In-Person Meeting

February 22, 2019

Participants

Jim Peterson, Adam Duarte, Erin McCreless, Chris Kwan, Denise Reed, Sheila Greene, Shelley Ostrowski, Mike Hendrick, Rod Wittler, You Chen Chao, Corey Phillis, Mike Urkov, Ben Geske, Emanuel Rodriguez, Scott Hamilton, JD Wikert, Shaara Ainsley, Mike Beakes, Matt Nobriga, Brad Cavallo, Chandra Chilmakuri, Campbell Ingram, Shawn Acuna, Javier Miranda, Sam Luoma

Recap: Where we are in the process

Jim reviewed the SDM process to date (see meeting slides for overview). We expect that all modeling will be completed by the end of February. We aim to evaluate and grade the DSM and produce a final report by the end of March.

Water availability results

Emanuel Rodriguez (FlowWest) demonstrated the app he has built to show water exports/deliveries and variability results for each management scenario. You can explore the app and compare results for the different scenarios here.

Delta Smelt models

Scott Hamilton - Limiting factors model

Scott Hamilton presented the results of his limiting factors model for delta smelt, for each of our management scenarios. A more complete description of the model and results for each scenario will be distributed to the group soon.

  • The goal of the LFM is to identify what the limiting factors are and how these vary across years
  • The challenge is to find a balance between a simple model, which is better statistically, and a more complicated one, which can help understand what factors are driving observed patterns and assess the impact of management scenarios. Scott's model is more in the second category.
  • Scott built submodels to address specific scenarios, e.g. adult salvage and food production
  • Silverside predation. Scott used the results from Baerwald et al. (2012) and Schreier et al. (2016), which investigated the presence of delta smelt DNA in the guts of silversides, to estimate silverside predation rates on delta smelt. His extrapolations suggest that silversides have a population-level effect. This influences the effectiveness of management actions because actions that don't address silverside predation may have limited impact.
  • Found no benefit from the X2 actions
  • Plugged flow data into the food model, which feeds into the LFM. Also looked at the effects of X2 actions on salinity and suitability of EC conditions. Found that increased food supply and better salinity conditions still didn't have a positive effect on smelt, and the model suggests that this is due to predation by silversides.
  • Discussion of Yolo notch scenarios. Ted Sommer believes that slow flows through the Yolo Bypass could lead to a big increase in food supply. But, during big flow events, food may get washed out of the channel – so counterintuitively, more flow could mean less food. Any effect probably wouldn't happen until water starts to drain. Ideally we would evaluate this with daily data, but CalSim outputs are monthly – so we have to interpret any results with caution. There was a question about what exactly this action is; are we just adding a notch, or controlling exactly how much water flows through the system at specific times? In the current phase we modeled the preferred alternative (notch). In the next stage of the project we should be more specific about how much water and when.
  • OMR flows. Scott developed a separate adult salvage model for this, in which salvage was estimated as a function of delta inflow, the interaction of OMR and turbidity and the strength of the first flush. The model predicted increases of 4-9% over the FMWT index resulting from the OMR actions.
  • Conservation hatcheries.
    • Modeling suggests that adding adults in November was beneficial but not hugely so. This is partly because added adults would have lower survival, and also because of silverside predation. Model suggests that silversides reduce survival by 92%, although this involves a lot of speculation.
    • Adding fertilized eggs – increased FMWT by 50% and had a large beneficial effect. However, the model does not account for mortality in the life stages between egg and adult. You can protect eggs but not free-swimming larvae (e.g. predation). Scott will add this into the model.

Rose/Smith Modified bioenergetics model

Jim Peterson presented the modifications we've made to the Delta smelt bioenergetics model, which was first published in Rose et al. (2013) and modified by Will Smith (FWS). See meeting powerpoint for an overview of the model, modifications, and results.

The data and R code for the models are available here for people to explore.

  • Model calibration. Calibrated to FMWT index. The model hit the middle values (years) well, but missed the very low and high values.
  • Result that removing the I:E ratio is actually bad for delta smelt. Matt said that the I:E guidelines were created for salmonids in the San Joaquin basin and increase flow in spring. Removing the I:E ratio means there's more pumping, which leads to higher salvage. There is some confusion here because the modeled entrainment takes place in June but the flow actions are in April-May. We need to know the actual observed flows with I:E removed.
  • Yolo big notch action. Didn't add food as a result of this action in the model. Scott thinks this action would have a big impact on food, especially in spring. Not sure how far this effect would reach geographically.
  • Discussion of Yolo pulse flow food action. Jim said it is problematic to evaluate food supplementation without including density dependence in the model – this isn't very realistic. Matt said there aren't enough delta smelt to generate their own density dependence. There is density dependence in the system, but not from food; rather, it's imposed on them externally. Our model won't catch that. Food supplementation could help get past Allee effects. Food supplementation will help other fish species as well, which will in turn influence delta smelt populations. Ideally we need to build an ecopath/ecosim model of the whole community, but that's outside the scope of the SDM effort.
  • The model doesn't capture the population booms that happened in 2001, 2002, and 2011. The 2011 data is consistent with Scott's model; the cohort of silversides was different than in other years. No one in the room had an explanation for 2001-2002, although it was noted that the population was high in 1999 and remained high for a couple years after that.
  • Scott commented that we need to be careful in using these models to evaluate actions, when we know the models can't explain important trends like the 2001/2002/2011 population booms. We need to consider uncertainties in the models and think about important things we might be missing. For example, the BiOp has reduced entrainment, and there's empirical support for this but the models don't account for it.

Chinook model

Adam Duarte presented the results of the chinook model, which was modified from the one developed for the CVPIA SIT project. See meeting powerpoint for background, details, and results. The routing and survival model, including data, R code, movement and survival parameters, maps, and routing and survival functions here for people to explore.

The visualization tool for routing and survival is available here.

  • We're trying to improve the calibration. It's possible we're modeling habitat incorrectly. Problems arise in places where fish interact – in the model, where fish interact we kill them by pushing them downstream. This is due to the rulesets for migration and habitat. We may need to change these rulesets.
  • The calibration for the winter run model is better because it just includes one location.
  • Spring chinook have two life histories we're modeling: The subyearling life history in which juveniles leave by May, and the yearling life history in which they leave the following fall.
  • Growth during summer. Some people in the SIT thought that productivity in the region is too low for fish to grow much, but there is disagreement within the SIT about this. Corey said that screw trap data suggest that size distribution is the same as in the fall.
  • Matt said that exit isn't linked to flow increases in the natal tributary. Adam said that based on screw trap and trawl data, fish tend to leave regardless, so the model uses this rule set.
Last updated on Dec 5, 2018