ICES mar. Sei. Symp., 195: 492-498. 1992 Possible interactions between salmon migrations and landings, smolt production, herring abundance, and hydrographic factors in the Gulf of Bothnia, 1976-199 Erkki Ikonen and Raimo Parmanne Ikonen, E., and Parmanne, R. 1992. Possible interactions between salmon migrations and landings, smolt production, herring abundance, and hydrographic factors in the Gulf of Bothnia, 1976-199. - ICES mar. Sei. Symp., 195: 492^198. The greater part of the natural and hatchery production of salmon smolts in the Baltic Sea occurs in the northern part of the Gulf of Bothnia, but the main feeding areas are situated in the surroundings of the Gotland and Bornholm Deeps. The post-smolt run from the home river to the Baltic Main Basin lasts from May/early June to December. ccording to catch statistics and reported tag recoveries, a varying proportion of postsmolts end their feeding migration in the southern part of the Gulf of Bothnia. One of the most important food items for salmon is the herring. The possible dependence of the salmon catch on the hydrographic factors, herring biomass, and herring year-class strength is considered. ccording to the analyses the factors best explaining the yearly variation of the salmon offshore catches in the Bothnian Sea were the herring biomass and the smolt production of the previous year. Erkki Ikonen and Raimo Parmanne: Finnish Game and Fisheries Research Institute, Fisheries Division, PO Box 22, SF-I5I Helsinki, Finland. Introduction Baltic salmon originating from the Gulf of Bothnia feed chiefly in the Baltic Main Basin. The main feeding grounds are situated in the area of the Gotland Deep and the Bornholm Deep. However, a proportion of the feeding salmon seems to stay and feed in the Bothnian Sea. This proportion varies widely, forming between 2% and 2% of the total offshore catches. Clupeids constitute the main food of salmon in the Baltic Sea. In the Baltic Main Basin sprat dominate in the salmon diet (Christensen, 1961; Thurow, 1966; Christensen and Larsson, 1979). In the northern part of the Baltic Sea, herring are more important (Christensen and Larsson, 1979; Linko etal., 1979; ndersson, 198; Salmi and Ikonen, 1982). The effect of the environmental conditions on the feeding salmon is poorly known. Reddin (1986) has shown that salmon feeding in the Northern tlantic prefer a temperature range of 4-8 C during feeding migration. The aim of this paper is to find a possible reason for the greatly varying choice of feeding ground and to determine whether the reasons are biological or hydrological. Salmon in the Bothnian Sea Salmon smolt production in the Gulf of Bothnia The salmon fisheries in the Bothnian Sea (Fig. 1) are based on the wild and hatchery smolt production in Subdivisions 3-31 (Fig. 2). In 1989, the contribution of natural production in the Gulf of Bothnia was.3 million smolts, whereas that of hatchery production was 3.7 million (non., 1991a). The greater part of the natural production came from Swedish rivers. The hatchery production was about equal on both sides of the Gulf of Bothnia (non., 198, 1984, 1987, 199, 1991a). The Swedish smolt production has been presented as the total figure for the whole Baltic Sea. The annual production in the Main Basin area has been estimated at 3 smolts, and this number has been deducted from the total figure. The most important feeding grounds of the salmon originating from the rivers emptying to the Gulf of Bothnia are situated in the Baltic Main Basin. Of the tag recoveries of hatchery-reared smolts released in 198-1989 in Subdivision 31 in Finland, 62% were obtained 492
66 2 3',/ m BOTHNIN BY ' NORTHERN QURK 63 BOTHNIN SE 6 GULF OF FINLND 47 G8 G9 HO HI H2 H3 H i H5 H6 H7 H8 H9 JO Figure 1. The Baltic Sea. from the Main Basin, 12% from the Bothnian Sea, 25% from the Bothnian Bay, and 1% from the Gulf of Finland. In Swedish tagging experiments carried out in Subdivisions 3 and 31 (non., 1991a) in the same period, the distribution was: Main Basin 85%, Bothnian Sea 11%, and Bothnian Bay 4%; recoveries from the Gulf of Finland were negligible. The Finnish salmon releases in the Bothnian Sea consisted of River Neva stock, which does not migrate so far. The proportion of recoveries made in Subdivision 3 was 87% (non., 1991a). Salmon fisheries in the Bothnian Sea Salmon are caught during the feeding migration with drift nets and drifting long lines. The salmon offshore fishery is permitted from the middle