DNA sampling between the shells: evaluating non-lethal DNA sampling methods in oysters for high-throughput genotyping

With the development of cost-effective high-throughput shellfish genotyping tools, the application of genomic selection in oyster breeding is quickly becoming a reality. A critical, but laborious step in this process is the non-lethal DNA sampling of candidate broodstock for genotyping. Benignly sampling oysters is complex, requiring anesthesia to induce incomplete adductor muscle paralysis, followed by DNA sampling via biopsy, hemolymph or tissue extraction which can take significant time per individual and physically damage valuable breeding candidates. In this study genotype concordance of multiple non-lethal, and less injurious DNA sampling methods in Eastern (Crassostrea virginica) and Pacific (C. gigas), oysters were evaluated. Genotypes obtained via swabbing, eDNA collection, and mantle tissue clipping were compared to a reference adductor muscle-derived genotype, which was obtained lethally. Eastern oyster (Crassostrea virginica) samples were evaluated on a 66K Single Nucleotide Polymorphism (SNP) array and Pacific oyster (Crassostrea gigas) samples were evaluated using PCR-based genotyping-by-sequencing. Retention of glue-on tags and oyster survival were also evaluated for 27 days in Pacific oysters in the field, after non-lethal sampling. Overall SNP genotype concordance rates were high in both oyster species (> 99% in samples passing quality control) with DNA quality having little impact on genotype concordance rate. Using a microhaplotype sequence workflow in two Pacific oyster genotyping panels resulted in a higher non-concordance rate (approx. 4%), however the missing (non-called) allele was found in raw sequencing data in the vast majority of cases, suggesting that rudimentary genotype calling parameters are likely driving the higher error rate. This is not surprising given the nascent implementation of microhaplotypes in aquaculture research and ongoing panel development. Post-swab survival was high, and all animals retained physical tags for subsequent identification. Overall, non-lethal swab DNA collection provided high quality genotype data similar to mantle or adductor tissue-derived genotype data; swab sampling was also faster and easier to implement and may be less stressful. The use of non-lethal swabbing for genotyping will be an important step for implementing genomic selection and using advanced genetic tools in research supporting oyster restoration and production aquaculture.

Molluscan Shellfish aquaculture has grown steadily over the last few decades, now representing the third largest group of farmed aquatic animals after finfish and crustaceans (FAO, 2024). From 2020-2022 molluscs accounted for approximately 16% of the net global increase in aquaculture production. Much of the growth in mollusc production is driven by cupped oyster aquaculture, which was the second largest farmed aquatic species group by weight globally, producing 6.2 million tons in 2022 (FAO, 2024). As oyster aquaculture has grown, so has the complexity regarding how oysters are produced, cultured and marketed (e.g. Willer et al., 2021, Botta et al., 2020; Hollenbeck and Johnston, 2018), and selective breeding of oysters for commercially desirable traits is now an important component of continued industry growth (Allen et al., 2021; Gjedrem et al., 2012; Degremont et al., 2015; Guo and Ford, 2016; Hollenbeck and Johnston, 2018; Jiang et al., 2024).

At the forefront of selective breeding efforts is the application of genomic selection (GS) to rapidly improve commercially desirable traits in oysters and other aquaculture species (e.g. Hollenbeck and Johnston, 2018; Houston et al., 2020; Jourdan et al., 2023; Marín-Nahuelpi et al., 2025). This approach has become more common as inexpensive, high-throughput genotyping of large numbers of broodstock candidates is now possible for oyster species (e.g. Guo et al., 2023; Delomas et al., 2023a; Delomas et al., 2024; Sutherland et al., 2024; Thompson et al., 2025). A key component of this approach is non-lethal tissue sampling for genotyping to calculate a genomic estimated breeding value (GEBV) for each broodstock candidate. For bivalves, sampling requires access to the soft tissues via anesthesia that induces the animal to gape, or shell clipping followed by careful removal of a piece of gill or mantle tissue (Berg et al., 1995; Liu et al., 2021). Multiple studies in oysters and other bivalve shellfish have focused on optimal methods or tissue types to sample non-lethally (Liu et al., 2021; Mao et al., 2013; Bai et al., 2024), as well as the optimal anesthesia protocol (e.g. Suquet et al., 2009; Hintz et al., 2017; Bai et al., 2024). However, excising tissue can be challenging as gaping varies among individuals and tissue removal may be harmful, potentially causing stress in the oyster or mortality (Song et al., 2025, Leung et al., 2026). Layered upon this challenge is the large scale at which breeding programs operate, requiring sampling of hundreds, if not thousands, of animals in a short amount of time, often by less experienced personnel such as interns, resulting in injury that has the potential to significantly impact animal health and complicating efforts to develop or implement GS methods. An alternative approach to tissue clipping is the swabbing of internal tissues which collects smaller tissue fragments and cells, similar to a medical biopsy. Swab sampling may be faster to implement than excising tissue, but swab-based DNA samples have not been evaluated for their effect on DNA quality and concentration, nor on the ultimate genotype quality when sequenced.

