Development of a species-specific qPCR assay for the detection of the threatened dugong (Dugong dugon)

The global population of dugongs (Dugong dugon) has declined by 20% over the last ninety years, with some regional populations facing total extirpation. Effective conservation depends on the ability to reliably monitor population trends and habitat use through time. Hence, there is a growing need for the development of reliable, scalable, cost-effective and non-invasive monitoring techniques that could inform dugong conservation. To this end, we developed a species-specific, non-probe-based qPCR assay from the 16S rRNA mitochondrial gene region to enable the detection of D. dugon DNA in environmental samples. The assay was validated both in silico and in vitro using non-target sympatric species. Synthetic DNA was employed to evaluate assay sensitivity and potential inhibition effects. A total of 40 samples, including 28 water and 11 seagrass roller samples from the Saudi Arabian Red Sea coast (n=39), as well as one water sample from dugong exhibit at SeaWorld Abu Dhabi (n=1) were tested using this newly developed assay. Our results showed that the assay is specific to D. dugon and analytically sensitive under controlled conditions, with a modelled limit of detection of 2.88 copies/reaction and a limit of quantification of 66 copies/reaction. In field applications, D. dugon was detected in water samples from the aquarium and two field sites, while no amplification was observed in the seagrass roller samples. This qPCR assay enables targeted detection of D. dugon, which will improve eDNA biomonitoring and conservation efforts by enhancing understanding of the species’ distribution.

The dugong (Dugong dugon) is a herbivorous marine mammal distributed across the Indo-Pacific region and the only extant member of its family. At a global scale, dugongs are listed as Vulnerable on the IUCN Red List of Threatened Species due to declining numbers across most of their natural range (Marsh and Sobtzick, 2019). However, regional assessments have identified subpopulations that are Critically Endangered (e.g., Nansei, Japan and East African coast) and Endangered (e.g., New Caledonia subpopulation) (Marsh et al., 2025). Population declines can be attributed to anthropogenic pressures, including habitat loss, marine pollution, bycatch mortality, alongside life history traits like low fecundity (Poommouang et al., 2021).

The Red Sea is home to the third largest population of dugongs (Poommouang et al., 2021), but little is known about the distribution and population ecology of dugongs within this basin (Marsh et al., 2011; Nasr et al., 2018). Most of the current knowledge comes from field observations (aerial, surface and underwater) (Preen, 2004; Nasr et al., 2018), photo identification (Shawky et al., 2017; Newman et al., 2025), analysis of feeding trails (Shawky, 2019; Khamis et al., 2022), and fishermen interviews (Nasr et al., 2018). Due to their timid nature and limited time spent near the surface (Chilvers et al., 2004), dugong field observations can be difficult, inefficient, and prone to non-detection error. The use of complementary non-invasive biomonitoring approaches could overcome some of these challenges by providing a reliable and cost-effective alternative for detecting dugong.

Environmental DNA (eDNA) techniques have revolutionized biodiversity tracking in various ecosystems, where traditional monitoring is inappropriate or ineffective due to invasive sampling or the difficulty of physical identification of species through visual surveys, especially when assessing elusive or rare species (Thomsen and Willerslev, 2015). One widely used eDNA-based biomonitoring approach are species-specific quantitative polymerase chain reaction (qPCR) assays, which are designed to amplify only the target taxon, therefore improving specificity and reducing false positives from non-target organisms (Andruszkiewicz et al., 2020). While probe-based qPCR assays can increase specificity and have been applied increasingly for monitoring a range of marine and invasive taxa in recent years (Andruszkiewicz et al., 2020; LeBlanc et al., 2020; Cook et al., 2025), they are also more expensive than non-probe-based qPCR assays (Piaggio et al., 2014; Piggott, 2016; Xia et al., 2018; Robinson et al., 2019), which may provide a more cost-effective alternative and enable broader use in resource-limited and remote regions.

Species-specific qPCR assays have been developed for manatees (Hunter et al., 2018), dugongs’ closest living relatives (Marsh, 2014); however, to the best of our knowledge, no assay targeting D. dugon has been published. This study aims to (1) develop a non-probe-based qPCR assay specific to D. dugon, (2) evaluate its specificity and sensitivity through both in silico and in vitro validation, and (3) test the assay using field samples collected along the Saudi Arabian Red Sea coastline.

