1. SALMONID (BROOK TROUT) POPULATION PERSISTENCE Development of a DSS Ben Letcher USGS, Conte Anadromous Fish Research Center, Turners Falls, MA Keith Nislow USFS, Northern Research Station, Amherst, MA

2. Why care about brook trout?  Widespread  Found in most northeastern streams with decent habitat  Small isolated streams, rivers, lakes, bogs, sea-run…  Indicator of water quality  Temperature, acidity  Sensitive to land use change  Mobile  Habitat connectivity important – what’s the key scale?  Important component of aquatic community  Abundant  Predation, food source, nutrient dynamics  Invaders in the west  Important to understand population dynamics  Important fishery  Native and stocked  Indicator of functioning habitat  Sensitive species, harbinger  Good data available  Distribution, local abundance  Individual-based studies

3. Who cares about brook trout?  Eastern Brook Trout Joint Venture  Coalition of state and federal managers  The Nature Conservancy  Connecticut River program  USFWS  LCC project  USFS  Long-term funding  Trout unlimited  Sea-run brook trout coalition

4. Threats to population persistence  Habitat fragmentation  Isolated populations  Water withdrawals  Seasonal effects of stream flow  Land use/land change  Riparian buffer, impervious surfaces  Climate change  Air temperature and precipitation affecting: Stream flow and temperature  Interactions with climate change

5. Overall goal  Understand how populations work  What affects local population persistence?  Endpoint – probability of persistence after x years  Body size distributions  Develop DSS tool for managers  Probability of population persistence under varying management scenarios Eastern brook trout joint venture, 2007

6. LCC project tasks  Task 1: Hierarchical modeling framework to account for multiple scales and sources of uncertainty in climate change predictions  Task 2: Statistical models to predict stream flow and temperature based on air temperature and precipitation.  Task 3: Incorporate climate change forecasts into population persistence models  Task 4: Develop a decision support system for evaluating effects of alternate management strategies in the face of climate change.  Task 5. Develop curriculum and run training workshops for users of the decision support system.

7. Approach DataAnalysisModelSimulation Management tool  Synthetic data collection and analysis to:  Account for multiple sources of uncertainty  Allow error propagation  Provide answers in form of statistical distribution  How certain are we of result? Uncertainties Measurement, Observation Process [survival…] Inputs [environment, GCC] Run-to-run Outcome [Persistence]

8. Approach DataAnalysisModelSimulation Management tool  Fine-scale data collection at multiple sites  ~ 1 km, 20-m units  Seasonal  Tagged individuals, >35,000 since 1997  Model dynamics and uncertainty using Bayesian estimation  Growth  Survival  Reproduction  Movement  Combine statistical models into simulations  Link components- interactions  Develop management tool - DSS  Web-based  Evaluate alternate management strategies

9. What questions can we address?  Habitat fragmentation  Which barriers do we prioritize for removal/repair?  Water withdrawal  How much water can be extracted?  Importance of water source  How does extent of groundwater input affect persistence?  Climate change forecasts  What are the effects of variation in stream flow, temperature?  Interactions  How much will effects of isolation and water supply be magnified under GCC?

10. Approach Reproduction Body growth Survival Movement Age structure Body size distributions Population processes Abundance N e, N b Environment Outcome Stream Temperature Stream flow Habitat Fish community Catchment scale model (< 1 Km) Density dependence

11. Probability of persistence Fish model Links to Terrestrial project Hydrologic model Drivers Climate change Fish model Seasonal setting Precip, air T Stream flow, water T Resulting DSS: evaluate alternate management strategies Drivers Urban growth, etc Decadal setting Impervious… Succession Scenarios HabitatCaps Probability of persistence Seasonal Decadal

12. Near-term linkages between projects  Working with terrestrial group  Develop models for catchments in three large watersheds South, James River, VA Middle, ~Westfield River, MA North, Kennebec River, ME  Expand models to entire watersheds Collaborate with Eastern Brook Trout Joint Venture to estimate occupancy in specific catchments Collaborate with Dept C+E Engineering and terrestrial group to generate downscaled predictions of P and T and to develop hydrologic models

13. Project components  USFWS LCC  Tasks 1-5 1 Post-doc, Paul Schueller (Feb 2012 - 2013) 1 PhD student, Krzysztof Sakrejda (current – 2013) 1 Programmer (2012-2013)  USFWS LCC holdback  Flow modeling 1 post-doc, TBD (2011 – 2013)  USGS LCC  Assist with tasks 1-5 1 post-doc, Doug Sigourney (current – 2013)  Add in evolutionary dynamics 1 post-doc, Michael Morrisey (Jan 2011 - 2013)  TNC fragmentation project  Barrier removal/repair prioritization 1 post-doc, Cailin Xu (2008 - 2010) 1 PhD student, Paul Schueller (2008 – 2012) 1 Technician  USFS  Air temperature/stream temperature relationship Several technicians  UMass  Hydrologic model Dept of Civil and Environmental Engineering 1 post-doc, ~Austin Polebitski

