BROWN BAG SEMINAR: Development of Recommended Flow Targets to Support Biological Integrity Based on Regional Flow-Ecology Relationships for Benthic Macroinvertebrates in Southern California Streams

Dr. Eric Stein is the principal scientist at Southern California Coastal Water Research Project where he oversees a variety of different kinds of projects focused on instream and coastal water quality, bioassessment, hydromodification, watershed modeling, and assessment of wetlands and other aquatic resources.

This brown bag seminar is the second of three seminars focused on flow targets and ecology (part 1 is here).  Dr. Stein noted his presentation is focused more on water quality impacts rather than fisheries.  The work presented in this seminar is part of an integrated framework that has been developed by a technical group working over the past year and a half trying to figure out how to approach environmental flows and develop assessment frameworks and tools to inform the variety of different policies in the state that are affected by environmental flow considerations.   He noted that environmental flow considerations transcend many agencies and many different programs, so the framework is an attempt to organize the science to support a variety of different agency programs.

Hydrology is an integrative driver of stream health.  “It’s the obvious,” he said.  “If you’re trying to protect rivers, you have to think of streams and you have to think about hydrology, and so it is really kind of an integrator.  We know that biological health and the ecological condition of these streams and rivers is ultimately affected by things like geomorphology, temperature, and water quality parameters, but all of these modulated through the hydrologic condition of the stream.  So when we think about management, managing the hydrology of these systems is really fundamental to helping ensure that their ecological condition is what we want it to be.  Sometimes that might be reflective of reference conditions and sometimes that might be reflective of some alternative state that we deem as managers or as society as an appropriate end point, but regardless, hydrology an important element to making sure we’re achieving those desired goals.”

California has a lot of complex environments due to its continental-scale variability in precipitation, temperature, and elevation, Dr. Stein said. The slide shows three different streams in the state, all of which have different environmental flow needs, so the framework needs to consider different stream types.  Many programs are attempting to set environmental flows, and they each have different endpoints and management needs.  Management concerns can vary depending on location: sometimes there are urban issues, such as water reclamation or stormwater capture and reuse; other times, there might be issues associated with dam operations; or there might be issues associated with agricultural water diversions or groundwater use.  Another significant challenge is coordination because the responsibilities are spread among many different agencies and programs, and mechanisms aren’t always in place for coordination between the programs for things like data sharing and technical tools, or on policy implementation.

All of these things make this quite a daunting challenge for a place as complex and diverse as California,” he said.  “There is a need for a coordinated framework to try and encapsulate all of these things and find a way to work more cooperatively and leverage the efforts.  There are a lot of streams and a lot of watersheds, and trying to do this one at a time throughout the whole state would take us a long time.  We want to try to be efficient with the resources that we have.”

The workgroup is working in coordination with several agencies around the state to develop a framework that could be applied statewide that considers hydrology and ecology together and utilizes that understanding to set environmental flow targets.  That framework can then transition into the policy implementation stage where competing beneficial uses are considered and the implementation challenges that go along with that, he said.

Another goal of the framework is to assess vulnerability.  For example, in Southern California, Dr. Stein said they were looking at climate change predictions on flow and what that means from a vulnerability perspective which is their reason for thinking about environmental flows.  The framework can also evaluate and inform management actions such as wastewater reuse or dam operations, as well as assist with assessing whether hydrology is the most important management concern or if it is another factor, such as temperature or water quality.  For all these reasons, understanding the relationship between flow and ecology is really important for answering all of these questions, he said.

So the framework starts with a statewide approach, which is really intended to provide a first tier way to look at setting environmental flow targets that could be applied anywhere in the state at a relatively coarse scale; the next tier would then provide a way to develop more region or location-specific flow targets that take into account some of the very local considerations.  This needs to be done in a way that utilizes standardized ways of collecting data and provides consistency so that the information can then be shared.

One of the first things the workgroup developed was a classification system for the state’s streams.  The state’s streams were binned or classified into nine hydrologic classes based on things such as catchment properties, rainfall, geology, and soil, which provide an organizational framework to work with for the first tier.  The Map is available at the Surface Water Ambient Monitoring Program (SWAMP_ bioassessment website.)

Within the nine hydrologic classes of streams, dimensionless unit hydrographs can be used to understand reference hydrology, set preliminary targets based on the reference hydrology, and then to think about what other appropriate ecological end points might be relevant for each of the hydrologic classes.

Dr. Stein said the rest of his talk would be focusing on the second tier of the process, which is getting into the more detailed site specific environmental flows approach; however, he will be discussing a specific application in Southern California that doesn’t have to do with salmon.