Assay development

Three mitochondrial regions were identified as potential candidates for qPCR development: the control region, 12S rRNA, and 16S rRNA. These regions were chosen because they are well represented in GenBank and/or display relatively high variability among species. All partial and complete DNA sequences for these three regions were retrieved for D. dugon.

A total of eight candidate primer sets were designed in silico, including four from the 12S rRNA, three from the 16S rRNA, and one from the control region (Supplementary Table 1). The candidate primer set from the control region was developed by aligning 568 available D. dugon sequences using the Muscle Alignment extension in Geneious Prime (version 2025.1.3; https://www.geneious.com). This assay was designed using Primer3 in Geneious Prime. For the remaining seven sets we targeted primer sites within the hypervariable regions of the mitochondrial 12S and 16S rRNA genes, which are known to exhibit high polymorphism across vertebrates (Valsecchi et al., 2020). Single nucleotide polymorphisms (SNPs) and indels were interrogated by performing a multiple sequence alignment of the D. dugong reference (GenBank accession number NC_003314) with 103 marine mammal mitogenomes, including 4 Sirenians, 31 Pinnipeds, 53 Cetacean Odontocetes and 15 Cetacean Mysticetes species (Supplementary Table 2). Putative primers were filtered for species specificity by requiring ≥5 variant positions within the annealing region, with stringent selection for at least one 3’-terminal discriminatory mismatch. Candidate primer sets developed from these two regions included either two (forward and reverse) dugong-specific primers (primer set combination Ddu_12S_04) or one dugong-specific primer (forward or reverse) coupled to one of the universal marine-vertebrate (reverse or forward) primer described by Valsecchi, Bylemans (Valsecchi et al., 2020) in the remaining cases (see Supplementary Table 1). Therefore, taxonomic discrimination in the tested assays is primarily conferred by the dugong-specific primer. This design was considered sufficient for two reasons: (i) extant sirenian species are allopatric (non-overlapping distributions), and (ii) the dugong-specific primers exhibited substantial sequence divergence from those of other marine mammals. Specificity was further evaluated using in silico analysis, as described in the following section. To accommodate the highly degraded nature of environmental DNA in aquatic systems (Bohmann et al., 2014), we designed candidate primer sets that produce short amplicons, ranging from 87 to 232 bp.

In silico assessment

In silico validation of primer sets, including hairpins, heterodimers, and melting temperature, was performed using OligoAnalyzer (Owczarzy et al., 2008). To evaluate primer specificity, we retrieved 499 16S, 420 12S, and 5,883 control region sequences from the NCBI GenBank database, representing dugong, closely related taxa (i.e., manatees (Trichechus manatus, Trichechus inunguis, Trichechus senegalensis) and the extinct Steller’s Sea cow (Hydrodamalis gigas), sympatric marine mammals, and some potential contaminant species. The list of species used in this test can be found in the (Supplementary Table 3). The search_pcr function of USEARCH (Edgar, 2010) was employed to test for sequences that would bind the primers with four or fewer mismatches.

Assay optimization

qPCR assays were performed using SYBR Green intercalating chemistry, with amplification specificity assessed by melt curve analysis. Synthetic DNA (sDNA) for each gene region was designed for assay optimization and use as a positive control (sDNA sequences are presented in Supplementary Data Sheet 1). All sDNA were based on dugong sequences and included a stretch of 4–5 bp that was altered to facilitate the identification of putative sDNA contamination of samples. The synthetic oligonucleotides were ordered as gBlocks DNA fragments from Integrated DNA technologies (IDT, USA) and resuspended according to manufacturer instructions, resulting in a starting concentration of 10 ng/μL.

A series of annealing temperatures, along with varying primer concentrations, was tested to determine optimal assay conditions. Testing was conducted using an estimated sDNA input of approximately 35,000 copies/μL (12S and 16S rRNA assays) and 47,000 copies/μL (control region assay). Copy number was calculated from DNA mass using the conversion (C) x (M) x (1x10 −15 mol/fmol) x (Avogadro’s number) = copies/µL, where C (current concentration) = 10 ng/µL and M (molecular weight) = 5.85, 5.91, and 7.94 fmol/ng for the 12S, 16S, and control region fragments, respectively. Primer concentration was optimized by evaluating four concentrations (200, 400, 800, and 1000 nM) in quintuplicate reactions including a non-template control (NTC). To confirm annealing temperature, a gradient qPCR was conducted in quintuplicate including an NTC across a range of 50 °C to 60 °C, with 2 °C intervals.