14. Decision support  Good understanding of catchment and sub-watershed population persistence models in MA  USFWS LCC and TNC funding to  Scale up to watershed models  Identify minimum data needs to scale up to among- watershed models  Evaluate GCC effects on the landscape  Develop tools for managers to use Not limited to well-studied systems Apply to specific sites to address management needs Can we apply models range-wide? Need test sites Better local data = more realistic simulations

15. Decision support  How will the DSS work?  Identify management question  Identify space and time scales  Pick stream segments on web-based map  Load local data Environmental conditions, size distributions, community, genetics, movement data, etc  Simulation will automatically fine-tune model to local conditions  Run simulations  Evaluate alternatives

17. Approach – working across scales  Hierarchical models  Scale up  Propagate error  Watershed  Sub-watershed Catchment  Among-watershed  Multiple study sites

18.  Fine scale (10 Km)  Westfield River, western MA  100-m long sample sites  12 microsatellites  Pairwise Fst 0.11 – 0.24  Assignment tests using Structure  Similar results in NH, VT, VA  Catchment and sub- watershed scales  Need detailed data, ~ 1 km Spatial population genetics – what’s the right minimal scale?

19. Approach Sub-watershed scale model (1-5 km) Outcomes Connected catchment scale models Sub- watershed abundance and body size Movement patterns and catchment- specific production Movement Movement is observed with repeat sampling and PIT tag antennas

20. Approach Connected sub-watershed scale models Watershed scale model (5-50 Km) Watershed- scale abundance and body size Meta- population and genetic population structure Outcomes Movement Movement is observed with radio-tagged fish and is inferred with genetic data

21. Approach – broad questions  Do we need a detailed tagging study for each catchment?  Define catchment types Size, connectivity  Apply type to each unstudied catchment  Use existing data to tune catchment type model to local conditions (Hierarchical Bayesian modeling)  Can we apply models across watersheds?  Minimum local data needs?  Existing studies in MA, ME, NH  Planned for VA, PA/NJ (DEWA)  Workshop in Feb Defining these relationships is key

22. Progress to date  Development of linear models for  Growth, survival, movement  Population dynamics simulation incorporating existing estimates  Climate change scenarios  Not hierarchical High Q Low Q Control T x Control F = 174 yrs Stronger climate change effect 

23. TaskYear 1Year 2Year 3 1. Hierarchical model development 1. Determine statistical model structure 2. Estimate statistical model parameters 3. Develop simulation model based on #2 4. Combine all statistical models into simulation model 5. Incorporate simulation model into user interface 2. Air temperature/ stream temperature model 1 Deploy paired temperature recorders 2. Develop statistical model for paired temperature recorder data 3. Apply statistical model to selected watersheds 3. Climate change modeling1. Obtain downscaled stream flow and temperature predictions for the West brook 2. Develop model to apply downscaled estimates to selected watersheds 4. Decision support system1. Develop web-based user interface 2. Incorporate simulation model into web-based user interface 5. Model use/application workshops 1. Develop training tools 2. Conduct training class at USFWS Region 5 office

24. Probability of persistence Fish model Links to Terrestrial project Hydrologic model Drivers Climate change Fish model Seasonal setting Precip, air T Stream flow, water T Resulting DSS: evaluate alternate management strategies Drivers Urban growth, etc Decadal setting Impervious… Succession Scenarios HabitatCaps Probability of persistence Seasonal Decadal

26. Big questions  Which barriers should be prioritized for repair/removal?  How much water can be extracted from a stream?  Minimum flows  How do populations with very low effective population size persist?  Adaptation to isolation?  What is the minimum patch size for persistence? Strongholds or hopeless?  How will brook trout populations respond to climate change?  Range contraction  Effects of stream flow and temperature  Interactions between fragmentation and GCC  What are the best strategies to mitigate future challenges?

27. Challenges  Scale  How to scale up?  Space Define a population – how big? Where are the fish? Importance of local adaptation? Can we apply models to unstudied or poorly studied systems?  Time Can we apply short-term studies (1-15 years) to long-range forecasts (>50 years)? Timing of local adaptation?  At what organizational level do we collect data? Population Individual Genotype  Uncertainty  How propagate across scales? For example, downscaled predictions of temperature and precipitation are uncertain in space and time Need an approach to propagate this (and other) uncertainty all the way to projections of population persistence Eastern brook trout joint venture, 2007

28. NA LCC Landscape Conservation Cooperative