LIFE HISTORY, TRO-PHIC RELATIONSHIPS, AND BATHYMETRIC :DISTRIBUHGN AND MOVEMENT OF AMERICAN SMELT, OSMERUS MORDAX (MiTCHlLLJ‘, zIN GULL LAKE, KALAMAZOO AND BARRY COUNTJES, MICHSGAN Ihesis for the Degree of M. S. iVliCHEGAN STATE UNWERSSTY RiCMR‘D QREIiB EURB-EDGE 2.967 11111111111111111111 1293008316733 1 LIBRARY f THaws 1W2 9 3-004 MAY 0 5 2010 $5, '3 3—31 7‘ 1‘3 ABSTRACT LIFE HISTORY, TROPHIC RELATIONSHIPS, AND BATHYMETRIC DISTRIBUTION AND MOVEMENT OF AMERICAN SMELT, OSMERUS MORDAX (MITCHILL), IN GULL LAKE, KALAMAZOO AND BARRY COUNTIES, MICHIGAN by Richard Greig Burbidge American smelt represent a fishery of considerable importance but like many prolific species, they have flourished and are sometimes a problem. Ecological relationships of smelt are a source of contro- versy. Smelt are taken commercially only during short seasonal periods in fluctuating amounts. More must be known of life history, trophic relationships, and bathymetric distribution and movement before smelt can be effectively controlled and utilized. The study of the natural history of smelt in Gull Lake was based on data from 286 smelt collected with gill nets, an Isaacs-Kidd midwater trawl, and hook and line, from December 1965 to November 1966. Physical measurements (length, weight, sex, age) were taken, and stomachs were analyzed. Echo recordings of fish distribution.were obtained with a Furuno F-850-A (200 kc/sec) Fish Finder, from October to December 1966. The number and position of fish traces in each transect were recorded to demonstrate vertical and horizontal movement. Temperature and dissolved oxygen profiles were determined. Gull Lake smelt were shown to be one genetic race; their growth was rapid, but mature individuals were smaller than smelt in other environments. Growth of young fish‘was comparable. A high natural mortality in the fourth year was indicated. There was a predominance of females at all ages, and females were larger than males. Richard Greig Burbidge Smelt in Gull Lake ate the most available food, but apparently had a preference for dipteran larvae. A change from night to daylight feeding, accompanied by a change to filter feeding, was indicated as larger dipteran larvae declined. Only a few fish were noted in stomachs of smelt. Evidence that smelt are prey of larger game fish was established. Smelt were restricted to the colder hypolimnion in summer; in colder months they were distributed at all depths. Diel vertical and horizontal movements of the smelt population were demonstrated. LIFE HISTORY, TROPHIC RELATIONSHIPS, AND BATHYMETRIC DISTRIBUTION AND MOVEMENT OF AMERICAN SMELT, OSMERUS MORDAX (MITCHILL), IN GULL LAKE, KALAMAZOO AND BARRY COUNTIES, MICHIGAN By Richard Greig Burbidge A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of ZOOIOgy 1967 Acknowledgments I would like to thank Drs. G.H. Lauff, D.C. McNaught, J.C. Braddock, and E.W. Roelofs for their time and effort spent as members of my graduate committee. Their comments on problems and methods were invaluable, and their suggestions greatly improved the presentation. I am especially indebted to Dr. Lauff for his concern and guidance throughout the duration of my graduate pragram; to Dr. McNaught for his help in the field; to the department of zoology for financial support; and to Mrs. B.R. Henderson for her kind assistance. Special thanks are due John L. Hesse who spent many sleepless nights helping collect data, and J. Whitfield Gibbons and Don L. McGregor who also helped with the field work. Information on the fishes of Gull Lake was provided by Dr. C.W. Huver and the Plainwell office of the Michigan Conservation Department. Finally, I wish to thank my wife, Elena, for her moral support and tolerance throughout a most difficult period, and for the many hours she spent typing the thesis. ii I. II. II. III. IV. V. VI. Table of Contents Introduction . . . . . . . Literature Review . . . . A. Life Hiatory B. Trophic Relationships . C. Bathymetric Distribution and Movement Materials and Methods Resalts and 13186088101! 0 o o o o o o o A. Life History 3. Trophic RelationShips o o o o e e o C. Bathymetric Distribution and Movement Summary and Conclusions Literature Cited 0 o o o o o o o o o o 0 Appendix................ iii 14 14 36 50 70 73 81 Table Table Table Table Table Table Table Table Table Table Table Table Table Table l. 3. 4. 5. 6. 7. 8. 9. 10. ll. 12. l3. 14. List of Tables Total length distribution of Gull Lake smelt . . . . Percent age composition of Gull Lake smelt . . . . . Mean total length (mm) of Gull Lake smelt . . . . . Mean weight (g) of Gull Lake smelt . . . . . . . . . Total length-weight relationship of Gull Lake smelt Sex ratio of Gull Lake smelt . . . . . . . . . . . . Sexual dimorphism of Gull Lake smelt . . . . . . . . Food of Gull Lake smelt expressed in estimated percent°ft0talw1m00000000000000 Food of Gull Lake smelt expressed in percent frequencyofoccurrence oeeeoeooeooeoe Seasonal feeding habits of Gull Lake smelt . . . . . Diel feeding habits of Gull Lake smelt . . . . . . . Seasonal bathymetric distribution of Gull Lake smelt Diel bathymetric distribution of Gull Lake smelt . . Number of fish in each sonar transect expressed as percent of maximum number for one transect . . . . . iv 15 23 25 26 3O 32 34 37 39 45 48 52 57 61 F180 F180 F180 F180 Fig. F180 F180 F180 F180 F18. F13. Fig. F180 F180 F180 F180 4. 5. 6. 7. 8. 9. l3. I4. 15. List of Figures Gull Lake, Kalamazoo and Barry Counties, Michigan . . Diagramatic representation of the Gull Lake sonar transect (Point A to Point B) showing quadrats included within the half-sound pressure beam of sound 0 o e o o o e o o e o o o o o o o o e o e e o 0 Total length distribution of Gull Lake smelt . . . . Total length distribution of age groups of Gull Lake smlt C O O O O O O O O O O O O O O O O O O O O O O 0 Total length distribution of male and female Gull Lake Smalt e o e o o o o o o e o o o o e o o o e o 0 Mean total length and weight of age groups of Gull Lake smelt o e o o e o o o o o o e o o o o o o o o 0 Total lengthpweight relationship of Gull Lake smelt . Sexual dimorphism of Gull Lake smelt . . . . . . . . Food of Gull Lake smelt expressed in percent frequency Of occurrence 0 o o o o o e o o o o e o e 0 Seasonal feeding habits of Gull Lake smelt . . . . . Gull Lake temperature (C) and dissolved oxygen (Mg/1) prOfiles e e o o o e o o e o o o e o o o o o o e e 0 Seasonal bathymetric distribution of Gull Lake smelt expressed in percent of catch for each month . . . . Diel vertical distribution and movement of Gull Lake fish illustrated by the quartile curve method . . . . Diel vertical distribution and movement of Gull Lake fish illustrated by the quartile curve method (11.12 OCtOber) o o o o o e o o o e o o o o o o o o o Diel horizontal distribution and movement of Gull Lake fish illustrated by the quartile curve method . Diel horizontal distribution and movement of Gull Lake fish illustrated by the quartile curve mathOd (11-12 OCtOber) o o o o o o o e e o o o o e o IO 13 17 19 20 27 31 35 40 46 51 53 58 59 67 69 Introduction The smelts (Osmeridae) comprise a group of anadromous fish of wide geographical distribution in northern latitudes. Their recorded distribution includes both North Atlantic, North Pacific, and Arctic coasts, forming an irregular, interrupted, distributional belt that encircles the Northern Hemisphere (Kendall, 1927). The American smelt, Osmerus £95935 (Mitchill), is found along the Atlantic coast of the United States from the Raritan River (400 30') to the Gulf of St. Lawrence. They are abundant along northern shores of New Brunswick, and are found in many fresh-water lakes of Maine, New Brunswick, and Nova Scotia, where they have become landlocked (Goods, 1884). The first shipment of smelt to the Great Lakes was in 1906, when eggs from New England were deposited in the St. Mary‘s River, Mich. (Van Oosten, 1937). This introduction.was not successful for the species was never taken in that region. In 1912, 16 million eggs from Green Lake, Maine, were deposited in Crystal Lake, Mich.; success became apparent when large breeding runs were observed in 1919 (Creaser, 1926). Smelt were first captured in Lake Michigan in 1923 (Van Oosten, 1937) and their ensuing dispersal and establishment in the Great Lakes recorded (Hankinson and Hubbs, 1922; Creaser, 1925; Savage, 1935; Van Oosten, 1937; Clarke, 1944; and Dymond, 1944). It is certain that the Crystal Lake population.was the only source of smelt now present in Michigan and all the Great Lakes except Ontario, where the species has been present since historical time (Van Oosten, 1953). Smelt have flourished in spite of a heavy mortality in 1942-43 which caused an estimated loss of 50 million lb. (Van Oosten, 1947). l 2 Great Lakes commercial smelt production was 16 million lb. in 1960 (Baldwin and Saalfeld, 1962). Smelt are taken with pound, gill, and fyke nets, principally during the winter; most fishing is done in Green Bay (Van.0osten, 1953; Swanson, 1955). Canadian fishermen have success- fully used trawls to harvest Lake Erie smelt (Ferguson and Regier, 1963). Great Lakes smelt normally spawn at night in accessible streams during April and May at the beginning of their third growing season (Stevenson, 1944; Baldwin, 1950; Van Oosten, 1953). Sportsmen employ dip nets to harvest a quantity of smelt estimated to be greater than commercial catches (Schneberger, 1937). Smelt have been a source of controversy since their establishment in fresh water. They are an item of considerable importance to commercial fishermen and sportsmen, but like many prolific species in a favorable new habitat they have flourished. In some areas smelt have become a menace to commercial fishermen (Schneberger, 1937). Knowledge of their life history may make effective control possible. Smelt are considered important because of their varied ecological relationships with game species (Rupp, 1959). ‘At this time these relationships are not certain. Extreme levels of abundance make it necessary to understand their role as predators, competitors, and prey (Gordon, 1961). The species is abundant in certain ares of the Great Lakes, but cannot be taken in quantity except during short seasonal periods. Knowledge of their movements and distribution must be extended if the smelt fishery is to be profitable (Carr, 1964). The present investigation of smelt in Gull Lake was undertaken to contribute to a description of the species and enhance the knowledge of its ecology. Information on the natural history of smelt may afford 3 a better understanding of their economic potential, and their present and future biological and ecological role. A total of 25,000 adult smelt from the Oden State Fish Hatchery were introduced in Gull Lake, Kalamazoo and Barry Counties, Mich., from 1950 to 1953 (Plaimell office of the Michigan Conservation Department, unpublished records). The first heavy spawning run was observed in 1956 in Prairieville Township Park Creek. Establishmsnt has been successful except for a moderate mortality in 1962, and a heavy mortality due to a pathogenic bacteria in 1965. Gull Lake smelt spawn at night in April in Prairieville Township Park Creek, Little Long Lake Creek, and along the shore between these streams at the beginning of their third growing season. Spawning smelt are protected so there is no dip netting. Sportsmen fish for smelt by hook and line in the winter. Gull Lake is among the larger and deeper lakes of Michigan (Taube and Bacon, 1952). The basin is of glacial origin, formed probably during the last ice invasion. Surface area is 8.2 kmz; maximum depth is 33 m. The drainage is relatively small; inlets from Miller and Little Long Lakes, springs, and several other streams contribute to the water supply. A dam at Gull Lake Outlet controls the outflow of