| Abstract/Description or Keywords |
Canada’s Wild Salmon Policy (WSP2) calls for the monitoring and assessment of all geographically, ecologically and genetically distinct populations of wild Pacific salmon, known as Conservation Units (CUs). The Wild Salmon Policy states that CUs will be assessed against specific reference points, or benchmarks, for indicators such as spawning abundance or fishing harvest rate. For each CU, a higher and a lower benchmark are to be defined so as to delimit ‘green’, ‘amber’, and ‘red’ status zones. As numbers of spawning salmon decrease, a CU moves towards the lower status zones and the extent of management actions directed at conservation should increase. The status of a CU does not dictate that any specific action must be taken, but instead serves to guide management decisions in conjunction with other information on habitat, ecology and socioeconomic factors. Over the past few years, significant headway has been made towards defining benchmarks for CUs in the Fraser River watersheds, however, there has been much less progress towards setting benchmarks for other areas of coastal British Columbia and the Yukon. It is in this context that the Pacific Salmon Foundation (PSF) commissioned the independent analysis presented here. This analysis provides a foundation towards developing abundance-based status benchmarks for CUs that spawn in the Skeena River basin, the second-largest watershed in BC and home to approximately 50 CUs of sockeye, Chinook, chum, coho, and pink salmon. The main objectives of this analysis are: derive and explore some possible options for status benchmarks, conduct a preliminary independent assessment of the status of Skeena CUs, and make available a foundation of computer code for performing stock-recruitment analyses that can be expanded and updated in the future. While this analysis presents benchmark options, it does not determine which benchmarks should be used for assessing Skeena CUs as that is the responsibility of DFO in consultation with First Nations and other affected parties. DFO has recommended that Wild Salmon Policy status assessments adopt an approach that combines several different classes of status indicators, including current abundance of spawners, time trends in abundance, geographic distribution of spawners, and fishery harvest levels relative to a CU’s productivity. The scope of the analysis presented here is limited to examining benchmark options related to current abundance and harvest levels. The approach used is based on CU-specific ‘stock-recruitment models’, which describe the relationship between the number of spawning salmon in a parental generation (the ‘stock’ or ‘escapement’) and the expected numbers of offspring produced and available for harvest or spawning (the ‘recruitment’). By fitting a stock-recruit model to actual data from a CU, one can estimate the average recruitment that is expected from a given number of spawners. The model can be used to estimate a variety of values that are commonly used as status benchmarks, such as the escapement and harvest rates that are expected to maximize long term fishing yield (‘Smsy’ and ‘Uopt’, respectively), and the escapement level that will allow a CU to return to Smsy within one generation in the absence of fishing (‘Sgen1’). As a lower benchmark option, this analysis determined Sgen1 for each CU, following recommendations used for Fraser sockeye. A second lower benchmark option, 10% of the virgin stock size (‘0.1*So’), is presented as a more intuitive and computationally simpler alternative. Setting reference points as a fraction of the virgin stock size is a common practice in many fisheries throughout the world, and 0.1*So corresponds to the level at which fisheries managed by the US Pacific Fisheries Management Council are closed. For the upper benchmark, Smsy was determined for each CU. Since escapements beyond Smsy may produce additional ecosystem benefits, ‘Smax’ (the escapement that produces the maximum recruitment) is also presented. As a benchmark for harvest level, Uopt is presented for each CU, as well as ‘Umax’ which is the harvest rate that exceeds the productivity of the stock and would eventually lead to extirpation. The current abundance status of each CU was determined by comparing its average escapement for 2006-2010 (the last five years of available estimates) to the Smsy and Sgen1 benchmarks. The current harvest rate status was computed by comparing fishing levels over this period to the Uopt benchmark. There are four major sources of uncertainty and potential bias that affect the evaluation of benchmarks and status. The relative size of these biases will vary between species and CUs depending in part on the amount and quality of data available for the stock-recruit analysis. However, the actual extent of bias is unknown. Each of the four biases is discussed in more detail below. It is important to be aware of uncertainty and possible bias when considering the results of this analysis, as they can have important consequences for management. This includes the possibility of overestimating productivity and Uopt, setting lower abundance benchmarks too low, and evaluating harvest rate and abundance status as being better than they actually are. If the benchmarks presented here were to be applied without considering this uncertainty, then future policies and management strategies could result in reduced conservation performance and loss of long term fishery yields. The numbers of recruits used for this analysis were determined using annual estimates of the number of salmon caught from each CU. For sockeye, these harvests are estimated by a model that uses information about when the salmon from each CU are thought to be migrating