Updated_Statement_of_Work_NFFP73001804342.doc

STATEMENT OF WORK

Purchase Request Project: Spatiotemporal modeling of stream networks Background

To estimate spatial variation in density for single or multiple species over time and place, Thorson and Barnett (2017) developed a Vector Autoregressive Spatio-Temporal (VAST) model that builds upon delta-generalized linear mixed modeling techniques (Thorson et al. 2015). VAST uses a delta model to decompose catch rates into spatiotemporal encounter-probability and positive catch rates. VAST is now used for univariate (single-species) or multivariate (multiple species/ages/sizes) spatiotemporal index standardization in stock assessments and population models around the world. VAST uses Gaussian Markov random fields to approximate spatial and spatiotemporal variation. This model assumes that correlation in spatial variation decreases as distance increases.

In stream networks, connected streams are probably more related in habitat and species distribution than unconnected streams in close proximity. Spatial and spatiotemporal models of stream networks are vital in the management and conservation of freshwater and anadromous fishes, often doubly threatened by terrestrial development along connected waters upstream. Here we propose to add stream networks as a spatial model in the VAST package due to the wide range of features allowing for flexible model specification already included in VAST.

The stream network spatiotemporal model would be applied to Columbia River salmonids, and Oregon coastal coho. Like other anadromous fish, salmon populations in Columbia River and the Oregon coast have been and continue to be impacted by development on land. The spatiotemporal stream network model would be used to estimate how Oregon coastal coho and Columbia River Chinook salmon populations have responded to management programs (e.g. channel restoration) along with other continued habitat changes.

Tasks

1) Include stream network as a spatial process in VAST with test dataset

2) Apply VAST with stream network feature on Columbia River salmonids (Oregon coastal coho and Upper Columbia River chinook).

Deliverables Overall Deliverables shall include the following:

· Code repository and documentation for use in Github.

· Written report that demonstrates the tool for different areas and scenarios, i.e. simulation tested and applied on Columbia River salmonids.

Schedule

The contract shall be completed within nine (9) months of awarding the contract. The period of performance is Date of Award plus 9 months.

The Contractor shall furnish all labor, equipment, and materials necessary to perform the requirements in the Statement of Work in accordance with the terms, specifications, and provisions of this solicitation and contract.

Required experience and skills:

· The contractor must have one year working knowledge of VAST software currently developed by Thorson et. al. (2015).

· The contractor must have at least five years experience working with Template Model Builder (TMB, Kristensen et. al. 2016) and complex spatial models is R.

References

Kristensen, K., Nielsen, A., Berg, C., Skaug, H., Bell, B. 2016. TMB: Automatic Differentiation and Laplace Approximation. Jr. of Statistical Software 70 (5): doi: 10.18637/jss.v070.i05 Thorson, J. T., and L. A. K. Barnett. 2017. Comparing estimates of abundance trends and distribution shifts using single- and multispecies models of fishes and biogenic habitat. ICES Journal of Marine Science 74(5):1311–1321.

Thorson, J. T., A. O. Shelton, E. J. Ward, and H. J. Skaug. 2015. Geostatistical delta-generalized linear mixed models improve precision for estimated abundance indices for West Coast groundfishes. ICES Journal of Marine Science 72(5):1297–1310.