water to the Kalamasoo River. Sand, gravel, and rubble are the principal shoal bottom types. Marl, muck, and pulpy peat are fond in deeper areas. A mixture of marl and pulpy peat is predominant. Gull is an alkaline dimictic lake that thermally stratifies in ether (Hutchinson and Lgffler, 1956). Water appears green due to fine particles of marl in suspension, and the penetration and back-scattering of predominantly green and blue light. Gull Lake is well supplied with vegetation; at least 24 species of 4 vascular plants have been recorded (specimens available in the Kellogg Biological station herbarium collection). A wide variety of fish is present in Gull Lake (manuscript in preparation, C. W. Buver). The most common forage fish are the brook silversides (Labidesthes sicculus), logperch (Percina caggodes), and bluntnose minnow (Pigghales notatus). The white sucker (Catostomus conersoni), yellow bullhead (Igtalurus natalis), and longnose gar (Episosgeus osseus) are cos-on. Game fish include the smallmouth bass (Microgerus dolomieui), largemouth bass (Microgerus salggides), green sunfish (Emmi; cyanellus), bluegill (Loggia agochirus), rock bass (ablgglites rupgstris), yellow perch (_P_g£c_a_ flanscens), cisco (Congonus artedii), rainbow trout (_S_a_l;o_ gairdneri), and lake trout (Salvelmg gym“). Rainbow trout have been stocked annually since 1942, lake trout since 1943. Fishing pressure is moderate. Emphasis is placed on rainbow trout, lake trout, and smelt. Ciscos, also cold-water fish, are native to the lake. Smallmouth and largemouth bass, bluegills, and yellow perch are the most important warm-water species. A study would be incomplete without a consideration of life history, trophic relationships, and bathymetric distribution and movement. There is a distinct relation between these three aspects; a discussion of one phase would be impossible without incorporating the others . Literature Review Life History Aspects of smelt life history are covered in the literature. Much has been based on samples collected during spawning runs, or within a restricted period, or that were too small. The nature and composition of a smelt population must be studied over a period of time if the fish is to be effectively controlled and utilised. Kendall (1927) su-arised most early information on smelt life history and included studies on both marine and landlocked smelt. The Fisheries Research Board of Canada initiated an investigation of smelt in the Miramichi River and Bay, New Brunswick. Various phases of this study were reported (McKenzie, 1943, 1944, 1946, 1947, 1948, 1956, 1958, 1964). Coastal smelt in Great Bay, New Hamp., were studied by warfel, Frost, and Jones (1943). New England smelt have been studied in lakes of New York (Greene, 1930; Zilliox and Youngs, 1958), New Hampshire (Hoover, 1936), and Maine (Rupp, 1959; Rupp and Redmond, 1966). Smslt in Crystal Lake, Mich., were investigated by Crosser (1926) and Beckman (1942). Studies were conducted on Great Lakes smelt in Lake Michigan (Creaser, 1929a; Schnaberger, 1937; Baton and Tack, 1952), Lake Huron (Baldwin, 1950), and Lake Superior (Bailey, 1964)- The present study is concerned with the length distribution, percent age composition, mean length and weight, length-weight relation- ship, sex ratio, and sexual dimorphism of smelt collected in Gull Lake over a 1 year period. The spawning phenomenon was not investigated in Gull Lake, but accounts are available (Creaser, 1926; Kendall, 1927; Greene, 1930; Schneberger, 1937; Hoover, 1936; Stevenson, 1944; 5 6 McKenzie, 1944, 1947, 1964; Baldwin, 1950; Van Oosten, 1953; Lievense, 1954; Rupp, 1959, 1965; Rothschild, 1961). Eggghig,Relationshigs Trophic relationships in lakes in.which smelt were introduced has been a source of dispute, arising from variations in reports of the food and feeding habits of smelt. These can be attributed to small samples collected in one season, incorrect interpretation of results, and differences in habitat and size of fish. Smelt are often introduced as food for game fish, but many believe they reduce the number of game fish fry. Some fear smelt as competitors for the food of game fish. Concern for the role of smelt began'with Creaser (1926, 1929a, 1929b),'who examined smelt from Lake Michigan and several Michigan lakes and found them feeding on crustaceans, insects, emerald shiners (Notropis atherinoides), and smelt and yellow perch fry. He concluded: "The smelt is...an enemy of all smaller fishes, including the young of the commercial species, as well as a competitor for the food of the adults of the larger species." Creaser was supported by Doolittle (from Kendall, 1927) and Greene (1930),*who observed small smelt in New England lakes feeding on seoplankton, but large individuals feeding primarily on other smelt and insect larvae. In contrast, other studies in New Hampshire (Hoover, 1936), Green Bay (Scmbergor. 1937), Lake Huron (Baldwin, 1950; Gordon, 1961), and Lake Erie (Ferguson, 1965) found smelt seldom eating fish, but feeding primarily on bfitt°l faun. and sooplankton. Van Oosten (1953) summarised: "...newhere have inyestigators found such species (game and commercial fish) present in the stomachs of smelt in any significant quantities." 7 Evidence that smelt are eaten by commercial and game fish‘was provided by Kendall (1927), Greene (1930), Dymond (1944), Tharratt (1959), and McCaig and Mullen (1960). It was the purpose of the Gull Lake study to analyse the food and feeding habits of smelt collected over 1 year to determine their position in the nutritional scheme of the lake, and understand their role as competitors, predators, and prey. Bathyggtgic Distribution and Movement Diel bathymetric movements of marine fish have been noted by Hickling (1935), Welsh, Chace and Nunnemacher (1937), Brawn (1960), Johnson (1961), Manzer (1964), and Clarke and Backus (1964). Similar movements have been reported in fresh water for the cisco (Cahn, 1927), and white bass (gggggg_chrzsopg) (McNaught and Hasler, 1961). Recording echo sounders have been used to find fish and measure their movements. Hodgson (1950) reported on the use of echo sounders for pelagic fishing; Balls (1951), Richardson (1952), and Valdes and Cushing (1966) utilised these instruments to record vertical movements. Cushing (1952, 1954, 1957, 1963) has done much to perfect their use in fish distribution studies. Schaefers and Powell (1958) correlated midwater trawl catches with echo recordings. Clarke and Backus (1956) measured vertical movements of deep scattering layers. Northcote, Lors and MacLeod (1964) recorded diel movements of freshdwater fish. Gordon and Larsen (1965) published on the application of echo recorders to fishery research. Most research on smelt distribution has been concerned with locating areas where the fish could be effectively harvested. Sampling was done primarily'with bottom nets and trawls so little is known of their 8 bathymetric distribution and movements. Vertical distribution of smelt has been studied in lakes of New York (Greene, 1930; Odell, 1932; Galligan, 1962) and the Great Lakes (Van Oosten, 1953; Gordon, 1963; Carr, 1962, 1964). That smelt have diel vertical movements was first reported by Kendall (1927),*who suggested that they may ascend in the evening to feed. Sand and Gordon (1960) and Ferguson (1965) found a dispersal of Lake Erie smelt to midwater in late afternoon, and a return to bottom in the morning. Materials and Methods This study was based on data from 286 smelt collected in Gull Lake from 17 December 1965 to 12 November 1966, and echo recordings obtained from 7 October 1966 to 15 December 1966. Smelt were collected with a 1/2-inch nylon gill net (100 x 6 ft), 1-inch nylon gill net (70 x 6 ft), 3-ft Isaacs-Kidd midwater trawl, and hook and line. A small otter trawl (3 m) and Gulf III sampler were utilised with no success. Gill netting operations were initiated in December 1965. The 1/2- inch gill net was fished on the bottom at Stations 1, 2, and 3 (Fig. 1) from December to June. During February it was set at Station 4 directly under the ice, and vertically from surface to bottom. The 1/2-inch gill net was marked in meters and fished vertically at Station 5 from July to November to detect diel fish movements. The not was attached to two buoys at the surface, and rods were fastened to both ends to keep it open. It was set and recovered primarily at night at 2 and 3 hr intervals. Fish were removed and their depth recorded. The gear was anchored to a permanent buoy to minimise drift. The 1-inch gill net was fished on bottom at Stations 6 and 7 in August. The Isaacs-Kidd trawl was towed at various depths from September to November. Runs from 10 min to 1 hr were made around Station 5, or from Station 5 to the north end of the lake. One February sample was obtained by fishing through the ice with hook and line at Station 8. The selectivity of the gear was obvious. The midwater trawl captured age-group O or small age-group I smelt, while gill nets captured age-groups I to V. Smelt were not collected in January and 9 .eduaoaees one wagon uoomaouu use oneness couueum .scwzog .mouugou human use ooucsgex .923 :30 .u .wum x. ..::z I! .‘ .... 1.1; . e Oe‘liea ..e 0. .‘Au.. 82.. .. S. M. .... ..h w.....A._ . . .... ...H..... . ........?... ..n." .. ”Wren”? . .. ....w...... . . ......m... .0 . m x < 4 ............ 1 Cum. . .... ......Hm. m w ...—i- _ Swan”. ......H..... A... @ “..m ......m: urn. 2. _..oa ... .... .0 d .. . .u....... ....... 0 20... ® ® .. ® ......” ..."... v.35 ...mxda 020- ....... 3:3,. 10 ..: 1!. 4 0.2.0 20 :45. ......G.H.................:.. .... .. 20.0306 000.39. is em”... gamma”... ..... ...HH...H..HH.... m3... w.._4_>m_m_¢m.nw..... 11 March when the lake was freezing and opening. A June sample was lost due to improper preservation. Smelt were preserved in 57. formalin and returned to the laboratory for examination. Total length (tip of snout to tip of tail, lobes compressed) and standard length (tip of snout to end of vertebrae) were measured to the nearest millimeter. Weight was determined to the nearest 0.1 g. Sex was determined by gross inspection of the gonads. Sex of age-group 0 fish could not be determined with confidence. Scales were removed from behind the dorsal fin, examined under a drop of water at 100 x, and aged according to a "shiny line" criterion (McKenzie, 1958; Bailey, 1964). A year of growth was considered to have begun after the April spawning run. Smelt captured in April had completed spawning. Stomach contents were removed from the lower esophagus to the pyloric sphincter and examined in water under a stereomicroscope. Percent fullness was estimated on the basis that a full stomach was completely distended. Food items were sorted and the percent which each item composed of the entire mass was estimated (Lagler, 1952). Volume was measured with a sedimentation tube graduated to 0.1 m1; winter samples were too small to be measured accurately. Remains included all unidentified digested matter. Temperature and dissolved oxygen were measured from July to November. Temperature was determined with a Whitney electric resistance thermometer. Water samples were obtained with a Van Dorn water bottle and analysed for dissolved oxygen by an unmodified Winkler technique. All field work was done from a 27 ft inboard vessel or small boats. Echo sounding recordings were obtained with a 200 kc/sec Furuno "Triton" Mode], 17-85041 White Line Fish Finder to determine the distribution 12 and movement of fish. The sound wavelength in water (15 C) was 0.7185 cm. Sound pulse frequency was 170/min. Duration of the sound pulse was 0.0015 sec, so the sound pulse length was 2.156. The beam of sound was cone-shaped and symmetrical.. The beam angle at half-power point was 5.50 (full-angle); at half-sound pressure it was 7.50 (full-angle) (personal communication, 8. Kunitomo, Furuno Electric Company). Transacts were run with the echo sounder between Points A and B (Fig. 1) every hour at various times from October to December; engine speed varied from 500 to 1000 rpm. The instrument was set Gain lo, at a depth range of 0-40 fathoms. At this gain only echoes from fish were received (see Appendix). Individual fish echoes were recorded on dry electrosensitive paper as vertical traces approximately 0.5 mm wide and 3 - long. Each recorded transect was divided into eight vertical depth intervals of 4 m, and eight horizontal units of equal length (Fig. 2)- The number of fish traces in each quadrat was recorded; the top of each trace was considered as the actual depth of the fish. The approximate volume of each quadrat included within the sound beam was calculated. The number of traces in each quadrat was divided by the volume to give a density of fish per cubic meter. The density in each quadrat was then converted to percent of the total number of fish in the transect. The depth interval frog 0 to I. m was not included because the artificial surface line was 2 m deep. Further, the 28 to 32 m depth interval was not included because the bottom contour varied considerably. All transects were run in calm weather so the volume of water covered by the scum beam was consistent. 