Loading...
HomeMy WebLinkAboutAPA1826Susitna Joint Venture Document Number Please Return To DOCUMENT CONTROL Studies of Alaska Red Salmon EDITED BY Ted S. Y. Koo SEATTLE • UNIVERSITY OF WASHINGTON PRESS • 1962 Studies of Red Salmon Smolts from the Wood River Lakes, Alaska BY ROBERT l. BURGNER Contribution No. 110, College of Fisheries, University of Wash- ington, Seattle, Washington. This is a revised version of the first of two installments of the author's. thesis accepted in 1958 by the University of Washington in partial fulfillment of the require- ments for the degree of Ph.D. Abstract introduction The Seaward Migration CONTENTS Age Groups in the Seaward Migration Determining the Migration Index Size and Age of Migrants Procedure for Processing Samples Correction for Shrinkage of Fish in Formalin Preservative Evidence Regarding Selectivity of the Fyke Nets Diurnal Changes in Size of Smolts Migrating Samples Taken for Length Frequency and Age Composition Studies Method of Age Determination Relationship between Parent Escapement and Seaward Migration Influence of Climate on Timing of Seaward Migration Sun1mer Growth of Yearlings Prior to Seaward Migration Growth as Indicated by Scales Relationship between Scale-Growth Increments and Seasonal Climate 249 251 251 253 253 254 256 256 257 258 260 262 264 265 268 271 271 274 t 250 Seasonal Changes in Length Frequency of Smolts Pattern of Seasonal Changes Effect of Lake-Ice Breakup and Distance from Outlet on Pattern of Seasonal Migration Origin of Size Groups in the 1955 Seaward Migration Causes of Difference between Lakes in Size of Smolts Produced Climate Level of Basic Nutrients Competition for Food Relation of Growth Rate to Survival Fresh-Water Survival Marine Survival Effect of Smolt Size on Adult Population Size Summary Acknowledgment Contents 276 276 278 279 283 283 284 284 288 288 290 291 292 294 Literature Cited Figures 1-27 inclusive 294 following page 298 6. Studies of Red Salmon Smolts from the Wood River Lakes, Alaska Robert L. Burgner ABSTRACT Red salmon (On.cm·hynchus n.erka) smolts of the Wood River lakes, Bristol Bay, Alaska, were studied to determine fluctuations in abundance, growth, and survival and to ascertain causes of the fluctuations. Indexes of seasonal and annual variation in numbers of smelts migrating seaward were obtained by fishing fyke nets ~t the outlet of the lake system. Sampling systems to determine changes in age and growth of smolts were developed, including a method whereby the majority of fish sampled were measured alive and released. Diurnal changes in size of smolts migrating were found. The relationship between numbers of parent spawners and numbers of smolts produced was found to be highly variable for the year classes 1949 through 1955. Sources of variability in growth and survival were therefore sought. Amount of growth attained by yearling srnolts during the early part of the growing season and timing of seasonal migrations were related to timing of breakup of la...lte ice and te:r,nperatures following breakup. Scales of smolts and of returning adults were utilized to show that size at seaward migration varied with lake of origin. Differences between years and between lake areas in growth attained during first year of life were associated with differences in srawning population density. Size of yearlings at time of seaward migration was shown to influence marine survival and age at return to the spawning grounds. INTRODUCTION THE STUDIES described in this paper are a part of the ::t•'isheries Re- search Institute program directed toward furthering knowledge of the causes of fluctuations in abundance of red salmon runs in Bristol Bay, Alask.a. This report concerns primarily studies of red salmon smolts, or seaward migrants, of the Wood River system in the Nushagak district of Bristol Bay. Sockeye or red salmon (Oncorhynchus nerka Walbaum) spend approxi- 251 252 Alaska Red Salmon mately half their life in fresh water, enter the marine environment at an advanced stage and do not come in contact with the coastal fishery until their return migration. Such distinctive environments allow the con- venient division of fluctuation in abundance into those which occur in fresh water, at sea, and as a direct result of the fishery. The anadromous life history of this species permits detailed study of spawning and early life. For studies to be effective in evaluating mor- talities at these stages, adequate systems of observation are needed to cover the fresh -water history year after year in such a manner as to be quantitative and comparative. Determination of t..he mortalities causing fluctuations in abundance is essential to proper management and sus- tained utilization of the very valuable red salmon resources of Pacific coast rivers. Seaward migrants are of key significance, for they have reached the threshold of the marine phase of their life history. Number of seaward migrants related to number of parent spawners provides a direct measure of production of young. This relationship provides a measure of vari- ability in survival rate resulting from interplay of natural causes and establishes the existence of possible cyclic patterns in fresh-water sur- vival. Evaluation of seaward :mlgrants with respect to their later return as adults in catch and escapement provides a measure of marine sur- vival. With the latter information as a basis, forecast of adult returns from seaward migrations of known relative magnitude follows logically. In this report the procedures used in determining abundance, size, age, and migration timing of Wood River smolts are described. Evi- dence is presented to relate differences between years in migration timing, growth and survival, to factors of climate, population density, and locale of embryonic and juvenile growth. The main lakes of the Wood River chain are piedmont lakes of glacial origin occupying essentially bedrock basins, and lie in a general east- west axis bordered on the west by the Wood River Range (Fig. 1). They are joined by short rivers, with the 18-mHe-long Wood River connecting the lowest lake with Nushagak Bay. Wood River drains a basin of ap- pro~5.mately 1,415 square miles in which the principal lakes cover 174 square miles (U.S. Army Corps of Engineers, 1957). The Wood River lakes are deep, temperate, and typically oligotrophic, with bottom water temperatures varying near that of maximum density. They are ice ... · covered for approximately 6 months of the year, but show distinct thermal stratification h'ith formation of a true thermocline in summer. Average annual precipitation recorded at Dillingham, Nushagak Bay, over a period of 35 years was 25.86 inches (U.S. Army Corps of Engineers, 1957). Elevation, dimensions, and depths of the primary lakes are recorded in Table 1 as seen on the following page. The Wood River lakes constitute the main spawning area for red salmon Smolts from Wood River Lakes TABLE 1. PHYSICAL DATA ON PRINCIPAL LAKES OF THE WOOD RIVER SYSTEM Maximum Area Surface Length Name Recorded {Square Elevation (Miles) Depth (Feet) Milas) (Feet) Aleknagik 20 360 34 34 Nerka 25* 538 78 70 Beverley 22 617 38 100 Kulik 17 525 24 140 *Longest arm only. So1.1rces: Vertie (193R); U.S. Army Corps of Engiueers (1957). Maximum soundings furnished by author. 253 in Nushagak district. Spawning is distributed over the lakes in tributary creeks, spring ponds, rivers between the lakes, and along lake beach areas. All four spawning area types-creek, spring pond, river, and lake beach-are of importance in the Wood River lakes; their relative impor- tance varies considerably from year to year. The following terms will be used in discussion of the stages of young red salmon in fresh water: Fry-stage from time of emergence in spring through first gJtowing -season in the lakes; synonymous with underyearling, and age 0 fish Yearling-fish in second year of life; synonymous with age I fish Age II fish-fish in third year of life Fingerling-general term applied to young salmon during the second or third year of fresh -water life prior to seaward rrdgration; young salmon after fry stage Smolt-seaward migrant in second or third year of life THE SEAWARD MIGRATION Age Groups in the Seaward Migration The annual seaward migration in the Wood River lakes is comprised of three age groups of young reds. These age groups are: (1) fry from the adult spawning of the previous year, (2) yearling smolts, and (3) smolts in their third year of life, age II. The validity of these age designations for young red salmon has been established by a combined study of sea- sonal length-frequency progression and scale structure (Koo, 1962). Fry migrations from the Wood River lakes were not sampled by the smolt indexing procedure used at the system outlet. However, wh...tttever the magnitude of the seaward migration of fry, the numbers of returning adults with no fresh-water annulus on their scales have been inconse- 254 Alaska Red Salmon quential in the Nushagak District. The highest percentage :return of this group was found in adult returns from the 1949 spawning (1950 fry), and amounted to 6.3 per cent of the return run. Yearling smolts have constituted the major portion of the seaward migration in every year. During the season of migration they are readily distinguished from fry by their size and scale pattern. There is no over- lap in size between fry in the lakes and migrating yearlings during June and July (see length-frequency curves prepared by writer in Koo, 1962). Smolts migrating after two years in fresh water form a smaller pro- portion of the total migration than do year lings. Determining the Migration Index Responsibility for studies of the seaward migration of red salmop. at Wood River was divided between Dr. Koo and the writer. Dr. Koo's work was primarily concerned with the enumeration of young salmon in the seaward migration and with the calculation of an index t;o show the rel- ative abundance between one year and the other. His work included in this report covered the years 1951 to 1958; the year 1952 was used as the base year in the computation of indexes. The method used to ob- tain an annual index to magnitude of the Wood River seaward migration will be described briefly, since the data are utilized in discussion of fresh -water survival and of size and age changes in the migration. Although fyke nets have long been used to obtain samples of seaward migrants (Babcock, 1905; Chamberlain, 1907), the Fisheries Research Institute was first to apply the gear as a means of indexing annual mag- nitude of smolt migrations in large rivers. This gear, installed at the outlet of the Wood River lake system in 1951, has been used in the major river systems in Bristol Bay and in the Chi~£1ik River. At Wood River a winged fyke net (Fig. 2) was fished annually during the seaward migra- tion period of approximately two months. The net was set daily for a fixed period during the evening migration in the same location on the gravel shoal at Mosquito Point {Fig. 3 ). · The method of fishing was kept as constant as possible from yea.r to year. Total catch for each season furnished the annual index to abundance of seaward migrants. The fyke net used for indexing was set on the river bottom in approx- imately 4 feet of water. The net intercepted a cross section of the river approximately 8 feet wide and 4 feet deep. According to calculations based on river measurements taken by Koo on June 15, 1955, the net intercepted about 1/75 of L'le river width and 1/145 of the cross-sectional area across the outlet in line with the net site. Maximum current ve- locities occur across the shoal where the fyke net was set. Average current velocities taken at the fyke-net location on the date stated above exceeded 3 feet per second. Smolts from Wood River Lakes 255 Most of the smolt migration occurs during evening hours. This was shown by Koo in studies of the hourly pattern in timing of seaward mi- gration at Wood River. During 17 days of around-the-clock fishing with fyke nets in 1951, over 95 per cent of the smolt.s were caught during the evening period from 2100-0200 hours. Hourly distribution of migration during this evening period ~or the years 1955 through 1957, based on data collected by Koo, is shown in Figure 4. Large fluctuations in smolt indexes have been found from 1951 to 1958 (Table 9, p. 267), but no direct test of the reliability of the smolt index was made. Evidence as to .reliability of the method has come primarily from consistency of results: (1) The seasonal trend of catches of the index net, daily changes in magnitude with fluctuations in water temperature and wind direction, and catches in relationship to observed abundance of smolts at the lake outlet have all shown a pattern of consistency. (2) The proportion of the catch taken early in the evening has been consistent from year to year. During the 1955 through 1959 seasons, the evening fyke-netting period was extended to include the hours 2100- 0200. Catches for the 2100-2300 period, used in former years, have averaged close to the same percentage of the total catch during the 2100- 0200 period in each of these years (Koo, 1959, 1960): 1955 42.1% 1956 43.4% 1957 39.4% 1958 46.4% 1959 42.0% (3) Magnitude of adult returns has shown a relation to index of a bun- dance at time of seaward migration. Deviations to date appear to be explainable largely on the basis of recognized changes in ocean mor- tality. (4) In 1952 a second fyke net identical in structure was fished for 28 days during the season in a position 25-50 feet offshore from the index net. The total calculated catches of the two nets during the 28-day period were in very close agreement-147,991 fish in the index net, 160,343 fish in the offshore net. This was in spite of the fact that two nets could not be tended as carefully or fished in as favorabls a. location and circum- stances as a single net. Comparison of fishing results of the two nets fished simultan~ously is presented in Table 2 (p. 256). Although differences between days in length of time the nets were fished somewhat obscure the day-to-day fluctua- tions in magnitude of migration, high catch of one net was associated with high catch in the other, and low catch with low. The coefficient ai cor- relation between the unadjusted catches day by day was +0.84e. 256 Alaska Red Salmon 1'ABLE 2. COMPARISON OF FISHING RESULTS BETWEEN 'tHE MOSQUITO POINT INDEX NE-T AND A Sii;coNil FYKE NET SET 25-50 FEET FURTHER OFFSHORE, 1952 Date Time Fished I Catch of R~d Sm0lts -~ .... Hmu·s M:i.p,t;i;~B ~.nqt;;ii; N~'t Offahora Net June }..u 4 40 6,612 13,774 22 3 10 10,780 4,480 24 3 10 51,128 51,216 28 3 30 806 1,096 29 2 30 258 342 30 3 0 13 13 July 1 6 40 0 0 2 6 40 15,960 28,462 6 4 0 2,590 981 7 2 45 34,020 15,660 14 3 25 132 58 15 2 25 33 7 16 3 15 142 90 17 3 20 1,041 77 18 2 45 2:304 25,698 19 3 15 12,722 11,957 21 3 15 140 32 22 2 5 172 114 23 3 20 2,368 536 24 2 0 1,210 350 30 3 15 1,193 1,753 31 3 25 1,751 1,534 Aug. 1 3 0 2,383 1,741 2 2 0 41 23 4 g 45 g7 56 5 1 53 8 18 7 2 0 21 45 10 2 0 136 230 TOTAL 147,991 160,343 RATIO 1 . 1.08 . Size and Age of Mig·rants Procedure for Proct~ssing Samples Preliminary studies were made of seasonal and annual size and age compositi.cm Qp,an~~g Q£~~rr:iii5 1r1 Wu-oo i'iiver smolt§ during the yeru.·s i949--53. Beginning 1fi 1954, more complete sampling was initiated, and the number of samples taken annually was increased severalfold. Length, rather than weight, was used to measure size of individuals in order that large samples of smolts could be processed rapidly. All length measure- ments were from tip of snout to fork of tail. Samples for study were drawn at the fyke-net site during the ~vening - Smolts from Wood River Lakes 257 index period, held overnight in live-boxes, and processed the following day. Care was exercised in taking samples at the fyke-net site to avoid selection resulting from possible size stratification of fish during the sampLing procedure. Tests of procedure showed no indication of size selection in drawing samples. In 1954 a procedure of measuring fish alive v.ras adopted in order to increase the number of length-frequency samples without unnecessary killing and preservation of smolts. The live fish were measured under anesthesia. Lengths of these fish were not directly comparable with those of fish measured after preservation in formalin. The measure- ments of preserved fish were converted to correspond to the live fish measurements, since by far the greater numbers were processed alive and released. Correction for Shrinkage of Fish in Formalin Preservative Measurements taken of a smolt sample collected on June 2, 1954, il- lustrate changes in length of red smolts accompanying preservation in 10 per cent formalin. Fish in the sample were measured alive under anesthesia, remea..c;ured approximately 24 hours after preservation, and again after 5 months in formalin. Table 3 summarizes the results: TABLE 3, COMPARISON OF LENGTHS OF SMOLTS MEASURED UNDER ANESTHESIA AND AFTER PRESERVATION Anesthetized Preserved 24 Hours 5 Months Age I--139 fish Mean length in millimeters 86.3 82.0 81.1 Mean decrease in length 4.3 5.2 Percentage shrinkage 5.0 6.0 Age II--15 fish Mean length in millimeters 106.3 101.7 100.4 Mean decrease in length 4.6 5.9 Percentage shrinkage 4.3 5.5 Measurements in Table 3 indicate that the major changes in length of red smolts occurred within a few hours after lr • .illing, in preservative. Roughly 80 per cent of decrease h1. measured length occurred during the first 24 hours in both age groups of smolts. While the actual shrinkage was slightly more in older fish, the percentage shrinkage in terms of total length was slightly less. Another sample, taken on June 9, 1954, is used to illustrate shrinkage in different length classes. The fish were measured under anesthesia, sorted into 5 mm length groups, preserved, and the sampJe remeasured over 5 months later. Tabulations by length class are given in Table 4. Difference in mean length between anesthetized and preserved fish is around f mm for each length class, percentage difference in length de- 258 Alaska Red Salmon creasL"'lg with increased size. Hile (1936) likewise found that shrinkage in length of ciscoes was greater on a per cent basis for smaller fish. His specimens, ranging from 150-350 mm, were first pre}3erved in formalin and l~ter transferred to alcohol. TABLE 4. SHRINKAGE OF RED SALMON FINGERLINGS IN LENGTH DURING STORAGE IN 10 PER CENT FORMAL!~* Mean Length (in mm) Difference in Length Class Number of 5i Months Mean Length (in mm) Specimens Measured Alive under After Mill!-Per Cent Anesthetic Preserved meters 73-77 4 76.5 71.3 5.2 6.8 78-82 53 80.8 76.0 4.8 5.9 83-87 82 84.6 79.6 5.0 ~.9 88-92 17 89.0 84.2 4.8 5.4 93-97 11 94.8 89.5 5.3 5.6 98-102 6 98.5 93.2 5.3 5.4 108-112 1 108.0 103.0 5.0 4.6 Total Age I 170 84.61 79.61 5.0 5.9 Total Age II 4 100.75 95.50 5.25 5.2 *Measured under anesthetic before killing, remeasured 5i months later. other measurements taken of red smolts show that the initial and most abrupt decrease in length is apparently not due to formalm preservative, but rather occurs as a result of rigor mortis. Shrinkage in length between fish measured alive under anesthesia and remeasured upon reaching a state of rigor mortis without being preserved in formalin averaged between 2 and 3 per cent. This agrees closely with results reported by Shetter (1936) for brook and brown trout averaging 6-7 inches in length. To compensate for total shrinkage of preserved specimens, 5 rom was added to length measurements of yearling smolts preserved several months in 10 per cent formalin. A slightly larger correction was needed for age ll fish which ranged over 100 mm. Evidence Regarding Selectivity of the Fyke Nets Since the smolt samples taken at the fyke-net site were to be used for size and age studies, it was necessary to determine whether the nets were size selective. This was done by comparing samples with those taken by other gear. Indirect evidence did not indicate that index fyke nets, as used at Mos- quito Point, tended to select small smolts. Sampling within the lake system by means of beach seines, lake traps, tow nets, and fyke nets failed to furnish any evidence of an abundance of larger or older finger- Smolts from Wood River Lakes 259 lings not represented in the samples taken by fyke net at the outlet of the lake system. The fyke nets used within the lake system were set in rivers between lakes, usually in swifter water than at Mosquito Point. Length-frequency study established that fish in samples captured within the lakes by this gear were similar in size or smaller than those sampled concurrently by fyke net at Mosquito Point. The differences in length frequency were those to be expected between Mosquito Point migrants a"'1d those reds still to migrate. More direct information regarding possible fyke-net selectivity was obtained in 1954. It had been observed that beach seines of the type used for sampling red fingerlings in. the lakes tended to capture a school in its entirety if the school was surrounded initially by the set of the seine. The fish were observed to herd toward the beach and to avoid contact with the seine until they were finally pocketed in the bag in shallow water .. Size selectivity was believed negligible once the young salmon were sur- rounded by the set. It was decided, therefore, to obtain samples simul- taneously by beach seine and standard fyke net at Mosquito Point for study of fyke-net selectivity. The seine used was tapered, 200 feet long, with a large bunt made of 1/8-inch-mesh saran-plastic screen. The sets were made with an outboard boat a short distance above the fyke-net site and in slower current than at the site. In order to make catches at the same time, all fish were removed from the cod end of the fyke net, the cod end re- placed, and as soon as a sufficient number of fish appeared in the fyke net, seine set was begun. The moment it was determined that smolts were caught in the seine, catches from both gear were removed. In this manner, two pairs of catches by seine and fyke net were obtaine.d on each of three nights, June 15, 16, and 17. The catches were held overnight in live-boxes and a 2-pound sample (about 150-170) processed from each in standard manner the following day. Runs were relatively light during the three evenings of sru.11pling and in- dividual samples taken by both beach seine and fyke net were composites from a few schools. Fyke-net catches were obtained by fishing a con- siderably longer time, and as a result contained components of more schools than did seine catches. Fyke-net catches were made during 6- minute fishing periods on June 15 and 16, and during a 14-minute and a 29-minute period on June 17. It required less than a minute to set the seine, and no schools could enter once the seine was set. Length fre- quencies obtained by the two types of gear were compared with com- posite frequencies of all samples taken during the same evenings. The frequencies of June 15 are compared in Figure 5, those of the smaller smolts running on June 16 and 17 in Figure 6. The feature at once apparent in both figures is the greater hetero- geneity in length frequency among samples taken by beach seine. The results support observations that there are marked differences between 260 Alaska Red Salmon schools in size composition of the individual fish. The fyke-net samples were composed of portions of a greater number of schools, hence tended more toward the average size composition of all schools passing, and thus showed less variability in size composition between samples. Comparisons of individual samples obtained simultaneously by the two types of gear are presented in Figure 7. The over-all mean length of smolts taken by fyke net was 86.07 mm, by beach seine, 86.95 mm, a difference of only 0.88 mm. The two seine hauls made on June 15, cap- tured several very la1:ge age n smolts of a size range not found in the fyke-net samples. Fewer fish of this length range (over 105 mm) were captured by either gear the following two nights. It is not known whether unusually large fish of this size avoided capture by the fyke net on June 15 or whether capture of these fish by beach seine and not by fyke net was a chance occurrence. It is unlikely that fish of this size range com- prised a substantial proportion of the run, since they have never been captured in numbers by any gear used. If selection exists, however, some error would arise in determining the proportion of age II smolts in the seaward migration in years when sizes of fish in this age group run large. In the size range below 105 mm the samples gave no evidence that the fyke-net set in the current velocities at Mosquito .Point selected smaller fish. There was, then, no evidence that the proportion of yearlings re- tained by the fyke net ·was influenced by the size of yearlings migrating. Comparison of beach-seine and fyke-net samples did bring out the heterogeneity in length frequency of smolts among schools. As a result of these tests, it was decided that when samples for size and age study were to be drawn from the catch, the fyke net should be fished long enough before lifting to make the resultant catch a composite of several schools. Diurnal Changes in Size of Smolts Migrating During early years of sampling at \Vood River, marked diurnal dif- ferences between samples were noted in size of smolts taken. The diurnal differences were first studied in detail in 1954 to determine possible causes and to evaluate their effect on the sampling schedule adopted. Hwrly samples of the run were taken on two evenings of heavy migration. Length frequencies of these samples are presented in Figures 8 and 9. A pattern of general decrease jn the size of smolts during the evening was evident in the samples. This was caused by a decrease in the size of yearlings, since very few older fish were present in the samples. Evidence that such was a regular feature of the migration in 1954 was indicated by additional samples taken during regular fyke-netting hours, which were from 2100-2300 each evening. The mean weight of fish in the first and last samples taken each evening is shown in Table 5. In 20 of 21 instances, average weight of individual migrants in the first sample was gr<'ater than that in the last. Smolts from Wood River Lakes 261 TABLE 5. MEAN WEIGHTS OF SMOLTS TAKEN IN FIRST AND LAST SAMPI.:ZS DURING ~E 2100-2300 EVENING FYKE-NETTING PERIOD, MOSQUITO POINT, MAY 30 THROUGH JUNE 19, 1954 - Time Time Difference Mean Weight of Fish ~n Grams Date First Last in First Last j S 1 Differenr.e Sample Sample Minutes Sample ampe May 30 2240 2300 20 7.03 6.55 0.48 31 2117 2300 103 9.65 6.54 3.11 June 1 2113 2300 107 6.72 6.05 0.67 ·- 2 2130 2300 90 7.05 5.89 1.14 3 2140 2300 80 6.05 5_.93 0.12 4 2218 2300 42 6.43 6.01 0.42 5 2200 2300 60 6.61 6.53 0.08 6 2200 2300 60 6.67 6.13 0.54 7 2154 2300 66 7.11 5.04 2.07 8 2106 2300 114 5.97 5.34 0.63 9 2130 2300 90 6.21 5.21 1.00 10 2104 2300 116 6.06 5.34 0.72 11 2105 2300 115 8.10 7.26 0.84 12 2104 2300 116 6.45 5.53 0.92 13 2104 2300 116 5.47 5.24 0.23 14 2107 2300 113 7.32 6.26 1.06 15 2120 2300 100 5.89 5.74 0.15 16 2133 2238 65 5.40 6.20 -0.80 17 2200 2243 43 5,2_7 5.10 0.17 18 2122 2300 98 i'i.'t8 5.50 0.28 19 2107 2300 113 6.05 6.01 0.04 TOTAL 1,827 13,87 MEAN 87 5.54 5.88 0.66 In 1956 and 1957 a substantial fraction of the migration was composed of age II smolts, and although there appeared to be a slight tendency for age n smolts to ru:n earlier in the evening, the pattern was far from consistent. The tendency for larger smolts to run e~lier in the evening is still under study, for in other years it was not as evident as in 1954. Possible diurnal change in selective action of the fyke net was dis- carded as a cause of diurnal change in size of fish in the cai:cho If change in efficiency of the net occurred, it would be expected in the direction of increase in efficiency with increasing darkness. This would be ex- pected to increase rather than decrease the proportion of large fish caught; however, a decrease in proportion of large fish was established. Further, as shown in Figures 8 and 9, not only do the proportions of large fish decrease, but size groups of small smolt s are found in late evening samples that were not present in samples taken earlier in the evening. The appearance of small fish in later samples cannot be ex- plained by either increased or decreased efficiency of gear. Suggested reasons for the phenomenon of diurnal change in sizes caught 262 Alaska Red Salmon are (1) that the larger fish reaching the lake outlet have a greater mi- gration stimulus because of their more advanced development, 1 and there- fore, show less hesitation in entering the river from the lake; or (2} that the final distance from the daytime milling area to the lake outlet is traversed more swiftly by larger fish because they tend to be stronger swimmers. Neither alternative can be explained solely on the basis of behavior of individual fish. Red salmon fingerlings exhibit a marked schooling habit. It is quite unlikely that the faster swimmers or individuals with a greater urge to migrate break away from their unit schools to reach the outlet during evening in advance of their slower members. Further, fingerlings migrating pa...~ Mosquito Point are in definite schools, cross- ing the shoal areas in distinct waves. The behavior of the school, made up of individua· :s, must therefore be considered in any explanation of the phenomenon of diurnal chanr.e rn size frequency. This requires no particular modification of the ~11ggested reasons advanced, for it has been shown repeatedly by sampling in the lakes that length-frequency composition varies greatly from one school to the next. Schools com- posed of larger fish may simply begin migration out of the lake earlier in the evening or may migrate faster than those composed of smaller fish. Migration of t.he individual is thus apt to be controlled to a considerable degree by size and impulse of its companions. The explanations advanced, particularly the latter, require the concept of a milling area or reservoir in the lake near the outlet where young tend to congregate during the course of the day or early evening prior to migration out of the lakes. Observations of the behavior of red smolts near the outlet o:f Lake Aleknagik, particularly in 1949, support a res- ervoir hypothesis. Schools of red fingerlings were observed during cer- tain days feeding and milling in considerable c;:>ncentrJ.tions in the more shallow lower end of the lake near the outlet. Evening observations were made from a boat anchored over submerged light-colored panels near shore about 300 yards above the lake outlet. During the day only occa- sional schools of fingerlings were observed over the panels. During early evenL"lg many more schools were observed crossing the panels, but movement was not decisiveJ.y directional. Schools crossed the panels moving away from the outlet as well as toward the outlet. However, from about 2140 on, movement was more decisive and all schools swam de- finitely and directly towa:r:d the lake outlet. SampJ es Taken for Length Frequency and Age Composition Studies Most of the smolt samples taken from 1954 through 1956 were 2-pound 1Svardson (1955, p. 246) suggests that the physiological development achieved by a fish cannot be measured properly by size or age alone, and states the need for a better measure of "physiological age.'' Smolts from Wood River Lakes 263 samples, averaging well over 150 fish per pound. In 1957, individual sample size was reduced to 1 pound, averaging over 100 fish per pound. Total number of samples and fish used in determining the season's length frequency and age distribution a..~e given in Table 6. 'l'ABLE 6. TOTAL NUMBER OF SMOLT SAMPLES USED FOR LENGTH- FREQUENCY AND AGE DETERMINATIONS, 1954-57 Total Season's Number of Number Year Catch of Smo1ts Samples of Fish during Index Period* 1954 745,832 52 8,923 1955 904,236 110 14,419 1956 1,325,910 63 11,134 1957 627,884 94 10,207 *Daily fishing hours: 2100-2300 in 1954, 2100-0200 in 1955 through 1957. During nights of heavy migration, weighing, recording, and releasing of the catch occupied the full efforts of the fyke-netting crew. However de- sirable, it was not at all feasible to draw a uniform fraction ot each catch as a sample. It was the practice to draw a minimum number of samples spaced at fixed intervals on a sampling schedule that could be followed each year. This sampling procedure did not fully take into account di- urnal changes in size and age composition that may have occurred. Samples taken at or between 2200 and 2300 were used in the present study for the 1954 period. The standard evening fyke-netting period was extended to 0200 beginning in 1955, and three samples were generally drawn at approximately 2200, 2300, and 2400. All three were proce:.;sed unles~ the evening catch was light (less than one index point, or 4,000 fish). If the days of light catches continued, a sample was processed every three days. Some flexibility was exercised, particularly if the evening run was unusually heavy early or late in the evening. Samples during an evening were not weighted according to the proportion of the evening catch they were to represent, since this would have been an unwarranted refinement beyond the accuracy of the fyke-netting and sampling procedure. Table 7 (p. 264) gives the proportions of the migra- tion each season that were represented by daily samples. In computing seasonal length-frequency curves, daily samples were weighted by the daily catches they represented to adjust for changes during the season in the magnitude of catches. From 1954 through 1956 this was done by combining frequencies for adjacent days within five- day periods in which no significant change in size of migrants was ob- served, then weighting the frequencies by the total fyke-net index catch for days included. Since this involved a decision as to which days could be grouped, the procedure was modified in 1957 so that each daily fre- quency was simply weighted by magnitude of daily catch. 264 Alaska. Red Salmon TABLE 7. PROPORTION OF THE WOOD RIVER SMOLT MIGRATION REPRESENTED BY DAILY LENGI'H-FREQID~CY SAMPLES, 1954-57 No Sample At Least One Sample Two or More from Daily Catch from Daily Catch Samples Daily Year Number Per Cent Per Cant Per Cent of Season Number of Se1son Number of Season of Days Catch of Days Catch of Days Catch 1954 5 1.0 30 99.0 20 73.7 1955 18 ) 12.9 36 87.1 23 78.4 1956 31 ! 4.0 32 96.0 19 86.3 1957 25 2.8 34 97.2 29 88.4 Method of Age Determination Beginning i~1 1954, nearly ,"'1 samples of smolts drawn during evening fyke-net sampling were. proc-essed alive and released, necessitating an age-sampling technique adaptable to this procedure. Groups of age I and age II smolts could generally be separated by length frequencies at the beginning of the migration season. Later in the season sizes of the two age groups overlapped. In order to determine age composition it was necessary to preserve for scale study those fish which fell in the range in which lengths of age I and age II fish over lapped. Those fish in the length frequency well outside this overlapping range needed merely to be counted if age composition of the sample was the only information desired. The zone of overlap in length frequency of the two age groups was determined in the field at intervals during the season by examination of smolt scales under a microscope. As shifts in length frequency were observed, scale samples were examined to detect shifts in the overlap range. Sample fish in the range of overlap were preserved at the time the daily live-smolt samples were measur~d. Complete smolt samples, preserved at frequent intervals during the season, provided a check against the range of overlap designated in the field. By this method, an adequate age breakdown of a large number of samples was obtained with a minimum of preserved specimens. Scales from fish in the preserved samples were mounted on glass slides and projected for age reading with methods and equipment de- scribed by Koo (1962). Age readings of smolt scales for the years 1951- 57 were made by the writer, the 1958 age readings by Koo. An index to annual age composition of smolts in the Wood Riyer sea- ward migration was obtained by weighting samples by the magnitude of the catches which they were to represent. Grouping of daily age samples and catches was the same as used in weighting length frequencies. Al- though changes during the season in age composition and in magnitude of migration were considerable, corrections in the season's age com- position attained by weighting the samples by magnitude of catch were Smolts from Wood River Lakes 265 not extreme. This may be illustrated by comparing the per.centages of age ll fish obtained in this manner with values obtained by giving equal weight to each day on which samples were taken. This comparison is made in Table 8 for two years when the percentages of age ll fish were high. TABLE 8. EFFECT OF WEIGHTING ON AGE DETERMINATIONS, 1956 AND 1957 Method Samples weighted by magnitude of fyke-net catch Daily samples given aqua! weight Percentage of Age II Smolts 1956 1957 21.6 19.3 17.6 20.3 Relationship between Parent Escapement and Seaward Migration Previous investigators have found considerable variability in fresh- water survival rates of red salmon under natural conditions. At Karluk Lake on Kodiak Island this variability was detected by indirect calcula- tions of fresh-water survival after measures of marine survival had been determined through fin-clipping experiments (Holmes, 1934; Barnaby, 1944). ?¥lore direct measurements have been ma.de by investigators in British Columbia lakes, with the most extensive work at Cultus Lake. From 1925 through 1936 tests of artificial versus natural propagation were conducted there. In experiments on natural propagation, adult spawners entering the lake and resultant numbers of seaward migrants produced were counted, average fecundity was determined from samples of adult females, and potential number of eggs available for deposition was calculated for each year. Survival to smolt stage from calculated numbers of eggs spawned varied from 1. 05 to 3. 23 per cent in three ex- periments conducted prior to initiation of predator control experiments (Foerster, 1938). However, natural spawning was permitted only in certain of the years involved, and eggs were shipped to other watersheds during the experimental period, which may have disturbed natural con- ditions in Cultus Lake even prior to predator control (Thompson, 1950, 1951). At Lakelse Lake, British Columbia, the same procedures as those at Cultus Lake were used to obtain calculated survivals from natural spawn- ing (Brett and McConnell, 1950). Range in survival from egg to smolt for seven brood years is given as between 1.0 and 4.9 per cent (Canada, Fisheries Research Board, 1957): "There has been no obvious rela- tionship between the number of eggs laid and the number of smolts pro- duced .... " Fresh-water survival rates for a small sockeye population at Port 266 Alaska Red Salmon John Lake in the central coastal region of British Columbia were studied by similar methods. Foerster {1955) reported survival values of 0.5, 3.0, 5.0, and 3.0 per cent. The smolt production was reported as having "·varied between 11,000 and 20,000 from egg depositions of 300,000 to 2,000,000" (Canada, Fisheries Research Board, 1957}. At Babine Lake in the upper Skeena system the number of smolts was estimated by marking and recovery methods rather than by means of a cow1ting fence (Withler, 1952). The calculated ratios of smolts produced to number of eggs available fur deposition range from 0.49 to 2.49 per cent for the brood years 1949 through 1954 {Canada, Fisheries Research Board, 1957a); with regard to the relation between initial population size and survival, they stated: "Not only did larger runs produce more smolts; they produced a better percentage survival from eggs to smolts, indi- cating that still better production J:night be expected from still larger runs.'' At Lcike Dalnee on the east Kamchatka Peninsula, Krogius and Krokhin {1956a) reported fresh-water survival of sockeye over 17 brood years (1935-51) to be highly variable, ranging from 0.04 to 1.3 per cent of eggs available for deposition. Krogius {1951) gave an even wider range of values {0. 005-1.04 per cent) for the same lake. Krogius and Krolrnin (1956) commented that marine mortality was much more constant than fresh-water mortality, and that there was no correlation between es- capement and return. Their data do show that the four parent spawning populations producing the greatest number of smolts per spawner were well below average in size. In summary, measurements of fresh-water survival have succeeded in showing variability between brood years in survival. No consistent relationship between magnitude ot initial population and percentage sur- vival to seaward migration has been demonstrated. At Wood River annual indexes of abundance of smolts, 1951-58, clas- sified into component indexes of age I and age n smolts (Table 9), provide a basis for study of the relationship between parent escapement of red salmon to spawning grounds of the Wood River lakes and relative num- bers of smolts produced. In the first three years, 1951 through 1953, age composition values were not based on a,s complete sampling and require a brief explanation. Both the limited smolt samples taken during the 1951 migration and the age composition in the adult returns of 1953 and 1954 to the Nushagak District indicated that the percentage age II fish in the smolt migration was less than 20 per cent. The smolt samples and sub- sequent adult returns of the 1952 seaward migration indicated less than 1 per cent age II smolts. All migrants in these years were assigned as yearling smolts in the comparisons made. This does not seriously alter the indicated survival relationships because of the small total migration in 1951 and the very low percentage oi age n smolts in 1952. The maxi- mum possible degree of error is trivial (Table 10). Determination of Smolts from Wood River Lakes 267 age co~;r.>sition in the 1953 seaward migration was more complete, and the ·v-alue of 4. 7 per cent age II smolts is believed to be a reasonable approximation. TABLE 9. INDEXES OF ABUNDANCE OF WOOD RIVER SMOLTS BY AGE (1952 = 100) - Year of Per Cent Wood River Smelt Seaward Abundance Ind<.:!x Migration Age II Age I Age II Total 1951 <20.0 9.9 1952 < 1.0 100.0 1953 app. 4.7 282.2 13.9 296.1 1954 4.2 420.4 18.4 438.6 1955 2.0 217.3 4.4 221.7 1956 21.6 258.2 71.2 329.4 1957 19.3 133.6 31.9 165.5 1958 35.0 150.0 80.8 230.8 TABLE 10 . WOOD RIVER ESCAPEMENTS AND SMOLTS PRODUCED Wood River Index Values of Index Units Year Escapement Smelts Produced per 1,000 Age I Age II 1949 101,000 :nso 452,000 100.0 13.9 1951 458,000 282.2 18.4 1952 227,000 420.4 4.4 1953 516,000 217.3 71.2 1954 571,000 258.2 31.9 1955 1,383,000 133.6 80.8 - *Using maximum values of 20 per cent migration and 1 per cent age II smelts index values would be: 1949 year cla.ss 1950 year class 8.9 11.?. 9 Total Spawners 9.9* .10 113.9* .25 300.6 .66 424.8 1.87 288.5 .56 290.1 .51 214.4 .16 age II smelts in 1951 in ~952, the smelt During the years 1949-52 Wood River escapement data were obtained by comprehensive ground sur\reys of the spawning areas, supplemented by aerial survey estimates. Since 1953 total escapement estimates have been obtained by the much more accurate method of trunk stream enu- meration from towers situated on Wood River. The spawning ground surveys were continued after 1953, and provided a basis for adjusting the 1949-52 escapement estimates to represent total escapement esti- mates for the system. Values given in Table 10 were determined and provided by John R. Gilbert, who has summarized the Fisheries Re ... search Institute spawning-survey data for this period. 268 Alaska Red Salmon The relations between values representing parent escapement of red salmon to the Wood River spawning grounds and smolts produced are given in Table 10 3..t,d graphed in Figure 10. The rate of reproduction is represented by the number of smolt-index units per 1,000 pa..t'ent snwn- ers. While survival values !or the brood years 1951, 1953, and 1954 are closely gr.ouped, values for the remaining four years fail to :fall in line. The best rate of reproduction, that of 1952, was 19 times that of the poorest, 1949. No definite reh.tion between size of escapement and smolt index is seen. The two extremes of escapement provided the poorest reproduction r-=ale. It appears that conditions for survival were more important than abundance of spawners in determining abundance of smolts. Minor changes in the relation between years would be expected il survival values were corrected for differences between years ii1 average fecundity of females and in sex ratio. The primary factor influencing average fecundity would be marine-age composition of female spawners. Mathisen (1962) showed an average fecundity of 3,639 eggs for female red salmon returning to Pick Creek, Lake NeJrlr..a, after two winters in the ocean, and an average fecundity of 4,290 for those returning after three Winters. The proportions of the two groups, small and large fe- males;-in the Wood River escapement varied between 23 and 84 per cent small females during the years 1949-55. Sex ratio of escapement during these years could not be determined with accuracy. Values obtained ranged between 52 and 60 per cent females. Approximated corrections for se;x ratio and fecundity were applied to the data but were found to modify only slightly the highly variable relation between parent escape- ment and smolts produced. Influence of Climate on Timing of Seaward lr1igratirm Causes for fluctuations in fresh-water survival rate must be sought in changes in the ecosystem to which salmon are susceptible. Varia- bility in seasonal timing of seaward migration and in growth attained by smolts is a reflection of differences between years in conditions of fresh-water environment. Foerster (1937) demonstrated that differ- ences between years in timing of seaward migration at Cultus Lake were largely associated with water temperatures in months immediately pre- ceding and during migration. As to lakes that freeze over, it was soon learned that seaward migration of sockeye followed shortly after breakup of lake ice (Babcock, 1904; Chamberlain, 1907), so that differences in timing of migration were associated with differences between years in dates of breakup of lake ice. Cessation of seaward migration has also been associated with rising water temperatures (Chamberlain, 1907; Ward, 1932; Foerster, 1937; Parker and Vincent [1956]). At Cultus Lake, Foerster found, over a Smolts from Wood River Lakes 269 period of nine years, a close relationship between rising mean daily temperatures in Sweitzer Creek and cessation of seaward migration. He stated: "When surface waters rise above 10° C the passage of young sockeye appears to be inhibited and consequently, as summer stratifi- cation is set up and the epilimnial v..-a.ters rapidly increase in tempera- ture, they form a temperature barrier or ~lanket to terminate the further migration of sockeye" (Foerster, 1937). This ntemperature barrier" to seaward migration of red salmon was first d~scribed by Ward in an attem].X to explai,n the origin af landlocked sockeye in Baker Lake, Wash- ington. His cornrnent is: ''When the surface water of the lake passes 10° C the ptigrators which have been continuously at or near the surface appear to desert that level and withdraw into deeper layers" (Wa,rd, 1932). Striking differences in seasonal timing of Wood River seaward migra- tions have occurred during the year~ un9,er study. These differences were measured by daily changes in magnitude of smolt catches made by the index iyke net at the outlet of the Wood River lakes. The cumulative- catch curves of Figure 11 show as much as a month's difference between years i..tl seasonal timing of pe'ak catches. At least four influences had a significant effect on timing of migration in t..~e Wood River lakes. The most influential factor was climate. How- ever, changes in rel~tive numbers in the component s'ubpopulations of smelts from the lakes, differences between years in size and age of smolts, and differences in rate of growth in early summer appeared to contribute. The last three factors will be discussed i.p a later section. Changes in smolt behavior associated with changes in water tempera- ture have been observed since the beginning of these studies in 1949. The start of substantial seaward migration has been consistently delayed until lake outlet temperatures reach 38°-39° F following breakup of lake ice in late May or early June. Timing of initial migration was closely associated with breakup of ice on Lake Aleknagik in spring. This is shown in Table 11 where the years are listed in order of date of brea.h.~p and dates by which time 5, 10, and 20 per cent of the season's index catch had been made. For the 5 per cent dates, the years fell essen- tially in the same orde:r as the date of lake ice breakup. This initial sequence was modified by the continuing influence of weather and other factors; yet by the time 20 per cent of the .mtgration was over, the re- lation to date of brea.lrup was still 2Lpparent. Cessation of seaward migration coincided quite closely with warming to over 50° F (10° C) of the lake-su.rface waters m.?asured at the Lake Nerka weather station. Records of lake-surface temperature for the seasons 1952 through 1957 are graphed in Figure 12. Dates by which time the lake-surface temperatures at the Lake Nerka. station passed and remained above 50° F varied over: a month between years, ranging from June 16 in 1954 to July 21 in 1955. These dat~.s were compared with dates by which time 90 per cent of the season's cumulative-index catch 270 Alaska Red Salmon TABLE 11. REI.ATION .bETWEEN DATES OF BREAKu'"P OF LAKE ICE ON LAKE ALEKN.<\GIK AND EARLY-SEASON SEAWARD MIGRATION OF RED SALMON SMOLTS AT WOOD RIVER INDEX SITE Breakup Dt~.te, Date 5% of Date 10% of Date 20% of Year* Migration Migration Migration Lake Aleknagik O'ver Over Over 1954 May 26 June 1 June 2 June 2 1953 May 27 May 31 June 3 June 11 1957 May 28 June 2 June 7 June 12 1951 May 30 June 4 June 7 June 9 1956 June 1-3 June 10 June 12 June 15 1952 June 7 June 11 June 12 June 14 1955 June 10 June 23 June 26 June 29 *Years listed in order of breakup date. had been taken at Mosquito Point. The smolt catch each year falls off rapidly beginning at about this time of year (see Fig. 11). In Table 12 it is demonstrated that in five of the six years, the 90 per cent level was passed within a very few days of the date the lake-surface temper- ature reached 50° F. Cessation of migration is probably closely as- sociated with a general thermal ((!hangs in the lake epilimnion. TABLE 12. RELATION BETWEE:N' LAKE-SURFACE TEMPERATURES AT LAKE NERKA WEAT".riER S'l'A'l'ION AND SEASONAL TIMING OF SMOLT SEAWARD MIGRATION AT WOOD RIVER Year 1954 1957 1953 1956 1952 1~55 Date Surface TempeJ:-atures at Cabin Bay, Lake Nerka First Rose over 50 11 F for Longer Than One-day Period Jwte 16 June 16 June 21 July B July 19 ,July 21 Date C"Umu1ative Smolt Catch Reached 90% at Wood River Index Site June 15 June 26 June 23 July 12 July 19 July 16 The relation between a~.nnual timing of smolt migration, breakup of lake ice, and la..lte temp~r~mres following breakup is shown in Figure 13 for the years 1952 through 1957. For each year, the time interval be- tween breakup and attainment of 50° Flake-surface temperature is com- pared with the 5 per cent and 90 per cent levels of migration. Delay in ice breakup and warming of the lake resulted in delay in smolt migration from the lakes. As to sockeye yearlings remaining behind in the lake, Ward (1932) and Foerster (1937) suggested that rising temperatures create an epi- limniallayer of warm water through which the fingerlings are reluc- Smolts from Wood River Lakes 271 tant to pass. No such reluctance on the part of red fingerlings has been observed in the Wood River lakes, for the greatest concentrations ob- served inshore and near the surface at Lake Nerka were seen during the week or ten days after surface temperatures exceeded 50° F (10° C). The record catch of red fingerlings made in surface lake traps fished for six years at Lake Nerka was made on a day the surface tempera- ture at the trap site read 61° F (16.1° C). Summer Grawth of Yearlings Prior to Seaward Migration Growth as Indicated by Scales While temperature has been shown to influence timing of seaward mi- gration, and is known to affect growth as well, there seems to be no pub- lished data showing that temperature differences between seasons directly affect growth of young sockeye in the lakes. Differences between years in size of smolts are usually associated with differences in population density or in food level rather than with climate. Indirect evidence of the effect of climate on growth was given by Krogius and Krokhin (1948, 1956a), who stated that at Lake Dalnee annual fluctuations in abundance of plankton were closely associated with variations in completeness and intensity of spring overturn and with depth of thermocline. Biogenic elements were considered to be more thoroughly circulated in years of more complete overturn and more available to phytoplankton in the epi- limnion if the epilimnion was thick. It was suggested further that magni- tude of the plankton . rop affected the size of smolts, although the evi- dence given does not appear to be at all clear-cut. In the preceding section it was shown that seasonal climatic conditions affected timjng of seaward migration. It will be shown in the discussion to follow that differences in climatic conditions have a definite bearing upon the amount of growth gained by red salmon yearlings :in June and early July prior to seaward migration. The amount of new summer growth that has occurred prior to seaward migration is .indicated on a scale of a yearling smolt by the increment of new growth formed beyond the annulus. As shown by the scales, growth is negligible in the spring prior to breakup of lake ice, but following breakup and warming of lake waters a rapid increase in size of yearlings occurs in the lakes. This is indicated on the scales of yearlings by a band of wide circuli laid down beyond the broken, narrowly spaced rings of the annulus (Fig. 14). Increase in scale radius, measured from the focus of the scale, is not strictly proportional to increase in length of fish over the size ranges with which we are concerned. Investigators have found that in juveniles of many different species of tish there is an extended period after scale formation when the linear dimensions of the scale increase at a rela- tively faster rate than does the length of the fish. Dunlop (1924) det.:t;r- 272 Alaska Red Salmon mined that sockeye of the lower Fraser River show a fairly constant increase in scale radius relative to length of fish between 35 and 100 mm, a lessening rate to 115 mm, a.'ld a decreasing ratio above 115 mm. Meas- urements were presu.mably made from preserved specimens. Clutter and Whitesel (1956) gave tht: relation between mean scale radius (greatest scale radius, magnified 200x, measured from focus to edge of scale) and .mean fork length as established by 12 samples of preserved Fraser River sockeye smolts with mean lengths ranging from approximately 50-150 mm. Their linear equation for the regression line was: radius = 1.323 fork length -33.11 The writer has also found a proportionately greater increase in scale radius than in body length of preserved fish over lengths encountered in age I and age II groups of fingerlings from Wood .River lakes. The per- centage increase in scale radius was about one-third greater than the percentage length increase between fish of 80 and 108 mm fork length. Smolt samples taken in 1953 at Mosquito Point illustrate changes in amount of summer growth appearing on scales of smolts at different times during the season. In each sample, scales were mounted from two fish of each millimeter length in the sample or of one fish of each length if two were not available. Several scales were mounted from each fish. The scales were taken from the area immediately above the lateral line slightly forward of the adipose fin. The scales were first projected at high magnification and the most symmetrical from each fish was chosen for measurem...;nt. This scale was then projected on line paper and the circuli marked off along a Hne, beginning at the center of the central plate and extending to the scale edge along the longest axis of the scale. The amount of new summer growth in the scale was calculated by measuring the distance from the narrowest. circulus, assumed to represent the outer edge of the annulus, to the sca1e edge. Percentage of growth made during the current summer was computed as 100 times this measurement divided by the total pro- jected distance from mid-central plate to scale edge. No measurable amount of summer growth was discernible in the first four samples, taken on June 1, 3, 6, and 12, 1953. The amounts of summer growth measured on the scales in samples taken after these dates are presented in Table 13. Percentages were grouped according to length of fish since it was conceivable that percentage of summer growth showing on the scales might vary with fish length. However, the percentage of summer growth on the anterior scale radius was about the same for all size groups in each sample (Table 13). Sample averages show a steady increase in summer growth between June 16 and July 15, on which date 30.9 per cent of the scale radius constituted new growti1. Increase in summer growth on the scales and TABLE 13. SUMMARY TABLE OF AMOUNT OF NEW SUMMER GROW'1\H APPEARING ON SCALES OF RED D:ARLINGS IN SEAWARD MIGRANT SAMPLES TAKEN AT MOSQU!!O POINT, 4AKE ALEKNAGIK, 1953 Length of June 16 June 22 June 28 ~'ish in Number Average Per Cent I Number Average Per Cent Number Average Per Cent Millimeters of Fish Summ.er Growth of Fish Summer Growth of Fish Summer Growth 73-77 8 3.6 8 10.1 6 15.7 78-82 10 1.9 9 12.3 9 16.8 83-87 8 4.5 9 14.4 9 15.8 88-92 18 2.7 10 11.7 8 14.7 93-97 4 3.4 8 10.2 7 15.3 98-102 1 o.o ---- 103-101 - - ---- TOTAL 49 44 39 AVERAGE 3.0 11.8 15.7 Length of July 4 July 9 July 15 Fish in Number Average Per Cent Number Average Per Cent Number Average Per Cent 'Hi1limeters of Fish Summer Growth of Fish Su;mner Growth of Fish Summer Growth 73-77 --1 27.6 -- 78-82 8 22.4 2 24.5 5 32.8 83-87 7 19.1 7 28.0 10 30.7 88-92 9 20.0 11 26.7 7 30.4 93-97 5 23.5 7 21.3 2 22.8 98-102 4 23.8 2 22.7 2 30.6 103-107 --1 22.8 9 31.4 108-112 ----1 33.6 113-117 __.. ---1 35.8 TOTAL 33 31 37 AVERAGE 21.4 25.3 30.9 -------~--··-·--·--· ----------------·----------------------"-- I 274 Alaska E!ed Salmon the dates involved are shown graphically in Figure 15. The delay in first appearance of summer growth is very similar to that found by Dombroski (1952) for 1951 Babine Lake smolts. Measurements of new growth on the scales illustrate summer growth increments are to be found in smolt samples taken at successive dates during the migration period. The inc:t'ement pattern is not necessarily the same as would be obtained by repeated sampling of a single population of fish within the lakes prior to migration, but does represent growth as ob- served in successive groups of seaward migrants. As will be discussed later, main seaward migration is generally completed before substantial summer growth appears on scales of migrating year lings. Relationship between Scale-Growth Increments and Seasonal Climate The time of the appearance of summer growth on scales V'a.ried between years. Much of the difference in timing could be related to differences in climate between years, just as the time of seaward migration was in- fluenced by climatic factors. The difference between years in time of appearance of summer growth is illustrated by comparing the average amount of growth beyond the an- nulus present on scales of yearling smelts sampled on the same date each season. In this instance July 5 was selected, since by this date summer growth is normally well differentiated on scal~s, yet some migraUon is still in progress. Smolt samples which had been preserved on or very close to July 5 were available for the years 1952 through 1957. Dates of samples, numbers of scales measured, and average percentages of summer growth found along the longest axis of the scale radius are given in Table 14. ?ABLE 14. PERCENTAGE OF NEW SEASON GRO~T.H ON SMOLT SCALES, 1952-57* Date of l{t•:!nber Per Cent New Season Year of Scales Growth on Scales Saniple Measured Mean Standard Error 1952 July 7 30 7,9 1.15 1953 July 4 33 21.5 1.13 1954 July 5 55 19.9 0.70 1955 July s). 50 6.0 1.02 July sf 1956 July 5 22 4.5 I 1.26 1957 July 6 32 20.7 0.88 *Samples taken at Wood River fyke-net site on or near July 5. Two rather simple measures of late spring climatic differences among these years are used for compu-ison with scale growth. The first meas- urement is date of lake-ice breakup at Lake Nerka, central lake in the Smolts from Wood River Lakes 275 Wood River chain. The second is based on temperature records be• ween breakup and the time when samples were drawn in earJy July. I.1 this case, daily mean air temperatures were used rather than water tenper- atures, first, because daily water temperature records were not avail- able for all years, and second, because surface temperatures can: 10t be expected to reflect too closely transfer of heat into the lake. It is prob- able that mean air temperatures give a more accurate index t >lake warming than do water surface temperatures. Differences betweell sea- sons in air temperature were measured by the cumulative number of air "temperature units" beginning with the date of breakup of lal:e ice. The number of temperature units per day was set as the differen!e be- tween the mean temperature for the day and 32° F. In Table 15 the years 1952-57 are listed in order Df time of l~ke-ice breakup. Included, in addition, are the date on which the smolt flample was drawn for scale measurement, average percentage summer ~owth found on the scales, number of days from breakup to date of sample, and cumulative number of air temperature units from breakup to date of sampling. The years 1953, 1954, and 1957 group together a~. years of early breakup with a large number of cumulative temperatuxe units and a large amount of new growth on smolt scales by July 5. Convarsely, the late breakup years 1952, 1955, and 1956 show fewer curr;ulative temperatm·B-units ;md little summer growth on smelt scales. A closer linear relation appea:rs to exist between cumulative temperat1re and scale growth than between length of time after breakup and scale growth (Figs. 16A, 16B). Year 1954 1953 1957 1952 1956 ].955 TABLE 15. RELATION BETWEEN AMOUNT OF NEW SUMMER GROWTH ON SMOLT SCALES, TIME OF LAKE-ICE BREAKUP 1 AND TEMPERATURE FOLLOWING BREAKUP Date of Per Cent New Season Growth Sample on Scale Radius July 5 19.9 July 4 21.5 July 6 20.7 July 7 7.9 July 5 4.5 July 6} 6.0 I July 8 Number of Days from Lake Nerka Breakup 40 35 34 30 24 22-24 Cumulat. ve Air Pate ple TU' s to of Sat\ 844 805 863 548 44& 382 The existence of t.l).e relation betweAn spring temperatu:..·es and g·rowth of red fingerlings should not be regarded as proof of direct camle and effect. Tempe.t'ature records are simply one measure of gener2.1 dif- ferences in environment caused by climatic conditions. Prior to spring breakup in Bristol Bay lakes, the surface layer of snow and ice insttlates the water below from increasing warmth of spring air, penetrat .on of light, and mixing action of wind. Following breakup, surface tern .. >era- 276 Alaska Red Salmon tures rise quickly to 38°-39° F. Spring overturn of the lake occurs ac- companied by rise of nutrients from fue lake bottom, and the amount of solar radiation inc:reases steadily toward a maximurn in Jtme. At break- up, and for about two and ..:me-half weeka, there is no indication of re- cent growth on fingerling scales. This is to be expected since the plank- ton bloom associated with sprjng overturn and increased light and tem- perature is not instantaneous. Lag in abqndance of copepod and cladoceran feed, combined with reduced metabolism and activUy of the fish until warming of their environment occurs, results in no growth on finger ling scales until an accumulation of about 300 temperature units has occurred. Weather and temperature before breakup have little apparent eifect1 then, on subsequent summer growth and activity of young red salmon in the lakes except by their influence on time of breakup of lake ice. How- ever, air temperatures after breakup are shown to bear a relationship to growth attained by young salmon. The amount of sunshine, the 84'ttent, direction, and timing of wind action, and the amount of rainfall are ad- ditional factors known to influence the summer temperature regimen in the lakes. These also affect to some degree the growth of phytoplankton 1 zooplankton, insect, and fish populations within the lakes. Seasanal Changes in Length Frequency of Srnolts Pattern of Seasonal Changes Seasonal changes in length frequency and age of smolts have been de- scribed by a number of biologists. In many red salmon populations, smolts migrating ear li&r in the season tend to be larger than later mi- grants of the same year. class. Ba't"naby (1944) found a ma.th.ed decrease in average size of age ill smolts du.t,ing successive weeks of sampling at Karl;,lk Lake in the years 1925-36. A less marked difference in age IT smo:ts was attributed to the fact that L'l.ter rrligrants had added new sum- mer growth. He concluded that "the urge to migrate seaward is related to the size a.nd growth rate of fingerlings .... " Gilbert (1916, 1918) also found a consistent seasonal decrease in yearling smolt size of Rivers Inlet sockeye durl11g the yea.i:'s 1914 through 1916. Clutter and Vlhitesel (1956) reported that ~ .,_ various sockeye populations of the Fraser River syr::tem, yearling smolts were of fairly consi.stent size dur]ng the main migratory period but frequently exhibited some trend towards size de- crease during the season. An exception to seasonal decrease in size was reported by Dombroski (1954), who found a definite seasonal trend of in- crease in length and weight of Babint~ Lake yearlings in the years 1950, 1951, and 1953. The inc.~. eases were probably ascribable to additional growth gains in the summer shortly before migration, shown by Dom- broski. (1952) for the later migrants of 1951, or to th,~ contribution to th.e migration made by Nilkitkwa Lake smolts, which may be first to appear in numbers in the migration because of close proximity to the Smolts from Wood River Lakes 277 system outlet, but which tend to be smaller in certain years because of poplla.tion density (Johnson, 1956, 1958). Parker and Vincent [1956] hypothesized that differential timing of migration of two different races was the cause of some of the seasonal changes found in size of smolts at Lake Kitoi in 1955. In the Wood River lakes, study of size changes in smolts occurring during the season revealed that the seaward migration of young red sal- mon passing Mosquito Point was by no means composed of a homogeneous stock of fish, and that distinct seasonal patterns of change occurred. Seasonal patterns of change in numbers of different size and age groups were of special interest in that they provided an insight into causes of the marked duferences found, and particularly in that a relation was indicated between size and survival rates. Length frequencies of smolt samples taken at Mosquito· Point and weighted as described previously (p. 263) are graphed by five-day periods in Figures 17 to 20. These graphs present seasonal changes that oc- curred in numbers, sizes, and ages of smolts migrating in the years 1954-57. In the graphs, the first period in 1955 and the last period in all four years have been extended to include more than five days because of low levels of migration at these times. In Figure 21 the season's fre- quencies for the four years are presented. The season's frequencies for 1958 and 1959, provided by Koo, are also included. Features to be noted in these graphs are differences in timing of mi- gration (already explained in part by differences between years in cli- mate), differences in age composition, and seasonal and annual differ- ences in size of smolts in the migrations. Especially noteworthy is the bimodal size frequency of age I smolts in 1955, 1957, and 1958. Migration during the years 1952 through 1957 followed a general pat- tern. Yearling migrants (age I) passing Mosquito Point during the first several days of the season were small, and the first waves of large yearlings reached Mosquito Point several days after migration had be- gun. Grouped frequencies of 1957 provide the clearest illustration of this phenomenon (Fig. 20). Particularly in the years when the size range was great, as in 1955 and 1957, yearling frequencies were dis- tinctly bimodal. The difference in size cannot be accounted for by dif- ferences in amount of recent summer growth. Initial appearance of large yearlings at the lake outlet occurred well before summer growth was registered on the scales, and the major share of the season's migration was over before appreciable summer growth on the scales was found. In the Wood River lakes a tendency was apparent, pronounced in cer- tain years, for the modal group of large yearling migrants to decrease in average length during the season until rapid new summer growth re- versed the trend. However, the modal group of small yearlings was first to appear at the outlet of the lake system-an apparent reversal of the pattern that larger smolts tend to migrate first (Barnaby, 1944; Gilbert, 278 Alaska Red Salmon 1916, 1918; Clutter and Whitesel, 1956). Furthermore, smaller smolts tended to complete migration earlier in the season. This, in addition to other evidence, led to the hypothesis that the migrations commonly are comprised of numerous groups of smolts which have experienced dif- ferent growing conditions in different parts of the several lakes. The small early migrants were believed to be fl~om Lake Aleknagik, the lower lake in the system, and the waves of larg:er migrants from Lake Nerka and other upper lakes. It was of particular interest to test this hypothe- sis, since it would aid in determining what conditions are responsible for size achieved by the smolts. Effect of Lake-Ice Braakup and Distance from Outlet on Pattern of Seasonal Migration The timing of ice breakup in the various lakes of the Wood River sys- tenl suggests that smolts from the upper la#:~'S would be later in reaching Mosquito Point. Substantial seaward migration has never been observed during spring fyke-netting operations at Wood River, Little Togiak River, and Aguluk.pak. River until the lake outlet temperatures have reached at least 39° F following breakup of the lake ice. This is in agreement with migration threshold temperatures of approximately 40° F, reported for sockeye fingerlings in lakes with winter t,emperatures falling below this level (Chamberlain, 1907; Foerster, 1937, 1952; Parker and Vincent, [1956)). Any delay in ice breakup of the upper lakes in the Wood River syste1n might therefore be expected to delay arrival of upper-lake mi- grants at Mosquito Point. Lake Aleknagik has in every year been the first to become ice free, although brealrup on Lake Nerka may occur the same day. The contrast between years in breakup sequence is some- times rnarked; for example, 1954 versus Jl956 (Table 16). TABLE 16. SEQUENCE OF LAKE-ICE BREAKUP IN 1954 AND 1956 Lake Aleknagik Nerka Little Togiak Kulik Breakup Date 1954 1956 May 26 May 26 Before May 30 (No record) June 1-3 June 11 June 15 June 20 (approx.) The colder 1956 season extended the breakup period, and the ice cover on the lakes delayed the arrival of threshold temperatures for migration in the upper lakes areas. The rate of smolt migration between the outlet of Lake Nerka and the outlet of Lake Aleknagik at Mosquito Point, a distance of 18 miles, is not known. How.ever, marking and recovery experiments conducted in 1956 shed some light on possible rates of migration. The maximum rate of movement of red fingerlings released from the Catherine Cove ..... Smolts from Wood River Lakes 279 (Lake Nerka) lake trap was recorded on July 23. Seven of these tattooed fish were observed the following day in Cabin Bay, Lake Nerka, a dis- t~ce of 5 miles from point of release. Thirteen fish tattooed .in marking experiments on red fingerlings migrating out of Little Togiak Lake were subsequently recaptured at the outlet of Aleknagik, 40 miles distant. The minimum length of time before ,"ecapture was 11 days. The shortest lapse of time between release of h~ttooed red fingerlings from a Lake Nerka lake trap and their hrrival at the outlet of Aleknagik, a distance of 34 miles, was 8 days. Tl:i.3Se experiments indicated a maximum rate of travel, measured in a direct water route, of 4-5 miles per day. Although the migration rates should not be taken too seriously because of the small numbers of !'ecoveries and the possible effect of tattooing on normal behavior, it is to be expected that fingerlings already in Alek- nagik before breakup woulcl be first to pass Mosquito Point, and also that over-all timing of seaward migration durir~ a season would be af- fected to some degree by relative numbers in 1nigrant populations orig- inating in different lake areas. Substantiating evidence that Aleknagik smolts are first to migrate seaward was presented by Koo (1962), who showed that later-migrating yearling smolts had more new-season, or ''plus," growth on their scales, and that the amount of fresh-water plus growth found on scales of re- turning adults was directly related to distance from their lake of origin to the outlet of the system at Mosquito Point. Origin of Size Groups in the 1955 Seaward Migration In a search for additional evidence of origin of different smolt size groups in the Wood River lakes, the writer made detailed studies of the 1953 year class of Wood River red salmon. This year class was chosen because yearlings migrating in 1955 exhibited the most striking bimo- dality in size frequency (Fig. 21) and hence offered the best chance of recognition in later life stages available for examination. In the 1955 yearling migration (1953 year class), small yearlings were first to appear at Mosquito Point, and during the first 10 days of fyke netting no other yearling sizes were caught. Although the largest migra- tion of this small size group did not occur until July 10, fairly late in the season, there was no evidence that this group had moved down from upper-lake areas. No small yearlings of the size group in question were encountered in samples from Lake Nerka lake traps or Little Togiak River fyke-net samples in 1955. A large contingent of small 1955 year- lings remained an extra year in the lakes, the survivors ma.ldng up the particularly heavy migration of small age II smolts at the beginning of the 1956 season (Fig. 19). These small age II fish also were not encoun- tered in samples from the upper lakes area~ including Little Togiak Lake, Lake Nerka, and Lake Kulik. Presumably they grew in Lake Aleknagik. 280 Alaska Red Salmon To settle the question oi the origin of small smolts migrating in 1955 as yearlings and in 1956 as age II fish, it was necessary to examine scales from adults of the 1953 year class returning to spawning grounds in 1957 through 1959. H the small smolts did indeed originate only in Lake Aleknagik and· the large smolts in the upper lakes of the Wood River system, the nuGlear area of the scales of adult reds from the 1953 year class taken on the spawning beds in the different lakes should show dif- ferences in amount of fresh-water growth because of the strong tendency of red salmon to return to their original birthplace for spawnifig. Scales used in this study were taken from the fish just above the lateral line and between dorsal and adipose fins, shown by Koo (1962) and Clutter and Whitesel (1956) to be in the area of first :.; .. ;3le formation. The meas- urement chosen for comparison of scales from 1955 yearling smolts with scales from returning adults was the anterior radius from focus to outer edge of the fresh-water annulus. Tfrls measurement was chosen in pref- erence to the radius from focus to edge of fresh-water growth because of the possibility that additional growth appearing after the fish migrated from the lakss may not always be distinguishable from the widening circuli of summer lacustrine growth. Scale-radius measurements were used instead of circulus counts because the former are a more direct index to fish size. For simplicity of procedure the longest radius from focus to edge of the first annulus was used. To facilitate measurement, the images of smolt and adult scales were projected onto a table with the apparatus described by Koo (1962) at a magnification of 230 times. In the analysjs presented here, scale samples were utilized from the first returning contingent of the 1953 year class, i.e., four-year-olds taken on the spawning grounds in 1957 that had Iiligrated seaward as yearlings in 1955 and returned as adults after two winters in the ocean. 2 Koo (1962) found that the fresh-water growth pattern determined from scales of adults returning after twc· winters in the ocean does not differ from that of those returning to the same spawning area after three win- ters in the ocean, if the scales have been collected from a single year class migrating seaward in the same year. Therefore, it was assumed that comparison of the scales of yearling smolts of the two size groups encountered in 1955 with scales of the single age group of adults that re- turned in 1957 would provide reliable data. (Analysis of scales of adults returning in 1958, after three winters in the ocean, has since confirmed this assumption.) The scale photomicrographs in Figure 22 illustrate dif- ferences between the two smolt sizes in scale growth and the similarity to be found in scale pattern of smolts and adults of the same year clasE:i. The 1957 scale samples from adult reds were collected from dead fish 2 Scale formula 1.2 (Koo, 1962) or 42 (Gilbert and Rich, 1927). This is one of the two main age groups in the Nushagak District, the other the five-year-olds, scale formula 1.3, or :>2 • .... Smolts from Wood River Lakes 281 on the various spawning grounds during spawning area studies under the supervision of J. R. Gilbert. In most instances, the marginal portions of adult scales were too resorbed to permit counting the number of marine annuli. Therefore, individuals that had spent but two winters at sea were identified by length measurement. This was possible bet~ause length is influenced by the number of winters the fish remain in the ocean. The smaller adults of both sexes are those fish that have spent two winters in the ocean; the larger fish have spent primarily thrue, occasionally four winters in the ocean (Koo, 1962). Because there is some overlap in length of the two groups, it was necessary to select fish of lengths falling well below the intercept of the two modal length-frequency groups to be reasonably certain that the adults used for scale study had spent only two winters at sea. Males of lengths between 430 and 510 mm and fe- males between 400 and 490 mm, mid-eye to tail fork, were used. Scales of 25 of these fish from each of the spawning ground~ sampled were measured if available. Sexes were combined in treatment since no ap- preciable difference in amount of lacustrine scale growth has been found between sexes of red salmon (Clutter and Whitesel, 1956). Several samples, collected at intervals throughout the main smolt migration period, were utilized for measurement of 1955 yearling smolt scales. Scale measurements were first made without reference to length of the fish, and were later classified by fish length. Scales from fish of lengths between 68 and 77 mm, inclusive, were used to represent the mode of small yearling smolts, and scales from fish of lengths between 83 and 102 mm, inclusive, to represent the mode of large 1955 yearlings (Fig. 23). Radius-measurement frequencies of 1955 yearling smolt scales, grouped to 5-millimeter intervals, are presented in Figure 24. Superimposed on the frequency of small-yearling scale measurements are corresponding frequencies taken from returning Aleknagik spawners, and on the fre- quency of large-yearling scale measurements, corresponding ft~equencies from returnees to the other Wood River lakes. The close correspondence in amount of scale growth to the first annulus between small smolts and returning spawners to Aleknagik and between large smolts and upper- lakes spawners establishes the fact that the small smolts migrating in 1955 did inhabit Lake Aleknagik during their fresh-water growth. Radius measurements of adult scales from focus to edge of the fresh- water armulus are presented classified by spawning ground in Figure 25. Locations of spawning areas repre3ented are shown in Figure 26. As will be discussed later, few scales of small adults from Lake Aleknagik were available so these are grouped 1:ogether in the graph. None of the adults sampled from Agulowak River, tributary to Lake Aleknagik, fell in the small size category, the 1957 escapement there being almost en- tirely of five-year-old Lsh with three years of ocean residence. The greatest number of scale samples were collected from Lake Nerka 282 Alaska Red Salmon where the majority of salmon spawned in 1957. Scale measurement data for samples from this lake were examined to determine whether there was any relation between type or location of spawning area and amount of growth achieved during the first year. An F test was first applied to test homogeneity of the set of scale measurement means of the samples from the Lake Nerka spawning areas (Table 17). A very large F ratio, significant at the 1 per cent level, was obtained; hence the hypothesis of homogeneity was rejected, indicating that one or more of the differ- ences between means was significant. TABLE 17. MEASUREMENT* DATA FROM SCALES OF ADULT RED SALMON, SCALE FORMULA 1.2, LAKE NERKA, 1957, SHOWING ANALYSIS OF VARIANCE TEST Spawning Area lti :LXii X;. I; x;j River Bay 25 2,529 101.2 258,649 Fenno Creek 25 2,608 104.3 275,414 Allah Beach 25 2,568 102.7 265,388 Stovall Creek 25 2,725 109.0 298,539 N-4 Beach 25 2,456 98.2 244,430 Pick Creek 25 2,624 105.0 277,088 Anvil Bay 25 2,348 93.9 223,334 Kema Creek 25 2,297 91.9 214,281 Agulukpak River 25 2,234 89.4 201,204 TOTAL 225 22,389 2,258,327 *Measurement: focus to outer edge of first annulus in mm; magnifica- tion 230x. Analysis of Variance of Scale Measurements Source of Degrees of Sum of Mean F Variation Freedom Squares Square Between areas 8 21,722 2,715.25 69.23 Within areas 216 8,573 39.22 TOTAL 224 30,295 - F.99 :::::: 2.60 In order to reach a decision as to which of the differences among the means could be considered significant and which not, a multiple-range test proposed by Duncan (1955) was applied. Values from his table of "special significant studentized ranges" were multiplied by the standard error to obtain a series of "shortest significant ranges" between means. The means are ranked in order, and the difference between any two means is considered significant if it exceeds the corresponding shortest significant range for the number of means involved in the range under Smolts from Wood R:iver Lakes 283 consideration. Results of the multiple range test are presented in Table 18. TABLE 18. St.Jm!ARY SHEE'r OF MULTIPLE RANGE TEST FOR SIGNIFICANCE OF DIFFERENCES llETWEEN SAMPLE MEANS, SCAL:>: RADIUS TO FIRST ANNULUS, OF RED SALMON IN LAKE NERKA SPAWNING AREAS* Standard error of mean, s~ = 1.253 Number of means in range 2 3 4 5 6 7 8 9 Shortest significant range, 5 per cent level 3.17 3.66 3.78 3.87 3.95 4.00 4.05 4.08 U~per Lake Nerka Lower Lake Nerka Beacn Spawning Creek Spawning Area nwnbers 11 10 9 6 2 4 3 7 5 Localities Agulukpak Kema Anvil N-4 River Allah Fenno Pick Stovall River Creek llay Beach Bay Beach Creek Creek Creek Heanst 89.4 91.9 93.9 98.2 101.2 102.7 104,3 105.0 109.0 *Measurement: focus to first annulus x 230. After Duncan, 1955, Table 4. tAny two means not und~rscored by the same solid line are significantly different, Any two means underscored by the same ~olid line are not significantly different. The underscorings in Table 18 indicate which means are and which are not significantly different. All six means of Lower La!ce Nerka samples were significantly larger than the means of samples from the three spawning populations in Upper Lake Nerka. This suggests that young in the two sections of the lake remained somewhat segregated during their first year, and that the Lower Lake Nerka feeding area provided better growing conditions. Among the six spawning groups of Lower Lake Nerka, the three beach-spawning groups showed less growth during the first year than did the three stream groups, but there was no clear demarca- tion between samples from the two spawning-area types: lake-beach and stream. For this year class, it is evident that rearing-area conditions exerted more influence on growth than did conditions for embryonic development. In a comparison between lakes (Fig. 27), samples from Lakes Beverley and Kulik suggest that growth conditions for young during their first year were similar to those in Lake Nerka. The contrast was very striking, however, between Lake Aleknagik and the other Wood River lakes. The scale growth of Aleknagik adults was markedly less for this period of early life. Causes of Difference Between Lakes in Size of Smozts Produced Climate Yearling smolts migrating in June or July have achieved most of their 284 Alaska Red Salmon growth during the previous summer. For the 1955 smolts, the pr~yious growing season, 1954, was characterized by early breakup of the lake ice and warm lake temperatures. New summer growth appeared early on the scales of yearlings migrating in 1954 (Table 14). As shown in a previous section, differenceR between years in climate resulted in dif- ferences in growth rate of you..1g salmon. However, within a single year the entire lake system is subjected to much the same climatic conditions, hence poor grovv-th of fry in one lake area, such as Aleknagik in 1954, would not be expected unless it was caused by conditions other than eli- mate. Differences between lakes in level of key nutrients or in degrees of competition for food are suggested. Level of Basic Nutrients It has been proposed that in red-salmon lakes, levels of nitrogen, phosphorus, or other biogenic elements may be limiting factors in pro- duction of phytoplankton, hence indirectly of zooplankton and young salmon (Barnaby, 1944; Nelson and Edmondson, 1955; Nelson, 1958; Goldman, 1958; Krokhin, 1957; Alaska Department of Fish and Game, 1959). No comparative information on levels of critical nutrients has been collected for the Wood River chain. The lakes are basically quite similar in mor- phometric features, but Lake Alel:Jlllgik as the lowest lake in the series may have the greatest amount of dissolved nutrients flowing through. It is slightly more turbid in summer and there is no reason to suspect a lower level of basic nutrients. Competition for Food Chamberlain (1907) was first to suggest that differences between years in population density of young sockeye during lake residence might be the cause of observed annual differences in yearling smolt sizes. Foer- stelr {1944) demonstrated the effect of population density on size of mi- grating smolts. He correlated weights of Cultus Lake yearlings in the migratio11 years 1927 through 1935 with lake population densities based on Pumbers and ages of seaward migrants. His conclusions were sup- ported by the inverse correlation found by Ricker (1937) between zoo- plankton abundance and population size of sockeye salmon in Cultus Lake, !"egl.!ctim~ in plankton being assumed to be the result of grazing by young sockeye. Rounsefell (1958) used additional Cultus Lake data from more recent ye:a.rs to correlate biomass of smolts versus the logarithm of the rnJmber pf smolts produced. He concluded that "the closeness of the semilogarithmic fit suggests-that the increasing competition between smolt~} as their numbers increase sets an asymptotic level on the bio- mass." Analyzing Karluk Lake data, he interpreted a linear relation betwP.en numbers and biomass of the annual smolt migrations as indi- cating no intraspecific competition for food in the years studied. Krogilis and Krokhin (1956) stated that in Lake Dalnee, Kamchatka, the correla- Smolts from Wood River Lakes 285 tion was inverse between size and weight of red smolts and the number of plankton consumers both intra-and interspecific. No relation between pop1lation density and size of migrants was found at Lakelse Lake for the seven brood years beginrJ.ing in 1946 (Canada, Fisheries Research Board, 1957). Studies at Babine Lake were first to show that unequal distribution of sockeye in a single lake can result in overcrowding and stunting (Johnson, 1956, 1958). Lake areas adjacent to main spawning concentrations were found by tow-net sampling to be more densely populated by "underyear- ling" young. Marked differences between years in size of young sockeye in these areas were attributed to differences in spawning density and consequent density of young. Abundance of zooplankton was re!ported to be reduced in areas where density of young salmon was high (Canada, Fisheries Research Board, 1958). With regard to intraspecific competition in the Wood River lakes, there was evidence that changes between lakes in relative population £i.ze of young red salmon were correlated with observed changes in length fre- quency of smolts. In various areas of the Wood River lakes, the only available measure of population size of young for the years concerned was magnitude of parent escapement. For this measure to be reliable, it is necessary that in a given year class, mortality rates. of young reds be reasonably paralleled between population groups defined by major divisions of the lake system. It would also be desirable, but not as es- sential, that difference in mortality rates between year classes during their first year not be large. Since the relation between magnit1·des of escapement and seaward migration has been shown to be erratic, definite differences in mortality rate must exist between year classes during fresh-water life. However, changes among lakes in relative magnitude of spawning populations have been large, and it is pertinent to examine the data on parent escapement and smolt size for indications of a rela- tion between parent population size and growth attained by young salmon. The red fry population (age 0) in Lake Aleknagik is assumed to consist only of fish originating within the lake and its tributaries, including the Agulowak River connecting Lake Aleknagik with the upper Wood River lakes. Emerging fry in the Agulowak River spawning area are prevented by the swift current from migrating upstream into Lake Nerka, and con- sequently must descend into Lake Aleknagik to feed. No appreciable downstream fry movements between the major lakes has been found, so that fry migration from Lake Nerka to Lake Aleknagik is assumed to be negligible. Two major area groupings will be used for the upper lakes~ (1) the Lake Nerka-Little Togiak Lake area, including the Agulukpak River between Lake Nerka and Lake Beverley, and (2) the Lake Beverle)'-Lake Kulik area. Data on density of spawning in the separate lakes of the Wood River system have been compiled by J. R. Gilbert Irom Fisheries 286 Alaska Red Salmon Research Institute records of aerial and ground surveys conducted an- nually since 1946.3 Estimates are presented in Table 1 of Report of Op- erations, 1958 (Fisheries Research Institute, 1959). This information from 1946 through 195'! is expressed in Table 19 in terms of number of parent spawners per surface square mile of lake rearing area for young in the three lake regions. The parent years 1952-57, for which smolt length frequencies are presented, are indicated below the broken line in Figure 21. TABLE 19. ESTIMATED ESCAPEMENT OF RED SALMON PER SURFACE SQUARE MILE OF LAKE REARING AREA, 1946-57 Escapement Lake Aleknagik Lake Nerka-Lake Beverley- Year (34 Sqnare Hiles) Little Togiak Lake Lake Kulik ~2 Square Miles) (62 Square Miles) 1946 10,600 25,000 21,100 1947 5,500 10,800 11,400 1948 4,600 11,800 5,900 1949 1,000 700 200 1950 3,800 2,700 1,600 1951 2,700 2,400 2,800 -------------------r--------------------------- 1952 2,000 1,600 500 1953* 7,400 2,900 500 1954 3,900 3,000 3:100 1955* 14,500 6,300 6,000 1956* 7,400 5,300 1,400 1957 2,700 I 2,400 200 - *Escapements producing bimodality in yearling smolt length fre- quencies. The heaviest spawning density shown in Table 19 for the years 1952- 57 occurred in the Aleknagik-Agulovtak rearing area in 1953, 1955, and 1956. These were parent years of the yearling smolt groups that ex- hibited the most pronounced bimodality in length frequency. The small yearlings were offspring of heavy spawning populations in Lake Alek- nagik and tributaries. The greatest contrast between growth of Aleknagik smolts and those from. upper lakes occurred in the 1953 year class (Fig. 21) when the escapement reached the highest percentage of the Wood Hiver lakes total. Although spawning was particularly heavy in all three lake areas in 3 Each seaso.1' s escapement estiniates were based on aerial estimates and ground coWlts taken at or near peak of spawning. Data most directly comparable were used to determine differences between lakes and be- tween years. Estimates since 1953 are considered more accurate, since trunk stream enumeration of the runs by use of towers in Wood River was begun in that year and provided a more accurate annual estimate of total escapement to the system. Smolts from Wood River Lakes 287 1955 alld the over-all size of smolts in the 1957 migration was reduced, distinct bimodality of yearling smolt sizes in 1957 was still present be- cause of the much greater parent spawning density in Lake Aleknagik. Thus, the smallest Aleknagik smolts observed in any year were pro- duced by the largest escapement, that of 1955. The large escapements of the same magnitude in 1953 and 1956 produced stunted smolts similar in average size. The parent populations of the other recent year classes in the Aleknagik-Agulowa,k area had been at a lower level, and there was not as much contrast in size between the small, early migrants, when evident, and the large migrants. The consistency of these results is strong evidence that the capacity of Lake Aleknagik nursery area was taxed by the progeny of large spawning populations. It will be shown in another paper that stunting occurs in other lakes of the system under similar parent spawning densities. There is danger in attempting to assign marked differences in smolt size simply to population pressures of fry alone. If intraspecific com- petition caused differences in growth, older age groups may well have played an important part. The largest migration of smolts measured by the fyke-net index method occurred in 1954 and 1956 (see Table 9). If early season feeding of these fish prior to migration seriously affected the lake\s food supplies, the lower lake in the system would presumably be most affected, since all migrants must either originate in or pass through it. The fry in this area migrating as yearlings the following years, 1955 and 1957, would thus be affected. Also yearlings remaining an additional year in L.'lese lake C'..reas and migrating at age n would com- pete to some extent with the fry for food. On this latter point, however, there is no consistent relation between yearling holdover and fry stunt- ing since the yearling holdover in 1954 was very slight, that in 1956 moderate, that in 1957 heavy. The competition between year classes may not be as heavy as within year classes because of differences in food and feeding area, particularly early in the summer when fry utilize the lit- toral areas more heavily. There is additional evidence supporting the hypothesis that smolt size differences are the result of intraspecific competition within the year group. This evidence is provided by a comparison between areas in amount of fresh-water growth associated with relative magnitude of parent spawning populations. Scale measurements presented of four- year-old adult salmon, 1953 year class, have indicated that growth during the initial year achieved by salmon originating in other 'Wood River lakes was much superior to that obtained in Lake Aleknagik (Fig. 25). This, however, is exactly the reverse of the relation found by Koo {1962) for adults of the 1948 year class returrJng to spawn in 1952 and 1953, and by the writer for scale measurements made of adults of the 1946 year class returning to spav,'n in 1950. The first year's growth in those two year classes was better in Aleknagik than in the Beverley-Kulik area. Fol- 288 AJ.aska Red Salmon lowing the hypothesis of population pressure, an explanation of this re- versal in rates of early growth lies in relative population sizes of young salmon present in the two areas during the years in question. Referring again to Table 19, the spawners were relatively much more numerous in the Beverley-Kulik area than in the AleknagLlt-Agulowak area in the years 1946 through 1948. In the years to follow, the Aleknagik-Agulowak area population increased greatly in relative imp:>rtance, overshadowing that of the Beverley-Kulik area in eight out of nine years, particularly in the years 1953, 1955, and 1956, which produced the bimodality of smolt sizes seen in 1955, 1957, and 1958, respectively. The relation discussed above between magnitude of adult populations and growth of smolts produced is surprisingly consistent. In view of the very large changes in relative magnitude of adult spawning popula- tions between the areas under discussion, it appears that spawning popu- lation estimates were a sufficiently sensitive index of relative size of fry p:>pulations produced within the major divisions of the lake system; and, therefore, that intraspecific competition of the young does play a determining role in growth achieved during the first year of lake life. In view of other possibilities, yet unmeasured, the relation found be- tween initial populations and smolt size in Wood River lakes must be further tested. The very real possibility of interspecific competition should not be overlooked, for a prominent competitor, rivaling the young reds in abundance, exists in the threespine stickleback fow1d in the Wood River lakes. This species is closely associated with young red salmon in the lakes and was found by the writer to feed on the same organisms. It will be shown in a later report that r1eduction in growth of sticklebacks also occurs when p:>pulation density of young salmon is high. Increase or decrease in ge.1erallevel of abundance of sticklebacks or annual fluc- tuations in their abundance would be expected to influence growth of young salmon if competition for food is indeed responsible for a major share of the size fluctuations observed. Relation of Growth Rate to Survival Fresh-water Survival Young red salmon that grow more slowly are probably subject to greater mortality from a number of sources. Ricker and Foerster (1948) have suggested that lacustrine predation is gnatest while the sockeye are small, and that if this vulnerable stage is, passed through rapidly the chances for survival are increased. It is likely that slower-growing young salmon may also suffer more damage from parasites. In Bristol Bay, the cestode, Triaenopho·rus crassus, was found by the writer to be. a common parasite of red salmon (Lawler and Scott, 1954). The plerocercoid stage of this cestode has been found on occasion to be detrimental to other species of fish, and may ..... Smolts from Wood River Eakes 289 affect young red salmon. At the fish culture station at Kalerna, s~ eden, plerocercoids of T. crassus were reported to ha,ve caused a widespread mortality of young rainbow trout, brook trout, and Atlantic salnon in the years 1921 through 1923 (Bergman, 1924; Scheuring, 1930). Ji'liller (1945) reported the cestode to havl~ affected growth and condition of com- mon whitefish (Coregonus clupeaformis) and tullibee (Leucichth: 1s sp .) at Lesser Slave Lake. In young red salmon the plerocercoids of T. eras sus destroy c 1nsid- erable tissue during' development and encystment in the museu .ature. The plerocercoids first appear in the flesh of red fry in late su .nmer. Weakening of young salmon is likely to be severe or even letha i if the fish is very small, since a high proportion of the musculatt1.re is then destroyed by a single plerocercoid. While direct proof is lacking that the cestode is a serious mortality factor in young salmon, the p uasite is prevalent enough to be wo:rthy of attention. Average annual in· !idence of parasitism by T. eras sus in red smolt samples collected at M Jsquito Point "'aS found to be 66.0 per cent for the years 1948 through 195S (Table 20). TABLE 20. PERCENTAUES OF WOOD RIVER RED SALMON SMOLTS PARASITIZED BY THE PLEROCKRCOID S rAGE OF THE CESTODE TRIAENOPHORUS CRASSUS IN THE YEARS 1948 THROUGH 1958* , , j Average Per Cent Range Total Number Number of I Year C1f .Fish Samples ...... ·;~·········· Samples E:~:aminec Examined All Samples Para 1948 100 2 48~ 52 88 8 1949 504 8 50-112 81 7 1950 350 7 so . 