of September to the middle of June. small number of standing lines have been used near the coast in spring. During the spawning run in spring and summer, trapnets have been used almost exclusively. In 1976-199, the offshore catches varied between 66 and 544 tonnes (Fig. 3). However, the fluctuation of these landings does not show the same pattern as the offshore catches in the Main Basin or the total catches in Subdivisions 24-31 (Fig. 3). The great majority of the feeding salmon have been caught during September- December with drift nets and long lines (Fig. 4). Material Yearly data concerning salmon and herring in the Gulf of Bothnia were available for the period 1976-199. The following variables were investigated: (1) Salmon catch 493
MILLION 5 4 3 2 1 1974 1976 1978 198 1982 1984 1986 1988 199 YER ORIGIN OF SMOLT FINLND SUBDIV.3 SWEDEN SUBDIVS.3-31 EE1 FINLND SUBDIV.31 M1 TOTL Figure 2. Wild and hatchery-reared salmon smolt production in the Gulf of Bothnia. 1 T o N N E S J 1976 1978 198 1982 1984 1986 1988 199 YER RE Wm OFFSHORE SUBDIV.3 (HI! TOTL SUBDIVS.24-31 E 3 OFFSHORE MIN BSIN I i Figure 3. Salmon landings in the Baltic Sea in 1976-199. 494
TONNES 8 6 4 2 <<ÿ;x-xs» W'-:} ' c rm i ; i d ' 2 3 4 5 6 7 8 9 1 11 12 MONTH YER 11986 1987 K l 1988 ÏÜÏÏIIJ1989 Figure 4. Timing of the landings in the Bothnian Sea offshore fishery. in the offshore fishery in the Bothnian Sea (ICES Subdivision 3) (Fig. 3). (2) Salmon smolt production of the previous year in the Gulf of Bothnia (ICES Subdivisions 3 and 31) (Fig. 2). (3) Herring year-class strength in the Gulf of Bothnia (non., 1991b). (4) Number of one-year-old herring in the Gulf of Bothnia (non., 1991b). (5) Herring stock biomass in the Gulf of Bothnia (non., 1991b). In addition, the severity of the winter was studied on the basis of: (6) The number of real ice days in the northern part of Subdivision 3 at Norrskär (Kalliosaari, 1978, 1982; Kalliosaari and Seinä, 1987; Kalliosaari, pers. comm.). Yearly water temperature records for 1974-1989 from the various depths were obtained from the Finnish Institute of Marine Research as follows: (1) Mean monthly water temperature in July in the northernmost part of Subdivision 3, at Valassaaret, from depths of m, 5 m, 1 m, 15 m, and 2 m. (2) Mean monthly water temperature in October in the northernmost part of Subdivision 3, at Valassaaret, from depths of m, 5 m, 1 m, 15 m, and 2 m. Methods The possible dependence of salmon offshore catches in the Bothnian Sea (ICES Subdivision 3) on the herring abundance and hydrographical factors was investigated by calculating the Pearson product-moment correlations (SS Institute Inc. 1985). The dependence of the salmon offshore catch in the Bothnian Sea on the smolt production of the previous year, Baltic herring abundance, severity of the winter, water temperature in October of the actual year and in July at the previous year was investigated by multiple regression analysis (SS Institute Inc. 1987). In the procedure the linear regression model is fitted by the least squares method. The ability of models including various independent variables to predict the salmon catch was studied with the coefficient of determination (R2)- s the regressor variables in the linear regression model seem to correlate, a principal component analysis (SS Institute Inc. 1987) was made in order to obtain uncorrelated independent variables for the model. The principal components were computed from the correlation matrix. The dependence of the salmon catch on the two first principal components was studied with the regression analysis. Results The salmon catches have been large in low water temperatures in October (Table 1). There is a positive correlation between the salmon offshore catch in the Bothnian Sea and the water temperature in July of the previous year (Table 1). The null hypothesis, that the data are samples from a normal distribution, cannot be rejected (Table 2). The absolute correlations between the salmon off- 495
