BAY DELTA SCIENCE CONFERENCE: The effects of water project operations on juvenile salmon survival in the Delta

Literature, data review, and recommendations for future salmonid investigations

Survival for migrating juvenile chinook salmon has been low for those on the San Joaquin River, averaging approximately 5% since 2002.  Survival is more variable for Sacramento River chinook salmon, and survival data are limited for steelhead.  While water export operations contribute to salmonid mortality by way of direct mortality at facilities, this does not account for the majority of the mortality of salmon in the Delta; and the contribution of various stressors to the high mortality is unknown.

To address these issues, the Collaborative Adaptive Management Team (part of the Collaborative Science and Adaptive Management Program) designated the Salmonid Scoping Team in 2013 and charged it with conducting a literature review and answering specific questions regarding salmon mortality in the Delta.

At the 2016 Bay Delta Science Conference, Dr. Rebecca Buchanan and John Ferguson gave these presentations highlighting their findings and recommendations.


Presenter: Dr. Rebecca Buchanan, Research Scientist with the University of Washington

One of the first primary tasks the Salmon Scoping Team was to conduct a literature and data review of the known effects of water project operations on juvenile salmon survival in the Delta.  The scope was fairly narrow, but that was necessary in order to get anything done.  “We examined the effects of various types of water operations first on hydrodynamic conditions in the Delta, flow and water velocity, then the effect of the hydro conditions on route selection and travel time, and then we cascaded down to effects on survival,” said Dr.  Rebecca Buchanan, noting that in the presentation, she will be talking about survival very generally in the Delta, and the effects of water project operations on survival.

The team conducted a literature review of journal articles, technical reports, green papers, white papers, and some dissertations; due to a lack of literature on some aspects, they did take some recent data and did some very basic data presentations.

The study was focused on survival in and through the south Delta.  “We acknowledge that water project operations may affect survival outside of this region, both upstream and downstream, but we were limited to this region in the south Delta,” she said.


Dr. Buchanan began with survival data, presenting a chart showing through-Delta survival for San Joaquin River salmon.  Most of the data comes from San Joaquin River fall-run Chinook; the data was wire-tagged data up through 2006, and then acoustic telemetry since then.  “Plotting them all on the same plot here, we see that survival has declined quite a bit from the late 1990s, and it’s been particularly low since 2002, with limited variability in that last decade,” she said.

For Sacramento River chinook, there are fewer estimates overall; and taken across all of the runs, they are more variable.  “We have some estimates of 0 for fall and spring run chinook, up to about 35% for winter run and then from about 17% to 64% for late fall run, so more variable over all the runs of chinook in the Sacramento, but we have fewer estimates, especially for any one particular run,” she said.

For steelhead, there are two years of estimates of through Delta survival for Sacramento River fish, and for San Joaquin fish, there is a six year study that concluded this year with the 2016 field season.  “We see quite a bit of variability for estimates for the steelhead,” she said.  “The plot shows release specific survival estimates for the San Joaquin steelhead, and I’m showing you that to demonstrate that there is variability within a season or within a year and then between years as well, and that is more variability than we are seeing for the fall run chinook in the San Joaquin.”

They also looked at what is known about route specific survival for San Joaquin fall run chinook.  They compared the two primary routes: the San Joaquin River route in the top plot and the Old River route in the bottom plot.

Historically, the coded wiretag data showed differences in survival between these two routes with higher survival for the San Joaquin river route,” Dr. Buchanan said.  “Acoustic telemetry data in more recent years have not continued that pattern of the two routes; instead we see really no difference in survival between the two routes.  It’s low in both of them.  But it’s particularly low for the Turner Cut subroute of the San Joaquin route.  Fish that enter Turner Cut, at least for Chinook, mostly don’t appear at Chipps Island later.  They may go through the facilities or they may just swim through the interior Delta.”

Dr. Buchanan noted that they do see some type of spatial heterogeneity in survival from this type of comparison; another way to look at spatial heterogeneity is on the reaches scale.  She presented a table showing the survival rate per kilometer for different reaches in the south Delta for fall run chinook over the different years, noting that it’s from the acoustic telemetry data so it only goes back to 2008,  The colors indicate relative level of survival, with red being very poor survival and the lighter colors indicating higher survival.

So what we’re seeing here is that we have particularly poor survival in the interior Delta route, the Turner Cut route, and also in the lower reaches of the Delta,” she said.  “There is higher survival in the upper reaches.  There is some variability from year to year, but we still have this general pattern of poor survival in the downstream reaches.”