The conditions that generated the lowest cycle threshold (CT) value, showed no amplification in negative controls, and displayed the optimal amplification curve were selected as optimal and applied for the remainder of the study.

Limit of detection and limit of quantification

For sensitivity evaluation, a standard curve sDNA was generated using seven to 11 replicates of a seven-step, 10-fold dilution series of sDNA (~0.71 copies to 711,800 copies per reaction); additional replicates at the lower concentrations were included to improve sensitivity estimates. The limits of detection (LOD) and quantification (LOQ) were defined following the criteria described by (Klymus et al., 2019). Specifically, the LOD was considered the lowest DNA standard concentration yielding positive amplification in at least 95% of replicates, while LOQ was defined as the lowest concentration that could be reliably quantified with a coefficient of variation (CV) below 35%. These thresholds were determined by analyzing replicate standard curve data in RStudio V 4.3.0 using the script provided by (Klymus et al., 2019), with all settings retained at their default values.

In vitro specificity validation

For inclusivity testing, we used multiple lines of evidence, including sDNA, DNA extracted from D. dugon tissue samples, and eDNA extracted from water in the SeaWorld Abu Dhabi dugong exhibit (for sampling and DNA extraction methods see “ In vivo assay validation using eDNA samples” section). Genomic DNA from a subset of sympatric marine mammal species were analyzed for exclusivity testing, including Grampus griseus (Risso’s dolphin), Stenella longirostris (Spinner dolphin), Sousa plumbea (Indian Ocean humpback dolphin), Stenella attenuata (Pantropical spotted dolphin), Tursiops aduncus (Indo-Pacific bottlenose dolphin), and Megaptera novaeangliae (Humpback whale). These non-target DNA extractions were obtained from the Reef Ecology Lab (REL) DNA collection, which were all sourced from Red Sea specimen. Additional sirenian and sympatric marine mammal species as listed in Supplementary Table 3 were assessed only in silico because tissue-derived DNA was not available. All qPCR reactions were performed in duplicate on DNA extracts, with the addition of positive and negative controls.

The identity of positive PCR amplicons obtained from sDNA, tissue, and eDNA samples used for validation was verified by Sanger sequencing. For each sample that showed amplification, the PCR replicate with the lowest CT value was submitted to the Australian Genome Research Facility (AGRF; Perth) for bi-directional sequencing. The resulting sequencing chromatograms were aligned in Geneious Prime, visually inspected for quality control, trimmed where possible, and then compared with publicly available reference sequences for assignment (Positive sequences are presented in Supplementary Data Sheet 1).

Inhibition test

To test for potential inhibition caused by the eDNA matrix, sDNA at 71.8 copies/reaction was mixed with either neat or diluted eDNA (1/10) from water and roller samples. To prepare these eDNA matrices three independent samples of each type were mixed at equal volumes creating two eDNA matrices (water matrix and roller matrix). All qPCRs were done in triplicate, using the conditions determined through the optimization testing. Samples used as eDNA matrices were confirmed not to contain target DNA through both qPCR (this study) and metabarcoding (unpublished results).

