Roughhead grenadier (Macrourus berglax) COSEWIC assessment and status report: chapter 17

Appendix A - Construction of abundance indices

Overview

The survey data used in this report are from the DFO fall survey of NAFO divisions 2GHJ3KLMNO and the spring survey of divisions 3LNO. The area covered by these surveys has varied considerably over time. In order to construct consistent time series of relative abundance indices, subsets of consistently sampled strata were used (i.e., strata sampled in most years). For each survey, two or three time series were constructed, a long time series covering a smaller area, and shorter time series covering a larger area.

In constructing times series of abundance indices from these data, it is important to note that the stratification scheme for these surveys was revised substantially between 1992 and 1993 (Bishop 1994). Stratum boundaries were extensively revised, resulting in substantial changes to stratum areas (and thus to the weighting factors used to calculate stratified means). A number of stratum codes were dropped and new codes were added, without any substantial change to the total area covered. This needs to be kept in mind when selecting strata for inclusion in consistent time series (e.g., a new code appearing in 1993 should be included if it covers an area included in the analysis under a different code prior to 1993). Additional deepwater strata were added to the surveys in 1996.

For strata sampled in most years but missed in a few years, predicted values were used for the missed years. These were obtained from a statistical model with year and stratum terms. Both year and stratum were used as factors or class variables in these models (as opposed to continuous covariates as in the models in Devine et al. 2006). A generalized linear model was used, with a log link and Poisson error, allowing for overdispersion. This model assumes that distribution does not change between years. In order to minimize effects of distribution change, models were restricted to the year with missing data and the preceding and following year or two. Narrower year ranges were used during periods of distribution change. Models that included a year´stratum interaction (which reflects changes in distribution) were examined to ensure that the importance of this term was small relative to the stratum term.

The following indices were constructed:

  1. Fall 2J3K, 1978-2003, Strata – 2J: 201-219, 222-224, 227-231, 234-240; 3K: 617, 620-642, 645-647, 650-652
  2. Fall 2J3KL, 1981-2003, Strata – 2J: 201-219, 222-224, 227-231, 234-240; 3K: 617, 620-642, 645-647, 650-652; 3L: 328, 341-350, 363-372, 384-392
  3. Fall 2J3KLNO, 1996-2003, Strata – 2J: 201:240; 3K: 617-631, 633-654; 3L: 328, 341-350, 363-372, 384-392, 729-736; 3N: 357-362, 373-383, 723-728; 3O: 329-340, 351-356, 717-722
  4. Spring 3LNO shallow, 1971-2002, Strata – 347-354, 359-366, 368-392
  5. Spring 3LNO deep, 1996-2005, Strata – 328:366, 368-392, 717-736

(This index was provided by M. Koen-Alonso and F. Mowbray, DFO, St. John’s).

Predicted values for missed strata

Statistical models were fit to the following subsets of data in order to obtain predicted values for missed strata. Unless otherwise noted, data were the tow-by-tow catch rates (as opposed to stratum means).