The Salmon Scoping Team was then asked to review what is known about the relationships between survival and Delta inflow, Delta exports, the I/E ratio and the E/I ratio.

Dr. Buchanan began by showing a plot of the mean San Joaquin River flow at Vernalis on the horizontal axis and mean exports from the VAMP period of each year of the VAMP study, which was from 2000-2011.  “That is one month in the spring typically from April 15 to March 15, and what’s obvious here is that for the most part, there’s a pretty tight relationship between exports and inflow, except for this one observation from 2006, where we had high inflow and low exports,” she said.  “The overall high correlation here means it’s going to be difficult to distinguish between an effect of inflow and an effect of exports using statistical methods.  So bear that in mind, as we look at the results.”

She then presented a slide showing Delta survival versus inflow for the San Joaquin fall run chinook.  She noted that the circles indicate the data is from wiretag data and the triangles are from acoustic tag data.  “There has been a positive association found between Vernalis flow and through-Delta survival which is based primarily on Delta and ocean recoveries of coded wire tags,” she said.  “This relationship is not observed if you look only at ocean recoveries of the coded wire tags, but that of course combines the Delta survival with ocean survival and ocean capture rates as well.  It’s also not necessarily consistent across all levels of inflow that we’ve observed.  We see somewhat messy but probably a positive association here for the lower levels of inflow, but for the higher levels of inflow, we’re not seeing this positive association anymore.”

The other plots show survival on different spatial scales.  In the lower left, it is showing survival from Mossdale to the Turner Cut junction, which is the upper reaches of the Delta for San Joaquin migrating fish.  “There we see what looks like a positive association,” she said.  “We didn’t do any formal analysis but I feel pretty confident in saying that’s a positive association.  That’s all from acoustic telemetry data.”

The lower right is survival from the Turner Cut junction to Chipps Island.  “That’s the lower reaches, and there that positive association disappears,” she said.  “It may not be a negative association but it’s not positive.  So we are seeing spatial heterogeneity in the relationship between inflow at Vernalis and survival, and that’s probably going to be important if we want to try to increase survival by doing something with inflow.”

There has also been a positive association observed between flow in the Sacramento and survival to Chipps Island for late-fall chinook and for fall chinook.  There are two different tagging technologies (acoustic versus wiretag) and different sizes of fish.  On the left column is Ryer to Chipps Island, and on the right column, it is Sacramento to Chipps Island.  The different rows in this plot are the different sizes of fish; so small fish are at the top; large fish at the bottom.

Generally what we see from these two different models and different types of fish is that as inflow increases, survival increases, but the effect levels out for high levels of inflow, and that’s consistent,” she said.  “In the Sacramento River, we also observed regional variation in the inflow-survival relationship, so they took late fall run survival estimates and compared them from high flow years to low flow years, and observed that for the high flow year, we had higher survival in the riverine reaches than in the low flow years, but that pattern did not continue into the tidal estuarine reaches.  It’s very similar to what we found for the San Joaquin.”

Next, Dr. Buchanan presented a slide for salmon on the San Joaquin, plotting survival against exports.  “There’s been some moderate evidence of a positive association between exports and survival through the Delta, based on Delta and ocean recoveries of wire tag data,” she said.  “In the scatterplot, you see kind of a messy positive association here for the lower levels exports which was not what we were expecting.  We were expecting a negative association, so is this meaningful? Is there something about exports that’s driving this, or are we seeing evidence of that correlation between inflow and exports?  We did not answer that question, so that’s still something to consider.”

For the higher level of exports, we have very low survival, which is what we expected, but we only have a few observations; that’s not enough to characterize the variability at that level of exports.”

For the Sacramento River, there is some evidence of a negative association between exports and Delta survival for fall run chinook using the Delta and ocean recoveries of the coded wire tag,” Dr. Buchanan continued.  “Again if you just look at the ocean recoveries, this association does not appear.  This was not observed for late fall run, so it’s not totally consistent across all runs.”

For late fall, however, there has been a negative association observed with the relative survival of an interior Delta route to the mainstem route, so how well do fish in Georgiana Slough do getting through the Delta compared to fish in the mainstem, that’s what this plot is showing,” she said.  “It does look like there’s a negative association there, but there’s quite a bit of noise in the data and the model predictions are fairly uncertain there.  Also, there is a competing model that was developed that did not have exports but had comparable weight, and found some evidence of a negative association or a negative relationship between exports and survival on a various scales, but it’s not really overwhelming evidence.”

For survival of Sacramento River chinook and the Delta E:I ratio (export:inflow), there has been some indication of a negative association between E:I and Delta survival for fall run.  “The E:I ratio is found to be useful in a stage-structured life cycle model by Cunningham; they did not find it to be useful for other runs of fish, however,” she said.  “Newman and Rice found a small effect, but it was not statistically significant for fall run using code wire tag data.”

There have been a couple of model investigations modeling the recovery rates of coded wire tag data, comparing models that use the E/I ratio versus using exports and inflow separately; she noted that the E/I ratio models were not supported over the models that used them separately.  “There’s not a lot of evidence that the E/I ratio is terribly useful for Delta survival, at least based on the available data, and these are all from coded wire tag data.”


Dr. Buchanan then summarized what they found.  “Our primary finding was that salmon survival in the South Delta is low, which is not a big surprise; we knew it was low, but what was somewhat surprising is just how low it is and how consistently low it has been, especially for the San Joaquin fall run chinook,” she said.  “Other things that had not been totally obvious before were that we do have spatial variability in their survival, not necessarily between routes or at least not as consistently as we had thought in the past, but certainly upstream versus downstream, we see this variability.”

As far as the lack of route differences in survival, maybe that’s because something changed in the system or maybe that’s because survival is so low that we cannot detect differences,” she continued.  “As far as the relationships between survival and these different management operations go, the one for Delta inflow is the most compelling; we did see it to be spatially heterogeneous for both river basins. It wasn’t totally consistent in that we had some conflicting data from really high flow years and we didn’t identify exactly what was going on there.”

The patterns we observed for I/E were pretty similar to what we observed for Delta inflow.  For exports, there was some evidence of a relationship, but it’s not the one we were expecting, and not terribly well supported by the data.”

  • Insufficient data on survival in Delta for steelhead, Sacramento River Chinook (all runs)
  • Tagging studies represent only part of life history, populations: “The tag studies that we have available to us represent only part of the life histories in populations that use the Delta, smelt-sized hatchery fish, so we’re missing the smaller fish and we’re missing the wild fish,” she said.
  • Limited understanding of indirect effects of water project operations on mortality in Delta:  “I haven’t been talking about mechanisms that might explain indirect effects of water project operations on mortality in the Delta, but we did identify some possible mechanisms, and we didn’t find much research on that, so there’s a need for some work there.”
  • Formal analysis of relationships between inflow, exports, I:E and survival is incomplete for existing data, especially on the San Joaquin.

Dr. Buchanan noted they were working on the last bullet point.  “Even when we have those analyses done, there will still be some constraints on our understanding.  One, all of the observations that we have are in the presence of the management operations, which is understandable, but it does make it difficult to assess their effectiveness because we’re lacking control and we’re lacking variability in the conditions; without that variability, it’s very difficult to identify a relationship.  We also don’t have very many observations at higher levels of exports or inflow.  The low overall survival makes it difficult to detect changes in survival because of low effect of sample sizes and the high uncertainty in the results.”

We are lacking a survival target that has been agreed upon by the region,” she continued.  “There have been some benefits suggested for different purposes, but without a commonly accepted target, we don’t really know how to plan studies to determine if a management action has been successful.


Presenter: John Ferguson, Anchor QEA

John Ferguson with Anchor QEA next highlighted the answers to the questions they had been given by the Collaborative Adaptive Management Team and reviewed the team’s recommendations, noting that in answering the questions, they were focused on only the Delta and not the whole life cycle; the focus was on the South Delta and in particular on exports and operations.

Their upcoming report from the team will have two volumes: Volume 1 is the key findings and gaps found from the literature review with five technical appendices; Volume 2 will be the responses to the eight questions.

Mr. Ferguson then presented a general conceptual model, noting that some of the cells are color coded.  Green are the things that they concentrated on, although the things in blue might also be affecting salmon, he noted.  “The point here is that based on the time and data that we had, we looked at a subset of what we think might be going on, so that’s just a framework to keep in mind,” he said.

First question: Effects of exports on flow and dynamics

The found that the export effects vary with distance from facilities (decrease), export level (increase), inflow, and tides, and the largest export effect was estimated in Old River near the SWP and CVP intakes.  They also found that they had a very small effect out in the distributary channels coming off the San Joaquin leading into the interior Delta.  “That led to quite a bit of discussion within the scoping team,” said Mr. Ferguson.  “Is that important? Do we understand what the thresholds are?  Small effect, no effect? We don’t know, so that formulated some of our thinking in our recommendations.”

