2994–3005. 1003 https://doi.org/10.1111/1755-0998.13681 1004 1005 Kuhn R, Böllmann J, Krahl K, Bryant IM, Martienssen M (2017) Comparison of ten 1006 different DNA extraction procedures with respect to their suitability for environmental 1007 samples. Journal of Microbiological Methods 143: 78–86. 1008 https://doi.org/10.1016/j.mimet.2017.10.007 1009 1010 Langlois VS, Allison MJ, Bergman LC, To TA, Helbing CC (2021) The need for robust 1011 qPCR-based eDNA detection assays in environmental monitoring and species 1012 inventories. Environmental DNA 3: 519–527. https://doi.org/10.1002/edn3.164 1013 1014 Lear G, Dickie I, Banks J, Boyer S, Buckley H, Buckley T, Cruickshank R, Dopheide A, 1015 Handley K, Hermans S, Kamke J, Lee C, MacDiarmid R, Morales S, Orlovich D, 1016 Smissen R, Wood J, Holdaway R (2018) Methods for the extraction, storage, 1017 amplification and sequencing of DNA from environmental samples. New Zealand 1018 Journal of Ecology. https://doi.org/10.20417/nzjecol.42.9 1019 1020 Loeza-Quintana T, Abbott CL, Heath DD, Bernatchez L, Hanner RH (2020) Pathway to 1021 Increase Standards and Competency of eDNA Surveys (PISCeS)—Advancing 1022 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 29 collaboration and standardization efforts in the field of eDNA. Environmental DNA 2: 1023 255–260. https://doi.org/10.1002/edn3.112 1024 1025 Lynggaard C, Bertelsen MF, Jensen CV, Johnson MS, Frøslev TG, Olsen MT, 1026 Bohmann K (2022) Airborne environmental DNA for terrestrial vertebrate community 1027 monitoring. Current Biology 32: 701-707.e5. https://doi.org/10.1016/j.cub.2021.12.014 1028 1029 McKee AM, Spear SF, Pierson TW (2015) The effect of dilution and the use of a post-1030 extraction nucleic acid purification column on the accuracy, precision, and inhibition of 1031 environmental DNA samples. Biological Conservation 183: 70–76. 1032 https://doi.org/10.1016/j.biocon.2014.11.031 1033 1034 Minamoto T, Miya M, Sado T, Seino S, Doi H, Kondoh M, Nakamura K, Takahara T, 1035 Yamamoto S, Yamanaka H, Araki H, Iwasaki W, Kasai A, Masuda R, Uchii K (2021) An 1036 illustrated manual for environmental DNA research: Water sampling guidelines and 1037 experimental protocols. Environmental DNA 3: 8–13. https://doi.org/10.1002/edn3.121 1038 1039 Miya M (2022) Environmental DNA Metabarcoding: A Novel Method for Biodiversity 1040 Monitoring of Marine Fish Communities. Annual Review of Marine Science 14: 161–1041 185. https://doi.org/10.1146/annurev-marine-041421-082251 1042 1043 Morisette J, Burgiel S, Brantley K, Daniel W, Darling J, Davis J, Franklin T, Gaddis K, 1044 Hunter M, Lance R, Leskey T, Passamaneck Y, Piaggio A, Rector B, Sepulveda A, 1045 Smith M, Stepien C, Wilcox T (2021) Strategic considerations for invasive species 1046 managers in the utilization of environmental DNA (eDNA): steps for incorporating this 1047 powerful surveillance tool. Management of Biological Invasions 12: 747–775. 1048 https://doi.org/10.3391/mbi.2021.12.3.15 1049 1050 Nykänen M, Louis M, Dillane E, Alfonsi E, Berrow S, O’Brien J, Brownlow A, Covelo P, 1051 Dabin W, Deaville R, de Stephanis R, Gally F, Gauffier P, Ingram SN, Lucas T, Mirimin 1052 L, Penrose R, Rogan E, Silva MA, Simon-Bouhet B, Gaggiotti OE (2019) Fine-scale 1053 population structure and connectivity of bottlenose dolphins, Tursiops truncatus, in 1054 European waters and implications for conservation. Aquatic Conservation: Marine and 1055 Freshwater Ecosystems 29: 197–211. https://doi.org/10.1002/aqc.3139 1056 1057 Paul JH, Jeffrey WH, DeFlaun MF (1987) Dynamics of extracellular DNA in the marine 1058 environment. Applied and Environmental Microbiology 53: 170–179. 1059 https://doi.org/10.1128/aem.53.1.170-179.1987 1060 1061 Pearman JK, Keeley NB, Wood SA, Laroche O, Zaiko A, Thomson-Laing G, Biessy L, 1062 Atalah J, Pochon X (2020) Comparing sediment DNA extraction methods for assessing 1063 organic enrichment associated with marine aquaculture. PeerJ 8: e10231. 1064 https://doi.org/10.7717/peerj.10231 1065 1066 Rees HC, Maddison BC, Middleditch DJ, Patmore JRM, Gough KC (2014) REVIEW: 1067 The detection of aquatic animal species using environmental DNA – a review of eDNA 1068 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 30 as a survey tool in ecology. Journal of Applied Ecology 51: 1450–1459. 1069 https://doi.org/10.1111/1365-2664.12306 1070 1071 Rojahn J, Gleeson DM, Furlan E, Haeusler T, Bylemans J (2021) Improving the 1072 detection of rare native fish species in environmental DNA metabarcoding surveys. 1073 Aquatic Conservation: Marine and Freshwater Ecosystems 31: 990–997. 