Host Fish Collection

Fish species and numbers varied by experiment based on our ability to collect fishes in the wild in early spring and on fish availability at hatcheries. We obtained salmonids from the following fish hatcheries: Nashua National Fish Hatchery, Nashua, New Hampshire (Atlantic Salmon, Salmo salar); Silvio O. Conte Anadromous Fish Research Laboratory, Turners Falls, Massachusetts (Brook Trout, Salvelinus fontinalis); and Massachusetts Division of Fisheries and Wildlife, Sandwich, Massachusetts (Brook Trout; Brown Trout, Salmo trutta; Rainbow Trout, Oncorhynchus mykiss). We collected all other fishes by seining and backpack electrofishing in the Fall River, Massachusetts (Slimy Sculpin, Cottus cognatus; Longnose Dace, Rhinichthys cataractae; Blacknose Dace, Rhinichthys atratulus; White Sucker, Catostomus commersonii) or the Connecticut River, Massachusetts (Banded Killifish, Fundulus diaphanous; Bluegill, Lepomis macrochirus). We collected fishes from river sections where mussels were absent or rare to avoid removing potential hosts and to reduce the chances that fishes had immunity from prior exposure to glochidia (O'Connell and Neves 1999; Rogers and Dimock 2003). We maintained fishes in aquaria and fed them black worms until the start of experiments.

Mussel Broodstock Collection and Glochidia Extraction

We collected Brook Floater broodstock from streams by snorkeling. We collected one gravid mussel from the Nissitissit River in Middlesex County, Massachusetts, in March 2017 (Experiment 1); three gravid mussels from Wesserunsett Stream in Somerset County, Maine, in April 2017 (Experiment 2); two gravid mussels from the West Branch Farmington River in Berkshire County, Massachusetts; and three gravid mussels from Wesserunsett Stream in October 2018 (Experiment 3). We transported mussels to the laboratory individually in aerated 3.7-L glass jars of water in a cooler. We maintained mussels in an environmental chamber at CARC at a temperature similar to stream temperatures at the time of broodstock collection (∼5°C) to inhibit glochidia release. We conducted experiments within 6 wk of broodstock collection. Immediately before extraction of glochidia for the experiments, we acclimated broodstock to 10°C, an approximate temperature at which glochidia are released in the wild (about 14°C; Wicklow et al. 2017).

We extracted glochidia for Experiments 1 and 2 by puncturing one or both gills with a 1-mL syringe and sterilized 18-gauge needle and flushing glochidia from the gills with water into a beaker. In Experiment 3 we used aquaria to immerse mussels in a water bath with serotonin (23 mg/L) for 2–3 h (Eads et al. 2010; Patterson et al. 2018) to induce the release of glochidia and avoid gill trauma associated with gill punctures. Glochidia from the serotonin bath were collected on a 150-µm screen and then resuspended in water in a beaker.

We determined glochidia viability for each mussel by evenly suspending glochidia in a 1000-mL beaker and collecting five 200-µL subsamples with a pipette. We placed all five subsamples together in a Petri dish, added a sodium chloride (NaCl) solution, and under a dissecting microscope counted the number of open and closed glochidia before and after exposure to NaCl. We calculated glochidia viability as

Glochidia viability across all broodstock individuals was 88%–93%; based on consistently high viability we used all broodstock in the experiments (see Hove et al. 2000). For each experiment, we combined glochidia from all broodstock, evenly suspended the glochidia, and then divided the total volume into equal stock solutions for each replicate inoculation based on target inoculation densities (see subsequent). We decanted water in each stock solution until there was only enough water to suspend glochidia in a Petri dish and then photographed the Petri dish containing the glochidia with a digital camera and macro lens (5D Mark 3S camera, 100 mm f2.8/L Macro IS USM Lens, Canon U.S.A. Inc., Huntington, New York, USA). Photographing allowed us to count glochidia added to each inoculation bath, resulting in a more accurate quantification of glochidia than volumetric estimates alone; these numbers were used to calculate attachment rates.

