In California’s Central Valley, studies have found that increased streamflow can improve the survival of imperiled juvenile salmon populations during their oceanward migration. However, these studies have not explored the potential nonlinearities between flow and survival, giving resource managers the difficult task of designing flows intended to help salmon without clear guidance on flow targets. A recent study analyzed salmon migration survival data from over 2400 acoustic-tagged juvenile Chinook salmon spanning differing water years and year types to extract actionable information on the flow-survival relationship in the Sacramento River.
At the 2022 Bay-Delta Science Conference, Cyril Michel, Assistant Project Scientist with National Marine Fishery Service’s Southwest Fisheries Science Center and the University of California Santa Cruz, presented the study results.
Over the past decade, many studies have used acoustic telemetry to determine what environmental covariates best predict the survival of out-migrating salmon in the Sacramento River. Those studies have consistently found that flow during out-migration is one of the best predictors of survival. Beyond that, other studies have also looked at what covariates best predict the year-to-year variability in overall escapement of salmon in the Central Valley; those studies have also found that flow during the rearing and out-migration period of salmon tends to be one of the best predictors of year-to-year variability and escapement.
“So it seems clear that flow during out-migration is a very important predictor of survival, and ultimate success of salmon populations in California, so it makes sense to do a deeper dive into this particular relationship,” said Mr. Michel. “A lot of these studies I just showed you looked at linear relationships between different environmental covariates in survival. And now that we know that flow-survival is one of the best, strongest relationships, we should look for the potential of nonlinearities in that relationship.”
For example, in the hypothetical plot on the slide, if there was a flow-survival threshold, or a flow value above which survival jumps up quickly, it could be a useful tool for management to maximize the use of water, a limited resource. So the researchers decided to explore if there were nonlinear relationships or a threshold.
The researchers selected the Sacramento River from the Deer Creek confluence down to the Feather River confluence as the region of interest. They chose this region because almost all salmon smolts in the Sacramento Valley transit through here, it tends to have very low flow during the late spring, and it’s an area that has high mortality.
They concentrated on the spring months when spring run and fall run out-migrate through the system. They used data from over 2400 acoustic tagged chinook salmon smolts from several studies spanning six different years.
They first took the data for the 2400 salmon and grouped them by the flows they experienced into five groupings. On the plot, the x-axis is the flow; the y-axis is survival for each one of the groups.
“What jumped out to us right away was there seems already to be strong evidence of a nonlinear relationship between flow-survival around 10,000-11,000 CFS where survival jumps from about 20% up to about 50%,” said Mr. Michel. “This looked very promising, but we wanted to do this more formally.”
The researchers next ran various mark-recapture survival models, allowing the flow survival relationship to take different shapes, such as a linear relationship, a log-linear, polynomial, cubic spline, and a simple threshold model. They then assessed the fit of the different models to the survival data using model selection. A summary table of results is shown on the slide below.
“What we found was the survival model that best fit the survival data was a threshold form; in particular, with three flow thresholds,” said Mr. Michel. “Those flow thresholds were 4300 CFS, 10,700 CFS, and 22,900 CFS as measured at the Wilkins Slough gauge, which is the lowest gauge located in the region of interest.”
The slide below shows the graphical representation of that best-fit model of all the different forms of the flow-survival relationship, the three threshold model. The x-axis shows the 4300, 10,700, and 22,900 CFS threshold, and the y-axis shows the predicted survival of fish experiencing flows above, in between or, beneath the different flow thresholds.
“What we see is for fish that experience flows beneath the 4300 flow thresholds, that survival is approximately 3%,” said Mr. Michel. “For fish that experience intermediate flows, they had survival of about 18 to 19%. And for fish that experience high flows, they had survival of approximately 50%. And finally, one interesting thing we found was for fish that had experienced very high flows, there was actually a slight drop in survival for those fish resulting in about 35% survival through the region of interest.”
“These are significant gains in survival,” he continued. “For example, the change in survival between the low flow and intermediate flow is 6.3 fold increase; between the intermediate flow and the high flow group, it’s a 2.7 fold increase. And overall, from the lowest survival rates to the highest survival rates, we saw a 16.9 fold increase in survival. So with all else being equal, if you were able to keep flows above a certain level versus below a certain level, you would see, for example, here, a 6.3 fold increase in the number of fish that make it out of the region of interest. That is pretty significant.”
What are the mechanisms to these thresholds? The researchers decided to take a deep dive into potential mechanisms. First, for the 4300 CFS threshold, the strongest evidence seemed to be for water temperature driving this threshold.
The plot shows the same survival data with the fish grouped into five temperature ranges based on the temperatures they experienced, not the flows.