Most people who look at environmental flows are thinking about a specific taxa like salmon, and are considering the life history requirements for the species, such as cues to migration and breeding requirements; next they think about the hydrologic conditions that are important to support the life history stages, how to manage the hydrology to support the life history, and how that affect things like dam and reservoir operation.  Dr. Stein said while that’s how the majority of environmental flows work in the state has progressed, it isn’t really relevant for a lot of the issues they face in Southern California.  He did acknowledge that there are some steelhead streams in Southern California, but for the most part, fish issues aren’t the driving issues for the region.

The driving issues that we’re dealing with in Southern California are things like use and reuse of treated effluent,” he said.  “We have a lot of systems that have been fundamentally modified over the last 100 or 150 years, and the hydrology is primarily driven by treated wastewater or stormwater discharge, and so now as we are in the era of increased recycled water use and reclamation and trying to become more independent of some of the imported water sources, there’s a lot of discussion if we want to reuse some of this water that have been discharged into the streams for other uses, what are the ecological implications of that and is that acceptable or desirable?

Another issue for Southern California is with stormwater retention where some of the peak flows are being used for groundwater recharge or potential reuse, and what are the implications of retaining those stormwater runoff flows from an ecological perspective.  There are also concerns about protecting some of the areas that are intact from future land use changes and what effect the projected fundamental shifts in the ecosystems associated with climate change will have on environmental flow requirements.

He presented a slidem with a graph, noting that what they try to do is when setting flow targets is to create graph similar to this where the y axis represents the potential change in the biological community, and the x axis is the change in flow regime.  The question is how much the biological community might change in relation to the change in flow regimes and what rate of change is acceptable.  “So if we can create this .., we can understand how does that change in some hydrologic measure related to a change in the biological community, and then use that to set management targets,” he said.

In Southern California, the waterways of concern are not rivers but instead are the smaller, wadeable streams, most of which are intermittent and dry up or are spatially intermittent or temporally intermittent over the course of the year.  They are focused on benthic invertebrates as the main ecological end point because most of the issues as it relates to environmental flows in Southern California are driven by the water quality programs, rather than by the water rights or water supply programs.

“The primary indicator we use from a water quality perspective at this point in time are benthic invertebrates, and that’s used statewide for all water quality programs, but in Southern California, that’s the key driver,” Dr. Stein said.  “The management issues are things like stormwater capture and water reuse, and so that’s kind of the framework for our tier 2 region specific environmental flows analysis that we’re working with in Southern California – so no dams, no salmon.”

The next challenge is to select the local approach to tackle environmental flows, and Dr. Stein noted that there are many options available for this.

As part of the statewide framework, one of the things that we’re working on is a literature review that builds off of this document, the Final EPA-USGS Technical Report: Protecting Aquatic Life from Effects of Hydrologic Alteration which came out last year where they summarized existing environmental flow programs throughout the country,” said Dr. Stein.  “So using that and using work that’s gone on in Australia and Europe and other places around the world where they’ve been dealing with these issues is to think about how do you begin to select the most appropriate technical approach given the issues that you’re dealing with; so given the stream types, the biological end points, and the management concerns, what would be the most appropriate assessment tool for you to use, so that’s part of what we’re working on now.”

There are basically two types of assessment tools with a lot of hybrids in between, he said.  “There are statistical models such as the graphic on the top where you’re trying to derive a statistical relationship between something like probability of occurrence or density and some sort of measure of hydrology, and then there’s more of a mechanistic approach which is trying to think about the ecological history and life history needs of an organism and match that with certain flow requirements.”

In Southern California, they use an approach called the Ecological Limits Of Hydrologic Alteration (ELOHA) which Dr. Stein described as a hybrid approach that has mechanistic underpinnings but is largely a statistical driven approach.   He said it was an appropriate choice because they are not focused on a specific taxa, like a certain bird or a certain amphibian, but are focused on a community of benthic macroinvertebrates as the assessment end point.

We’re really interested in a measure of the entire community as an index, so it’s a community based approach,” he said.  “The other thing about the ELOHA approach that was attractive for us is we don’t want to do this on a reach by reach or catchment by catchment basis; we wanted to develop flow targets that are applicable for the entire Southern California region, and so the ELOHA approach is appropriate because it’s really well suited for a community-based analysis versus individual organisms, and it’s also well suited for a regional versus a location specific type of analysis.”

