512

Hypotheses on the Origin of Exploited
Skipjack Tuna (Katsuwonus pelamis)
in the Eastern and Central Pacific Ocean

By Brian J. Rothschild

i -i^iar.nc C;

SPECIAL SCIENTIFIC REPORT-FISHERIES Na 512

^UNITED_STATES DEPARTMENT OF THE INTERIOR

FISH AND WILDLIFE SERVICE

UNITED STATES DEPARTMENT OF THE INTERIOR

Stewart L. Udall, Secretary
Frank P. Briggs, A.^sistant Secretary for Fish and Wildlife

FISH AND WILDLIFE SERVICE, Clarence F. Pautzkc, Commissioner

Bureau op Commercial Fisheries, Donald Tj. McKenian, Director

Hypotheses on the Origin of Exploited

Skipjack Tuna ( Katsuwonus pelamis )

in the Eastern and Central Pacific Ocean

By

BRIAN J. ROTHSCHILD

United States Fish and Wildlife Service
Special Scientific Report--Fisheries No. 512

Washington, D. C,
April 1965

CONTENTS

Page

Introduction^ 1

Evidence___ 3

Distribution of spawning 3

Spatial distribution of larvae 3

Temporal distribution of larvae 3

Gonad indices 4

Length-frequency distributions of Hawaii samples 5

Modal sizes 6

Size distribution 6

Length-frequency distributions of eastern Pacific samples 7

Comparison of eastern and central Pacific length-frequencies 8

Movements 9

Tagging in the eastern Pacific 10

Long distance tag recoveries 10

Prediction of Hawaiian skipjack catch 10

Subpopulations^ 11

Hypotheses 11

Skipjack exploited in eastern Pacific 11

Skipjack taken in Hawaiian waters 15

Conclusions^ 17

Summary ___^ I7

Acknowledgments ^ . 19

Literature cited ^ .19

Hypotheses on the Origin of Exploited

Skipjaclc Tuna (Kotsuwonus pelamis)

in the Eastern and Central Pacific Ocean

By

BRIAN J. ROTHSCHILD

Fishery Biologist (Research)

Bureau of Commercial Fisheries

Biological Laboratory, Honolulu, Hawaii

ABSTRACT

A set of hypotheses has been formulated to account for the origin and move-
ment of exploited groups of skipjack tuna (Katsuwonus pelamis) in the eastern
and central Pacific Ocean. The hypotheses take into account the available evi-
dence on larval distributions, gonad indices, size distributions, tag recoveries,
catch predictions, and immunogenetic studies. The evidence suggests that most
skipjack taken by the eastern Pacific skipjack fisheries originate in the central
Pacific. It is likely that the equatorial region of the central Pacific contributes a
major portion of the recruitment stockfor the eastern Pacific. Large proportions
of the Hawaiian catch also may originate in the Equatorial Zone. Skipjack catch
predictions are discussed in the context of evidence which indicates that year-
class-strength phenomena affect the Hawaiian landings. The need for more evi-
dence on the origin and movements of harvested skipjack is emphasized.

INTRODUCTION

Skipjack tuna (Katsuwonus pelamis) are distri-
buted throughout the tropical and subtropical Pa-
cific Ocean. In recent years yields from Pacific
skipjack fisheries have been of the order of 550
million pounds annually. There are now three
major fisheries: the eastern Pacific fishery, in
the coastal waters of North, Central, and South
America from Baja California to Peru (161.4
million pounds in 1962); the Hawaiian fishery
(9.4 million pounds in 1962); and the fishery off
the Japanese archipelago (375 million pounds in
1962). Formerly there was a fishery in Micro-
nesia, which produced 72.8 million pounds in 1937
(Shapiro, 1948: p. 59). Although this area has
been virtually unfished for skipjack over the past
2 decades a fishery is now recommencing.

The apparent decline in eastern Pacific yellow-
fin tuna stocks beingfished (e.g., Schaefer, 1961,

1962,1963) and the increasing world demand for

1/
tunas (Chapman)- suggest that fishing for skip-
jack may become greatly intensified, both in
existingfishing areas and in areas not now fished.
Management may eventually become necessary to
ensure continued high yields of skipjack.

To place future investigations and possible
management measures on a rational and eco-
nomically sound basis, we need knowledge of the
origins and relationships of the groups of skip-
jack that are being harvested or may be harvested
in the future. A first step in organizing and en-
larging this knowledge is the construction of a
set of hypotheses incorporating available data
into a testable, integrated picture of the popula-
tion biology of the Pacific skipjack. The presen-
tation of such hypotheses is the purpose of this
paper. Since informationfrom Japanese sources
(e.g., Kawasaki, 1955) indicates that the skipjack
caught off Japan originate from spawnings in the
Ryuku-Izu-Bonin Islands area, far from the areas
of direct concern to American fishermen, I will
discuss only the origin and relationships of the
skipjack of the eastern and central Pacific. I do
not attempt to review all the literature on skip-
jack populations because adequate reviews have

— ' Chapman, W. M. Recent trends In world tuna
production and some problems arising therefrom.
Paper presented to the Symposium on Scombroid
Fishes. Marine Biological Association of India,
Mandapam Camp, South India, 12-15 January 1962.
(In press.)

recently been published by others (Waldron, 1963;
Jones and Silas, 1963; Postel, 1963).

This paper collates the available evidence per-
tinent to the problem of the origin of the exploited
groups of skipjack in the eastern and central
Pacific Ocean and forwards hypotheses which, I
believe, are consonant with such data. Its sec-
tions will cover, first, a brief statement of the
hypotheses; secondly, the evidence upon which
they are based, and finally a discussion of the
hypotheses and the sometimes intuitive reason-
ing which led to their development.

The set of hypotheses has been generated from
a consideration of a wide variety of data. Many
of these data have been published by the Inter-
American Tropical Tuna Commission (lATTC)
and the Bureau of Commercial Fisheries. Other
data have been obtained from the files of the
Bureau's Biological Laboratory, Honolulu, and
analyzed for this paper.

The hypotheses are based on a consideration
of three conceptual zones from which the skipjack
harvested in the eastern Pacific might originate
(fig. 1). The hypotheses postulate that large pro-
portions of skipjack which enter the eastern Pa-
cific fishery originate in the Equatorial Zone.
The predominant flow pattern for these fish is
suggested in figure 2. Atprerecruitsize(< about
35 cm.) skipjack are continually dispersing from
the equatorial central Pacific. A large compo-
nent moves east toward the coast of the Amer-
icas; near the coast, one portion enters the Baja
Californiafishery area while another moves into
the Central and South American fishery areas.

Figure 1 . --Schematic representation of the
conceptual areas that were evaluated in
this paper.

The skipjack remain in the eastern Pacific for
several months and may attain sizes of 55-65 cm.
The attainment of this relatively large size (for
the eastern Pacific) seems correlated with an
offshore spawning movement. This offshore
movement terminates in equatorial waters of the
central Pacific where spawning occurs. An ap-
parent north-south component of movement is
superimposed on the generally easterly move-
mentof "adolescent" prerecruits to the eastern
Pacific fishery and the generally westward di-
rection of the eastern Pacific spawning escape-
ment. This north-south component may be a
general dispersal affected by population density
or it may be associated with the seasonal shift
of the physical environment (e.g., temperature).

MEXICAN FISHERY

CENTRAL AND
SOUTH AMERICAN
FISHERY

/

SEASONAL

AND DIFFUSION

MOVEMENTS

Figure 2. — Diagram showing the flow of skipjack between the central
Pacific and the Mexican and South American fisheries of the eastern
Pacific.

particularly in the more temperate latitudes.
The maximum northward extension of the Equa-
torial Zone skipjack occurs during the northern
summer, thus also accounting for a large pro-
portion of the Hawaiian catch.

EVIDENCE

Data that seem pertinent to the problem of the
origin of the exploited groups or subpopulations
of skipjack have been selected from the litera-
ture. The data selected involve the distribution
of spawning, length-frequency distribution of
Hawaiian and eastern Pacific samples, a com-
parison of eastern and central Pacific length-
frequency distributions, movements, prediction
indices, and subpopulation structure. This sec-
tion is, in effect, an appendix to the hypotheses
section, but is presented here so that the reader
can readily examine the background material
before encountering a discussion of the various
hypotheses.

Distribution of Spawning

Postlarval skipjack and advanced gonad indices
have been used as evidence of skipjack spawning.
Gonad indices are based on the relative size of
the largest ova or the ratio of body weight to
gonad weight. These indices are difficult to in-
terpret, because running ripe skipjack are prac-
tically never taken; therefore, gonad or egg size
indices only indicate maturity up to the level at
which prespawning fish cease to be available to
sampling. The time elapsing between cessation
of availability and actual spawning is not known.

The distribution of the earliest identifiable
larvae or postlarvae may be a better index of the
temporal-spatial distribution of spawning than
are gonad indices. This belief is based on the
reasonable assumption that larvae and even post-
larval skipjack have a much smaller degree of
mobility than the adults and on a less defensible
assumption that mortality in skipjack at these
stages is approximately uniform with respect to
time of year and location.

Evidence, based on the occurrence of larvae
and juveniles, that skipjack spawn during a wide
range of months over the enormity of the tropical
Pacific should be viewed with caution. It is nec-
essary to determine whether the capture of skip-
jack larvae or juveniles in incidental numbers
in certain areas indicates that the spawning which
occurs in those areas is great enough to make a
measurable contribution to the exploited stocks.