A. Fall data, all sizes
1978-2003 2J3K index
  1. 1977-1982, 2J strata 201:219, 222:224, 227:231, 234:240. Used for missed strata in 2J in 1978-1982.
  2. 1979-1983, 3K strata 617, 620:642, 645:647, 650:652. Used for missed strata in 3K in 1979-1983.
  3. 1983-1985, 2J strata 201:219, 222:224, 227:231, 234:240. Used for missed strata in 2J in 1984.
  4. 1985-1987, 3K strata 617, 620:642, 645:647, 650:652. Used for missed strata in 3K in 1986.
  5. 1987-1990, 3K strata 617, 620:642, 645:647, 650:652. Used for missed strata in 3K in 1988 and 1989.
  6. 1988-1990, 2J strata 201:219, 222:224, 227:231, 234:240. Used for missed stratum 236 in 2J in 1989.
  7. 1995-1997, 2J strata 201:219, 222:224, 227:231, 234:240. Used for missed strata in 2J in 1995.
  8. 1999-2001 3K strata 617, 620:642, 645:647, 650:652. Used for missed stratum 650 in 3K in 2000.
1981-2003 2J3KL index
  1. 1981-1984, 2J3KL strata 201:219, 222:224, 227:231, 234:240, 617, 620:642, 645:647, 650:652, 328, 341:350, 363:372, 384:392. Used for missed strata in 3L in 1981-1983.
  2. 1985-1987, 2J3KL strata 201:219, 222:224, 227:231, 234:240, 617, 620:642, 645:647, 650:652, 328, 341:350, 363:372, 384:392. Used for missed stratum 388 in 3L in 1986.
  3. 1992-1994, 2J3KL strata 201:219, 222:224, 227:231, 234:240, 617, 620:642, 645:647, 650:652, 328, 341:350, 363:372, 384:392. Used for missed stratum 348 in 3L in 1993.
1996-2003 2J3KLNO (deep) index
  1. 1996-1998, 2J3KLNO strata 201:240, 328:392, 617:631, 633:654, 717:736. Used for missed strata in 3O in 1996.
  2. 2000-2002, 2J3KLNO strata 201:240, 328:392, 617:631, 633:654, 717:736. Used for missed stratum 650 in 2000.
B. Spring data, all sizes
1971-2002 3LNO shallow index (1972, 1983, 1984 omitted)
  1. Missed strata where roughhead grenadier have never been caught were filled with 0 (347-349, 351-354, 359, 360, 364, 371 373, 376, 384).
  2. 1971-1974, 3LNO strata 347:354, 359:366, 368:392. Used for missed strata in 1971 and 1973.
  3. 1974-1977, 3LNO strata 347:354, 359:366, 368:392. Used for missed strata in 1976.
  4. 1977-1979, 3LNO strata 347:354, 359:366, 368:392. Used for missed stratum 368 in 1978.
  5. 1990-1992, 3LNO strata 347:354, 359:366, 368:392. Used for missed strata in 1991.

1996-2005 3LNO deep index

Used time series provided by Mariano Koen-Alonso and Fran Mowbray (DFO St. John’s).

C. Fall data, adult females
1978-2003 2J3K index
  1. 1977-1981, 2J strata 201:219, 222:224, 227:231, 234:240. Used for missed strata in 2J in 1978-1980.
  2. 1978-1981, 3K strata 617, 620:642, 645:647, 650:652. Used for missed strata in 3K in 1978-1979.
  3. 1981-1984, 2J3K strata 201:219, 222:224, 227:231, 234:240, 617, 620:642, 645:647, 650:652. Used for missed strata in 2J3K in 1982-1984.
  4. 1985-1987, 3K strata 617, 620:642, 645:647, 650:652. Used for missed strata in 3K in 1986.
  5. 1987-1991, 2J3K strata 201:219, 222:224, 227:231, 234:240, 617, 620:642, 645:647, 650:652. Used for missed strata in 2J3K in 1988-1990, except for stratum 647.
  6. 1995-1997, 2J strata 201:219, 222:224, 227:231, 234:240. Used for missed strata in 2J in 1995.
  7. 1999-2002, 2J3K strata 201:219, 222:224, 227:231, 234:240, 617, 620:642, 645:647, 650:652. Used for missed strata in 2J3K in 2000.
  8. Predicted values for stratum 647 in 1988-1990 were clearly unreasonable (too large) using model 23. In 1990, there were tows in this stratum but no length frequencies were taken. A prediction for the mean catch rate of adult females in this stratum in 1990 was obtained by multiplying the mean total catch rate in 647 in 1990 by the average proportion mature (>275 mm AFL) in this stratum in 1985-1992. Predictions for stratum 647 in 1988 and 1989 were obtained by running the model on stratum means (including the prediction for 647 in 1990 given above) for 1987-1992.
1981-2003 2J3KL index
  1. 1981-1984, 2J3KL strata 201:219, 222:224, 227:231, 234:240, 617, 620:642, 645:647, 650:652, 328, 341:350, 363:372, 384:392. Used for missed strata in 3L in 1981-1983.
  2. 1985-1987, 2J3KL strata 201:219, 222:224, 227:231, 234:240, 617, 620:642, 645:647, 650:652, 328, 341:350, 363:372, 384:392. Used for missed stratum 388 in 1986.
  3. 1987-1992, 2J3KL strata 201:219, 222:224, 227:231, 234:240, 617, 620:642, 645:647, 650:652, 328, 341:350, 363:372, 384:392. Used for missed strata in 1988-1990.
  4. 1992-1994, 2J3KL strata 201:219, 222:224, 227:231, 234:240, 617, 620:642, 645:647, 650:652, 328, 341:350, 363:372, 384:392. Used for missed stratum 348 in 1993.
1996-2003 2J3KLNO (deep) index
  1. 1996-1998, 2J3KLNO strata 201:240, 328:392, 617:631, 633:654, 717:736. Used for missed strata in 3O in 1996.
  2. 2000-2002, 2J3KLNO strata 201:240, 328:392, 617:631, 633:654, 717:736. Used for missed strata in 2000-2002.

