Shallin Busch 1. Cameron Ainsworth 2, Jameal Samhouri 1 William Cheung 3, John Dunne 4, Thomas Okey 5

Transcription

1 Shallin Busch 1 Cameron Ainsworth 2, Jameal Samhouri 1 William Cheung 3, John Dunne 4, Thomas Okey 5 West Coast Vancouver Island Aquatic Management Board 1. NOAA - Northwest Fisheries Science Center 2. Marine Resources Assessment Group Americas Inc. 3. University of East Anglia 4. NOAA Geophysical Fluid Dynamics Lab 5. WCVI Aquatic Management Board

2 Uncertainty in estimating response to climate change IPCC 2007

3 Biological impacts Primary productivity Plankton community structure Geophysical Fluid Dynamics Laboratory CM2.1 Deoxygenation Species range shifts Hake Ocean acidification Cheung et al (Fish and Fisheries) Orr et al 2005 Ainsworth et al., in review

4 Three NE Pacific food web models Southeast Alaska Northern British Columbia Northern California Current Guénette 2005 Ainsworth et al., 2008 Field 2004

5 Ecopath with Ecosim Polovina, Christensen, Pauly, Walters

6 Addressing uncertainty Climate Model DO Range shift Ocean acidif. 1 0 prod. Plankt. comm. Biomass Vuln.

7 Model uncertainty Climate Model DO Plankt. comm. Range shift 1 0 prod. Ocean acidif. Biomass Vuln. Climate Model Climate Model

8 Model uncertainty For all impacts, we ran scenarios of 3 different strengths: Scenario strength Nominal Moderate Severe Effect 50% Effect Effect + 50% Forcing function Time Ainsworth et al., in review

9 Model uncertainty Models vary in their sensitivity to climate scenarios Nominal Moderate Severe Ainsworth et al., in review

10 Model uncertainty Functional s vary in their sensitivity to climate scenarios Southeast Alaska N. British Columbia N. Cali. Current Ainsworth et al., in review

11 Process uncertainty Climate Model DO Plankt. comm. Range shift 1 0 prod. Ocean acidif. Biomass Vuln.

12 Cumulative effects Additive= + + Geomean=( x x ).3 Synergy=Additive + Geomean

13 Cumulative effects Effect A Geometric mean Effect B Additive Synergy

14 Diversity scores impacted by summation technique 20% SE Alaska N. BC N. Cali Current % change from baseline 0% -20% -40% -60% -80% -100% Kempton s Q Shannon Kempton s Q Shannon Kempton s Q Shannon Geomean Additive Synergy

15 (Dis)agreement among techniques on magnitude of change SE Alaska N. British Columbia N. Cali Current Geomean Additive Synergy Geomean Additive Synergy Geomean Additive Synergy Forage fish Shrimp Pelagics Salmon Marine mammal Demersal fish Squid Zooplankon Rockfish Flatfish Crabs Seabird Shellfish Sharks Primary producers Jellies % change from baseline <-50% -50% -10% >50% 10%-50% -10% 10% Not calculated

16 (Dis)agreement among techniques on magnitude of change SE Alaksa N. British Columbia N. Cali. Current 20% 47% 13% 13% 7% 19% 31% 13% 25% 12% 25% 19% 6% 19% 31% All same All different Synergy & Additive Synergy & Geomean Additive & Geomean

17 Process uncertainty Combinatorial techniques greatly affect magnitude of predicted change Agreement among techniques when results ranked is slightly better Must consider how well combination techniques capture physiological and population-level process of interest

18 Effect size uncertainty Climate Model DO Plankt. comm. Range shift 1 0 prod. Ocean acidif. Biomass Vuln.

19 Ocean acidification scenarios Forcing function Time Certain Shrimp Euphausiids Epifauna Infauna Carn. epibenthic inverts Uncertain unidirection Crabs Copepods Infaunal detrital inverts Epibenthic inverts Carnivorous zooplankton Small zooplankton Marine plants Uncertain multidirectional Jellyfish Phytoplankton Microzooplankton Large zooplankton Busch et al., in prep.

20 Ocean acidification scenarios Linear change over 50 yrs. Climate scenario Effect size Nominal Moderate Severe Small 5% 10% 15% Large 25% 50% 75% Ainsworth et al., in review

21 Certain s Uncertain positive Uncertain negative Change from baseline biomass 100% 80% 60% 40% 20% 0% -20% -40% -60% -80% -100% Total ecosystem Shrimp Northern British Columbia Shellfish Crabs Zooplankton Primary producers Jellies Squid Fished species Unfished species Demersal fish Pelagic fish Forage fish Salmon Rockfish Flatfish Sharks Marine mammals Seabirds

22 Certain s Uncertain positive Phytoplankton positive Uncertain negative Phytoplankton negative Change from baseline biomass 100% 80% 60% 40% 20% 0% -20% -40% -60% -80% -100% Total ecosystem Shrimp Northern British Columbia Shellfish Crabs Zooplankton Primary producers Jellies Squid Fished species Unfished species Demersal fish Pelagic fish Forage fish Salmon Rockfish Flatfish Sharks Marine mammals Seabirds

23 Effect size uncertainty Importance of understanding climate change impacts on primary producers Ecosystem impacts of change in non-primary producers can be dwarfed by change in primary producers

24 Parameter uncertainty Climate Model DO Range shift Ocean acidif. 1 0 prod. Plankt. comm. Biomass Vuln.

25 Biomass (tonnes) Forage fish 8 Monte Carlo methods Biomass Ecopath with Ecosim Monte Carlo routine Coefficient of variation dependent on functional 15% phytoplankton, zooplankton & benthic meiofauna 5% all others Uniform distribution, 200 trials Vulnerability Manually implemented in Ecosim Every predator-prey interaction varied independently Coefficient of variation: 50% all s Uniform distribution, 200 trials Year Results: 1) functional, 2) aggregated functional

26 Monte Carlo results Biomass Monte Carlo: Southeast Alaska, benthic invertebrates Biomass Monte Carlo: N. Cali Current, small pelagic fish Mean 25%, 75% Compounding impacts of trophic dynamics

27 Monte Carlo results About 30% of the functional s and aggregated functional s have >10% change in biomass Biomass change for vulnerability Monte Carlos less than forced variation Magnitude of change Number of s % change biomass for sensitive s SEA NBC NCC 75% 25% Biomass Vulnerability

28 Sensitive functional s: biomass change >10% Functional s All models 2 models Salmon Sharks Pacific Ocean perch Forage fish Flatfish Herring Halibut Rockfish Crabs Pacific cod Shrimp Arrowtooth Epifaunal inverts Phytoplankton Infaunal detritivores Aggregated s All models Benthic inverts 2 models Large and small pelagics Birds Flatfish 1 model Elasmobranches

29 Parameter uncertainty Knowing the biomass and vulnerability of some species s matters more than others Sensitivity to biomass estimates is greater than to vulnerability estimates Sensitivity of specific functional s to parameter uncertainty is fairly well captured in aggregated s

30 Conclusions Modeling ecosystem impacts of multiple climate change impacts is messy Understanding how uncertainty impacts our results can help us target assumptions and data inputs that may matter most Recognize that modeling exercises tell us the type of changes we may expect, not the magnitude or the specific effects