Aquarium experiment and sampling
To examine the effect of changes in species composition on the behaviour of eDNA, we conducted aquarium experiments using two mock fish communities comprising H. neglectus, C. temminckii, O. latipes, R. flumineus, and M. anguillicaudatus. Mock community 1 (MC1) consisted of one individual of each of the five fish species, whereas mock community 2 (MC2) consisted of three H. neglectus individuals and one individual of each of the other four fish species (Fig. 2). We used two aquaria (A and B). Each aquarium was used four times, twice for each mock community, giving two replicates (R1 and R2). This resulted in eight experimental units (2 mock fish communities × 2 aquaria × 2 replicates). Figure 2 shows the experimental setup used in this study.
Experimental setup of the aquarium experiments.
To set up the aquaria, 20 L of tap water was added into each aquarium (GEX Co. Ltd., Osaka, Japan) and heated with a heater (Spectrum Brands, Wisconsin, US) until the water temperature reached 25 °C. Water in the two aquaria was maintained at 25 °C and constantly circulated with an aeration device. Before adding fish to the aquaria, the water was sampled for the negative control. The first experimental samples (day 0) were taken 1 h after adding the fish and subsequent samples were taken each day until day 4. At each sampling, two 1-L samples of surface water were collected from each aquarium and then 2 L of tap water was added to each aquarium to maintain the volume of water. The weight of individual fish species was measured using an electronic balance immediately after the final sampling. After each experiment, the two aquaria were bleached before being reused.
In Japan, experiments on fish do not require any legal procedures or permission. However, in order to avoid causing pain to the specimens, the experiments in this study were conducted in accordance with the ARRIVE guidelines, Japanese laws and guidelines for mammals, birds, and reptiles as below; Act on Welfare and Management of Animals (Notice of the Ministry of the Environment No. 105 of October 1, 1973), Standards relating to the Care and Keeping and Reducing Pain of Laboratory Animals (Notice of the Ministry of the Environment No. 88 of 2006), Fundamental Guidelines for Proper Conduct of Animal Experiment and Related Activities in Academic Research Institutions under the jurisdiction of the Ministry of Education (Notice of Ministry of Education No. 71, 2006), and Guidelines for Proper Conduct of Animal Experiments (established by the Science Council of Japan on June 1, 2006).
DNA extraction
Each 1-L water sample was filtered immediately through a GF/F glass fibre filter (nominal pore size = 0.7 μm, diameter = 47 mm; GE Healthcare Japan Corporation, Tokyo, Japan). Filter funnels and measuring cups were bleached after filtration to prevent cross-contamination among the water samples. All filters were stored separately at − 20 °C until DNA extraction. Total eDNA was extracted from each filter using a DNeasy Blood and Tissue Kit (QIAGEN, Hilden, Germany) and Salivette tubes (Sarstedt AG & Co. KG, Nümbrecht, Germany). Extraction methods were as previously described18 with modifications. A filter sample was placed in the upper part of the Salivette tube and 220 μL of solution containing Buffer AL (200 μL) and Proteinase K (20 μL) was added. The tube containing the filter was incubated at 56 °C for 30 min, then centrifuged at 5000 × g for 3 min, and the solution was collected in the base of the tube. To increase eDNA yield, 220 μL Tris-EDTA (TE) buffer was added to the filter sample and centrifuged at 5000 × g for 1 min. Then, ethanol (200 μL) was added to the collected solution, and the mixture was transferred to a spin column. Total eDNA was eluted in buffer AE (100 μL), following the manufacturer’s instructions. All eDNA samples were stored at − 20 °C prior to qSeq and dPCR.
