International Council for the
Exploration of the Sea
•
C.M. 1991/M:28
Ref.J
Anadromous and
Catadromous
Fish Cornrnlttee.
VARIATION OF GROWTH RATE, TAG-RECOVERY RATE AND TEMPORAL
DISTRIBUTION OF TAG-RECOVERIES IN BALTIC SALMON TAGGING
EXPERIMENTS
by
Matti Salminen
Finnish Garne and Fisheries Research Institute
Fisheries Division
P.O.BOX 202, 00151 Helsinki Finland
Abstract
The pattern of variation of three basic variables acquired from
Baltic salmon tagging experiments - length increment (growth
rate), return rate (survival) and temporal distribution of tag
recoveries - was analysed. Year of release (year-class) and
length at release (smolt size) were used as independent
variables, i.e. sources of variation. The data eonsisted of 352
groups (296613 ind.) of reared smolts released in the northern
part of the Gulf of Bothnia (ICES 31) in 1970-1988.
•
Year-class related factors - probablY including changes in
marine environment as weIl as in smolt quality - aeeount for a
major part of variation in tagging data. After a eorrection to
smolt size, "year-class" explained 11% of the remaining
variation of individual" length increment (until A1+ stage), and
54% and 65% of the variation of adult return rate and temporal
distribution (Al+pereentage) of tag reeoveries for eaeh smolt
group, respectively.
A significant positive eorrelation was found between mean length
increment and adult return rate for eaeh smolt year-elass,
indicating that same faetors, that regulate growth, affect
survival as weIl - probably via growth rate. Furthermore; the
.positive eorrelation between growth rate and A1+ pereentage
indicates, that the sooner the fish attain legal size and fully
recruit to the fishery, the sooner they are eaught.
The importanee of smolt size is clearly demonstrated by the
significant positive regressions of mean size at reeapture,
return rate and A1+ pereentage on smolt length. Length at
recapture (A1+) inereases by 3.3 cm, A1+ pereentage by 13 and
return pereentage by 3.2 for 5 cm increased smolt length (from
15 to 20 cm). During the study period there was a sligth
positive trend in the mean length of A1+ salmon, probablY owing
to simultaneous rise in smolt size.
The role of gear seleetivity, maturity and tag loss as possible
soure es of biase in the tagging data are diseussed.
2
1. Introduction
Data from tagging experiments (Carlin-tag) are widely used in
the assessment and management of Baltic Salmon stocks. Estimates
of growth-rate, post-smolt mortality, adult mortality and the
distribution of salmon catches by fishing seasons and fishing
areas, e.g., are often based on tagging data. Surprisingly
little attention has, however, been paid to the pattern of
variation of, and - especially - the connection between the
parameters used as the bases of these estimates. The stochastic
(year-to-year) variation of post-smolt survival (LINDROTH 1965),
growth rate (CARLIN 1969, THUROW 1973) and many indices of catch
data, is weIl known, but the correlation between these
variables, as weIl as the consequences of the variation itself,
•
have mainly remained obscure.
I analysed the pattern of variation of three basic variables
acquired from tagging experiments -
length increment ("growth-
rate"), recapture-rate ("survival") and the distribution of tagrecoveries between fishing seasons - and the correlation between
these three. Year of release (year-class) and length at release
(smolt size) were used as independent variables, i.e. sources of.·
variation. Data from Finnish tagging experiments in the northern
part of Gulf of Bothnia in 1970-1988 were used in the analyses.
2. Material and methods
2.1. Tagging data
In all, 352 groups (296613 individuals) of tagged reared salmon
smolts were released in the northern part of the Gulf of Bothnia
(ICES 31) in 1970-1988 (Table I). All groups belonged to salmon
stocks exhibiting a similar pattern of long feeding migration
(CARLIN 1969, LUNDIN 1974, IKONEN and AUVINEN 1984) to the
Baltic Main Basin (ICES 22-29).
Until May 1991, a total of 24903 tag-recoveries (8,4 %) were
reported from these experiments. I devided the recoveries
according to the age-group of the fish and fishing season as
foliows:
•
3
Time of
recovery
Age-groups
Year of release 2nd year April
A+,
2nd year May 3rd year April
Al+, A2
Al+
3rd year May 4th year April
A2+, A3
A2+
Al
Later referred
to as
Post-smolts
and so on ..
4t
•
For the growth analysis, all A1+-recoveries (defined above)
reported from the Main Basin (leES 22-29) between November and
April were extracted from the data. During this period the
growth is assumed to be minimal (LARSSON 1973) and the fish are
assumed to be in same state of sexual maturation. The length
increment of these fish after release (=length at recapture individual length at release) was used as a variable describing
"growth-rate". Length at recapture was reported for 2675
recoveries. The freguency distribution of length increment is
given in Fig 1.
The return-rate of adult salmon (Al+ and older) for each group
(352) of tagged smolts was used as a variable describing
"survival". The freguency distribution of return-rate Is given
in Fig 1.
The percentage of Al+ -fish of total adult (A1+ and older)
recoveries for each group of tagged smolts was used as a
variable describing "temporal distribution" of tag recoveries.
Groups with less than 10 adult recoveries (46) were rejected
from the data. The frequency distribution of A1+ -percentage is
given in Fig 1.
2.2. Statistical methods
Differences in length increment, return-rate and Al+ -percentage
were tested by ANOVA for each year-class and between yearclasses. Length at release was used as a covariate in the
analyses. The frequency distributions of return-rate and A1+ percentage were normalized by arcsinV-transformation.{Fig 1).
4
Before the application of ANOVA, Bartlett's test of homogenity
was used to test for possible differences in the variances for
each year-class.
The relationship between average length increment, return-rate
(transformated) and A1+ -percentage (transformated) for each
year-class was tested by correlation analysis. The dependancy
was assumed to be linear (Pearson correlation coefficient).
Regression analysis (linear regression) was used to test the
dependency of length-increment (and length at recapture) on
individual length at release, and the dependency of return-rate
and A1+ -percentage on the mean length at release for each
smolt-group. - Same regressions were already included in ANOVA
(above), where length at release was used as a covariate.
3. Results
3.1. Between year-class component of variation
Between year-class variation accounts for a substantial part of
the total variation (counted from the regression line) of
return-rate and A1+ -percentage (54 and 65%, respectively). In
the variation of length increment the between-year component is
not as important (11 %), though statistically significant (Table
II).
3.2. Growth-rate, return-rate and A1+percentage
The correlation between average length increment, return-rate
and A1+percentage for each year-class~is significant (Table
III). Year-classes with good growth (or high mean length at
A1+stage) give good returns and high percentages of A1+ fish
(Fig 2).
Highest mean values of length increment, 54-56 cm, were obtained
in year-classes 1972-1973 and again 1983-1984 and 1988. The
A1+percentage for these year-classes varied between 62 and 75,
and return-rate between 8 and 24. Growth-rate was, on the other
hand, poorest in year-classes 1977-1981 (length increment 47-50
tt
5
cm), glvlng only 31-48 % A1+-fish of all adult recoveries
(return-rate 3-7%). Year-class 1987 was an exception: low length
increment (48 cm) resulted, as expected, in a low return-rate
(4.1%), but also relatively high Al+ percentage (66.6%).
3.3. The effect of smolt size
The regression of length increment (and length at recapture),
return-rate and A1+ -percentage on smolt size (length at
release)
is highly significant (Table 11, Figures 3 and 4),
accounting for 1.6 % (5.4%), 7.3 % and 9.4 % of the total
variation, respectively. Length at recapture increases by 3.3
cm, Al+ percentage by 13 and return percentage by 3.2 for
5 cm increased length (from 15 to 20 cm).
The regression of length increment on smolt size is negative,
while the regression of length at recapture on smolt size is
positive (Fig. 3, Table IV). The pattern is about the same also
for separate year-classes (Table IV), but the the values of
regression coefficients (slope) and probability seem to vary
from year to year.
For year-classes with average or good growth (1984,1985, 1986
and 1988) the regression of length at recapture on smolt size is
highly significant (Table 111) and the regression coefficient
tt
aproaches 1.0, while the regression of length increment on smolt
size is not significant. For year-class 1987 with poor growth,
on the other hand, the (negative) regression of length increment
on smolt size is highly significant, while the regression' of
length at recapture on smolt size is not as clear.
During the study period there was a slight positive trend in the
mean length of A1+ salmon (Fig 2).
4. Discussion
4.1. Evaluation of the growth data
Growth estimates based on tagging data may be biased for several
reasons. Fishermen, e.g., don't often have time for accurate
length or weigth measurements. This becomes evident from the
6
fact, that the reported lenghts and weights don't always "fit"
together. Active size selection is another source of biase:
undersized fish are released. or. if taken. not always reported.
At Al+ stage 60 cm size limit cuts, at least in year-classes
with poor growth. a substantial part of the lower edge of the
length distribution (Fig 3, Fig 5). Third, Carlin-tag is known
to have an adverse effect on both length (SAUNDERS and ALLEN
1967) and weight increment (ISAKSSON and BERGMAN 1978).
Growth data may be biased by maturity and ge ar selection, too
(THUROW 1973). At Al+ stage (open sea, November-April) all fish
were assumed to be in the same stage of sexual maturation. This
migth be true, but the grilse, known to be among the smallest
individuals bY November (SAUNDERS et.al. 1983) have already left·
•
for spawning grounds, which may lead to an overestimate of
length.
The spring emigration of spawners of the year, starting
in Mars-April, may, on the other hand, lead to an underestimate
(LARSSON 1973).
Male salmon grow faster (CHRISTENSEN and LARSSON 1979) and
attain maturity earlier than female. Almost all grilse are known
to be males. If the proportion of grilse varies from year to
year (see e.g. LARSSON and SVENSSON 1974), so does the sex-ratio
of A1+ fish caught offshore. This emphasizes the importance of
separating the sexes in growth analysis. Unfortunately. the data
. doesn't allow of sex separation. Besides precocious males, the
salmon can not be sexed as smolts, and there is not much data on
sex in the recovery files either.
At present, some 80 % of the totaloffshore catch of salmon is
taken by 160 mm drift nets (Report of ... 1991). which are highly
selectiye" gear, and 20 % by long lines, which are somewhat less
selective (CHRISTENSEN and LARSSON 1979) The ratio is, however,
by no means constant. Besides "normal" alteration. some major
changes have taken place in the salmon fishery during the study
period (e.g. 1978). and. probably in the composition and
selectivity of off-shore fishery, too.
Because of the scarcity
of the data. all recoveries regardless to the gear were put
together in this study. This probably adds to the variation of
length data. but makes it less size selected as weIl.
•
7
At A1+ stage, drift nets select for the largest and fastest
growing individuals (CHRISTENSEN and LARSSON 1979, see also Fig
6.). The intensity of selection depends on the size and, hence,
on the growth of the fish belonging to a given year-class. Thus,
gear selection (passive), as weIl as the active selection by the
fishermen (size limit), tends to "dampen" the real size
variation present in the population. - Besides, the largest
individuals are probably caugth already in September-October,
when the growth still continues.
As a matter of fact, drift nets don't select for fish length,
but rat her for fish girth. There is some evidence, that the
condition factor of salmon varies from year to year. KARLSSON
and ERIKSSON (1990), e.g., reported a condition factor 1.24 for
the salmon caught offshore in the autumn 1990, which is an
exceptionally high figure if compared with a long-term (19771990) average (1.10) during september and October. This
variation is another source of biase in the growth data~
irrespective of whether it is correlated with the length (or
growth history) of the fish or not.
•
Size selection of drift nets may change within a winter season
as weIl. Length increment between November and April- is supposed
to be minimal (e.g. THUROW 1966, 1973), but CHRISTENSEN (1961)
has reported a considerable decrease in the condition factor of
salmon during the winter. As a concequence, growth data may be
biased by between year chances in the timing of winter fishery.
4.2. Variation of growth rate
Fluctuations in the age-spesific size of salmon have drawn much
attention. LINDROTH (1964 and 1965) demonstrated a long lasting
size decrease starting from the 1939 year-class. THUROW (1973)
found a slight negative trend in the length of Al+ salmon in
1957-1972. According to Swedish tagging experiments in 19531972, there was a positive trend in mean weight of older salmon
from year-class 1953 to year-class 1959, a negative trend for
year-classes 1959-1964, and, again, a positive trendstarting
from year-class 1964. (LARSSON 1973).
8
For year-classes 1970-1988 a slight upward trend in the mean
length of A1+ salmon was noticed (Fig 1.). However, no such
trend can be found in the mean length increment, only ups and
downs. This suggests, that the positive trend in mean length of
A1+ salmon is mainlY caused by the rise in smolt size (Table I).
The negative regression of length increment on smolt size (Fig
3, Table 4) in the whole data suggests better growth rate for
small than for large smolts. The same analysis for separate
year-classes (1984-1987) reveals, howewer, that the regression
is clearly pronounced only for year-class 1987 with poor growth
rate, indicating that the regression is an artifact caused by
intensive size selection (Table IV).
•
Here, again, we have to remember the connection between growth
(and size) and maturity and, on the other hand, between maturity
and migration as a possible source of biase. The proportion of
both precocious males and grilse may vary from year-class to
year-class (LARSSON and SVENSSON 1974). Precocious males are
known to be among the smallest smolts (CARLIN and OTTOSSON
1967), and there is evidence, that the same holds true for
grilse, too (RITTER 1972, LARSSON and SVENSSON 1974 and Fig 6).
CARLIN (1969) explained the 1962-1966 decrease in the mean size
of older sa1mon by intensified offshore fishery, rejecting the
possibility of growth variation. More recent study, suggests,
however, a major environmental component in determination of
growth rate and, accordingly, the age-spesific size of sa)mon.
KUIKKA (1991), e.g., found a positive correlation between the
growth rate of salmon and the size of sprat stock. The problem
of this kind of approach is, that we don't know, whether the
size differences arise already in the post-smolt phase during
the first summer, or later. Unfortunately, the scarce post-smolt
data doesn't allow of a thorough growth analysis.
If post-smolt phase is critical, one might expect significant
differences between the growth rate of southern and northern
stocks of salmon. Before taking any regulatory measures, which
would affect gear selectivity, e.g. changing the minimum mesh
..
9
size of drift nets, a comparative growth analysis is of vital
importance.
4.3. Growth rate and temporal distribution of tag-recoveries
The positive correlation between growth rate (length increment)
and A1+ percentage (Table 111), as weIl as the positive
regression of A1+ percentage on mean length of each smolt group
(Fig 4), is easy to explain: the sooner the fish attain legal
size and recruit to the fishery, the sooner they are caught.
•
•
There are, however, some signs that the connection between mean
length increment and A1+ percentage is breaking down. Yearclasses 1986 and 1987, despite of their average or poor growth,
gave exceptionally high percentages of A1+fish (Fig 2.).
This might be explained by large smolt size and/or high
condition factor of A1+ fish.
So far, there is no information on the connection between
temporal and spatial distribution of tag recoveries. One might
assurne, that the sooner the fish are caught, the smaller is the
number of spawning migrants, and vice versa. The connection
between growth and maturity makes the picture, howewer, more
complicated. Fish belonging to a year-class with good growth
rate probably mature and leave for spawning grouds earlier than
fish belonging to a year-class with poor growth rate.
4.4. Growth rate and tag return rate
Positive correlation between growth rate and tag return rate
(Table 111) indicates,
same factors, that regulate growth,
...
. that
..
affect survival as weIl. In addition, the correlation between
growth rate and temporal distribution of tag· recoveries
suggests, that there is one more factor to be taken into account
in this context, i.e. the effect of tag loss.
ISAKSSON and BERGMAN (1978), studYing the salmon releases in
Iceland, reported a 10% tag loss for smolt year-classes 1974 and
1975 between the time of release and time of recovery. The data
collected from the Baltic main basin by an leES programme
~
10
(Report of ... 1991) indicated a 20 % tag loss for the season
1988/89 (smolt year-classes 1986-1988), 10 % tag loss for the
season 1989/90 and 30% tag loss for the season 1990/91.
The problem arlslng from this kind of approach is, that it
doesn't separate between age groups. One might assurne that tag
loss is not instantaneous, but rather goes on during the whole
life span of tagged fish, perhaps with varying speed. As a
consequence, the incidence of tag loss migth depend on the
length of the sea phase be fore recapture, which in turn would
lead, e.g., to an underestimate of the survival of year-classes
with poor growth and to a biased estimate of the distribution of
salmon catches between fishing areas.
•
Indeed, a small data collected from the mouths of rivers
Kokemäenjoki and Karvianjoki (trap net -fishery) suggests, that
tag loss is a major source of biase in tagging data (Table V).
In age group Al+ the number of tagged fish was even higher than
expected, in age group A2+ 50 % and in age group A3+ only 25% of
the expected number. Besides tag loss, these numbers may, of
coarse, indicate, that the tag makes the fish more vulnerable
for (trap net) fishery at A1+ phase. It'is also possible, that
the tagged fish have been larger than the untagged fish - which
is not unusual in tagging experiments. In any case, the
•
phenomenon should be taken seriously, and all possible data
concerning this issue should be collected and analysed.
4.5.
"Year class -effect"
Year class related factors explain a substantial part of.the
variation in tagging data. But what are these factors ? It is
obvious, that the fluctuation of growth rate is mainly caused by
changes in marine environment, e.g. in temperature regime and
food resources (KUIKKA 1991). The connection between environment
and return rate is somewhat more complex. Same factars obviously
affect return rate via growth rate - as suggested by the
correlation between length increment, return rate and Ai+
percentage - some others probably directly. The latter include,
e.g., the size of predatory fish stocks (LARSSON 1977) and -
in
the broader sense of the word "environment" - changes in salmon
11
fishery and tag reporting activity, too.
Changes in smolt quality - and in releasing practice - may add
to the year-to-year variation of tagging data as weIl. One
might, however, assurne, that if compared with "environmental"
chances, these factors are of minor importance - perhaps with
the exception of smolt size. On the other hand, differences in
smolt quality and releasing methods, etc., probably explain a
major part of within year-class variation of tag return rate. In
this respect tagged fish are probably more heterogenous than
untagged fish belonging to same year-class .
•
•
4.6. Smolt age and size
Smolt size is connected with smolt age. In these data the mean
length of different smolt groups varied between 132 and 170 mm
(mean 155, n=30) for 1+ smolts, between 141-240 (mean 176,
n=217) for 2+ smolts and between.·143-261 (mean ·196, n=105) for
3+ smolts. The regressions of mean size at recapture, Al+
percentage and return rate on smolt size indicate, that the
consequences of a large scale shift to younger and smaller
smolts, perhaps with a simultaneous change in the minimum mesh
size of drift nets from 160 to 180 mm, would be far reaching.
During periods of poor growth, practically no Al+ salmon of Gulf
of Bothnian origin would be caught from the Main Basin .
References
Carlin, B. 1969: Salmon tagging experiments. In: Arsberättelse
för Ar 1968. Laxforskningsinst. Medd., 3.
Carlin, B. and ottosson, Y. 1967: Laxungarnas tillväxt under
andra sommaren. Laxforskningsinst. Medd. 16.
Christensen, o. 1961: Preliminary results of an investigation on
the food of Baltic salmon. ICES, Doc. C.M.1961/No.93.
Christensen, o. and Larsson, P-O. (ed.) 1979: Review of Baltic
Salmon Research. A synopsis compiled by the Baltic Salmon
Working Group. lCES Coop. Res. Rep. 89. 124 pp.
Ikonen, E. and Auvinen, H. 1984: Migration of salmon in the
Baltic Sea, based on Finnish tagging experiments. lCES C.M.
1984/M:4. Anadromous and Catadromous Fish Committee.
~-----~--------------
---
----
-
12
!saksson, A. and Bergman, P. K. 1978: An evaluation of two
tagging methods and survival rates of different age and
treatment groups of hatchery-reared Atlantic salmon smolts.
J. Agr. Res. lceland 10(2):74-79.
Karlsson, L. arid Eriksson, C. 1991: Experimental fishery With
salmon drift nets of different mesh sizes in the Baltic in the
autumn 1990. Report of Baltic Salmon and Trout Assessment
Working Group. C.M. 1991/Assess:13. Appendix 1.
Kuikka, S. 1991: Effects of some external factors on the
predictability and production capacity of Baltic salmon stocks.
lCES, Doc. C.M. 1991/M:29.
Larsson, H-O. 1977: The influence of predation after release on
the result of salmon smolt planting. lCES, Doc. C.M. 1977/M:44.
Larsson, P-O. 1973: Laxen i östersjön. Proc from Salmon Rearing
Conf., Visby 1973. Paper No.7.
tt
Larsson, P-O. and Svensson, K. M~ 1974: Studies on the possible
influence of early maturity on grilse frequency by means of
tagging experiments in the river Lule. ICES, Doc. C.M.
1974/M:28.
Lindroth, A. 1964: östersjölaxens storlek. Svensk Fisk.
Tidskrift., 73:29-34.
Lindroth, A. 1965: The Baltic salmon stock. Mitt.int.verein.
theor.angev.Limnol., 13:163-192.
Lundin, H-E. 1974: Sammanställning över fördelning pa omrader av
äterfängster fran märkningsförsök. Laxforskningsinst. lnf. 5.
Report of Baltic Salmon and Trout Assessment Wqrkinq Group. C.M.
1991/Assess:13. Copenhagen, 19.-26. March 1991:
~.
Ritter, J. A. 1972: Preliminary observations on the influence of
smolt size on tag return rate and age at first maturity of
Atlantic salmon (Salmo salar). ICES. Doc. C.M. 1972/M:14.
Saunders, R. L. and Allen K. R. 1967: Effects of tagging and of
fin-clipping on the survival and growth of Atlantic salmon
between smolt and adult stages. J. Fish. Res. Bd. Canada, 24(12)
Saunders, R. L., Henderson E. B., Glebe, B. D. and Loudenslager,
E. J. 1983: Evidence of major environmental component in
determination of the grilse:larger salmon ratio in Atlantic
salmon (Salmo salar). Aquaculture, 33: 107-118.
Thurow, F. 1966: Beiträge zur Biologie und Bestandkunde die
Atlantischen Lachses (Salmo salar L.) in der Ostsee. Ber.dt.
wiSSe Komm. Meeresforsch., 18(3/4):223-379.
Thurow, F. 1973: Growth parameters of Baltic salmon (Salmo salar
L.). Ber.dt.wiss .. Komm.Meeresforsch., 23(4):445-451.
•
Table I. Releases of t.!Igged smolts in the river mouths of ICE S 31 in 1970-1988
Releases
'y'ear
•
1970
1871
1872
1973
1974
1975
1976
1977
1978
1979
1880
1981
1982
1983
1984
1885
1986
1987·
1988
TOT.
Recover.... data
No of
No of
groups smolts
7
4
5
11
7
•
11
6
18
11
13
31
16
........
i:..:J
17
28
41
42
29
8300
4000
4983
8998
5500
8427
5446
11873
9559
8529
13840
13303
20803
14895
27395
30854
35622
35509
28667
352
296613
........
Je:.
t"'1ean (X
size (mm)
154.7
166.8
175.2
159.9
168.6
167.1
169.9
174.7
164.6
177.8
178.8
183.5
177.8
178.2
172.2
183.1
190.5
191.1
190.4
180.1
Adult
return-~';
8.0
14.5
21.8
15.2
9.3
n n
0.0
4.4
3.6
3.6
5.2
6.4
3.7
h?
- .'-
' 8.3
9.1
9.2
6.8
4.1
.11.8
Al + recüveries
Mean length (mrn) at
Release
Rec.:;pture
157.9
170.6
179.5
163.8
173.5
172.4
178.1
182.7
177.8
182.4
193.9
198.9
178.3
187.1
179.3
195.4
208.8
212.6
186.6
77
I
.'
(X Weighed mean of the means für different group:s:
•
673.3
687.5
716.4
706.9
672.4
668.8
687.8
652.1
670.8
646.3
663.5
680.6
713.9
740.7
718.6
706.1
720.2
692.9
758.1
t·.·1ean
A1+
length
percenincrement
tage
515.3
516.8
536.8
543.1
498.8
496.5
518.8
469.4
493.1
463.9
469.6
481.7
k ....... "7
--.I.,JI .......
553.6
539.3
510.7
511.4
480.3
562.4
55.8
74.4
64.6
68.1
54.2
48.9
51.5
38.5
37.1
31.2
47.9
44.8
51.0
62.0
67.7
65.8
76.5
66.6
"7 ....
n
.-":-.0
63.6
Table 11. R esults of AN OVA
Source of variation
COVARIATE
length at release
ANOVA for lenqth increment
Mean sauare
Sum of Squ·:ues
dJ.
F--ratio
Siqn.level
,,:>< '"
253657.4
1
253657.4
47.90
1659697.4
18
92205.4
17.41
RESIDUAL
14060895.2
2655
5296.0
TOTAL
15974250.0
2674
MAIN EFFECT
}Ie.:ir üf release
Source of v.:iriation
COVARIATE
mean length at rel.
ANOVA für talJ-return rate farcsin\l~~]
Sum of Square:s
dJ.
t·.·1 ean :square
~
x x
F--ratiü
Siqn.level
884.6
1
884.6
58.24
);~;.:
MAIN EFFECT
year of rele.:ise
6049.5
18
336.1
22.13
~x~
RESIDUAL
5042.9
...............
15.2
11977.0
351
F--ratiü
Sian.level
TOTAL
Source of variation
COVARIATE
mean length at rel.
.J~.c..
AN OVA for A 1 +oercentaqe f arc:sinV~~l
Mean square
Sum of 5 auares
dJ.
2428.2
1
2428.2
85.27
15158.5
18
842.1
29.57
RESIDUAL
8144.4
286
28.5
TOTAL
25731.1
305
MAIN EFFECT
year of release
~
): =-e
xx~
•
•
T able 111,
Correlalion belween mean lenglh increment (growlhJ lag relurn-rate
(survivalJ and A 1+percenlage for each Jlear-class
Correlalion anal...sis
•
GROw'TH
SURVIVAL
A1+ -PERCENTAGE
GRUw'TH
1.000
19
0.000
0.622
19
0.004
0.653
19
0.002
SLI R\lIV.t..L
0.622
19
0.004
1.000
19
0.000
0.613
19
0.006
Al+PERCENT.
0.653
19
0.002
0.613
19
0.006
1.000
19
0.000
Coeffic:ient ~ample ~ize.. ~ignificance level
Table IV
Par ameler~ fm lhe
regre~~ion of
individu.:,llength al rec.:ipture
.:ind length increment on individual length at release
'y'ear.~Iass
1984
1985
1986
1987
1388
70-88
Length at recapture
Re
R 2
A
F
11.9"'''''''
42.4"'''''''
40.0"'''''''
0.990
0.038
0.093
0.097
0.042
0.086
0.653
0.054
0.660
0.910
0.840
OA30
Length increment
R"'2
RC
21.7"'''''''
-0.344
-0.088
-0.159
-0.571
-0.012
0.011
0.001
0.004
0.072
0.000
""',,
1 s:;:....
-'.~
-0.347
0.016
7••• ....
.:J "''''
F
Intercept
n
3.3t-I~':
O.Ons
600.9
528.9
544.7
601.1
564.7
301
443
374
169
234
43.1 "'''''''
585.13
21375
OAns
l.4ns
12)3"'''''''
T able V. T he expected (exp) and observed (obs) number of lagged salmon in c,:stch
sampies fron... the mouths of rivers Kokemäenjeki and Karvianjeki in 1984-1987
(trap-net fishery)
Releases
Catch sampies in 1984-1 987
0/
Year
Al+ salmen
A2+ salmen
A3+ salmon
A4+ salmon
Al+ -A4+
dass tagged n
exp obs n
exp obs n
exp ob:::- n
exp obs
n
exp obs
,~
1982
1983
1984
1985
1986
Total
7.8
1.6
?7
'-.'
2.6
4.1
63
22
59
18
50
0.4
1.6
0.5
2.0
149 4!"1
.-
?'-
6
1
........)
4.9
385 6.2
230 R?
- .....
117 3.0
12 785 20.3
?
'-
6
1
0
9
146 11.4
228 3.6
128 3.5
3
0
1
6
11
0.5
0.2
0
0
215 16.8
646 10.4
417 11.3
k
135 ...... ••_I
50 2.0
~
502 18.5
4
l'
I'
0.7
0
5
8
8
1
........)
1463 44.0 25
•
•
e
e
>rJ
~.
lQ
CI
~
t1>rJ
P! t1
rt<1l
...Nor'_)of groups
..,.
I::J
(.JJ
I::J
Ci
Ci
(_Tl
t::J
No
0:1
-••J
I::J
CI
15-19,9
< 1.0
20- 24.9
3.0 -
,.C
.... _1-
L
oJ.::t
<1l
PJ :J
:J ()
0.-.::
:0.
;t:< ~.
~Cfl
+rt
t1
'0 .....
(1)tr
t1 r:::
() rt
<1l ~.
:J 0
rt:J
PJ
30- 34,9
35- 39.9
: 0
t1
~
<1l
:J
lQ
rt
::r
45- 49,9
LI
50- 54.9
lTl
-
5:1- 59.9
lTl
()
t1
CD
3
CD
:J
rt
t1
(1)
rt
r:::
t1
:J
:C>
Z...l.
...,+
60-64,9
:C>
(;:1
lTl
6:1-69.9
70-74,9
t::J
(\J
t::J
Ul
r..)
Ul
No of individuals
l.'
Ci
12'.012'.9
1:',01 :1,9
1:::,018.9
2'1,021,9
2'4 ~(J24,9
=-
JJ
JJrn
» ...,
...,e
ITlJJ
Z
~
-
27 . 9
80- 84.9
30,030,9
33.033.9
85-89 . 9
90- 94,9
0
u'
r;
~.
~:
\.)
0,
•,N!
(.JJ
I~Tl
t::J
Ul
Ci
...
Ci
Ci
...
Ul
t::J
N
t::J
t::J
N
Ul
Ci
W
Ci
Ci
W
Ul
t::J
<
2'7,0-
75-79,9
~.
:J
Ul
...
9.9
40- 44.9
Cfl
lQ
(1)/'"1)
f groups
g;:
3q
,6,C> 6,9
9,0 -
·...'1.) 1""'1
(1)..0
: r:::
CI
0
...
b
o •
~
('J
~,
...
<:
I)\?
.N~
26
26- 27 ,9
28-29.9
30- .31.9
32-33,9
34-.35.9
36- 3-7,9
.38-.3-9,9
40-41.9
42-43.9
44-45.9
46-47,9
48-49,9
5U-51.
52- 53,9
54-55,9
56- S,7.9
58-59,9
60- 61,9
62- 6.3.9
64- 65,9
6G-67,9
68-69,9
70-7'1,9
72-73,9
74-75.9
76-77.9
78-79.9
80-/3,1.9
82- 8.3. 9
84-/3,5,9
> = 86
Z
Cl r
JJITl
rn Z
3: Gl
rn""
ZI
...,
n
3:
r - - - - - - - - - - - --
80.0
l\ .
70.0
i
w 60.0
,. '1
0;
.I
I
\
/
/
l
40.0
\
'-0--
•
~-
----I-I',.,.
r· r-.
:8
_.....
." ... , .....
.
Vi
Ol
r-
,--
Vi
Ci)
1-· 1-·
.,.-
.-
...
r-.
Gl
t[)
Il
__ ' ~ I••
"..1--
\,....
t r)
15.0 ~
4
'\
10.0
u
[J,,--
~'-11/
20.0
('J
20.0
/#' -""'-0-"
,
\\.
~·I
.0
\0/
JJ
\
\
'L
0.0
",.0
jl
\
/
r-
.
_U
\
I ,I
-{l
n-,,/D--••
..
\
h".....
",
I
a. 30.0
+
,/\\
\
/
,/
w
~J::
I
A 1+
~~--i{,.pE R CE NTAG E
d
~ 50.0
zw
ü
25.0
(1\
r-. F-
r·
r~
en m m
,-.,.-
-
..'
'\
~
10.0
t-
:
W
'\. .,/1
I.RETURN
R.t.. TE
,./
ü:
'" . "
5.0
'"
....."
'.
CJ)
0
Gl
Vi
Ci)
.,.-
r-... r-...
r-
co
0:)
Vi
.- r- .,.rYEAR CLASS
D:)
ti')
aj
":t"
0'.,)
l{)
[r.,)
Ol
Vi
m
.-
Gl
(".J
,--
tD
U0
cn
r· co
co co
Ci)
Vi
80.0
75.0
75.0
70.0
70.0
65.0
65.0
60.0
60 . U-
55.0
LENGTH
50.0
t
0
r-.
Vi
r-
('.J
'"
Ol
,--
rr)
r-.. r-..
Ci)
Vi
.-
.-
...
t{)
'" cn....'"
Gl
r-
r-..
r::: r-..
(1\
m
.- .VI
t
t
t INCREMEt'~T
50.0
+-S.E.
45.0
40.0
()'j
1-·
Gl
r-
ü'
0
r-
('J
Vi
Vi
01
co co
r·
Vi
.,.-
.,.-
r-
YEAR CLASS
Fig 2.
:8
U
55.0
t
40.0'
•
0.0
a)
80.0
45.0
z
~
~.
17)
,--
If)
tr \
":t"
er)
er.,)
Vi
Ci)
0'1
a:5
.,.-
....
'b
n:)
cn
,--
r-.. (Yj
co co
Ci)
Vi
Mean return rates and A1+ percentages (above) and mean
lengths at recapture (A1+) and mean length increments
(below) for year-classes 1970-1988
•
LENGTH INCREMENT (y=-O.347x+585.6. R ....2=O.02)
1000
.
. ..
800
.
'.
SOO
. ...
MM
400
..
..
200
SMOLT SIZE
0
100
150
200
300
250
mm
LENGTH AT RECAPTURE (y=O.653x+585.6. R .....2=O.05)
1200
1100
..
1000
90ü
-
e
700
600
.
• :. - .
...J._·...-::·.-rl.i
;} •••
••••~ "lI .:-. 7. • • -·~t-..r:~
.~._.
. : ::.~ ~~~~...~~.....r. --;::=a:",:.. - ......
800
MM
... ···Vi •... ./ .;.. ~::... .. . ... '" ... .
.
• :~
. .
-
.r.
_ _ _ -.:. -=--~L~:rt'1j~
-
•
..
-
•
.1 ""
11 _
-.
.~
~
:"C"o"...
... ~ : . -v•• _ • • • •
'11:..••• '"
fI ...
.r-~.
L.--••
_~
.
-:...
..
';/.- -.-- ~-=~- =-- .:......_-- -SIZE LIMIT
." "
••••
500
..
400
300
200
100
Fig 3.
SMOLT SIZE
150
200
250
300
The regressions of length increment (above> and length
at recapture (belowl on individual length at release
mm
RE T U R N RAT E (v=0.076x+1.10. R ··.." 2=0.07)
40.0
35.0
.'.
30.0
~c
'Cij
.
25.0
•
20.0
15.0
10.0
•
•
• • '.1
'
Ir;"
~. r,'. I
•
I~:
111,
i
_li ..
• • r.
...e ..
. .•
••
•
."
,,-...
...........-..-.
=".~.
•• •11.
•
I:.. .~i"~i-'·'
'..
• . ...,.r-oI.............. •I!.:." !..!.' •• ' •
I.·... .'·. '.
.....-..--.-. • ' .
·lfl ' . 1i.1
.....
• • '.
5.0
'. I• ."
• •
I'
I .. t
I·~II••
• 11
:
!
t~"'
i
0.0
120
I
•
•••
•
• 'tl _'.1
" .• iI.~- ,.1. _." '. .• ' _
• • :~ "
' ; ; . • r. ,I
•
•
I, -.•
<..l
Ci
I
•
I
t'
i
.'
I
•
I
•
••••
.
.! • ••'
i
•"
140
160
180
rn rn
200
220
240
260
280
MEAN LENGTH
A 1 + PERCENTAGE (v=O.136x+25.9. R ....2=D.09)
80.0
70.0
I
t,~
>c:
'(Ji
o
.I
. '
• !--I 'r.'i='.II.
• ,I. r. .1..
•
• • ~•.
•" ••1
•
•••
~
.1
_. I .
-I
•
• '!~I
60.0
••
50.0
()
'-
40.0
•
• ".'
I.
••••••• !
.'"
_I
••••
•• '
I.
1
.. 1 t
, .•..;....,• . . - .
I.-!.. ~••----:. • r'
-•-.',,,'
.• • •
•
'I
i
•':", •
r.":""'-'=-••
• I •
" ';..!.~.ii--....
Ir' " ....
•
'.~
:'! '.
iii..
," •
....
" •• '
",
I.
"
I.
.: •
••
'.'
•
.a
i •
'1 11 -li
I I I ," I
"
30.0
I
I
11
..I
I
I
20.0
120
rn rn
140
160
180
200
220
240
260
280
MEAN LENGTH
Fig 4.
The regressionsof return rate (above) and Al+
percentage (below) for each smolt group on mean
length of the group at release
.
30
S'~ ze
limit
Ul
.....
ro
:1
'd
I
25
I
I
I
I
I
I
I
I
20
'rl
> 15
'rl
'd
s::
-rl
10
~
0
0
5
z
0
\t
~.
01
f',
OJ
l"
Cl
01
'<'"""
t<)
A
.
. 0-.'. 01
Ci
0;'
l.CI
f',
OJ
0r-
("
A
A
01
l[)
m
CD
I
m
CD
I
tD
<.0
I
I
0Cl
lf'I
I
I
I
J
(D
(D
f',
lf)
I .1 ••
0
N "':t <.0
CO ID <.0
t<)
.... r---
OJ 0
J
Oi
Ci
Oi
l.f':.
....
'<'"""
t<)
("'"
("-.
("....
QJ
I
I
I
I
C'J "'ot
.... ("'"
("
. 0101
0;0
(D
....
("
("
ca
0
I.•
~00
I• •
.
01
Oi
01
("-.
Ü1
o::J
l()
{D
QJ
OJ
I
cn
GI
I
I
('-J
I
I
I
I
"'ot
(D
('-J
"'ot
Ci Oi
01
A
.... ro ro D:l ro
("
LENGTH GROUP
Fig 5.
n_ ..
I
n"
088
.11 n
I
(D
lf)
I
Ci
Oi
'<'"""
t<)
A
ca
ca
01
(D
l[)
Oi
A
0
m
"
/\
CM
Length distribution of A1+ salmon for year-classes
1987 (poor growth) and 1988 (good growth)
200
180
160
"U
QJ 140
11'
~
0 120
(.)
<ll 100
..... 80
0
60
0
z
40
20
RE CO''.JE RIE..JC
ELEAS;E.S r-
/-lZ
•/
/
/
lo..
/
~
"......
r-
2000
l'
"0
I
/
Cl1
r-
"•, 1\
~
.....
r-
1000 Ö
o
,
I.,
r-
z
rr-
f\
500
1'1.
/
/.0
Ü
~
1500
1\
r-
/
I
111
1'-,
I
/
•
2500
I'-
-.- -~ ...,
.
o
120
100
80
0
z
A1+
IGRI
60
OIMM
•
40
20
0
40
'"'k
~._I
r-
30
25
A2+
I t·..1AT
0
:z 20
15
Olt·,·1t·..1
r-
10
r-
5
nl
0
c..~
1'<-1
f-'l
-::t
..-
c..~
U-I
I
<:..~
(D
..-
I
0
1'<-1
0
-.:t
0
U-J
()
(f)
..
r.~
1'-.
.-I
0
1-.
<:..~
OJ
0'1
0'1
<:..~
I • •
c..~
C' .- ,-
..--
.--
(,~
t"~
I
I
I
0
0
I
0
("~
N
0
(0
cJ)
0
<:..~
0'1
01
'"'t
...rI.....
0'1
('I
1'<-1
U-J
(Tl
<:..~
(,~
(,~
(,~
01
1'-.
(,~
(,~
01
(0
(,~
(""~
I
I
I
0
I
0
I
0
0
I
0
0
I
0
('I
(,~
0
t<1
c-~
-.::t
N
U-I
('I
(D
1'-.
(ü
(,~
N
(,~
0'1
<:..'tl
01
C=l
l>~
~l
I
I
0
0)
(,~
Cl
1""')
SMOLT LENGTH MM
Fig 6.
Frequency distributions of smolt length for releases in
the mouth of river Iijoki in 1985. Above: All releases
and all recoveries. Center: Grilse and immature Al+
salmon. Below: Mature and immature A2+ salmon
•