In this study we examined the effect of multiple non-lethal sampling methods in two species of oysters on DNA quality, quantity, and genotyping success on both a single nucleotide polymporphism (SNP) array (Guo et al., 2023) and two amplicon genotyping-by-sequencing panels (Sutherland et al., 2024, Thompson et al. In prep). The study goal was to evaluate the utility of swabbing for collecting DNA from candidate oyster broodstock for genotyping to confirm pedigree and contribute to GS analyses. Genotype concordance between DNA collected from non-lethal biopsies (mantle clip or swabs) and adductor muscle (a lethal sample) from the same individual was determined to test the impact of tissue or sample type on genotype quality and repeatability, which is of great importance in GS applications, imputation, pedigree evaluation and general research.

Eastern oyster (Crassostrea virginica) tissue sampling and SNP array analysis

Tissue and non-lethal sampling

To evaluate the effect of non-lethal biopsy sampling types on DNA quality/yield and genotype concordance in Eastern oysters (Crassostrea virginica), two genotyping experiments were conducted using oysters from the Choptank River, MD USA (henceforth referred to as Eastern oyster experiment 1 and Eastern oyster experiment 2). In Eastern oyster experiment 1, three non-lethal sampling methods were tested: a mantle tissue clip from the edge of the shell (e.g. Berg et al., 1995; N = 8), a cytology brush swab of the oyster viscera (7-inch Puritan Histobrush cat # 25-2199, Puritan medical products, Guilford, ME, USA; N = 8), and collection of oyster eDNA via water samples of individual oysters housed in 1 liter beakers for 8 hours (N = 10). For the tissue or viscera sampling (cytology brush swab and mantle), oysters were relaxed with 5% magnesium sulfate (Epsom Salt) in 20-liter buckets with aerated ambient seawater (8-10 ppt) for 4-8 hours to allow oysters to gape and allow access to internal tissues (e.g. Hintz et al., 2017). Cytology brush swabs were taken by briefly rotating the brush tip against the adductor muscle and the mantle for 2-3 seconds and then the brush portion was clipped with electrical wire cutters (diagonal cutters) into 1.5 ml Eppendorf tubes prefilled with 400 ul of EZNA Tissue lysis buffer (Omega Biotek, Norcross, GA, USA) and frozen at -20 °C for subsequent DNA extraction. Wire cutters were cleaned between samples with 70% EtOH and disposable tissue (Kimwipes). Similarly, mantle tissue clips taken with dissecting micro scissors and forceps were placed directly into 1.5 mL microcentrifuge tubes with EZNA tissue lysis buffer and stored immediately at -20 °C until extraction. For eDNA sampling, one liter of beaker water was filtered through a 10% bleach-washed 500 ml magnetic filter housing (Sterlitech, Auburn WA, USA) attached to a PVC manifold on a 47mm Whatman 1-micron cellulose nitrate filter and stored immediately at -20°C until extraction. After sampling non-lethally to simulate real-world conditions of collecting tissue or DNA from live oysters (swabs, clips or eDNA), oysters were shucked and a portion of the adductor muscle was sampled with forceps and a scalpel (30-50 mg) and placed in tissue lysis buffer and frozen at -20°C. Shucking occurred within minutes following the non-lethal sampling. Adductor tissue served as the reference tissue type for comparing extraction quality, DNA yield and for estimates of genotype concordance between paired non-lethal and adductor samples (N = 26).

For Eastern oyster experiment 2, cytology brush swabs (7-inch Puritan Histobrush; 30-40 cents per swab) were compared to two less expensive swab alternatives that are widely available: a cotton tipped swab (6-inch cotton tipped swab: Item #782010, Dealmed medical supplies, Brooklyn, NY, USA; ~4 cents per swab) and a disposable nylon eye lash brush (wood disposable mascara swab, Item # 20220428000017, TbestMax, Shenzen, China; ~2 cents per swab). Sample size was N = 10 for each swab type. In experiment 2, the same viscera and adductor swabbing approach used in experiment one was implemented, but swab heads were clipped into 95% ethanol and stored for later extraction, which was implemented to be more compatible with the core genotyping facility’s standard process sampling workflow. Swabs were taken non-lethally after oysters were immersed in 5% magnesium sulfate for 4-8 hours and oysters were then shucked and adductor tissue preserved to provide paired adductor/swab samples for each individual to estimate genotype concordance (N = 30 adductor samples). Experiment two swabs and adductor tissues were shipped in ethanol to the Center for Aquaculture Technologies (CAT) in San Diego, CA USA for DNA extraction and processing.