Experiment 1

In Experiment 1, we tested the host suitability of three fish species: Slimy Sculpin, Longnose Dace, and Atlantic Salmon. We inoculated Slimy Sculpin (mean length = 72 mm ± 5.0 SD) and Longnose Dace (87 mm ± 7.0) by placing six individuals of each species in 200 mL of water in a McDonald-type hatching jar (similar to those produced by Global Aquaculture Supply Co., Sioux Falls, South Dakota, USA; hereafter, McDonald jar). Our target inoculation density was 200 glochidia/fish; however, counts of glochidia in photographs indicated true inoculation densities of 121 glochidia/ fish (3.64 viable glochidia/mL; Table 1) for Slimy Sculpin and 150 glochidia/fish (4.50 viable glochidia/mL; Table 1) for Longnose Dace. Air injected into the bottom of the McDonald jars suspended the glochidia, facilitating glochidia contact with fishes. We exposed fishes for 20 min, removed the fish, and then filtered the water over a 150-µm mesh sieve to collect unattached glochidia. We counted the number of unattached glochidia and subtracted this number from the estimated number of glochidia in the inoculation bath to estimate the attachment rate (the percentage of viable inoculated glochidia that attached to each fish; Table 2).

Atlantic Salmon (mean length = 180 mm ± 1.0 SD) were too large for the McDonald jars; therefore, we pipetted glochidia directly onto the gills of two individuals. Before inoculating fish, we photographed the Petri dish containing the glochidia that we pipetted onto the gills of each fish. We anesthetized fish with tricaine methanesulfonate (MS 222) and pipetted the entire glochidia stock solution onto the left or right gills to obtain a target inoculation density of 300 glochidia/fish. We conducted the inoculation over a tray to collect unattached glochidia, and we counted glochidia in the tray to estimate the number of glochidia that attached to each fish by subtracting the number counted in the tray from the number counted in the photographs (Table 1).

After inoculation, we placed Slimy Sculpin and Longnose Dace in 3-L Aquatic Habitat (AHAB) tanks (Pentair Aquatic Ecosystems, Apopka, Florida, USA), for a total of three tanks/ species (two individuals/tank). We placed individual Atlantic Salmon in separate 9-L AHAB tanks. We inspected the contents of each tank every 1–3 d, beginning the day after inoculation. We collected sloughed glochidia or juveniles by increasing the flow in the AHAB tanks for 10 min and collecting flushed material on a 150-µm filter. We placed sloughed glochidia or juveniles from each tank and collection event in a Petri dish and counted glochidia and juveniles under a dissecting microscope. Starting day 7 postinoculation, most juveniles exhibited a foot and two adductor muscles but lacked movement; thus, we left material in Petri dishes overnight at room temperature (∼18°C) and inspected it the next morning. Mussels that exhibited foot movement the next morning were considered metamorphosed juvenile mussels, and all other individuals were considered sloughed glochidia. We estimated the metamorphosis rate of attached glochidia by dividing the total number of live juveniles recovered from tanks by the total number of glochidia collected from tanks (Rogers et al. 2001).

If no juveniles were collected after 5 d, we inspected a subsample of fish, and if no glochidia were attached, we terminated the experiment. We sacrificed all fishes at the completion of all experiments and inspected the fishes under a compound microscope for remaining glochidia. The duration of the experiments was 37–40 d. Using room-controlled temperature we slowly increased the water temperature in the AHAB tanks from 13°C to 19°C (average rate = 1°C/d for 6 d) to facilitate glochidia metamorphosis. The initial AHAB temperature (13°C) was chosen to reduce thermal stress during transfer of glochidia and fishes from the holding and inoculation chambers. We measured dissolved oxygen in a subset of the AHAB tanks every 3 d with a YSI Professional Plus multiparameter water quality meter (Xylem, Inc., Yellow Springs, Ohio, USA); dissolved oxygen was >7.0 mg/L for all measurements.