“What we see is that the fish that experience the highest temperatures here, about higher than 19.5 degrees Celsius, are the fish that had by far the lowest survival – of almost zero,” said Mr. Michel. “Similarly, if we plot the flow that all of our 2400 fish experienced versus the respective temperatures they experienced, as measured at Wilkins Slough, what we see is the fish that had the lowest survival, the ones that had flows beneath 4300 CFS, shown on the x-axis with that vertical dashed line – those are the same fish that experienced almost strictly water temperatures above about 19.5 degrees Celsius. So essentially, these fish here are the same fish that experienced flows beneath 4300 CFS. So that is strong evidence that water temperatures are likely the driver here.”
Is this a persistent trend in the Central Valley? It does seem to be so, said Mr. Michel. Back in the 70s and 80s, Pat Brandes saw a similar relationship in the Delta survival of Chinook salmon using coded wire tags, where above 20 degrees Celsius survival crashed to almost zero. In another study currently in review, the results were similar in the Delta using acoustic tagged data that showed that above 19.5 degrees Celsius, survival goes to almost zero.
“Yet, the physiological limit is 25 degrees Celsius, so why are we seeing a 20 degree Celsius threshold,” said Mr. Michel. “Since this session is about temperatures, I decided to give some additional ideas as to why this 20 degrees Celsius threshold exists.”
In work currently under review, the researchers have been investigating the role of how the prey responds to temperature and how the predator response to temperatures might lead to nonlinear survival relationships.
The plot on the lower left shows how the maximum swimming performance of Chinook salmon relates to water temperature from a study by Lehman et al. in 2017.
“What we see is with higher water temperatures, small stream performance really declines,” said Mr. Michel. “At the same time, from data from a study I did, published in 2020, we saw that predation rates increase rapidly with increasing water temperature. So the plot on the left represents prey evasiveness, and the plot on the right represents predator activity; these are happening at the same time across the same ranges of temperatures. So you could imagine how ultimate predation and mortality really accelerates above certain water temperatures.”
“On top of that, there might exist some synergistic impacts of two different predators in the Delta: the largemouth bass and striped bass and how they react to different water temperatures. And ultimately, there seems to be a lot of evidence for a nonlinear temperature versus survival relationship as mediated through predation.”
Looking at the 10,700 CFS threshold, the researchers thought that one of the highest potential covariances that might be leading to this threshold was the travel time of the fish. The plot on the slide below shows flows at Wilkins Slough for the fish and their respective median travel time.
“What we see is for fish experiencing flows above 10,700 CFS, they have much shorter travel times than fish experiencing lower flows,” said Mr. Michel.
“Now looking at the travel time distributions (below) for not just the 5% bins but for all for fish broken out by the flow thresholds, we do see significantly faster travel times for fish experiencing flows above 10,700 CFS versus fished experiencing flows below 10,700 CFS.”
“It seems clear that 4300 CFS at Wilkens Slough should be considered as a low flow standard for chinook salmon smolts during the out-migration season,” Mr. Michel said. “Similarly, water temperatures above 20 degrees Celsius should be avoided at all costs when possible during out-migration season. Finally, 10,700 CFS as measured at Wilkens Slough could be considered a survival-increasing measure when the conditions allow it. And this could be done through the use of pulse flows.”
So in the same paper, they investigated the potential for using pulse flows. First, the researchers ran different scenarios for 2013-2016 and 2018; they next created alternative hydrographs that utilized the use of pulse flows and ensured minimum low flow thresholds above 4300 CFS. Then, to see how that compared to the actual hydrograph in those years, they used the number of out-migrants as counted from the Red Bluff river screw traps and applied the respective survival rates to the different thresholds to ultimately find out a cohort-wide survival estimate for each one of the years.
“What we see is we see a 55% to 132% increase in survival by utilizing these different flow thresholds without adding any water to the water budget that was used that year,” Mr. Michel said. “With just a modest addition to the water budgets, we saw increases of 79% to 330% in overall migration survival. And finally, even in years where the existing water budget did not allow for a pulse, namely 2014 and 2015, just maintaining the low flow standard at Wilkens Slough was enough to increase survival by 83 to 132%.”
Finally, the slide shows results from a different study that looked at predation rates in the Delta and used that to estimate the daily through-Delta survival of salmon based on these predation rates.
“This is day by day through-Delta survival through the spring period,” he said. “What we tend to see is in most springs, predicted survival salmon through the Delta does go to almost zero.”
How can pulse flows be useful? “The great thing about pulse flows is we believe they should also increase the number of out-migrants at any one time,” he said. “Many studies have shown when flow increases in the river, it usually triggers fish to out-migrate, so pulse flows can not only be used to trigger fish to out-migrate, but also give them the higher survival as they out-migrate.”
“So we could use tools such as climate forecasts to decide that it seems like there might be a good period coming up in the Delta of low air temperatures, and we could have a pulse flow occur here to ensure that fish are triggered out-migrate and also experience good conditions both in the river and in the Delta,” said Mr. Michel. “Then similarly, you could consider having a last chance pulse flow when you know from the climate models that water temperatures are about to get really warm in the river and in the Delta. And you could give them one last chance to out-migrate, have a pulse flow here to move fish through the system before the Delta gets too warm for salmon.”