through areas where fisheries are occurring (i.e., a ‘run-reconstruction model’). Relatively minor changes to run-timing for sockeye can lead to significant differences in the estimated harvest and, ultimately, the estimated number of recruits. If harvest from a CU has been overestimated, then this would in turn lead to inflated estimates of recruitment and productivity. This type of a bias could lead to the incorrect conclusion that a CU is not overharvested when it really is. All of the available information on sockeye run-timing by CU, harvest timing in all marine and freshwater fisheries has been taken into account so the harvest rates estimated for each Skeena sockeye CU are as accurate as possible. For Chinook and coho, harvest rates in fisheries are estimated directly based on coded-wire tag data for a few tagged stocks (‘indicator stocks’ assumed to represent other CUs of the same species). The assumption that one tagged stock can be representative of several others is an additional source of uncertainty and possible bias. For most salmon species, fish vary in age-at-maturity, the age at which they mature and return to their natal streams. Salmon returning within a particular year will be from different years of spawning (i.e., brood years), with the exception of pink salmon that have a fixed age-at-maturity of two-years. In order to perform a stock-recruit analysis, the recruits returning in a given calendar year must be apportioned by age so that the correct number of progeny is attributed to the number of parental spawners in a brood year. This is necessary to arrive at the total number of recruits produced by each parental cohort, information which is the foundation of a stockrecruit analysis. However, the proportion of ages among recruits can vary substantially from year-to-year, and it is rare that this information is available for every year. Therefore, stockrecruit analyses frequently uses an average age composition calculated from available data for a CU, and apply this average to all years in order to apportion annual recruitment among the various parental cohorts. However, the practise of using a single average age composition produces biases in the recruitment reconstruction. Specifically, these biases are expected to lead to productivity estimates that are inflated, Uopt harvest rate is then too large, and Sgen1 would be set too low. These biases would lead to management advice that would lead to overharvest and reduced future production. In the Skeena, this bias potentially affects all CUs except for pink salmon (which all return to spawn at the same age), the Babine system sockeye CUs (for which annual age data are available for every year), and possibly the Kalum-late Chinook CU (for which annual age data is available for returns from 1988 onwards). Age composition-related bias is a potential issue for all other CUs examined in this report. Unfortunately, wide variation in the extent of age-related bias among populations (e.g., Babine vs. Nass, six Columbia Chinook stocks vs. Babine) does not allow one to compute a reliable correction factor. Measurement error bias leads to overestimation of both stock productivity and the magnitude of density dependence. Like age composition effects, this bias will result in overestimation of the harvest rate benchmark and underestimation of the lower abundance benchmark. The size of the bias will increase with the extent of error in escapement measurements, when the time series is short, and when there is limited contrast in the range of escapements over time. Time series bias will likely result in a slight overestimation of the harvest rate benchmark and potentially larger underestimation of the lower abundance benchmark and other abundance-based benchmarks. The magnitude of measurement error and time series biases in the Skeena-based stockrecruitment estimates is uncertain, however, there are many characteristics of the data which suggest these biases could be quite large. Use of the benchmarks developed here for future management assumes that the historical data used to estimate the benchmarks is representative of future conditions. This would not be true if the productivity of a CU is changing over time. This analysis found statistical evidence of declining productivity in the last decade for a few CUs, including important ones such as the Babine sockeye wild runs. The fundamental question is whether such conditions are permanent or temporary. While it is possible that CUs which show recent declines in productivity will continue to show this pattern over the next few years, a longer view of the same data may reveal multiple cycles of decline and recovery. Several possible abundance and harvest rate benchmark options have been calculated for 34 of the 53 salmon CUs identified for the Skeena watershed. These benchmark values can be found within the main report and accompanying tables. However, for 19 CUs (mainly sockeye), there was insufficient data available to determine benchmarks. In addition, the analysis presents the results of a preliminary status assessment that compares current (2006-2010) escapement and harvest rates to a subset of these benchmarks (Sgen1, Smsy, and Uopt). For many of the 34 CUs where benchmarks have been calculated, the stock-recruit data is poor, and the benchmarks and status assessments presented here are uncertain and potentially biased. The results described here should, therefore, be interpreted with caution until additional studies can be undertaken to verify these results. Sockeye: Five CUs were assessed as being most likely in the red abundance status zone (i.e., below the Sgen1 benchmark, meaning that it would take more than a generation without fishing for these CUs to return