Age-specific mortality and transport of larval walleye pollock Theragra chalcogramma in the western Gulf of Alaska

Sarah Hinckley, Kevin Bailey, Susan Picquele

NOAA, National Marine Fisheries Service, Alaska Fisheries Science Center, 7600 Sand Point Way NE, Seattle, Washington 98115, USA

Mary Yoklavich

NOAA, National Marine Fisheries Service, Pacific Fisheries Environmental Group, Southwest Fisheries Science Center, PO Box 831, Monterey, California 93942, USA

Phyllis Stabeno

NOAA, OAR, Pacific Marine Environmental Laboratory, 7600 Sand Point Way NE, Seattle, Washington 98115, USA

Marine Ecological Progress Series, 98, 17-29.
Copyright ©1993 Inter-Research. Further electronic distribution is not allowed.

DISCUSSION

Estimating larval mortality rates for marine fish is difficult due to the many sources of sampling bias and error. In this study, we have attempted to avoid many of the problems which have complicated earlier studies. Use of multiple surveys instead of the more usual catch curve analysis made it possible to avoid the assumptions of constant egg production and constant mortality. Comparisons of several gear types and day/night sampling indicated that retention and escapement of walleye pollock larvae from sampling gear was not a problem. Transport of larvae through the sampling area was accounted for by sampling virtually the entire distribution of walleye pollock larvae, and boundaries were refined through the use of a model of advection and diffusion developed for this area. By analyzing daily growth increments of otoliths, we avoided the uncertainty of using length as a proxy for age, ensured that the same cohorts were identified in each survey, and were able to estimate mortality rates for specific 3 d cohorts of larvae. Ageing also made it possible to truncate the range of cohorts that had estimated mortality rates to those that were adequately sampled in each pass. Also, Somerton & Koyabashi (1992) show that truncating the age distribution avoids bias due to low capture probabilities. Larvae hatched after completion of the first pass could also be excluded this way. Since virtually the complete vertical and horizontal distribution of larvae was sampled in this study, differences in spatial distribution of larvae, for example with age, were not a factor.

The range of mortality rates estimated from this study (–0.0574 to 0.0757) was low compared to other published mortality rates for walleye pollock during this developmental period. Rates from prior studies have ranged from z = 0.065 to 0.11 d (Table 4). Estimates of mortality rates reported for the larval stage of other gadoids also have been somewhat higher (e.g. for Atlantic cod: z = 0.08 d, Houde 1987; and z = 0.12 d, Sundby et al. 1989). The fact that the daily mortality rates for all of the 3 d cohorts were consistently low with none exceeding z = 0.0747 d and that mortality rates determined from catch curve analysis were also low supports the conclusion that the low overall mortality rates found in this study were realistic.

If your browser cannot view the following table correctly, click this link for a GIF image of Table 4

In accordance with the low mortality rates observed for larvae, there is evidence that 1988 may be a relatively strong year-class for walleye pollock spawned in Shelikof Strait. There was a high relative abundance of 0-age juveniles caught in an August–September 1988 juvenile survey (Bailey & Spring 1992) indicating that recruitment was established early. Furthermore, the 1988 year-class has made a strong showing in the annual March hydroacoustic surveys of the Shelikof spawning stock in both 1990 and 1991 (Hollowed et al. 1991; J. Traynor, AFSC, pers. comm.). Low larval mortality rates during the spring of 1988 could have been a contributing factor to the strength of the year-class.

It is interesting to note that wind speeds during May of 1988 were much lower than normal for an extended period of time including the time period of this survey. The mean May wind speed (derived from atmospheric pressure gradients estimated from National Meteorological Center data) for 1988 was 4.63 m s, whereas the average for May of 1980 to 1987 and 1989 was 6.70 m s. The mechanism explaining a relationship between wind speed and larval survival is not known, although several authors have speculated on possible causal relationships in other species (Lasker 1975, Sundby & Fossum 1989, MacKenzie & Leggett 1991).

Based on the power test done previously, we did not expect to resolve mortality rates below 0.037 d (for a power of 0.95). The 4 significant mortality rates were all greater than this (0.041 to 0.067 d). Mortality rates of 9 of the 13 oldest cohorts were not significantly different from zero. There are several possible reasons for this. First, patchiness in the larval distribution could have caused the variance to be too large to make low mortality rates indistinguishable from zero. Better stratification and increased sampling in areas of high abundance could have reduced this source of variability. However, the patchy distribution of larvae and the uncertainty in predicting areas of high larval abundance make survey design difficult. Second, the fact that each pass was not synoptic (i.e. each took from 4 to 6 d to complete) and that the pass duration was greater than the duration (3 d) of each cohort caused an increase in the variance in the age-length key, as it was a composite from several days. Adding more passes and more days between the passes would be alternative ways to increase our ability to detect low rates of mortality. Third, there is high natural variation in the age-length relationship. A larger ageing sample would be helpful in reducing this source of variation. And fourth, although we tried to eliminate the possibility of advection of larvae into the survey area between Passes 1 & 2 through the survey design and use of the model of advection and diffusion, it is still possible that such immigration did occur.

It is interesting to note that the 4 significant mortality rates were for cohorts of young larvae. Mortality rates actually are probably higher for young fish. A decrease in mortality rate with age or size is to be expected (Cushing 1974), and has been reported for walleye pollock larvae by Yoklavich & Bailey (1990).

