Distributional Ecology of the Cisco (Coregonus Artedii) in Indiana'

Distributional Ecology of the Cisco (Coregonus artedii) in Indiana'

DAVID G. FREY

Indiana University

ABSTRACT

The cisco is known with considerable certainty to occur in 41 lakes of northern Indiana, of which Shriner Lake at a latitude of 41° 14' N. is the southernmost known natural occurrence of the family Coregonidae. There are three and possibly more instances where lakes with cisco populations several decades ago no longer contain the species, presumably as a result of altered limnological conditions. The few recorded attempts at stocking the cisco in lakes in which they did not otherwise occur have apparently been quite uniformly unsuccessful, and it is concluded that the pre- Columbian distribution has not been extended by man's activities.

The species generally occurs in the deep water of a lake in summer. As oxygen conditions become progressively more severe, the population is forced into the upper part of the hypolimnion, and finally into the lower or even the upper part of the thermocline. It is assumed in this paper that the species can tolerate water temperatures as high as 20° C., and an oxygen content as low as 3 ppm., or possibly even slightly • lower. As the species is forced progressively higher in the lake it comes into increas- ingly greater content with species of fish commonly occurring in the epilimnion, and may even be preyed upon by the bowfin and the northern pike.

The summer bathymetric distribution of the species in several Indiana lakes was determined by means of gill nets, and in the other lakes it was inferred from the extremes of temperature-oxygen stratification of summer. Data available for 37 cisco lakes as compared with 53 non-cisco lakes demonstrate that in general the cisco lakes have a thicker stratum of water with temperatures below 20° and an oxygen content greater than 3 ppm. The shallower cisco lakes tend to have an oxygen maximum in the thermocline, and in one lake such a photosynthetic maximum has occurred regularly over four successive summers. This type of oxygen distribution is believed significant in enabling the species to survive in small and shallow lakes.

The facts that the cisco lakes in Indiana occur in definite clusters and that these lakes tend to have a different type of oxygen distribution than the non-cisco lakes point to the importance of regional geologic and geomorphic controls in the limnology of the lakes and consequently in the persistence of cisco populations in them.

Contribution No. 585 from the Department of Zoology, Indiana University. The field studies reported on here were financed largely by the Indiana Department of Con- servation. The assistance of the following temporary employees of the Indiana Lake and Stream Survey is gratefully acknowledged: Edwin B. Davidoff, Christopher Davis, William R. Eberly, Allan E. Johnson, Harold R. McReynolds, George Naryshkin, Russell Noyes, Jr., William D. Ploughe, Thomas W. Pape, William H. Schleicher, Keith V. Slack, Tom A. Stombaugh, and Richard B. Williams. INVEST. INDIANA LAKES & STREAMS VOL. IV, No. 7. OCTOBER 1955.

177 178

TABLE OF CONTENTS Page Introduction ...... 178 Distribution of the cisco in Indiana ...... 180 Indiana cisco chronology ...... 180 Additional records of ciscoes in Indiana ...... 183 Distribution of cisco lakes by major drainages ...... 186 Artificial stocking of ciscoes 188 Factors governing the summer bathymetric distribution ...... 191 Influence of age on oxygen-temperature requirements ...... 196 Summer bathymetric distribution in Indiana lakes ...... 198 A. Indian Village chain of lakes (Noble Co.) ...... 199 ( A) 1. Indian Village Lake ...... 199 (A) 2. Hindman Lake ...... 201 (A) 3. Gordy Lake ...... 201 (A) 4. Knapp Lake ...... 201 5. Tippecanoe Lake (Kosciusko Co.) 202 6. Oliver Lake (LaGrange Co.) ...... 203 7. Myers Lake (Marshall Co.) ...... 203 Comparison of temperature and oxygen stratification in lakes with and without ciscoes 207 Types of oxygen distribution in cisco and non-cisco lakes ...... 212 Constancy of oxygen typology during given summer ...... 212 Oliver Lake ...... 212 Myers Lake ...... 213 Little Tippecanoe Lake ...... 215 James Lake ...... 217 Oxygen type-frequency distribution ...... 217 Relation of oxygen typology to summer surfacing and mortality 221 Discussion and conclusions ...... 222 Literature cited ...... 225 INTRODUCTION The coregonid fishes, which are closely allied to the trouts and sometimes are included in the same family with them, occur widely distributed in the lake districts of the northern portions of both North America and Eurasia. Except for its absence from Greenland, the group is continuously circum-boreal in distribution. The southernmost latitudinal limit of the natural distribution of the family is reached in the lake district of northern Indiana (Meek 1916:136) and in fact, the southernmost known natural occurrence of the family is in Shriner Lake in Whitley County, Indiana, at a latitude of 41° 14' N. The various types of coregonids are so plastic in their phenotypic expression that almost every lake has its own particular race, nation or subspecies of the species concerned. The most widely distributed coregonid in the glacial lakes of the northern United States and Canada is the cisco, Core gonus artedii Le Sueur. This species extends through the glaciated portion of the United States at least from New York to Minnesota, and in Canada it extends northward to Hudson Bay and environs (Dymond 1933) and northwestward into the Yukon and Northwest Territories (Dymond 1943). This is a pelagic, plankton-feeding species with a terminal mouth, 179 belonging to the subgenus Argyrosomus of the European terminology. Like the other members of the genus it has been considered to be a cold stenotherm requiring a fairly high level of dissolved oxygen, although its oxygen requirement is apparently less demanding than in the whitefishes. The present study, however, demonstrates that local cisco populations can be much more tolerant of high temperatures and low concentrations of oxygen than is commonly assumed. The phenotypic plasticity of the cisco is demonstrated by the fact that Koelz (1931) recognized 24 groups (subspecies) within this one species, although Hile (1936b and 1937) later concluded that there was insufficient basis for establishing these infra-species categories. Moreover, as Hile pointed out, there is a considerable environmental control of phenotypic expression, so that, for example, different year classes within a single body of water are able to differ significantly from one another in various morphometric characters. The present study is concerned particularly with the distribution of the cisco, and the factors controlling its distribution. If we assume for North America, as Thienemann (1950) has done for several of the coregonids in northern Europe, that in late-glacial, or at least by early post- glacial, time all the glacial lakes (including ice-block basins without out- lets?) in the appropriate watersheds contained ciscoes, then the absence of the cisco from particular lakes must have resulted from the development of conditions unsuitable for the continued survival of the species. Thienemann (1918) concluded for Coregonus maraena in northern Ger- many that the species was eliminated from certain of these lakes as the oxygen content of the hypolimnion declined to certain minimum values, accompanying the natural typological aging of the lakes. Willer (1924) reached similar conclusions regarding Coregonus albula in East Prussia. A few lakes which did not naturally have coregonids now have vig- orous populations of various species established through artificial stocking. This is true of all the large, deep lakes along the southern slope of the Alps, and also of a very few lakes in northern Germany. The great bulk of the attempted plantings in this latter region, however, were unsuccessful. The ability of these fishes to survive and maintain their populations in a lake, particularly in one in which the epilimnion reaches temperatures too high for the various species, is regarded as an indication of the presence of adequate levels of dissolved oxygen in the hypolimnion throughout the summer. Indeed, in northern Germany the lakes containing the whitefish-type coregonids (particularly Coregonus maraena) be- long to the early stages of the profundal midge succession. Coregonus maraena and Tendipes plumosus are mutually exclusive. On the other hand, Coregonus albula, which is very closely related to our cisco in mor- phology and physiology, apparently reaches its maximum abundance during the mesotrophic stage of lake development (Willer 1924). So far as known, Indiana does not have more than one or two lakes in which the hypolimnial oxygen levels remain fairly high throughout the 180 summer, at least consistently so from one year to another. Usually there is a very marked depletion. Yet the cisco in Indiana is known to occur in lakes having a surface area as small as 2 ha., or a maximum depth as small as 6 m. Such morphometry must certainly provide environmental conditions near the survival limit for the species. Because of the great limnological interest in the factors controlling the distribution of the various species of coregonids, it was decided to initiate a study on the particular species occurring in the inland lakes of Indiana.

DISTRIBUTION OF THE CISCO IN INDIANA Reference to the cisco of Indiana has been in the literature since 1875 when Jordan described the population occurring in Tippecanoe Lake as a separate species, Argyrosomus sisco. Subsequent references have listed additional lakes in which the species occurred, described the growth rates of particular populations, or discussed the synonymy. Koelz (1931) listed the Indiana cisco as subspecies sisco of Leucichthys artedi, characterized by a relatively short head, snout, maxillary, and paired fins. However, the number of specimens from Indiana lakes available to him for ex- amination was so small that definitive conclusions could not be reached. An annotated listing of the Indiana cisco chronology follows below. The complete literature citations are given in the bibliography. Indiana Cisco Chronology 1875a. Jordan. Specimens from Tippecanoe Lake collected about Nov. 25, 1875, were described as a new species, Argyrosomus sisco, thought to be distinct from the related species in Lake Michigan. Reported to occur also in nearby Barber Lake (now known as the Barbee Chain), although not known to occur elsewhere in Indiana. Apparently identical with the cisco of southern Wisconsin, known to occur in lakes Geneva and Mendota. 1875b. Jordan. Same material as in Jordan 1875 a. No new records. 1875c. Jordan. Mentions Argyrosomus sisco as one of four species occurring in the State (the other three only in Lake Michigan). 1876. Levette. Argyrosomus sisco "abounds" in all the lakes tributary to the Tippecanoe River. Also occurs in great numbers in Lake James (Steuben Co.), where large numbers are caught annually during the spawning period. 1877. Jordan. Fifteen specimens listed from Region V, comprising Lake Manitou in Fulton Co., and lakes Center and Tippecanoe in Kos- ciusko Co. No closer locality information is given. An attempt to locate these specimens in various museum collections was unsuccessful. Possibly the specimens were destroyed during the disastrous fire in Owen Hall at Indiana University in 1883. This citation is the earliest known reference to the possible occurrence of ciscoes in Manitou Lake. Blatchley and Ashley (1901) definitely list the cisco from this lake on the authority of a Dr. Vernon Gould. Gerking 181

(1945) erroneously ascribed the latter record to Eigenmann and Beeson (1894). It is not too likely that the specimens concerned came from this lake, and it is certain that none of the specimens came from Center Lake, because of the extreme summer oxygen depletion in this body of water. Most probably the specimens came from Tippecanoe. 1878. Jordan. Core gonus artedi var. sisco listed as occurring in "Lake Tippecanoe, etc." 1886. Jordan and Everman. List the cisco as occurring "in the deep lakes of Northern Indiana (Tippecanoe, Eagle [= Winona], Maxin- kuckee, etc.) . . ." This reference to the presence of ciscoes in Winona and Maxinkuckee is surprising coming from Jordan and Evermann, because there is no tangible evidence that the species has occurred in these two lakes in modern time. 1887. McDonald. Quotes a letter from F. M. Baker of Rome City, In- diana, that ciscoes occur "in some of our lakes in Kosciusko and Noble Counties." This is the first reference to the occurrence of the cisco in Noble County. Although the lakes are not specified, it is likely that the Indian Village group is being referred to. 1892. Jordan. This paper is identical with Jordan (1875b), except that the figure is omitted. 1894b. Kirsch. Three specimens from Shriner Lake in Whitley Co. Re- ported to be common in nearby Cedar Lake, but not present in nearby Round Lake or in Blue Lake three miles to the east. Common in nearby Crooked Lake, and present in small numbers in the west end of Big Lake ( Noble Co.). These two latter lakes are at the head of the Tippecanoe Drainage, just over the divide from Cedar, Shriner, and Round Lakes, which are collectively known as Tri Lakes. 1894. Eigenmann and Beeson. List the records cited by Kirsch. Also quote a paper ascribed to Jordan in 1877, reporting ciscoes as occurring in lakes Eagle [= Winona] and Maxinkuckee, but in the supposedly complete bibliography of Indiana fishes in this same volume there is no paper by Jordan containing these references. The only known reference to the occurrence of ciscoes in these two lakes is that by Jordan and Evermann (1886). 1895. Kirsch. Cisco not listed from any lakes of the Maumee Drainage, except Lake Erie. Fish Lake [-= Hamilton Lake] in Steuben Co. and lakes Indian and Cedar in Dekalb Co. are specifically mentioned. 1896b. Kirsch. More or less repeats information contained in Kirsch (1894b). Emphasizes that ciscoes do not occur in Round Lake. Ciscoes in Cedar Lake reported to be smaller than in Shriner Lake. 1896. Evermann and Smith. Ciscoes reported as occurring only in Tip- pecanoe, Crooked, Shriner, and Cedar Lakes in Indiana. Measure- ments are given on four specimens from Crooked Lake (Whitley Co.). 182

1901. Blatchley and Ashley. List cisco as occurring in Blue River Lake [ , Blue Lake] in Whitley Co. on the basis of a collection made by Kirsch in 1892, even though Kirsch (1894) definitely states that the cisco does not occur in Blue Lake. This is undoubtedly an erroneous reference. Other valid records by Kirsch for lakes in Whitley and Noble counties are cited. The cisco is also listed as occurring in Lake Manitou (Fulton Co.) on the authority of a Dr. Vernon Gould. Another list cited here of 19 species collected in the lake by Eigen- mann and Norman does not contain the cisco, and in light of other known instances of Blatchley's uncritical and sometimes erroneous references to the literature, the record for Manitou must be regarded with considerable suspicion. 1902. Blatchley. Repeats the list of 19 spp. of fishes "known" to occur in Blue River Lake as cited in the 1901 paper by Blatchley and Ashley. 1902. Hay. In Indiana, the cisco is listed as occurring in the Tippecanoe River and its tributary lakes, and in lakes of the Eel River system. 1908. Meek. Cisco merely reported to occur in lakes of northern Indiana and southern Wisconsin, without any specific localities being mentioned. 1911. Jordan and Evermann. Cisco known to occur in lakes Tippecanoe, Barber, Crooked, Shriner, Twin, Cedar, and James. It is not certain whether the Twin Lakes listed is the group in Marshall Co. or in LaGrange Co., although the latter is more likely. Three specimens from James Lake (Steuben Co.) collected by W. S. Blatchley are recorded. 1913. Lower. This is one of the most significant studies on the cisco of Indiana. Lower lists the species as occurring in Tippecanoe, Barbee, Crooked, Shriner, Twin Lakes, James, Dallas, Royer, Long, Lake of the Woods, and Little Balls (location ?). Has two photographs of three specimens from Lake of the Woods. Gives measurements on specimens from Lake of the Woods, McLish, and Royer Lakes. Describes late summer mortality of the cisco in Long Lake. 1915b. Miles. Several ciscoes planted in Crooked Lake (Steuben Co.), but none known to have been subsequently seen or caught. 1920. Evermann and Clark. Attempted unsuccessfully to locate ciscoes in Maxinkuckee with gill nets in July of 1899. Concluded the species was not present. 1931. Hile. Caught 62 ciscoes in Indian Village Lake, 24 in Gordy, and 11 in Hindman, all of which are small lakes in the Indian Village chain in southwestern Noble Co. Hile studied the age and growth of these populations. 1931. Scott. Reports ciscoes from lakes Snow, James, Indian Village, and Tippecanoe. Discusses the relation of the oxygen notch in the thermocline to the surfacing of ciscoes in Snow and James lakes in September. 183

1931. Koelz. Examined four specimens from Gordy Lake, three from Indian Village, one from James, and two from Tippecanoe. Clas- sified them as subspecies si,sco of Leucichthys artedi, although the small number of specimens prevented a final decision on the matter. They seemed to be closer to subsp. artedi. 1938. Blatchley. Lists the species as occurring only in Tippecanoe, Shriner, and Cedar lakes, and in the Tippecanoe River. Obviously he had not searched the literature very thoroughly. 1945. Gerking. Reports ciscoes in the Twin Lake group (Lawrence and Myers lakes) in Marshall Co. and in Clear and Gage lakes (Steuben Co.), in addition to records already in the literature. The occurrence of the species in Lake Manitou is erroneously ascribed to Eigenmann and Beeson (1894), and in Oswego Lake to Jordan 1875b).

TABLE 1. Known and reported occurrences of the cisco in Indiana lakes through 1945. The original source of each record is listed.