Second question: Use of available models

Models are all tools, and they work for the purposes that they were designed to achieve, he said.  “You wouldn’t want to use a 1D flow routing model for the system to look at channel junctions; you would want to look at the system with a 3D model. You apply the tool that you need at the time to answer the question that you’re trying to answer.  For us, that led us to think that what we really need to do is more integration between our biological studies, for example: fine scale hydrodynamic modeling at channel junctions to learn how fish are responding better,” he said.

Third question: The effects of exports and inflows on juvenile salmonid survival.

Mr. Ferguson noted that this was previously covered by Dr. Buchanan.

Fourth question:  OMR flow management: Is the January 1st onset date working?

The answer is relatively well, said Mr. Ferguson.  “We did some analysis on this, looked at 20 years of data, and we found that the January 1 onset coincides with listed species present,” he said.  “But we also reported to CAMT that based on monitoring at the Sac trawl and the beaches above the Delta, if you initiated the OMR restriction based on first detection, you would start earlier in many years, particularly for winter run.

Fifth question: Are the salvage density based export restrictions working?

We said, based on the direct effects, it appears that they likely reduce direct mortality or entrainment in the pumps, but this then generates a whole question that we have which is what’s the effect of the pumps away from the pumps, out in the interior Delta?,” he said.  “This is an area of indirect mortality, and we haven’t really been able to inform that very well with the data that we have.  And that led to quite a bit of discussion in our scoping team as to what the effects might be.”

Mr. Ferguson noted that they were asked to catalog all of their scientific debate.  “Here you see the litigation coming into to effect where there was a lot of debate and a lot of scientific disagreement but when we put ourselves in the room, you really kind of come together after three years, and we did not have that many disagreements,” he said.  “Where we did disagree, we didn’t have the data, so it was more of opinion based disagreement like this example here.”

Sixth question:  Can the team come up with some new flow metrics?

Can you look at the data, but your heads together and say, are there additional physical or biological metrics that we should pursue?,” he said.  The team came up with five physical ones:  Net flow in the lower San Joaquin River (QWEST);  hydraulic residence time in the South Delta; percent positive flow in the OMR Corridor; relative proportion of CVP exports; and proportion of Sacramento River water in exports.

We identified these; we did not analyze them, but they could be analyzed, and one of the things I hope we do in the near term is gather all of our data into one dataset – our coded wire tag data, acoustic tag data, and do a multivariate analysis of these kinds of metrics to see if they might work better for managing flow and exports in the basin.”

Seventh question: Biological response metrics

They also came up with eight additional biological metrics:  Fish routing into the interior Delta; survival at the route and reach scale; survival at the Delta scale; condition of fish entering and leaving Delta; contribution of fry rearing to survival and adult production; probability of export facility entrainment; direct (salvage) mortality relative to population abundance; and juvenile abundance exiting Delta.

We don’t have a marker out there.  We don’t have that bar identified.  There’s some numbers that have come through different forums, like the BDCP but it isn’t codified, so we researchers don’t know how well an action is really making progress towards achieving a certain goal.  You’ll see that come up in our recommendations.”

Eighth question: Addressing concerns about surrogacy.

Honestly, we didn’t do that much with it as there’s not much we can do,” he said.  “There are not that many comparative studies.  We basically concluded that you’re going to use surrogates until the test fish that you really desire are abundant enough that regulators will let you use them, and the questions regarding the use of surrogates will continue.  We have to put it on the researchers; we said, ‘if you’re a researcher and you’re proposing to use steelhead for chinook, you better justify that in your study plan, and managers, you’d better be okay with that before you fund that action.’  We concluded that generally don’t compare chinook and steelhead, and use chinook to chinook and get as close as you can to your target species in a common sense approach.”


Mr. Ferguson then summarized their findings and overarching recommendations.  How do you translate science into management discussions into actions?  “That’s the beauty of CAMT; in my opinion, it’s the beauty of this collaboration,” he said.  “We’ve come together and we have a foundational document that people can use.”

The team had some overarching considerations that came out of these data that influenced how they thought about what they want to tell CAMT is important.

