Science Response 2018/034

Pre-season run size forecasts for Fraser River Sockeye (Oncorhynchus nerka) salmon in 2018


Fraser Sockeye survival has been low in the last three years (2015-2017). Most of these salmon would have been deposited in the gravel as eggs from 2011 to 2013, migrated to the Northeast (NE) Pacific Ocean as smolts from 2013 to 2015, and returned to spawn as adults from 2015 to 2017. During this period, unusually warm land and ocean temperatures have been observed. In the NE Pacific, a ‘warm blob’ developed late 2013 and persisted through to 2016, characterized by anomalously warm waters 3-4°C above seasonal averages and extending down to 100 m depths in some regions. A strong El Niño also occurred from 2015 to 2016, which also contributed to warmer land-ocean temperatures in local Pacific and British Columbia waters. Fraser River temperatures were warmer than average in spring and summer months during this period, and spring freshets were particularly early in 2015 and 2016. Although exact temperatures experienced by salmon throughout their life will vary, along with their specific responses, warmer conditions are generally linked to poorer salmon survival. In the ocean, for example, warm coastal ocean temperatures during and following salmon ocean entry are associated with reduced survival of salmon stocks in BC and Washington (Mueter et al. 2005), and warm conditions as early as one year prior to outmigration may influence juvenile growth (Beamish and Mahnken 2001). Fraser Sockeye that will return in 2018 will have experienced many of these same freshwater and marine warm conditions as recent returns (2015-2017).

Fraser Sockeye forecasts are presented as standardized cumulative probabilities (10%, 25%, 50%, 75%, and 90%) to capture inter-annual random (stochastic) uncertainty in returns, largely attributed to variations in stock survival. The probability distribution for each stock generally spans the range of survival it has exhibited historically for a given brood year escapement, with forecast values at the lower probability levels representing lower observed stock survivals, and values at the higher probability levels representing higher observed survivals. Total Fraser Sockeye returns have either been near or below the 50% probability level forecast for the past 13 years, excluding 2010, indicating average to below average survival for this aggregate during this period. In the last three return years (2015-2017), total Fraser Sockeye returns have fallen at the lowest (10%) probability level forecast, consistent with very poor survival

The combined Fraser Sockeye forecast for 2018 has a median estimate of 14 million, which is similar to the long-term average return for this cycle line (13.1 million), and above the all-year average (7.4 million). At the lower end of the forecast distribution, the forecast for 2018 is 5.3 million at the 10% probability level and 8.4 million at the 25% probability level.  Warm temperatures in BC freshwater and Pacific marine ecosystems have occurred in recent years, coinciding with poor total Fraser Sockeye survival. If these recent trends in Fraser Sockeye survival persist, then 2018 total returns are likely to be at the lower (10%-25%) probability level forecast. At the individual stock level, however, responses are likely to vary. Pre-season fisheries’ planning considers a range of possible return outcomes based on the forecast distribution, with some emphasis on lower p-levels (10%-25%) in recent years. In-season management decisions are based on in-season data, compared with forecast parameters.

The effects of extremely warm water temperatures on survival have been incorporated quantitatively into the forecasts for eight stocks where temperature covariate models historically perform well (Early Stuart, Bowron, Chilko, Quesnel, Raft, Cultus, Weaver, Birkenhead). Together, these eight stocks account for 24% to 32% of the total 2018 forecast, depending on the probability level. Incorporating temperature covariates into these forecasts reduced the total forecast by only 17%, since most of the total forecast was not informed by these warmer temperature data. Additionally, since not all stocks will exhibit similar survival, the forecast distribution for total Fraser River Sockeye salmon will likely over-estimate total returns, particularly at the high probability levels, and it is more appropriate to reference individual stock forecast distributions, versus the total summed forecast.

The Late Run Shuswap stock dominates the 2018 forecast (50% at the median probability level). Unlike the 2016 and 2017 forecasts, the Larkin model was not specifically chosen to generate the Late Shuswap forecast, as the extremely large 2010 brood year escapement is no longer expected to impact survival of this stock. Instead, the Ricker-cyc model (see Table 4) was used based on the model selection criteria. The Larkin model forecast is slightly lower than the selected Ricker-cyc model, though it is similar to the Ricker model forecasts, which means that this difference is not due to delayed-density dependence.

The Summer Run Chilko (16%) and Quesnel (8%) stocks are the next largest contributors to the 2018 forecast. Usually a smolt-power model is used to generate Chilko forecasts. However, given smolt data were only available for the 2014 brood year, and not the 2013 brood year, a blended smolt-sibling model was used. Specifically, four year olds were forecast with a smolt-power model, and five year olds were forecast using a four-to-five year old sibling model (using preliminary four year old returns in 2017 to predict five year old returns in 2018). The five year old forecast for Chilko is very small, due to the low return of four year old fish in 2017. Quesnel was forecast using the same model as 2017: a Ricker model using sea-surface temperature measured at the Entrance Island lighthouse station. This addition of the Entrance Island covariate in this model has a strong effect on the forecast for Quesnel, reducing it by more than 50%. However, preliminary return estimates for 2017 indicate that this model performed very well under similar environmental conditions as those used to generate the 2018 forecast.

The remaining stocks are expected to contribute a total of 26% of the forecast, dominated by Stellako (4%), Seymour (4%), and Scotch (2%).

This Science Response Report results from the Science Response Process of December 15, 2017 on Pre-season abundance forecast for Fraser River Sockeye returns in 2018. The 2018 forecast relies on methods of past CSAS processes and publications (Cass et al. 2006; DFO 2006, 2008, 2009, 2011, 2012, 2013, 2014a, 2014b, 2015a, 2015b, 2016, 2017; Grant et al. 2010; Grant and MacDonald 2012, 2013; MacDonald and Grant 2012).

Accessibility Notice:

This document is available in PDF format. If the document is not accessible to you, please contact the Secretariat to obtain another appropriate format, such as regular print, large print, Braille or audio version.

Date modified: