Regulation of the Abundance of Pink Salmon Populations

University of British Columbia Institute of Fisheries • H. R. MacMillan Lectures in Fisheries 9 ? q Symposium on Pink Salmon 1960 Published 1962

W. E. RICKER Fisheries Research Board of Canada Biological Station, Nanaimo, B.C.

INTRODUCTION

RECRUITMENT CURVES AND INEQUALITIES OF ABUNDANCE

SOME YEARS AGO I attempted to review a matter that has been in the minds of many people for many years: what is the nature of the relationship between the size of a fish stock and the annual reproduction, or recruitment? This sounds a bit academic, but actually it is an intensely practical matter. The form of this relationship largely determines a fish stock's average yield potential — that is, how many thousands or millions of pounds it can be expected to yield in an average year.1 A graph relating size of spawning stock to average recruitment has two logically necessary characteristics: it must pass through the origin, and it must sooner or later cross the replacement diagonal, or "45°-line". Within the framework of these requirements, quite a variety of shapes is possible. Except for A, the types shown in Fig. 1 have all been suggested, at least tentatively, by different fish stocks. If recruitment be defined as the number of fish of a brood that survive to near maturity, then Pacific salmon are among the easiest of all fish in which to study recruitment, because typically each fish is exposed to capture during only one short part of its life— near the end. At any rate this was true before high-seas fishing became important, and it is still substantially true for chum, sockeye and pink salmon along a good deal of the North American coast. Among the Pacific salmons, the pink salmon should be easiest of all to work with, because practically all individuals mature at the same age—age 112. In spite of these favourable conditions, we do not yet have any abundant body of reliable data indicating the relation between spawners and recruitment for various pink salmon stocks. What usually happens is that when there are good data for catch the data

'In any particular year, of course, the prevailing environmental conditions are often more important than parental stock size in determining the abundance of the fish—comparing the usual range of variation in the effects of both. zAlmost all salmon biologists on both coasts of the Pacific now regard this as demonstrated, although opposing arguments were marshalled as recently as 1954 in a tour de force by A. P. Vedensky. 155 SYMPOSIUM ON PINK SALMON

150

100

50 100 150 PARENTS FIGURE 1. SOME POSSIBLE REPRODUCTION CURVES. ALL ARE SCALED SO THAT 100 IS THE ABUND- ANCE OF SPAWNERS WHICH ON THE AVERAGE REPRODUCES ITS OWN NUMBERS.

for escapement are poor--and vice versa. This of course stems from the nature of the material: escapements are easiest to count or estimate accurately when they are small, whereas it is only when its stock is large that a particular river's contribution to the commercial catch can easily be separated from its neighbours' contributions. However, good catch and escapement information for pink salmon have been gathered for some years now on the Skeena River, and more recently on the Fraser and other rivers, so that more recruitment curves, and more reliable recruitment curves, are becoming available as time goes on. Meantime, Fig. 2 shows an older example for illustration, a series of estimates of the catch and escapement of the central region of the British Columbia coast as assembled by Dr. Neave (1953). Among other things, this graph illustrates that when statistics apply only to the mature period of a fishery, you are not likely to have information concerning the production obtained from any of the really heavy densities of spawners. This in turn means that it is very difficult to draw an average curve reliably: two quite different ones are sketched in on Fig. 2, and either seems equally possible. We are thus left in considerable uncertainty about the best level of escapement for pinks in this region. In order to cover a wider range of stock sizes, I once obtained data from Mr. C. E. Atkinson for the

156 Regulation of the Abundance of Pink Salmon Populations

4 MILLIONS

0 0 2 4 6 8 MILLIONS OF SPAWNERS FIGURE 2. RECRUITMENT (CATCH PLUS ESCAPEMENT) PLOTTED AGAINST ESTIMATED SPAWNING STOCK OF PINK SALMON FOR THE CENTRAL REGION OF THE BRITISH COLUMBIA COAST (FISHERY AREA 6-8). THE YEAR OF SPAWNING IS INDICATED FOR EACH POINT. CURVES ARE DRAWN FREEHAND (SEE THE TEXT). DATA FROM NEAVE (1953, P. 460).

Karluk pink salmon escapements (Fig. 3). These lack the corresponding catches, however; if catch were added to the ordinate values all the points would be shifted higher so as to represent total recruitment—perhaps twice as high, on the average. In spite of this weakness, it is clear that an extremely large spawning population in 1924 produced very few recruits—apparently because of wholesale smothering of eggs in the nests. Figure 3 has another peculiarity—that all the large runs occurred in even-numbered years. This illustrates a remarkable aspect of pink salmon biology: the alternation of strong and weak lines or "cycles" which exists in many places. This phenomenon is most inconvenient for industry, and has been puzzling to science. At first glance it seems to make nonsense of relationships of the sort shown in Fig. 1 and 2. We normally expect a natural law to apply to all like objects with equal probability. If this is not so, something very peculiar is going on. Consequently, it will be profitable to compare catches and recruitment data for pink salmon wherever they can be obtained, in order to see whether and how they may be reconciled with normal reproduction theory. The best treatment of the odd-year even-year problem is in two papers by Neave (1952, 1953). Since these were written, about 8 years' additional information has

157 Symposium on Pink Salmon

become available on the American side of the Pacific, and we have obtained further publications concerning west Pacific stocks.

AI 26 0 33 50 100 THOUSANDS

2 3 4 PARENTAL STOCK (millions)

FIGURE 3. REPRODUCTION RELATIONSHIP FOR PINK SALMON OF THE KARLUK RIVER, ALASKA, BASED UPON FENCE COUNTS EXCEPT IN 1922 AND 1924. THE SHADED SQUARE IN LOWER LEFT IS SHOWN ON A LARGER SCALE IN THE UPPER RIGHT CORNER. COMMER- CIAL CATCHES ARE NOT INCLUDED IN THE PROGENY. THE CURVE IS DRAWN FREEHAND. FROM FIGURE 20 OF RICKER (1954); DATA ARE FROM THE UNITED STATES FISH AND WILDLIFE SERVICE, PACIFIC SALMON INVESTIGATIONS.

DEFINITIONS Any more or less homogeneous group of pink salmon spawning in a particular tributary will be referred to here as a stock. All stocks are divided into two lines—the odd- year line and the even-year line. A group of adjacent stocks, which arrive and spawn at about the same time of year, is a run, of which there may be one to several in a large river: for example, an upriver and a downriver run, or an early and a late run. A group of runs in one geographical region, more or less distinctive in respect to certain mor- phometric or meristic characters or in the nature of their alternation of odd-year and even-year abundance, will be called an assemblage or group of runs—what in some Russian papers is called a natio or nation. Marked inequality of two pink salmon lines is here called dominance of the larger

158 Regulation of the Abundance of Pink Salmon Populations

line: there is no implication of any necessary direct physical interaction of the two lines, though that is not excluded. Any definition of when dominance should first be recog- nized is bound to be arbitrary, but clearly a definition is needed. I propose to speak of the existence of dominance when one line is at least twice as great as the other and this condition lasts for at least 8 years (4 generations of each line). Line ratios from 2:1 to 4:1 will be considered as "weak" dominance; 4:1 to 8:1 will be called "moderate" dominance; 8:1 to 16:1 "marked" dominance; and anything larger will be "extreme" dominance or extreme inequality.

ASSUMPTION The one general assumption used in what follows is that pink salmon stocks return to their native streams with considerable fidelity. The amount of wandering is likely variable, but it probably rarely exceeds 10% for indigenous stocks to judge by experi- ments with other anadromous salmonids. Transplanted stocks may possibly be more in- clined to wander, but this has not yet been fully proven.

DATA SOURCES In what follows commercial catches are used extensively as a measure of abundance in the "fully-developed" period of each fishery (that is, after it has once reached a peak or plateau of production). They can also be used to compare adjacent years even in the developmental stage of a fishery. Everyone recognizes, of course, that landings will rarely be a perfect index of stock size. However, we are interested only in rather large differences between the two lines, or large changes in the abundance of a single line, so ordinary year-to-year fluctuations in rate of fishing will be of little importance for our purposes. In fact, the range of sizes of the catches in any area is usually so great that logarithmic paper has been used almost exclusively for the graphs which portray them. There is often more than one version extant of the catch statistics for a particular area. Also, statistics are sometimes presented in weight units, sometimes in numbers of fish. It has not been necessary, however; to try to make all the graphs conform to a uniform system, because in no case are such differences sufficient to affect the conclu- sions drawn. In a number of instances different bodies of data have been plotted together on one graph, using conversion factors or overlapping points, in order to get long series. Most data are cited from published sources, but recent USSR catches by regions are from the VNIRO [Central Research Institute for Marine Fisheries and Oceano- graphy], as supplied to the International North Pacific Fisheries Commission.

CHARACTERISTICS OF LINE DOMINANCE AND RELATED MATTERS From published data the list below has been compiled of some of the pieces of information which might have a bearing on inequality of the two lines of pink salmon. Point 1. Dominance occurs in all degrees, from practically complete absence of one line, to practical equality of the two lines. An example of practically complete dominance is the odd-even alternation of spawning populations in McClinton Creek, Masset Inlet (Fig. 4). Extreme dominance is repre- sented by the catch of pinks in the Fraser River region (Fig. 5). A condition of marked dominance is shown by the pink runs of Cook Inlet up to 1927 (Fig. 6). Mostly moderate dominance is exhibited by the Rivers Inlet pinks from 1919 to 1931 (Fig. 17). For an example of weak dominance, the period from 1921 to 1931 in the Skeena River fishery will qualify (Fig. 18). Finally, the catches of the southern part of Clar- ence Strait district of southeastern Alaska (Fig. 7) will serve as an example of absence of dominance. A borderline condition is that shown for Frederick Sound (Fig. 8) :

159 Symposium on Pink Salmon during the period 1918-27 the even-year landings were substantially greater than the odd except for 1922-23.

100,000

10,000 RS S PAWNE

1 000

1 00

1 0

From fry transplanted

7 1 1 930 32 34 36 38 40 42 FIGURE 4. SPAWNING RUNS OF PINK SALMON COUNTED AT THE FENCE AS THEY ENTERED MCCLIN- TON CREEK, MASSET INLET, QUEEN CHARLOTTE ISLANDS. OF THE ODD-NUMBERED YEARS, THE FENCE WAS MAINTAINED IN 1933 AND 1937 ONLY, WHEN RETURNS WERE EXPECTED FROM TRANSPLANTED FRY-SOME MARKED BY FIN-CLIPPING AND SOME UNMARKED. OF THE 7 ADULTS FOUND IN THOSE YEARS, 4 WERE MARKED. DATA FROM PRITCHARD (1938, P. 146; 1948, P. 130). 160 Regulation of the Abundance of Pink Salmon Populations

i I I I I

•

1000

1900 1910 1920 1930 1940

FIGURE 5. LANDINGS OF PINK SALMON FROM TRAPS AT POINT ROBERTS AND BOUNDARY BAY, W.A,SHINGTON, JUST SOUTH OF THE MOUTH OF THE FRASER RIVER. SALMON TAKEN FROM THESE TRAPS CONSISTED OVERWHELMINGLY OF FISH BOUND FOR THE FRASER. DATA FROM ROUNSEFELL AND KELEZ (1938, P. 807) .

161 162 square metreofspawningarea). not fundamentallyincompatible--many streamsintheCentralRegion[ofBritish abundant inabsoluteterms (forexample,onthebasisofnumberadultsper erous BritishColumbiastreamsandconcludedthat"in general theindividualstreams reflect closelytheratiosshownbycommercialcatches inthesea". central partoftheBritishColumbiacoast,exhibitonlya slightlygreaterdifferencebe- lines candifferradicallyinindividualstreams,butthisappears nottobetypical,atany be compared(Fig.9).Neave(1952,p.58)reviewedthe spawningcensusesfornum- rate inCanada.Forexample,thespawningrunsofHooknose Creek,KingIsland,inthe tween thetwolinesthandoescatchoftheirArea(No. 8),overtheyearsthatcan a - Neave (1953,p.476) Point 2.Wheredominance doesnotexist,thetwolinesmaybeeithersparseor Within aregionofgenerallyequalabundance,therehave beenreportsthatthetwo

THOUSANDS 100 I 000 LANDINGS OFPINKSALMONINTHEREGIONCOOKINLET,ALASKA. DATAFROMRICH I

1 900 points outthat"largeruns in bothevenandoddyearsare Symposium onPinkSalmon AND BALL 1 910 FIGURE (1931, 1 920 6. P. 700). 1930 REGULATION OF THE ABUNDANCE OF PINK SALMON POPULATIONS

Columbia coast] have heavy runs of adults in both cycles". In his 1952 paper he cites several examples, including the Neekas River, where pink spawners have exceeded 2 per square yard (2.5 per square metre) of total stream bottom, over a long series of odd and even years. In a region where marked inequality does exist, at least a substantial part of the stocks are really numerous in the "on" years—almost by definition. Point 3. The dominant years usually tend to be in the same line in all streams over a fairly wide area, sometimes over a very wide area ; but where two such areas meet, the dominant lines may be different in adjacent streams. Figure 10 is a map of the distribution of dominance in various stocks during the 1920's. The sharpness of the boundary that can exist between dominance types is perhaps best illustrated in the Amur River region (Fig. 11), where the My River on the south and the !ski on the north have differed from the massive alternation of the Amur

1000

U_ 100

1900 1910 1920 1930

FIGURE 7. LANDINGS OF PINK SALMON FROM THE SOUTHERN PART OF CLARENCE STRAIT, SOUTH- EASTERN ALASKA. DATA FROM RICH AND BALL (1933, P. 608). 163 Symposium on Pink Salmon

0 2

z .5

0.2 MILL IONS

0.1

0.05

0.02 1906 1910 1914 1918 1922 1926 1930

FIGURE 8. LANDINGS OF PINK SALMON FROM THE FREDERICK SOUND DISTRICT, SOUTHEASTERN ALASKA. DATA FROM RICH AND BALL (1933, P. 508) .

