Effect of Oxygenation and Location on Survival and Growth of Endangered Lost River Suckers in Net Pens Free

Conservation aquaculture, which seeks to cultivate aquatic organisms to prevent extinction of species and restore ecosystems, is becoming increasingly popular in endangered species management (Schreier et al. 2012; Froehlich et al. 2017; Tave et al. 2019; Chandra and Fopp-Bayat 2021). Unlike conventional aquaculture that maximizes productivity, conservation aquaculture emphasizes the importance of behavioral development for improving post-release survival (Brown and Day 2002; Tave et al. 2018). Rearing practices involving fish receiving natural prey are intended to produce fish that are better equipped to survive in the wild compared to those involving conventional rearing (Roberts et al. 2009; Tave et al. 2019). Soft releases, another component in conservation aquaculture, is the practice of providing an acclimation period at the release site to allow fish time to adjust to environmental conditions, adapt to consumption of natural prey, and increase growth prior to liberation (Brown and Day 2002; Golden et al. 2006; Furr 2018). Cage culture is a commonly used method for soft releases that can increase rearing capacity while providing a protected, seminatural rearing environment for captively reared endangered species (Sowka and Brunkow 1999; Billman and Belk 2009). However, concentrating large numbers of fish in a single location during soft releases poses significant risks, as fish are subject to unpredictable environmental conditions they cannot escape (e.g., hypoxia, disease, predation). To mitigate these risks, careful site selection is critical, and incorporating technology such as oxygenation systems may be necessary to enhance survival and support successful acclimation.

Two species of endemic fish to the Klamath River basin, the Lost River Sucker Deltistes luxatus and Shortnose Sucker Chasmistes brevirostris, are jointly listed as endangered (U.S. Endangered Species Act, ESA 1973, as amended; USFWS 1988). Lost River Suckers have two remaining populations: Clear Lake, California, and the larger population in Upper Klamath Lake, Oregon (National Research Council 2004; Hewitt et al. 2018; see Krause et al. 2023 for decline in Upper Klamath Lake Lost River Sucker population). Wild Lost River Sucker populations in Upper Klamath Lake are primarily comprised of adults that were spawned in the early 1990s (Hewitt et al. 2018; Bart et al. 2021; Krause et al. 2022). Few captures of age-1+ Lost River Suckers coupled with negligible recruitment of young adults into the spawning populations indicate high juvenile mortality is limiting the recovery of this species in Upper Klamath Lake (Burdick and Martin 2017; Hewitt et al. 2018). Hypothesized causes of mortality include parasites, pathogens, avian and fish predation, interactions with nonnative species, toxic substances from cyanobacteria or agricultural practices, water quality related to massive cyanobacterial blooms, or some combination of these factors (USFWS 2019). Of these, only avian predation on juvenile Lost River Sucker mortality has been quantified at 4.4–14.9% annually (Evans et al. 2022). The most frequently hypothesized cause of mortality for young-of-year of the two sucker species in Upper Klamath Lake is poor water quality related to massive blooms and subsequent death of the cyanobacterium Aphanizomenon flos-aquae (Bortleson and Fretwell 1993; Kann and Welch 2005; USFWS 2019; Eldridge et al. 2012a). Cyanobacterial blooms cause high pH, and the subsequent decay of cyanobacteria causes hypoxia coupled with high un-ionized ammonia.

To offset juvenile mortality for imperiled Lost River Sucker and Shortnose Sucker populations, the U.S. Fish and Wildlife Service (USFWS) began to develop and implement the Sucker Assisted Rearing Program (SARP) in 2015 (Day et al. 2017; Day et al. 2021). This program follows the examples of other Catostomidae recovery programs in the western United States, including those for Cui-ui Chasmistes cujus, June Sucker Chasmistes liorus, and Razorback Sucker Xyrauchen texanus, that incorporate captive rearing (Day et al. 2017). Larval Lost River Suckers are captured from the lower Williamson River, a known spawning tributary, and grown to ≥ 200 mm total length prior to release into Upper Klamath Lake (USFWS 2019; Day et al. 2021). The goal of SARP is to raise juvenile Lost River Suckers and Shortnose Suckers to a size that improves survival during the poor summertime water quality of Upper Klamath Lake. These fish are then introduced into Upper Klamath Lake by direct releases into the lake, or through a soft release into protected net pens. These soft releases provide an opportunity to monitor responses to initial reintroduction and time for additional growth and acclimation to lake conditions prior to release (Caldwell et al. 2023).

Hypoxia can impair recovery efforts using soft releases or cage culture of Lost River Suckers in Upper Klamath Lake. In laboratory tests, juvenile Lost River Suckers succumb to sustained dissolved oxygen concentration <1 mg/L at 22°C within an hour, and ≤1.34 mg/L at 22°C within 24 hours (Saiki et al. 1999; Martin and Saiki 1999; Stone et al. 2017). Whereas, 98% survived for at least 14 d at 2.1 mg/L at 22°C (Meyer and Hansen 2002). The lowest dissolved oxygen concentration in which growth, blood ion loss, and swimming performance were not affected was 2.1 mg/L at 22°C (Meyer and Hansen 2002). Lethal and sublethal oxygen deficiencies occur annually near the benthos and to a lesser extent the entire water column from July to September (Burdick et al. 2020a; Burdick et al. 2021).

Oxygenation and aeration have both been considered to improve survival and growth of juvenile Lost River Suckers in net pens within Upper Klamath Lake. Because the atmosphere is only about 21% oxygen, aeration primarily reduces hypolimnetic hypoxia by mixing the water column and is inefficient at transferring oxygen from air to water (Beutel and Horne 1999). Oxygenation is five times more efficient at transferring oxygen to water, decreases the probability of super saturation with dissolved nitrogen, and is less likely to mix the water column or disturb sediments than aeration (Beutel and Horne 1999). Oxygenation can successfully improve growth (Welker et al. 2019), survival (Welker et al. 2019; Dwyer and Peterson 1993) and increase allowable rearing densities in raceways (Dwyer and Peterson 1993; Miller et al. 1995) in cold water fish hatcheries (Oncorhynchus spp., Liboriussen et al. 2009). Cone and box shaped contact chambers that use pressure and turbulence to infuse water with oxygen have been used to maintain hypolimnion oxygen concentrations at or above 2 mg/L in deep lakes (30 m average depth, Beutel and Horne 1999; Preece et al. 2019), but in shallow lakes (≤ 7 m average depth) with thin hypolimnion, oxygenation is less effective (Liboriussen et al. 2009). Use of oxygenation in large, warm, hypereutrophic lakes less than 3 m deep (average depth), to address periodic hypoxia caused by cyanobacterial blooms, has not been evaluated. Given that water holds less oxygen at warm rather than cooler temperatures (Wetzel 2001), it is unclear if oxygenation will work in the shallow warm Upper Klamath Lake and improve juvenile Lost River Sucker survival.

