rs H^ Ai6 DEPARTMENT OF COMMERCE AND LABOR 3UREAU OF FISHERIES GEORGE M. BOWERS, Commissioner SOME OBSERVATIONS % ON SALMON ANKTROUT IN ALASKA Bureau of Fisheries Document No. 627 WASHINGTON GOVERNMENT PRINTING OFFICE 1907 Pass u Book '4 J L. SOME OBSERVATIONS ON SALMON AND TROUT IN ALASKA. By F. M. Chamberlain,^ Naturalist, U. S. Fisheries Steamer Albatross. Bureau of Fisheries Document No. 627. ^7^ DEPARTMENT OF COMMERCE AND LABOR lA ,^ BUREAU OF FISHERIES GEORGE M. BOWERS, Commissioner SOME OBSERVATIONS ON SALMON AND TROUT IN ALASKA Bureau of Fisheries Document No. 627 WASHINGTON GOVERNMENT PRINTING OFFICE 1907 DEC COT^TEI^TS. Page. Introduction 5 Dist inctive characters of the species 6 Common names • 6 Differentiating marks in adults 6 Designations of young salmon 7 Hybridization 10 Resemblance of the youngs 11 Detailed descriptions of the young 13 Humpback salmon 13 Dog salmon 14 King salmon 15 Coho salmon 16 Sockeye salmon 17 Steelhead trout 19 Charr, or Dolly Vardeu trout . , 19 The basins studied 20 Conditions controlling the work 20 Method used to determine migratory mnvements 21 The Naha 22 Character of the stream 22 Yield of salmon 23 Catch of young salmon in the trap 25 Yes Bay Stream 27 Karluk River 28 Suitability as a spawning stream 28 Movements of young salmon as shown by trial calches 28 Conditions reported in British Cohiml)ia 30 Summary of observations 31 Young salmon in fresh water 31 The sockeye 31 Movement of fry above the lakes 31 Food and feeding 32 Food supply in relation to the hatching season 34 Growth in fresh water 34 Migration of yearlings ^ 36 King salmon 40 Age and season of migration 40 Effect of change from fresh to salt water 42 Food '. 43 Coho salmon 44 Earliest migrations -. . . . 44 Food and habits - 44 3 4 CONTENTS Young salmon in fresh water — Oontinued. Dog salmon Humpback salmon Trout, and t^harr Sea habits of young salmon Notes afforded by collections and records The sockeye King salmon Coho Dog and humpback salmon C(jnclusions from available data Abundance of food Return of adults to fresh water Approach of schools Food and feeding The sockeye King salmon Coho Humpback and dog salmon Relation of food supply to number of adult salmon Age of adult salmon Salmon-marking experiments Methods Regeneration of lost parts Salmon in the Trocadero, at Paris Factors influencing return to fresh water .^ Sex instincts versus condition of nutrition The different runs Temperature Currents Ascent of streams Interval between arrival and spawning i . . Parent streams "Introduction into streams not previously lVc(|U(nte(r ' ' 'Return of marked salmon" ' 'Distinctive and characteristic runs' ' Variations in weights and measurements Variations in counts Streams not utilized by sockeyes Relation of size of run to spawning area Selection of spawning ground Conditions required by the sockeye Preferences of the king salmon Spawning streams chosen by the coho, dog. ami humpback salmon. . . Nature of spawning beds selected Deposit of eggs Completeness of spawning 100 Percentage of natural production 101 Relation of spawning habits to number of fish 102 Changes incident to maturation , 104 Return of adults to salt water 106 Enemies of young salmon 107 Geographical glossary - 109 SOME OBSERVATIONS ON SALMON AND TROUT IN ALASKA. By F. M. Chamberlain, Natiirnlist , U. S. Fisheries Steamer Albafrnss. INTRODUCTION. To assist in the solution of various problems, it was deemed desir- able in connection with the Alaska salmon investigation of 1903 to establish shore stations. The work at these stations was to comprise not only the study of the habits of the salmon in fresh waters and adja- cent bays, but, in addition, a reconnaissance of all the neighboring basins with reference to available hatchery sites, observation of the methods employed in taking fish for the canneries, an inquiry into the sea habitat and the factors influencing the return of the adult fish, an inquiry into the efficiency of the hat6heries then operated, and a gen- eral study of the biological features of territory immediately adjacent to the stations. The facilities offered by the establishments of the Alaska Packers' Association at Loring and at Karluk determined the adoption of the Naha and Karluk rivers as localities for this work.*^ The greater part of the data obtained at these shore stations is con- tained in unpublished reports. In the present paper are presented such of the facts as bear upon the natural history of the salmon and, with a view to the application of these results in future work, some notes on the methods used in the inquiry. Most of the material contained herein relates to young salmon, but the known facts in the life of the adult, including the spawning period, are considered, and mention is made also of the trout as associated with the salmon. A chapter differentiating and describing the species, particularly in the finger- a The observations in the latter region were made between the early part of May and September, 1903, by the late Cloudsley Rutter, naturalist of the steamer Albatross, assisted by M. H. Spaulding, of Startford University. At Loring the work was carried on during 1903 and 1904 in charge of the writer, assisted at different times by E. L. Goldsborough and H. D. Aller, of the Bureau of Fisheries, and H. C. Fassett, fishery expert on the steamer Albatross. During the summer of 1905 observations along some of the above lines were continued by the writer at Yes Bay, in connection with other workof the ^Zfeafross, with the assistance of Mr. Fassett and of J. S. Burcham,of Stanford University. 5 LOWER PART NAHA BASIN Based on Stenometer and Compass Survey B Y E.L.Goldsborouqh AND F.M. Chamberlain scale of fe et O lojo B«o 3oo« Grace Peak eioo ft. Clif€ "^CvKoosnek I . ortmann Hatcherj^[ /CO. 6 SALMON AND TROUT IN ALASKA. ling stage, is followoMl l)y a detailed record of the field ol5servations on the young, and upon these tlata, which are largely statistical, is based the succeeding discussion of the habits of the salmon from the time of the migration of the young imtil the return of the adults to the spawning beds. Liberal use has been made in this discussion of the results of previous workers on the sul)ject, and, in addition to the collections and notes made by the shore parties in Alaska in 1903, 1904, and 1905, all of the Albatross collections of young salmon now in possession of the Bureau have been studied. The concluding chapters of the report are given to the discussion of questions relating to the adult salmon, and contain in tabular form the statistics of weights and measurements of nearly 10,000 sockeye salmon, with anatomical counts of about 4,600 of these. In the entire paper the sockeye, as the most valuable commercial salmon of Alaska, has been made the main feature. DISTINCTIVE CHARACTERS OF THE SPECIES. COMMON NAMES. The most common appellations of the various species as they are known in Alaska are used m this paper. There seems to be no reason why, for example, the words ''quinnat," ''chinook," and "king," which in a loose way pertain, respectively, to the Sacramento, Colum- bia, and Alaskan tschaunjtscha, should be continued to the confusion of readers. The "blue-back" of the Columbia differs in no specific essential from the "sockeye" of Alaska. Names based on characters common to several species, such as "red," "silver," etc., are especially apt to be confusing. The name "trout" as here applied to small fingerlings may include steelhead, rainbow, and cutthroat. Individual variations in both trout and salmon overlap the limitmg points in most characters, but in the salmon the sum of various char- acters sets the five American species distinctly apart. Careful exami- nation has so far failed to show any distinguishing character to differentiate the young of these trouts. Indeed, apparently no specific difference is constant in the adult. DIFFERENTIATING MARKS IN ADULTS. Fishermen and large handlers of salmon roughly but very accu- rately distinguish the adults of different species by certain obvious characters. The king salmon is knowft by the small black spots on the tail. The tail of tlie humpback is spotted, but with larger oblong spots. The backs of both king and coho are commonly spotted, the spots of the coho being as a rule smaller than those of the king, but these spots are not noted by fishermen. Rarely the sockeye shows a few spots, particularly on the tail, but these are never distinct as in the other species. (To the sea-run form only does this statement SALMON AND TROUT IN ALASKA. 7 apply; the dwarf lake-dweller is spotted.) The steelhead, also a spotted fish, is at once recognized by its slimness, the square tail fin, and the deep caudal peduncle. It is difficult to pick up a steelhead by gi'asping the tail, wliereas a salmon may be readily so held. Sal- mon without spots — sockeye, dog salmon, and sometimes the coho — are somewhat confusing. The sockeye is usually distinguishable by the blue back after death, the sharp nose, and the narrow maxillary, while the tail fin lacks the produced pointed lobes of the dog salmon; also the scales are firmer and show a clean-cut paving, and the flesh is intensely red. The coho and dog salmon scales when silvery appear to be of a finer and more delicate texture than those of the sockeye. This is especially notable in the dog salmon. The female dog salmon is usually very deep, both dorsoventrally and in lateral thickness — ''plumpness." The caudal peduncle is less compressed than in other species and the curves joining it to the body are shorter than in the coho. In the Alaska fisheries the dog-salmon males are seldom taken until the secondary sex characters begin to be developed — the "hook- bill" and dirty coloration. '^ The produced caudal rays mentioned above and the light color of the flesh are distinguishing marks in this species. The coho is distinguished from the dog by less delicate scales and deeper peduncle, by its small pupil, and in general by the occurrence of spots often at first overlooked. In closer examination one will consider the greater length and fineness of the gillrakers in the sockeye, the increased number of anal and branchiostegal rays in the king, the fine scales of the hump- back, the large and few pyloric coeca of the coho. The dog salmon possesses no single diagnostic mark, but difl^ers from the other species respectively in each character as mentioned above. Changes incidental to the spawming period will be noted under that head. DESIGNATIONS OF YOUNG SALMON. The lack of a distinctive terminology for the young of fishes has led to much confusion in the interpretation of reports of fish cultur- ists and investigators, more or less consequent acrimonious debate, and some legal entanglements. In an efl^ort to settle this matter for the benefit of American writers and readers, the American Fisheries Society in 1905 adopted the following nomenclature:'' Fry=fish up to the time the yolk sac is absorbed and feeding begins. Advanced fry = fish irova. the end of the fry period until they have reached a length of 1 inch. Fmgerlings=iish between the length of 1 inch and the yearling stage, the various sizes to be designated as follows: No. 1, a fish 1 inch in length and up to 2 inches; o " In handling a large number of dog salmon in 1907 it was observed at Juneau early in September that many green silvery males were being taken in the traps at Shelter Island and vicinity. Among these were several dog-salmon grilse." (Fassett.) *> Report of the Commissioner of Fisheries for fiscal year ended June 30, 1906, p. 24. 8 SALMON AND TROUT IN ALASKA. no. 2, a fish 2 inches in length and up 4-o 3 inches; no. 3, a fish 3 inchesln length and up to 4 inches, etc. Yearling s={is[\ that are 1 year old,, but less than 2 years old from the date of hatching. These may be designated no. 1, no. 2, no. 3, etc., after the plan described for finger- lings. • These definitions have been generally adopted in government and state reports and are the ones used in this paper. The use of the French term "alevin" instead of "fry" for the larval stage of sal- monids has been abandoned here for the reason that the French writers do not restrict the term to that period of development of the young fish. Though the use of "alevin" has had the support of such authorities as Francis Day«, Livingston Stone^, and Cloudsley Rutter'^, it seems unnecessary, if not even absurd, to continue the use of a foreign word and give it a meaning not recognized in the language from which it is drawn, more especially as even the writers mentioned above did not make a strict application of the term. Some French writers'^ have apparently endeavored to make a technical use of the terms "alevin" and "fretin," but their example has not been followed by later writers, and the word alevin seems to be used now to designate the young of the salmon in the most general way.^ In the same manner the German word "Brut," or "Jung- brut," has about the same latitude as has been given hitherto to the word " fry " in English, nor does the German term " Setzlinge " admit of strict application.^ The words "larva" and "larval" have been used by many writers in descriptions of the young salmonids. Others would restrict these terms to fishes exhibiting a greater change in the stages, such as the eels and the ladyfish. As in neither case is there a complete meta- morphosis, this limitation is scarcely tenable. The terms, however, are not yet current among fish culturists. The great diversity of size among species and among individuals of a given species at the time of hatching, as well as the intimate dependence of fish growth upon environment, in some cases may a Francis Day, British and Irish Salmonidge, p. 43, 44, and 82, 1887. & Livingston Stone, Domesticated trout, p. 151, 6th ed., 1901. c Cloudsley Rutter, Natural history of the quinnat salmon. Bulletin U. S. Fish Commission, vol. xxii, 1902, p. 69 and 72. d Larbale trier, Albert, Traite-Manuel de Pisciculture d'eau device, p. 220, 1886: "Alevins. — Les jeunes poissons venant d'eclore portent le nom (Valevins. Toutefois, il est a remarquer que cette denomination s'applique surtout aux jeunes dessaumons, truites et ombres-chevalier, tant qu'ils n'ont pas resorbe la vesicule; aprfes, ils consti- tuent \q fretin; pour les carpes et autres cyprins, quelques auteurs prefferent I'appella- tion defeuilles. Cette distinction ne nous semble pas n^cessaire; d'ailleurs, nous ne sommes pas seul a penser de la sorte, car le nom d 'alevin tend a se generaliser." «C. Raveret-Wattel, La Pisciculture, vol. ii, 1907, p. 185. /Paul Vogel, Ausfiihrliches Lehrbuch der Teichwirthschaft, p. 334, 341, 347, 349, 1898. SALMON AND TROUT IN ALASKA. 9 seem to introduce inconsistency into the definitions adopted. Thus under some circumstances a yearling charr might be found of less size than a salmon fry; but if is believed that by use of the above nomen- clature an exact interpretation of language will always be made possible. As in all instances involving individual characters, physio- logical and physical limits may overlap. Salmon fry usually begin to feed before the complete absorption of the yolk, a remnant of the yolk being persistent even for some time after the ventral walls have united and all outward appearance of the sac has been lost; and hence the migrating young of the salmon could with equal propriety be termed fry, since many still have yolk remnants, or fingerlings, since they have begun to a small degree to feed and have become over an inch in length. In this paper, in all cases where the schools contain many individuals with the embryonic fin membrane still evi- dent, the term "fry" is retained. In case of the humpback and dog salmon young which were taken in salt water, it is sometimes impos- sible to know whether "fingerling" or "yearling" is the proper desig- nation, but since it seems probable that the greater number were less than one year of age, the term "fingerling" is used. In the case of the coho the distinction is even more doubtful, but is applied with as much discrimination as the state of our knowledge permits. The terms "parr" and "grilse" have come into American use from the British writers on Salmonidse, but the latter designation has attained a meaning somewhat dift'erent from the original. "Grilse," as applied to the Atlantic salmon (Salmo solar) by both American and British writers, refers to the incompletely grown fish which return from the sea to the rivers to spawn. Unlike the Pacific salmon (OncorhyncTius) , the Atlantic salmon, both male and female, mature perfect sex products before completing their growth, and after spawning in the same manner as the grown fish of one or more years' greater age, return to the sea as "grilse-kelts" to continue their growth and return the following season (or second year after) as "salmon." What are known as "grilse" among Pacific salmon are the small males, presumably lacking at least one year of the usual age of adults of the species, which leave the sea for the spawning beds, mature per- fect milt, but after "spawning" die in the same manner as fully grown males. Females in no instance show this precocity. While there are sometimes small females among mature fish, they grade into the regular size in such manner that they can only be supposed to be those individuals which by heredity or unfavorable environ- ment have failed by a greater or less degree to reach the standard size. (See p. 86-87.) The term "parr" is applied in general to young fish still in fresh water and showing the dark bars or parr marks. They may be fin- gerlings, or yearlings, or even adults. The males of this stage in the 10 SALMON AND TEOUT IN ALASKA. case of the Pacific salmon in some instances mature perfect sperma- tozoa, but whether they spawn and die immediately afterwards is not known, nor is it known definitely whether fingerling parrs may develop the sex product. The terms "smolt" (frequently still spelled and pronounced "smelt") and "kelt" as used for stages of the Atlantic salmon hardly have parallels in the case of the Pacific salm'on. If any of the genus OncorJiynchus return to sea as kelts, it has not yet been fully demonstrated, and almost the entire weight of evidence is against the belief that it ever occurs. The term "smolt" (French "tacon") is applied, in contradistinction to parr, to that stage of Salmo solar when, in fresh water, the parr marks are lost and the young fish assumes its livery of silver in preparation for its descent to the sea." It might be used with some propriety of the yearling migrating sockeye, but it seems undesirable to confuse further the meaning of words which have their proper use only with the eastern species. HYBRIDIZATION. The question of natural hybridization has never been investigated, though it has long been well known that trout may be artificially crossed and fertile hybrids produced.'' That the species of salmon may be variously crossed with success has also been demonstrated, but owing to the difficulty of retaining them in fresh water until of breed- ing age the fertility of salmon hybrids has not been proved. Rarely adult salmon are taken which seem to possess characters of two spe- cies, but on the basis of predominating characters they have been assigned to one or the other of the species and the possibility of a hybrid ignored. The differences in time and place selected by the different species of salmon for spawning minimizes the possibility of natural hybridization; and the deficient vitality of crossed eggs and hybrid fry is, perhaps, sufficient to account for the failure of most if not all such accidental product when natural vicissitudes must be overcome. Moreover, the young of the different species of salmon are distinct and show characteristically distinct habits. This is not so evident, however, in Alaskan trout. Rainbows and steelheads spawn together in the Naha. Spawning cutthroats have not been noted there because they do not happen to inhabit that portion of the Naha which was under observation; but their segregation is inconstant. One species or the other may be most numerous in the lower or upper reaches of a stream. In the Naha basin cutthroats are more numerous a Day, op. cit., p. 90. b For a full discussion of hybridization of trout, see " British and Irish Salmonidae" by Francis Day, p. 47-50, 254-270, pi. x and xi, 1887, and Paul Vogel, op. cit., p. 308, 311. SALMON AND TEOUT IN ALASKA. 11 in the upper sections of the system; at Yes Bay the case is reversed. If the fry and fingerhngs of the three recognized species possess any distinctive marks or habits these have so far escaped detection. RESEMBLANCE OF THE YOUNG. By one who knows the adult salmon, the young of the salmon are not apt to be confused with any fishes except trout. Of the fishes having an adipose dorsalthey are readily distinguishable from the capelin, smelt, and eulachon by the siphon-shaped stomach and numer- ous coecal appendages; the grayling is known by its high dorsal fin; the whitefish by its comparatively small mouth. The young of the true trout very greatly resemble the spotted species of salmon, and are distinguishable mainly by the short anal,fin. The salmons usually show at least 14 fully developed — that is, full length — rays, exclusive of the 3 to 5 short, simple (un- branched) rays in the front of the fin. Sometimes there are but 13, and in rare instances only 12, while the number may be increased to 17 in the sockeye, which normally has 14, and in the king salmon, with 16 as a normal, the extreme will doubtless be still greater. The trout have normally 10 to 12 developed rays in the anal, but while this seems little separated from the number given for the salmon, as a rule the extra length of the fin in the latter may be noted by even a casual comparison. When spread the fin of the trout fingerling shows a rounded outline, the front rays somewhat more evenly graduated than in the salmon and the hinder rays much higher in proportion, causing it to resemble the dorsal in general outline. The anal of the salmon fingerling usually has a slightly concave outer margin, the hinder rays being shortened. The charr (Dolly Varden, or "salmon trout"), in addition to the characters of the anal as just described for the true trout, is distinguishable by the peculiarly mottled coloration, and in the advanced stages also by the less com- pressed body. The different species of salmon (Oncorhynchus) are not usually difficidt to distinguish from one another, yet individual variations sometimes confuse the determination. While in adults the sum of various differentiating characters makes it possible to decide the species, the late development of some characters greatly increases the problem in immature examples. It is possible that some local variations obtain; for example, sockeye fingerlings from Wood River seem to have a much smaller eye than fingerlings of the same size and species in vSoutheast Alaska. There is also much variation in the length and number of gillrakers in the young of this species and perhaps also in those of the humpback salmon. From the number of specimens at present available for study it can not be determined whether or not this variation is of geographical origin. 