2024 Outside �dogfish assessment

Technical Working Group Meeting #2

​

Quang Huynh, Blue Matter Science

April 4, 2024

1

Summary

• Dogfish are long-lived, late-maturing, low fecundity species
• Fished hard during the 1940s-1950s, apparent recovery by the 1970s
• Notable recent declines in indices despite much lower retention/discards today

​

• Fit a 2-sex, age-structured model (Stock Synthesis 3) to catch, survey indices, and length composition for the outside stock in BC
• Today’s agenda includes overview of data and biological inputs and update of modeling from February meeting

​

• Powerpoint and references available at: https://pbs-assess.github.io/dogfish-assess/

​

​

2

Data�Catch

3

• Commercial catch (1-5) organized by gear/retention due to differences in size composition
• Assuming 100% discard mortality due to handling

​

• Add survey catch
• Recreational (iRec) and salmon troll catch, although incomplete series

​

​

High trawl catch (20 kt landings) pre-1950 have never been repeated

​

Current TAC = 10k tonnes outside, 5k inside

Catch units:

• Thousands of pieces for IPHC, HBLL, iRec, Salmon
• Tonnes for all others

Data - Fishery Lengths

• Commercial size composition
• Retained dogfish are mostly large females
• Discards are smaller with balanced sex ratio
• Assume recreational catch has selectivity of trawl discards

​

4

Data�Indices of abundance

• Currently, Dogfish are caught in all three major surveys (HBLL/PFMA, IPHC, and Synoptic trawl)
• Rapid declines in the HBLL and SYN indices since 2010

​

• Surveys are very recent compared to dogfish longevity

​

• Added bycatch fishery CPUE (since 1996) and Hecate Strait Multispecies Assemblage (HS MSA) (1980-2002)

​

5

Data�Survey Lengths

SYN trawl

• Annual size composition is the nominal sum over sampled areas
• Explored standardization model to impute size composition in missing survey areas. Model was insensitive to either composition series

​

6

Data�Survey Lengths

• IPHC survey
• Lengths are not measured in the HBLL survey in outside waters.
• Selectivity is mirrored to the IPHC survey, based on comparison of HBLL and IPHC sets in 4B

7

Growth

• Length-age samples throughout BC, variety of fishery, tagging, and survey samples from the 1980s-2000s
• Add 2010 U.S. NWFSC samples to get better estimate of t0

​

8

Dotted line = 85, 95 cm; size of 50 % maturity

Growth curve

• High residual variation in growth curve (SD = 0.18-0.20)
• High K parameter relative to longevity, size is not a good predictor of age (M/K around 1)
• Dogfish difficult to age (erosion of bands, deposition stops during pregnancy)

9

Growth curve

Different growth curve compared to U.S. assessment

US assessment has lower K/higher Linf with lower residual SD

​

Differences could be due to:

• Real differences in growth
• Sampling (area, time)
• Reader skill

10

Maturity

11

​

Maturity at length from synoptic trawl survey:

Code 55 = Females with mature gonads

Code 77 = Females carrying pups (50% maturity at 95 cm)

​

The latter is more appropriate to calculate spawning output

Maturity

12

Maturity at size depends on growth curve

​

If there is uncertainty about growth curve, propose to use maturity at age from literature to be explicit about assumptions

​

​

Explore 2 maturity ogives:

• Taylor and Gallucci samples from Washington collected in 2000s (“early maturity”)
• McFarlane and Beamish samples from Strait of Georgia in 1980s (“late maturity”)

​

​

​

Maturity

13

Maturity at age from synoptic maturity at size with BC growth curve very similar to Taylor and Gallucci 2000s (early maturity)

​

Fecundity

14

​

Fecundity (2-16 pups/pregnant female) is a linear function of size (Ketchen 1972):

Converted to age-based values:

Natural mortality

Maximum observed age:

73 for females, 70 for males (4B) M = 0.074 (per Hamel and Cope 2022)