Another way to ensure higher oxygen concentration for imperiled juvenile Lost River Suckers would be to place net pens in locations with less frequent hypoxia. In Upper Klamath Lake these locations are in deeper water far from shore, which makes routine maintenance of net pens logistically challenging and time consuming. Therefore, a solution that includes oxygen supplementation that would allow net pens to be located in shallow water near shore, where hypoxia frequently occurs is desirable.

We compared net pen location and oxygenation as methods for improving oxygen concentration and thus survival and growth. We examined the effect of a low-cost portable oxygenation system on survival of juvenile Lost River Suckers in net pens placed at two locations with different depths and ambient oxygen profiles, in the warm, shallow (≤ 4.2 m), hypereutrophic waters of Upper Klamath Lake. We hypothesized that ambient oxygen content would vary with net pen location within the lake and that oxygenating net pens would decrease the frequency and severity of hypoxic events at both locations. Finally, we hypothesized that oxygen concentration would improve survival and growth of juvenile Lost River Suckers.

Upper Klamath Lake is the largest freshwater lake by surface area in Oregon (approximately 305 km²; NRC 2004) and has an average depth of 2.6 m, except for a 6.4–9.5 m trench that runs parallel to the western shore. As a result of wetland loss and anthropogenic eutrophication, the phytoplankton community shifted from diatoms in the early 1900s to a monoculture of A. flos-aquae that shapes the current water quality conditions in the lake (Bortleson and Fretwell 1993). As daily mean summertime water temperatures reach about 21°C, A. flos-aquae blooms cause pH to exceed 9.5 and dissolved oxygen concentration to fluctuate by 7 mg/L a day (Eldridge et al. 2012b). Upper Klamath Lake has an unstable and thin hypolimnion that only thermally stratifies for days to a week at a time. Because water holds more oxygen at cooler temperatures and hourly summer temperatures range from about 12.5 to 28.7°C, water can be 100% saturated at dissolved oxygen concentrations from 10.5 mg/L to 7.6 mg/L, respectively. When the lake is stratified, dissolved oxygen concentrations can be greater near the surface than the benthos. Wind events that mix the water column result in death and decay of A. flos-aquae that leads to a decline in dissolved oxygen concentrations (<3.4 mg/L), especially near the benthos (Kann and Welch 2005; Eldridge et al. 2012b). Un-ionized ammonia (NH₃) rarely exceeds known sublethal thresholds of Lost River Suckers, but can be higher within the pore water of the benthic sediments (Kuwabara et al. 2016; Burdick et al. 2020a).

We installed two pairs of net pens in Upper Klamath Lake, equipped with and without oxygen, during June 2019. Net pens were approximately 3 × 3 m square enclosures constructed of 0.63 cm nylon netting stretched across ridged poly vinyl chloride (PVC) pipe frames that extended 1 m above the water column at each site (Figure 1). Net pens rested on the bottom of the lake and were surrounded by buoyant docks (Figure 1). Mammalian predation was prevented with 2.54-cm² poultry wire which surrounded the outside of the net pen, and bird predation was prevented with 2.5-cm² nylon netting on the exposed part of the pen above the water.

Net pens were located at Mid North (MDN), an offshore site devoid of vegetation, and Fish Banks (FB), a nearshore site bordering a wetland and surrounded by submerged aquatic vegetation (Figure 2). Sites were chosen for their contrasting oxygen concentrations, and survival and growth of juvenile Lost River Suckers in two previous net pen studies conducted in 2016 and 2017. In these studies, survival was lower and growth slower at FB than at MDN (Hereford et al. 2019; Burdick et al. 2020b). Although hypoxia occurs at both sites, the frequency and duration of hypoxic events tends to be less at MDN than FB, which may be attributed to the shallower depth and warmer temperatures at FB. In previous net pen studies in Upper Klamath Lake, fish grew throughout the summer study period (Hereford et al. 2019; Burdick et al. 2021), and prey was not thought to be a limiting factor in their growth. Therefore, we did not feed fish in net pens. Depth in the net pens declined from 1.7 m to 1.1 m at FB and from 4.2 m to 3.1 m at MDN over the study period. The net pens at each site were approximately 100 m apart, with the oxygenated net pen located down current from the non-oxygenated net pen (Wood et al. 2008).

To increase dissolved oxygen concentration, we used Water Spikes (Warren Water Broom MFG, Astoria, Oregon) that consisted of pressurized turbulence chambers that infused water with oxygen. Water Spikes consisted of an aluminum box (10.2 × 30.5 × 61 cm) with two perforated internal baffles that created pressure, turbulence, and increased the concentration of dissolved oxygen gas according to Henry’s Law. Henry's Law states that the amount of gas that can be dissolved in a liquid is directly proportional to the partial pressure of that gas in equilibrium with that liquid (Brown et al. 2012). By increasing the pressure inside the Water Spikes, more oxygen dissolves into the water. We chose to use oxygenation rather than aeration to reduce turbulence near the benthos, which can increase diffusion of ammonia from the sediments into the water column (Beutel and Horne 1999; Zhang et al. 2018). Disturbing the sediments in Upper Klamath Lake is a concern because ammonia concentrations in the sediment can be high, contributing as much as 42.7 mg of NH⁴ per (m²)⁻¹ d⁻¹ to the water column (Kuwabara et al. 2016). A 373-watt, 95 liter per minute (lpm) waterproof well pump moved subsurface water into the Water Spike. An Invacare Perfecto 2 oxygen concentrator extracted oxygen from the air to produce a 5 L per min flow and inject about 7,200 L a day into the Water Spike through a valve near the water input (Figure 1). Prior to deployment, output from the oxygen concentrators was confirmed to be ≥95% pure oxygen. The concentrator had an emergency warning indicator to notify users if oxygen content declined below 85%, which was not triggered during our study. The Water Spikes were mounted to the side of the dock partially submerged at the water line and oxygenated water was diffused into the net pen near the benthos with perforated acrylonitrile butadiene styrene pipe (Figure 1). The water pump and oxygen concentrator were powered with a gasoline generator with an external tank to allow for extended run time (Figure 1). Approximately every 3 d, the whole system was briefly shut down (<5 min) for maintenance.