1074 https://doi.org/10.1002/aqc.3514 1075 1076 Rourke ML, Fowler AM, Hughes JM, Broadhurst MK, DiBattista JD, Fielder S, Wilkes 1077 Walburn J, Furlan EM (2022) Environmental DNA (eDNA) as a tool for assessing fish 1078 biomass: A review of approaches and future considerations for resource surveys. 1079 Environmental DNA 4: 9–33. https://doi.org/10.1002/edn3.185 1080 1081 Shane SH, Wells RS, Würsig B (1986) ECOLOGY, BEHAVIOR AND SOCIAL 1082 ORGANIZATION OF THE BOTTLENOSE DOLPHIN: A REVIEW. Marine Mammal 1083 Science 2: 34–63. https://doi.org/10.1111/j.1748-7692.1986.tb00026.x 1084 1085 Sievers F, Wilm A, Dineen D, Gibson TJ, Karplus K, Li W, Lopez R, McWilliam H, 1086 Remmert M, Söding J, Thompson JD, Higgins DG (2011) Fast, scalable generation of 1087 high‐quality protein multiple sequence alignments using Clustal Omega. Molecular 1088 Systems Biology 7: 539. https://doi.org/10.1038/msb.2011.75 1089 1090 Stoeckle MY, Mishu MD, Charlop-Powers Z (2018) GoFish: A Streamlined 1091 Environmental DNA Presence/Absence Assay for Marine Vertebrates. Ecology. preprint 1092 https://doi.org/10.1101/331322 1093 1094 Thalinger B, Deiner K, Harper LR, Rees HC, Blackman RC, Sint D, Traugott M, 1095 Goldberg CS, Bruce K (2021) A validation scale to determine the readiness of 1096 environmental DNA assays for routine species monitoring. Environmental DNA 3: 823–1097 836. https://doi.org/10.1002/edn3.189 1098 1099 Thomas AC, Nguyen PL, Howard J, Goldberg CS (2019) A self-preserving, partially 1100 biodegradable eDNA filter. Methods in Ecology and Evolution 10: 1136–1141. 1101 https://doi.org/10.1111/2041-210X.13212 1102 1103 Thomsen PF, Willerslev E (2015) Environmental DNA – An emerging tool in 1104 conservation for monitoring past and present biodiversity. Biological Conservation 183: 1105 4–18. https://doi.org/10.1016/j.biocon.2014.11.019 1106 1107 Tsuji S, Takahara T, Doi H, Shibata N, Yamanaka H (2019) The detection of aquatic 1108 macroorganisms using environmental DNA analysis—A review of methods for 1109 collection, extraction, and detection. Environmental DNA 1: 99–108. 1110 https://doi.org/10.1002/edn3.21 1111 1112 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447 31 Untergasser A, Cutcutache I, Koressaar T, Ye J, Faircloth BC, Remm M, Rozen SG 1113 (2012) Primer3—new capabilities and interfaces. Nucleic Acids Research 40: e115. 1114 https://doi.org/10.1093/nar/gks596 1115 1116 Wallinger C, Staudacher K, Sint D, Thalinger B, Oehm J, Juen A, Traugott M (2017) 1117 Evaluation of an automated protocol for efficient and reliable DNA extraction of dietary 1118 samples. Ecology and Evolution 7: 6382–6389. https://doi.org/10.1002/ece3.3197 1119 1120 Whitehead H, Gordon J, Mathews EA, Richard KR (1990) OBTAINING SKIN SAMPLES 1121 FROM LIVING SPERM WHALES. Marine Mammal Science 6: 316–326. 1122 https://doi.org/10.1111/j.1748-7692.1990.tb00361.x 1123 1124 Wickham H (2020) reshape2: Flexibly Reshape Data: A Reboot of the Reshape 1125 Package. Available from: https://cran.r-project.org/web/packages/reshape2/index.html 1126 (March 4, 2024). 1127 1128 Wickham H, François R, Henry L, Müller K, Vaughan D, Software P, PBC (2023) dplyr: 1129 A Grammar of Data Manipulation. Available from: https://cran.r-1130 project.org/web/packages/dplyr/index.html (April 21, 2024). 1131 1132 Wickham H, Chang W, Henry L, Pedersen TL, Takahashi K, Wilke C, Woo K, Yutani H, 1133 Dunnington D, Brand T van den, Posit, PBC (2024) ggplot2: Create Elegant Data 1134 Visualisations Using the Grammar of Graphics. Available from: https://cran.r-1135 project.org/web/packages/ggplot2/index.html (March 4, 2024). 1136 1137 Wilcox TM, McKelvey KS, Young MK, Jane SF, Lowe WH, Whiteley AR, Schwartz MK 1138 (2013) Robust Detection of Rare Species Using Environmental DNA: The Importance of 1139 Primer Specificity. PLOS ONE 8: e59520. https://doi.org/10.1371/journal.pone.0059520 1140 1141 Ye J, Coulouris G, Zaretskaya I, Cutcutache I, Rozen S, Madden TL (2012) Primer-1142 BLAST: A tool to design target-specific primers for polymerase chain reaction. BMC 1143 Bioinformatics 13: 134. https://doi.org/10.1186/1471-2105-13-134 1144 1145 Zhang Q, Wang J, Deng F, Yan Z, Xia Y, Wang Z, Ye J, Deng Y, Zhang Z, Qiao M, Li 1146 R, Denduluri SK, Wei Q, Zhao L, Lu S, Wang X, Tang S, Liu H, Luu HH, Haydon RC, 1147 He T-C, Jiang L (2015) TqPCR: A Touchdown qPCR Assay with Significantly Improved 1148 Detection Sensitivity and Amplification Efficiency of SYBR Green qPCR. PLOS ONE 10: 1149 e0132666. https://doi.org/10.1371/journal.pone.0132666 1150 Author-formatted, not peer-reviewed document posted on 28/05/2024. DOI:  https://doi.org/10.3897/arphapreprints.e128447