STEVE ROMAINE - PUBLICATION ABSTRACTS 

Abstract

A methodology combining acoustics and net tows was tested and evaluated on euphausiid biomass and distributions in two B.C. inlets adjoining the Strait of Georgia from 1991 to 1996. Better information concerning the reliability of seasonal and interannual variability of euphausiid stock size estimates as well as the within-time-period statistical reliability of these estimates was the goal of this thesis. Repeated surveys were conducted primarily in the spring and fall in both Jervis and Saanich Inlets, and monthly surveys were conducted in Jervis Inlet between June 1994-June 1995. Acoustic samples were the primary data source for biomass estimates, with net tows providing species composition and size-frequency distributions.

Transects provided closely-spaced acoustic sample points along lines. Estimation of total stock size from samples required point observations to be converted into maps, and maps to be converted to integrated biomass estimates for the survey area. Block averages estimated stock sizes from a mosaic of equally weighted sample averages integrated over the survey area. The geostatistical interpolation method of kriging estimated unknown points within an area based on a function of spatial autocovariance and unique weightings applied to neighbouring measurements. Both acoustic estimation methods provided similar results, but kriging was the preferred method of choice as data points were considered spatially dependent for estimates and high resolution spatial distributions could be further displayed and analyzed.

One source of uncertainty in stock size estimates based on linear acoustic transects was their repeatability. This uncertainty was evaluated by comparing results from independent survey grids between both transects and days. On a daily basis, estimates agreed to within 31-59% (2 standard deviations) and between survey grid agreement was within 46-63%.

Seasonal variation of stock size due to predation, growth, and migration was on average about 1.3-2.6x greater than within time period variations due to statistical uncertainty. Monthly surveys of Jervis Inlet noted a bimodal biomass peaking in spring, followed by a large drop in early summer. A second peak occurred in October followed by a large winter decline. Less frequent sampling of Saanich Inlet showed higher average euphausiids stocks in the fall compared to the spring, and a similar winter decline. Broader coverage of Jervis Inlet noted an over-winter biomass decline similar to Saanich Inlet.

Conversion from acoustic return to biomass was an additional source of uncertainty. A hybrid acoustic target strength model was developed to minimize this uncertainty by using seasonal size-frequency information combined with Macaulay's (1994b) model for euphausiid acoustic returns. The strong seasonal variation of size-frequency caused target strength conversion factors to vary up to a magnitude of four.

Top of Page

Abstract

Management of euphausiid and other Strait of Georgia fisheries requires better information about seasonal and interannual variability of euphausiid stock size and the within-time-period statistical reliability of these stock size estimates. To provide this, we conducted repeated euphausiid surveys in Jervis Inlet over the past five years: twice per year (Oct-Nov and Jan-Feb) in most years since November 1991; and monthly June 1994 - June 1995. Underway echosounding gave spatially detailed measurements along zig-zag transect lines. Net tows (mostly discussed previously, Mackas and Moore 1994) provided species identification and size-frequency distributions.

The steps in our analysis were:

• Generate snapshot 'maps' of euphausiid biomass density within the Inlet by averaging/interpolation of data from the closely-spaced acoustic transect lines making up individual survey grids.
• Integrate these biomass density estimates over the inlet area, giving time series estimates of total stock size that track seasonal and year-to-year changes in stock biomass and distribution.
• Compare observed changes in stock biomass vs. removals by the commercial fishery.
• Evaluate statistical consistency of the stock size estimates through within-time-period comparison of replicated survey grids, and alternate data collection and analysis methodologies.

Overall, we believe our within-time-period stock size estimates are reliable to about factor of 2.5. We partition the uncertainty as follows:

• Within-time-period-repeated complete survey grids done with the same echosounder and the same data processing method agreed on average to ±25-30%. Partial surveys agree to about factor of 1.65.
• For replicate surveys done with two different echosounders (operating at 104 and 200 kHz from different ships), along-line correlation was r² = 0.74. Total biomass estimates from the 200 kHz averaged 27% greater.
• Alternate data processing methods (subarea block averages vs. kriging with and without forcing to zero biomass density along the inlet shoreline) showed similar spatial and seasonal patterns. Block averages were higher (about 15%) than kriging with shoreline zero-forcing but we believe that kriging is more conservative and robust with respect to aliased patchiness.
• Conversion from acoustic return to biomass was the largest single source of uncertainty. Estimates based on alternate calibration models differed by factors of up to 3 for individual time periods, typical variation was about factor of 2.

Total stock estimates ranged from a minimum of <1000 tonnes (Feb. 93) to a maximum of>10000 tonnes (April 95). Both seasonal and interannual components of variation were large (3-10 fold). In most years, acoustically-estimated stock biomass declined during the winter. Timing of the decline is coincident with the fishing season. But the magnitude of the winter decline (3000-6000 tonnes) is much greater than the removal by the fishery (less than 500 tonnes). We therefore attribute most of it to seasonal variation in the balance between growth and natural mortality. We conclude that present harvest levels are a small fraction of present stock size. 

Top of Page

Abstract

Currently, most of the commercial harvest of euphausiids in British Columbia has come from either the outer portions of Jervis Inlet or the surrounding Strait of Georgia area. In order to provide better information regarding within-time-period statistical consistency of results from acoustic survey grids and seasonal variability of euphausiid stock size within Jervis Inlet, we obtained replicated monthly surveys between June 1994 and June 1995. Underway echosounding gave spatially detailed measurements along zig-zag transect lines. Net tows provided species identification and size-frequency distributions. We wanted to evaluate the variability of euphausiid stock size estimates due to the following sources: 1) Incomplete survey coverage (variability among within-time period replicate and alternative calculation methods); 2) Target strength assumptions; and 3) Month-to-month and seasonal changes in stock size. Various methods were employed to estimate the total stock size, including moving averages, linear kriging, and extrapolation from net tows at a small number of locations.

Four different sets of biomass estimate comparisons were considered in this study:

1. Repeated identical transects taken during most months allow the determination of statistical stability of the total biomass estimate assuming the population does not change over a one day period. Our estimates found that repeat surveys agreed on average to ±17.2%.
2. On two occasions, two different echosounders (operating at 100 and 200 kHz from different ships) ran parallel survey tracks (separated by 100-500 m). Along-line correlation was r² = 0.74 between the two echosounders. Total biomass estimates differed by an average of 27.3%.
3. For comparison between interpolation / integration methods, moving averages gave higher estimates (about 30%) than kriging calculations; however, we believe that interpolation is more robust with respect to skewing from the highly concentrated euphausiid patches that were commonly observed within Jervis.
4. Average euphausiid body length ranged from 12 to 21 mm with an overall average of 15 mm. Various target strength models exist to estimate euphausiid biomass from scattering strength; in this study, the basic modified Johnson and a model devised by Macaulay were used to estimate biomass within the inlet area based on the size-frequency averages from net tows. Variations in monthly biomass estimates based on the Macaulay model ranged from -40% to +150% of the basic Johnson model, with a net overall annual biomass difference of zero between the two models.

Top of Page