Each net pen was equipped with a Biomark Multiplexing Transceiver System (Biomark IS1001 Master Controller, Boise, Idaho) and three antennas (1.2 × 0.6 m). Each was installed horizontally with one near the surface (hereafter “top antenna”), one at mid-water column, and one on the bottom to detect passive integrated transponder (PIT) tagged Lost River Suckers (Figure 1). The top antenna floated and lowered as the lake surface dropped over the period of the study. Antennas were tuned to detect Lost River Suckers up to a distance of approximately 0.3–0.5 m so that the detection areas did not overlap at the beginning of the study. To prevent overloading the data capacity of the system, tags were recorded no more than once every 10 min per antenna, unless the tag was detected at a second antenna. A tag detected at two or more antennas within 10 min would reset the record delay for that tag, which resulted in more frequent detection of Lost River Suckers that moved rapidly among antennas. Cessation of movement was interpreted as death.

Four hundred and forty juvenile Lost River Suckers were selected for PIT tagging on May 31, 2019, from the SARP rearing facility. Lost River Suckers were randomly assigned to one of four tanks until each had 110 individuals. Fish were sedated with tricaine methanesulfonate (MS-222) and PIT-tagged using MK25 implant guns and 12-mm, 134.2-kHz, full duplex PIT tags in preloaded needles (Biomark®, Inc. Boise, Idaho) Tags were implanted along the ventral surface in an anterior-to-posterior direction to keep the needle and tag away from sensitive organs such as the heart (Burdick 2011). After tagging, fish recovered in circular tanks for one month where they were monitored daily for tag loss and mortality until they were moved to net pens. Less than 5% died or lost their tag following tagging in each group (Table 1). Surviving fish that retained their PIT tags were transported to net pens on July 1, 2019, in coolers equipped with aerators and maintained at 18–20°C. Fish from each of the four tanks were placed into four net pens (MDN oxygen, MDN control, FB oxygen, and FB control). Fish were acclimated to lake conditions by slowly mixing lake water with water in the coolers over a period of 45 to 60 min, measured to standard length, and scanned for their PIT tag prior to release into net pens.

Hourly dissolved oxygen (DO) concentration, water temperature, and pH measurements were recorded by two YSI EXO2 sonde with an optical DO probe (YSI, Inc., Yellow Springs, Ohio) within each net pen. Sondes were deployed in a vertical position with the probe ends approximately 0.3 m above the benthos and 0.3 m below the surface. They were retrieved, cleaned, and (or) replaced regularly with a laboratory-calibrated replacement sonde according to USGS protocol (Wagner et al. 2000). These instruments were deployed on July 1, 2019 and retrieved between September 23–25, 2019 at all four net pens. Because of a data server crash, all water quality data is missing for FB oxygen from July 19 (13:00) to July 24 (10:00). Water-quality spot checks and profiles were conducted with a freshly calibrated sonde to ensure the Water Spikes were working correctly and to determine the spatial extent of oxygenation within the net pens. Spot checks of dissolved oxygen concentration were conducted weekly at the outflow of the Water Spikes. Horizontal profiles near the benthos were conducted weekly to ensure that highly oxygenated water was mixing throughout the net pen.

We collected weekly filtered water samples 0.3 m above the bottom of the net pen for total ammonia that followed established USGS protocols (USGS 2018). Samples were analyzed at the Sprague River Water Quality Lab (Chiloquin, Oregon). Un-ionized ammonia concentrations were derived using equations developed by Emerson et al. (1975) based on ammonia, pH, and temperature measurements taken at each site at the time of sampling. We compared the frequency among net pens where un-ionized ammonia concentrations exceeded the threshold for gill lamellae swelling (0.2 mg/L; Lease et al. 2003), and the median lethal 96-h (0.78 mg/L) and 24-h (1.02 mg/L) concentrations for juvenile Lost River Suckers (Saiki et al. 1999). Given that no effects on juvenile Lost River Suckers exposed un-ionized ammonia concentrations less than 0.2 mg/L (Lease et al. 2003), we considered levels below this threshold benign.

Dissolved oxygen concentration (mg/L) values were plotted by sonde depth, net pen, and treatment. Oxygen levels within each net pen was compared to three laboratory-determined lethal thresholds for imperiled juvenile Lost River Suckers in the Upper Klamath Basin: 1) total mortality within an hour (<1 mg/L, Stone et al. 2017); 2) median lethal level within 24 hours (1.34 mg/L, Saiki et al. 1999); and 3) the lowest level at which any juvenile sucker mortality was reported within 14 d (2.1 mg/L, Meyer and Hansen 2002). All of these thresholds were determined at a static temperature of 22°C, whereas temperature fluctuated in our study. Using dissolved oxygen concentration, temperature, and an altitude correction factor (Benson and Krause 1980, 1984; Rounds 2011), we calculated the percent oxygen saturation ≥100%. We summarized periods of supersaturation because we assumed that oxygen conditions for fish could not be improved after water was fully saturated with oxygen. We noted periods in each net pen when the entire water column (upper and lower sondes) reached of the aforementioned dissolved oxygen concentration thresholds.

Water temperature and pH were evaluated as water-quality indicators, including mean, minimum, and maximum of daily water temperatures and pH by sonde depth, net pen and treatment. We summarized the number of hours that the whole water column was at or exceeded 22°C, the lowest mean temperature that resulted in death of juvenile Lost River Suckers at 27 d in laboratory trials (Martin et al. 2021). We summarized the number of hourly and whole water column measurements in which pH exceeded 10, a level when juvenile Lost River Suckers have shown erratic swimming behavior, and 10.3, the 96-h median lethal concentration (Saiki et al. 1999).

After fish were released into net pens, subsequent genetic analysis (USFWS 2023) of a small (2 mm²) sample of tissue from the upper caudal fin indicated that 97.0% of our study animals were juvenile Lost River Suckers. A total of nine Shortnose Suckers, three hybrid suckers (Lost River Sucker, Shortnose Sucker, and/or Klamath Largescale Sucker Catostomus synderi), and one fish with no genetics sample were removed from further analyses. The death date was assigned based on cessation of movement of PIT-tagged juvenile Lost River Suckers among antennas. In some cases, fish movement between antennas slowed before stopping completely, making the timing of death only determinable to the day. In other cases, movement stopped suddenly, and the hour of death could be determined. Net pens were visited 6 d a week to check for moribund fish. Dead and floating individuals were removed, scanned for a PIT tag, and measured to standard length. The timing of death, as determined by remote detections, was corroborated for most fish by collecting the bodies within a day or two. We tabulated number of fish that were released in each net pen, number of fish that died within 7 d of release, confirmed deaths (bodies recovered), remote deaths (bodies not recovered), and survivors until the end of the study.

