Longfin smelt (Spirinchus thaleichthys) is a threatened anadromous fish species that spawns in freshwater to moderately brackish (i.e. 5-10 ppt) reaches of the upper San Francisco Estuary and has declined to approximately 1% of its pre-1980s abundances. Despite 50+ years of population monitoring, the efficacy of 10+ years of conservation efforts for longfin smelt remain uncertain due to a limited understanding of how the species responds to environmental variation, such as salinity. For example, high mortality during larval stages has prevented culture efforts from closing the life cycle in captivity. Here, we investigated the effects of salinity on longfin smelt yolk-sac larvae. Newly hatched larvae from four single-pair crosses were acutely transferred to and reared at salinities of 0.4, 5, 10, 20, or 32 ppt. We compared whole-body water and sodium ion (Na⁺) content, notochord length, and yolk-sac volume at 12, 24, 48, 72, and 96 h post-transfer for each salinity treatment. We found that larvae maintained osmotic and ionic balance at 0.4-10 ppt, whereas salinities ˃10 ppt resulted in decreased water and increased whole-body Na⁺ content. We also found that larvae grew largest and survived the longest when reared at 5 and 10 ppt and that yolk resorption stalled at 0.4 ppt. Finally, there were significant but small interclutch variations in responses to different salinities, with clutch accounting for <8% of the variance in our statistical models. Overall, our results indicate that longfin smelt yolk-sac larvae likely perform best at moderately brackish conditions, thus yielding a mechanism that explains their distribution in field surveys and providing key information for future conservation efforts.

2.1 Broodstock and Embryo Collection

Adult broodstock longfin smelt were collected from the wild by the U.S. Fish and Wildlife Services Chipps Island trawl with a midwater trawl, the FCCL using a lampara net in the Sacramento San Joaquin Delta, and the UC Davis Otolith Geochemistry and Fish Ecology Laboratory using an otter trawl in the Alviso Marsh in South San Francisco Bay. Broodstock collections were approved under California Department of Fish and Wildlife MOU ID: Hobbs_LFS_2021 and Specific Use Permit IDs: S-191990002-19, 199-001, and D-0021521915-6. Fish were held in 76 L cylindrical carboys during collections and transported to the FCCL at the end of each sampling day and held at 10 ppt at 12℃ until visually gravid and then strip-spawned. A total of four clutches from separate single-pair crosses were obtained between January and April 2019 (see Supplementary Table 1 for parental information) and embryos were maintained using methods described in Yanagitsuru et al. (2021). Fertilized embryos were held at 12℃ in pre-treated Delta water (0.2 ppt; see Supplementary Table 2 for additional water constituents) at the FCCL until 5 or 7 days post-fertilization (dpf) when fertilized and unfertilized embryos could be differentiated.

Fertilized embryos were transported to UC Davis fish conservation physiology laboratory in conical tubes within coolers chilled with two ice packs. The ice packs were separated from direct contact with the tubes with a Styrofoam pad to prevent temperatures from cooling too much. By the time embryos arrived, water temperatures within tubes were between 11.7 and 13.2℃. Upon arrival, embryos were placed in 500 mL nylon mesh (100 µm mesh size) containers within gently aerated 2-gallon buckets receiving flow-through (approximately 4 L h^-1) freshwater (well water; 0.4 ppt; see Supplementary Table 2 for additional water constituents). Temperature was lowered gradually to 9℃ at 0.5℃ h^-1 as this temperature yielded high hatching success in a previous study (Yanagitsuru et al., 2021). Embryos were exposed to a 12:12 light:dark cycle. Embryos from all clutches were maintained in these conditions and checked twice daily between 8:00-9:00 and 22:00-24:00 to remove any dead embryos (indicated by a milky opaque coloration, similar to delta smelt (Tsai et al., 2021), until hatching. Temperature (8.3-9.2℃), dissolved oxygen (94.8-99.2%), salinity (remained constant at 0.4 ppt), and pH were measured daily with a YSI 556 (YSI, Inc., Yellow Springs, OH, USA). Ammonia (total ammonia nitrogen) was also measured daily with a salicylate colorimetric water test kit (API, Calfont, PA, USA) and remained below detection threshold (<0.25 ppm) throughout the duration of the experiment (see Supplementary Table 3 for water parameters). All clutches were monitored for hatching at 8:00 and between 22:00-24:00. No hatching was observed between 22:00-24:00, indicating that all hatching occurred between 24:00-8:00, and thus all individuals used for experimentation were between 0 and 10 hours post-hatch (hph).

2.2 Salinity Exposures

Clutch 1 was used to measure survival over time at five different salinities: 0.4 (freshwater), 5, 10, 20, and 32 ppt. Water of different salinities were made by mixing well water with Red Sea Coral Pro salt mix (Red Sea Fish Pharm, Ltd., Herzliya, Israel) and stored in heavily aerated and mixed 300 L reservoirs. Between 8:00-9:00, hatched larvae were divided into five groups of 30 individuals and acutely transferred to 500 mL mesh containers in separate, gently aerated 2-gallon buckets receiving flow-through water (approximately 4 L h^-1) of one of the five experimental salinities. Because yolk-sac larvae do not feed exogenously and to ensure water quality issues from excess feed did not affect survival, larvae were left unfed. Larvae were monitored every 24 hours thereafter, and dead larvae were recorded and removed to determine survival until 192 hours post-transfer (hpt).