Spatial distribution of larvae. --The distribu-
tions of skipjack larvae in the eastern and central
Pacific are different, because larvae are taken
in relatively high numbers in the central Pacific
with 1-m. plankton nets, but almost none are
caught with similar gear in the eastern Pacific.
Klawe (1963) has undertaken the most complete
study of the distribution of tuna larvae in the
eastern Pacific Ocean. He examined a large
number of plankton tows, representing samples
taken by several types of gear at various depths.
Only a few larval (about 0.02 larvae per tow)
and juvenile skipjack were found. Klawe (p. 466)
concluded that for skipjack there is "...very
limited spawning in coastal and oceanic waters
of the area of our study. Young forms /werej
collected only on a few occasions, off Central
America and as far north as the Gulf of Tehuan-
tepec. Young /we r-e_7 also collected off Ecuador."
Since only a few skipjack larvae were taken in
the eastern Pacific, no temporal patterns were
ascertained from the data.

The relative absence of skipjack larvae in
Klawe's data concurs with Matsumoto's (1958:
p. 59) figure 31, which shows a drop in skipjack

o

larvae per 1,000 m. of water strained in the
vicinity of long. 120° W. as compared with areas
to the west.

With regard to latitudinal distribution of skip-
jack larvae, Matsumoto's (1958: p. 58) figure 30
shows apparently high densities of skipjack lar-
vae in the equatorial region, with a reduction in
numbers of larvae per 1,000 m. of water strained
as one proceeds northward from the Equator.
Data in the files of the Bureau of Commercial
F isheries Biological Laboratory at Honolulu, the
appendices of Matsumoto (1958) and Strasburg
(1960), and Matsumoto's (1958) figure 33, for
example, suggest that the Hawaiian Islands re-
gion, particularly in the summer, is also a locus
of high densities of larval skipjack.

Temporal distribution of larvae. --The tem-
poral distribution of tuna larvae in the central
Pacific is known from Matsumoto (1958: p. 64),
Nakamura and Matsumoto^, and miscellaneous
data in the files of the Bureau's Biological Labo-
ratory at Honolulu. It appears from these data
that skipjack spawn the year-round in the equa-
torial region, with peak numbers of larvae per
sample from May to September. In the Hawaiian
Islands, spawning occurs primarily during the
northern summer, whereas in the Marquesas,

2/

Nakamura, Eugene L., and Walter M. Matsumoto.

Manuscript. Distribution of larval tuna in Mar-

quesan waters. Bureau of Commercial Fisheries

Biological Laboratory, Honolulu.

spawning occurs primarily during the northern
winter.

Gonad indices.--Orange (1961) considered
skipjack gonad indices for the several statistical
areas defined by lATTC. In the northernmost of
these areas, off Baja California, skipjack up to
73 cm. were examined; all of these fish had
gonads in latent or resting states. Interestingly,
the most nearly mature females in all areas of
the eastern Pacific covered by that study were
found just a short distance from Baja California,
in the Revillagigedo Islands. Another area that
yielded fish in a comparatively high state of
maturity (although not as nearly "mature" as
the fish from the Revillagigedo region) was the
Cocos Island region. This region and the Revilla-
gigedos are the only places in the eastern Pacific
where relatively high proportions of maturing
skipjack were found. None of the coastal areas,
from Baja California to Peru, have yielded skip-
jack with gonads in as advanced a state of matu-
rity, even though many fish taken in these locali-
ties are of a size (usually over 40 cm.) at which
full maturity and spawning are possible.
Orange's data are summarized in figure 3, which
shows the percentage of females classified as
"maturing" for the various areas that contained
maturing fish. Orange shows that the offshore
insular areas, the Revillagigedos and Cocos
Island, have higher percentages of mature
females than areas nearer to the coast. A much
larger percentage of fish in the Revillagigedos
mature above 68 cm. than at smaller lengths,
but fish of this size are unusual.

The Revillagigedo area also is unique in that
it exhibits a peak incidence of maturity during
the second, third, and fourth quarters, whereas
all the other areas considered by Orange (coasts
of Mexico, Central America, Colombia. Ecuador.
Galapagos, and Cocos Island) show tendencies
toward winter maturity.

As Orange (1961) points out, the minimum size
at which "spawning" occurs in the Revillagigedos
is 55 cm., which is smaller than some of the
fishfoundin a condition of latent maturity in the
Baja California fishery. Orange's use of the
word "spawning'' is unfortunate, because the
developing gonads described by him were not in
such an advanced state as to be considered per se
indicators of imminent spawning; the amount of
time required for these fish to become running
ripe is not known.

The concept that skipjack in imminent spawn-
ing condition are unavailable to the usual meth-
ods of capture in any of the fisheries operating
in the spawning areas is based on the difference
in ova diameters: between ova of running ripe
skipjack which are rarely taken, and those ot
fish with mature but not running ripe gonads.
Brock (1954: p. 100) , for example, points out that
the larger ova from maturing skipjack taken in
the Hawaiian fishery range between 0.4 and 0.9
mm. in diameter with an average maximum modal
size of 0,7 mm. He indicates, however, that the
single running ripe skipjack that was available
for his examination had ova with an average
diameter of 1.125 mm. Brock's observations on
mature ova diameter concur with those of Raju
(1963) and Bunag (1956). Brock's observations
on ova diameter from running ripe skipjack con-
cur with Yoshida^. The time interval required
for a 0.7 mm. skipjack ovum--which is 60 per-
cent of the attainable spawning diameter or 20
percent of the attainable spawning volume--to
reach extrusion size is unknown. The possibility
that small fractions of the total skipjack egg com -
plement become ripe and are extruded over long
periods of time seems contrary to Brock's ob-
servation of a ripe female where "...a gelatinous
mass of eggs came foaming out of the visceral
cavity in sufficient quantity to make a fair double
handful." It thus is evident that skipjack are
unavailable to the usual methods of capture for
some interval prior to their spawning.

QUAKTKR OF VF.VR

Figure 3. --Percentage of females classified as
maturing during each quarter of the year
from various areas of the eastern Pacific.
(Data are from Orange, 1961.)

_3/ Yoshida, Howard 0. Skipjack tuna spawning
in the Marquesas Islands and Tuamotu Archipelago.
Bureau of Commercial Fisheries Biological Labora-
tory, Honolulu. (In press.)

Length-Frequency Distributions
of Hawaii Samples

Use of length-frequency distributions to deter-
mine age and population structure of skipjack is
made difficult by the lack of ancillary aging meth-
ods. The problem is further complicated by the
skipjack tuna's presumably extensive migrations
and by fluctuations in availability which suggest
that the length-frequency distributions may not
be representative of the populations that are
sampled.

Length-frequency samples were obtained by
systematic sampling from Hawaiian skipjack
landings during various months from 1948 to
1954 and again from 1959 to 1963. For economy
of space, data for 1952-63 only are presented in

figure 4 as percent frequency. These percent
frequencies are based on many samples and
measurements for most months, so chance fluc-
tuations are likely to be small. These frequen-
cies are intended to show the size distribution
of skipjack for each month and should not be
construed as representing the relative magni-
tudes of the catches among the various months.

An examination of the length-frequency distri-
butions shows that during the winter three domi-
nant modal groups may be evident. Although dur-
ing the summer there are typically two modal
groups, on occasions such as in August 1959 and
June 1961, three were noted. Typical modal
lengths for the winter are 35, 50, and 70 cm. (the
modal lengths given here and elsewhere in the
text, unless specified, give the approximate mod-
al values for these typical groups) . while those for

40 60 80 100 40 60 80 100
FORK LENGTH (CM I

40 60 80 100

Figure 4. --Percent frequency distributions of skipjack samples
caught by the Hawaiian fishery during various months.

the summer typically are 45 and 70 cm. Modes
also occur in intermediate positions. The 35-cm.
group does not always appear in the catches; the
cannery which receives the bulk o£ the Hawaii
catch has established a policy which discourages
fishermen from taking fish of this size. The
same length-frequency pattern persists for sev-
eral months in many years; for example. May
through July of 1954 showed relatively small
proportionsof fish smaller than 60 cm., whereas
summer samples for 1952 contained relatively
large proportions of fish under 60 cm. The con-
sistency of the frequency distributions for some
adjacent months is a further suggestion that they
are representative of the population as it is sam-
pled by the fishery.

Another interestingfeature of the distributions
is the occasional occurrence of a modal group
in the April-September samples between the 45-
and 70-cm. groups. Definite examples of this
group are found, for instance, in the August 1959
distributions. This intermediate group also is
evident in a less pronounced degree in other
summer samples. The occurrence of this group
probably is not due to chance because the samples
are large and the group is evident in some cases
for at least 2 months.

Modal sizes. --The modal sizes for the domi-
nant modal groups were determined from length-
frequency distributions (fig. 4). The determina-
tion of which modal groups were dominant was
subjective but based on a conservative criterion
because only prominent modes well separated
from one another were selected to the exclusion
of minor irregularities in the frequency profile.
These selected modal points were plotted in
figure 5 for each month and year for which data
were available. It should be emphasized that
other modal groups are evident and probably
real, but they were not considered dominant and
are therefore not included in figure 5.