To estimate survival of Lost River Suckers in each of the four net pens, we applied the nonparametric Kaplan-Meier estimator (Kleinbaum and Klein 2012) using the survival package (Therneau 2024) in R (R Core Team 2022). We assumed that all individuals had independent fates, and that fish that lived to the end of study survived at the same rate as those that died during the study (Kleinbaum and Klein 2012). We assumed that mortality in the net pens was not caused by treatment of fish prior to being placed in the net pens. To ensure that our assessment of survival was unbiased by potential transport related stress, we only used movement data collected at least 7 d after fish were introduced to the net pens. An important assumption of our analysis is that we were able to correctly identify the day of death for each fish. This assumption would have been violated if PIT tags were lost, as this would indicate fish were dead. All fish that died during our study retained their PIT tags, similar to related studies that found less than 1% tag loss for PIT-tagged Lost River Sucker and Shortnose Sucker (Hewitt et al. 2018). We also assumed no fish escaped from the net pens because this would result in false death date assignments. We found no evidence of holes in net pens and most fish were detected multiple times on various antennas in a day. We may have erroneously assigned an early death date to fish that stopped moving for a day or more before death. This would not cause bias in the overall estimated survival rates, but it could affect the timing of estimates mortality by a few days. Median duration of survival among net pens was compared using a chi-squared test and plotting 95% confidence intervals (CI). We plotted survival by treatment for each site and compared it to dissolved oxygen concentrations recorded at the top and bottom sondes.

To compare growth of Lost River Suckers between oxygenated and control net pens, all individuals were measured to standard length at the beginning of the study on July 1, 2019, survivors were measured at the end of the study between September 23 and September 25, and fish that died were measured upon recovery. Using notched box plots (version 3.5.1; R Core Team, 2022) we confirmed that juvenile Lost River Suckers had similar starting standard lengths in all net pens. A two-sample t-test was used to compare mean change in standard length ± standard deviation (SD) between treatment groups at each site and between sites for all treatments. To determine if Lost River Suckers that died had differences in growth compared to individuals that survived the entire study, we compared length of fish that survived to those that died in each treatment and site.

Site-specific differences in dissolved oxygen concentration were recorded between oxygen and control net pens (Burdick et al. 2023). Although dissolved oxygen concentration in the outlet of the Water Spikes was always > 90% during spot checks, it did not always result in greater dissolved oxygen concentration throughout the net pens. Mean (± SD) dissolved oxygen concentration was slightly higher in the oxygenated net pen than the control net pen at FB at both sonde locations (lower sondes: FB oxygen 8.0 ± 3.3 mg/L compared to FB control 7.0 ± 3.4 mg/L, Upper sondes: FB oxygen 8.9 ± 3.0 mg/L compared to FB control 8.5 ± 4.0 mg/L, Figure 3). At the MDN lower sonde location, mean (± SD) dissolved oxygen concentration was slightly higher in the oxygenated net pen (6.1 ± 1.9) than in the control net pen (6.0 ± 1.9; Figure 3). At the MDN upper sonde location, mean (± SD) dissolved oxygen concentration was slightly lower in the oxygenated net pen (8.3 ± 2.1) than the control net pen (8.6 ± 2.7; Figure 3). The ability of the Water Spikes to increase dissolved oxygen concentration in the oxygenated net pens relative to controls was diminished when the ambient levels were high (Figure 3). When ambient oxygen saturation was ≤100%, Water Spikes increased dissolved oxygen concentration most frequently in the FB lower water column, FB upper water column, and MDN upper water column (Figure 3; Table 2).

Dissolved oxygenation concentrations varied by depth, site, and treatment. Supersaturation was more prevalent than hypoxia at the sonde location with the most extreme conditions of each. Consecutive hours of supersaturation at the upper sonde (range of maximum consecutive hours 66–162) lasted longer than those of hypoxia at the lower sonde (range of maximum consecutive hours 0–24 hours), except for at MDN lower sondes (Table 2; Burdick et al. 2023). Hypoxia below aforementioned lethal thresholds was comparatively less frequent and varied among net pens (Table 2). Oxygenated net pens had a lower frequency of dissolved oxygen concentration values below lethal thresholds at all sites and depths except at the MDN upper water column, where oxygen levels always remained above lethal thresholds in both control and treated net pens (Table 2).

Water Spikes also decreased the duration of dissolved oxygen concentration below thresholds throughout the entire water column. Although there were only 8 hours when the entire water column was below 1.0 mg/L dissolved oxygen at FB control (July 16, 17, and September 15 on Figure 4B and C), there were no times when the entire water column at FB oxygen was below this level (Table 2; Figure 4). The FB oxygen net pen also experienced fewer occasions when the entire water column was less than or equal to 1.34 mg/L and 2.1 mg/L dissolved oxygen compared to FB control (Figure 4). There were no occasions at MDN when the entire water column was below any dissolved oxygen concentration threshold for both the control and oxygen net pens (Table 2; Figure 5).

Mean daily temperature, pH, and un-ionized ammonia were similar among net pens. For temperature and pH, greater diel fluctuations were observed at FB compared to MDN (Figures 6 and 7; Burdick et al. 2023). Water temperatures that exceeded 22°C were more frequent at FB compared to MDN (Figure 6). The entire water column was at or above 22°C for 570 hours (range: 1 to 140 consecutive h), 545 hours (range: 1 to 93 consecutive h), 140 hours (range: 1 to 101 consecutive h), and 155 (range: 1 to 118 consecutive h) h at FB control, FB oxygen, MDN control, and MDN oxygen, respectively. The pH exceeded 10.0 and 10.3 more frequently at FB compared to MDN (Figure 7). The entire water column exceeded a pH of 10.0 at FB control for 32 hours (range: 1 to 8 consecutive h) and at FB oxygen for 13 h (range:1 to 5 consecutive h), and never at MDN. The pH did not exceed 10.3 throughout the entire water column in any of the net pens. Un-ionized ammonia never exceeded median lethal concentrations (0.78 mg NH₃ – N/L) in any samples. At MDN, it only exceeded 0.2 mg/L on August 27 at MDN when it was 0.55 mg/L in the control net pen and 0.46 mg/L in the oxygenated net pen (Burdick et al. 2023). There were no occurrences when un-ionized ammonia exceeded 0.2 mg/L at FB net pens.