DNA extraction

For Eastern oyster experiment 1, DNA was extracted from all tissue types (adductor, mantle, brush swabs, and filters) using the Omega Biotek EZNA Tissue DNA spin column kit (Norcross GA). For tissues (30-50 mg) and brush swabs (whole swabs), buffer was brought to room temperature and samples were digested with proteinase K at 56°C for 6 hours, after which extraction followed the manufacturer’s instructions. For extraction of the eDNA filters, filters were cut into small strips and double the volume of initial lysis buffer (400 ul) was added to account for the extra volume from the filter pieces – 50ul proteinase K was then added to the buffer for digestion. After digestion at 56 °C for 6 hours, 440 ul of BL buffer was added. Following that step, the manufacturer’s instructions for the tissue kit were followed for the remaining of the extraction. Extracted DNA was then shipped on dry ice to CAT for genotyping.

For Eastern oyster experiment 2, DNA was extracted from the ethanol preserved swab samples at CAT. Swabs were removed from ethanol, suspended in lysis buffer and then processed using a custom magnetic bead DNA extraction kit (CAT and MolGen, San Diego, CA, USA). Automated processing and liquid handling steps of the extraction protocol were performed using PurePrep 96 units (Molgen, San Diego, CA) according to kit and instrument-specific guidelines. Resulting genomic DNAs were assessed for yields and quality by spectrophotometry (Nanodrop) and 2% agarose gels with a 1Kb Hyper ladder (Meridian Bioscience, Cincinatti OH, USA). Individual samples with the majority of DNA appearing below 500 bp were considered to be ‘low molecular weight’ DNA, samples with the majority of DNA fragment sizes between 500 and 2000 bp were considered ‘medium molecular weight’, and samples with the majority of DNA fragment lengths above 3000 bp were considered ‘high molecular weight’.

Array genotyping and SNP analysis

Single nucleotide polymorphism (SNP) genotyping of Eastern oyster DNA on the 66K HD oyster array (Guo et al., 2023) was conducted at CAT, on the Affymetrix GeneTitan (Thermo Fisher Scientific, Waltham MA, USA). SNP array data were analyzed with Axiom Analysis Suite 5.0 software following recommended threshold settings (DishQC ≥ 0.82, QC call rate ≥ 97%). SNPs were retained if they were included on the ‘recommended’ SNP list generated with SNPpolisher 3.0 software (ThermoFisher). The recommended SNP list was created by applying a series of stringent quality control filters and statistical thresholds to ensure high-confidence SNP selection, including a call rate cutoff of 97%, and various thresholds for metrics such as heterozygosity and homozygosity scores, ensuring well-separated genotype clusters.

Genotyping-by-sequencing analyses of Pacific oysters (Crassostrea gigas)

DNA sampling and experimental designs

Two experiments were conducted using Pacific oysters (Crassostrea gigas) to evaluate multiple swab types. Adult Pacific oysters representative of broodstock candidates were obtained from the Pacific Shellfish Research Unit (PSRU; USDA ARS) shellfish repository in Yaquina Bay, Oregon. In both experiments, after animals were transferred onshore and before swabbing, external detritus/fouling was removed by stiff wire brushing and each shell was dried using compressed air. Paired 4 x 8 mm glue-on tags (FPN, Hallprint North America Wenatchee WA, USA) were affixed to each shell using glue (Premium Coral Frag Glue, Polyplab, Overland Park, KS, USA) for identifying individuals. After tagging, an 8% MgSO 4 relaxation treatment (Supplementary File S1) was conducted on each experimental replicate to gain access to viscera. The relaxation treatment was a modified protocol developed from Proestou et al. (2025) to increase the success rate of anesthetization (gaping) in Pacific oyster.