Several characteristics of figure 5 are evident.
Most months contain two modes, other months
three or four modes. Certain years (e.g., 1950,
1959) have typically three modes for many
months. The intermediate mode (about 60 cm.)
is most common during the nonsummer months.

The positions of the modes occur at about
45 cm., 60 cm., and 70 cm. Modal lengths for
the 45-cm. group, which is presumed to be the
fastest growing of the various modal groups, do
not necessarily increase with time; in fact there
are some months when the modal length stays
the same or even decreases. There is a tendency
for modal lengths to follow a U-shaped curve
with a minimum in length occurring during the

summer. The lengths of the 70-cm. group seem
to have no consistent pattern except that there
has been a trend in some years for these fish to
increase in size. It is evident that the month-to-
month change of modal sizes in many instances
cannot be representative of growth; these data
suggest passage of successive groups through the
fishery. The varying rate of change in modal
length suggests that this passage is not uniform
with respect to time.

7T

•>. '

1950 1951 1952 1953 1954 1959

1961 1962 1963

Figure 5. --Size distribution of dominant modes
in the length- frequency samples of skipjack
taken during various months from the Hawaiian
fishery. The shaded portion shows the sum-
mer peak fishing period.

Size distribution. — The proportions of skipjack
in the length frequencies (fig. 4) over 60 and
70 cm. are plotted in figure 6. The proportion
of skipjack over 60 cm. is shown invariably to
exceed the 50th percentile during the summer.
The total yearly catch is weighted toward sum-
mer catches, since a large proportion of the
total landings is taken during the summer. Skip-
jack larger than 60 cm. are often less available
in the presummer and postsummer periods.
Occasionally fish larger than 60 cm. reach a
peak in availability in December, January, or
February; this peak occurs most often near
February.

Proportions of fish larger than 70 cm. also
are plotted in figure 6, and they likewise show a
peak for most summer months. In some years,
such as 1960, the proportion of fish over 60 cm.
is a fairly close parallel to the proportion of fish
over 70 cm. In other years, such as 1951, 1953,
1954, and 1959, there are relatively large devia-
tions between the proportions of fish over 60 and
70 cm.; in these years the commercial catch
was relatively large, indicating that the peak

100

(8.4)

(9.9)

(9.5)

(12.9)

(7.3)

(12.1)

Figure &. --Proportions of skipjack greater than 60 and 70 cm. long, from the length-
frequency samples taken during various months by the Hawaiian fishery. Total catch in
millions of pounds accompanies each panel. The shaded portion shows the summer peak
fishing period.

catches are correlated with relatively high pro-
portions of fish between 60 and 70 cm. The oc-
casional influx of large fish during the nonsum-
mer period consists of fish larger than 70 cm.

Length -Frequency Distributions
of Eastern Pacific Samples

Length-frequency distributions for various
regions in the northern-eastern Pacific fishery
area, compiled by quarter years, are available
in Broadhead and Barrett (1964). The most com-
plete set of these data is from the coastal area
of Baja California and the Revillagigedo area.
These quarterly length-frequencies have been
superimposed for each year in figure 7.

Figure 7 shows that an increase in size is not
always evident from quarter to quarter: size may
decrease, remain stable, or increase from one
quarter to the next. A decrease in size during a
subsequent quarter-year can be interpreted as an
influx of smaller size fish replacing the larger
fish of the previous quarter or the removal of
the latter by emigration or mortality. Stability
in size could be interpreted as a lack of growth,
a balance of growth and removal of fish as they
become larger . or a constant replacement of one
group by another of equivalent size in any one
quarter. An increase in size between quarters
may be due to growth of the present inhabitants
of an area, to the influx of large fish, or the re-
placement of smaller size fish by larger size
fish.

The data of figure 7 were examined to deter-
mine the nature of the predominant shift in size
between quarters. The between-quarter differ-
ences in size were classified either as an in-
crease, a stability, or a decrease in size. These
classifications, which, of course, are subjective,
are presented in table 1. They show that in-
creases occur less frequently than decreases or
stability. Many of the increases occur between
the third and fourth quarters. The Baja Califor-
nia samples have more increases.

A pattern seems evident for the Baja California
samples in that the fewest increases in size occur
between the second and third quarters. .Also for
some sets of years, the fourth-quarter size dis-
tribution seems related to the first-quarter size
distribution of the next subsequent year. Fur-
thermore, the data tend to indicate that the sizes
are more similar within years, particularly for
the second, third, and fourth quarters, than be-
tween years.

There are also differences in skipjack size
among the areas of the eastern Pacific. Broad-
head and Barrett (1964) show that small skipjack
(modal length < 50 cm.) are predominant in the
Baja California area, large fish (modal size
> 60 cm.) are most common in the Gulf of Cali-
fornia and the Mexican coast at least in the first
quarter of the year, and finally fish of interme-
diate size are found in the Revillagigedo area.
However, the size composition of the Revilla-
gigedo samples has shifted relative to the Baja
California samples. From 1958 to 1960 the sizes

1ST QUARTER 2D QUARTER

•3D QUARTER

4TH QUARTER

BAJA CALIFORNIA

REVILLAGI6ED0 ISLANDS

600
LENGTH (MM)

600

LENGTH (MM)

Figure 7.--Superimposition of quarter-year length- frequency distribu-
tions of skipjack taken in the Baja California and Revillagigedo
regions of the eastern Pacific during 1954-60. Data are from Broad-
head and Barrett (1964).

of the Revillagigedo and the Baja California skip-
jack are about the same. From 1955 to 1957 the
Baja California fish are generally smaller than
those from Revillagigedo. It seems that the dis-
appearance of the size differential is due to a
decrease in the size of the Revillagigedo fish
rather than to an increase in the size of Baja
California fish.

Comparison of Eastern and
Central Pacific Length -Frequencies

A comparison between the sizes of skipjack
taken in Hawaiian waters and those from the
eastern Pacific is also of interest. Data for var-
ious regions of the northern area of the eastern
Pacific arranged by quarters are available in
Broadhead and Barrett (1964). Data from both

stability or decreases in size

Table 1. --Between quarter Increases in size (+) and ^^-^^^j.^^ „^ „^^i

(-) for major portion of skipjack length distributions sampled from fish taken in
the Baja California and Revillagigedo region of the eastern Pacific. Blank spaces
indicate no data. These changes in size are interpreted from data presented by
Broadhead and Barrett (1964)

Year

1954

1955

1956

1957

1958

1959

1960

Quarters

1
1
2

2

1

3

3

4

4
1

1

1
1
2

2
1
3

3
1
4

4
1

1

1
1

2

2
3

3

4
1

1

1
2

2
1
3

3

1
4

4
1

1

1
1
2

2
1
3

3

1
4

4
1
1

1
1
2

2
1
3

3

1
4

4

1
1

1
1

2

2

3

3

1
4

4

1
1

Baja California

-

-

+

+

+

-

-

+

+

-

-

+

-

+

-

-

-

+

-

-

+

-

-

+

Revillagigedo Islands

+

-

-

+

+

-

-

-

+

-

-

-

+

-

-

+

-

-

+

-

the eastern Pacific and Hawaiian fisheries for
only 3 years (1954, 1959, 1960) have been pub-
lished. As previously noted the most complete
set of data from the eastern Pacific is from Baja
California and Revillagigedo samples. Since the
fairly consistent relationship of sizes between
areas of the northern-eastern Pacific already
has been indicated, it will suffice to compare the
Baja California size distribution with that of the
Hawaiian fishery. To place the material on a
com par able basis with that of the eastern Pacific,
the Hawaiian data have been compiled by quarter-
years and plotted on an increased scale (fig. 8).
These composite curves are, of course, weighted
by the number of fish in each size for each month
or subregion included in the sample. The large

600 700

LENGTH (MM )

Figure 8. --Length- frequency distributions of
skipjack taken during the third quarter of
the year from fisheries off Baja California
and the Hawaiian Islands. Note that the
Hawaiian length- frequencies are plotted on
an increased scale. Data on Baja California
samples are taken from Broadhead and Barrett
(1964).

number of samples and measurements suggests
that the curves are fairly representative of
lengths in the population for the considered time
period and that they have an accuracy suitable
to illustrate the points to be discussed.

In general the curves for the Hawaiian samples
arebimodal; those for the Baja California sam-
ples for the most part have a single mode. The
connotations of unimodal and bimodal, in this
case, refer to the general appearance of the
curves and are not intended to exclude the possi-
bility of more than one modal group within a
particular mode.

Discrepancies in the length-frequency distri-
butions, during the various quarters, between
Hawaiian and Baja California samples may be
attributed to a variety of causes. Some of these
causes are probably related to differences in the
time sequence--between Hawaii and the eastern
Pacific--of the previously mentioned phenomena
which affect stability in size (i.e. , balances among
growth, immigration, and emigration). Of par-
ticular interest are the third quarter length-
frequencies because these generally reflect peak
catches in the Hawaiian and eastern Pacific. It
is likely that they, more than the samples at
other times of the year, most accurately repre-
sent the total population.

The data for 1960 and 1954 show remarkable
similarities between the eastern Pacific and the
small-size Hawaiian fish. The minimum sizes
of fish from both fisheries are about the same,
and their distributions appear to have several
parallelisms in frequency. The 1959 distribu-
tions are, however, quite different. It would
seem, at least in some years, that the Hawaiian
and Baja California-Revillagigedo fish have com-
mon elements in size-frequency distribution.