A greater proportion of juvenile Lost River Suckers survived to the end of the study at FB oxygen (0.73) than FB control (0.47), whereas a similar proportion survived at MDN oxygen (0.91) and MDN control (0.93, Table 1; Burdick et al. 2023). Based on estimates with nonoverlapping 95% Confidence Intervals (CI) at FB, survival was higher in the control net pen than the oxygen net pen between August 14 and September 15. However, between September 16 and the study’s conclusion, survival was 0.72 CI (0.64, 0.82) in the oxygen net pen and 0.48 CI (0.39, 0.58) in the control net pen (Figure 4). On September 15, hypoxia throughout the water column coincided with the mortality of 48 fish at the FB control net pen (Figure 4). Survival declined at a slow but steady rate in the FB oxygen net pen from August 14, 2019 to the end of the study. At the conclusion of the study, survival probabilities at MDN were similar between the control 0.94 CI (0.89, 0.98) and oxygen 0.93 CI (0.88, 0.98) net pens (Figure 5). Mortality occurred on 6 d at MDN control and 8 d at MDN oxygen (Figure 5). No more than one fish died per day at MDN control and oxygen.

Although some fish died at times when dissolved oxygen concentration was >1.0 mg/L (Figure 4 and Figure 5), times when dissolved oxygen concentration was <1.0 mg/L throughout the entire water column at FB control corresponded with a significant mortality event (Burdick et al. 2023). There were 6 total h on July 16 and 17 (two consecutive 3-hour periods, 06:00–08:00 and 0:00–02:00) when the entire water column was below the 1 mg/L (0.50 – 0.86) at the FB control net pen and no mortality was observed. Hourly dissolved oxygen concentrations at FB control were below the lethal level of 1.0 mg/L (range = 0.46–0.61) throughout the entire water column for 2 hours on September 15 (03:00 and 04:00; Figure 8). In the FB control net pen, most of the fish that died during this September 15 event (44 of 48 or 92%) ceased movement sometime between 03:01 and 05:03. In contrast, dissolved oxygen concentrations were never < 2.1 mg/L in September for the FB oxygen net pen. Hourly dissolved oxygen concentrations at the FB oxygen net pen were low, particularly at the lower sonde in the early morning on September 15, although they never fell below the stress threshold of 2.1 mg/L at the upper sonde (Figure 8).

At the beginning of the study, average standard lengths (mean ± SD) were similar (95% confidence intervals for medians overlapped for all sites) across all sites (118.4 ± 13.5 mm for FB oxygen, 119.4 ± 12.8 mm for FB control, 118.7 ± 12.4 mm for MDN oxygen, and 120.8 ± 15.2 mm for MDN control; Burdick et al. 2023). For fish that survived the entire study, growth rates (mean/day ± SD) were faster at FB oxygen (0.62 ± 0.10 mm) than FB control (0.58 ± 0.09 mm; p = 0.024) and faster at MDN control (0.82 ± 0.12 mm) compared to MDN oxygen (0.76 ± 0.13 mm; p = 0.0047). When comparing locations with pooled treatments, MDN had a faster daily growth (0.79 ± 0.13 mm) compared to FB (0.60 ± 0.10 mm; p < 0.001).

The number of hours with lethally low dissolved oxygen for juvenile Lost River Suckers in net pens was most affected by the location of the net pen, but also affected by the presence of a Water Spike. Oxygen concentration below lethal thresholds was less frequent at MDN than at FB. This finding is corroborated by previous studies, which showed greater dissolved oxygen concentration and survival of captive juvenile Lost River Suckers at MDN and another site in the middle of Upper Klamath Lake near Rattlesnake Point (RPT), than at FB (Hereford et al. 2019; Burdick et al. 2020b). The differences in dissolved oxygen concentration among sites may be associated with the decay of cyanobacteria that can accumulate on the northwest shoreline where the FB net pen was located. High biological oxygen demand in the sediments during periods of cyanobacterial decomposition, have a proportionally greater effect on smaller volumes of water, such as in shallow locations (Pace and Prairie 2005). Therefore, water depth and volume also may have played a role in the difference in dissolved oxygen concentration and pH among locations in Upper Klamath Lake.

Temperature differences between the two net pen locations are unlikely to be the cause of differential mortality. Water temperatures were slightly higher, more variable, and exceeded the stress threshold of 22°C more frequently at FB than at MDN. We selected the 22°C threshold because it was the lowest temperature associated with the mortality of juvenile Lost River Suckers (Martin et al. 2021). However, Lost River Suckers only died in laboratory studies when infected with a gill parasite and held at a constant 22°C for 27 d. The longest duration that water temperature exceeded 22°C in our study was 4 d at FB (Figure 6). Median lethal water temperatures over 96-hour periods are much higher (30.5°C; Saiki et al. 1999) and were not recorded in our study. Therefore, we consider the highest water temperatures recorded in our study to be a mild stressor for Lost River Suckers.

Our oxygenation system appeared to increase juvenile Lost River Sucker survival by elevating dissolved oxygen concentration above lethal levels during a documented hypoxic event at FB. Oxygenating the FB oxygen net pen may have improved survival for juvenile Lost River Suckers by preventing the occurrence of lethal hypoxic conditions throughout the entire water column or by providing a refuge from hypoxia at the bottom. The mass mortality event following a lethal hypoxic event for ≤2 h that occurred at the FB control net pen in September did not occur at FB oxygen where dissolved oxygen concentration remained above 2.1 mg/L during the same time period. Although this study does not establish a lethal oxygen threshold, the significant mortality event in the FB control net pen was documented in September, occurring only when dissolved oxygen concentration throughout the entire water column fell below 1.0 mg/L, as occurred in previous studies (Martin and Saiki 1999; Stone et al. 2017). Hypoxia was associated with high mortality in September but not in July, indicating in addition to hypoxia that months of exposure to other stressful conditions also may have played a role in mortality. White foam that formed on the surface of the water in the oxygen net pens may have indicated cyanobacterial mortality, but were not accompanied by decreases of pH typically associated with cyanobacterial death. Cyanobacteria in Upper Klamath Lake may release harmful compounds during cell senescence that also may have contributed to juvenile Lost River Sucker mortality (Burdick et al. 2020a).

