Science Stories: Adventures in Bay-Delta Data

  • September 28, 2021


Smelt population was crashing for sure,

While managers franticly searched for a cure.

A synthesis team was tasked with the goal

Of testing if outflow could fill in the hole.


Though nothing is ever as clear as it seems

The FLOAT-MAST was faultless in chasing their dreams.

They looked at the plankton, the fish and the flow,

But temperature left them with nowhere to go.


Fall outflow may manage a critical part,

But ecology complicates things from the start.

High outflow alone was never enough,

To find a solution may always be tough.


But FLOAT will continue in showing the way

For science and synthesis even today.

Act I

It was a dark and stormy night. Actually, it wasn’t stormy, which was the problem. The curtain rises on the intrepid scientists of the Interagency Ecological Program (IEP) in 2016 as California’s drought continues. They are wrestling with a critical question: Is high fall outflow the key to improving habitat conditions for Delta Smelt and benefiting the population? Scientists and resource managers in the San Francisco Estuary are desperately trying to save the endangered Delta Smelt, but the best research to date had resulted in “It’s complicated” (Sommer et al. 2007). However, many studies pointed to high Delta outflow (freshwater flowing out of the estuary) was an important part of the story (IEP-MAST et al. 2015; Thomson et al. 2010). Delta Smelt had positive population growth rates only twice since 2002, and both years had high flows.

Enter from stage left, our heroes: The Flow Alteration Management Analysis and Synthesis Team (FLOAT-MAST). Their task was to evaluate all aspects of Delta Smelt population, health, life history, and habitat to see whether high flows would again allow Delta Smelt to rebound. This effort would build on previous synthesis efforts summarizing what we know about Delta Smelt and fall low-salinity-zone habitat (Brown et al. 2014; IEP-MAST et al. 2015). The report on their work “Synthesis of data and studies relating to Delta Smelt biology in the San Francisco Estuary, emphasizing water year 2017” (FLOAT MAST 2021), just came out, but the punchline was clear from the contents of the [empty] fish nets. High outflow alone was not enough to cause population recovery for Delta Smelt. However, the team was able to show which of our ideas about Delta Smelt habitat requirements still have merit, and which need to be revised. This is their story.

If the goal was Science for Science’s sake, it would have been one thing, but the science surrounding flow in the Delta is tightly tied to water management. The stakes were high. In 2016, USFWS asked for increased Delta outflow in the fall to benefit Delta Smelt above what was required by Water Rights Decision 1641 (D-1641). There wasn’t enough water available to increase Delta outflow in 2016, but IEP scientists began laying the framework to analyze future actions. In 2017, they lucked out. It was the wettest water year (measured from October 1st to September 31st) on record in Northern California (DWR 2019 (PDF)), so Delta Smelt were going to have lots of outflow! Now was the chance for FLOAT-MAST to test their hypothesis that high fall outflow is helpful for Delta Smelt!

Why did they think fall outflow was important? Well, previous work investigating ideal habitat for Delta Smelt discovered that smelt like water in a particular range of salinity and turbidity (Sommer and Mejia 2013). They like to hang out in the “Low Salinity Zone” (LSZ, between 0.5-6 ppt) during the fall, though some fish hang out in fresh water year round (no one told them that smelt prefer 0.5-6ppt; (Hobbs et al. 2019c)). Salinity in the estuary can swing widely if the amount of freshwater outflow increases or decreases, meaning the Low Salinity Zone is dynamic, sometimes occurring upstream in the Sacramento and San Joaquin rivers, sometimes occurring downstream in Suisun Bay and Suisun Marsh. However, good habitat for Delta Smelt is about more than just salinity. Areas with stationary habitat aspects (things that are not dynamic or changing over time) that include lots of extended shallows, narrow channels with tidal wetlands, and sandy shoals are also thought to be better for smelt because they have more places to rest and more food (Bever et al. 2016; Hammock et al. 2019). High Delta outflow lets the dynamic region of good salinity overlap with the areas of good stationary habitat with tidal wetlands and shallow shoals, creating the perfect spot for smelt (Figure 1).