In vivo assay validation using eDNA samples

A total of 40 eDNA samples were collected from 28 sites across the Red Sea (including, 28 seawater, and 11 seagrass roller samples) (Figure 1; Supplementary Table 4) and from the SeaWorld Abu Dhabi dugong exhibit (1 water sample). At each site, 10 L of water was collected using 5 L Niskin bottles at a depth of 1 m and transferred to decontaminated plastic buckets (10% bleach, 90% ethanol, followed by multiple rinses with site water). Three replicates per site were collected by filtering 2 L of the water immediately using a Masterflex drill pump and 1.2 µm cellulose acetate filter cartridges (Interlab). Similar methods were followed at SeaWorld Abu Dhabi, except that surface water was collected directly into decontaminated buckets. Filters were preserved and sealed using a syringe filled with 400 µL DNA Shield (ZYMO Research). Additional sampling was undertaken at sites associated with seagrass meadows. Dugongs are seagrass grazers, and therefore the grazed surfaces may retain their DNA. Consequently, sampling the surface of seagrass meadows could provide valuable insight into whether water samples or feeding-associated surfaces yield a stronger eDNA signal. To this end, after water collection at each site, sterile mini microfiber rollers (50mm, 4mm nap) (3 replicates) were used to swab the seagrass surface via snorkeling or using an extendable paint roller handle from the boat. Rollers were immediately stored in 25 mL conical tubes and preserved with 5 mL of MagMax Microbiome Lysis buffer (Applied Biosystems, ThermoFisher). Due to field conditions, including high temperature, samples were stored in a cooler until they reached the lab, where they were stored at room temperature. DNA was extracted from filters and rollers using the MagMax Microbiome Ultra Nucleic Acid Isolation Kit (Applied Biosystems, ThermoFisher) on the KingFisher Flex platform (ThermoFisher, USA), as previously described by Jarman, Alexander (Jarman et al., 2024). Laboratory extraction controls were run alongside samples, with only the extraction reagents included. Amplification of the field samples by qPCR was performed using 25 µl reactions with 5 µL of template using the reaction and cycling parameters of the optimized assay. Reactions were conducted in triplicate, and all experiments included a negative qPCR control (no template DNA added) and a positive control of sDNA. All eDNA samples were tested neat and 1:10 dilution. The fragment size of the positive qPCR products was determined by using a QIAxcel Advanced System (Qiagen, Hilden, Germany).

Assay design, optimization, and validation

Out of eight candidate primer sets initially designed, the top five were selected based on the in silico test results; primers with a high likelihood of forming self- or heterodimers, or binding to non-target species, were excluded from further analysis (for details see Supplementary Table 1). In vitro validation and optimization of the top five primer sets led to the selection of the optimal assay herein called Ddu_16S_07 (MarVer3F/DduMV3R), based on lowest CT, lowest variation in replicate efficiency, and no amplification observed in negative control samples. This assay successfully detects the target D. dugon and amplifies an 87 bp fragment of the 16S rRNA mitochondrial gene region (Table 1).

In silico primer specificity testing showed no binding of the assay to non-target species with up to four mismatches. At five mismatches, the assay was predicted to bind to Hydrodamalis gigas in addition to the target species. Because H. gigas is extinct and the high number of mismatches would be expected to reduce primer efficiency (Stadhouders et al., 2010), the assay was considered specific to dugongs. Empirical exclusivity was demonstrated for the six cetacean species tested in vitro, whereas exclusivity with respect to the closest sirenian relatives and other potential sympatric marine species was supported only by in silico analysis due to the unavailability of tissue-derived DNA for those species.

Assay optimization resulted in a 25 μL qPCR assay containing the following: 2 mM MgCl 2 (Applied Biosystems, USA), 1x PCR Gold buffer (Applied Biosystems), 0.25mM dNTPs (Astral Scientific, Australia), 0.4mg/mL bovine serum albumin (Fisher Biotec, Australia), 800 nM forward and reverse primers (IDT, Australia), 0.6μL of a 1:10,000 solution of SYBR Green dye (Life Technologies; USA), and 1U AmpliTaq Gold DNA polymerase (Applied Biosystems), and 5 μL DNA template. All qPCRs were performed on either StepOnePlus or QuantStudio Real-Time PCR instruments with the following cycling conditions: 95 °C for 5 min, followed by 50 cycles of: 95 °C for 30 s, 56 °C for 30 s, 72 °C for 45 s, melt curve and finishing with a final extension stage at 72 °C for 10 min. We observed no amplification in any non-target species when using the species-specific assay, confirming its specificity.

Assay LOD and LOQ

Following the analysis of the standard curve using the modeled calculation of Klymus, Merkes (Klymus et al., 2019), the assay showed strong analytical performance, with high standard-curve linearity (R 2 = 0.985) and an amplification slope of -3.28 (Figure 2), translating to an efficiency of 101.78%, which is within the acceptable range (i.e., 90 to 110%) for qPCR reactions (Rogers-Broadway and Karteris, 2015). The lowest standard amplification was 0.71 copies/reaction, where 7 out of 11 replicates successfully amplified. The modelled LOD was estimated at 2.88 copies/reaction, and the LOQ at 66 copies/reaction (as determined by decay modeling using a CV of 0.35). These values were derived from pooled results across two independent runs to increase the number of observations near the detection threshold. This step was justified by low inter-assay Coefficient of Variation (%CV = 2.18%) based on CT values across independent runs, confirming that instrument and operator differences were minor and did not introduce technical bias (Ahmed et al., 2016; Cui et al., 2018).