We cannot be certain of the cause of gradual decline in survival at FB oxygen prior to the hypoxic event at FB control on September 15, but suspect it could be associated with unmeasured side effects of oxygenation at this location. Water temperatures were nearly identical between the two FB net pens and greater dissolved oxygen concentration in FB oxygen than FB control, was expected to increase not decrease survival. The Water Spikes caused turbulence and mixed the water column, but un-ionized ammonia was not elevated in oxygenated net pens compared to control ones.

Hypoxia can slow fish growth. Juvenile Lost River Suckers avoid benthic oxygen concentration ≤1.0 mg/L in net pens (Hereford et al. 2019), and in extreme cases exhibit aquatic surface respiration that is both energetically intensive and separates fish from their benthic prey (Foott et al. 2007; Roberts et al. 2009; He et al. 2015). Depending on the duration of hypoxic events throughout the entire water column, physiological adaptations such as air breathing (Chabot and Claireaux 2008), hyperventilation increasing oxygen transfer to blood and gill remodeling may help fish survive hypoxia. However, these adaptations have energetic costs that may affect growth (Mikheev et al. 2014; Chen et al. 2017). Under extreme conditions, fish may conserve oxygen by reducing movement, feeding, and growth. However, this in turn may result in an indirect cost such as inability to evade predators, cessation of feeding, and/or depleting anaerobic energy reserves (Chabot and Claireaux 2008; McBryan et al. 2013). Benthic hypoxia may limit access to prey resources on the bottom of the lake. The effects of hypoxia on juvenile Lost River Sucker prey availability and consumption warrant further study.

Growth and survival of juvenile Lost River Suckers in our study varied among net pen sites. Similar findings were documented in previous studies, where growth was faster at MDN than FB and even greater growth at RPT where dissolved oxygen concentration was not recorded below the lethal threshold (Burdick et al. 2020b; Hereford et al. 2019). Hypoxia and high temperatures in our study were more severe and frequent at FB than MDN, which may have increased energetic demands at FB that slowed growth and decreased survival relative to MDN. Differences in prey, not quantified in this study, also may have played a role in growth and survival. However, good water quality may not always equate to the fastest growth rates. Caldwell et al. (2023) found the slowest growth for the closely related Shortnose Sucker at net pens in Pelican Bay, a known cold-water refuge in Upper Klamath Lake, as compared to RPT and the Williamson River mouth. They also found that growth of Shortnose Suckers was positively correlated with survival (Caldwell et al. 2023). Therefore, choosing sites that consistently result in faster growth and larger fish at release could enhance survival and recruitment, as seen with the closely related June Suckers (Billman et al. 2011; Fonken et al. 2023). This information could help SARP managers decide where to place net pens in Upper Klamath Lake for juvenile Lost River Sucker and Shortnose Sucker acclimation and release.

Oxygenation did not have a measurable effect on pH in our study. We expected turbulence caused by the oxygenation system to disrupt cyanobacteria buoyancy causing death. Field staff observed accumulation of white foam on the water surface that was presumed to indicate phytoplankton death. This was only seen at oxygen, but not at control sites. However, pH was similar between oxygenated and control sites, indicating that cyanobacteria were still photosynthesizing despite physical disruption. Fortunately, pH high enough to cause erratic swimming behavior (10.3) only occurred briefly (<8 consecutive h) and never at both the upper and lower sondes in the same net pen simultaneously.

Although our oxygenation system did not cause oxygen supersaturation, naturally occurring hyperoxia was much more frequent than hypoxia in our study. Hyperoxia is generally benign or even beneficial to fish and at high temperatures it can increase aerobic capacity, improve heart function, moderate stress during anaerobic activity, and increase thermal tolerance limits (McArley et al. 2021). However, it also has the potential to increase oxidative stress under warm conditions (McArley et al. 2021). Hypoventilation from hyperoxia increases retention of metabolically produced CO₂ and thus acidifies blood. However, because naturally in Upper Klamath Lake where hyperoxia coincides with high pH, this effect is likely reduced. Although most studies report growth in fish to be improved or unaffected by hyperoxia, it is not clear if daytime hyperoxia makes fish more tolerant of night time hypoxia (McArley et al. 2021). Some evidence indicates that mortality of fish challenged with pathogens increases with hyperoxia, possibly because additional oxygen benefits pathogens (Caldwell and Hinshaw 1995).

Our study corroborated previous research that indicated that the effectiveness of oxygenation is limited in large, unstratified, shallow, warm hypereutrophic lakes (Liboriussen et al. 2009). There are several reasons why oxygenation in Upper Klamath Lake, characterized by these same conditions, was only moderately effective on a small spatial scale in net pens. In contrast to deep lakes where oxygenation systems inject oxygen below the hypolimnion and stratification retains it (Preece et al. 2019), shallow polymictic lakes like Upper Klamath Lake lack the stratification that helps maintain benthic oxygen levels. Water Spikes may be more effective in very shallow water (less than about 1 m deep) because of an added oxygen-to-water ratio. For example, when ambient water unsaturated with oxygen (p≪0.01 paired t test) and the water depths ranged from 1.1–1.7 m, the hourly dissolved oxygen concentration was 1.2 ± 1.6 mg/L (mean ± SD) greater in the oxygen net pen than the control net pen at FB lower and 0.9 ± 1.3 mg/L at FB upper. A smaller increase in dissolved oxygen concentration was associated with Water Spikes at MDN oxygen at the lower sonde (0.1 ± 0.6 mg/L, p≪0.01, paired t test) when ambient water was unsaturated with oxygen and which had higher water depths that ranged from 3.1–4.2 m. Oxygen concentration was not significantly altered by the Water Spike at the MDN oxygen upper sonde.

Because warm water holds less oxygen than cool water, oxygenation is minimally effective at water temperatures greater than about 15°C (Moore et al. 2012), which were common throughout our study. Despite our system delivering 7,200 liters per day of oxygen, the Water Spike used in our study may have been undersized for the high sediment oxygen demand of Upper Klamath Lake. The demand can spike as high as 9 g/m²/day in localized areas following cyanobacterial die-off events that are common during summer in Upper Klamath Lake (Wood 2001). Future use of oxygenation systems in the lake should consider scaling the systems to exceed the benthic demand. Finally, given that Upper Klamath Lake is hypereutrophic, oxygenation may have increased oxygen demand by inducing a higher rate of decomposition and bacterial respiration, thus resulting in an effect opposite of what was desired (Moore et al. 2012; Gerling et al. 2014).