Table 1. Correlation of offshore catches of salmon in the Bothnian Sea with monthly mean water temperature 1974-1989. Predicted salmon catch in tonnes ctual year Previous year 6 Depth (m) July October July October -.19 -.66.9.11 5 -.16 -.69.31.7 1 -.12 -.66.42.6 15 -.26 -.69.32.3 2 -.4 -.67.7.6 4 shore catch in the Bothnian Sea and the regressor variables are highest with the herring biomass, mean water temperature in October, number of real ice days, and number of group 1 herring (Table 3). The correlation with the actual herring year class is small and the variable was excluded from further analyses. ccording to the coefficients of determination of the linear regression models with varying independent variables, the proportion of the salmon catch variation explained by the herring biomass and the smolt production of the previous year is 89% (Table 4). If the number of independent variables in the model is increased from 2 to 6, the coefficient of determination increases by 3%. Thus, in the material analysed, the herring biomass and smolt production of the previous Table 2. Means and standard deviations of the salmon, herring, and hydrographic variables in 1976-199 and the probability that the characters are samples from a normal distribution (n = 15). Normal Mean s.d. probability Salmon catch (t) 224.3 145.3.53 (thousands) 322 859.375 Herring year class 3972 2543.143 Group 1 herring (millions) 414 2572.153 Herring biomass (1 t) 423.9 16.1.61 13.7 42.6.428 October 5 m, C 7.6 1.1.548 9.76 1.9.9 Salmon catch in tonnes Figure 5. Predicted salmon catch in the offshore fishery in the Bothnian Sea (ICES Subdivision 3) according to the linear regression model based on herring biomass and smolt production. year seem to be sufficient to explain the greater part of the yearly variation of the salmon catch in offshore fishing in the Bothnian Sea (Fig. 5). The equation of the regression model is: Salmon catch (t) = 62.7 + 1.59 * herring biomass (t).17 * smolt production (in thousands). The model suggests that a large herring stock biomass Table 3. Correlations between salmon, Baltic herring abundance, and hydrographic factors in 1976-199 (n = 15). Salmon catch Smolt production Herring year class Group 1 herring Herring biomass Number of ice days October 5 m, C Previous July 1 m, C Salmon catch Herring year class 1-group herring Herring biomass October 5 m, C -.27 -.84.425.53 -.448 -.499.58.351.633 -.35.49.211.288 -.48.217.63 -.612 -.61 -.629 -.21 -.4 Previous July 1 m, C.418 -.87 -.58.166.157 -.263.14 496
Table 4. The biggest coefficients of determination (R2) and the error sum of squares of the regression models between the salmon offshore catch in the Bothnian Sea and various regressor variables in 1976-199. Predicted salmon catch In tonnes I '45 + No. of regressor variables in the model Variables R2 Error sum of squares 4 + 35 + 1 Herring biomass.281 21263 2 Herring biomass 3 Herring biomass 4 Herring biomass Group 1 herring 5 Herring biomass Group 1 herring 6 Herring biomass Group 1 herring October 5 m, C.889 32657.913 25 729.919 23795.92 23564.92 23 516 in the actual year and small smolt production in the previous year predict large offshore salmon catches. Under such circumstances a maximum amount of herring is available per salmon, and a large proportion of salmon may stay in the Bothnian Sea, instead of extending the feeding migration to the Baltic Main Basin. Principal components were computed from the smolt production, herring abundance, and hydrographical variables (Table 5). The first component accounts for Table 5. The eigenvectors and the cumulative variance of the first three principal components based on smolt production, herring abundance, and hydrographic factors in the Bothnian Sea in 1976-199. Principal components 1 2 3.435.514 -.87 Group 1 herring.482 -.237 -.159 Herring biomass.539 -.56.181 -.5.24 -.52 October 5m, C.117.71.465.155 -.377.849 Eigenvalue 2.672 1.362.968 Cumulative variance.445.672.834 3 + 25 + 2 + 15 + 1 + 5 + + L_ + 1 2 3 4 5 Salmon catch in tonnes Figure 6. Predicted salmon catch in the offshore fishery in the Bothnian Sea (ICES Subdivision 3) according to the linear regression model based on the first two principal components of the smolt production, herring abundance, and hydrographical variables. 45% of the variance and the first two components explain 67%. ccording to the eigenvectors (Table 5), the second principal component is mainly a measure of sea temperature and smolt production and the first component a measure of other characters. ccording to the coefficients of determination of the linear regression model, the first two principal components explain 69% of the yearly salmon catch variation in offshore fishing in the Bothnian Sea (Fig. 6). Discussion The regressor variables used in the models obviously do not predict the salmon catches, but the amount of salmon in the Bothnian Sea. The salmon catch is used in the model because yearly data on the amount of salmon in the sea are lacking. C.p.u.e. data are available for the salmon drift net and long line records made in the Bothnian Sea in the period 198-1989. Their correlations (r) with the offshore salmon catch in the Bothnian sea are.74 and.53, respectively. The hypothesis that the offshore salmon catch in the Bothnian Sea is connected with the feeding conditions is 497
