MAIN-STEM AND TRIBUTARY SAMPLING
OF RED SALMON SCALES
FOR POPULATION STUDIES
Marine Biological Laboratory
■■■ ' ' lyeo
WOODS HULE, MASS.
SPECIAL SQENTIFIC REPORT-FISHERIES Na362
United States Department of the Interior, Fred A. Seaton, Secretary
Fish and Wildlife Service, Arnie J. Suomela, Commissioner
Bureau of Commercial Fisheries, Donald L. McKernan, Director
MAIN-STEM AND TRIBUTARY SAMPLING OF
RED SALMON SCALES FOR POPULATION STUDIES
by
Ted S. Y. Koo, Research Assistant Professor
and
Howard D. Smith, Senior Research Biologist
Fisheries Research Institute
College of Fisheries, University of Washington
Seattle, Washington
This work was financed by the Bureau of Commercial
Fisheries under Contract Nos. 14-19-008-2336 and
14-19-008-9346, with funds made available under the
Act of July 1, 1954 (68 Stat. 376), commonly known as
the Saltonstall-Kennedy Act.
United States Fish and Wildlife Service
Special Scientific Report- -Fisheries No. 362
Washington, D. C.
August I960
CONTENTS
Page
Introduction 1
Sannpling areas 2
Kvichak Rive r 2
Spawning grounds 2
Sampling methods 2
Kvichak River sampling 2
Spawning ground sampling 2
General methods 4
Freshwater age composition of Kvichak River and spawning ground
samples 4
Discussion of results 6
Kvichak River sampling 6
Spawning ground sampling 7
1956 samples 8
1957 samples 8
1958 samples 8
C onclus ions 9
111
MAIN-STEM AND TRIBUTARY SAMPLING OF
RED SALMON SCALES FOR POPULATION STUDIES
by
Ted S. Y. Koo and Howard D. Smith
ABSTRACT
Comparison of freshwater age composition by scale studies was
made on red salmon sannpled at two locations: (1) in the Kvichak River
before fish dispersed into the lake system, and (2) on spawning grounds
after fish had spawned. Kvichak River samples in 1956 and 1957 had higher
percentages of 1 -winter -in-lake fish than did spawning ground samples.
The discrepancy was believed to be due to difficulties in obtaining repre-
sentative samples and proper weighting from spawning grounds. The 1958
samples from both locations were comparable, probably as a mere coinci-
dence.
It is concluded that although spawning ground sannpling is essential
to studies of population density, distribution, and subpopulations in a
lake system, it is difficult, if not impossible, to obtain and weight the
samples so that they are representative of the whole escapennent. The
proper place to sample the entire escapement is in a trunk river, such as
the Kvichak River in the Kvichak system.
INTRODUCTION
Age and size composition of red salmon
(Oncorhynchus nerka) runs of Bristol Bay,
Alaska may be determined only if both
catch and escapement are sannpled, be-
cause the commercial gear, restricted
by regulation to gill nets of 5 1/2-inch
mesh or larger, is highly selective on
size, and therefore on age and sex of
the fish. In the average runs, the com-
mercial gear takes a higher proportion
of the larger fish which have spent 3
years in the ocean, whereas more of the
small fish escape. Catch is usually
sampled at a cannery where fish are
landed; escapement may be sampled at
two places--first in the main river before
dispersing to the spawning grounds, and
second on the spawning grounds after
fish have spawned.
The purpose of this paper is to deter-
mine whether or not spawning ground
sampling can represent the entire escape-
ment. If the objective is to study popula-
tion density, time and geographical
distribution, and subpopulations, spawn-
ing grounds are the place to sample. If
it is to study sex ratio, age composition,
and other characteristics of the popula-
tion as a whole, the logical place to sample
is in the trunk river or main stem where
the entire escapement ascends within a
relatively short period and can be ade-
quately sampled.
iContribution No. 58, College of Fisheries, University of Washington.
1
This paper compares scale samples
collected both in the trunk river and on
spawning grounds in the Kvichak River
system during the 1956 to 1958 field
seasons. Because resorption erodes much
of the peripheral part of scales of fish on
the spawning ground, leaving only the
central portion legible, comparison is
confined to freshwater age composition.