Movements

This section considers skipjack movements as
inferred from tagging studies conducted by the
lATTC. These studies have yielded many short
distance recoveries and a total, to date, of three
midocean recoveries.

Tagging in the eastern Pacific. --Most skipjack
tagging in the eastern Pacific has been under-
taken by the lATTC. Tag recoveries from skip-
jack tagged in the eastern Pacific show essential-
ly two patterns of movement: a pattern that
seems typical for skipjack tagged to the north of
the Gulf of Tehuantepec and another pattern for
those to the south of this area. Interchange of
tagged fish in the north-south direction across
the Gulf of Tehuantepec (lat. 15° N.) has not yet
been reported.

In the northern region, studies by the Califor-
nia Division of Fish and Game (Blunt and Messer-
smith, 1960) have indicated, with limited data,
that there is northward movement of skipjack in
May and June, northward and southward move-
ment in July, and southerly movement in Septem-
ber. Schaefer, Chatwin, and Broadhead (1961),
with large numbers of tag returns, have shown
there are essentially no interarea movements
among the lATTC statistical areas north of the
Gulf of Tehuantepec.

In the southern region, there appears to be
relatively greater coastwise movement than in
the northern region. Blunt and Messersmith
(1960) have shown interchange between the Four-
teen Fathom Bank (long. 80° W., lat. 9° S.) and
the Gulf of Guayaquil (long. 80° W., lat. 3° S.).
More detailed studies by lATTC (Schaefer et al.,
1961) have shown interchange among the regions
off Colombia, Ecuador, and Peru. The fish
seemed to move toward the Fourteen Fathom
Bank in the summer and away from it in the
winter. These authors also give data which indi-
cate that skipjack marked off Central America
may be recovered in South American fisheries.

Long distance tag recoveries. --To date three
eastern Pacific tagged skipjack have been recov-
ered in the central Pacific (fig. 9). The two re-
coveries in Hawaiian waters were of fish tagged
in the spring and fall. The skipjack recovered
in equatorial waters was tagged in the winter,
was at large for a shorter period of time, and
was smaller than the two tagged fish recovered
in Hawaiian waters.

These three recoveries present problems of
interpretation. How valid are inferences regard-
ing the movements of large numbers of fish based
on a few tag returns? To provide some tangibility
for this question, the probabilities of recovering
a skipjack from a region different from that in
which it was tagged should be considered. This,
however , requires a discussion of various satel-
lite problems such as the distribution of fishing
effort on tagged fish, tagging mortalities, the
distribution of tagged fish among untagged fish.

etc. Many aspects of these problems have been
discussed in lATTC publications (e.g., Schaefer
etal., 1961, Barrett and Connor, 1962). I cannot
evaluate all aspects of these problems. Further-
more, many of the data necessary for such an
evaluation are unattainable with available tech-
niques. Therefore, in this study, the probabili-
ties of capturing a tagged interregional-migrant
skipjack can only be appraised in a gross and
qualitative fashion.

LONG RANGE TAG UCOVERIES OF SKIPJACK

■^SAN fX.4NCJSC0
~J.OS ANGELES

JUNE 19U
77 CM.

«UG.1962
78 CM.

j .-i--SffI.

*♦••

APR. 1H3 y
70 CM.^-

160- "CHRISTMAS!

Figure 9. --Schematic representation of three
long-distance recoveries of tagged skipjack
showing time of tagging, time out, time of
recovery, and size. These fish were tagged
by lATTC.

The likelihood of recovering a tagged interre-
gional-migrant skipjack is dependent on, among
other things, the amount of effective fishing effort
expended for skipjack in the region of potential
recapture. A fairly intensive (for the central
Pacific) skipjackfishery exists in the immediate
vicinity of the Hawaiian Islands; in other areas
of the central Pacific, however, relatively few
skipjack are taken. Therefore, the recovery of
an eastern Pacific tagged skipjack in the equa-
torial central Pacific may appear more likely to
represent a large movement of fish than the two
Hawaii recoveries. Of the several thousand
skipjack tagged in Hawaiian waters during 1957
and 1958 none have been recovered in the eastern
Pacific fishery.

Prediction of Hawaiian Skipjack Catch

Seckel and Waldron (1960) have derived an
index which predicts the magnitude of the annual
catch of skipjack tuna in the Hawaiian fishery.
In essence it forecasts the magnitude of the sum-
mer catch (June, July, and August) because a
considerable proportion of the Hawaiian catch is
taken during the summer. The index itself is
based on the time during January, February, or
March when the sea-surface temperature off
Oahu, Hawaii, changes from cooling to warming.
The prediction states that the annual skipjack
catch will be above average when the shift from

10

cooling to warming occurs early in the January-
February-March period; when this shift is late
in the period catches will be below average.

The quality of the predictive index bears an
important relation to the biology of the skipjack.
For example, if the time of warming index ac-
counts for a high degree of variability in catch
(in the Hawaiian fishery, catch is closely corre-
lated with catch-per-unit-of-effort, R. Uchida,
personal communication), then variations in
catch are probably related to features of the en-
vironment that operate not more than several
months prior to the summer peak fishing period.
If this is the case it is unlikely that year class
strength, for example, would affect the catch
since year class strength of the fishable stock
is probably fixed at least several months prior
to the period of peak fishing. As the degree of
variability in catch, accounted for by its regres-
sion on time of warming, is decreased, the pos-
sible operation several months prior to the sum-
mer peak fishing period of mechanisms unrelated
to the time of warming become increasingly
apparent.

It seemed that a statistical evaluation of the
relation between sea-surface warming time and
skipjack catch would be appropriate. For this
analysis sea-surface temperatures taken about
every other week off Koko Head, Hawaii, were
used. Least squares parabolas were fitted to
the temperature curves for the period between
December 1 and May 31 for the years for which
data were available (1956-63). The time of shift
from cooling to warming was obtained by taking
the first derivative of each parabola, setting the
derivative equal to zero, and solving to obtain
the minimum temperature. The point on the
abscissa where temperature is at a minimum is,
of course, the time of shift from cooling to warm-
ing.

The correlation coefficient between time of
warming and total catch was 0.664, which is non-
significant at the 5-percent level with six degrees
of freedom. This neither confirms nor denies
the existence of this relation, but if a relation
does exist the available degrees of freedom are
not sufficient to demonstrate significance. Let
us assume, however, that two more degrees of
freedom are acquired and that the relation is
adjudged significant with a correlation coefficient
of 0.664. We would then say that 44 percent of
the variation in catch is accounted for by time of
warming; the other 56 percent is unexplained and
could be due to events which are uncorrelated
with time of warming such as inter alia year
class strength and size of fish.

Subpopulations

Evidence for the existence of subpopulations of
skipjack in the Pacific has been advanced by
Sprague and HoUoway (1962) and Sprague, Hollo-
way and Nakashima (1963). This evidence is
based on the existence of immunogenetically sep-
arable groups of skipjack. The location of these
subpopulations is charted by Sprague (1963).
His figure shows two subpopulations (I and II) in
the Hawaiian region, one (III) in the equatorial
region, three in the South Pacific (IV, V, VI),
and one (VII) in Micronesia. A cursory examina-
tion of subpopulation data shows no grossly ap-
parent relation between subpopulations I and II
and size of fish and time of year for the Hawaiian
samples. Attention is called to Hennemuth's
(1959) paper which shows morphometric differ-
ences in skipjack among areas within and between
the eastern and central Pacific.

HYPOTHESES

In this section the material on spawning, size
distribution, movements, and gonad indices is
synthesized into a set of hypotheses concerning
the origin of the skipjack exploited in the eastern
Pacific, their mode of entrance into the eastern
Pacific, the variables associated with their de-
parture from the eastern Pacific, and the origin
of the fish exploited in Hawaii. I am not the first
to present some of the ideas on which these hy-
potheses are based, viz., "...it appears that some
of the west coast population /of skipjack/ range
far to the westward" (Schaefer, 1963: p. 50) and
"these /tag/ recoveries are strong support for
the hypothesis, which we have frequently men-
tioned, that the skipjack of the eastern Pacific
may undertake long offshore-inshore migra-
tions" (Schaefer, 1963: p. 57). These allusions
are similar in spirit to some of the hypotheses
entertained in this paper, but it will be seen that
they are somewhat different in concept.

The phenomena which the hypotheses of this
paper attempt to account for are undoubtedly
complex, and for this reason I suspect that in
some cases the explanations offered are over-
simplifications. I hope that the simplified pic-
ture presented here will generate further ques-
tions on the biology of the skipjack tuna.

Skipjack Exploited in Eastern Pacific

It appears as though skipjack tuna that are
taken in the eastern Pacific Ocean are generated
from spawnings in the central Pacific Ocean.
This concept is based on the distribution of skip-
jack larvae, which indicates apparently negligible
spawning (relative to the numbers of skipjack

11

spawned in the central Pacific) to the east of
about long. 120° W. It could be asserted that this
negligible spawning in the eastern Pacific, never-
theless, contributes a large recruitment to the
eastern Pacific fishery. This, however, seems
unlikely; therefore, I conclude that those skipjack
taken in the eastern Pacific fisheries are immi-
grants from the central Pacific. The size of
skipjack that immigrate into the eastern Pacific
is not certain, but some of the fish may weigh
less than 4 pounds (Schaefer 1961: p. 67-69).

A consideration of skipjack tuna in the central
Pacific suggests that these fish do not comprise
a single, homogeneous, population unit. This
suggestion is derived from at least two lines of
evidence. First, immunogenetic studies by
Sprague (1963) demonstrate the existence of sev-
eral subpopulations in the central Pacific. Sec-
ondly, a cursory examination of the distribution
of skipjack larvae shows their densities are not
uniformly distributed in space or time through-
out the central Pacific. The lack of uniformity
suggests the existence of potential isolating
mechanisms or the possibility of several spawn-
ing groups implying the existence of several sub-
populations.

Despite a poor definition of the mechanisms
that isolate the various subpopulations of skip-
jack in the central Pacific, the spatial-temporal
distributionof larvae and the results of immuno-
genetic studies suggest that the central Pacific
can be considered, from a hypothetical point of
view, to consist of three possible zones of origin
for those apparently adolescent skipjack caught
in the northern- and southern-eastern Pacific
fisheries. These, in a very broad sense, are the
Hawaiian Zone, the Line Islands or Equatorial
Zone, and the Marquesas Zone.

Our consideration of the origin of those skip-
jack caught in the eastern Pacific will be limited
to the Hawaiian and Equatorial Zones. This limi-
tation results from the speculation that large
numbers of fish from the Marquesas Zone do not
enter the eastern Pacific fishery area. (The
speculation is perhaps more likely for skipjack
that originate in the Tuamotus than for skipjack
that originate in the Marquesas Islands per se.)
Three observations support the speculation.
First it is likely that those skipjack in the South-
ern Hemisphere exhibit the same apparent tend-
ency toward net south-poleward movement (which
would reach maximum southward limits in the
northern winter) as is found for the apparently
net north-poleward movement in the Northern
Hemisphere. In order for Marquesas Zone skip-
jack to enter the Central and South American
fishery areas they would have to exhibit at least a

slight northern component in movement and this
seems unlikely for large numbers of these fish.
The second observation concerns reports of ex-
tremely large skipjack taken in the Society Is-
lands. Since these reports had not been verified
an attempt was made to secure some length-
frequency distributions from Tahiti. Thirty-
three lengths were obtained. These lengths were
compared with 125,105 length measurements
made during the last several years on samples
from the Hawaiian fishery. Only in an extremely
minute possibility could the Tahitian and Hawai-
ian samples be drawn from the same population
of lengths. The Tahitian sample contained fish
several centimeters, on the average, larger than
the Hawaiian samples. This could result from a
variety of phenomena. One possibility is that the
Tahitianfish are unexploited relative to the Ha-
waiian fish. I will postulate later in this paper
that large components of the Hawaiian fish do
not originate in the Hawaiian Islands, but else-
where— probably in the equatorial central Pa-
cific. Now, if the Hawaiian lengths are repre-
sentative of equatorial central Pacific lengths
then it could be inferred that the Tahitian or
Marquesas Zone fish are fished to a less degree
than the Hawaiian component of central Pacific
fish. Depending on the relative magnitudes of
fishing Intensity and stock sizes etc., this might
indicate that the Marquesas fish do not enter the
eastern Pacific. Thirdly, it is again noted that
skipjack in the Marquesas Zone comprise differ-
ent subpopulation(s) than those of the rest of the
central Pacific and, while it is not necessary, it
is likely that their migratory behavior is differ-
ent than the subpopulation(s) to the north. If it is
different and the subpopulation(s) to the north
enter the eastern Pacific, then the subpopula-
tion(s) of the Marquesas Zone do not. These
three observations then, taken together, have
generated the speculation concerning the Mar-
quesas Zone fish.

Since, by hypothesis, the Marquesas Zone does
not contribute significant numbers of individuals
to the skipjack taken in the eastern Pacific, it
becomes appropriate to examine the relative
likelihoods of Hawaiian or Equatorial Zones as
points of their origin for the eastern Pacific
skipjack. These likelihoods are reckoned from
inferences drawn from composite length-fre-
quency distributions, changes in length-frequen-
cy distribution and gonad maturity within the
eastern Pacific fishery areas, and from tagging
studies.

A first inference concerning the origin of east-
ern Pacific skipjack is based on the spawning
destination of eastern Pacific emigrants. If it
canbe established that emigration is associated

12

with spawning, then the destination of the spawn-
ing subpopulations should, on the average, be
coincident with the zone of their natality. The
concept of spawning-associated emigration of
eastern Pacific fish is based on the relative im-
maturity of slcipjack in the eastern Pacific, an
apparent reduction in the number of larger skip-
jack in the eastern Pacific fishery, the fact that
the onset of advanced maturity In the eastern
Pacific area Is correlated with attainment of
large size, an incidence of more nearly mature
fish near the offshore islands than near the
American coasts, and an apparently negligible
spawning in the eastern Pacific.

As for their spawning destination it is perhaps
significant that skipjack at the northern and
southern extremes of their range in the eastern
Pacific have gonads in a resting stage of maturi-
ty. Also, there is a tendency toward an extensive
period of summer maturity in the Revillagigedos
and winter maturity south of the Revillagigedos
to Cocos Island. This implies that, depending on
the time Interval from an advanced stage of ma-
turity to spawning, skipjack in the Revillagigedos
spawn well into the fall. As the time of spawn-
ing proceeds from summer through fall to winter,
the northern limit of this activity in the central
Pacific moves south. In other words, in the
northern fall there is a tendency toward more
equatorward spawning than Hawaiian-Zone
spawning. For the winter-maturing fish south
of the Revillagigedos. spawning must take place
south of or in the Equatorial Zone, because there
is almost no winter spawning in the Hawaiian
Zone (Brock, 1954, indicates that Hawaiian-Zone
spawning ends in September). Skipjack south of
the Revillagigedos to Cocos Island, however, are
less mature than those near the Revillagigedos
proper. This might indicate a northward move-
ment of skipjack along the southern Mexican
coast, but tagging data have shown little inter-
change between these areas. Furthermore, those
found north of the Gulf of Tehuantepec in the
eastern Pacific may be composed of two contin-
gents: one contingent exhibits summer maturity
and may spawn in the Hawaiian or Equatorial
Zone; the other exhibits winter maturity and
spawns at least as far south as the Equatorial
Zone.

A second inference, relating to the size dis-
tribution of skipjack which immigrate into the
eastern Pacific, is based on the observation that
Baja California and Revillagigedo length-fre-
quency samples often show no marked increase
In length over several months. If compensatory
or differential growth does not operate, then
these fish must have been spawned over several
months. Since the predominant spawning period

in Hawaiian waters Is less than several months
the Baja California and Revillagigedo fish must
contain a component which is of non-Hawaiian
origin. It seems likely that this component orig-
inates in the equatorial region. A degree of cau-
tion should be retained, however, in considering
the three-quarter year size replacement and the
gonad maturity indices because these indices
might reflect phenomena that represent relative-
ly few fish at either extreme of the three-quarter
year period.

A third inference concerns the observation
that size distributions of skipjack taken in the
third quarter of the year in the northern-eastern
Pacific fishery often exhibit a size component
slightly larger or smaller than the small-size
modal group of Hawaiian fish. If these differ-
ences are not due to sampling errors, then they
might arise from several possibilities. These
might include variations in growth rate, slight
differences in age (of the order of a few months) .
and size specific movements. While there is no
strong evidence to support any of these possi-
bilities, the size range of any ''year class' (bar-
ring compensatory growth) from equatorial wa-
ters would be expected to have more larger and
smaller elements than a "year class" generated
in Hawaiian waters because of the more extended
period of spawning in equatorial waters.

A final inference concerning the relative mer-
its of either the Hawaiian or Equatorial Zone as
a place of birth for the skipjack harvested in the
eastern Pacific is based on the three mid-Pacific
tag recoveries that resulted from extensive tag-
ging in the eastern Pacific. Although it is diffi-
cult to place a great amount of weight on three
tag recoveries, it can be indicated that if skip-
jack emigrating from the eastern Pacific all enter
the Hawaiian Zone, then a greater number of tag
returns should be obtained from the Hawaiian
Islands. The pattern of recoveries, considering
the location of fishing effort, is not inconsistent
with the hypothesis that the major movement of
skipjack in the northeastern portion of the Pa-
cific is usually between the equatorial central
Pacific and the Baja California region.

As mentioned previously a segment of the equa-
torial fish, potential recruits to the eastern Pa-
cific fishery, disperses to the east and north.
As the fish move east they are "split'' into a
northern group that enters the Mexican fishery
area and a southern group that enters the South
American fishery area. The mechanisms by
which the skipjack are split into northern and
southern contingents may be manifold and prob-
ably are not fixed in time or space. The extent
to which the mechanisms partition skipjack Into

13

'■■ " FEB .<^

"S5.. y-'^

Figure 10. --Location of 21° and 28° isotherms in the eastern Pacific according to Wyrtki
(in press). Blackburn (1962) has suggested that the distribution of skipjack harvested
along the coasts of the Americas is coincident with surface temperatures between 21° and
28° centigrade.

northern and southern contingents is a function
of the north-south and temporal distribution of
eastward-moving skipjack as these fish encoun-
ter the splitting mechanisms. One possible split-
ting mechanism is the warm water cell (surface
temperature greater than 28° C.) in the vicinity
of lat. 15° N. off the coast of Central America.
The location of this warm water cell for average
conditions is contained in Wyrtki's^ recent
paper (see fig. 10) and has been discussed by
Blackburn (1962: p. 39-40), who suggested that

it impedes north-south movement of skipjack
across lat. 15° N. to an extent which varies with
surface temperature. This is undoubtedly the
case, but it appears that the cell has a more
important effect; it is one of the mechanisms
that serve to split skipjack into northern and
southern groups.

A/ Wyrtki, Klaus. Thermal structure of
eastern Pacific Ocean. Deut. Hydrograph. Z.
press . )

the
(In

14

Skipjack Taken in Hawaiian Waters

The Hawaiian skipjack fishery uses pole-and-
line gear and live bait. The average annual catch
is about 10 million pounds, fluctuating in recent
years between 7 and 14 million pounds. The bulk
of the catch (usually about 70 percent by weight)
is taken during June, July, and August, and gen-
erally consists of large numbers of "season-
size fish" {>than 60 cm.). "Season-size fish"
also occur sporadically during the winter but
are not nearly as abundant as during the summer
"season." The magnitude of each year's catch
has been predicted for several years (Seckel and
Waldron, 1960). This prediction is based on
changes in sea-surface temperature that occur
in February and March and, in essence, foretells
the magnitude of the catch during the "season"
in June, July, and August. Several problems
arise in interpreting the nature of the fluctua-
tions in the Hawaii catch. A consideration of
these problems generates several hypotheses
pertinent to the origin of those skipjack taken by
the Hawaiian fishery.

Skipjack spawn in the Hawaiian Zone so it is
likely that at least some of the skipjack caught
in the Hawaiian Zone fishery are generated in
this Hawaiian Zone. Immunogenetic studies have
shown that those skipjack caught in the Hawaiian
Zone fishery are comprised of two subpopula-
tions. On this basis I postulate that at least one
of these two subpopulations has its origin in the
Hawaiian Zone. The other subpopulations may
or may not originate in the Hawaiian Zone, but in
any event the two subpopulations must maintain
a degree of Isolation at spawning time.

A wide variety of isolating mechanisms that
tend to reduce interbreeding between populations
of many organisms are described by various
authors (see, for example, the concise resume
by AUee, Emerson, Park. Park, and Schmidt,
1949: p. 606). Some possible isolating mecha-
nisms that could maintain immunogenetic dis-
tinctness between two groups of skipjack are
distance between spawning units, differences in
the time of spawning between breeding units,
island association of spawning, island association
of relatively high larval survival, length segre-
gation at spawning time, and the association of
breeding groups with particular water types.
Both distance and time of spawning could operate
as isolating mechanisms between the Hawaiian
and Marquesas Zones because these Zones are
at the extremes of the north-south spawning
range; spawning in the Hawaiian Zone opcurs
predominantly during the northern summer
whereas spawning in the Marquesas takes place
predominantly during the northern winter. These

mechanisms also probably operate between the
Hawaiian and Equatorial Zones. Included in the
concept of distance as an isolating mechanism is
the apparent island-association of spawning or
high larval survival, suggested by what appears
to be relatively high densities of skipjack around
major island groups.

Another possible isolating mechanism is length
segregation of skipjack schools (Brock, 1954:
p. 98-99). Perhaps at spawning time small dif-
ferences in length, on the average, may tend to
segregate schools of skipjack. Any differences
in length between the two populations would act
as isolating mechanisms. Since present sam-
pling techniques do not capture skipjack that are
spawningor are in an imminent spawning condi-
tion this hypothesis cannot be tested. One final
type of isolating mechanism involves the hypoth-
esis that subpopulations of skipjack are associ-
ated with water types as indicated by tempera-
ture and salinity at the sea surface (Sprague,
1963).

Another characteristic of the skipjack taken
in the Hawaiian Zone that may indicate a non-
Hawaiian originfor some of the fish taken in the
Hawaiian fishery is the roughly equal numbers
of fish in the 45- and 70-cm. modal group. Catch-
es representative of the full range of sizes for
any population should have considerably more
fish in the smaller size groups. Several expla-
nations can be offered for the relatively few
small fish in the Hawaiian catch. Small fish
might not be caught in the Hawaiian pole-and-
line fishery; however, there are no data that
describe this selectivity. A second possibility
is that the large size group of fish contains sev--
eral age groups indistinguishable with respect to
size. This may be partially true, but 1 believe
its extent is insufficient to explain the equal
numbers of large and small fish. A third type of
selectivity may cause mature fish to be unavail-
able to pole-and-line gear during the spawning
season. The unavailability of imminent spawners
already has been pointed out. If the proportion
of larger fish in which spawning is imminent is
smaller than the proportion of smaller fish in
which spawning is imminent, then the numbers
of larger fish in the catch will be greater than
would be expected if there were no differences
in maturation between large and small fish. No
data are available on the extent to which immi-
nent spawning affects the representativeness of
skipjack samples.

Other mechanisms that might produce rela-
tively large numbers of large fish are the possi-
bility of a disproportionate increase in the density
of large fish in the immediate vicinity of the

15

Hawaiian Islands (i.e., the fishery area) or that
equal numbers of large and small fish are pro-
duced by an emigration of small fish rather than
by an immigration of large fish. The latter pos-
sibility is contradicted, however, by the appar-
ently tremendous influx of large fish into the
Hawaiian fishery during the summer. If we as-
sumed negligible fishery selectivity and minimal
effects of maturity-associated size-specific
availability, then the samples indicate there are
roughly equal numbers of large and small fish in
Hawaiian waters during the summer. This prob-
ably is caused by an immigration of large fish
that did not originate in the Hawaiian Zone.

Evidence from subpopulation studies and the
distribution of skipjack lengths then admit the
hypothesis that one of the subpopulations enter-
ing the Hawaiian fishery did not originate in the
Hawaiian Zone and that this subpopulation is com-
prised at least, in part, of large numbers of
season-size skipjack. As previously noted a
prediction has been available for forecasting the
magnitude of the annual catch of skipjack in the
Hawaiian fishery. A large annual catch is an
index of a high apparent abundance of season-
size fish. I have shown that certain years with
relatively large catches, and thus a high apparent
abundance of seasonfish, were characterized by
a reduction in size of the season-size fish. Thus,
in years of relatively poor catches the proportion
of fish over 60 cm. consists mostly of fish over
70 cm. long (fig. 6). In years of relatively large
catches, the season-size fish are mostly between
60 and 70 cm. long and their proportion in the
total catch is apparently independent of the pro-
portion of fish over 70 cm. Therefore, since the
magnitude of the catch could be predicted and
the size distribution is related to the magnitude,
the size distribution also could be estimated in
the sense that in years of high catch the size of
the season-size fish would be slightly reduced.

These predictions imply that the season catch
of skipjack is controlled by events that are asso-
ciated with the warming of the sea surface in the
vicinity of Oahu several months before the onset
of peak fishing. By this reasoning it would seem
that the biomass and size distribution of prere-
cruits to the Hawaiian fishery are oriented in
space and that this orientation is controlled by
the absolute position and rate of movement of
either the isotherms or properties related to the
isotherms.

On the other hand, a study in the degree of
variation in the relation between time of warm-
ing and catch indicated that a large component of
variability in total catch might not be attributable
to the sea-surface time of warming index. There-

fore, the unexplained variation is due to events
other than the time of warming index which is an
evident part of the temporal change in sea-sur-
face temperature and occurs a few months prior
to eachfishing season in the vicinity of Oahu. A
possible component of the heretofore unexplained
variation is the strength of the various year
classes that enter the skipjack fishery. The
formation of large or small year classes of the
large skip jack taken in the Hawaii an fisheryprob-
ably takes place 2 or 3 years prior to the recruit-
ment of this large size group and therefore seems
independent of events, such as the time of warm-
ing, that occur a few months before the fishing
season. The hypothesis that year class strength
influences the summer apparent abundance of
large skipjack is strengthened by the observation
that a small-size component of the large-size
group becomes particularly evident in years of
relatively high catches. It is postulated that this
small-size component represents an unusually
strong year class. If year class strengths are
important components of variability in skipjack
catches, then it is also likely that slight reduc-
tions in average size of the skipjack during years
of high apparent abundance may also reflect den-
sity dependent growth that is typical of large
year classes for many species of fish.

Additional inferences pertinent to the origin of
skipjack taken in the Hawaiian Zone are based
on size distribution and growth data from the
Hawaiian fishery. Growth of skipjack tuna based
on modal progressions within size distributions
of Hawaiian samples is often less, on a month-
to-month basis, than that estimated from tag
returns (Rothschild^/). lnfact,for some months
modes remain the same or decrease in length.
These observations suggest that, as in the east-
ern Pacific, the fishery is constantly being re-
plenished withfish of smaller or equivalent size
while at the same time larger fish leave the
fishery area. During some years there is a
U-shaped distribution of lengths with a minimum
in the summer. This could be interpreted as a
removal of the larger fish of the small size group
during the summer fishery. A second possibility
is that smaller fish enter the fis.iery initially,
then emigration is reduced, the ascending right-
hand limb of the U-shaped distribution reflecting
growth of those that are residual. Thus it is
apparent that all exploited subpopulations of skip-
jack exhibit less than expected increases in
length. It is hypothesized that this characteristic
is due to certain sizes of fish moving through
the fishery.

Rothschild, Brian J. Manuscript. Estimates
of skipjack tuna (Rats uw onus pelamis) growth in
the Hawaiian Islands. Bureau ot Commercial Fish-
eries Biological Laboratory, Honolulu.

16

A comparison of length-frequency distribu-
tions taken during the third quarter of the year
by the Hawaiian and eastern Pacific fishery pro-
vides further clues to the origin of skipjack
caught in the Hawaiian fishery. A similarity in
size distribution exists between the small-size
group of Hawaiian and of eastern Pacific fish
for 2 of 3 years for which data are available.
This similarity in size could be indicative of a
common origin for eastern Pacific and the small-
size group Hawaiian fish. By hypothesis the
eastern Pacific fish originate in the central Pa-
cific; therefore, the concept of a common origin
suggests that those taken in Hawaiian waters
also originate in the equatorial central Pacific.
The lack of apparent similarities between third
quarter length-frequencies of Hawaiian and east-
ern Pacific fish during the third year for which
data have been published is interpreted as an
environmentally induced failure of equatorial
central Pacific skipjack to enter the Hawaiian
Zone during this year.

A consideration of various features of the
biology of the skipjack taken by the Hawaiian
fishery yields no immediately discernible con-
clusion regarding the origin of those skipjack
taken in the Hawaiian Zone. It seems fairly
evident, however, that a likely hypothesis would
suggest that one of the two subpopulations that
enter the Hawaiian fishery originates in equa-
torial waters of the central Pacific and is com-
prised of fish that make up the bulk of the season
catch. The magnitude of the season catch may
be related to the physical environment, but it is
likely that dynamic features of the skipjack popu-
lations are also important.

CONCLUSIONS

The purposes of the hypotheses presented in
this paper were to assemble data pertinent to
understanding the movements of skipjack tuna in
the central and eastern Pacific and to provide a
guide for future studies. The need for more
evidence on the problems of the origin and move-
ment of skipjack tuna is evident. The accumula-
tion of this evidence should be based on a consid-
eration of alternative sets of hypotheses, a sub-
stantiation of the assertions made in this paper,
and the design of critical experiments to test the
hypotheses.

In considering sets of hypotheses other than
those proposed in this paper it is important to
note that the conceptual division of the central
Pacific into three zones of origin puts immediate
constraint upon the nature of solutions. Should
the central Pacific be considered as a single
zone (one huge population of skipjack that extends

from the Hawaiian Islands in the north to the
Tuamotu Archipelago in the south), as two zones
(a northern and southern zone, both or only one
contributing to the eastern Pacific fishery), or
finally as many discrete zones (each contributing
variable proportions of fish to the eastern Pacific
fishery), then the nature of the hypotheses based
on these considerations might be different than
that of the three-zone concept. A consideration
of these alternatives and several others has,
however , continued to suggest that the three-zone
approach presented in this paper is most con-
sonant with the evidence as it is now available.
When new and pertinent data are acquired, then
it is possible that alternatives will be accepted.

Some of the assertions made in this paper
need to be substantiated. Perhaps the weakest
assertion involves the exclusion of Marquesas
Zone fish from the Central and South American
fishery area. The temporal-spatial distribution
of skipjack larvae certainly needs to be studied
more intensively. One of the more perplexing
features of the evidence is the large numbers of
immunogenetic ally distinct groups that have been
identified. What are the isolating mechanisms?

Finally, which experiments provide critical
tests of the hypotheses? In general, studies of
larval and juvenile skipjack are indicated since
these studies may provide clues to the year class
strength phenomena mentioned earlier in this
paper , elucidate isolating mechanisms, and pro-
vide samples for subpopulation studies. More
specifically an experiment to test the hypotheses
is based on the definition of a critical area. This
critical area lies along a line between the Hawai-
ian Islands, the Line Islands, the Marquesas
Islands, and the Tuamotu Archipelago. Juvenile
skipjack (since adult skipjack appear to be un-
available to capture for some interval prior to
spawning; samples of juvenile skipjack are more
likely to be from a single genetic pool than the
adults) should be taken along this critical area,
and their immunogenetic affinities determined
with already developed reagents. The immuno-
genetic affinities of the eastern Pacific adoles-
cent skipjack could then be compared with central
Pacific samples to determine the loci of their
origin. The conduct of this research will pro-
vide added benefits by substantiating various
assertions and providing new information on the
temporal-spatial abundance of larval and juvenile
skipjack.

SUMMARY

A set of hypotheses has been generated to ac-
count for the origin and movement of exploited
skipjack in the eastern and central Pacific Ocean.

17

The hypotheses postulate that most recruits to
the eastern Pacific fishery result from spawning
in the equatorial central Pacific. This recruit-
ment stock remains in the fishery for several
months, begins to attain sexual maturity, and
leaves the eastern Pacific for the equatorial
central Pacific where spawning takes place.
Some of the evidence upon which the hypotheses
are based, and its major features are summa-
rized as follows:

1. It appears that skipjack spawning in the
eastern Pacific Ocean is only incidental whereas
in the central Pacific spawning is relatively in-
tense. This suggests that the skipjack in the
eastern Pacific come from the central Pacific.

2. It was hypothesized from immunogenetic
studies and the lack of uniformity in the time-
space distribution of skipjack spawning that the
skipjack of the central Pacific are not a single
homogeneous population unit.

3. The central Pacific was divided into three
arbitrarily designated, but likely, zones of origin
for skipjack taken in eastern Pacific. These
zones are the Hawaiian Zone, the Equatorial
Zone, and the Marquesas Zone.

4. It was speculated that skipjack from the
Marquesas Zone do not enter the eastern Pacific
fishery areas because an unlikely northward
component would be required for their migratory
path, the Marquesas Zone fish comprise different
subpopulations than any of the other areas stud-
ied, and a size difference between Tahitian and
Hawaiian Zone fish suggests the possibility that
fish from the Marquesas Zone may not undergo
a relatively high degree of exploitation.

5. A consideration of the relative merits of
the Hawaiian and Equatorial Zones as loci for the
origin of eastern Pacific exploited fish involved
establishing that emigration from the eastern
Pacific was spawning-associated. It appears
that skipjack samples in the eastern Pacific ex-
hibit relatively high maturity indices for about
9 months each year. This period extends well
into the fall suggesting an equatorial destination
for the eastern Pacific emigrants.

6. Length-frequency distributions taken from
some regions of the eastern Pacific show quar-
ter-to-quarter changes which may be indicative
of a more-or-less continual movement of skip-
jack through the fishery. It appears that this
size replacement is of a duration such that it is
unlikely that these fish are solely generated in
the Hawaiian Islands, thus again suggesting an
Equatorial Zone origin for a majority of these
fish.

7. There have been three long-distance tag
recoveries to date. All three resulted from fish
tagged oft the Mexican coast. Two were recov-
ered in Hawaiian waters, one in equatorial
waters. When compared with the rest of the

central Pacific, the Hawaiian region maintains a
relatively intense fishery, and therefore the re-
covery in equatorial waters is more likely to
represent a large movement of fish than the two
recoveries in Hawaiian waters.

8. It is postulated that as the adolescent skip-
jack approach the coast of the Americas they are
split into a northern and southern contingent;
the northern contingent is harvested off the
Mexican coast and the southern contingent off
Central and South America.

9. Length-frequency distributions from the
Hawaiian fishery do not exhibit typical trends
such as consistent growth increases; these dis-
tributions do, however, show a consistent pattern
for adjacent months. Only 3 years of data were
available to compare the essentially bimodal
Hawaiian length frequencies with the essentially
unimodal eastern Pacific length frequencies.
For two of the years it appears that skipjack in
the small size Hawaiian group have common
elements with eastern Pacific size frequencies.

10. The observations of equal numbers of
large and small fish in the Hawaiian fishery, a
tremendous seasonality in catch, and the exist-
ence of at least two subpopulations in Hawaiian
waters suggest that there is a component of fish
enteringthe fishery that doesn't originate in the
Hawaiian Zone.

11. A prediction of skipjack tuna catch in the
Hawaiian Islands has been utilized over the past
several years. Large components of variability
are not associated with the availability-associ-
ated predictive index, and it seems likely that
year-class associated phenomena play an im-
portant role in controlling the abundance of skip-
jack in Hawaiian waters.

12. In some years when apparent abundance
in Hawaiian waters is high there is a smaller
size component of fish in the Hawaiian fishery.
This component may be evident as early as Feb-
ruary and may in itself serve as a predictive
index. An implication is that if year class phe-
nomena can be predicted from the Hawaiian
fishery and if the Hawaiian and eastern Pacific
fisheries catch skipjack of common origin then
it should be possible to predict the abundances of
year classes that enter the eastern Pacific fish-
eries.

13. There is some time interval prior to
spawning during which skipjack are unavailable
to capture. This has a bearing on sampling
problems.

14. The hypothetical nature of this work was
emphasized.

15. A critical line for testing the hypotheses
in this paper was established among the Hawaiian
Islands, the Line Islands, and the Marquesas
Islands.