Diagram of Delta Smelt habitat showing how low outflow puts the good salinity zone in the Sacramento River (which is mediocre habitat, like a Motel 6), while high outflow puts good salinity in Suisun, which is good habitat (like the Hilton).

Figure 1. The relationship between Delta outflow and Delta Smelt habitat. When outflow is low, the area of good salinity conditions (dynamic habitat) is upstream in the Sacramento River, where the stationary habitat is mostly narrow channels, which aren’t too comfortable to Delta Smelt. When outflow is high, the low salinity zone is pushed into Suisun Marsh, where higher turbidity, extended shoals, and marshes provide better habitat.

Act II

The FLOAT-MAST team was made up of scientists, but each came from different organizations with different expertise. To remain as scientifically objective and independent as possible, it was critical to let the data tell the story. Dr. Larry Brown from the US Geologic Survey was chosen to lead the team because he was a foundational leader within the scientific community who was broadly trusted to provide the best scientific, independent leadership.

Larry wanted to let the data speak for themselves, but data tend to babble. The team started by concentrating on fall habitat conditions but realized conditions in summer were also important. So was spring. Likewise, flow was important but was really just an index of temperature and salinity and food and habitat. There were SO MANY THINGS GOING ON! Herding the cats and getting all the various pieces to tell a coherent story was more difficult than previously thought. Ecology is always more complex than we expect, and even the best management action is unlikely to work the same way every time.

To make matters worse, when the final data from 2017 were in, it was clear there would be no happy ending. High temperatures in the summer 2017 caused the Delta Smelt population to crash before the high fall flows could have any benefit. The population was so low going in to 2017 that they needed a really fantastic year, and instead had the lowest population index on record. Much of the exciting analysis the team had planned never made it into the final report because the high temperatures swamped any potential benefits of flow, rendering those analyses pointless, or impossible. The low fish catch meant they could not really look at fish health, size, or distribution because they just couldn’t catch enough of them.

The team had chosen to divide up the work and each write chapters on salinity, temperature, turbidity, phytoplankton, clams, zooplankton, smelt health, smelt distribution, and smelt survival and summarize the different lines of evidence into a final conclusion. Unfortunately, people with different perspectives added these lines of evidence up in different ways. Some lines showed that high fall outflow helped smelt, other lines showed high fall outflow had no effect. Some lines were inconclusive in regards to flow, and some showed a negative relationship. However, they could say conclusively that 2017 was not a good year for smelt, probably due to high temperatures (Table 1).

While they couldn’t make conclusions on the effectiveness of the 2017 flow action, the team achieved a lot of other important wins along the way.

  • By partnering with UC Davis researchers, the team could look at details that trawl surveys can’t tell us, like how Delta Smelt eat, grow, move from place to place, and how the environment influences their health (Hammock et al. 2020; Hobbs et al. 2019a; Hobbs et al. 2019b; Teh et al. 2020).
  • The data analysis and special studies initiated by the FLOAT team helped to build life-cycle models for Delta Smelt so they could predict smelt responses in future years (Polansky et al. 2019; Smith et al. 2020).
  • The team itself provided an awesome opportunity for scientists to learn from one another and think broadly about the big picture. As one team member said “I learned more from one two-hour meeting than a decade of workshops.” Putting a team of experts together allowed them the freedom to talk and think creatively about solutions (and it was fun too!). While most of the ideas and analyses did not make it into the final report, they increased the capacity of our team to tackle similar analyses faster in the future.

As a scientific community we learned the importance of looking at the big picture. To make it easier to look at the big picture, Larry and the Delta Science Program put together a “Smelt Conditions Report” updated annually to see how the year shaped up for Delta Smelt (DSP, 2020).