SAMPLING AREAS
Kvichak River
The trunk river of the Kvichak system
is the Kvichak River. In their upstream
migration, red salmon follow the banks
of this river and pass steadily into the
lake along both sides of the outlet. They
seldom use the deep midstream section
as a migratory route. Escapement can
therefore be sampled from either shore in
the river or at the lower end of Lake
niamna (fig. 1).
Samples of Kvichak River escapement
were taken at Igiugig (1),' which is situated
on the upper reaches of the river. Owing
to the topography of the bottom and wave
action caused by a prevailing wind, most
sampling took place on the left or south-
eastern shore in Lake Iliamna immediately
above the outlet.
Spawning Grounds
There are three principal types of
spawning habitat in the Iliamna-Clark
system. First in importance are the
tributary streams and rivers, which sup-
port about 70 percent of the visible
spawning population. Next in importance
are the beaches on Lakes Iliamna and
Clark and on the accessory lakes tributary
to thenn. Last are the spring-fed ponds,
such as Hudson Ponds (6), in which only a
few salmon spawn.
All fish do not appear on the various
spawning grounds at the same time nor
within a short period of time. Although
the entire escapement has passed through
Igiugig and entered Lake Iliamna by the
end of July, some populations do not
appear on the spawning grounds until
October or later. While attempts were
2 The Arabic figure in parenthesis immediately following the
locality is the number used in figure 1.
made to obtain measurements and scale
samples from all known important spawn-
ing areas in the system, it was indeed a
formidable task to effect such a compre-
hensive coverage over a wide area and
for an extended period of spawning.
SAMPLING METHODS
Kvichak River Sampling
The escapement on the Kvichak River
was sampled with a beach seine of
cotton webbing with 3 -inch mesh (stretch
measure). The net was 200 feet long and
12 feet deep and was set with a skiff
powered by an outboard motor and hauled
in manually.
The escapement past Igiugig usually
extends over the month of July. Except
for the first and last few days of the
migration, when only a few fish were
running, fish were sampled daily. Each
day 3 or 4 hauls were made, which
normally took 4 to 6 hours. Fish were
measured and tagged, and a scale was
removed from each fish before it was
released. When possible, 40 or more
scale samples, about half of them from
each sex, were collected each day.
Seining was carried on during a period
of about 3 weeks each season. During
this period, 87 to 99 percent of the
escapement made its way into the lake,
as revealed by daily counts (table 1).
Spawning Ground Sampling
Individual spawning grounds were sur-
veyed by air, whenever possible, at a
time when the greatest number of fish
was present. A visual estimate of the
population was taken then, but the actual
sampling was deferred until later when
enough dead fish for an adequate sample
were available along the banks. At that
time, a ground survey was made to take
the sample.
The total spawning population seen dur-
ing our aerial surveys constitutes only a
fraction of the population counted at
Igiugig. In the 3 years, 1956 to 1958,
this fraction ranged from 10 to 18 per-
cent. This relatively low estimate of the
spawning population is partly due to the
numbers of fish seen represent only the
(1) Igiugig
(2) Lower Talarik
(3) Newhalen River
(4) Tazimina River
(5) Kijik Lake
(6) Hudson Ponds
(7) Flat Islands
(8) Knutson Bay
(9) Surprise Creek
(10) Lonesome Bay
(11) Iliamna River
(12) Finger Beaches
(13) Tommy River
(14) NickG. Creek
(15) Copper River
( 16) Kokhanok Creek
(17) Belinda Creek
Figure l.--Iliamna - Clark system, showing the sampling locations.
Table 1.— Samples of adult red salmon from the Kvichak River escapement, 1956-58
Total
Percent
Sc
ale
sample
Year
escapement
enumerated at
Igiugig
of escapement
during sam-
pling period
Number of
scales
collected
Percent
of total
escapement
1956
9,d^3,000
95.9
1,673
0.018
1957
2,965,000
99.1
1,742
0.059
1958
535,000
87.1
720
0.135
peak counts and partly due to the incom-
plete coverage of spawning grounds be-
cause of certain conditions. For instance,
Lake Clark and many of its tributaries,
important as they are as spawning areas,
were not surveyed because of their turbid
glacial water. Comparative sampling data
on the spawning grounds are shown in
table 2.