Our conclusions are limited by the lack of replication among treatments at the site level. We employed a 2 × 2 experimental design in which each site and each treatment were repeated only twice and never under the exact same conditions due to cost constraints. Consequently, we cannot conclusively determine whether oxygenation or location were the driving factors in the survival of Lost River Suckers. Nonetheless, our results highlight the importance of careful placement of net pens and the use of supplemental oxygen in net pen rearing of imperiled Lost River Suckers in Upper Klamath Lake.

There are several considerations that should improve the effectiveness of oxygenated net pens for fish rearing within warm, shallow hypereutrophic lakes. Using less permeable material would reduce circulation with the lake and increase the efficacy of oxygenation, but also reduces natural recruitment of prey. Oxygenation systems in this study were useless when water was super-saturated with oxygen which occurred between 16–60% of the time depending on net pen and sonde location. Therefore, substantial reductions in the energy expended powering oxygenation systems could be accomplished by running the system only during periods of low oxygen, especially late in the season. For example, using oxygenation later during the cyanobacterial cycles in summer (July–October) may be more beneficial to Lost River Sucker survival because that is when low dissolved oxygen levels contributed to the mass mortality event at FB control. Automating the oxygenation systems in remote net pens to activate when oxygen levels approach threshold levels could enhance efficiencies of these systems. To eliminate carbon emissions and risk of fuel spills from gasoline generators, solar panels could power the oxygenation systems in locations like Upper Klamath Lake with over 300 d of sun annually.

We adapted an oxygenation technology initially developed for salmonid (Oncorhynchus spp.) hatcheries, with cool temperatures and confined raceways, to enhance the survival of imperiled juvenile Lost River Suckers in net pens within a warm, hypereutrophic lake. The benefit of this technology was only apparent during very extreme hypoxia events in a small net pen. Specifically, at one of our two net pen locations in Upper Klamath Lake, fish in our study showed significantly higher survival rates during a brief episode of severe ambient hypoxia (dissolved oxygen concentration ≤0.61 mg/L) when provided supplemental oxygen. Given the robust hypoxia tolerance of juvenile Lost River Suckers, we were unable to demonstrate a benefit of supplemental oxygen when ambient dissolved oxygen concentration was >1.0 mg/L. Considering the risk of high mortality during occasional extreme hypoxia in Upper Klamath Lake (<1.0 mg/L; Burdick et al., 2020) even at sites with relatively infrequent extreme hypoxia events, oxygenation is a critical component of soft release strategies in Upper Klamath Lake to improve post-stocking survival for juvenile SARP Lost River Sucker and Shortnose Sucker. Improvements in efficiencies of oxygenation systems will be critical for maximizing the survival in net pens for imperiled fish in hypereutrophic systems.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Data S1. Data were collected to assess survival of Lost River Suckers Deltistes luxatus in net pens with and without addition of Oxygen in Upper Klamath Lake, Oregon. To determine how a low-cost oxygenation system affected survival of captively reared fish, we introduced Passive Integrated Transponder (PIT) tagged juvenile Lost River Suckers into four net pens in Upper Klamath Lake. The fish originated from the U.S. Fish and Wildlife Service's Sucker Assisted Rearing Program in Klamath Falls, Oregon which rears suckers collected as larvae in Upper Klamath Lake including Lost River Sucker (Deltistes luxatus), Shortnose Sucker (Chasmistes brevirostris), and Klamath Largescale Sucker (Catostomus synderi). Suckers were monitored continuously by PIT antennas. These data contain all remote detections from the PIT antennas, all physical captures, and the water quality data associated with each net pen. Data were collected from May to September of 2019.

Available: https://doi.org/10.3996/JFWM-24-011.S1 (44.22 MB) and https://www.sciencebase.gov/catalog/item/6530389bd34edd15305a9ee2 (December 2023)

Reference S1. Bart RJ, Kelsey CM, Burdick SM, Hoy MS, Ostberg CO. 2021. Growth, survival, and cohort formation of juvenile Lost River (Deltistes luxatus) and Shortnose Suckers (Chasmistes brevirostris) in Upper Klamath Lake, Oregon, and Clear Lake Reservoir, California—2019 Monitoring Report: U.S. Geological Survey Open-File Report 2021–1119.

Available: https://doi.org/10.3996/JFWM-24-011.S2 (2.33 MB PDF) and https://pubs.usgs.gov/of/2022/1099/ofr20221099.pdf (June 2024)

Reference S2. Bortleson GC, Fretwell MO. 1993. A review of possible causes of nutrient enrichment and decline of endangered sucker populations in Upper Klamath Lake, Oregon: U.S. Geological Survey Water-Resources Investigations Report 93-4087.

Available: https://doi.org/10.3996/JFWM-24-011.S3 (853 KB PDF) and https://pubs.usgs.gov/wri/1993/4087/report.pdf (June 2024)

Reference S3. Burdick SM, Conway CM, Ostberg CO, Bart RJ, Elliott EG. 2021. Survival and growth of suckers in mesocosms at three locations within Upper Klamath Lake, Oregon, 2018: U.S. Geological Survey Open-File Report 2021–1036.

Available: https://doi.org/10.3996/JFWM-24-011.S4 (1.62 MB PDF) and https://pubs.usgs.gov/of/2021/1036/ofr20211036.pdf (June 2024)

Reference S4. Burdick SM, Martin BA. 2017. Inter-annual variability in apparent relative production, survival, and growth of juvenile Lost River and Shortnose Suckers in Upper Klamath Lake, Oregon, 2001–15: U.S. Geological Survey Open-File Report 2017–1069.

Available: https://doi.org/10.3996/JFWM-24-011.S5 (3.67 MB PDF) and https://pubs.usgs.gov/of/2017/1069/ofr20171069.pdf (June 2024)

Reference S5. Eldridge SLC, Wood TM, Echols KR. 2012a. Spatial and temporal dynamics of cyanotoxins and their relation to other water quality variables in Upper Klamath Lake, Oregon, 2007–09. U.S. Geological Survey Scientific Investigations Report 2012–5069.

Available: https://doi.org/10.3996/JFWM-24-011.S6 (2.56 MB PDF) and https://pubs.usgs.gov/sir/2012/5069/pdf/sir20125069.pdf (June 2024)

References S6. Eldridge DB, Eldridge SLC, Schenk LN, Tanner TQ, Wood TM. 2012b. Water-quality data from Upper Klamath and Agency Lakes, Oregon, 2009–10, U.S. Geological Survey Open-File Report 2012–1142.