to Smsy levels: Babine late-timing wild (Ň Nilkitkwa CU), Bear CU, Gitanyow (=Kitwancool) CU, Morice CU, and Swan CU. Four CUs are most likely in the amber status zone: Babine early- and mid-timing wild (Ň Babine CU and Ň Tahlo/Morrison CU, respectively), Lakelse CU, and Motase CU. Seven CUs were most likely in the green status zone (meaning that these CUs are likely at or above levels that would be expected to produce the maximum long term yield): Alastair CU, Azuklotz CU, Damshilgwit CU, Kitsumkalum CU, McDonell CU, Stephens CU, and the Babine enhanced stock. The remaining 12 lake-type and two river-type sockeye CUs could not be assessed due to insufficient data. Based on recent (2006-2010) harvest rates, the Damshilgwit and perhaps the Babine mid- and late-timed wild CUs are likely being harvested at levels above Uopt. Prior to 1997, three CUs (Bear, Morice, and Babine late-timing wild) appear to have been harvested at rates above Uopt. The average 2006-2010 harvest rate across assessed CUs was 28%, considerably less than the average Uopt (50%). There is very wide variation in productivity among CUs, indicating wide variation in Uopt. The estimated Uopt for the enhanced Babine stock was 64% compared to 45-47% for the three wild Babine system CUs. Bias for these estimates should be relatively low owing to the long time series, accurate escapement estimates, and use of year-specific age composition data for the recruitment reconstruction. Historical trends (1960-2010) in harvest rates indicate that wild CUs have been overharvested as the Babine aggregate was fished at rates close to Uopt for the enhanced component. Recent (2006-2010) harvest rates are lower than the long-term average (1960-2010) and closer to rates that are appropriate for the wild Babine CUs. However, the three wild CUs show signs of reduced productivity in recent years, which would imply that the Uopt benchmark calculated in this report may be too high. Chinook: Seven CUs were assessed as most likely being in the green abundance status zone: Ecstall, Kalum-early, Kalum-late, Lower Skeena, Middle Skeena Large Lakes, Middle Skeena Mainstem Tributaries, and Upper Bulkley. An additional four CUs could not be assessed due to insufficient data. In this analysis there was no evidence that any of the seven assessed CUs have been harvested above Uopt. Less productive CUs (Upper Bulkley and Kalum-early) are currently harvested at very low rates (4%). More productive stocks have Uopt ranging from ~60-75% and in recent years have been harvested at approximately 40%. Coho: The Lower Skeena CU was assessed as most likely being in the amber abundance status zone, while the Middle Skeena and Upper Skeena CUs are most likely in the green zone. The Skeena Estuary CU could not be assessed due to a lack of data. Up to the mid-1990’s, all CUs were likely harvested at or above Uopt (53-72%), but recent rates are lower (34% for 2006-2010). Middle Skeena and Upper Skeena CUs have shown positive trends in escapement since harvest rates have been reduced, but the Lower Skeena CU has not recovered, possibly due to declining productivity. Chum: The Skeena Estuary and Lower Skeena CUs were assessed as most likely being in the red abundance status zone, while the Middle Skeena CU is most likely in the amber zone. Prior to the mid-1990’s all CUs were likely harvested at or slightly above Uopt (30-44%), but recent rates have averaged 14%, near or below Uopt. Productivity for the Skeena Estuary and Lower Skeena CUs has apparently declined over time, which may be the reason that these CUs have not recovered despite reductions in harvest. It should be noted that escapement estimates for chum are of low reliability given the poor survey coverage in recent years. Pink: The Middle-Upper Skeena Even-Years CU was assessed as most likely being in the red abundance status zone. The Nass-Skeena Estuary Even-Years CU is most likely in the amber zone. The three odd-years CUs are likely in the green zone (Nass-Skeena Estuary, Lower Skeena, and Middle-Upper Skeena). Historically, pink salmon CUs were harvested at or slightly above Uopt (42-57%), but recent rates are almost half of this value. This analysis found no evidence of reduced productivity over the last decade. Conclusion and Future Work: Additional exploration of sources of uncertainty and bias in the benchmarks and status assessments is recommended. As discussed above, biases common to stock-recruit models can potentially lead to overestimating productivity and appropriate harvest rates, and setting the lower abundance benchmark too low. This can result in overharvest, reduced conservation performance, and reduced long term harvest yields. Several of these biases are expected to be larger when data quality and quantity is poor, which is the case for many of the CUs in the Skeena Watershed. Results from management strategy evaluations can provide considerable guidance on the utility of the various benchmarks computed from this analysis. Management strategy evaluations use simulations to test the expected effectiveness of alternative management approaches for meeting management objectives (such as protecting weak stocks and providing fishing opportunities) under a range of uncertainties. A logical next step in implementing the Wild Salmon Policy in the Skeena is to use the estimated stock-recruitment parameters and benchmarks developed here in a management strategy evaluation to provide an explicit and rational way of setting harvest rates which address a set of agreed management objectives. However, these evaluations are also dependent on the accuracy of the parameters estimated, which further argues for a more quantitative assessment program to validate parameter values estimated for key CUs. |