Only the more important spawning areas
were sampled each year. An area im-
portant as spawning grounds in one year
may not be so in another. Also, owing to
weather or other difficulties, even locali-
ties that were important as spawning
grounds were not visited in some years.
Therefore, not all localities sampled in
the first year were sampled in succeed-
ing years. Of the 16 localities sampled
in the 3 years, only 3 were sampled
consistently in all 3 years; 6 in 2 years,
and 7 in 1 year (table 3). The names and
locations of the sampling areas are shown
in figure 1. In 1956, the summed peak
count of the scale sampling areas (marked
by X in table 3) was about 81 percent of
the total from all localities; in 1957, 56
percent; and in 1958, 73 percent.
General Methods
All scales were removed from a speci-
fied area on the body about midway between
the dorsal and adipose fins and within 5
scale rows above or below the lateral
line. They were mounted on gummed cards,
and impressions were made on cellulose
acetate for examination under the micro-
scope and projector after the method
described by Koo (1955).' Scales with
^Biology of the red salmon, Dncorhynclius nerhn (Walbaum)
of Bristol Bay. Alaska, as revealed by a study of their scales.
Ph.D. Thesis. University of Washington, Seattle. 164 pp.
(Available on interllbrary loan or by microfilm service.)
regenerated nuclear areas were dis-
carded. Those collected from the Kvichak
River were read by one author and those
from spawning grounds by the other. A
certain portion of the samples were read
by both authors to determine whether the
two authors agreed, in general, in their
interpretation of scale marks. The validity
of scale markings read as annuli is not
discussed here because it is not important
to this study. Table 4 shows the percentage
of agreement in the two readings.
Only freshwater age was determined,
and only two age groups were involved,
either one annulus (designated here as 1.),
or two annuli (designated here as 2.). Two
types of errors resulted in disagreement
between readers: (1) Scales read as 1. by
the first reader and 2. by the second, and
(2) scales read as 2. by the first reader
and J., by the second. For the 23 dis-
agreements encountered over the 3-year
period (table 4), 12 were errors of the
first type, and 1 1 were errors of the second
type. Disagreements between reader s thus
tended to average out.
Comparative readings by the two authors
indicate a high degree of agreement.
Therefore, any major difference in age
composition between Kvichak River and
spawning ground samples cannot be
attributed to difference in interpretation
of scale marks.
FRESHWATER AGE COMPOSITION OF
KVICHAK RIVER AND SPAWNING
GROUND SAMPLES
In Kvichak River samples, percentages
of age groups \. and 2. were calculated
for each day a sample was taken. They
were then weighted according to the daily
Table 2. — Samples of adult red salmon from the Kvichak spawning grounds, 1956-58
Year
Peak population
estimate from
aerial surveys
Number of
scales
collected
Scale sample
Percent
of total
escapement
1956
1957
1958
1, "440, 000
301,000
96,000
1,050
910
0.011
0.031
0.159
Table 3. — Peak population counts on various Kvichaic spavming grounds, 1956-58
[(x) indicates scale sample taken; — indicates no survey made]
Number of fish, peak counts
Locality-
1956
1957
1958
Copper River (15)^
450,000
(x)
70,000 (x)
20,000 (x)
Nick G. Creek (U)
5,000
(x)
3,000 (x)
1,200 (x)
Knutson Bay (8)
155,000
(x)
23,500 (x)
14,000 (x)
Belinda Creek (17)
6,000
(x)
2,000 (x)
Finger Beaches (12)
10,000
(x)
6,000 (x)
2,700
Tazimina River (4)
32,500
(x)
10,000 (x)
650
Kijik Lake (5)
86,000
(x)
27,000 (x)
6,000
Kokhanok Creek (16)
190,000
(x)
30,000
2,800 (x)
Lower Talarik Creek (2)
32,000
(x)
9,500
250
Iliamna River (ll)
200,000
(x)
26,500
10,000
Newhalen River (3)
23,000
25,000 (x)
30,000 (x)
Tommy River (13)
12,000
2,100 (x)
2,050
Hudson Ponds (6)
3,000
500 (x)
0
Flat Islands (7)
41,000
3,700
1,200 (x)
Surprise Creek (9)
135 (x)
Lonesome Bay (10)
500 (x)
Miscellaneous
194,500
62,200
4,515
Total
1,440,000
301,000
96,000
""■ Number refers to location in figure 1.
Table 4. — Scale reading agreement by the two authors
Year
material
collected
Source of
material
Number of
scales
read by
both
Scales read as
belonging to
same age
Number
Percent
1956
1956
1957
Kvichak River
Spawning grounds
Spawning grounds
196
117
207
189
109
199
96.4
93.2
96.1
escapement counted from the towers
erected in the stream at Igiugig. Finally,
the total percentages of the two age
groups for the entire season were calcu-
lated (table 5).
In spawning ground sanriples, percent-
ages of the two age groups were calculated
for each locality. These were weighted
according to estimated population size
for each locality, and the totals for all
spawning ground samples were obtained.
These are compared in table 5 with those
from Kvichak River samples.
were properly weighted, there should be
no difference in the freshwater age
composition between the two samples.
The discrepancy revealed in 1956 and
1957 samples is obviously due to sampling
difficulties, which may have occurred at
either or both places. Let us examine
each sampling closely.
Kvichak River Sampling
Theoretically, several factors here may
cause nonrepresentative sampling of the
population.
In 1956 and 1957, there was a marked
difference in age composition between
samples from the Kvichak River and from
the spawning grounds, but in 1958, the
age composition was practically identical.
The differences in the first 2 years are
consistent in that there is a higher propor-
tion of I. fish in Kvichak River samples
than in spawning ground samples. In other
words, proportionately more 2. fish were
found on the spawning grounds than were
first sampled in the Kvichak River. This
is especially pronounced in 1956.
DISCUSSION OF RESULTS
Scale samples from both the Kvichak
River and the spawning grounds naturally
canne from the same population. If they
were representative of each place and
1. The sampling period: The run lasts
about a month. Our sampling covered only
about 3 weeks. The early and tail ends of
the run were not sampled. However, most
of the migration took place during the 3-
week sampling period; in both 1956 and
1957, more than 95 percent of the run
occurred within this period (table 1).
This coverage can certainly be considered
adequate.
2. The weighting: The percentages of
the two age groups as determined from
each day's samples were weighted ac-
cording to the number of fish counted
from towers in the Kvichak River. Tower
enumeration of the adult run has been
proved an accurate method of assessing
the number of fish by this Institute's
work on the Wood River and by the Fish
and Wildlife Service's work on the Egegik
Table 5. — Freshwater age composition between Kvichak River and spawning ground samples
of adult red salmon, 1956-58
Year
Locality
Age groups, in percent
1.
1956
Kvichak River
Spawning grounds
83.6
53.6
16.^
46. A
1957
Kvichak River
Spawning grounds
57.4
44.1
42.6
55.9
1958
Kvichak River
Spawning grounds
11.6
11.2
River.* Therefore, weighting by the daily
counts should be reliable.
3. The sampling locality: Ideally fish
should be seined from both shores in any
day of sampling in order to avoid any
possible selection of segregated groups
of fish. Seining on both shores with equal
effort was not achieved because of the
difficulties mentioned earlier, and there-
fore may be a source of error. However,
from tagging studies, the junior author
(unpublished study) found that there was
no evidence of stratification of fish on the
spawning grounds, and at the same tinne
individual areas contained tagged fish
from each shore about in proportion to the
numbers tagged there.
4. The sampling gear: Is it possible
that the beach seine might be selective
so that more 2. fish avoid the net and
appear on the spawning grounds ? This
might happen if 2. fish were larger than
1. fish, either by their original difference
in size or by their differential associa-
tion with marine age, namely 63 and 42-
The original difference in length, which
amounts to about 20 percent during smolt
stage, diminishes rapidly in the adults
because of the overwhelming marine
growth. Therefore, it is improbable that
net selection could result on that account.
The possibility of differential association
of marine and freshwater ages can be
ruled out since in 1956 the escapement
was composed almost entirely (99 per-
cent of 3} fish.
Fron-i the above considerations, we
conclude that sampling in the Kvichak
River was probably adequate, and the
discrepancy in age composition between
Kvichak River and spawning ground
samples must be traced to the latter.
Spawning Ground Sampling
Possible sources of sampling error are
as follows:
1. Although the sampling crew had tried
to cover most major spawning areas, a
number of places were left unsurveyed,
either because of time and personnel
limitations or because of practical diffi-
culties. In 1957, for instance, sampling
areas accoxinted for only 56 percent of
the peak population estimated from aerial
surveys.
2. Only a small portion (10 to 18 per-
cent, table 2) of the actual escapement
could be accounted for by spawning ground
estimates. The weighting of age groups
by localities was done according to rela-
tive population estimates, the accuracy
of which could not be measured.
•4 The following is quoted from "Progress Report and Recommendations for 1957"byAdministrationof Alaska Fisheries, Juneau,
Alaska, November 1956, p. 19:
"2. Evaluation of Towers for Counting Adult N4igrant Red
Salmon in Bristol Bay.
"A critical comparison of weir and tower counts was made
on the Egegil< River during the period July 12 to July 30 when
somewhat more than a million adult fish migrated upstream
(Figure 34).
"After finding out the habits of the fish and locating towers
in the proper places, a completely satisfact jry count of the
migrants was obtained. On two days during the early part of
the run, fish by-passed the towers which were not located
properly. During the balance of the migration, estimates of the
total run from tower averaged only 1.6 percent lower than
estimates made at the weir."
EGEGIK RIVER RED SALMON
ENUMERA TION — 1956
WEIR COUNTS
UPSTREAM TOWER
COUNTS
12 14 16 18 20 22 24 26 28 30
JULY
3. The important spawning streanns
are quite long, often 10 to 20 miles, and
have fish distributed over the entire
length. Ground surveys and sampling
could normally cover only relatively short
sections. If distribution of age groups
was uneven over the whole length of the
river, then sampling one or two sections
could lead to biased results.
4. In some areas spawning took place
over a long period of time, and there
were instances of early and late popula-
tions. Owing to practical difficulties,, re-
peated visits to a spawning ground were
not made, and therefore the data used
may be representative of one population
but not the other.
5. The two age groups were not evenly
distributed over all spawning grounds.
While 1. fish predominated in one local-
ity, 2. fish predominated in another
(table 6).
listed were not representative of the
population; they could have been selective
in favor of Z, fish because of timing,
selection of a certain section of a stream,
or other reasons.
1957 Samples. --Samples from only two
spawning ground areas, Nick G. Creek
and Belinda Creek, had significantly higher
percentages of I. fish than samples from
the Kvichak River (table 6), but both
these localities are insignificant spawn-
ing creeks (table 3). Unless the population
in these creeks had been grossly under-
estimated, which is improbable, weighting
cannot be a main cause for the discrep-
ancy here.
The two factors that caused nonrepre-
sentative samples in 1956 might also have
caused them in 1957. Kokhanok Creek and
ELiamna River, for instance, while both
important as spawning areas, were not
sampled.
The variable ratios of the two age groups
on the spawning grounds occur in each of
the 3 years and are occasionally extreme.
This fact, coupled with the uneven distribu-
tion of population size on the various
spawning grounds, makes proper sampling
and weighting extremely difficult, if not
impossible.
The above are general considerations
of some theoretical factors that may
cause spawning ground sampling to be
nonrepresentative of the escapement pop-
ulation. Let us now examine each year
individually.
1956 Samples. --Samples from only one
spawning grovind area, Tazimina River,
has a higher percentage of 1_. fish than
samples from the Kvichak River (table 6),
but Tazinnina River was relatively unim-
portant as a spawning area (table 3).
Therefore, weighting cannot be considered
here as the main cause of the discrep-
ancy between spawning ground and Kvichak
River sampling.