Inhibition test results

When the neat eDNA matrices were spiked with 71.8 copies/reaction to test for inhibition only two replicates out of the three amplified no matter the sample type. Once the eDNA matrix was diluted 1/10 all replicates amplified indicating that all sample types have inhibition and it is best practice to test all eDNA samples as neat and 1/10. There was also a delay of 3 cycles when the sDNA was mixed with the eDNA matrix. ΔCT or CT shift of ≥ 3 cycles of the synthetic controls spiked in the eDNA matrices from the sDNA alone reactions indicates the presence of inhibition (Hartman et al., 2005).

Environmental DNA samples

Of the 40 eDNA samples (29 water samples, including 28 Red Sea samples and one SeaWorld sample, as well as 11 seagrass rollers) tested with the Ddu_16S_07 assay, two field water samples produced amplification. In both cases, amplification occurred only in the neat extracts, with no amplification observed in the 1:10 dilutions. Each sample amplified in two out of three technical replicates, with CT values ranging between 27.46 to 34.28, and 27.16 to 32.94, respectively. Melt curve analysis yielded peak temperatures of 79.41 °C and 78.52 °C, consistent with the expected product. Both water samples with amplification were paired with seagrass rollers, which did not result in amplification.

In contrast, the dugong exhibit water sample amplified in all neat technical replicates, with CT values between 28.71 and 33.22, and melt curve of 76.51 °C. Minor variation in melt temperatures was observed between qPCR runs and is likely reflecting inter-run and instrument-related differences. Within each run, melt temperatures of field and aquarium samples were consistent with their respective positive controls. In addition to their melt-curve profiles, amplified products were further verified by product size evaluation through QIAxcel (Qiagen, Germany) analysis, and Sanger sequencing, supporting the utility of the primer set for detecting the target species in eDNA samples. No amplification was observed in roller samples or negative-template controls, while all positive controls amplified successfully.

Dugong populations have been decreasing across most of their natural range (Marsh and Sobtzick, 2019). While they are well monitored via aerial surveys in parts of north-eastern Australia due to the spatial extent of their range and relatively large population (Marsh et al., 2025), there is not much information available about their distribution and population trends in other regions, such as the Red Sea (Nasr et al., 2018). Given the rise of anthropogenic pressures, such as climate change, pollution, habitat destruction, and intensive commercial fishing, which has subsequent impacts on marine fauna (O’Hara et al., 2024), it is more critical than ever to find scalable solutions for the biomonitoring of vulnerable species.

In recent years, targeted qPCR-based eDNA detection has become one of the most accessible and cost-effective methods that has been successfully applied in the biomonitoring of marine species (Hunter et al., 2018; Knudsen et al., 2019; Andruszkiewicz et al., 2020; Valsecchi et al., 2022; Boldrocchi et al., 2024). This growing body of work highlights the potential of eDNA to overcome many of the logistical challenges associated with traditional survey methods (Thomsen and Willerslev, 2015; Pawlowski et al., 2021; De Brauwer et al., 2022), particularly for elusive or cryptic species such as marine mammals (Boldrocchi et al., 2024; Ushio et al., 2025). However, despite these advances, the development of robust, species-specific assays remains limited for several taxa, and assay performance can vary substantially depending on target region and primer design (Langlois et al., 2021; Thalinger et al., 2021).

In this context, our study contributes to ongoing efforts by developing and validating a species-specific, non-probe-based qPCR assay targeting an 87 bp fragment of the D. dugon 16S rRNA region, which addresses a key gap in molecular monitoring tools for this species. Although sequence variation was present within the amplicon, in silico analysis indicated that the primer-binding regions were conserved across the dugong sequences examined. However, as the geographic origin of most sequences could not be confirmed through GenBank records, the extent to which these results capture the full haplotype diversity across the species’ range remains uncertain.