for protracted periods (Eniutina, 1954b). Little less sharply differentiated were the various parts of the coast of Kodiak and Afognak Islands, at least up to 1926: the southern, eastern and northeastern sectors showed little trace of consistent dominance in the catches, while the northwestern and western coast of Kodiak from Kupreanof Strait to the Ayakulik River exhibited a marked even-year dominance, similar to the mainland opposite (Rich and Ball, 1931). Some British Columbia examples may be seen in Neave's (1952) tables I and II. The inset of Fig. 10 shows the small wedge of streams having their two lines more or less equal, situated in the middle of the otherwise strongly even-year Queen Charlotte Islands. Point 4. Dominance existed in some regions before commercial fishing began. Information on relative size of the lines during the earliest history of a fishery tends to be poor, because often catch records were poorly kept or even non-existent. Neverthe- less, data for several regions indicate quite clearly that regular alternation of abundance 164 Regulation of the Abundance of Pink Salmon Populations

AREA 8 LANDINGS 1 000

500

200 0 2

0 100

O. 50

HOOKNOSE CREEK 0 SPAWN E AS 20

1 0

1948 1950 1 952 1954 1956 1958 FIGURE 9. ABOVE: LANDINGS OF PINK SALMON FROM FISHERY AREA 8 OF THE CENTRAL PART OF THE BRITISH COLUMBIA COAST. BELOW: NUMBER OF SPAWNERS ENTERING HOOKNOSE CREEK, ON KING ISLAND (IN AREA 8). SPAWNERS WERE COUNTED AT A FENCE EXCEPT IN 1959, WHEN THEY WERE ESTIMATED BY INSPECTION OF THE STREAM. LANDINGS ARE FROM RECORDS OF THE CANADA DEPARTMENT OF FISHERIES, VANCOUVER; SPAWNERS ARE FROM HUNTER (1960, P. 840) AND MANUSCRIPT DATA. of the two lines existed in the earliest days of catch records, hence presumably also prior to commercial fishing. The Fraser River stock shows this clearly (Fig. 5). For northern British Columbia, the Queen Charlotte Islands and much of the mainland are lumped together in published catch records for earlier years, but an odd-even alternation occurs regularly back to 1908, when the northern pack first reached any appreciable size. In Cook Inlet, Alaska, the even-year pink salmon packs exceeded the odd years consistently from an early period of exploitation, though the record is incomplete; the average ratio was about 10:1 (Fig. 6). Bristol Bay records give the same impression (Fig. 13). Fin- ally, in the records of the Amur estuary from 1902 to 1914, when the catches of both

165 Symposium on Pink Salmon

140° 160. 180°

140. 160. ISO.

FIGURE 10. DISTRIBUTION OF PINK SALMON DOMINANCE AS OF ABOUT 1925. HORIZONTAL SHAD- ING: EVEN-YEAR DOMINANCE; VERTICAL SHADING: ODD-YEAR DOMINANCE; CROSS- HATCHING: ABSENCE OF DOMINANCE. NO UNEQUIVOCAL INFORMATION WAS FOUND lines were still increasing, the even-year catches were weakly but consistently dominant, 3 the average ratio being about 2:1 (Fig. 12) . This consistent alternation could not be a result of the play of economic factors, for there is no known economic cycle having that periodicity. Instead, it must reflect a real and primitive alternation of abundance of the two lines. In some of these examples the actual difference in size of the two lines might considerably exceed the difference in recorded catches (if market demand were limited), but it could scarcely be less (cf. Hypothesis 5, below). Point 5. In several regions the initial dominance became intensified after commercial utilization became fairly intensive; in other regions dominance has appeared, or has first become recognizable, only after commercial fishing became considerable. One of the best examples of intensification of an initially rather weak alternation is again the Amur River (Fig. 12) . After 1914 the discrepancy between the lines increased rapidly, and by 1923 supplies were so low that a closed season on commercial fishing in odd years was imposed for 4 generations. An Alaska example is Bristol Bay (Fig. 13), where the even-year catch dropped practically to nil by 1919. In British Columbia

sOne of the earliest records I have seen of regular alternation of catches is for Hokkaido, as figured by Vedensky (1954, fig. 13) following Lindberg. This indicates consistent dominance of the even-numbered years, moderate or weak in degree, from 1883 to 1897. Considering the fishing boats available at that time, these figures must represent catches in the streams of the island or immediately offshore, but I do not have information on how long or how actively the fishery had been prosecuted before 1883. 166 Regulation of the Abundance of Pink Salmon Populations

(D 0 1-1 •-< C+ N CD CO (1) (1) L ÇU. U) (1) Q- C+ T-R1 < C CD WY U) CD )UTHERN KURIL ISLANDS, THE ANADYR REGION, OR ALASKA 3 0 CD a o H. a) , THE ALEUTIAN ISLANDS, OR WASHINGTON SOUTH OF PUGET CD (R) C O RIOUS AUTHORS; INSET AFTER NEAVE, 1952, FIGURE 1. C 0 1— 1-1 o B 0 a rI CD CD 0- (-+- the best example (Fig. 5). - - CD 0 0 A) O 0 sr there are several examples of situations where dominance of CD (-1- (T. (R) at any rate became recognizable, only after commercial fishing CO C+ '175 C+ H. the Maritime Province dominance of the odd-numbered years -< (.1) 4 (D H. U) H• id persisted to 1951 (Fig. 14). In the Bolsheretsk region of CD < H. c CL years dropped behind in 1919 (Fig. 15), and stayed in second H (1) Q- (1) (R) W B CD CL) landings figures are confirmed by data on catch per trap (Fig. U) 0-0 0 u) 1-H N 0 - 1— 0 0 A) (-1 :an shift from one line to the other. • I 1— CO (T) and eastern Pacific pink stocks provide examples of the very a m al of dominance. For example, in the Maritime Province there r to even-year dominance in 1954 (Fig. 14). In British Colum- Fishery Areas north of Vancouver Island had a weak to mod- :e up to the early 1930's. Subsequently some of them switched for part or all of 1932-47 (Fig. 17; see also Neave, 1952, ar alternation (Fig. 17) or back to a moderate even-year super-

LOUTH OF THE BOLSHAIA (OR BYSTRAIA) RIVER, CENTER OF THE LARGEST SALMON FISHERIES OF

167 Symposium on Pink Salmon

of 5 E.

Ni kolagva

Cope Boss Nix hnol Cape Prong• Prong. ox hoer A'.fr R. Z•leny■ Oa i Vark•

Chomp NOVOISSOVO St. Chomi R

Cape Neva I s kovo

Mariinak Ty•

DI Kastri

00ge KloSter — Kaimp

STRAIT of

TARTARY

SAKI4ALIN

Poronai B.

Data Bay

TERPENIE GULF

SEA of JAPAN

FIGURE 11. LOWER COURSE OF THE AMUR RIVER AND ADJACENT REGIONS.

Point 7. In one instance reversal of dominance seems to have spread like a wave from one region to adjacent ones. In most rivers bordering the Sea of Okhotsk there was a reversal of superiority of the lines during the 1930's, according to Kaganovsky (1949, pp. 39-45), Dvinin (1952, p. 69), Eniutina (1954a, p. 335) and others. It began in east Sakhalin in 1933, when odd-year superiority appeared there, displacing the previous even-year superiority in Ter-

168 Regulation of the Abundance of Pink Salmon Populations

50 50

20 20 0 2

I 0 1 0 OF PINK SALMON

NC z 5 5 0 Q. U- 0 cr 2 2

2 0 I U- —1 0

2 0.5 0.5

0.2 0.2 THOUSANDS

1 910 1 920 1 930 1 940 1 950

FIGURE 12. LANDINGS OF PINK SALMON IN THE NIKOLAEVSK REGION (AMUR RIVER AND ESTUARY). DATA FOR 1902-49 ARE IN MILLIONS OF FISH; THOSE FOR 1948-59 ARE IN THOUSANDS OF METRIC TONS (THE POINT FOR 1948 HAS PRACTICALLY THE SAME POSITION IN BOTH UNITS). DATA FROM SMIRNOV (1947), VEDENSKY (1954), AND VNIRO. penie Gulf and a condition of no marked alternation in Aniva Bay.5 In the Bolsheretsk region of southwest Kamchatka the odd years displaced the even from first place in 1937 (Fig. 15) ; west Kamchatka as a whole followed in 1939, as did the northern and western coast of the Okhotsk Sea, down to the Iski River (Fig. 11). Only the Amur River preserved its even-year superiority, but even there the odd years increased and even years declined through the 1930's until the 1939 and 1940 catches were practically equal (Fig. 12). West Sakhalin, though not directly bordering on the Sea of Okhotsk, also seems to have been affected; it retained its even-year superiority to 1934, following which there has been no steady pattern (at least to 1946), but there were very large odd- year catches in 1939 and 1941. Point 8. Dominance can disappear, temporarily at least, if the big line becomes reduced in numbers; but there is no known example of a weak line increasing to a condition of equality with the dominant line while the latter remains abundant. In Canada the Skeena River is a good example of the disappearance of dominance (Fig. 18). Following 7 generations of even-year superiority, the 1932 catch dropped abruptly; since then the two lines have exhibited no consistent difference, while fluctuating about a reduced level of abundance. Similarly, dominance disappeared during the period 1949-55 in the Arnur region, after the big even-year line fell off abruptly (Fig. 1 2 ) .

,This account follows Dvinin, Kaganovsky (p. 39) lists east Sakhalin and the southern Kuril Islands as regions of odd-year dominance from 1917 to 1935 ; however, his figure 9 shows even-year dominmce in east Sakhalin in 1927-31. 169 Symposium on Pink Salmon

0 00 -

1 900 1 910 1920 1 930

FIGURE 13. LANDINGS OF PINK SALMON IN THE BRISTOL BAY REGION OF WESTERN ALASKA-ALMOST WHOLLY FROM THE NUSHAGAK AREA. STARTING IN 1922 THE FISHING SEASON CLOSED ON JULY 25, EARLIER THAN FORMERLY, MAKING MUCH OF THE PINK SALMON RUN UNAVAILABLE TO THE NETS. DATA FROM RICH AND BALL (1928, P. 57).

170 Regulation of the Abundance of Pink Salmon Populations

5000

2000

o 1000 2

; 500

z 200

t 00

1 910 1915 1 920 1 925 1 930 1 935 1940 1 945 1 950 1955 FIGURE 14. LANDINGS OF PINK SALMON IN THE MARITIME PROVINCE [PRIMORE] OF THE USSR. DATA FOR 1926-34 ARE FROM MILOVIDOVA-DUBROVSKAIA (1937, P. 102), AND THOSE FOR 1940-59 ARE FROM VNIRO. DATA FOR 1909-25 AND 1935-40 ARE FROM THE CHARTS KAGANOVSKY (1949, PP. 42, 45), SCALED APPROXIMATELY TO METRIC TONS BY COMPARISON WITH MILOVIDOVA-DUBROVSKAIA'S DATA FOR 1926-34. The same occurred in 1925-30 and again in 1945-50 for the major stocks of the west coast of Bering Sea (Fig. 19)—in each case as a result of a marked decline in the dominant odd-year catches. We would, of course, like to see the off-year line increase to a level of productivity equal to that of the dominant years, in any and every area. To date this has not been observed, as Vedensky (1954) has emphasized. Point 9. Studies of predation on pink salmon fry in small rivers indicate that predators eat a larger percentage of the smaller hatches. Studies of freshwater mortality of pink salmon eggs and fry have been made by Prit- chard (1936, 1948), Wickett (1952), Neave (1953), Semko (1954) and Hunter (1959). There is the usual large variability associated with variable environmental conditions, but in every instance (McClinton Creek, Hooknose Creek, Karimai Spring Creek) there is a tendency for smaller spawnings to have a larger percentage fry delivery to salt water—as is normal, and indeed is necessary if the stock is to be maintained. What is more interesting is that one component of this freshwater mortality has exactly the opposite behaviour. Pritchard's (1936) data for McClinton Creek show that in the stomachs of Dolly Varden char (Salvelinus malma), cutthroat trout (Salmo clarki) and yearling coho salmon (Oncorhinchus kitsutch) taken just above the fry counting fence (where the fry were artificially concentrated), the average number of pink salmon fry per stomach was practically the same in 1931 and 1933, in spite of the fact the fry migration was about twice as large in 1931. Neave (1953) and Hunter (1959) extended this observation to natural situations. In Hooknose Creek pink and

171 172 chum frywereeatenprincipallybysculpins salmon tostreamsthatwerebarrenofpinksin"off"years. Pritchard(1938)reports ham Island,B.C.)fromthe brief periodofdownstreammigration,whichinthisstreamtypicallylastsonlyonenight populations becausethepredatorsbecamesatiatedduringtimesofheavyfrymigration. ing fryhatchedfromeggsofpinksimportedotherregionsin less thanthatofnativefryhatchedinthedominantyears. for eachindividualfish,percentagelossesweregreateramongthe to seawitheveryappearanceofhealth.Howeveronlyone adultreturnedfrom878,000 the resultsoftransplantingpinksalmoneggstoMcClinton Creek,MassetInlet(Gra- Creek, werehatchedinhatcherytroughs,releasedatthefree-swimming stageandwent creek ofthesefrythetransplanted stockin1956and1958was0.26%0.031%, completely discouraging.EggsfromLakelseRiverofthe Skeenasystemwereplanted, stream havebeenexcessive inthe"off"year,thoughnotassevereatMasset.For island (Fig.10).Ineachof1931and1935"green"eggs weretakentoMcClinton fry releasedthefirsttime,andsixfrom506,000second time(Fig.4). 1958, andCanadaDepartment ofFisheriesrecords).Againlossesafterleavingthe as comparedwith0.89%and 0.59%forthenativestockof1957and1959(Wickett, and theemergingfrywerecounted astheyleftthecreek.Thepercentagereturnto There havebeentwomajorwell-controlledattemptsin Canadatointroducepink Point 10.Whereextremedominanceexists,thesurvivaltoadultstageofmigrat- A laterattemptmadeinWahleach(Jones)Creek,aFraser tributary,hasbeenless MILLIONS OF PINK SALMON 0 05 0.2 05 0.1 2 5 CATCH OFPINKSALMONINTHEBOLSHERETSKREGION,ASNUMBERSFISH. KAMCHATKA, INTHOUSANDSOFMETRICTONS.BOLSHERETSKDATAFROMDVININ POINTS JOINEDBYBROKENLINESATRIGHTREPRESENTTHECATCHOFALLWEST

1 1954—FIGURE 915 1 920 2 Symposium onPinkSalmon AND TABLE 1925 Tlell River,only30milesawaybutontheeastcoastof i 930 6); FIGURE WEST KAMCHATKADATAFROMVNIRO. 1935 (Cottus 15. 1 940 spp.), troutandcoho.Duringthe 1 94 5 1950 "off" less numerous years,ismuch 1955 5 20 50 0

fry

•

Regulation of the Abundance of Pink Salmon Populations

200

>- CC

CD- I 00 CR 4

CC 50

(-)

4 (S) 20

0 _J

5 1910 1920 1930 FIGURE 16. RELATIVE CATCH PER TRAP NET IN THE BOLSHERETSK REGION OF WEST KAMCHATKA (1920 = 100). DATA FROM SEMKO (1939, TABLE 6).

0 2 20

,c 1 0

0. 5 0

Ofl O 2 u_

• 0.5

0.2 1920 1930 1 940 1 950 FIGURE 17. PINK SALMON CANNED PACK AT RIVERS INLET (FISHERY AREA 9 OF BRITISH COLUM- BIA). DATA FOR 1919-52 ARE CANNED PACKS, FROM ANNUAL REPORTS OF THE BRITISH COLUMBIA DEPARTMENT OF FISHERIES; THOSE FOR 1951-60 ARE TOTAL LANDINGS FROM THE CANADA DEPARTMENT OF FISHERIES STATISTICS, CONVERTED TO CASES AT THE RATE OF 1 CASE = 70 LB. (31.8 KG) OF RAW FISH.

173 • •

Symposium on Pink Salmon

500

97.

•z 200

2 1 00

v z 20

▪ 10

5 1 915 1 920 1 925 1 930 933 1 940 1 945 1 950 1 95 5 FIGURE 18. CANNED PACK OF PINK SALMON FROM THE SKEENA RIVER AND ADJACENT BRITISH COLUMBIA WATERS (FISHERY AREA 4), IN MILLIONS OF 48-LB. CASES. DATA AS FOR FIG. 17; FIGURES AFTER 1956 REPRESENT TOTAL LANDINGS. comparison, the geometric average spawning-ground returns of fry of native pinks at 4 other British Columbia localities have been 0.82% to 3:74% (Neave, 1953, p. 458). Similar transplants have been made at various times to Puget Sound streams of the State of Washington. About these it is known only that there has been no permanent success to date. There is unfortunately an ambiguity in interpreting these results. They might result from either excessive mortality in off years, or from a failure of the transplanted fish to return to the stream where planted. Pritchard (1938) discovered some pinks having the same mark as his transplanted fry in remote regions (notably the Fraser River area), but whether wandering was sufficient to account for the almost complete failure of Masset returns remains unclear. At any rate, almost complete failure of a transplanted salmon stock to return to the new site is very unusual, when the transfer has been made at the egg stage and the fry hatch has been successful. Often such transfers give very good returns. For example, in spite of considerable wandering, the pink salmon transplanted from Sakhalin to rivers of the Barents and White Seas returned to those rivers in large numbers (Kossov et al., 1960). Point 11. In many regions of Asia the dominant lines tend to have consistently smaller fish than the weak lines. This phenomenon has not been identified in America. Soviet scientists early discovered that the size of the pink salmon in the big years was less than in the off years, in most parts of Asia (Pravdin, 1929, 1932; Kaganovsky, 1949). The most extensive uniform series of data I have found is for 1930-49, published by Semko (1954) for the Bolshaia River region of west Kamchatka; it can be extended backward a bit by considering figures in an earlier paper of his, for 1925-27 and 1929 (Table I, Fig. 20) . The linear correlation coefficient between catch and average length for 1930-49 is r = -0.705; between catch and weight it is r —0.785. Both figures are highly significant, and the latter suggests that 62% of the variation in fish weight during 1930-49 was associated with density of stock. If the runs of 1925-29 174 Regulation of the Abundance of Pink Salmon Populations

PINK SALMON 20

a 0.5 CF)

1910 1920 1930 1940 1950 FIGURE 19. PINK SALMON CATCHES OF EAST KAMCHATKA, INCLUDING KARAGIN AND ADJACENT REGIONS OF THE BERING SEA COAST. DATA AS FOLLOWS: 1944-59 FROM VNIRO ; 1934-43 FROM KAGANOVSKY'S (1949) FIGURE 7, TOP GRAPH, EACH LINE SCALED SEPARATEI.Y TO THE 1944-47 VNIRO DATA; 1926-33 FROM MILOVIDOVA-DUBROVSKAIA'S (1937) TABLE 2; 1909-1925 FROM SEMKO'S (1939) FIGURE 5, SCALED TO MILOVIDOVA- DUBROVSKAIA'S FIGURES. be included, the relationship becomes definitely curved (Fig. 20). The second-degree regression of average weight in kilograms (w) on catch in millions of fish (C) is: 2 w = 2.006 — 0.09123 C 0.002447 C The multiple correlation coefficient between w and C is R = 0.884; and the fraction of the (greater) variability in weight during 1925-49 that is related to stock density is 2 estimated as R A = 77%. Similar but less extensive data are available for the Amur and other regions. Birman (1956) presents the data of Table II. This series is much too short to be convincing by itself, and in any event the average size in 1949 is closer to 1948 than to 1947. How- ever, Birman and other authors all imply that this alternation of size was as well known on the Amur as in west Kamchatka. I have not located any detailed information on the relative sizes of the fish of groups of stocks that show moderate or extreme dominance in America.

Point 12. In major Asian assemblages of stocks the superiority in size of the fish has shifted to the new weak line when dominance was reversed, and fish of both lines become large when both are much reduced in abundance. When two lines of pinks in west Kamchatka reversed their relative numerical posi- tions, the size of the fish followed suit at exactly the same time — in 1936-37 (Fig. 21). Krykhtin (1958) shows that pinks in the Amgun River (the principal pink salmon 175

Symposium on Pink Salmon

TABLE I. AVERAGE FORK LENGTH AND WEIGHT OF PINK SALMON (SEXES GIVEN EQUAL WEIGHT) IN CATCHES MADE NEAR THE BOLSHAIA RIVER OF WEST KAMCHATKA, AND THE NUM- BER OF FISH CAUGHT THERE. DATA FOR 1930-49 FROM SEMKO (1954, p. 20); DATA FOR 1925-29 FROM SEMKO (1939, p. 40).

YEAR Length Weight Catch Year Length Weight Catch mm kg millions mm kg millions 1925 534 2.011 1.12 1926 472 1.301 18.12 1927 546 2.171 1.22 1928 6.12 1929 516 1.791 0.72 1930 472 1.40 5.7 1931 526 1.90 3.7 1932 463 1.20 19.0 1933 520 1.80 1.2 1934 429 0.81 16.6 1935 477 1.33 6.9 1936 496 1.46 9.7 1937 477 1.37 13.2 1938 480 1.43 11.8 1939 444 1.05 18.8 1940 483 1.38 10.0 1941 460 1.24 15.7 1942 473 1.33 12.8 1943 453 1.10 20.3 1944 450 1.15 14.0 1945 473 1.28 12.5 1946 483 1.31 7.8 1947 471 1.18 21.1 1948 505 1.47 4.3 1949 468 1.23 20.2 1 Estimated from the average length. 2 Approximate estimate from Semko, 1954, figure 1, scaled to the values tabulated for subsequent years.

o 27

2.0 025

1.8

05 2 4 'CO 1.6 0

1.4

0 1.2 LA o 39 1 .0

A VERAGE 0.8 .34

0.6

1 1 2 4 6 8I 1 0 1 2 14 18 20 CATCH I N MILLONS

FIGURE 20. AVERAGE WEIGHT OF PINK SALMON OF THE BOLSHERETSK REGION (MEAN OF AVER- AGES FOR MALES AND FEMALES EACH YEAR), PLOTTED AGAINST THE SIZE OF THE CATCH IN THE SAME YEAR. THE TREND LINE IS THE QUADRATIC EXPRESSION COM- PUTED IN THE TEXT, DATA FROM TABLE I; 176 Regulation of the Abundance of Pink Salmon Populations

TABLE II. CATCH AND AVERAGE LENGTH OF PINK SALMON IN THE AMUR ESTUARY. DATA FROM BIRMAN ( 1956, FIGURE 1). Year Catch Av. length 10 3 tons mm 1947 1.3 507 1948 12.3 440 1949 1.9 456

2.0

1 5

1.0

15 I

1 0

1930 i932 1934 1936 1938 1940 1942 1944 1946 1948 FIGURE 21. AVERAGE WEIGHT OF PINK SALMON IN THE BOLSHAIA RIVER REGION OF WEST KAM- CHATKA, 1930-49 (ABOVE ) , AND THE LANDINGS IN EACH YEAR (BELOW ) . DATA FROM TABLE I.

TABLE III. COMPARISON OF SIZE OF THE AMGUN RIVER PINK SALMON WITH THE TOTAL COM- MERCIAL CATCH OF THE AMUR RIVER AND ESTUARY. DATA FOR FISH SIZE FROM KRYKHTIN (1958, TABLE 3 ) ; CATCH DATA FROM VNIRO.

Amur Anzgun gorbuscha Y ear catch Av. length Av, weight 10 3 tons min kg 1950 1.2 463 1.44 1951 2.3 446 1.20 1952 0.9 475 1.62 1953 0.9 469 1.40 1954 0.3 494 1.56 1955 0.4 485 1.60 Average 1.0 472 1.47 177 Symposium on Pink Salmon

tributary of the Amur) exhibited no alternation in size during 1950-55, when Amur catches of both lines were low (Table III). The 1950-55 year-classes were also caught by a high-seas fishery, so that the catches of Table III do not constitute an index of abundance on exactly the same basis as pre-war Amur catches ; nevertheless the stocks of 1950 to 1955 are obviously all small relative to previous even-numbered years (Fig. 12). The fish of Table III average much larger in size than the 440 mm shown in Table II for Amur pinks of the dominant year 1948. Point 13. In one Asian assemblage the dominant line has consisted of fish which are consistently larger than those of the weak line. The Maritime Province group of stocks is an exception to the rule that Asian pink salmon are smaller in the big years. Streams flowing into the western side of the Strait of Tartary at the northern end of the Sea of Japan, plus the Chomi, Tymi and My Rivers on the southwestern coast of the Amur estuary (Fig. 11), have their dominant runs in odd-numbered years (Fig. 14) ; but these consist of larger fish than the even- numbered years (Kaganovsky, 1949; Eniutina, 1954b; Birman, 1956). The most abundant run of this assemblage is apparently the one that inhabits the Tumnin River flowing into the Bay of Data. Table IV shows that during 1947-49 the size of the Tumnin pinks varied directly as their own abundance and inversely as those of the Amur. Be- cause of this, Birman (1956) suggests that the Maritime Province fish must associate in the sea with the much more numerous Amur fish—his views are discussed further in Appendix I. TABLE IV. AVERAGE LENGTH OF DATA BAY PINK SALMON AND THE CATCHES OF NEIGHBOURING AREAS.

Maritime Length of Amur Province pinks estuary Year catch ( Data Bay) catch 10 3 tons mm 10 3 tons 1947 3.6 494 1.3 1948 2.0 463 12.3 1949 5.1 489 1.9

With the decline of the Amur even years after 1948 (Fig. 12), the Maritime Prov- ince pinks too lost their alternation in size. I know of no recent data for Data Bay, but Krykhtin gives information for the My River, the most northern of the Maritime Prov- ince stocks (Table V). These seem to be about the largest Asian pinks—substantially

TABLE V. AVERAGE LENGTH AND WEIGHT OF PINK SALMON OF THE MY RIVER. DATA FROM KRYKHTIN (1958 ) . Length Weight Year mm kg 1950 507 1.67 1951 508 1.56 1952 524 1.80 1953 528 1.85 1954 519 1.70 1955 542 2.00

178 Regulation of the Abundance of Pink Salmon Populations larger than the Data Bay pinks in 1947 and 1949, or the Amur (Amgun) pinks in 1950-55 (Table III). Like the latter, they exhibit little trace of odd-even alternation in size during this period, but it is interesting that the weights of Amgun and My pinks tend to vary in parallel fashion during these six years, presumably because of common environmental variability.

Point 14. The various assemblages of pink salmon in Asia differ in their morphological characteristics, some of which almost certainly have a genetic basis. In Canada and southeastern Alaska the odd-year and even-year lines differ in average size. Detailed morphological studies of pink salmon have been made as follows:

Assemblage Stock or site Author West Kamchatka Bolsheretsk, 1926( ?) Pravdin, 1929 Okhotsk Iski R., 1950 and 1951 Eniutina, 1954a Amur Amur estuary and river, 1928 Pravdin, 1932 Amur Amgun R., 1950 and 1951 Eniutina, 1954a Amur Amgun R., 1949-52 Eniutina, 1954b Maritime Province My R.; Uarke, Cape Lazareva, Eniutina, 1954b De Kastri (all 1949-52) West Sakhalin Rachisi R.; Antonovo fishery Dvinin, 1952 (1947) East Sakhalin Poronai R. (in Terpenie Gulf) Dvinin, 1952

Table VI compares some of the characters measured ; more detail is given in the papers above. A lot of the differences discovered are statistically significant; and some are quite large. Many, however, pertain to characters that are known to be greatly affected by temperature and other factors during development, and usually there are no comparisons of different years in the same locality (except sometimes in respect to size, weight and condition factor). Eniutina's (1954b) figures for the Amgun River apparently include her earlier (1954a) series, but even so, the differences between the two tabulations give some idea of year-to-year variability (Table VI). There are considerably larger differences between Pravdin's (1932) data for the Amur estuary in 1928 and Eniutina's Amgun figures. She suggests that these may arise, partly at least, because Pravdin used material which could have been a mixture of three different "nations": Maritime Province, Amur and Okhotsk. Another possibility is that the various Amur River stocks themselves might not be completely homogeneous. Although more comparisons between years and between individual stocks would be desirable, for all of the above stocks or assemblages there appear to be one or a few divergent characters that are unlikely to be wholly attributable to environmental or sampling variability. Among these are: Bolsheretsk: thin caudal peduncle. Iski River: long snout, long anal fin. Amgun River: deep caudal peduncle, long pectoral fin. My River: short tail, short anal insertion, short lower jaw, short and narrow maxillary. Rachisi River: short pectoral fin, long ventral fin. Poronai River: short snout, long dorsal insertion, short anal fin. In both of the last two localities the fish had long lower jaws. The short pectorals and long ventrals of the Rachisi River pinks are very distinctive—one almost wonders whether the measurements for the two fins could have become transposed! 179 Symposium on Pink Salmon

TABLE VI. MERISTIC AND MORPHOLOGICAL CHARACTERISTICS OF PINK SALMON FROM DATA OF THE AUTHORS INDICATED. FIGURES TABULATED ARE ARITHMETIC MEAN VALUES, WITH THE STANDARD ERROR OF THE MEAN GIVEN BELOW EACH. ITEMS 1-7 ARE COUNTS; ITEMS 8-28 ARE PERCENTAGES OF THE FORK LENGTH OF THE BODY. METHODS OF MAKING COUNTS AND MEASUREMENTS, AND THE NUMBERING OF THE CHARACTERS, ARE AS DESCRIBED IN PRAVDIN (1929, 1932). ENIUTINA'S DATA ALL APPLY TO SILVERY' FEMALEFISH OF 45-46 CM FORK LENGTH. En'utina En'utina Pravdin Pravdin Dvinin (1954a) (19548) (1932) (1929) (1952) River or West West East region Iski Amgun Amgun My Amur Kam- Sakhalin Sakhalin Year 1950-51 1950-51 1949-52 1949-52 1928 chatka Rachisi R. Poronai R. Character 1. Lateral line 180.60 175.30 170.00 174.05 175.64 172.9 scales 1.47 1.16 1.07 1.26 0.53 0.25 2. Branchiostegal 12.24 12.65 11.35 11.87 12.18 12.12 rays 0.12 0.08 0.11 0.06 0.07 0.08 3. Rakers on 29.80 29.92 29.18 29.27 30.03 29.93 1st arch 0.27 0.18 ______. 0.10 0.15 0.13 0.15 6. Vertebrae 68.84 68.82 69.92 70.33 69.12 70.19 0.14 0.12 0.14 0.06 0.18 0.15 7. Pyloric caeca 136.80 129.15 129.00 128.88 127.8 3.26 3.05 0.25 8. Snout length 7.20 6.34 6.32 5.97 6.00 6.36 6.54 4.95 0.08 0.05 0.06 0.05 0.42 0.10 0.057 0.09 10. Head length 21.60 20.99 21.26 21.91 21.79 22.20 0.10 0.06 0.15 0.14 0.12 0.09 12. Maxillary 8.60 7.86 8.19 7.84 8.65 8.46 8.31 8.57 length ' 0.11 0.04 0.07 0.09 0.16 0.09 0.06 0.10 13. Maxillary 1.91 1.76 1.81 1.68 1.59 1.90 2.16 2.06 width 0.03 0.02 0.03 0.03 0.04 0.04 0.17 0.01 14. Lower jaw 13.60 12.81 12.96 12.46 13.38 13.27 14.22 14.45 length 0.12 0.06 0.09 0.11 0.13 0.12 0.12 0.09 15. Depth of head' 10.76 10.78 8.69 9.41 0.16 0.09 0.10 0.08 16. Depth of body 22.80 24.47 25.02 23.58 23.65 23.30 24.87 24.78 0.21 0.13 0.23 0.19 0.25 0.16 0.13 0.16 16a. Thickness of 10.60 9.74 11.70 body 0.11 0.07 0.07 17. Predorsal 46.36 45.52 45.68 44.50 46.24 45.62 45.74 distance 0.15 0.13 0.20 0.28 0.14 0.14 0.13 18. Postdorsal 39.74 40.96 ...... 40.05 39.40 38.25 distance 0.26 0.37 0.15 0.14 0.14 19. Preventral 50.90 50.61 50.95 50.57 50.87 51.77 distance 0.19 0.02 0.20 0.14 0.15 0.18 20. Preanal 67.43 67.76 67.12 66.55 69.82 49.21 distance 0.21 0.14 0.23 0.16 0.20 0.16 21. Pectoral length 14.26 14.46 12.18 14.12 0.11 0.08 0.09 0.10 22. Ventral length 10.86 11.66 14.72 12.30 0.09 0.07 0.08 0.10 23. Caudal peduncle 17.24 18.01 17.15 17.26 17.16 17.18 length 0.12 0.18 ...... 0.15 0.09 0.09 0.18 23a. Caudal peduncle 7.42 6.74 6.72 6.60 6.93 depth 0.05 0.06 0.07 0.05 0.27 24. Dorsal 10.48 10.62 10.48 9.91 10.55 10.32 11.41 12.28 insertion 0.12 0.08 0.12 0.09 0.13 0.08 0.07 0.11 25. Anal insertion 12.20 11.59 11.85 9.95 11.38 11.41 11.97 11.55 0.12 0.15 0.09 0.33 0.08 0.09 0.07 0.13 26. Height of anal 1136 10.63 ...... -- 9.03 9.18 9.62 8.81 0.13 0.08 0.14 0.12 0.08 0.09 28. Middle rays 6.16 5.41 5.64 5.19 5.50 5.90 5.69 5.65 of tail 0.11 0.12 0.09 0.08 0.10 0.07 0.03 0.08

1 From middle of eye 180 Regulation of the Abundance of Pink Salmon Populations

The My River stock, living next door to the Amur, is the most distinctive of all. In addition to the features listed in Table VI, over the years 1950-55 its fish were consistently much larger than the Amgun River pinks, and also were always larger than the Iski River pinks, though by a small amount (Krykhtin, 1958, table 3). The fecundity of a female of a given length was considerably less in the My River than in the Amgun (Eniutina, 1954b, figure 1). Eniutina suggests that the deep caudal peduncle of the Amgun fish might represent muscle needed for their longer upriver migration. There were no significant differences in pyloric caeca counts among the 5 Asian stocks studied; the Iski River has a higher average (136.8) than any of the others, but apparently few fish were involved. Gill rakers were significantly more numerous in west and east Sakhalin, less numerous in west Kamchatka and in Pravdin's Amur-region samples of 1928 (but not in Eniutina's Amgun fish). The only published comparative work on pink salmon morphology in America is that of Pritchard (1945), who counted the gill-rakers and pyloric caeca of a number of stocks. He found some significant average differences in both characters, differences which tend to be positively correlated.6 Again there is no direct measure of year-to-year variability in a given stream, either within lines or between lines, so it is impossible to assess the role played by environment; also, the relation of the above two characters to fish size should be studied. (However, in related species gill-rakers and caeca are apparently less affected by environment than are such things as number of scales or vertebrae.) Pritchard found that pinks of five streams tributary to Masset Inlet did not differ significantly in respect to caeca or rakers, whereas two from Moresby Island did: Haans Creek (Skidegate Inlet) had fewer caeca and fewer rakers, while Pallant Creek (Cum- shewa Inlet) had higher-than-average caeca but fewer rakers. Samples taken from a restricted region of the Harrison River and its tributaries (Fraser system) differed among themselves in both caeca and rakers to a degree that was significant statistically, but the actual stock to which the fish of the different samples belonged is not completely clear. Although some of Eniutina's morphological studies are based on material collected in both odd-numbered and even-numbered years, I have not found any direct comparison of odd and even lines in her work, or in any other from the western Pacific. In the Bol- sheretsk region there is evidently little or no difference between the two lines in average weight at a given stock density, as shown by the fact that there is no consistent separation of odd-year and even-year points along the trend-line in Fig. 20. In British Columbia, however, Hoar (1951) for the period 1927-42, and Godfrey (1959) for 1944-58, both found that odd-year pinks were larger than even-year pinks by 0.5-1.5 lb. (0.2-0.7 kg), in spite of the fact that catches (and presumably stocks) for the coast as a whole were less in the even years (Godfrey, 1959, figure 3). As a result, in this section of the pink salmon's range there is a positive association between abundance and size. This is the reverse of the Asiatic picture, and the reverse of what would be expected to result from within-line competition. Hence either the major even- line runs in British Columbia and adjacent Alaska are congenitally faster-growing than the odd-line runs, or else they spend more time in parts of the ocean that are favourable for growth. HYPOTHESES TO ACCOUNT FOR DOMINANCE We have, then, 14 points to consider. You will recall that when Woodrow Wilson presented his 14 points, Clemenceau complained that "le bon Dieu n'avait que dix". By

°Pritchard inadverently shows the correlation as negative. 181 Symposium on Pink Salmon coincidence, it is only the first 10 of our points that seem to have a direct relation to reproduction theory and the phenomenon of dominance. But we were bound to examine the others, and perhaps someone else will discover an "angle". Meantime, keeping the above information in mind as best we can, let us see how it bears on the various hypotheses that seek to account for a regular alternation of pink salmon abundance. Several of these hypotheses call attention implicitly or explicitly to the phenomenon which Ferris Neave (1953) has called depensatory mortality. What is depensatory mortality? The word "depensation" was invented by Dr. Neave to describe the situation where a particular cause of mortality kills a greater percentage of fish when the stock is small than when it is larger. Depensatory mortality is the opposite of the more usual compensatory mortality, in which a greater percentage of fish dies as a stock becomes denser.

1. HYPOTHESIS OF DEPENSATORY PREDATION ON FRY IN FRESH WATER. As outlined in Point 9 above, it has been discovered that the consumption of pink salmon fry by other fish is depensatory, as the fry move downstream; the smaller fry populations suffer much greater percentage mortality at this time. The net effect of this is to flatten the left limb of the reproduction curve, so that a wider range of stock densities is reasonably stable. In the situation shown in Fig. 22B the only true equilibrium point is at A, but if one line becomes quite small because of unusually unfavourable weather, it tends to stay small for a considerable time. Recovery of abundance is slow because the average surplus produced is very small between M and N. The situation of Fig. 22A is more serious, because if a stock becomes reduced below 13 it will not replace itself and will tend to decline to extinction.

100

C9 50 0 CC 0_

0 50 100 0 0 100 SPAWNERS FIGURE 22. HYPOTHETICAL REPRODUCTION CURVES INFLUENCED BY NATURAL DEPENSATORY MORTALITY. ABSCISSA-SPAWNERS; ORDINATE-PROGENY.

In the streams actually studied, compensatory effects of absence of crowding, such as using only the best nesting sites at low population levels, is more than enough to over- come the greater susceptibility to predation of small numbers of fry. Thus depensatory freshwater predation as far as observed to date has not set up situations which actively induce inequality of the two lines ; rather, the observed situations favour the preserva- tion of inequality whenever it arises as a result of environmental variability.

182 Regulation of the Abundance of Pink Salmon Populations

It is also possible, though it has not yet been observed, that a particular combination of depensatory and compensatory factors in freshwater survival could produce two "humps" in a recruitment curve. In the situation described by Fig. 23A, there would be only one equilibrium point, at A, in the absence of fishing. In Fig. 23B there are three equilibrium densities of spawners, A, B and C, of which B is unstable. In the absence of fishing the stock would tend to fluctuate about either A or C, but jumps froth one to the other could occur rarely, following exceptionally good, or poor, reproduction. If the two lines in a river happened to be at different loci—one near A, the other near C— there would be dominance of one line which could persist indefinitely.

50 i 00 0 50 100 S PAWNERS FIGURE 2 3 . HYPOTHETICAL RE PRODUCTION CURVES INFLUENCED BY NATURAL DE PENSATORY MORTALITY.

Depensatory mortality in fresh water is a local mechanism, which helps to preserve dominance as it arises in each river or stream individually. However, Dr. Neave points out that, in British Columbia at least, there have been abrupt declines in pink salmon lines over fairly extensive regions of the coast, caused by widespread unfavourable clima- tic conditions. This of course can initiate a marked discrepancy between the two lines where none already exists, and even reverse some existing alternations, so that a whole region is brought into phase (Point 3).

2. HYPOTHESIS OF DEPENSATORY MORTALITY OF FINGERLINGS IN SALT WATER. There are some stocks—those in large muddy rivers—for which depensatory predation may not be important because predation itself is unimportant. Also, while the synchronization mechanism described in the previous paragraph may be adequate to explain some local uniformities, it might not account for the actual situation in a region where all the rivers and tributaries share the prevailing dominant cycle, including streams of very diverse size and character. For example, in the Puget Sound and Fraser River region pinks spawn in many small streams, and also in the main channel of the Fraser and the Thompson (Ward, 1959). An unusually cold, dry winter along the coast in this region might dry out and freeze up numerous pink spawning streams of Puget Sound and the lower Fraser tributaries, but it could scarcely have any disastrous effect on the Fraser main-stream spawners, still less those of the Thompson and Seton Rivers, situated far upstream in the "dry belt" where cold winters are nothing unusual. For such situations as these, might it be possible to locate the area of operation of 83 Symposium on Pink Salmon

natural depensatory mortality out in salt water? In the southern Strait of Georgia and Puget Sound there is a series of islands separated by channels through which the young pinks pass on their way out to the Strait of Juan de Fuca. In this region upwards of 60 million adult pinks were present in odd-numbered years early in this century,, producing about a billion fry migrants in the even years. Even today, in the spring following the odd-year spawnings dense bands of young pinks can be seen lining the margins of the narrow passages in this region, so it seems at least possible that the protection of large numbers should accompany the young fish some distance into salt water. That is, in local areas the shallow-water marine predators might quickly become satiated with fry at high fry densities, whereas they would continue picking them off indefinitely when there were only a few. In this way the present-day excess mortality of the even-year pinks transplanted into Wahleach Creek might have its origin in salt water (Point 9). Similarly the Masset Inlet streams, where extreme dominance apparently existed from the earliest times, all flow into a common sea basin 200 square miles in extent, and must then pass out through a narrow channel 20 miles long to get to the open sea. This provides exactly the set-up for depensatory activity of the type just postulated. Details were given above of the disappearance of off-year transplants in this area (Point 9). If this was due to depensatory predation, it could only have been in salt water, because the fry were counted just as they left the river. A third region of primitive (though much less extreme) inequality of the two lines is the Amur River. Here the estuary, 70 miles long by about 25 miles wide and mostly less than 30 feet deep, provides another region in which depensatory predation might easily be significant—though the fact that its water is often entirely fresh and presumably muddy might restrict predation substantially. Similar argument might apply to the Nush- agak estuary in Bristol Bay.

3. HYPOTHESIS OF PREDATION BETWEEN THE TWO LINES (CANNIBALISM) . A special case of marine depensatory mortality is predation of one pink salmon line on the other. It is sometimes said that the two lines are entirely separate throughout their life history, but this is an overstatement. It is true that the fry of one line normally leave fresh water before the adults of the other line arrive there, but in the ocean they meet each other. The paths of the outbound and the inbound migrations overlap, potentially at least. Manzer (1956) shows that off British Columbia there are many age 0+ pinks in coastal waters at the time the age 1+ pinks pass through on their way to their spawning streams. Dvinin (1952) says that large numbers of age 0+ pinks are taken in trap nets as late in the season as autumn along the southwest coast of Sakhalin. If both the inbound and outbound fish dawdle a bit in coastal waters, they may be in con- tact for some weeks. There seem to be no published records of young pink salmon occurring in the stomachs of older pinks, but there have been only a few studies of the ocean food of this species, and these mostly concern fish taken far out from shore. It is known that age 1+ pinks feed extensively on fish, and in many regions fish constitute their major food (Synkova, 1951; Allen and Aron, 1958; Andrievskaia, 1958). Cannibalism, if it occurs, would almost certainly approximate to the "B" type predation of Ricker (1952), in which the absolute number of kills at a given moment is pro- portional jointly to the abundance of the consumers and of the consumed. This would

'During the period 1925-33 the odd-year catch from this region averaged 10.2 million pinks (Rounsefell and Kelez, 1940, p. 806), so that the total adults present then were possibly 15 million (at 70% utilization by the fishery). From Rounsefell and Kelez estimate of decline in catch per unit effort (p. 813), this represents about 25% of the number of fish present prior to the obstruction of Hell's Gate in 1913, so that the total was of the order of 60 million pinks in the odd years before 1915. If we consider this as the primitive abundance, and assume a relatively inefficient reproduction rate, for example that 2% of the eggs in females produce fry, the fry production was 2% of 30,000,000 X 1700 eggs = 1,000,000,000 fry each even-numbered year. 184 Regulation of the Abundance of Pink Salmon Populations correspond, for example, to a situation where the older fish meet the younger ones cas- ually and randomly, neither of them changing their habits because of increased or decreased abundance of the other. Cannibalism of this sort is inherently depensatory for the smaller line, at all levels of abundance of the two lines. Only if the two lines happen to be exactly equal are they in a state of equilibrium, but this is unstable. If one line exceeds the other even a little bit, it tends to set up an oscillation whose amplitude increases at an increasing rate, and which continues until some compensatory mortality factor counteracts it. Quantitatively, if A and B represent the abundance of the two lines, and a and b are the fractions of each line killed by cannibalism, then from expression (2) of Ricker (1952) the mortality rate in each generation of each line is affected by the abundance of the immediately previous generation of the other line as follows: log(1 — a) = kB log(1 — b) = kA The same proportionaity constant, k, applies to both situations. The rapidity of divergence of line A from line B depends on the degree of initial displacement, on the magnitude of a and b at a given density, and of course on the magnitude of the compensatory mortality factors which oppose divergence. To get an idea of what would happen if compensation were absent, consider a situation where the two lines are equal, so that A = B and a = b; specifically, let the initial mortality rate due to cannibalism be 20% (a = b = 0.2). Let there now be an "accidental" displacement that makes line A twice as large as B, such as could readily be caused by environmental influence. With these initial conditions calculations from the formulae above (Appendix II) show that after 10 years the ratio of the two lines will have increased from 2:1 to 16:1, and they will have gathered "momentum" for a much more rapid divergence during the ensuing 10 years. Repeated application of the equations above will reduce the B stock to zero, while A continues to increase. In most actual situations, the intensification of compensatory factors (such as mortality from overcrowding of the spawning grounds by the dominant line, and greater successs in reproduction by the weak line) would presumably be sufficient to counteract the depensatory influence of cannibalism and stop the divergence of the two lines before the smaller becomes extinct. Thus if it were to be found that there is any appreciable consumption at all of small pinks by large pinks, cannibalism would automatically become a prime suspect as a factor which perpetuates and intensifies inequality between the lines. Experimental observation is seriously handicapped, of course, by the fact that areas where the effect of cannibalism is most important would be those where extreme dominance prevails naturally, so that the scarcity of the off-year fish makes it difficult to find them in the stomachs of the dominant line. It is tempting to imagine that the "lost" McClinton transplanted fry may have tarried in Masset Inlet, and there fell prey to the large accumulation (some millions) of even- year pinks which congregate in the basin before entering fresh water. At any rate the information described in Point 10 above would follow from a condition of cannibalism as well as, or better than, from depensatory marine predation by other fishes. One characteristic of cannibalism, not shared by most other depensatory mechanisms, is that the two lines tend to change in abundance inversely. There are in fact several instances where a weak line became substantially more numerous just after the dominant line suffered a severe reverse; and where a dominant line has decreased following an exceptionally good "off" year. Both the Bolsheretsk region (Fig. 15) and the Amur 185 Symposium on Pink Salmon region (Fig. 12) illustrate this reciprocal tendency.8 Particularly suggestive are (1) the failure of the Amur off line to increase during a 4-year closed season on commercial fishing when the dominant line was at maximum abundance; and (2) the subsequent increase of this line, in spite of active fishing, after the dominant line had decreased. Wherever cannibalism is important, the two lines cannot be at maximum abundance simultaneously—and this is in accord with observations to date (Point 8). In addition, if some cannibalism occurred offshore where different stocks mix, it could account for, or at least could have assisted in, the progressive shift from even-year dominance to odd- year dominance which occurred during the 1930's in rivers bordering the Sea of Okhotsk (Point 7). Similarly, following the tremendous overpopulation of 1924 the Karluk even-year run was at a very low ebb in 1926 and especially in 1928, being about the same size as the odd year 1929 and much less than the odd years 1931 and 1933 (Fig. 3). Nevertheless it was the even line that recovered, and had resumed its extreme dominance by 1932. This too would be understandable if the adult pinks of other runs in that region, which still had large populations in the even years during this period, consumed some considerable fraction of Karluk young of the odd-year broods. 4. HYPOTHESIS OF FOULING OF THE REDDS BY LARGE EGG DEPOSITIONS. It has been suggested that the eggs of a large spawning may affect spawning grounds unfavourably—either by persistence of dead eggs from one year to the next (which has been observed), or by favouring the increase of a large stock of egg predators in the gravel (which is hypothetical). In this way a numerous line would have a depressing influence on a smaller line as long as it maintained its superiority. Hunter (1959) has developed this theme with reference to Hooknose Creek, B.C., but the sequence of percentage hatches (of pinks and chums) in that stream does not fit the hypothesis too well, at least not in any simple form. Also, as Neave (1952) emphasizes, there are numerous streams in central British Columbia which have large spawnings every year, and such also occur abundantly in southeastern Alaska. . Nevertheless it is probably too early to dismiss the idea of spawning ground contamination as everywhere inapplicable. For example, it is possible that a large population is always able to clean its redds well enough to wash out last year's dead eggs and so avoid any ill effects from them, while a smaller population might fail to do this, and could become progressively reduced as a result. Or some types of spawning beds may be sus- ceptible to this effect, whereas others are not. Hunter (1959) suggests that contamination of redds by eggs of a very large population may even extend to the second year following, and so depress the survival rate of their progeny's eggs as well as those of the alternate line. In this connection, it is interesting that at Karluk the few pinks produced by the big year 1924 had a very poor reproductive season in 1926, so that the spawning stock decreased further in 1928 (Fig. 3). But with the progeny of the 1928 spawning the even-year line shot up again, and by 1932 it had reasserted its dominance. This sequence may merely reflect environmental variability, but it does correspond with Hunter's hypothesis.9 9There are also, of course, numerous examples of catch trends in which both lines move in the same direction. This is to be expected as a fishery develops in any area, and also if it falls off because of a deteriorating environment or overhshing—hence such situations do not require special explanation. 9Still another explanation seems possible. The descriptions I have heard of disastrous overpopulation by pinks or sockeye indicate that the (normally) most favourable spawning sites produce practically no progeny, because of wholesale suffocation of eggs, fungus infection, etc. At the same time, the relatively few fish that spawn in unfavourable and normally unused portions of the stream hatch some fry; and these produce almost the whole of the following generation of spawners. There is a pronounced (though by no means perfect) tendency for salmon to return to and spawn in the part of a stream where they were hatched. Thus the spawning of the first generation following a major overpopulation might largely be dissipated on poor sites, so that survival rate is low in spite of the sparse population. Detailed knowledge of overseeding and its consequences is of importance to this and other problems. Because it happens rarely nowadays, no chance should be lost to make such a study if one occurs.. 186 Regulation of the Abundance of Pink Salmon Populations

5. HYPOTHESIS OF DEPENSATORY FISHING. The four mechanisms above can produce or maintain line dominance "naturally". When commercial fishing comes along, at first glance we might expect that it would go after large supplies of fish most intensively and thus gradually even out any differences between the two lines. The fact is, however, that just the reverse often happened (Point 5). To explain this it has been suggested that in many places the fishery itself operates, or used to operate, in a depensatory manner—taking a larger percentage of the smaller lines than of the larger ones. This idea is excellently presented by Rich and Ball (1928, p. 66), for example. Hav- ing in mind the Alaska situation up to 1927, these authors concluded that "there is no doubt that fishing operations ordinarily operate so that the spawning escapement in good years is better in proportion than in poor years, which is just the reverse of what sen- sible conservation would call for." Similar views had been expressed by Gilbert and O'Malley (1921), and doubtless others. Under what conditions has fishing been depensatory ? It seems to have occurred mainly during the earlier stages of commercial utilization, and particularly in remote areas. In general, this would be at times and places where an individual or firm outfitted himself for the season with salt, barrels, cans and other equipment. took them by steamer to a remote station, got busy salting and/or canning, and then had to close down the operation when available supplies of either containers or fish were used up. Particularly for a relatively low-priced species like pink salmon, there was a tendency for an operator to take in only as many cans or barrels as he could be reasonably certain to fill, and this would be governed by the normal minimum runs of the area in question. Thus in a poor year the rather small supply of fish would be utilized quite intensively, while in a good year the run might experience only half as large a percentage utilization, or less. In this way the difference between smaller and larger runs became intensified.10 Even in settled regions where materials were not a limiting factor, large runs used to get re- lief from fishing at times during the season when canneries became glutted by the abundance of fish, or when buying was discontinued because potential markets seemed satur- ated. In years of small runs, on the other hand, the small boats formerly in use could not readily be deployed from one fishing region to another, so that fishermen simply fished harder where they were in trying to make a reasonable income for the season. To-day, in Canada at least, all this has changed. Large purse-seiners increasingly dominate the fishery, and these can be swiftly moved from one region of abundance to another. Gill-netters too have become much more mobile. Also, prices for canned salmon are high and demand is reasonably steady, so that serious market gluts have practically disappeared. Thus Neave (1953, p. 461) concludes that during the post-war period "the catch seems to constitute a rather constant percentage of the run, whether the latter be large or small." In addition, an attempt is being made, with increasing success, to regu- late the catch so as to provide a more or less constant and optimum escapement from year to year, wherever possible. Insofar as this is successful, instead of being a depensatory mortality factor fishing becomes a compensatory factor, as Rich and Ball recom- mended.

',The system of assigning exclusive fishing rights in specific areas to individual firms by annual auction or tender can be particularly favourable to depensatory fishing of this sort. That is, each year's lessee will tend to take as many fish as his capacity permits, if there is no guarantee of his being able to fish there another year and hence no strong incentive to save a reasonable spawning stock. By contrast, the arrangement which prevailed on parts of the Canadian coast up to about 1917, where a particular region was assigned to the same firm year after year, set the stage for a prudent operator to assure his future supplies by foregoing a large part of his potential catch in poor years. However this arrangement terminated before the need for restraint became acute. In any event, in many areas it would be geographically difficult to set fishing boundaries so that the man who conserved a supply of spawners would be the principal one to benefit from it. 187 Symposium on Pink Salmon Support for the view that salmon fishing formerly was depensatory on the Amur River is provided by a graph of Smirnov (1947, p. 38) in which catches of summer chum salmon in the estuary are compared with the escapement of this run to the spawning grounds of the Beshenaia River, an Amur tributary (Fig. 24). The estuarine catch and the escapement vary in parallel fashion over the period 1925-1940, but the fluctuations in escapement are relatively much more severe: spawners declined practically to zero in a number of 'Kars.

