Exploration approaches
Bruno S. Carturan, Stephanie Peacock
2025-04-28
1 Chinook
1.1 Comparison of the benchmarks
The goal is to compare the DFO’s Sgen, Smsy, 20% and 40% Smsr with Sgen and Smsy benchmarks obtained fitting a simple HBSR Ricker model to the same data. DFO’s benchmarks come from the benchmarks_posteriors.rds in github.com/Pacific-salmon-assess/yukon-CK-ResDoc
Import benchmark_posteriors.rds:
benchmarks_posteriors <- readRDS(file = paste0(wd,"data/benchmark_posteriors.rds"))## CU Sgen Smsy Umsy Seq Smsr S.recent
## 1 Big.Salmon 233.0164 237.0502 0.01673658 476.1064 14163.596 2122.864
## 2 Big.Salmon 1132.8005 1568.0577 0.22996020 3349.9051 6818.822 2087.906
## 3 Big.Salmon 1054.2850 1777.5635 0.32346183 3924.9950 5495.435 2189.614
## 4 Big.Salmon 2311.0797 5379.0707 0.44592883 12501.5920 12062.621 2123.634
## 5 Big.Salmon 870.4047 1216.3595 0.23496336 2602.8712 5176.805 2116.316
## 6 Big.Salmon 1147.4328 1286.0920 0.09830776 2639.8751 13082.303 2153.766
## Smsr.20 Smsr.40
## 1 2832.719 5665.438
## 2 1363.764 2727.529
## 3 1099.087 2198.174
## 4 2412.524 4825.049
## 5 1035.361 2070.722
## 6 2616.461 5232.921Some edits on the name of certain CUs:
## name_old name_new cuid
## 1 Big.Salmon Big Salmon 1201
## 2 YukonR.Teslinheadwaters Teslin 1212
## 3 UpperYukonR. Upper Yukon 1211
## 4 NorthernYukonR.andtribs. Northern Yukon 1207
## 5 Whiteandtribs. White 1206
## 6 MiddleYukonR.andtribs. Middle Yukon 1204Import the reconstructed recruits and spawners data in brood_table_long.csv (emailed by Hannah Hunter in Wednesday 23 April 2025):
## CU_f BroodYear S_lwr S_med S_upr R_lwr R_med
## 1 YukonR.Teslinheadwaters 2019 5928.75 6714.07 7622.48 3394.98 5015.65
## 2 YukonR.Teslinheadwaters 2020 2651.68 3031.48 3460.40 2609.80 4991.72
## 3 YukonR.Teslinheadwaters 2021 1703.36 2040.96 2424.85 NA NA
## 4 YukonR.Teslinheadwaters 2022 1467.77 1909.85 2471.85 NA NA
## 5 YukonR.Teslinheadwaters 2023 2631.17 3154.81 3788.59 NA NA
## 6 YukonR.Teslinheadwaters 2024 3232.22 4057.37 5072.80 NA NA
## R_upr CU
## 1 7806.28 Teslin
## 2 10165.71 Teslin
## 3 NA Teslin
## 4 NA Teslin
## 5 NA Teslin
## 6 NA TeslinPlot spawner abundance, smoothed generational avergave and current spawner abundance:
The figure above also shows the current spawner abundance (csa) in 2024
for (1) the smoothed generational average vs. (2) the value in the rapid
status assessment (RSA). The values differ, especially for White.
Question: How come the current spawner abundance is different? Are we not using the same dataset? Answer: no they were not but now they should.
Using the same SR dataset above, we fit the HBSR Ricker model like we do for the PSE.
IMPORTANTLY, recruits estimates are incomplete after 2017 so we remove observations more recent than 2017 to calculate the benchmarks.
Question: Did you also not consider data after 2017 to generate the benchmark density distributions? Answer: Yes and no, it’s a more complicated modelling approach that takes into account those uncertainties.