Pacific oyster experiment one involved 48 animals using the Puritan PurFlock Ultra (#3318-U) swabs. To obtain non-lethal samples, swabs were rotated along the viscera of the anesthetized, gaping oysters, then placed into a 1.8 mL microcentrifuge tube and rinsed with 95% ethanol. Each swab was clipped using wire cutting pliers to separate the handle from the brush. Pliers were sanitized between oysters using a 10% bleach dip and freshwater rinse. After each oyster was tagged and swab sampled, it was placed into a tank with 20-micron filtered seawater at ambient conditions for recovery. Oysters were given a 72-hour recovery period to assess if mortality occurred due to the acute stress of relaxation and swabbing procedures. For experiment 1, after the recovery period ended, each animal was shucked, and an adductor sample taken and preserved in 95% ethanol. No extended observation of survival was done with Pacific oyster experiment 1.

For Pacific oyster experiment 2, two swab types were evaluated and an extended survival study was performed to assess the longer-term effects of DNA sampling on survival. Control (oysters subjected to cleaning, tagging and relaxation, but no DNA sampling) and a mantle tissue clipping treatment (described below) were also conducted specifically for use in the extended survival portion of the experiment.

The two swabs evaluated in Pacific oyster experiment 2 were the Puritan Cytology Brush (#1008560) and a Microbrush Plus Fine Bendable Micro Applicator (#PF400, Microbrush International, Grafton WI, USA). One-hundred oysters were used for each treatment, split into two replicates of 50 oysters each. The replicates were performed in series, such that all replicate 1 treatment groups were sampled, then the second replicate treatment groups were created and sampled. Sampling procedures for the cytology and microbrush swab were identical to those employed in Pacific oyster experiment 1. Oysters in the mantle clipping treatments had a small piece of tissue excised from the oyster using forceps and scissors, which was then preserved in 95% ethanol (but not used in genotyping). After each treatment replicate was created (50 oysters per group) they were placed into a recovery bath by experimental replicate, and allowed to recover onshore for 72 hours in ambient condition seawater.

Extended survival study after DNA sampling

For Pacific oyster experiment 2, after the 72 hour recovery period the oysters were placed into 25-liter oyster baskets (SEAPA, Edwardstown SA, Australia) by experimental replicate, with all replicate 1 groups in one basket stack and all replicate 2 groups in another basket stack. Vertical position within the basket stack was randomly assigned, and the four basket stack measured 0.68m tall. The two basket stacks containing tagged oysters were transferred to the PSRU shellfish repository and held subtidally, experiencing natural conditions in Yaquina Bay from 3 July – 1August 2025. On 1 August all baskets were brought onshore to assess tag retention and survival per replicate. Oysters that lost tags were identified and survival was determined by counting live animals within each basket. To assess differences in survival chi-square tests of homogoneity were used within each basket stack comparing a DNA sampling method to the control. For example, within basket stack 1, three chi-square tests were evaluated, (1) microbrush vs. control, (2) cytology swab vs. control and (3) mantle clip vs. control. After survival and tag loss were evaluated, 24 oysters per replicate of microbrush and cytology brush treatments were randomly chosen for lethal sampling using Pacific oyster experiment 1 protocols to obtain adductor tissue samples. This resulted in 96 samples for the cytology brush and microbrush treatments in Pacific oyster experiment 2.

DNA extraction

For experiment 1, ethanol preserved swabs were shipped to CAT (San Diego, CA). A magnetic bead DNA extraction kit (CAT and Molgen) was used for extracting all samples on a single plate. For experiment 2, DNA was extracted using an Omega blood and tissue kit on a Kingfisher robot (Thermo Fisher Scientific) following manufacturer’s protocols. DNA concentration was quantified using a Nanodrop spectrophotometer (Thermo Fisher Scientific).

Library prep, sequencing and data processing

For Pacific oyster experiment 1, GBS library preparation was conducted using a 592 locus amplicon panel (Sutherland et al., 2024) and sequenced on an Ion Torrent Machine (Thermo Fisher Scientific) at CAT. Reads were processed using Torrent Suite software (Thermo Fisher Scientific) to produce genotype calls. Briefly, reads were demultiplexed by sample, then aligned to a reference genome (GCA_000297895.1; Zhang et al., 2012). Torrent Suite software (https://github.com/iontorrent/TS) was used by CAT to further processed the data and generated a SNP output file.

To investigate the utility of a microhaplotype genotyping approach, a second round of processing and analysis was conducted following methods of Thompson et al. (2025) for Pacific oyster experiment 1 data. Binary alignment map (BAM) files were produced by mapping reads to the oyster genome (Zhang et al., 2012) using Torrent Suite software (github.com/iontorrent). The Genome Analysis Toolkit (GATK; Van der Auwera and O'Connor, 2020) was used to generate variant call format (VCF) files using HaplotypeCaller, GenomicsDBImport, and GenotypeGVCFs functions with default settings. Variant call format (VCF) files were filtered to retain genotypes with GQ >= 20, depth >=5 and SNPs were filtered to eliminate SNPs with MAF<0.01 and QUAL < 99. Only bi-allelic substitutions were retained. Microhaplot (Ng, 2019) called microhaplotype alleles, which were filtered by removing any allele with depth < 5, allele balance (ratio of allele B to allele A) ≤ 0.30, and genotype depth < 20 reads. Alleles with any N or X nucleotide calls were also removed.