54 for females, 53 for males (outside waters)

​

Proceed with M = 0.074 for both sexes based on meta-analytic relationships

​

​

​

15

Natural mortality

Due to low fecundity, there is an upper bound on natural mortality such that the slope of the unfished replacement line does not exceed one. Otherwise, a female can not produce enough offspring over lifetime to replace itself

​

For consistency, not looking to sample M from a prior distribution

16

Questions – Data inputs and biology

17

Assessment history

Wood et al. 1979:

• Estimated unfished vulnerable biomass (BC to Oregon) of 200,000 tonnes from Delury and Leslie depletion models
• Assumed pup survival and recruitment was density-independent. M = 0.094, solved from the maturity and fecundity schedule
• MSY was estimated to be 9-11 thousand tonnes where MSY = FMSY * 200,000 and FMSY = 0.5 M
• Natural mortality was proposed as the density-dependent mechanism for the yield curve. Alternative hypotheses of density-dependent growth and fecundity were biologically implausible to achieve the proposed MSY value.�

Gallucci et al. 2010:

• Surplus production model, was not used for advice

​

18

Current modeling approach

• Fit Stock Synthesis age-structured model to catch, indices, length composition (downweight length composition sample sizes with McAllister-Ianelli method)
• Assumption of closed population in BC
• Explicitly incorporates low fecundity and lags in age structure, but must model must convert ages to sizes

• Constant growth, fecundity, and natural mortality in most scenarios. Re-scale max maturity = 0.5 to accommodate 2-year gestation period
• Estimate selectivity for fishing gears and surveys with size data. Mirror selectivity otherwise.
• Selectivity priors
• Normal prior for length of full selectivity, mean from modal length of aggregate composition with CV = 0.3
• Width of ascending limb from 5 %ile with CV = 0.3

​

19

Current modeling approach

• Use Ian Taylor’s stock-recruit relationship where recruitment explicitly can not exceed spawning output, designed for elasmobranchs. Low fecundity 🡪 maximum steepness value can be less than 1
• Estimate unfished recruitment (R0) and stock-recruit productivity parameter (zfrac, 0-1), both uniform priors

​

�Illustrative example:

• Unfished population N = 100 adults, which produce 10 pups/adult, that's 1,000 pups (B0)
• Density-dependent unfished pup survival is 0.6, then we get 600 recruits (R0)

​

• A population of N = 20 produces 200 pups (0.2 B0). If steepness is 1 in a closed population, we'd have to conjure 600 recruits from 200 pups. Thus, steepness is capped at 0.33 (200/600)

​

20

Stock recruit relationship

21

Sensitivity grid across growth and maturity

Four models:

​

(A1) BC growth curve + early maturity

(A2) US growth curve + early maturity

(A3) BC growth curve + late maturity

(A4) US growth curve + late maturity

​

22

Estimates

• Stock biomass is low today because declines in the indices occur with low catches relative to the 1940s
• Note we don’t observe the spawning output

​

23

Problems

• Can not fit HBLL and SYN indices. The declines are too extreme with dogfish longevity and catch series

​

24

Problems

Erratic selectivity values

• IPHC selectivity curve has no inflection with BC growth curve. Model must generate composition with lots of large fish
• Can’t fit hook and line length composition with US growth curve? Doesn’t fit data at all

​

25

Corresponding�age selectivity

• Most mortality today is assumed to be of immature animals (selectivity in row 2-3)

​

​

26

**New slide**

Colours = selectivity, black = maturity

Problems

Erratic selectivity values

• Can’t fit hook and line length composition with US growth curve? Doesn’t fit data at all

​

27

Problems

• No density-dependence estimated in the model (no catch advice)
• The model must make the stock as unproductive as possible to fit the declining indices

28

Likelihood profile of zfrac and corresponding steepness

Problems

• Length data do not inform depletion
• Little change in predicted size composition and mean length between 1937 (unfished) and recent years, despite low B/B0 in model
• Low M/K life history 🡪 mode in length comp represents many cohorts. Shape of the length comp not responsive until stock is very low