While the mitochondrial control region is widely used for dugong population genetic studies due to its high variability (Plön et al., 2019), only a single candidate primer set targeting this region was evaluated in our assay development. During in vitro testing, this control region assay performed less robustly, showing higher CT values and greater variability among replicates. In contrast, the 16S rRNA assay provided more consistent and reproducible amplification across replicates, likely reflecting the more conserved nature of its primer-binding sites. Within the set of assays tested here, this made the 16S rRNA region the most effective target for dugong detection.

Our assay demonstrated high specificity and analytical sensitivity, with a modeled LOD of 2.88 copies/reaction and a modeled LOQ of 66 copies/reaction. To contextualize the analytical performance of our dugong-specific assay, we compared its capabilities with a validated probe-based genus-specific eDNA assay developed for the three vulnerable manatee species (Trichechus spp.) developed by Hunter, Meigs-Friend (Hunter et al., 2018), which targets a 69 bp fragment of the mitochondrial cytochrome b gene. Our assay demonstrated comparable analytical performance, with an efficiency of 101.78% (R 2 = 0.985), which aligns closely with the manatee assay’s efficiency of 95.25% (R 2 = 0.993). While the manatee qPCR assay reported an LOD of 13 molecules µl -1 (approximately 52 copies/reaction given their 4 µl template volume), our modeled calculation yielded a lower 95% LOD of 2.88 copies/reaction. However, it must be noted that our study employed a curve-fitting modelling approach (Klymus et al., 2019) to determine the LOD, whereas the manatee assay employed a discrete threshold approach where LOD values are restricted to the specific standard concentrations included in the tested dilution series (Klymus et al., 2019). LOQ was reported for the ddPCR platform in the manatee study, but not for qPCR; therefore, LOQ could not be directly compared between the two assays. Overall, these metrics suggest that our assay performs comparably to the validated manatee assay and support its suitability for sensitive dugong detection.

In field applications, D. dugon was detected in, two complex environmental water samples, despite the limitations of validation in areas where the target distribution is patchy and largely unknown. This limited field detection likely reflects low and spatially heterogenous eDNA concentrations, as well as potential effects of DNA degradation and inhibition. However, it should be noted that filtering larger volumes of water and sampling at greater depths, where the animals are more active, may increase the probability of detecting rare and elusive species such as dugongs (Faure et al., 2023).

The detection of dugong DNA in water samples but not in seagrass roller samples from the same sites, most likely reflects substrate-specific differences in how eDNA accumulates, persists, and is recovered (Simpson et al., 2023; Dawkins et al., 2024). Although seagrass surfaces may retain dugong-derived DNA during feeding, the signal may not have been present in the exact sampling area and may have been diluted by background material (i.e., plant material and surface debris), leading to non-detection in roller samples despite positive in-water detection. Detection could potentially be improved by increasing the number of roller replicates, sampling a broader area of the seagrass bed, or targeting areas with signs of feeding activity.

To enable accurate biomonitoring application and interpretation, targeted eDNA assays require rigorous validation spanning in silico, in vitro, and in vivo evaluations (De Brauwer et al., 2022). Thalinger, Deiner (Thalinger et al., 2021) proposed a framework that provides a standardized approach for reporting eDNA assay validation, with categories spanning from Level 1 (Incomplete) to Level 5 (Operational). Based on demonstrated assay sensitivity, along with expanded in vitro and field-based testing, our assay meets the criteria for Level 4 (Substantial), indicating strong performance under field conditions. Achieving Level 5 (Operational) would require additional steps, including statistical estimation of detection probabilities, assessment of environmental variables affecting assay performance, and exhaustive evaluation of assay specificity, which are beyond the scope of the current study.

Nevertheless, our findings highlight the utility of this assay for non-invasive detection of D. dugon, while also indicating that detection success is strongly influenced by sampling design. Future applications could benefit from optimized sampling strategies that incorporate factors such as filtered water volume, sampling depth, and ecological context (e.g., areas of recent feeding activity) to improve detection probabilities for this elusive species.

The development of a species-specific eDNA detection assay for D. dugon represents an important step forward in scaling up biomonitoring of this threatened species. A method that allows rapid and reliable detection of D. dugon is especially valuable in areas where information on its distribution remains limited. Widespread adoption of this approach could enable more proactive, evidence-based conservation by providing the timely information needed to detect population change, guide management interventions, and evaluate their effectiveness across the species’ range.