Stock Composition

Catches of herring in the Kattegat, the Skagerrak and the Eastern part of the North Sea are taken from a mixture of two main spawning stocks (ICES 1991/Assess 15): the Western Baltic spring spawners and the North Sea autumn spawners.

In addition, several local stocks have been identified (Jensen, 1957). These have however been considered to be less abundant and therefore of minor importance to the herring fisheries (ICES 1991/Assess 15).

The North Sea autumn spawners (NSAS) enter Skagerrak and Kattegat as larvae and migrate back to the North Sea at an age of 2–3 years (Rosenberg & Palmén, 1982). The Western Baltic spring spawners (WBSS) spawn around the Baltic Island Rügen. They migrate through the Belt Sea, to the Kattegat and the Skagerrak as adults after spawning (Biester, 1979) and the juveniles starts migration to the same areas as 1- and 2-ringers.

The herring stocks in the Kattegat and the Skagerrak have been identified within samples by a number of different methods. Some of them are presented below. In a number of scientific papers the average counts in number of vertebrae in herring samples have been considered (Rosenberg & Palmén, 1982; Gröger & Gröhsler, 1995 and 1996). NSAS have a mean number of 56.5 vertebrae while the WBSS has traditionally been considered to have a mean number of 55.8 vertebrae (ICES 1992b/H:5), a more recent investigation (Gröger & Gröhsler 2000, Gröger & Gröhsler WD 2000) points to a somewhat lower value of 55.63. The most abundant local spring spawning herring, the Skagerrak spring spawners (SSS), are represented by a higher mean number, 57.0 vertebrae.

Following the tradition from Heinke (1898), several other morphometric and metric variables have been used to separate herring stocks (Rosenberg & Palmén, 1982). The use of most of these variables was evaluated by an ICES workshop in 1992 (ICES 1992b/H:5). This group concluded that a simple modal length analysis of the relevant 1–2 age groups would be precise enough for routine assessment purposes.

However, modal length analysis has proved to be an imprecise measure requiring a large sampling effort. Experience within the Herring Assessment Working Group showed that the separation procedure often failed. The amounts of herring catches that were allocated to the NSAS stock have varied between 30 to 50% of total annual landings during the last 10 years. There was an apparently very high among years variation in the proportion of spring spawners applied for the Skagerrak in quarters 3 and 4 (ICES 1999a/ACFM 12, Table 2.1). Errors in the estimate of these proportions clearly affected the quality of the assessment of the WBSS stock. A more precise measure was needed.

Otolith microstructure analysis has also been tested to separate spring and autumn spawned larvae (Moksness &

Fossum, 1991) and adults (Zhang & Moksness, 1993). Otolith growth in the larval stage (which can be inferred by microscopically examine the otolith centre) is significantly slower for autumn spawners. The processing speed of the method can be accelerated by image analysis and training (Mosegaard & Popp-Madsen 1996). The disadvantage of a lower number of measurements is outweighed by a higher precision. Efficient grinding methods have opened up the possibility to include all ages in a routine examination. From 1996 the method using otolith micro-structure to separate Baltic spring spawners from North Sea autumn spawners has therefore increasingly been applied to the Division IIIa samples.

For the HAWG 2001 the years from 1991 to 1996 have been reworked applying common splitting keys for all years in the series, but two different methods depending on the availability information for different quarters and age groups.

For quarters one and three in the years 1991 to 1997 otolith microstructure information was available for all age groups for quarters two and four a least squares’ minimisation method fitting to observed frequency distributions of VS counts has been applied (see Section 3.2.4). For 1998 and 1999 the split presented in last years Working Group report (ICES 2000b/ACFM:10) has been adopted, and for the present year the otolith-based method has been applied (see the following Sections 3.2.1 and 3.2.2).

3.2.1 Treatment of spring spawning herring in the North Sea

The split was performed on age classes 2, 3, and 4+ WR using proportion of spring spawners f(sp) calculated from VS-counts using the equation:

f(sp) = [56.5-VS(sample)]/[56.5–55.8]

3.2.2 Treatment of autumn spawners in Division IIIa

For commercial landings in 2000 the split of the Swedish and Danish landings was conducted using an age-class stratified random subsample of herring where analysis of individual otolith microstructure determined the spawning type (Mosegaard and Popp-Madsen 1996). A total of 1024 otoliths from the year 2000 were analysed for spawning type in Div. IIIa. Samples were taken from Danish and Swedish landings constituting 51% and 49% of the analyses respectively. Data were disaggregated by area (Kattegat and Skagerrak), age group (0–4+ WR) and quarter (1–4).

The available number of individual otolith microstructure analyses by SD, quarter and age group for year 2000 is given in the following text Table:

Skagerrak Kattegat total

Age: 0 1 2 3 4+ Subtot 0 1 2 3 4+ subtot

Quarter

1 37 37 1 29 13 43 80

2 10 32 22 18 82 207 154 22 20 403 485

3 90 54 36 34 31 245 3 56 29 8 4 100 345

4 30 28 6 64 2 32 16 50 114

120 129 74 56 49 428 5 296 228 43 24 596 1024

Despite a reasonable coverage of the fishery, some of the age, area and season combinations had to be estimated as an average of the proportions in adjacent areas, age groups or years. These data are indicated in Table 3.2.1.

