2.1. Column experiments

We injected a solution containing eDNA from bluegill sunfish (Lepomis macrochirus) into flow-through columns packed with substrates of contrasting size, before and after biofilm colonization, to determine how substrate size and biology might impact the retention of eDNA in porous substrates. In all cases, we wet-packed substrate into chromatography columns of dimension 4.8 (inner diameter)× 60 (length) cm glass (Chromaflex) with Teflon fittings and attached them to a peristaltic pump via Masterflex L/S tubing [16,24]. We ran the experiments using four different porous media: clean quartz sand (hereafter S, D₅₀ = 1 mm), clean pea gravel (PG, D₅₀ = 1 cm), sand colonized with biofilm (SB) and pea gravel colonized with biofilm (PGB). Our experimental design is depicted schematically in figure 1 [33]. We chose PG and S substrates as they are representative of the small benthic substrate typical of a low-gradient streams. Prior to each substrate packing, we sterilized the glass column, tubing and fittings with a 10% bleach solution for 10 min to avoid possible eDNA contamination [34]. We rinsed the column and materials vigorously before each experiment to remove any residual bleach. For substrate with biofilms (PGB, SB), we incubated them in low-nutrient experimental streams and full sunlight for three weeks at the Notre Dame Linked Ecosystem Experimental Facility in South Bend, IN, USA (http://research.nd.edu/core-facilities/nd-leef/).

Schematic of the experimental set-up for column studies under saturated conditions (modified from Anders & Crysikopoulos [33]).

For each individual experiment, we pumped at least four pore volumes (PVs) of deionized water through the wet-packed column at 150 ml min⁻¹ with a peristaltic pump and then transferred the inflow tube to a sodium chloride solution (concentration = 25 mg NaCl l⁻¹). We measured solute breakthrough curves (BTCs) using a YSI 3200 conductivity meter and probe cell to understand conservative transport through each column. We used chloride as a conservative tracer, which is commonly used in similar experiments; on short time scales, the low concentration of chloride used should not have been a concern (Nerenberg R 2013 (Department of Civil and Environmental Engineering & Earth Sciences, University of Notre Dame). Oral communication 2013 October). Additionally, additions of NaCl are used as conservative tracers for nutrient transport studies with no significant influence on biofilms; chloride is a biologically essential solute that typically exists in streams in concentrations that exceed biological need, and is not stressful to biofilms [35]. Additionally, in some very early experiments in our eDNA studies, we also took measurements in cases where NaCl was absent with no notable influence (CL Jerde, BP Olds, AJ Shogren, EA Andruszkiewicz, AR Mahon, D Bolster, JL Tank 2013, unpublished data).

After characterization of flow-through columns using the conservative tracer, we conducted experiments with an eDNA solution, which came from established tanks holding a steady population of bluegill sunfish. We allowed eDNA to accumulate by turning off the flow-through filtration on the tanks for 12 h prior to influent solution collection; we expected the shedding rates from bluegill fry to remain similar over time [36]. To remove any large particles (e.g. tissue, scales, etc.) in the influent solution, we filtered the influent solution through clean 0.5 mm mesh. To keep the collected solution mixed and avoid settling and segregation, an aquarium pump continuously mixed the solution containing eDNA throughout all column experiments (pump speed = 0.15 ml s⁻¹). We followed a timing schedule determined using the conservative tracer additions. For the PG experiments, we pumped eDNA solution through each individual column (150 ml min⁻¹) and sampled every 30 s for the first 10 min, at 1 min intervals for 15 min and at 5 min intervals for 35 min thereafter. For the sand experiments, we sampled every minute for 40 min, then at 5 min intervals for 20 min. We designated time zero (t = 0) at the start of when we began to pump in the eDNA solution, and at t = 18 (PG, PGB) or 24 (S, SB) minutes we placed the tube back into fresh water to flush the column of suspended eDNA, but continued to monitor outflow to capture eDNA retention for 42 and 36 additional minutes, respectively. At 10 min intervals throughout each experiment, we took 15 ml samples directly from the influent solution (n = 6) using a sterilized 30 ml syringe, treating them the same as the effluent samples (see below). We performed each experiment twice with each substrate treatment, yielding two replicate runs for each substrate, for a total of eight separate experiments.

We collected all effluent samples (n = 50 per experiment) as 15 ml in 60 ml centrifuge Falcon tubes with 33.5 ml aliquot of 100% ethanol and 1.5 ml of a 3 M sodium acetate solution. We determined the pump flow rate of 150 ml min⁻¹ using Stokes settling velocity for a 200 µm particle and estimated that 150 ml min⁻¹ yielded a sufficient velocity to suspend eDNA particles (less than 200 µm). Following collection, we stored samples at −20°C until centrifugation, following the methods of Ficetola et al. [1] and Thomsen et al. [9].

After each column experiment, we collected four approximately 20 ml substrate samples directly from the column in 160 ml specimen cups for estimation of chlorophyll a (chl a—represents a relative quantification of autotrophic portion of biofilm) and OM using ash-free dry mass (AFDM—represents the total organic mass in attached biofilm) of attached biofilm. For chl a, we extracted chl from each substrate sample and measured fluorometrically using standard methods [37], expressing chl a per unit surface area of substrate. From the remaining two samples, we estimated OM by placing the approximately 20 ml of sampled substrate into 100 ml of water and mixing vigorously to loosen biofilm from the substrate; the water plus biofilm slurry was filtered onto a pre-ashed and weighted glass fibre filters (GF/F, Whatman) filtered and dried for 48 h at 60°C to measure dry mass. The filters were then ashed at 550°C for 1 h, re-wet and dried for 48 h at 60°C to measure ash-free dry mass. There was no statistical difference between chl a (t-test, t = −0.41, d.f. = 5.9, p = 0.69) or AFDM (t-test, t = −0.46, d.f. = 3.7, p = 0.67) between biofilm-colonized substrate treatments (PGB versus SB). Mean chl a was 5.32 mg cm⁻², and mean AFDM was 0.464 g cm⁻². We took biofilm and OM samples after experiments to more accurately estimate the biofilm that remained in the column over the course of the experiment, though we recognize that biological activity of the natural biofilm may have been inhibited after exposure to deionized water.