18

ACKNOWLEDGMENTS

Many people have contributed to this study.
John C. Marr of the Bureau of Commercial Fish-
eries, Honolulu, suggested this investigation and
gave advice. Detailed criticism was received
from O. Elton Sette, Bureau of Commercial Fish-
eries Biological Laboratory, Stanford; Vernon
Brock, University of Hawaii, Marine Laboratory;
and Milner B. Schaefer, Scripps Institution of
Oceanography. Valuable suggestions were re-
ceived from David N. Cushing, Fisheries Labo-
ratory, Lowestoft; William F. Royce, Fisheries
Research Institute, University of Washington;
and William E. Ricker, Fisheries Research
Board of Canada. The suggestions and encour-
agement of my colleagues at the Biological Lab-
oratory of the Bureau of Commercial Fisheries
in Honolulu, especially John J. Magnuson, Tamio
Otsu, Gunter R. Seckel, Richard S. Shomura,
Lucian M. Sprague, and Richard N. Uchida, are
appreciated. I am responsible for the arguments
forwarded in this paper and do not wish to infer
that those acknowledged necessarily agree with
my conclusions or methods.

Brock, Vernon E.
1954. Some aspects of the aku, Katsuwonus
pelamis, in the Hawaiian Islands. Pac. Sci.
8(1): 94-104.

Bunag, Daniel M.
1956. Spawning habits of some Philippine tuna
based on diameter measurements of the
ovarian ova. Philippine J. Fish. 4(2): 145-
175.

Hennemuth, Richard C.
1959. Morphometric comparison of skipjack
from the central and eastern tropical Pa-
cific Ocean. Interamer. Trop. Tuna
Comm. Bull. 3(6): 241-304.

Jones, S., and E. G. Silas.
1963. Synopsis of biological data on skipjack
Katsuwonus pelamis (Linnaeus) 1758 (In-
dian Ocean). Proceedings of the World
Scientific Meeting on the Biology of Tunas
and Related Species, 2-14 July 1962, FAG
Fish. Rep. (H. Rosa, Jr., Ed.) 6(2): 663-
694. (Species Synopsis 21.)

LITERATURE CITED

AUee.W. C, A. E. Emerson, O. Park, T. Park,
and K. P. Schmidt.
1949. Principles of animal ecology. W. B.

Saunders Co. Philadelphia and London,

837 p.

Barrett. Izadore, and Anne R. Connor.
1962. Blood lactate in yellowfin tuna, Neo-
thunnus macropterus. and skipjack, Katsu-
wonus pelamis. following capture and tag-
ging. Interamer. Trop. Tuna Comm. , Bull.
6(6): 233-280.

Blackburn. Maurice, and Associates.
1962. Tuna oceanography in the eastern trop-
ical Pacific. U.S. Fish Wildl. Serv., Spec.
Sci. Rep.-- Fisheries 400, 48 p.

Blunt. C. E., Jr., and James D. Messersmith.
1960. Tuna tagging in the eastern tropical Pa-
citic, 1952-59. Calif. Fish Game 46(3):
301-369.

Broadhead, Gordon C. and Izadore Barrett.
1964. Some factors affecting the distribution
and apparent abundance of yellowfin and
skipjack tuna in the eastern Pacific Ocean.
Interamer. Trop. Tuna Comm., Bull. 8(8):
419-473.

Kawasaki, Tsuyoshi.
1955. On the migration and growth of the skip-
jack, Katsuwonus pelamis (Linnaeus), in
thelzuand Bonin Sea areas and the north-
eastern sea area along the Pacific coast of
Japan. Bull. Tohoku Reg. Fish. Res. Lab.
4: 101-119. (In Japanese with English sum-
mary.)

Klawe. Witold L.
1963. Observations on the spawning of four
species of tuna (Neothunnus macropterus,
Katsuwonus pelamis, Auxis thazard and
Euthynnus lineatus) in the eastern Pacific
Ocean, based on the distribution of their
larvae and juveniles. Interamer. Trop.
Tuna Comm., Bull. 6(9): 449-540.

Matsumoto, Walter M.
1958. Description and distribution of larvae
of four species of tuna in central Pacific
waters. U.S. Fish Wildl. Serv., Fish. Bull.
58: 31-72.

Orange, Craig J.
1961. Spawning of yellowfin tuna and skipjack
in the eastern tropical Pacific, as inferred
from studies of gonad development. Inter-
amer. Trop. Tuna Comm., Bull. 5(6): 459-
526.

19

Postel, Emile.
1963. Expose synoptique des donnees biologi-
ques sur la bonite a ventre raye Katsu-
wonus pelamis (Linne) 1758 (Atlantique et
Mediterranee). Proceedings of the World
Scientific Meeting on the Biology of the
Tunas and Related Species, 2-14 July 1962,
FAO Fish. Rep. (H. Rosa, Jr., Ed.) 6 (2):
515-537. (Species Synopsis 15.)

Raju, G.
1963. Spawning of the oceanic skipjack Katsu-
wonus pelamis (Linnaeus) in the Laccadive
Sea. Proceedings of the World Scientific
Meeting on the Biology of Tunas and Re-
lated Species, 2-14 July 1962, FAO Fish.
Rep. (H. Rosa, Jr., Ed.) 6(3): 1669-1682.
(Exper. Pap. 40.)

Schaefer, Milner B.

1961. Appendix A, Report on the investigations
of the Inter-American Tropical Tuna Com-
mission for the year 1960. Interamer.
Trop. Tuna Comm., Annu. Rep. 1960: 40-
107.

1962. Appendix A, Report on the investigations
of the Inter-American Tropical Tuna Com-
mission for the year 1961. Interamer.
Trop. Tuna Comm., Annu. Rep. 1961: 44-
103.

1963. Appendix A, Report on the investigations
of the Inter-American Tropical Tuna Com-
mission for the year 1962. Interamer.
Trop. Tuna Comm., Annu. Rep. 1962: 35-
88.

Schaefer, Milner B., Bruce M. Chatwin, and
Gordon C. Broadhead.
1961. Tagging and recovery of tropical tunas,
1955-1959. Interamer. Trop. Tuna Comm.,
Bull. 5(5): 341-455.

Shapiro, Sidney.
1948. The Japanese tuna fisheries. U.S. Fish
Wildl. Serv., Fish. Leafl. 297. 60 p.

Sprague, Lucian M.
1963. Subpopulations. In: Van Campen, W. G.
(Ed.) Progress in 1961-62, Bureau of
Commercial Fisheries, Hawaii Area, Bio-
logical Laboratory, Honolulu, Hawaii. U.S.
Fish Wildl. Serv., Circ. 163: 10-12.

Sprague, Lucian M., and James R. HoUoway.

1962. Studies of erythrocyte antigens of the
skipjack tuna (Katsuwomus fsicj pelamis).
Am. Natur. 96(889): 233-238.

Sprague, Lucian M., J. R. Holloway, and L. 1.
Nakashima.

1963. Studies of the erythrocyte antigens of
albacore, bigeye, skipjack, and yellowfin
tunas and their use in subpopulation iden-
tification. Proceedings of the World Scien-
tific Meeting on the Biology of Tunas and
Related Species, 2-14 July 1962. FAO Fish.
Rep. (H. Rosa, Jr., Ed.) 6(3): 1381-1393.
(Exper. Pap. 22.)

Strasburg. Donald W.
1960. Estimates of larval tuna abundance in
the central Pacific. U.S. Fish Wildl. Serv.,
Fish. Bull. 60: 231-255.

Waldron, Kenneth D.
1963. Synopsis of biological data on skipjack
Katsuwonus pelamis (Linnaeus) 1758 (Pa-
cific Ocean). Proceedings of the World
Scientific Meeting on the Biology of Tunas
and Related Species, 2-14 July 1962, FAO
Fish. Rep. (H. Rosa, Jr., Ed.) 6(2): 695-
748. (Species Synopsis 22.)

Seckel, Gunter R., and Kenneth D. Waldron.
1960. Oceanography and the Hawaiian skipjack
fishery. Pac. Fisherm., 58(2): 11-13.

ADDENDUM

A recent paper received in our Laboratory on June 1, 1965 (Kawasaki,
T. 1964. Population structure and dynamics of skipjack in the North
Pacific and its adjacent waters. Bull. Tohoku Reg. Fish. Res. Lab.
24:28-47) appears to contain similar conclusions to those in this publi-
cation; the present author has not, however, had the opportunity to
examine Kawasaki's paper in detail.

B.J.R.

MS*1417

20

MBL WHOI Library - Serials

5 W

SE 01638

Created in 1849 the Department of the Interior — a depart-
ment of conservation — is concerned with the management,
conservation, and development of the Nation's water, fish,
wildlife, mineral, forest, and park and recreational re-
sources. It also has major responsibilities for Indian and
Territorial affairs.

As the Nation's principal conservation agency, the De-
partment works to assure that nonrenewable resources are
developed and used wisely, that park and recreational re-
sources are conserved for the future, and that renewable
resources make their full contribution to the progress, pros-
perity, and security of the United States — now and in the
future.

UNITED STATES
DEPARTMENT OF THE INTERIOR

FISH AND WILDLIFE SERVICE

BUREAU OF COMMERCIAL FISHERIES

WASHINGTON, D.C. 20240

OFFICIAl. BUSINESS

Return this sheet to above address, if you do
NOT wish to receive this material Q, or if
change of address is needed □ (indicate
change). '

POSTAGE AND FEES PAID
U.S. DEPARTMENT OF. THE INTERIOR

liibrarian,

Marine Eiolosical Lab.,
2,29, J ' ^02-3 Hole, I-ass.