76 e 1951 423 8 I so-73 82 e 1952 31::0 7 50 82 7 1953 393 8 32-78 78 6 1954 445 7 50-110 63 5 1955 810 7 99-152 63 3 1956 400 4 100 74 6 1957 263 3 76-104 23 2. 1958 276 ., 74-111 16 . ~ -- M~an annual percentage = 6J betwe-l!n n Per Cent itized ·-90 1•92 1-94 1-96 t-:30 1~90 -76 -80 -86 -25 -22 ---------------- *Only samples taken between dates of lake breakup and July 31 are included, Delay in seaward migration is another probable cause of mort a.lity in slower-growing ycung salmon. The tendency for smaller fish to remain an additional year or more in the lake before migrating seaw~ ~rd has been well established (Foerster, 1937; Barnaby, 1944; Koo, 1962; l(rogius and Krokhin, 1956a; Krogius, 1957). The additional mortality s .tffered by those fish residing an extra year .in the lakes is not apt to bE offset by increased ocean survival gained through their greater size at S•!award migration. This view is shared by Krogius and Krokhin (1956c: ), who emphasized the need to understand causes of difference in age at down- stream migration in order to develop means of management to •educe length of residence in fresh water. In summary, slower growth of young in fresh water is believed to result in greater over-all mortality during fresh-water stages .. \.ctual 290 Alaska Red Salmon proof that such is the case is lacking for red salmon of Wood .l.tive:r lakes. Marine Survival It i..,:; generally accepted that marine survival rates vary dii'ectly with sh!:e of anadromous salmonoids at time of seaward migration. In fin- clipping experiments on Karluk Lake red salmon smelts; higher marine survival values were obtained in four of five years for the older (and larger) of the two major cmolt age groups (Barnaby, 1944). However, it must be noted that siza alone at seaward migration may not have been responsible for these results. A close exam:ination uf the Karluk data indicates that differential S11rvival m the ocean was probably influenced by at least three distinct factors oi unknown re.lrative importance: (1) size and age at seaward migration, older smolts being larger in size; (2) timing of entry into .I!'.arine environment, older smolts tending to migrate earlier in the season; and (3) length of stay in the ocean, older smolts having a greater tendency to remain only two years at sea, but at the same time ter.rHng to return later in t:hot=.! season in the year of return. Strong evidence that size of smolts at time of seaward migration is actually an important qualitative factor in survival during seaTVard mi- gration and ocean residence was presented by Foerster (1954) for Cultus Lake sockeye. Data were presented gi ring numbers and mean size of smolts in the 1927 through 1944 seaward migrations, and the number of adults in return spawning escapements t0 the lake. Through analysis of these data by multiple-correlation treatment, Foerster concluded that the negative correlation between magnj:ude of smolt migration and percentage return of adult spawners to the lake. was related principally to size nf smolts at time of seaward migration. Krogius ~nd Krokhin (1956a) did not find a relation between size and marine survival of Lake Dalnee sockeye smoits, but considered that this lack of coincidence with Foerster's findin;;ri;; may be explained by the larger size of Dalnee smolts and less size varlaoility between years than at Cultus Lake. Distinct bimodality of sizes at Wood; iver in the 1955 vearling mi- gration presnnted a unique opportunity to study the effeci: o~· a jifference in smolt size on marine survival and length of stay in the t ...... ean, for in this instance ~he size difference existed within a single year class en- tering the nl:iu:lne environment in the same season. Peretmtages of small and large migrants found in the 1955 seaward migration were compared wlth percentages of returning adults in catch and escapement possessing the fresh-water scale pattern characteristic of small and of la.1'ge sea- ward migrants. Small smolts, originating in the Aleknagik.-Agulowak area, were calculated to have totaled 29 per cent of the 1955 yearling seaward migration, and were produced by a parent escapement calculated at approximately 49 ,r,ar cent of the Wood River lakes total. Adult returns in catch and escapement in 1957 and 1958 provided evidence that stunting Smolts from Wood River Lakes 291 of young reds in fresh water had affected both marine survival and age at return (Table 21 ). TABLE 21. RATIOS OF SMALL A.~D LARGE YEARLINGS IN 1955 SEAWARD MIGRATION A..~D IN RETURN AB ADULTS IN 1957 A.lffi 1958 Small Large Yearlings Yearlings 1955 seaward migration 29 71 1957 adult return 7 93 1958 adult retm"n 18 82 Tbtal adult return 15 85 The ratio of small to large yearlings in the 1957 adult return was cal- culated to be 7:93 and in the 1958 adult return, 18:82, for a total weighted ratio of 15:85. This indicates that survival of small Aleknagik yearlings was about half that of large yearlings from the remainder of the Wood River system, and that there was a definite tendency for smaller smolts to rc,main an extra year at sea. It should be noted that the computations involve determinations of smolt age composition, spawning population estimates, age analysis of catch al!ld escapement, and an assumption as to percentage of Wood River fish in the entire Nushagak red salmon cat\!h. Return ratios may also have been affected by differential fishing mortality from the Japanese high-seas fishery. Selection by the Bristol Bay fishery may conceivably have had an effect, although tagging conducted by Bureau of Commercial Fishf'ries personnel has in·li.cated that tirr.t.ing of the Aleknagik run through the fishery does not differ hom the remainder of the Wood River race.s. There was also no indicathm from length frequencies that Aleknagik adult fish differed in size from those of the same age group in there- mainder of the system. Although confidence intervals of the ratios pre- sented cannot be determined, the trend of results is strongly suggestive. Further, the 1959 and 1960 :.:·eturns of adult red salmon from the 1S57 seaward migration again indicated a lower survival of sh,nted smolts from Lake Aleknagik. Effect of Smolt Size on Adult Popu:i.ation Size The above discussion touches on possible effects of changes in la- c.ustrine growth conditions on fresh-water and marine survival of red salmon. The inverse '!'elation observed between population size and smolt size in subdivisions of tr1e Wood River lakes system may explain the mechanism of some fluctuations in population level that have occurred. When spawning· populations produce numbers of young beyond the capacity of a lake rearing area, stunting of yc,ung, low survival, and delay in return from the o~ean may follow. With low survival, the return runs d:rop in numbers and fry production may decrease, again permitting 292 Alaska Red Salmon good growth. Individual rearing areas may thus tend to fluctuate in popu- lation level somewh&t 1Gdependently of other rearing areas in the lake system. This sequen~e of events appears to have held in recent years in the Wood River la...l{es. SUMMARY 1. The early life history studies discussed in this paper are an integral part of the Fisheries Research Institute program on red salmon runs of the Nushagak District in Bristol Bay, Alaska. The research described deals with smolts in ·wood River seaward migrations of 1951 through 1959. 2. A smolt enumeration system adaptable to large rivers was estab- lished in 1951 at the outlet of Lake Aleknagik. A winged fyke net was fished throughout the migration seasons: 1951 through 1958, its catch furnishing the annual migration index. A comparison of catches made by two nets fished simultaneously during the same period as well as indirect evidence attest to reliability of the index method. 3. Smolts used for length frequency study were measured alive under anesthesia. Those fish preserved in 10 par cent formalin undergo shrink- age in length. Shrinkage is greater in percentage for smaller fish. A correction of +5 mm in length was determined necessary to convert lengths of pr9served yearling smol1:s to correspond to measurements of live fish. 4. Size selectivity of the index fyke net was tested by comparison of smolt samples taken simultaneously at Mosquito Point by beach seine and fyke net. Heterogeneity between schools in size composition was made evident by the tests. There was no indication that larger yearling smolts are less likely to be caught by tr.e fyke net than smaller fish. 5. A pattern of decrease in size of yearling smolts during evening migration was established. Net selectivity was ruled out as a causa. It is proposed that the phenomenon of diurnal change in size was due to difference in swimming speed or migration .stimulus related to fish size, and that schools of larger fish tended to reach the outlet earlier in the evening from the daytime milling area. 6. Agr. determination of smolts was based on a combination of length- frequenc,:ll' a.11d scale study. In determining age it was necessary to pre- serv,e for scale study only those fish of lengths which fell in the range of overlap between age I and age II fish. 7, No definite relationship was found between magnitudes of parent escapement of red salmon :.n Wood River lakes and seaward migrants produced for the year classes 1S49 through 1955 .. Rates of survival of young in fresh water varied as mu~h as 19 times. 8. Climate was determined to be a dominant factor in over-a~l1 timing Smolts from Wood River Lakes 293 of seasonal migration. Timing of early season migration was closely associated with breakup of lake ice, and termination of migration with rise of lake surface temperatures above 50° F. 9. Seasonal increase in new ~ummer growth on scales of yearling smolts was demonstrated, and differences between years in amount of new growth attained by a given calendar date during the season was shown to be related to time of breakup and subsequent lake tempera- tures. 10. The pattern of seasonal changes in smolt length frequency at Wood River, 1954 through 1957, suggests that L~e first smolts to migrate were from the lowest lake in the Wood River chain, and that in certain years these fish were smaller in size at time of migration than were those from lakes above. Delay in migration of smolts from the upper lakes is attributed to delay in ice oreakup and distance between lake of origin and Wood River. 11. Length-frequency and age studies reveal a distinct bimodality in size of yearling smolts in the 1955, 1957, and 1958 seaward migra- tions. Smaller yearlings were first to begi1.1 migration. The origin of size groups,was studied by comparison of scale measurements of 1955 year- ling smolts with measurements of fresh -water scale grmvth in adults of the same year class returning to the spawning grounds in 1957 and 1958. The small smolts were found to h.ave originated in Alek.rmgik, confirming the hypothesis based on the seasonal smolt size sequence observed at Mosquito Point. 12. In Lake Nerka, scale measurements of adult fish of the 1953 year class established the existence of significant differences in fresh -water growth between fish spawning in upper and lower arms of the lake. This indicated limited circulation of young and difference between lake areas in conditions for growth. Significant differences in fresh-water growth were a.1so found between adult samples from different spawnL11g grounds in Lower Lake Nerka. 13. Small size attained by Lake Aleknagik smolts migrating in 1955, 1957, and 1958 was related to heavy density of parent spawning popula- tions in the Lake Aleknagik-Agulowak River area. Differences between lakes in degree of competition between fry for food were apparently a major source of differences in growth attained. The possibility of com- petition between age classes as well as interspecific competition with threespine sticklebacks is discussed. 14. Change in smolt size, hence in fresh-water and marine sur·. ~val rate, is suggested as one cause of fluctuations in relative spawning pop- ulation levels among the rearing area divisions in the Wood River lakes. The low returns of adults in 1957 and 1958 from the small-size group of smolts migrating sBaward in 1955 .;:,1~strate the effect of smoh: size on marine survival and age at return. 294 Alaska Red Salmon ACKNOWLEDGMENT It is a pleasure to express my sincere appreciation toW. F. Thomp- son, who, as the first director 7 organized and developed the Fisheries Research lnstltute. It was he who foresaw the need for the early life his- tory studies as a part of the basic program of research on red salmon in Bristol Bay. His constant encouragement, inspiration, and helpful advice are gratefully acknowledged. During the course of the studies I received valuable cooperation and assistance from other memberto:. oi the staff working on closely integrated problems. These included {)!".v:i.tcularly, Ted S. Y. Koo, Ole A. Mathisen, and John R. Gilbert. Wilb1..l~ A. Church gave competent assistance in the conduct of the sampling pr•Jg;ram at Lake Nerka. I also wish to thank Miss Pea;rl Mooney, Institute librarian, for her assistance in preparation C!f my original thesis, and William F. Royce for his review and criticism of the revised manuscript. This study was financed by the Bristol Bay salmon canning industry. LITERATURE CITED Alaska Department of Fish and Game. 1959. Annual Report for 1957. Juneau. 124 p. Babcock, J. P. 1904. Fisheries Commissioner's Report for 1903, British Columbia Dept. of Fisheries, Victoria. 15 p. -----. 1905. Fisheries Commissioner's Report for 1904. BriUsh Co- lumbia Dept. of Fisheries, Victoria. 12 p. Barnaby, J. T. 1944. Fluctuations in abundance of red salmon, Onco- rlz;,nchus nerka (Walbaum), of the Karluk River, Alaska. U.S. Fish and Wildl. Serv., Fishery Bull., 50(39):237-95. Bergman, A. M. 1924. Larven von Triaenoplzor.ts robustus Olsson und einer Dibothriocephalusart als Ursache eines Massensterbenr.; unter jungen Lachsfischen. Deutsche tiedirztliche Wochschr., 32(21):292-~a; 32(22):311-16. Brett, J. R., and J. A. McConnel. 1950. Lakelse Lake sockeye survival. J. Fish. Research Board Canada, 8(2):103-10. Canada, Fisheries Research Board. 1957. Annual Report for 1955. Ot·· tawa. 182 p. -----. 1957a. Annual Report for 1956/1957. Ottawa. 195 p. -----. 1958. Annual Report for 1957/1958. ottawa. 195 p. Chamberlain, F. M. 1907. Some observations on sa1mon and trout in Alaska, 112 p, In Rept. U.S. Commissioner of Fisheries for 1906. Clutter, R. I., and L. E. Whitesel. 1956. Collection and interpretation of sockeye salmon scales. Int. Pacific Salmon Fisheries Comm., Bull. 9, 159 p. New Westminster, B. C. Smolts from Wood River Lakes 295 Dombroski, E. 1952. Sockeye smolts from Babi!te Lal~e· in 1951. Fish- cries Research Board of Canada, Pacific Biol. Sta. Progr. Rept., (91):21-26. -----. 1954. The sizes of Babine Lake sockeye salmon smolt emi- grants, 1950-1953. Fisheries Research Board of Canada, Pacific Biol. Sta. Progr. Rept., (99):30-34. Duncan1 D. B. 1955. Multiple range and multiple F tests. Biom~trics, 11(1):1-42. Dunlop, H. A. 1924. The growth-rate of the scales in the sockeye salmon, Oncorhynchus n.erka. Contrib. Canaclian BioL, 2(10):151-59. Fisheries Research Institute .. 1959. Report of Operations, 1958. Univ. Washington. 24 p. Foerster, R. E. 1937. The relation of temperature to the seaward mi- gration of young sockeye salmon (0nc(;rhynchus nerka). J. Bioi. Board Canada, 3(5):421-38. ------. 1938. An investigation of the relative efficiencies of natural and artificial propagation of sockeye salmon (Oncorhynchus nerka) at Cultus Lake, British Columbia. J. Fish. Research Board Canada, 4(3):151-61. -----. 1944. The relation of lake population density to size of young sockeye salmon (Oncorhynchus nerka). J. Fish. Research Board Can- ada, 6(3):267-80. -----. 1952. The seaward-migrating sockeye and coho oalmon from Lakelse Lake, 1952. Fisheries Research Board of Canada, Pacific Bioi. Sta. Progr. Rept., (93):30-32. -----. 1954. On the relation of adult sockeye salmon (Oncm~hynchus nerka) returns to known smolt seaward migrations. J. Fish. Research Boat'd Canada, 11(4):339-50. -----. 1955. The Pacific salmon (genus Oncorhynchus) of the Canadian Pacific coast, with particular reference to their occurrence in or near fresh water. Int. North Pacific Fisheries Comm. Bull., (1): 1-56. Gilbert, C. H. 1914. Contributions to the life-history of the sockeye salmon. (:No. 1) p. 53-57 . .Jn Fisheries Commissioner's Report for 1913. British Columbia Dept. Fisheries. Victoria. -----. 1915. Contributiong to the life-history of the sockeye salmon. (No. ?.) p. 45-75. In Fisheries Commissioner's Report for 1914. Brit- ish Columbia Dept. Fisheries. Victoria. -----. 1916. Contributions to the life-history of the sockeye s~lmon. (No. 3) p. 27-64. Jn Fisheries Commissioner's Report for 1915. Brit- ish Columbia Dept. Fisheries. Victoria. -----. 1918. Contributions to the life-history of the sockeye salmon. (No.4) p. 33-80. In Fisheries Commissioner1 s Report for 1917. Brit- ish Columbia Dept. Fisheries. Victoria. Gilbert, C. H., and W. H. Rich. 1927. Investigations concerning the red Alaska Red Salmon salmon runs tn the Karluk River, Alaska. Bull. U.S. Bur,. Fisheriess 43(Pl. 2):1-69. Goldman, C. R. 1960. Primary productivity and limiting factors in three lakes of the Alaska Peninsula. Ecol. Monographs, 30:207-30. Hile, R. 0. 1936. Age and growth of the cisco Leu.cichthys artedi (Le Sueur), in the lakes of the northeastern highlands, Wisconsin. Bull. U.S. Bur. Fisheries, 48(19):211-317. Holmes, H. B. 1934. Natural propagation of salmon in Alaska. Fifth Pa- cific Science Congress, Victoria and Vancouver, B. C., Proc. for 1933, 5:3585-92. Johnson, W. E. 1956. On the distribution of young sockeye salmon (On- corhynchus nerka) in Babine and Nilkitkwa Lakes, B. C. J. Fish. Re- search Board Canada, 13(5):695-708. -----. 1958. Density and distribution of young sockeye saimon (On- corh:ynchus nerka) throughout a multibasin lake system. J. Fish. Re- search Board Canada1 15(5):961-82. Koo, T. S. Y. 1959. Red salmon smolt enumeration in the Wood River system. Report of field operations and results, 1958. Univ. Wash- ington, Fisheries Research Inst. Circ. 103. 11 p. -----. 1960. Abundance, size and age of red salmon smolts from the Wood River system, 1959. Univ. Washington Fisheries Research Inst. Circ. 118. 9 p. ----. 1962. Age and growth studies of red salmon scales by graphical means. See elsewhere in this volume. Krogius, F. V. 1951. 0 dinamike chislennosti krasnoi. [Oncorhynchus nerka (Walbaum)]. !zvestiia Tikhookeanskovo Nauchno-issledovatel- skovo Inst. Rybnovo Khoziaistva i Okeana~., 35:1-16. [On the dynamics of abundance of the sockeye salmon Oncorhynchus nerka (Walbaum) .] Fisheries Research Board of Canada, Transl. Ser., No. 101, 1957. ottawa. Krogius, F. V., and Eo M. Krokhin. 1948. Ob urozhainosti molodi kras- noi (Oncorhynchus nerka Walb. ). Izvestiia Tikhookeanskovo Nauchno- issledovatelsiwvo Inst. Rybnovo Khoziaistva i Okeanog. 28:3-27. [On the production of young sockeye salmon Oncarhym:h~t.s nerka (Wan· .:.urn).] Fisheries Research Board of Canada, Transl. Ser., No. 109, 1958. Ottawa. -----. 1956. Rezultaty issledovanii biologii nerki-krasnoi, sostoianiia ee saposov i kolebanii chislennosti v vodakh Kamchatki. Voprosy Ikhtio- logii. (7):3-20. [Results ~fa study of the biology of sockeye salmon, the conditions of the stocks and the fluctuations in numbers in Kam- chatka waters.] Fisheries Research Board of Canada, Transl. Ser., No. 176, 1958. Ottawa. -----. 1956a. Priehiny kolebanii chislennosti krasnoi na Kamchatke. Trudy Problemnekh i Tematicheskikh Sovesbchanii ZIN, (6): 144-49. [Causes of the fluctuations in abundance of sockeye salmon in Kam~ Smolts from Wood River Lakes 297 chatka.] Fisheries Research Board of Canada, Trans!. Ser., No. 92, 1959. Ottawa. Kroki1in, E. M. 1957. Istochiki obogashcheniya nerestovykh ozer bio- gennymi elementami. Izvestiia Tikhookeanskovo Nauchno-issledova- telskovo Inst. Rybnovo Khoziaistva i Okeanog., 45:29-35. [Sources of enrichment of spawning lakes in biogenic elements.] Fisheries Re- search Board of Canada, Trans!. Ser., No. 207, 1959. Ottawa. Lawler, G. H., and W. B. Scott. 1954. Notes on the geographical distri- bution and the hosts of the cestode genus Triaenophorus in North Amer- ica. J. Fish. Research Board Canada, 11(6):884-93. Mathisen, 0. A. 1962. The effect of altered sex ratios on the spawning of red salmon. See elsewhere in this volume. Mertie, J. B., Jr. 1938. The Nushagak District, Alaska. !r S. Geol. Survey, Bull. 903. 96 p. Miller, R. B. 1945. Effects of Triae1Wphorus on growth of two fishes. J. Fish. Research Board Canada, 6(4):334-37. Nelson, P. R. 1958. Relationship between rate of photosynthesis and grmvth of juvenile red salmon. Science, 128(3317):205-7. Nelson, P. R., and W. T. Edmondson. 1955. Limnological effects of fertilizing Bare Lake, Alaska. U.S. Fish and Wildl. Serv., Fishery Bull. 56(102):415-36. Parker, R. R. , and R. E. Vincent. (1956]. Progress report on research studies at the Kitoi Bay Research Station, p. 25-67. In Alaska Depart- ment of Fisheries Ann. Rep:. for 1955. Ricker, W. E. 1937. The food and the food supply of sockeye salmon (Oncorhynchus nerka Walbaum) in Cultus Lake, British Columbia. J. Biol. Board Canada, 3(5):450-68. Ricker, W. E. , and R. E. Foerster. 1948. Computation of fish production. Bull. Bingham Oceanog. Collection, 11(7):173-211. Rounsefell, G. A. 1958. Factors causing decline in sockeye salmon of Karluk River, Alaska. U.S. Fish and Wildl. Serv., Fishery Bull., 58(130):83-169. Scheuring, L. 1930. Beobachtungen zur Biologie des Genus TriaeJwfthm'US und Betrachtungen iiber das Jahreszeitliche Auftreten von Bandwiirm- ern. Z. Parasitenkunde, 2:157-77. Shetter, D. S. 1936. Shrinkage of trout at death and on preservation. Copeia, (1}:60-61. SWrdson; G. 19.55. Salmon stock fluctuations in the Baltic Sea. Drott- ningholm, Sweden. Statens undersoknings-och forsoksanstalt for sotvattensfisket. Ann. Rept. and Short Papers for 1954. (36): 226-62. Thompson, W. F. 1950. Some salmon research problems in Alaska; pre- pared for the meeting held by the National Research Council on scien- tific research in Alaska. Washington; D. C., November 9, 1950. Univ._ Washington Fisheries Research Inst., Circ. 11. 19 p. -----. 1951. An outline for -salmon research in Alaska; prepared for 298 Alaska Red Salmon the meeting of the International Council for the Exploration of the Sea at Amsterdam, October 1-9, 1951. Univ. Washington Fisheries Re- search Inst., Circ. 18. 49 p. U.S. Army Corps of Engineers. 1957. Harbors and rivers in southwest- ern Alaska. U.S. GO",;i:. Printing Office, Washington. 89 p. (House Document No. 390] Ward, H. B. 1932. The origin of the landlocked habit in salmon. Proc. Nat. Acad. Sci. 18(9):569-80. Withler, F. C. 1952. Estimation of the size of the sockeye smolt run, Babine Lake, 1951. Fisheries Research Board of Canada, Pacific Bioi. Sta. Progr. Rept., (91):17-19. 1 N WOOD RIVER LAKE SYSTEM BIUSTOL BAY, ALASKA 01.2345 fA I LE 5 Fig. 1. Wood River lake system1 Bristol Bay, Alaska. Fig. 2. The fyke net used at Mosquito Point index site, Wood River, for smolt sampling. (Photo by W. F. Thompson) Fig. 3. The outlet of Lake Aleknagik, showing fyke-net location (arrow) in the river current at Mosquito Point; Wood River is in the foreground. (Photo from color transparency by Ole A. Mathisen) 25 20 IS 10 s 0 25 3) IS ~ "'10 " w a ~ 5 .. ... 0 25 20 15 1? s 0 !- !- !- i- r-- - 2100- 2200 r-- r--- 2200- 2300 I '1 S ~ 1956 - 1957 - 2300- 2400 Houu r--- r-- 2~- 0100 r-- :---'] 0100- 0lOO I I I I I Fig. 4. Mean percentage distribution of hourly fyke-r.et catches during daily fishing period, 2100-0200 hours, at Mosquito Point smelt- enumeration site 1955-57. .. -;. s Tttn• -21J5 ""''2150 leoeb-tebt.,mpl.., I L ~-->-<:-z..., I 0 Tim• ·-2135 10 ...... ..,2150 s s 0 . l'yke-oet .,mpl.,, dmull<ultollll wltb above !: ~ ... ~ol ..d ~ I .. .. s Compudtt, alloampl.,, Jun• IS (i sample1 907 flab 0 1 1 L r 1 •' • 1 I 70 80 90 100 I I 0 I 20 Longth Ia mllllm•tort Fig. 5. Length-frequency samples of red smolts taken at Mosquito Point, June 15, 1954. (All fish measured alive. All frequencies smv..>thed by moving averages of threes. Frequencies at each millimeter interval graphed as per cent of total sample.) 10 5 IO ~ -g., s tl "' -tl ... 5 Q .. .... 0 ... = Ill u .. Ill 0 II. 10 5 ... ...... Time -2205 June 16 --2238 June 16 ·-·-2214 Jun~ l7 •••••••• 2240 June 17 Beach-teine samplea Time -2~5June 16 ---2238 June 16 ·-·-~214 June 17 •••••••• 2240 Junt-17 Fyke-net samples, simultaneo1.11 ~-~will>abm ~~~--~~ ~ Composiee, all samples, Junr 16-17 10 samples 1132 fiah Length in millimeters 120 Fig. 6. Length-frequency samples of red smolts taken at Mos- quito 1?oint1 June 16 and 17, 1954. (All fish measured alive. All frequencies at each millimeter interval graphed as per cent of total sample.) Ill ... p. E tl "' .... 0 ... ~ (lj 0 .. (lj ll., 5 5 10 5 5 0 10 s 0 10 5 5 Fyke net Beach seine Xf= 88.72 mm "'ib • 88.99 mm 2135, June 15 Xf= 87.94 mm xb•93.70 mm 2150, Junf" IS ~fm84.76 mm xb= 82.62 mm 2205, June 16 Xf• 85.23 mm Xb=86.19 mm 2238, June 16 ~=85.77 mm xb ""'86.60 mm 2214, June 17 xf= 84.44 mm xb=85.07 mm 2240, June 17 Xf:::r 86.07 mm xb =86.95 mm Composite of above liamples, June 15-17 140 Fig. 7. Comparison of pairs of length-frequency samples taken simultaneously by beach seine and fyke net at Mosquito Point, Lake Aleknagik, June 15-17, 1954. .. -0. s. 0 .. ... 0 ... = .. u .. .. g. lO 5 0 10 5 5 0 10 5 0 L~ngth in millimeters t.., 2100-2106 D • 152 w=5.-97 1 .. 88.50 D .157 'il. 5. 78 r. 87.87 D • 170 w c 5.34 T. ss.n t & 2358-2400 n • 183 w-4.95 r-83.04 t • 0058 -QlOO n • 178 w •5.10 T • 84.17 t. 0158-0m<l D • 179 1f -5.07 i = 84.65 Fig. 8. Length frequencies of red smolts in samples captured by fyke net at Mosquito Point at hourly intervals during evening of June 8 and early morning of June 9, 1954. (All fish under 95 mm are yearlings. Measured under anesthesia. Frequencies smoothed by moving averages of threes.} t = fishing period w = mean weight in g n = sample size l = mean length in mm .. -Q, s a n 0 ... 0 ... ... 0 ... n .. u .. " 0:. 10 s t ... 2·1oo-2·104 n = 125 w. 6.45 T = 90.34 0~--L-----~L-------~----~~~-----L--~ 10 5 0 5 0 10 5 0 10 5 0 70 80 length in millimeters t = 2159-2200 n • 157 w: 5.78 T • 88. ·14 t = 2259-2300 n • 164 • w = 5.53 r = 86.81 t :: 2359-2.(00 n • 157 w :11:5.78 r. 87.99 t = 0059-Q 100 n • ·134 w. 5.27 r = 85. 1s 100 110 Fig. 9. Length frequencies of red smolts in st~.mples ca¢ured by fyke net at Mosquito Point at hourly intervals during evening of June 12 and early morning of June 13, 1954. (Measured under anesthesia. Fre- quencies smoothed' by moving averages of threes. All fish under 100 mm were yearlings.) t = fishing period n = sample size w = mean weight in g l = mean length in mm 80 c 60 .. u " .. c. .. > ~ 40 :; E " u zo May 1000 j .5 ~ 500 .. E ~ 0 ... Ill 0 1949 1950 Index of smolts produced fZ1Zl Agr II D Ag .. I Parent escapement 1951 1953 !954 Year af escaprment 1955 Fig. 10. Wood River escapement countpr and indexes of smolts produced. June I i 195Z j ,.~ I ,. . ../ 11956 .. ... July Fig. 11. Cumulative daily fyke-net catches, Mosquito Point, ir> per cent of season's total. Fig. 12. Daily lake-surface temperatures: Lake Nerka station, June-Septennber, 1952-57. 1954 1953 1957 1956 195Z 1955 --Fniod brtwun lGlc~ hrt>ekup and wormlo.g of tm.,_. aurfac~ to so• F utntnua ?rriod brtw~rn date-s then S cmd 90 pPr c:rnt of the ucucus'• cumulatlvr indtx; catch orr frocbtd. 1111111111111111111 IIIIIIIIIIIIIIIIIIIIIIIIUIIUII 11111111 ,. ...................... . II IIIIIIIIIIIIIIIIIIIIIIIUIIIIIIIIIIIIIIIII tlllllllllllllltlillllllltllllllltlll"'''''''"''''' 111111111111111 f:,!lllll I IIIII ..... S 10 IS a. July Fig. 13. Relationship between spring climate and timing of the red smolt migration, Wood River lakes, 1952 through 1957. (Years arranged in sequence of breakup dates at Lake Aleknagik.) Fig. 14. Scale of a red salmon yearling 99 nun long, collected at Mosquito Point on July 5, 1954, showing new summer growth beyond the narrow rings of the annulus. Magnification 113x. 30 .c .. 3: 0 ~ C'l ~ .. a 20 E :I w .. r:; .. u ~ " 10 ... Jun.-July Fig. 15. Changes during 1953 migration season in amount of new summer growth on scales of yearling smolts sampled at Mosquito Point, Lake Aleknagik. (Summer growth is shown as percentage of total anterior scale radius.) 25 t~ fi s 15 il ~ i 10 u J! 5 Oj I L! I I I 10 20 3D •n 25 ~ 20 G. ll e 15 ~ lt :! 10 i .. t c>. s Number of dayt Irom date of lee breakup to date of mmple. 1953 / • ~1957 ·19~ OL-------~--~~--~--------~------~~----~ 0 200 800 Number of aJ.r TU11 accumulated from date ~ lee breakup to date of ~ample Fig. 16. A. Relation between time of ice breakup, Lake Nerka, and percentage new summer growth on scales of red yearlings in above samples. B. Relation between cumulative air temperature following ice breakup and percentage of new summer growth on scales of red yearlings in samples collected on or about July 5 in the years 1952-57. " w 0 s " "' 0 -.. .a s " ~ ~ > w " -" ..: 60 80 Length In mllllmet•n -ngrl ---a~~ U May 2!1-Jun~ 2 \ " Juu• 8-12 June 18-22 Jun• 23•July l 120 Fig. 17. Length frequencies of red smolts, i.n 1954 Wood River seaw, rd migration, weighted by magnitude of 2100-2300 catches in index fyke net and grouped by five-day periods. (Frequen- cies smoothed by moving averages of threes.) 0 li ... 0 M . ... E ~ ~ u .. ~ . ..: Jun• 1-22 --Agd ---Ag•ll July 8-12 60 120 Lfl'ngth ln mllllmco~cra Fig. 18. Length frequencies of red smolts in 1955 Wood River seaward migration, weighted by magnitude of 2100-0200 catches in index fyke net and grouped by five-day periods. (Frequen- cies smoothed by moving averages of threes.) -·c--·-----~------~v-----. ---•-.• ~--.,_.,r---~-----_,....,.._....,. ''\ ,! ' I \ / \ ' ' I ' .. --Age[ ---Agel& Ju,. 1 -12 Juue 13 ... 17 June 18-22 June 23-U E I ' , I ... ........,...,.. 1 d I 0 li ... 0 : 1 t 4" u ... e ~ A .. > ~ a .. ..: 60 June 28 -July 2 July 3-7 July 8-12 July 13 -Augwt 8 so 120 Length In mUUmcters Fig. 19. Length frequencies of red smolts in 1956 Wood River seaward migration, we ... ghted by magnitude of 2100-0200 catches in index fyke net and grouped by five-day periods. (Frequen- cies smoothed by moving averages of threes.) ~ N 0 May 29 -1""• 2 ,_, ... "-. -~ .... ~~-'--- ]oUle 3-7 ]uoe8-12 ...... _ ! I ,J,t1t ,..,____ 'CI..., I ---....--. __ , I ! ... 0 .. ]1111< 18-22 = I ......,..... ,Q ..,_...__.,,=+-~ -I I .. . ... N ! . a: !50 -Agel ---·AgeD ]liDO 2l • 27 ]tme 28 • July 2 July 3-28 80 100 120 Length In m11Umoten Fig. 20. Length frequencies of red smolts in 1957 Wood River seaward migration, weighted by magnitude of 2100-0200 catches in index fyke net and grouped by five-day periods. (Frequen- cies smoothed by moving 2.verages of threes.) r-6~ (\ -Aoe 1 4 ~ ---·Age II 1954 2 19!5!5 2 0 4 ....... t:l 19!56 .... ~ 2 .., ~ 0 ~ 4 ~ 'II) 1957 ..... 2 (l 1958 4 19!59 2 0 6'0 120 Fig. 21. Season composite length frequencies of red salmon smolts, 1954-57, Wood River. A B Fig. '22. Photomicrographs of scales from small and large yearling smolts in the 1955 seaward migration, and of nuclear areas of scales from adults in the 1957 return escapement, Wood River lakes (magnification 43x): A. Scale from smolt 72 mm long; B. Nuclear area of scale from adult female, Hansen Creek, Lake Aleknagik, showing small amount of growth to first annulus; C. Scale from smolt 90 mm long; D: Nv.clear area of scale from adult female, Fenno Creek, Lake Nerka, showing good growth to first annulus. 1 ... 0 5 u ll "" Length In mllUmetm Fig. 23. Season composite length frequency of red salmon year-. ling migrants, Wood River, 1955, showing length groups from which scale radius measurements were taken, and number of scale measurements made. _______ .,.. _____ _...•~M,09tW4¥V ;u:w;a:::az 144 az.s:;::t:U.,tWii$~4}$lW¥f.., ... 4iPfMt£Ji.?h¥!¢.k:H*$,.~;;)!CJi'Xjtid4?ftW'!.l;%::i;;l\6ZpUtJ;Q;t:E)il;i;%i!IAljifll'f>K1&Jj.._ff~,.*ifMfk .. Ui'JJt\ti:;;:;JiM1Ilk_:cg::;:}MUJ£Q44JAifU4JVWQ4MRWii4f.~ " :z: a) 10 A ·--Aleknagik adults -Small yrarllngs n•29 D"S6 o~ ~ I B 20 10 ----adults Jram lakrs above Alrknaglk n=29Z -l.Grgr yearlings ll"lZ2 t-, I , __ \ , \ I \ I ' . \ I \ I \ ' \ I \ I \ I \ I ' I l I \ ' ' I ' I 0 I 'rrC"?C:' :( • • I ' 'I ,...,. SO 6S 80 95 110 1ZS 140 Magnlflrd seal., radius to first annulus, In mllUmrtrrs Fig. 24. Comparison betwr-en scale radius measurements of yearling smolts ancl of four- year-old adult red salmon (scale formula 1.2) from 1953 year class, Wood River lakes: A. Modal group of small yearlings, 1955, com- pared with adult fish from Lake Aleknagik spawning grounds, 1957; B. Modal group of large yearlings, 1955, compared with adult fish from Wood River lakes above Lake Aleknagik, 1957. .. w 0 .. e .. .. " .. 0 " e .. " ., 0 .. ... 0 .. .. .0 e " :z: 3-L. Nrrka Frnno Crerk 4-t. Nrrka Allah Beach S-L. Nerka Stovall Crrrk 1 6-L. Nerka N-4 B~ach I 7-L. N~rka Pick Crrei< I 8-Llttlr Togiak l.Gke West end I 9-L. Nrrka Anvil Boy I 10-L, Nerka Krma Crrek 13-L. Beverley SllvnHorn 14-L. Kulik Crant Rlv<r no:29 ..,.zs .,.zs ...... zs naZS n-ZS ,.,zs n•ZS n•ZS nmZS no<ZS ...... zs n•ll n•ll SO 65 80 95 110 IZS 140 Magnlflrd scalr radius to flm annulus, In mllllmrtrn Fig. 25. Frequencies of radius measurements of scales from four-year-old adult 1·ed salmon, year class 1953, scale formula 1.2. (Measure- ments from focus to edge of fresh-water annulus x 230, grouped in 5 mm-class intervals. Scales taken from spav.'!led-out fish.) Fig. 26. Spawning grounds represented in the study of first-year growth attained by adults, 1953 year class, returning 1957. (See Figure 25 for names of numbered areas.) 10 40 10 10 Ulwn LGiol' Nrrkcaad UttlrTogiak Lak• .... •• 170 50 65 95 110 125 140 Ma!Jillfird seal .. tad.ha lD fint ~~~U~Ullll, la miWmrcm Fig. 27. Difference between lake areas in scale radius measurements of four-year-old adult red salmon, scale formula 1.2, 1953 year class.