Each year fishery managers convene to plan the next season of salmon fisheries. This pre-season planning process is generally known as the “North of Falcon” (NOF) process, referencing the region north of Cape Falcon, Oregon which marks the southern border of active management for Washington salmon stocks. The NOF planning process coincides with the March and April meetings of the Pacific Fishery Management Council (PFMC), but includes additional public meetings as well as government to government negotiations. Final FRAM model runs are a joint exercise which encompass fisheries from both the PFMC and NOF pre-season negotiations and are usually completed at the end of the April PFMC meeting.
A wide range of model inputs are compiled and processed for FRAM pre-season model runs. Model inputs are received from the respective government agencies/tribes in Canada (Department of Fisheries and Oceans, DFO), Washington state (Washington Department of Fish and Wildlife, WDFW), twenty Western Washington treaty Indian tribes (member tribes of the Northwest Indian Fisheries Commission, NWIFC), the Columbia River Intertribal Fish Commission (CRITFC), Oregon (Oregon Department of Fish and Wildlife, ODFW), and California (California Department of Fish and Game, CDFG).
The process of developing FRAM model inputs begins with the forecasting of hatchery and wild/natural stock abundances during January-February. From February through April, fishery inputs for FRAM are generally developed and updated through the pre-season process following technical staff discussions and manager agreement for ocean fisheries through the PFMC and for Puget Sound tribal and state fisheries through NOF. Fishery-related mortality parameters, such as release mortality rates, drop-off, and mark-selective fishery parameters, are reviewed and confirmed at the start of the annual management cycle.
A variety of methods are used to forecast Coho abundances. These forecasts are usually developed by local/regional technical staff during one or more technical meetings where relevant forecasting information is exchanged. Final agreed-to forecasts are published in the PFMC Salmon Technical Team Pre-season Report 1; Stock Abundance Analysis and Environmental Assessment Part 1 (Stock Assessment and Fishery Evaluation; SAFE). Columbia River, Oregon, and California stock forecasts are provided through the Oregon Production Index Technical Team (OPITT) to FRAM modelers. Although Alaskan stocks and fisheries are in the Coho FRAM model, they are not modeled dynamically pre-season and instead use static scalers to represent a proxy of stock abundances and base period fishing. Canadian stock abundances are provided by PSC Coho Technical Team and Southern Panel representatives through an annual exchange meeting between the U.S. and Canada. This exchange meeting regularly occurs just prior to the first NOF salmon co-manager meeting in mid-March (between March and April PFMC meetings). Until these Canadian stock inputs are available, values from the previous year must be used which creates greater uncertainty during the annual planning process.
Abundance forecasts vary in units of measure. For example, there are forecasts of salmon returning to a terminal area (which implies some accounting for pre-terminal fishery levels), forecasts of ocean abundance (which is commonly fishery impacts plus escapement), and forecasts of abundances prior to any fishing impacts (which includes natural mortality and non-landed fishery related mortality). Each of these different types of forecasts need to be converted to the “unit of measure” used by FRAM, which is the abundance of each stock prior to fishing vulnerability and natural mortality. This starting Coho FRAM unit of measure is characterized as January-age-3 fish (JA3). FRAM stock abundances are input as a stock recruit scaler, which is calculated as the forecasted number of fish divided by the FRAM base period average abundance for each stock.
Many U.S. domestic and Pacific Salmon Treaty (PST) International Coho management objectives are specified in ocean age-3 abundance units (OA3, fishery impacts plus escapement). Thus, most of the forecasts are produced in terms of ocean age-3 abundance, which are then expanded by 1.2317 to account for natural mortality and provide an estimate of age-3 fish abundance in January of the fishing year (OA3 * 1.2317 = JA3). The value 1.2317 is a result of a two-step conversion due to a prior December age-2 (DA2) unit of measure used in modeling before FRAM (1.2317 = 1.333 (convert OA3 to DA2) * 0.924 (convert DA2 to JA3)). Any non-landed fishery related mortality that occurs is ignored in this ocean abundance-to-total abundance FRAM conversion step.
For Puget Sound and the Washington coast, data and methods used to generate fishery inputs for FRAM and TAMM are shared, discussed, and agreed-to by salmon co-manager technical staff in the NOF annual process. Fishery inputs are compiled by the FRAM modelers using submissions from the respective tribes (ceremonial & subsistence, commercial and test fisheries) and WDFW (recreational, commercial and test fisheries). In general, fishery inputs can be either provided as scalers, quotas, or TAMM harvest rates (mark-selective or non-selective). For the calculation of recreational Puget Sound marine fishery inputs, an external file is used. Based on historical catch record card and creel survey-based estimates of mortalities, recreational fishery inputs are calculated depending on the fishery type (mark-selective, non-selective, catch and release and non-retention), months or days open for the respective fishery, and any regional or fishery specific adjustments (e.g. two pole adjustment). Ocean fishery inputs for PFMC-managed fisheries are provided to the FRAM modelers by staff from Washington coastal Indian tribes, CRITFC, WDFW, ODFW and CDFG.
PFMC-managed Coho retention fisheries are modeled as landed fish quotas or mark-selective fishery quotas (i.e., retained marked fish). PFMC-managed Coho non-retention fisheries are modeled using external estimates of mortalities generated from historical Coho to Chinook ratios of landings when retention of both species was allowed. In some fisheries, like the troll fisheries South of Cape Falcon, these external mortality estimates are adjusted downward to account for shifts in effort away from the species that cannot be retained.
Fisheries outside of PFMC jurisdiction are modeled using quotas as a landed catch expectation, as catch (or occasionally effort) scalers, or as terminal area harvest rates used during TAMM processing. Canadian fisheries are modeled using fishery scalers generated from post-season PST Coho Technical Committee (CoTC) Coho FRAM modeling exercises, which are intended to represent similar fishery expectations or regulations in the upcoming fishing year. Alaskan fisheries are modeled pre-season using static scalers to represent proxy stock abundances relative to Coho FRAM base period fishing.
Although Coho FRAM is primarily used for pre-season fishery impact assessment, it can also be used in a post-season context. These post-season model runs are primarily conducted for two purposes: as a tool to evaluate the model’s performance by comparing model estimates to independently derived estimates, and to evaluate the performance of the fishery management system towards meeting conservation objectives for key stocks. Post-season model runs are occasionally used as a “one-stop-shop” to get a historic time series of Coho escapements or pre-fishing abundances of hatchery and natural stock units for the purpose of conducting population studies or evaluating forecasts.
This context uses the FRAM model to estimate what occurred in a previous year based on observed input values and the associated estimated abundances. Post-season FRAM model runs use the “backwards” BkFRAM utility that finds FRAM starting abundances using inputs of observed estimates of escapements and fishery catches (landed and non-retention). Post-season model runs are conducted by the Pacific Salmon Commission (PSC) CoTC every year during the PSC annual meeting (February) and used to evaluate management obligations under the PST. Post-season model runs are often lagged by several years due to data availability issues (e.g., in 2021 for fishery year 2019).
To cite this page:
Salmon modeling and analysis
workgroup. 2023. Coho FRAM Applications in FRAM Documentation.
https://framverse.github.io/fram_doc/ built September
21, 2023.