Table 1 - Results of analyses of each response variable assessed as part of the FLOAT-MAST report for the most recent high flow-years (2006, 2011, and 2017), low flow years, and specifically for 2017. Arrows represent direction of trend, with sideways arrows indicating varying results. Solid green symbols signify the variable responded as predicted. Red checked symbols signify the variable did not respond as predicted. Grey circles indicate insufficient data to evaluate the variable.
Physical Habitat Low Flows High Flows 2017
Fall LSZ Location Confluence Suisun Suisun
Area of LSZ Decreased as expected Increased as expected Increased as expected
Turbidity Could not evaluate Could not evaluate Could not evaluate
Water Temperature Unexpected ambiguous response Unexpected ambiguous response
Biotic Habitat Low Flows High Flows 2017
Phytoplankton Unexpected ambiguous response Unexpected ambiguous response increased as expected
Harmful algal blooms increased as expected decreased as expected decreased as expected
Zooplankton Unexpected ambiguous response Unexpected ambiguous response increased as expected
Clams increased as expected decreased as expected decreased as expected
Water Hyacinth increased as expected decreased as expected decreased as expected
Delta Smelt Low Flows High Flows 2017
Distribution Could not evaluate Could not evaluate Could not evaluate
Growth and Survival Unexpected ambiguous response Unexpected ambiguous response Unexpected decrease
Health Metrics Unexpected ambiguous response Unexpected ambiguous response Unexpected ambiguous response
Feeding Success Unexpected ambiguous response Unexpected ambiguous response Unexpected decrease
Life History Diversity Unexpected ambiguous response Unexpected ambiguous response Could not evaluate


The first draft of the report was completed by spring of 2019. Larry and the team had done their best to connect the pieces and provide a scientifically robust, comprehensive view on why we did not see the predicted response to fall outflow. The draft report was distributed for peer review. The lengthy review process began, with comments coming in from the IEP Science Management Team, Flow Alteration Project Work Team, various members of the authors’ management chains, and USGS’s external peer review process. Larry took it upon himself to tackle addressing the bulk of the comments and had the report almost ready for distribution when tragedy struck. Larry passed away from a massive heart attack early in 2021, just before his scheduled retirement. Larry was one of the most experienced, respected, and prolific scientists in the Bay-Delta community. He was an important mentor to hundreds of younger scientists, and his loss is felt deeply (Herbold et al. 2021).

The remaining team members finalized the document and distributed it far and wide. They also produced a two-page summary (PDF) that boiled down a 265 page report (with 475 pages of appendices) to a quick fact sheet designed for managers that are in a hurry. They are currently planning their next steps. The new environmental regulations for the State and Central Valley Water Projects include a number of fall flow actions, and the FLOAT-MAST team’s experience evaluating the high flow of 2017 may help evaluate these new actions as well.

What can we take from the FLOAT-MAST experience? A few clear lessons stand out.

  • If you are going to have a huge synthesis project tackling lots of hypotheses and topics, and that includes many experts, you better have a stellar champion. You need the leader, the orchestrator. Larry was so excellent at that and this project is just one example of why he is so missed. If we are going to tackle wide-ranging synthesis questions, find the leader first. AND – we need to cultivate synthesis leaders in our community.
  • Huge synthesis projects take a long time. If you need a fast answer, projects should be smaller in scope with a clear management question. They should produce quantitative results that answer quantitative questions, when possible, instead of relying on descriptive analyses.
  •  Partnerships yield tremendous benefits in synthesis projects, especially large ones like this one. We can’t be afraid to reach out to experts that might make our work better, in ways we can’t even predict.
  • Ecology is always more complicated than we want to admit. We have an anthropogenically altered system, and climate change will decrease our ability to predict the outcome of our management actions. Flow is one of the few variables we can change through management actions, but other stressors (over which we may not have as much control), such as temperature, will frequently mask the effect of flow.
  • Scientists need opportunities to think creatively about the big picture. Large synthesis projects are opportunities to train early-career scientists to put together multiple analyses to answer a management question.


Tiny Delta Smelt

Need more than Delta Outflow

Water must be cool.

Further Reading

Categories: General