Available: https://doi.org/10.3996/JFWM-24-011.S7 (4.83 MB PDF) and https://pubs.usgs.gov/of/2012/1142/pdf/ofr20121142.pdf (June 2024)

References S7. [ESA] US Endangered Species Act of 1973, as amended, Pub. L. No. 93-205, 87 Stat. 884 (Dec 28, 1973). Available at: Endangered Species Act of 1973, As Amended through the 108th Congress (fws.gov) (June 2024)

Available: https://doi.org/10.3996/JFWM-24-011.S8 (308 KB PDF)

Reference S8. Foott JS, Stone R, Fogerty R. 2007. Lack of Disease Response in Juvenile Upper Klamath Lake Suckers (age-0+) to Adverse Water Quality Conditions – pilot study August 2007. U. S. Fish and Wildlife Service, Anderson, CA.

Available: https://doi.org/10.3996/JFWM-24-011.S9 (412 KB PDF)

Reference S9. Furr DW. 2018. San Juan River Razorback Sucker (Xyrauchen texanus) and Colorado Pikeminnow (Ptychocheilus lucius) population augmentation: 2017. Final Report to the U.S. Fish and Wildlife Service, San Juan River Basin Recovery Implementation Program, Albuquerque, NM.

Available: https://coloradoriverrecovery.org/sj/wp-content/uploads/sites/3/2022/03/prpstk_Augmentation_CPM_RBS_San_Juan_River_2013_OCR.pdf (June 2024)

Reference S10. Golden ME, Holden PB, Albrecht B. 2006. Retention, growth, and habitat use of stocked Colorado Pikeminnow stocked as age-0 Fish in the San Juan River from 2002–2005: final summary report. Final report to the U.S. Fish and Wildlife Service, San Juan River Basin Recovery Implementation Program Biology Committee, Albuquerque, NM.

Available: https://coloradoriverrecovery.org/sj/wp-content/uploads/sites/3/2022/04/sml_Retention_growth_habitat_use_age-0_CPM_2003-2004_OCR.pdf (June 2024)

References S11. Hereford DM, Conway CM, Burdick SM, Elliott DC, Perry TM, Dolan-Caret A, Harris AC. 2019. Assessing causes of mortality for endangered juvenile Lost River Suckers (Deltistes luxatus) in mesocosms in Upper Klamath Lake, south-central Oregon, 2016: U.S. Geological Survey Open -File Report 2019-1006.

Available: https://pubs.usgs.gov/of/2019/1006/ofr20191006.pdf (June 2024)

References S12. Hewitt DA, Janney EC, Hayes BS, Harris AC. 2018. Status and trends of adult Lost River (Deltistes luxatus) and Shortnose (Chasmistes brevirostris) Sucker populations in Upper Klamath Lake, Oregon, 2017: U.S. Geological Survey Open-File Report 2018-1064.

Available: https://pubs.er.usgs.gov/publication/ofr20181064 (June 2024)

References S13. Kuwabara JS, Topping BR, Carter JL, Carlson RA, Parchaso F, Fend SV, Stauffer-Olsen N, Manning AJ, Land JM. 2016. Benthic processes affecting contaminant transport in Upper Klamath Lake, Oregon: U.S. Geological Survey Open-File Report 2016–1175.

Available: https://pubs.er.usgs.gov/publication/ofr20161175 (June 2024)

Reference S14. Stone R, Jacobs J, Som NA, Foott JS, Phillips B, Ross J, Taylor D, Tyler T. 2017. 2017. Lost River Sucker fry sentinel survival in Upper Klamath Lake (July–September 2015). U.S. Fish and Wildlife Service, California–Nevada Fish Health Center, Anderson, CA.

Available: https://www.fws.gov/sites/default/files/documents/UKL%20Lost%20River%20sucker%20fry%20sentinel%20survival%20report%20FINAL%2003May2017.pdf (June 2024)

Reference S15. [USFWS] U.S. Fish and Wildlife Service. 2019. Species status assessment for the endangered Lost River Sucker and Shortnose Sucker. Klamath Falls, Oregon: U.S. Fish and Wildlife Service, Klamath Falls, OR.

Available: https://doi.org/10.3996/JFWM-24-011.S10 (6,136 KB PDF)

References S16. [USFWS] U.S. Fish and Wildlife Service. 2023. An improved genetic tool for species identification of Klamath River basin suckers. U.S. Fish and Wildlife Service Abernathy Fish Technology Center, Longview, WA and Klamath Falls Fish and Wildlife Office, Klamath Falls, OR.

Available: https://doi.org/10.3996/JFWM-24-011.S11 (2,145 KB PDF)

Reference S17. [USGS] U.S. Geological Survey. 2018. U.S. Geological Survey Techniques and Methods. Editors USGS.

Available: https://doi.org/10.3133/tm9A0 (June 2024)

Reference S18. Wagner RJ, Mattraw HC, Ritz GF, Smith BA. Guidelines and standard procedures for continuous water-quality monitors: site selection, field operation, calibration, record computation, and reporting. U.S. Geological Survey, Water-Resources Investigation Report 00-4252, Reston, VA.

Available: https://pubs.usgs.gov/tm/2006/tm1D3/ (June 2024)

Reference S19. Wood TM. 2001. Sediment oxygen demand in Upper Klamath and Agency lakes, Oregon. U.S. Geological Survey, Scientific Investigations Report 2001-4080.

Available: https://pubs.usgs.gov/wri/2001/4080/wri01-4080.pdf (June 2024)

Reference S20. Wood TM, Cheng RT, Gartner JW, Hoilman GR, Lindenberg MK, Wellman RE. 2008. Modeling hydrodynamics and heat transport in Upper Klamath Lake, Oregon, and implications for water quality. U.S. Geological Survey, Scientific Investigations Report 2008–5076.

Available: at https://pubs.usgs.gov/sir/2008/5076/pdf/sir20085076.pdf (June 2024)

Caylen Kelsey (U.S. Geological Survey) assisted with data management. Ross Clyma, Carolyn Malecha, Rachael Paul-Wilson, and Brian Hayes of U.S Geological Survey conducted field work. Joel Ophoff, Zach Tiemann, and Michelle Jackson of U.S. Fish and Wildlife Service reared Lost River Suckers from larvae and assisted with PIT tagging. The associate editor and three anonymous reviewers provided comments on this manuscript. Funding was provided by the U.S. Bureau of Reclamation as part of its mission to manage, develop, and protect water and related resources in an environmentally and economically sound manner in the interest of the American public.

Any use of trade, product, website, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.