For Pacific oyster experiment 2, a different amplicon panel with 288 loci selectively distributed on all 10 chromosomes was used. The 288 locus panel is under active development, with this sequencing run being part of the overall evaluation of the panel utility and reliability.

For Pacific oyster experiment 2, PCR amplification followed methods of Campbell et al. (2015) and Anderson et al. (2025) using dual indexing to uniquely identify samples. Swab and adductor samples from the same oyster were amplified in the same plate, removing any potential library preparation bias. Sequencing used a 100-cycle single-end kit on an Illumina NextSeq sequencer. Reads were demultiplexed, aligned to amplicon references using bwa-mem (Li and Durbin, 2009), and variants called using GATK. Genotyping was conducted using MICROTYPER (Delomas et al., 2023b) with resulting data filtered on allele balance, read depth, genotype depth and removal of reads containing N or X nucleotides.

Calculations and statistical analyses

For all Eastern and Pacific oyster genotype datasets, genotype concordance was assessed by comparing genotypes of paired non-lethal samples (swabs or mantle clips) to adductor-derived genotypes. Genotype non-concordance (1 - rate of concordance) was estimated as the number of non-matching genotypes per locus divided by the total number of loci with called genotypes for both tissue types (i.e. excluding loci with missing values in either or both genotypes).

For non-matching genotypes in the Pacific oyster microhaplotype data, further analyses were conducted to determine what genotypes classes were involved in the mis-matching cases. The number of homozygote;heterozygote mismatches and homozygote;homozygote mismatches were calculated per individual. To investigate non-matching further, the missing allele (for homozygote;heterozygote mismatches) was identified and raw genotype call data was queried to determine if the missing allele was present or missing completely.

To statistically compare SNP genotype call rates and concordance rates among swab or tissue types, non-parametric Wilcoxon signed rank tests (2 groups) or Kruskall-Wallis tests (3 or more groups) in base R v. 4.5.2 (R Core Team, 2024) were used. In the event of a significant global Kruskall Wallis test, post-hoc Dunn tests were run from the Package FSA (Ogle et al., 2026) for pairwise comparisons among treatment groups using the “holm” method to correct for multiple pairwise comparisons.

Eastern oyster DNA sampling and SNP array results

In eastern oyster experiment one, DNA extractions from the cytology brush swabs yielded the highest DNA concentrations with an average of 423.2 ng/ul (sd=144.5), followed by adductor muscle at 77.3 ng/ul (sd=42.3), mantle at 50.8 ng/ul (sd=27.8), and eDNA at 18.9 ng/ul (sd=11). Swabs collected in experiment two (mascara, cotton and cytology brush swabs) were stored in ethanol for several weeks prior to extraction and yields were generally lower for all swab types, but the two brush swabs (cytology and mascara) generally had higher extraction concentrations. For example, nylon mascara/eyelash swabs averaged 151.3 ng/ul (sd=82.4) and cytology brush swabs averaged 121.7 ng/ul (sd=50.8), while the cotton tip swabs had lower average concentrations at 28.3 ng/ul (sd=12.8).

DNA quality across swab types as assessed by the presence of high molecular weight DNA on agarose gels for experiment two samples, was variable, with adductor samples showing the most consistently high molecular weight DNA extracts (Figure 1). For the cytology brush samples, some showed high molecular weight bands while others produced a low to medium molecular weight smear indicating lower quality (Figure 1). High molecular weight DNA was present in some samples from all swab types, but was inconsistent across samples within a swab type.