Other factors that could have caused
the discrepancy are: (1) Localities not
sampled for scales had large populations
with higher than 84 percent 1_. fish.
(2) The samples obtained from localities
1958 Samples . - -Age composition of
spawning ground samples agreed with
that of Kvichak River samples. While in
2 previous years most spawning localities
had higher percentage of Z. fish than
Kvichak River samples, in 1958 most
spawning localities had higher percentages
of 1_. fish (table 6). In fact, the unweighted
mean is 21 percent for J_. fish in 1958,
compared with the weighted nnean of 11.
Obviously then, weighting here has played
a big role in bringing about the agreement
between spawning ground and Kvichak
River samples. This can be noted espe-
cially in the case of Newhalen River
(table 3).
Only 8 locations were sampled in 1958
compared with 10 in either 1956 or 1957,
and the sampling areas in 1958 were, in
general, quite different from those of the
other 2 years (table 3). Three new areas.
Flat Islands, Surprise Creek, and
Lonesome Bay, which were insignificant
producers, were added in 1958, while
some of the more important areas, Iliamna
River, Kijik Lake, and Finger Beaches,
were not sampled. Therefore, the excellent
agreement between Kvichak River and
spawning ground samples in 1958 is
probably a fortuitous occurrence rather
than a result of more representative
sampling.
Table 6. — Number (n) of scales read and percentage (^) of 1. adult red salmon from
various Kvichak spawning grounds, 1956-58
1956
1957
1958
Locality
n
$
n
i
n
1o
Copper River (15)-'-
156
35.3
188
44.2
133
9.0
Nick G. Creek (K)
51
64.7
59
72.9
31
22.6
Knutson Bay (8)
53
81.1
24
54.2
84
31.0
Belinda Creek (17)
^9
79.6
52
78.9
—
Finger Beaches (12)
24
62.5
26
26.9
—
—
Tazimlna River ("4)
48
89.6
41
41.6
—
Kijik Lake (5)
4-7
48.9
38
44.7
___
Kokhanok Creek (16)
125
76.8
82
7.3
Lower Talarlk Creek
(2)
40
62.5
Iliamna River (ll)
39
48,7
Newhalen River (3)
53
34.0
127
1.6
Tommy River (13)
— _
52
28.9
Hudson Ponds (6)
45
57.8
Flat Islands (7)
126
16.7
Surprise Creek (9)
29
20.7
Lonesome Bay (10)
24
58.3
All spawning ground
samples.
weighted:
53.6
44.1
11.2
Kvichak River samples, weighted:
83.6
57.4
11.6
■"■ Number refers to location in figure 1.
CONCLUSIONS
The difference in freshwater age com-
position between Kvichak River and
spawning ground samples is due to diffi-
culties in obtaining representative
samples and in proper weighting from
the spawning grounds. Of the total escape-
ment counted in the Kvichak River, only
about one -fourth of it could be seen on
the spawning grounds. Most of the popu-
lation was therefore not available to
sampling on the spawning grounds. It is
not known whether the missing fish were
near lake shores, in deep-water spawn-
ing areas, or hidden behind glacial water
or other natural protection. Then, too, a
large portion of the population seen was
not sampled because extensive areas and
lengthy spawning periods were involved.
Further, it is widely recognized that
there are subpopulations of red salmon
within one lake system, and that these
subpopulations segregate on the spawn-
ing grounds. The fact that various locali-
ties in the ELiamna-Clark system showed
different age composition is clear evidence
of such segregation. Sampling the segre- place to sample is in the trunk river,
gated subpopulations to represent the such as the Kvichak River in the Kvichak
whole population cannot be achieved with- system, while the fish are migrating
out prohibitive increases in cost and upstreann before dispersing to the spawn-
effort, ing grounds. This cannot adequately be
replaced with spawning ground sampling.
We conclude, therefore, that to deter- It must be added that trunk river sampling
mine the sex ratio and the size and age cannot replace spawning ground sampling
composition of the escapement of red either, if the objective is to study sub-
salmon into a lake system, the proper populations in a lake system.
10
GPO 899451
MBL WHOILIbrarv - serials
)rarY - =>
llHwlf'
5 WHSE 0
490