Temperature-specific development rates of eggs of walleye pollock are needed to estimate their abundance, distribution, production and mortality by age intervals (Kendall and Kim 1989, Kendall and Picquelle 1990, Kendall et al. 1994, Picquelle and Megrey 1993, Brodeur et al. 1996). These estimates are necessary for individual-base modeling and biophysical modeling of the ecosystem of the Southeast Bering Sea as a means of discerning the natural processes that support walleye pollock as the dominant species. Existing models of temperature-specific development rates of eggs of walleye pollock are inappropriate for use in Southeast Bering Sea modeling because researchers used eggs of populations geographically distant from the Bering Sea or incubation temperatures were too high. Eggs need to be obtained close to the area to be studied in order for results to be accurate (Blood et al. 1994) and incubation temperatures must encompass those documented in the study area. Temperatures as low as -1.5EC were recorded within walleye pollock egg patches April 1995. These same egg patches were within the top 30 m of the water column, therefore the effect of light on incubation rates must be determined. This new information will be used to develop a new model incorporating data from the nearby Shelikof Strait population. This new model will span -1 to 7.7EC and will improve the results of walleye pollock egg vertical distribution research and IBM and biophysical modeling of the Southeast Bering Sea.

In April 1997, eggs from walleye pollock that were collected northwest of Unimak Island were incubated in constant dark at average temperatures of -0.6, 0.4, 2, and 3.8EC. An additional batch of eggs was incubated at 3.9EC under diel light conditions. Development of embryos was normal for all temperatures except -1EC; gross abnormalities included malformation of the tail, similar to that reported by Nakatani and Meada (1984), and the absence of eyes. Eggs incubated in constant dark conditions hatched at similar relative rates; 50% hatch of eggs occurred approximately midway through the period of time required for all eggs to hatch at each temperature. However, hatching of eggs incubated under diel light was delayed; 50% hatch occurred after 90% of the hatching period had elapsed. These data were used to develop a model for temperature-specific development rates from -0.6 to 3.8EC (figure above).

While reviewing the data, it was decided to repeat the incubation of eggs at the coldest temperature (-0.6EC); so many of the eggs had died at that temperature that not enough survived to hatch and the midpoint of the last developmental stage could not be used for the model. In addition, eggs collected from walleye pollock in Shelikof Strait would be incubated at the same low temperatures as the Bering Sea eggs in order to compare development and hatching rates. In light of efforts to determine whether Bering Sea walleye pollock populations are distinct from those in the Gulf of Alaska, comparison of egg development and hatch rates would be a valuable addition to what is known.

In April 1998, efforts to locate spawning walleye pollock in the Bering Sea were unsuccessful, and plans to repeat the ñ0.6 C incubation were dropped. However, eggs were collected from Shelikof Strait walleye pollock and incubated at average temperatures of 0.2, 1.8, and 2.8C. Using raw data, a temperature-specific model of development rates was constructed and compared with the low-temperature model of the Bering Sea eggs (figure at left). Preliminary data analysis indicates that three of the five parameters within the component predicting stages 7-21 are significantly different between the two models. Also, when the time predicted to reach the midpoint of four different stages (7, 13, 18, and 21) was compared between the models, results were consistently higher (more time required) for the Shelikof Strait model at 1.8 and 2.8C. Results were inconclusive for predicted times at 0.2C.

Mean length of the larvae at hatch, hatching pattern, and midpoint (point at which 50% of larvae had hatched) were compared for two similar temperatures: 0.2 and 1.8C (Shelikof Strait larvae) and 0.4 and 2.0C (Bering Sea larvae)(figure left and below left). For both the 0.2 Vs 0.4C (left) and the 1.8 Vs 2.0C (below left) comparisons, hatching began earlier, midpoint of hatch was earlier, and overall mean length was higher for the Bering Sea larvae.

Whereas there are direct relationships between increasing temperature and both beginning of hatch and mean length of larvae at hatch, a temperature difference of 0.2C between the Bering Sea and Shelikof Strait groups does not appear to be the sole reason for the difference in these parameters. Difference in the onset of hatch between the 3.8 and 3.9C groups during the 1997 Bering Sea was only 8 hours. In contrast, commencement of hatch for the Shelikof Strait eggs lagged behind the Bering Sea eggs by 80 hours in the 1.8 Vs 2.0C comparison and by 96 hours in the 0.2 Vs 0.4C comparison. Blood et al. (1994) reported an increase in mean length of newly hatched larvae of only 0.4 mm over a temperature span of 3.9C. In contrast, for the 0.2 Vs 0.4C comparison mean length of Bering Sea larvae was 0.4 mm longer than that for Shelikof Strait larvae, and 0.9 mm longer for the 1.8 Vs 2.0C comparison. Comparing midpoint of hatch between the two groups, the midpoint was 24 hours earlier for Bering Sea eggs in the 1.8 Vs 2.0C comparison but was 224 hours earlier at colder temperatures (0.2 Vs 0.4C). The degree of difference between them indicates a delay in the hatching mechanism of Shelikof Strait eggs at the colder temperature.

Although further analysis is necessary to verify these results, there do appear to be real differences in Bering Sea and Shelikof Strait egg development at cold temperatures. Fine-tuning of egg development models will proceed after verification, at which point the models can be incorporated into the individual-based and biophysical models.

Revised timeline:

Nov 98 - Jan 99 verify data and fine-tune regression models (models available for biophysical modeling and vertical distribution work)

Feb 99 ñ June 99 write manuscript

July 99 submit for publication (outlet not selected at this time)