Tides Leave Little Trace on Fish eDNA Signals in Japanese Coastal Waters

Environmental DNA has become one of the most powerful tools in modern marine ecology. Instead of chasing fish with nets or waiting for divers to spot them, scientists can now simply collect a bottle of seawater, extract the genetic fragments that organisms continuously shed into their surroundings, and read out a snapshot of the entire community. A new study published in Metabarcoding and Metagenomics tested whether one of the ocean’s most obvious sources of noise — the daily rise and fall of the tide — actually distorts that snapshot. The answer, drawn from two very different Japanese coastlines separated by ten degrees of latitude, is reassuring for practitioners: tides matter far less than where you sample, even when the water is moving by one and a half to two metres.

The research team, led by Masaki Miya and colleagues, focused on two peninsulas with sharply contrasting marine faunas. The temperate Boso Peninsula, at 35 degrees north on the Pacific side of Honshu, hosts rocky shores swept by the northward-flowing Kuroshio Current. The tropical Motobu Peninsula, at 25 degrees north on Okinawa-jima Island, faces the East China Sea and is ringed by coral reef lagoons. These two regions were deliberately chosen because prior work on tidal effects on eDNA had been concentrated in high-latitude fjords and estuaries, leaving lower-latitude, species-rich environments largely unexplored. Whether tidal currents scramble eDNA signals in warm, biodiverse waters was an open question with real consequences for how coastal monitoring programs are designed.

The sampling design was tightly choreographed around the spring tide phase of 16 and 17 May 2022, when daily tidal height differences reached approximately 1.5 to 2 metres. At each peninsula, three sites within a 20-kilometre radius were visited over two consecutive days, with four rounds of daylight sampling per site timed to capture both outgoing and incoming tides. Surface seawater was collected with sterilised buckets, transferred into sterile plastic bags, and preserved immediately with benzalkonium chloride to halt DNA degradation. Three 1000-millilitre replicates were collected per event, yielding 72 samples in total. Water temperature and salinity were logged on-site with a portable meter, and samples were transported in cooling boxes with ice packs before being filtered the same day through Sterivex cartridges with a nominal pore size of 0.45 micrometres.

Laboratory work followed the MiFish pipeline, a well-established protocol for fish eDNA metabarcoding. DNA was extracted from the filter membranes using a DNeasy Blood & Tissue kit after proteinase K lysis, and a two-step PCR approach was used to prepare paired-end libraries for sequencing on an Illumina MiSeq platform. A mixture of three primer pairs targeting a hypervariable region of the mitochondrial 12S rRNA gene — approximately 172 base pairs, known as the MiFish sequence — was applied in a 0.9:1.0:1.0 ratio. Eight technical replicates per template minimised PCR dropouts, and a series of field, extraction, and PCR blanks was carried through the entire workflow to detect contamination. The eight blank libraries yielded no reads after denoising, confirming that the contamination controls held firm.

Sequencing performance was exceptional. Two MiSeq runs produced 8,594,144 reads, with 96.6% of base calls reaching Phred quality scores of 30 or above — far exceeding Illumina’s own guideline of 80% for this read length. After merging paired-end reads, quality filtering, dereplication, and denoising to generate amplicon sequence variants, 7,335,844 reads (85.4% of the initial total) survived for taxonomic assignment. Of these, 99.5% were identified as fish sequences using MiFish DB ver. 43, a reference database encompassing 8,523 species across 489 families. Sequences with less than 80% identity — mostly from terrestrial mammals such as cows, pigs, and humans — were excluded as non-fish. After removing putatively chimeric sequences and ostensibly exogenous detections from deep-sea or freshwater species, the final dataset comprised 4,446 detections assigned to 595 species, 295 genera, and 92 families.

The between-region results were striking. Species richness per sample averaged 40.4 on the Boso Peninsula but 83.1 on the Motobu Peninsula, reflecting the well-known latitudinal gradient in marine biodiversity. Only 32 of the 595 species were common to both peninsulas, including widely distributed Kuroshio-region fish such as Kyphosus bigibbus and Mugil cephalus cephalus. Non-metric multidimensional scaling ordination based on Jaccard indices clearly separated the two regions in two-dimensional space. On the Boso Peninsula, six temperate reef or sandy-bottom species — including Acanthopagrus schlegelii and Takifugu alboplumbeus — were detected in all 36 samples. On the Motobu Peninsula, no single species appeared in every sample, but coral reef-associated fish such as Siganus spinus and Chrysiptera cyanea dominated the most frequent detections, underscoring the ecological distinctness of the two faunas.