Experiment 2

In Experiment 2, we retested Slimy Sculpin (mean length = 72 mm ± 10 SD) and Longnose Dace (72 mm ± 11) using different individuals than in Experiment 1 and tested five new fish species: Blacknose Dace (mean length = 67 mm ± 7 SD), Banded Killifish (75 mm ± 9), Bluegill (77 mm ± 2), White Sucker (122 mm ± 5), and Brook Trout (375 mm ± 109). We inoculated 12 individuals each of Longnose Dace, Blacknose Dace, and Banded Killifish, with each species divided into two replicate inoculations in separate McDonald jars with six individuals/jar. We inoculated six Slimy Sculpin together in a single McDonald jar. We inoculated three White Sucker and four Bluegill, with each species in a single McDonald jar. Water volume in all McDonald jars was 250 mL (50 mL higher than in Experiment 1). Inoculation methods and duration in McDonald jars were as described for Experiment 1 using a McDonald jar.

Table 1.

Inoculation methods for three host identification experiments for Brook Floater (Alasmidonta varicosa). Fish species without entries under “Replicate” were held in a single chamber. Water volume is the volume of the inoculation bath. The stock solution represents the glochidia solution used to inoculate fishes. The target inoculation density was determined volumetrically. The actual inoculation density and stock solution glochidia density were determined later by counting glochidia in photographs of the inoculation stock to which fishes were exposed. Scientific names for fishes are in Table 4.

Table 2.

Glochidia attachment rates and juvenile metamorphosis rates of Brook Floater (Alasmidonta varicosa) on fishes in three experiments. Attachment rate is the percentage of inoculated glochidia that attached to fishes. Metamorphosis rate is the percentage of attached glochidia that metamorphosed into juvenile mussels. Average juveniles/fish is based on the daily number of juveniles produced/the number of fish surviving, summed across experimental days. Mean values and SD were calculated only from replicate chambers in which fishes survived to produce juvenile mussels (see Fig. 2). Scientific names for fishes are in Table 4.

Our target inoculation densities were 250 glochidia/fish for Longnose Dace, Blacknose Dace, Banded Killifish, and Slimy Sculpin; 300/fish for White Sucker; and 200/fish for Bluegill. Photographic counts indicated that inoculation densities differed slightly from our targets (Table 1). For example, replicate inoculations for Longnose Dace contained 1,284 viable glochidia (214 glochidia/fish; 5.14 viable glochidia/mL; Table 1) and 1,338 viable glochidia (223 viable glochidia/fish; 5.36 viable glochidia/mL; Table 1).

We inoculated Brook Trout together in a single bucket with 23 fish in 4,000 mL of water. We exposed fish for 20 min, removed the fish, and then filtered the water over a 150-µm mesh sieve to collect unattached glochidia. Our target inoculation density was 1,000 glochidia/fish, but photographic counts indicated a density of 743 glochidia/fish (4.27 viable glochidia/mL).

After inoculations, we separated fishes into AHAB tanks that consisted of three 3-L tanks for Blacknose Dace (4 fish/ tank), Longnose Dace (4 fish/tank), Banded Killifish (4 fish/ tank), Slimy Sculpin (2 fish/tank), and White Sucker (1 fish/ tank). We placed Bluegill (2 fish/tank) into two replicate 3-L tanks and Brook Trout into one 260-L circular tank.

We collected glochidia and juvenile mussels from AHAB tanks following methods described for Experiment 1. We collected glochidia and juveniles from the Brook Trout tank by siphoning 60 L of water from the tank bottom with a 2-cm hose every 1–3 d; we collected siphoned material on a 150-µm filter. We estimated metamorphosis rate and measured dissolved oxygen as described for Experiment 1. Experiment 2 ended on days 24–34.

Experiment 3

In Experiment 3, we tested new individuals of Brook Trout (mean length = 145 mm ± 13 SD), Rainbow Trout (146 mm ± 11), and Brown Trout (140 mm ± 10). We inoculated fishes with glochidia following methods described for Brook Trout in Experiment 2, except that we inoculated each fish species in three replicate inoculation baths, each with 15 individuals. Our target inoculation density was 200 glochidia/ fish, but photographic counts indicated densities of 200–315 glochidia/fish (0.75–1.18 viable glochidia/mL; Table 1). We calculated glochidia attachment rate as in Experiments 1 and 2.