Additional time series

Mark Simpson and Dave Kulka (DFO St. John’s) have developed time series of abundance and biomass indices that control for variation in the area sampled by either deleting missed strata from the time series or combining them with adjacent strata in a similar depth zone. Trends are similar to those described in Figures 5 and figure6, except that the decline in the 1980s and early 1990s is slightly steeper in these indices (Figure A1). This may reflect the exclusion of deep strata (>750 m) from these indices for years before 1996. Deep strata (750-1000 m) are included in the index in Figure 5a.

Mariano Koen-Alonso and Fran Mowbray (DFO St. John’s) have also developed abundance and biomass indices from these surveys. Their indices are based on a subset of core strata sampled in most years. Because not all core strata are sampled in all years, the area covered by their index varies slightly from year to year. Also their indices include deepwater strata sampled only since 1996. The time series based on their index is shown in Figure A2. Their spring abundance index (1996-2005) is used in Figure 6 of this report.

References

Bishop, C.A. 1994. Revisions and additions to stratification schemes used during research vessel surveys in NAFO subareas 2 and 3. NAFO SCR Doc. 94/43.

Devine, J.A., K.D. Baker, and R.L. Haedrich. 2006. Deep-sea fishes qualify as endangered. Nature 439: 29.

Figure A1. Abundance and biomass indices for roughhead grenadier from the fall and spring bottom-trawl surveys of NAFO divisions 2J3KLNO, supplied by Mark Simpson and Dave Kulka (DFO, St. John’s). Engels indices exclude strata with depths >750 m.

Figure A1.  Abundance and biomass indices for roughhead grenadier from the fall and spring bottom-trawl surveys of NAFO divisions 2J3KLNO, supplied by Mark Simpson and Dave Kulka (DFO,St. John’s). Engels indices exclude strata with depths >750 m.

Figure A2. Abundance and biomass indices for roughhead grenadier from the fall and spring bottom-trawl surveys of NAFO divisions 2J3KLNO, supplied by Mariano Koen-Alonso and Fran Mowbray (DFO, St. John’s). Results for core strata with more than 2 sets. Note that deepwater strata are added to the core strata in 1996.

Figure A2.  Abundance and biomass indices for roughhead grenadier from the fall and spring bottom-trawl surveys of NAFO divisions 2J3KLNO, supplied by Mariano Koen-Alonso and Fran Mowbray (DFO, St. John’s). Results for core strata with more than 2 sets. Note that deepwater strata are added to the core strata in 1996.

Page details

Date modified: