BULLETIN OF THE FLORIDA STATE MUSEUM BIOLOGICAL SCIENCES Volume S Number 2 THE AMERICAN CYPRINID FISHES OF THE SUBGENUS LUXILUS (GENUS NOTROPIS) Caner R. Gilben d.:0 m *. 0 0 UNIVERSITY OF FLORIDA Gainedvllle 1964 Numbers of the BULLETIN OF THE FLORIDA STATE MUSEUM are pub- lished at irregular intervals. Volumes contain about 800 pages and are not necessarily completed in any one calendar year. , OLIVER L. AUSTIN, JR., Editor WALTER AuFFENBERG, Managing Editor Consultant for this issue: Donn Eric Rosen Communications concerning purchase or exchange of the publication and all manuscripts should be addressed to the Managing, Editor of the Bulletin, Florida State Museum, Seagle Building, Gainesville, Florida. . 1 Published 2 June 1964 ~ Price for this issue $1.45 THE AMERICAN CYPRINID FISHES OF THE SUBGENUS LUXILUS (GENUS NOTROPIS) CARTER R. GILBERT 1 SYNOPSIS: The cyprinid fishes of the subgenus Luxilus (genus Notropis) comprise three well-defined species groups of eight species and one subspecies. They occupy clear, small to medium-sized streams over most of temperate North Amer- ica east of the Rockies, being absent only froni areas southwest of the Red River drainage in Texas, the upper Missouri basin, the southern Piedmont plain, and peninsular Florida. The present study defiries the subgenus, reviews its nomen- elatural history, explains its patterns of variation, describes the various forms and their distributions, and discusses their probable zoogeographic and phylogenetic histories. TABLE OF CONTENTS Introduction --__-__-_--_-___._.-_-_--_.... 96 Notropis zonistius ____.______.___ 126 Acknowledgements ...._-_.._-.-_-_._ 97 Notropis zonatus 129 Materials 98 Notropis Pilsbryi 188 Methods of Counting and Notropis cerasinus 186 Measuring ___..._...._...__..__. 99 Notropis cornutus --_._-- _._.-- 140 Hybridization __._..._.__._.__._._. 101 Notrovis albeolus ___._.__._.. 151 Zoogeography and Phylogeny ____ 104 Notropis chr!/socephalus ____-___ 157 Generic and Subgeneric Notropis chrusocephalus Nomenclatural History ......_. 110 chrt,socephalus .- 158 Subgeneric Diagnosis and Notropis chrusocephatus: Specific Relationships ___ 114 chnmocephalus X isolepis 166 Key to Species and Subspecies Notropis chrusocephalus of the Subgenus Luxilus _-__ 117 isolepis . __._.____-_._._._.- 167 Species and Subspecies Accounts 122 Tables 178 Notropis coccogenis ._____.- 122 References Cited ---_....._.--.... -.. 185 1 This paper is a revision of a dissertation submitted to the University of Michi- gan in 1960 in partial ful81]ment of the Ph.D. degree. The work was carried out while the author served as a teaching fellow in the Department of Zoology and as a research assistant in the Division of Fishes of the Museum of Zoology. Now at the University of Florida, he is Assistant Professor of Biological Sciences and Assistant Curator of 'Natural Sciences in th6 Florida State Museum. Manu- script received 5 July 1963.-ED. Gilbert, Carter R. 1964. The American cyprjnid fishds of ' th€ ~gtnus-Luxilus (Genus Notropis). Bull. Florida State Mus., vol. 8, 80.2,· pp. 195-194.· / 96 BULLETIN FLORIDA STATE MUSEUM Vol. 8 INTRODUCTION Of the many species of cyprinid Ashes inhabiting eastern North America, few are better known to the layman than the Common Shiner, Notropis cornutus. This is due to its abundance, wide distribution, large size, and striking coloration. Nevertheless the systematic re- lationships of the forms referred to this and to closely related species have long perplexed students of North American freshwater fishes. The group that includes Notropis comutus and its most closely related forms (here referred to as the subgenus Luxilus) is in a rel- atively early stage of evolution. Within the Luxilus complex may be found all stages of speciation, and in at least two cases considerable subjective judgment is needed to determine whether forms are in- cipient or actual species. The Luxilus group thus furnishes an ex- cellent opportunity for observing and understanding the basic proc- esses of change which occur during evolutionary history. The members of the subgenus Luxilus have had an unstable ·no- menclatural history. Their recent evolution and consequent close inter-relationships have made the arrangement of species difficult. D. S. Jordan, who was mainly responsible for previous attempts to classify these groups, moved species back and forth among genera and subgenera, usually with little apparent reason. Although morphological criteria have been the primary bases for the resolution of various taxonomic problems encountered in the pres- ent study, zo6geography, ecology, and clinal variation also have been taken into account. This has resulted in the recognition of 8 species and 1 subspecies instead of 6 species and 4 subspecies as in the cur- rently accepted classification. Both arrangements are listed below. Current classification Proposed classification coccogenis group N.otropis coccogenis (Cope) Notropis coccogenis ( Cope) Notropis Zonistius (Jordan) Notropis. zonistius (Jordan) zonatus group Notropis zonitus zonatus (Agassiz) Notropis zonatus (Agassiz) Notropis zonatus pilsbrt/i Fowler Notropis pitsbryi Fowler cornutus group Not¥opis cerasinus (Cope) Notropis cerasinus (Cope) Notrepis cornutus cornutus (Mitchill) Notropis cornutus (Mitchill) 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 97 Current classification Proposed classifcation cornutus group (continued) Notropis cornutus frontah (Agassiz) Notropis albeolus Jordan Notropis a?beotus Jordan Notropis cornutus chr!/socephakts Notropis chrusocephaZus chrusoceph- (Rafinesque) alus (Rafinesque) Notropis cornutus isolepis Notropis chrusocephalus boleps Hubbs and Brown Hubbs and Brown The subgenus Luxilus is one of the widest ranging of the minor cyprinid groups in North America. Its species occur in fresh waters from the Gulf Coast north to 52° N. latitude in Manitoba and from Nova Scotia to the eastern slope of the Rocky Mountains. The only major area east of the Mississippi River from which it is absent is the southeastern Atlantic coastal plain from central North Carolina south- ward through peninsular Florida. To the southwest it does not oc- cur beyond the Red River drainage. All species ordinarily inhabit rather clear, small or medium-sized streams with gravel to rocky bottoms and moderate to swift grad- ients. Although some populations may be found living in extremely turbid waters, this is abnormal and usually temporary. Hybridization between species of the Luxilus complex is frequent and occurs in many combinations both within the subgenus and with species of other subgenera and genera. These are reviewed, and the possible causes and effects of hybridization are discussed. ACKNOWLEDGMENTS This study was made possible by the aid and cooperation of many people. I am especially grateful to Dr. Reeve M. Bailey for his ad- vice, criticism, and professional guidance throughout the course of the study; to my wife, Nancy, for her constant encouragement and invaluable help in the preparation and completion of this thesis; to Mrs. Elizabeth M. Anthony who prepared the drawings as well as the letters and margins for the distribution maps; to Dr. Ernest A. Lachner for much helpful advice regarding the preparation of this manuscript; and to Dr. Clarence L. Smith for Beld assistance and for most of the photographs included in this paper. I am indebted to the following persons for the use of specimens and equipment under their care: Dr. Edward C. Raney, Cornell University; Drs. Leonard P. Schultz, Ernest A. Lachner and William Ralph Taylor, United States National Museum; Dr. James E. B6hlke, Academy of Natural Sci- 98 BULLETIN FLORIDA STATE MUSEUM Vol. 8 ences of Philadelphia; Dr. Milton B. Trautman, Ohio State Archeo- logical Museum; Dr. Royal D. Suttkus, Tulane University; Dr. Philip W. Smith, Illinois Natural History Survey; and Dr. Shelby D. Gerking, Indiana University. Sincere thanks are extended to Mrs. H. Williams Diamond, Jr., who helped type the manuscript; to Dr. Edward J. Crossman, Royal Ontario Museum of Zoology, for supplying certain locality records for Canada; to Dr. George A. Moore, Oklahoma State University, for several Oklahoma records; to Dr. Clark Hubbs, Uni- versity of Texas, for all Texas records; to Dr. Frank B. Cross, Univer- sity of Kansas, for several Kansas records; and to Dr. Ralph W. Yer- ger, Florida State University, for several Florida records. Apprecia- tion is extended to Dr. Robert R. Miller, University of Michigan, for helpful advice and for the use of an unpublished list of hybrid fish combinations; to Dr. Karl F. Lagler, University of Michigan, for the use of equipment under his care; to Dr. Louella Cable, for the use of facilities at the U. S. government fish hatchery, Northville, Michigan; to Dr. Robert D. Ross, Virginia Polytechnic Institute, for supplying color notes for Notropis albeolus; and to Dr. William Ralph Taylor, for advice regarding the study. Finally, I am grateful to my fellow graduate students, Thomas L. Poulson, Teruya Uyeno, R. Jack Schultz and Howard F. Huddle for their help in the field and laboratory. MATERIALS The collections examined were selected to include samples from all parts of the ranges of the various forms, with special emphasis being placed on Series from critical areas. Specimens have been examined or recorded from the following museum collections: Acad- emy of Natural Sciences of Philadelphia (ANSP); Chicago Natural History Museum (CNHM); Cornell University (CU); Illinois Natural History Survey (INHS); Indiana University (IU); Kansas University (KU); Museum of Comparative Zoology, Harvard University (MCZ); Nova Scotia Provincial.Museum (NISPM); Ohio State University (OSU); Royal Ontario Museum (ROM); Stanford University (SU); Texas Natural History Collection, University of Texas (TNHC); Tulane University (TU); University of Florida (UF); University of Michigan, Museum of Zoology (UMMZ); United States National Museum (USNM). Locality records of specimens examined may be found in the original dissertation. 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 99 METHODS OF COUNTING AND MEASURING To reduce sample bias usually no more than 10 specimens were counted from a single series, except in lots from certain critical or poorly collected areas, where larger numbers were used. As certain relative body proportions change with increase in size, body measure- ments were made only on specimens falling within certain size ranges in each species. For the larger forms (N. cornutus, N. albeolus, N. chrysocephalus, and N. coccogenis) specimens measuring approxi- mately 60 to 80 mm in standard length were used; for the smaller species (N. zonistius, N. cerasinus, N. zonatus, and AT. pilsbryi) the limits were 55 and 75 mm. Counts ahd measurements were taken as outlined by Hubbs and Lagler (1947: 8-15) except that the mouth gape was determined by the transverse distance between the posterior angles of the mo.uth, instead of the greatest trahsverse distance across it. To eliminate difficulties in counting fine-scaled species, a new method was de- vised for taking scale counts on the anterior dorsal part of the body. The usual predorsal scale count on the dorsal midline directly in front of the dorsal fin is entirely satisfactory for coarse-scaled species, but overlaps and irregularities in fine-scaled forms make accurate deter- minations difficult. The scale rows along the side of the body are more regular than the predorsal scales, and provide a more accurate and effective assessment of differences in scale size among popula- tions. Anterior dorsal-lateral scale counts were made along the longi- tudinal row of scales lying approximately three-fourths of the distance from the lateral line to the mid-dorsal line. This varied from the third to the sixth row above the lateral line, depending on the size of the scales. The scales were counted from a point directly under the origin of the dorsal fin forward to their termination. The sum of this count and the body circumferential scale count is of great value in interpreting differences between the various species, par- ticularly those in the N. Cornuttls species group. Scale counts for both the circumferential and the caudal peduncle series are usually expressed as total numbers, but may also be writ- ten to indicate the relative disposition of scales above and below the lateral line. Thus a circumferential count written as 14-2-10 signifies 14 scales between and above and 10 scales between and below the lateral lines on each side of the body. 100 BULLETIN FLORIDA STATE MUSEUM VoL 8 Proportional measurements are expressed in thousandths of the standard length (shortened hereafter to SL). These measurements were tabulated initially in the form of frequency distributions by ma- jor drainage areas for each member of the subgenus except N. cerasinus. When inspection revealed no consistent geographical var- iation in morphology except in N. albeolus, the proportional measure- ments for all the others were combined in the final tabulations. Certain meristic counts for N. cornutus, N. albeohts, N. c. chryso- cephalus, and N. c. isolepis were separated in the tables by major drainage areas. This was done because the wide ranges of these forms increases the possibilities for clinal or genetic variation. The better to show the overall differences between N. cornutus living east and west of the Appalachians, all populations in each of these two areas were combined, with three exceptions: Those in the upper Susquehanna and the St. Lawrence river systems are on or near the drainage divide between the Atlantic Ocean and the Mississippi; specimens from the streams in Maine and the Maritime provinces of Canada show rather unusual meristic characters. Counts from all areas were combined in the overall comparison of the Luxilus group. Vertebral counts were determined from X-ray photographs. The procedures follow the methods discussed by Bailey and Gosline (1955). The Weberian apparatus, regarded as containing four fused vertebra, and the hypural plate were included in the counts. All photographs were made with type M film and a «hard ray" machine (Miller, 1957). The angle of the mouth to the top of the head was determined with the aid of a special ocular as described by Smith (1956: 250). With the pointer set at a 0° reading, the Bsh is placed on the stage so that one crosshair (arbitrarily called the horizontal hair) in the ocular lies directly over the length of the lateral opening of the mouth, with the intersection of the crosshairs slightly anterior to the tip of the snout. The ocular is then rotated clockwise until the vertical crosshair touches the head at the posterior edge of the occiput. The angle is then read in degrees to the nearest half degree. Inspection revealed that usually about 90 percent of the counts showed relatively little deviation from the mean of a normal distri- bution, while the extreme values dropped away rather sharply. As it is manifestly desirable to show this condition in a species account, the method shown in the following hypothetical examples was adopted to express it: A frequency distribution expressed as· (22) 25 to 29 (33) indicates that 90 percent of the counts fall between 25 and 29, with 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 101 the extremes 22 and 83; a distribution expressed as (8) 9 or 10, usually 9, shows that less than 10 percent of the counts number 8 and, while a count of 10 is not particularly rare (more than 10 percent) 9 is a much more common value. To determine possible sexual variation, approximately equal num- bers of males and females were measured for the various morpho- metric characters. In most cases the means and range of variation differed so little and the overlap was so great that measurements were lumped in the final tabulations. Tables and lists of locality records in the original dissertation not regarded as essential have been omit- ted. Only new names or new combinations of names have been re- tained in the synonymy. HYBRIDIZATION Numerous hybrid combinations involving members of the Luxilus complex are known. Three of these, N. c. chrysocephalus X N. cor- nutus, N. c. chrysocephalus X N. rubellus, and N. cornutus X N, rubellus, are the most common cyprinid hybrids in eastern North America. The cornutus X rubellus hybrid was described by Jordan (1889a: 354) as a new species, Notropis macdonaldi, and the chryso- cephalus X rubellus cross was erroneously recorded by Forbes (1909: 886) and by Forbes and Richardson (1920: 149) as Notropis pilsbryi. A number of papers have appeared dealing entirely or in part with various aspects of hybridization in certain species of Luxilus. Among the more comprehensive and important of these are Raney's (194Ob) account of the comparative spawning behavior of N. cornutus and N. rubellus in New York, and Miller's (1962; 1968) papers deal- ing with the comparative morphology, sexual development, and hermaphroditism of N. cornutus X N. rubellus. In my own (196lb) treatment of the relationships of N. cornutus and N. c. chrysocephalus I list the arguments for and against specific differentiation of col'nutus and chr!/socephalus, and favor their recognition as species, although the two forms often act as subspecies. This evidence is as follows: 1. Intergrading populations of subspecies normally show a per- feet blending of morphological characters. Although this is ap- proached in some associations containing both comutus and Chryso- cephalus, in other populations the two forms remain morphologically distinct, with few apparent intermediates. In addition many popu- lations within the zone of "intergradation" contain only one form. 102 BULLETIN FLORIDA STATE MUSEUM Vol. 8 This mosaic pattern is not concordant with the normal expectation for subspecies. 2. Competition is, in effect, nonexistent between subspecies, and where one has moved into a region formerly occupied solely by an- other, intergrading populations occur. The presence of isolated relict populations of cornutus surrounded by populations of chr!/so- cephalus far to the south of the principal range, together with the observed gradual replacement of cornutus by chrysocephalus in parts of Ohio, indicate that competition exists where these Rshes occur together. 8. Chrysocephalus apparently has disappeared from several mixed populations in southern Michigan following construction of dams near the mouths of the creeks where it formerly occurred. If cor- nutus and chrysocephalus were only subspecifically distinct, the phe- notypic character of the populations would be expected to remain more or less constant. 4. The experimentally-verified differences in temperature toler- ance between cornutus and chrysocephalus in the northern limits of their respective ranges are greater than one might normally expect in subspecies. . The apparent preference of cornutus for headwaters and of chrysocephalus for the lower reaches of streams furnishes fur- ther evidence of this difference. 5. The distributional patterns of these forms suggest a long sepa- ration of cornutus and chiysocephalus, ample to permit evolution to the species level, and possibly dating back to late Pliocene. Raney's paper includes a discussion of the morfhology, fertilit:y, hybrid vigor, and sex ratio of the cornutus X rubellus hybrids. He decided, as did Hubbs and Brown (1929: 86-87), that hybridization between these species is accidental, and usually results from simul- taneous spawning over the nests of other species , such as Hybopsis micropogon, Semotilus atromaculatus, and Exoglossum maxillingua. In the absence of such nests rubellus and cornutus spawn in different parts of the same riffie, rubellus in mid-water near the head, and cornutus on the bottom toward the base. Milt from rubellus males is washed downstream by the current, where it unites with eggs of cornutits. The latter observations are similar to those of Moore and Paden (1950: 92) for Notropis pilsbryi and Dionda nubila. Hybrids of this combination apparently arise as a result of the simultaneous 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 103 spawning of Notropis pilabryi in rifiles and Dionda nubila in pools at the bases. The following is a list of the known hybrids involving members of the subgenus Luxilus. Some have been reported in the literature and are so indicated; those which have not yet appeared in the pub- lished record are indicated by catalog numbers only. I have made n6 attempt to verify the accuracy of all these identifications, nor have I included all literature references. Current nomenclature is used in all cases. Notiopis albeolus X Notropis mbetlus (UMMZ 174754) Notropis cerasinus X Chrosomus oreas (Raney and Lachner, 1946: 226) Notropts chrusocephabs chrt/socephdus X Campostoma anomalum (Trautman, 1957: 114) Notropts chrusocephalus chrusocephalus X Hybognathus nuchalis (Luce, 1988: 116; O'Donnell, 1985: 482) Notropis chrysocephalus chrusocephalus X Hybopsis biguttata (Luce, 1988. 115; O'Donnell, 1985: 482) Notropis thrusocephalus chrusocephalus X HI/bopsis micropogon (Trautman, 1957: 114) Notropis chrysocephalus chrusocephalus X Notropis chrosomus (UMMZ 174883, UMMZ 175614) Notropis chrysocephalus chrusocephalus X Notropis cornutus (Gilbert, 196lb: 181-192) Notropis chrusocephalus chrusocephalus X Notropis Zeuciodus (UMMZ 175194) Notropis chrysocephalus chrgsocephalus X Notropis rubellus (Forbes, 1909: 886 [as N. pilibrvil; Forbes and Richardson, 1920: 149 [as N. pilsbrvi]; Hubbs, 1955:10; Hubbs and Brown, 1929: 35-86; Hubbs and Moore, 1940: 96 [identification of N. pilsbr~/il; Murray, .1988: 84; O'Donnell, 1985: 482; Trautman, 1948: 171 [in part]; Trautman, 1957: 114-115 [in part]) Notropis cornutus X Campostoma anomalum (Trautman, 1957: 114) Notropis cornutus X Chrosomus eos (UMMZ 78827) Notfopis cornutus X Chrosomus oreas (Raney, 1950: 167) Notropis cornutus X Clinostomus elongatus (Greeley, 1988: 51; Greene, 1985: 88-89, Koster, 1939: 207; Raney, 1940a: 7) Notropis cornutw X Clinostomus funduloides (Raney, 1950: 167) Notropis cornutus X Hubopsis biguttata (Taylor, 1954: 48) Notropis coniutus X HI/bopsia micropogon (Creeley, 1988: 51; Raney, 1940a: 7; Raney, 1940c: 185; Trautman, 1957: 114) Notropis cornutus X Notropis photogenis (Greeley, 1988251) Notropis cornutus X Notropis rubettus (Goldsborough and Clark, 1908: 85 [as N. macdonaldil; Greeley, 1988: 51; Hubbs, 1955: 10; Hubbs and Brown, 1929: 35-86; Hubbs and Moore, 1940: 96 [identification of N. macdonaldi]; Jordan, 1889a: 854 [original description of N. macdonaldi]; Lachner, 1952: 486; Raney, 1940a; 7; Raney, 194Ob: 861-367; Raney, 1940c. 135; Raney, 104 BULLETIN FLORIDA STATE MUSEUM Vol. 8 1950: 165, 167; Taylor, 1954: 43, Trautman, 1948: 171 [in part]; Trautman, 1957: 114-115 [in part]) Notropis cornutus X Semotilus atfomaculatus (Bailey and Oliver, 1939: 172, Greeley, 1988: 51-52; Koster, 1939: 207; Raney, 194Oa: 7; Raney, 1940c: 135; Simon, 1946:91; Taylor, 1954: 42) Notropis cornutus X Semotilus corporalis (Bailey, 1988: 150-151; Greeley, 1988: 52; Raney, 194Oa: 7) Notropis pilsbrvi X Dionda nubila (Moore and Paden, 1950: 92) Notropis pilsbFyi X Notropis rubellits (Cross, 1954:813, Moore and Paden, 1950: 92) ZOOGEOGRAPHY AND PHYLOGENY The zo6geographic and phylogenetic history of a group of animals, often the subject of speculation, is usually reconstructed only with difficulty. Because the genus Notropis has evolved 56 recently, most of the forms linking the included natural groups are still extant and allow a relatively clear interpretation of the phylogenetic relation- ships. On the other hand, the high degree of speciation in the genus makes the zo6geographical relationships extremely complex. The subgenus Luxilus probably evolved in the Mississippi basin, perhaps during the early to middle Pliocene. This hypothesis is based first on the present geographical distribution of the Luxilus complex and, second, on the assumption of a Miocene invasion of North America by the family Cyprinidae (Romer, 1945: 588). The primary division within the subgenus most likely involved the precursor of Notropis coccogenis and Notropis zonistius. Judging from its present distribution, the coccogenis group probably evolved in the Tennessee River system. The marked differentiation of cocco- genia and zonistius from the other species of Luxilus indicates that isolation from stocks in the Mississippi Valley lasted for a consider- able time. The reasons for isolation, however, are unknown at pres- ent. Later, presumably still during the Pliocene, a segment of this stock reached the Chattahoochee River basin, there evolving into N. zonistius, while the stock remaining in the Tennessee system gave rise to N. coccogenis. Presumably this transfer was effected through a localized stream capture rather than by direct connection of the Chattahoochee with the Tennessee. No evidence indicates such a connection, and these river systems show little close faunal similarity. Also during the Pliocene another segment of the Mississippi Val- ley stock presumably became is6lated in the Ozark region. This eventually gave rise to the zonatus group, which includes Notropis 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 105 zonatus and Notropis pilsbrgi. The cause of this isolation also is un- known, although it probably is related to the fact that the Ozarks contain clear, cool, rather swift-flowing streams. Unlike some other Ozarkian endemics, Notropis zonatus and Notropis pilsbryi have no immediate relatives living in upland areas east of the Mississippi River. Consequently these species are believed to be autochthonous for the Ozark region. The distributions of N. zonatus and N. pilsbryi suggest a long sep- aration between the Missouri and White river systems that has pre- vented gene interchange. This interpretation is supported by the parallel distribution patterns of other species pairs, such as £theo- stoma euzonum-E. tetrazonum. The presence of N. zonatus in the Blaek River system strongly suggests that this river at one time was tributary to the Missouri, and that a reversal of flow has occurred rather recently, long after zonatus and pilsbryi separated. Records of zonatus from the headwaters of the St. Francis and Little rivers of southeastern Missouri are best explained on the basis of stream captures, although the possibility exists that these streams also once Rowed to the north. Probably the most perplexing aspect of the zo6geography of these species has to do with the presence of pilsbryi in the Arkansas and Red river drainages. The distribution of sev- eral species (Notropis spilopterus and Lepomis megalotis) indicates a definite faunal relationship between the Illinois and Neosho rivers (tributaries of the Arkansas) and the Ozarkian tributaries of the Mis- souri River. The absence of Notropis zonatus from the Illinois and Neosho is an apparent contradiction, inexplicable save for the possi- bility that if zonatus did invade these river systems it could not com- pete with the well-established population of pilsbrgi. Hubbs and Moore (1940: 94) have suggested the isolated populations of pilsbrui in the Red River system may have been introduced. It is equally possible that they represent relict populations, especially as they have been recorded from three separate localities in this drainage. Prior to the Pleistocene epoch the drainage pattern of the Ohio Valley differed considerably from that of today. The Ohio River extended little east of Cincinnati, and what is now the upper Ohio flowed northward, probably to a river in the area of the present Lake Erie basin. Larger than either of these streams was the Teays River, a major prolongation of the Kanawha River. The Teays flowed north- westward across· Ohio, Indiana, and apparently Illinois, and Illence southward to the present Mississippi Valley. It occupied a wide, deep trench comparable in size with that of the lower Ohio River today (Flint, 1947: 166). 106 BULLETIN FLORIDA STATE MUSEUM Vol. 8 A number of species that are completely or largely restricted to the Roanoke River system, which drains into the Atlantic Ocean, have affinities not with other Atlantic coastal forms, but with species in the Ohio Valley. The headwaters of the Kanawha River are narrowly separated from those of the Roanoke over a fairly extensive area in western Virginia. Much of the present Roanoke fauna, including Notropis cerasinus and, possibly, N . albeolus, almost certainly reached this drainage by headwater transfer from the oid Teays system' dur- ing preglacial time (Wright, 1984: 65), though whether the invasions by these species occurred simultaneously cannot be determined. Evolution of the common N. cornutus-N. chrysocephalus stock probably took place in the Mississippi Valley. This hypothesis is sup- ported by the distribution pattern of these two species, which cen- ters ih this region. Some of the common cornutus-chrysocep/talus stock is believed to haye become separated from the main popula- tion in late Pliocene in one of the following areas: the Atlantic coast, the upper Mississippi Valley, the Great Plains, or the Great Lakes. The relative distributions of the two species suggest this segment evolved into N. Cornutus, while that part of the population to the south became N. chrysocephalus. There is little evidence that N. cornutus evolved in either the Mis- sissippi Valley 6r on the Atlantic coast, except for its presence there today. Geological evidence (Flint, 1947: 163-167; and 1957: 168-170) indicates that before the Pleistocene the middle and upper parts of the Missouri did not flow into the Mississippi River as they do today, but instead "flowed north into the Souris-Assiniboine system, perhaps discharging via the Lake Winnipeg depression, the Nelson, and the Hudson Bay region. The advent of the ice sheet flowing from the northeast blocked all this drainage and detoured the Missouri along the ice margin. Contact was then established with the Kansas River below Kansas City, thus effecting a new outlet for the upper Missouri River. The upper Missouri therefore provided an isolated region where evolution of Notropis cornutus could have occurred, but its apparent absence from the upper Missouri drainage today argues strongly against this area as a place of origin. Radforth (1944: 10-11) describes and pictures a hyp6thetical course for the Laurentian River, which drained the present Great Lakes re- gion (with the possible exception of Lake Superior) in pre-Pleistocene time. Possibly a segment of the cornutus-chrusocephalus ancestral stock entered this river system during late Pliocene and there evolved into N. cornutus. This Laurentian center of origin for N. cornutus 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 107 appears to be the best of the several possibilities. In addition to other factors cited previously, spread to such distant areas as the Atlantic coast and the Rocky Mountains could have been accomplished most readily from a geographically intermediate area. Following its initial invasion of the Atlantic coast the species spread southward, possibly to the Roanoke River. The population may then have become isolated from the remaining cornutus stock farther north, resulting in the eventual evolution of a new form, Notropis albeolus. The presence of other closely related species and subspecies pairs in the James and Roanoke systems, such as Etheo- stoma longimanum-Etheostoma podostemone, Moxostoma fhothoecum- Moxostoma hamiltoni, and Notropis procne procne-Notropis procne longiceps, tends to confirm this idea. It is equally plausible, as in- dicated earlier, that albeolus reached the Atlantic coast via the Teays River. This is substantiated by the similarity in distribution pattern of this form with other species , such as Notropis ardens and N. ma- tutinus, which are restr~cted to the Roanoke and neighboring systems on the east coast, but whose afBnities are with the Ohio Valley fauna. The differentiation of albeolus into two rather distinct forms, one occurring in the Roanoke and New drainages, the other in the Neuse, Tar, and Cape Fear systems, suggests that albeolus may even have evolved in some way from a union of stocks coming both from the Atlantic coast and the Ohio Valley. The first glacial advance completely inundated the Laurentian River and thus split the range of cornutus into two major segments, one along the Atlantic coast, the other west of the glacier. Whether cornutus reinvaded the glaciated area during all the interglacial pe- riods, which is a possibility, cannot be determined. Some remixing of segments of the Atlantic coastal and western populations of cor- nutus probably did occur during the Pleistocene, for no taxonomic separation can be made between the eastern and western popula- tions of cornutus at the subspecific level. If these populations did not come into contact throughout the entire Pleistocene, the long pe- riod of isolation should have produced more pronounced morpho- logical differences. Furthermore cornutus is so widespread and adapt- able that it would be among the most likely species to expand into newly available territory. Dispersal of N. chrysocephalus to various parts of the Mississippi Valley evidently occurred throughout the -Pleistocene. The Missis- sippi River was then slightly cooler, less turbid, and less sluggish, and probably did not pose the barrier to small-stream fishes that it 108 BULLETIN FLORIDA STATE MUSEUM Vol. 8 does today. Certain parts of the Mississippi could have once acted as "filter bridges" for their movement. Lowered ocean levels during the glacial periods permitted many freshwater streams to join before reaching salt water, for example those draining into the Gulf of Mexico from the Alabama sy*tem west to the Mississippi. During one of the glacial advances some chruso- cephalus stock in the Mississippi Valley apparently moved through the lower reaches of the Mississippi into some of the Gulf coastal streams. When, with the retreat of the glacier, the sea level rose and restored the Mississippi to its former proportions, the chryso- cephalus stock isolated in the coastal streams evolved into a new form, isolepis, here regarded as a well-defined subspecies of chrysocephalus One can only speculate when the original invasion of the coastal streams by chrysocephalus occurred, but the estimited time neces- sary for a new form of this complex to evolve suggests it was probably during one of the first two glacial periods. Much later, after the morphological separation of chrusocephalus and isolepis was com- plete, the two forms came together again when the Alabama-Coosa system captured certain tributaries of the Tennessee. Van der Schalie (1945: 357-858) discusses the geological and malacological evidence showing the relationships of these two systems. Notropis cornutus either survived in the upper Ohio Valley throughout the Pleistocene or reached there during one of the inter- glacial periods. Absence of this species from the upper Kanawha River system suggests a fairly recent invasien, either from headwater capture or through overflow streams from ponded lakes formed by the advancing Wisconsin glacier. The further advance of the Wis- consin ice sheet (Flint, 1947: 288) isolated the Ohio Valley popula- tions of cornutus from both the Atlantic coast and the western popu- lations. As the Wisconsin ice Sheet retreated, meltwater lakes formed at the glacial margin. Because the glacier prevented northward flow, the rising lakes began to cut outlets to the south. One of the largest, Lake Agassiz, covered an extensive area and drained into the Mis- sissippi Valley by means of the Warren River outlet (Radforth, 1944: 18), near the present southern extremity of the Red River of the North. As the glacier retreated farther, a channel finally opened to the northeast, the origin of the Nelson River. This stream was situated at a lower elevation than the older Warren River, and thus formed a new outlet for Lake Agassiz, which consequently decreased 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 109 rapidly in area and was superseded by Lake of the Woods, Lake Win- nipeg, and many lesser lakes. Soon after formation of Lake Agassiz, there appeared farther to the east glacial lakes Duluth, Chicago, and Maumee, the precursors, respectively, of lakes Superior, Michigan, and Erie. These bodies, also plugged at their northern extremities, drained into the Missis- sippi Valley by way of the St. Croix outlet (Lake Duluth), the Fox and Chicago outlets (Lake Chicago) and the Fort Wayne outlet (Lake Maumee). With further glacial retreat changes continued until the present-day Great Lakes were eventually formed. N. cornutus probably moved northward via the Warren River outlet into Lake Agassiz, and reinvaded the newly-formed Great Lakes from the south through glacial channels such as the St. Croix and Fox rivers (Greene, 1985: 15) and from the north via glacial Lake Agassiz through the Ft. William Outlet in western Ontario (Underhill, 1957: 31). Notropis cornutus apparently was able to follow the re- treating glacier more closely than many other species and, as a result was again able to invade the east coast, this time via the Horseheads outlet (in the Lake Ontario drainage) into the Upper Susquehanna system (Bailey, 1945: 125-126). There it again mixed with the popu- lation of cornutus which had remained on the Atlantic coast presum- ably throughout the Pleistocene. At this time ocean levels still were much lower than at present. The shallow bottom of Chesapeake Bay was above water, and all streams from the Susquehanna south to and including the James were interconnected. Although influence of the cornutus stock that crossed onto the Atlantic coast may have extended into the Chesapeake Bay area, the bulk of the western gene pool probably remained near the point of original entry. Proof that the preceding sequence of events occurred depends in large part on the usual absence of pigmentation on the chin and gular region in cornutus from the upper Susquehanna. This character, while not consistent enough to be taxonomically name- worthy, nevertheless is of value in the determination of zo6geograph- ical relationships among populations. Pigment usually is lacking west of the Alleghenies, but not in coastal streams. N. comutus may possibly have utilized a temporary outlet between the forerunner of Lake Ontario and the Mohawk River. Apparently some species did enter the Mohawk in this way, but the history is blurred by the construction of the Mohawk-Erie canal across the drainage divide in the early 1800's . Notropis cornutus seems to have moved a considerable distance eastward from its western glacial refu- 110 BULLETIN FLORIDA STATE MUSEUM Vol. 8 gium, and this stock may have given rise, at least in part, to the present population in the Canadian maritime provinces. With the openings of the Chicago and later the Maurnee outlet into glacial lakes Chicago and Maumee, N. chrysocephalus also reached the Great Lakes drainage, there to liv6 sympatrically with Notropis cornutus, possibly for the first time since the stocks separated in the Pliocene. Apparently chrysoceplmlus did not enter the Lake Ontario drainage till well after cornutus. By this time the Horse- heads connection was broken, which prevented chrysocephalus reach- ing the Atlantic coast via the upper Susquehanna. The general warming of streams in N6rth America since the re- treat of the Wisconsin ice sheet, and particularly with man's activities during the past century or so, has wrought significant changes in the relative distributions of N. cornutus and N. chrysocephalus which are still continuing. In the Ohio Valley at the height of the Wisconsin advance, cornutus was apparently the dominant, if not the sole repre- sentative of the Luxilte complex above roughly the mouth of the Wa- bash River. As stream conditions changed chrysocephalus invaded this area, and the resulting competition virtually eliminated cornutus from the lower Ohio Valley, though isolated relict populations re- main in the lower Kanawha River system in southern West Virginia and in the White River system in southern Indiana. The gradual re- placement of cornutus by chrusocephalus in eastern Ohio during the past 85 years has been well documented by Trautman (1957: 857). GENERIC AND SUBGENERIC NOMENCLATURAL HISTORY Luxilus Rafinesque, 1820a: 47-48 (original description of' genus and subgenus; type species, Luxilus chrysocephalus Rafinesque, by subsequent designation of Jordan and Gilbert, 1877). Girard, 1856: 202-203 (name mistakenly allied with Notemigonus). Jordan, 1876a: 94 (Luxilus chrusocephatus syn- onymized with Luxitus cornutus) Jordan, 1876b: 286-287 (Hypsilepis syn- onymized with Luxilus) Jordan and Copeland, 1876: 184, 153 (type species said to be Luxilus cornutus (Mitchill), but that species was not among those included in Luxilus by Rafinesque; Hypsitepis a synonym) Jordan, 1877: 12, 28-31 (Luxilus chrusocephalus a synonym of Luxilus cornutus; review of Rafin- esque's species; type species, Luxitus cornutus). Jordan and Gilbert, 1877: 86 (type species, Luxilus chrusocephalus Rafinesque). Jordan, 1882: 852-854 (Luxilus divided into three subgenera: Luxilus, Coccotis, and Alburnops). Jordan, 1885: 814 (subgenus of Notropis). Jordan, 1929: 82 (Luxilus restored to generic status). H!/psolepis Agassiz, 1854: 859 (original description of genus Hypsotepis; type species, by original designation, Leuciscus cornutus; name attributed to 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 111 Baird). Cope, 1864: 279 (spelling emended to Hypsilepis). Jordan, 1876b: 283 (Hypsilepis synonymized with Luxilus). Jordan and Copeland, 1876: 158 (Hypsilepis synonymized with Luxitus). Plargyrus Rafinesque, 1820a: 50-51 (substitute name for Rutilus if included spe- cies are not in that genus; type species, by absolute tautonymy, Rutilus plar- gyrus). Kirtland, 1845: 26 (refers to species Leuciscus plargyrus). Storer, 1846: 158 (refers to species Leuciscus plargyrus). Girard, 1856: 196 (Plar- g!/rus accorded generic status). Jordan, 1876a: 94 (Rutilus plargyrus Rafines- que wrongly synonymized with Luxilus kentuckiensis). Jordan, 1876b: 281 (incorrectly allied with part of present subgenus Cyprinella). Jordan, 1877: 28-29 (Plargyrus Girard [not of Rafinesquel placed in synonymy of Luxilus; review of Rafinesque's Species). Coccotis Jordan, 1882: 852 (proposed as a subgenus of Luxitus; type species, by original designation, Luxilus coccogenis). Jordan, Evermann and Clark, 1930: 127 (geheric status). Coccogenia Cockerell and Callaway, 1909: 190-191 (proposed as a subgenus of Notropis on basis of scale characters; type species by original designation, Notropis coccogent4. The genus Notropis Rafinesque (1818) is a large and diverse group, containing numerous closely related, morphologically similar forms. At various times in the past its principal components have been re- garded as distinct, though poorly-defined genera. Gilbert (1884b) was the first to propose that a number of these be combined under the name Notropis. Jordan (1885b) relegated these groups to sub- generic status within the broad genus Notropis, where they have remained ever since despite sporadic efforts (Jordan, 1929; Jordan, Evermann and Clark, 1930) to separate them. Although some of the chief subgroups of Notropis are difficult to separate and the relationships of some are not yet clear, several clusters of species share morphological characters and obviously con- stitute natural units. The subgenus Luxilus Rafinesque (1820a) is one such group. The species of Luxilus are characterized by common diagnostic features, but the diversity of their past assignment to genera shows that ichthyologists have not always recognized their close interrelationship. Some have been moved indiscriminantly from group to group without apparent reason. Part of the difficulty regarding the status of Luxilus stems from the original description (Rafinesque, 182Oa: 47-48) which, as is often true of Rafinesque's accounts, is quite general and vague. Of the four species he included in this group, only one (L. chrysocephalus) is now assigned to Luxilus as presently defined. 112 BULLETIN FLORIDA STATE MUSEUM VoI. 8 His description of Rutilus plargyrus in the same paper shows this species probably is identical to L. chi'ysocephalus, despite its assign- ment to a different genus. The description is accompanied by the following statement: "I call this genus Rutilus in the supposition that Cyprinus rutilus may be the type of it; if it should be otherwise, it may be called Plargyrus." In 1817 Mitchill described Cyprinus cornutus from the Hudson River drainage in New York. This species is not closely related to the genus Cyprinus, a fact that was soon recognized, and it was as- signed to the old-world genus, Leuciscus. Agassiz (1854), realizing that the affnities of col'nutits were not with Leuciscus either, erected a new genus, Hypsolepis (this name has been attributed to Baird), with col'nutus as type species. Plargyrus Rafinesque was employed by Girard (1856: 31-82) to include the species cornutus. Girard states: "Since Rdnesque's genera are to be restored, his name Plargyrus is to take the precedence over the genus Hypsolepis of more modern coining. The name of Plargyrus was provided for in the IchthyoZogia Ohiensis, to replace that of Rutilus in the eventuality that Cyprinus rutilus of Europe, which was the type of the genus Rutilus, should prove generically distinct from Rutilus plargyrus and similar American species, and which is the case. Girard evidently misunderstood the true relationships of Lux- ilus chrysocephalus, for he regarded this species as a close relative of Notemigonus crysoleucas, a species with no close relatives among the North American Cyprinidae. Apparently Girard was alone in his use of the name Plargyrus, for subsequent investigators such as Putnam (1868), Cope (1864, 1867, 1869a, 1869b, 1870, and 1871), Abbott (1870 and 1874), and Jordan (1875) continued to use HUps61epis (spelling changed to Hypsilepis by Cope, 1864) for cornutus and its related forms. Jordan (1876a: 94), in his first review of Rafinesque's Ichthyologia Ohienms, concluded that both Luxilus chrysocephalus and Rutilus plkirgyrUS Of RaBnesque were synonyms of Hypsilepis cornutus . The genera Luxilus and Rutilus both were described prior to Hypsitepis, however, and as Rutilus contains only old-world species not closely related to N. cornutus, Hypsilepis must Obviously be placed in the synonymy of Luxilus. He made no mention of Girard's (1856) paper which resurrects Plargyrus as a genus for cornutus. Jordan (1882: 852) redefined the genus Luxilus and included in it "three well-marked subgenera": Luxilus, Coccotis, and Alburnops. "He defined Luxilus as large species, with the scales very closely im- bricated, and much deeper tlian long; the dorsal over the ventrals 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 118 and jaws even; type L. coinutus. In Coccotis he proposed to include "rather large species, with the scales less closely imbricated, but still deeper than long; dorsal behind ventrals and lower jaw projecting; type L. coccogenis." The third subgenus, Alburnops, he described .as small species with the scales normal; the jaws equal and the dor- sal somewhat behind the ventrals; type L. blennius." The name Coccotis was not used for many years, and its existence was overlooked by Cockerell and Callaway (1909) when they erected the subgenus Coccogenia for the sole reception of N. coccogenis; based entirely on a supposed difference in the character of its scales. The name has not been used since. Jordan and Gilbert (1888) combined all the forms now considered to belong to Notropis and which have 2,4-4,2 teeth into the genus Minnilus. As extensive variation in dental arrangement exists among closely related species of Cyprinidae (Gibbs, 1957: 186), this scheme of classification was unrealistic and resulted in. the splitting of a number of natural groups. Jordan (1884: 284) again referred cornutus and its related forms, including coccogenis, to Luxilus. Gilbert (1884b: 210) proposed that Notropis Rafinesque (1818) replace Minnilus as the generic name for a number of species previously referred to many separate genera. In this Jordan (1885b: 810-814) concurred and published a subgeneric classification to distinguish the various groups within Notropis This system was followed with minor changes by virtually all ichthyologists until 1929, when Jordan again split Notropis and ele- vated its numerous subgenera to full generic status. He again re- garded Luxilus as a genus for cornutus and its closest relatives, but he placed N. coccogenis, solely on the basis of Scale characters, in the genus Hydrophlox, an artificial group containing a number of unre- lated species. Jordan, Evermann and Clark (1930) retained Luxitus as a genus. They also resurrected as a genus Jordan's Coccotis, which had been proposed as a subgenus 48 years earlier, and placed in it the species coccogenis, zonistius, zonatus, brimle!/i, and macdonaldi. No reasons were given for setting up this group. Few compilers and no practicing ichthyologists have followed either Jordan's or Jordan, Evermann and Clark's classification con- sistently. No structural or other basis has been discovered for a meaningful division at the generic level of this young and cohesive group. The current trend is to defer judgment on the relationships between the various subgenera of Notropis until a thorough system- atic study is made of the entire genus. 114 BULLETIN FLORIDA STATE MUSEUM Vol. 8 SUBGENERIC DIAGNOSIS AND SPECIFIC RELATIONSHIPS The subgenus Luxilus is characterized by the fdllowing diagnostic features: anal rays modally 9 (10 in specimens of N. zonistitis from southern part of range); pectoral rays 14 to 17; lateral-line scales usually 89 to 41; caudal peduncle scales usually 14 to 16 (7-2-5 to 7); peritoneum black or heavily speckled with black; pharyngeal teeth 2,4-4,2; dorsal fin inserted directly above or slightly behind anterior insertion of pelvic fin; anterior lateral-line scales and adjacent scales taller than wide, notably so in some forms; dark pigmentation present on scale pockets along side of body in most species, resulting in the appearance of black crescent-shaped vertical streaks; mouth large, - terminal and oblique; size relatively large; body form in most forms more or less compressed. SPECIES ~ SUBSPECIES N. cornutus N. albeolus N. chrysocephalus chrysocephalus N. chrysocephalus isolepis N.cerasinus N. zonatus N. pilsbryi N. coccogenis N.zonistius Figure 1. Hypothetical Phylogeny of the Subgenus Luxuus. 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 115 Three well-defined species groups are recognizable within the subgenus: The coccogenis group contains two species (N. coccogenis and N. zonistius), as does the zonatus group (N. zonctus and N. pils- bryi); the cornutus group consists of Bve forms (N. cornutus, N. albe- olus, N. chrysocephalus chrysocephalus, N. chrysocephalus isolepis, and N. cerasinus). The first two species pairs are completely allo- patric, but cornutus and its immediate relatives exhibit variable geo- graphic relationships. The coccogenis group is defined as follows: body strongly com- pressed or not; peritoneum heavily speckled with black pigment but never uniformly black; dorsal fin inserted slightly behind pelvic; scales along anterior-lateral part of body somewhat taller than wide; poste- rior borders of scales along side of body smooth and without scallops; nuptial tubercles weakly developed, nearly granular in some areas; tubercies present in two or three rows along lower jaw, present on tip of chin, on upper part of head concentrated mainly along side and tip of snout; tubercles either present or absent on top of head or back; tubercles present in both sexes, though poorly developed in females; crescentic vertical streaks moderately developed, not evi- dent in some individuals; a red bar present on cheek or on margin of opercle and preopercle; caudal spot present or absent; a consistent pattern of dark pigmentation in the dorsal fin, either present or absent in the caudal fin; circumferential scales usually 18 to 16-2-10 to 12; caudal peduncle scales normally 7-2-5 or 6; no black lateral stripe on side of body. The zonatus group has the following characters: body terete, not strongly compressed; peritoneum uniformly black; dorsal En inserted slightly behind pelvic; scales along anterior-lateral part of body only slightly taller than wide; posterior edge of scales along side of body each with four or five definite scallops; nuptial tubercles weakly de- veloped; tubereles granular, present in a single row along lower jaw, absent from tip of chin, present on top of head; tubercles pres- ent in both sexes, though weakly developed in females; crescentic vertical streaks absent from side of body; no red bar present on cheek or on margin of opercle and preopercle; caudal spot absent; no per- sistent pattern of dark pigmentation in either dorsal or caudal fin; circumferential scales usually 12 or 18-2-12 or 18; caudal peduncle scales normally 7-2-5 or 6; a distinct black lateral band present along side of body and encircling snout; a narrow, short secondary stripe running parallel and slightly dorsad to main lateral band. The cornutus group may be diagnosed as follows: body deep and compressed, this character being quite variable among different popu- 116 BULLETIN FLORIDA STATE MUSEUM Vol. 8 lations of albeolus; peritoneum uniformly black; dorsal fin inserted directly above pelvic; scales along anterior-lateral part of body much taller than wide, becoming more pronounced with age; posterior edges of scales along side of body smooth and without scallops; nuptial tubercles well developed, the tips usually straight, sometimes curved forward; tubercles present in a single row along lower jaw, absent from tip of chin, present over entire head and along back as far as, and including, anterior part of dorsal fin; tubercles normally present only in breeding males (weakly developed in females of N. cerasinus); crescentic Vertical streaks usually well developed along side of body, but either absent or weakly developed in N. albeolus; no red bar pres- ent on cheek or on margin of opercle and preopercle; caudal spot ab- sent; no consistent pattern of dark pigmentation present in either the dorsal or caudal Bns; circumferential scales usually 12 to 17-2-11 to 15; caudal peduncle scales usually 7-2-7 (7-2-5 or 6 in N. cerasinus); a black lateral stripe sometimes present on body of young but not consistently, present throughout life. The three species groups do not seem extremely close to one an- other. Their morphological differences are fairly sharp, and no con- necting links exist between them. The relationships within each group, on the other hand, appear generally quite close, particularly in the zonatus and, to a lesser extent, the coccogenis groups. With one exception this holds true for the cornutus group also. N. cerasi- nus, although obviously much more closely related to N. cornutus and N. chrysocephalus than to any of the other species of Luxilus, differs notably from them in the character of the nuptial tubercles and their normal presence in females, in the slight difference in appearance of the scales, and in the number of caudal peduncle scales. The closest relatives of Luxitus are those species of Notropis that share with it the largest number of similar or identical morphological characters. To base a relationship on only one or two shared features can be misleading. Some of the most constant diagnostic characters in the forms of Luxilus-pharyngeal tooth counts, anal ray counts, color of peritoneum, shape of scales along side of body, size of eye, and relative position of dorsal and anal fins-vary considerably among other closely related Nokopis species . Notropis ariommus seems rather close to the subgenus Luxilus. This species has in common with Luxilus: 2,4-4,2 teeth, a black peri- toneum, large eye, dorsal fin inserted slightly posterior to the pelyics, and crooked parallel stripes running along the upper side of the back. In addition arioinmus lives under ecological conditions quite similar to those preferred by all forms of Luxilus, particularly N. zonatus and 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 117 N. pilsbrgi, which it resembles superficially so closely that it has been confused with zonatus in the literature (Evermann and Hildebrand, 1916: 444). N. ariommus differs from typical forms of Luxilus in the form of the scales on the side of the body, which are normal in shape instead of high and diamond-shaped. From N. zonatus and N. pilsbryi it also differs in lacking prominent scallops on the posterior edge of the lateral scales and the black stripe along the side of the body. While no other Notropis species shares so many features with Luxilus as ariommus does, Notropis pyrrhomelas of the headwater areas of the southern Piedmont Plain, which Gibbs (1957) placed in the subgenus Cyprinella, shares the following characteristics with N. coccogenis and N. zonistius: Dark pigmentation on the dorsal fin; a dark crescent-shaped blotch of pigment in the fork of the caudal; high diamond-shaped scales on the side of the body; a large eye; a prom- inent dark scapular bar; a peritoneum heavily znecked with black; and 10 anal rays. The last character is found in only a few populations of zonistius. N. pyrrhomelas differs from coccogenis and zonistius in having a 1,4-4,1 dental formula. The similarities suggest fairly close ties between the subgenera Cyprinella and Luxilus. KEY TO SPECIES AND SUBSPECIES OF THE SUBGENUS LUXILUS la. A dark band on dorsal fin (may be orange-red in young). Peri- toneum heavily speckled with dark melanophores, but not uni- formly black. A red stripe usually present in life on cheek or on margin of opercle and preopercle. Nuptial tubercles in two or three rows along lower jaw extending to tip of chin. Dorso- lateral scales heavily margined with dark. Circumferential scales usually 18 to 15-2-10 to 12 2 2a. Black band on posterior half of dorsal fin almost parallel to distal margin. A wide black border in fork area of caudal fin. No caudal spot. Prominent red bar at margin of opercle and preopercle. Body more slender and terete. No nuptial tubercles on top of head, back, or dorsal fin. Angle between mouth and top of head usually 50 to 52°. Vertebrae 40 to 42, usually 41 - Notropis coccogenis Tennessee, Savannah, Catawba and New drainages;. Ala- bama, Tennessee, Georgia, South Carolina, North Carolina, and Virginia. 