## region species_name cuid cu_name_pse curr_spw sr_lower sr_lower_025
## 1 Yukon Chinook 1201 Big Salmon 2130.983 671.5070 283.68031
## 2 Yukon Chinook 1204 Middle Yukon 5564.849 1440.5314 666.22865
## 3 Yukon Chinook 1202 Nordenskiold 1414.348 143.4859 73.38141
## 4 Yukon Chinook 1207 Northern Yukon 2773.023 1150.9164 405.80889
## 5 Yukon Chinook 1203 Pelly 2716.552 1451.9279 589.53019
## 6 Yukon Chinook 1205 Stewart 1322.545 500.5560 253.42296
## 7 Yukon Chinook 1211 Upper Yukon 1801.376 490.6149 262.10389
## 8 Yukon Chinook 1206 White 1545.884 665.1495 280.82129
## 9 Yukon Chinook 1212 Teslin 3170.409 662.2060 344.92775
## sr_lower_975 sr_upper sr_upper_025 sr_upper_975
## 1 1900.5962 3176.248 2443.0379 4909.230
## 2 3050.0700 6662.050 5425.1511 9403.231
## 3 288.5519 1128.838 955.1067 1440.231
## 4 3646.7495 4528.539 3269.3925 8557.082
## 5 3369.1850 5498.808 4319.6108 8192.238
## 6 1027.4630 2200.123 1776.2843 3041.087
## 7 925.9375 2005.287 1641.6218 2741.998
## 8 1913.0461 2562.412 1825.8377 4445.311
## 9 1304.2388 3468.408 2851.8780 4558.508Calculate the HDP and 95%CI from the posterior density distribution for the DFO’s benchmarks and corresponding biological status probability:
## cu_name_pse status_Smsy_red status_Smsy_amber status_Smsy_green
## 1 Big Salmon 9.3625 71.0875 19.5500
## 2 Middle Yukon 5.3125 72.0625 22.6250
## 3 Nordenskiold 0.0000 33.3875 66.6125
## 4 Northern Yukon 40.8750 43.1875 15.9375
## 5 Pelly 13.9750 64.0500 21.9750
## 6 Stewart 15.5500 76.2875 8.1625
## 7 Upper Yukon 0.0125 71.5125 28.4750
## 8 White 3.2875 78.3375 18.3750
## 9 Teslin 0.1000 75.7000 24.2000
## status_Smsr_red status_Smsr_amber status_Smsr_green
## 1 17.0750 70.1500 12.7750
## 2 10.0000 73.4125 16.5875
## 3 0.0000 0.9000 99.1000
## 4 58.8375 40.1375 1.0250
## 5 30.6625 69.2875 0.0500
## 6 27.8250 71.1875 0.9875
## 7 0.0125 3.7375 96.2500
## 8 7.2000 78.4125 14.3875
## 9 0.2250 20.3750 79.4000Figure to compare the approaches:
In the figure above, the horizontal black line is the current spawner abundance (geometric average of spawner abundance over the last generation since 2024); the green and red horizontal dashed lines are the absolute 10,000 and 1,500 fish benchmarks used in the rapid status assessment (RSA); the coloured boxes show the biological status for each method, and the % are the corresponding probabilities.
Summary:
same status for all nine CUs for the Sgen & Smsy between DFO and PSF;
same status for five out of the nine CUs for all three benchmarks;
Sgen for PSF tend to be smaller than DFO’s Sgen and 20% Smsr
the Smsr benchmarks are less conservative than the Sgen & Smsy benchmarks for Upper Yukon and Teslin, and more conservative for Northern Yukon.
Important note: the biostatus for Nordenskiold and Stewart is actually RED because current spawner abundance < 1500 (the lower absolute benchmark which is part of the new rule).
Question: is there a justification/reference for using 20% and 40% of Smsr (vs. 50% for instance)? Answer: there is a bit of a history with Carrie Holt.
1.2 Compare results for Big Salmon & Middle Yukon for year > 2004
Big Salmon and Middle Yukon cannot be dissociated genetically from one another. The sonar at Big Salmon allows to quantify Big Salmon CU abundance and one can then deduce Middle Yukon CU abundance, but only since 2005. Before that, a constant relative proportion was used:
Source: run-reconstruction-CU-spawners.25Apr2025.csv in Pacific-salmon-assess/yukon-CK-ResDoc/tree/main/analysis/data/raw.
The goal is to refit the HBSR Ricker model to the recruits per spawner data but discarding data prior 2005 for both Big Salmon & Middle Yukon and compare the results with the ones above:
!!! WORK NOT SHOWN BECAUSE NOT RELEVANT ANYMORE !!!
Conclusion: removing data prior 2005 leads to less conservative benchmarks, and even changes the status for Middle Yukon.
Question: Sensitivity analysis worth it?
Answer: No need to do anything because these assumption prior 2005 have been considered in the modelling.
1.3 Teslin: other source of data
Run reconstruction abundance is likely biased high for the Teslin CU (because the CU is far from the Eagle sonar and many fish certainly die on the way; local reports attest of much lower abundances). So we use instead the spawner survey data from Nisutlin and Wolf rivers to obtained a index of spawner abundance for Teslin. But this index is a partial assessment of the total number of spawners because there are a few other streams where individuals of the CU go. The current spawner abundance can consequently not be compared to absolute benchmarks. For more details, see the csas-teslin-reconstruction; the Teslin_CU_expansion_2025-05-15.csv datasets can be downloaded from the previous file:
## yrs_teslin nisutlin wolf sum expanded_index generational_mean
## 1 1969 105 NA 105 159 NA
## 2 1970 615 NA 615 929 NA
## 3 1971 650 750 1400 1400 NA
## 4 1972 237 13 250 250 NA
## 5 1973 NA NA NA NA NA
## 6 1974 NA NA NA NA 477
We use the expanded_index to determine Teslin CU’s
benchmarks and biological status using the percentile method (with the
25th and 50th percentiles of the lower and upper benchmarks,
respectively):
Conclusion: PSF to use expanded index and percentile benchmarks, leading to red/poor status outcome.
Question: What percentiles best correspond to the Smsr benchmarks used for other Yukon Chinook CUs? The 25th and 75th percentiles were chosen to represent Sgen and Smsy.