BESHENA RIVER Z 1 0 0 - 0 SPAWNERS 7

80-

I U • 60 - 0

40 - HOUSANDS

T 30 -

AMUR ESTUARY • 40- CATCHES 0

2 30 - I

0 20 -

- 1

2 10 -

0 I I I I i 1925 1930 1935 1940

FIGURE 24. BELOW: CATCH OF SUMMER CHUM SALMON IN THE AMUR ESTUARY. ABOVE: NUMBER OF CHUM SPAWNERS IN TIIE BESHENAIA RIVER, AN AMUR TRIBUTARY. REPLOTTED FROM FIGURE 2 OF SMIRNOV (1947).

188 Regulation of the Abundance of Pink Salmon Populations

6. HYPOTHESIS OF THE INFLUENCE OF DENSITY-INDEPENDENT FISHING. The effect of depensatory fishing in promoting inequality of two pink salmon lines is very easy to visualize. Somewhat less obvious, hence quite often overlooked, is the fact that fishing favours inequality of the two lines even when it takes a constant percentage of the salmon present every year. Figure 25 is an idealized reproduction curve of a type

0 0

FIGURE 25. HYPOTHETICAL REPRODUCTION CURVE" (OMAN), TO ILLUSTRATE THE EFFECT OF A FISHERY IN STABILIZING INEQUALITY OF TWO LINES (SEE THE TEXT). that is favourable to this turn of events. In this example, the straight left limb (OM) has a slope of 2.5, and maximum yield is obtained with a 60% fishery and a spawning stock 40% of replacement size. Under the conditions shown an unfished stock will fluctuate about the average replacement density, A, which is designated 100. If in a year of favourable environmental conditions the reproduction is twice as good as average, producing a spawning stock of 200, the expectation of progeny from that large stock is no greater than from only 100. Similarly, survival only half as good as average, producing a return of 50, does not reduce the average reproduction potential. If a very poor reproduction year reduces recruitment to a tenth of normal, the expected reproduction from 10 spawners is 25; however (with average reproductive success) it goes up to 62 in the following generation and is back to 100 in the subsequent one. Thus there is little likelihood of a 2-year cycle appearing in the unfished stock: in fact dominance could be maintained only by unfavourable reproduction occurring repeatedly in one only of the two lines. Consider now the situation when the same stock is being fished at the rate of exploitation, 60%, which gives maximum yield for average reproduction conditions. The average spawning stock is 40 producing (on the average) progeny equal to 100 (at M) , of which 60 is captured and 40 becomes spawning escapement. If in a particular year reproductive success is aboye average, producing say 150 units of progeny, the catch

189 Symposium on Pink Salmon

goes up to 90, and the escapement to 60; but this extra spawning stock does not increase the expectation of progeny because MN is parallel to the base line. On the other hand, suppose an exceptionally poor year comes along, so that reproduction is a tenth of normal, or 10: catch and escapement are reduced in proportion, and the spawners are now only 4 units. This many spawners on the average produce 10 units of recruits, the fishery taking 6 and leaving 4 for reproduction—so that there is no increase. Thus there is no hope for substantial recovery unless an exceptionally good survival year happens to come along in this line; this may easily take 10 to 20 years, and in the meantime the stock has a very convincing 2-year 'cycle".11 Strictly, the above analysis applies only when the reproduction curve has a straight ascending limb (OM) and a flat top (MN). But if a similar study be made of other possible curves, such as those of Fig. 1, 3, 22 and 23, it will appear that for all of them without exception, recovery of an accidentally-depressed stock is retarded by the existence of a fishery. For many types there is considerable quantitative similarity to what was found from Fig. 25—this is true of all types that have a portion of a flat segment or a dome to the left of the 45°-line, particularly if most of the ascending limb has a slope not much greater than the diagonal to the origin from the point marking the optimum equilibrium rate of exploitation.12 With all such curves, fishing (at a constant rate) greatly enhances the likelihood that a reduced level of abundance will persist, once it has become depressed by unusually unfavourable environmental conditions or for any other reason. And the greater this rate of fishing, the slower is recovery likely to be. On the other hand, dominance which is maintained in this manner should not persist indefinitely, if only because (with no relaxation of exploitation) sooner or later the abundant line is apt to tumble down near the level of the poor one and stay there. In- crease of a depressed line to near optimum abundance again would be much less common, but not impossible, in spite of the fishing handicap—thus making reversal of dominance a theoretical possibility. The operation of this mechanism of maintaining inequality is of course facilitated and reinforced by the situations described above under hypotheses 1 and 2. The general effect of depensatory predation is to make the ascending limb of a reproduction curve less convex upward than it would otherwise be—thus favouring the persistence of any accidentally-established low level of abundance when rate of exploitation is steady. In a context which implies a reproduction curve like one of those of Fig. 22 or 23, Neave (1953, p. 476) has pointed out that intensive fishing may "tend to prevent a low population from reaching the numerical strength necessary to 'break through' and achieve a higher general status." With the curve of Fig. 23A, the general region of M becomes a stable level of population when a moderate fishery is in operation, more stable even than N because it tolerates a larger equilibrium rate of exploitation (though its catch is less than at N). It would be very easy for one line of a stock to slip down to M, while the other maintained itself more precariously in the neighbourhood of N. Considering hypotheses 5 and 6 together, the effects of fishing seem potentially cap-

151n addition to its bearing on pink salmon dominance, the above considerations show that fishing constantly at the optimum rate for average reproduction must (in the presence of random environmental fluctuations whose effects are symmetrically distributed about the average) result in a reduction of the average abundance of stock below what is needed for maximum reproduction. That is, the theoretical maximum sustained yield cannot actually be maintained with a constant rate of exploitation. The reason is that unfavourable environmental conditions tend to depress the stock permanently and in proportion to their effectiveness ; whereas favourable conditions, producing spawning stocks greater than 40, have no lasting effect on stock size. Thus the equilibrium point will tend to shift down to some position on OM where years of favourable environmental factors can balance unfavourable years. Similar reasoning applies, less precisely of course, to other types of reproduction curves, but the effect becomes quantitatively unimportant only for curves that rise gradually to the replacement level and for a considerable distance beyond it (curve A. Fig. 1). "That is, the point on the reproduction curve which has the greatest vertical distance from the 45-line. The relation between this slope and equilibrium rate of exploitation is described by expression A18 of Ricker (1958). 190 Regulation of the Abundance of Pink Salmon Populations able of explaining most of the observed characteristics of dominance. Nevertheless the fact is that in many places dominance was well developed before commercial fishing started.

7. HYPOTHESIS OF COMPETITION FOR FOOD BETWEEN THE LINES. Points 10 and 11 make it impossible to deny that great abundance can reduce the size of pink salmon, at least in Asiatic waters. Presumably this results from competition for food. If food competition within a line is all that severe, we are bound to enquire whether there is any way its effects could be passed along to the alternate line in the form of an increased mortality rate of the latter. This would presumably have to take the form of partial exhaustion of the marine food supply by the dominant line, so that the succeeding brood would forage in depleted waters during its first few months at sea, and might suffer from extra mortality due directly or indirectly to malnutrition. How- ever, continuing the hypothesis, the food supply must then snap back within a year, so as to let the off-year pinks reach the larger final size which they ultimately attain. All this is imaginable, but rather far-fetched.

8. SUPPLEMENTARY HYPOTHESIS: SEPARATION OF STOCKS AT SEA OR ALONG MIGRA- TION ROUTES. Separation of groups of pink salmon stocks at sea into a number of more or less dis- crete units would not of itself account for the origin of dominance. It could, however, help to account for examples of the occasional close juxtaposition of rivers having extreme dominance beside those which lack dominance or have the opposite alternation. Birman's interpretation of the different sea homes of the Amur River runs and some neighbouring runs is described in Appendix I. On the British Columbia coast, some dis- tinctness of the marine distribution of runs from neighbouring rivers might make it easier to understand how a wedge of streams lacking dominance can maintain itself in the middle of a strongly even-year region on the Queen Charlottte Islands (Fig. 10) : that is, the rivers of the wedge might have stocks allied in marine distribution to those of the adjacent mainland rather than to their more immediate neighbours, at least in the odd-year line. However, available published information gives little or no support to this hypothesis for America. In the Strait of Georgia there is a sharp transition from strong odd-year dominance at the south end to absence of dominance at the north end. Accordingly, it would not have been surprising to find the south-end fish migrating almost exclusively through the Strait of Juan de Fuca, and the north-end fish through Johnstone Strait— thus setting the stage for partial or complete separation of the stocks at sea. For all we know, perhaps the fry do behave in this way. The south-end adults, however, when they come back from the sea, use both the northern and the southern entrance to the Strait of Georgia in substantial numbers, as do some at least of the north-end fish (Morrison Creek at Courtenay—Pritchard, 1944). Summarizing, there is no evidence of marine separation of adjacent runs in America. In Asia there is indirect evidence that the Amur River runs may live in the Sea of Japan, at least in part, rather than mixing with the other Okhotsk Sea stocks.

CONCLUSION The pink salmon "cycle" is popularly regarded as an unsolved and unsolvable mystery —something unexpected and unfathomable. To science, of course, there is no problem that is insoluble in principle. But science, as reflected in the customary training of scientists, may be handicapping itself by too strong a predilection for solutions that are at 191 Symposium on Pink Salmon once simple and comprehensive. An hypothesis gains support in proportion to the number of facts it explains, the number of problems it solves. The opposite situation, where one problem may require several hypotheses, is much less acceptable to our conditioning. Yet after many minds have searched for the cause of a well-known phenomenon, and when all the hypotheses suggested explain only a part of the facts, it becomes time to consider the merits of multiple causation. In pink salmon dominance, the basic fact is that the fish mature in their second year of life. Given this, there are several possible natural mechanisms that produce or en- courage dominance, as does man's fishing unless it is specifically managed so as to avoid this. I for one am about ready to give up the search for a unique cause of dominance, and concentrate rather on identifying which cause or causes operate on each individual stock. If there is a remaining mystery, it may be this: when dominance can be caused by so many factors, why are there some pink salmon stocks in which it does not occur and has never occurred?

PINK SALMON MANAGEMENT IN RELATION TO DOMINANCE If several causes of dominance are possible, is there any practical value in knowing which are the important ones in individual areas? The answer is yes, for the different hypotheses give different answers to two very important questions: (a) Is maximum yield obtained when the two lines are equal in numbers, or when they are unequal, and at what level of population for each? (b) Will the establishment of a commercial-size population in a blank line be easy, or difficult, or impossible? Let us see what the various hypotheses have to say concerning these questions. Hypotheses 1 and 2. (a) Where dominance is maintained by depensatory predation —in fresh water or in the sea—we should aim at a selective regulation of rate of exploitation so as to have equal and rather large spawning populations (whatever is needed for maximum yield) in both lines. If the combination of depensatory and compensatory mortality factors in early life produces recruitment curves like Fig. 23A or B, then the move from the lower to the higher dome becomes more difficult, because intermediate levels of abundance are relatively unproductive. The change would best be done by stop- ping fishing completely on the smaller line for a year, combined possibly with temporary but intensive destruction of fry predators (Wickett, 1952), in the hope that the gap could be bridged in a single season. (b) If one line is absent altogether, stocking on a sufficiently massive scale should get an off-year line started (cf. Neave, 1952, p. 70), especially if the introduced fish could be protected from fishing during the first few generations. Hypothesis 3. (a) Where dominance is a result of cannibalism of one line on the other, the weak line simply cannot be brought up to the level of the dominant one by regulatory measures: because if the weak line is built up, the dominant one must decline. Although, in theory at least, a condition of near-equality of the two lines (at an intermediate level) could be maintained by careful regulation, this would not necessarily provide the greatest total catch; furthermore, it would be difficult to accomplish because of the diverging tendency of the cannibalism. (b) Starting up an off-year run where none now exists will be particularly difficult, probably impossible, under a regime of active cannibalism. Hypothesis 4. (a) If dominance is found to be a result of contamination of the redds by large egg depositions, it should first be discovered by actual study whether the redds can be cleaned artificially, and also whether a large stock of spawners in the alternate line might be able to do this naturally once it reached a high level of abundance. If so, 192 Regulation of the Abundance of Pink Salmon Populations

regulation for two large lines should be attempted; if not, it may be necessary to write off the weak line. (b) The above studies would also indicate the prospects of success in establishing a population in a blank line. Hypotheses 5 and 6. (a) Where dominance is caused or maintained by fishing— whether by heavier utilization of smaller runs, or merely a uniform rate of exploitation from year to year—this can be corrected by a more flexible application of fishing effort. For this purpose regulations should reduce the rate of utilization when stocks are low and increase them when they are abundant — aiming at a fixed optimum supply of spawners every year. This is, I believe, the present objective of pink salmon fishing regulations in most Canadian waters. Unfortunately there are often considerable practical difficulties in the way of realizing it—stemming from imperfect prediction of supplies, the need for treating many stocks as a unit, and the conflicting requirements of other salmon species. (b) In this situation there shot.id be no obstacle to starting up populations in off years from plantings of modest size, once fishing pressure in the off year is temporarily reduced or withdrawn. Because of the different prescriptions indicated by the various diagnoses above, it is of practical importance that research be continued to identify the cause or causes of dominance wherever it occurs. SUMMARY The 2-year life history of pink salmon divides each stock into 2 lines. Frequently one line exceeds the other consistently over a period of years, by a small amount or by a great amount. This "dominance" of one line has prevailed, at one time or another, over much but not all of the fish's range. It reached its most extensive and most extreme development during the middle 1920's, when even-year lines were dominant in most Asiatic regions (except the mainland coast south of the Amur, and the western coast of the Bering Sea), in most of Alaska except the southeastern portion, in British Columbia except for eastern Vancouver Island and the adjacent mainland. At that time odd-year dominance characterized the mainland of Asia south of the Amur River, the western coast of the Bering Sea, and extreme southern British Columbia and adjacent Washing- ton (particularly the Fraser River). No consistent dominance characterized the northern half of eastern Vancouver Island and adjacent mainland, all of southeastern Alaska, Afognak Island and southern and eastern Kodiak Island, and possibly some areas for which information is poor, northern Alaska and adjacent Siberia, Aleutian Islands, Kuril Island, Hokkaido). Dominance existed in some regions before commercial fishing began, but it usually became intensified after large-scale fishing started. Elsewhere dominance first became recognizable some time after fishing was well developed ; and in several regions it has persisted for a while, to disappear later. A shift in dominance from one line to another has frequently been observed, though several assemblages have retained the same dominance continuously since the beginning of usable records. Several studies show that, in smaller streams at least, there is a greater survival rate of the fry of larger stocks and of fry which go to sea during the height of the season of migration, because predators become satiated when fry are numerous ("depensatory" predation). In regions of extreme dominance, off-year fry from imported eggs survive less well after leaving the spawning tributary than do native fry of dominant years. Morphological and meristic differences exist between typical stocks of different regions of Asia; there are no comparable studies for America. In at least two abundant stocks of Asia, the size of the individual fish has been much less in years of abundance than in off