Quantitative sequencing
Simultaneous quantification and sequencing of the extracted eDNA were performed by qSeq as previously described15,16. First, SPE was performed. The SPE reaction mixture (20 µL) consisted of 1 × PrimeSTAR Max premix (Takara Bio Inc., Kusatsu, Japan), 300 nM of the primer qSeq-MiFish-U-F (Table 1), and extracted DNA (2 µL). The SPE primer qSeq-MiFish-U-F contains an eight-base length random sequence tag, which creates 65,536 different variations, enabling the quantification of up to approximately 1.0 × 105 copies of DNA15. This amount of variation was sufficient to quantify the abundance of eDNA in this study. SPE was initiated by denaturation at 94 °C for 1 min, followed by cooling to 60 °C at 0.3 °C/s, incubation at 60 °C for 1 min, and final extension at 70 °C for 10 min. Subsequently, the excess primer was completely digested by adding exonuclease I (4 µL, 5 U/µL; Takara Bio Inc.) to the SPE mixture. The digestion was performed at 37 °C for 120 min, followed by inactivation of the exonuclease I at 80 °C for 30 min. The first-round PCR mixture (25 µL) contained PrimeSTAR Max premix (12.5 µL), primers qSeq-MiFish-U-R and F2 (300 nM each; Table 1), and the SPE product (2 µL). Following 40 cycles of amplification at 98 °C for 10 s, 55 °C for 5 s, and 72 °C for 5 s, the amplification product was subjected to agarose gel electrophoresis, and the band of the expected size was removed and purified using Nucleospin Gel and PCR Clean-up column (Takara Bio Inc.). The qSeq-MiFish-U-R primer also contains eight N bases to increase the complexity, which improves the sequencing quality, and thus PhiX was not added in this study. Finally, a 2nd-round PCR was performed to add an index for Illumina sequencing as described elsewhere15. The indexed PCR amplicon was purified using AMPure XP beads (Beckman Coulter, Indianapolis, IN) followed by sequencing using a MiSeq platform with MiSeq Reagent Kit v3 for 600 cycles (Illumina). The sequence data obtained in this study were deposited in the DDBJ database under accession numbers SAMD00219124–SAMD00219214.
Data analysis
First, all sequences were assembled and screened by length and quality of reads using the mothur software package (v1.39.5)22. The processed sequence reads were classified using the MiFish pipeline (http://mitofish.aori.u-tokyo.ac.jp/mifish/), with the parameters as previously described23. Subsequently, the representative sequences of individual operational taxonomic units (OTUs) were extracted using the Usearch program (http://www.drive5.com/usearch/). The random sequence tags (RST) at the end of sequences in the OTUs were counted to quantify the environmental DNA from each fish species as described elsewhere16. For comparison, the relative proportion of eDNA from individual species in each sample was calculated from the composition of the sequences of the fish species obtained by qSeq.
Microfluidic digital PCR
Quantification of eDNA was also performed by microfluidic dPCR using the BioMark Real-time System and 12.765 Digital Array (Fluidigm Corporation, South San Francisco, CA, United States) as previously described13. For each sample, the PCR mixture (6 µL) contained 2 × Probe qPCR mix (3.0 µL; Takara Bio Inc.), 20 × binding dye sample loading reagent (0.6 µL; Fluidigm Corporation), forward and reverse primers (900 nM), TaqMan probe (125 nM), ROX solution (0.015 µL), and sample DNA (1.0 µL). We used three sets of primers and probes to quantify the eDNA of H. neglectus, O. latipes, and M. anguillicaudatus (Table 1). PCR was initiated at 98 °C for 2 min, followed by 50 cycles of 98 °C for 10 s and 60 °C for 1 min. The amplification curves obtained from individual reaction chambers of the microfluidic chip were analysed using Fluidigm Digital PCR analysis software (Fluidigm Corporation) to obtain abundance of DNA molecules.
Statistical analysis
We employed Gaussian Type II regression models with the standardised major axis method to determine the relationship between the log10 eDNA abundances obtained from qSeq and dPCR analyses with the “sma” function of the “smatr” ver. 3.4.8 package in R ver. 3.6.024. Zero values were disregarded for the modelling. We employed the Gaussian Type II model because our preliminary evaluation showed higher R2 values for Type II regression models with a Gaussian distribution than for those with a logarithmic distribution in all cases. We compared the differences in the coefficient values by overlapping the 95% confidence interval (CI) ranges.
Source: Ecology - nature.com
