1. The Salmonid Population Viability Project Lahontan Cutthroat Trout Viability

2. The Salmonid Population Viability Project Team
Seth Wenger, Doug Leasure, UGA River Basin Center
Helen Neville, Dan Dauwalter, Robin Bjork, Kurt Fesenmyer and Jean Barney, Trout Unlimited
Erin Landguth, University of Montana
Jason Dunham, Nate Chelgren, USGS
Dan Isaak, Charlie Luce & Zach Holden, USFS Rocky Mountain Research Station
Mary Peacock, University of Nevada-Reno
Joe Glassy, Lupine Logic, Inc.
Doug Peterson, US Fish & Wildlife Service

3. The motivation: better conservation planning
How “healthy” is each population?
Which would benefit most from what management actions?
Which ones are hopeless?
Where could we reintroduce?
How might climate change affect populations?

4. What we really want: Estimates of population viability

5. In practice… We often use surrogates or rules of thumb
Habitat size (Hilderbrand and Kershner 2000: 8-25 km)
Land cover, land use
Climate vulnerability
These surrogates may not be validated, and often cannot be validated
Trout Unlimited’s Conservation Success Index for Westslope Cutthroat trout

6. Why aren’t we using PVA more?
Data-intensive: years of consecutive abundance estimates are the minimum for each population.
Each population is modeled separately. Limited options for extrapolation.
Alternative modeling approaches (e.g., RAMAS) require parameters that must be estimated from literature values– make limited use of field data.

7. Alternative approach: STPVM
The Spatio-Temporal Statistical Population Viability Model (STPVM)
PVA on many populations at once
Provides quantitative extinction estimates for each population
Can test effects of management actions
preliminary

8. How STPVM works:
Models are fit using actual observations of organisms across populations simultaneously
Applicable to any species with appropriate data; not limited to fish P1 P3 P2 P4 ?
NPS.gov

9. The advantages
Models produce actual estimates of viability for each population– this is what we really want
Enable range-wide quantitative assessment of overall status – also what we really want
They use all available field *data*
Simultaneous estimation across multiple populations lets us borrow information to estimate viability for populations with little/no data
Allow us to explore which variables across the landscape most affect viability
S. Walsh

10. Applying ST-PVM to Lahontans
Lahontan cutthroat trout are ideal for ST-PVM because they are found mostly in small, isolated populations
They have been intensively sampled (lots of data)

11. Building the fish database
Can point out new eradication and metapop tables
The full database will be available to project partners

12. 154 populations total 63 FWS conservation populations
1,907 sampling sites
6,227 sampling events
33,750 individual fish
Fish data sources: NDOW, UNR (M. Peacock, J. Dunham), ODFW, CDFW, TU
So far modeled 53 populations of LCT with 1,115 sampling sites, 3,209 sampling events, and 114,019 individual fish collected.

13. ST-PVM: A Hierarchical Model
Observation Model
Sampling Model
Process Model

14. ST-PVM: Observation Model
Sampling Model
Process Model
1995
2001
Site
Pass 1
Pass 2
Pass 3 A 10 7 2 B 23 12 C 1
Site
Pass 1
Pass 2
Pass 3 B 17 D 9 2 E 15 1
Population 1 …
1991
2010
Site
Pass 1
Pass 2
Pass 3 A 14 7 6 B C 9 8
Site
Pass 1
Pass 2
Pass 3 D 20 E 14 12 F 1
Population 2 … … …

15. ST-PVM: Observation Model
Sampling Model
Process Model 1
Detection at pass 1:
Drainage Area + Flow Conditions
Detection Probability
0.5 1 2 3
Pass Number

16. ST-PVM: Observation Model
Sampling Model
Process Model
Year: 1995
Fish: 12 ± 5
Fish: 31 ± 11
Fish: 42 ± 15
Fish: 8 ± 4
Lets us use # of fish caught to estimate # not caught, with error
Fish: 10 ± 7

17. ST-PVM: Sampling Model
Observation Model
Sampling Model
Process Model
Year: 1995
80 m
80 m
100 m
30 m
Population Size in 1995:
757 ± 132 fish
Proportion of Population Sampled:
13.6%
± 5
50 m
Total fish at sampled sites: 103 ± 42
Total sampled length: 340 m
Population Extent: 2,500 m

18. ST-PVM: Process Model
Observation Model
Sampling Model
Process Model
Based on a modified Ricker Model, with covariates
Pop. Growth Rate (r)
Carrying Capacity (K)
Density-Independent
Mortality (d)
Stream Temperature
High Flow Magnitude
Low Flow Magnitude
Population Extent
Brook Trout
Low Flow Magnitude
NDVI
Severe floods
Wildfires

19. ST-PVM: Population Assessments
Observation Model
Sampling Model
Process Model
Population Size
preliminary

20. ST-PVM: Population Assessments
Observation Model
Sampling Model
Process Model
Population Size
preliminary

21. Population Viability: 2025 Extinction Probability
For current LCT Conservation Populations as well as other historic habitats
preliminary

22. The Covariates
Ecological Forecasting

23. Covariates: Normalized Difference Vegetation Index (Landsat, 1985-present)
Active photosynthesis and vegetation
1992 – Dry year
2011 – Wet year

24. Maximum NDVI (for catchment) by Population
Dry year
Wet year
high precipitation year

25. Covariates- Stream Temperature
Developed new stream temperature database,
also available to partners
Data from 14 agencies
502 thermograph sites
741 site-years

26. Covariates- Stream Temperature
Annual estimates of mean August stream temperature for every 1km stream segment
Current and future climate conditions

27. Covariates- Stream flow
Currently from the VIC macroscale hydrologic model
Working on alternative, updatable approach

28. Stream drying Many streams are intermittently dry
Jason Dunham is deploying sensor networks range-wide to model desiccation
Will incorporate results in future

29. Covariates: Brook Trout Densities

30. Scenarios: what if we…

31. Management scenarios Brook trout suppression/removal scenarios
Reintroduction potential of different habitats
Metapopulation reconnection
Cattle fencing/riparian restoration?
Thermal degradation
We are building a tool to let users examine these alternatives.

32. What if we remove brook trout?
BKT: SD = +4
preliminary

33. What if we remove brook trout?
BKT: SD = 0
preliminary

34. What is best way to reintroduce?
10 fish
100 fish
preliminary

35. Metapop Reconnect

36. Metapop Reconnect
preliminary

37. Climate Change We can estimate changes in covariates in coming decades
Still working on flow projections
We can’t do desiccation very well, yet but coming soon…

38. Next phase
We hope to work with LCT management teams to provide a useful tool for a “Recovery Roadmap”
Hand over LCT database for continued update
Publish in peer-reviewed journal
Continue to update models and scenarios
Beginning data phase for Bonneville cutthroat trout and interior redband trout

39. Acknowledgements: Funding
NASA grant number NNX14AC91G
Bureau of Land Management, National co-op
Trout Unlimited
US FWS
That’s it, thanks!