29

Boxed numbers = mean length

Problems

• Length data do not inform depletion
• Little change in predicted size composition and mean length between 1937 (unfished) and recent years

​

• Estimated selectivity is useful to describe impacts of fishing to age structure, e.g.,
• Recovery of mature cohorts in the 1970s following the liver fishery
• Changes in sex ratio due to targeting of females

30

Likelihood profile with zfrac

31

**Updated figure**

zfrac and reference points

• zfrac determines the value of FMSY and MSY (height of yield curve)
• zfrac = 1 does provide upper bound on MSY
• MSY calculated on current selectivity (discards of much younger fish compared to retention in earlier fishery)
• Ratio of BMSY/B0 robust to zfrac

​

32

Retrospective pattern

• Model (A1), least of our problems

​

33

Alternative fits

Fit to three separate sets of indices:

(A6) IPHC + CPUE

(A7) SYN

(A8) HBLL

​

Model says there’s no stock with SYN and HBLL?

34

**Updated slide**

Exploring an increase in natural mortality

• Increase in M to explain additional contribution to index decline from any combination of pinniped predation, behavior, migration offshore (not related to fishery)
• No data in the model validate whether this is an appropriate approach
• Explored stepwise and linear increase:

35

Pinniped census: Res Doc 2018/006

Exploring an increase in natural mortality

• Timing of M change affects perceived impact of 1940s fishery. If M increase occurs in the 1980s, early fishery has little impact on stock

​

36

Exploring an increase in natural mortality

• Timing of M change affects stock size, but no density-dependence estimated

​

37

**New slide**

Summary and discussion

1. Do we have an acceptable assessment?

• Can’t fit HBLL and SYN index well, decline too rapid for long-lived life history. Need a reason to reject HBLL and SYN, and accept the IPHC and CPUE series
• Important dynamics
• Size compositions indicate current mortality on immature females
• Low fecundity and low M creates “stiff” model for sensitivity analysis
• Demographic modeling gives us upper bounds in
• Natural mortality
• Steepness 🡪 MSY and FMSY

​

2. Is catch advice possible?

• Age-structured production model approach assumes density-dependence in the population, but estimates little to no surplus production
• Does density-dependence even matter? See zfrac profile shows historical F < FMSY when catch = 0

38

**Updated slide**

Extra slides

39

Fit to length data

40

Numbers = sample size

Fit to length data

41

Numbers = sample size

Fit to length data

42

Numbers = sample size

Standardized trawl CPUE by depth

• Steepest declines in shallow waters

​

43

Standardized trawl CPUE by month

• Some variability by month (steepest in August, least steep September-January)
• Trawl surveys in May-July representative of average
• HBLL OUT survey in August-September

​

​

44

Survey Timing

Thinking about whether survey catchability is changing due to seasonality in dogfish behavior/migration..

​

Haida Gwaii: Even years, August/September

Hecate Strait: Odd years, May/June

QCS: Odd years, July/August

WCVI: Even years, May/June

​

HBLL N: Even years, August/September

HBLL S: Odd years, August/September

​

45

Steller sea lion abundance 1971-2017

46

DFO. 2021. Trends in Abundance and Distribution of Steller Sea Lions (Eumetopias Jubatus) in

Canada. DFO Can. Sci. Advis. Sec. Sci. Advis. Rep. 2021/035.

Overwintering California sea lion abundance

• A tripling between 2009 and 2021

47

DFO. 2023. California Sea Lion Abundance Estimation in Canada, 2020–21. DFO Can. Sci.

Advis. Sec. Sci. Advis. Rep. 2023/016.

Coastwide trends from trawl surveys

• Declines over last 20 years have been coastwide, although steepest off the west coast of Vancouver Island
• (biomass is a relative value here and not meaningful; rightmost legend is proportion change not %)

​

48

Davidson et al. In prep.