Jan Horbowy
Sea Fisheries Institute,  Kołłątaja 1, 81-332 Gdynia, Poland

Key words: age-length key, comparison, Baltic, herring, sprat.

Abstract.
In this paper, the age-length keys (ALK) of herring and sprat sampled from Polish commercial and research catches in 1995 and 1996 are compared with respect to time (quarters 1-4), catch location (Subdivision 24, 25, and 26) and fishing gear type (trawl, gillnet, research). Two methods have been applied to determine similar ALKs: cluster analysis and the c2 test.
The herring ALKs from Subdivision 25 and 26 do not differ significantly. They differ, however, from the ALKs collected in Subdivision 24. The ALKs from the first half of the year differ significantly from the ALKs from the third quarter. The comparison of herring ALKs with respect to gear is inconclusive: two out of four comparisons of trawl and gillnet ALKs show significant differences. There is no significant difference in the sprat ALKs from the first three quarters. On the other hand, the ALKs from the third and fourth quarters differ significantly.  The comparison of ALKs with respect to subdivision is inconclusive: two out of four comparisons of ALKs from Subdivision 25 with ALKs from Subdivision 26 show significant differences.
The results of analyses show  which ALKs can be used interchangeably, which may be helpful in both filling in the gaps in sampling for age and planning the sampling method.
Conclusions based on cluster analysis were confirmed by statistical tests; the distance between ALKs compared with cluster analysis and the probability from statistical tests showed a negative correlation.