After inoculations, we transferred fishes from each bath into a 113-L circular tank with flow-through well water; we used three replicate tanks for each species, each containing 15 individuals. Unlike in Experiments 1 and 2, for Experiment 3 we kept all fish from each replicate inoculation bath in the same holding tank throughout the experiment, which allowed us to examine the relationship between attachment rate and metamorphosis rate. We increased the tank temperature from 15°C to 18°C using a heater (average increase = 1°C/d). We inspected tanks for glochidia and juveniles as described for Brook Trout in Experiment 2. We estimated metamorphosis rate as described for Experiment 1 and measured dissolved oxygen daily. Experiment 3 lasted 25 d.

Table 3.

Results of generalized linear models (GLMs) assessing factors that predict Brook Floater glochidia attachment and juvenile metamorphosis rates in Experiment 3. Inoculation density is the number of viable glochidia/mL to which fishes were exposed (see Table 1). Attachment is the estimated number of glochidia attached/fish calculated as the chamber-wide attachment rate divided by the number of fish in the chamber. The top models are in bold.

Data Analysis

We created sets of generalized linear models (GLM) to assess how well attachment rate (Experiment 3) and metamorphosis rate (Experiment 1 and 3) were predicted by various factors. We did not assess metamorphosis rate for Experiment 2 because of high fish mortality resulting in insufficient replication for analysis. For Experiment 1, we created a model to assess how well metamorphosis rate was predicted by host species (fixed factor). We excluded Atlantic Salmon from these models because of insufficient replication. For Experiment 3, we compared models to assess how well attachment rate was predicted by host species and inoculation density (number of viable glochidia/mL in the inoculation bath) individually, and when both factors were modeled together as an additive term (Table 3). For Experiment 3 we also created models to assess how well metamorphosis rate was predicted by host species and attachment rate, individually and together. For this model, we expressed attachment rate as the number of glochidia attached to the fish.

For each experiment, we created a separate model for each factor or combination of factors and included a null model (a model with no explanatory factors; Table 3). We fit all models with a logit link function and a quasi-binomial error structure; this error structure accounted for overdispersion that resulted from clustering in the data. We evaluated models by fitting them twice: we first extracted the log-likelihood from the binomial model, and then we extracted the dispersion parameter from the quasi model to calculate the likelihood; these were used to calculate a quasi-corrected Akaike Information Criterion (qAIC) (Bolker 2021). We calculated explained deviance by subtracting the residual deviance from the null deviance and dividing by the null deviance (Zuur et al. 2015). We selected the best model as the most parsimonious model with high explained deviance and low qAIC (Burnham and Anderson 2004; Wagenmakers and Farrell 2004). We contrasted marginal means using 95% confidence intervals to compare fixed factors in models, and we back-transformed standard error intervals from the logit scale using package “emmeans” (Length et al. 2022; R package version 1.6.0.). All data analyses and models were created in R v4.0.2 software package (R Core Team 2020, Vienna, Austria).

Figure 1.

Number of sloughed glochidia or juvenile mussels produced by Brook Floater (Alasmidonta varicosa) in Experiment 1. Data points and bars represent the mean and standard deviation, respectively, among replicate fish holding chambers on each day standardized by the number of fish in each chamber.

Experiment 1

Glochidia attachment rate was high for all fish species (range = 78.1%–84.0%, Table 2). For Slimy Sculpin and Longnose Dace, most sloughed glochidia appeared within 5 d of inoculation (Fig. 1). For Atlantic Salmon, large numbers of sloughed glochidia appeared within the first 5 d, but this was followed by another peak shortly before juveniles began to appear on day 15 (Fig. 1).