Lake County Source I. Based on actual specimens: Tippecanoe Kosciusko Jordan 1875a Shriner Whitley Kirsch 1894 Crooked Pl Evermann and Smith 1896 James Steuben Jordan and Evermann 1911 Royer LaGrange Lower 1913 Lake of the Woods PP McLish Steuben Indian Village Noble Nile 1931 Gordy Hindman II. Reported to occur: Barber [= Barbee] Kosciusko Jordan 1875a James Steuben Levette 1875 Eagle [= Winona] Kosciusko Jordan and Evermann 1886 Maxinkuckee Marshall Cedar Whitley Kirsch 1894 Crooked >, Big Noble Manitou Fulton Blatchley and Ashley 1901 Twin LaGrange ? Jordan and Evermann 1911 Dallas LaGrange Lower 1913 Long Little Balls 21 Snow Steuben Scott 1931 Lawrence & Myers Marshall Gerking 1945 Clear Steuben Gage PI III. Questionable reports: Blue River [= Blue] Whitley Blatchley 1902 Crooked Steuben Miles 1915 Oswego Kosciusko Gerking 1945 (erroneously citing Jordan 1875b) 184 Table 1 summarizes the distribution of the cisco in Indiana lakes, as reported in the literature through 1945. Specimens of fish from ten lakes were examined by competent persons and reported in the literature. The species was reported to occur in 14 additional lakes, excluding the "Twin Lake" listed by Jordan and Evermann (1911), since the exact identity of this lake is not known, although it is believed to be the group in LaGrange Co. In addition, there are questionable or definitely erroneous reports of ciscoes occurring in three other lakes. Additional Records of Ciscoes in Indiana Since 1937 it has been lawful to fish for cisco with nets in Indiana lakes during the period November 1 through December 31, both dates in- clusive. Upon payment of a license fee of $2 annually, each person is per- mitted to fish one gill net not more than 150 feet long, with an extension 1 measure of not less than 2 /2 inches. It is illegal to fish with the bottom of the net more than 15 feet below the surface of the water. All nets must be surrendered to the Conservation Department for storage during the closed season. Each fisherman is supposed "to strip or otherwise remove the milt and spawn from each cisco, as it is taken from the net, and to return such milt or spawn to the waters from which the cisco was taken." There is no direction that the eggs and sperm be mixed together before being placed in the water ! The laws governing the catching of cisco in the inland waters of Indiana have been changed drastically a number of times. In 1875, when Jordan described the population from Tippecanoe, there were no fish laws, and people caught ciscoes in shallow water during spawning season in a variety of ways. The Office of State Commissioner of Fish (later, Fisheries and Game) was established in 1881. Soon thereafter the first fish laws were enacted, including one permitting the taking of ciscoes by net or spear during a certain specified period in November and December, the limits of which were changed by various commissioners. In 1901 a closed season for all fishes from December 1 to April 1 was established, and the use of nets and spears at any time of the year was abolished. This closed season was in effect until the present law was enacted in 1937. In 1951 there were 129 persons licensed to fish for cisco in Indiana in the manner prescribed above. A questionnaire soliciting information on the occurrence, abundance, annual catches, and spawning behavior of the cisco was sent to all these persons. Replies were received from 44, distributed quite satisfactorily over the entire lake district. According to these returns, cisco are definitely fished for in 29 lakes, and are reported to occur in an additional 12 lakes. The data are summarized in Table 2, with each fisherman concerned being listed by his license number. The distribution of these lakes is shown in Figure 1. It is immediately apparent from the table that cisco fishing is largely a local cultural activity. A fisherman not only tends to concentrate his efforts in the lakes close to his place of residence, but also shows little TABLE 2. Summary of questionnaire to cisco fishermen as to which lakes contain ciscoes. The post office address and 1951-license number of each fisherman is given. Symbols: X = ciscoes actively fished; R = ciscoes reported to be present but not fished. m — Steuben LaGrange Noble Whitley Kosciusko

Post Office

Location of

]ounty o. Cisco .4 . Fishermen z0 `g -5 ,.. Vii

14 g 0 Tippecanoe

.= . Tippecanoe u N 0 ''. E ".. Q g g, --5 . r. E — -" - > • — Shock Secrist .'' r c, c.5 CC :4 4 4' :'. c4 5 ,i Big Little Oswego iteuben ' Ray 122 X R Hudson 219 X X R It XRRR R 175 X X X 141 R XRRR Fremont 233 X . 163 X 164 XXXXXX X Angola 150 R X R 223 X XRX XX XRR R 209 XX XX X RR It 217 XRXXRR 208 X X XXX X XRRR 193 X XRXRXR X LaGrange LaGrange 384 R R X Noble Albion 136 R X X X X 224 X X XXX X R X 201 X R X Kimmel 112 X X Wolcottville 125 X X XR XR X X RR EL 115 X XRXR R Ligonier 227 X R 11 X Columbia City 190 X X X X 147 R R X X? 137 X 225 X X ? X X X 113 X XRR Larwill 188 X X R R X X 134 X 133 X R R X 114 X X X XX X X X X South Whitley 109 (Fishing club—no ciscoes) Marshall Argos 159 X X St. Joseph Mishawaka 154 X South Bend 111 X X 153 X Kosciusko Leesburg 199 X 192 XRR 106 X North Webster 198 X X 126 X XXR X 107 X R 127 X X 128 X X 129 X X 185 knowledge as to the occurrence of ciscoes in lakes outside his own particular region. This limited knowledge of cisco distribution among the fishermen makes it probable that there may be additional lakes containing small populations of ciscoes. However, because of the relation between chemical stratification of the water and the presence of ciscoes to be discussed later, it is doubted that any lakes actively fished are omitted. Only five lakes listed are fished by just one person, but in two of these ciscoes are reported by other persons as being present. This may be another example of the very local nature of the knowledge concerning cisco lakes, and it may also indicate that the fish population in a lake must be at least a certain minimum level to make it worth the effort necessary to catch the fish. On the other hand, for eight lakes the occurrence in them of non-fished populations is reported by single fishermen. The list of 41 lakes includes 19 of the 25 lakes previously reported in the literature as containing ciscoes. Hence, the number of known and reported cisco lakes has been doubled by the present study. It might be well to point out here that ciscoes have been caught in lakes Tippecanoe, Myers, Knapp, Indian Village, Gordy, Hindman, Oliver, and Cedar (Whit- ley Co.) by Lake and Stream Survey crews in 1951 and 1952 (Table 4). This increases to 14 the number of lakes in which the occurrence of ciscoes is substantiated by actual specimens. Regarding the six lakes (Big, Winona, Maxinkuckee, Manitou, Twin, Little Balls) reported in the literature to have cisco populations but not listed by the cisco fishermen, the identities of the Twin Lakes reported by Jordan and Evermann (1911) and of Little Balls Lake reported by Lower (1913) are not known. Since, however, the two groups of lakes in Mar- shall and LaGrange counties known as "Twin Lakes" both contain ciscoes, the exact identity is not needed in this instance except for historical chronology. As already stated in the cisco chronology, there is no tangible evidence, contrary to the report by Jordan and Evermann (1886), that the cisco has ever occurred in modern time in lakes Maxinkuckee and Winona. Not one of the 11 persons reporting being active in the Tri Lakes- Crooked Lake region listed the species as occurring in Big Lake (Noble Co.), so that it is entirely possible the cisco has died out in this lake in the past 60 years. Likewise, if the report by Blatchley and Ashley (1901) of the former occurrence of the cisco in Manitou Lake is correct, then here too the species may have died out within the past 50 years. The cisco has also been reported by Mr. Lowell C. Reinhart, an old time cisco fisherman from Hudson, Indiana, to have occurred many years ago in Big Long and Turkey Lakes in Steuben Co., but the populations have apparently died out. For a species such as the cisco tending to be restricted to youthful or middle age lakes, typologically speaking, one would expect the number of suitable localities to become progressively smaller with time. Thienemann (1918 and 1928) has already pointed out several instances where north German lakes formerly containing populations of Core gonus 186

maraenus or C. lavaretus no longer have these fishes because of changes in the limnological conditions affecting the ecology of the species. The fish populations in the various lakes do not all have the same characteristics. According to the fishermen, where the populations are small the individual fish tend to be large, as in the Indian Village lakes. Conversely, where the populations are large, as in Myers and Oliver lakes, the individual fish (known locally as "spikes") tend to be small. To a certain extent, the population in a given lake is said to retain its size- abundance characteristics from one year to another, but a number of experienced fishermen also reported that there are quasi-cyclical fluctuations in size and abundance in individual lakes : when the fish are abundant they are small, and when they are less abundant they are larger in size. Hile (1936a) found this relationship to hold true among the cisco populations of four lakes in northeastern Wisconsin, noting that the occurrence of dominant year classes depended on purely local conditions affecting spawning and survival, and Jarvi (1950) has been observing similar fluctuations in the coregonid populations in Finnish lakes for many years. This variation in size and abundance from lake to lake and from year to year within a lake may account for some of the lakes reported as having the species present but not fished. Certainly, a 2%-in. (extension measure) gill net would not adequately sample some of these populations. In fact, a number of fishermen contrary to the law as written (but not necessarily contrary to the sound management of the species) somehow manage to have a variety of gill nets available : the particular one used in a given lake in any year depends upon the average size of the fish in that year. Distribution of Cisco Lakes by Major Drainages The Lake District of Indiana, shown in Figure 1, lies largely within the Erie-Saginaw Interlobate Moraine, also known as the Packerton Moraine, extending in a general southwest-northeast direction for a distance of 100 miles to the northeast corner of the State and into southern Michigan. The moraine is of complex origin, dating from Cary Time within the Wisconsin Glacial. In present time, all the major drainages in the northern part of the state originate in lakes in this moraine. Earlier drainage relationships and changes in drainage during and immediately after Cary Time unfortunately are only very imperfectly known, which complicates the task of explaining the present distribution of cisco lakes. Cisco lakes occur in all the major drainages arising in the Packerton Moraine, both those entering the Great Lakes and those ultimately entering the Mississippi River (Table 3). The Kankakee River, which unites with the Des Plaines River in Illinois to form the Illinois River, and this in turn joins the Mississippi just north of St. Louis, has two cisco lakes. The Tippecanoe Basin contains five lakes with ciscoes, and the Eel River

CISCO LAKES OF INDIANA A MICHIGAN • 1 CLEAR 21 ATWOOD 31 HINDMAN 4I• ■■•■ • • 111■4, • • ■■•11 , —110 q. • • 01■1 2 GREEN 22 WH 1 TMER 32 GORDY 3 MARSH 23 SLOAN 33 I NDIAN VILLAGE • 25 c'11 7 4 YOUNGS 24 FISH 34 SHOCK 8 5 NORTH OT TER 25 SOUTH TWIN 35 SECRIST vt 1 0 6 SNOW 26 ROUND 36 LITTLE TIPPECANOE 46 7 KRIELBAUM 27 SHRINER 37 TIPPECANOE ST JOSEPH 8 JAMES 28 CEDAR 38 OSWEGO RIVER 24 9 JIMERSON 29 CROOKED 39 LAWRENCE 1 0 GAGE 30 KNAPP 40 MYERS 23 OHIO II GOOSENECK 1 7 41 42 H 1 2 Mc LISH 18 6 ,1 19 STEUBEN 1 3 LAKE OF THE WOODS 0 14 45 22 1 4 BIG LONG LAGRANGE , 1 5 EVE A 16 MARTIN ; MAUMEE RIVER 1 7 OLIVER - 18 HACKENBERG M-ARS-HAL- L BASIN 19 MESS1 CK ELKHART 20 DALLAS 4•

• 33 34 • 31 -- 32 -- _30 • KANKAKEE 37-' 38 a 36 RIVER / " 40 35 39 , NOBLE 29 ' 2 6 4, 27 4 • \_,,,1 0.

• FORMER CISCO LAKES

STARKE 41 TURKEY • 42 LONG 43 BIG

411, 44 MANITOU TIPPECANOE , R/ VER

44 KOSCIUSKO CISCOES PLANTED

_ wHITLEy j 45 HAMILTON, in 1947 - L 1\ 46 CROOKED, ca. 1915

EEL RIVER

FULTON „

FIG. 1. Distribution of cisco lakes in Indiana. 187

TABLE 3. Distribution of cisco lakes by county and river system. Questionable reports and lakes formerly having cisco populations are in parentheses.

River system County Kankakee Tippecanoe Eel St. Joseph Maumee Marshall Lawrence Myers Fulton ( Manitou )