Within each peninsula, the three sites also turned out to be far from interchangeable. Species richness across sites ranged from 67 to 120 on Boso and from 219 to 324 on Motobu, with statistically significant differences among sites on both peninsulas (Kruskal-Wallis test, p < 0.0001). The intermediate sites, BP2 and MP2, consistently recorded the lowest totals and the lowest numbers of unique species, while the third site on each peninsula — BP3 in Tateyama Bay and MP3 near a shorter shore-to-reef-edge distance — showed markedly higher richness. At BP3, tropical species such as Parascorpaena mossambica and Pomacentrus coelestis were detected exclusively, likely because an offshoot of the Kuroshio Current flows directly into the bay, carrying larvae and juveniles of southern species northward. At MP3, the close proximity of both reef-edge and shore-side habitats within the lagoon may have boosted diversity. These findings highlight that even sites within 10 to 15 kilometres of each other can host meaningfully different fish communities.

When the team compared outgoing and incoming tides at each site, statistically significant differences in community composition emerged at five of the six sites (p < 0.05), with only BP1 showing no significant tidal effect. Indicator species analysis identified 10 species linked to specific tidal phases — five for outgoing and five for incoming tides — primarily on the Motobu Peninsula, though these species showed no clear ecological pattern. Species richness differed significantly at two sites (BP3 and MP3, p = 0.035), though the differences were marginally significant. When fish were classified into four ecological categories — benthic, benthopelagic, fossorial, and pelagic — benthic and benthopelagic species together comprised over 70% of detections at every site, and no significant differences in these proportions were found between tidal phases (chi-squared test, p > 0.05).

The decisive test came from variation partitioning, which quantified how much of the observed community variation was explained by site, tidal phase, and sampling date. Across both regions combined, site explained the largest share — 46% on Boso and 33% on Motobu — while tidal phase accounted for only 2.8% and 1.5% respectively. Within each peninsula, tidal phase explained between 3.4% and 12.2% of variation, comparable to the date factor, which explained 0% to 14.5% across sites. In other words, even though tides produced statistically detectable shifts in the eDNA signal, those shifts were dwarfed by the influence of location. The team also examined presumably exogenous eDNA from 14 oceanic deep-sea and six pure freshwater species, which together represented just 0.65% of all detections. Freshwater fish were detected more often during outgoing tides, consistent with influx from adjacent rivers, while deep-sea fish detections during outgoing tides may reflect larval stages residing in surface waters.

The practical message for coastal monitoring is clear. In regions with substantial tidal differences, fish community surveys using eDNA metabarcoding can capture community characteristics effectively without needing to account heavily for tidal levels at this spatial and temporal scale. Sampling plans comparing fish community structures across sites need not be constrained to a single tidal phase, which simplifies logistics considerably. Nevertheless, the authors caution that significant statistical differences between outgoing and incoming tides were observed at most sites, and the underlying mechanisms — whether individual fish move in response to tidal changes or eDNA is transported by tidal flow — remain unresolved. Species richness also failed to reach saturation even with 12 samples per site over two days, suggesting that further research is needed to determine the appropriate sampling effort for comprehensive community assessments. The study opens the door to more efficient coastal biodiversity monitoring across temperate and tropical waters alike, while reminding practitioners that the ocean still holds surprises in how its genetic fingerprints move with the tide.

Subject of Research: Impact of tidal changes on fish environmental DNA metabarcoding in temperate and tropical coastal regions of Japan

Article Title: Assessing the impact of tidal changes on fish environmental DNA metabarcoding in temperate and tropical coastal regions of Japan

Article References: Oka, S.-I., Sado, T., Ushio, M., & Miya, M. (2024). Assessing the impact of tidal changes on fish environmental DNA metabarcoding in temperate and tropical coastal regions of Japan. Metabarcoding and Metagenomics, 8, Article e135461. https://doi.org/10.3897/mbmg.8.135461

Image Credits: AI Generated

DOI: 10.3897/mbmg.8.135461

Keywords: Assessing, impact, tidal, changes, fish, environmental, metabarcoding, temperate, tropical, coastal, regions, Japan