Mean metamorphosis rate of attached glochidia varied by host species and was highest for Slimy Sculpin (80.9% ± 2.6 SD), followed by Atlantic Salmon (35.2% ± 13.7) and Longnose Dace (29.1% ± 21.9) (Table 2). Metamorphosis rate was similar across the three replicates for Slimy Sculpin, but it varied for Atlantic Salmon and Longnose Dace (Fig. 2). Production of juveniles on Slimy Sculpin and Longnose Dace began on days 17 and 15, respectively, and Slimy Sculpin peaked on day 24; production of juveniles on Longnose Dace did not indicate a clear peak (Fig. 1). Juvenile production on Atlantic Salmon began on day 15 but appeared to occur over a more protracted period with no distinct peaks.

Fish species was a good predictor of metamorphosis rate. When comparing modeled probability of metamorphosis using 95% confidence intervals among fish species, Slimy Sculpin had a higher probability (0.81; 95% confidence interval = 0.57–0.93) than Longnose Dace (0.22; 95% confidence interval = 0.09–0.43) (P < 0.05); this model explained 79.5% of the deviance.

Experiment 2

Attachment rate varied among fish species (Table 2). The lowest attachment rate of glochidia was on Bluegill (51.0%) and the highest was on Brook Trout (80.3%), with the other species having attachment rates of 61.1%–77.6%. Sloughed glochidia appeared mostly in the first 5 d after inoculation for all species except for Brook Trout, which sloughed glochidia until day 10 (Fig. 3).

Metamorphosis rate varied greatly among fish species and was highest for Brook Trout (71.6%) and Slimy Sculpin (72.6% ± 5.2 SD) and lowest for Bluegill (4.9%) (Table 2). Metamorphosis rate was similar across the three replicates for Slimy Sculpin, but it varied among replicates for all other species (Fig. 2). Production of juvenile mussels began on days 10–13 for all species except Bluegill, from which one juvenile appeared on day 24. Production of juvenile mussels peaked on day 11 for Brook Trout and on days 20 and 21 for Slimy Sculpin and Banded Killifish. Juvenile production from fish species that had a low metamorphosis rate (e.g., Longnose Dace, Blacknose Dace, White Sucker) did not display conspicuous peaks (Fig. 3), and Bluegill produced only a single juvenile.

Experiment 3

Attachment rate was similarly high among the three trout species (83.2%–84.6%, Table 2). Sloughed glochidia appeared mostly before day 11 for Brook Trout and Brown Trout and before day 7 for Rainbow Trout (Fig. 4).

Figure 2.

Juvenile metamorphosis rate (number of juveniles/number of glochidia) of Brook Floater (Alasmidonta varicosa) on fishes in three experiments. Replicates refer to individual fish holding chambers. Numbers above each bar refer to the number of fish in each chamber that survived (left number) out of the initial number inoculated (right number).

Metamorphosis rate varied widely among species and was highest for Brown Trout and lowest for Rainbow Trout (Table 2), but metamorphosis was similar among replicates for all three species (Fig. 2). Production of juvenile mussels began on days 11–12 for all three species and peaked on day 12 for Brook Trout and days 14–16 for Brown Trout and Rainbow Trout (Fig. 4).

The top model for predicting glochidia attachment included host species + inoculation density and explained 54.9% of the deviance (Table 3). In the top model, contrasts among attachment rates for host species did not differ (P > 0.05), and inoculation density alone was only a marginally significant factor (P = 0.07). The model including only host species explained 8.3% of the deviance, and the model including only inoculation density explained 28.2% of the deviance. Overall, models with host species + inoculation density and inoculation density alone were within two qAIC units of the null model, and thus models were not considered strong predictors of glochidia attachment.

The top model for predicting glochidia metamorphosis contained host species only, explained 98.7% of the deviance, and had the lowest qAIC (Table 3). Brown Trout had the highest probability of metamorphosis (0.62 ± 0.02 SD), followed by Brook Trout (0.13 ± 0.02; P < 0.001) and Rainbow Trout (0.06 ± 0.01; P < 0.001).