Results Directory UMM :Data Elmu:jurnal:A:Aquaculture:Vol184.Issue3-4.Apr2000:

Specific Growth Rate in length between measurement periods was calculated as: SGR s 100 = ln FL y ln FL r t y t Ž . Ž . Ž . t 2 t1 2 1 Ž . where t s first sampling day; t s second sampling day and FL s fork length mm . 1 2

3. Results

No fin damage of any kind was visible until the end of May, when it was evident in small numbers of fish. There was a rapid rise in the incidence of damage between June Ž . and July Fig. 2 . From July onwards, splitting was the most common type of fin Ž . damage Fig. 2a . It then declined during the photoperiod winter, but increased to previous levels by December. Ž . Thickening and tissue loss were always less prevalent than splitting Fig. 2b–c . Neither category of damage was particularly severe. Thickening category 2 was never recorded, while only two fish ever had less than 30 of the dorsal fin remaining, and Ž . most of those affected 130 out of 139 had 60 to 90 remaining. There were some Ž . differences between groups in the frequency of fin damage Table 1 , but all four groups Ž . showed similar patterns of development of fin damage Fig. 2 . Since fin splitting was the most common category of damage, and it is the best indicator of current levels of aggression, all subsequent analyses are based only on this measurement. Ž . By November, 2.2 7 fish were very small parr that had failed to smolt, while Ž . 4.1 13 fish were sexually mature male parr. As the growth patterns of both of these Ž . categories of fish differed from the majority of the fish which were immature smolts they have been excluded from the following analysis, except where indicated. Ž . Table 2 gives the mean fork length SE of fish with and without split fins in July, September and December when fin splitting was evident in large numbers of fish in all groups. Experimental group had a significant effect on fork length at all times due to the effect of the temperature manipulation, while fish that had split fins were, on average, Ž larger than fish without split fins in July and September, but not in December Two-way ANOVA with group as a factor in: July, F s 46.4, P - 0.01; September, F 3, 286 3, Ž s 37.5, P - 0.01; December, F s 34.8, P - 0.01; with fin condition split or not 286 3, 286 . split as a factor in: July, F s 81.8, P - 0.01; September, F s 24.9, P - 0.01; 1, 286 1, 286 . December, F s 1.6, n.s. . In December, there was a significant interaction between 1, 286 group and fin condition, as the fish in group C that had split fins were still larger than those without split fins, but there was no longer any such relationship within the other Ž groups Interaction between group and fin condition in: July, F s 2.5, n.s.; 3, 286 . September, F s 1.1, n.s.; December, F s 3.1, P - 0.05 . 3, 286 3, 286 The relationship between fin damage and length within each group of fish in July was very strong, as shown by logistic regressions relating fin splitting to fish length for each Ž . group Fig. 3a–d . In all groups, there was a strong and significant positive relationship between the probability of having split fins and the size of the fish, with the probability Ž . calculated from the logistic regression lines rising from less than 0.15 for the smallest Fig. 2. Changes over time in the percentage of fish affected by three measures of dorsal fin damage in four Ž . Ž . Ž . groups of juvenile Atlantic salmon: a splitting, b thickening and c tissue loss. fish in each tank, to over 0.9 for the largest fish. The effect was related to the size range within the tank, not the absolute size of the fish: for instance, an 85-mm fish would be one of the largest fish in Group B or C, and would have a 0.86 or a 0.88 probability of having split fins, respectively. A fish of the same size in Group A or D would be in the Table 1 Comparisons by x 2 test of frequencies of three categories of fin damage between groups of Atlantic salmon of different mean length on five sampling dates. x 2 values were regarded as invalid and are omitted if cross-tabulation yielded expected frequencies of less than five in one or more cells. df s 3 in all cases. 2 Category Sampling date n x Splitting June 600 0.69 July 314 7.70 U September 314 10.73 UU November 314 25.84 UU December 314 15.54 UU Thickening July 314 18.19 UU November 314 31.74 December 314 2.41 U Tissue loss June 600 7.99 U July 314 15.25 September 314 2.58 UU November 314 37.96 UU December 314 28.38 U Level of statistical significance: P - 0.05. UU Level of statistical significance: P - 0.01. middle of the size range and would be considerably less likely to have split fins Ž . probabilities of 0.62 and 0.61, respectively . To adjust for the differences in size among the four groups, fish size is expressed as Ž . the deviation from the mean length of the group as a proportion of the mean Fig. 3e . Table 2 Ž . Ž . Mean fork length SE of fish with and without split fins in four groups A to D of Atlantic salmon in July, September and December, 1997 Ž . Month Group n Mean fork length mm SE of fish with fins: intact split July A 76 771.3 831.2 B 53 701.8 801.5 C 82 701.1 780.9 D 83 822.0 951.2 Sept. A 76 1162.2 1221.5 B 53 983.4 1112.0 C 82 1132.1 1181.7 D 83 1263.1 1331.4 Dec. A 76 1592.9 1561.5 B 53 1465.8 1462.0 C 52 1472.4 1571.7 D 83 1682.1 1691.8 Fig. 3. The effect of body size on the probability of having split fins in juvenile Atlantic salmon in July 1997 Ž . Ž . Ž . Ž . Ž . in a Group A, b Group B, c Group C, d Group D, and e all four groups combined. The effect of Ž . deviation from mean fork length for fish in a group see text for explanation on the probability of having split Ž . Ž . fins is shown in e ; symbols squares are the actual proportion of fish that had split fins, within each 0.05 Ž . range. The logistic regression lines have been calculated using only immature smolts ns 294 , but symbols Ž . Ž . representing non-smolting parr ns 7; open triangle and mature male parr ns13; open circle have been Ž . Ž . added to e for comparison. Lines are the logistic regression line given by the equation: Y s ar 1q a , where wq z X Ž Ž . Ž .. Ž Ž .. ase , and X s fork length in a to d or deviation from mean fork length in e . The values of w Ž . Ž . Ž . and z for each graph are as follows: a w sy7.9442, z s 0.0994, b w sy9.4217, z s 0.1322, c Ž . Ž . 2 w sy12.5787, z s 0.1720, d w sy11.0913, z s 0.1357, e w s 0.4002, z s10.4572. The values of x Ž . Ž . Ž . Ž . Ž . were as follows: a 10.56, b 14.04, c 22.39, d 25.59, e 71.44, and P - 0.025 in all cases. Parr that would not smolt were as likely as other fish of their relative size to have fin Ž 2 . damage x s 0.002, 1 df, n s 29, n.s. , while mature male parr were somewhat more Ž 2 likely to have fin damage than immature fish in the same relative size range x s 4.42, . 1 df, n s 122, P - 0.05 .

4. Discussion