Clutches 2, 3, and 4 were used to measure wet and dry mass (used to calculate whole-body water content), notochord length, yolk-sac volume, and whole-body sodium ion (Na+) content over time post-transfer at the five experimental salinities (methods described below). For each clutch, a subset of hatched larvae was collected between 8:00-9:00 to gather pre-transfer measurements. All other larvae were divided into groups of 100 individuals into 500 mL mesh containers within 2-gallon buckets of different salinities as described above and left unfed. Because the survival trial indicated that substantial mortality began at 96 hpt and because of the limited availability of this imperiled species, we only sampled up to 96 hpt. Subsets of larvae from each salinity treatment were subsequently sampled at 12, 24, 48, 72, and 96 hpt to gather post-transfer measurements. Temperature (8.3-9.1℃), dissolved oxygen (93.5-100%), salinity (0.4 ppt: 0.4 ppt; 5 ppt: 5.0-5.7 ppt; 10 ppt: 9.9-10.5 ppt; 20 ppt: 19.9-20.8 ppt; and 32 ppt: 31.6-32.4 ppt), pH (8.2-8.6), and ammonia were also measured at this time (Supplementary Table 3). Ammonia remained below detection threshold throughout the duration of the experiment.  

2.3 Whole Body Water Content and Na+ Content

For each clutch, five groups of 10 larvae each (n=15) were sampled pre-transfer and at 12, 24, 48, 72, and 96 hpt to measure wet and dry masses and further used to calculate whole-body water content. Due to high mortality between 48 and 96 hpt in the 32 ppt treatment group for all clutches, sample sizes were reduced at 72 (n=5) and 96 hpt (n=5) for this group. Likewise, clutches 2 and 3 experienced high mortality in the 20 ppt treatment group before 96 hpt and sample sizes were reduced at 96 hpt (n=13; see Supplementary Table 4 for all sample sizes). Larvae were pooled for each mass measurement to ensure measured masses fell within the working range (minimum load: 10,000 µg; repeatability: 50 µg) of the Veritas HPB 625i semi-micro balance (Veritas Technologies LLC, Santa Clara, CA, USA) used. Minimum loads were achieved by measuring larvae within 1.5 mL centrifuge tubes. Upon collection, each group of larvae was euthanized with 0.5 g L^-1 tricaine methanesulfonate (MS-222) buffered to pH 8.0 with sodium bicarbonate (NaHCO₃) in accordance with UC Davis IACUC protocols and rinsed over a 100 µm nylon mesh screen with 1 mL of Millipore water to remove any residual salts on body surfaces. Larvae were gently dried by wicking away water with a KimWipe (Kimberly-Clark Worldwide, Inc., Roswell, Georgia, United States) and then weighed for wet mass. During preliminary testing, consecutive measures of wet mass were found to decrease with each repeated measure, suggesting that the samples were quickly drying and all wet mass for experimental samples were thus measured only once. This procedure was completed within two minutes of euthanizing larvae. The same groups of larvae were then dried in a convection oven at 60 ℃ over 48 h and weighed for dry mass. We chose 60 ℃ as this fell within the recommended temperature range by Schmidt et al., (2013) and was a temperature close to that used by a similar study by Gallagher et al., (2013). Dry mass was measured three times for each sample and the average of masses was used for analyses. Wet and dry masses were divided by 10 (number of larvae per group) to calculate average individual larval wet and dry masses. Whole-body water content was calculated as the difference between wet and dry mass and reported as a percentage of wet mass.

The same dried groups of larvae previously used to measure dry mass were then used to measure whole-body Na+ content following modified methods described in Gallagher et al. (2013). Larvae were placed in 0.2 mL microcentrifuge tubes and digested in 10x the wet mass of the pooled group in 1 mol L^-1 nitric acid (V/W) at 65 ℃ for one week. Larvae were dissociated with daily mixings with a vortex shaker (Scientific Industries, Inc., Bohemia, New York, United States) to aid digestion. After digestion, tubes were centrifuged to separate the solids and supernatant was analyzed on a Sherwood Scientific M360 flame photometer (Sherwood Scientific Ltd., Cambridge, United Kingdom) and standardized to dry mass for whole-body Na⁺ content. Whole-body Na⁺ content values were divided by 10 (number of larvae per group) to calculate an average individual whole-body Na⁺ content per sample.

2.4 Morphometrics

Subsets of larvae (see Supplementary Table 4 for sample sizes) were imaged with a Canon EO6 Rebel T6 (Canon, Tokyo, Japan) mounted on a Leica S8APO stereomicroscope (Leica Microsystems, Chicago, IL, USA) and analyzed for morphometrics with Fiji ImageJ software following methods described in Yanagitsuru et al. (2021). Whole-body larval images were taken next to an electron microscopy grid (200 µm grid, 50 µm bar) to provide a scale for calibrating pixels to mm to measure notochord length. Lateral and ventral images of yolk-sacs were all taken at 12.8x magnification and calibrated to a separate image of an electron microscopy grid at the same magnification to measure yolk-sac length, depth, and width. Yolk-sac volume was approximated as an ellipsoid (V = ⁴/₃πabc, where V is volume, a is half-length, b is half-depth, and c is half-width of the yolk-sac).

2.5 Statistical Analyses

R 3.6.3 (R Core Team, 2013) was used for statistical analyses. Linear models were used to analyze all metrics. Wet mass, dry mass, whole-body water content, whole-body Na+ content, notochord length, and yolk-sac volume were each modeled separately as functions of the fixed categorical effects of salinity, time post-transfer, clutch ID, and their interactions. Whole-body Na+ content was log-transformed to meet the assumptions of normality. Tukey’s multiple comparison post-hoc tests were used to examine differences between salinities and times post-transfer. Statistical significance was accepted at p < 0.05. All values are reported as mean ± SEM unless otherwise stated.

Some image files were missing and morphometrics were thus unable to be measured for those fish.