Aquarium experiment
The aquarium experiment was performed to examine how the relationships between fish abundance and target eDNA concentration varied depending on filter pore size and DNA fragment length. Five replicates of 25-L tanks, filled with 20 L of aged tap water aerated by a pump, were prepared. The water temperature was regulated (around 25°C) using a thermostat (DS150, GEX, Japan), and a 12-h/12-h light/dark cycle was adopted throughout the experiment. After a week, 500 mL of tank water was collected from each tank in advance, using a 1-L plastic beaker as a blank sample, and subsequently, five individuals of the juvenile zebrafish (total length: around 30 mm; no sex identification) were introduced into each tank. Then, after 5 days, 3 L of water samples were collected from each tank using a 10-L plastic container, introducing an additional five juvenile zebrafish (i.e., ten individuals per tank in total). After another 5 days, we collected 3 L of water samples similarly as above and introduced an additional ten individuals (i.e., 20 individuals per tank). Similarly, we collected water samples after 6 days and introduced an additional ten individuals (i.e., 30 individuals per tank). After 10 days, 3 L of water samples were collected, and ten additional individuals were introduced (i.e., 40 individuals per tank). Finally, after 5 days, we collected 3 L of water samples.
After water collection, the water samples were thoroughly mixed and 250 mL of them was filtered using a 47-mm-diameter polycarbonate membrane filter (0.2 μm pore size; Merck Millipore, Germany) using magnetic filter funnels (500 mL capacity; Pall Corporation, US) and a filtration manifold (three-place; Sanplatec Corporation, Japan). A similar procedure was performed for 500 mL of the samples with the same material filters with different pore sizes (0.8, 3, and 10 μm pore size). This alternative was performed as a result of clogging of smaller pore size filters. Additionally, 250 mL of distilled water was filtered with the same material filter at each sampling time, with 0.2-µm pore size filter used as a negative filtration control. All filtered samples were covered with a commercial aluminum foil and kept at −20°C until eDNA extraction. The fish in experimental tanks were fed a small amount of commercial bait every 2 or 3 days. The bottom of each tank was cleaned immediately after feeding to eliminate the effect of the feces. Tank water was filled up to the original water volume after water sampling and bottom cleaning was performed. We supplied replacements for the dead or dying individuals as soon as possible. However, we replaced them after water sampling when the dead or dying individuals were observed on the sampling day. Disposable gloves were worn to collect and filter water samples throughout the experiments. All the equipment for fish rearing (pomp, air stones, tube, and acrylic tanks), water collection (beakers, containers, polyethylene tanks), and water filtration (filter funnels and tweezers) were bleached before every use in 0.1% sodium hypochlorite solution for at least 5 min (Yamanaka et al., 2017).