Basic genotyping results, including call rate and QC data, showed that brush swabs performed fairly well overall compared to adductor muscle samples. In terms of SNP array quality control (QC), cotton swabs had the highest rate of samples failing QC at 4/10, while only one cytology brush swab sample (of 19) failed QC. The mascara brush swabs fell in between (2/10 samples failed QC). Overall SNP call rates were high and similar between brush swabs, mantle clips, and adductor samples despite the apparent differences in molecular weight (Figure 2). Cytology brush swabs had the highest average QC call rate (98.9%, sd=0.28), followed by mascara swabs (98.6%, sd=0.6), adductor muscle tissue (98.6%, sd=0.7), mantle clips (98.3%, sd=0.3), and cotton swabs (97.9%, 1.16). The eDNA samples had the lowest call rates (average 90.2%, sd=3.2) and only four of 10 samples passed quality control and produced a call rate. A Kruskall-Wallis test revealed a significant difference in call rate among sample treatments (p=0.004), with post-hoc Dunn tests showing a significant difference in call rate between eDNA and all other treatments except for the eDNA vs mantle comparison (p=0.11). No significant difference in call rates was found between pairwise comparisons of any of the swab and tissue treatments (p>0.15). Due to the high QC fail rate and low call rate of eDNA samples, they were not examined for genotype concordance.

An average of 57, 300 SNPs were assessed for genotype concordance across all samples in the Eastern oyster experiments. Genotype concordance analysis from both experiments showed that the different swab and tissue sample types were highly concordant with adductor samples from the same individual overall (Figure 3; 0.56% mean non-concordance across all sample types). Mascara/eyelash swabs had the lowest mean non-concordance rate at 0.30%. Non-concordance for cytology brush swabs was also low (0.58%), while mantle clips produced slightly higher mean non-concordance rates of 0.82% (Figure 3). There was no significant difference in non-concordance rate among tissue/swab types (Kruskal test p=0.0538). Finally, there was no relationship between molecular weight category (low, medium, high) and non-concordance rate across samples in the round two sampling (Kruskal Wallis test, p = 0.90; Figure 4).

Pacific oyster experiment 1

In Pacific oyster experiment 1, 12 samples did not pass QC (< 70% of loci scored). One sample matched identically to a different individual’s adductor and swab samples - this sample was removed from analysis as it was a sampling or library preparation error. The number of loci scored ranged from 124-428 per sample out of 440 total loci. The Sutherland et al. (2024) panel has 592 total loci, but 152 loci consistently failed to genotype well in the service providers laboratory and were not reported. Genotype calling percentage was high (Figure 5a) for adductor and swab samples. After removing poor performing sample pairs (the 12 oysters that failed QC) no significant difference in genotype call rate existed between swab and adductor samples (Wilcoxon signed-rank test, V = 236, p = 0.82).

SNP analysis

Genotype concordance for SNPs was high between swab and adductor samples from the same individual. The only individuals having greater than 1% non-concordance were samples with low quality (i.e. samples with < 70% loci scored), which were assigned as failed samples via QC (Figure 5b). Of the genotypes that were non-concordant, all cases included a homozygote and heterozygote genotype. No cases of reference homozygote to alternate homozygote were observed. In the mismatching genotype cases there was no strong bias for a heterozygote or homozygote within a single sample type (swab or adductor). More specifically, there were 44 cases of heterozygous swab and homozygous adductor, and 40 cases of the swab being homozygous with the adductor being heterozygous.

Microhaplotype analysis

Non-concordance was slightly higher with microhaplotypes (Figure 5C) but remained low overall, ranging from 0-10.5% (average = 3.3%). The majority of non-concordant cases involved a homozygote-to-heterozygote mismatch (n= 301), and there were 2 cases of a heterozygote-to-heterozygote mismatch. In one of the heterozygote-to-heterozygote mismatches, the differing allele was present in raw sequence data but was 3 rd ranked and did not pass allele balance filtering. The other heterozygote-to-heterozygote mismatch had identical raw genotypes, but a haplotype with an N nucleotide, which was either ranked 1 st or 2 nd, and was filtered out of both swab and adductor samples because of the N base call. In the 301 homozygote-to-heterozygote mismatches, 295 cases had the missing haplotype present in the raw sequencing data, and 6 did not have the missing haplotype in raw data. Of the 295 cases, 31% were ranked 2 nd allele, 19% were ranked 3 rd allele, 12% were ranked 4 th allele, 8% were ranked 5 th allele, 8% were ranked 6 th allele, 4% were ranked 7 th allele, 8% were ranked 8 th allele, 4% ranked 9 th allele and the remaining 4% were ranked 10 th or higher. This pattern indicates that PCR amplification variability and/or data filtering criteria are likely responsible for the heterozygote to homozygote mismatch.