He then described the four steps to the ELOHA approach:

  • The first step is to estimate the degree of hydrologic alteration. The basic premise is to relate hydrologic change to biologic change, so you have to first estimate the degree of hydrologic change between natural and current conditions by calculating a series of flow metrics.
  • That hydrologic change is compared to some response in the biological community. In this case, they use the benthic macroinvertebrate index in the CSCI to establish thresholds of biological response.
  • Develop a regional index of hydrologic alteration based on priority metrics: That relationship between hydrologic change and biological change is then used to develop indices or scores that can then be used to assess and prioritize how flow is managed in different streams.
  • That is then applied to evaluate management actions and set management targets.

He then presented a flowchart showing the process they use, starting with the statewide classification, compiling the bioassessment data, estimating the hydrologic change and comparing the two,  selecting the metrics, and then evaluating management actions.

Dr. Stein said that California has a robust bioassessment and biomonitoring program, with about 5000 sites that have been evaluated over the last 15 years.  In Southern California alone, there are about 800 sites that have been assessed in the last seven or eight years that they had to work with for understanding the biological condition using benthic macroinvertebrates; the problem is that very few of the sites are gauged, so the first challenge was how to estimate hydrologic change across the broad array of sites, some of which have no gauge data.

When you don’t have all the data, you go to models – and that’s possible, but we still don’t really have the resources or the time to develop 800 models; that’s a lot of modeling,” he said.  “So we decided to go with an ensemble model approach, which essentially means that we’re going to pick a subset of models that are representative of the range of watershed conditions that we have in Southern California: some large, some small, some steep, some not so steep, some higher rainfall, some lower rainfall.  We want to encapsulate a range of watershed types, and pick a representative set of models, and calibrate and validate those models with the data that we have.”

This yields essentially a library of models to choose from.  Dr. Stein explained how they next developed a model assignment tool which is essentially a statistical tool to pick the correct model:  “We can go to any new catchment of interest with a bioassessment site and we delineate the catchment draining to that; I can calculate some basic properties of that catchment, such as size and flow, and then based on the physical properties of that watershed, I can pick the right model out of my ensemble and tune very few – about 3 of 4 different parameters get adjusted in the model, and then I can run it, so basically it allows me to create this library that I can apply in a very parsimonious way across a broad range of sites in a fairly efficient manner.”

He then explained how they assign a model to an ungaged site.  “We picked 26 models.  These were at gauged locations.  We have a range of catchment sizes, a range of elevations, and a range of impervious cover.  We calibrated and validated these models with field data, and then we created this clustering algorithm using basically a variation of a boosted regression tree analysis to essentially create this binning that allows us to go to any new site, select the right model from our library, and assign it to our site of interest.”

So if we have some sites of interest where we have bioassessment data as shown here on the map with the red dots, I can use my model assignment tool, I can pick the right model off the shelf, I basically run it under current land use conditions and so that gives me a daily flow series that is shown in yellow (above, right) which would be the current hydrology,” he continued.  “We run a six year simulation which is a synthetic simulation that includes wet, dry, and average conditions so it encapsulates the range of conditions and I get a measure of a current daily flow over sort of a six year period that represents a range of hydrologic conditions.

Then I go into the model, and I turn that catchment natural, which means I make the land use natural, I change some of the storage parameters in the model, I change a couple of the routing parameters in the model, and then I rerun the model using that same six-year rainfall record so the exact same synthetic precipitation synthetic record into the model and run it under natural conditions.  That gives me the blue hydrograph, so now for each of my 800 sites.  What I essentially have is reference hydrology and current hydrology from the model … I now have some way to compare current to reference hydrology for each of those 800 sites, and that allows me essentially to calculate that hydrologic change, so the Delta age or the hydrologic change is essentially just the difference between those two hydrographs.  Simple.  Pretty straightforward.”

The next challenge is to create some sort of flow metric or measure that aggregates different elements of the flow, he said.  “If you go into the literature and look at flow metrics that have been used over the last 6 or 10 years, there are literally hundreds of different flow metrics that people have published on,” Dr. Stein said.  “These metrics can fall into these basic categories.  There are metrics that relate to the magnitude of flow, the variability of flow, the duration of flow, and the timing of flow.  We wanted to have some metrics that we looked at that were representative of all those different elements of the hydrograph, so we’re not just capturing magnitude, but different elements of the hydrograph.  So we compiled about 300 of the metrics that we felt represented the range of these different elements of the hydrology and were things that had been previously published on, and then we calculated those metrics using those hydrographs I just showed you, and now we have essentially current and reference hydrology.”