The first one is that the system is complicated,” he said.  “But more importantly, what we’re telling CAMT is, you shouldn’t think of the Delta as the Delta; we keep referring to it as the Delta, but it isn’t one uniform area.  We all know that, but how do we translate that knowledge into thinking about actions for the Delta that require different actions in different areas?  The three that we came up with are the mainstem river and inflow; the tidal area in the lower San Joaquin that may require a different set of actions; and the interior Delta where you have a lot of export, inflow, and tidal effects.  So we need to think about the Delta regions as regions and come up with actions to improve actions for salmon survival that will differ among the regions.”

Another one of the overarching considerations is that the questions need to be integrated more with science, management and operations.  “The approach shouldn’t be just about researchers coming up with an idea of how we think we ought to do the monitoring; we need to have this integration between the scientists, the managers, and the operators all in the room with the hydrodynamicists, the statisticians, and define what needs to be tested, what are the operations that are really of concern to managers and to the operators, what are the questions that the managers have, and how can the scientists inform how we set up those tests and make those observations.”

And lastly, there’s a need for more modeling for salmon, either with existing models or develop new ones because of the uncertainty.  “We have to make decisions in the face of uncertainty, and we at the SST recognize that, so we’re saying use the tools that we have, develop new ones, but we’re not going to have time and we’re not going to have the money to inform all of the questions the managers have to a level that scientists would like, so therefore we have to incorporate modeling as well.”

Mr. Ferguson said that the survival plot for salmon was one thing that stuck with the team through their deliberations.  “It showed that since 2002, survival has been low and it’s stayed low; it doesn’t matter what the flow year is or what the water year is, so we’ve got an issue there,” he said.  “This drove our thinking into actions, and you’ll see that in our recommendations.”

The team had four recommendations they communicated to CAMT:

One – Continue existing survival studies, monitoring, and analysis of data (foundation for expanded, future studies):One is to continue our existing studies; that provides a time series of data.  We also have quite a bit of data that’s been collected through 2016, and we need to analyze those datasets.

Two – Implement short-term actions to improve salvage facility operations (disagreement on whether to recommend short term actions or premature to do so):This caused quite a bit of discussion in the scoping team, because some people were saying, ‘we know we need to act because survival is so low, but what do we need to do, and do we have the information to say that if we do that, it’s really going to help survival, because we didn’t do those analyses,’” he said.  “That’s one of our disagreements, but we coalesced around this recommendation to focus on project operations, and we list six to eight things in our report that could be done at the pumping plants.”

Three and 4: Develop and implement a long-term monitoring, research, and adaptive management plan: We need a longer-term vision or strategy for where we’re going with salmon science in the Delta that’s based on monitoring, modeling, and direct manipulation and experimentation where the funding is stable source, so there’s continuity between years,” he said.  “It’s going to require a commitment; it’s going to require some discussions.  What are the key management questions and to what precision do you need the researchers to test.  It’s going to require agreement among the operators, the managers, and the scientists as to what can be achieved.”

It also needs to be adaptive, and that means a lot of different things to a lot of different people,” he continued.  “In my mind, we would be fairly straightforward in saying adaptive management means we need to design, we need to implement, we need to analyze the data, we need to interpret the data, and then we need to adapt our study every year; so we have this annual cycle of studying, analyzing, debating, and forming, adjusting our experiments as we go forward with a focus on not just monitoring, but experimentation and getting at the causal mechanisms.”

Mr. Ferguson said that CAMT realizes that they need to take actions to improve salmonid survival, whether that’s at the export facilities or tackling predation.  “We also need to do a better job of identifying what the next questions are and developing objectives for salmon survival in the Delta, they listened to us loud and clear on that,” he said.  CAMT also understands that we need better turnaround of our data; for example, we have six years of steelhead acoustic tag data, we have access to two of those years.  We’re not faulting anybody; we’re saying we need to do a lot better job of turning the data around more quickly so we can use it, because there is a sense of urgency in CAMT.  Survival is low; we need to take some actions.”

Mr. Ferguson said he thinks CAMT will coalesce around CAMT certain operations that are of interest to managers, such as OMR and I/E.  “CAMT’s not interested in a brand new program here,” he said.  “We’re looking to see where can CAMT add value and where can these insights provide additional information, working within existing programs to do so; the CVPIA has been proposed as one such vehicle for that.”

For more from the Bay Delta Science Conference …

Daily emailsSign up for daily email service and you’ll never miss a post!

Sign up for daily emails and get all the Notebook’s aggregated and original water news content delivered to your email box by 9AM. Breaking news alerts, too. Sign me up!

(Visited 749 times, 1 visits today)

Leave a Reply