INFORMATION TO USERS
This reproduction was made from a copy of a document sent to us for microfilming. While the most advanced technology has been used to photograph and reproduce this document, the quality of the reproduction is heavily dependent upon the quality of the material submitted.
The following explanation of techniques is provided to help clarify markings or notations which may appear on this reproduction.
1. The sign or “target” for pages apparently lacking from the document photographed is “Missing Page(s)”. If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting through an image and duplicating adjacent pages to assure complete continuity.
2. When an image on the film is obliterated with a round black mark, it is an indication of either blurred copy because of movement during exposure, duplicate copy, or copyrighted materials that should not have been filmed. For blurred pages, a good image of the page can be found in the adjacent frame. If copyrighted materials were deleted, a target note will appear listing the pages in the adjacent frame.
3. When a map, drawing or chart, etc., is part of the material being photographed, a definite method of “sectioning” the material has been followed. It is customary to begin filming at the upper left hand corner of a large sheet and to continue from left to right in equal sections with small overlaps. If necessary, sectioning is continued again-beginning below the first row and continuing on until complete.
4. For illustrations that cannot be satisfactorily reproduced by xerographic means, photographic prints can be purchased at additional cost and inserted into your xerographic copy. These prints are available upon request from the Dissertations Customer Services Department.
5. Some pages in any docum ent may have indistinct print. In all cases the best available copy has been filmed.
Universi^ Micrcxilms International
McGeehan, Lawrence Thomas
MULTIVARIATE AND UNIVARIATE ANALYSES OF THE GEOGRAPHIC VARIATION WITHIN ETHEOSTOMA FLABELLARE (PISCES: PERCIDAE) OF EASTERN NORTH AMERICA
The Ohio State University
University Microfilms I ntern eti on elæ o N. zeeb Road, Ann Arbor. Ml 48106
Copyright 1985 by McGeehan, Lawrence Thomas All Rights Reserved
In all cases this material has been filmed in the best possible way from the available copy. Problems encountered with this document have been identified here with a check mark V
1. G lossy photographs or p a______g e s
2. Colored illustrations, paper or print______
3. Photographs with dark backgroundi / '
4. Illustrations are poor copy______
5. Pages with black marks, not original copy
6. Print shows through as there is text on both sides of page______
7. Indistinct, broken or small print on several pages
8. Print e x c e e d s margin req u irem en______ts
9. Tightly bound co p y with print lo st in sp______in e
10. Computer printout pages with indistinct print______
11. P a g e (s______) lacking when material received, and not available from school or author.
12. P a g e (s)______seem to b e missing in numbering only a s text follow s.
13. Two p ages n u m b ered______. Text follow s.
14. Curling and wrinkled p a g______e s
15. O ther______
University Microfilms International
MULTIVARIATE AND UNIVARIATE ANALYSES OF THE GEOGRAPHIC
VARIATION WITHIN ETHEOSTOMA FLABELLARE (PISCES:
PERCIDAE) OF EASTERN NORTH AMERICA
DISSERTATION
Presented in Partial Fullfillment of the Requirements for
the Degree Doctor of Philosophy in the Graduate
School of The Ohio State University
By
Lawrence T. McGeehan
The Ohio State University
1985
Reading Committee: Approved By
Dr. Paul C. Baumann
Dr. Ted M. Cavender
Dr. Willard C. Myser Adviser
Dr. Barry D. Valentine Department of Zoology Copyright by
Lawrence T. McGeehan
1985 ABSTRACT
A series of multivariate discriminant function analyses, utilizing 31 meristic, morphometric, and pigmentation characters, indicated Etheostoma flabellare Rafinesque is a widespread polytypic species. There are five recognizable (92 to 96 percent of the time) allopatric forms within E. f l a b e l l a r e . The five allopatric forms are combined into two multivariate groups : a Montane-East Slope group, including E. f. robustum (n. s s p . ), E. f . h u m e r a l i s , and E. f. brevispina; and an Interior group, including an Ozark-Tennessee River assemblage and E. f . flabellare. The nominate form E. f . flabellare inhabits the Mississippi, Ohio, and Great Lakes basins, and isolated localities in the Mohawk, Hudson, and upper Susquehanna Rivers; indicating the subspecies ability to transgress drainage divides, especially in terrain formerly covered by Wisconsinan glacial ice. Etheostoma f. lineolatum (Agassiz) is reduced to a synonym of E. f. flabellare; since there is clinal variation between the two forms, any division along this Cline would be arbitrary. Another questionable form (recognized via discriminant analysis 92% of the time from adjacent populations), herein termed the Ozark-Tennessee group, inhabits the White River system of the Ozark Uplands in Missouri and Arkansas, and tributaries of the lower Tennessee River in the southern Appalachians. Further analysis of this form (including specimens from the geographically intermediate populations, i.e., the Duck, Clarks, and Buffalo Rivers) is needed to clearly define its geographic distribution and relationship with the nominate form. Etheostoma f. robustum (new subspecies) inhabits the upper Tennessee River system (upstream of the confluence of the Little Tennessee River), New River, and as a localized population in headwaters of Shavers Fork Cheat River (Monongahela River drainage). Etheostoma f . flabellare and E. f . robustum are both present in Shavers Fork, but are not syntopic. Intergrades between the two subspecies are identified from headwaters of the Guyandot, Coal, and Elk Rivers, West Virginia. Etheostoma f . humeralis inhabits the Atlantic slope drainages, including the lower Susquehanna, Potomac, James, Roanoke, Neuse, and probably Cape Fear Rivers. Etheostoma f . brevispina is the most southern representative of the barred fantail group on the Atlantic slope, inhabiting the Catawba, Broad, and Pee Dee River drainages in North and South Carolina. ACKNOWLEDGMENTS
I wish to thank my adviser. Dr. Ted M. Cavender, for
his encouragement and guidance during the course of my
graduate studies at Ohio State University. His enthusiasm
for Ichthyology is apparent and inspirational. Without his
tenacious support this work would not have been completed.
I would also like to thank my reading committee,
including Dr. P. Baumann, Dr. T. Cavender, Dr. W. Myser, and
Dr. B. Valentine. Their suggestions were most helpfull.
This work also benefited from the assistance, during
several field trips, of Miles Coburn, William LeGrande, Ted
Cavender, Rich Carter, and Kathy Chan.
I also thank the following for specimens borrowed or
locality information provided: B. Branson, Eastern Kentucky
University; T. Buchanan, Westark Community College; T.
Cavender, Ohio State University Museum Zoology; D. Cincotta,
West Virginia Department of Natural Resources-Elkins; E.
Cooper, Pennsylvania State University; D. Etneir, University
of Tennessee; C. Hocutt and J. Stauffer, Appalachian
Environmental Laboratory University of Maryland Frostburg;
R. Jenkins, Roanoke College; D. Lee, North Carolina State
Museum; W. LeGrande University Wisconsin Stevens Point; E.
Menhinick,.University North Carolina; L. Page, Illinois
Natural History Survey; W. Pflieger Missouri Department of Conservation; R. Schoknecht, Cornell University; M.
Warren, Kentucky Nature Preserves; and A. White, John
Carroll University.
I am especially gratefull to my wife, Susie. Her unwavering support is deeply appreciated.
The understanding and support offered by my parents and family, especially my sons Brian and Miles, were valuable sources of inspiration during my studies. August 1, 1951 Born - Bethlehem, Pennsylvania
Received primary and secondary education in St. Marys, PA
1 9 7 3 ...... B.S., John Carroll University
1 9 7 5 ...... M.S., John Carroll University
1 9 7 3 - 1 9 7 5...... Graduate Assistant, John Carroll University
1 9 7 5 - 1 9 8 2 ...... Teaching Assistant, Ohio State University
WORK EXPERIENCE
1985 Special Assistant to Governor's Science Advisor
1983-1984 Researcher for Ohio Environmental Protection Agency Office of Planning
1973-1975 Field Coordinator, Environmental Resource Associates
FELLOWSHIPS
Mary 0‘sburn Summer Fellowship, 1980 Graduate Research Associate, 1979 (Summer)
RESEARCH INTERESTS
Evolution of North American freshwater fishes
Multivariate analyses of morphologic variation of populations
Master's Thesis: The effects of acidic effluents on the freshwater fishes of the Clarion River basin, Pennsylvania
PRESENTATIONS
Multivariate analysis of geographic variation in E. flabellare, 1982 Annual Meeting of American Society of Ichthyologists and Herpetologists TABLE OF CONTENTS
Page ACKNOWLEDGMENTS ...... Ü
VITA ...... iv
TABLE OF CONTENTS ...... V
LIST OF TABLES ...... viii
LIST OF FIGURES ...... x
INTRODUCTION...... 1
Brief Historical Overview ...... 2
METHODS ...... 10
RESULTS ...... 22
PART I: MULTIVARIATE ANALYSES ...... 22
Analysis of E,. f,. flabellare and E. f. l i n e o l a t u m...... 29
Taxonomic Conclusions; Interior Group ...... 42
Discriminant Function Analysis of Montane-East Slope Group ...... 43
Taxonomic Conclusions: Montane-East Slope Group. 49
Discriminant Function Analysis Of All 1,114 Specimens In Five Groups: Ozark-Tennessee, E. f . f l a b e l l a r e , E. f. robustum, E. f. humeralis and E. brevispina ...... 50
Principal Component Analysis: JE. f^. flabellare and E.. f . robustum...... 55
PART II: PIGMENTATION, MERISTIC, AND MORPHOMETRIC (RATIO) VARIATION...... 60
Pigmentation Variation ...... 60
Lateral Bars ...... 60 Page L i n e s ...... 63
Pigmentation Of The Ozark-Tennessee Group ...... 75
MERISTIC VARIATION ...... 76
S c a l a t i o n...... 76
Lateral Line Scales ...... 76
Pored Lateral Line S c a l e s ...... 79
Index of Pored to Total Lateral Line Scales ...... 82
Fin Ray Elements ...... 82
Dorsal Spines ...... 82
Dorsal Soft Rays ...... 85
Pectoral Rays ...... 85
Anal R a y s ...... 87
MORPHOMETRIC VARIATION ...... 87
Spinous Dorsal Fin Base Length ...... 87
Spinoup Dorsal Height ...... 95
Pectoral Ray Length ...... 95
Caudal Peduncle Depth ...... 95
Eye Diameter ...... 97
Head W i d t h ...... 97
Head Length ...... 97
Morphologic Convergence: Southwestern and Atlantic Slope Groups ...... 101
TAXONOMIC DISCUSSION ...... 102
ZOOGEOGRAPHIC DISCUSSION ...... 119 Page
DISCUSSION: EVOLUTION WITHIN ETHEOSTOMA FLABELLARE ZOOGEOGRAPHIC AND TAXONOMIC CONCLUSIONS...... 133
SUMMARY ...... 139
LITERATURE CITED ...... 142
APPENDIX A Material Examined...... 149 LIST OF TABLES
1. Names of Sample Localities in Figure 3 ...... 14
2. Standardized canonical variate coefficients for variables used in discriminant analysis of all 50 samples of JE. flabellare. Standardized coefficients indicate the relative contribution of each character to each variate ...... 27
3. Standardized coefficients of variables used in the discriminant analysis of 37 Interior samples ...... 30
4. Standardized coefficients for variables used in the discriminant analysis of 32 samples historically recognized as E. flabellare and f.. l i n e o l a t u m...... 33
5. Standardized coefficients for variables used in the discriminant analysis of "ancestral" samples from south of the glacial terminus ...... 37
6. Jackknifed classification table for "ancestral" and "descendant" samples. The DFA was defined utilizing "ancestral" samples only ...... 39
7. Standard deviations for characters most influential in discriminant function analysis between E.. f . flabellare and _E. f_. lineolatum...... 41
8. Standardized coefficients of variation for 13 Montane-East Slope samples...... 46
9. Jackknifed classification matrix of thirteen Montane-East Slope samples pooled into three groups: Montane, Mid-Atlantic and Southern Atlantic...... 48
10. Standardized coefficients for variables used in the discriminant analysis of 1,114 specimens of flabellare in five groups ...... 54 Table Page
11. Jackknifed classification matrix of all 1,114 specimens pooled into five g r o u p s ...... 56
12. Factor loadings for principal component analysis of E. f. flabellare and JE. f. robustum. Mensural variables were log transformed...... 58
13. Mean differences in meristic and pigmentation characters for subspecies of E_. flabellare. Standard deviations are in parentheses. The last two columns are results of parled t-test among three f o r m s ...... 77
14. Mean differences in proportional measurements of mensural characters for subspecies of E^. flabellare Expressed in thousandths of SL. (standard deviations are in pa r e n t h e s e s ) ...... 94
15. Proportional measurements of Etheostoma flabellare robustum expressed as thousandths of standard l e n g t h ...... 117
16. Average differences in meristic characters for the Ozark-Tennessee form and adjacent populations. Standard deviations are given in parentheses...... 126 LIST OF FIGURES
Figure Page
1. Distribution of Etheostoma flabellare (modified from Lee et al., 1980; and Page, 1983).... 3
2. Iconotype for E. flabellare. Drawing from Rafinesque's notebook, as reprinted in Collette and Knapp (1967)...... 6
3. Sample localities (numbers refer to names in Table 1) 13
4. Canonical variate analysis of all 50 samples. Sample centroids are projected onto the first two canonical variâtes. Lower diagram provides a vector sketch of the canonical coefficients...... 24
5. ÜPGMA cluster diagram of 50 sample centroids as projected onto the first three canonical variâtes... 25
6. Canonical variate analysis of the 37 samples included in the Interior group ...... 28
7. Canonical .variate analysis of the 32 samples historically recognized as E. f. flabellare or E. _f. l i n e o l a t u m ...... 31
8. Canonical variate analysis of "ancestral" populations, with "descendant" populations projected onto this discriminant space ...... 35
9. Canonical variate analysis of the thirteen Montane- East Slope samples ...... 45
10. Three dimensional ordination of canonical analysis of the five forms recognized in this study. Darkened lines denote one standard deviation. Not drawn in perspective ...... 53 Figure Page
11. Principal component analysis of E. f. flabellare and E. f. r o b u s t u m . E. flabellare = black dots; E. f. robustum = open squares (Greenbrier and Gauley River specimens) and open circles (Shavers Fork Cheat River specimens). Intergrades from Coal River = black stars ...... 57
12. Variation in lateral bars. The diagrams indicate the mean (center point), two standard errors of the mean (black rectangle), one standard deviation on either side of the mean (outer limits of open rectangle), and sample range (basal line) ...... 62
13. Variation in lateral bar number within the New River d r a i n a g e ...... 64
14. Geographic variation in mean percent of scale rows which are lined. Solid black line approximates the Wisconsin glacial boundary ...... 66
15. Variation in lines. Samples are arranged west to east in three groups, representing transects; south of the Wisconsinan glacial terminus, north of the glacial terminus, and across the Great Lakes ...... 68
16. Variation in lines in selected populations. Samples are generally arranged in a west to east transect... 70
17. Variation in lateral line scales ...... 78
18. Variation in pored lateral line scales ...... 80
19. Variation in pored lateral line scales in Interior populations of fantails ...... 81
20. Variation in the ratio of pored lateral line scales to lateral line scales ...... 83
21. Variation in dorsal spines ...... 84
22. Variation in pectoral r a y s ...... 86
23. Variation in anal rays ...... 88 Figure Page
24. Variation in the ratio of gape width to standard length. Samples are arranged in two transects, south and north of the glacial border ...... 89
25. Variation in proportion of body depth to standard l e n g t h ...... 90
26. Variation in caudal peduncle depth to standard l e n g t h ...... 91
27. Variation in soft dorsal fin length (expressed in thousandths of standard length)...... 92
28. Variation in the proportion of soft dorsal fin length to standard length...... 93
29. Variation in the proportion of dorsal fin height to standard length ...... 96
30. Variation in proportion of caudal peduncle depth to standard length ...... 98
31. Variation in proportion of eye diameter to standard l e n g t h...... 99
32. Variation in proportion of head width to standard length..'...... 100
33. Distribution of E. flabellare subspecies and an Ozark-Tennessee form ...... 103
34. Adult males of Etheostoma f. flabellare: A. E. f^. flabellare B. E. f. robustum C. E. f. humeralis D. E. f. b r e v i s p i n a...... 105
35. Adult males of Etheostoma flabellare: A. Intergrade E. f^. flabellare x E. f. robustum B . f^. flabellare Middle Cumberland R. C. E. f. flabellare? North Fork White R. D. E. f.. flabellare lined s p e c i m e n ...... 109
36. Variation in the ratio of pored lateral line scales to total lateral line scales...... 129
xii: Figure Page
37. Head profiles of adult males (drawn from photo projections): A. Etheostoma f. r o b u s t u m . B. E. f. robustum x E. f . f l a b e l l a r e . C. E. f. f l a b e l l a r e...... 131
38. Geographic range of E. flabellare (enclosed by dark line) and the ranges of the other nine members of the subgenus Catonotus (stippled areas). Ranges are from Page (1983)...... 137 INTRODUCTION
Although Etheostoma flabellare is a geographically variable species, multi- and univariate analyses of the variation indicates the differing forms should be considered as one species. A series of discriminant function analyses, utilizing 31 characters on 1,114 specimens (from 50 sample areas), supplemented by principal component analysis and standard morphometric and univariate comparisons, supports the recognition of four geographic subspecies, with a fifth form needing additional research to ascertain its validity.
These forms are all allopatric with no evidence of syntopy, although intergrades between two forms are recognized. There are numerous geographic areas where individual forms are found contiguously across headwater divides.
The analyses of character variation supports the reduction of E. f . lineolatum as a synonym of E. f . flabellare, and together with the fantail populations inhabiting the White and lower Tennessee Rivers (which comprise a questionable taxonomic form) are phenetically lumped into an "Interior" assemblage.
Populations inhabiting the upper Tennessee, New, and
Shavers Fork Cheat Rivers are described as a new subspecies, 2
E. f . robustum. Etheostoma f. brevispina is herein limited to the east slope drainages of the
Saluda-Catawba and Pee Dee Rivers of Tennessee, North and
South Carolina, and Virginia. Etheostoma f. humeralis is resurrected and is applied to fantail populations inhabiting the Susquehanna (lower sections), Potomac, James,
Roanoke, and Neuse Rivers, and possibly the Cape Fear River
(only published photographs were examined). Etheostoma f . r o b u s t u m , E. f . b r e v i s p i n a , and E. f . humeralis comprise a second phenetic grouping, termed the
"Montane-East Slope" assemblage.
Phenetically intermediate populations, interpreted as intergrades between E. f . flabellare and E. f . robustum, are recognized from the Guyandot, Coal, and Elk
River systems in West Virginia.
Brief Historical Overview
The barred fantail darter, Etheostoma flabellare
Rafinesque, is one of the most common and widely distributed percids in eastern North America (Fig. 1). This member of the subgenus Catonotus (established by Bailey and Gosline,
1955; diagnosed by Kuehne and Small, 1971; and modified by
Braasch and Page, 1979; and Page, 1981) typically inhabits gravel riffles of small to medium sized upland to montane streams, but has also been collected over wave swept gravel Fig.^ 11.' * Distribution M*Cof ‘C^ViAnr^-f'nmoEtheostoma f1flabellare q$^a1 1 a?*a (modified from Lee et , 1980; and Page, 1983). bars in lakes (Lake Erie and Cayuga Lake). It has been reported from the Hudson River drainage of New York, south in the Atlantic slope drainages to South Carolina, west through much of the Mississippi River basin to eastern
Oklahoma, north to Minnesota, and east through the Great
Lakes and St. Lawrence River basins.
As is common with other widespread aquatic species,
E. flabellare is geographically variable. Recently
(Jenkins et ^ . , 1972; Kuehne and Barbour, 1983; Page,
1983) the barred fantail darter has been considered polytypic with three recognized subspecies, E. f. flabellare, E. f . lineolatum, and E. f. brevispina; based primarily on pigmentation characters.
The nominate subspecies has been defined by Hubbs and
Lagler (1958) as typically having no or inconspicuous lengthwise rows of spots or dashes. Ross and Carico (1963)
included the presence of 8-14 well defined dark vertical bars extending from the dorsum to at least the lateral line
series. This form has been reported from the upper Ohio
River drainage, the eastern Great Lakes region (Hubbs and
Lagler, 1958), as well as the east slope of the Appalachians
(Ross and Carico, 1963). Pflieger (1971) noted a disjunct
population of E. f . flabellare inhabiting the Upper
Current and Black Rivers in southeastern Missouri. Etheostoma flabellare was originally described by
Constantine Rafinesque (1819) from specimens he collected
during a summer trip down the Ohio River from Pittsburgh,
Pennsylvania to the Falls of the Ohio, near Louisville,
Kentucky (Call, 1899). Although Rafinesque did not keep type material, his description (Call, 1899) and drawing (Fig. 2)
(Collette and Knapp, 1967) appear similar to fantail
populations inhabiting the upper Ohio River basin.
The second subspecies, E. f. lineolatum, has been
distinguished by the presence of conspicuous lengthwise rows
of spots or dashes forming a series of horizontal lines
across the flanks (Hubbs and Lag1er, 1958). Louis Agassiz
(1854) originally described this form (as Catonotus
lineolatus) from specimens collected by Dr. L. Watson, in
streams trilputary to the Mississippi River, near Quincy,
Illinois. This form has been subsequently recognized from
the Arkansas, Mississippi, Missouri, lower Tennessee, and
lower Ohio Rivers, the Great Lakes drainages in Wisconsin
and Upper Peninsula of Michigan, and as a disjunct
population inhabiting the headwaters of the Guyandot River,
West Virginia (Hubbs and Lagler, 1958; Jenkins et al.,
1972) .
Although individuals from the extremes of their ranges
can be readily identified to either subspecies (i.e., lined
or unlined), specimens with intermediate patterns have been J ' c /a ’- / £ -
Fig. 2. Iconotype for E, flabellare. Drawing from Rafinesque*'s notebook, as reprinted in Collette and Knapp (1967). recognized from a broad geographic zone extending from
eastern Illinois (Hubbs, 1926), Indiana (Hubbs and Lagler,
1941), western Ohio (Trautman, 1981), and central Kentucky
(Clay, 1975). Also, the lined and barred patterns are not mutually exclusive and vary independently, thereby creating overlapping geographic distributions and broad disjunctions.
Therefore, these characters were analyzed and interpreted separately, and in some analyses excluded completely.
The recognition and distribution of the third subspecies, E. f. brevispina, has also been problematic. This darter was originally described by Coker
(1926) as Richia brevispina, from three specimens collected in Paddys Creek, North Carolina, a tributary of the Catawba River. A year later, Hubbs (1927) synonymized this form with E. flabellare, indicating it's similarity with specimens from Maryland drainages.
Ross and Carico (1963) indicated races of brevispinna
[sic] are found in the Saluda-Catawba River of North and
South Carolina, the Yadkin and Neuse Rivers of North
Carolina, and the Tennessee River of Tennessee, North
Carolina, and Virginia. Jenkins, Lachner, and Schwartz
(1972), however, indicated that brevispina inhabits the
Roanoke, Pee Dee, New, and upper Tennessee River drainages.
They also noted the presence of the nominate form in the
Tennessee and Roanoke River drainages, although exact localities were not given. Ross and Carico (1963) believed the populations inhabiting the New and Roanoke Rivers represented intergrades between flabellare and brevispina.
In addition to these three recognized forms several ichthyologists (Pflieger, 1971; Lee et aJ., 1980; Kuehne and Barbour, 1983) have recently speculated that other populations currently recognized as E. flabellare probably warrant taxonomic recognition, both as species and subspecies. Pflieger (1971) noted the fantail populations in the White River system (Missouri) possibly warranted taxonomic recognition, although no diagnosis or distributional information was given. Kuehne and Barbour
(1983:148) state "questions concerning the fantail darter and possible relatives remain the largest species problem among darters of the genus Etheostoma."
The geographic variation, distributional limits, and biologic relationships of these differing forms have never been satisfactorily investigated. Previous analyses of variation within the E. flabellare complex have been limited due to the small number of populations (Ross, 1958;
Buhan, 1966; Pflieger, 1971) and/or characters examined
(Bailey and Gosline, 1955).
Therefore, the objectives of this study are: 1) characterize and describe via multi- and univariate 9 statistical techniques the variation and distribution of characters throughout the known range of E. flabellare as they apply to recognition of taxonomic forms; and 2) interpret the biologic and zoogeographic evidence regarding population variation, gene flow, dispersal, and natural selection theory. A total of 31 counts and measurements were made on
1,114 specimens of Etheostoma flabellare from 316 collections (listed in Appendix A). The analyses of meristic, morphometric, and pigmentation variation used the following characters (character abbreviations used in text are given parenthetically): lateral line scales (LL); pored lateral line scales (PL); scales above the lateral line
(SAL); scales below the lateral line (SBL); caudal peduncle scales (SACP); transverse scales (TRS); ventral scales
(VSQ); dorsal fin spines (Dl); dorsal fin rays (D2); pectoral fin rays (Pi); anal fin rays (AN); lateral bars
(LB); spots on caudal (SPC); horizontal lines (LINES); pre-isthmus length (ID ; head length (HL); head width (HW); body depth (BD); caudal peduncle depth (CPD); pectoral fin
length (PlL); pelvic fin length (P2L); anal fin length
(AND ; spinous dorsal fin base length (DID; spinous dorsal fin height (DlH); soft dorsal fin transverse length ( D2D; predorsal length (PROD; snout length (PROD; interorbital
10 11 width (lOW); eye diameter (ED); gape width (GW); and
standard length (STDL or SL).
Counts and measurements follow Hubbs and Lagler (1958),
except as noted below. Scales above the lateral line were
counted from the base of the soft dorsal fin downward and backward to, but not including, the lateral line scale.
Transverse scales are counted from the anal anterior
insertion to the spinous dorsal fin base, plus the scales
from the anal origin to the soft dorsal fin base. Ventral scales are counted in a U-shaped series just cephalad to the vent, starting and ending at the scale adjacent to the
lateral line series. The number of lateral bars is the number of vertical lateral blotches (extending to at least the lateral line series) counted from the posterior edge of the opercle to the caudal fin base. Spots on the caudal fin are the number of spots on the median ray of the caudal fin.
The number of lines is the number of scale rows (counted downward and backward from the origin of the soft dorsal fin base) showing five successive spots in a lateral scale
series. Pre-isthmus length is the shortest distance from the tip of the lower jaw to the median edge of the branchiostegal membrane. Body depth is measured just anterior to the pelvic fin insertion. Soft dorsal fin length
is measured transversely from the insertion of the first ray to the tip'of the longest posteriorly projecting ray. Gape 12 width is the transverse distance between the posterior edges of the mandibular joints.
Measurements were taken to the nearest 0.1 mm on alcohol preserved specimens with needle-tipped Helios dial calipers. Counting and measuring were aided by a Bausch and
Lomb zoom (1-2X) dissecting microscope and a fine jet of compressed air.
To reduce analytic complications associated with allometric growth and sexual dimorphism, only adult male specimens 027.7 mm STDL) were included in this study. Sex was determined by shape of genital papilla, body coloration, development of spinous dorsal bulbs, and examination of g o n a d s .
The 1,114 specimens examined were from 316 collections throughout t,he known range of Etheostoma flabellare.
Collections were initially pooled into 50 representative samples primarily by drainage basin (Fig. 3, Table 1).
Intrademe gene flow is assumed to occur within the individual samples. Whenever the homogeneity of interdrainage collections was suspect, the samples were analyzed individually. For the initial analysis, samples were not pooled from across natural geographic barriers
(i.e., waterfalls and drainage divides) or where zoogeographic disjunctions were obvious. Wêê^
Fig.Siii 3. Sample localities (numbers refer to names in Table 1). Table 1. Names of sample localities in Figure 3.
# Sample Name N
1. Lower Susquehanna River 12 2. Potomac River 40 3. James River 17 4. Western Roanoke River 15 5. Eastern Roanoke River 13 6. Dan River 21 7. Neuse River 16 8. Pee Dee River 33 9. Catawba River 16 10. Upper Tennessee River 28 11. Southern New River 55 12. Northern New River 24 13. Cheat River 10 14. Mohawk River 10 15. Lake Ontario drainages 16 16. Genesee River 16 17. Central Lake Erie drainages 31 18. Western Lake Erie drainages 32 19. Upper Susquehanna River 11 20. Allegheny River 19 21. Monongahela River 25 22. Youghiogheny River 16 23. Mahoning River 17 24. Muskingum River 38 25. Little Kanawha River 22 26. Coal River 11 27. Elk River 8 28. Guyandot River 32 29. Big Sandy River 11 30. Little Sandy River 18 31. Scioto River 78 32. Miami River 28 33. Licking River 33 34. Central Ohio R. drainages 8 35. Kentucky River 58 36. Salt River 15 37. Green River 32 38. Wabash River 29 39. Middle Cumberland River 17 40. Lower Cumberland River 21 41. Lower Tennessee River 19 42. Upper Current River 7 43. Lower Current River 4 44. North Fork White River 15. 45. Upper White River 7 Table 1 (con't)
Sample Name
46. Missouri River 27 47. Arkansas River 38 48. Illinois River 12 49. Rock River 5 50. Wisconsin River 28 since Etheostoma flabellare exhibits abundant geographic variation in many characters, data were analyzed with multivariate statistical techniques. Due to the relatively large number of variables (31) and initial groups
(50), a stepwise discriminant function analysis was considered most appropriate (Dixon, 1983). This technique is widely used in studies of geographic variation to establish the phenetic relationship among a priori groups (Gould and Johnston, 1972).
The stepwise multiple discriminant function analysis
(Dixon, 1983: Biomedical Data Programs BMDP 7M) selects linear combinations (classification functions) of optimal characters which best discriminate between groups. The process begins by selecting the variable which has the largest F-statistic generated in a one-way ANOVA. This initial variable is then paired with each of the other variables, one at a time, and a Wilks lambda is computed.
This is a measure of group discrimination which considers the difference between the group centroids and the cohesion
(homogeneity) within the groups. The next variable in conjunction with the previously entered variable(s ), which minimizes the Wilkes lambda, is then selected to enter the classification function.
This procedure is repeated until all variables are entered or no additional variables provide improvement of 17 group discrimination. Successive functions are derived such that their coefficients maximize group differences while being uncorrelated with previous functions. The maximum number of unique functions is equal to the number of groups minus one, or the number of discriminating variables, which ever is fewer (Klecka, 1980).
The second procedure of the discriminant function analysis, which is useful in the analysis of geographic variation, is the classification of individuals into groups.
Classification assigns each specimen to the group it most closely approximates, based on the classification functions.
Each individual is assigned to its own group (a hit) or to another group (a miss) (Pimentel and Frey, 1978). This analysis of Etheostoma flabellare presents jackknifed classification tables. This procedure computes the classification function for each specimen with that specimen omitted from the computations. The function thus generated is then used to classify the omitted specimen to the group it most resembles, resulting in a classification with less b i a s .
Classification is another means of checking the importance of the variables chosen by the discriminant analysis. The procedure is considered successful if few specimens are misclassified (Dixon, 1983). Classification, like a multivariate range test, can be used 18 for interpretation. The hits and misses, plus the probabilities of individuals being assigned to each group, indicate the uniqueness of individual groups and the phenetic interrelationships between groups (Pimentel and
Frey, 1978).
The final procedure of discriminant analysis which allows interpretation is the ordination of individuals in discriminant (canonical) space. Discriminant space is derived by rotating the orthogonal variable axes of
Euclidean space so the angle between any possible pairs of variable axes has a cosine that is the correlation coefficient between the variable pairs. These axes are selected so the first will disclose the greatest possible difference between groups. Each successive axis is selected such that it is orthogonal to all proceeding axes. Since there is one less axis than the total discriminating axes, they are termed canonical variable axes (Pimentel and Frey,
1978 ) .
The canonical variables are adjusted so the pooled within-group variances are one and their overall mean is
zero. The scale of the canonical variable graphs is in
standard deviation units (Dixon, 1983). The group centroids
are plotted in two or three canonical variate dimensions for
spatial interpretation. Included in the two-dimensional
canonical graphs is a diagram of the vectors 19 of the original variable axes, as they occur after rotation into the particular canonical space. The length and direction of these vectors generally disclose their amount and direction of positive influence, although the variables act synergistically in discriminant space, and the location of a group represents a complex interplay of the variables
(Pimentel and Frey, 1978). The first three canonical variâtes usually account for the greatest percent contribution of group discrimination (Pimentel, 1979). The group centroids, as plotted in the first three canonical dimensions, were utilized in an unweighted pair group method
(UPGMA) clustering procedure (Sneath and Sokal, 1973), which evaluates phenetic distances between groups.
The assumptions of discriminant analysis are homogeneity of variance-covariance matrices of groups and multivariate normality in sample distributions. Several authors have indicated that discriminant analysis is a robust technique, which can tolerate deviations from these assumptions (Lachenbruch, 1975; Pimentel, 1979). All mensural characters were log^ transformed to increase their linearity and homogeneity of the variance-covariance matrices (Pimentel and Frey, 1978).
To negate allometric variation, only adult specimens were measured and mensural characters were first log^ transformed, then converted to residuals (size free characters), by a partial covariance technique (Dixon, 1983:
BMDP 6M) where standard length was held as the independant variable (Thorpe, 1976). Discriminant function analyses
(DFA) were also computed using ratios of mensural characters
( e.g., body measurement/standard length). However, the results were virtually identical, therefore, ratios are only presented in tabular form to ease comparison of final r e s u l t s .
Principal component analysis (PCA) was also utilized to analyze patterns of character covariation. In PCA a set of original variables is used to generate new sets of variables, which are linear combinations of the original variables but are uncorrelated with each other. The new variables (principal components) define independent patterns of variation among the original variables, which may then be interpreted separately. The principal components are hierarchically ordered so the first one accounts for the maximum percentage of variance possible for a single linear combination of the original variables. The second represents the combination of variables (uncorrelated with the first) which explains the second largest amount of variance, and so on (Pimentel, 1979; Page and Cordes, 1983). PCA does not assume group membership or differences among samples, but produces coordinates along axes of greatest variation and is therefore an unbiased technique for discovering patterns of 21 character covariation (Humphries et aJ.. , 1981; Todd et al., 1981).
Abbreviations used herein for institutions are;
Appalachian Environmental Laboratory, University of
Maryland, Frostburg (AEL); Cornell University (CU); Eastern
Kentucky University (EKU); Illinois Natural History Survey
(INHS) ; Museum Comparative Zoology (MCZ); Ohio State
University Museum of Zoology (OSUM); Pennsylvania State
University (PSU); United States National Museum (USNM);
University of Wisconsin Stevens Point (UWSP); Westark
Community College (WJC); West Virginia Water Resources
Department of Natural Resources (WVWR). PART I: MULTIVARIATE ANALYSIS
The results of the initial discriminant function
analysis of all 50 samples of Etheostoma flabellare are presented in Fig. 4, and represent their primary phenetic
relationship. The first two canonical axes account for 43.0%
of the total dispersion (26.5% and 16.5% respectively). All
characters recorded were initially included in this analysis, except LINES, due to its random geographic occurrence (see discussion on pigmentation: LINES). A subset
of 20 mensural and meristic characters was selected by the
analysis, as optimal for discrimination among groups (Table
2) .
The samples ordinate into two phenetic clusters along
the first canonical axis (CV 1). Geographic data were not
included as variables, although clusters generally represent
aggregates of eastern versus western samples (Fig. 4);
herein termed the "Montane-East Slope" and "Interior"
groups. This division is further substantiated by an UPGMA
clustering analysis of the sample centroids (Fig. 5), as
22 Fig. 4. Canonical variate analysis of all 50 samples. Sample centroids are projected onto the first two canonical variâtes. Lower diagram provides a vector sketch of the canonical coefficients. CANONICAL VARIATE 1
Fig. 4. 3 4 C. Ohio 32 Miami
Fig. 5, UPGMA cluster diagram of the 50 sample centroids as projected onto the first three canonical variâtes. 26 defined by the first three canonical variâtes (accounting for 54.3% of the total dispersion). The only discrepancy between the two methods is the inclusion of the Upper
Tennessee River samples with the Interior group. This inclusion is considered spurious due to constraints of the
UPGMA clustering procedure, and in light of the jackknifed classification and Mahalanobis distance data (not given), which indicate that this sample is pheneticly most similar to the Montane-East Slope group, as shown in Fig. 4.
Characters most influential in this discrimination are PL and LL (Table 2). The Montane-East Slope samples tend to have a greater number of pored lateral line scales and fewer total lateral line scales.
There is also considerable variation among samples within the Montane-East Slope and Interior groups, primarily along the second canonical variate. Several univariate and multivariate analyses were applied to the two groups to further analyze within group variation.
In order to evaluate other characters and character combinations which may clarify the phenetic relationships among Interior samples, a discriminant function analysis was applied to the 37 samples included in this group.
The DFA clustered the 37 Interior samples (812 specimens) into two phenetic groups (Fig. 6); the first includes the Lower Tennessee, Upper White, and Current River Table 2. Standardized canonical variate coefficients for variables used in discriminant analysis of all 50 samples of flabellare. Standardized coefficients indicate the relative contribution of each character to each variate.
Character CV 1 CV 2 CV 3
PL -.611 -.375 .276 LL .516 -.076 .002 -.327 .302 -.034 Dl -.032 -.548 -.507 ED -.296 -.134 -.298 D2L .157 .566 -.343 SPC .061 -.519 .087 Pi .102 -.079 -.216 1RS .348 -.166 -.013 row .022 -.091 -.318 DlL .399 .347 .123 AL ' .122 -.122 .029 GW -.016 .109 -.131 DlH .012 -.297 -.223 IL .293 .243 .343 PlL -.276 .085 -.005 HL .253 -.131 .052 D2 .010 .002 .116 -.114 -.029 -.246 HW -.255 -.090 -.320 % Variance explained 26.5 16.5 11.3 : I i
CANONICAL VAHIATE 1
Fig. 6. Canonical variate analysis of the 37 samples included in the Interior group, . 29 samples (termed the Ozark-Tennessee group) and the second includes the remaining Interior samples, generally arranged in an east-west orientation paralleling canonical variate 1
(CV 1). The stepwise procedure included 16 of the original
28 mensural and meristic characters entered (Table 3); LINES and LB were not originally included. The first two canonical variâtes explain 41.6% of the total dispersion, 23.1% and
18.5% respectively.
Collectively the five samples in the White (Upper White and Current Rivers) and Lower Tennessee Rivers are separated from the remaining 32 Interior samples 92% of the time by the jackknifed classification procedure. This assemblage is primarily distinguished by typically having a greater number of rays in the soft dorsal fin, fewer dorsal spines, less spots on the caudal fin, and fewer lateral line scales
(Table 3). '
Analysis of flabellare and E. f. lineolatum
The second cluster depicted in Figure 6 includes samples which have been historically recognized as either flabellare or lineolatum. In order to evaluate the phenetic relationships of these 32 samples (including 762 specimens), a separate discriminant function analysis was employed (Fig. 7), utilizing the same 28 initial variables as in Figure 6. The first two canonical variâtes account for Table 3. Standardized coefficients of variables used in the discriminant function analysis of 37 Interior samples.
-.234 -.058 .171 -.427 TRS .143 .084 D1 .074 .454 SPC .676 .238 D2L -.643 .156 ED -.079 .297 PI -.274 -.044 low -.124 .262 P2L -.043 .083 GW -.187 .422 D2 -.138 -.263 IL -.019 -.337 LL .284 .117 DIL . -.124 -.066 DIH .191 .316
% Variance explained 23.1 18.5 Total =41.6 19^0,6'^^ 26
17 27
36 40 35
CANONICAL VARIATE 1
Fig, 7. 'Canonical variate analysis of the 32 samples historically recognized as E. £. flabellare or E. £, lineolatum, 32
42.3% of the total dispersion, 26.2% and 16.1% respectively.
As in the previous analysis, the 32 samples ordinate into an approximate east-west multivariate dine paralleling CV 2.
Although there is a strong polarity between the eastern and western samples (sensu flabellare and lineolatum) from the extremes of their ranges, there are no large disjunctions anywhere along the multivariate dine.
Characters most influential in this east-west dine, ranked in decreasing order of importance to CV 1, are D2L, SPC, PL, and GW (Table 4).
The Ohio basin samples tend to have more pored lateral line scales, narrower gapes, thinner bodies and caudal peduncles, and shorter soft dorsal fins. A noted exception to the ordination of samples into an east-west dine paralleling CV 1 is the intermediate place of the Arkansas
River samples (Fig. 7). This is primarily due to their relatively large number (x=34.2) of pored lateral line scales, in comparison to geographically adjacent populations, e.g., Missouri River (x=26.7).
Not only is there an east-west orientation of the sample centroids as projected in the discriminant space, there is a north-south orientation as v/ell. Characters most
influential in developing the north-south polarity paralleling,CV 2 are CPD, POL, and SPC (Table 4).
Northwestern samples tend to have greater caudal peduncle Table 4. Standardized coefficients for variables used in the discriminant analysis of 32 samples historically recognized as E. f. flabellare and E. f. lineolatum.
low -.209 -.272 PL .440 .146 1RS .054 .115 CPD -.141 .523 Pi -.183 -.251 SPC .478 .497 D2L -.627 .017 POL -.104 .479 P2L -.051 -.365 DIL -.133 .311 HL .072 -.460 GW _ -.333 .030 Dl -.149 -.168 DIH -.010 .108 AN .234 -.212
%Variance explained 26.2 16.1 Total =42.3 34 depths, increased preorbital lengths, and more spots on the caudal fin.
The range of E. flabellare in the Mississippi and
Ohio River basins is traversed (east to west) by the terminal margin of the Wisconsinan continental glaciation.
Since the retreat of the Wisconsinan glaciation, approximately 10,000 years B.P. (Goldthwait et al.,
1965), the drainages north of the glacial maximum would be available for recolonization by fantails which survived glaciation in southern réfugia.
In order to investigate hypotheses concerning ancestral-descendent populations, glacial réfugia, dispersal routes, and gene flow, a discriminant function analysis
(Fig. 8) was calculated using eight populations (including
216 specimens) from south of the glacial terminus, representing' presumed glacial réfugia stocks, i.e., the
Missouri, Green, Salt, Kentucky, Licking, Little Sandy, Big
Sandy, and Little Kanawha Rivers. These populations also represent a transect across the transition zone from the
Missouri River, with a conspicuously lined population
(=lineolatum), to the Little Kanawha River, with an unlined population (=flabellare).
This discriminant function was then utilized to classify 156 specimens from five samples located north of the glacial terminus ("descendant"), i.e., the Illinois, 31
24 38 32
CANONICAL VARIATE 1
Fig. 8. Canonical variate analysis of "ancestral" populations, with "descendant* populations projected onto this discriminant space. 36
Wabash, Miami, Scioto, and Muskingum Rivers, to the a priori southern ("ancestral") groups. This analysis assumes that the "descendant" populations north of the glacial boundary will be most similar (ordinate closest) to their "ancestral" populations south of the glacial terminus.
The "ancestral" samples generally ordinate along CV 1 in an east-west multivariate dine, extending from the Green to the Little Kanawha River. However, the Missouri and Big
Sandy River samples do not ordinate along CV 1 in their respective geographic positions. The first two canonical variâtes explain 70.0% of the total variation, 42.6% and
27.4% respectively. Characters most influential in determining this multivariate dine, ranked in decreasing order of importance to CV 1, are DlL, AN, D2L, Pi, and SAL
(Table 5). The western samples tend to have longer dorsal fin bases, fewer anal rays, more pectoral fin rays, and fewer scales above the lateral line.
The Kentucky and Salt River samples are displaced in a negative direction along CV 2 (Fig. 8), indicating unique patterns of variation. Characters most influential in determining this dimension are IL and PL (Table 5). The
Kentucky and Salt River samples have the lowest mean number of pored lateral line scales of all populations surveyed
(x=22.5 and 21.9 respectively). They also tend to have Table 5. Standardized coefficients for variables used in the discriminant analysis of "ancestral" samples from south of the glacial terminus.
DIL .565 -.105 SAL -.328 -.419 PI .371 -.344 PL -.199 .456 low .222 -.331 CPD .174 .259 IL .049 -.486 D2 .233 -.072 D2L , .429 .401 AN -.483 -.084 SPC -.172 .070 BD .218 .126 SBL -.019 -.374
% Variance explained 42.6 27.4 Total =70.0 38 greater isthmus lengths, and more scales above and below the lateral line series.
The accuracy of identification, as determined by the jackknifed classification procedure (Table 6) of samples from south of the glacial terminus, varies from a low for the Salt River specimens (33.3%) to a high for the Licking
River samples (87.9%). The misidentified specimens are generally classified into neighboring (adjacent) drainages inversely proportionate to the geographic distance between drainages. This result would be expected if there were gene flow between adjacent populations; essentially in accord with an isolation by distance model (Endler, 1977).
In contrast, the "descendant" samples from north of the glacial boundary generally ordinate in an intermediate position along the multivariate dine paralleling CV 1 (Fig.
8), excluding the Illinois River samples, which have a strong affinity with the Missouri River samples. The northern samples are generally not juxtaposed with their nearest geographic neighbors from south of the glacial terminus (as determined by drainage confluence or geographic distance), but are clustered together in an intermediate position, with specimens often incorrectly classified (Table
6) to geographically distant populations.
The unknowns ("descendant" demes) may have undergone differentiation after dispersal from their ancestral stocks. Table 6. Jackknifed classification table for "ancestral" and "descendant" samples. The DPA was defined utilizing "ancestral" samples only.
Group % Correct Miss. Green Salt Kent Lick LSand BSand Lkc
"Ancestral" Samples 81.5 22 5 0 0 0 0 0 0 78.1 4 25 1 2 0 0 0 0 Salt 33.3 0 1 5 7 1 1 1 0 Kent 82.8 1 0 3 48 3 2 1 0 Lick 87.9 0 1 0 1 29 1 1 0 LSand 55.6 0 1 0 1 1 10 1 4 BSand 72.7 0 0 0 0 1 1 8 2 LKana 63.6 0 0 0 2 4 1 1 14
"Descendant" Samples Illi 0.0 8 2 0 0 0 0 0 2 Wabas 0.0 1 6 7 2 7 2 0 4 Miami 0.0 2 1 0 9 6 1 0 9 Sciot 0.0 18 11 1 6 9 8 3 22 Muski 0.0 4 8 1 4 4 8 0 9
Total 74.5 therefore not aligning with their nearest neighbors. This explanation is supported by the multidimensional ordination of the population centroids in the initial DFA (Fig. 4).
Another possibilty is the unknowns are intermediates due to post-glacial mixing of several stocks from different glacial réfugia; an alternative also supported by Fig.4. The multivariate similarity of Illinois and Missouri River samples implies an ancestor-descendent relationship, thereby supporting a hypothesis of fantails surviving glaciation in an Ozark Upland refuge, with postglacial dispersal northward.
An analysis of the sum of the standard deviations for the most influential characters defining the east-west multivariate dine (Table 7), indicates the intermediate populations,• both north and south of the glacial terminus, are not more variable.
If the "descendant" populations are a mixture of several ancestral stocks, then any increased variation of the descendants due to introgression could have been dampened by subsequent selection. Endler (1977), using computer simulation models, found that unless a zone of intergradation was observed within a few hundred generations after secondary contact, it would be impossible to distinguish'secondary from primary intergradation. Assuming a two to three year average generatiçn time for E. Table 7 . Standard deviations for characters most influential in discriminant function analysis between E. flabellare and E. f. lineolatum.
"Ancestral" Populations DIL SAL PI lOW CPD D2 D2L AN SPC BD Sum
Misso 0.03 0.56 0.52 0.03 0.03 0.67 0.02 0.42 1.11 0.03 3.39 0.02 0.66 0.43 0.04 0.02 0.57 0.02 0.51 0.83 0.02 3.12 0.02 0.60 0.46 0.05 0.03 0.51 0.01 0.46 1.80 0.03 3.97 Kent 0.03 0.67 0.50 0.05 0.03 0.71 0.02 0.57 0.87 0.02 3.47 Lick 0.03 0.52 0.43 0.03 0.02 0.35 0.02 0.38 0.81 0.02 2.61 0.04 0.57 0.42 0.03 0.02 0.47 0.02 0.50 0.92 0.02 3.01 0.02 0.40 0.30 0.02 0.04 0.75 0.02 0.65 0.67 0.03 2.90 LKana 0.03 0.52 0.48 0.03 0.03 0.49 0.02 0.61 1.37 0.02 3.60
"Decendant " Populations
Illin 0.02 0.29 0.49 0.02 0.01 0.67 0.02 0.43 0.67 0.02 2.64 0.Ô3 0.53 0.51 0.03 0.02 0.58 0.02 0.46 0.78 0.02 2.98 0.03 0.96 0.57 0.03 0.03 0.74 0.02 0.59 1.17 0.03 4.17 0.03 0.57 0.45 0.07 0.04 0.63 0.02 0.49 1.01 0.04 3.35 Muski 0.03 0.68 0.39 0.03 0.02 0.53 0.02 0.58 1.03 0.03 3.34
Reference average SD == 3.49 (Misso + LKana) Intermediate average SD = 3.18 (Green to BSand) Intermediate average SD = 3.29 (Illi to Muski) 42 flabellare (Lutterbie, 1979), and 11,000 years since déglaciation, the three to five thousand possible generations of fantails would presumably have ample opportunity to experience the effects of selection and gene
Taxonomic Conclusions: Interior Group
The series of discriminant function analyses, on the
Interior populations of fantails, indicates the possibility of two recognizable forms. A widespread form, referable to
E. f . flabellare, is distributed from the
Mohawk-Hudson and upper Susquehanna River drainages through much of the Ohio, Mississippi, and Missouri Rivers, and less extensively through the Arkansas River and Great Lakes drainages. The uninterrupted multivariate dine, across the reported transition zone separating populations heretofore recognized as E. f. flabellare and E. f. lineolatum, is regarded as evidence supporting the synonymization of E. lineolatum with the nominate
Another possibly distinct form is suggested by the ordination of the population centroids in the canonical variate space and by the UPGMA clustering procedure. This second form is distributed in the White River basin of the 43
Ozark uplands in southern Missouri and northern Arkansas, and the lower Tennessee River basin.
When compared to adjacent populations, the
Ozark-Tennessee assemblage (White and Lower Tennessee samples) is accurately identified 92% of the time, and possibly deserves taxonomic recognition (as suggested by
Pflieger, 1971). However, the taxonomic rank of this form awaits further analysis, including examination of additional material from within the Ozark region and lower Tennessee and Cumberland Rivers. Drs. W. Pflieger and H. Robison
(1982, personal communication) have indicated their intentions to study this and related forms in the Ozark r e g i o n .
Discriminant Function Analysis of Montane-East Slope Group
Character variation within the thirteen Montane-East
Slope samples (Fig. 5) was analyzed via a separate step-wise discriminant function analysis. In the discriminant space
(Fig. 9), the 13 original samples ordinate into three major clusters. The discrimination is primarily along the first two canonical variâtes accounting for 64.1% of the total dispersion, 36.8% and 27.3% respectively. The stepwise procedure includes 16 of the original variables (Table 8).
The three clusters include: a Montane assemblage, i.e., the
Cheat, Upper Tennessee, and New River (northern and Fig, 9, Canonical variate analysis of the thirteen Montane-East Slope samples. 10 13 1 1 I 12 < -1 z o
-3
-3 -2 -1 1 2 3 CANONICAL VARIATE 1
Fig. 9. Table 8. Standardized coefficients of variation for 13 Montane-East Slope samples.
Character CV 1 CV 2
LB .130 .748 Dl -.451 + .175 PL -.321 -.459 ED .002 -.352 AN -.250 -.243 LL -.148 .419 SPC -.399 .102 DIH -.307 .253 VSQ -.223 .265 D2L .284 .155 HW .007 -.474 IL .216 .466 CPD ' .319 .069 HL -.184 -.023 PIL . 146 .041 PI -.296 -.007
% Variance explained 36.8 27.3 47 southern) samples; a Mid-Atlantic Slope assemblage, i.e., the Lower Susquehanna, Potomac, James, Roanoke (western and eastern), Dan, and Neuse River samples; and a Southern
Atlantic Slope group, including the Catawba and Pee Dee
River samples.
The first canonical axis (Fig. 9) separates the Montane samples from the Mid- and Southern Atlantic Slope groups and ordinates an approximate north-south multivariate dine in the Mid-Atlantic Slope samples. Characters contributing most to this dimension, in decreasing order of importance, are
Dl, SPC, PL, CPD, and DlH.
The second canonical axis (Fig. 9) differentiates the
Catawba and Pee Dee River samples, i.e., the Southern
Atlantic Slope group, from the remaining Montane-East Slope samples. Characters contributing most to this dimension, in decreasing order of importance, are LB, HW, IL, PL, LL, and
ED (Table 8).
When the Montane-East Slope samples are pooled into three groups (Montane, Mid-Atlantic Slope, and Southern
Atlantic Slope), based on their ordination in Figure 9, the pooled jackknifed classification matrix (Table 9) indicates a correct identification of unknown specimens 95.0% of the time (96.2%, 93.9%, and 94.9%, respectively). The three forms are allopatric and separated by drainage divides.
However, individual forms are distributed across such Table 9. Jackknifed classification matrix of thirteen Montane-East Slope samples pooled into 3 groups: Montane, Mid-Atlantic and Southern Atlantic.
Group Percent Number of specimens classified into Correct groups
Montane Mid-Atlantic Southern Atlantic
Montane 94.9 111 6 0
Mid-Atlantic 96.2 5 129 0
Southern 93.9 1 2 46 Atlantic 95.0 barriers, e.g., the Montane group is found in the headwaters of the Tennessee, New, and Monongahela Rivers.
Taxonomic Conclusions: Montane East Slope Group
The discriminant function analysis (Fig. 9) accurately distinguishes three distinct groups within the Montane-East
Slope samples with 95% accuracy. According to Mayr
(1969:210) a subspecies is "an aggregate of phenotypically similar populations of a species inhabiting a geographic subdivision of the range of a species and differing taxonomically from other populations of the species." Mayr
(1970) defined differing taxonomically as differing by sufficient diagnostic morphologic characters. Simpson
(1961:175) suggested the usage of the "75% rule"; where 75% of individuals of adjacent subspecies would be unequivocally determinable. Hubbs and Hubbs (1953) and Pimentai (1959), in an attempt to be even more stringent, chose an 84% separation level as adequate for subspecific recognition.
Bailey and Smith (1981:1557) felt subspecies "are usually allopatric or contiguous populations that are morphologically distinct at about the level of 95%."
The levels of accuracy (95%) of identification of the three eastern groups (Montane, Mid-, and Southern Atlantic
Slope), as defined by the discriminant function analysis plus their allopatric distributions,*are consistent with the subspecies concept visualized by Mayr (1970), Simpson
(1961), Hubbs and Hubbs (1953), and Bailey and Smith (1981).
The Mid-Atlantic Slope group is referable to E. f .
humeralis (Girard). Populations inhabiting the
Catawba-Saluda and Pee Dee Rivers (Southern Atlantic Slope
Group) are referable to E. f. brevispina (Coker). The
Montane fantail populations inhabiting the upper Tennessee,
New, and Shavers Fork Cheat Rivers are considered a new
subspecies here called E. f . robustum.
. Discriminant Function Analysis Of All 1,114 Specimens In
Five Groups: Ozark-Tennessee, E. f. flabellare,
E. f. r o b u s t u m , E. f. h u m e r a l i s , and E. f. brevispina
A final discriminant analysis was performed which
utilized 28 variables (excluding LINES and LB) on 1,114
specimens lumped into five groups; Ozark-Tennessee, E.
f . f l a b e l l a r e , E. f . r o b u s t u m , E. f.
humeralis, and E. f. brevispina. Lumping was in
accordance to canonical variate ordinations in previous
a n a l y s e s .
The overall test of the hypothesis of no geographic
variation was rejected (p<.001 for Wilks lambda; F= 39.5;
DF= 84, 4^03). The first three canonical variâtes extracted
from the variance-covariance matrix accounted for 94.0% of 51
the dispersion, 57.3%, 27.5%, and 9.2% respectively (Fig.
10). The relative influence of the 21 characters, chosen by
the stepwise procedure to the three canonical variâtes, is
summarized in Table 10.
The ordination of the five sample centroids in the
three dimensional discriminant space is very similar to the
ordination in the initial discriminant analysis (Fig. 4),
which utilized 50 groups. The Interior (Ozark-Tennessee and
flabellare) forms are differentiated primarily along CV 1
from the three Montane-East Slope forms (robustum,
humeralis, and brevispina). Characters most influential
to this dimension, in decreasing order of importance, are
PL, DIL, LL, HL, and PlL (Table 10).
Canonical variate 2 differentiates robustum from
humeralis and brevispina, and the Ozark-Tennessee form from flabellare. The ordination of the sample means in the
second dimension implies parallel trends of character
co-variation in Montane-East Slope and Interior samples. The
Ozark-Tennessee and humeralis forms both ordinate in a
similar direction along CV 2, relative to the ordination of robustum and flabellare. Characters most influential to this dimension, in decreasing order of importance, are Dl,
IL, DlL, HL, and CPD (Table 10). The third dimension separates' brevispina from humeralis. Characters most Fig. 10. Three dimensional ordination of canonical analysis of the five forms recognized in this study. Darkened lines denote one standard deviation. Not drawn in perspective. *01
2 AO L AO Table 10. Standardized canonical variate coefficients for variables used in the discriminant analysis of 1,114 specimens of E. flabellare in five groups.
Character CV 1 CV 2 CV 3
LL 0.50 -0.162 -0.144 PL -0.690 -0.285 0.240 Dl -0.161 -0.827 0.605 D2 -0.035 0.30 0.550 DIL 0.532 0.426 -0.617 ED -0.215 -0.203 -0.106 AN 0.079 -0.159 0.055 CPD -0.233 0.342 0.091 DIH 0.017 -0.244 0.258 PI 0.067 -0.109 0.199 IL 0.188 0.428 0.314 HW -0.232 -0.239 -0.149 VSQ 0.260 0.060 -0.10 PIL -0.339 0.001 -0.228 AL 0.180 -0.088 -0.260 D2L ' 0.250 0.258 0.187 HL 0.453 -0.369 -0.230 low -0.072 -0.213 0.278 POL -0.133 0.072 0.298 GW -0.030 -0.054 0.403 SAL 0.190 -0.040 -0.058
% Variance explained 57.3 27.5 9.4 Total = 94.0% influential in this dimension are related to the spinous dorsal fin (dorsal spines and DlH).
The jackknifed classification matrix (Table 11) indicates correct identification of samples into five a priori groups, 92.1% of the time. This matrix also indicates high levels of accuracy in identification with respect to Interior versus Montane-East Slope groups, 94.2% of the time (97.4% and 91.0% respectively).
Principal Component Analysis:
E. f . flabellare and E. rob u s turn
A principal component analysis (Fig. 11), utilizing 53 specimens, further demonstrates the differentiation of E. f . flabellare from E. f . robustum. Etheostoma f . flabellhre is represented by 23 specimens from three drainages: Youghiogheny River (11), Monongahela River (5), and Little Kanawha River (7). Etheostoma f. robustum is represented by 27 specimens from two drainages : New
River(19) and Cheat River (8). Three specimens of the presumed intergrade (Coal River) were also included. The first dimension (PC 1) includes 54.7% of the variation and is interpreted as. a general size vector with little taxonomic value (Humphries et ^ . , 1981). PC 2, representing 11.6% of the variation, separates the two forms with no overlap. The coefficients.(Table 12) indicate the Table 11. Jackknifed classification matrix of all 1,114 specimens pooled into five groups.
Group Percent Number of specimens classified into group Correct Ozark flab. robu. hume. brev., Total
Ozark 78.8 41 8 1 2 0 52 flabellare 96.5 9 735 5 13 0 762 robustum 83.8 0 16 98 1 2 117 humeralis 85.8 0 11 5 114 4 134 brevispina 83.7 0 0 2 9 38 49
Total 92.1 1,114 PC I I
• /
PC I
Fig. 11. Principal component analysis of E. f. flabellare and E. f. robustum. E^. f. flabellare = black dofs; E. f, robustum = open squares (üreenbrier and Gauley River specimens) and open circles (Shavers Fork Cheat River specimens). Intergrades from Coal River * black stars. Table 12. Factor loadings for principal component analysis of E^. flabellare and E. f_. robustum. Mensural variables were log transformed.
Factor 1 Factor 2
0.983 0.0 LPDL 0.978 0.0 LHL 0.965 0.0 LAL 0.950 0.0 LD2L 0.949 0.0 LCPD 0.934 0.0 LIL 0.918 0.0 LPOL 0.893 0.0 0.888 0.285 0.884 0.0 0.875 0.0 LGW 0.870 0.0 LIOW 0.863 0.0 LPIL 0.861 0.0 LP2L 0.821 0.287 SPC 0.797 0.0 LDIL 0.680 -0.418 PL 0.0 0.852 PI 0.0 0.723 LB 0.0 -0.700 D2 0.0 -0.604 59
relative contribution of each character to the values of the
principal components. Large coefficients, regardless of
sign, indicate a greater contribution to numerical value for
the principal component with which the character is
associated. Characters most heavily loading to the second
dimension are PL, Pi, LB, and D2 (Table 12). Etheostoma
f . robustum generally has more pored lateral line scales
and pectoral rays, but less lateral bars and dorsal soft
rays than E. f . flabellare. The clustering of two
distinct groups substantiates discriminant function analysis
.recognition of E. f . robustum as a taxon distinct from
E. f. flabellare. Coal River specimens ordinate within
the cluster representing robustum, contrary to their
ordination in the DFA. PART II: PIGMENTATION, MERISTIC, AND
MORPHOMETRIC (RATIO) VARIATION
Pigmentation Variation
Lateral Bars
Typically, 9-14 lateral bars are readily apparent on specimens collected east of the Wabash River, Indiana, and the Licking River, Kentucky. The barring pattern is less pronounced or absent in specimens collected west of these drainages; usually it is difficult to discern, being masked by pronounc.ed dark horizontal lines. In any given population, dark vertical bars are most evident in males defending territories. Seifert (1963) noted, while studying territorial behavior (in aquaria) of lined specimens of Iowa fantails (=lineolatum according to Seifert), that males would develop dark vertical bars when left undisturbed.
Pflieger (1971) noted specimens of the barred form inhabiting the headwaters of the Current and Black Rivers of
Missouri. He referred these specimens (P-64-123) to the nominate form E. f . flabellare (personal communication, 1982), presumably due to their lack of
60 conspicuous lateral lines and obvious barring. This southeastern Missouri population is disjunct from the main range of the barred form, usually found several hundred kilometers to the northeast, and surrounded by populations of the lined fantail darter. The lower Current River is inhabited by lined fantails, which Pflieger (1971) considerd either lineolatum or possible intergrades, lineolatum x flabellare (Pflieger, 1982; personal communication).
The reliability of dark lateral bars as a taxonomic character for the discrimination of E. f . flabellare from E. f . lineolatum is questionable, since it is a labile character. The intensity of the bars varies seasonally, sexually, and ontogenetically, as well as geographically. The character also appears in disjunct populations (e.g.. Upper Current and Licking Rivers). This character is presumably primitive within the subgenus
Catonotus, based on out group comparison, as most members share dark lateral bars. Their occurrence may be due to retention of an ancestral character state, as evidenced by their disjunct geographic distribution. The presence of dark lateral bars is not considered a useful taxonomic character within the western samples of E. flabellare.