supported by a positive correlation (r =.67) between the catch and the mean weight of salmon age group.1+ in the Bothnian Sea in October-December 1976-199. In 1991 the herring stock biomass in the Gulf of Bothnia is 72 tonnes (non., 1991b) and the smolt production of the year 199 was 3 29 (Fig. 2). ccording to the first model presented, a salmon catch of 65 tonnes can be predicted in the offshore fishery in the Bothnian Sea in 1991. The predicted catch is higher than earlier, which is in accordance with the data obtained from the fishery, where catches have been unusually high in the Bothnian Sea in January-May 1991. s presented earlier the abundance of the herring population and number of group 1 herring seem to be the main variables required to predict the salmon catches in the Bothnian Sea. During the post-smolt run, this herring year class is at the age of +. s herring spawn in coastal waters, the migrating post-smolts and + herring occur in the same area. possible indication of a large number of salmon feeding in the Bothnian Sea may be an abundance herring year class. In these circumstances a larger proportion of salmon post-smolts than average prefer to stay in the Bothnian Sea, utilizing herring which were born in the same spring when the smolts entered the sea, and whose size therefore remains suitable as the food of the growing post-smolts. If the herring biomass is small, the feeding conditions are poor and the post-smolts prefer to migrate to the Baltic Main Basin, where food is always available. The correlation between low smolt production and good landings is difficult to explain. possible reason may be competition for food. However, as the annual changes in the number of post-smolts are quite small compared with those in the herring biomass, the post-smolts must have a negligible effect on the herring biomass. More data are needed in order to clarify the interactions between salmon, herring, and hydrographical factors. References ndersson, E. 198. Merilohen ravinnosta syyspyyntikauden aikana itäisellä Suomenlahdella. Riista- ja kalatalouden tutkimuslaitos. kalantutkimusosasto. Tiedonantoja, 15: 4-47. non. 198. Report of the Baltic Salmon ssessment Working Group. ICES CM 198/M: 3. non. 1984. Report of the Baltic Salmon and Trout ssessment Working Group. ICES CM 1984/ssess: 11. non. 1987. Report of the Baltic Salmon and Trout ssessment Working Group. ICES CM 1987/ssess: 21. non. 199. Report of the Baltic Salmon and Trout ssessment Working Group. ICES CM 199/ssess: 21. non. 1991a. Report of the Baltic Salmon and Trout ssessment Working Group. ICES CM 1991/ssess 13. non. 1991b. Report of the Working Group on ssessment of Pelagic Stocks in the Baltic. Copenhagen, 15-25 pril. ICES CM 1991/ssess 18. Christensen. O. 1961. Preliminary results of an investigation on the food of the Baltic Salmon. ICES CM 1961/68. Christensen. P., and Larsson, P-O. 1979. Review of Baltic Salmon Research. Coop. Res. Rep. Cons. Int. Explor. Mer, 89: 124 pp. Kalliosaari, S. 1978. Ice winters 1971-1975 along the Finnish coast. Finn. Mar. Res., 245: 63 pp. Kalliosaari, S. 1982. Ice winters 1976-198 along the Finnish coast. Finn. Mar. Res., 249: 3-61. Kalliosaari, S. 1987. Ice winters 1981-1985 along the Finnish coast. Finn. Mar. Res., 254: 3-63. Linko, R., Lemmetyinen, R., and Rautamäki, P. 1979. Saaristomeren ravintoketjun myrkkyjäämätutkimus. 117 pp. Turun yliopisto. Reddin, D. G. 1988. Ocean life of tlantic salmon in the Northwest tlantic. In tlantic salmon: planning for the future. Ed. by D. Mills and D. Piggins. Portland, US 1988: 483-511. Salmi, J., and Ikonen, E. 1982. Food and feeding of salmon in the Bothnian Sea. ICES CM 1982/M: 41. SS Institute Inc. 1985. Procedures Guide for Personal Computers, Version 6 Edition. Cary, NC: SS Institute Inc., 373 pp. SS Institute Inc. 1987. SS/STT Guide for Personal Computers, Version 6 Edition. Cary, NC: SS Institute Inc., 128 pp. Thurow, F. 1966. Beiträge zur Biologie und Bestandskunde des tlantisches Lachses in der Ostsee. Berichte der Deutschen Wissenschaftlicher Komission für Meeresforschung 18: 3/4, 223-379. 498