For the years 1991 to 1996 average values of proportions by quarter and age–class were applied. The averages were obtained by weighting by sample-size.

For the 1^st and 3^rd quarters 1991-1995 proportions of WBSS were based on otolith microstructure analyses from Swedish IBTS samples. For the 2^nd and 4^th quarters proportions were based on VS-counts. Primarily samples from commercial landings were used when available and supplemented by samples from cruises when necessary. In 1996 otolith microstructure analyses were extended with material from Danish cruises in the 3^rd quarter.

From 1997 and on otolith microstructure analyses were increasingly available from commercial landings, and from 1998 the proportions were based only on otolith microstructure analyses. Where there was a lack of information averaging over adjacent years, age-classes and quarters were used to estimate proportions.

3.2.3 Autumn spawners in the fishery in Sub-divisions 22 and 24

In the western Baltic a small percentage of the herring landings consisted of autumn spawned individuals. Compared to the 1997 years assessment (ICES 1998a/ACFM 14) the magnitude of the problem in later years appears minor. Juvenile autumn spawned herring of the age group 1 was found in landings’ samples from Sub-division 22, quarter one, year 2000 comprising about 17% in numbers. The small size at age however, indicated that the herring were local autumn spawners rather than originating from the North Sea stock. Since this problem is of limited influence and since it only affects the younger age classes (0 to 2 WR), the landings were treated as coming from the Western Baltic spring spawning stock. The existence of varying proportions of autumn spawners in sub-divisions 22–24 however, indicates a potential problem for the assessment that should be kept in mind.

3.2.4 Accuracy and precision in stock identification

Stock identification and splitting methods for the years 1991–1996

During the last decade the HAWG has encountered a suite of difficulties in the assessment of the Western Baltic Spring Spawning (WBSS) stock, as it was impossible to separate the WBSS from some North Sea stock component (autumn

analysis. Different stock components not easily distinguished by their otolith microstructure (OM) are considered to have different mean VS as e.g. winter spawning Downs herring: 56.6 (Hulme 1995), and the small local stocks, the Skagerrak winter/spring spawners: 57.0 (Rosenberg & Palmén 1981). Further, the estimated stock specific mean VS count varies somewhat among different studies; North Sea: 56.53, Western Baltic Sea: 55.60 (Gröger & Gröhsler 2000) and North Sea: 56.5, Western Baltic Sea: 55.80 (ICES 1992b/H:5).

For the years 1991 to 1999 individual otolith microstructure information from Swedish IBTS surveys in quarters 1 and 3 has been worked up, categorising individuals into three spawning groups autumn-, winter- and spring- spawners.

From these the information from 1991 to 1996 was used in the analysis. VS-counts from the same individuals were used to compare VS-count distributions with identified OM-based spawning type.

From 1640 individuals analysed from the Skagerrak and the Kattegat 1991-2000 where both VS counts and otolith microstructure analyses were available distributions of VS counts were analysed by spawning type and subdivision.

Since no information on pure stock VS counts was available for the time series in question, as a pragmatic solution we estimated each VS-count distribution using the observed counts according to individual spawning type determined from otolith microstructure, assuming a composite but among years constant stock structure of each spawning type.

In the present analysis no difference in VS-count distribution was found between winter and autumn spawners in the Skagerrak, indicating that the Downs herring may be the dominating winter spawning type in this area. In the Kattegat area, however, a different VS distribution was assigned for an apparently composite winter spawning stock.

The split between spawning types was then estimated by the sum of squares minimisation of the following function:

2 estimated. f(i) is the relative frequency of the i^th VS number, where the subscripts AS, WS, SS refer to the present attributed proportions of autumn, winter, and spring spawners, and obs refer to the observed sample distribution with unknown spawning type proportions.

From landing samples in the period 1991 – 1996 the total number of VS counts available were 12166 for the Skagerrak and 15750 for the Kattegat areas.

For the first and third quarter in the period 1991-1996 an average for each quarter, sub-division and age-class (0-4+), over all years was calculated based on individual otolith microstructure information from the Swedish IBTS samples and in 1996 supplemented with quarter three Danish acoustics samples. A total of 2005 otoliths were analysed for this period with 820 from the Skagerrak and 1185 from the Kattegat.

An effort was made to compare proportions of spring spawners from different sources of information. The estimated proportion of spring spawners from VS counts in surveys and commercial landings were plotted versus the estimated proportion spring spawners using otolith microstructure from surveys (Figures 3.2.1 and 3.2.2). The data were disaggregated by year-group, division, age group, and quarter. The year-groups were 1991-1996 and 1997, sub-divisions were Kattegat and Skagerrak, age-classes were 0,1, 2, 3, and 4+; and quarters 1 and 3. For survey versus survey comparisons the average number otolith microstructure analyses and VS counts per data point were 115 and 887 respectively, for landings versus survey comparisons were based on 118 and 698 analyses respectively. A minimum of 25 and 75 otolith microstructure analyses and VS counts respectively was applied in both comparisons to reduce random errors.

Figure 3.2.1 shows a high degree of correspondence between the two methods when applied to a reasonable number of individuals sampled from the same sources of samples.

Also when using VS counts from landings’ samples versus otolith microstructure analyses from surveys a reasonably

The applied proportions of spring spawners by sub-division, quarter, and age-class for the period 1991-1996 are given in Table 3.2.2.