In our package, we can just churn out those metrics for every one of those 800 bioassessment sites for both current and reference conditions, and then we bin them by average, wet, dry, and all years and now we have the ability to look at the overall hydrology and then under different climatic conditions,” he said.  “So essentially now what we have is a set of metrics that speak to hydrologic change, so we have change in magnitude, we have change in variability, change in duration and change in timing for all of those 800 sites under different climatic conditions.”

The next step is to determine which metrics to use relative to the biology, and so using statistical analysis, they created a logistic progression curve.  “Basically on the y axis, we’re asking what is the probability of change in biology, and on the x axis we’re looking at the change in hydrology,” Dr. Stein said.  “This is the same curve that I showed you in the beginning, change in biology, change in hydrology, only now I’m just showing it to you with real data.”

For the y axis, the measure used is the California Stream Condition Index, which is an index of biotic integrity that aggregates measures of biological community health.  “The important thing about the California Stream Condition Index is that it’s a predictive index, which means that the way you calculate the California Stream Condition Index is you basically pick a site, you simulate what would the biology of that site look like under natural conditions, and that creates what called an e or expected value for that site.  Then you go out and measure the bugs like we normally do, you see what’s there, and you compare the observed to that expected value, and the ratio of the two gives you that CSCI score, so it always goes from 0 to 1, with 1 being the site matches expectations under reference conditions.”

That’s important for this analysis because that essentially is the biologic change; it is the current condition relative to a reference expectation, he said.  “What I’ve done is I’ve used the CSCI which gives me a measure of biologic change for reference on the y axis, and on the x axis I have my measures of hydrologic change, so now we can create that graph of delta biology versus delta hydrology.  So that’s essentially the analysis we used.”

The reason logistic regression analysis was used is because the y axis is reflective as a probability of change.  The established threshold in literature for the CSCI is 0.79 which is sort of the breakpoint between reference and non-reference conditions; Dr. Stein pointed out that the number isn’t regulatory but rather it’s based on a comparison to the distribution of reference sites throughout the state.  Sites above .79 are considered good sites; below that, not so good sites.

What the y axis here shows is what’s the probability that my CSCI score would be above that .79 value,” he said.  “So what I’m showing you in this graph is basically the 50% probability, so I can tell you that for a certain hydrologic metric such as high duration and flow days, so if I want a 50% likelihood that I’m going to have good biology in my streams, I want to make sure that my high flow duration does not decline by more than about six or seven days.  So basically out of a 50% probability, if I come down onto the x axis, so it’s a minus 6 days, so if I have anything minus six days or less, then I have a 50% probability of things being in good shape.   Now you as a manager might decide 50% isn’t the right number, maybe you want 75%, so just go up to 75% on the y axis, go across and pick a different number.”

The logistical regression plots can be generated for all of the biology-hydrology combinations; the shape of these curves can be different so all of the plots must be looked at independently.  “So we produce a lot of these plots with different shapes,” Dr. Stein said.  “Then we go through a selection process where we basically go through all of those plots and we pick the ones that have the strongest relationships with the biology.  Then we run them through another set of filters because we want to make sure that the metrics could differentiate reference from non-reference, so if they can’t, that’s not very useful for us.  We eliminate redundant metrics so we want to have metrics that are non-redundant with each other, and we want to have metrics that are amendable to management, meaning you can actually do something from a management perspective to influence that element of the hydrology.”

After the filters are applied, the final set of metrics is produced.  “In this case, we looked at a series of seven metrics, we identified in some cases increasing or decreasing thresholds in need of the increasing direction or the decreasing direction, and we identify the critical precipitation condition,”   he said.  “So we can use a matrix like this to create sort of management template to say, if I can manage the stream within these certain different parameters, then I can have in my case a 60% probability of having good biology in that stream.  The important thing about the thresholds is that they are not absolute; they are relative to some sort of baseline condition, so they can be tailored for whatever stream you happen to be interested in.”

The metrics are aggregated and binned into categories A, B, C, and D.  “Now I can go back to my map of the bioassessment sites, and for each site, I can put them into a category of hydrologic condition, based on whether the current hydrology has deviated from reference condition to a point where we think it’s going to create a hydrologic alteration, and then I can color code them with the blue being good and red being bad, and a few gradations in between.  This first step allows us to get a sense of the hydrologic landscape within the region.”

