1. Can we have our herring and eat our salmon, too
Can we have our herring and eat our salmon, too? A qualitative approach to modeling trade-offs in Puget Sound
Tessa Francis
Puget Sound Institute, University of Washington Tacoma
Chris Harvey Northwest Fisheries Science Center, NOAA, Seattle, WA
Mike Carey USGS, Anchorage, Alaska, USA
Acknowledge coauthors

2. Understanding tradeoffs
Rodger Jackman/ gettyimages.com
As you know, we are increasingly turning to ecosystem-based approaches to managing the world’s ecosystems, including marine ecosystems.
And as you also know, the major task of EBM is to simultaneously consider and manage multiple ecosystem components to protect and sustain the services they provide.
As we get deeper into this process, a challenge that is emerging is to identify and evaluate the tradeoffs that exist among different ecosystem services, or specific recovery targets.
One obvious example of this in coastal ecosystems is letting people live on coastlines, and protecting habitat, or water quality, or ecosystem function.

3. Climate and oceanography Assessment and policy decisions
EcoPath with EcoSim
Harvey et al Estuaries and Coasts
Biogeochemistry
Hydrographic
submodel
Community
Habitat
Fisheries
Climate and oceanography
Management
Assessment and policy decisions
Slide courtesy of Isaac Kaplan
The Atlantis Ecosystem Modeling Framework
We are developing approaches to assessing these tradeoffs, and one such approach is the development of ecosystem models, some of which you may be familiar with, including EwE and Atlantis.
The downside to these models is that they are time-intensive to build, data-hungry, and so complex that drilling down to individual interactions that we might be interested in is pretty tricky.
So we are working to develop a simpler tool for use in evaluating tradeoffs, and specifically food web tradeoffs.
Isaac Kaplan, Chris Harvey, Phil Levin, Jason Link, Howard Townsend
NOAA NMFS
Beth Fulton CSIRO Australia

4. Are there tradeoffs among recovery targets?

5. Qualitative Modeling (-) (+) Community Matrix -1 1
So we are developing a qualitative modeling approach to assessing foodweb tradeoffs in Puget Sound. And the way this works is very simple.
You start with your food web of interest. Here we have a predator and two prey species.
Community Matrix -1 1

6. Qualitative Modeling Predictions and Reliability-1 1 20 10
-10 21
-20 1
0.8
0.7
0.5
0.6
0.4
So we build the community matrix based on diet data, and use this community matrix to predict how an increase in the population of any species in the food web, will affect each other species. So for example, how will an increase in rockfish affect forage fish, taking into account all the feedback loops between each species pair. From diet matrix we create community matrix. From community matrix we calculate the adjoint, which is the sum of all feedback loops between each species pair. The sign of the number is the directional response. For example, if we wanted to know
Community Matrix A
Predicted Response (pos, neg, neutral)
Σ Links between pairs
adj(-A)
Reliability
Σ Links between pairs
Total links in web
adj(-A)
perm(minor(abs(A)))
Dambacher et al. 2003, 2007

7. Again, a reminder that I’m interested in the tradeoffs between herring and chinook salmon in puget sound, both of which are in decline in puget sound and both of which have recovery targets set for them.

8. Qualitative Modeling Step 1: Simplify
So our first step is to create a simplified food web that contains just the species that we’re interested in, because we are focusing on specific key interactions.
Harvey et al Estuaries and Coasts

9. Diet > 0.1 Predation > 0.1 Pinnipeds Salmon Herring Forage fish
Juv. salmon
Juv. herring
Macro-
Zooplankton
And the way we create that simplified food web is to start with our species of interest, chinook and herring.
And we create their food web based on a diet threshold of 10%.
So we include predators for whom salmon/herring comprise 10% of the diet, and species that represent 10% of the diet of herring/chinook.
So once we had our food web constructed, we explored the impact of management actions that could potentially be employed to increase salmon or herring, and asked whether management actions to increase one or the other species in recovery would force us to consider a tradeoff with the other recovery target.
Copepods
Phytoplankton

10. ? ? Scenario: increase herring Pinnipeds Salmon Herring Forage fish
Juv. salmon
Juv herring
Macro-
Zooplankton
Copepods
Phytoplankton

11. ? ? Scenario: decrease pinnipeds Pinnipeds Salmon Herring Forage fish
Juv. salmon
Juv. herring
Macro-
Zooplankton
Copepods
Phytoplankton

12. Tradeoffs between recovery targets
Scenario
Winners
Losers
Tradeoff
Pinnipeds
Juv. herring
Phyto
Herring - No
Pinnipeds
Adult salmon - No
Juv. salmon
Herring
Yes
Pinnipeds
Zooplankton
Copepods
Forage fish
Copepods
Juv. herring
Herring
Zooplankton
Phyto
Adult salmon
Yes
So my take-away from this is that there are some management actions, most notably those affecting the juvenile stages of salmon and herring, where we may want to consider that recovery of one species may hamper recovery of the other.

13. EcoSim scenario: Increase eelgrass by 100%
Plummer ML et al Ecosystems

14. Compare to Ecosystem Model
Pinnipeds
Salmon
Herring
Forage fish
Juv. salmon
Juv. herring
Macro-
Zooplankton
Copepods
Eelgrass
Phytoplankton

15. Bottom Line
Qualitative models can be used to identify tradeoffs in EBM.
Qualitative models can corroborate results from fancy ecosystem models.
These models are useful in data-poor systems, and to help guide recovery strategies.

16. Opportunities Suggest implications of recovery actions
Spatially-explicit food web scenarios
Lingcod
Crabs
Birds
Explore hypotheses about food-web interactions, management actions… easily
Compare with/groundtruth complex ecosystem models
Rodger Jackman/ gettyimages.com

17. Thanks.

18. Strong Diet Overlap during Critical Growth Period
Between juvenile Salmon spp. and Herring
High overlap between pink, coho, herring and chinook salmon.
Euphausiid
Chum Salmon show
Least overlap with
Other salmon & herring 18

19. Simulated Predation Demand by Resident Chinook in Puget Sound
FL > 300 mm
-Herring are the predominant prey of Chinook >300 mm throughout the year
-Over 1,500 metric tons of herring consumed by resident Chinook annually in PS

20. Source: WDFW

21. Size-selective predation on Herring
by Juvenile & Resident Chinook Salmon
Herring found in age-0
Chinook stomachs
are significantly smaller
than those sampled
Concurrently with MW Trawl
Same for larger resident
Chinook

22. Conclusions
Tradeoffs between salmon and herring exist under some circumstances
Via competition for macrozooplankton
Dependent on abundance of salmon
Qualitative modeling can highlight specific key interactions/biological parameters that may determine tradeoffs
Zooplankton abundance
Top-down pressure by juvenile salmon

23. Compare qualitative model with ecosystem model
California Current groundfish: key prey species
Small phyto
Micro zooplankton
Surface seabirds
Diving seabirds
Pinnipeds
Deposit feeders
Large phyto
Shallow Small Rockfish
mackerel
Euphausiids
Chinook
Juv Bocaccio
Juvenile Shallow Rockfish
Bocaccio
Benthic Grazers
copepods
Ecosystem Model (Atlantis)
Qualitative Model
Krill
Forage Fish
Other rockfish
Hake
Myctophids
Benthic grazers
Krill
Forage Fish
Other rockfish
Compare with straight diet data: prey
Compare with Atlantis, an ecosystem model of CalCurrent, with a food web module that estimates predation mortality on 4 groundfish species based on predator diets, consumption rates, biomass, ontogenetic shifts in diet and migration patterns.