The mortality rates for the older cohorts were very low, and this, combined with very high variance, made them statistically indistinguishable from zero. The lack of significance of the mortality rates for older cohorts could also be explained by an underestimation of the abundance of older larvae during Pass 1 or an overestimation during Pass 2, either of which would have caused an apparent decrease in the mortality rate (or negative rates). There was no increase in the variance of the mortality rate relative to the mean with age, indicating that the distribution of older larvae was not patchier (which would have increased the variation in abundance estimates; Pielou 1977).

Transport of larvae was examined by displacement of larvae, by satellite drifter trajectories, and by satellite imagery. During the spring, daily mean current speeds in Shelikof Strait have been measured from 10 to 50 cm s (Schumacher et al. 1989); however, the current structure in Shelikof Strait is complex in both time and space (Stabeno & Incze unpubl.). Circulation is dominated by the Alaska Coastal Current (ACC), a narrow (less than 20 km wide), buoyancy-driven current system. There is much variability in current speeds and location and direction of the ACC within the study area (Reed et al. 1989). Eddy-like features on the order of 20 km in diameter are common and add complexity to the current structure. Due to the complex wind regime, regions of wind-driven convergence and divergence are common. Both eddies and regions of wind-generated convergence probably play a role in the observed patchiness of larvae.

Larval drift rates estimated by several different methods were all similar. The average drift rate from these 3 methods was ca 3.1 to 3.9 cm s. This rate is comparable to the larval drift rates estimated from other studies for larval walleye pollock (Incze et al. 1989: about 4 cm s; Kim & Kendall 1989: 3.2 cm s; Reed et al. 1989: 5 cm s; Kim & Bang 1990: 3.13 cm s). Other studies have estimated transport rates as high as 8 to 9 cm s (Incze & Campbell 1989, Incze et al. 1990).

Daily average velocities usually are weak (<10 cm s) in the region near Sutwik Island and the Semidi Islands, the areas where the larval patches were found for this study. This flow is not in the ACC, where currents can be as fast as 50 cm s. The low net drift of the larvae in the patches was small because they are not in the ACC. The model of advection and diffusion was able to retain larval patches if an eddy was included in the current field (Stabeno & Incze unpubl.). Both cyclonic and anticyclonic eddies have been observed in the Shelikof Strait region several times during the past 5 yr (Vastano et al. 1992, Schumacher et al. 1993, Stabeno & Incze unpubl.) and larval patches appear to be associated with these eddies.

Also, it is notable that there was no apparent along-strait gradient of ages with older larvae further downstream, as might be expected if all cohorts were spawned in the same location (Kim & Kendall 1989, Kendall & Picquelle 1990) and no mechanism for retention of cohorts was present.

Inferences can be made about the patterns and velocities of surface flow based on sea surface temperature (SST) pattern displacements derived from satellite images (Vastano & Borders 1984). A preliminary examination of SST patterns for the survey period in May of 1988 (A.C. Vastano, Dept. Oceanography, Texas A&M Univ., College Station, TX 77843, USA, pers. comm.) showed a circulatory feature to the northeast of the Semidi Islands on May 23 to 24 and another to the south of the Semidi Islands on June 4 to 5. The direction of flow between these 2 surface features would be consistent with the assumption of coherence between the larval patch features found in this area. It is not clear from satellite images whether any circulatory features existed in the Sutwik Island area during our study that could have entrained the buoy and the larvae. The SST photographs do, however, indicate several significant meanders in the surface flow, particularly during the Pass 1 time period. The images also indicate a pattern of surface flow along the Alaska Peninsula near Sutwik Island in the same direction as the movement of the buoy and the inferred drift of Patches 1 & 2.

Larval patchiness may play a significant role in overall rates of larval mortality. Patch-level studies of larval growth, feeding and predation, and their importance to larval survival are being conducted for walleye pollock. The techniques described in this study will also continue to be used to obtain age-structured mortality rates, which, along with studies of the mechanisms and causes of larval mortality of walleye pollock in Shelikof Strait, will promote a better understanding of the dynamics of the recruitment process for this stock.

In summary, realistic and precise rates of larval pollock mortality were estimated for individual cohorts during the period of downstream drift in 1988. These estimates were low compared to those from other years, coinciding with calm ocean conditions during the larval drift period, and corresponding to relatively strong recruitment of the 1988 year class. These data reinforce the concept that good larval survival is a necessary, but perhaps not sufficient, factor for development of strong year-classes. Smaller and younger cohorts experienced significant mortality over the 12 d period between sampling dates, while mortality of older and larger cohorts was insignificant. The results support the concept of intra-specific size-dependent mortality.

Acknowledgements. This research is contribution number FOCI 0158 to NOAA's Fisheries Oceanography Coordinated Investigations. We thank Annette Brown, Nazila Merati and Stella Spring for assistance in the laboratory, Debbie Siefert for shrinkage data, Michiyo Shima for help with the gear comparisons, and other FOCI staff for general assistance. Drew Vastano gave useful advice on interpretation of the satellite imagery. The authors also thank Bob Francis, Anne Hollowed, Art Kendall, and Dave Somerton for their helpful comments on drafts of this manuscript.

Return to previous section or go to next section

PMEL Outstanding Papers

PMEL Publications Search

PMEL Homepage