To illustrate how this can be used for management, Dr. Stein then showed an example of how they used this in the San Diego River Basin.  “We went in and looked at the current hydrologic conditions using the same categories under the current land use, 2016,” he said.  “We next overlayed the County’s general plan with what their proposed development scenario is like for 2050.  (below, left). Then we can project the hydrologic conditions as they relate to biology and we can show where in the watersheds we think conditions are going to be getting worse or which areas are at risk from a biological condition perspective.  So if you allow that land use to go forward and you don’t mitigate those hydrologic impacts through BMPs and things like that, then where do we think the areas are that are going to be at the highest risk, and so you can use that to prioritize management decisions.”

Another application is to assess conditions in the watershed.  “We can ask the question, where is the hydrology good or bad and where is the biology good or bad, and we create this contingency table at the bottom,” he said (referring to the slide above, right).   “We can then look at that and say in the areas where biology is poor (it’s below the .79 threshold) and the hydrology is also poor, those are areas where I want to prioritize my flow management actions because I’ve got poor biology and poor hydrology both occurring.  In contrast, if I have poor biology and good hydrology, with the other stressors, that may be the case where hydrology isn’t the most important factor, maybe there’s some other stressor at play there, and so we can use this to prioritize management actions.”

The information can be used for scenario analysis as well.  “This is the Alvarado Creek sub-watershed within the San Diego River,” Dr. Stein said, referencing the slide below, left.  “In this case, this is an area that is already built out.  The current effective impervious cover for that built out watershed is about 50%.  What we want to do is we want to try and manage that hydrology down to a level where we think it’s going to be protective of the instream biology.  So we evaluated a series of different BMP implementation scenarios that change the effective impervious cover – it’s either 5, 10, or 25%, as well as looking at this 85th percent capture, which is what’s in their current permit.  We asked how do we think that’s going to change the hydrology and will it affect the biology and so you can see here is the model that shows the current hydrology and what it would look like under 2% impervious cover, which is what we consider reference conditions.”

The table (above, right) picks out a few of those flow metrics that I showed you.  It really just shows what the target is on the far right, and then it shows you under the 2, 5, 10, and 25% impervious cover BMP scenarios, how would those metrics perform relative to the target.  The result is then color coded.  If they are in green, it means you’ve met your target, so you can start to look at these types of different scenarios and say how much BMP implementation do I need to put in in order to manage the hydrology down to a level where it’s going to create those healthy bioassessment scores.  And then we also look at the 85% capture which in this case, was probably the best performing BMP option.”

Dr. Stein said that as an output of the workgroup, they have a series of data products that are available, including the GIS for the stream classification, all of those hydrologic models, the ensemble of models and the model selection tool so they can be used for other purposes, as well as all of the biological models.  They are all available so they can continue to be built on and used for other regions, he said.

In terms of next steps, the workgroup is continuing their work.  Similar to the work that has been done for benthic invertebrates, they are starting to use benthic algae as another bioindicator.  They are also doing some preliminary work looking at some of the local fish and general riparian habitat using different models, and doing additional causal assessment work to improve the ability to discern flow versus habitat.

One of things that we are particularly interested in is this concept of shifting baselines,” Dr. Stein said.  “We know that the flow regimes of our streams are shifting due to climate change and changing water use practices, so we have to think about how this sort of slow gradual change in the hydrologic regime might affect the biology, and then ultimately is to integrate this concept into the statewide overall framework.”


Question: Can you list a few of the entities who are actually being able to use some of your findings?  And then, secondly, you mentioned the effects of water not going into streams from stormwater capture and so forth, but I’m curious … has anybody looked at the implications of adding State Water to the Santa Clara River watershed? 

Dr. Stein:  “The first question has to do with the agency entities, so in Southern California, most of the folks I’ve worked with have been regional boards and the local municipalities, so the regional boards on the regulatory side and the municipalities of sort of the management entities.  A little bit with the Department of Fish and Wildlife, but those are the main folks.”

Statewide, there are basically four entities we’ve been working with, which are Department of Fish and Wildlife and their instream flows program, DWR, and there are two parts of the water board that we’ve been working with.  One is the Division of Water Rights who has a process that is going on to develop regional and instream flow evaluation methods, and then the Division of Water Quality which is more focused on some of the things I talked about.  Those are the main entities that we’ve been interacting with.”

In terms of the second question, so the hydrologic change going forward, it can be an augmentation or a depletion essentially, so that’s why I showed the logistic regression curve, as we split those at sort of the zero point and we generate one of those analysis for both the decreasing direction and the increasing direction, because we want to try and understand biologically, do we see more of a relationship in the declining where you have less water or an augmentation where you have more water.  The patterns and the shapes of those curves can be quite different, so we’re trying to get a little bit of understanding of in which direction is there more sensitivity.”



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!

Print Friendly, PDF & Email