193 Symposium on Pink Salmon years. (The not-too-abundant Maritime Province stock assemblage is an exception, for which a special explanation has been advanced.) In one instance the size difference between two lines became reversed when dominance was reversed, and it has disappeared when dominance disappears. Changes in fish size related to abundance have not yet been detected in North America. Along the coast of British Columbia and southeastern Alaska the odd lines have had the larger fish since 1930, in spite of being (on the whole) the more numerous line. Hypotheses to explain the various aspects of dominance are all based fundamentally on the fact that these fish mature in their second year; some of them also depend on the occasional occurrence of an exceptionally poor or good survival year to initiate a period of dominance. 1. Depensatory predation in fresh water is an observed condition which tends to delay the restoration of a line to its equilibrium abundance. If it produced a reproduction curve of the form of Fig. 23B, it could perpetuate inequality indefinitely even in the absence of fishing, but this type of relationship is probably exceptional if it occurs at all. 2. Depensatory predation in salt water has no support from observation, but it is plausible enough hypothesis for certain areas. It can account for the observed synchronization of line dominance in diverse types of streams on a scale which seems puzzling in terms of hypotheses 1, 4 or 5. 3. Cannibalism between lines also can account for the synchronization of dominance in stocks that spawn in diverse freshwater habitats. Indeed, it can explain all other known characteristics of naturally-occurring dominance, including the reciprocal change in numbers of dominant and weak lines on the Amur and in west Kamchatka. It provides a dynamic force that tends directly and continuously to make the two. lines more different—instead of merely conserving differences that arise by accident. At present, however, no quantitative information concerning its role is available. 4. Prolonged fouling of the redds by superabundant spawnings lacks a clear-cut demonstration so far, and probably has only local importance, at most. Nevertheless it might, like hypothesis No. 3, provide a mechanism whereby reduction of one line would increase survival of the other, and vice versa., 5. At certain places and times in the past fishing has almost certainly intensified or perhaps even initiated dominance, by taking a larger percentage of the smaller line (depensatory fishing). Today this is much less important or has even been reversed, because of increased facilities for processing large runs and because of improved information for fishery management and better enforcement of regulations. 6. Fishing which takes a constant percentage of a stock has been widely influential in helping to preserve inequalities of natural origin because it reduces the surplus available for restoration of a depressed stock. Unlike effects No. 3 and 5, it cannot initiate or intensify dominance. It reinforces and is reinforced by the effects of depensatory mortality (No. 1 and 2) in conserving differences between lines. With a reproduction curve of the (hypothetical) type of Fig. 23A, a stock has two stable equilibrium levels when rate of exploitation is moderate. 7. Food competition between lines seems everywhere to be an unlikely cause of dominance, though within-line competition can have an important effect on fish size and hence yield. For pink salmon management, the principal dichotomy is between hypotheses 1, 2, 5 and 6 on the one hand, and 3 (possibly also 4) on the other. The former suggest that 194 Regulation of the Abundance of Pink Salmon Populations management should everywhere be directed toward having two lines both at the maximum level of productivity for the region concerned. By hypothesis 3, however, it would be impossible to have the two lines simultaneously at maximum productivity in a region characterized by natural dominance, and it would be very difficult to maintain them at any moderately large nearly-equal level of productivity. Furthermore, the condition of extreme inequality may well provide the maximum average catch per year for such stocks.

LITERATURE CITED ALLEN, G. H., and w. ARON. 1958. Food of salmonid fishes of the western North Pacific Ocean. U.S. Fish and Wildlife Serv., Spec. Sci. Rept.-Fish., No. 237, 11 pp. ANDRIEVSKAIA, L. D. 1958. [Food of Pacific salmons in the northwestern part of the Pacific Ocean.] Materialy po Biologii Morskovo Perioda Zhizni Dalnevostochnykh Lososei, pp. 64-75. Published by Vsesoiuznyi N.-I. Institut Morskovo Rybnovo Khoziaistva i Okeanografii (VNIRO), Moscow. [Fish. Res. Bd. Canada, Translation Series No. 182.] BIRMAN, I. B. 1956. [Concerning causes of a peculiarity of the pink salmon of the Sea of Japan.] Zoo!. Zhurnal, 35(11) : 1681-1684. [Fish. Res. Bd. Canada, Translation Series No. 142.] DAVIDSON, F. A., and ELIZABETH VAUGHAN. 1941. Relation of population size to marine growth and time of spawning of pink salmon (Oncorhynchus gorbuscha) of Southeastern Alaska. J. Marine Res., 4(3): 231-246. DVININ, P. A. 1952. [The salmon of south Sakhalin.] Izvestiia Tikhookeanskovo N.-I. Inst. Morskovo Rybnovo Khoziaistva i Okeanogr. (TINRO), 37: 69-108. [Fish. Res. Bd. Canada, Translation Series No. 120.] 1958. [New information on the migration of pink salmon in the Sakhalin region.] Rybnoe Khoziaistvo, 34(1): 12-15. 1959. [Salmon of Sakhalin and the Kuril Islands.] Glavnaia Gosudarstvennaia Inspektsiia po Okhrane Rybnykh Zapasov i Regulirovaniiu Rybolovstva pri Sovete Ministrov RSFSR. 36 pp. Moscow. ENIUTINA, R. I. 1954a. [Morphobiological and morphometrical characteristics of the pink salmon of the Amgun and Iski Rivers.] Izvestiia TINRO, 41: 333-336. [Fish. Res. Bd. Canada, Transla- tion Series No. 289.] 1954b. [Local stocks of pink salmon in the Amur basin and neighbouring waters.] Voprosy Ikhtiologii, No. 2, pp. 139-143. [Fish. Res. Bd. Canada, Translation Series No. 284.] FAIRBRIDGE, R. W. 1958. Dating the latest movements of the Quaternary sea level. Trans. New York Acad. Sci., II, 20(6): 471-482. 1960. The changing level of the sea. Sci. American, 202(5): 70-79. GILBERT, C. H., and H. O'MALLEY. 1921. Special investigation of salmon fishery in Central and West- ern Alaska. Rept. U.S. Comm. Fish. for 1919, App. IX, pp. 143-160 [U.S. Bur. Fish. Doc. 891.] GODFREY, H. 1959. Variations in annual average weights of British Columbia pink salmon, 1944- 1958. J. Fish. Res. Bd. Canada, 16(3): 329-337. HOAR, W. S. 1951. The chum and pink salmon fisheries of British Columbia, 1917-1947. Bull. Fish. Res. Bd. Canada, No. 90, 46 pp. HUNTER, J. G. 1959. Survival and production of pink and chum salmon in a coastal stream. J. Fish. Res. Bd. Canada, 16(6): 835-886. Japanese Fisheries Agency. 1955. On the salmon in waters adjacent to Japan. Bull. Int. North Pacific Fish. Comm., No. 1, pp. 57-92. KAGANOVSKY, A. G. 1933. [Commercial fishes of the Anadyr River and the Anadyr estuary.] Vestnik Dalnevostochnovo Filiala Akademii Nauk SSSR for 1933, No. 1-3, pp. 137-139. 1949. [Some problems in the biology and dynamics of abundance of pink salmon.] Izvestiia TINRO, 31: 3-57. KOSSOV, E. G., M. S. LAZAREV and L. V. POLIKASHIN. 1960. [Pink salmon in the basins of the Barents and White Seas.] Rybnoe Khoziaistvo, 36(8): 20-25. [Fish. Res. Bd. Canada. Translation Series No. 323.] KRYKHTIN, M. L. 1958. [Biologically uniform groups among Pacific salmon.] Voprosy Ikhtiologii, No. 10, pp. 3-11. [Fish. Res. Bd. Canada, Translation Series No. 195.] LINDBERG, G. U. 1953. [Principles of the distribution of fishes and the geological history of the Far- eastern seas.] Ocherki po Obshchim Voprosam Ikhtiologii, pp. 47-51. Akademiia Nauk SSSR, Ikhtiologicheskaia Komissiia, Moscow/Leningrad. [Fish. Res. Bd. Canada, Translation Series No. 141.] MANZER, J. I. 1956. Distribution and movement of young Pacific salmon during early ocean residence. Fish. Res. Bd. Canada, Pacific Prog. Rept., No. 104, pp. 24-28. MILOVIDOVA-DUBROVSKAIA, N. V. 1937. [Information on the biology of and fishery for pink salmon of the Maritime Province.] Izvestiia TINRO, 12: 101-114.

195 Symposium on Pink Salmon

NEAVE, F. 1952. "Even-year" and "odd-year" pink salmon populations. Trans. Roy Soc. Canada, V, 46: 55-70. 1953. Principles affecting the size of pink and chum salmon populations in British Columbia. J. Fish. Res. Bd. Canada, 9(9): 450-491. NIKOLSKY, G. V. 1953. [Contribution to a biological foundation for the salmon fishing industry of the Amur basin.] Ikhtiologicheskaia Komissiia Akad. Nauk SSSR, Trudy Soveshchanii, No. 4, pp. 160-168. [Fish. Res. Bd. Canada, Translation Series No. 84.] PRAVDIN, I. F. 1929. [Morphometrical characteristics of the west Kamchatka pink salmorr, Oncorhyn- chus gorbuscha Walbaum.] Izvestiia Tikhookeanskoi Nauchno-promyslovoi Stantsii, 4(1), 132 pp. Vladivostok. 1932. [The Amur pink salmon, Oncorhynchus gorbuscha Walbaum natio amurensis nova.] Izvestiia Vsesoiuznyi N.-I. Inst. Ozernovo i Rechnovo Rybnovo Khoziaistva (VNIORKh), 14: 53-98 + 3 plates. 1940. [A review of investigations of far-eastern salmon.] Izvestiia TINRO, 18: 1-105. PRITCHARD, A. L. 1936. Stomach content analyses of fishes preying upon the young of Pacific salmon during fry migration at McClinton Creek, Masset Inlet, British Columbia. Canadian Field- Naturalist, 50: 104-105. 1938. Transplantation of pink salmon ( Oncorhynchus gorbuscha) into Masset Inlet, British Columbia, in the barren years. J. Fish. Res. Bd. Canada, 4(2): 141-150. 1944. Return of two marked pink salmon (Oncorhynchus gorbuscha) to the natal stream from distant places in the sea. Copeia, 1944(2): 80-82. 1945. Counts of gill rakers and pyloric caeca in pink salmon. J. Fish. Res. Bd. Canada, 6(5): 392-398. 1945. Counts of gill rakers and pyloric caeca in pink salmon. Ibid., 6(5): 392-398. 1948. Efficiency of natural propagation of the pink salmon ( Oncorhynchus gorbuscha) in Mc- Clinton Creek, Masset Inlet, B.C. Ibid., 7(5): 224-236. RICH, W. H. and E. M. BALL. 1928. Statistical review of the Alaska salmon fisheries. Part I. Bristol Bay and the Alaska peninsula. Bull. U.S. Bur. Fish., 44: 41-95. 1931. Statistical review of the Alaska salmon fisheries. Part II. Chignik to Resurrection Bay. Ibid., 46: 643-712. 1932. Statistical review of the Alaska salmon fisheries. Part III. Prince William Sound, Copper River and Bering River. Ibid., 47: 187-247. 1933. Statistical review of the Alaska salmon fisheries. Part IV. Southeastern Alaska. Ibid., 47: 437-673. BICKER, W. E. 1952. Numerical relations between abundance of predators and survival of prey. Cana- dian Fish Culturist, No. 13, PP. 5-9. 1954. Stock and recruitment. J. Fish. Res. Bd. Canada, 11(5): 559-623. 1958a. Maximum sustained yield from fluctuating environments and mixed stocks. Ibid., 15(5): 991-1006. 19586. Handbook of computations for biological statistics of fish populations. Bull. Fish. Res. Bd. Canada, No. 119, 300 pp. ROUNSEFELL, G. A. and G. B. KELEZ. 1940. The salmon and salmon fisheries of Swiftsure Bank, Puget Sound and the Fraser River. Bull. U.S. Bur. Fish., 49(27): 693-823. SEMKO, R. S. 1939. [The Kamchatka pink salmon.] Izvestiia TINRO, 16: 1-111. 1954. [Stocks of west Kamchatka salmon and their commercial utilization.]/bid., 41: 3-109. SMIRNOV, A. G. 1947. [Condition of the stocks of Amur salmon, and causes of their fluctuations in numbers.] Izvestiia TINRO, 25: 33-51. SYNKOVA, A. 1. 1951. [Food of Pacific salmon in Kamchatka waters.] Izvestiia TINRO, 34: 105-121. VEDENSKY, A. P. 1954. [Age of pink salmon and the principles underlying its fluctuations in abundance.] Izvestiia TINRO, 41: 111-195. WARD, F. J. 1959. Character of the migration of pink salmon to Fraser River spawning grounds in 1957. Int. Pacific Salmon Fish. Comm., Bull. No. 10, 70 pp. wicKErr, W. P. 1952. Production of chum and pink salmon in a controlled stream. Fish. Res. Bd. Canada, Pacific Prog. Rept. No. 93, pp. 7-9 1958. Adult returns from the 1954 Fraser River planting. Ibid., No. 111, pp. 18-19. APPENDIX I. FLUCTUATIONS IN SIZE OF THE AMUR AND MARITIME PROVINCE GROUPS OF PINK SALMON STOCKS, AND THEIR MARINE JOURNEYS The larger size of the Maritime Province pinks in their years of greater abundance has puzzled a number of USSR authors, because this is the reverse of conditions on the Amur and in Kamchatka. To account for it Birman (1956) puts forth the hypothesis that the Tumnin River and related pinks probably live and compete with the more num-

196 Regulation of the Abundance of Pink Salmon Populations

erous Amur pinks in the Sea of Japan, both probably going south in winter to near the Korean coast. Furthermore, new data from Data Bay in 1954 showed that the pinks then averaged 491 mm (cf. Table IV) ; that is, like the Amur pinks they had increased in size in the even years, in conformity with the decreased abundance of even-year Amur runs. A corollary of this hypothesis is that the size and the abundance of the two lines of these fish must be controlled by two different mechanisms; further, that the observed size differences were determined by competition in the sea, whereas abundance is determined either in fresh water or at some early stage of marine life (before the Amur and Maritime Province pinks become mixed together). The above hypothesis has some interesting aspects. If the Amur and Maritime Prov- ince pinks really mix together in the Sea of Japan, it would seem logical that the former should go there by the direct route southward through Nevelskovo Strait, and back again the same way (Fig. 11). But apparently the return migration, at least, does not use this route to any extent: the pinks taken in the fisheries at De Kastri and Cape Lazareva (just below Nevelskovo Strait on the mainland side), and even at Cape Uarke in the estuary itself, have the seasonal timing, the physical appearance, and the odd-even alternation of abundance that is characteristic of the Maritime Province pinks (Eniutina, 1954b) ; while on the west coast of South Sakhalin the prevailing movement of maturing pinks is southward (Dvinin, 1952). Thus the Amur pinks, if they do go into the Sea of Japan, probably must swim outside of Sakhalin and then double back westward through Laper- ous Strait which separates that island from Hokkaido, returning the same way. To do this, they would have to pass through waters used by east Sakhalin pinks, though not necessarily at the same times of year. Most Amur pinks return to that river early in the season, a month or more before any runs reach Sakhalin streams. In any event the east Sakhalin pinks, whose dominant runs since 1931 have been in odd-numbered years and whose fish have been distinctly smaller in the odd years, could not be involved in the same competitive feeding situation as the Amur pinks. Why might so indirect a migration route be used by the Amur pink salmon ? Lindberg (1953, etc.) has urged that during geologically recent high-water phases the Amur flowed southward into the Sea of Japan.' Birman accepts this idea, and concludes that the Amur pinks would therefore originally have had their marine home in the Sea of Japan. The Kadi channel was closed by falling sea levels in the later Pleistocene, and the mouth of the Amur moved north to its present position. Early post-glacial sea-levels were much below the present, however, so that the shallow Amur estuary was dry land joining Sakhalin to the mainland. The Amur pink salmon were thus forced to go northward instead of southward on leaving the river, but the argument must be that some of them were still able to locate their customary feeding grounds by rounding the Sakhalin peninsula. There might be a substantial advantage in retaining the habit of foraging in the Sea of Japan, for there the Amur pinks would presumably compete only with the relatively small Maritime Province and (probably) west Sakhalin stocks; whereas those which went out beyond the Kurils to escape the winter cold of the Okhotsk Sea would mingle and compete with the vast hordes of pinks from the northern coast of that sea, from both sides of Kamchatka, the Kuril Islands, east Sakhalin and northeastern Hok- kaido. 'Presumably the river flowed out through the Kadi River divide, which to-day is less than 100 metres in elevation, entering the sea at Cape Yuzhnyi south of Nevelskovo Strait (Fig. 11). According to Fairbridge's (1958, 1960) interpretation, during the Pleistocene the world sea level was substantially higher than at present almost continuously up, to the time of the Illinoisan glaciation. However during both the Illinoisan and Wisconsin glaciations it fell to 80-100 metres below the present level ; this would put the Amur estuary well out of water. Following the Wisconsin, the estuary became flooded about 5000 B.C., if the world sea level were the principal factor involved. 197 Symposium on Pink Salmon

In the Amur estuary, soundings indicate a low ridge from Dzhaore east to Zelenyi Gai. Hence while the estuary was out of water, up to about 7000 years ago, the west-shore rivers north to and including the My would have flowed south to the Strait of Tartary, whereas streams north of Dzhaore, including the Amur, flowed northward. And the My River today marks the northern limit for pink salmon having the Maritime Province morphological characteristics (Eniutina, 1954b). Tagging should be able to throw light on the problem of where Amur pinks go, but its evidence so far is only indirect, because (as of 1954) no pinks tagged at a distance had been recaptured in the Amur River or the Northern Amur estuary. The summary of distant returns of Japanese tags (Japanese Fisheries Agency, 1955, p. 76) shows that none of the tags applied along the coast of eastern Hokkaido and the Kuril Islands were retaken in the Amur area; however, it is not stated whether tagging was done early enough in the season to intercept Amur fishz. On the other hand, tagging off northwestern Korea produced two distant returns from the Maritime Province area, in or near the Tumnin River and the My River; another return, however, was from the northwest coast of the Sea of Okhotsk. Tagging along the west coast of Sakhalin near the south end, by both Japanese and Russians, is said to show that two-thirds to three-quarters of local recaptures are made south of the tagging site (Dvinin, 1952, 1958) ; a few of these tagged fish went out of Laperouse Strait and were retaken northward along the east coast of Sakhalin (Japanese Fisheries Agency, 1955).

APPENDIX II.

COMPUTATION OF STOCK SIZES UNDER CANNIBALISM The equations on page 185 are: log(1 — a) = kB (1 > a > 0; B > 0) (1) log(1 —b) =kA (1 > b > 0; A >0) (2) A and B are the numbers of maturing fish in each line, and a and b are the corresponding mortality rates due to cannibalism (which occurred in the year previous to maturity). The conditions given also imply a relationship between successive generations of each line which may be written: K(1 — an +1) All (3) B11+3 = K(1 —b 2) Bn + (4) The subscripts n, n+1, etc., represent successive calendar years, and K is the geometric mean of the factors by which successive generations of each line would be multiplied in abundance, if cannibalism were to be removed. For the purpose of examining trends which would be favoured by cannibalism, we may postulate that K is constant for a series of generations and see what happens; but it is of course necessary to have K vary initially in order to produce the original difference between A and B. For the example of the text, there is an (unstable) equilibrium position with A= B and a = b = 0.2. Let us first evaluate K and k. At equilibrium A. can be equated to

A014 in (3) ; hence: K = 1/(1-0.2) = 1.25 Let A. = B. = 1000 at equilibrium, so that from (1) : log (1 — 0.2) = kB k = —0.0969/1000 = —0.0000969

'Dr. H. Kasahara of the International North Pacific Fisheries Commission tells riv that in recent years numbers of Japanese hooks from the Kuril region have been taken from pink salmon in the Amur estuary. This suggests, but does not necessarily prove, that part at least of the Amur pinks do not enter the Sea of Japan. 198 Regulation of the Abundance of Pink Salmon Populations Now let the A-line be increased so that in year 1 A, = 2000. From (2) : log (1 -131) = -0.0000969 x 2000; 1 - b, = 0.640 From (4), in year 2: B, = 1.25 X 0.640 X 1000 = 800 Since the B-line has now decreased to 800, from (1) and (3) : log (1 - a,) -=. -0.0000969 X 800; 1 - a, = 0.837 A, -= 1.25 X 0.837 X 2000 = 2090 Next, the A-line now being 2090, from (2) and (4) : log (1 - b,) = -0.0000969 X 2090; 1 - b, = 0.627 B, = 1.25 X 0.627 X 800 -= 627 Continuing the calculations, the following series of population sizes is obtained: B-line A-line Bo = 1000 A, = 2000 B, = 800 A, = 2090 B, = 627 A, = 2270 B, = 473 A, = 2550 B, = 334 A, -= 2960 B„= 216 A„= 3530 The ratio 3530:216, or approximately 16:1, is established 10 years after the initial increase of the A-line to twice the size of B. A feature of some interest above is the fact that as B decreases and A increases, the sum of A and B in adjacent years eventually increases. In other words, the greater the difference becomes between the lines, the, greater becomes the total abundance of the two lines considered together. It does not necessarily follow that inequality of the lines is necessary to obtain maximum yield, for in practice K would sooner or later have to de- crease for the A-line (as a result of greater crowding), and it would probably increase for the B-line, until a new (and stable) equilibrium was set up; and the size of A and B at that point would be partly determined by the amount of change in K for each line —something which could differ in each locality. Nevertheless it is also clear that the condition of equality of the lines does not necessarily produce the maximum stock size for the combined lines, when cannibalism is a factor causing dominance. The above example is artificial in that it contains only a single arbitrary element—the initial displacement of line A. In a real stock there would be an "accidental" component in the survival rate of every year, mostly a result of environmental variability. If it were large enough, this could of course make the trend due to cannibalism unrecognizable in a series of any ordinary length. However, the environmental fluctuations usually observed in nature seem not to be large enough to conceal an effect of cannibalism of the magnitude indicated above, and in any event a long time scale is available (some thousands of generations). What would happen if the initial mortality rate due to cannibalism were really large, say 90%? Proceeding as above, with Ao = Bo = 1000, at equilibrium: K 1/(1 - 0.9) =10 log (1 - 0.9) k 1000 -0.001 199 Symposium on Pink Salmon

When A is increased to 2000: log (1 —13,) =-0.001 x 2000; 1 — = 0.01 B, =10 X 0.01 X 1000= 100

log (1 — a2) =-0.001 X 100; 1 — a, = 0.794 A, = 10 X 0.794 X 2000 =45,880 -16 log (1 —130 = —0.001 x 15,880 = 16.12; 1 — 1)3 = 1.32 X 10 -13 B4 =_ 10 X 1.32 X 10--16 X 100 = 1.32 x 10 - Under these conditions the B stock becomes exterminated in the second generation after the initial displacement. Examples like this show that any condition involving a high or even moderate rate of cannibalism is extremely unstable and could only lead to rapid extermination of one of the two lines.

DISCUSSION DR. KASAHARA: One of the important matters may be just a matter of chance. Any reproductive curve has considerable fluctuations in itself. DR. RICKER: I am sure that chance—that is, effects of a variable environment—is important in many situations. In fact, in all situations. On the other hand, there are these examples where something more than chance seems to be involved; where the effect persists for many long years or has a long uniform trend and it is very difficult to understand why repeated planting of eggs in the off years has not given the expected results. MR. MAGNUSON: Is there any indication that in these years of different densities that not only the growth rate is depressed but that the variability in growth is different at the different densities? DR. RICKER: I believe that information may be available in Russian publications, but I don't have it at my fingertips. MR. GILHOUSEN: I want to point out one item concerning the extreme dominance in the Fraser River pink salmon, which is that the dominant line occurs in the odd years, and also that in the early years of Fraser River sockeye fishery an extreme form of dominance existed amongst the sockeye, also on the odd years. Of course, every fourth year, not every second year. MR. GILBERT: One minor point. I recently came across some old records of Bristol Bay kept by Mr. Tom Wooton that would indicate that the odd year run in Bristol Bay in the early years may have been an artifact in that these records indicate that chums were called pinks and packed as pinks in the early years. Only on the even years do hump backs show up in the early catch records of Mr. Wooton. It may modify both the odd and even. DR, RICKER: There was also another artifact in that chart of the Bristol Bay runs (Fig. 13). The very steep decline was partly due to the fact that the regulations had stopped fishing earlier in the last two years and cut off most of the pink salmon. MR. VERHOEVEN: Northern Clarence Strait has an odd-year dominance, and when I tried to examine this once rather superficially by comparing the success of fishing at similar dates on odd and even years, the runs would be equal until the latter part of the season. Then suddenly in Northern Clarence Strait you would get a sud- den influx of fish which would make the odd year dominant over the even year. DR. RICKER: That happened in the odd years only, is that right? MR. VERHOEVEN: Only in the odd years. In fact, in talking to the fishermen, they told me that they wouldn't put their traps in on even years in the area if it wasn't necessary in order to maintain their rights to the trap sites. DR. RICKER: The data I had were only up to 1927. This difference has persisted, has it, in recent years? MR. VERHOEVEN: Yes, it has. DR. RICKER: Incidentally, if anyone is looking for a really useful project, he might continue the statistical analysis of Rich and Ball down to the present day. DR. SHEPARD: I think one factor in this phenomenon of difference in size in odd and even years involves two things: One is the actual contribution of the various spawning 200 Regulation of the Abundance of Pink Salmon Populations

streams to the total stocks, and also the timing of the runs to the different part of the streams, as we have already heard yesterday. At least in British Col- umbia one reason for the difference in even and odd years in size of fish is the fact that in one year you will have a late run, and in another year you will tend to have an earlier run. Inherently, the late run fish will tend to be larger because they spend a longer time at sea than do the ones that come early. The fry go out earlier in the first place and they come back later. Therefore, their total duration at sea may be as much as 20 percent longer than early-run fish. I think this, in part, explains some of these differences in size from the even and odd years. DR. RICKER: Do you think that would apply in these Asiatic examples? DR. SHEPARD: I am not familiar with the exact timing of the runs, but we know that even within our own river systems, and in even tributaries to river systems, that there may be early and late run segments. The fish will differ in size in the way I have described, the later running ones being larger.

201