Validation of regional ocean model hindcasts for zooplankton biomass via comparisons with thirty years of field sampling data in a rapidly changing ecosystem

The eastern Bering Sea (EBS) is one of the most productive ecosystems on the planet and
produces the largest fish catch by volume in the United States. The EBS is currently
undergoing rapid ecological change, having experienced record-low seasonal sea-ice in the
past two years. To effectively manage fisheries in the face of rapid change, predicting the
response of the ecosystem to this forcing is imperative and requires the application of
modeling. The Regional Ocean Model System (ROMS) is widely used in the EBS, North
Pacific, and nationwide to conduct forecasts and hindcasts of ocean physics. In the EBS, the
10km resolution ROMS-NPZ model (Bering10K) includes multiple functional groups for
primary and secondary producers; however, there has been relatively little work validating
these components of the model. We propose to compare spatial and temporal patterns in
Bering10K predictions of secondary producers with field-sampling records of spatial and
temporal variation. In particular, field observations of zooplankton community have not
been compared to model output. A rich time series of data on zooplankton abundance,
collected with OAR, NMFS, and NOS support are available for this comparison. The lack of
comparison is critical in that zooplankton link changes in primary production to fish
through interactions, beginning with the first year of life. Survival of juvenile fish through
their first winter relates strongly to eventual recruitment and zooplankton availability
plays a key role; zooplankton remain key diet components of many commercially and
ecologically important species after recruitment to the fishery. This lack of model to
observational comparisons also makes it difficult to assess whether Bering10K hindcasts
and forecasts of secondary producers capture observed spatial and temporal variability,
and therefore provide credible estimates in locations and times without field data. This
uncertainty then makes it impractical to use the Bering10K model to inform the short-term
(1-5 year) forecasts/hindcasts of fish distribution that are currently used in the Alaska
pollock stock assessment and to develop adaptive sampling designs for fisheries (cod,
pollock) and protected species (e.g., subarctic cetaceans, including the critically
endangered N. Pacific right whale). This proposal aims to evaluate the Bering 10K output
by comparison to field observations of the zooplankton community. This comparison will
be evaluated using several statistical approaches and will identify areas for model
improvement. Furthermore, we will use the comparisons to test short time-scale scenarios
designed to simulate recent warming trends. The overall outcome of this research will be to
advance the capability of Bering 10K to predict zooplankton biomass, providing vital
ecosystem information that can be used to estimate fisheries production in a rapidly changing
Bering Sea.

Climate Risk Area: Marine Ecosystems