QUESTION: Does the survival benefit calculation account for fry migrants or only smolts?
Mr. Michel: “The survival benefit calculation is based on rotary screw trap counts. I did not try to tease apart run or sizes; I essentially grabbed all of the rotary screw trap counts for the spring period. So it’s not necessarily a population-specific or life stage-specific estimate of out-migration survival; it’s more of a seasonal estimate. So for spring out-migrating fish of all different sizes and populations, those were the population gains as predicted by our survival model.”
QUESTION: Is your pulse flow hypothesis similar to that used to suggest pulse flows on the Stanislaus River and San Joaquin River?
Mr. Michel: “A lot of pulse flow work has been going on in the San Joaquin Basin, and I’m just starting to get up to speed on how those are implemented and the research behind it. So I can’t necessarily speak too much to that. But I would like to say that the Sacramento system is ripe for this kind of work too, it’s where most of the fish come from in California, and we have a lot of control over flow in that river. So I’m hoping we can move in that direction soon.”
QUESTION: If a 10,000 CFS flow pulse in the Sacramento River were diverted from the south Delta, would that provide a net benefit to salmon?
Mr. Michel: “There could be kind of potential win-win here where perhaps when a pulse flow occurs in the river, it’s not “lost” for operating purposes. Maybe some of it can be diverted once it arrives in the Delta. I think more research could likely help us figure that out. But it certainly would help survival in the river, which may offset any losses due to getting exported in the Delta. So I think that’s definitely something to look into. It’s always nice to have the potential for a win-win here.”
QUESTION: What are the bass densities in the region of interest?
Mr. Michel: “I don’t know if anyone has a good answer for that. It depends on the year; it depends on the time. Striped bass make a spawning run into the region of interest every year, and the timing of that spawning run and the size of that spawning run changes. There’s also resident striped bass and largemouth that seemed to occupy that region, but I don’t think anyone has a good idea on the abundances there, unfortunately, or the variation in the abundances.”
QUESTION: Have you have any ideas on how additional flows might be obtained, especially in critically dry years?
Mr. Michel: “As part of the new biological opinions and proposed actions for the Central Valley Project and State Water Project; there are pulse flows that are supposed to happen out of Keswick reservoir into the Sacramento River when storage is above 4 million acre-feet. So we’re making progress, and we’re hoping in a year soon, we will be able to implement a pulse flow. But, unfortunately, the formula for the acre-feet storage standard does not really allow for pulses to happen in dry and definitely not critically dry years, and those are likely the conditions under which salmon need the pulse flows the most. So I still think there’s room to improve on when we can potentially implement these pulse flows. But certainly being able to get one under our belts under any kind of year class would be good to learn more about if they work and how well they work, and that will help us implement them in the future.”
Cyril J. Michel, Jeremy J. Notch, Flora Cordoleani, Arnold J. Ammann, Eric M. Danner
Water is a fundamental resource in freshwater ecosystems, and streamflow plays a pivotal role in driving riverine ecology and biodiversity. Ecologically functional flows, managed hydrographs that are meant to reproduce the primary components of the natural hydrograph, are touted as a potential way forward to restore ecological functions of highly modified rivers, while also balancing human water needs. A major challenge in implementing functional flows will be establishing the shape of the managed hydrograph so as to optimize improvements to the ecosystem given the limited resources. Identifying the shape of the flow–biology relationship is thus critical for determining the environmental consequences of flow regulation.
In California’s Central Valley, studies have found that increased streamflow can improve survival of imperiled juvenile salmon populations during their oceanward migration. These studies have not explored the potential nonlinearities between flow and survival, giving resource managers the difficult task of designing flows intended to help salmon without clear guidance on flow targets. We used an information theoretic approach to analyze migration survival data from 2436 acoustic-tagged juvenile Chinook salmon from studies spanning differing water years (2013–2019) to extract actionable information on the flow–survival relationship.
This relationship was best described by a step function, with three flow thresholds that we defined as minimum (4259 cfs), historic mean (10,712 cfs), and high (22,872 cfs). Survival varied by flow threshold: 3.0% below minimum, 18.9% between minimum and historic mean, 50.8% between historic mean and high, and 35.3% above high. We used these thresholds to design alternative hydrographs over the same years that included an important component of functional flows: spring pulse flows. We compared predicted cohort migration survival between actual and alternative hydrographs. Managed hydrographs with pulse flows that targeted high survival thresholds were predicted to increase annual cohort migration survival by 55–132% without any additions to the water budget and by 79–330% with a modest addition to the water budget.
These quantitative estimates of the biological consequences of different flow thresholds provide resource managers with critical information for designing functional flow regimes that benefit salmon in California’s highly constrained water management arena.