118 BULLETIN FLORIDA STATE MUSEUM Vol. 8 2b. Dark diagonal band (orange-red in young individuals, black in adults) on dorsal fin, not paralleling distal margin, but beginning on lower third of anterior margin and terminating at middle of posterior margin. No black border in fork area of caudal fin. Caudal spot present, about equal to diameter of eye. Faint red bar sometimes present on cheek. Body deeper and more compressed. Nuptial tubercles present on top of head, back, and dorsal fin. Angle between mouth and top of head usually 57 to 61°. Vertebrae 88 to 40, usually 39. Anal rays usually 10 in specimen from southern part of range. Notropis zonistius Chattahoochee and upper Savannah drainages of Alabama, Georgia, and Florida. lb. No dark band on dorsal fin. Peritoneum uniformly black. No red stripe on cheek or on margin of opercle and preopercle. Nuptial tubercles in single row along lower jaw not extending to tip of chin. Dorso-lateral scales usually not broadly margined with dark pigment. Circumferential scales usually 12 to 17- 2-11 to 15 8 Sa. A prominent black stripe along length of body and encircling snout; a thin but distinct black stripe parallel to and slightly above this stripe on body. Body slender and terete. No black crescent-shaped bars on side of body. Insertion of dorsal fin slightly behind insertion of pelvic. Nuptial tu- bercles weakly developed, granular. Lateral scales with 4 or 5 prominent scallops on posterior margin __ -_-_-_-__.__-_--4 4a. No black pigment on posterior-dorsal margin of opercle or on area bounded by or between opercle and anterior ~ part of lateral line. Pigment covering cleithrum usually very dark in adults. Area between lateral line and ~ lateral stripe on middle part of body unpigmented. Pig- ment unequally distributed on caudal peduncle, very little present below lateral line. Fairly dense pigment around edges of dorso-lateral scales, forming parallel crooked lines that run length of back. Breeding colors usually more pronounced -_-_-_-_-_--_----Notropis zonatus Missouri, Black, St. .Francis, and Mississippi drainages; Missouri and Arkansas. 4b. Black pigment present on posterior-dorsal margin of opercle and on area between opercle and anterior part 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 119 of lateral line. Pigment present only on upper half of cleithrum, usually relatively light in adults. Area be- tween lateral line and lateral stripe on middle part of body pigmented. Pigment equally distributed above and below lateral line on caudal peduncle. Little pig- ment present around edges of dorso-lateral scales, with no parallel crooked lines along back. Breeding colors usually more subdued ---_-_______..--_______Notropis pilsbryi Red, Arkansas and White river systems excluding the Black River; Arkansas, Missouri, eastern Kansas, and eastern Oklahoma. 8b. No prominent black stripe along length of body and head (some darkly pigmented juveniles have an indistinct dark stripe). Body usually deep and compressed (often slender and terete in N. albeolus). Black crescent-shaped bars usu- ally present on side of body (ordinarily absent in N. albeo- lus). Insertion of dorsal fin directly above insertion of pel- vic. Nuptial tubercles well developed, not granular. Lat- eral scales without prominent scallops on posterior margin_-.5 5a. Black, crescent-shaped marks on side of body very prom- inent, numerous, and distinct. Dark upper half of head sharply contrasting with light lower half, particularly in breeding males. Nuptial tubercles present in both sexes though much better developed in males; tips of tu- bercles in breeding males curved forward at tips. Scales on dorso-lateral part of body·with a fine line about a quarter of the way from and paralleling outer edge. Caudal peduncle scales usually 14 or 15 (7-2-5 or 6). Size relatively small, not exceeding 90 mm. -_-_---_.---_.._-_- Notropis cerasinus Roanoke and New river systems in Virginia and North Carolina. 5b. Black, crescent-shaped marks on side of body usually prominent and distinct, though less so than in N. cerasi- nus (marks ordinarily absent in N. albeolus). Dark up- per half of head not contrasting so sharply with lighter lower half. Nuptial tubercles normally present only in males; tubercles not curved forward at tips. Scales on dorso-lateral part of body without line paralleling outer edge. Caudal peduncle scales usually 16 (7-2-7). Size larger, reaching 175 mm. 8 120 BULLETIN FLORIDA STATE MUSEUM Vol. 8 6a. Scales relatively large, the anterior dorso-lateral scales usually 13 to 16, circumferential scales usu- ally 24 to 29. Scale pockets in dorso-lateral area with considerable dark pigment, forming three or four prominent straight or crooked parallel stripes along length of back; these meet posterior to the dorsal fin, forming distinct 'Vs." No broad, light area lateral to mid-dorsal stripe in breeding male. Mid-dorsal stripe not attaining a bright golden hue in live breeding males; narrow longitudinal stripe not present beneath parallel stripes on side of back. Pigment usually present on chin and anterior part of gular area - 7 7a. Parallel dark lines along dorso-lateral part of back crooked, sometimes fairly indistinct. Pre2 dorsal and anterior dorso-lateral scale rows always crooked. Anterior dorso-lateral scales (18) 14 to 16 (19); circumferential scales (23) 26 to 29 (32) _-_.__..Notropis chrysocephalus chr!/socephalus Northeastern Oklahoma east to northern Geor- gia and north to southeastern Wisconsin and western New York. Intergrading with isolepis in central Alabama. 7b. Parallel dark lines along dorso-lateral part of back straight, always distinct. Predorsal and anterior dorso-lateral scale rows straight and even, rarely crooked. Anterior dorso-lateral scales (12) 18 or 14 (15); circumferential scales (28) 24 to 28 - ___.__.--_-_--_.-----Notropis chrysocephalus isolepis Southeastern Oklahoma and northeastern Texas to west-central Georgia and the Gulf coast. In- tergrading with chrysocephalus in central Ala- bama. 6b. Scales relatively small, the anterior dorso-lateral scales usually 17 to 25, circumferential scales usu- ally 26 to 35. Scale pockets in dorso-lateral area with little or no dark pigment; no parallel longitudi- nal stripes running along back. A broad, light stripe Lateral to mid-dorsal stripe, particularly evi- 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 121 dent in breeding male. Mid-dorsal and narrow longitudinal stripes (the latter just below broad, light stripe) attaining a brilliant golden hue in live breeding males, disappearing soon after death (not yet observed in N. albeolus). Pigment either pres- ent or absent on chin and anterior part of gular area 8 Sa. Scales smaller; those in anterior dorso-lateral region (16) 19 to 24 (80), those in circumferen- tial series (26) 80 to 85 (89). Crescent-shaped bars usually present on side of body. General tone of body silvery-bronze. Dorsal region without a deBnite greenish cast in life. Breed- ing males with prominent, broad, dorso-lateral light stripe and much red on fins, less on body. Body usually deep and compressed ______-_______- Notropis cornutus Northern United States and southern Canada from northeastern Colorado to southeastern Saskatchewan and the Atlantic coast as far south as southern, Virginia. 8b. Scales larger; those in anterior-dorso region (15) 17 to 19 (23), those in circumferential series (24) 26 to 80 (88). Crescent-shaped bars usu- ally absent on side of body. General tone of body silvery with a deBnite greenish caste in life. Breeding males with the broad dorso- lateral light stripe not particularly evident; lit- tle or no red on fins, none on body except oc- casionally on tip of snout. Body usually slen- der and terete, more so in specimens from Roa- noke and New river systems; less so in speci- mens from Neuse, Tar and Cape Fear sys- tenns Notropis albeolus Roanoke, New, Neuse, Tar and Cape Fear drainages in North Carolina, Virginia, and West Virginia. 122 BULLETIN FLORIDA STATE MUSEUM Vol. 8 Notropis coccogenis (Cope) Warpaint Shiner, figures 2, 14D, map 1. , =1 U I i -, 11' ··t . ilip / 1. tr * It-n St / Figure 2 . Adult male Notropis coccogenus, UMMZ 181488 , 113 mm SL, from Cosby Creek, Cocke County, Te nnessee. Hypsilevis coccogents Cope, 1867: 160 (original description; type locality Holston River, Virginia). Leuciscits coccogenis, Giinther, 1868: 258 (description; specimens from Vir- ginia in British Museum). Lu:Titus coccogenis, Jordan and Copeland, 1876: 158 (upper Tennessee River). Minnilus coccogenis, Jordan and Gilbert, 1888: 188 (description; range). Notropis coccogenis, Jordan, 1885b: 814 (range). Hydrophlox coccogenis, Jordan , 1929 : 88 (description ; range). Coccotis coccogenis, Jordan , Evermann and Clark, 1980 : 127 (reference to original description; range). Notropis brimle!/1 Bean, 1908: 918-914 (original description; Cane River, North Carolina; based on atypical specimen of N. coccogenis). Hydroplilox brimlevi. Jordan, 1929 : 83-84 (description ; range) Coccotis brim/eul, Jordan, Evermann and Clark, 1980: 127 (reference to orig- inal description). TypE, A series of 100 specimens (ANSP 8561-3660) bearing the data "Holston River, Virginia, collected by E. D. Cope," have been assumed to represent the syntypes of Hupsilepis coccogenis Cope (Fowler, 1910: 285). Jordan and Evermann (1896b: 285) listed as the type a specimen (USNM 36849), also collected by Cope, from the "Holstoii River, North Carolina." As the Holston River closely ap- proaches but does not enter North Carolina, the label on the latter specimen is assumed to be partly in error and should probably read Virginia. Neither the U. S. National Museum nor Academy of Nat- ural Sciences of Philadelphia has any record showing how or when this specimen was transferred, although judging from the entry in the National Museum catalogue Cope himself probably sent it late in 1884. The readily explained difference in locality data, the fact that Jordan and Evermann presumably had a valid reason for labeling the 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 123 specimen as type, and the similarity in quality of preservation of specimens from the ANSP and USNM series strongly indicate that they were collected at the same time. Therefore, I think it logical to assume that the National Museum specimen is from Cope's original type series, and I hereby designate it as a lectotype of Hypsilepis coccogenis. The lectotype specimen is well preserved and is typical of N. coccogenis. Following are the more important meristic counts: anterior dorso-lateral scales 16; circumferential scales 14-2-10; lateral- line scales 89; caudal peduncle scales 7-2-6; pectoral rays 15 on each side; anal rays 9; pharyngeal teeth 2,4-4,2. An examination of the syntypes in the Academy of Natural Sciences revealed they actually comprise five species. In addition to 90 para- types of Hypsilepis coccogenis which have been recatalogued as ANSP 8561-8650, the following are also present: Notropis ariommus (6 speci- mens)-ANSP 8651-8656; Notropis rubricroceus (2 specimens)- ANSP 3657-3658; Notropis chrysocephalus chrusocephalus (1 speci- men)-ANSP 8659; Notropis galacturus (1 specimen)-ANSP 3660. DIAGNOSIS. An attenuate and terete, moderately large species of Luxilus, reaching 116 mm SL (Fowler, 1936: 192). Notropis cocco- genis differs from other members of its subgenus, except N. zonistius, in the following characters: 2 to 8 rows of nuptial tubercles on lower jaw, extending to tip of chin (figure 14D); dorso-lateral scales out- lined with dark pigment; peritoneum speckled with dark pigment, not appearing uniformly black. N. coccogenis is unique among forms of Luxilus in having: outer third of dorsal fin with a moderately oblique band which nearly parallels distal edge of fin; lower two-thirds of dorsal Bn yellowish; a wide black border on posterior edge of caudal fin; a red maxillary spot; a prominent red bar at margin of opercle and preopercle; nuptial tubercles present on side and tip of snout, absent from top of head and back; an extremely well-developed black scapular bar; lower jaw usually projecting slightly. A comparison of N. coccogenis and N. zonistius is presented in table 1. DESCRIPTION. Fin ray and scale row counts and proportional measurements are listed in tables 4 and 9 to 15. Characters men- tioned in the diagnosis are not repeated in the following description: Angle of mouth to top of head usually 50° to 52° (range 46° to 55°); anterior dorso-lateral scales 15 to 18 (19); circumferential scales 25 to 29 (82) (relative distribution of circumferential scales usually 13 to 15-2-10 to 12; range 18-2-10 to 17-2-13); sum of anterior dorso- lateral and circumferential scales (40) 41 to 46 (50); caudal peduncle scales 14 to 16, usually 15; anal rays (8) 9 or 10, usually 9; pectoral 124 BULLETIN FLORIDA STATE MUSEUM Vol. 8 rays 14 to 16 (17), usually 15; vertebrae 40 to 42, usually 41; back dark olive; pockets of some scales along side of body with dark pig- ment, forming a few vertical crescent-shaped bars; breast, belly, and side metallic-silvery, with a rosy hue in breeding males; prominent gray-blue mid-dorsal stripe present; snout and upper lip red in breed- ing males; lower fins white; nuptial tubercles weakly developed, pres- ent in both sexes, but better developed in males; tubercles present on upper surface of pectoral fin and on side and tip of snout, absent from rest of head and back; females metallic-silvery, with only traces of red markings. VARIATION. Hildebrand (1982: 63-64) compared the holotype of Notropis brimlet/i Bean (USNM 50601) with specimens of N. COCCO- genia, and concluded that differences in appearance of the lower jaw in these two forms (projecting in coccogenis, included in brimleyi) warrant specific separation. His conclusions are not substantiated by the present study; examination of many individuals of coccogenis reveals that the supposed differences fall within the normal range of variation of the species. Sexual dimorphism is less pronounced than in some Luxilus forms. Males are usually large, with most individuals over 100 mm long belonging to this sex, although one of the largest examples reported by Outten (1957: 75) was a female. During the breeding season males are more brightly colored than females, particularly as regards the red on the snout and side of the body. Although nuptial tubercles are present in both sexes, they are somewhat better developed in adult males. DISTRIBUTION, The center of distribution of Notropis coccogenis is in the upper half of the Tennessee River system in eastern Ten- nessee, western North Carolina, northern Georgia and western Vir- ginia, with a reliet population present in the Tennessee system of western Alabama. It occurs in the headwaters of the Savannah River in North and South Carolina (Pickens, 1928: 80), of the Catawba River in North Carolina, and (one record) of the New River in north- western North Carolina (USNM 162281). Its presence in the Catawba and New systems may possibly be due to introduction. Its apparent absence from the upper Elk drainage in south-central Tennessee is possibly due .to inadequate collecting. Contrary to numerous accounts (Jordan and Evermann, 1896a; Jordan and Evermann, 1896b), N. coccogenis is not known from the Cumberland River system in Kentucky. Specimens recorded by Wool- man (1892: 288) from the Big Sandy River of Kentucky have not been located, and from the known distribution this record is probably in 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 125 error. A reexamination of material recorded by Fowler (1922: 12; 1928: 11) from the Dan River, Virginia, and from the Neuse River system, North Carolina, shows these specimens are Notropis albeolus. LIFE HISTORY AND ECOLOGY. Most of the following information has been taken from Outten's (1957: 68-84) detailed study of this spe- cies' life history. Notropis coccogenis is an inhabitant of the swift riffies and flow- ing pools of medium to large upland streams. It occurs character- istically in the middle to upper levels of water and prefers a bottom composed of rocks and gravel. Most individuals attain maturity When 2 years old. At this time they usually are about 75 mm SL, with the male averaging slightly larger than the female. Outten observed spawning activity in the French Broad River system of North Carolina from early June to middle July when the water temperature was 68° or higher. This is somewhat later than the period of spawning for N. cornutus and N. chi·ysocephaZus in Michigan. As the temperatures that N. coccogenis requires for breed- ing do not appear appreciably different from those for cornutus and chrysocephalus, the late spawning of coccogenis may be due to the cooler stream temperatures at the high elevations of the Tennessee uplands. Spawning takes place in rapids 6 inches or more in depth or in more slowly flowing water up to 4 feet deep. In the shallow rapids small groups of up to 8 or 10 males were observed to hold territories over Hybopsis micropogon nests 2% feet long by 2 feet wide, with the largest male occupying the upstream position. Occasionally one of the females lingering in a group farther back in the pools moved for- ward near a male. After remaining side by side a short time, they settled to the bottom where, with considerable vibration of their bodies, they spawned. This behavior is similar to that Raney (1940a: 7-8) observed for Notropis comutus in New York. Age-growth studies show the main increase in length to take place in the first and second years of life. During the first year males and females show little size difference, but in the second and third years males grow considerably more than females. Apparently little or no difference in growth rates occurs during the f6urth year. Counts of males and females in each age group show equal num- bers of each sex present during the Brst year of life, but a higher proportion of females among older fish (66.2 percent in the second year of life, 76 percent in the third). This is paralleled by my ob- servations on breeding populations of N. cornutus and N. chryso- 126 BULLETIN FLORIDA STATE MUSEUM Vol. 8 cephalus in Michigan and by those of Pfeiffer (1955: 101) on N. ru- bellus, in which females appear always to dominate markedly. N. coccogenis feeds on terrestrial and aquatic animals, principally insects, with occasional small amounts of vegetable matter, usually taken at the surface or in the upper water levels. Food seems se- lected largely according to its relative abundance, with little or no species-preference evident. 91' 89' 87' 85' 83' 81' 79' 77' v A . s - .4.17 ki >24 2-1-9 -/4 vt.. 4 >4. . 37~ 35' ' . I. 3 f 492 3 . N.coccogenis T---· '. .= O 25 sO .O zOO 0 N.zonistius ~ . -Ill SCALE OF MILES 911 89' 87' 85' 83' 81 79' Map 1. Notropis zonistius (Jordan) Bandfn Shiner, figures 8,148, map 1. / t ts. 3. . 1er *- , Figure 8. Adult male Notropis zonistius, UMMZ 175580, 85 mm SL, from Cane Creek, Dahlonega, Lumpkin County, Georgia. 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 127 Codoma eurustoma Jordan and Brayton, 1878: 42, 84-85 (partim, type series con- tained specimens of N. zonistius and C. eurystmna [= N. uenustus]. Luxilus zonistius Jordan, 1880: 239-240 (original description; type series consists of young specimens forming part of Jordan and Brayton's type series of Codoma eurystoma; holotype USNM 28452, 68 , 5 SL ; type locality Su- wannee Creek, Gwinnett County, Georgia). Minnilus zonistius, Jordan and Gilbert, 1888: 188-189 (description; range). Notropis zonistius, Jordan, 1885b:·814 (list; range). Coccotia zonistius, Jordan, Evermann and Clark, .1980: 128 (reference to orig- inal description; range). Hydrophlox zonistius, Schrenkeisen, 1988: 127 (characters; related to Hydro. phlox coccogenia; notthern Georgia). DIAGNOSIS, A deep, compressed, small species of Luxilus, the males rarely exceeding 85 mm SL. Notropis zonistius differs from other members of its subgenus, except N. coccogenis, in the following charactersi 2 to 8 rows of weakly-developed ,nuptial tubercles on lower jaw, extending to tip of chin (figure 14B); dorso-lateral scales outlined with dark pigment; peritoneum speckled with dark pigment, not ap- pearing uniformly black. N. zonistius is unique among forms of Luxilus in having: dorsal fin with a strongly oblique black band (orange-red in younger specimens) beginning on lower third of an- terior margin and terminating at middle of posterior margin; a distinct caudal spot, about equal to diameter of eye; a faint red bar along cheek. N. zonistius and N. coccogenis are compared in table 1. DESCRIPTION. Fin ray and scale row counts and proportional measurements are listed in tables 4 and 9 to 15. Characters men- tioned in the diagnosis are not repeated in the following description: Body deep and compressed; mouth large, terminal, and oblique, usually forming an angle of 57° to 61° with top of head (range 51° to 61.5°); anterior dorso-lateral scales 15 to 18 (19); circumferential scales (25) 27 to 30 (relative distribution of circumferential scales usually 15 or 16-2-11 or IP; range 13-2-10 to 16-2-13); sum of anterior dorso-lateral and circumferential scales (40) 48 to 48 (49); caudal ped- uncle scales 13 to 15, usually 14; anal rays (8) 9 or 10, usually 9; pectoral rays 14 to 16, usually 15; vertebrae 88 to 40, usually 89; color of body olivaceous to steel-blue above; side of body with consider- able coppery luster; pockets of some scales along side of body with dark pigment, forming a few vertical crescent-shaped bars; prominent gray-blue mid-dorsal stripe; tubercles on upper part of head best developed on side and tip of snout, granular on top of head, back, anterior part -of dorsal fin, and upper surface- of pectoral fin; tubercles present on posterior edges of scales on upper half of body. 128 BULLETIN FLORIDA STATE MUSEUM Vol. 8 The following color description by Reeve M. Bailey is based on live spedimens (UMMZ 157882) collected 28 August 1939 from a tributary of the Chattahoochee River 2.5 miles southwest of Chipley, Harris County, Georgia: side silvery, tinged with pinkish-orange on lower parts in large adults; upper parts dark olivaceous; lower fins white; dorsal fin white, with a band in middle, the band pink in young, reddish-orange in small adults, becoming entirely black in large adults; narrow reddish edge at base of caudal fin in smaller breeding males; tail bright red in large adult, with a whitish base and distal margin; lowermost ray of caudal fin whitish; reddish-orange bar extending upward and downward from base of pectoral, this bar black in breeding males; cheek pale silvery blue. VARIATION. Nuptial tubercles are present in both males and fe- males, but they are better developed in the males. Though no de- tailed comparison of breeding colors has been made between the sexes, one may assume that the male colors are more pronounced, as in other species of Luxilus. Notropis zonistius apparently is the only species of Luxilus with a gradient in anal ray count (table 18). Specimens from the southern part of the range have a modal count of 10, as compared to 9 in specimens from farther north. This situation is comparable to that found in Notropis cummingsae, in which individuals from upland areas usually have 9 anal rays and those from lowland areas 10 or 11 (Hubbs and Raftey, 1951: 18), and in Notemigonus cr!/soleucas, in which specimens from farther south have a greater modal number of anal rays than those farther north (Schultz, 1927). Martin (1949: 65) has shown that "a fast growing group of indi- viduals may have relatively large body parts, such as head and fins." Thus in individuals developing in a warm environment, the anal fin gr6ws more rapidly into a longer fin with more fin rays than in indi- viduals from colder waters. As the variation in anal-ray counts in N. zonistius probably has an environmental rather than a genetic basis, the populations differing in this character have not been ac- corded subspecific status. DISTRIBUTION. Notropis zonistius -holo- type From 1885 to 1896 the specific name megalops was used in place of cornutus. The change was instituted by Jordan on the grounds that Mitchill's (1817: 289) brief diagnokis of Cyprinus cornutus, which was published in July or August, 1817, did not constitute an adequate description. Though Mitchill published a detailed description of the species 6 months later, Rafinesque in the meantime had described the same form as C!/prinus megalops. Jordan (in Jordan and Evermann, 1896b: 281) later reversed himself, and in 1916 the matter was Rnally resolved by the International Commission of Zo6logical Nomenclature, 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 148 which ruled (Jordan, 1916: 28) that a diagnosis, no matter how brief, constitutes a description providing there is ne reasonable doubt as to the identity of the species involved. .DIAGNOSIS. A species of Luxilus distinguished from other mem- bers of the subgenus by the following characters: Anterior dorso- lateral scales relatively small, numbering (16) 18 to 24 (80); circum- ferential scales (26) 80 to 35 (39); sum of anterior dorso-lateral and circumferential scales (42) 48 to 59 (69); mid-dorsal stripe gray-blue, turning to brilliant golden in live breeding males; a light olive stripe, slightly wider than dorsal stripe extending along body on either side of dorsal stripe (figure 15D); below this is another dark stripe, the same width as the dorsal stripe, which also becomes brilliant golden in breeding males; pigment usually absent from chin and gular region (figure 9A), particularly on specimens from west of the Alleghenies. Further distinguished from the closely related N. atbeolus by the following characters: Anterior dorso-lateral and circumferential scales smaller (for counts see preceding paragraph, table 7, and tables 9 through 11); considerable red pigment always present on body and fins of nuptial males, usually absent or subdued in breeding females; color of dorsal region olive-brown; scale pockets along side of body usually with considerable pigment, forming prominent crescent-shaped bars, particularly in breeding males; side of body with a bronzy aspect in life, as contrasted to the more silvery aspect of N. albeolus; body deep and compressed; maximum size apparently somewhat larger. DESCRIPTION. Fin ray and scale row counts and body propor- tions were taken on varying numbers of specimens; these appear in tables 6,7, and 9 through 15. Characters mentioned in the diagnosis are not repeated in the following paragraph: One of the largest forms 06 Notropis, the males sometimes attain- ing a length of 175 mm or more; sides of body silvery-bronze, with a pink wash in breeding males; breast and belly silvery, -with little or no pink in breeding males; pigment wholly or partly absent on ehin and gular region in most populations west of the Appalachian Moun- tains, variable in most populations east of the mountains; top of head blue-brown, becoming deep blue-gray in breeding males; sides of head lighter gray than top; branchiostegals,ordinarily. colorless, but light rosy in breeding males; dorsal fin slightly dusky, often becoming notably darkened in breeding males; other fins normally plain; breed- ing *ales pink or red on distal third or f6urth 6f all Bns. 144 BULLETIN FLORIDA STATE MUSEUM Vol. 8 VARIATION. Geographical variation is noteworthy only in number of anal rays, pigmentation on chin and gular area, and in size of an- terior dorso-lateral scales. No where is variation consistent enough to be deemed worthy of taxonomic recognition. A -Populations in Maine, Nova Scotia, and, presumably, New : Brunswick often have many indi- viduals (usually 40 to 60 percent) with 8 anal rays, a variation found only occasionally in other parts of the range (table 13). An analysis of Other morphological characters from the Maine-New Brunswick- Nova Seotia area reveals surpris- ingly low anterior dorso-lateral and circumferential scale counts (table 7), though, as discussed below, they fall within the overall variation of this form. In no other way do 8 these Bsh differ from typical cor- nutus. The chin and anterior part of the gular area are unpigmented in populations west of the Appala- chian Mountains, but this character shows little apparent consistency in consistently absent and, as dis- populatiens east of the Appala- chians. Only in the upper Sus- quehanna River system is pigment cussed in the zoogeographical sec- Figure 9. A. Notropis cornutus. In- tion, this is of value in determming dividuals living west of the Alle- phylogenetic relationships. There ghenies usually lack pigment on the can be little doubt that this charac-gular areas and most of the chin; those from the Atlantic side often ter is largely under genetic control, have pigment here. B. Notropis a premise substantiated by the fact chrysocephalus chrusocephalus. This that this part of the head usually is pattern remains rather consistent pigmented in specimens of N. throughout most of th6 range of this chrysocephalus living sympatrically form, It is often lacking in speci- mens from the Tennessee River drain- with N. cornutus. In many popu- age. lations over the northern part of the 1964 GILBERT: FISHES OF THE SUBCENUS LUXILUS 145 range specimens of cornutim are heavily pigmented not only on the underside of the head, but over the entire body. This heavy pigmen- tation, often so intense that individuals are almost black, is without doubt a direct result of environmental conditions. Many northern waters are clear but heavily stained or tea-colored, which apparently causes an expansion and multiplication of melanophores over the entire body. The size of the predorsal scales shows the greatest geographic variation of any character encountered in this species. Within a given population scale size is usually more or less uniform. This is not invariable, and in several series scale size has been found to vary considerably. There may even be great variation between popu- lations in adjacent streams. Scale size generally decreases gradually from south to north, a gradient that is particularly noticeable in Atlantic coast populations. Northeastern populations, however, show an abrupt increase in average scale size (a decrease in scale counts), which reverses the trend seen in other coastal populations of cornutus. Individuals on the Atlantic coast have larger anterior dorso-lateral and circumferential scales, on the average, than those living west of the Appalachians. In the upper Susquehanna River system scale size averages somewhat smaller than for other coastal populations, a fur- ther indication of a post-Wisconsin invasion of the Susquehanna by western populations of cornutus. Difference in scale size has been the basis for a subspecific separation of populations from opposite sides of the Appalachians (Hubbs, 1926: 45-46). The desirability of maintaining this separation, however, has been questioned by some authors (Hubbs and Brown, 1929: 89; Greeley, 1988: 67), and their observations are confinned by the present study. Sexual dimorphism is pronounced in this species, as in most other nest-building or territorial cyprinids, such as Semotilus ati·omaculatus, Campostoma anomalum, and Hybopsis biguttata. Some differences are not evident until the breeding season, when males develop a deep red coloration on body and fins, as well as prominent nuptial tubercles on the head, back, anterior part of the dorsal fin, and upper surface of the pectoral fin. Females, in contrast, have a more sub- dued coloration and usually lack tubercles. Other sexual differences are present throughout the year and become pronounced in males only during the breeding season. This is particularly true of the black crescent-shaped markings on the side of the body, which be- come notably darker and more prominent. Also the mid-dorsal stripe and the narrow, light, parallel, longitudinal stripes on the lower part of the back become relatively lighter, contrasting markedly with the 146 BULLETIN FLORIDA STATE MUSEUM Vol. 8 rest of the body, and present a definite golden caste in life (Raney, 194Oa: pls. 1-4). The golden color disappears shortly after death, and the stripes again appear darker than the surrounding areas. Sim- ilar changes have been noted by Lachner (1952 : 488) for Hybopsis biguttata. Other sexual differences are present at all seasons: The adult male has a deeper and more compressed head and body, longer pectoral and pelvic fins (table 6), a greater amount of fatty tissue on the fins, and reaches a larger maximum size (Raney, 1940a: 5). DISTRIBUTION. Occurs over a large part of the northern half of the United States and southern Canada from northeastern Colorado, southeastern Wyoming, eastern parts of North and South Dakota and southeastern Saskatchewan eastward in the southern parts of the Hud- son Bay drainage, the upper Mississippi and Ohi6 valleys, and the Great Lakes-St. Lawrence drainage to the Atlantic coast as far south as south-central Virginia. On the Atlantic slope it ranges from the Gaspid Peninsula and Nova Scotia as far south as the James River system in Virginia; it is common in the St. Lawrence and Great Lakes drainages and occurs in the southern part of the Hudson Bay drainage. The northern- most record is from a tributary of the Red River of the North in Man- itoba (Keleher, 1956: 265); the westernmost records are from Wyoming and Colorado. As indicated by Radforth (1944: 89) the northern limits of its range coincide closely with the 65° F. July isotherm. It occurs in the upper part of the Mississippi River drainage, including parts of the Missouri and Ohio systems, but is not recorded from the Missouri River system in Montana, and is known from this system in North and South Dakota only from the James River and other eastern tributaries. It occurs sparingly in the lower Missouri system of Nebraska, Iowa, and Missouri, with most records in central Mis- souri. In the Platte system it is known mostly from upland tributaries in northeastern Colorado, southeastern Wyoming, and western Ne- braska, with a few scattered records from the lowest parts of this drainage. In the Kansas River system it is abundant in places, the westernmost records being from isolated spring-fed streams. It is absent from the greater part of the Ohio River system, but is found in headwater tributaries in Indiana, Ohio, Pennsylvania and New York, being especially common and widespread in the Allegheny and Shenango drainages of New York and Pennsylvania. Isolated relict populations occur to the south in the lower Kanawha River system of West Virginia (Paint Creek, F.ayette County), and in a tributary of the middle part of the White River in Indiana (Mill Creek, Hendricks and Morgan counties). loi. .95 85" 75' 65 - I I\1 1 1 W~,/ 1 \\\\ © ' 2 sr '-·/ < Ofts. %* .]T ~ 47 .,-2647#- )1 5 . ~ 1 »\ - =s,r.<, F>~Jj *~~K~ ~~,<~~F)~~t11,~~& <4'*L-, A©-c-,-»/0, C i - 42 4 .i.*.©16·941Y>ff_i·i* 24 5,5~ff< :-:.: ~' . *' 44 -./ / 41' -t i...0-0 G ILB E R T : FIS H E S O F TH E S U B G E N U S LUXILUS 147 /37 15. 33' SCALE OF MILES 4 6 1 75, Map 4. 148 BULLETIN FLORIDA STATE MUSEUM Vol. 8 A specimen of N. cornutus in the United States National Museum (USNM 86211) bears the locality "vicinity of Medicine Hat, Alberta". The presence of several eastern species such as Semotilus corporalis and Pimephales notatus in the same collection suggests the locality data may have been transposed. A single specimen supposedly collected in the Black River, Clark National Forest, Missouri (UMMZ 117557) also is regarded as suspect . There are no other records of this species. from the Black River basin, and the next closest populations are in the Missouri River system ap- proximately 150 miles to the north. N. chrysocephalus, which was taken in the same collection, is common throughout the Black drain- age and, from the known relative distribution of these two forms it is rather unlikely that they should occur sympatrically so far south. This record may have resulted either from an introduction or, more likely, from an accidental transfer of specimens in the laboratory. Two records of N. comutus from the Arkansas River drainage in Kansas (UMMZ 122075, Winfield, Cowley County; KU, Rock Creek, near Iola, Allen County) are also thought to be due to transposed data. Other collections from Kansas have been found with faulty labeling, and until these records are substantiated it seems best to treat them as erroneous. The populations in the headwaters of the Platte and Kansas rivers, those from the middle part of the White River, Indiana, and in the lower Kanawha River system, West Virginia, are the most interesting and significant ones in the range. Apparently N. cornutus once was much more widely distributed than at Present, and its disappearance from large areas of the Great Plains can be attributed to the drying this area has undergone since the Wisconsin glacier retreated north- ward. The absence of N. cornutits from a large part of the Ohio River system is due partly to natural conditions and partly to man's inRuence. Trautman (1957: 857) has shown that cornutus is now missing from areas where man's activities have caused the streams to warm and silt and the flowing springs to disappear. Commonly it has been replaced by N. chrysocephalus, which reflects the two species' close ecological relationship and the tolerance of chryso- cephalus for warmer, more turbid conditions. Early collections from the Ohio River system show that chrysocephalus has long been the dominant form and evidently started to replace cornutus before man started altering the landscape. Apparently a warming of the streams occurred, and though cornutus probably could have survived had no other factor been involved, chrusocephalus was favored by 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 149 the warmer conditions and largely dispossessed it hi the southern part of its range. As' these species occupy similar ecological niches, competition apparently extirpated cornutus from most of the Ohio Valley. The presence of isolated populations of cornutus in the middle part of the White River system in south-central Indiana and in the lower Kanawha River system of West Virginia strongly supports this view. Although these populations may have been introduced by man, the evidence strongly favors their presence as natural. Though populations of chrusocephalus live only short distances away (see maps 4.and 5), there is no evidence that it has ever occurred in these pockets occupied by cornutus. I collected in these areas in August and September 1958, some 25 years after the cornutus colonies were discovered, and failed to find chr!/socephalus. Apparently some bar- rier has prevented its moving in here as it has elsewhere. In the Indiana population the barrier is almost certainly eco- logical. Mill Creek is unusual in having a shifting-sand bottom almost-throughout its length, with short stretches of rubble and gravel riffle in a few places where sand has not covered the bottom. N. col'nutus undoubtedly spawns in these limited stretches and has been able to survive by this thin margin. Apparently the stream is fed by a number of small springs, for on 26 August 1958 the water tempera- ture in a shallow, unshaded section of the stream was only 70° F., much lower than might be expected considering the weather at the time. This stream is also one of the few places in southern Indiana where Rhinichthys atratulus can still be found. The habitat in Mill Creek is marginal for col'nutus and has not allowed the population to reach large numbers. Paradoxically it has also permitted the species to survive, for the long stretches of shifting sancl bottom have appar- ently effectively barred the movement of chr!/socephalus into the area. The reasons for chrysocephalus' absence from Paint Creek in the lower Kanawha system are less obvious. The area where cornutus was collected appears suitable for chrysocephalus, which has been collected near the mouth of Paint Creek. Thus some barrier, either physical or ecological, must exist in the lower half of the stream. In its lower reaches Paint Creek is a gravel-bottomed stream consisting almost entirely of shallow riffles with few or no deep pools. Perhaps this habitat is unsuitable for chrusocephalus. LIFE HISTORY AND ECOLOGY. N. cornutus prefers small to medium- sized streams with clear, cool, weedless water, a moderate to swift current, and alternating pools and riffies, the latter with a gravel or 150 BULLETIN FLORIDA STATE MUSEUM Vol. 8 rubble bottom. Though the species is confned to this type of habitat over most of its range, populations in prairie and plains areas, particu- larly in Missouri, Iowa, and Nebraska, tolerate extreme turbidity and seasonally warm water provided a rubble or gravel bottom and a good current afe still present. These factors seem more critical for spawning than clarity of water. Two other Bshes, Rhinichthys cataractae and Hybognathus hankinsoni, which are commonly thought to be limited to clear, cool waters, are also able to survive under these same conditions of turbidity and high temperatures. N. cornuttis is usually found at the head or foot of riffies, most often the latter, but it tends to avoid the actual rifile except at spawn- ing time. It likewise avoids still waters, though individuals may occasionally be found in them, especially if a slight current is present. Although this species is generally a stream fish, it is also common in lakes in the northern part of its range. The apparent geographical bias probably has several causes. Lakes are relatively few in the southern part of its range and ecologically unsuitable. The northern lakes differ in usually having clean, weedless, wave-washed shores with gravel or rubble bottoms, thus in part duplicating the preferred stream conditions. The species is not known to spawn in lakes, and apparently enters them from feeder streams. Strangely neither cor- nutus nor chrysocephalus occurs, except as strays, in western Lake Erie where ecological conditions similar to those of inland streams permit the presence of such usually stream-inhabiting fishes as Noturus #aous, Etheostoma blennioides, and Etheostoma #abellare. Because cornutus is so common and so easily observed, much has been written about its ecology and life history. Numerous food study accounts exist, many of which Adams and Hankinson (1928) reviewed. Largely carnivorous, it apparently prefers aquatic insects and their larvae. The diet is controlled largely by food availability, and plant material is not uncommon fare. This shiner usually feeds at the surface or in the upper water levels, but it may also take food on the bottom. This versatility in feeding habits probably partly explains its great success as a species. Numerous accounts of its spawning behavior have appeared. Raney (1940a) reviewed all prior literature and also presented de- tailed new observations. The species may spawn from mid-May to mid-July, the time depending mainly on the water temperature; a minimum of at least 64° F. apparently must be attained before spawning can occur (Raney 1940a: 4). Although spawning has been observed in water as warm as 88° F., this is unusual, for spawning is normally completed before such temperatures are reached. 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 151 Spawning takes place usually in shallow to moderately deep riffies (8 to 8 inches) with a gravel or rubble bottom. The fish may excavate small depressions in gravel or sand in running water, or utilize the nests built by other species, even when these are in pools. It appears to prefer the nests of other species when available, with those of such nest-building forms as Exoglossum maxillingua, Semotilus atromacu- latus, Semotilus corporalis, Campostoma anomatum, and species of Hybopsis (subgenus Nocomis) most frequently utilized. Males arrive first on the riffles, sometimes in large numbers (Fow- ler, 1909; 540; Raney 1940a: 5). There each individual attempts to hold a small territory. Females meanwhile remain a short distance downstream, usually in the first pool. When ready to breed the females move upstream, take a position over a male, and shortly thereafter the spawning act is consummated. Notropis albeolus Jordan White Shiner, Figures 10, 11, 15C, 15F, map 4 C: ·4 E : ' 't:ld frft.f **r))4~i~79; 1 111¢ffil;11'1.1~37:70. 4 4,(f,I!;fi/i' Figure 10. Adult male Notropis albeolus, CU 20'822, 120 mni SL, from South Fork, Roanoke River, Allegheny Springs. Roanoke County, Virginia. (This speci- men is typical of those in the Neuse, Tar, and Cape Fear drainages.) Hupsilevis corm,tus corm,tus, Cope, 1870: 459 (misidentification; Neuse River, Nortli Carolina). Luxilus cornlitus, Jordan, 1876b: 286-287 (misidentification in part; range). Notropis comutus. Goldsborough and Clark, 1908: 85 (misidentification; Horse- pen Creek, Virginia). Notropis megal(ms, Jordan , 1888 : 58-59 ( misidentification in part , description : range). Notropis megalops albeolus jordan, 1889b : 128, 125, 129 (original description, comparison with N . megalops; type locality Roanoke River , near Roa- noke, Virginia; also in Tar and Neuse river drainages, North Carolina). Notropis albeolus, Evermann and Cox, 1896 : 805 , 809 (synonymy ; Neuse River system, North Carolina). 152 BULLETIN FLORIDA STATE MUSEUM Vol. 8 Luxitus cornutus .aibeolus, Jordan, 1929: 82 (description; range). Luxilus albeolus, Jordan, Evermann and Clark, 1980: 129 (range; reference to original description). Notropis cornutus albeolus, Fowler, 1945: 29 (Roanoke and Neuse rivers, North Carolina). Notropis coccogents, Fowler, 1922 : 12 (misidentification ; tributary of Dan River, Virginia). TypE. Notropis megalops albeolus was described on the basis of specimens David Starr Jordan collected in the Roanoke River at Roanoke, Virginia, 81 July-2 August 1888. Although no types were designated, the United States National Museum has a series of 16 specimens with the following data: "Roanoke River, Roanoke, Vir- ginia, 1888, U. S. Fish Commission, D. S. Jordan;" the catalogue entry is dated 8 April 1889. As these seem to comprise at least part of the syntypic material of N. albeolus, I hereby designate as lectotype of Notropis megalops albeolus an adult, 76 mm SL, which retains the original catalogue number (USNM 40177). Fourteen other specimens of the original syntypic series now become paratypes and are reeata- logued as USNM 177840. The remaining individual is Notropis ardens, and has been recatalogued under that name. DIAGNOSIS. A species of Luxilus distinguished from other mem- bers of the subgenus by the following characters: Red pigment usually lacking on body and fins of nuptial males, always absent in females; dark pigmentation absent from dorsal and caudal fins; scale pockets along side of body with little or no pigment, resulting in few or no crescent-shaped bars; side of body silvery in life; no prominent, par- allel, longitudinal stripes running along length of dorso-lateral region of body (figure 15Ch a light olive stripe, slightly wider than dorsal stripe, extending along length of body on either side of mid-dorsal stripe. N. albeolus is further distinguished from the closely related N. cornutus by the following characters: scales relatively large, those in the anterior dorso-lateral series (15) 17 to 20 (23); those in the circum- ferential series (24) 26 to 80 (88); sum of anterior dorso-lateral and circumferential scales (41) 48 to 50 (55); color of dorsal region olive- green; side of body silvery in life, as contrasted to the more bronzy aspect of N. cornutus; broad light stripes on either side of mid-dorsal stripe less evident in breeding males; body usually slender and terete in northern populations, body form in southern populations not ap- preciably different from N. cornutus, maximum size apparently some- what smaller, adult males attaining a maximum length of about 180 mm. 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 158 A comparison of N. albeolus and N. corntitus is presented in tables 8,6,7, and 9 through 15, and in Bgures 150 through 15F. DESCRIPTION. Fin ray and scale row counts and proportional measurements are listed in tables 6 through 15. Characters men- tioned in the diagnosis are not repeated in the following paragraph: Mid-dorsal stripe grayish-black, slightly more than half as wide as eye, and extending around base of dorsal fin; a light streak about the same width as dorsal streak extending along body on either side of dorsal streak (this character not nearly so clearcut and distinct as in N. comutus); below each of these light streaks is another dark streak of the same width and slightly lighter color than the dorsal streak (this and the preceding character evident only in breeding males); pigment on chin and anterior part of gular region either present or absent; pinks and reds usually absent on body, rarely present on snout of breeding males; pink or red coloration usually absent from all fins at all ages, occasionally present on dorsal, caudal, anal, and pelvic Bns of breeding males; dorsal fin slightly dusky, but never greatly darkened in breeding males; other fins plain; nuptial tubercles present over head, the tips pointing straight upward, possibly not as well developed as in N. cornutus and N. chrysocephalus (relatively few tuberculate specimens of albeolus have been examined). VARIATION. Notropis albeolus exhibits the most interesting mor- phological variation found in any form of Luxilus. Differences in body proportions in particular contribute to a rather perplexing prob- lem involving its correct taxonomic status. In general specimens from the New and Roanoke drainages seem similar in appearance, as do those from the Neuse, Tar, and Cape Fear systems. This similarity is particularly noticeable in body shape; specimens from the New and Roanoke tend to be more sIender and terete, those from the other drainages are usually deeper and more compressed. Propor- tional measurements for individuals from the New and Roan()ke aver- lu~ 16..,1 ~~j: i L .A ./ Figure 11. Adult Notropis albeolus, USNM 1018,89, 76.5 min SL, from Waqua Creek at highway 140, Brunswick County, Virginia. 154 BULLETIN FLORIDA STATE MUSEUM Vol. 8 age less for a number of characters, particularly depth of body, depth of head, and distance from dorsum to lateral line (table 6). Also depth of caudal peduncle, length of head, mouth gape, width of body, width of head, width of eye, length of upper jaw, and all fin lengths average proportionately slightly less; the caudal peduncle, on the other hand, averages slightly longer. Although these differences are usually rather marked, atypical specimens can be found in both the northern and southern populations (figure 10). Morphological differences of the above magnitude · exist among individuals of N. cornutus. In that form these differences can always be explained by environment, slender-bodied fish usually being found where a shortage of food exists. As food is apt to be scarcer in cool, relatively sterile northern streams, slender-bodied individuals of cor- nutus tend to occur to the north. The morphological differences in albeolus cannot be explained by environment. While streams of the Roanoke and New River systems are in general clearer, swifter, and presumably somewhat more sterile than those of the Neuse, Tar, and Cape Fear drainages, this can hardly explain such sharp differences in body proportions. In addition the closely related Notropis cerasinus, a common species in the headwaters of the Roanoke, is characterized by a body form fully as deep and compressed as either that of N. cornutus or of the southern, population of N. albeolus. This strongly suggests that the differences in body form in albeolus have a genetic rather than an environmental basis. Consequently N. albeolus may best be regarded as containing two distinct recess, one inhabiting the New and Roanoke drainages, the other the Neuse, Tar, and Cape Fear systems. The slight difference in scale counts between the two races is sub- ject to considerable overlap. Table 7 shows specimens from the New and Roanoke drainages average about one scale more in the dorso-lateral and circumferential scale series than do those from the other systems. The upper limit (55) for this count in specimens from the more northern drainages is 4 more than for specimens from the Neuse, Tar, and Cape Fear; the lower limit (42) is 1 more. The crescent-shaped markings that Occasionally appear on the side of the body are more common in specimens from the southern drainages. Other kinds of variation apparently occur equally in both races. The lack of red pigment on both the body and fins, including the breeding males, is one of this fish's most distinctive characters. Though this feature holds true most of the time, Robert D. Ross in- forms me in a letter that occasionally breeding males from both the 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 155 New and Roanoke show some red on the fins or body. Large adult males taken in the Tar River 20 August 1946 by Reeve M. and Joseph R. Bailey (UMMZ 147588) also showed some red on the caudal fin. Red pigment when present occurs only on the tips of the fins, not including the pectorals. I have never seen specimens of albeolus so colored, so I cannot assess possible differences from cornutus in. distri- bution of red on the fins. Apparently the only part of the body on which red pigment appears is the snout. Some individuals, particularly those from the Neuse, Tar, and Cape Fear systems, have a few vertical crescent-shaped bars along the side similar to those in cornutus. cerasinus. and chrysocephalus. Such specimens are extremely difficult to distinguish in preservative from sparsely pigmented individuals of cornutus. Analysis of vertebral variation in albeolus (table 8) shows that specimens from the Roanoke and New river systems usually have 40 vertebrae, individuals from the Neuse generally 89. Specimens from the Tar drainage have either 89 or 40 vertebrae with equal frequency. The Cape Fear population could not be analyzed because of lack of material. At Brst glance the differences in vertebral counts appear to be directly correlated with variation in body depth, though the presence of either 89 or 40 vertebrae in Tar River specimens indi- cates otherwise. Possibly these counts reflect ecological differences between the several streams. RELATIONSHIPS. The forms albeolus and cornutus are so close to the species-subspecies line in their degree of evolution that it is diffi- cult to decide their proper taxonomic relationship. In the past al- beolus has been treated both as a subspecies and, in recent years, as a species, which Moore (1957: 126) has questioned. I myself have had doubts about the correct taxonomic status of albeolus. My first inclination (196la: 2411) was to regard it as a sub- species of N. cornutus, primarily because not all preserved specimens of the two forms can be distinguished. This situation is by no means unique, and may be noted in other North American Cyprinidae, for example in species of the genus Hybopsis (subgenus Nocomis). Mem- bers of this group are very similar in over-all appearance, and several forms are positively distinguishable only by means of the number and arrangement of the n(lptial tubercles on the heads of breeding males. Thus though only a small percentage of individuals of these species can be told apart, the constant differences in the character of the tubercles leaves little doubt that their relationships are at the spe- cific level. Many other pertinent examples exist in fishes, as well as in other groups of animals. 156 BULLETIN FLORIDA STATE MUSEUM Vol. 8 The fact that some individuals of cornutus and albeolus cannot be distinguished does not disprove their specific distinction. Differ- ences in color and in other features of body pigmentation, in size of the anterior dorso-lateral scales, and in body shape all argue for specific recognition. DISTRIBU'nON. Restricted to the Roanoke River system (including the Chowan) in Virginia and North Carolina, parts of the New River system in Virginia and West Virginia, and to the upper parts of the Neuse, Tar, and Cape Fear systems in North Carolina. The apparent scarcity of albeolus from the more lowland tribu- taries of the Neuse, Tar, and Cape Fear drainages probably is due mainly to lack of collecting. Most of the specimens examined from these systems were taken during the spring when the streams are usually high, particularly those in the area below the Fall line. Re- cent autumn collections from the lowland areas indicate that this species is, in fact, fairly common there. In recent years N. albeolus apparently has come to occupy a much more extensive area of the New River system than formerly. Jordan (1889b: 140-141) did not record it in his 1888 collections from Reed Creek, near Wytheville, Virginia, yet the species is plentiful there today. Carl L. Hubbs failed to take it in Wolf Creek, Bland County, Virginia in 1986, where it was collected both in 1958 and 1956 (UMMZ 169176 and UMMZ 174488). A series of 25 specimens (USNM 177876), Frank J. Schwartz collected in the Greenbrier River near Caldwell, Greenbrier County, West Virginia 18 October 1956 also represents a record for an area where this form was not previously recorded, despite rather intensive collecting both by Hubbs and John Addair during the 1930's. Although these range extensions may have resulted from human introduction, the fish could equally plausibly have found their way without help. LIFE HISTORY AND ECOLOGY. No detailed work on the ecology or life history of N. albeolus has been published. The only data that have appeared consist of general habitat descriptions or collection sites. This species seems to prefer the same habitat as N. cornutus, cool, clear, medium-sized streams with alternating riffies, Rowing pools, and a rubble to rocky bottom. Although it is usually collected at·tbe foot of riffles in midwater, it may on occasion be found in quite dissimilar situations. Some 28 specimens (UMMZ 174749) I collected in the Otter River, Campbell County, Virginia on 12 June 1956 were living in very warm, turbid, swift water on a bottom com- 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 157 posed entirely of shifting sand. Interestingly I also found here the closely related Notropis cerasinus, which is ordinarily restricted to headwaters of the Roanoke and New systems. Possibly both species had temporarily moved or been waihed out of a nearby tribut~ry where conditions were more to their liking. Populations of albeolus in the Neuse, Tar, and Cape Fear drain- ages are found more frequently in turbid situati6ns than are those in the New' and Roanoke systems. This probably reflects ecological differences rather than racial dissimilarities in physiological toler- ances, which are nevertheless possible, especially in view of the rather marked morphological differences between the two races. N. albeolus probably breeds in late April or early May, an estimate based largely on circumstantial evidence. All spring collections I examined were taken either in early April, late May, or early June, and none contained gravid females or high tuberculate males. Notropis chrysocephalus Striped Shiner. DIAGNOSIS. A species of Luxilus distinguished from the closely related Notropis cornutus by: larger scales, those in the anterior dorso- lateral series (12) 13 to 16 (19); circumferential scales (28) 24 to 29 (32); sum of anterior dorso-lateral and circumferential scales (86) 87 to 45 (48); dorso-lateral scale pockets darkest around edges, which, with the intensification of pigment caused by the overlapping scales, results in three parallel stripes running lengthwise along dorso-lateral part of back and meeting posterior to dorsal fin, thus forming distinct rs" when seen from above; pigment usually present on chin and gular region; mid-dorsal stripe not turning brilliant g6lden in breed- ing males; dorso-lateral part of back without a light olive stripe ex- tending along back just lateral to dorsal stripe. DESCRIFI'ION. A large species of Luxilus, the males sometimes attaining a length of 175 mm or more; body and head deep and com- pressed, especially in adult males; pectoral fIt rays 15 to 17 (18); lateral-line scales (86) 87 to 40 (42); dorsal region brown; scale pockets along side of body with considerable pigment on some scales forming a number of vertical crescent-shaped bars; side of body metallic- bronze in life; mid-dorsal stripe gray-black, slightly more than half as wide as eye, and extending around base of dorsal fin. 158 BULLETIN FLORIDA STATE MUSEUM Vol. 8 Notropis chrysocephalus chrysocephalus (Rafinesque) Northern Striped Shiner, figures 98, 12,158, map 5. - #.. I % + PY A i *. ... Figure 12. Adult male Notropis chrusocephalus chrusocephalus, UMMZ 154507, 140 mm SL, from Clear Creek, Wildie, Rockcastle County, Kentucky. Luxilus chrusocephalus Rafinesque, 1820a: 48 (original description; type locality Kentucky). Notropis comutus chrusocephalus, Hubbs, 1926: 46 (recognized as a distinct form; intergradation with N . c. frontalis; name frontalls applied to chru- soceph,mis by Jordan and Evermann; nominal forms plumbeolus and gibbus of Cope are synonyms; Great Lakes drainage). Luxilus comutus chi·ysocephalus, Jordan, 1929: 82 (description; range). Notropis chrusocephalus chrusocephalus, Gilbert, 196la: 2411-2412 (general account). Notropis chr!/socephalus, Larimore and Smith, 1968: 828, 828, 832-888, 885, 388-889,848, 845, 858, fig. 81 (Champaign County, Illinois). ?Rutilus plargurus Rafinesque, 1820;2 50-51 (original description; type locality Kentucky), Leitciscus plargynts, Kirtland, 1845 : 26-27, pl . 8, fig . 2 ( in part ; description ; range; Ohio). Luxilus plargurus, Jordan, 1876a: 94 (review of Rafinesque's species). ?Leuciscus gibbosus, Storer, 1846: 166 (in part; description; Huntsville, Alabama). ?Hypsolepis gibbosus, Agassiz , 1854 : 859 (Storer's Leuciscus gibbosus placed in genus Hypsolepis). ?Plargulus gibbosus, Girard , 1856 : 196 (in part ; list of related species ; Ala- bama). Plargyrus tvpicus Girard, 1856: 196 (in part; name replaces Rafinesque's Rutilus plargyrus). Plargyrus cornutus, Girard, 1856: 196 (misidentiEcation in part; list of species). HI/psolepis cornutus, Putnam, 1868: 7 (misidentification in part; in Harvard collection). Leuciscus comutus, Gunther, 1868: 249 (misidentifcation in part; specimens in British Museum from Michigan, Lake Erie, and Miami River). Hypsilepis cornuttis, Cope, 1869b: 872-873 (misidentification in part; descrip- tion; variation; range; Pennsylvania). Luxilus cornutus, Jordan, 1876a: 94 (misidentification; review of Ratines(lue's 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 159 species; Luxitus chrusocephalus Rafnesque synonymized with Cyprinus cornutus Mitchill). Minnmis cornutus, Jordan and Swain, 1888: 249 (misidentifcation; Clear Fdrk of Climberland River, Kentucky). Notropis cornutus, Gilbert, 1884b: 202 (misidentification; restoration of gehus name Notropis; East Fork of White River, Indiana). ?Cyprinus cornutus, Call, 1896: 15 (misidentification; Falls of the Ohio, Louis- ville, Ky.; possibly refers to another species since Notropis megalops also listed). Notropis comutus cornutus, Knapp, 1958: 61 (misidentification; replaces N. cornutus isolepis immediately to north). Notropis cornuta, Hubbs, 1955: 10 (misidentification in part; hybridization with N. rubetta). Athrnops plumbeolus Cope, 1864: 282 (original description; type locality Saginaw River, Flint, Michigan). Hybopsis plwnbeolus, Jordan, 1876b: 288 (description; range). Minnilus plumbeolus, Jordan and Gilbert, 1888: 192 (description; range). Luxilus plumbeolus, Jordan, 1884: 294 (description; range; synonymy). Hypsilepis frontalis, Cope, 1864: 279 (misidentification, description; southeastern Michigan). Hypsilepis cornutus frontalis, Cope, 1867: 159 (misidentification; description; Holston River, Virginia). Minnilus cornutus frontalis, Jordan and Gilbert, 1883: 187 (misidentification; description; range). Notropis megatops frontalis, Jordan, 1885b: 814 (misidentification; in list; synonymy). Notropis cornutus frontalis, Jordan and Evermann, 1896a: 256 (misidentifica- tion; range; reference to original description). Hypsilepis cornutus gibbus Cope, 1867: 158 (original description; type locality southeastern Michigan). Minnitus cornutus gibbus, Jordan and Gilbert, 1888: 187 (description; range). Hybopsis lacertosus Cope, 1869a: 280-282 (original description; type locality Bear Creek, tributary to Middle Fork of Holston River, Virginia). Alburnops lacertosus, Jordan, 1878: 420 (range). Hydrophlor tacertosus, Jordan and Brayton, 1878: 64, 84-85 (Holston River, Virginia). Minnilus lacertosus, Jordan and Gilbert, 1883: 192 (description; headwaters of Tennessee River). Luxilus Zatertesus, Jordan, 1884: 294 (description; range; synonymy). Notropts lacertosus, Jordan, 1885b: 814 (range). Luxilus cornutus lacertosus, Schrenkeisen, 1988: 129 (headwaters of Tennes- see River). Notropis.megalops, Jordan, 1885a: 128 (misidentifcation in part; Plargyrus tt/Pi- cus a synonym). Notropis cornutus: chrusocephalus X frontalis, Hubbs and Brown, 1929: 86-89 (in part; discussion Of intergradation; hybridization; records for Ontario). TypE. Raftnesque designated no type specimen for this species. As N. chrysocephalus has often been confused with N. cornutus, par- 160 BULLETIN FLORIDA STATE MUSEUM Vol. 8 ticularly in areas where the two species occur sympatrically, I feel the selection of a neotype is desirable. I therefore designate as neo- type of Luxilus chr!/socephalus Rafinesque (1820) an adult male (UMMZ 174589) 99 mm SL, collected in a creek about 6 miles SSW of Danville, off highway 85, Lincoln County, Kentucky 5 April 1958 by R. M. Bailey and party. Six other topotypes (UMMZ 165232) were collected with the neotype. The more important meristic counts for the neotype are: Anterior dorso-lateral scales 14, circumferential scales 28, lateral-line scales 88, caudal-peduncle scales 16, pectoral rays 17-17, anal rays 9. NOMENCLATURE. The 'large-scaled form" of N. cornutus has long been recognized by ichthyologists. Many chose to regard it merely as a variant of Cornutus, although some separated it subspecifically. Consequently the identification of N. chrysocephalus in the literature often must be resolved by geography. Cope (1864: 279) believed Agassiz's description of Hypsilepis frontalis was based on this spe- cies, and as a result the name frontaBs was incorrectly applied for more than 50 years. Not until 1926 did Hubbs point out that frontalis represents the fine-scaled northern form, and that the name chryso- cephalus, which Jordan (1876a: 94) had placed ih the synonymy of Cyprinus cornutus Mitchill, applied to the coarse-scaled southern fornn. Cope's (1867: 158) description of Hypsilepis cornutus gibbus pos- sibly was based on the hybrid Notropis cornutus X Notropis chryso- cephalus. Although Cope recorded only 16 predorsal scales for this form, individuals of the cross sometimes have almost as large scales as does chrysocephalus. Unfortunately the types of gibbus appar- ently no longer exist so this question is impossible to resolve. An examination of the seven syntypes of Alburnops plumbeolus Cope (ANSP 2055-2061) shows that six specimens are N. chrpoceph- alus, the other Notropis heterodon. To Bx the status of the specific name plumbeolus and to prevent possible confusion with Notropis heterodon I designate as lectotype of Alburnops plumbeolus a speci- men 40.5 mm SL, catalogued as ANSP 2055. Five specimens become paratypes (ANSP 2056-2060), while the last is recatalogued as Notropis heterodon (ANSP 2061). Cope (1869a: 280-282) described Hybopsis lacertosus on the basis of Bve specimens from Bear Creek, a tributary of the Middle Fork of the Holston River, Virginia. Nothing in the original description indi- cates how this form differs from Notropis c. chrysocephalus or No- tropis coccogenis, the species of Luxilua occurring in this area, al- though the name -lacertosus signifies a slender body. Unfortunately, 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 161 the types (ANSP 2885-2889) are in extremely poor condition, the bodies having become too decomposed as to be of aid in recognition. The heads, including the pharyngeal teeth are still intact, as are most of the fins, The head is deep in all specimens, in no case does the lower jaw project notably beyond the upper, and there is no sign of a black band on any dorsal fin. As all the above characters are indica- tive of Notropis chrysocephalus, I regard Hybopsis lacertoms Cope as a synonym of that species. DIAGNOSIS. A subspecies of N. chrysocephalus distinguished from N. c. isolepis by: Slightly smaller scales, those in anterior dorso-lateral series (18) 14 to 16 (19); circumferential scales (28) 26 to 29 (82); sum of anterior dorso-lateral and circumferential scales (38) 40 to 45 (48); predorsal and anterior dorso-lateral scale rows more crooked and un- even; anterior dorso-lateral stripes crooked, more poorly defined, and not unif6rmly parallel (figure 15B); pigment usually present on chin and gular region (figure 9B often absent in specimens from Tennessee drainage). DESCRIPTION. Fin ray and scale row counts and body proportions were taken on varying numbers of specimens. These appear in tables 6 and 9 through 15. Characters mentioned in the description of Notropis chrysocephalus and in the diagnosis are not repeated in the following paragraph: Arrangement of circumferential scales above and below lateral line (11-2-10) 12-2-12 to 14-2-18 (15-2-15); dorso-lateral scale pockets darkest around edges which, with the intensification of pigment caused by the overlapping scales, results in three crooked, fairly well-defined parallel stripes running lengthwise along dorso-lateral part of back; side of body silvery-brodze with a deep rose-red in breed- ing males; breast and belly silvery with little or no pink in breeding males; branchiostegals colorless, with a decided rosy wash under- lain with gray in breeding males; distal third or fourth of dorsal, cau- dal, anal, and pelvic fins pink in breeding females, deep pink to red in bfeeding males, except for extreme distal edges which are white; outer fifth of pectoral fin rosy, distal edge white in nuptial males; basal two-thirds of dorsal fin normally pallid, with a pinkish blush in nuptial males; basal two-thirds to three-fourths of anal, pelvic, pec- toral, and caudal fins pallid. VARIATION. Relatively little morphological variation occurs in this subspecies. Populations containing individuals with slightly smaller scales are found throughout the range, but the resultant high counts fall well within the normal range of variation for the species and pose no nomenclatural problem. A series of 85 specimens from the Coosa 162 BULLETIN FLORIDA STATE MUSEUM Vol. 8 rn ..' '9 r~1 EE - sa 'r,W IO ·lv ls -h~'. ~.-'4- s,di M --f„ TZI'*:3.4-0-< -, us 94 = . . Z O ..1 r.. r \ 55:, A ~·Y k 0.0 0 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 168 River system, Alabama (UMMZ 175746) which showed the highe5t counts for any individuals of this form exhibited a number of abnor- malities such as deformed fins and "bulldog" snout. These anomalies were also found in other species from the same locality, suggesting that some factor such as the chemical composition of the water was responsible, probably during early embryological development. Occasionally the ordinary dark pigment on the anterior dorso- lateral scale pockets is either faint or virtually absent. In such cases chrusocephalus approaches cornutus in appearance and looks like a hybrid between these two species. The resemblance is merely super- ficial, for such individuals are found in areas where it seems unlikely that cornutus bas ever occurred. As these are typical chr!/socephalus in all other respects, the suppression of pigment along the upper side of the body apparently is of no special significance. Individuals occasionally lack pigment on the chin and gular re- gion. This is particularly common in the Tennessee River system, where the pigment is wholly or partially absent in approximately half the individuals. DISTRIBUPON. Found throughout a large part of east-central United States from the upper Coosa River system in Alabama and Georgia and the lower Mississippi River drainage in Alabama, north- eastern Mississippi, Tennessee, northern Arkansas, and northeastern Oklahoma, nofthward to the lower Great Lakes region from south- eastern Wisconsin to western New York. Common in the upper Coosa River system of Georgia and Ala- bama, but becomes increasingly less common in the lower parts of that drainage. It is abruptly replaced by N. c. isolepis a short dis- tance north of the confluence of the Coosa and Tallapoosa rivers, with no apparent evidence of intergradation. It also occurs in the eastern- most tributaries of the Cahaba and Black Warrior rivers and is re- placed by isolepis in the lower parts of these drainages also. Collec- tions are at present not extensive enough to permit an accurate pie- ture of the relationships of the two forms, although intergradation is proved by a series of 16 specimens (UMMZ 168613) from the head- waters of the Black Warrior system. Found in the headwaters of the Illinois and Neosho rivers (trib- utaries of the Arkansas) in northeastern Oklahoma, northwestern Ar- kansas, and southwestern Missouri. It has not yet been recorded from Kansas, but should occur in the extreme southeast corner of that state. To the east chrysocephalus is common throughout most of the White and Black river systems of northern Arkansas and southern Missouri, but occurs sparingly in the St. Francis River system and other small 164 BULLETIN FLORIDA STATE MUSEUM Vol. 8 tributaries of the Mississippi River in eastern Arkansas and Missouri, Apparently absent from western Tennessee west of the Tennessee River system, although it is quite common throughout most of the Tennessee drainage itself in Tennessee, extreme northeastern Missis- sippi, northern Alabama, northwestern Georgia and southwestern Vir- ginia. It has n6t yet been reported from North Carolina, To the north it is one of the most abundant cyprinids in the Ohio River system, except for the upper Kanawha (New) River drainage, where. it is -practically absent-the single record for this drainage (USNM 177878)' is believed an introduction. Common ill tributaries of the Gasconade and Meramec rivers in Missouri, it is -rate in the Osage drainage of that state. It is found in some of the smaller tributaries of the Mississippi River in eastern Missouri both north and south of the Missouri River, and occurs in a few scattered localities in northeastern Missouri. There are as yet no records for Iowa. It is common in the Illinois River system in Illinois and in suitable habitats of the more eastern and southern parts of the state, but it has not yet been found west of the Illinois drainage. It occurs throughout most of the lower Great Lakes region and is entirely absent only from the Lake Superior drainage. Its appar- ent absence from the more northern tributaries of the other Great Lakes seems to be related primarily to temperature, as its northern dis- tributional limit coincides closely with the 70°F July isotherm (Rad- forth, 1944: 89). In the Lake Michigan basin it is known from ex- treme southeastern Wisconsin, northeastern Illinois, northern Indiana, and from western Michigan. It occurs sparingly in the Lake Huron basin, having been recorded only from certain tributaries of Saginaw Bay in Michigan and from a few localities in Ontario. It is wide- spread and common throughout most of the Lake Erie drainage. In the Lake Ontario basin its occurrence is spotty, with all known records coming from streams flowing into the south side of the lake. The lack of records from any streams draining into the Atlantic Ocean indicates that this species probably was a late entrant into the Great Lakes drainage and failed to reach the Horseheads Outlet into the upper Susquehanna River system (Bailey, 1945: 125-126) before this connection was eliminated. In recent times N. chrysocephalus seems to have replaced the closely related N. cornutus throughout much of the latter's former range.. This is substantiated by the presence of isolated populations of cornutus in West Virginia and Indiana, and by Trautman's (1957) observations over the past 85 years on the relative distribution of 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 165 these two species in Ohio. A more complete discussion of the rela- tionships of the two forms appears in the account of cornutus and in the section on zo6geography and phylogeny. LIFE HISTORY AND ECOLOGY. N. c. chrysocephalus prefers an en- vironment similar to that of N. cornutus. Both are ordinarily found in small to medium-sized streams having clear, weedless water, a moderate to swift current, and alternate pools and riffies, the latter with a gravel and/or rubble bottom. Individuals usually are found at the head or foot of riffies, more often the latter, but tend to avoid the riflies themselves exdept at spawning time. They likewise avoid quiet water, but individuals occasionally may be found there espe- cially if a slight current is present. As a rule chrysocephalus seems somewhat more tolerant than N. cornutus of warm, turbid conditions and has replaced it in a number of places where the environment has been so modified. The apparent preference of chrysocephalus for warmer waters is shown by the dis- tribution of the two species when both occupy the same stream; the main body of the col'nutits population is more apt to occur toward the headwaters, chrusocephalus usually is more common in the lower parts (Trautman, 1939: 285). When an obstruction was built across one such stream (Gilbert, 196lb: 188) N. cornutus took over completely, apparently because that segment of the population of chrysocephalus occurring toward the headwaters could not compete successfully un- less continually reinforced from the main population downstream. N. chrysocephalus is rarely found in lakes, partly because they are scarce throughout most of the species' range. Only to the north is chrysocephalus found with any consistency in standing waters, and here clear, gravel or rubble-bottomed, wave-washed shores are usually present. As in N. cornutus, spawning probably does not occur in the lakes themselves, but in tributary streams. References to food habit studies of N. cornutus listed by Adams and Hankinson (1928) refer in part to N. chrusocephalus. Further ac- counts of the food habits Of chrysocephalus are discussed by Langlois (1954). These studies reveal no apparent differences in food habits between cornutus and chrusocephalus. Raney (1940a) described in detail the spawning behavior of N. cornutus and assembled the literature on breeding both in this form and N. chrysocephalus. This shows no essential differences in spawn- ing behavior between the species, a finding supported by my own observations and those of others (Hankinson, 1932; Lachner, 1952). Whatever differences exist in spawning requirements most likely in- volve 'optimum temperatures and/or ranges of spawning temperature. 166 BULLETIN FLORIDA STATE MUSEUM Vol. 8 The northern limits of distribution of the two species seem to indicate that such differences do exist. Both species hybridize with other species with roughly equal frequency. Notropis chrysocephalus: chrysocephalus X isolepis Notropis megalops, Gilbert, 1891: 157 (not of Rafinesque; in part; collections from area of intergradation in Black Warrior River system, Alabama). The only collection I have seen that contains positively identified intergrades (UMMZ 168618) is from Blackburn Fork (9 miles south- west of Oneonta, Alabama), a headwater tributary of the Black War- rior River. Further collecting will probably reveal intergrading pop- ulations in a number of localities, most likely in other headwater trib- utaries of the Black Warrior and Cahaba systems. Other areas where intergrades might occur are the lower Coosa drainage where the two subspecies appear to be allopatric, the headwaters of the Tallapoosa River, the more westerly headwater gulf tributaries bordering the Tennessee River drainage, and the Arkansas River system in Arkansas. The chief basis for identification of UMMZ 168618 as an inter- grading population is the presence of typical individuals of both sub- species together with intermediates. Several specimens have pro- nounced, straight, dorso-lateral stripes running along the side of the back, while in others these stripes are crooked and poorly defined. The predorsal and anterior dorso-lateral scale rows in the former (isolepis type) individuals are regular and even, while these rows are irregular and uneven in the latter (chrysocephalus type) specimens. The subspecies chrysocephalus and isolepis are readily distin- guished not only on the basis of the criteria discussed above, but also by notable differences in both the anterior dorso-lateral and circum- ferential scale counts (tables 9 through 11). These differences by themselves are not sumciently great to permit identification of small series of intergrades, such as the above. The apparent scarcity of intergrading populations of chr!/socepha- lus and isolepis, together with the rather unusual distribution of these forms in the Coosa system, suggest that these forms are close to the specific level of separation. I maintain them as subspecies because, in addition to the intergrading population previously cited, specimens of isolepis living well eut of the area of potential intergradation occa- sionally have rather uneven anterior dorso-lateral stripes, causing the individual to appear more or less intermediate. Also past collections from near the known area of intergradation usually contain too few specimens or the individuals are too small to permit accurate identi- 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 167 fication. Hence some series of specimens identified to subspecies may actually consist of intergrades. Notropis chr!/socephalus isolepis Hubbs and Brown Southern Striped Shiner, figures 18, 14E, 15A, map 5 . 4,< p, 9, -r t. f. ii wirl; 541'T# j ,W v j B 0 .--ee f A* S-F®Zoe\= 1 . 0 (Be# L Wed t52\ ® Weee- ~\JAY\-t>- Figure 14. Tubercle patterns in Luxilus: A. Notropis zonatus, UMMZ 116878; B . N. zonistius, UMMZ 157882 ; C . N . cerasinus, UMMZ 174710 , D . N . COCCO- genis, CU 18558; E. N. thrysocephalus isolepis, UMMZ 161262; F. N. cornutus, UMMZ 95626. 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 169 and uniformly parallel; pigment always present on chin and gular region, often with a streak 6f black running posteriorly down gular membrane. A comparison of N. c. isolepis and N. c. chrysocephalus is presented in table 8. DESCRIPTION. Body proportions and counts of fin rays and scale rows are listed in tables 6 and 9 through 15. Characters mentioned in the description of Notropis chrysocephalus and in the above diagnosis are not_repeated in the following paragraph. Arrangement of circumferential scales above and below lateral line (10-2-11 or 11-2-10) 11-2-11 to 18-2-18 (15-2-18); dorso-lateral scale pockets darkest around edges which, with the intensification of pig- ment caused by the overlapping scales. results in three straight, well- defined, parallel stripes running lengthwise along dorso-lateral part of back, the stripes converging posteriorly with those from opposite side to give the appearance of long, narrow, parallel ..VS" when viewed from above (figure 15A); pigment always present on chin and gular area. The following color description is based on an adult male (UMMZ 161262) collected on 6 June 1949 from a tributary of the Ouachita River 0.3 mi. southeast of Vernon, Jackson Parish, Louisiana: snout, branchiostegals, lower surface of head, and belly rosy, shading to scarlet along sides of body; dorsal and caudal fins of breeding males washed with yellowish, outer margin clear, bordered inside by a nar- row band of crimson; distal half of pectoral, pelvic, and anaI fins brilliant scarlet, except for the edges which are clear; proximal half of anal Bn cream, distal half a brilliant scarlet except for the narrow border which is colorless; proximal half of pectoral and pelvic fins slaty gray, distal half scarlet, with the border colorless. VARIAT[ON. This form shows relatively little geographic variation. Occasionally an unevenness is evident in the arrangement of the pre- dorsal and anterior dorso-lateral scales, but this is slight and never approaches the usual situation in N. c. chrusocephalus. Anal ray counts vary to some extent, the extremes in this subspecies being the greatest encountered in any form of Luxilus. A specimen with 7 anal rays (UMMZ 128116) was found from the Red River drainage of Arkansas, and one with 12 anal rays (UMMZ 161396) from a tributary of the Mississippi River in northern Mississippi; other specimens in these series have normal counts. Occasional populations contain an unusually high percentage of individuals with an anal ray count other than 9, a condition also encountered in certain populations of Notropis cornutus. Counts of 81 specimens of isolepis (UMMZ 157780) 170 BULLETIN FLORIDA STATE MUSEUM Vol. 8 * is ./ K:/n. 14; 1. - 4741 -'. '2 & ; Y. rk15' 9 -i:N**,- - leo.1. 4 . 50>..1../ ]$*44. N#'42 0**53-5- *trt<#as ~ 6-54*: 0 0• 0 00 A 8/C D F t»»BS»*6%**2&,10~ix*6 - Figure 15. Pigmentation patterns and arrangement of dorso-lateral scales in Luxihis. A. NOtropts chn/socephalus isolepts, UMMZ 161262, Red River drain- age, Louisiana. B. N. c. chr!/socephalus, UMMZ 100877, Wabash River drainage, Indiana. C. N. atbeolus, CU 20322, Roanoke River drainage, Virginia. D. N. comutus, UMMZ 9526, St. Lawrence River drainage, New York. E. N. cornutus, UMMZ 174818, James River drainage, Virginia. F' N. albeolus, UMMZ 174749, Roanoke River drainage, Virginia. 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 171 from the Chickasawhay drainage in Mississippi show 14 with 9 anal rays and 17 with 10. Sexual variation is pronounced in N. c. isolepis, as in the closely related N. cornums and N. c. chrysocephalus. Differences presumably are identical to those occurring in the latter form and very similar to those in the former. Breeding males develop a deep red coloration on body and fins, the black crescent-shaped markings on the side of the body become notably darker and more pronounced, and prom- inent nuptial tubercles develop on the head, back, anterior-part of the dorsal and upper surface of the pectoral fins. Females have a more subdued coloration and usually lack tubercles. Adult males also de- velop a deeper and more compressed head, have slightly longer pec- toral and pelvic Bns, and an increased amount of adipose tissue sur- rounding the Bn rays, but apparently do not differ in other respects from the females (table 6). RELATIONSHIPS. N. c. chrysocephalus is the only form of Luxilus that occurs with N. c. isolepis in the Gulf drainage. Both are found in the Coosa River system and, although they occur in adjacent tribu- tary streams a few miles apart, they have not yet been found to- gether. Specimens from one collection taken. in an eastern tributary of the neighboring Black Warrior system have been identified as in- tergrades, and several other collections in the same drainage have been identified as N. c. chrusocephalus. Perhaps transfer of chruso- cephalus into the Black Warrior has been effected through a localized stream capture. The apparent lack of intergradation of chryso- cephalus and isolepis in the Coosa basin is difficult to explain, though this may in some way be related to the fact that, in this area at least, the ranges of the two forms appear to be separated by the s6-called Fall Line which marks the transition from an upland to a lowland environment (Fenneman, 1988: 126-131). On the other hand these forms may be incipient species which, because of ecological similari- ties, have each prevented the other from moving into the areas it occupies. Should this explanation be correct the intergradation in the Black Warrior drainage would be difficult to explain. DISTRIBUTION. Confined to tributaries of the Gulf of Mexico from the eastern half of the Red River system (including the Ouachita River) in Texas, Louisiana, Arkansas and Oklahoma, eastward to the Tallapoosa drainage in western Georgia (UF 9568). Its range west of the Mississippi River is, with one exception, re- stricted to the Red River system. A single record from the Arkansas drainage in Arkansas (USNM 165878) is interpreted as the result of either human introduction or localized stream capture. This form 172 BULLETIN FLORIDA STATE MUSEUM Vol. 8 apparently is fairly common in favorable habitats in southern Ar- kansas, extreme southeastern Oklahoma, and extreme northeastern Texas, but farther west is known only from the Blue River system of south-central Oklahoma. To the east it occurs in a number of smaller tributaries of the Mississippi River in western Mississippi, and re- cently has been recorded as far north as southwestern Tennessee (USNM 179780-179781). It probably does not normally occur in any independent river system east of Mobile Bay, although there are un- published records (not verified by me) from the Escambia and Choc- tawhatchee drainages in Alabama. In the Coosa River system isolepis occurs as far north as Chilton County, Alabama; here its range dove- tails with that of N. c. chlysocephalus. It is one of the most common fishes in upland tributaries to the Gulf of Mexico. LIFE HISTORY AND ECOLOGY. No studies dealing with the life history and ecology of N. c. isolepis have been made. The form probably differs little, if at all, from N. c. chrysocephalus in these regards. 1964 GILBERT: FISHES OF THE SUBGENUS LUXILUS 178 TABLE 1 Comparison of Notropis coccogenis and Notropis zonistius Character Notropis coccogenis Notropis zonistius Position of band Confined to posterior part Not confined to posterior on dorsal fin of fin; almost parallels distal part of fin; slopes obliquely margin forward and intersects lower third of anterior margin Color of dorsal Black at all ages Orange-red in young; black band in adults Caudal spot Absent Present BIack border on Present Absent distal margin of caudal fin Red bar on side Prominent; situated at mar- Not prominent (often ab- of head gin of opercle and pre- sent); situated on cheek opercle Nuptial tubercles Absent ' Present on top of head and back Body forrn Attenuate and terete Deep and compressed Angle of mouth to Usually 49 to 58' Usually 57 to 61« top of head Number of 40 to 42; usually 41 88 to 40; usually 89 vertebrae Maximum length Ca. 115 mm. Ca, 85 mni. of males 174 BULLETIN FLORIDA STATE MUSEUM Vol. 8 TABLE 2 Comparison of pigmentary characters in Notropis zonattis and Notropis pilsbryi Character Notropis zonatus Notropis pilsbrvi Pigment on posterior- Not bordering posterior- Bordering posterior- dorsal part of opercle dorsal margin dorsal margin Pigment on cleithral Extending ventrally to Extending ventrally about area pectoral insertion; be- halfway to pectoral in- coming quite dark sertion; not becoming particularly dark Relation of lateral stripe Not extending below Extending below lateral to lateral line (anterior lateral line line to caudal peduncle) Distribution of pigment Unequally distributed; Equally distributed on caudal peduncle more present above lat- above and below lateral eral line than below line Dorso-lateral stripes Present Absent on back Red pigmentation on Usually more Usually less pronounced body and Rns pronounced Black pigmentation More intense; appearing Less intense; appearing on body deep black more dusky TABLE 8 Comparison between forms of the Notropis cornutus species group 1964 G ILBER T: FISH ES O F TH E SUBG ENUS LU XILU S N. c. chruso- Character N. cornutus N. albeolus cephalus N. c. isolepis N. cerasinus Anterior dorso-lateral (16) 18 to 24 (30) (15) 17 to 20 (28) (18) 14 to 17 (19) (12) 13 or. 14 (15) (14) 15 to 19 (22) scales Circumferential scales (26) 30 to 85 (39) (24) 26 to 30 (38) (28) 26 to 29 (82) (23) 24 to 28 (23) 25 to 28 (30) Sum of anterior dorso- (42) 48 to 59 (67) (41) 43 to 50 (55) (88) 40 to 44 (48) (36) 37 to 41 (48) (87) 41 to 47 (51) lateral and circumfer- ential scales Caudal peduncle (13) 15 or 16 (18); (18) 15 or 16 (17) (12) 15 or 16 (17); (14) 15 or 16; (12) 13 to 15 (16); scales usually 16 usually 16 usually 16 usually 14 Parallel stripes on Absent Absent Present; Present; Absent anterior dorso-lateral crooked straight part of back Crescent-shaped bars Present Usually absent Present Present Present onside of body Reds on body and fins Present and Usually absent; Present and Present and Present and of breeding males widespread occasional on snout widespread widespread widespread and fins Color of dorsum Olive-brown Olive-green Olive-brown Olive-brown Olive-brown Nuptial tubercles Males only Males only Males only Males only Both sexes present in: Appearance of nuptial Straight at tips Straight at tips Straight at tips Straight at tips Curved forward tubercles on upper at tips part of head Body forrn Deep and Usually slender and Deep and Deep and Deep and compressed terete in north- compressed compressed compressed ern race; usually deep and compressed in southern race 3 Maximum size of males Ca. 175 mm. Ca. 180 mm. Ca. 175 mm. Ca. 175 mIn. Ca. 90 mm. 176 BULLETIN FLORIDA STATE MUSEUM Vol. 8 TABLE 4. COMPARISON OF PROPORTIONAL MEASUREMENTS