12 SALMON AND TKOUT IN ALASKA. The humpback (0. gorhuscha) is unique among the salmons in never developing the parr marks. It is usually entirely silvery in all the young stages, and is the only one of the salmons whose young up to some 6 inches in length, taken in fresh water, wdll show" no parr marks even under the scales. As possible exceptions to this must be noted the dwarf sockeye, whose young are as yet unknown, and perhaps small mature fishes returning for spawning. The fingerlings of the dog salmon {0. Iceta) have the parr marks sometimes much reduced, and those readily disappear in poorly preserved specimens. But where several examples are at hand the greater average size in the younger stages, the slightly different out- line, and the presence of these marks will always distinguish them from the humpback. Of the four species of strongly marked fingerlings the steelhead (Salmo gairdneri) may be recognized by its short anal, as noted above for trout, and by its small size taken at the season of migration. The sockeye {0. ner-kci), king (0. tscliawytscha) , and coho (0. TcisutcTi), being all for some time resident in fresh water and hence very vari- able in size, require close inspection for trustworthy identification. In general, in the fingerling stage the sockeye wiU be recognized by its more slender and tapering form, and by the more nearly circular outline of the parr marks, though this latter does not always hold true. The coho is usually distinguish-able by the orange tints of the lower fins (a character rarely absent) and by the white front margins of both anal and dorsal, but especially by the falcation of the anal through the extreme production or elongation of the first developed rays. The king, very similar to the coho in general outline, does not exhibit tliis extreme form of the anal, and in the specimens examined from the Karluk River the parr marks are larger and the marking of the back much more notable. In a careful examination the sockeye can almost always be identified by the greater length and number of the giUrakers, and the king by the greater number of branchiostegal and anal fin rays. (See detailed descriptions following.) In salt water the parr marks are rapidly covered by the bril- liancy of the silver, so that, except the king and coho, which are spotted, all the species soon become plain. It is not known when the spots of the adult humpback first appear. Whether this obliteration of the parr marks by the silver overcast is caused by the salt water may be questioned, although in- the trout, which are known to run indifferently in salt and fresh water, the change is marked, fish from salt water being much more silvery. I have taken one example of sockeye yearling in Jordan Lake that seems almost as silvery as the salt-water individuals of the same size, but it is entirely unlikely that it had returned from the sea."^ "Sec under smolt, p. 10. Day, op. cit., p. 90. SALMON AND TROUT IN ALASKA. 13 Another change with residence in the sea is a rapid increase in proportional depth. This is most marked in the coho. Seven exam- ples from the head of Naha Bay, May 31, length 98 to 117 mm., had depth 4.6 to 5, while in four examples taken August 2, at the cannery wharf, 154 to 210 mm. length, the depth is 3.75 to 4.4, which approaches the normal depth of the adult fish with matured sex products (3.5 to 4 + ). DETAILED DESCRIPTIONS OF THE YOUNG. The following key will assist in the identification of small fingerlings. In larger examples — as large fingerlings and yearlings — the specific characters of the adult must" be the main reliance. Key for Identification of Young Salmon and Trout Between 1 and 2 Inches IN Length. With adipose fin, large mouth, moderate dorsal fin, siphon-shaped stomach. a. Anal fin long, at least 12 developed rays, the last of these much shorter than the first, giving the fin a straight or concave margin or outline Salmon. b. Showing no distinct parr marks. Back dark in dead examples gorbitscha. bb. Usually with distinct parr marks. c. Parr marks less distinct, mainly above lateral line, body comparatively slender; gillrakers short, equal to less than 2 interspaces; eye small. keta. cc. Parr marks more distinct, showing below lateral line; body rather slen- der or deep; eye large, d. Gillrakers long, equal to or greater than 2 interspaces; body rather slender; parr marks tending to become circular nerka. dd. Gillrakers short; body deep; parr marks well defined bars. e. Parr marks narrower than interspaces, often orange coloration on fins; branchiostegals and anal rays few, 13 or 14; anal with marginal stripes kisutch. ee. Parr marks wider than interspaces,' branchiostegals and anal rays many, 15 to 16, anal not striped tschawytscha. aa. Anal fin short, fewer than 12 developed rays, the last not much shorter than the first, the entire fin thus having a convex outline or margin, the height great in proportion to length of base Trout and Charr. b. Parr marks as bars, body compressed, depth carried well toward tail. .Trout. bb. Parr marks as roundish blotches with mottling above and below, body less compressed, tapering rapidly toward tail Charr. The Humpback Salmon, Oncorhynchus gorbuscha (Walbaum). The humpback fry (pi. i, ifig. 3) at the time of closure of the ventral walls average about 35 to 36 mm. in total length. Depth, greatest just beliind pectorals, 6 in length (exclusive of caudal rays) ; in ill-nourislied examples the slenderness becomes marked, this usually quite notable in late migrants; head about 3.5; eye 2.5 in head; nose round and blunt, tip of lower jaw scarcely reaching profile; contour in front of dor- sal fin little arched; lunation of caudal slight; pectoral rounded, 2 or a little more in head (measured from axil to tip of longest ray); longest ray of dorsal about 2.5 to 3; longest ray of anal 3 to 3.75; greatest length of ventral equal to or somewhat greater than greatest height of anal; gillrakers long and close-set, about 5-13 developed, longest equal to half diameter of pupil, and to the distance from first to fourth (3 interspaces) at upper end of lower limb. 14 SALMON AND TEOUT IN ALASKA. In life, back green to l)ronzy_ changing to indigo bliu; upon deatli; the sides sil- very with brassy luster and green tints; ventral surface silvery white, usually with- out marks of any kind, but rarely with a few faint short parr marks above the lateral line; a narrow median dark line on back; membranes of caudal greenish, with black punctulations; dorsal similar with a narrow darker front margin; pectorals and lower fins colorless; iris greenish silver. In preserved specimens, particularly formalin specimens, a magnifier shows minute punctulations over almost the entire side, the back, and the dorsal and caudal mem- branes, and sometimes on the maxillary, chin, and throat. The humpback fingerling (pi. ir, fig. 2) is little changed from the fry described, except in size. It is the most slender species of the genus, with head long and pointed. In 20 examples from Karluk Beach 65 to 92 mm. in total length, the average depth was 5.57 in length to base of middle caudal rays, extremes 5.12 to 6; in 6 examples 83 to 92 mm. long the depth was 5.54, head 4.4, eye 3.8 in head, pectoral a little less than 2, dorsal 2.26, anal a little more than 3, ventral 2.46. Branchiostegals 11-12 to 12-13; gillrakers 10 to 13 upper limb and 16 to 19 in lower limb, longest from IJ to IJ times diameter of pupil and spanning 4 to 5 interspaces. The length and number of the gillrakers is not infrequently greater than in the sockeye fingerling, but the absence of visible scales in the smaller individuals, and their delicacy and small size in the larger of the humpbacks, as well as the more slender body outline, will distin- guish these fingerlings from the sockeye. In the adult humpback the gillrakers are 11-17 to 13-19, the longest If to 2 times diameter of pupil, covering 4 interspaces. Colors of the fingerling: The dusky of the dorsal shows as a diffuse blotch on the front and distal portions, the dusky of caudal more intense toward the points of the lobes and at the base, but less marked than in the dog salmon; a Uttle dusky appears in the axil of pectoral. A few small black blotches on the upper side were noted in a few of some fingerlings reared at the Clackamas station. At this stage the scales, though very thin and delicate, may be made out with a magnifier or a good eye. The lateral line is a mere furrow and shows no tubing. In the dog salmon of the same or much less size the scales are evident and tubes of the lateral line distinct. In the Karluk specimens examined the ovaries are ribbon-like, whereas in other species the ovaries are more cylindrical and usually somewhat swelled at the ante- rior end. The Dog Salmon, Oncorhynrhus keta (Walbaum). The dog-salmon fry (pi. i, fig. 5) attains a length of about 40 mm. by the time the ventral walls are fully united, perhaps a gi-eater length than the fry of any other salmon of the genus. It resembles the humpback in general shape. Greatest depth just behind pectorals, 5.5 to 6 in body; head 4; eye about 2.66 in head; pectoral 1.75 to 2; dorsal about 2; anal about 2.5; ventral equal to anal; gillrakers little more than tubercles, about 5-11 to 12 may be made out, longest about one-fourth diameter of pupil, equal to about one interspace. In life, ground color bright grass green, becoming slightly darker on back and paling on lower side to an overcast on the silver; occasionally a brassy luster on back; lower parts silvery with the palest green iridescence; back with very fine black punctula- tions, fusing into numerous small black spots about the size of pupil from nape to base of caudal; a median dark line on back in front of dorsal, not marked back of dorsal. The punctulations cover sides to axil of pectoral in front, to about halfway between lateral line and ventrals, and surround caudal peduncle, bt>coming coarse and scat- tered below. The parr marks, from G to 12 in number, lie mainly above lateral line. SALMON AND TKOUT IN ALASKA. 15 Pectoral and lower tins uncolored, caudal and dorsal greenish, fading distally, black punctulations on membranes, dorsal with blackish front margin. Iris brassy. In preserved specimens the general color is silvery, with the dusky shades as described for the humpback, but never so dark on the back. Parr marks along lateral line elliptical or oval in shape, narrower and longer (deeper) toward the head; greatest diameter equal to eye, usually more numerous and naiTower than in the sockeye, and not becoming bars as in the coho and king; also more subdued by the silver overcast than in the two last-mentioned species ; along median dorsal line a row of small blotches sometimes coalescent into a mere stripe, the area lietween this and the parr marks usually spotted with round dots less than half diameter of pupil; occasionally a few broken blotches below. Dorsal and caudal membranes and first ray of pectoral dusky; other fins immaculate. The dog-salmon fiiigerling (pi. ii, fig. 3) is less slender in outline than the humpback, which, except that the dog salmon is obviously scaled, it otherwise resembles. In 26 examples from Whidby Island, taken June 30, 1903, 78 to 122 mm. total length, sexes in equal numbers, the average depth is 5.1 m body (4.87-5.4); head 4.4; eye 3.3 in head; pectoral 2—; dorsal 2+; anal 3; ventral 2.4; developed anal rays 13 and 14; branchiostegals 13-13 to 14-15; gillrakers 8-12 to 9-15, longest about equal to pupil, spanning about 3 interspaces (sometimes only 2); scales 130 to 146, average 139. In specimens from Dvmdas Bay, July 24, 67 to 103 mm., the gillrakers are somewhat shorter, f to once the diameter of pupil, covei-ing 2 to 3J interspaces; scales 129 to 136, average 132. In fingerlings from Naha Bay, July 2, 50 to 65 mm. in length, the parr marks are still apparent. In sea-run examples from Karluk Beach, July 24, up to 100 mm., they scarcely show through the scales, but are quite evident when the scales are stripped. The other coloration much as in the late fry stage; chin dusky; several rays of the pectoral with punctulations; tip of dorsal, except last ray, distinctly black; tip of caudal black, marked even in fork, this black tip of caudal distinguishing them roughly from the humpback of the same size, though the scaling must be the final test. The King Salmon, Oncorhynchus tschawytscha (Walbaum). The young king salmon (pi. i, fig. 4, figure and description from Karluk River speci- mens) at the time of the complete union of the ventral walls are from 35 to 37 mm. aver- age length. In general outline they are less slender and tapering than the humpback or dog salmon. The depth, greatest in the region of front of dorsal, 4.25 in body; head about 3.33 to 3.5; eye 2.5 in head; pectoral 1.75, vertical fins high, dorsal 1.5; anal 2; ventrals 2.33; gillrakers about 4-11, short and well separated, longest equal to one-half diameter of pupil, spanning 1 interspace. The increased number of branchiostegal and anal fin rays (about 16 in each) help to distinguish this species. Ground color of specimens in alcohol silvery, and except breast and in front of ven- trals, with fine punctulations; about 9 to 12 long narrow parr marks usually equal to or greater in width than the silvery interspaces and lying about equally on either side of lateral lure; a dark median dorsal line in front of dorsal fin, not so evident in caudal region ; between this and the parr marks numerous round or oval blotches about size of pupil, the larger of these alternating with the parr marks so that it gives the upper end (*f these the appearance of being symmetrically margined by the silvery of the ground color; caudal lunation comparatively deep. In the Sacramento the caudal some- times has a reddish tinge, and shades of yellow may appear on the ventrals and anal. Yearlings (pi. in) from Karluk Lagoon, about 137 mm. long, have the follow- ing measurements: Depth 4.4; head 4; eye 3.5-4; pectoral 1.6-1.7; dorsal 1.8-2; anal 2.3-3; ventral 2.2-2.3; greatest depth about midway between pectoral and dorsal, gracefully tapered to tip of the rather sharp nose. Gillrakers 10-13; longest equal to or less than diameter of pupil, spanning about 2 interspaces. (In adults 10731—07 2 16 SALMON AND TROUT IN ALASKA. from Oregon the longest gillraker is about If to 2 times diameter of pupil, covering 3 interspaces.) Color silvery, dark above; top of head, back, and sides to lateral line more or less thickly spotted with small roundish black spots about half diameter of pupil and less; distal portion of dorsal dusky, sometimes nearly black; caudal dusky; inner side of outer rays of pectoral dusky; anal and ventrals unmarked; parr marks visible: under silver. Distinguished at sight from coho by the longer anal and the usually somewhat smaller eye, but definitely by the more numerous and finer branchioste- gals and greater number of pyloric cceca. The Coho Salmon, Oncorhynchus kisutrh (Walbaum). The coho fry (pi. i, fig. 6) at the time of hatching is about 27 mm. in total length, the sac about 10 mm. The greater size and the peculiar shape of the yolk sac distin- guish the coho from the sockeye. These fry average about 35 mm. t^tal length at closure of ventral walls. The ven- tral membrane on either side of the ventral fins is persistent for a long period. In general shape the coho fry and small fingerlings much resemble the king salmon. Greatest depth just in front of dorsal, about 5 in body; thence diminishing toward nose; dorsal outline arched; head about 3.75; eye 2.5 in head; pectoral 2; vertical fins high, the front rays in both soon becoming extended; dorsal about 1.75; anal 2; ven- trals 2.75; caudal lunation shallow. Gillrakers very short, little more than tubercles.^ about 5-10. In Steelhead Creek, May 29, migrating fry had a ground color of smoky green with brassy iridescence, black punctulations everywhere except on throat and breast; these punctulations finest and most numerous on back, becoming coarser and more scattered on lower side; lower side with a red overshade or iridescence; mem- branes of fins with punctulations, but on pectoral and ventral these punctulations only close to body; remainder of these fins brownish orange; caudal-fin membranes orange ground, the fine black punctulations giving a dark effect, upper and lower (dorsal and ventral) margins clear brown, most pronounced on rudimentary rays, membrane between ray branches colored and dotted as between rays; membranes of dorsal with very pale orange shade deepening on rear, front with black margin equal in width to a central membrane, sometimes a narrow orange margin in front of this; anal membranes distinctly orange with punctulations as in caudal, these becoming close near margin at extremity of long rays and forming a black band about equal in width to two rays and joining membrane, tips of these rays and membranes white, giving the fin a white front margin and a black submargin equal to two or three times the width of white; adipose with orange front and top margin; dentary siu-face of mandible orange-brown at tip; maxillary uncolored, with scattered punctu- lations; iris bronzy to brassy; cheek with bronzy ground; opercular face dark; parr marks one-third to one-fourth as wide as long, increasing posteriorly in relative width (antero-posterior), and extending about equally above and below lateral line, the first partly under opercular flap, the last roundish and usually about under adipose fin, the penultimate sometimes under adipose; a narrow dark median line on back; smaller round spots on back sometimes alternate with parr marks. Viewed from above in water, the back shows ground color bronzy, with a few scattered dark spots,* narrow supraorbital stripes ending at nostril, the median dark line showing distinctly with bronze bands of about equal width on either side. In alcohol entire surface dusky, with punctulation except on a little of breast, some of the lower surface of head, and the paired fins. Parr marks present as 8 or 9 short and narrow bars, about one-half the width of the interspaces (variable but not usually so broad as in the king), and equal in length (depth) to about half depth of body, lying about evenly divided by lateral line, becoming more nearly round toward SALMON AND TROUT IN ALASKA. 17 the caudal ; a dark blotch at base and front of dorsal ; tips of front anal rays immacu- late, forming a distal white stripe along front of fin. A fingerling 58 mm. in length (pi. ii, fig. 4) has a depth of 4 in body; head 3.5; eye 3 in head; pectoral 1.5; dorsal (longest ray) 1.33; anal 1.25; ventral 2 (some of the first dorsal and anal rays are variably produced) ; gillrakers, 8-14, longest one-half diameter of pupil, spanning about 2 interspaces. Markings much as in earlier stages, but with additional markings on back. Short oval or triangular blotches appear between the parr marks, and numerous small, round, dark spots become irregularly scattered over entire upper surface, including top of head. The orange tints of the fry and smaller fingerlings largely disappear at about this size, the time of the change varying greatly in different localities. In the same degree the falcation of anal and dorsal shows local and individual variation, but it is always notable. The tips of the first 4 to 6 rays of both these fins, with the outer membrane, retain the orange color or become white, to form a stripe about half the width of the pupil at the margin of the fin ; this lies upon a submarginal band of black of similar width. The distinctness of these bands is variable, but in no case are they entirely absent in fingerlings of 40 to 100 mm., or even more, when in fresh water. At a certain point the elongated rays seem to stop growth in extension while the remainder of the fin continues its normal increase in size until the permanent form has been reached, after which all the growth proceeds together. Examples of 130 to 150 mm. show traces of the peculiarity of both growth and color, but later there is no variation from the normal shape as found also in the king. A sea-run yearling (pi. iv) from Karluk Beach, June 18, 1903, 150 mm. in length, had depth 4.75; head 4; eye 3.66 in head; snout nearly equal to eye; pectoral 1.5; dorsal 2; anal 2.8; ventral 2; gillrakers 9-14, longest equal to diameter of pupil, spanning about 3 interspaces (in adults from 1^ to 2 times pupil, spanning 2f to 4 inter- spaces). Life colors (Rutter) "back olive brown, thickly spotted with black, dorsal dusky, except last ray is pale. Caudal yellowish by transmitted light, tip dusky, the dusky portions greater on lobes; pectoral yellowish; parr marks distinct; iris some- wlijit golden." In alcohol 11 parr marks quite distinct, a little broader than inter- spaces and depth al^out half depth of body, spots on back round, the largest about J diameter of pupil, covering top of head and back as far as lateral line — extending on to upper edge of caudal; anal and pectorals with very few punctulations, lower parts silvery. The Sockeye, Oncorhynclms nerka (Walbaum). The sockeye fry at time of hatching measures some 24 mm. in total length, the sac itself about 9 mmi, varying in individuals. The yolk sac is approximately cylindri- cal in general outline, sometimes slightly deeper (dorso-ventrally) at the posterior end. It has little tendency to become pointed at the posterior ventral extremity, as in the coho. The ventral walls become completely united and the yolk disappears externally when the young sockeye measures about 32 mm. in total length. At this stage both dorsal and ventral outlines are arched (the dorsal the more in alcoholic specimens by reason of the contraction of the softer ventral tissues). The greatest depth is near the middle of the body just in front of the dorsal, about 4.75 to 5.5 in length to base of caudal rays (end of scaling). In poorly nourished examples the depth is greatly decreased in proportion to the length. Head about 3.75; nose rounded, blunt, length about equaling half diameter of eye; eye about 2.5 in head; pectoral and dorsal about 2; anal and ventral about 2.66 in head; gillrakers 5-14, in length about two-fifths diameter of pupil, spanning 2 to 3 interspaces. General color silvery, becoming dusky above. There is less tendency to metallic iridescence than in the coho, and the fins do not show the orange tints of the latter. 18 SALMON AND TKOUT IN ALASKA. The ventral surface is iniinaculatc, including all the lower fins. The fine punctula- tions with which the upper parts are shaded extend from just behind the gill-openings to slightly below the lateral line and posteriorly approach more and more the ventral surface until, at the caudal peduncle, they reach the lower fin membrane. The indi- vidual punctulations are larger and more separated on the lower side; the close setting above gives th(> back a marked dusky color. The membranes of the rayed dorsal and the caudal are similarly dusky. A single row of small blotches sometimes merging together occupies the median dorsal line from top of head to base of caudal. Along the lateral line or a little below it anteriorly the parr marks appear, from seven to ten small, rounded, or elliptical blotches about half the diameter of the eye and more or less evenly distributed between the gill-openings and base of caudal ; the depth of color and the approach to circularity of outline increase toward the caudal region; a dark spot on the opercle and one at the base of caudal complete this row. A second row of similar but smaller blotches sometimes appears between the lateral row and the median dorsal line. In transparent specimens the black of the inner ventral linings may show through slightly. Tip of chin dusky. The small fingerlings (pi. i, fig. 2), 35 to 50 mm. in length, show some change in shape. The caudal peduncle becomes slim in propor- / _ 2 ^-^''"^ \^-0i^ to the length and the general outline is more ^^^"iX^ Il^^^^^r^ \ tapering; the greatest depth, at a point about midway \J \y^ ^^^O between base of pectorals and front of dorsal, is about . & See report by W. N. Lockington, in Report Commissioner of Fisheries, California, 1878-79, p. 50. 40 SALMON AND TROUT TN ALASKA. been bred in fresh water- foi- several generations, the 4-year fish weigh from 2 to 4^ pounds. Tliese are from the Sacramentcr'fish which average between 15 and 20 pounds. At the same time the number of eggs has diminished from a normal number of about 6,000 to only 1,300 to 1,400. These examples demonstrate a possible effect of fresh-water resi- dence. Dwarf sockeyes occur in such lakes as Alturas, Wallowa, Seton, Ozette, etc., in all of which fish may come and go at will. No other species of the genus breeds in lakes or exclusively in lake tribu- taries. It may be that the sockeye is in process of evolution from an anadromous form to a permanent fresh-water resident. Part of the young apparently possess the primal instinct to return to sea at once the first year, another and greater part have changed to become one- year residents, and the smallest part have lost the to-ocean instinct entirely and remain to reach maturity in the lakes. Experiments as to the action of artificially reared young of the two races would be of the utmost interest and no little value to fish-culturists. The exist- ence of marine and fresh-water forms of the same species is well known in Atlantic Salmonida?, and as well in other families. The effect of change of habitat in the two forms respectively has in no instance been worked out, and remains for solution, one of the most interesting and impoi'tant j^roblems in ichthyology and fish-cultiu'e. KING SALMON. AGE AND SEASON OF MKIRATION. The observations on the Sacramento liiver have demonstrated that Ihe king salmon young in that stream for the most part leave the fresh water as soon as they are able to swim and feed. All the spawning occurs in the upj^er j^arts of the river and its tributaries, yet fry with unabsorbed yolk were sometimes taken only a short dis- tance above tidal influence." This same action was noted in the Karluk. On July 3 a trap wliich had been set overnight at the mouth of that river just above the lagoon took 1,300 fry, nearly all of which were king ; there were only 4 humpback and 7 sockeye fry among the examples saved. As the collector did not differentiate species it is impossible to say whether 7 were all of the sockeye fry in the 1,300, or whether that number was only the natural proportion among the 180-odd specimens saved. The humpback being of striking appear- ance, however, it is probable the 4 saved were the entire catch of that haul. It should be stated that the haul also contained 5 sockeye year- lings, 25 coho fingerlings, a few sticklebacks, flounders, and young trout. No other sets of this trap were made; hence it is impossible to state " Rtitter, Natural liistory of tlie quinnat salmon. Bulletin U. S. Fish Commission, vol. XXII, 1902, p. 92. SALMON AND TROUT IN 'ALASKA. 41 what period the run of Idng fry occupied, but inasmuch as a few were taken in dayhght, and the total number taken was so Large, it would seem probable that the migration at that time was near its height. No fry of this species were taken in seine hauls made in the river May 22. On the lower Sacramento the principal migration of the king fiy occurs in March and April. Ice never forms on their breeding grounds, and they are free to travel as soon as they are able, which the tempera- ture of the water and consequent time of hatching make possible at the above date. The adults reach the river in April, May, and June, and in August and September. They spawn in numbers from the middle of August to the end of September, and from the beginning of October until in December. The first eggs deposited begin to hatch about September, and fry begin to migrate about October, or even earlier, and continue throughout the winter, since no obstacle is offered by ice, and the fall and winter rains put the stream into the most favorable condition. The run is practically over by April 1 on the lower river. From that date on it will be noted that the examples taken gradually increase in size, showing them to be in a manner summer residents." In the Karluk the adults first appear about the mouth of the river in May and continue in small numbers into August. They are known to spawn in the river below the lake latein August, or at approximately the same season as in California. The different time of migration of the fry is accounted for by the difference in average water temper- ature. In the Sacramento the eggs are deposited in water of a tem- perature of 56° to 46° F., the winter temperature rarely reaching as low as 40°. In the Karluk the lake surface in August varies between 40° and 50°, or slightly above, and as the spawn of tliis species is deposited in the stream below the lake this may be taken to be the approximate temperature of the spawning beds. It is improbable that any great number enter the lake as adults to spawn in the lake tributaries. At an average temperature of 45° (it is probably less) during August and September eggs deposited in the latter part of August could not hatch before the middle of November. By tliis time it is probable that the temperature has dropped to freezing, and it is doubtful whether even the earliest eggs hatch before the advent of spring. The lake probably clears of ice and the water begins to warm up late in April or early in May. With the increased warmth, due to the long, periods of day in that latitude, the fry would have developed at about the date noted, namely, late in June and early in July. As there is little rain at that season, there would seem to be no cause for the downstream movement except age (development) and instinct. oRutter, op. cit., p. 93. 42 SALMOISI. AND TROUT IK ALASKA, As in the Sacramento, it was found that a small number of .young king salmon remain in the river until they reach the size of finger- lings. Five yearlings averaging 111 mm. total length were taken in Karluk Lake June 5; 2 females slightly smaller, June 23 and 30; 2 males, 115 mm., were taken in an upstream trap at the mouth of the river Ju\j 3, and 2 slightly smaller examples on the 5th. This would indicate that they were feeding in the river at this date. In a seine haul on the lagoon 4 males averaging 130 mm. and 4 females aver- aging 135 mm. were taken July 24, in company with large numbers of sockeye and king fry and a few coho and sockeye fingerlings (or yearlings) and small trout. It has not been noted that these feeding fingerlings migrate in schools, but Rutter has shown that in the Sac- ramento, with the growing scarcity of food in the fall and the oppor- tunity offered by fall freshets, they gradually forsake their summer quarters. The "summer residents" were also found in abundance in the Columbia River work of 1895. The altitude of these breeding waters and the proximit}^ of snow tends to reduce the temperatures below those of the Sacramento, altitude doubtless effecting for them what latitude does for the Karluk. The bottom temperatures given for Alturas Lake " can not be relied on, owing to the character of the instrument with which they were taken. It is probable that the bottom temperature at 150 feet is little over 40° F., as shown by later observations in Wallowa and other lakes. The falling air temperatures of the autumn must be closely followed by the water. The circulation in these lakes must be comparatively small and the cooling of the waters rapid. Dr. Evermann's researches show that the lake temperatures, and consequently those of the spawning waters, fall rapidly after August. Since the king salmon deposits its eggs late in August and early in September, it is likely that the early freezing of the water arrests the development of the fry, so that they can not arrive at the migrating stage until some time in the spring. But, in so far as the tempera- tures taken in Idaho are somewhat above the corresponding obser- vations in Karluk Lake, it is probable that the downstream migration of the main schools would have been completed before the arrival of the investigating party in July. The size and growth of the finger- lings found exhibit a condition exactly analogous to that described by Rutter for the Sacramento and observed in a less degree for the Karluk. EFFECT OF CHANGE FROM FRESH TO SALT WATER. Rutter demonstrated that the king salmon fry is unable to sustain immediate transfer from fresh water to sea water. In addition to his experiments on the California salmon, he made a test at Karluk. oEvermann, Bulletin U. S. Fish Commission, vol. xvi, 1896, p. 157. SALMON AND TROUT IN ALASKA. 43 August 2 about 600 fi\v and iingerlings were placed in a live-box 3 by 3 by 7 feet and towed from the river through the lagoon to the sea, a trip of 5 hours. About 50 died en route, 150 died during the fol- lowing night, and all but 70 in the next 24 hours. These were proba- bly all king salmon. The totals show that all under 2 inches died; 6 percent of the fiiigerlings up to 2h inches lived; 94 per cent of the larger fingerlings lived. This residt is singular enough when it is noted that the king salmon 3'^oung migrating out of the river are but If inches in average length. Tt is inconceivable that they remain in the fresh-water end of the lagoon until they have added 50 per cent to their size. Two-inch fish were taken by Eigenmann in 1890 at Mare Island; 2^-inch examples were taken by the writer in 189S in San Pablo Bay in brackish water. There is every reason to believe that fingerlings of the latter length under natural conditions are perfectly able to endure a standard salinity. The fact that this species' breeds only in rivers of considerable volume insures to the young the opportunity of making the transition with the necessary gradualness. In this they differ from the frequenters of short streams, the fry of which species are doubtless able to endure a sudden entry into salt water.'' The tendency of the resident king fingerlings to advance upstream has been pointed out by Evermann in the report of the Columbia River inquiry. The same habit was noted by Rutter, and the pres- ence of king salmon in the catches at Karluk Lake verifies the earlier observations, showing that the location of the breeding ground is not closely indicated by the presence of the young. FOOD. The food of this species is almost wholly insects, in large part from the surface. The yearlings mentioned above, taken in Karluk Lake, contained only insects, as did also the young taken at the mouth of the river; but the 8 taken in the lagoon contained amphipods in addi- tion. These large fish must be yearlings. Their presence in the lake can be accounted for only on the assumption that the fry or finger- lings migrated into the lake the previous season, or that the yearlings ascend the stream in the spring and summer. Since their natural food would become scarce earlj^ in the fall it would seem to be improbable that they would remain in the fresh water, whereas the length of the river and the unnaturalness of an instinct that would impel female yearlings to ascend it from salt water lay that supposition open to serious question. They remain to be accounted aberrant individuals « Questions regarding the change from fresh to salt water, and vice versa, made by all the salmon, have been discussed in detail by Sumner (Bulletin U. S. Bureau of Fisheries, vol. xxv, 1905. p. 53-108) and by Greene (Bulletin U. S. Bureau of Fisheries, vol. xxiv, 1904, p. 429-456). 44 SALMON AND TEOUT IN ALASKA. that have obeyed neither the instinct to descend as fry nor that to seek the sea in self-preservation upon the approach of winter. Migrating fry at the river mouth were largely empty, but a few contained insect remains. The small fingerlings taken in a seine haul in the lagoon also showed a purely insect diet. The lot taken by the trap at the river's mouth averaged about 42 mm. (37-53) ; those seined in the lagoon averaged 47 mm. (40-62), showing a growth of about 5 mm. since reaching brackish water perhaps not over a month previously. It w^as noted that late in July the number of these small fish seen about the margins of Karluk Lagoon had greatly decreased, indicating that the stay of the main body in brackish water is short. COHO SALMON. EARLIEST MIGRATIONS. As in case of the king salmon, a considerable number of coho young remain as summer residents in the streams of their birth or in the connecting lakes; but the greater part seek the sea as soon as they become free-swimming. In the trap at station 2j on the Naha, the fingerlings and yearlings taken largely exceeded the fry in number. Both were taken from the middle of May until in June — the migration period coinciding with that of the sockeye. It is possible that this migration is in large part influenced by the sockeye movement. Wlien yearlings of the latter species travel at the surface the cohos of similar size maj^ be impelled, to some extent, to school with them; but the general absence of yearlings in the lakes early in the summer indicates that the spring migration is instinctive and general for the species. The fry reach the swimming stage somewhat later in the season than the sockeye. The eggs, though requiring a slightly shorter incubation period than those of the sockeye, are deposited much later in the fall. By means of a trap which was set in Steelhead Creek on Naha Bay in 1904 the migration was found to be heavy as early as May 19, water at 48°. On this date over 1,100 fry were taken, the net spanning the entire stream. This run reached its maximum ten days later, when over 3,000 fry were taken in a single night. It continued, however, until sometime in Jul}^, when the temperature had reached 54°. In this creek the humpback and the dog fry left in Ma}', the coho from the middle of May through June, and the steelhead in July. These dates are approximate for these species in the Loring district. FOOD AND HABITS. The fry taken in the trap during the time from May to July 9 showed no appreciable increase in size, the catch average varying irregularly between 37 and 40 mm. Of 600 measured, 85 per cent were SALMON AND TROUT IN ALASKA. 45 between 3G and 40 — extremes 33 and 43 mm. A few fingerlings, less than 1 per cent, were taken at the same time. The food at that time, in the few that contained any, was insects. The main movement was early in the evening, the lifting of the trap at 1.30 a. m. and 9.30 a. m. of the same day showing a catch of 2,015 between dusk (about 10 p. m.) and 1.30 a. m. and 50 during the morning twilight. In the Karluk in 1903 the first specimens of coho fry were obtained May 22. At that time salmon hj and small fingerlings were said to be numerous in the little sloughs at the edge of grassy marshes where the current was slight. Over 1 ,000 were taken in a single haul of the seine. Of the 42 examples preserved from this haul only 2 were cohos, about 38 mm. long, 1 with insect food. Of the 17 fingerlings saved from the same haul 15 were cohos, 12 males averaging 68 mm. ami 3 females averaging 52 mm., all feeding on insects and larvae. It should be noted that the small fingerling sockeyes taken in this haul contained crustaceans, some also insects and insect larvae, while the 2 yearling sockeyes contained only insects and larvjB. Small fingerlings were taken in Karluk Lake July 27, 30, and 31, and as late as August 22. In all but 2 examples examined, which con- tained Crustacea, the food was insects and insect larvae. Coho young may be found in almost every brook of Alaska through- out the summer. They linger along the margins and in the pools, with no apparent seaward movement. As the adults penetrate all these small streams to spawn, the upstream movement of the young, if there be any, does not excite attention. They are resident in the lakes as well as in tlie tributary streams. With the fall rains these residents are swept out of the streams into the lakes or the sea in the same manner as the king. In the summer of 1905 many coho fingerlings were taken in Yes Lake. August 22, 15, averaging 95 mm. in length (63-122 mm.), were taken in a night haul. Two of these (13 per cent) contained sticklebacks, one having eaten 12, all less than 20 mm. in length; 66 per cent were feeding on flies, etc.; 13 per cent contained beetles, and 26 per cent snails; 13 per cent had eaten caddis larva^, and a like number other larvae. August 24, 55 were taken in night hauls of a shore seine, average length 85.6 mm., extremes 53-130. Of these, 91 per cent contained winged insects, as flies and ants; 42 per cent beetles; 14 per cent mites, eggs, etc.; 7 per cent sticklebacks; 7 per cent snails; and 5 per cent caddis larvae. September 10, 88 were secured in the same manner, average length 83 mm., extremes 51-120. Of these, 44 per cent contained ^^dnged insects and the same number caddis larvae, 20 per cent beetles, 11 per cent mites, etc., about 2 per cent sticklebacks, 5 per cent snails, and 7 per cent other larvae than caddis. It will be noted that the average size of 46 SALMON AND TROUT IN ALASKA. the latter lot is sliglitly smaller, but the decrease is not sufficient to indicate a movement of the larger individuals out of 4lie lake. A more significant fact is the growing scarcity of surface food and greater amount of bottom material — the caddis found in their stomachs. As early as April 6 (1903) coho yearlings of 145 to 165 mm. were gilled at the Fortmann Hatchery in Heckman Lake at the mouth of the river. They were attracted by the waste eggs thrown in the river. May 17, 1903, 11 males taken at station 2 averaged 119 mm., and 8 females 112; May 24 of the following year 12 males averaged 125 and 18 females 123. If these may be regarded as typical there appears to be a growth of about 50 per cent between September and May, or about twice that which from similar data was estimated for the sockeye. The yearling cohos taken in the Naha were found to eat the young salmon fry whenever taken with them in the nets. That they sometimes were able to pre}^ upon them in a natural state was evidenced by the presence of digested fry in some examples that were seined in Roosevelt Lagoon in May. September 12, 1897, the writer seined a number of 3^oung cohos at the mouth of Klawak River. No other fish except sticklebacks were noted. Of the 71 cohos saved the 37 males averaged 85 mm. in length (50-125); the 34 females averaged a small fraction of a milli- meter greater (47-135). At this time the hatcher}^ was operating and sockeyes were spawning in the lake tributaries above. Many of these young cohos contained salmon eggs. A more common food was a large maggot, probably the blowfl}^ larvae from dead fish along the stream. One stomach contained 2 leeches, and many of the smaller had insects. Examples taken with a hook in brackish water at the Klawak cannery wharf contained insects and a few beach crustaceans; 50 examples from the lake, October 14, 1905, averaged about 75 mm. These were taken with a dip net and do not give the average size of lake residents of that date. (See also record in salt water, p. 53.) Like the king fingerling the coho is readily taken with a hook, either with a fly or bait, salmon spawn being especially attractive in season. During the spawning season the coho in a few places may do some damage to the sockeye eggs. Large numbers are attracted to the scene of the hatchery work at Loring by the washings fyom the freshly spawned eggs. They were nowhere noted in any number on the natural beds. As the result of the wide diversity of spawning regions, the coho may be said to possess three movements seaward, first as fry, second as fall fingerlings in the same manner as the king, and third as yearlings, lake winter residents leaving in the spring with the sockeyes. This SALMON AND TEOUT IN ALASKA. 47 may be true also of the king salmon, but it is not believed, from the observations on the Columbia, that the king spawns above the lakes ordinarily, hence only the wandering young would winter in them. DOG SALMON. The dog-salmon young, so far as known, all leave the fresh water as soon as they are able to swim. The records of the occurrence of larger individuals in streams have not been authenticated. The dog-salmon fingerlings. reported in the rivers of Washington by Gil- bert and Evermann" on later examination were found to be cohos. A similar find by Davis'' likewise proved erroneous. This species breeds in the Naha in too limited numbers to permit observations of value. Fry were taken in only two instances at sta- tion 2. A few were taken at station 1 the middle of April, and again in Gibson and Emma creeks the middle of May. They were found in abundance, however, running out of the creeks of Deep (Moser) Bay, April 30, temperature 37°,- many of them still with remnants of yolk. A few were found in Steelhead Creek after May 7 and up to June 6. In the main their fresh-water habit seems identical with that of the humpback. Early in June, 1903, immense schools of small Imgerling dog salmon were seen leaving the Karta River. Examples taken on the 8th of that month about the margins of the upper Kasaan Ba}^ average about 40 mm. They were feeding on insects. It was reported by the workmen building the Alaska Packers' Association trap at that point that shoals of these young could be seen at times well out in the bay, where they were pursued by larger fish, apparently Dolly Varden trout. The great number of adult dog salmon spawning in the Karta River makes this easily credible. In the Karluk lagoon, 1903, fry and small fingerlings were ob- served in large numbers May 12-14. They lay close inshore by the spit, moving about in schools, but not going out v/ith the tide. Some of these still contained yolk, others were feeding on insect larvae, amphipods, and surface material. Slightly larger young were taken in the lagoon June 9-12, feeding on crustaceans and insects. June 18 they were noted as abundant outside on Karluk Beach, 6 taken averaging 50 mm. in length ; they were feeding on insects, crustaceans, and small cottoids. July 24 neither dog nor humpback young were present among the examples taken in a seine haul in the lagoon, all apparently having sought the sea. (See p. 52.) "Gilbert and Evermann, Bulletin U. S. Fish Commission, vol. xiv, 1894, p. 198. b Davis, Pacific Fisherman, vol. i, no. 4, May, 1902, p. 9. 10731—07 1 48 SALMON AND TROUT IN ALASKA. HUMPBACK SALMON. The fry of the humpback salmon leave the fresh water with the breaking up of ice. In 1903 they were first taken April 12 at station 1 on the Naha, but the number secured, 240, indicates that this was not the beginning of the run, though all the lakes were still covered with ice. The number migrating increased from this date to the end of the month, and they were found leaving the lagoon throughout May. During this time there was no increase in average size, the fry apparently leaving as rapidly as they attained sufficient strength. They traveled almost wholly b}" night, the heaviest movement apparently being on the first of falhng water after a rise. At these times the average size was slightly less than at others. On April 19 a trap set near station 3 took over 1 ,500 fry. There is but a very small spawn ng ground below Jordan Lake above the point where this net was set, and it is thought that the larger part of the run indicated by this catch had passed through Jordan Lake under the ice. The average size was a trifle greater than that of the fry taken at the same time at station 1, wholly because of fewer undersized-fish. Very few "humpback fry were seen at Karluk, a few were taken at the mouth of the river May 11, and again July 3, but at no time did they appear abundant. One example occurred in an experiment testing for endurance of salt water and was found to survive the test. While the young humpbacks are in fresh water, feeding is only incidental. A few have been found to contain remains of insects, larvae, and crustacea. As with the dog salmon and coho, the descent from the small creeks in which many are hatched makes it necessary that they be able to stand a quick change to salt water. In carrying a number from the trap in fresh water to the lower Naha Bay it could not be observed that immediate immersion in salt water caused them the least inconvenience. TROUT (STEELHEAD?) AND CHARR. The first trout fry appear in Steelhead Creek about July 1. On that date in 1903, 255 were taken in the trap, and on the following day 295. About the same number were present again in 1904, 283 being taken on July 9. While no further trials of the trap were made in 1903 the fry were noted to continue in the creek in large numbers until swept out by the fall rains about the middle of September. During this period their abundance was almost incredible when viewed with regard to the number of adults seen in the spawning season. In the lower Karta River July 26, 1903, trout fry were very numer- ous. At this time the temperature was 64° and the stream very low. Along the sandy bars in places the receding of the water had left SALMON AND TKOUT IN ALASKA. 49 pools in wh cli large numbers of fry were imprisoned. The tempera- ture in these pools was much higher than in the stream, and in many cases the'water had evaporated, leaving the fry to die. All of them were swept out by the first rise. At Karluk, August 29, 1903, in a small stream tributary to the small lake east of the main lake, Spaulding noted the occurrence of fry imprisoned in the same manner. From the date and place it is believed that these were trout fry. No examples were preserved. Small fingerlings and fry of trout may be seen during the late summer almost as widely distributed as the coho. So far as observed the food of the trout fmgerling is insects. The observations were made on July 11 in a pool about 6 to 18 inches deep at the foot of the fall in Flume Creek, where numbers of both trout and coho young fmgerlings were resident for several weeks. The coho were schooling near the surface; the trout inclined rather to scatter and occupy the bottom. One trout was observed to take a position on a rock, from which it would dart for food to either side and forward for some 20 inches, returning each time to the same resting place. During the eight or ten minutes it was observed, it made over twenty trips, once attacking and driving away a brother fingerling that tried to take a place upstream a.nd cut off the food supply. The trout rarely came to the surface, striking mostly at submerged drifting particles. Upon return to the pool two hours later the fingerling was found to be gone. The fry of the charr occur in very much smaller number. A slight migrqijtory movement was noted in Steelhead Creek in 1904. Two were taken in the trap May 29, 16 on June 5, and on the 13th 30, the last observed. Their habit is to burrow in the gravel in the small streams. In these places their protective coloration and agility permit them readily to elude observation. No notes were made on the food of these young. Fingerlings of all sizes of trout and charr were taken in the trap at various times, but no indication of any migratory movement appeared. Both species pass indift'erently to and from salt water. The seaward movement of the fry is unciuestionable, but individuals that remain in the streams after the fry period are doubtless governed only by convenience in obtaining protection and food. In the pools called Trout Ponds on Trail Creek only cutthroats and charrs were taken with a hook. These never exceeded a certain size, about 8 inches, though both male and female ripe cutthroats were taken May 30. In the brook below larger individuals were taken, a ripe female cutthroat being secured April 25. No steelheads were noted in Trail Creek, but a few trout fry may be found there. In Steelhead Creek only rainbow and Dolly Varden trout are taken below the falls, but 50 SALMON AND TROUT IN ALASKA. in the lake above the falls cutthroats are abundant. These falls are probably now impassable at all stages of water, and it isiiardly to be believed that fry of the trout in the lakes help to account for the abundance of trout fry in the stream. It is to be observed that their numbers decrease toward the upper reaches, so that for some distance below the falls few are seen, and none in the few yards of the stream between the lake and the head of the falls. SEA HABITS OF YOUNG SALMON. NOTES AFFORDED BY COLLECTIONS AND RECORDS. THE SOCKEYE. The sea habitat of the young sockeye so far has not been studied. The only observations available are the residts of occasional and irregular seine hauls made by the Albatross parties at various times, accompanied usually by no notes regarding exact locality or asso- ciated forms. These scant collections throw little light upon the sea habits or habitat. The gear used was capable only of taking exam- ples in comparatively shallow water, close inshore, on smooth beaches. Larger examples would scarcely be taken under these con- ditions even if present in the same waters, and smaller fry would not be held by the web ordinarily used in the vessel's seines. The smallest specimens in the collection are 8 examples aver- aging 41 mm., from Svimner Harbor, taken July 2, 1896. This harbor is a small bay northeast of the town of Unalaska. It receives a small creek, the outlet of a lake. The fry were taken in company with coho fry and fingerlings. The sockeyes were feeding on Crustacea, the cohos on both Crustacea and insects. If they were taken in the bay, they were doubtless recent migrants from the lake. These sock- eyes differ somewhat from the more southerly specimens in having shorter and less numerous gillrakers — about 10-17; in coloration they resemble some of the examples from Karluk, the parr marks being longer and more bar-like than in those from Southeast Alaska. A number of fmgerlings from Wood River, Bristol Bay, taken July 23, 1903, averaged about 41 mm. in length. The stomachs were filled with small crustaceans and insects. Wood River has little fall from the lake which it drains, and in spring tides is affected as far as the lake. No notes accompany the specimens, so they are of little biological significance. They seem to differ from the Southeast Alaska examples of similar size in being of less tapering outline and having a smaller eye. Six yearlings, average 98 mm., are in the same lot, with the same food present in the stomachs. The main run from the Kvichak River was reported to Mr. John N. Cobb, of the Bureau of Fisheries, as occurring from the first to the middle of June, fish from SALMON AND TROUT IN ALASKA. 51 3J to 4 inches in length. These are taken for eating by the Chinese at the cannery. Some are said to be pink-meated. The next smallest examples in the salt-water collections are 12 from Alert Bay, British Columbia, taken June 5, 1895, probably from the beach near the cannery. With these are a number of small dog fingerlings of about 40 mm. and possibly a few humpback. Many of the specimens are in poor condition and not positively identifiable. The sockeyes average 62 mm. (53-68). Their food is all pelagic material, small adult copepods, ostracods and amphipods, crab larvae, and a worm-like marine form — Sagitta. The other species were also feeding on Crustacea. From Nikolski, Bering Island (Komandorskis), there are 5 year- lings taken July 3, 1895; 2 males, 121 and 130 mm., 3 females, 133 to 135 mm. ; all with amphipods in the stomachs. From Kiska Island, June 7, 1894, 1 male, 196 mm., containing copepods, and 2 females, 230 and 245 mm., with stomachs empty. These are said to have been taken in a small lake, but data for this are incomplete. They are apparently not dwarfs. From Isanotski Straits, at the extremity of the Peninsula of Alaska, 5 examples were obtained July 15, 1894; 3 males, 123 to 135 mm., and 2 females, 132 and 143 mm. The only stomach left in the specimens by the collector contained small crustacea, schizopods, and amphipods. These fishes were all infested with a parasitic round- worm occurring in masses in the region of tlie air-bladder. They are also noteworthy for the shortness of tlie gillrakers. These (on the right side) average less than 33 in number, the longest equal to about the distance between 6, or spanning 5 interspaces. The speci- mens from Nikolski average 34, with a length equal to about 5J interspaces. Those from Kiska run high, averaging 36 in the 3 examples- at hand. Dundas Bay specimens, in a count of 12 some- what smaller indiAaduals, give an average number of 33 + with aver- age length a little less than 5 interspaces, whereas Karluk fish, in 10 examples taken June 8, 1903, show an average number of 35+ and a length equal to nearly 9^ interspaces. The examples from Dundas Bay were taken July 24, 1903, in a seine haul on the tide flats with mud and sand bottom. With them occurred dog and humpback fingerlings of about the same size, per- haps in the same school. The individuals saved are 6 males, average 80.6 mm. (71-91), and 6 females, average 77.7 (73-85). They were feeding on crustaceans, insects, and insect larva;. At Uganuk young salmon are taken in large numbers in seine hauls for sand launces. Of 25 pounds of these small fishes taken in this manner June 15, 1903, 36 per cent were young salmon, mainly sock- eyes, but in addition some cohos; 4 per cent were young herrings; the remaining 60 per cent were sand launces, or, as they are known 52 SALMON AND TROUT IN ALASKA. locally, candlefish (Amrnodytes). The sockeyes of this lot whose stomachs were examined contained crustaceans, a few with fragments of insects also; several had small herring fry. Seventy-two males averaged 82.1 mm. in length (47-105); 70 females averaged 81.3 nnn. (48-110). It is a curious fact that in these collections, where there are large numbers of apparently regular-sized examples, usually the males are found to average slightly the larger, whereas in cases of only a few large examples the females almost always exceed the males in size. At Karluk young salmon are found abundantly in the waters in which seining for the cannery is carried on. It is said these yearlings sometimes rush through the seines in clouds, and even in the large- meshed seines numbers are drawn ashore with the adults. From the catch on June 8, 1903, were saved 67 sockeyes, averaging 181 mm. (123-207). These were feeding on small Crustacea; of 20 examined none had fry of any sort in the stomach. The distribution of food was peculiar in that individuals were feeding almost exclusively on a giv«n form. For example, one was filled with ostracods, others had but few; some had almost nothing but copepods, others as exclusively amphipods; many, however, had such a mixture as would be expected. This seems to indicate that the crustaceans form schools to a certain extent, either in zones or otherwise, which enables the fish, acting as a tow-net, to obtain more or less nearly pure masses of a given form. Among the specimens saved from this haul other than sockeyes were only 2 cohos. One of these had eaten a young cottoid of 18 mm. length. On July 3 from a similar haul were saved 30 specimens; 12 males averaging 136 mm. (122-156), and 18 females averaging 139 mm. (125-164). Some of these were empty, most had been feeding on small crustaceans, several contained in addition small coiled sliells (pteropods), and 2 had small blennies and sticklebacks in some number. One coho occurred in this lot. July 24 young salmon were very abundant in Karluk Lagoon. A seine haul, with a 75-foot seine, covering about 250 square yards, took over 2,000 salmon, 30 to 150 mm. in length, and a few small trout and charrs. It is noteworthy that in this lot there seem to have been no humpbacks or dogs, whereas among specimens taken at the same time on the outer side of the spit, in the caniiery seine, no sock- eyes were saved. Wliether this was a peculiarity of the distribution at that time or whether it arose from some other reason it is impos- sible to state. It is known that the collector did not closely differ- entiate the various species. The young sockeyes taken in the lagoon varied from 30 to 145 mm. in length. The smaller are doubtless, from the hatchery, the yearlings may be from the lake; all were feeding on Crustacea and insects, tlie smaller fish more on insects SALMON AND TROUT IN ALASKA. 58 and their larvae, the larger on Crustacea, mainly an amphipod. In several of each size were masses of intestinal worms. It is stated that small salmon are seen in the seines throughout the canning season, but mainly in the earlier part. It will be noted that the larger specimens were obtained in June, and that after July 3 no collection of sockeyes was made from the cannery seines. It may be that the continual hauling in time frightens thein away; but it seems more reasonable to believe that with increasing age they move farther out from shore. The vast numbers observed in June must be the yearling migrants from the lake. As to what becomes of the fry of the same season after reaching salt water, there is no clue, and can be none until search is made for them with small-meshed nets. The only other specimens in the collection of interest in this con- nection are 2 males of 91 mm. and 5 females 84 to 104 mm., average 92.4 mm., taken in Karta Bay June 26, 1897. These contained mainly insects; some had also Crustacea, and one had a few young flatfish, THE KING SALMON. It is a significant fact that in the collections at hand there occurs but one young king salmon taken in salt water. It would seem there should have been some in the Karluk Beach collections if the species is accustomed to tarry near its parent stream. They are sometimes taken in San Francisco Bay and the region outside the straits, but none of these examples have become part of the Bureau's collection. They have been reported also from the region of Kil- lisnoo, Alaska. The example mentioned is a 215 mm. male, which was taken with the cohos (see below) at the Loring cannery wharf August 2, 1904. As this was preserved for a coho, it is possible that others of the 45 mentioned below were of this species. THE COHO. The young coho in salt water is more easily observed than the other species. It readily takes the hook, and apparently is less timid than the others in approaching surface and shore. In 1904 45 were taken at the Loring cannery wharf August 2. They aver- aged 190 mm. (158-226). On July 10 at the same place about 30 were taken. No measurements were made except of the largest, which was 138 mm. On August 2, 1905, a scattered school came about the Albatross vvhile anchored at the extreme head of Yes Bay; 26, averaging 202 mm. (152-237), were taken with a hook over the ship's side. Only a few, 6 or 8, would appear at once, and they took the hook baited Avith bits of meat, etc., very shyly in the per- fectly clear water. Most of the stomachs contained offal from the ship's messes; 5 contained fishes up to 65 mm. in length, all that 54 SALMON AND TROUT TN ALASKA. could be identified being sand launces; 2 contained young sticklebacks, one of them 10 individuals; 2 had isopods, and only i> had taken insects from the surface. Another example taken later, a male of 265 mm., contained 4 small herring. At Karluk young cohos are occasionally taken in the cannery seines; two, 180 mm. long, preserved from the catch of June 8, con- tained 2 species of amphipods and one a young cottoid; one, 158 mm., preserved from the seine July 3, was an empty female; July 24, another, 175 mm. long, contained Ammodytes. As will be seen, these records indicate the presence of very few young cohos about Karluk Beach. The genera! collections of the Albatross afford the following data: A number of cohos were taken at Karta Bay with larger sockeyes and smaller dog salmon on June 26, 1897. Of the specimens preserved 8 males average about 80 mm. (56-100), and 14 females average nearly 100 mm. (80-140). They were feeding mainly on insects and Crustacea. At Thorne Bay, July 5, of a number of small cohos together with a few dog salmon, seined probably at the mouth of the river, 24 males averaged about 55 mm. and 50 females about 56 mm., the high average of the latter being due to the presence of a few slightly larger individuals (extremes, males 45-65 mm., females 45-78 mm.). The stomachs examined contained insects for the most part; a few had small crustaceans and 2 had flatfishes. At Port Alexander, July 3, 1903, many young cohos were taken in the seine; 4 males and 2 females were preserved; average about 150 mm. They were feeding on young herring and sand launces, also larval crabs and amphipods. Of the specimens saved from Uganuk, June 15, 1903, 5 are males, averaging 138 mm., and 8 females, 130 mm. All but 3, which were empty, were feeding on young herring, each containing from 1 to 5 individuals. (See p. 51-52.) At Unalaska 6 examples, taken July 23, 1888, average 148 mm., contained insects, crustaceans, grubs, and in one case a small fish like a salmon fry. One humpback fingerling was in this lot. Twelve examples, taken at Sumner Harbor July 2, 1896, averaged about 60 mm. and were feeding on insects and Crustacea. They were in company with the smaller sockeyes. (See p. 50.) Isolated examples in the collection, not worthy of fuller notice, are: 1, Kilisut Harbor, July 1, 1903; 5, Tribune Bay, May 17, 1894; 3, Union Bay, British Columbia, June 22, 1903; 3, Cleveland Passage, July 13, 1903; 1, Pavlof Harbor, July 25, 1903; 1, Humboldt Harbor, Shumagin Islands, July 31, 1888; 3, New Morzhovoi, July 17, 1894, and 2, Kiska Island, June 7, 1894. From the above data it is seen that the cohos remain for some months about the shores near the streams whence they issue. They SALMON AND TROUT IN ALASKA. 55 may be found about the mouths of the streams in brackish water perhaps soon after their descent of the stream. It may be they remain about the streams for a time to accustom themselves to the salt water, but this is not evident in case of the fry. The sea-run exanxples are readily distinguished by the silvery appearance and usually by the greater depth of body which follows habitual disten- sion of the stomach. In some cases, v/hile near in shore, insects appear to continue a staple article of diet, as in the fresh water. The cohos feed less on Crustacea than the sockeyes, perhaps inhabit- ing slighter depths; correlated to this is the abundance of small fishes found in their stomachs — sticklebacks in brackish water and herring and sand launces in more open regions. From the catches at Naha and Yes bays it would seem that the cohos continue to school after reaching salt water. The results of the seine hauls indicate that the difl'erent species of salmon school together, or at least in the same waters. DOG AND HUMPBACK SALMON. On May 29, 1903, a small school of salmon, about 50 individuals, was seen along the shore about 1 mile below Naha Bay. Seven of these were secured — 5 humpbacks 42 mm., and 2 dog slightly larger. The former were feeding on small Crustacea and pteropods, the lat- ter on Crustacea and insect larvae. On July 1 and 2 a few dog salmon fingerlings were taken in the net at station 2. These may have been either fishes that had remained in Roosevelt Lagoon since early spring, or stragglers from the bay coming in on high tide. Ten were males, averaging 56 mm., and 4 females, averaging 62. Those with food had been taking insects. (See also p. 47.) At Karluk June 18 the fry of both dog and humpback salmon were abundant along the beach. Six preserved examples of the dog aver- age about 50 mm., and contained insects, crustaceans, and small cot- toid fry. Thirteen humpbacks saved average about 47 mm. ; of the 6 larger ones examined 4 contained crustaceans, the other 2 nothing. July 24 there were saved from the cannery seines 5 dog-salmon young, averaging 83 mm. (70-100), which contained a few insects and small fishes (blennies), but mainly crustaceans. Also 47 hump- back young, averaging 77 mm. (60-95), the main food in which was copepods, but in addition were found flies, insect larvae, amphi- pods, and in a few cases small fry, probably Lumyenus, up to 18 mm. in length; 1 stomach contained 22 of these. Only the following collections were made by the Albatross: In Alert Bay, British Columbia, June 5, 1895, about 50 dog salmon small fingerlings were taken, along with a number of small sockeyes; these averaged about 40 mm. and contained only small Crustacea. 56 SALMON AND TROUT IN ALASKA, In Admiralty Inlet, Wliidb}" Island, on June 30, 1903, many young dog salmon were taken in a shore seine on gravel bottom. Of these 13 of each sex were preserved, the males averaging 99 mm. (83-117), females, 98 mm. (78-122). In none of the 9 examined were any fry found. They were feeding wholly on plankton material, Crustacea, principally amphipods, and Sagitta. They appear to have been schooling alone. In Otter Bay, Pender Island, British Columbia, 1 1 dog salmon were preserved from the collection of JMay 31, 1895. These averaged a little over 70 mm. and contained Crustacea. June 29, 1897, 4 dog salmon, 2 of each sex, averaging about 56 mm., were taken at Loring, it is believed in a seine haul on the seining beach of Roosevelt Lagoon. They contained only insects. These specimens, considered in connection with those obtained at station 2, noted above, point toward a continuance of some of the young of the dog salmon in brackish water for a period and suggest the desira- bility of investigating such waters with suitable gear. A haul was made in Roosevelt Lagoon on the night of October 1, l!)()r), but no salmon were obtained. At Thorne Bay, July 5, 1897, many dog-salmon young were taken in seine hauls on the beach just below the river; 64 of the specimens are males, averaging about 65 mm. (42-82), and 84 are females about 2 mm. longer (44-83). The food in those examined was Crustacea, mainly ostracods. A few smaller ones taken in the river mouth were feeding on insects. In Karta Bay, June 26, 1897, dog, coho, and sockeye young were taken in the same hauls. Of the first, 19 males, about 60 mm. average, and 10 females, about 53 mm., were preserved. They were feeding mainly on insects and amphipods; a few contained small flatfish, 1 being filled with them. In Dundas Bay, July 24, 1903, many young salmon — sockeye, dog, and humpback — were seined on the tide flats. The specimens of dog salmon are 4 males, about 75 mm., and 5 females, 82 mm. long. They were feeding on Crustacea; a few had eaten insects and larvae in addition. The 3 humpback, 2 males and 1 female, average about 80 mm.; food the same as the other species. The 3 species were apparently schooling and feeding together. In Pavlof Harbor, July 25, 1903, the seine was hauled at the mouth of a small stream in deep water, gravel beach. Many young dog salmon, a few cohos, sand launces, and other small fish were taken. The specimens of the first preserved are 4 males, about 100 nnn. long, and 7 females, slightly smaller. Their food was Crustacea, except, in one instance, a few flies. At New Morzhovoi, July 17, 1894, 34 dog salmon young were pre- served. These average about 80 mm. in length. The stomachs con- tain Crustacea and gastropods; only 1 had insects. These specimens show the peculiar segregation of food noted in some sockej^es at SALMON AND TROUT TN ALASKA. 57 Karluk (p. 52) ; some had eaten almost exclusively large copepods, others ostracods, and yet others a peculiar Caprella-like form. Four specimens from Isanotski Straits, July 1.5, 1894, show no pecul- iar features. At Sucia Island, May 6, 1894, 3 humpbacks and 1 dog salmon were taken, the former 47 mm. and the latter 54; food, Crustacea. At Metlakatla, July 10, 1903, 2 humpbacks were taken, 64 mm. in length. One was empty, the other contained flies. Single specimens of humpback fingerling are recorded from Kodiak, August 14, 1888, and Unalaska, July 23 of the same year. The first, a female 113 mm. long, contained sinall fry and a few flies; the last, a male slightly larger, crustaceans. CONCLUSIONS FROM AVAILABLE DATA. From the above notes it may l)e concluded that many young sock- eye*, after reaching saltwater as yearlings early in the spring, remain, in company with other young salmon, for a few months about the shores near the mouths of the streams from which they are derived ; that during this time they feed principally upon the small crusta- ceans which are found from the surface to an unknown depth and, like the crustacean forms found in the lakes, have a diurnal move- ment from and to the surface. In tows made by the Albatross in the open ocean it has been found that many of these forms tend to be most abundant at the surface about dark, again decreasing in number within an hour or so. Whether the period repeats again at daylight was not tested. Surface tows in daylight made in Yes Bay and Behni Canal during July, August, and September, 1905, showed an almost entire absence of food material. At the same time it was found at depths of 20 to 75 fathoms and greater, the deeper the more abundant. It is known that this pelagic life is ultimately dependent upon the shore for food supply. The open sea far distant from land contains little life, unless it be conveyed by currents originating near land. The narrow and deep channels of Southeast Alaska furnish a superior environment for plankton life ; to the wholl}^ free-swimming forms, such as copepods, ostracods, etc., there is added the innumerable progeny of the littoral forms, such as crabs, worms, mollusks, and shore-spawning fishes. ABUNDANCE OF FOOD. As a test of the abundance of this salmon food material, during the summer of 1905 numerous hauls were made throughout Yes Bay and into Behm Canal. Yes Bay is a long, narrow inlet less than one-half mile in width and over 4 miles in length, the depth varying from 50 feet at the head to over 50 fathoms at the mouth. It receives sev- eral small creeks at the head and the main stream from Yes Lake, about midway between the two extremities. During the summer 58 SALMON AND TBOUT IN ALASKA. months the surface temperatures vary between 55° and 60°, the lower nearer the mouth. The surface densities depend almost wholly upon the precipitation. During ordinary weather the head of .the bay, from the pushing up of the dense bottom layers of water by the tides, acquires about half the density of sea water; midway, on account of the volume from the main stream from Yes Lake, the density is reduced to about one-fourth that of sea w^ater, increasing to one-third or more as the mouth of the bay is approached. During freshets the surface density is much reduced, the water becoming practically fresh at times. The bottom temperatures vary from about 50° at the head in 50 feet of water to 44° in 330 feet at the mouth, with corresponding densities of about 1.018 — or nearly three-fourths the density of sea water — to 1.0235, or closel}^ approximating standard density. The tows exhibit a very scant population of food forms in the upper bay, with a gradual increase as the mouth is approached and a sudden rise upon entrance to the main channel outside. In the fresher waters jellyiishes predominate; with increasing density and depth the crus- taceans increase in number. This whole subject of the distribution of pelagic forms is of funda- mental importance in the study of the sea habits of the salmon. With the pelagic forms, of course, must be considered also the shore forms, such, for instance, as the common amphipods (beach lice), which may be found under stones at low water; perhaps, also, certain mollusks, etc. The factors of temperature, density, light, depth of bottom, proximity of shore, set and velocity of currents, influence of land drainage, etc., require to be carefully studied. As yet nothing has been attempted in Alaskan waters toward the solution of these problems. It will be noted that the Karluk sockeyes taken in salt water in June are of greater average size than those taken on subsequent dates at the same place. Further, that the average size of the June spec- imens is about twice that of the migrants from the Nalia taken only a few weeks earlier at Loring. If these Karluk sockeyes are accounted yearlings of average size for the locality, which have left the lake not earlier than April, more likely in Ma}^, they either have increased very rapidly in size since reaching salt water or were much larger than Naha yearlings at the time of migration. But for the uniformity of size it would seem more probable that the specimens procured are the selected larger examples, the smaller escaping through the meshes of the seine. There is the possibility that these larger individuals were fry that reached salt water the previous sum- mer, and that the absence of smaller specimens in the collections is due merely to the escape of the fry and fingerling migrants of the same season through the large meshes of the seme. That year- lings from Karluk Lake are larger than those of the same age from the small lakes of Southeast Alaska may well be believed in view of SALMON AND TKOUT IN ALASKA. 59 the larger size of examples noted in Wallowa and Alturas lakes and the Eraser River. The dog-salmon young, cited above, vary from a little over 40 mm. to 120 mm.; the humpback show approximately the same figures. These, are unquestionably of the fry that reached salt water in March and April just previous. The catches date about three months later. This indicates an increase of about 100 per cent in that length of time, or about twice the rate of growth of the sockeyes that remain in fresh water. In other words, the species (or individuals) that reach salt water as fry make about the same growth in the first three months that the lake residents (sockeyes) make in twelve. The com- parison of fall catches of cohos and sockeyes in the lakes shows the former to exceed the latter greatly in size, but this is only a natural result of the colio's greater size from the fry stage to the adult. In the case of the humpback, however, the fry is scarcely larger than that of the sockeye, and the adult somewhat smaller. Its more rapid growth can be attributed only ^ to the abundance of food in salt water and the nature of the fish. The sockeye young which were taken in salt water must be yearlings. Owing to the lateness of the migration of that species it is impossible for them to have attained such size in the time since they could have reached salt water as fry of that same season. It will be seen that in general the four species of salmon are closely associated in sea habit. Wliile small they remain much of the time close inshore, to some extent feeding on the insects that fall on the water or are carried down by the streams. The greater part of the food, however, is the various crustacean forms, though the small fry of marine species make up no small part, particularly of the coho's food. With increasing age the salmon doubtless move into deeper water. If the habit of preying on small fishes continues and grows with the increase of size, it would seem that the young salmon would be remarked in pursuit of the schools of young herring, sand launces, and such forms that abound in the bays and coast waters. No note of this has been recorded. Occasional catches of slightly larger fish are made from time to time, possibly fish of the second year. Such cohos are well known in Puget Sound, and a run of some small sal- mon was observed in Naha Bay during the winter of 1896-97. But unless such fish in large numbers api)roached the surface of some small body of water near a settlement, they would scarcely attract attention though the channels were teeming with them. The salt-water habits of young trout and charrs are entirely unknown. RETURN OF ADULTS TO FRESH WATER. From the yearling stage to the adult little is known of any of the salmons, and nothing of the sockeye. There are reports of grown fish of that species taken in the winter in various places, as in the 60 SALMON AND TROUT IN ALASKA. vicinity of Karliik and near Union Bay, Cleveland Peninsula; but the identification of these catches has not been authenticated. The only pertinent fact regarding their place of. residence after leaving the vicinity of the parent streams, if they leave it, is that the pres- ence of adult fish is first noted at the time when they are apparently rounding some point in their progress from more open water. Such places are Point Iliggins, at the entrance to Behm Canal ; Cape Cha- con, at the lower end of Clarence Straits, and Otter and Sheringham points, on Juan de Fuca Straits. Karluk River is apparently approached directly from the open straits. APPROACH OF SCHOOLS. The presence ot salmon can bo noted only by their habit of leaping from the water as they approach the land. It is often possible by this means not only to recognize the presence of a school, but also to distinguish the species. In jumping, salmon do not leave the water with their ventral surface downward, as-do flying-fishes. They always jump sidewise with one side at an acute angle to the water surface. Sockeyes seldom entirely clear the water, but let the tail drag for some distance, fall on the side, and then perhaps skim the water for a short space. They may make two or three successive jumps, apparently at random and in varying directions. Cohos usually leave the water entirely, falling back on the caudal peduncle held rigid with the fin directed upward. The tail may then drag through the water a short distance till the fish falls on its side and disappears. The humpbacks jump very agilely and characteristically. They leap clear of the water, shaking the tails vigorously while in the air, sometimes turn- ing completely with a corkscrew motion. On falling they strike on the side. The cause of this jumping has been much discussed. It was noted that a school of kings feeding in Naha Bay in December, 1903, jumped in much the same manner as fish on the way to the spawning grounds, but apparently not so frequently, probably because of the small num- ber of individuals under observation. The sockeye travels in large and compact schools when approaching the spawning regions, and the jumping should be easily accounted for in the natural playfulness of groups of animals or the struggle for preferred position. Stomachs of netted fish often contain salmon scales and teeth, the latter frequently the individual's own. But these are doubtless swallowed during the struggles in the net and not by reason of previous combats. Under certain conditions the schools exhibit little tendency to jump. In 1903 large numbers of sockeyes entered Karluk River without such announcement. Of course it is not known whether this species feeds in schools in salt water, or whether it is only the approach of sex maturity that impels the gathering into schools. From the SALMON AND TROUT IN ALASKA. 61 arrangement of tlie gillrakers, and the few observations made on feed- ing fish, it may be surmised that the larger part of their nutriment is derived from pelagic Crustacea and associated forms. They are known to feed at times on small fishes. In striking into a school of small fish from below they might be led to broach the siu'face. In towing for plankton food there would be no such occasion. No authentic observation of sockeyes in open ocean has been reported, but inasmuch as no search has been made for them the matter remains without evidence. It is probable that they do not jump in feeding, and tlius they are unobserved up to the time they start for the spawning beds. FOOD AND FEEDING. THE SOCKEYE. It is reported that the Fraser River schot)ls, even before leaving Juan de Fuca Strait, have ceased to feed. In Southeast Alaska it was found that a certain percentage of the earlier catches, even at the mouths of the rivers they were presumed to bo entering, were still feeding. This is shown in the following table: — Food of Adult Sockeyes. Total num- ber ex- am- ined. 120 3* 512 400 <510 692 .'■,00 /.'•)12 100 .511 100 a. 514 ^•200 511 Number having eaten fish. Number having eaten Crustacea. Number having eaten worms. Total with food. Per cent of to- tal with food. Locality and date. Male. 4 1 3 Fe- male. 2 ""'3' Total. Male. Fe- male. Total. Male. Fe- male. Total. Karta Bay: July 1-4, 1903 July 23-26, 1003... July 1 (J, 1904 6 a 6 5 1 6 1 1 2 1 * 6 1 11 Yes Bay: July 17-20, 1903 . . . .\ug. 16-21,1904.. July26toAug. 14, 190,5 1. 3 112 "3 85 .S3 6 11 1 6 197 18:5 13 23 74 1 1 ■' 1 16 1 6 1 32 1 7 i 13 d 197 ! '>s d Boca de Quadra: July 30 to Aug. .5, 1903 100 1-' 32 e 184 36 8 Aug. 24-29, 1904.. Kogan, Moira Sonn; ... .It Jjl . J^ - • •' , 1 " ?• 1 ' •* /K* -j- - - , _ _ _ -- ^ T,»ir Ji g J T ..T " "'" i: *• " ..^f ~n~ " T it ^i >^ - ??L ; ^'i' r -t-^r it ._ ■ J. • ^2 _ _ i -"di"- " " "1 ' ix -i,,* />V -.- _u«- ' ' / ° 1 ' , ' • *•// 1 ? * " ' i. ' • ' / ^ ' ' 1 ■*"* 1 • ' / *gi ' 1* It "'i -I"-; ^, - /-i^l-it:,!'" 1 ,* r^' I i Ij."' ''i/ t = ' "' ' ' j^ /I Jr '^ / 1 I*T 1 1 ' ' \ 1 \ ^r'' ' 1 \' 1 T i '• ' i ' r Li !" \ 'o.*'V-* ' ■ ; » / .1 '■ .. '■ .. .- ,-4' 9<" l-jl|..'.4_, 4.4 <^ 1 i ** • / ' B ] ' " * / */ t_ 4 , - " • / -. . : 1 >■ J 1 o'l 1 : • /\ sr ^ c 1 r. ' liiv 1 i %' ' ' ii .,4; ^ .. ""iiirir , !."Tti^iirr' ni ^ : " • . i "i \! ^ » ' - ^ -■ -'■ ■■ ■■ ?. ' . ' \ f •/ 1 >i '5 7M /!, 1 ' " , I '•i 1 -^^ -,- -1- M-/,Ji^H I it . _ _ iji_ _i _ r>'^ ^i;:^ -- '"1 ol" 1 ^Wl °«i. 1 'k . - ' ' ' "' ^r ^ " i V ^^-r-. +^ ^ ^ r^ ,,^ ., ^' ^^-•-" -jj-^- ^ !!!i 9* ' ■- " ^■■" * - ]" 11 -\--v- '^",■•- / ■■' . . \ \ 1"' - — - -■ ^^■--.■' ' "V \ " »' '- t\ '»« ^ \- "' ----- 41 r- ^ ""••.. ""'tc It "'" !i^^;: P- I1 - .- . ..-I X . I=ir^B.lt-- -. - sv'^ [.J „ II ■ 4^ -- X "■ -^ - ^J^ 1 , I "-^ ^ 1 ''o ' - . -- _ ,.-.^ ^ — .. -„ ^.^ X i: -- ^ ai'^>^ - X - - it ^^-^ t IE -.. . -. _^ B ■> I " " 1 _4X- - - i" ~ ■ " -■-— ■■^- s . -1--+-] t- ^ ^s e ! ^ t ilti ' X X ^ a si » _4 ^ X ._ - j_ ' -Xiti}: X u T - " - '. ~ " ^ ^T XI _ _ ^ / 1 i ' 1 1. . 1 1 1 ' ■ o o o - - g,0 ;- _^^ljih^1: ► "'•IH,- ^- X ^i^ 4_ " "'Uh^ -.- - 1^- . ^< X "I __ X -- --. '•j'li ' V^o"^ ' " * • *^ o ' 1 •. ° T X *'°^ ^x X _ "O., t i' _ ± Gl °°i- **J : i,. ?qj> T l« H'l^' -■ X > 'H'L -■>, ! <»- "(>•• T STl o° •, ^., o° V X o° !. IT 1^)' ■■ ' • '' ^- ±2t 4X"- *-• ' ^- ^ X X X *. X : ••,; ",. ■ » it X^ _ ^> 1 x V ^ V rf >< m m ' ' 1 > ' - a m ^ ° a J "-- "I '1 I^ _4_ <=i t -^i^ I - ^<^o ■ -- 1 --e. "i^ rt _^ wvt^ •>!:- -W it X ^x z -h ^ oH^ -« -^ ' O fc . . X- m 1^ ^ --^ _^ -. ^= V- it ts i fit- _1 ? 1 • X J*- 1 V -j n>c, V n X ~ ■> 1 -t. It T 't - x - -4- T + 1 1 1 , _i " " 14" ' " ' ' "Ji " " " "JJ W w SALMON AND TKOUT IN ALASKA. 85 Table 5 shows the depths of the males tabulated in the same manner as the lengths and weights, showing these measurements in percentages of the total number of fish in each lot measured. Dia- gram E presents the same information for 1904 shown by curves. Table 5. — Depths op Male Sockeyes, Shown as Percentages op the Total Number op Examples Examined for Each Locality. Bepth. Tam- gas. Dolomi. Port Ches- ter. Quadra. Nowis- kay. Kegan. Karta Bay. Yes Bay. 1904 1904 1903 1904 1904 1903 1904 1903 1904 1903 1904 1903 1904 1903 1905 0.0 .6 1.0 1.2 2.0 7.7 1G.-4 24.0 19.2 13.2 7.6 3.4 2.0 .6 "^ j . 90 mm 0.4 .4 .6 .8 6.1 19.0 26.3 22.7 12.5 7.0 2.8 1.4 .2 •2 1 , 1 100 mm 0.8 4.0 4.0 7.4 22.0 18.0 13.0 14.0 0.5 4.8 4.0 .8 2.6 1.3 7.3 10.6 15.3 11.3 1,5.3 8.0 8.0 8.0 8.0 .2 1 1 0.2 .2 .6 .4 5.2 13.4 19.6 25.2 16.4 9.2 5.0 4.0 1.0 110 mm 0.4 2.9 10.6 19.9 17.0 20.1 13.0 8.7 0.0 2.3 0.4 .8 2.3 6.5 11.2 14.9 16.0 1.5.8 17.0 8.3 6.0 1.0 1.8 16.' 7' 5.3 8.9 36.0 18.0 8.9 8.9 I "3.'2" 5.4 8.6 10.8 14.0 16. 1 24.9 13.0 3.2 0.4 2.5 5.5 10.7 13.5 12. 3 15.1 20.6 11.9 5.3 2.5 "i."i' 1.3 4.0 4.8 11.7 14.0 28.3 20.0 10.6 3.1 .6. 0.9 1.4 1.9 7.7 8.2 11.0 24.0 17.3 15.4 3^3 1.9 120 mm 125 miii 1.4 2.4 5.8 15.4 13.8 16.4 11.6 14.6 9.6 6.8 2.6 'i."2' 1.8 6.0 12.8 20.0 25.6 17.2 10.0 4.9 .6 .6 0.3 130 mm •6 1.8 2.3 4.7 150 mm 7.4 . 2 14.5 160 mm . 20.9 21.8 ] 13 .2 7 4 180 mm 4.1 185 mm .6 I12" 140- 518 147- ^503~ 151- No. of specimens Average depth, mm. 528 114+ 505 121 122 125- 150 126+ 503 138+ 265 145- 93 140- 509 141 + 56 134 208 144+ 164 159 339 158 A close correspondence will be seen in all the curves for Dolomi and Tamgas." The greatest number of individuals in each are 4-pound fish. The balance of the average weights of the two lots is destroyed by the varying numbers under and over this maximum. The weight curve, however, for Dolomi is very similar to that for Tamgas, though it occupies a place about 0.5 pound higher in the scale, the average weight for Dolomi being 4.5 pounds as compared with 3.9 pounds. An examination of the length curves (diagram e) also shows a very close parallel between these two localities, w:hile the difference between the curves for depths (diagram f) will explain the greater weight of the Dolomi examples, the depth curve for Dolomi standing about 5 mm. higher on the scale than that for Tamgas. Comparing the two lots, it is shown that while the average length is about the same for each locality, and the number of examples of any given length is about the same for each, the Dolomi fish average about 5 mm. greater in depth and therewith 0.5 pound greater in weight. These two localities seem to be set off clearly from all others examined. a In the following discussion of the tabular data reference is to the males only unless otherwise specified. 86 SALMON AND TROUT TN ALASKA, The Dolonii fish of 1903, only 200 in number, were much heavier than those of 1904, but in this they but follow the rule jexhibi ted by other localities, in almost all of which the 1904 fish are lighter than those of the previous year. It will be noted that in 1904 the examina- tion of Dolomi fish preceded that of Tamgas fish by about two weeks. The latter were reported to be quite immature sexually, apparently more so than the Dolomi fish of the earlier date. This may account in a measure for their slenderness. In both these localities running small fish, two type forms are noted more distinctly than in streams carrying the larger fish, viz, the slender terete form, usually dark, with scales more or less embedded, but not showing the other marks of sex maturity; and the deeper, more compressed form, with scales dis- tinct and brightly colored. Fassett noted at Tamgas that one or the other of these forms predominated on different days, indicating a difference in schools. The dark fish may be fish which have been feeding in the brackish water for some time. The fish of Karta and Yes bays, the heaviest of all, in 1904 approached one another closely in average weight, varying only 0.3 pound. Exami- nation of the curves for weight, however, shows great difl'erence. The curve for Karta Bay is comparatively simple, rising to over 24 per cent on 7 pounds, extremes falling between 3.5 and 9.5 pounds and showing little tendency to indicate a double node or maximum, while Yes Bay shows 2 maxima, one on 6 pounds and one on 8 pounds, with small numbers on 7 and 7.5 ; in addition the upper extreme rises to 10.5 pounds. The length curves exhibit this variance increased. In this the line for Karta, showing some tendency to form a double node, rises on the same maxima with Yes Bay, 490 mm. for the lower, and 550 mm. for the upper, but in the case of Yes Bay the percentages on each are about equal (12 per cent), while for Karta they are, respec- tively, 3 + and 21 .5 per cent. This shows that the relatively short fish are found in much greater number among the Yes Bay examples, reducing the total average length (std.) in this locality to 525 mm. as against 538 at Karta, in the face of a slightly higher average weight for Yes Bay. Reference to the depth curve again offers the explana- tion. The two length types of Yes Bay cause the depth curve for that locality to be somewhat truncated, rising to about equal height (14 to 16 per cent) on 140, 150, 160 mm. and reaching an extreme of 180 mm. This indicates the prevalence of short deep fish in Yes Bay and long slender fish in Karta Bay. In this year the Karta fish were examined about six weeks earlier than those of Yes Bay. They were noted as nearly all bright, but some were of advanced maturity, with the usual changed form. The Yes Bay fish were noted as being of "advanced" maturity in out- ward ap])earance, and it was said they were so from the beginning of the run. The catches made in Behm Canal averaged smaller fish and 1 1 • * J I ' 1 * ^-i^t - -1 -t* A^ + -p" , ,-w jf- 9- ' 4. /J ^ •'"■'._.-,? V J t - ^Li • ^^t. .-Silt 41^^ J Xlt^- ^V :,*■ \-^%' J •E ^'' il'' " jo"' H 1 ! '^^^ - J"^ ^ _l_ _- -1- - ^ ♦- x'^41'*'^ -x -5 _ 3 _ _ - "'^,, . . y. 1. e « r '^•j, ^di Xll -. • ^ E - •f^^-S.At^ltP ••.,^ *r 1 ; T i;^'^.- li z" . _L y _x, , . , --,-5- - 'V -H H-, . ,0* ll _i ^^ ^ i> 4l± .._ *'«^/tj-, I ^ . ^X- IXT'^.-^.^I .^' r- ° ^\ 1 , T^' . • : . r !i: ip.4/ ^j^t,^ '2I-lttll hI ^'*S-^ "^[Jl .-pc „at" '^ . Jl Jj» iXv^'*L. 4 ° \'' ^ - *r < " " ■ ■" \ *i*U iSl^ L " • ' ■ ^ " • '^' ■ ■ 1 ^ " .' ^\ ». lo >■ ^(Ss. \ • il' I!I„_ _7[- ir>?p ; ^ ^^3 -- \ o '~f- ' ^ % - + ^»i-: - It -,«- ^ -t-> '^ --^ --, 5-« ^ .' ■ - ^olt l^ ^^ ^ - \X ^ > '« X 1 J jS V . ^ ^ ' ir -^, "" ;;^ ^^ z\jr - -i • ^ 2^ 4 o -it ^- t' t ->- °J^lt o : >-. ._ ^#,..' ^-- ►« sit _r:!: It s= -« t ; _ _ - _ __ _ % I 'i ~ iL cj It. _.i}: " " "" II : i}i 31 IT ^ i .-._t;^ ij : it _ _ __ "I u ' . -^ :j~ ■ ^ 1 " ' ■ ^ ' , N r 1 it it itt s iL it It " - -(--t-- -^ ]r 5 J : _. ..i __iji E •s j_ 1 . „ _, , T- _ _ [ lo 2} ><: s < O < Q o "J (D tr ■z. o — u. O X C/5 SALMON AND TTtOTTt TN ALASKA. 87 brit!:hter. The fishermen thought these were ''Umik River fish." Wliether regarded as fish traveUng to another basin or as a hiter run not yet ready to enter the bay, these smaller fish account for the abundance of 6-pound fish and the double node in the curve. In 1903, the fish of these two localities, examined at approximately the same time, difi"ered much more in averages, the Yes Bay fish exceed- ing those of Karta Bay by more than 1 pound in weight, by 12 mm. in length, and about 15 mm. in depth. In neither is there an indication of more than one type. The curves for length, weight, and depth show a single maxinuim, that for length in both localities on 560 mm., for weight on 7.5 and 8 pounds, respectively, for Karta and Yes bays, and for depth on 145 and 160 mm. Over 25 per cent of the Yes Bay fish reach the maximum, while of the Karta fish less than 20 per cent are of maximum weight and length and 24 per cent are of maximum depth. This indicates, again, that increased weight is largely due to increased depth. In condition the Karta fish of this year were noted as being quite as- much, if not more, advanced than the Yes Bay. This applies rather to the last lot of 153, examined July 23 to 26, in which tlie shorter fish predominated, only 45 per cent being over 550 mm. in length, while of the first lot of 55, examined Jul}^ 1, 62 per cent were of that length. In the two lots from Yes Bay the curve for that of 1903, taken earliest — about a month earlier than the 1904 lot — is much more like the curves for Karta than like that of Yes Bay for 1904, rising to over 25 per cent on 560 mm., with extremes between 480 and 610. In 1905 the values are more distributed, varying between 10 and 15 per cent from 520 to 570 mm., with extremes between 420 and 590. The average lengths, respectively, of the 1903 and 1905 examples are 555 and 532 mm., the average depths approximating 158 mm. in both cases, while the weights are 8.3 and 8.2 pounds. This indicates that the fish of 1905 were more of the deep form, while those of 1903 were rather long and comparatively slender. These 1903 fish were from the early part of the run, those of 1905 some weeks later, giving another example of the long slen- der type form in the early run. On the' whole, it seems that Yes Bay fish may be said to be deeper than the Karta fish of corresponding maturity, and that the early fish of both localities are of less depth and weight for corresponding length than later fish. Quadra and the Moira Sound streams, Kegan and Nowiskay, show an almost exactly equal average weight in the fish examined in 1904, as well as close similarity in the averages of length and depth. Inspec- tion of the curves, however, shows this similarity to be factitious. The Quadra fish are a symmetrical group, while the Moira Sound fish are, like Yes Bay, composite, apparently made of a short deep type, and another correspondingly longer and more slender. 88 SALMON AND TROUT IN ALASKA,' In Quadra fish taken in 1903, about one month earher than the 1904 lot, the length curve shows much variation (see.iJiagram g). The average length is nearly 20 mm. greater than in 1904, and the curve zigzags between 500 and 570 mm. across the 10 per cent line, while in 1904 it shows a single rise to about 17 per cent on 500 and 510 mm. The difference in average length, weight, and depth of Quadra fish in the two seasons of 1903 and 1904 is very close to that of Yes Bay fish for the same years. The curves, however, are quite unlike, that of Quadra for 1903 being compound, while that for Yes Bay is simple. It was noted that the Quadra fish of 1903, though examined later than those of Yes and Karta bays, were less developed sexually. In almost all cases the curves for the females, though quite similar in general form, exhibit less complexity than the curves for the males. (See tables 4 and 4a, p. 84.) In no case is the curve of such a character as clearly to indicate fish of difl^erent ages, though it is possible that such curves as that for Yes Bay in 1904 may be so accounted for. The maxima in this case, separated by 60 mm. (80 total length) and 2 pounds, would seem to be about what might be expected for the difference between a 3-year and a 4-year fish. In the females the maxima of length are separated by the same interval and the much lower percentage of the lower maximum accords with what is known of the development of the king salmon in captivity, in which the males exhibit much the greater precocity. Of the occurrence of the grilse form, or Arctic salmon, we know little, for the reason that many doubtless pass the meshes of the nets, and many are unreported by the fishermen, since they are not taken as "counts." At Quadra, of the sockeyes delivered at the scow about 1 in 300 were grilse. This is, perhaps, not far from the average ratio in the Loring district. They were not noted in unusual number at Dolomi. The Moira Sound fish examined in 1903 are too few to be of any value in the present comparisons. The increase of relative depth in the later fish may indicate only the change incident to sex maturation. It carries with it, however, a corresponding increase in weight. If it is not a typical form, but merely a sex character, then the early run ascends the stream with a smaller amount of tissue deposit than later runs, else they do not first enter the bay for the purpose of ascending the stream, but return again, if not hindered, to the feeding grounds. Such observations as could be made in opening the fish for deter- mination of sex did not indicate that the difference in sex maturity of the different parts (seasons) of the run are material until after the main body has arrived. The late arrivals — the last week or so of the regular run and belated sections — do show marked advance of sex maturity. It is probable that careful measurements would show a gradual increase in relative maturity, but not proportionate SALMON AND TROUT IN ALASKA. 89 to the difference in time at which the (hfferent schools arrive at the breeding stream. Variations in counts. — For a further test of the similarity of fishes from a given basin, counts were made of the dorsal and anal fui rays, the branchiostegal rays, and of the tubes of the lateral line and number of longitudinal rows of scales. These various counts are tabu- lated in the same manner as the measurements, except that sexes are not differentiated, percentages being used instead of the number of examples in order to make comparable the results for the different sized k)ts that were obtained. In the fin-ray counts the totals of rudimentary and branched rays are used, but the terminal half ray, which varies greatly in development, is in all cases omitted. In table 6, showing the counts for dorsal rays, the similarity of the two years for each of the various localities is striking; Yes Bay, Karta, and Quadra exhibit the maximum on 15 rays, while Dolomi and the two Moira Sound points give it to the next lower number, or 14 rays; 12 and 17 are the extremes. The averages differ very slightly; Karife Bay is highest with 14.7 and Nowiskay lowest with 14.3. It is noteworthy that these two localities were examined in sequence in 1904, and the field notes call attention to the exact identity of method in recording the fin counts. Table 6. — Dorsal Fin-Ray Count. Locality. Date. Number exam- ined. Average number of rays. Percentage of fish with specified numbers .of rays— 12 13 14 15 16 17 18 1904 1903 1904 1903 1904 1903 1904 1903 1904 1903 1904 1903 512 420 509 300 512 500 511 100 511 200 513 100 14.7 14.8 14.5 14.6 14.5 14.5 14.4 14.4 14.4 14.4 14.3 14.3 "6.'2" 0.6 .5 1.7 1.6 2.3 2.6 2.5 2.0 2.5 3.0 5.4 7.0 31.6 29.0 41.5 36.3 44.0 42.4 54.2 57.0 53.7 53.5 58.2 61.0 60.9 63.1 53.8 61.0 51.7 51.2 41.3 40.0 41.2 41.0 34.1 30.0 6.8 7.1 2.5 .6 2.1 3.8 1.9 1.0 1.9 2.5 1.9 2.0 Yes Bay 6.2 .4 .3 2 Quadra 6.2 Kegan .0 In the anals (table 7) the parallel is less sharply defined but still characteristic. All reach the maximum on 18 except Dolomi and Nowiskay in 1904, which rise highest on 17. The extremes are 15 and 20. Quadra shows the lowest percentage on 17 and highest on 19. The averages vary from 18.2 at Quadra to 17.3 at Nowiskay. The order of localities as to number of rays is changed from that of the dorsals, though the Moira Sound streams are still lowest. 90 SALMON AND TROTTT TN ALASKA. Table 7. — Anal Fin-Ray Count. Locality. Date. Number exam- ined. Average number of rays. Percentages of fish with specified numbers of rays — 15 16 17 18 19 20 1904 1903 1904 1903 1904 1903 1904 1903 1904 1903 1904 1903 512 499 508 300 512 420 510 100 511 200 513 100 18.2 18.2 18.0 18.0 17.8 17.8 17.7 17.8 17.3 17.4 17.3 17.4 0.4 10.9 13.0 16.1 14.0 25.9 27.1 29.4 32.0 48.5 42.5 50.1 44.0 64.1 55.4 63.4 69.0 60.0 63.8 61.7 56.0 46.6 48.0 41.3 46.0 32.6 29.2 19.0 16.3 13.8 8.8 7.4 8.0 2.9 3.0 1.7 3.0 1.6 ? " 6.'2' '".'2' .6 .3 .2 .2 1.1 2.0 2.0 6.5 6.4 7.0 1.0 .3 .2 '".'2 The branchiostegal rays exhibit httle local variation in number. The maximum in practically all cases falls on 14/13, the left over- lapping membrane carrying the extra ray; about one-third the total occurs on this maximum. Extremes are 12/11 and 16/15. Quadra and Kegan resemble one another in rising high on 13/13 and failing low on 14/14 and the higher counts. Karta and Dolomi exhibit the highest counts on 14/14. In no instance was a clearly defined case of right overlapping seen, though occasionally the right membrane carries the higher number of rays. Table 8. — Branchiostegal Ray Count. Date. Num- ber exam- ined. Percentage of fish with specified brancliiostegal counts — Locality. 11/11 11/13 13/15 14/11 12/11 12/14 13/U 12/13 12/12 14/12 13/14 1904 1903 1904 1903 1904 1903 1904 1903 1904 1903 1904 1903 511 420 511 100 513 100 511 200 512 500 508 300 0.2 0.4 0.4 0.4 1.7 2.5 2.0 1.0 0.2 1.1 2.9 2.0 .2 1.0 1.0 2.0 2.1 3.4 .4 .3 1.0 3.1 3.5 2.0 .4 2.0 .8 .5 1.7 .8 2.7 1.3 3.1 0.2 1.9 0.2 0.2 "i.o" .2 .4 .8 4.0 Nowiskay .2 .4 2.1 1.0 3.0 .2 .5 .2 .2 .2 1.5 .8 2.2 .4 .6 1.1 .2 " '.4' .4 3.5 0.2 .2 2.0 3.0 1.7 .6 .3 2.3 Date. Percentage of fish with specified branchiostega counts— Locality. 13/12 i 13/13 14/13 14/14 15/14 15/13 15/15 16/15 14/15 16/14 15/12 15/16 1904 1903 1904 1903 1904 1903 1904 1903 1904 1903 1904 1903 5.5 9.0 15.0 17.0 4.7 4.0 8.8 7.5 11.1 12.0 4.7 5.3 16.8 21.7 22.5 26.0 15.2 22.0 18.0 15.0 27.3 21.8 13.3 15.0 32.5 35.0 35.2 29.0 34.3 34.0 30.7 27.0 33.6 30.0 37.6 32.3 26.2 15.0 12.1 11.0 21.4 18.0 26.2 28.0 13.7 14.6 19.7 21.6 10.0 7.9 2.5 4.0 14.2 10.0 10.1 8.5 4.5 7.2 12.6 13.3 2.0 1.4 1.3 1.0 2.5 2.0 1.1 1.0 1.0 1.8 3.1 3.6 0.8 .7 .2 0.2 "o.'s" 0.2 .2 0.2 .2 .2 1.0 .4 1.1 3.0 .8 3.5 .6 1.2 2.0 1.6 1.0 .6 0.2 '""."5' .6 1.0 .2 .2 "".2 .4 .2 LO .8 .4 .6 .6 SALMON AND TROUT IN ALASKA. 91 In the number of tubes in the lateral Ime (table 9) no great local variation was found, the averages varying from 132.1 to 135, with extremes 126 and 143. The value of this count is somewhat lessened by the varying degree to which the tubes extend on the caudal. This does not imply that all the high counts are due to this; in some instances high counts are due to the finer scaling. In other cases even low counts continued well on to caudal. The averages for the different years, however, show a striking similarity. The maxima vary from 132 for Karta in 1904 to 135 for Dolomi in 1903, but mostly fall on 133. Table 9. — Tubes of Lateral Line. Locality. 'Ka^a Nowiskay. . . Yes Bay. . . . Quadra Kegan Dolomi Date. Num- ber. Aver- age. Percentage of fish with specified numbers of lateral line tubes— 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142143 1904 512 132.1 0.2 0.4 1.4 3.2 12.7 17.5 23.4 20.5 12.7 5.0 2.5 0.2 0.2 1 1903 419 132.5 .5 1.1 1.7 9.3 18.6 19.3 20.0 15.0 9.3 3.3 .7 1.4 1904 513 133.1 .2 .8 1.1 5.6 11.3 17.7 23.0 15. S 13. 6 6.2 3.1 .40.8 1903 100 132.9 .1 .1 9.0 11.0 15.0 24.0 21.0 13.0 4.0 . . 1 i.o 1904 509 133.0 .6 2.0 4.9 11.2 20.8 21.6 20.2 11.8 3.7 2.0 . 6' . . . ...0.2 1903 298 132.9 .3 2.6 7.6 13.3 19.0 21.4 13.3 12.3 5.0 2.6 I.3I .3 1 1904 512 134.0 .4 .2 1.9 5.2 14.6 15.2 22. 19.5 11.5 6.2 1.7! .8 .4 1903 496 133.6 .4 1.4 .6 2.2 7.2 15.2 20.0 21.8 14.6 9.2 4.4 2.2 .2 1904 511 132.9 .6 2.1 5.4 11.1 21.0 22.9 19.5 9.7 4.9 1.3 . / 0.2 1903 99 133.0 1.0 9.0 6.0 19.0 23.0 17.0 13.0 8.0 2.0 1.0 1904 511 134.1 .4 .9 6.2 12.5 15.6 23. 5 18.6 12.5 5.6 1.7 . / .6 .4 6.2... 1903 200 135.0 ... .5 1.5 9.5 9.0 13.0 25.0 20.0 11.0 6.5 2.0 .5 .5 In table 10, showing rows of scales, it will be seen that the counts for 1903 regularly fall below those for 1904. This is due to the vary- ing procedure regarding the fine rows in front of dorsal and ventral fins, where the counts were initiated. A considerable variance is thus introduced. The branching of the rows under the front of the dorsal also affected the accuracy of the count. In some specimens a few scales' difference in the point at which this branching occurs adds or excludes a row in the count, which was made for the top rows from front of rayed dorsal down to lateral line, and for lower rows from fi'ont of ventrals up to lateral line, the lateral row being included in the total. Table 10. — Rows of Scales. Locality. Dolomi . . Kegan . . . Karta. . . Qiuvlra. Yes Hmv. Nowiskay Date. 1904 1903 1904 1903 1904 1903 1904 1903 1904 1903 1904 1903 Num- ber exam- ined. 511 200 5111 100 512I 420, 512j 496| 509, 300I 513 100 Aver- Percentage of fish with specified numbers of rows of scales - 1 age. 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 44.0 0.2 0.4 1.3 6.6 11.5 18.5 21.1 20.5 11.3 5.6 1.9 0.6 43.0 2.5 3. 5 16. 21.5 24. 18.5 9.0 3.5 1.0 .5 44.5 .6 .6; 3.5 10.0 14.0 18.7 22.5 16.4 8.5 3.5 .6 6.8 44.1 2.0 1.0 13.0 21.0 20.0 23.0 15.0 3. 1.0 1.0 45. .4 1.1 4.3 9.7 18.1 25.0 24.0 12.3 3.9 .2 .4 0.2 44.0 . 2 .2 3.3 13.3 19.3 17.1 20. 9 14.8 7.1 3.3 .2 45.2 ....| 1.5 6.2 10.3 16. 2 22. 1 19.3 13.1 7.5 2.5 .8 9 44.9 .01 .8 6.0 11.3 20. 8 24.4 20. 4 9.3 5.2 .8 .4 .2 45. 5 .2 1.2 3.9 10.6 15. 5 17.4 21.2 16.6 8.4 3.1 1.2 .6 44.3 .3 1.3! 2.0 10.3 17.6 21.3 23.6 12.6 (;. 6 3.0 .6 .3 45.8 8 3.3 8.7 11.3 19.3 18.9 19.1 11.1 4.7 1.9 .4 0.2 43.8 1.0 6.0 9.0 27.0 2fi. 17.0 9.0 2.0 3.0 92 SALMON AND TROUT IN ALASKA. The average number of rows of scales varies from 43 to nearly 46, with extremes 38 and 52. It was noted that the slender type form usually showed a slightly smaller number of rows. In conclusion, it may be stated that the various counts, wliile not amounting to demonstration, point to similarity in fishes from a given basin rather than to a heterogeneous mixture of schools. In general the environment at any one locality is very like that at any other; hence if the output of each stream were entirely segregated there would scarcely arise great differences of character among the different lots. The range of variation is, in fact, so small that it appears to fall within the possible effect of the personal factor if the counts were made by different individlials ; or of changing schools in any given stream if made by the same individual in sequence. With measure- ments, however, there is a difference. A sufficient number of fish for the purpose can be measured in a few days spent at any one locality, and the time element may be controlled by frequent change of place and renewed examinations at each. The remarkable difference found between the fish of such adjacent localities as Dolomi and the Moira Sound streams clearly proves the value of such data.'' STREAMS NOT UTILIZED BY SOCKEYES. It is unfortunate in the study of Alaska salmon that almost no streams not known to be frequented by sockeyes have been examined. The only exception of importance is the Anan of Bradfield Canal, which was visited on August 31, 1905. This stream is noted as the earliest and most productive humpback stream in Southeast Alaska. It is slightly less than the Naha in volume and about 3 miles in length below the first lake. Nowhere in the course are any impedi- ments to salmon at ordinary stages of water. The lake has about 160 feet of elevation, and on the above date had a surface temperature of 59.5°, about 1 degree higher than the stream. A tributary of the stream was 54°. At that time humpback and king salmon in small numbers were seen, and there were a few scattering red fish which were thought to be sockeyes. Two humpbacks were seen to jump in the lake. There is no apparent reason why sockeyes should not ascend this stream in numbers et^ual to the runs in such streams as those of Moira Sound. An examination of the lake might reveal some obstacle to their natural propagation. There was no tiling at that season apparent in the surface densities or temperatures to deflect a run entering Clarence Strait. From this strait through Ernest Sound and into Bradfield Canal the temperatures were in general increasing, 54° to 58°, and densities decreasing, 1.020 to about 1.010. In Behm a For a similar study of the winter flatfish {Pleuronectes aviericanus) , see Hermon C. Bumpus, On the identification of fish artificially hatched. American Naturalist, vol. XXXII, June, 1898, p. 407-412. SALMON AND TROUT IN ALASKA. 93 Canal at the same time temperatures rose from 55° at the head of Tongass Narrows to 58° off Spacious Bay, densities falHng from 1.0188 to 1.0124 at the same time. Subsurface tows demonstrated an abundance of plankton food throughout all the channels. Ketchikan Creek is a larger stream than Helm Bay Creek, which carries a few thousand sockeyes. It drains a lake and is frequented by humpbacks and cohos. No sockeyes are known to enter it. They could not reach the lake owing to falls about a mile from the mouth, but a school of fish looking for a suitable stream would learn tliis fact only after ascending the river. A few hundred yards below its head Naha Bay receives a small creek known as Steelhead Creek. This creek drains lakes of considerable size, and should carry water suitable for sockeyes, though they could not enter the lakes on account of an intercepting fall. The volume is small, but during the rainy season is ample for the ascent of fish, and many coho, dog, and humpback salmon, as well as steelheads, spawn there. No sockeyes are known to enter it, nor were any sockeye fry seen among the thousands of salmon fry taken there. It is probable that examination of the unstocked streams of Alaska would disclose others of interest in the question of stream selection. RELATION OF SIZE OF RUN TO SPAWNING AREA. There seems to be no relation between the size of the run at any given stream and the extent of spawning ground. Hetta, consid- ered good for from 50,000 to 150,000 fish, has comparatively little area of beds. The fish spawn mainly in one small creek about a mile in length, and along the lake shores. The Naha, as noted above, has a small area compared to its natural productiveness. Kegan has almost no spawning bed — ^only about a hundred yards of the main stream. At Nowiskay the borders of the lake are used almost entirely, none of the entering streams being suitable. Yes Bay and Karta streams both have excellent and extensive beds. On the other hand, the stream at Ward Cove has a greater area of good spawning ground than any of these streams except Karta and Yes bays, yet it yields too few sockeyes to pay for fishing. Karluk Lake has many tributary creeks that are used by spawning fish, but the total area seems scarcely commensurate with the enormous productiveness. With the exception of the few streams just mentioned, little is known of the spawning grounds of the Alaskan sockeye. Up to 1903 no attempt was made to arrive at the natural fecundity of the spawn- ing beds; hence the investigator is absolutely without standard or means of accurate comparison. 94 SALMON AND TROUT IN ALASKA. SELECTION OF SPAWNING GROUND. CONDITIONS REQUIRED BY THE SOCKEYE. Interesting preferences are shown by the mature salmon on coming out of the lakes to spawn. The first run of sockeyes at Fortmann Hatchery (usually of somewhat smaller fish) enters McCune Creek. This creek is a mile or so in length and drains a slope to the southeast- ward of the lake, the mouth, as will be seen from the map, being but a few yards from the entrance of the main stream. The lower course is over fine gravel and has but a moderate fall. On July 28, 1903, Heckman Lake at the surface was 64°, Naha River at the hatchery 60°, McCune Creek 53°. There were no spawners in sight. August 27 the river was 61° and the creek 52§°, and fish had been spawning a week. At this time some fish were showing a preference for the creek, which was fenced. Considerable numbers of ripe fish were in the river. October 15 the river was about 46°; November 17, 36*°. Selective discrimination is also shown in Jordan Lake. Emma Creek, tributary to that lake, has a small lake near its source in which its w^ater is somewhat warmed. Gibson Creek, a larger stream, evidently has no expansion in its course, for its waters are always cold. Sockeyes are never known to spawn in Gibson Creek, which carries a temperature of 48° or less, while Emma Creek, between 50° and 60°, is to a small extent made use of. August 18, at Nowiskay, North AiTii of Moira Sound, the main stream above was 51°, the lake over the spawning beds 62^°. A few fish were in sight, but none spawning. This stream is 44° and the lake 46° to 45° during the spawning season. No sockeyes enter the stream, though dog and humpback salmon do. The sockeyes spawn about the lake shores. At Karta Lake, September 11, Willow Creek was 50^°; Alder Creek, 47^°. Sockeyes were spawning in both in about equal numbers, or slightly preferring Willow Creek, perhaps for its greater size. The main river belov/ the lake was 53 2° and full of dog and humpback salmon. At Yes Bay, September 19, 1903, the river above was-full of spawn- ing sockeyes, temperature 50°; sockeyes were also entering the flooded creeks at 70°. September 14, 1904, the surface of the lake was 52°, the water over the beds, 49.5°; in the pool above, 48.5°; a feeder of this pool, 46°. May 2, 1905, the temperature of the main stream had reached 49°. Early in September of that year salmon were spawning in water of 51°. At Quadra, August 1, 1903, the lake surface was 63° to 65°. Osten Creek, at the head of the lake, in which the fish spawn, 51° at 3 p. m. ; the sockeyes had just begun to appear at tliis stream. A fine creek SALMON AND TROUT IN ALASKA. 95 farther down the lake was 57°; it is said few sockeyes enter this. The hatchery brook was 46°. The bay just off the stream August 5 was 64° at 6 p. m., or practically the same as the outflow from the lake. At Kegan, October 3, fish were spawning in the main stream above at 47° and refusing a smaller stream at 43° flowing through the same mouth; later they were reported to be spawning in both, tempera- tures unlaiown, but doubtless equalized to a degree. At Ward Cove, September 6, a few sockeyes were spawning in the stream at 55^°, but this is not a^sockeye stream, properly speak- ing, though it has the required lake and an extensive stretch of good spawning bed. In the Wallowa Dr. Kendall observed that the dwarf sockeye preferred the warmer water after temperatures fell below 45°, though spawning continued until the temperature had dropped to 40°. These observations indicate that the natural spawning temperature for the sockeye is between 47° and 55°, probably no.t over 50° by preference. (See p. 97.) The natural fall temperature for creeks not draining lakes or extensive swamps is about 45° to 48°, lake outlets 55° to 60°. The presence in a stream's course of a lake of any large size and a low altitude may usually be at once discovered by tem- perature, except, of course, when the lake outflow is mingled with another stream of low temperature, as from a glacier. Ordinary springs have a summer temperature of 44° to 48°, winter around 40°. Creeks, of course, fall with the air temperature to freezing. In the same manner in which they approach the ascent of a river from the sea, the sockeyes school about the spawning stream a short time before entering, and after entering the current proceed by easy stages to the bed finally occupied. Frequently they are seen to drift back over a riffle they had almost surmounted, and appear in most cases to enter swift water with hesitation and caution. It was noted at the Fortmann Hatchery when high water prevented the usual seining of ripe fish that the natural spawning did not com- mence for a day or so after the fish covered the beds. A short time seems to be spent in selecting nests. The sexes of the Alaska sockeyes as found during the fishing sea- son are approximately equal, or perhaps there is more commonly a slight excess of males. Of 1,025 examined at Karta Bay June 24 to Jul}^ 8, 1904, the sexes were equal; 548 of the first arrivals at Nowiskay July 14-22 had an excess of 12 males, while of 551 at this same place August 10-31 males were still in excess by 19. At Kegan from July 24-29, 988 examples showed 28 in favor of the females; Yes Bay July 26-28, 1905, had an excess of males by 10 in 510, while on August 14 among 172 fish the females exceeded by 24; August 16-20, 1904, in 976 examples there were 31 more males 10731—07 7 96 SALMON AND TROUT IN ALASKA. than females. Dolomi August 4-11 showed a surplus of 21 males in 990; at Quadra August 24-29, 967 examples had an excess of 45 males. These figures show an average preponderance of males by 1.3 per cent. There seems to be no difference in the time of arrival of the two sexes. One sex may be more numerous on one day's fishing and the other on the following day. It is stated that in the Eraser during the first of the run, the males are greatly in excess. The presence of unmated males is to a cer- tain extent damaging, since they go through the spawning move- ments alone, and thus disturb the beds. Under wholly natural con- ditions there should be practical equality in numbers of the two sexes. PREFERENCES OF THE KING SALMON. The king salmon spawns in higher temperatures than the sockeye. On the McCloud River it begins spawning at a temperature of about 56°, the summer run completing the season within 3° or 4° of that temperature. In Battle Creek the fall run spawns in practically the same range. That they habitually seek a less depressed tem- perature than the sockeye is shown by their spawning below the lakes into or beyond which the latter species invariably continues. This has been noted at Karluk, on the Wallowa, and on the Salmon and Payette rivers in Idaho, though in short streams like the Yes Bay Stream the straggling king salmon may also go above the first lake. But wliile this temperature is suitable a higher can not be endured. It has been found that the summer run can not be suc- cessfully confined in lower Battle Creek. These fish naturally travel to the cooler upper reaches, and retention in the warmer water of the lower course causes their death. At the Trocadero Aquarium it was found that fish of 3 and 4 years can not endure a summer and fall temperature of more than 61°; an addition of 3 or 4 degrees is fatal. Moreover, at that period an abundance of water is essential. This latter condition, perhaps, enables the fish safely to make the ascent of the heated waters of the Sacramento. During May, when the height of the summer run is passing, the river temperatures range closely about 60°, at which temperature the fish are able to move at the rate of 10 miles per day. In August, when the fall run is traveling up the river, the temperatures are decreasing from about 80° to 74°. At this temperature they travel but half as fast, though at a more advanced stage of maturity. During October and November the temperatures continue to fall, varying at Sacramento City from 70° to 60°, and 60° to 50° during the respective months. At tliis time the spawning temperatures of 50° to 55° are found in the middle course of the river, and in seasons when the depth of water is suitable it is found that the salmon spa^vn largely on the shallows below Tehama, thus withdrawing large num- SALMON AND TROUT IN ALASKA. 97 bers from the propagation stations of the upper river. A careful study of the king sahnon's movements on the Columbia is much to be desired. The greater length of the river, more numerous branches, and higher latitude and altitudes make that basin an extra^ordinary field for research, and its very great economic value demands the results. SPAWNING STREAMS CHOSEN BY THE COHO, DOG, AND HUMPBACK SALMON. The coho apparently demands a yet lower temperature for spawn- ing than the sockeye. Where it enters lake outlets early, it presses into the higher waters in much the same manner. It differs in that it often selects streams not connected with lakes, but in all cases its later arrival from the sea finds the temperature low. Wliile the most active of any species, long journeys do not find favor with it. Wal- lowa and Baker lakes seem to be at about the limit to which it travels. In the Naha in 1903 the cohos were spawning October 27 in a tem- perature of 46°. This is probably about the temperature at which they begin. During October and later the small streams fall rapidly in temperature, so that scarcely any running water after that date would be too warm for spawning purposes. Since in favorable places the coho continues spawning throughout most of the winter, any temperature above freezing does not seem to be prohibitive. At Hetta it is reported that the cohos sometimes spawn in small num- bers throughout the winter, even as late as March. This lake is said to have areas of open water, perhaps due to warm springs, about which sockeyes spawn until February. Alexander (ms. notes) states that in the winter of 1902-3, while the lakes and stream were frozen over, one of the lakes of South Olga stream, Kadiak Island, was visited by a party of white men and Indians who found a large number of sockeyes and cohos frozen in the ice. These were thought to be late-run fish entering after the canneries had closed. Many carcasses of salmon that had been frozen in the ice were seen by Rutter at Karluk Lake in the spring of 1903. In that stream the occurrence at times of late runs arriving after the usual closing time of the cannery early in September is well known. They more frequently occur in seasons of a small early run, and at such times the cannery is held open for them. The carcasses of winter spawners were still to be seen on Trail Creek the last of May, 1904. The dog and humpback salmons apparently exercise less discrimi- nation than other species. The humpback was noted as working upstream in the Naha at 61° August 25, 1903, and in the Anan at 58° August 30, 1904; yet it had completed spawning in the outlet of Yes Lake September 11, 1905, in temperatures around 56°. Karta River, September 12, 1903, was full of both dog and humpback sal- 98 SALMON AND TROUT IN ALASKA. moil spawning at a temperature of 54°, and Stieelhead Creek contained many spawning humpbacks August 25, 190.3, at about tJie same tem- perature. It is probable that the optimum temperature for the humpback is about the same as for the king. The dog salmon seems to accept a slightly lower degree ; a few were spawning in a creek of Moira Sound as early as August IS in 48°. NATURE OF SPAWNING BEDS SELECTED. With the exception of the coho, the salmons appear to require a depth of water and a fineness of bed material somewhat in corre- spondence with their relative size. The king salmon spaA^ois in water of a depth up to 2 or 3 feet, with a bottom of gravel and coarse sand. No exact conditions seem to be required. In the Salmon River, Idaho, the kings are reported to avoid the coarser gravel and swift currents; in the Sacramento they will spawn on bowldery rapids. The sockeye affects shallower water for spawning than the king. On most Alaskan beds many occupy water which does not completely cover the male fish. In the selection of spawning beds this species is unique in requiring the adjacency of a lake; also it is the only species of the genus that ever spawns in still water. The lake shores chosen are more frequently portions through which a small rill flows or where seepage comes through, but sockeyes are known to spawn where no inflow of any kind is apparent. In the streams the beds chosen are similar to those well known for various species, but, as mentioned above, in water of less depth than is usual for the king. Sometimes bottoms of fine gravel are occupied, sometimes bottoms of larger bowlders where the current is so swift as to make the holding of position precarious. No observations are recorded as to whether the first comers occupy the more accessible grounds. They apparently distribute from the beginning, and there is no evidence that a spawner ever recedes from a position once chosen and occupied. Unless carried out by the current, even the spawned- out fish remain until they die about the pool where they have spawned. At Yes Bay stream in 1903 fish could be found throughout the stream as far as the falls, but strikingly few examples were attempting to pass higher. The writer has seen no fish attempt the dam on the Naha just above the spawning beds, though cohos are said to do so. A few individuals usually enter all the small unsuitable brooks during times of freshets, apparently returning with the subsidence of the water. In the Karluk tributaries many fish occupy beds where none of the material is of a size they can move, hence the naked eggs are swept dowTi the current and doubtless most of them lost. In bottoms of finer material occupied by many fish dead eggs in aljundance are usually to be seen in the eddies behind the hillocks formed by the spawners working up the gravel. SALMON AND TROUT IN ALASKA. 99 The coho, perhaps in its effort to reach low temperatures, frequently continues up the small streams to a point where the bottom material is coarse. In Steelhead Creek it passes humpbacks and dogs on the gravelly lower part of the stream and spawns in the rough upper por- tions. In all the small Alaskan streams it doubtless occupies the sockeye beds later in the season and perhaps does some damage, but the greater part of the run spawns below those beds and in creeks not entered by the sockeyes. Cohos, like the smaller species, while mak- ing their way upstream or when frightened out of the "nest," may frequently be seen in water too shallow to cover them. Whether these "nests" are excavated in shallows where the original depth is less than that of the fish, or whether the small amount of water is not more usually due to a fall in the stream subsequent to the fish's entry, has not been fully ascertained. Humpback and dog salmon may often be seen spawning on the same beds, but in general the former is more frequently seen on finer bottoms and in shallower water, while the more powerful dog salmon sometimes occupies bottoms of coarse gravel in some depth of water. The humpback is said sometimes to fail to get beyond the reach of high tide in the selection of its spawning place. DEPOSIT OF EGGS. The deposit of spawn on riffles and in small streams, where it is most subject to the action of floods and frost, at first sight seems to be poor economy. But the necessities are twofold, first, aeration, and second, protection from light and enemies. In water of much depth relative to volume the bottom current is retarded, \vith the result that fine sediment is deposited and the circulation through the gravel or other material is impeded if not entirely cut off. If the eggs were placed in such a region of little current, they would not be carried from under the mother fish and the material for covering and protecting them would not be carried down over them, hence any such movement as the spawning fish now makes would only lead to stirring up eggs and gravel, with the result of destroying the first spawn within a few days after its deposit. For this reason the lake spawning of the sockeye must be of less proportionate value. The action of the spawning fish in breaking away the gravel by means of forcing the body and tail against it achieves the double purpose of working the sex product down to the genital opening and of setting the fine gravel and sand in motion to cover the eggs. When first extruded the contents of the egg envelope do not quite fill it, the egg is soft and will bear considerable rough handling. Its specific gravity, only slightly greater than the water, holds it to the bottom, but allows it readily to drift along with the current. Upon finding an early lodg- ment the osmotic absorption of water "freezes" it to the object against which it rests, and any drifting particles of sand not heavy 100 SALMON AND TROUT IN ALASKA. enough to dislodge it readily cover and conceal it from the light. It is probable that the dead eggs referred to above as seen in abundance about a well-covered ground are as often the earlier deposits dug up by later arrivals as eggs that have failed to be covered. Live eggs are almost never seen among these, as should be the case if they are of recent deposit. Uncovered eggs must eventually perish by the light even if not consumed by egg-eating fishes and birds. In ordinary cur- rents the eggs drift perhaps but a few feet. Very early the ridge below the ''nest" rises to a height that creates a bottom eddy, causing the downward motion of the eggs to cease on its lower slope. For this reason when one of these ridges is opened the greater number of eggs ordinarily is found in its lower portion. During the season of spawning a well-covered ground becomes thrown, through the joint action of the fish and the current, into high ridges, or hillocks, sometimes as much as 2+ feet above the corre- sponding depressions or holes. Eggs may usually be found buried in these ridges, but their semibuoyancy and spherical form make it quite difficult in a current to recover them without a screen. The abun- dance of dead eggs which at times collects in the- eddies below these ridges has been mentioned above. On Karluk Lake in 1903, in the creek on which the spawners were counted, these beds were examined by digging in the gravel to find the condition of the deposited eggs. Between August 5 and September 2, 58 "nests" were so examined. In these were found 4,005 good eggs and 2,022 dead ones, or, in other words, about two-thirds of the buried eggs were found to be in good condition. On the latter date 587 eyed eggs were found under about 10 inches of gravel, with only 13 dead ones. This demonstrates that eggs will live and develop under proper conditions when deeply buried. In another stream, in the center of a nest, under 6 inches of gravel, only 29 of 620 eggs recovered were living. In a third bed of 1,140 eggs taken from the lower half of the nest, in a light current and from under 7 inches of gravel, only 28 were dead. In general, the observer records few eggs from locations in strong current; this was possibly in part from failure to find the eggs as well as from their scarcity. Most beds show a decided balance in favor of the good eggs. In two examinations of the connecting stream from one of the tributary lakes less than 4 per cent of the eggs were dead. COMPLETENESS OF SPAWNING. In 1903 a careful count was kept of the sockeyes spawning in one stream of Karluk Lake, the second from the outlet on the right or east side. This is a small creek, averaging some 10 feet in width, about 1 mile of which is used for spawning beds. From August 5 to Sep- tember 5 of that year, 21,756 spawned fish were examined in this creek, presumably the total number spawning there in that time and practically the total for the season. Of these, males were in excess by SALMON AND TROUT IN ALASKA, 101 about 3 per cent, the numbers being 10,723 females and 11,033 males. But one unspawned female was found dead. Dead unspawned males are more common. Of 636 females opened, about 80 per cent were entirely spawned out, i. e., with no loose eggs in the abdominal cavity; the; remaining 20 per cent had an average of 97 eggs unspawned, with the most in any instance noted 1,246. The sockeye carries between 2,500 and 4,000 eggs, an average, perhaps, of about 3,500. This remnant, then, amounts to about one-half of 1 per cent of the total number of eggs matured. The product of this one stream on the same basis of estimate is 37,000,000 eggs. It is believed that less than one-tenth the number of fish entering the lake spawned in the above- mentioned creek. Thus approximately 400^000,000 sockeye eggs were spawned in Karluk Lake basin in 1903. Sockeyes are reported by natives to spawn late in the winter, even under the ice, but it is doubtful whether it is usual for any noteworthy number to occur as a fall run, as with other species and in more southerly streams. In 1903 the spawning season was practically over early in Septem- ber. Since the fishing continues ordinarily into that month, the spawning should last much later. The double operation of cannery and hatchery, perhaps, accounts for its early close. . On the Naha the season continues into November. At Hetta the fish are reported to spawn in small numbers until later in the winter. PERCENTAGE OF NATURAL PRODUCTION. The percentage of natural production of fry is a matter of the utmost interest. In daylight observation of a well-populated spawn- ing ground one is sometimes struck by the absence of factors to cause damage. At Yes Bay, Karta Bay, Kegan, the Naha, Wards Cove, and other places one may see hundreds of various species spawning uninterruptedly. At times on the humpback beds where fish are numerous the dead eggs lie in numbers in the eddies for days and even weeks untouched, not a trout or sculpin in sight — though it is certain that at least a few are in protected places near by. The ducks arrive late after most of the spawning in that region is over. It is believed by the writer that the natural loss of spawn' has been overestimated. In Steelhead Creek in 1903 comparatively few cohos were noted — it is believed not more than 50 pairs ; yet over 3,000 fry of that species were sometimes taken in a single night as they were leaving for the sea. The run begins early in May and lasts until July. In six sets of the net in May and June an average of about 1,300 of these fry were taken, which indicates a run of between 50,000 and 75,000 for the season, or a product of at least 30 per cent. Similar facts regarding steelheads were noted in the same creek, which was observed care- fully in both 1903 and 1904 for spawning steelheads in an effort to take sufficient eggs for-an experiment. Not over a dozen pairs could 102 SALMON AND TROUT IN ALASKA. be seen either season, yet early in July the fry were migrating at the rate of 200 to 300 per night and the creek was full of them till the fall rains swept them out. Regarding the migration of sockeye young from Seton Lake in 1903, Babcock states : ' ' The sight was amazing, and impressed one with the fact that the percentage of natural fertilization of ova and the survival of the residting fry was greater than has been generally believed by the authorities." It will be noted that eyed eggs were found in the Karluk creeks on September 2. On this date the lake surface was 44°, having fallen about 10° in as many days. With the shortening of the days and prevalent cloudy weather the temperatures must fall rapidly from this date, and it is questionable whether any but the earliest deposited and hence most advanced eggs hatch before the onset of freezing weather. Fr}^ were found as late as August 1 in the stream just below the lake; species unidentified. It is improbable that any consider- able number of sockeye eggs hatch in time for the young to seek the shelter of the lake for the winter. With the congealing of the waters in the mountains the streams will become reduced in volume and the entire bed must at times become frozen over, part by its exposure and the remainder by the formation of anchor ice. The habit of the fish of spawning on the shallow riffles must unfailingly subject the spawn to this influence. That it is not an extermi- nating influence is due to the covering of gravel which the eggs have received. The depth to which the bared beds freeze in winter has not been ascertained. At points where a constant current perco- lates through, such as must occur on the spawning riffles, it is im- probable that this depth is so great as might be suspected in view of the low air temperatures. The effects of freshets are largely nullified by the season at which they occur. Late fall rains produce comparatively small floods for the reason that the precipitation in the hills falls as snow. The spring floods which come with the melting snows will fmd the eggs in great .part hatched, and fry are largely able to rise above the deposit. RELATION OF SPAWNING HABITS TO NUMBER OF FISH. The spawning habits of a species bear a suggestive relation to its abundance. The humpback in Southeast Alaska far outnumbers all others. This is a region of small streams, and practically all that have a suitable bed and are accessible from the sea are utilized by this salmon. The small size of the adult humpback makes spa%vn- ing possible in even shallower water than the sockeye or coho re- quires. The number of eggs, about 2,000, is perhaps less than in other species. The incubation period is less, than that of the sock- SALMON AND TROUT IN ALASKA. 103 eye or king. This, with the early arrival on the beds and conse- quent high temperature of the incubating water, brings the hump- back egg to a hardy stage early in the season and the fry to a stage of development that sends them seaward with the opening of spring. The small size and light color of the egg also may be advantageous. In regions of larger rivers, such as Bristol Bay, the number of hump- backs is much less in proportion to other species. The same is true for more southerly regions. In the Bristol Bay region the average size of humpbacks is considerably less than in Southeast Alaska. Whether these facts are due to the character of the streams can only be surmised. The limited number of cohos can not be attributed to scarcity of spawning beds of the nature required. These are scarcely less widely distributed than those suitable for the humpback. The spawn incubates in about the same time, but is somewhat larger and more brightly colored. The increased number of eggs, about 3,000, should offset the disadvantage of size. The cohos coming last to the beds, their eggs are undisturbed by other fishes, but since much of the spawning is late the loss from physical causes may be greater. The fry are probably less active, reach the sea later in the season, and perhaps suffer greater loss from predatory enemies. The dog salmon, while widely scattered, occurs in numbers only in selected streams. The eggs are very large, rather light colored, and run about 3,000 to the fish. The incubation period probably differs little from that of the humpback. The fry reach the sea early, and there is no obvious reason why this should be one of the scarcer species. The spawning habits of both king and sockeye are obviously advantageous. The king, resorting to large rivers, is able by its strength to reach waters where the conditions are most favorable. Other species do not work over its beds; and though limited in the main trunks entered, the king secures extent of territory in the nu- merous branches ; the habits of the young permit them to ol)tain the necessary food in the streams, and the distance of the trip to be per- formed insures their sufficient age and size for adequate protection upon reaching the sea and its rapacious inhabitants. The summer residence conserves the species to an extent, while the migration of the main body as fry prevents overpopulation of the limited fresh- water area. In this connection it is of interest to note that the fish found in the Columbia headwaters in 1895 were of less than the aver- age size for that stream, weighing only 10 to 14 pounds. Even allow- ing for tissue expenditure during the ascent of the stream this does not indicate that the stronger fish travel farthest, but rather the con- trary. It seems possible that at a certain point the beds are occupied by the stronger fish and that the less powerful go beyond in search of unpreempted grounds. 104 SALMON AND TKOUT IN ALASKA. The sockeye profits by its lake habit. An unoccupied food supply is thus made available, and protection for the immatere young is obtained in waters not naturally inhabited by many predaceous animals. In basins of large productiveness the overflow of fry, migrating as such, conserves the lake food supply. In the main the species secures immunity from disturbance on the spawning beds, passing beyond all but coho and king and avoiding the latter by selection of colder water. CHANGES INCIDENT TO MATURATION. The changes that are associated with maturation of the reproductive element vary in extent in the sexes, in the species, and with the age. The production of the jaws and growth of canine teeth that is more or less characteristic of all male salmons is seen also in the trout and the charr. The change in body form — the increase in depth by the ridging of the back — is much more developed in the salmon. The changes seem to be greater in the larger, and presumably older, fish. The grilse of the king show very little of it, scarcely more than the female, and in the case of fingerlings in the Sacramento which develop mature spermatozoa there is no evident external change. In the trout and charr the upper jaw elongates somewhat but does not become hooked. The lower jaw elongates and becomes knobbed or hooked; this swelling is usually received in a corresponding recess or notch in the upper jaw, but sometimes is sufficiently large to prevent complete closure of the mouth. After the spawning season is over these growths are to an extent resorbed, but the jaws never fully recover their original shape. In the Sacramento the king salmon of the spring run show none of these characters until some time after entering the fresh water. The late runs, being more developed, show some changes by the time they have reached the mouth of the river in Suisun Bay. The various changes, which have been fully described by Rutter, are progressive with the development of the reproductive elements. Little of the red color is shown by this species except in the large (old?) males, and even in these never approaches the brilliancy of the sockeye or coho. Neither do they as fully develop the body depth as species which frequent shallower waters. The Alaskan sockeye at the time it is first taken in the fisheries has already begun to show the hooked jaw, so that with care one can distinguish the sexes in the larger fish. Cases will occur, however, in which the female has jaws somewhat prolonged, and there are cor- responding cases of males with neat heads, so that in statistical work it is essential that the fish be opened to make sure of the sex. Late in the season, or in occasional schools at any time, the males are more readily distinguishable. SALMON AND TROUT IN ALASKA. 105 When the sockeyes emerge from the lake en route to the spawning beds the change has become complete. The entire fish, except the head and fins, has changed from its original green and silver to a bril- liant vermilion or maroon. The head retains its bright green color and the fins become variously dark. At this time the sockeye rivals the tropical fishes in beauty of coloration. The elongation of jaws and development of depth are extreme in large males; the females show little of it and acquire less brilliance of color, in some cases scarcely any. The grilse (Arctic salmon) exhibit less change than the large fish, but show all the characters to a degree. The dwarf sockeye exhibits very little change; sex of ripe fish taken in Wallowa Lake can not be distinguished by external appearance. At the time of entering streams for spawning they are thus described by Kendall : Back olive green, more or less spotted with black; dorsal and caudal somewhat spotted; sides dusky metallic blue or smoky with faint marks like parr marks; belly white or slightly dusky. Pectorals, ventrals, and anal black, tipped with white; outer ray of pectorals white; a general brassy luster after the fish has been out of water some time. Colors, spotting, etc., vary in different individuals, but there is never any red. In Alturas Lake in larger examples of the dwarf the red is sometimes present and the other changes are somewhat more marked. The coho shows more change in the female than any other species, and the jaw develops more knob, as in the trout. It is sometimes difficult to distinguish the sexes among spawning fishes in the Naha. Both may acquire a red almost as brilliant as the sockeye' s, so that in the water it is easy to mistake either species for the other. The extreme height of the back in the male of the humpback and the characteristic color blotching of it and of the dog salmon are well knowTi. The females of the two species, particularly the latter, retain silvery colors and neat forms more fully than other species. The useful purpose, if any, which these changes subserve is difficult to surmise. It may easily be believed that the development of teeth, and of jaws to render them effective, is a matter of defense with the male. It does not hesitate to make use of the weapon to drive away intruders, but its awkward and slow movements render its armature of no avail so far as trout and other egg-eating fish are concerned. The only service is in preventing other males of the same species from mating with the female, and even in this the ability is not always equal to the emergency. A small active male has been seen to slip alongside a spawning female during the absence of her consort in pur- suit of another interloper and administer all the attention shown by her regular attendant. In general it may be assumed that in this man- 106 SALMON AND TROUT IN ALASKA. ner nature secures the offspring of the well-developed mature males, which offspring is known to be stronger and more thrifty than that of the younger fish. The exhibition of beauty in color and form in the male, which has been thought to be the result of sex selection, can hardly be so con- sidered in the case of these fish. The female apparently has no option in the matter. The larger male will drive away the smaller and take possession. The peculiar compression of the male, with the flattening of the ventral aspect, increases its ability to work through the shallow water, which the small size of the female, without such change, allows it to enter. RETURN OF ADULTS TO SALT WATER. To test the question of the return of salmon to salt water after once entering a river system, Rutter, in 1903, made two experiments. Eight hundred salmon (species not noted, but probably mainly if not all sockeyes) were tagged and released in Karluk Lagoon (brackish water) 1 mile from the sea. Four hundred of these were released June 12 and 1.5. One was retaken outside in a cannery seine 1 hour after release, 3 were retaken entering Karluk Lake after 11, 13, and 34 days, respectively; 2 were retaken by seines in the lagoon, and 1 died in the lagoon. Doubtless many of the tags were lost, and perhaps many fish entered the lakes or were retaken outside and unobserved. Five hundred and fourteen were tagged and released on June 30. One of these was retaken in New Red (Ayakulik) River, 40 miles down the coast, 3 days later. One reached the lake in 10 days; another was found spawning there August 2; one was seined in the lagoon 2 days after release; and another, a green female somewhat fungused and worn, 27 days after. No others were reported. This experiment is very incomplete and most desirable data are lacking, especially as to whether the fish were taken into the lagoon in a live car or were seined in the lagoon after voluntary entrance. The influence of the tag, clamped to the lower jaw by a strong ring, in irritating the fish and stimulating unnatural action can not be estimated. The results point to the probability that some fish remain in the lagoon for several days, some reenter the ocean, and others spend varying times somewhere between the lagoon and lake. Another experiment was the tagging of 255 individuals at the mouth of the lake, from July 3 to 25. Of these, 123 were males, 64 of which were ripe and 66 showing more or less of decay, fungus, etc.; 132 were females, 23 ripe, and 70 in poor condition. These fish were taken in a trap built at the outlet of the lake, but on the shallow or small current side. This location unfortunately led to the securing of a large percentage of weak fish, and made the results less vahiable. In general, fish reach the lakes in sound condition. Of these tagged SALMON AND TROUT IN ALASKA. 107 jfish, 3 females and 1 male were retaken in the ocean by the cannery seines in from three to eleven days after tagging, 1 was retaken in the trap the following day, and 2 males were found in the spawning creeks. The main interest attaching to this result is in the number of ripe fish obtained at the entrance to the lake. The percentage of ripe fish in the number trapped is of no point, since the strong green fish entered in the heavj^ current beyond the trap. It may be that these fish were hindered from making an early ascent of the river by the seining carried on for the cannery, and that such a condition of ripe- ness is unnatural. If there is no purpose in a lake residence of about one month it is difficult to see how it should have been brought to be the normal habit. That these exceptions were abnormalities is shown by the failure to find any of these tagged fish in the nearly 22,000 spawned fish examined at a creek which these ripe fish would mogt naturally seek, the nearest good ground about the lake. The main body of fish entering Karluk Lake came in against the stronger current and in the deeper water. During the day they entered in small schools of 20 to 60 or more, at intervals of a minute or less. They seemed to linger about the foot of the lake some days. ENEMIES OF YOUNG SALMON. In the trap at the mouth of Karluk Lake in 1903, 190 charrs were taken between June 5 and July 25. The biggest catch (20) was made on June 5, after which the number fell off rapidly. The next highest catch was made June 26, about ten days after the arrival of the first salmon. On the 27th a charr of 535 mm. length was taken with salmon eggs in its stomach. On July 1 a steelhead, also containing salmon eggs, was taken. One hundred and fifty-seven, or about 82 per cent, of the stomachs were empty; 14, or 7^ per cent, contained remains of fish, mostly cottoids; 13, or 7 per cent, con- tained insects or their larvfe; 6, or over 3 per cent, contained mol- lusks, snails, and clams; 3 in addition had eaten eggs, 2 salmon eggs, and the third those of some small fish. It will be observed that these fish were taken in an "upstream" trap into which the fish would come from the river. In the river at this time were numerous schools of salmon fry and small fingerlings, yet there is no record that any stomach contained a single identifiable young salmon. During May and June many of these trout (?) were seen jumping near the foot of the lake. On the first day the trap was installed. May 29, 10 were taken; on the 31st, 9, etc. No young salmon are reported in these. Upon the occupation of the spawning beds by the salmon the trout follow to feed on their spawn. This probably forms a large part of their food until late in the season, when the maggots from the flies breeding in the dead salmon become more common. 108 SALMON AND TROUT IN ALASKA. On August 19, 131 cliarrs were taken in a seine haul in a spawning creek. One hundred and twenty-eight of these were feeding on maggots; 3 contained insects, and 1 a cottoid; 54 were without sal- mon eggs; but the remaining 77 contained from 1 to 110 eggs each, an average of 16. At this time there were few live spawning salmon in the creek, and few dead eggs in sight in the eddies, where earlier there had been an abundance. The trout were more numerous than at any previous time, in schools of 30 to 50 as well as many isolated individuals. In the salt water about Karluk Beach they were even more abun- dant than in the lake. Sometimes as many as 2,000 are taken in a single haul, and ordinarily at least 500. No effort is made to destroy them. July 20, 220 of these were examine S y F.— Doc. 627. Plate V. LEJe'08