1.4 Upper Yukon: other source of data
Similarly, run reconstruction abundance is likely biased high for the Upper Yukon CU. It has been estimated that the counts at the Whitehorse Fish ladder and the Takhini sonar together account for nearly all of the spawners in the Upper Yukon CU. We use a spawner index representing the sum of the two populations. The time series has only height data points because there are only six sonar counts at Takhini between 2017 and 2024 (for more details, see the csas-UpperYukon-reconstruction; the UpperYukon_CU_expansion_2025-05-15.csv datasets can be downloaded from the previous file):
## yrs whitehorse whitehorse_wild takhini sum expanded_index generational_mean
## 1 2017 1226 748 1872 3098 3098 NA
## 2 2018 691 435 1554 2245 2245 NA
## 3 2019 282 245 NA 282 1227 NA
## 4 2020 216 164 NA 216 940 NA
## 5 2021 274 175 247 521 521 NA
## 6 2022 165 99 476 641 641 1178
## 7 2023 154 80 355 509 509 872
## 8 2024 479 263 1049 1528 1528 818There is not enough data points to calculate biological status with the percentile benchmarks, but we can compare current spawner abundance to the absolute lower benchmarks, which yield a red status (vs. amber with the RR data):
Conclusion: PSF to use expanded index from Takhini + Whitehorse for 2017-2024, with only wild portion of Whitehorse counts. Current abundance compared to absolute benchmark of 1,500 yields poor/red status outcome.
Question: Is there general consensus that these two surveys represent “nearly all” spawners for the CU? Answer How much spawning occurs on the mainstem below the dam or in tribs of Lake Labarge? Initial discussion indicates <10%, but worth confirming.
1.5 Porcupine
We use the escapement count from the JTC report 2025 (Appendix B12)
(i.e. the estimate column in trib-spwn.csv):
## CU system year estimate
## 1 Porcupine porcupine 2014 3066
## 2 Porcupine porcupine 2015 4851
## 3 Porcupine porcupine 2016 6665
## 4 Porcupine porcupine 2017 1191
## 5 Porcupine porcupine 2018 3414
## 6 Porcupine porcupine 2019 4740
## 7 Porcupine porcupine 2020 NA
## 8 Porcupine porcupine 2021 409
## 9 Porcupine porcupine 2022 349
## 10 Porcupine porcupine 2023 512
## 11 Porcupine porcupine 2024 468There is not enough data points to calculate biological status with the percentile benchmarks, but we can compare current spawner abundance to the absolute lower benchmarks, which yield a red status:
Conclusion: PSF to use the escapement count and compare it to absolute benchmark of 1,500 yields poor/red status outcome.
2 Chum
2.1 Estimated spawner abundance from the JTC report
Import the Appendix B25 (proportion of CUs) table:
## Year Sample.Size Mainstem Yukon.Early White Teslin
## 1 2005 491 67.7 2.1 29.8 0.4
## 2 2006 728 41.0 1.0 54.9 3.1
## 3 2007 735 46.9 0.5 52.1 0.5
## 4 2008 745 48.0 0.1 49.9 2.1
## 5 2009 366 68.3 0.1 30.6 1.0
## 6 2010 337 52.8 0.7 46.3 0.2Import the Appendix B16 (total escapement estimate) table:
## Year Spawning.escapement.estimate
## 1 1980 22912
## 2 1981 47066
## 3 1982 31958
## 4 1983 90875
## 5 1984 56633
## 6 1985 62010Some CU name corrections:
## name_old name_new
## 1 Mainstem Middle Yukon
## 2 Yukon.Early North YukonDetermine the spawner abundance for each CU using both tables:
## Year Middle Yukon North Yukon White Teslin
## 1 2005 296186.15 9187.4580 130374.404 1749.9920
## 2 2006 90568.18 2208.9800 121273.002 6847.8380
## 3 2007 111146.90 1184.9350 123470.227 1184.9350
## 4 2008 80591.04 167.8980 83781.102 3525.8580
## 5 2009 63946.56 93.6260 28649.556 936.2600
## 6 2010 62192.59 824.5230 54536.307 235.5780
## 7 2011 105249.79 205.5660 98671.680 1438.9620
## 8 2012 65114.13 137.6620 72410.212 137.6620
## 9 2013 101132.31 400.5240 97928.118 801.0480
## 10 2014 77614.02 313.5920 78554.796 0.0000
## 11 2015 67259.30 869.2640 40420.776 108.6580
## 12 2016 101686.90 0.0000 42563.231 871.6020
## 13 2017 191957.63 401.5850 208824.200 401.5850
## 14 2018 59615.94 3159.5830 86896.239 4469.6540
## 15 2019 41384.47 2366.8242 51490.910 4623.7958
## 16 2020 12997.43 145.7744 9832.718 536.0736
## 17 2021 15060.50 0.0000 7924.140 162.1900
## 18 2022 15494.31 0.0000 5620.873 918.8178
## 19 2023 15423.24 207.6460 6004.062 455.0540
## 20 2024 12206.52 168.2096 3079.530 719.7430Plot spawner abundance:
Determine the benchmarks and associated biological status using the percentile method (25% and 50% percentiles):
Conclusion: biostatus is red either because current spawner abundance is below the lower absolute or relative benchmarks in all four CUs.