Pacific oyster experiment 2

Microhaplotype analysis

Extracted DNA concentrations were sufficient for PCR-based amplicon sequencing, with the microbrush swab having lower concentration and variation than the cytology swab (Figure 6A). No difference in genotyping success (number of loci scored) existed between paired adductor and cytology brush samples (Wilcoxon signed-rank test, V = 454.5, p = 0.48), but a difference did exist between microbrush samples and paired adductor tissue (Wilcoxon signed-rank test, V = 310.5, p = 0.007) with microbrush samples having 175 loci scored on average, and adductor samples having a mean of 166 loci scored. After sequence data processing and quality control, 96.9% of samples were retained with no obvious difference in the number of loci successfully genotyped by sample type (Figure 6B). Only 2 samples sequenced poorly (having fewer than 50 loci scored) and 1 adductor sample was found to not have a corresponding swab sample (human error during the experiment), resulting in 93 oysters for evaluation of swab to adductor genotypes.

Non-concordance rate ranged from 0 to 26%, but the majority of samples (all but 2) had non-concordance rates below 6.1% (Figure 6B) with average rates of 4.0% and 3.9% for the cytology and microbrush, respectively. No statistically significant difference in non-concordance rates exists between microbrush and cytology brush samples (Wilcoxon signed-rank test, W = 975, p = 0.42). Of the 563 total mis-matching genotype cases, only 10 included both genotypes being heterozygous. In these 10 heterozygote-to-heterozygote mismatches, the highest depth allele was identical between samples, while the lower depth allele differed but was present in the sequence data for the paired sample. In homozygote-to-heterozygote mismatches, the missing allele was not found in the sequence data for 16 cases, 13 of which came from a single sample that had low read depths overall. The 537 other homozygote-to-heterozygote mismatches had the missing allele identified in the sequencing data, most often being the 2 nd ranked allele (96.0%), but also being identified as the 3 rd ranked (3.4%), 4 th ranked (0.4%) or 5 th ranked allele (0.2%).

Survival and tag retention in Pacific oyster experiment 2

There was low mortality overall after the 27 day deployment in the Yaquina Bay oyster repository, with mean survival of all replicates of 93.25%. No statistical difference existed within basket stack on survival compared to control treatments (all p values > 0.05 using chi-square test of homogeneity). Control treatments survived at 100% and 84% for basket stack 1 and 2 respectively, microbrush treatments survived at 96% in stack 1 and stack 2, cytology brush survived at 94% and 86% and mantle clipping treatments experienced 100% and 90% survival in stack 1 and 2 respectively. No oysters lost both glue-on tags, and only 3 oysters lost a single tag during the 27 day period in Yaquina bay.

As genomic selection (GS) becomes increasingly applied in shellfish breeding programs to advance commercially desirable phenotypes, efficient and rapid methods for non-lethal biopsy are essential in generating the high-quality genotype data needed for accurate breeding value estimation. Previous studies have utilized mantle or gill tissue for non-lethal sampling of DNA in shellfish (e.g. Berg et al., 1995; Mao et al., 2013; Liu et al., 2021; Bai et al., 2024), but gaining access to mantle or gill tissue is difficult and care must be taken to not damage the oyster. Brush swabs provide a quicker and gentler means of collecting DNA, but this method has not been thoroughly tested for its effect on genotyping quality (e.g. genotyping concordance) in previous investigations. In this study different non-lethal and putatively less harmful sampling approaches in oysters were evaluated on how they affected DNA and genotype quality. Specifically, genotypes from swab samples were compared to genotypes from adductor tissue samples from the same individual. The adductor muscle is a standard tissue for DNA extraction in oysters, but one that requires lethal or acutely stressful sampling.

Eastern oyster genotyping experiments

For the SNP array genotyping results (eastern oyster experiments 1 and 2), the cytology brush swab performed well, generating genotype data with low non-concordance rates (mean of 0.58%) when compared to adductor muscle-derived data. The mascara swab also performed very well, and is roughly 10-fold cheaper than the cytology swab, producing a non-concordance rate of 0.30%, the lowest recorded. The non-concordance rate of the mantle clip samples was slightly higher (0.82% non-concordance) but similar, indicating that any of the three methods could be applied. However, less experienced staff are less likely to seriously injure animals when using a softer swab compared to clipping tissue using sharp implements, as swabs are easier and faster to implement than tissue excision.

Other sample types such as the cotton tipped swabs, produced less consistent genotyping results, and eDNA samples produced very low-quality data overall for the SNP array genotyping, with more than half of the samples failing QC. Environmental DNA samples had much lower yields (mean ~18.8 ng/ul DNA) compared with the other swabs or tissues, which may explain why genotyping quality was severely reduced. It is likely that the eight hours of incubation time was insufficient to accumulate enough oyster DNA for successful genotyping on the array platform. Holman et al. (2019) found that eDNA sampling of European flat oysters (Ostrea edulis) produced high genotype concordance in Fluidigm SNP genotyping assays, but oysters were incubated for 72 hours before sampling water, which likely increased the concentration of oyster DNA in the water sample. Such long residence times present substantial logistical challenge for breeding programs that require rapid sampling of large numbers of animals.

An interesting result for the swab samples run on the SNP array (Eastern oyster experiment 2) is that there was no effect of DNA molecular weight category (low, medium, or high) on the rate of genotype concordance. It was hypothesized that higher molecular weight DNA would be needed for successful genotyping and high concordance to the adductor genotypes; however, this was not observed. This finding could be explained by the fact that the probes for the HD oyster array are relatively short (~20-60 base pairs), thus low molecular weight DNA (i.e. highly fragmented DNA) may still be intact enough to produce quality genotypes. Cytology brush and mascara brush samples also produced very high yields of DNA, which may have helped increase genotype quality. Overall, the cytology and mascara brush swabs appear to represent a reliable alternative to mantle biopsy or gill clipping (e.g. Mao et al., 2013; Liu et al., 2021), providing high concentration DNA and high genotype concordance when compared to adductor tissue, while likely causing less acute stress to the animal.

Pacific oyster genotyping experiments

Genotyping-by-sequencing (GBS) for Pacific oyster samples resulted in very similar patterns of high concordance for SNP genotypes, with only samples that failed quality control having greater than 1% non-concordance. Microhaplotype genotypes displayed higher levels of non-concordance; however, most genotype mismatch cases involved a heterozygote and homozygote pair (as also happened with SNP genotypes). Given the use of microhaplotypes in shellfish research is nascent, the higher non-concordance rates are not surprising, especially considering that the Sutherland et al. (2024) panel was not designed with microhaplotype variation as a target and the USDA POGS panel is under active development. The presence of the missing allele within the raw microhaplotype sequence data is encouraging and suggests that as microhaplotype panels become more refined with improved genotype calling methods, that non-concordance rates will drop to those observed in SNP genotyping or lower.

The PCR-based amplicon panel method does not appear to be biased on genotype class, given the number of heterozygote to homozygote mismatches were approximately equal when comparing swab to adductor. This result is encouraging, as it is likely that PCR did not preferentially amplify one allele, but had a random chance of amplifying either allele within heterozygotes. Failed samples (those with less than 70% of loci called in Pacific oyster experiment 1) did not have a large difference in SNP genotype non-concordance rate compared to samples passing QC. This result is somewhat surprising as we expected a larger effect of sample quality on genotype concordance, but GBS appears robust to variability in sample quality. A similar result was found with the USDA POGS microhaplotype panel, with no strong relationship between non-concordance rate and the number of loci evaluated. These are significant findings because, if swabbing for non-lethal sampling occurs for a high number of animals in a short time period, any samples with poorer genotyping calling rates can be treated as reliable and not be considered more error prone. Overall, any of the three brush types used in the Pacific oyster experiments are viable options for non-lethally sampling DNA from oysters. No statistical difference in genotype non-concordance existed between cytology and microbrushes, such that the choice of which brush to use will likely be predicated upon cost, brush availability and user preference. The oyster tagging protocols were robust with only a small percentage of animals losing tags during the offshore deployment, and non-lethal sampling produced low mortality overall, similar to Leung et al. (2026). While this represents a relatively short amount of time, potential broodstock candidates in a genomic selection program are likely to be selected and brought onshore for artificially-induced maturation (i.e. broodstock conditioning) within a few months from the time of swab sampling. Thus, the high survival and tag retention observed over the first month is promising, and suggests the swabbing and tagging methods implemented are scalable, and high-throughput application in a breeding program or large-scale research study is feasible and would produce high quality genotype data equivalent to adductor tissue sampling.

A challenging facet of implementing GS in shellfish aquaculture is obtaining non-lethal DNA samples for genotyping due to the inaccessible nature of oyster viscera. This study determined that multiple swab types can be used to generate high quality DNA for sequencing and genotyping on multiple, commonly used platforms (array and genotype-by-sequencing). The high post-sampling survival rates, genotype concordance rates, and tag retention suggest that swab sampling can be used as an effective, high-throughput and relatively benign method for obtaining DNA necessary to generate genomic estimated breeding values. The less-injurious nature of swabs compared to tissue excision is amenable to less-experienced staff or volunteers participating in large-scale DNA capture efforts that are often undertaken in breeding program operations. Overall, the use of swab DNA collection provides a viable, scalable and robust method for furthering the application of genomic selection in shellfish breeding.