Kosciusko Tippecanoe Shock Oswego Little Tippecanoe Secrist Whitley Crooked Cedar Shriner Round Noble (Big) Knapp Gordy Hindman Indian Village LaGrange Lake of the Woods Big Long Eve Martin Oliver Hackenberg Messick Dallas Atwood Whitmer Sloan Fish South Twin Steuben Green Clear Marsh (Hamilton) Youngs North Otter Snow Krielbaum James Jimerson Gage Gooseneck McLish (Turkey) (Long) (Crooked) Little Balls? Riddle? Basin contains three. Both of these rivers are tributaries of the Wabash, which enters the Ohio River at the Ilinois-Indiana-Kentucky corner. Most of the cisco lakes, however, 31 in all, lie in the St. Joseph drainage basin, 188 which enters Lake Michigan from the southeast at Benton Harbor, Michi- gan. Even the Maumee River system, which drains into Lake Erie at Toledo, contains one cisco lake in Indiana. The distribution of cisco lakes in the lake district and within the various drainage basins is obviously not uniform or random (Fig. 1) . Rather, there are definite groups or clusters of cisco lakes. Seven major clusters can be recognized : 1) the Twin Lakes group in Marshall Co., 2) the Tippecanoe group in Kosciusko Co., 3) the Indian Village group (perhaps including Shock Lake as well) in Noble Co., 4) the Tri Lakes group in Whitley Co., 5) the Oliver Lake group in southern LaGrange Co., and finally 7) the James Lake group in Steuben Co. No more than a half-dozen of all the reported cisco lakes occur independently of one or another of these clusters. That the distribution is not random is well illustrated by the fact that whereas the North Branch of the Elkhart River has numerous cisco lakes in the vicinity of Oliver Lake, the South Branch, which drains central Noble Co. and has many lakes, some of them reputed to be fairly deep, has no reported cisco populations. If the assumption is correct that all or nearly all the lakes in the lake district contained ciscoes in early post-Cary time and that populations have been eliminated by the gradual development of limnological conditions unsuitable for their continued survival, then the persistence of cisco populations in such definite clusters of lakes can be explained only by the operation of certain factors controlling the regional limnology of the lakes. It is interesting that Willer (1924) reached a similar conclusion for the lakes of East Prussia containing Core gonus albula. He distinguished three types of lakes : 1) the ermlandische type, being the deepest and most strongly oligotrophic ; 2) the oberlandische type, consisting of rather strongly eutrophic lakes located largely in woodland regions and having a better developed littoral shelf and a higher organic content in the offshore sediments ; and 3) the masurische type, an intermediate mesotrophic type with intermediate morphometry as well. The coregonid populations are confined almost exclusively to the masurische type, being completely absent from the oberlandische lakes. The three types of lakes are quite well delimited geographically, indicating the importance of regional factors in their limnology. Artificial Stocking of Ciscoes For several decades at the turn of the century the United States Fish Commission attempted to bolster the stock of whitefish in the Great Lakes by hatching large numbers of eggs in various Federal hatcheries, and then planting the fry in the lakes. Less extensive culture of the cisco was carried out. Occasionally, at the request of state officials or of other influential persons, fry were planted in various inland lakes. There is also the possibility, of course, that fry or adult fish from one lake were 189 transplanted into neighboring lakes by interested fishermen. It is necessary, therefore, to examine the available records on stocking to determine the influence of man in extending the pre-Columbian distribution of the cisco in Indiana. The whitefish, Coregonus clupeaformis, has been stocked in several Indiana lakes on a number of occasions. This species is more oligotrophic in its ecology than the cisco, and hence would be even less likely to survive in the generally productive lakes of Indiana. Locally around the Indian Village chain of lakes in Noble Co., Shock Lake in Kosciusko Co., and lakes Gage, Clear, and Krielbaum in Steuloen Co., occasional coregonids as large as 4 lbs. in weight are reported. These are sometimes referred to as "whitefish" by the local fishermen, as are also specimens only 1 to 2 lbs. in weight, even though these have the same appearance as the still smaller fish, which are designated "ciscoes." Although none of these large individuals has been personally examined, it is believed that all of them are ciscoes, and that the whitefish does not occur in any of the Indiana lakes. The largest cisco caught by the Lake and Stream Survey was a female from Knapp Lake (Indian Village chain) weightng 1007 grams. The largest ciscoes caught by Hile (unpublished data) from the several lakes of the Indian Village chain were 940 g. from Gordy Lake, 775 g. from Hindman, and 810 g. from Indian Village. According to Levette (1876) the whitefish was introduced into Winona Lake from Lake Michigan about the year 1860. The fish were reported to be occasionally caught in the lake, although never abundant. Possibly the fish that were planted lived for a while without reproducing, and then died out. The species is definitely no longer present in the lake. The same can be said for Wawasee, in which 2,000,000 whitefish fry were planted in 1896 at the request of Eigenmann (Kirsch 1896a). According to Clark (1894), a total of 215,000 whitefish eggs and fry were planted in the inland lakes of Indiana during the period 1872 through 1893. The specific lakes that received these plantings are not mentioned, and there is no additional information in the early biennial reports of the State Fish Commissioner of Indiana. During 1890 and 1891 for certain, and possibly in a few subsequent years as well, experiments were conducted at a hatchery at Warren, Indiana, on the raising of whitefish in ponds (Dennis 1892, and Kirsch 1894a) . There was some measure of success, and the hatchery superintendent enthusiastically predicted that the whitefish would become the popular pond fish in the Midwest, and that once its requirements were understood, it could be grown in any water in which bass will thrive ! Regarding the cisco, there is even less information about the stocking of the species in the lakes, and the general conclusion is that the pre- Columbian distribution has not been extended at all, or at least not appreciably, by these practices. At the hatchery at Warren, Indiana, previously mentioned as raising whitefish fingerlings in ponds, during the 1893-1894 biennium there were 190 hatched and distributed to the waters of Indiana and adjoining states 100,000 fry listed as whitefish x cisco hybrids (Kirsch 1894a). Even accepting this report at face value, it is obvious that the distribution of the cisco in Indiana was not affected by the survival of any of these fry. Construction of several state-owned fish hatcheries in Indiana was begun during the 1911-1912 biennium, and completed during the 1915- 1916 biennium. The first hatcheries established were at Wawasee and Tri Lakes. In the winter of 1914, and apparently only in this one winter, a large number of cisco eggs were stripped from females at the Tri Lakes Hatchery, and after being fertilized were "replaced in the nearby lakes at the most favorable locations" (Miles 1915a). Of the "nearby" lakes, Big, Crooked, Cedar, and Shriner were already known to have populations of ciscoes, whereas Round and Blue probably did not. The cisco has apparently disappeared from Big Lake, so that any stocking in this lake had no lasting effect. The only hint of benefit is in the case of Round Lake at Tri Lakes. One of the 11 fishermen reporting fishing for ciscoes in the Tri Lakes region stated that the species occurred in Round Lake but was 'not fished. This is at best a very uncertain record. Hence, we may conclude that the operations described above did not materially affect the distribution of ciscoes in Indiana. A few other plantings of ciscoes have been made that are recorded or reported, and one cannot know how many other similar plantings have gone completely unrecorded. Miles (1915b) mentioned that several ciscoes were placed in Crooked Lake (Steuben Co.), but that over a period of several years nothing was heard further about these fish, and no one caught any. In late November of 1947, Mr. Floyd DeLancey of Angola and the personnel of the Fawn River Fish Hatchery transplanted eggs stripped from Clear Lake ciscoes into Hamilton Lake (Steuben Co.). In consideration of the limnological characteristics of this lake, the prospects for at least long term survival of this planting are very poor. Not until the current cisco law was adopted in 1937 were the fishermen required to return the eggs of the cisco to the waters from which the fish were taken. It is not known how generally the fishermen comply with this provision of the law, nor how stringently the Conservation of- ficers attempt to enforce it. Under ordinary conditions of compliance, this practice would be of doubtful value in influencing the fish populations in the lakes being fished, and very likely would be completely in- effective in extending the distribution of the species. In the discussion of the temperature-oxygen relations in cisco lakes and non-cisco lakes, it will be seen that there are very few lakes without cisco populations at present which likely are capable of supporting the species. Even if the stocking of ciscoes had been appreciably more extensive than the meager literature records indicate, it is still certain that most, and perhaps all, of the plantings would have met with failure. Be- cause of the considerable number of lakes still in a rather youthful stage of development in northern Germany, stocking attempts there would be 191

expected to have a greater probability of success than in northern Indiana, and yet even in that region success was very limited.

FACTORS GOVERNING THE SUMMER BATHYMETRIC DISTRIBUTION Except for the Great Lakes where the species occurs primarily in shallow water, the cisco is generally regarded as a fish of the deep, cold water of lakes in summer. It tends to occupy the coldest water that still has sufficient oxygen for its requirements. Because of its needs for both cold water and fairly high levels of dissolved oxygen, it comes to occupy an intermediate depth position in summer, limited above by high temperature and below by insufficient oxygen. The thickness of the habitable zone is largely a function of the typological age of the lake. Perhaps the species, as Hile and Juday (1941) suggest, can descend temporarily into the underlying oxygen-deficient water for feeding, and perhaps also during a respiratory crisis can penetrate the thermal barrier of the thermocline into the epilimnion, and there survive after readjusting physiologically to the radically new conditions (Scott 1931). Birge and Juday (1911) made several observations regarding the vertical distribution of the cisco in a number of Wisconsin lakes. On Sept. 12, 1909, six specimens were collected in Green Lake at a depth of 67 m. where the oxygen content was very little more than 1 cc. per liter. The authors concluded that the species does not hesitate to enter water with this low an oxygen content. In Okauchee Lake ciscoes live just below the thermocline in summer. When the oxygen content of this water becomes too small there can be severe mortality of the fish. Of the four years for which observations are available, there was a severe mortality in 1909. The oxygen content of the hypolimnion was lowest in this year, reaching an observed maximum in late summer of only 0.44 cc. per liter. It should be mentioned here that the cisco mortality in Okauchee Lake is perhaps also associated with the almost regular development of a minus-heterograde oxygen distribution.2 Scott (1931) suggested that this type of oxygen distribution was responsible for the September surfacing of the cisco in Snow Lake, Indiana. Lower (1913) studied the cisco in northern Indiana during the period 1909-1912. Without presenting any supporting data he concluded that a temperature of 60° F. (15.6° C.) is fatal to the fish, and that as a consequence the fish are restricted to the hypolimnion in summer. The first detailed study of the ecology of the cisco was made by Cahn (1927), but unfortunately the section relating to the bathymetric distribution of the species is most confusing and difficult to interpret. By means of gill nets Cahn studied the summer distribution of the species in 14 lakes of Waukesha County in southeastern Wisconsin. Although he determined experimentally in the laboratory that the species if possible

'See p. 198 for a discussion of the Aberg-Rodhe system of classifying the oxygen distribution in lakes. 192 avoids water temperatures greater than 17° C., he concluded that in summer the cisco takes a position just above the thermocline. As substantia- tion for this conclusion he stated that if one determines the location of the thermocline with a thermometer, one can set gill nets just above this depth and catch ciscoes all summer long, whereas in 43 sets made below the "thermocline" in nine different cisco lakes, Cahn did not catch a single cisco. Cahn did not present any detailed temperature and oxygen data, but he stated that on one occasion in an unspecified lake, he caught no ciscoes at 9 m., 11 at 10 m., 2 at 11m., and no ciscoes in repeated attempts in deeper water. The thermocline at this time was stated to lie at 11 m. In all probability, Cahn used the word "thermocline" to mean the lower boundary of the thermocline, as we commonly understand the term. If this interpretation is correct, then Cahn found the ciscoes in all these lakes to occur in the lower portion of the thermocline (usual meaning) in summer, and only very seldom in the hypolimnion. This interpretation is substantiated in part by the series of data presented by Birge and Juday (1911) for three of the cisco lakes studied by Cahn. Lake Okauchee has already been discussed. On Sept. 20, 1906, Lake Nagowicka had an oxygen content of only 0.1 ppm. at 10 m., with even less at greater depths. The temperature at this level was 16.2° C., well up in the thermocline. At 8 m. there was an oxygen content of 4.1 ppm., but the temperature was 20.1°, which is generally considered to be at or even above the upper limit of tolerance of the cisco. Hence, on this date, only a thin stratum possibly less than a meter thick at the uppermost part of the thermocline provided suitable conditions of temperature and oxygen for the cisco. In the west basin of North Lake on Sept. 4, 1907, there was essentially no oxygen from 6 m. to the bottom. Yet the temperature at 5 m. was already 20.5°, and the oxygen content 6.2 ppm. In this lake very likely the ciscoes were even more sharply restricted in their late summer bathymetric distribution. Several other examples of cisco lakes in southern Wisconsin might be mentioned here, although they were not studied by Cahn. In Lake Men- dota at Madison the cisco population is forced by insufficient oxygen in deeper water to live at fairly high temperatures in the upper part of the thermocline (Birge and Juday 1911). Pearse (1921a) fished the lake with gill nets from June 24 to Sept. 5, 1919. The four ciscoes caught on Aug. 20 and Sept. 3 were from depths of 10 to 12 m. Quite a number of perch (Perca flavescens) were caught at the same and somewhat greater depths during this period. Pearse did not make any determinations of dissolved oxygen, but it is known from the studies of Birge and Juday (1911) that the oxygen of the hypolimnion usually becomes completely used up during summer stagnation. In 1906, the year for which the best series of records is available, there was no oxygen at all at 12 m. for a period of three weeks in August, and even at 10 m. the supply was so 193 small (less than 0.1 cc. per liter) "that animals requiring a moderate supply of oxygen, such as fish, could not have lived at this depth during this time." This stratum was located at the bottom of the thermocline in August and near the middle in early September. Water temperatures of less than 20° C. extended upward to 9 m. at this time, and the ciscoes presumably were living in this very thin stratum of warmer water with somewhat less severe conditions of dissolved oxygen. Lake Geneva in southern Wisconsin is a large lake, with an area of 2210 ha. and a maximum depth of 43.3 m. (Juday 1914). The oxygen content of the hypolimnion becomes considerably reduced in summer, although it evidently remains adequate for fish life down to about 30 m. (Birge and Juday 1911). During the period Aug. 8-25, 1920, Pearse (1921b) caught ciscoes at depths between 15 and 25 m. Other species caught in this stratum were the rockbass ( Ambloplites rupestris), smallmouth bass (Micropterus dolomieui), yellow perch, and walleye (Stizos- tedion vitreum). Pearse observed the northern pike ( Esox lucius) to be feeding on the cisco, even though the pike at that time was never taken below 15 m. In 1941 Nelson and Hasler (1942) likewise found that the larger individuals of northern pike were feeding almost exclusively on ciscoes, even though the bathymetric ranges of the two species barely overlapped. These investigators also captured a number of other species along with the cisco : yellow perch, rainbow trout (Salmo gairdneri), northern pike, largemouth bass (Micropterus salmoides), smallmouth bass, and white bass (Morone chrysops). Only ciscoes were caught below a depth of 60 ft. (18.3 m.). Hence, in Lake Geneva an upward displace- ment of the cisco in late summer brought about presumably by declining oxygen levels below, brings the species into contact with quite a number of other species, and subjects them to predation by such a typically shallow-water fish as the northern pike. Finally, in Green Lake with a maximum depth of 72.2 m., making it the deepest inland lake in Wisconsin, oxygen levels remain high almost to the bottom throughout the summer. Pearse (1921a) found that at this season the ciscoes occur exclusively at depths greater than 40 m., where they live alone. Hence, where oxygen levels are adequate the species may prefer to live far below the thermocline, utilizing only a relatively small portion of the total volume of water physiologically available. In a somewhat similar instance of summer bathymetric distribution correlated with high oxygen levels, Thienemann (1933) found a race of Core gonus albula, feeding on benthos rather than plankton. He postulated that this species was originally a bottom feeder, but that it was forced to change its habits by the declining oxygen levels of the deepest water with increasing typological age. Hile (1936a) and Hile and Juday (1941) described the summer bathymetric distribution of the cisco in four lakes of northeastern Wisconsin in relation to the temperature and oxygen stratification of the water and in relation to other species occurring along with the cisco. In Trout 194 and Clear lakes having fairly high levels of oxygen throughout the hypolimnion, the species occurred quite generally in the hypolimnion. In Silver Lake, which experiences a moderate oxygen deficit, the species be- came concentrated in a stratum from 10.5 to 15.5 m. In Muskellunge Lake, the lake with the most severe oxygen depletion in the hypolimnion, the ciscoes in 1931 were largely confined to a stratum only a half-meter thick, from 9.5-10 m. On Aug. 26, 1931, this stratum was located in the middle of the thermocline, with a temperature range of 16.3 to 14.8°, and an oxygen range of 4.6 to 1.6 mg./1. In 1930 and 1932, conditions were not quite so extreme, the fish being largely concentrated in each case in a stratum about 2 meters thick. This stratum in 1932 had thermal limits of 13.8 and 10.4°, and oxygen limits of 2.7 and 0.4 mg./1. No chemical data are available for 1930. By virtue of the severe oxygen utilization in the hypolimnion, the fish are forced upwards into the lowermost portion of the thermocline. It is interesting that at least in Muskellunge Lake even at severely reduced levels of dissolved oxygen the negative tropism of the species to the thermal gradient continues so strong that the fish do not ascend very high into the thermocline, where temperatures are still adequate and oxygen levels much more satisfactory. Hile (1936a: 259) concluded, "As a result of this deficiency of oxygen in the deeper strata, the ciscoes are forced out of the cooler strata and by reason of their preference for cool water and their need of oxygen become concentrated in that stratum that has the lowest temperature available and yet contains sufficient oxygen to support life." Hile thought this lower limit might be as little as 1 mg./1. In a later paper, however, Hile and Juday (1941) suggested that the cisco, like the perch, may make temporary sorties for food into deeper water with insufficient oxygen, and hence the lower bathymetric limit as determined by capture may well be below the lower physiological limit for oxygen and carbon dioxide. In these four Wisconsin lakes, various other species of fish were found associated with the cisco in summer, varying with the typological age of the lake and the dependent bathymetric distribution of the cisco. The four lakes arranged in order of decreasing oxygen content in the hypolimnion are : Trout, Clear, Silver, and Muskellunge. In Trout Lake, the ciscoes are associated with such polyoxybiontic cold-stenotherms as the lake whitefish (Coregonus clupeaformis), lake trout (Salvelinus namay- cush), and the burbot (Lota iota). In Clear Lake the walleye was caught at all depths with the cisco, and in addition a rock bass and five yellow perch were taken. None of the more oligotrophic stenotherms was present, however. In Silver Lake where there is considerable depletion of oxygen in the lower hypolimnion, the ciscoes are forced into the upper portion of the hypolimnion, and here they live largely by themselves. Only one perch was caught with them. Finally, in Lake Muskellunge, which develops the most severe oxygen deficit of these four lakes, the ciscoes are intimately associated with small perch in late summer, and may well be in competition with them for food. Hence, with decreasing 195 content of oxygen in the hypolimnion, the ciscoes are forced higher in the water and come into progressively greater contact with warm water species. Hile and Juday emphasize the fact, however, that the lower bathymetric distribution of any particular warm water species seems to vary considerably from one lake to another. Allequash Lake in northern Wisconsin is perhaps even more interesting as an example of the extreme environmental conditions the cisco can tolerate, and of the other species of fishes the cisco can be brought into contact with as it is forced progressively higher in the thermocline. The lake has an area of 142 ha. and a maximum depth of only 7.5 m. (Juday 1914) . On Aug. 20, 1907, the only date for which data have been published, the oxygen content at 5 m. where the temperature was 19.2° C. was 5.7 ppm., and at 6 m. at a temperature of 12.4° only 0.9 ppm. (Birge and Juday 1911). Certainly the ciscoes must have been concentrated in a very thin stratum at this time. Restriction of the fish to the upper part of the thermocline brought them into direct contact with other species. Hile (1936a: 317) records that, "Along with the 70 ciscoes captured in 1930 were taken 1 blue gill, 15 suckers, 22 pike-perch, 32 rock bass, and 182 perch." Conditions necessary for the survival of the cisco must be near the tolerance limit in this lake. Fry (1937) studied the seasonal movements of the cisco in Lake Nipissing, a large lake in southern Ontario. and correlated these with chemical and physical conditions. In this lake, the species moves into deep water in late spring and early summer, different size classes and sexes migrating at different times. Through depletion of oxygen in the hypolimnion, the fish become concentrated immediately beneath the thermocline in late summer. In general, the species is confined to temperatures below 20°, which Fry regarded as a critical limit for the species. At no time during six seasons of study were any considerable numbers of ciscoes taken in water that had been above 20° for very long. Moreover, once the fish get into the hypolimnion, their negative response to the sharp thermal gradient of the thermocline is so pronounced that even during the initial stages of autumnal cooling, as the thermocline sinks deeper the fish continue to remain below this thermal gradient in water quite low in oxygen, even though the epilimnion by this time has cooled below 20°. More recently Smith3 has described the seasonal bathymetric distribution of the species in Green Bay of Lake Michigan. In late May the fish are confined to depths less than 30 ft., most of the fish being concentrated in the uppermost 15 ft. In June, and especially in July, the fish tend to occur only at depths greater than 30 ft. In July the water temperature of the upper 30 ft. varied between 18.3 and 21.5° C., and Smith concluded that the only relationship of the lake herring to temperature was the avoidance of water warmer than about 20° C. However, the

3 Smith, S. H. 1955. Life history of the lake herring of Green Bay, Lake Michigan. 167 typed pages. To be published as a Fishery Bulletin by U. S. Fish & Wildlife Service. 196 influence of increasing water temperature in bringing about the downward movement of the species is somewhat obscure from the data. Thus, on June 10-12 the fish were already largely confined below 30 ft., although surface temperatures at this time varied from 12.9 to 15.1°, and the temperature at 30 ft. was 12.3-13.7°. The upper water temperatures in late May were almost this high, and yet at this time, as already stated, the fish were largely above 30 ft. in their bathymetric distribution. The oxygen content of the water does not seem to be influencing these movements. In summary, the cisco tends to occupy the entire hypolimnion during the summer if adequate levels of oxygen are present. In clinograde lakes (see below) the fish are forced progressively higher by depletion of oxygen below, until they become confined largely to the uppermost hypolimnion and lowermost thermocline. With still further depletion, the fish may be forced into the upper part of the thermocline, although judg- ing by Muskellunge Lake and even Nipissing they seem to resist this dis- placement as long as possible. Some degree of stenothermy persists in the species, however, because there are no records of the cisco occurring in the warm surface watters of our lakes during the summer. INFLUENCE OF AGE ON OXYGEN-TEMPERATURE REQUIREMENTS Among the coregonids, and apparently among fishes in general, physiological plasticity and adaptability declines with age. Thus, although the coregonids as a group are considered typically to require relatively cold water and high levels of dissolved oxygen for their survival, first summer young of at least three species have been successfully reared in shallow, warm-water ponds. At the early fish hatchery at Warren, Indiana (Dennis 1892, and Kirsch 1894a), whitefish (Coregonus clupeaformis) were reared in ponds to an age of 6 months before transplanting. The first attempts were not too successful, in that only 800 fingerlings 2 inches long still survived 3 months after the stocking of 100,000 fry in a small pond (Dennis 1892). Apparently the later attempts were more successful, although no data are presented. Furthermore, no data are presented on the temperature extremes to which the fish were subjected in these ponds. Slastenko (1931, as quoted by Sukhoverkhov 1943) was able to raise whitefish (species not stated by Sukhoverkhov, although very likely Coregonus lavaretus) in carp fattening ponds during their first summer. The most extensive experiments, reported on in considerable detail, were carried out by Sukhoverkhov (1943) on Coregonus albula and one of its varieties with a faster rate of growth. Fry from hatchery jars were stocked in carp ponds up to 150 ha. in area at the rate of 200 to 5000 per ha. No places in these ponds were deeper than 3 m., and the average depths were only 0.8-1.5 m. Permanent stratification was therefore out of the question. Temperatures in these ponds reached maxima of 26-28° C. on individual days in July and August, and occasionally the oxygen 197 content of the water fell to rather low levels over night. Survival on the whole was good, and the rate of growth tended to be much faster than in natural waters. The fish fed on a variety of littoral entomostraca that normally are not encountered by wild populations. Moreover, Dryagin (1939, as reported by Sukhoverkhov 1943) found C. albula in some of the lakes of the Kalininskoi District which were so shallow that summer stratification did not occur at all or was only transitory. These results on C. albula are of particular interest in the present study because of the close morphological and physiological similarity of the species to C. artedii, which, according to Bauch (1953), is somewhat intermediate morphologically between C. albula and the species collectively known as "grosse Marane." The greater tolerance of young coregonids to high temperatures is indicated by other observations as well. Fry (1937) observed in Lake Nipissing that the young ciscoes are the last to move into deep water as the lake warms up in spring. Kennedy (1941) noted that in the movement of C. clupeaformis from one lake to another, apparently induced by unfavorable temperatures in the one, the smallest individuals were the last to migrate. There is also some evidence that young coregonids are more tolerant of low oxygen levels than are larger and older fish. Willer (1929) pointed out that this is known for trout, and that presumably the same holds true for C. albula. He ascribed the mortality of the majority of individuals of this species at the end of their second year of life in Nariensee to the low oxygen levels in the hypolimnion. Furthermore, Bauch (1949) observed an almost complete disappearance of C. albula 3-5 summers old from Mochelsee at the beginning of summer stratification in 1944, which he attributed to the disappearance of oxygen from the hypolimnion. Younger fish continued to live in the lake. Eichler (1940) observed a heavy mortality of the species in LOwentinsee in 1938, which he ascribed to the disappearance of oxygen from the hypolimnion. The kill was not complete, however, because in 1944 Bauch (1949) again caught fish in this lake. He believes that only the young fish are able to withstand the severe hypolimnial oxygen depletion that develops. One of the interesting regional differences between continental Euro- pean lakes and the mid-continental lakes of North America is that the summer stratification of the former is usually much less sharply developed. The epilimnion frequently has temperatures below 20° throughout the summer, and the gradient of the thermocline is much less steep than in the North American lakes. As a result the thermocline does not seem to be such a pronounced barrier to the upward movement of C. albula as it is to C. artedii in their respective lakes. There is abundant evidence based both on visual occurrence of schools of fish and on food habits that C. albula makes food sorties into the too-warm epilimnion, much as the Wisconsin ciscoes apparently make food sorties into the deep water with insufficient oxygen (Hile and Juday 1941). 198

In lieu of evidence to the contrary we can assume that the young of C. artedii likewise are more tolerant of high temperatures and low oxygen than are the larger and older individuals in the populations. As environmental conditions become progressively more severe for C. artedii and C. albula in their respective lakes, one can suppose that in years of extreme oxygen deficit, involving the thermocline as well, many fish will be unable to adapt to the extreme conditions of oxygen and temperature and will perish. The individuals involved will be primarily the large and old members of the population. As long as some of the smallest individuals survive, however, the population can continue to exist. Com- plete elimination of the population might result when such severe environmental conditions develop in two or more years in succession, or when such a severe year follows one or more years without recruitment to the population, resulting in an age-frequency structure truncated below by the lack of the younger and physiologically more adaptable individuals.

SUMMER BATHYMETRIC DISTRIBUTION IN INDIANA LAKES In the summers of 1951 and 1952 the Indiana Lake and Stream Survey began a study of the summer bathymetric distribution of the cisco. A total of 348 specimens was caught in seven different lakes during the two seasons, as summarized in Table 4.

TABLE 4. Ciscoes netted by the Indiana Lake and Stream Survey during the summers of 1951 and 1952. Lake County 1951 1952 Indian Village Noble 6 1 Hindman 1 Gordy ,t 10 1 11 Knapp 4 Tippecanoe Kosciusko 26 - Oliver LaGrange - 48 Myers Marshall 53 203

Gill nets of various kinds were fished in several ways. During the second summer the standard procedure was to set two gill nets in tandem on the bottom and two in tandem diagonally through the water from surface to bottom in the manner described by Fry (1937). Sometimes "experimental" gill nets were used, each containing successive webbing sections of 11/2, 2, 21/2, 3 and 4-inch extension measure, and at other times standard gill nets of either 2 or 21/2-inch extension measure were used. The position of the ciscoes in the nets was kept track of by relating each fish to the nearest numbered float of the corkline. Depth soundings at both ends of the net, and usually in the middle as well, served as the bathymetric controls. In the rest of this paper, where it is necessary to typify or describe the type of oxygen distribution in a lake, the system proposed by Aberg and Rodhe (1942) will be used. The following types of oxygen distribution are recognized in stratified lakes : ortho grade, in which there is no 199 decline (subtype a) or only a very gradual decline (subtype b) in oxygen content with increasing depth ; clino grade, in which the oxygen curve is similar in shape to the temperature curve of a temperate, stratified, holomictic lake ; plus-heterograde, in which there is a pronounced peak of oxygen concentration in or near the thermocline ; and minus-heterograde, in which there is a pronounced decline in oxygen content in or near the thermocline (the so-called "oxygen notch"), with recovery at greater depths. In many respects this system is a restatement of other descrip- tions of oxygen distribution proposed earlier, and yet there is a simplicity in the concepts and terminology which recommends it. One of the ad- vantages of this system is that the types of oxygen distribution are described without reference to the primary trophic level of the lake, even though the orthograde and clinograde curves are usually associated with oligotrophic and eutrophic lakes, respectively. There are all gradations between the various types, however, so that it is sometimes difficult to assign a particular distribution to any one of them. A. Indian Village chain of lakes (Noble Co.) Hile (1931) recorded the cisco as occurring in three of this chain of six small lakes—Indian Village, Hindman, and Gordy. No ciscoes were caught in Rider or Duley, and Hile did not fish Knapp Lake. The cisco populations in Indian Village, Hindman, and Gordy are still maintaining themselves (Table 4), and in addition it is now known that Knapp Lake also contains the species. The Survey crews caught no ciscoes in either Rider or Duley. This chain of lakes is interesting in that all of them except Knapp are quite shallow and small in area (see Table 5), and exhibit severe oxygen deficits in the hypolimnion which force the ciscoes into the uppermost hypolimnion or even into the thermocline in summer. (A) 1. Indian Village Lake The only previous report as to the depth at which ciscoes occur in Indiana lakes in summer, based on actual netting records and the relation of this stratum to the temperature and oxygen characteristics of the water, is that of Scott (1931) for Indian Village Lake. Hile set a 31/4- inch extension measure gill net, which was 12 ft. in depth, in the deepest part of the lake on July 21, 1929, and caught 18 ciscoes in the upper two feet only of the net. The depth of the central portion of the lake is 6.3 m. According to these relationships, the fish were caught in the stratum from 3.64 to 4.25 m. below the surface. There is no assurance from the catch data that there were no ciscoes at shallower depths than this, although that is not too likely because of the temperature relationships. Scott presents an oxygen curve for the lake for June 27, about a month before the ciscoes were netted. At that time the oxygen content varied from 5.6 ppm. at the upper level of this stratum to 3.4 ppm. at the lower. A month later, there would be no reason to expect any appreciable change 200 at the upper level, although at the lower boundary the oxygen content might have been still smaller. Temperature data for this period are not available. Since, however, thermal stratification is largely a function of the surface features of a lake and its watershed topography, and for such small lakes, at least, varies relatively little from one year to another, more recent thermal data can be used. On Aug. 18, 1951, the interpolated temperature at 3.6 m. was 18.8°, and at 4.2 m., 16.4°. This zone lies in the upper portion of the thermocline. On July 31, 1952, water temeperatures at the same levels were 19.0° and 16.2°, respectively. In spite of the considerably steeper thermal gradient in 1952 than in 1951, the thermal limits of this stratum are very similar in the two years. As closely as could be determined from the field data, the six specimens caught in 1951 were taken at a depth of approximately 4.6 m. The oxygen content at this level was approximately 4.2 ppm., and the temperature was 14.5°. If the temperature tolerance in this year was 19°, as it seemed to be in 1929, then the fish could have ranged upward to 3.5 m. In 1951 the lake had a plus-heterograde oxygen distribution, with the maximum occurring at 4 m. The decline below this depth was very sharp, so that the fish would not have been able to remain continuously in water much more than a few decimeters below this level. Such a photosynthetic maximum of oxygen in the thermocline is probably of great significance in facilitating the survival of the cisco in this lake and in a number of other small Indiana lakes. There is evidence from at least one lake (Myers) that a photosynthetic maximum is of regular occurrence in some of these lakes, although in 1952 the oxygen distribution in Indian Village was clinograde. The depth at which the single cisco of this year came from could not be determined from the field data. The marginal conditions under which the ciscoes are forced to live in this lake are strikingly illustrated by the warm water fishes occurring with this species. One of the six specimens caught in 1951 was partially eaten, supposedly by the bowfin (Amia calva) 58 cm. long which had become entangled nearby in the same net. In four other sets of gill nets at this depth, or at most 2 ft. shallower, no ciscoes were caught, but there were instead six white suckers (Catostomus commersoni), one brown bullhead (Ictalurus nebulosus), and one black crappie (Pomoxis nigromacula- tus). In another set only a little shallower than this there were ten common suckers, one grass pickerel (Esox vermiculatus), seven bluegills (Lepomis macrochirus), one golden shiner (Notemigonus crysoleucas), and two warmouth bass (Chaenobryttus coronarius). Thus, not only is the bathymetric distribution of the cisco greatly restricted in this lake in late summer, but it lies immediately below or even overlaps to a slight extent the shallow water habitat of the warm water species. The stratum occupied lies in the upper portion of the thermocline roughly between temperature limits of 19 and 15°, and at oxygen 201 concentrations roughly between 8 and 3 ppm. (A) 2. Hindman Lake Hile (1931) caught 11 ciscoes in Hindman Lake on Aug. 12-13, 1929, but unfortunately for the present purpose he did not discuss their bathymetric distribution. The single cisco caught in 1951 is also from an indeterminable depth. However, the late summer stratification in this lake is so extreme there is only a very thin stratum in which the fish can live. If one assumes that the upper thermal limit of the species is 20° C., as Fry (1937) concluded, and that the lower oxygen limit is approximately 3 ppm., then only the stratum from 3.2 to 3.8 m. provided these conditions on July 26, 1952. On Aug. 26, 1953, the depth limits of this zone were 3.7 and 4.3 m. In both years the lake exhibited a strong plus- heterograde oxygen distribution early in the summer, somewhat less strong later. Since the odor of ILS was already pronounced in water samples from 5 m., it is likely that the photosynthetic release of oxygen in the upper thermocline is important in maintaining the species in this lake, too. (A) 3. Gordy Lake Hile (1931) studied 24 ciscoes from Gordy Lake, 11 of which were caught on July 23, 1929. Again, regretfully, no data are given on bathymetric distribution. The 10 ciscoes caught by the Lake and Stream Sur- vey in 1951 came from a depth of about 4.5 m., as near as can be determined. The single specimen caught in 1952 came from about 5 m. Its head had been eaten off, and nearby in the net in somewhat shallower water were two garfish (Lepisosteus sp.). It is not believed, however, that the gars ate the cisco, but rather that a bowfin did, as in Indian Village Lake, because another set from approximately the same depth yielded a bowfin and a bullhead. In 1951 nine bluegills were taken in the same set as five ciscoes. Hence, the bathymetric ranges of these two species if not slightly overlapping at this time were at least contiguous. Both in 1952 and 1953 Gordy Lake had a strong plus-heterograde oxygen distribution, with the oxygen maximum occurring in the upper part of the thermocline at a depth of 4 m. and a temperature of 19° C. Contrary to the conditions in Indian Village and Hindman lakes, there was not the extremely rapid cutoff in oxygen below this maximum, which may well be related to the greater hypolimnial volume in Gordy Lake. As a result concentrations of oxygen adequate for ciscoes persisted down to 6 m. or even a little deeper. Based upon toleration of temperatures up to 20° and dissolved oxygen down to 3 ppm., the cisco stratum in these two years extended from 3.8 to 6.3 m., and from 3.9 to 6.2 m., respectively, in late summer. (A) 4. Knapp Lake During the period Aug. 13 to 17, 1951, four ciscoes were caught at 202 depths of 6.0 to 8.5 m. Nets set at depths greater than 9 m. did not catch any fish. Chemical and physical data are available for 1952 and 1953. In both years the oxygen distribution was plus-heterograde, although in 1953 this had changed to clinograde by late August. On July 23, 1952, the stratum from 4.5 to 12.1 m. had conditions assumed to be suitable for cisco, and on Aug. 26, 1953, from 4.9 to 8.8 m. These zones extend through the lower 2/3 of the thermocline into the uppermost part of the hypolimnion. It is interesting and probably significant that all four lakes in the Indian Village chain which maintain cisco populations tend to develop a plus-heterograde oxygen distribution. This is important particularly in the two shallowest lakes in helping to maintain adequate levels of oxygen in the upper and middle thermocline, where the cisco lives in late summer. Here the temperature is so high that warm water species from the epilimnion, including predators, can occur along with the ciscoes. 5. Tippecanoe Lake (Kosciusko Co.) Tippecanoe Lake was fished during the period July 10-27, 1951, with various combinations of two experimental nets and two standard 3-inch nets. The nets were set off the bottom as well as on the bottom. Many sets yielded no fish at all. Nets set offshore at water depths of 10 m. or less did not catch any ciscoes, although one specimen each of rainbow trout, bluegill, and white sucker was taken at depths almost this great. Ciscoes first began appearing at 12 m. (there is one somewhat questionable record for 11 m.), reached their maximum concentration at 15-18 m., and then continued in small numbers to depths at least as great as 25 m. It is quite likely that this early in the season when the oxygen deficit is not yet very extreme, the ciscoes occur throughout the hypolimnion. No other species of fish were caught along with the cisco. Tippecanoe is a large lake and is the deepest in Indiana, with a maximum depth of 37.5 m. Moreover, the hypolimnial volume (that below a depth of 10 m.) is large, comprising 46 percent of the total volume of the lake. In this respect, it is almost on the border line between oligotrophic and eutrophic lakes in Thienemann's original statement of the concept. However, the lake is definitely not oligotrophic. It regularly develops an oxygen notch in the thermocline early in the summer, and as the season advances, the oxygen content of most or even all of the hypolimnion can be reduced to very low values. In 1951, the oxygen notch was less strongly developed and the oxygen content of the hypolimnion remained higher than for any other year on record. On Aug. 26 the oxygen content at 10 m. was 1.2 ppm., but below here it varied between 3.4 and 4.5 ppm. down to at least 30 m. There is little reason to believe that these levels of dissolved oxygen presented any difficulties for the fish. Birge and Juday (1911) present a series of observations on Tippe- canoe for Sept. 17, 1909, in which the oxygen content at 10 m. was only 0.16 ppm., and the maximum oxygen content of the hypolimnion was only 203

3.3 ppm. Scott (1916) likewise found a more extreme stratification than in 1951, although not quite so extreme as in 1909. On Aug. 12, 1912, the oxygen content at 10 m. was 0.27 ppm., and the maximum concentration in the hypolimnion was 3.66 ppm. Apparently, ciscoes can survive quite well at these higher oxygen levels. In some years, however, for which observations on the water chemistry are unfortunately lacking, conditions become extreme enough to bring about extensive mortality. 1954 is reported to have been such a year. It is believed that the hypolimnial oxygen content below the 10-meter level becomes so greatly reduced at such tims that the fish are unable to survive, and they rise to the surface. This seems to be a much better example of the phenomenon reported by Scott (1931) for Snow Lake, in which the September surfacing is believed to result from the fish becoming trapped below the oxygen notch and suffering temporary asphyxiation (see p. 221) . 6. Oliver Lake (LaGrange Co.) Oliver Lake is a deep, marl-type lake, which tends to maintain quite high levels of dissolved oxygen in most of the hypolimnion throughout the summer. Of all the lakes in Indiana it, along with James Lake in Steuben Co. and Crooked Lake in Whitley Co., most closely approaches the oligotrophic type in terms of summer oxygen distribution. This summer oxygen pattern is reflected in the summer bathymetric distribution of the ciscoes. A total of 48 ciscoes was caught in Oliver Lake in 1952, 40 during the period July 15-18, and 8 during Aug. 12-15. These fish were all taken at depths varying from 9.8 to 23 m. Two specimens may possibly have been caught in water as shallow as 8.5 m. ; but since these were from a diagonal set, the records are not above suspicion. Only two other fish were taken with the ciscoes—a rainbow trout at 11 m., and a white sucker at 13 m. The former tends to be a cold water species, and the latter frequently extends its range into cold water. Based on these catch records, the cisco in Oliver Lake lives at tempera- 0 tures 10 and below in summer, and at oxygen concentrations generally between 5 and 8 ppm. Even as late as September 12, there were still 2.5 ppm. of oxygen at a depth of 20 m. Contrary to the lakes of the Indian Village chain, the ciscoes here are not driven upward into the thermocline in late summer by reduced levels of oxygen in the hypolimnion. The fish remain in the hypolimnion all the time and apparently do not enter the thermocline, even though conditions here are adequate for survival, indicating that the hypolimnion is the preferred habitat of the species, and that the warmer water of the thermocline is used only when the hypolimnion is no longer habitable. 7. Myers Lake (Marshall Co.) This is one of the more interesting lakes in Indiana. Regularly during five successive summers (1951-1955) it has developed and maintained 204 a very pronounced photosynthetic oxygen maximum centered at about 8 m. in the lowermost part of the thermocline, resulting in the highest levels of dissolved oxygen known from any Indiana lake. The highest concentration recorded is 23.2 ppm. at a temperature of 12.1°, and the organism largely responsible for this condition is Oscillatoria Agardhii Gom. Moreover, below this oxygen maximum at a depth of 9.5 to 10 m., there is usually an extremely sharp decline in oxygen content to just a trace or none at all within a depth of a half-meter (Fig. 6). W. R. Eberly has been attempting to discover the factors responsible for the consistent development of this oxygen maximum, and its effect on the overall pro- ductivity of the lake. Using Birge's definition of a thermocline in temperate lakes as a thermal gradient equal to or exceeding 1° C. per meter of depth, then the mid-summer boundary between the thermocline and hypolimnion lies at 9.0±- 0.2 m. Because of insufficient oxygen in deeper water, the ciscoes are limited in their summer bathymetric distribution to the uppermost meter or two of the hypolimnion, plus whatever portion of the lower thermocline they are forced to invade. A total of 256 ciscoes was captured in the lake during 1951 and 1952. Certain data for these fish have already been partially organized by Davidoff,4 although the analysis of the bathymetric distribution is modi- fied and developed further in the present paper. During the period July 30 to Aug. 10, 1951, 53 ciscoes were netted at depths between 7.6 and 10.7 m., with the bulk of the fish apparently occurring right around 9 m. On Aug. 13, the thermal limits of this stratum were 12.9° and 8.8°. Because of the location of this stratum with reference to the thermocline, it may be presumed that there were no significant changes in temperature during the preceding two weeks. The peak of the oxygen maximum occurred at 8 m. on this date, with a value of 18.9 ppm., and a concentration of 8.1 ppm. at 10 m. Although no oxygen sample was collected at 11 m. on this date, it is likely from subsequent more detailed studies on the limnology of the lake that the oxygen cutoff began at 10 m., and that by a depth of no more than 10.7 m. there was probably very little dissolved oxygen remaining. Hence, in midsummer of 1951 the Myers Lake ciscoes were inhabiting a stratum only 2-3 m. thick at the thermocline-hypolimnion junction. The fish were probably reacting negatively to the thermal gradient, which kept them from occurring at higher levels in the thermocline, as in Indian Village and Hind- man Lakes. A few warm water species were occurring with the ciscoes. A bottom set at a depth of 10 m. caught three brown bullheads and one bluegill along with three ciscoes. In 1952, the lake was fished at three different times during the summer, and greater care was taken to ascertain the particular depth at which

4 Davidoff, E. B. 1953. Growth, response to netting, and bathymetric distribution of the cisco, Leucichthys artedii (Le Sueur), in Myers Lake, Indiana. Univ. Michigan, MS-thesis manuscript. iv + 31 typed pp. 205 each cisco was caught. The results for the bottom sets are more reliable than for the diagnonal, because the latter as a result of sagging did not always fish at the calculated depths. Eighty-six ciscoes were caught during the period June 26 to July 3, 1952. Most of these fish were captured at depths varying from 7.0 to 10.4 m. Two specimens were apparently caught at 10.7 m. and one at 11.9; these three specimens were all dead when hauled to the surface. At lesser depths, one-half to three-fourths of the fish were still alive after a 12-hour set. Eighteen ciscoes were taken in the upper of two experimental nets set on the diagonal. The bottom of this net was at a depth of 7 m. If the net was stretched straight along the diagonal line from this depth to the surface, then some of these fish could have been caught in depths as shallow as 5.2 m. Because of known instances in which nets set diagonally sagged, it is entirely possible that this net was sagging a bit. Furthermore, in view of what has already been stated about the bathymetric distribution of the cisco in 1951, it is considered quite un- likely that any of the fish were actually living in water shallower than 6.5 m. this early in the summer. On June 26, the temperature at 7 m. was 11.9° C., and at 10 m., 7.5°. Oxygen concentrations were 17.0 ppm. at 7 m. and 0.8 ppm. at 10 m. The oxygen maximum occurred at 8 m. on this date, with adequate concentrations (4.7 ppm.) down to 9.5 m. (see Fig. 6). If the three specimens calculated as having been caught at depths greater than 10.4 m., where the oxygen content at this time was only about 0.5 ppm., actually came from these depths, than it is possible that they had been making quick sorties into this lower water, such as already postulated for Lake Muskellunge in Wisconsin (Hile and Juday 1941). The fact that all three fish were dead when recovered lends support to this idea. Two rainbow trout were caught in approximately 9 m. of water, and were alive when taken. During the period July 22 to July 25, 1952, one diagonal set yielded 27 ciscoes from the upper net. If the assumption is made that the net was not sagging on this occasion, then these fish came from depths between 5.2 and 6.5 m. This statum was near the top of the thermocline. Its thermal limits were 18.0° and 12.5°, and its oxygen limits 9.3 ppm. at the top and 14.4 ppm. at the bottom. On this date, the oxygen maximum of 20.2 ppm. occurred at 8 m., and oxygen concentrations 3 ppm. or greater extended downward to 10.5 ppm. In light of what has already been reported in this paper concerning the strong negative tropism by the cisco to the steep temperature gradient of the thermocline, and the con- sequent pronounced tendency of the species to remain in the deepest water with adequate levels of dissolved oxygen, it is difficult to understand why these fish would vacate the lower half of the thermocline when the oxygen concentrations here were still much more than adequate for their needs. Possibly they were responding negatively to the large concentrations of Oscillatoria at depths of 8 m. and greater, particularly the flocculent 206 senescent algae near the lower limit of the thermocline, or perhaps the fish were actively avoiding the excessively high concentrations of oxygen in the middle thermocline. Sukhoverkhov (1943), for example, found that in one pond in which the oxygen content reached an extreme value of 39.2 mg./1., there was a loss of Core gonus albula, supposedly from the high oxygen levels, but the carp occurring in the same pond were not appreciably affected. That the ciscoes occur higher in the thermocline in late summer, even though oxygen levels for some distance deeper are still adequate, is further demonstrated by the catch data of Aug. 18-21. Eighty-four ciscoes were caught at this time. The calculated depths of 34 ciscoes caught in diagonal sets (4.9-6.1 m.) were somewhat less than those of the 50 fish caught in bottom sets (4.9-8.5 m.), which may be accounted for in part by the tendency of the diagonal nets to sag. Even allowing for this possible source of error, however, it is significant that no specimens were caught below 8.5 m. It is true that at this time the oxygen conditions in the upper part of the hypolimnion were greatly reduced over what they had been only a week previous, so that, for example, at 9.5 m. the oxygen content was only 2.4 ppm. There is no apparent reason to question the bathymetric distribution as calculated from the field data, and one must conclude that the ciscoes were avoiding the stratum from 8.5 to 9.5 m. for some reason. This higher bathymetric distribution in late summer brought the cisco into direct contact with several warm water species. One warmouth bass was taken at 7.3 m., and a brown bullhead was taken at 6.1 m., flanked on either side in the net by ciscoes. Bullheads were fairly abundant at depths down to 4.3 m. Rainbow trout, however, tended to occupy the same stratum as ciscoes throughout the summer. Four specimens were captured during this late-summer period at depths of 5.2-8.2 m. Thus from the available bathymetric data for Indiana and Wisconsin it is apparent that the cisco prefers to live in the deep, cold water of the inland lakes. Where oxygen concentrations are adequate, as in Trout and Clear Lakes in Wisconsin, the fish continue to occupy the entire hypolimnion throughout the summer, or only the bottom half of the hypolimnion in Green Lake. As the oxygen content of the lower hypolimnion declines, the fish are forced into progressively higher levels with the ad- vancing season, as in Silver Lake (Wisconsin) and perhaps Oliver Lake. With still further utilization of oxygen, the fish become concentrated immediately below the thermocline, as in Lake Nipissing. The negative response to the thermocline is very strong, however, and the fish stay out of this stratum if possible. Nevertheless, in instances of extreme stratification the ciscoes do enter the thermocline and may even live in the upper half of the thermocline in late summer, as in Indian Village, Hindman, Myers, Mendota, and Allequash lakes. Here they come into contact with various warm water species of fish. Although the cisco avoids the warmer temperatures of the thermocline if possible, when 207 forced to live in this stratum, it can tolerate temperatures up to 19° or even 20° C. The requirements of the fish for dissolved oxygen are less well understood. The fish may be able to tolerate concentrations as low as 1.5 ppm., although it seems more likely that the level for sustained activity of the larger fish at least is somewhat higher, possibly about 2.5 or 3 ppm. Almost nothing is known concerning the effect of the CO2-content of the water on the respiratory process of the species ; that is, the same concentration of dissolved oxygen may have different physiological values at different concentrations of free CO,. Black, Fry, and Black (1954), for example, have shown that all 16 species of freshwater fishes studied require greater amounts of dissolved oxygen at high CO, tensions than at low tensions. COMPARISON OF TEMPERATURE AND OXYGEN STRATIFICATION IN LAKES WITH AND WITHOUT CISCOES During the summers of 1950 through 1953, Lake and Stream Survey crews visited a total of 33 cisco lakes and 36 non-cisco lakes for the purpose of obtaining basic limnological data. Some of the lakes were visited only once, others were visited several times, and a few have been studied intensively for one to three summers. In addition there are stratification data given by Birge and Juday (1911) and Scott (1916 and 1931) for a number of Indiana lakes, so that in all temperature and oxygen data are available for 37 lakes now or formerly containing ciscoes, and 53 lakes without ciscoes in modern time. Relatively few of these lakes have had hydrographic surveys made. In the absence of such maps the Official Indiana Lake Guide (Gutermuth 1938) is the only ready source of approximate information as to the areas and depths of the lakes. In general, the areas listed in the guide are more reliable than the depths. Most of the non-cisco lakes visited by the Survey crews were selected for study because the Lake Guide includes them among the deeper lakes of the State. Many of them were found to be much shallower than listed. The surface areas and maximum known depths of these 90 lakes are presented in Table 5. Lakes represented by hydrographic maps are designated by an asterisk, and the area and depth of these lakes are taken from the maps. Areas of all other lakes have been taken from the Lake Guide. The depths listed, however, are the maximum depths found by the Lake and Stream Survey crews, or reported by Birge or Scott. They thus represent depths known to occur, and are believed in most instances to more closely approximate the actual maximum depths than do the figures listed in the Lake Guide. Each series of lakes is arranged in order of decreasing known depth. It is apparent immediately that there is no relation between the maximum depth of a lake and the presence or absence of ciscoes. Deep lakes may or may not have the species, and likewise shallow lakes. In fact, two 208

TABLE 5. Known maximum depths and actual or approximate areas of Indiana lakes for which temperature and oxygen data are available. denotes hydrographic map available. Counties are designated by the first letter of their names: Fulton, Kosciusko, LaGrange, Marshall, Noble, Steuben, Whitley. Cisco lakes Non-cisco lakes Max. Max. depth Area depth Area Lake County ( m.) (ha.) Lake County ( m.) ( ha.) *Tippecanoe K 37.5 286 Loon W 29 321 Crooked W 33 77.7 *Adams L 28.5 119 Clear S 32 310 *Maxinkuckee M 26.5 751 *Oliver L 28.5 145 *George S 25.3 206 *Snow S 27.5 120 *Winona K 24.4 215 *James S 26 534 Olin L 24 38.5 Dallas L 26 107 Pretty L 23.5 74.5 Fish L 24.8 32.8 Diamond N 23.5 38.8 *Big Long L 24.7 148 *Wawasee K 23.5 1200 Big N 23.7 80.5 *Crooked S 23.1 325 *Shriner W 22.3 48.6 *Dewart K 23 145 *Cedar W 22.3 58.3 *Yellow Creek K 22 57.7 *Gage S 21.3 132 Goose W 21.5 35.6 Lawrence M 19.7 24.7 Hamilton S 20.7 309 *Little Tippecanoe K 18.9 108 Balls S 20.5 31.6 *Round W 18.9 53.0 Round N 19.5 33.6 Myers M 18.5 34.8 Troy Cedar W 18.5 38.1 Lake of the Woods L 18.5 51.4 Fox S 17.5 57.5 Knapp N 17.5 31.2 Bear N 17 50.2 McLish S 17.2 6.1 *Webster K 15.8 237 Shock K 17 4.0 Little Long N 15.5 25.9 *Secrist K 17 40.1 Cook M 15 23.1 Martin L 16.8 6.1 *Manitou F 14.9 289 *J imerson S 16.5 82.2 Blue W 14.5 95.5 Messick L 16.5 22.3 Round S 14 6.1 South Twin L 15 47.3 *Big Barbee K 14 92.3 Royer L 14 21.9 Waldron N 13.5 80.1 Eve L 13.5 6.1 Pleasant S 12 19.4 Whitmer L 12.5 87.0 Pretty M 12 34.4 Big Turkey L 12.5 182 Thomas M 12 2.0 Hackenberg L 12 15.0 *Chapman K 11.9 168 Little Otter S 11.5 ? Bixler N 11.5 45.3 *Gordy N 10.5 11.5 Latta N 11.5 14.2 Long S 9.5 38.0 Black W 11 8.5 Atwood L 9 63.2 North Twin L 11 53.0 *Indian Village N 6.5 2.0 *Silver S 11 37.9 Hindman N 6 6.1 Center K 10 48.2 Hill K 9.5 23.9 Pike K 9.5 74.5 Golden S 9.5 47.8 *Cline L 9.2 8.0 Hog S 9.0 35.6 Otter S 9 58.3 *Shoe K 9 16.0 Walters S 8.5 15.0 Hollem M 8 8.1 Hogback S 8 50.6 Big Cedar L 8 43.7 Stevens K 6 5.7 Duley N 6 10.1 *Weir L 5.8 2.4 Center S 5 14.6 Loon S 4 57.1 209 of the shallowest studied (Indian Village and Hindman) harbor the species. In order to ascertain what general differences in summer oxygen content and temperature stratification might be present between the two groups of lakes, certain more or less arbitrary physiological limits were established for the cisco. The species was considered capable of surviving at any depth where the temperature was 20° C. or less and the oxygen content was 3 ppm. or more. All lakes, except a few with a very pronounced minus-heterograde oxygen distribution (Tippecanoe and Mar- tin), could then be divided into three bathymetric zones : a supra-cisco layer (essentially the epilimnion) with adequate levels of dissolved oxygen, but with water temperatures above 20°; an infra-cisco layer with suitable low temperatures, but with oxygen contents less than 3 ppm.; and an intermediate cisco layer, varying greatly in thickness from one lake to another and sometimes completely lacking, in which the temperature is less than 20° and the oxygen content greater than 3 ppm. In Figures 2, 3, and 4, the thickness of these three layers and the

la L.) 0 , u z o •,,, < cc < N In 0n 0 4 La-u_ rn C . g w20., z ,I E =_ 05 lax , 2 - 5•6 11?1 0 0 6 -1 IT:Aj f,°_i !2 '3'],.; 2=2

25 . a, ?, 30 1, .., . ,.„, > si _ . „, 13 z o ■n r,g 3C . , a. < = 5 '' LI,..T! gt <, ,E 0 >. ;2 . LJ ; E=g - Is >. 21 5. ONY Eiglj =,,,iu?Ji 32i iii

10 THERMOCL1NE

ED TEMP. >ze 15 TEMP. <2e; 02 >3 PPM.

E21 02 <3 PPM.

FIG. 2. Temperature-oxygen relations in known cisco lakes under conditions of maximum recorded summer stagnation. The middle stratum in solid black is by definition the cisco layer, with temperatures of 20° C. or less, and an oxygen content of 3 ppm. or greater. •

210

• z 0 co a N cr 0 'I) ,7) ni 0 ,n • z = 0 > ..-. Z >. 5 • 0 co I— • — 0 nj • . o C co N _1 CTI CC — FIACKENBERG

15 I THERMOCLINE 1 TEMP >20° 20 11111 TEMP. <20% 02 >3 PPM.

EZI 02 <3 PPM.

FIG. 3. Temperature-oxygen relations under conditions of maximum recorded summer stagnation in lakes reported to have ciscoes present but not actively fished. The middle stratum in solid black is by definition the cisco layer, with temperatures of 20° C. or less, and an oxygen content of 3 ppm. or greater. mean oxygen content within each layer is shown for each lake. Figure 2 is for lakes known to contain ciscoes, Figure 3 for lakes reported to now contain ciscoes but not fished or to formerly have contained the species, and Figure 4 is for lakes in which the cisco is not reported to be present. Where several series of observations are available, the series with the greatest depletion of oxygen in the hypolimnion was selected as representing the most extreme conditions of survival for the cisco. The heavy line to the left of each figure shows the vertical extent of the thermocline, as defined by a thermal gradient equal to or greater than 1° C. per meter. The differences in temperature-oxygen relationships between lakes containing cisco populations that are being actively fished and those in which the cisco is not reported to occur are striking and quite consistent. The stratum of water having conditions suitable for the species as here defined tends to be thicker in the cisco lakes and also to have a higher mean content of dissolved oxygen. Many of the non-cisco lakes do not have any stratum suitable for ciscoes according to the arbitrary limits established, and many others have only a thin cisco layer. The diagrams for the individual lakes cannot be regarded as absolute expressions of the stratification in the lakes, because there can be considerable variation from one year to another. For one lake the vertical series available may approximate the extremes of oxygen utilization, whereas in another lake the oxygen utilization represented may be even less than average. The survival of ciscoes in a lake is necessarily con- 211

2 12 42 o o 02.2; ErgE,5

1 0

23 _ la 60' 2, E. ■-•' 6; .5> o - 3rvi

10 15

1[1] I I l k-13 II I THE RMOC LINE TEMP. >2ð 0 3 6 9 TEMP. <20., 02 >3 PPM PPM 02 1ESa 02 < 3 PPM

Fie. 4. Temperature-oxygen relations under conditions of maximum recorded summer stagnation in lakes not reported to have ciscoes in them. The middle stratum in solid black is by definition the cisco layer, with temperatures of 20° C. or less, and an oxygen content of 3 ppm. or greater. trolled by the extreme and not the average conditions. Hence, on the basis of the single series of observations plotted one might wonder how ciscoes can maintain themselves in Royer Lake, or why for example they are not present in Olin or Fox lakes. Perhaps they are present in the latter lakes, even though the fishermen did not report them. Olin Lake, for instance, is closely connected to Oliver Lake and might be expected to receive stock from this lake. In any event, the data show that the lakes containing ciscoes have on the whole a greater buffer capacity against extreme oxygen utilization than do the lakes without this species. The loss of this margin of safety may well have resulted in the elimination of the species from the latter lakes during extreme years, as conjectured on p. 198. The final critical limits and associated conditions that result in the extinction of a cisco population in a particular lake are not known. Very likely some individuals are able to survive at oxygen concentrations considerably less than 3 ppm. in extreme years. In Lake Mendota, Wiscon- 212 sin, for example, on Aug. 24, 1906, no stratum of water at temperatures below 20° had an oxygen content as great as 3 ppm. (Birge and Juday 1911) . The same was true on July 26, 1951, in Hackenberg Lake, which is reported to have a non-fished population of ciscoes.

TYPES OF OXYGEN DISTRIBUTION IN CISCO AND NON-CISCO LAKES The main controls of the oxygen distribution in a lake are the depth of the photosynthetic zone, the intensity of oxygen utilization in the deeper water, and the stability of thermal stratification. Lakes having a large volume of hypolimnial oxygen in relation to the amount utilized during the summer tend to have an ortho grade distribution. Conversely, lakes having only a relatively small volume of hypolimnial oxygen develop a clino grade distribution. Where the transparency of the water is sufficient so that at least the upper part of the thermocline is included in the photosynthetic zone, a plus-hetero grade distribution can develop. And finally, in instances where a net positive assimilation does not extend to the thermocline, and where perhaps the thermal gradient is particularly steep, an oxygen notch ( minus-heterograde distribution) can develop. This last type of oxygen distribution is probably the least well understood.

Constancy of Oxygen Typology During a Given Summer The stability or instability of these types during a single season is illustrated by the detailed studies on temperature and oxygen distribution made on Oliver, Myers, and Little Tippecanoe lakes in the summer of 1952. The bathymetric distribution of the cisco in Oliver and Myers Lakes during 1952 has already been described. In Figures 5, 6, and 7, the oxygen curves are shown chronologically, and in each curve the solid line represents the "cisco layer" with a temperature below 20° C. and an oxygen content of 3 ppm. or greater. The limits of the thermocline at the beginning and end of the period of observation for each lake are given to enable the oxygen curve and the location of the cisco layer to be related to the thermal gradient.

Oliver Lake Oliver Lake (Fig. 5) had a plus-heterograde oxygen distribution during July and August, with the maximum gradually descending from 6 m. to 8 m. below the surface. On Aug. 22 the maximum was barely discern- ible, and the distribution on this date might be classified as b-orthograde because of the gradual decline in oxygen content nearly to the bottom. Except for a thin stratum on July 21, an oxygen content exceeding 3 ppm. extended all the way to the bottom of the lake until Aug. 22. Even by Sept. 12, however, only the lowermost 3 m. had an oxygen content of less than 3 ppm. In 1953 Oliver Lake also had a plus-heterograde distribution on July 8, which had changed to a b-orthograde (or a very weak plus-heterograde) 213

0 Cu >—

27.0° I 20.9° thermocline Sept. 12 — 1 0 8.4°

1 5

38 2.4 3 8 1.0 OLIVER LAKE — 1952 3.0 ' .3 1.2

25 4.1 0 , . ppm. oxygen

FIG. 5. Oxygen distribution in Oliver Lake during the summer of 1952. distribution by Aug. 25. On the latter date the stratum from 7.1 to 20.8 m. had conditions suitable for ciscoes, comparing favorably with 1952. On Sept. 6, 1930 (Scott 1931), the oxygen distribution present might be classified as weakly clinograde, with the cisco layer occurring from 6.5 to 17.0 m. On July 26, the only analysis in 1951, the oxygen distribution was minus-heterograde, and in late August, 1950, weakly clinograde. Hence, the type of oxygen distribution not only can vary within a single season, but can also vary from one year to another, depending upon the several factors controlling the bathymetric production and consumption of oxygen. As a consequence, the thickness of the zone habitable by ciscoes can also vary from month to month and from year to year. The "oxygen buffer capacity" of a lake is therefore important in helping the ciscoes survive through seasons of extreme oxygen utilization. Oliver Lake consistently has one of the most adequate oxygen supplies known for Indi- ana lakes, in spite of these variations in the type of oxygen distribution. Myers Lake Of all the lakes for which data are available, Myers Lake has the most pronounced plus-heterograde distribution. As evident in Figure 6, an oxygen maximum in the lower half of the thermocline was already well developed on June 26, and this prominent maximum persisted to at least Sept. 11. These rough data suggest that there were two peaks of photosynthesis at 8 m., one in late July and the other in late August. The cutoff in dissolved oxygen at times was very rapid below the photosynthetic to CO N 0 CO = to N N 0 N N I- I- Li >- ).- )- )- I-: 0_ ..°2 -I ...1 J Li Li Li 6 0. D -D D D IJ g n 1 , < < < < CI)

25.4° /

1 21.7° therrnochne June 26 thermocline 17.1 18.5 1 8.5 20. 19.8 16.7 1 6.4 Sept. 11

* 10 7.5° ...... 8.3 ..

0 15

MYERS LAKE — 1952 ppm. oxygen

FIG. 6. Oxygen distribution in Myers Lake during the summer of 1952. 215 maximum, with the result that there was usually virtually no oxygen below a depth of 10 or 11 m. The ciscoes at this time were necessarily living in the thermocline, and this is where they were caught with gill nets (see discussion on bathymetric distribution) . Not only did the plus-heterograde distribution in Myers Lake persist throughout the period of summer stratification, but it is also developed each year to about the same extent. A plus-heterograde distribution was observed on all occasions during the summers of 1951 through 1954. The factors involved in the development and maintenance of an oxygen maximum are not yet too well understood, but the consistent development of such a maximum is believed to be of very great importance in enabling the survival of cisco populations in some of the smaller lakes of the State. Little Tippecanoe Lake Little Tippecanoe (Fig. 7) is representative of a small number of cisco lakes and of the bulk of the non-cisco lakes with its clinograde oxygen distribution. The lake develops a very pronounced oxygen deficit in the hypolimnion. In middle August of 1952 there was very little oxygen below a depth of 7 m. The situation improved somewhat in late August and early September, although still not sufficiently to permit the survival of ciscoes below a depth of 7 m. Early in the summer the lake had a rather sharp oxygen maximum in the upper part of the thermocline, but this was completely dispelled within a space of only two weeks. Conditions such as occur in Little Tippecanoe very likely approach the extreme conditions that the species can tolerate. Even the shallower lakes of the Indian Village chain in Noble County do not confront the species with such severe demands for survival. On the basis of the arbitrary definition set up in this paper, the cisco layer in Little Tippecanoe was compressed by warm temperatures above and oxygen utilization below to a stratum only 0.1 m. thick on Aug. 11, and 0.2 m. thick on Aug. 18. One wonders how the species can possibly survive under such extreme conditions of stratification. Six fishermen contacted in the questionnaire reported that they have fished for ciscoes in Little Tippecanoe during the spawning season in late November and early December. In 1952 the Lake and Stream Sur- vey crew fished the lake with four experimental gill nets during the periods July 8-10, and Aug. 5-9. Bottom sets were made in water depths from 1.6 to 12.5 m., and diagonal sets were made from the surface to the bottom in water as deep as 10.6 m. The 'nets were certainly fished at all possible depths at which ciscoes might have been present. Yet not a single cisco was taken in the ten times the nets were lifted. Other species, however, were caught in or just above the cisco layer. One rockbass and one warmouth bass were caught at 5.8 m. Another warmouth was caught at 4.9 m., and five white suckers were caught at 5.2 m. Specimens of these and other species were taken quite commonly in still shallower water. ,r) 28 >- >- D D D D D D LI

JUNE —) < < < (/) - . .". . ..." ° 25.9 , : ...... ° 5 thermoc line ...... •• ...... 20.7 ... .. June 28 r ... thermodline Sept. 8 8.8° 1 0 8.3°

1 5

0.2 0.2 0.2 0.1 O. 02 0 0 5 20 ppm. oxygen LITTLE TIPPECANOE LAKE - 1952

Fro. 7. Oxygen distribution in Little Tippecanoe Lake during the summer of 1952. 217

What is not known is if the fish caught in Little Tippecanoe in late autumn have spent the summer in the lake. There is a possibility that they may have come from Big Tippecanoe, because Jordan (1875a) quoted the observations of Judge J. H. Carpenter that at spawning time the ciscoes "come in myriads into the streams which enter the lakes." Cahn (1927) also noted that ciscoes may ascend streams at spawning time and actually spawn in the rivers. As a specific example, some fish ascend the Oconomowoc River from Fowler Lake, even though the majority of the population spawns in the lake. It is difficult to understand, however, how such a "run" from Big Tippecanoe into Little Tippecanoe could maintain itself, because if the mid-summer oxygen conditions are inadequate for the survival of the species, then any fingerlings develop- ing from the eggs spawned in the lake would be doomed to early death. On the other hand, the fish might return to Big Tippecanoe Lake in spring before the unfavorable oxygen stratification develops. Kennedy (1941), for example, found that a number of species, including the closely related whitefish (Coregonus clupeaformis), moved in spring from the relatively shallow White Lake into the considerably deeper Big Trout Lake. The movement was believed to be induced by the warmer temperatures in White Lake. It is interesting that the largest individuals migrated first, a pattern similar to that observed by Fry (1937) for the downward spring movement of the cisco in Lake Nipissing. James Lake A further example of the oxygen and temperature distribution in cisco lakes, emphasizing this time the differences that can develop from one year to another, is afforded by James Lake (Fig. 8). On Sept. 20, 1930, the oxygen distribution was minus-heterograde, with only a trace of oxygen at a depth of 24.5 m. On Aug. 21, 1953, the distribution was plus-heterograde, and on Aug. 5, 1951, essentially b-orthograde. The vertical extent of the cisco layer is indicated by the solid portion of the oxygen curve. At no time was the temperature-oxygen distribution critical for the species, unless individuals were trapped below the oxygen notch on September 20, in the manner Scott (1931) postulated for Snow Lake. Thus, although probably all temperate holomictic lakes have an orthograde oxygen distribution during the spring overturn, relatively few of them are able to maintain this distribution over the summer. The most common condition in shallow stratified lakes is for a clinograde distribution to develop, either directly or after first passing through one of the heterograde types. A plus-heterograde distribution, especially when it is maintained throughout the summer, appears to offer the most favorable situation for maintaining the cisco in shallow lakes. Oxygen Type-Frequency Distribution In order to determine if there are any significant differences in the a significantdifferenceintheoxygendistribution betweentheciscoand oxygen distributionofciscolakesascomparedwithnon-ciscolakes,a same transparency,therefore,smalllakestheoretically aremorelikelyto less frequentlyaminus-heterogradedistribution. Themajorityofthe non-cisco lakes,andthisisparticularlytruefor thesmallerlakes.Al- type-frequency distributionwaspreparedforeachlake,basedonallthe cisco lakeslessthan60ha.insurfacearearegularly haveanoxygen cline creasing depthtotakeaccountofthefactthatinsmalllakes FIG. more frequentlytheyhaveaplus-heterogradedistribution, andsomewhat though anumberoftheciscolakesexhibitclinograde distribution,much large lakes.Themeandepthofthethermocline islistedforeachlake. have aphotosyntheticmaximumofoxygenin theirthermoclinethan years forwhichchemicaldataareavailable.Thesepresented this factorislargelyresponsible forthesurvivalofciscoinlakes maximum inthecoldwater ofthethermocline.Theauthorbelievesthat in Tables6and7.Ineachtablethelakesarearrangedorderofde- DEPTH— meters 25 20 1 1 5 5 0

Even asuperficialcomparisonofthetablesshows thatthereisindeed 20° portion oftheoxygencurvesdesignatesstratumwithatemperaturebelow in typeofoxygendistributionthatcandevelopfromoneyeartoanother.Thesolid

8. generallyliesclosertothesurfacethaninlarge lakes.Withthe

C.andanoxygencontentgreaterthan3ppm. Oxygen curvesforJamesLakeinthreedifferentyears,showingthevariation 2 46810 HETEROGRADE ppm. oxygen MINUS 20.IX.30 LAKE JAMES-3RDBASIN HETEROGRADE temperature 8 PLUS 218 21.V111.53 16

° C. 24

B -ORTHOGRADE thermo- 219

TABLE 6. Frequency of types of oxygen distribution in the cisco lakes of Indiana. Counties are designated by the first letters of their names.

Frequency of 0, types Mean depth of b- plus- minus- Area No. of thermocline ortho- hetero- hetero- clino- Lake County (ha.) years (en.) grade grade grade grade I. Lakes with cisco populations that are fished. James S 534 6 5.7-10.7 2 1 2 1 Clear S 310 5 6.6-11.2 4 1 Tippecanoe K 286 4 5.7- 9.7 4 Oliver L 145 5 5.4-10.7 2 1 2 Gage S 132 5 5.9-10.5 2 3 Snow S 120 5 5.2- 9.4 3 2 Little Tippecanoe K 108 2 4.5- 9.5 2 Dallas L 107 2 5.5- 8.5 2 Crooked W 77.7 4 4.8-10.5 2 Cedar w 58.3 2 3.5- 9.5 2 Lake of the Woods L 51.4 1 4-9 1 Shriner W 48.6 2 4-9.5 2 South Twin L 47.3 1 6-11 1 Secrist K 40.1 1 4-8.5 1 Myers M 34.8 4 4-10.5 4 Fish L 32.8 1 4-8 1 Knapp N 31.2 3 3.3-7.3 2 1 Lawrence M 24.7 1 3-9 1 Royer L 21.9 1 3-7 1 Gordy N 11.5 3 3.3-7.7 2 1 McLish S 6.1 2 3.5-8.5 2 Martin L 6.1 1 2-7 1 Eve L 6.1 2 4-9 2 Hindman N 6.1 2 2-5.5 2 Shock K 4.0 2 2-7.5 2 Indian Village N 2.0 3 2-6.3 1 2 II. Lakes reported to have ciscoes now (although not fished), or to have had ciscoes formerly. Big Turkey L 182 1 4-8 1 Big Long L 148 2 5-10 1 1 Whitmer L 87.0 1 4-9 1 Jimerson s 82.2 2 4.5-8.7 2 Big N 80.5 2 4.5-8 2 Atwood L 63.2 1 5-9 1 Round W 53.0 2 4.5-10 1 1 Long S 38.0 1 2-7 1 Messick L 22.3 1 3-9 1 Hackenberg L 15.0 1 0-8 1 as small as these. Even quite a number Of the larger lakes have a plus-heterograde distribution. Almost equally common among these larger, and for the most part deep, lakes, however, is a minus-heterograde distribution. If the oxygen notch results from the decomposition of current organic product arrested in its rate of sinking by the sharply increasing viscosity and density associated with a particularly steep thermal gradient, then the 220

TABLE 7. Frequency of types of oxygen distribution in the non-cisco lakes of Indiana. Counties are designated by the first letters of their names.

Frequency of 02 types Mean depth of b- plus- minus. Area No. of thermocline °mho. hetero- hetero. clino- Lake County (ha.) years (in.) grade grade grade grade Wawasee K 1200 1 9-12 1 Maxinkuckee M 751 2 7-10 2 Crooked S 325 2 6-10.5 1 1 Loon W 321 2 3-7.5 2 Hamilton S 309 2 5-9.5 2 Manitou F 289 1 3-8 1 Webster K 237 1 4-9 1 Winona K 215 2 5.5-10 1 1 George S 206 1 5-10 1 Chapman K 168 1 C-10 1 Dewart K 145 1 6-11 1 Adams L 115 4 5.8-10 1 3 Blue W 95.5 1 6-9 1 Big Barbee K 92.3 1 3-8 1 Waldron N 80.1 1 4-8 1 Pretty L 74.5 4 5.3-10.1 1 3 Pike K 74.5 2 3-7 2 Otter S 58.3 1 3-9 1 Yellow Creek K 57.7 3 3.3-8.7 3 Fox S 57.5 1 3-9 1 North Twin L 53.0 1 5-11 1 Hogback S 50.6 1 3-6 1 Bear N 50.2 1 4-8 1 Center K 48.2 3 4-8.3 3 Golden S 47.8 1 3-8 1 Bixler N 45.3 1 4-9 1 Diamond N 38.8 2 4-9.5 2 Olin L 38.5 1 4-9 1 Troy Cedar W 38.1 1 3-8 1 Silver S 37.9 1 4-10 1 Goose W 35.6 2 3-7.5 2 Hog S 35.6 1 3-9 1 Pretty M 34.4 2 4-8 2 Round N 33.6 2 4.5-9.5 2 Balls S 31.6 2 4-9.3 1 1 Little Long N 25.9 1 5-9 1 Hill K 23.9 1 5-9.5 1 Cook M 23.1 2 4.8-9.8 2 Pleasant S 19.4 2 5.5-10 2 Shoe K 16.0 1 4-8 1 Walters S 15.0 1 2-6 1 Center S 14.6 1 3-5 1 Latta N 14.2 1 5-9.5 1 Duley N 10.1 2 2-6 2 Black W 8.5 1 2-7 1 Hollem M 8.1 2 4-8 2 Cline L 8.0 1 4-9 1 Round S 6.1 3 4.3-10 1 1 1 Stevens K 5.7 1 2-6 1 Weir L 2.4 1 4-5.5 1 Thomas M 2.0 1 1-7 1 Note: Loon Lake (Steuben, 57.1 ha.) and Big Cedar Lake (LaGrange, 43.7 ha.) are 3rd order lakes with no water as cold as 20° C. 221

oxygen notch can be looked upon as a conserver of hypolimnial oxygen. The presence of an oxygen notch would then mean that the levels of oxygen in the hypolimnion are higher than they would be without such a notch, particularly if it is true that the hypolimnial oxygen deficit de- veloping in a given year is more closely controlled by the production of that year than by the accumulated organics of past years in the uppermost sediments. Einsele (1941) concluded this was true for Schleinsee and probably also for all other lakes in which allochthonous organic matter is at a minimum. A clinograde distribution is developed primarily in the deepest cisco lakes. In many of these, such as James, Clear, and Oliver, the clinograde distribution is only weakly developed, being somewhat intermediate between the clinograde type (sens. str.) and the b-orthograde type. It is interesting that in the smallest cisco lake known, Indian Village Lake, the oxygen distribution in two of three years was clinograde, rather than plus-heterograde, as in the other small cisco lakes. Hence, a plus-heterograde distribution is not an absolute sine qua non for the survival of the cisco in small lakes. The non-cisco lakes are quite uniformly clinograde, usually strongly so. A few were weakly plus-heterograde in one or two years. Adams and Pretty lakes in LaGrange Co. were each plus-heterograde in only one of four years. This condition was associated with a somewhat thicker stratum habitable by ciscoes. The plus-heterograde condition listed for Fox Lake was observed on July 10, and for Hog Lake on July 23. It is not known if this condition persisted through August into the partial fall overturn, although this is likely for Fox Lake, because the oxygen maximum here was strongly developed. Olin, Pretty (Marshall Co.), and Latta lakes seem to be suitable for ciscoes if the weakly plus-heterograde condition observed in them develops regularly each year, as it may well do in Pretty Lake. Aside from these few lakes and Manitou, which although clinograde still had a fair amount of oxygen in deep water in the single year examined, there are no lakes included in this study which very likely contain cisco populations even though not reported by the fishermen. Thus, the oxygen-temperature stratification in the various lakes attests to the general reliability of the questionnaire reports by the fishermen. • Relation of Oxygen Typology to Summer Surfacing and Mortality Scott (1931) was particularly interested in Snow Lake because the ciscoes there are reported to surface early in September in almost every year. He postulated that the fish become trapped in the oxygenated layer below the oxygen notch, and that, as the oxygen content of this lower layer is still further used up, the fish, unable to tolerate the conditions any longer, undergo temporary asphyxiation and rise to the surface. Scott did not believe that these fish died, but rather that after adjusting physiologically to the new conditions, they remained in the epilimnion 222 until fall turnover. Scott diagrammed the oxygen distribution in Snow Lake on Sept. 14, 1929, and Sept. 7, 1930, to illustrate his point. On the former date there was no oxygen below a depth of 9 m., except for a trace of about 0.15 ppm. at 12 m. According to the opinion of the present paper the ciscoes would have been eliminated from the hypolimnion long prior to this time. On Aug. 9, the last date in 1929 for which data are tabulated, there is no suggestion of a minus-heterograde distribution. Likewise, on Sept. 7, 1930, there is a very minor nick in the oxygen curve at 11 m. amounting to only 0.43 ppm. (not even listed in the table for this date), and in the other August and September series for this year (except possibly Sept. 2) there is no trace of an oxygen notch. The oxygen distributions might best be classified as clinograde, or even weakly plus-heterograde because of the slight yet distinct increase in oxygen content from the surface down to depths of 7 and 6 m., respectively, in the two years. In neither year does the very weak oxygen notch in September represent, therefore, a final stage of a persistent minus-heterograde oxygen distribution. The data presented by Scott do not seem to support the theory he developed. The cisco fishermen reported similar early September surfacing of the species (with mortality) to occur irregularly in Tippecanoe, Crooked ( Whitley Co.), James, Jimerson, Snow, Clear, Lake of the Woods, Dallas, Little Tippecanoe, McLish, and even Myers lakes. More fishermen did report them surfacing in Snow Lake than any of the others, however. A number of specific instances are : Tippecanoe in 1949, 1950, and 1954; Clear Lake on Labor Day a few years ago ; Little Tippecanoe in 1948; Snow Lake in 1947 and 1948; Dallas Lake in August of one year ; Myers Lake in 1953; etc. The point is that many of these lakes according to the data available do not develop a minus-heterograde distribution. Tip- pecanoe Lake, which has the most pronounced minus-heterograde oxygen distribution known from Indiana lakes, represents the only lake that might correspond in fact with Scott's theory to explain late summer surfacing and mortality. Hence, regardless of how reasonable the theory seems to be, there are no satisfactory data from Snow Lake or elsewhere to support it. The oxygen notches in Snow Lake considered by Scott do not seem to have much real physiological significance for the species, and September surfacing (and mortality) of the cisco occurs in quite a number of Indiana lakes representing a variety of types of oxygen distribution.

DISCUSSION AND CONCLUSIONS The coregonids as a group are generally considered to be polyoxybiontic cold stenotherms, primarily boreal in distribution. The cisco of North America and its European counterpart (C. albula) conform less closely to this generalization than any other species reported on in the literature, although probably the young of all species are less demanding in their environmental requirements than the adults. Contrary to the belief prevalent among limnologists and fishery biologists, the cisco is 223 not restricted in its distribution to cold water with high levels of dissolved oxygen. The species obviously is as plastic physiologically as it is morphologically, and, particularly among residual 200)populations, its toleration of fairly high temperatures (up to perhaps and of fairly low levels of dissolved oxygen (certainly down to 3 ppm., and possibly even lower) is well documented by both the present and a number of previous studies. In our inland lakes the cisco spawns in shallow water in November- December when the water temperature has dropped to slightly below 4° C. (Cahn 1927) . During the winter the fish are apparently generally distributed through the lake, occurring abundantly towards the surface. As the water warms in spring the fish begin to migrate from the surface water into the colder water below the incipient thermocline. In Lake Nipissing (Fry 1937) at least, the entire population does not migrate downward at the same time, but rather the migration occurs in a quite orderly sequence according to size and sex. The largest individuals of either sex migrate earliest, with the males of a particular size moving in general before the females. The two youngest age groups move downward last. Once the fish get into the deep water in this lake and in other cisco lakes, they tend to remain there until the time of the fall partial overturn, or even until the complete overturn. In any given lake the species tends to occur in the coldest water having adequate levels of dissolved oxygen. When high levels of oxygen are available, the summer bathymetric distribution may in some instances involve only the lowermost part of the hypolimnion and in others the entire hypolimnion. As the lake ages typologically and the oxygen content of the hypolimnion gradually declines, the fish are forced progressively higher in the hypolimnion. The steep temperature gradient of the thermocline, however, constitutes a formidable block to further upward dis- placement, and apparently only under conditions of extreme oxygen deficit in the hypolimnion does the species penetrate the overlying stratum of rapidly changing temperature. A number of instances are known based in part upon the presumed tolerance limits of the species to temperature and oxygen, where the fish do live in the thermocline in summer, and even in the upper part of the thermocline. The most extreme conditions of survival occur in those lakes where during the height of summer stagnation there is no stratum in the lake with a temperature less than 20° C. and an oxygen content as great as 3, or even 2, ppm. Even under these extreme conditions, however, the fish still react negatively to high temperatures. The epilimnion-thermocline junction in Indiana lakes usually has a temperature in the neighborhood of 20°. Towards the surface temperatures are higher. The downward transport of oxygen by turbulent diffusion and by more general circula- tion is such that in all lakes known to contain ciscoes this boundary stratum has adequate levels of dissolved oxygen for the species. Possibly the cisco, particularly the young of the year, can actually survive over 224 extreme summers in a thin stratum above the thermocline, as defined by Birge, although beyond this the species even with its great physiological adaptability cannot go. There are no known lakes in Indiana or elsewhere in which adult ciscoes are generally distributed through0 the epilimnion in summer when the water temperatures are above 20 . At some point in the aging of a lake the tolerance limit is surpassed, and the local population is snuffed out. Probably long before this time is reached the population has become greatly reduced in numbers, and significant reproduction occurs only at relatively infrequent intervals. Such a residual population with an atypical age structure, including the complete absence of one or more of the younger age groups, would be expected to be less capable of surviving a period of unusually extreme thermal and chemical stratification than a more normal population. Cisco populations in northern Indiana are not confined to the largest and deepest lakes ; indeed, four of the cisco lakes have a known maximum depth of less than 10 m. Indian Village and Hindman lakes with mixi- mum depths of 6.5 and 6 m., respectively, are the shallowest known cisco lakes. It is not believed, however, that the populations in these and many other of the shallower cisco lakes are existing under the most extreme conditions the species can endure, because there is a tendency in the shallower cisco lakes in particular for significant photosynthesis to occur in the region of the thermocline during summer, thereby providing adequate levels of dissolved oxygen at fairly low water temperatures. This factor is believed to be of importance in maintaining the species in many of the shallower lakes. The most extreme conditions are believed rather to occur in those lakes that exhibit a steadily declining oxygen content below the epilimnion. In these lakes the hypolimnion and even much of the thermocline can be completely lacking in oxygen or have only very low levels remaining by the height of summer stagnation. In a number of such lakes represented in this report, both in Indiana and Wisconsin, there was on occasion in summer no stratum of water with a temperature less than 20° C. and an oxygen content as great as 3 ppm. Assuming that all the natural lakes in the region concerned in this report either actually contained ciscoes or were accessible to colonization by the species in early post-Cary time, regardless of whether the particular lakes were ultimately located in the Great Lakes or the Mississippi River drainage basin, then those lakes still containing ciscoes represent the only ones in which conditions within the tolerance range of the species have persisted during the intervening millenia. The interesting fact regarding the distribution of cisco lakes in Indiana is that the lakes are not randomly distributed through the lake district but are grouped together in definite clusters, suggesting that particular patterns of geologic and geomorphic factors constitute the chief controls in their distribution. Through the action of these regional factors the cisco lakes tend to have a greater buffer capacity or safety factor for the cisco regarding oxygen 225 depletion in cold water, and also to have a more frequent positive assimilation balance in the thermocline in summer, resulting in higher levels of oxygen here than would otherwise occur. Dr. W. J. Wayne of the Indiana Geological Survey has suggested that the clusters of cisco lakes tend to be located in regions of glacial outwash rather than of till. The surround- ing surficial material would therefore tend to be size sorted, probably with a greater permeability and a relatively smaller percentage of very fine particles. The manner in which these factors, or others associated with them, might influence the development of the oxygen stratification described in this paper is not yet comprehended. Thienemann (1950) believes that the European equivalent (Corego- nus albula L.) of the North American cisco originally fed on benthos, but that as the oxygen content of the hypolimnion declined it was forced upward away from the bottom, and consequently had to modify its feeding habits. The fine gill rakers preadapted the species for feeding on zooplankton. In the oligotrophic lakes where there is adequate oxygen in the deep water but little plankton the present plankton-feeding species does not do very well. In the strongly eutrophic lakes where there is insufficient oxygen for the species in deep water, the surface waters even though containing adequate oxygen and an abundance of zooplankton are either so warm that the species has been eliminated altogether, or else the populations are very small and marginal. At the intermediate mesotrophic stage of lake development a large zooplankton occurs along with adequate oxygen in the upper hypolimnion, and it is in such lakes that the populations tend to be densest and most successful. The same lim- nogenetic controls are claimed to operate for the stint (Osmerus eperlanus L.) in the north German lakes. Certainly of all the coregonids in North America the cisco is best able to maintain itself under conditions of relatively high temperature and relatively low oxygen. During the gradual typological aging of a lake the cisco population probably alters its physiology correspondingly. If the matter were properly studied we would surely find the populations in different lakes to be as variable physiologically as they are morpho- metrically. Moreover, a considerable part of this variability might well be the same kind of direct and short-term response or adaptation to environmental conditions that Hile (1936b, 1937) discovered for the morphometry of the species in northern Wisconsin.

LITERATURE CITED

Aberg, Mile, and Wilhelm Rodhe. 1942. tber die Milieufaktoren in einigen siidschwedischen Seen. Symb. Bot. Upsalienses 5(3) :1-256. Bauch, Gerd. 1949. Untersuchungen iiber das Wachstums der kleinen Marane (Coregonus albula L.) in den Gewdssern Mitteleuropas. Abhandl. aus d. Fischerei u. deren Hilfswissensch., Lief. 2:239-326. . 1953. Die einheimischen Siisswasserfische. Neumann Verlag: Radebeul and Berlin. 187 pp. 226

Birge, E. A., and Chancey Juday. 1911. The dissolved gases of the water and their biological significance. Wisconsin Geol. & Nat. Hist. Surv., Bull. 22: xx + 259 PP. Black, E. C., F. E. J. Fry, and Virginia S. Black. 1954. The influence of carbon dioxide on the utilization of oxygen by some fresh-water fish. Canadian Jour. Zool. 32:408-420. Blatchley, W. S. 1902. Lake systems of Indiana. Bienn. Rept. Comm. Fisheries and Game for Indiana, 1901-02: 146-232. . 1938. The fishes of Indiana. Indianapolis: Nature Publ. Co., 121 pp. Blatchley, W. S., and G. H. Ashley. 1901. The lakes of northern Indiana and their associated marl deposits. 25th Ann. Rept. Dept. Geol. and Nat. Resources of Indiana, 1900:31-321 Calm, A. R. 1927. An ecological study of southern Wisconsin fishes. The brook silverside (Labidesthes sicculus) and the cisco (Leucichthys artedi) in their relations to the region. Illinois Biol. Monogr. 11 (1) :1-151. Clark, F. N. 1894. History and methods of whitefish culture. Bull. U. S. Fish Comm. for 1893, 13:213-220. Dennis, W. T. 1892. Bienn. Rept. State Fish Comm. of Indiana, 1891-92:49-53. Dryagin, P. A. 1939. Ryapushki Pustoshkinskikh ozer Kalininskoy oblasti. [Coregonus albula of the Pustoshkinsky lakes in the Kalininsky district.] Tr. VNIORKh, t. 22. (Original not seen.) Dymond, J. R. 1933. Biological and oceanographic conditions in Hudson Bay. 8. The coregonine fishes of Hudson and James Bays. Contr. Canadian Biol. and Fish. 8(1) :1-12. 1943. The coregonine fishes of Northwestern Canada. Trans. Roy. Canadian Inst. 24(2) :171-231. Eichler, H. 1940. Ein bemerkenswertes Maranensterben. Zeitschr. f. Fisch. 38(3):387-390. Eigenmann, C. 11. and C. II. Beeson. 1894. The fishes of Indiana. Proc. Indiana Acad. Sci. for 1893:76-108. Einsele, Wilhelm. 1941. Die Umsetzung von zugefiihrtem, anorganischen Phosphat im eutrophen See und Ihre Rtickwirkungen auf semen Gesamthaushalt. Zeitschr. f. Fisch. 39:407-488. Evermann, B. W., and H. W. Clark. 1920. Lake Maxinkuckee. A physical and biological survey. Indiana Dept. of Conservation: Vol. I, 660 pp.; Vol. II, 512 pp. Evermann, B. W., and IL M. Smith. 1896. The whitefishes of North America. Rept. U. S. Comm. Fish and Fisheries, 20:283-324 Fry, F. E. J. 1937. The summer migration of the cisco, Leucichthys artedi (LeSueur), in Lake Nipissing, Ontario. Univ. Toronto Studies, Biol. 44. Publ. Ontario Fish. Res. Lab. 55:1-91. Gerking, S. D. 1945. Distribution of the fishes of Indiana. Invest. Indiana Lakes and Streams 3(1) :1-137. Gutermuth, C. R. 1938. Official Indiana Lake Guide. Indiana Dept. Cons., Div. Fish and Game. 56 pp. Hay, 0. P. 1902. The lampreys and fishes of Indiana. Bienn. Rept. Comm. Fisheries and Game for Indiana, 1901-02:62-119. Rile, Ralph. 1931. The rate of growth of fishes of Indiana. Invest. Indiana Lakes 1(2):7-55. . 1936a. Age and growth of the cisco Leucichthys artedi (LeSueur) in the lakes of the Northeastern Highlands, Wisconsin. Bull. U. S. Bur. Fish. 48:212-317. . 1936b. Summary of investigations on the morphometry of the cisco, Leucichthys artedi (LeSueur), in the lakes of the Northeastern Highlands, Wis- consin. Pap. Michigan Acad. Sci., Arts & Let. 21:619-634. 227

. 1937. MORPHOMETRY of the CISCO, Leucichthys artedi ( LESUEUR), in the lakes of the Northeastern Highlands, Wisconsin. Int. Rev. HYDROBIOL. 36(1/2):57-130. Hile, Ralph, and Chancey JUDAY. 1941. BATHYMETRIC distribution of fish in lakes of the Northeastern Highlands, Wisconsin. Trans. Wisconsin Acad. SCI., Arts & Let. 33:147-187. JIIRVI, T. H. 1950. Die KLEINMARANENBESTANDE in IHREN BEZIEHUNGEN ZU der DMWELT ( Coregonus albula L.). Acta ZOOL. FENNICA 61:1-116. Jordan, D. S. 1875A. The SISCO of Lake Tippecanoe. Amer. Nat., March 1875, 9(3):135-138. . 1875B. The SISCO of Lake Tippecanoe and its relatives. 6TH Ann. Rept. Geol. SURV. of Indiana, 1874:187-196. . 1875C. Synopsis of the genera of fishes to be looked for in Indiana. 6TH Ann. Rept. Geol. SURV. of Indiana, 1874:197-228. 1877. On the fishes of northern Indiana. Proc. Acad. Nat. SCI. Phila- delphia for 1877, 29:42-82 . 1878. Catalogue of the fishes of Indiana. Ann. Rept. Indiana State Bd. AGRIC. for 1877, 19:362-369. 1883. Catalogue of the fishes of Indiana. BIENN. Rept. Comm. Fisheries of Indiana, 1881-82:96-103. . 1892. The CISCO of Indiana and its relatives. BIENN. Rept. Comm. Fisheries and Game for Indiana, 1891-92:135-142. Jordan, D. S., and B. W. EVERMANN. 1886. The food fishes of Indiana. Ann. Rept. Indiana State Bd. AGRIC. for 1885:156-173 . 1911. A review of the SALMONOID fishes of the Great Lakes, with notes on the whitefishes of other regions. Bull. U. S. Bur. Fish. 29:1-41. JUDAY, Chancey. 1914. The HYDROGRAPHY and MORPHOMETRY of the lakes. Wisconsin Geol. & Nat. Hist. SURV., Bull. 27:XV -I-- 137. Kennedy, W. A. 1941. The migration of fish from a shallow to a deep lake in spring and early summer. Trans. Amer. Fish. Soc. 70:391-396. Kirsch, P. H. 1894A. BIENN. Rept. State Fish Comm. of Indiana, 1893-94:25FF. . 1894B. A report upon explorations made in Eel River basin in the northeastern part of Indiana in the summer of 1892. Bull. U. S. Fish Comm. 14:31-41. . 1895. Report upon investigations of the Maumee River basin during the summer of 1893. Bull. U. S. Fish Comm. 14:315-337. 1896A. BIENN. Rept. State Fish. Comm. of Indiana, 1895-96:134. . 1896B. A report upon investigations made in the lakes and streams of Whitley County, Indiana. BIENN. Rept. State Fish Comm. of Indiana, 1895-96:21-79. KOELZ, Walter. 1931. The COREGONID fishes of northeastern America. Pap. Michigan Acad. SCI., Arts & Let. 13:303-432. LEVETTE, G. M. 1876. Observations on the depth and temperature of the lakes of northern Indiana. 7TH Ann. Rept. Geol. SURV. of Indiana for 1875:467-503. Lower, W. I. 1913. The CISCO of the deeper water lakes of northern Indiana. BIENN. Rept. Comm. Fisheries and Game for Indiana, 1911-12:21-27. McDonald, Marshall. 1887. Notes upon fish and the fisheries. Bull. U. S. Fish Comm. 7(3):33-48. Meek, Alexander. 1916. THE MIGRATIONS OF FISH. London: Edward Arnold, XVIII 427 pp. Meek, S. 0. 1908. List of fishes known to occur in the waters of Indiana. BIENN. Rept. Comm. Fisheries and Game for Indiana, 1907-08:134-171. Miles, P. H. 1915A. BIENN. Rept. Comm. Fisheries and Game for Indiana, 1913- 1914 :18. . 1915B. The lakes of Indiana. BIENN. REPT. Comm. Fisheries and Game for Indiana, 1913-14:72-230. 228

Nelson, M. N., and A. D. Hasler. 1942. The growth, food, distribution and relative abundance of the fishes of Lake Geneva, Wisconsin, in 1941. Trans. Wisconsin Acad. Sc., Arts & Lett. 34:137-148. Pearse, A. S. 1921a. Distribution and food of the fishes of Green Lake, Wis., in summer. Bull. U. S. Bur. Fish. 37:253-272. 1921b. The distribution and food of the fishes of three Wisconsin lakes in summer. Univ. Wisconsin Stud. Sci., No. 3:1-61. Scott, Will. 1916. Report on the lakes of the Tippecanoe Basin (Indiana). Indiana Univ. Stud. 3 (31) :1-39. . 1931. The lakes of northeastern Indiana. Invest. Indiana Lakes 1(3):57-145. Slastenko. 1931. Opyt akklimatizatsii siga v prudovykh khozyaistvakh Ukrainy. [Experiment on the acclimitization of the whitefish in the pond farms of the Ukraine.] Za Sotsialisticheskoye rybnoye khozyaistvo, 11-12. (Original not seen.) Sukhoverkhov, F. M. 1943. Opyt vyrashchivaniya ripusa i ryapushki v prudakh. [Experiment on the culture of Coregonus albula and its variety vimba in ponds.] Zool. Zhurnal 22(2) :77-91. Thienemann, August. 1918. Untersuchungen iiber die Beziehungen zwischen dem Sauerstoffgehalt des Wassers und der Zusammensetzung der Tiefenfauna in norddeutschen Seen. Arch. Hydrobiol. 12:1-65. 1928. tiber die Edelmarane (Coregonus lavaretus forma generosus Peters) und die von ihr bewohnten Seen. Arch. Hydrobiol. 19(1):1-36. . 1933. Coregonus albula lucinensis, eine Tiefenform der kleinen Marane aus einen Norddeutschen See. (Zugleich em n Beitrag zur Rassenbildung bei Coregonus albula.). Ztschr. Wiss. Biol., Abt. A, Ztschr. Morph. okol. Tiere 27(4):654-683. . 1950. Verbreitungsgeschichte der Siisswassertierwelt Europas. Die Binnengewasser 18:xvi + 809 pp. Willer, Alfred. 1924. Die kleine Marane (Coregonus albula L.) in Ostpreussen. Int. Rev. Hydrobiol. 12: 248-265. 1929. Neune biologische Beobachtungen iiber die kleine Marane (Coregonus albula L.). Zeitschr. f. Fisch. 27:251-269.