12 .essom we soon ousmmoua mssomuufims ecu enemas wounded“ mueuomae mausoam Am usfiom ou < usuomv uooeaeuu meson oxcg daze ecu no couueusomoueou owuuaeuwcua ON Owflm Results and Discussion Life History The length distribution of age-group 0 smelt did not overlap either sex of age-group I (Table 1). The gap between the two was attributed to the time they'were collected; age-group 0 smelt were taken in fall, but age-group I fish‘were not captured until summer. The size range of age-groups I to IV varied from 69 to 29 mm. The sise range of I-group males overlapped the II-group by 39 mm; II-group males overlapped the III-group by 29 mm. Age-group I females almost completely overlapped II-group females. An obvious difference existed between age-group II and III females, except for one individual in the 145-149 mm interval. Females of age-groups I, II, III, and IV were larger than males in these groups. The 5 mm length range intervals included individuals from as many as three age groups. Each interval from 140 to 199 mm included at least two age groups, and five age groups (I to V) were represented at 180 to 189 mm. Except for age-group 0, length‘was a poor index of age. McKensie (1958) investigated the marine smelt that spawn in the Mira-ichi River, New Brunswick, and found one size range. The largest seelt was 288 mm. The length distribution and maximum length attained by freshpwater smelt in different lakes is highly variable (Rupp, 1959). Kendall (1927) was aware of this and compared smelt from 21 Maine lakes. Mean lengths ranged from 58 to 305 mm. The largest smelt was 401 mm. Kendall indicated two races of smelt in some New England lakes, "large smelts" and "small smelts." Some have only the large race, or only the small race, or both. 14 15 Table 1. Total length distribution of Gull Lake smelt. Sex undeter- mined for age-group 0. M, male; F, female Age-group and sex has“ 0 I II III IV v (m) MP M F M F M F M r M 40-44 1 45-49 1 50-54 2 55-59 1 60-64 1 65-69 4 70-74 75-79 1 80-84 85-89 90-94 95-99 100-104 105-109 110-114 115-119 120-124 125-129 130-134 135-139 140-144 145-149 150-154 155-159 160-164 165-169 170-174 175-179 180-184 185-189 190-194 1 195-199 200-204 205-209 ...-e ”H “F‘NQNNQNV’N§H pa eat- F‘UOF‘UUUUH pm H UINOUlmJ-‘NN .... Hr— NUIOGWUUIOO u-e [Or-00.01:— .— 0— Total 11 47 82 51 63 6 l7 2 5 1 16 Greene (1930) presented evidence for large and small races in Lake Champlain. He said: "Differing growth rates, sizes at maturity, food selectivity and the fact that their apparent integrity has been preserved even though the two groups occur together throughout the lake, all point to a genetic rather than an environmental basis for the separation of the two groups." He did not find an anatomical basis for the two groups, and he found smelt with intermediate growth rates that could not be assigned to either race. Greene recorded a range for age-group II of 100 to 160 mm for the small race, and 150 to 240 mm for the large race. Zilliox and Youngs (1958) studied Lake Champlain smelt and accepted the separate race theory‘without discussion. Rupp (1959) reported on large and small races in some Maine lakes, and found two distinct spawning runs; the large race spawned first and the smaller second. The largest smelt was 279 mm. Creaser (1926) discovered that smelt in Crystal Lake matured at a large size and reached a length (305 mm) as great as some New England and salt water smelt. He concluded that large race smelt were introduced into the Great Lakes. Kendall (1927) supported Creaser by pointing out that Crystal Lake smelt came from a large race cultured at Green Lake Fish Hatchery, Maine. Evidence for the large race in the Great Lakes was given.by Schneberger (1937) who found Green Bay smelt at 305 mm, and Dymond (1944) who found one 367 - specimen in Lake Huron. Bailey (1964) recorded Lake Superior smelt at 280 mm. The length distribution of Gull Lake smelt (Fig. 3) indicated that lengths were roughly divisible into four modes, with most fish in the first two. This could not be satisfactorily compared to the smelt of New England because of the absence of two size ranges. Gull 16 62:52: you 3 o muchwuow< .ueaam mane ease no coausnauumne sundae Hauoa .m .w.a 17 Cf»: IHGZMJ VON mow CON mm. Om. mm. 0m: mt Oh. no. 00. mm. on. mi 0.3 mm. on. em. ON_ 0: a v .|.l_ Flo III. an a. . . lllllll, I. O— .- In. II ION I now I ..om I 100 .l TIL. J 0* 9v £138 WON 18 Lake smelt were also far short of the maximum size attained by easternnsmelt. The length distribution of Gull Lake smelt was similar to Great Lakes, Miramichi River, and Crystal Lake smelt for there was only one size range, but maximum length in these areas was greater. This indicates that one genetic race was introduced in Gull Lake. This is probably a safe assumption because they came from the Oden State Fish Hatchery near Petosky, Mich. The small race seldom reaches a size over 150 mm (Creaser, 1926), so Gull Lake smelt are probably large race fish. If Gull Lake smelt are the large race, they did not have a comparable growth rate, which suggests that Gull Lake is less favorable for the growth of smelt. It may be a physically harsh environment, or there may be differences in diet to account for size differences. Gull Lake smelt have a growth rate between the two races of New England smelt. The length distribution of age groups (Fig. 4) demonstrated that the first mode consisted primarily of age-group I smelt, and the second of age-groups I and II. The third mode resulted chiefly from age-group III, and the fourth from age-groups III and IV. Length curves for age-groups I and II overlapped and had the same slope and range. Individuals in the 160-164 mm interval were most numerous. The length range of age-groups I through IV'was 110 to 214 mm, or a 104 mm difference. Age-groups 0 and V‘were not included because of a small sample. Only two individuals of age-group V*were captured. The length range of age-group I Gull Lake smelt‘was approximately the same as Lake Superior smelt (104 to 213 mm) (Bailey, 1964). Gull Lake age-groups II, III, and IV were smaller, indicating larger Gull Lake smelt fared less well than younger fish. Fig. 5 illustrates the length distribution of Gull Lake male and 18 .uaoam oon game no mascuw own «0 douuanuuumdu summed deuce eeV ewdh 3...“. 2V IFS Z m .4 gm 8 com 8. om. mm. ow. P: 8.. 8. 8. mm. 8. me. oi mm. 0... mm. 0.0.. m: . w n , n l9 1 _ . _ “ FIAIIfiI: ll+lbllc ello , a l o. . macaw Iwo< II , ..II. {ON L on 1 . a a v e a . O I o. : a:omw-mo< [om ”LII-EL Till I u d Ti 1 — — u — O 1 o. .= Q30m01wo< low 1 on _ _ FIWIL, TIL, . _ _ . a . _ . q _ o 1 o. >_ m20m0lwo< 1 ON on HBBWHN l9 .mov:~od~ uo: mu 0 esoumaow< .335 £3 :5 238 use 22. no 533333 cameos egos .n .wum i/ 0'- CII’B \\ ‘- ’ /////[/[//////////// j I. 7//////////////////// I W////// W m o o N a: “-3 HBBWflN I [n O 2I0 II5 l20 I25 I30 I35 I40 I45 l50 I55 I60 I65 I70 I75 I80 I85 I90 I95 200 20.3 LENGTH(MM) 21 female smelt. There were fewer males than females in only one interval (150-154 mm). There was an equal number of each sex in the 130-135 and I60-165 mm intervals. The greatest difference existed in the 165-169 mm interval (19 females and 8 males); this difference was not highly significant (P<0.10) and could have resulted through sampling. Females had a total length range of 104 mm, and males 84 mm. The curves indicated that the sex ratio remained relatively constant for most length intervals. The consistent predominance of females at all length ranges suggests that forces selecting females were important at all sizes. McKenzie (1958) and Bailey (1964) found a predominance of female smelt at all lengths. Smelt generally spawn at the beginning of the third growing season (Van Oosten, 1953). A pronounced mortality often occurs among fresh-water smelt during or shortly after spawning (Kendall, 1927). If males are in poorer condition than females during spawning, they may be more susceptible to disease, parasites, predation, starvation, or extreme‘weather. This could account for the fewer numbers of mature males. The fewer numbers of immature males could have resulted from behavioral differences that made them more susceptible to predation or less susceptible to capture. That males are in poorer condition has not been proven, although it is known that males are more active than females during spawning, and may remain in the spawning stream for an.entire night or throughout the day (Hoover, 1936). McKenzie (1964) said: "Mortality among the males (smelt) at spawning is at least partly responsible for this increase in the Proportion of females with age.” The increase in proportion of Gull Lake females may have been due to the accumulated effect of males dying'while spawning, or decreased predation on larger 22 females. The percent age composition of Gull Lake smelt (Table 2) varied considerably with the month. April and May samples included only age-groups II and III. The July collection consisted of age-group II (84%) and older fish. Age-group I appeared in August, but most of the sample was age-group II (76%). The September sample was young due to the selectivity of the Isaacs-Kidd trawl. Age-group I smelt captured by gill net increased in the fall, and older individuals decreased. The December sample was comprised primarily of age-group I (89%). For the entire year, age-group I'was dominant (45%), followed by age-group II (40%). The small number of age-group 0 smelt was not an indication of their numbers in Gull Lake, but of the inability of gill nets to capture them. The percent age composition of Gull Lake smelt shows that few age- group I fish were taken until fall, and in fall the number of age-group II fish decreased. In spring and early sulmaer age-group I fish were not much larger than the previous winter, and would pass through the nets. As they grew they were captured in increasing numbers, and were most numerous in the winter samples. Age-group II smelt were caught in decreasing numbers during the growing season because of a probable natural mortality. Bigelow and Welsh (1925) indicated that smelt, as schooling fish. Probably associate in age groups. This would affect the catch during the year. particularly in winter and spring‘when nets were on the lake bottom where smelt were likely in schools. The predominance of agfl-group I smelt in Gull Lake, excluding age-group 0, agreed with reports by Greene (1930) for Lake Champlain, Warfel 2§,£;, (1943) for Great Bay, New Hamp., and McKenzie (1964) for the Miramichi River. In these areas, as in Gull Lake, the number of 23 flaws.o Asvn.u Anuvo.w Aeuuvo.on Aouuvu.ne Auuvm.n deuce Asvo.~N Aemve.us hummuaoh A8063 magnum Auvo.n Auvo.n Auvo.n Aonvu.om meolooeo Cvm.n 3:6 33 «.9» 350. 5 3v «8 meamgoz A~v~.N Auavn.c~ Anuvn.nn Asvo.a~ menouoo vaa.~a Anon.- doaaouaom Anv~.e Amvn.- Aenvo.ch onn.m uasms< aavo.e asvo.e Auvo.m ALNVo.ew amen Aavo.on Advo.on an: asco.o Aoavo.mo sauna > >H man an H o sumo: amouwcom< 33093 5 33339: no ueolsz .323 exam :5 no commence-co one uuoouem .« edema 24 smelt decreased with age, with few individuals in age-group V. This indicated that the Gull Lake sample was representative and not affected by sampling. The mean total length (Table 3) and weight (Table 4) of Gull Lake smelt were computed. Age-group 0 smelt were collected during 3 months, so it was not possible to determine their total length or weight for one growing season. Increase in size of age-group I was best represented because active growth does not begin until early summer (McKenzie, 1958). ‘Age-group II smelt were not collected in January and February, so their size at the growing season and was not known. The mean size of age-group III fish captured in July provided an estimate of the size of II-group fish at the end of the previous season; the number of age-group III smelt were too few to draw conclusions concerning their growth. Fig. 6 represents the mean total length and weight of age groups of Gull Lake smelt. Differences between the mean lengths of age-groups I and II, and II and III were 23 and 24 mm. Differences between the mean weights of these groups was 5 and 12 g. The length and weight of Gull Lake smelt age groups were not comparable to either race of Lake Champlain smelt. Mean total length for age-group III of the small race was 158 mm; for age-group I of the large race it was 208 mm. Starting with age-group I, Lake Superior smelt were 152, 185, 200, and 218 mm (total length), and 21.7, 43.5, 59.1, and 71.5 g (Bailey, 1964). Females were even larger. Starting with age-group I, Miramichi River smelt averaged 166, 195, 220, 244, and 260 mm (total length), and 29.4, 50.8, 73.0, and 90.5 g (McKenzie, 1958). The length and weight of age-group I Gull Lake smelt were similar 25 33.03 :vséz 3892: 733:5: 32:63 Cuvuéo demos at n. a: Aeuv «£2 hues-woo..— Anv h. an g hues—nah. van.~ou Auvc.ood Auvo.~¢~ Acnvn.nn~ nonleuon 33.02 $3.03 35m. «3 359.3— 23.?» meant-oz Advc.no~ Auuv¢.~o~ Aan¢.e¢~ Anvs.¢o menouuo 3362 5a.: aces-use..." Anvh.oou vaa.~o~ Amnvo.co— onn.nn~ umaus< 2362 23.2: and: 2823“ 22. :3.ch :3. and he: auvo.n<~ Aouvo.cn~ uuun¢ > >H Hum Hm H o 50:62 anomwuow< muoxuoua an unsavu>uuflu no monamz .uumlm mama uufiU «0 Allv nausea umuou moo: .n edema 26 33.2“ :Vo.oe Suffice 72.80.: 323.3 CSeJ "coon. ahvh.ue Aemvm.hu mueauaom Anvm.en monsoon Auvc.me «www.ne Amvo.- Aonv~.n~ uoefiooon ASN.3 33.3 33.2: 350.3 2:”.— moon—252 asco.ne AaLc~.n~ “may“.oa aaca.m hapouoo van.o~ Amvw.o wooaeudom Anvo.nn vac.~d Aemvo.ou Acv¢.c~ unawa< 33.3 33.: 3:6». Cuvuéu 33. 3:83 33.3 as: amvs.wm aomvo.o~ sauna > >H man an H 0 name: amoumcow< 303093 5 33339: no noes-.2 .uumlm axe." Zoo no A3 undue: moo: .e e33. 26 It. 1! Gull Lake sue t: 1 length and weight of age groups 0 to a F180 60 Man 27 A09 PIG—m3 z woe.c an on n N >H nnn.o «s cm a as 0 Hum o~m.o an me no an an nun.o «0 on a «m be m m\z o~uem ARV mo~oaom ARV mono: uoaoaom acme: nsouwuow< £8. 33 2.6 no 6:8 x8 .o emcee 33 dominated in all groups except V. The preportion of females increased with age; the ratio of males to females was closest in age-group II. The total sex ratio was 168 females (61%) to 107 males (39%). This difference was highly significant (P(0.005). There was no evidence that the indicated sex ratio was influenced by smelt schooling according to sex. Most published data on sex ratios were based on spawning season samples. Hoover (1936) and Langlois (1935) showed that smelt sex ratios fluctuate during the breeding season and even during the spawning night. Beckman (1942) sampled Crystal Lake smelt before and after the spawning season and found 61% females and 39%.males, which agreed with the Gull Lake sample. Both Beckman.and McKenzie (1964) found an increase in abundance of females with.age. The sex ratio of Gull Lake smelt strengthened the conclusion that females were selected during spawning because of a mortality of males. The decreasing proportion.of males over the years would result from mortality during additional spawning seasons. There is one record (Michigan Conservation Department) of a large smelt kill in Gull Lake immediately after a spawning run (1960). The significant difference in age-group I could have resulted from a selective predation or a behavioral difference that made males less susceptible to capture. These factors could, of course, have been important at all ages. Size differences between the sexes of Gull Lake smelt were apparent (Table 7). Females were larger (mean length and weight) than males in each age group (Fig. 8). The smallest difference between sexes was in.sge-group II. Sexual dimorphism‘was most obvious and only significant (P<0.05) in age-group III, with a difference of 17.7 mm and 12.3 g. At this age smelt were sexually mature and had Soévmv 8:86.33 £33351... .- 34 n.3, «fen 3363 A3053 > min n.n< Anvoiom vaoéa >u .- 0.9 nén .- :Ceéod 3:3: H: mfiu n33 33.2.): :nang = 5.3 Q: Awmvanu Asynés H Amy nouns—om Amy mega: A35 eeueaom A65 mode: nsouwuomd 33223 5 33338. .3 .3382 .28. 33 :96 so 523an aqua-... .a .3: 34 Fig. 8. Sexual dimorphism of Gull Lake smelt. 35 Row PIG—m3 zcz mu m on m we us uoaouoo m cod nu mm cod uoaaouaom “o n we um unsws< em co co hush H cod me: o a“ an an on naadc nonesz uHoEm oeu~eoa~osoa asuoaoaao ecooeuuoo evodedoo euoocceuo coco: accouusooc no moccaceuu accouoa cu cocoouaxo uuaam oaeq “goo no coca .o oaaee 39 Fig. 9. Food of Gull Lake smelt expressed in percent frequency of occurrence 0 40 100 50 CLADOCERA COPEPODA OSTRACODA 50*- O lOO CHAOBORUS 01 O FREQUENCY OF OCCURRENCE oz.) O [OO- TENDIPEOIDAE 50- O I l I I I APR JUL AUG SEPT OCT NOV DEC JAN FEB MONTH kl size and age. The character and arrangement of their teeth indicate carnivorous propensities, while their comparatively close-set gill rakers suggest minute planktonic food at certain stages of growth (Kendall, 1927). The food habits of Gull Lake smelt were in part similar to those of smelt in other areas, for the smaller and younger fish fed primarily on small crustaceans and insect larvae. However, similarities ended as adult smelt in other environments changed to a fish diet . Crystal Lake smelt fed on the emerald shiner in fall and winter (Creaser, 1926; Beckman, 1942). Smelt in Mountain Lake, Mich., fed on young smelt and yellow perch in summer, in addition to an equal amount of copepods and insects (1929b). Doolittle (from Kendall, 1927) found smelt over 150 mm in Sebago Lake, Maine, eating fish and insect larvae while smaller smelt fed primarily on sooplankton. Host fish eaten were smelt of all sizes. Reports in Kendall (1927) stated that the prevailing genera of zooplankton prey were Cyclops, Diaptomus, Daphnia, and Bosmina. Greene (1930) found young and small race smelt in Lake Champlain feeding primarily on cladocerans and copepods. The prevailing genera were Cyclops, Dia tomus, Qgphnia, Bosmina, and Leptodora. The larger race smelt fed primarily on other smelt and insect larvae. Greene suggested that the fish-eating habit of the larger race is not a characteristic acquired by their larger size, but "...size for size, they take many more fishes than the smaller race, and conversely, far less plankton." He concluded that there must be a racial basis for a physiological difference in food preferences. Mayfly (Hexaggnia) larvae comprised most of the insect larvae eaten in Lake Champlain. In Saranac Lake Chaoborus and tendipedid larvae were prominent. 42 Great Lakes smelt are similar to those in Gull Lake because they do not feed extensively on fish. This was true in Lake Michigan (Creaser, l929a), Green Bay (Schneberger, 1937), and Lake Huron (Baldwin, 1950). Fish were not eaten in significant numbers; only the sculpin (gppppg), burbot (pppp_lppg), and smelt were taken occasionally. In Lake Huron young smelt fed primarily on Cyclopg and Diaptgpps (Gordon, 1961). Daphnia and Bosmina were taken in amounts varying with the season. Older smelt fed most frequently on crustaceans, and insects accounted for the greatest volume. Cladocerans and copepods comprised most of the crustaceans, except in November when smelt from deep water had eaten Mygig exclusively. Mayflies dominated the insect food. Dipterans were consumed only in July and August in pronounced numbers. Fish were eaten infrequently, usually shiners in summer and smelt in fall. Observations on the food of Gull Lake smelt agreed most closely with food habits of Great Lakes smelt. Fish of both areas fed mainly on crustaceans and insects, and more often consumed their own species rather than.other fish. With the exception of deep‘water Great Lakes fish that fed on,M sis, smelt from both areas depended on copepods, cladocerans, and dipteran larvae for the major portion of their diet. As in the Great Lakes, food habits of Gull Lake smelt varied with the season, suggesting that they were eating fauna most abundant at the time. In su-aer large pupating dipteran larvae were eaten extensively. Pupation of the larvae in super probably removed the largest individuals, so in the fall only smaller larvae were present. Total biomass may not have changed, but individual larvae would have been considerably smaller. Further, low temperatures may have affected the visual pigments of larvae, thus altering their migratory behavior. In the fall, therefore, 43 vertical migrations of smaller larvae were significantly less extensive. In the fall cladocerans and copepods appeared in the stomachs of smelt. The peak of copepod consumption occurred when cladocerans were eaten least frequently, which‘would have occurred if there was a fall and late winter bloom of copepods. Cladocerans could have bloomed in summer and early'winter. These data indicate that Gull Lake smelt fed on food most available, but may have had a preference for dipteran larvae since crustaceans were abundant at all depths in summer. When larger dipteran larvae disappeared and vertical migrations of the remaining larvae decreased, smelt turned to a crustacean diet dominated by the most abundant organism. That there were no crustaceans eaten in sunset indicates that smelt were seeking and eating individual dipteran larvae. This would not have been a great energy expenditure since large larvae were extremely abundant in the water column at night. In the fall smelt appeared to change to filter feeding for capturing crustaceans. This would explain the consumption of only larger genera; smaller animals and plants would have escaped through their gill rakers. There was no evidence that Gull Lake smelt were feeding on the bottom; there were no bottom dwelling amphipods or clams eaten, though these are often found in bottom feeding Great'Lakes smelt (Ferguson, 1965)- Low oxygen conditions in summer may have limited the distribution of these food organisms on a large portion of the bottom of Gull Lake. Gull Lake smelt did not prey on other fish even though they had the physiologic ability to do so. They were not generally as large as "fish eating" smelt in other areas, but larger individuals were comparable to smaller "fish eating" smelt. Smelt in Gull Lake were restricted to the colder hypolimnion in summer (Fig. 12), and could not associate with 44 minnows and young fish in warm surface waters. In fall when adult smelt were near surface, young of smelt and other species would have been too large to be preyed upon. The only time smelt are in the habitat of small fish is during Spawning when their consumption is limited. The formidable smelt dentition, which suggests fish capturing capabilities, evolved in the ocean where smelt feed on fish and relatively large fast-moving crustaceans (shrimp). The dependence of smelt on crustaceans and insects suggests that they compete with the young of all associated species and adults of many (Gordon, 1961). In order to adversely affect other species by competition, however, smelt would have to have a rate of consumption so great that a food scarcity develops. There has been no evidence that this has happened. In Saginaw Bay where smelt have become plentiful in recent years, yellow perch collected from 1943 to 1955 were heavier, length for length, than perch collected from 1929 to 1930 (El-Zarka, 1959). Feeding habits of Gull Lake smelt varied with the season (Table 10). Data from Tables 8 and 10 are plotted in Fig. 10. There were no smelt captured by gill net in daylight during the entire study period; evidently smelt could see and avoid the white net. It is probably safe to assume that most smelt were captured at night, and that feeding activity data for the year were comparable. Sampling methods must be considered in a comparison of feeding activity. From July to November smelt were captured at night and fixed every 2 or 3 hr, so values, particularly percent remains, were not comparable. The low water temperature during those months would have slowed the rate of digestion so values for percent remains may not have been too high. The February value was correct because these 45 o.- e.nh cum own money we “.8 a an maniacs n.m~ o.co~ n m huesmen ¢.n~ o.wh we on noneeoon «.mu m.en «u an noeao>oz o.e~ m.en mm ms nooouoo ~.o~ h.~n w mu neeaounom o.sm o.nw “e an ausws< «.ws 0.00 cm nu mush o.n~ o.on ~ N has m.o n.en 0 ma uuue< ARV enemausu eoou coon noessz some: modicum sums usoouem mums aeoasz :25 9.3 :5 no 322. 2:33 "Season .3 .2: 45 .usoam mace ease no nuance massage secomuom .os .w.a 46 mun. IFZOZ coo... 1:; 5mm...” .\. 25. can >02 ....00 Hamm 6:4 ..:... mad. . . q _ . _ _ -0 U/ ..\a Rh ’l/ \\.\\III \\\/ .\. o’ \ O— /\\.\ ll\\.\. // .\ . \ .. .\ / ..ON // . . .. .lx/ \ .. II..I..I. mzszm \ II v on . I . . lllll mmmzja... roasoem .\ \ ’ ..oe \ //\\ .’ .LNBOHBd 45 .u~oam exam masu uo nudes: wadeoou Hedomeom .os .m.m 46 mum ITPZOZ 25. omo >02 .60 Kum 32 42. Ea _ 1 q - - - _ \ho U/ ..\/ 1 [ll \‘llol/l \il/ .\. . \ O— /\\\\ I]\\.\ // .\ I. \ .. .\ I \ now // x. . x .. .II .. ll.l..l mzszm \ II I\ on o l o .. lllll mmmzjaa :oqsoem \.\ \ I .9. . l . .. coon. 5:5 .553 .\ /\ I. on cm 2. e8 om oo. .LNBOHBd 47 smelt were captured at night and frozen immediately. The highest proportion of stomachs with food occurred in July and January. The proportion decreased steadily from July to October. Feeding activity increased slightly in winter. Percent fullness was highest in July, and decreased through April. There was not an appreciable increase in amount of food eaten in winter; more fish may have been feeding but individuals were not eating more. Percent remains was least in summer when most stomachs were full. The proportion of digested material increased in fall, peaked in November and December, decreased slightly in January and February, and was greatest in April. These data indicate that smelt changed from feeding at night in summer to late afternoon or early evening in the fall. If smelt were all captured at night, the amount of digested matter would not have increased if they had continued to feed at night. The decrease in the number with food and stomach fullness was probably in part due to a decreased metabolism associated with lower temperatures, but mainly due to a change in the time of feeding. Further evidence that the peak feeding activity of smelt changed from night to late afternoon or early evening was obtained by evaluating diel feeding habits (Table ll). Results from July did not indicate a change in diel feeding activity. The percent of empty stomachs, fullness, and remains did not change significantly, although in early morning of 29 July smelt appeared to feed less. In August there was a definite change in.feeding activity from early evening to morning, which indicated that smelt were feeding most intensely in late evening or early morning. In October and November there was a strong indication that smelt were feeding in late afternoon or early evening. It was not possible to predict from these data when smelt were specifically feeding, but a 48 Table 11. Diel feeding habits of Gull Lake smelt Stomachs Stomach Date and time Number empty (2) fullness (Z) Remains (Z) 15 July 2115-0005 9 0.0 65.6 5.6 0005-0310 2 0.0 50.0 7.5 0310-0555 4 0.0 63.8 3.0 29 July 0845-0545 5 20.0 20.0 14.8 3‘August 2005-2240 17 11.8 17.6 15.7 2240-0730 13 0.0 65.4 7.5 9 August 2000-2200 8 37.5 13.8 54.0 2200-0005 10 10.0 26.5 9.4 0355-0730 9 22.2 28.9 18.6 11 October 1945-2240 4 0.0 30.0 70.0 2240-0035 8 50.0 12.5 55.0 0035-0230 11 27.2 19.5 48.8 0230-0430 8 62.5 3.1 85.0 11 November 2080-2230 6 16.7 36.7 60.0 2300-2400 10 20.0 30.0 71.3 49 change from night to an earlier feeding peak was indicated. That smelt were feeding only at night in summer was further confirmed by the fact that only dipteran larvae were eaten. Smelt could not have eaten dipterans during the day because the larvae inhabit the mud on the lake bottom at this time (Woodmansee and Grantham, 1961). In Gull Lake it appeared that once larger dipteran larvae decreased, night feeding offered no advantage, so smelt fed on crustaceans during the day. If they changed to filter feeding, they would have benefited from the light of day in locating aggregations of sooplankton. Small numbers of dipteran larvae were eaten in fall so there was still same feeding in early evening. This agreed with Ferguson (1965) who found Lake Erie sooplankton-eating smelt feeding most actively at dusk and dawn, with minimal feeding in early morning darkness. Little was done in Gull Lake to determine the role of smelt as food for game fish. A northern pike (§§2§_lggigg)‘weighing approximately 4 lb. had eaten two adult smelt. In a 5 year study of Gull Lake fish, Huver (personal communication) found that lake trout, yellow perch, smallmouth bass, largemouth bass, and rainbow trout had preyed on smelt. There is some evidence that smelt in other areas are eaten by the lake trout, yellow perch, northern pike, rock bass, largemouth bass, burbot, brook trout (Salvelinus fontinalis), lake whitefish (Cogggonus clupgaformis), walleye (Stizostedion vitreum vitreum), white perch (£3222§_americanus). Atlantic salmon (§£l!2.§£lé£)s sauger (Stisostedion canadense), and chain pickerel (Q5 £189.!) (Kendall, 1927; Greene, 1930; Dymond, 1944). Tharratt (1959) suggested that smelt may be the primary food of Lake Huron yellow perch. McCaig and Mullen (1960) studied fish in Quabbin Reservoir, Mass., after smelt were introduced, and found a 50 significant increase in the growth of lake and brown trout (Salmo trutta). This suggests that smelt may be important to the growth and success of some game fish as their individual size and abundance make them ideal prey. Bathymetric Distribution and Movement Temperature and dissolved oxygen were measured in Gull Lake during the period of gill netting operations (Fig. 11). The lake was thermally stratified in July and August, with little change in upper and lower temperature limits. The thermocline was between 6 and 15 m (11 and 24 C). In September the upper'waters cooled; in October the thermocline was less definite and decreased in depth until 4 November‘when.the lake was homothermous. In July dissolved oxygen was constant from surface to bottom (8.5 mg/l). In August high oxygen concentrations in the thermocline were due probably to increased phytoplankton production; oxygen decreased to a near critical level in deep water. In September and October there was little change in high oxygen levels at the surface, but there was little or no oxygen below 20 m until the lake "turned over" in late October. By 5 November oxygen levels were high (over 10 mg/l) and constant from surface to bottom. The bathymetric distribution of Gull Lake smelt was determined during 4 months of vertical gill netting (Table 12). These data were expressed as percent of total catch for each month (Fig. 12). smelt were captured primarily during darkness. In July and August almost all smelt were captured between 8 and 20 m. In August no smelt were taken.below 20 m. Smelt captured in October were evenly distributed from surface to bottom. In November smelt captures were weighted 50 Fig. 11. Gull Lake temperature (C) and dissolved oxygen (Mg/1) profiles. DE P TH (M) IO- 15- 25- 20- 51 3O l l5 OCT 1 l l l8 OCT 1 l ll OCT 4 NOV 1 I l O 5 IO IS 20 25 o 5 l0 TEMPERATURE (C)—OXYGEN (MG/L) I5 20 2.50 5 l0 IS 20 25 52 as on oo nu sauce an-o~ u n manna I a e~.- n o as a o~-a~ n a mu m o~-n~ m 0 ea n N.-a n e a-“ a m and aoaao>oz aoeouoo unease been Ase mason uses. «and menu eo namesa.uua.e oaauoamea-a masseuse .«I oases 52 .nucoa some now mouse uo success :. venues axe u I06 m exam Imso no noun spawns. o o—uuoeh muse Homo meow eNm emu...” ezmomma on 3.0 ommm o 83 0 8mm 0 >02 ... ...00 1 0:4 .. 435 m Nm ¢N (W) HidBO 52 .susos some you mouse no assumes :I common axe u mos m smog Iuso uo :qu somuumm o cannon» tune as: omsom .NI .mae on mm. o >02 1 ...zmommm Om mm 0 FOO fil I 00 mm 0 034 .. om mm 0 mm 4:... mN mm P ON ‘ m. N. m .. ¢ #0. (W) HidEO 54 toward the surface, although numbers were too few to draw conclusions. The observed distribution showed a predominance of smelt in midwater at night. No fish were collected below 28 m because the net was short of the bottom. Echo recordings obtained at night in July and August 8111313013304 the gill netting results and showed little smelt activity of any kind below 20 13. Although smelt are sometimes fetmd in warm water (Kendall, 1927), they are a cold water species confined to deeper water during su-er thermal stratification. In summer smelt have been recorded in lakes of New York below 12 :- (Odell, 1932), and in the Great Lakes from 18 to 34 m (Van Oosten, 1953) and 27 to 45 m (Gordon, 1963). They were found at a thermal preference of 12.8 C in Lake Champlain (Greene, ”30% 6.6 to 8.3 C in Cayuga Lake (Galligan, 1962). The bathymetric distribution of Gull Lake smelt appeared to be controlled by levels of dissolved oxygen, temperature, and their own feeding habits. Except for 11 October, smelt captured in the 4 months illustrated in Fig. 12 were collected under conditions of sufficient oxygen at all depths. On 11 October there was 0.0 mg/l below 27 m, but fish were counted on echo recordings at 26 m where oxygen was 0.8 mg/l. In September and early October there were no fish observed below 1.0 mg/l. As the 1.0 mg/l isopleth was depressed with fall turnover, fish moved downward with it. Pearse and Achtenberg (1921) and Bardach (1955) demonstrated that an oxygen deficiency limited the depth of yellow perch, and that fish are dependent on a minimum level of oxygen tension. It was seemed that Gull Lake smelt did not venture much below a certain minimum level of oxygen (1.0 mg/l). Since there was sufficient oxygen during the 4 months of vertical 55 gill netting, temperature and feeding were the two primary factors that influenced smelt bathymetric distribution. In July and August smelt were feeding extensively onlghggbggg§_and Tendipedidae larvae; larvae were observed by echo recorder concentrated between the surface and 20 m at night. This would explain why smelt captured at night were almost all above 20 m, but the fact that they were not above 8 m had to be attributed to a temperature (22 C) beyond which they would not pass. This suggested that temperature was a more critical factor than their feeding habits. Ferguson (1958) in laboratory experiments showed that temperature acting alone could control the distribution of fish, and suggested that a thermocline may act as a thermal or density barrier. He also presented evidence that although light, feeding activity, and social behavior may interfere with precise temperature selection, temperature is the main controlling factor of fish distribution when oxygen is not a consideration. Fry (1947) defined the preferred temperature of fish as the "region in an infinite range of temperature, at which a given population will congregate with more or less precision." He defined the temperature corresponding to the maximum point of an activity curve as a "true optimum, a temperature at which the internal economy of an organism can function best to give a particular response to a given stimulus." Fish, therefore, tend to congregate at a temperature at which, in terms of energy gain and loss, a given, activity can be most efficiently carried on. Smelt generally do not penetrate the thermocline and move into warmer'water. The advantage smelt in Gull Lake would have gained by moving into warmer'water to increase their feeding space, assuming they could have adapted to the temperature, would have been offset by an increased metabolism and energy loss. In October and November, when 56 the lake was not stratified and surface temperature was 14 C, smelt were collected at all depths. Smaller dipteran larvae were not concentrating in upper waters at that time. Smelt were not in upper waters for feeding purposes, but were feeding primarily on crustaceans at all depths. The vertical gill net was lifted every 2 or 3 hr from July to November to determine if there was vertical movement in the smelt population (Table 13). Since most smelt were captured in midwater in July and August, movement was not observed. From October to November the number of captured smelt was small, but a movement and dispersal downward at night was indicated. In early evening most smelt were captured in upper waters, but in early morning they were captured in deep water as well. Pennak's (1943) quartile curve method was applied to data obtained from the echo recordings between Points A and B (Fig. 2) to depict vertical movement of smelt. One modification of the method was made: to permit the determination of the first, second, and third quartile depths on the basis of percent ages in 4 m intervals, linear interpolation was employed so quartile curves could be tabulated to the nearest meter. The quartile depths were determined from a cumulative frequency column and plotted in Figs. 13 and 14. Pennak's method was devised for zooplankton migrations, but since a finite number of fish in a known volume of water was considered, it was considered a valid method for recording fish movement. A total of three dates were combined for the 24 hr transect series in Fig. 13. Although the dates were widely spaced, other echo recordings from this study showed that the trends illustrated were representative for those times at all dates. There was a marked diel variation in 57 Table 13. Diel bathymetric distribution of Gull Lake smelt Depth (m) Date and time 1-4 5-8 9-12 13-16 17-20 21-24 25-28 29-32 15 July 2115-0005 5 4 0005-0310 2 0310-0555 1 1 2 30 July 2310-0345 0345-0545 HF. N N 3 August 2 005-2240 1 0 5 2 2240-0740 8 8 10 9 August 2000-2200 2200-0005 5 y-e b0 11 October 1945-2240 2240-0035 0035-0230 0230-0430 l-‘Ht-‘H NNNH F‘HHF‘ rouge-0rd NHN y-e 18 October 1930-2130 2130-2330 1 0130-0730 1 1 pop-o 4 November 2230-0015 1 0015-0215 2 1 7 November 1730-0845 1 1 l 1 1 1 November 20304230 2 0030-0900 2 l 2 2 .xua oo~L-oooav nonouoo m~ «Au; OOQOoooofiv monso>oz one «an: oowanOmuv monsoooa ma «voucomouaeu one mouse omens 7 5 .om—ucsm use uomcsm ouocov nomad magnum docuuuo> .suuu uo someone so consumed moose momenm .monuoa o>uso oduuueso one he moueuumsgau smflu axed afino no accesses use souusnuuumum docuuuo> Home .md .wflm W (W) Hld30 TIME 58 .e ow semen mosouo~uoe sommxo .cmuu no cocoons so museums“ noose movesm .Aaonouoo Nduudv eczema o>uso ouuuueso on» an mouuuuusaau menu oxen “use no ucoao>oa use acnusnuuueue ~eo«uuo> down .3 .mE 59 \ \ \ m2_._. 0000 0000 00¢0 00m0 00N0 00.0 00¢N 00mm 00_N 000m . a . . p . . . . . 7// / 0N 0N .vN INN ION (W1Hld30 60 numbers of fish found within the transect (Table 14). Fig. 13 and Table 14 show that at 1300 hr the few fish (7) recorded were below 16 m. Numbers did not increase appreciably until 1800 hr. There was a vertical rise of fish until 1900 hr when 75% was above 12 m. The absence of fish in upper and lower waters during that period indicated they were moving simultaneously toward the surface. After 1900 hr fish increased and had an apparent downward movement and dispersal. By 2400 hr they were evenly distributed from surface to bottom, except for a greater density from 4 to 8 m that did not change until 0700 hr. At 0100 hr there was another rise that peaked at 0500 hr. The maximum number of fish was at 0600 hr. After 0600 hr numbers decreased and fish moved toward bottom; by 0900 hr 75% were below 20 m. Another sharp rise at 1000 hr resulted from a school in shallow water. They may have been smelt, but should not be considered because the rest of the fish were concentrated in deep water. It was probably a littoral school moving to deeper water. Because the three quartiles coincided in the evening and morning: all fish appeared to move simultaneously in one direction. AI: night a constant proportion remained above 6 to 8 m, while the remainder dispersed and moved downward. The results in Fig. 14 strengthened the previous data. Numbers of fish during this series (ll-12 October) were larger, but a similar movement was observed. From 2000 to 2300 hr there was a general movement downward ending at 0200 hr. The 507. quartile dropped from 9 to 14 m. The upper half appeared to move downward more than the lower half. The largest number was at 0200 hr; beyond 0200 hr the apparent numbers decreased while fish again moved toward the surface. The absence of fish on bottom was attributed to an oxygen deficiency. 61 Table 14. Number of fish in each sonar transect expressed as percent of maximum.number for one transect Date and time Number Percent of maximum 15 December 1300 7 3.2 1400 5 2.3 1500 9 4.1 1600 5 2.3 1700 8 3.7 1800 59 27.2 4-5 November 1900 46 21.2 2000 67 30.9 2100 61 28.1 2200 59 27.2 2300 87 40.1 2400 86 39.6 0100 98 45.2 0200 93 42.9 0300 140 64.5 0400 144 66.4 0500 133 61.3 0600 217 100.0 0700 136 62.7 0800 83 38.2 0900 36 16.6 15 October 1000 56 25.8 1100 55 25.3 1200 42 19.4 11-12 October 2000 195 49.8 2100 260 66.3 2300 232 59.2 2400 200 51.0 0100 338 86.2 0200 392 100.0 0300 372 94.9 0400 313 79.8 0500 347 88.5 0600 110 28.1 62 Work with echo recorders should be considered with reservation because of the difficulty of relating echo traces to species. In Gull Lake differences between individual traces and schools could not be distinguished. Conclusions concerning species recorded had to be drawn after considering the behavior and habitats of all species present, and by utilising other sampling methods when.traces were recorded. Of the major species present in Gull Lake, the logperch, bluntnose minnow, yellow bullhead, longnose gar, smallmouth bass, largemouth bass, rock bass, green sunfish, and bluegill were restricted to the littoral sons; the white sucker was restricted to the benthic zone; brook silversides and young bluegills were found in the upper limnetic region; and the cisco, rainbow trout, lake trout, and smelt were considered lower limmetic forms. The yellow perch was most abundant in shallow water but ranged into all parts of the lake (Greene, 1930; Odell, 1932; Lagler 23 51,. 1962). Fish associated with the upper limnetic (surface) region and those restricted to the littoral zone were not included in the analysis of fish movement because the 0 to 4 m interval, including the major acne of rooted vegetation, was omitted. If the littoral aone fish had been counted, thewaould not have contributed to an observed vertical movement because they‘were restricted to bottom. SCUBA operations at night revealed most littoral fish resting on bottom, so there was little movement in any direction. Benthic fish are restricted to bottom in deeper water, except during oxygen depletion. Their contribution to numbers of fish counted was lessened by excluding all fish traces below 28 m. Very few traces 'were recorded close to bottom at any depth, so benthic fish were probably scarce. 63 Ciscos exhibit diel vertical movements (Cahn, 1927). In Gull Lake, however, the few captured by fishermen and gill nets indicate the population is small. Rainbow and lake trout are not abundant in Gull Lake (Kalamazoo Gazette, 5 March 1967). These species do not reproduce successfully in lakes of Michigan (Taube and Bacon, 1952), so they are stocked annually. Hasler and Villemonte (1953) showed that physoclistous yellow perch cannot have an extensive vertical migration, and that at night they remained close to shore in contact with bottom in a state of "sleep". In Gull Lake yellow perch would not have been in midwater at night; they may have been recorded in daylight but would not have been moving vertically. The 1/2 inch gill net, which captured yellow perch when on bottom, was fished vertically at night at Station 5 (Fig. 1). In 4 months 138 (88.4%) Smolt. 16 (10.2%) yellow perch, and 2 (1.4%) ciscos were captured. In 8.7 hr the Isaacs-Kidd trawl captured 34 smelt, 1 brook silversides, and 1 yellow perch between 4 and 16 m, and 95 young bluegills between 5 and 10 m. Most bluegills were less than 20 mm, and were captured primarily during September before echo recordings were evaluated. These data and knowledge of the habits and behavior of other fish in Gull Lake provide strong evidence that most fish recorded were smelt, and that recorded vertical movements applied to the smelt population. Smelt are known to be schooling animals (Kendall, 1927). They were probably in deep water during the day and accustomed to seeing one another. As darkness approached they may have lost their tendency to school and moved independently in the water column, toward surface in the evening and back toward bottom at dawn. In approaching daylight they probably schooled again in deep water, leaving the transect free of individual fish. The fish recorded in Gull Lake, assumed to be mostly smelt, exhibited vertical movements similar to most other aquatic animals, which according to Clarke and Denton (1962) "...are thought to be deepest in the daytime, to ascend at dusk, to sink or become more scattered during the night, and to rise again at dawn before descending for the new day." They suggested that animals in the sea which find upper waters unsafe in the bright light of day, hide in the depths and move into food-rich upper waters at night, thus extending their feeding space and food source while reducing the effectiveness of predators. Clarke and Denton (1962) emphasized that vertical migrations of animals are more closely correlated with the strength and rate of change of penetrating daylight than any other stimulus (e.g. temperature and pressure). Clarke and Backus (1964) felt it was unlikely that temperature has an effect on vertical migrations. Clarke and Backus (1956) suggested that aquatic animals do not remain at or move with an.optimum illumination, but are stimulated to move vertically by a reduction in light. Once the stimulus is received, animals move steadily toward surface even though it means moving into higher intensities. This possibility“was referred to by McNaught and Hasler (1964). Brawn (1960) indicated that changes in light intensity'may have initiated diel changes in depth of Atlantic herring (glgpgg_harenggs harenggs). Richardson (from McNaught, 1961) concluded that diel vertical movements of Atlantic herring and sprat (glgpgg spgttus) were in direct response to light. Many studies on fish behavior show a close correlation between diel vertical movements of fish and their food. Welsh g§_gl. (1937) found pelagic fish in the Sargasso Sea moving vertically”with their 65 primary food (copepods). McNaught and Hasler (1961) suggested that white bass followed vertically moving aggregations of Daphnia by means of an acute sense of smell. Northcote £2.2l° (1964) found the peamouth (Mylocheilus gaggingg) moving‘with and feeding on rising insect larvae and cladocerans at night. The question arises as to whether fish anticipate the rise of sooplankton and move toward surface in response only to light, if they follow dense aggregations of sooplankton upward that are known to respond to light, or if they respond to other unknown or combination of factors. Ferguson (1965) found that the smelt in Lake Erie ascended from bottom in late afternoon, remained dispersed in midwater during the night, and descended abruptly to bottom at dawn. The evening ascent was accompanied by active feeding on sooplankton. The morning descent may or may not have been accompanied by sooplankton feeding. Since smelt were not feeding at night, Ferguson could not justify their movement to midwater on the basis of food procurement, but suggested light as the factor to which they'were responding. He felt that smelt were feeding‘while ascending and descending simply because they'were most active at those times, for as Swift (1964) has demonstrated for the brown trout, it ”feeds becuase it is active, and is.not active because it is feeding." In the fall the diel vertical movements and feeding activity of Gull Lake smelt were similar to those of Lake Erie smelt. It was shown that Gull Lake smelt were not feeding at night, probably because of an absence of larger dipteran larvae concentrating in upper waters in the fall. Since they were not feeding at night, they would not have been moving into upper waters for food procurement. It appeand that the only 66 factor to which smelt were responding was light. This observed behavior may have evolved for feeding in the summer months when great concentrations of pupating dipteran larvae migrate to upper waters at night. During these months smelt may have actively followed dense aggregations of larvae upward, but they were probably responding mainly to light since they continued their vertical movements in the fall. This migration behavior insured that smelt had access to an abundant supply of food where and when they were relatively safe from predators. The quartile curve method was used to measure horizontal movements toward or away from shore. Quartiles were determined from a horizontal cumulative frequency column. The position of each quartile was determined to the nearest 0.1 of an interval by linear interpolation. Fig. 15 illustrates the horizontal movement of fish for the 24 hr series considered in Figs 13. In the early afternoon fish'were concentrated near the east shore. In late afternoon there was a dispersal and movement to the west until 1700 hr'when 75%wwere concentrated in the‘west half of the transect. From 1800 hr until morning fish were evenly distributed, and fluctuated from east to west. There was a movement toward the east shore in the morning. Except for the period from 0700 to 0800 hr, the closeness of fit of the three quartiles suggests there was little movement toward or away from opposing shores by significant numbers of fish at any one time. In other words, fish did not appear to move from both shores outward in the evening at the same time, but seemed to shift either east or west together. Only between 0700 to 0800 hr was there a universal movement toward both shores. From 1000 to 1200 hr a majority returned back to the east shore, where they‘were first 66 At. 82-08: 8.330 2 2.5 88-08: uoaagoz ml» 2.2 82-03.: 3%88 mm «mousomoumou one mouse omens .ommuasw use nomasm ouocoo nomad vacuum usumouuuom .eonuos o>uso ouuuueso on» he voueuumsauu than oxeq muse no usoao>oe one tomusouuumuo Heumoumnon mean .3 .3m 67 Am V meIm ...mm? 23 mmOIm Pmdm lilijii llle ll‘ IIITT—I 3WIJ. 68 observed at 1300 hr the previous day. The same horizontal movement was demonstrated on 11-12 October (Fig. 16). Fish were concentrated toward the west shore the entire evening, and exhibited periodic east or west fluctuations. The fish again appeared to move as a whole; equal movement away from opposing shores was not observed. Between 0500 and 0600 hr there was universal movement toward both shores. These data suggest that although there was not a simultaneous movement away from both shores in the evening, there may have been a slight movement toward the shores at dawn. There was a possibility that fish moved out from the west shore in the evening, were concentrated more toward the middle at night, and returned to the east shore in the morning. Although movement toward or away from shore may not have been universal, schools may have dispersed in the evening and formed in the morning near either of the shores. Unfortunately, there were too few data to determine whether schools were forming and dispersing along just one shore, or if they were on both shores away from the echo recorder. 68 .Auoaouoo ~a-aav corpus o>u=o oumuuuso on» he nauseousdun than used Hfisu no unmamsoa one counseumuudm meusowuuo: demo .oa .wam 69- Efimozm emu; 2. mmorm 54m oomo Como ooeo oomo OONO cozy Doom comm OO_N OOON BWIL Summary and Conclusions The study of American smelt in Gull Lake revealed distinct relationships between aspects of their life history, trophic patterns, and bathymetric distribution and movement. The maximum size and mean length and weight of age groups of Gull Lake smelt were less than smelt in other environments. Younger fish were of comparable size, but differences increased with age. The physical environment may have adversely affected growth, but the primary influencing factor was apparently a difference in diets of mature fish. Smelt of similar size and genetic background in other areas changed their food habits with age and fed on large crustaceans (flxgig) and fish. Smelt in Gull Lake fed on small crustaceans and insect larvae, and did not feed on large crustaceans and fish at any time. hygig is not found in Gull Lake, but the absence of fish in their diet was probably the major factor responsible for their smaller size. 4A study of the bathymetric distribution of Gull Lake smelt demonstrated that they were restricted from feeding on fish. In.summer they were confined to the colder hypolimnion, and did not penetrate the thermocline to feed on young fish in.warm surface waters. By fall when adult smelt were in upper waters, potential prey had grown so they'were too large to be fed upon. Therefore, the mid-summer distribution of smelt and the kinds and distribution of other fish in the lake indirectly affected smelt growth rate. Percent age composition of Gull Lake smelt was related to natural mortality and selectivity of sampling gear. Few age-group III and older fish indicated a high mortality in the 4th year. Fewer mature males than females could have been caused by a mortality of males in 70 71 poor condition during spawning. Fewer immature males could have resulted from behavioral differences that made them more susceptible to predation or less susceptible to capture. A smaller male could have resulted from a higher metabolism. Gull Lake smelt fed on the most available food, but probably had a preference for large dipteran larvae. Pupating dipterans were eaten exclusively in summer when.crustaceans were also abundant. Data indicated a change from night to daylight feeding, accompanied by a change to filter feeding, when the composition of dipteran larvae populations changed so that only smaller larvae with limited vertical movements were present. Smelt in Gull Lake exhibited diel vertical movements that could have evolved from feeding on dipteran larvae in upper waters at night. However, it was most closely related to changes in light intensity because the migrations continued after smelt no longer fed on insect larvae. Life history data give insight into the population dynamics of smelt. Their rapid growth and relatively short lifespan emphasize the need for harvesting smelt after they first spawn (beginning of third growing season) and before natural mortality is too great (4th year)- Gear could be deve10ped to exclude younger smelt. Food habits studies suggest that smelt could be introduced as food for game and commercial species in cold deep lakes that strongly stratify in summer, without fear of competition for food or predation on valuable fry. Data on bathymetric distribution and movement indicates that smelt may be effectively captured using midwater trawls at night. This would extend the fishing season and reduce fluctuations in existing catches. 72 It is clear that additional work is needed before the role smelt play in the aquatic environment is completely understood, and before they can be effectively controlled and utilized. Literature Cited Bailey, M.M. 1966. Age, growth, maturity, and sex composition of the American.smelt, Osmerus aggggx, of western.Lake Superior. Amer. Fish. Soc., Trans. 93:382-395. Baldwin, N.S. 1950. The.American smelt, Osmerus aggggx (Mitchill), of South Bay, Manitoulin Island, Lake Huron. Amer. Fish. Soc., Trans. 78:176-180. Baldwin, N.S., and Raw- Saalfeld. 1962. Commercial fish production in the Great Lakes 1867-1960. Great Lakes Fish. Comm. Tech. Rep. 3. 166 p. Balls, R. 1951. Environmental changes in herring behavior: A theory of light avoidance, as suggested by echo-sounding observations in the North Sea. J. Cons. perm. int. Explor. Mar. 17:274-298. Bardach, J.E. 1955. Certain biological effects of thermocline shifts. Hydrobiologica 7:309-326. Baten, W.D., and P.I. Tack. 1952. Relationships of weight and body measurements of adult smelt, Osmerus m5 (Mitchill). Progr. Fish-Culturist 14:50-55. Beckman, w.c. 1962. Lengtheweight relationship, age, sex ratio and food habits of the smelt (Osmerus 595225) from Crystal Lake, Benzie Ceunty, Michigan. Copeia:120-124. Bigelow, 3.8., and V.V. Welsh. 1925. Fishes of the Gulf of Heine. 0.8. Bur. Fish. Bull. 40:1-567. Brawn, V5“. 1960. Sfieional and diurnal vertical distribution of herring (gm m L.) in Passamaquoddy Bay, 14.13. J. Fish. Res. Ed. Canada 17:699-711. 73 74 Cahn, A.R. 1927. An ecological study of southern.Wisconsin fishes. Illinois Biol. Monogr. 11:1-152. Carr, I.A. 1962. Distribution and seasonal movements of Saginaw Bay fishes. 0.5. Fish and Wildlife Service, Spec. Sci. Rep: Fish. 417. 13 p. Carr, I.A. 1964. Lake Erie fisheries explorations, May-November 1960. 0.8. Fish and Wildlife Service, Com. Fish. Rev. 26(4):1-8. Clarke, C.H.D. 1944. Gleanings from the natural history of Huron County. Canadian Field-Natur. 58:82-84. Clarke, G.L., and R.H. Backus. 1956. Measurements of light penetration in relation to vertical migration and records of luminescence of deep sea animals. Deep Sea Res. 4:1-14. Clarke, G.L., and R.H. Backus. 1964. Interrelations between the vertical migration of deep scattering layers, bioluminescence, and changes in daylight in the sea. Inst. ocefinogr. Menace Bull. 64:1-36. Clarke, G.L., and B.J. Benton. 1962. Light and animal life, p. 456 to 468. lgzM.N. Hill (ed.) The sea. Interscience Pub., New York. Creaser, c.w. 1926. The establishment of the Atlantic smelt in the upper ‘waters of the Great Lakes. Pep. Mich. Acad. Sci., Arts, and Lett. 5:405-424. Creaser, C.W. 1929a. The smelt in Lake Michigan. Science 69:623. Creaser, c.w. 1929b. The food of yearling smelt from Michigan. Pap. Mich..Acad. Scio. Arts, and Lett. 10:427-431. cu‘hlnsa D.H. 1952. fiche Surveys of fish. J. Cons. perm. int. Explor. Mar. 18:45-60. Cushing, D.H. 1954. Some echo-sounding experiments on fish. J. Cons. ”no int. ”Flore “to 203266-275e 75 Cushing, D.H. 1957. The interpretation of echo traces. Fish. Invest. Lend. Ser. 2, 17(4). 16 p. Cushing, D.H. 1963. The uses of echo sounding for fishermen. HMSO, London. 25 p. Cushing, D.1-1., and 1.0. Richardson. 1955. Echo sounding experiments on fish. Fish. Invest. Lend. Ser. 2, 21(4). 33 p. Dymond, J.R. 1944. Spread of the smelt (93333; $53195) in the Canadian waters of the Great Lakes. Canadian Field-Natur. 58:12-14. El-Zarka, S.E.D. 1959. Fluctuations in the population of yellow perch, Eggs; flavesgns (Mitchill), in Saginaw Bay, Lake Huron. U.S. Fish and Wildlife Service, Fish. Bull. 59:365-414. Ferguson, R.G. 1958. The preferred temperature of fish and their mid- su-er distribution in temperate lakes and streams. J. Fish. Res. Bd. Canada 15:607-624. Ferguson, R.G. 1965. Bathymetric distribution of American smelt, Osgrus m, in Lake Erie. Great Lakes Res. Div., Univ. Mich. Pub. 13:47-60. Ferguson, R.G., and HA. Regier. 1963. Selectivity of four trawl cod ends toward smelt. Amer. Fish. Soc., Trans. 92:125-131. Fry, F.E.J. 1947. Effects of the environment on animal activity. Univ. Toronto Stud. Biol. 55, Ont. Fish. Res. Lab. Pub. 68. 62 p. Galligan, J.P. 1962. Depth distribution of lake trout and associated species in Cayuga Lake, New York. 11.17. Fish and Game J. 9:44-68. Geode, 6.8. 1884. The fisheries and fishery industries of the United States. See. 1. Govt. Printing Office, Washington. 895 p. Gordon, W.G. 1961. Food of the American smelt in Saginaw Bay, Lake Huron. Amer. Fish. Soc., Trans. 90:439-443. 76 Gordon, W.G. 1963. A trawling survey of southern Lake Michigan (August- November 1960). U.S. Fish and Wildlife Service, Com. Fish. Rev. 25(2):l-6. Gordon, H.G., and A. Larson. 1965. Application of the echo sounder to Great Lakes fishery research. Great Lakes Res. Div., Univ. Mich. Pub. 13:61-68. Greene, c.w. 1930. The smelts of Lake Champlain, p. 105 to 129. _In_ A biological survey of the Champlain.watershed. Suppl. to the 19th Annu. Rep. N.Y. Cons. Dept. (1929). Hankinson, T.L., and C.L. Hubbs. 1922. The establishment of smelt in the Great Lakes waters. Copeia:57-59. Hasler, A.D., and J.R. Villemonte. 1953. Observations on the daily movements of fishes. Science 118:321-322. Hickling, C.F. 1935. The hake and the hake fishery. Edward Arnold, London. 142 p. Hodgson, H.C. 1950. Echo-sounding and the pelagic fisheries. Fish. Invest. Lond. Ser. 2, 17(4). 16 p. Hoover, 8.8. 1936. The spawning activities of freshwater smelt with special reference to the sex ratio. Copeia:85-9l. Hutchinson, G.E., and H. Lgffler. 1956. The thermal classification of lakes. Proc. Net..Acad. Sci., Hash. 42:84-86. Johnson, v.3. 1961. Aspects of the ecology of a pelagic, sooplankton- eating fish. Verh. int. Ver. Limnol. 14:727-731. Jones, F.R.H., and G. Pearce. 1958. Acoustic reflection experiments with porch to determine the proportion of echo returned by the swim bladder. Exper. Biol. 35:437-450. Kendall, H.C. 1927. The smelts. U.S. Bur. Fish. Bull. 42:217-375. 77 Lagler, K.F. 1952. Freshwater fishery biology. Wm. C. Brown Co., Dubuque, Iowa. 421 p. Lagler, K.F., J.E. Bardach, and R.R. Miller. 1962. Ichthyology. Join: Wiley and Sons, Inc., New York. 545 p. Langlois, T.H. 1935. Notes on the spawning habits of the Atlantic smelt. Copeia:141-142. Lievense, S.J. 1954. Spawning of American smelt, Osmerus 5211135, in Crystal Lake, Bensie County, Michigan. Copeia:232-233. Manzer, J.I. 1964. Preliminary observations on the vertical distrib- ution of Pacific salmon (Genus Oncorhmchus) in the Gulf of Alaska. J. Fish. Res. Bd. Canada 21:891-903. McCaig, R.S., and J.R. Mullen. 1960. Growth of eight species of fishes in Quabbin Reservoir, Massachusetts, in relation to age of reservoir and introduction of smelt. Amer. Fish. Soc., Trans. 89:27-31. McKenzie, R.A. 1943. The effect of box-net mesh on the catch of smelt. Fish. Res. Bd. Canada, Atlantic Progr. Rep. 34:7-9. McKenzie, R.A. 1944. Stream clearance for spawning smelt. Fish. Res. Bd. Canada, Atlantic Progr. Rep. 35:24-25. McKenzie, R.A. 1946. The smelt fishery of northeastern New Brunswick. Fish. Res. Bd. Canada Bull. 70. 20 p. McKenzie, R.A. 1947. The effect of crowding smelt eggs on the production of larvae. Fish. Res. Bd. Canada, Atlantic Progr. Rep. 39:11-13. McKenzie, R.A. 1948. A new celluloid cpercular tag. Amer. Fish. Soc., Trans. 78:114-116. McKenzie, R.A. 1956. Difficulties in grading smelt. Fish. Res. Bd. WC. Atlmtic Progr. RCPe 65:17.20e 78 McKenzie, R.A. 1958. Age and growth of smelt, Osmerus mordax (Mitchill), of the Miramichi River, New Brunswick. J. Fish. Res. Bd. Canada 15:1313-1327. McKenzie, R.A. 1964. Smelt life history and fishery in the Miramichi River, New Brunswick. Fish. Res. Bd. Canada Bull. 144. 77 p. McNaught, D.C., and A.D. Hasler. 1961. Surface schooling and feeding behavior in the white bass, _R_g<_=_c_:_u_s_ chrysogs (Raf inesque), in Lake Mendota. Limology and Oceanography 6:53-60. McNaught, D.C., and A.D. Hasler. 1964. Rate of movement of populations of Daphnia in relation to changes in light intensity. J. Fish. Res. Bd. Canada 21:291-318. Northcote, T.G., H.W. Lors, and J.C. MacLeod. 1964. Studies on diel vertical movements of fishes in a British Columbia Lake. Verh. int. Ver. Limol. 15:940-946. Odell, T.T. 1932. The depth distribution of certain species of fish in some of the lakes of How York. Amer. Fish. Soc., Trans. 62:331-335. Pearse, A.S., and H. Achtenberg. 1921. Habits of yellow perch in Wisconsin lakes. U.S. Bur. Fish. Bull. 36:293-366. Pennak, R.W. 1943. An effective method of diagraming diurnal movements of zooplankton organisms. Ecology 24:405-407. Richardson, LB. 1952. Some reactions of pelagic fish to light as recorded by echo-sounding. Fish. Invest. Lend. Ser. 2, 18(1). 20 p. Rothschild, B.J. 1961. Production and survival of the eggs of the American smelt, Osmergg mordax (Mitchill), in Maine. Amer. Fish. Soc., Trme 903‘20480 Rupp, R.S. 1959. Variation in the life history of the American smelt in inland waters of Maine. Amer. Fich. Soc., Trans. 88:241-252. 79 Rupp, R.S. 1965. Shore-spawning and survival of eggs of the American smelt. Amer. Fish. Soc., Trans. 94:160-168. Rupp, R.S., and M.A. Redmond. 1966. Transfer studies of ecologic and genetic variations in the.American smelt. BcoIOgy 47:253-259. Sand, R.F., and W.G. Gordon. 1960. Exploratory fishing in Lake Erie, September 1958-November 1959. 0.5. Fish and Wildlife Service, Com. Fish. Rev. 22(6):1-12. Savage, J. 1935. Smelts in the Canadian‘waters of Lake Huron. Copeia:194. Schaefers, R.A., and 0.3. Powell. 1958. Correlation of midwater trawl catches with echo recordings in the northeastern Pacific. U.S. Fish and unant- Service, Com. Fish. Rev. 20(2):7-15. Schneberger, E. 1937. The biological and economic importance of the smelt in Green.Bay. Amer. Fish. Soc., Trans. 66:139-142. Stevenson, J.C. 1944. The smelt situation in the upper Great Lakes, Ontario, May 1943. Canadian Field-Natur. 58:128-129. Swanson, A.A. 1955. Winter smelt fishing out of Escanaba, Michigan. U.S. Fish and Wildlife Service, Com. Fish. Rev. 17(8):6-8. Swift, D.R. 1964. Activity cycles in the brown trout (m m L.). J. Fish. Res. Bd. Canada 21:133-138. Taube, C.M., and E.H. Bacon. 1952. Gull Lake. Inst. Fish. Res., Mich. Dept. Cons. Lake Inventory Summary 2. Tharratt, R.C. 1959. Food of yellow perch, 23:23 flavescens (Mitchill), in Saginaw Bay, Lake Huron. Amer. Fish. Soc., Trans. 88:330-331. Valdez, V., and D.B. Cushing. 1966. The diurnal variations in depth and quantity of echo traces and their distribution in area in the southern Bight of the North Sea. J. Cons. perm. int. Explor. Her. 30:237-254. 80 Van.Oosten, J. 1937. The dispersal of smelt, Osmerus £25925 (Mitchill), in the Great Lakes region. Amer. Fish. Soc., Trans. 66:160-171. Van Oosten, J. 1947. Mortality of smelt, Osmerus £25335 (Mitchill), in Lakes Huron and Michigan during the fall and winter of 1942-43. Amer. Fish. Soc., Trans. 74:310-337. Van Oosten, J. 1953. The smelt, Osmerus 595235. Mich. Dept. Cons., Fish Div. Pamphlet 8. 13 p. Warfel, H.E., T.P. Frost, and W.H. Jones. 1943. The smelt, Osmerus 595955, in Great Bay, New Hampshire. Amer. Fish. Soc., Trans. 72:257-262. Welsh, J.H., F.A. Chace, and R.F. Nunnemacher. 1937. The diurnal migration of deep*water animals. Biol. Bull. 73:185-196. Woodmansee, R.A., and B.J. Grantham. 1961. Diel vertical migrations of two sooplankters (Mesocyclopg and Chaoborus) in a Mississippi lake. Ecology 42:619-628. Zilliox, R.G., and W.D. Youngs. 1958. Further studies of the smelt in Lake Champlain. N.Y. Fish and Game J. 5:164-174. Appendix The beam of sound from the Furuno echo sounder gave an uninterrupted picture of fish distribution within the Gull Lake transect (Fig. 2) because of the slow boat speed and the high number of soundings per min. The methods utilized were determined to be valid after experimentation with empty glass vials. The sound pulse is reflected by any plane that represents a boundary between two media, and the strength of reflection is determined by the difference in density and speed of sound in the original and new medium (Midttun, U. S. Bureau of Commercial Fisheries, unpublished report). Studies on acoustic characteristics of fish demonstrate that, although the swim bladder occupies only 5 to 8% by volume in fresh-water species, it accounts for 40 to 80% of the signal reflected back. This is explained by the fact that acoustic reflectivity of fish flesh is about 4.6%, bone about 26%, and air 100% (Jones and Pearce, 1958; Cushing and Richardson, 1955). The appearance of the trace is dependent primarily on the size of the swim bladder. Air filled glass vials comparable in size to swim bladders gave approximately the same echo response as fish. A glass vial (60 by 7 mm) was fastened to a weighted line so the vial rested horizontally like the swim bladder of a fish. The vial was suspended at various depths and scanned with the echo sounder. At 30 m and gain 4, the vial returned an echo comparable to traces that were counted as fish, at that depth and gain. It was assumed, therefore, that all smelt were recorded at all depths at that gain, since the surface area of air bladders of age-group I smelt were comparable in size to the plane surface area of the vial. Age-group 0 smelt are usually associated with the surface; at those depths an echo would have 81 82 been reflected without difficulty. At a high gain an individual Chaoborus, because of its hydrostatic organs, is recorded as strongly as a fish echo. After experimentation it was decided that at gain 4 Chaoborus larvae were not recorded; only random strong echoes assumed to be fish remained. Sound energy was sufficient to return an echo from all fish at all depths within the half-sound pressure angle because of the narrow sound beam. Due to the directivity of sound from the transducer face, however, weaker sound waves were emitted over an area larger than the half-sound pressure angle, particularly in upper waters. That the sounder was on gain 4 tended to reduce this effect, but values for fish numbers in upper waters may have been too high. This was lessened by not including the 0 to 4 m interval. The scattering of sound from the transducer face was of some value because it probably recorded fish between the main sound pulses. At 500 rpm the boat traveled 1.1 ft between pulses, at 1000 rpm it traveled 2.0 ft, which.meant that a triangle of water 8 ft and 15.3 ft deep respectively was not included within the half-sound pressure pulses. However, fish in this water were probably reached by sound waves outside the half-sound pressure beam. HICHIGQN STQTE UNIV. LIBRQRIES ll IIHII ‘H 1 312930083 6733