Lateral bar number is also highly variable within the
Montane-East Slope group (Fig. 12). Within Atlantic slope drainages there are two barring patterns in allopatric LATERAL BARS
Choal
N. New
S. Now U. Tennessee
L. Susquehanna
Potomac Jamas
W. Roanoke
E. Roanoke
Dan
N buso
Pee Dee
Catawba
Fig. 12. Variation in lateral bars. The diagrams indicate the mean (center point) , two standard errors of the mean (black rectangle), one standard deviation on either side of the mean (outer limits of open rectangle), and sample range (basal line). 63
populations (Fig. 12). The first includes specimens from the
Catawaba and Pee Dee basins (E. f . brevispina), which
have reduced numbers of LB, x=5.7 and 6.0 respectively [see
Page (1983, Plate 33B) for photograph]. The second includes
populations north of the Roanoke drainage (E. f .
humeralis), which average greater than 9.6 bars. The Neuse
River specimens show intermediacy in bar number; however,
this may be in error. Lateral bars in the Neuse specimens
are sometimes reduced to small blotches along the lateral
line series and are difficult to discern; thereby causing
• subjectivity in counting [see Page (1983, 330 for a
photograph of a similarly marked specimen from the Cape Fear
drainage. North Carolina].
Within E. f . robustum, lateral bars are
geographically variable (Fig. 13). In the New River basin,
there is a north-south clinal decrease in LB (Fig. 13), with
the steepest portion of the dine occurring in Giles and
Montgomery Counties, Virginia.
Lines
Agassiz (1854:305), in the original description of what
is now considered E. f. l i n e o l a t u m , noted specimens
with "...close narrow interrupted black longitudinal lines."
These specimens were collected in tributaries of the
Mississippi River, near Quincy, Il].inois. The presence of L A T E R A L BARS
County
P o c a h o n t a s Nicholas
Montgomery
Wythe
Grayson
Fig. 13, Variation in lateral bar number within the New River drainage. 65
"conspicuous rows of spots or dashes" (=LINES), has subsequently become the key diagnostic character for recognition of lineolatum (Hubbs and Lagler, 1958:106).
Figure 14 illustrates geographic variation of LINES in selected populations of fantails. The western populations of fantails, from drainages tributary to the Mississippi and
Missouri Rivers, are all strongly lined, e.g., Missouri
River (x=14.2), Arkansas River (x=12.2), Illinois River
(x=13.7), and Wisconsin River (x=13.6). The unlined population, inhabiting headwaters of the Current and Black
Rivers in southeastern Missouri, is the exception, as noted by Pflieger (1971).
There is a distinct gap in average number of LINES between samples from tributaries of the Mississippi River
(Fig. 15), e.g., Missouri River (14.2) and Green River
(10.7). This may be due to restriction of gene flow across the Mississippi embayment, an adverse environment for small montane stream dwelling fishes.
In northern tributaries of the lower Ohio River, which drain Wisconsinan glacial terrain, there is a rapid west to east clinal decrease in LINES (Fig. 15), e.g., Wabash River
(x = 7 .0), Miami River (x=5.0), and Scioto River ( x = 2 .3).
Wabash River specimens have been considered intergrades
(lineolatum X flabellare) by Hubbs and Lagler (1941), and this interpretation was followed by Gerking (1945). Fig. 14. Geographic variation in mean percent of scale rows which are lined. Solid black line approximates the Wisconsin glacial boundary. Fig, 15. Variation in lines. Samples are arranged west to east in three groups, representing transects: south of the Wisconsin glacial terminus, north of the glacial terminus, and across the Great Lakes. Kentucky
Guyandot
Fig. 15. 69
Trautman (1957:601) noted that fantails from western Ohio prairie streams "... had rather well-developed lengthwise rows of dusky spots along their sides, suggesting possible striped fantail darter (E. flabellare lineolatum) influence...," while few specimens from east of the flushing escarpment "...contained spots and the longitudinal rows of spots were entirely absent." A similar transition zone (Fig. 15), where a moderately lined form changes to an unlined form, occurs across the Great Lakes, primarily in southern tributaries to Lake Erie, e.g.. Western Lake Erie
(x=8.2) and Central Lake Erie (x=4.2). Samples (CU 6642) from Cattaraugus Creek, a tributary of Lake Erie in western
New York, show no indication of LINES.
In the Ohio River basin, south of the Wisconsinan glacial border (Fig. 16), there is an upstream clinal reduction in LINES from the Green River (x=10.7) to the
Kentucky River (x=8.2). Concordant with the clinal reduction in the number of LINES, there is a clinal reduction in LINE intensity. For example, a Kentucky River specimen is typically paler than a comparably lined Green River specimen. Drainages east of the Kentucky River and south of the Ohio River are inhabited by fantails which lack or have few pale LINES. An exception is populations in the montane portion of'the Guyandot, Coal, and Elk Rivers, West Virginia
(Fig. 15), which are conspicuosly lined. The Guyandot River DRAINAGE N O 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7
Missou
N.Fk. Whit
L. Cumberla
Kentucky
W abash
Muskingum
Fig, 16. Variation in lines in selected populations. Samples are generally arranged in a west to east transect. 71
population has been referred to as lineolatum by Jenkins
et al. (1972), presumably due to conspicuous LINES.
Although samples from headwaters of the Guyandot River
(Wyoming County, West Virginia) are lined, specimens from
the Mud River (a lower tributary to the Guyandot River) are
unlined. There are no known geographic barriers between
these two populations, which could impede gene flow.
Geographically intermediate populations were not found either during field surveys in 1978 [LTM, 1978 field notes] or in museum collections. The Guyandot River basin is being
commercially mined for coal. Due to this activity, many of the tributaries which had suitable habitat for fantails are
chemically degraded and depauperate of fish.
Cope (1868:213) noted lined specimens of E. flabellare (as Poecilichthys flabellatus), with
"...the centre of each scale with a dark line forming together numerous longitudinal striae...," inhabited Walker
Creek, a tributary to the New River adjacent to the Guyandot
River basin. Specimens from Little Walker Creek (CU 42137),
a tributary to Walker Creek, are also lined.
Clay (1975:339) noted specimens from the Levisa Fork
Big Sandy River drainage showed "...a stronger lineolatum
tendancy." Clay does not mention which characters he
examined; presumably, he was referring to the presence of
lines. Specimens from the adjacent Tug Fork Big Sandy River 72 drainage (Fig. 15) tend to have more lines (x=1.8), in comparison to specimens from drainages to the immediate west; however, the lines are faint.
A smooth clinal decrease in LINES (Fig. 15) extends approximately 650 km, from the moderately lined Lower
Cumberland River population to an unlined Muskingum River population, including the intervening Green, Salt, Kentucky,
Miami, and Scioto River drainages. The steepest portion of the dine occurs in the Miami and Scioto River drainages.
According to Endler (1977), this would be the widest known dine width for a freshwater fish. Endler maintained that such broad gentle dines may be the result of a combination of heterozygous advantage and selection gradients.
In order to evaluate the variability of LINES, between either lined or unlined populations versus intermediate samples, a non-parametrie Sign Test was calculated for matched pairs, testing the difference between the average standard deviation in LINES of a reference sample versus the standard deviations of intermediate populations. The intermediate populations (both north and south of the glacial terminus) have significantly greater standard deviations in LINES (P=.05). Missouri and Little Kanawha
River specimens were utilized as references in the analysis of sample^ from south of the Wisconsinan glacial terminus, and Illinois and Muskingum River specimens were utilized in the analysis of samples from north"of the terminus. 73
The intermediate samples for these tests are those which are considered moderately lined, i.e., the Lower
Cumberland, Green, Salt, and Kentucky Rivers from south of the glacial terminus. Wabash, Miami, Scioto, Western, and
Central Lake Erie samples were considered as intermediates for the analysis of populations from north of the glacial t e r m i n u s .
Populations with the greatest standard deviations in
LINES were found in the Miami, Scioto, and Central Lake Erie drainages (the areas which encompass the steepest portion of the dine), and the Arkansas River. Within the Scioto River basin of central Ohio, specimens from unglaciated plateau regions typically showed no or few LINES, e.g.. Salt Creek,
Hocking County, OSUM 7215 (x=0.2, N=15). However, samples from the g [facial till plain drainages often exhibited faint
LINES, e.g.. Big Darby Creek, Pickaway County, OSUM 6224
(x=6.8, N=6). There are no known barriers to gene flow within this basin, indicating possible environmental influences on pigmentation or ecophenotypes.
The reliability of LINES as a taxonomic character is also suspect. The lines are formed by dots or horizontal bars on individual scales. The intensity of these dots or horizontal bars varies; thereby causing subjective decisions 74 during counting, particularly in populations which have faint or reduced LINES. A reduction in LINES is typically accomplished by gradual fading of the ventralmost LINE.
Moderately lined (x= 6-8) populations are the most problematic in accurately counting, and also show the greatest standard deviations in LINES.
The recognition of E. f. lineolatum (Agassiz), based solely on the presence of conspicuous rows of spots or dashes, is problematic. Populations west of the Mississippi
River are conspicuously lined, but a decreasing dine in
LINES exists from the Green and Wabash Rivers eastward. Such geographic variation is often an adaptive response to regional climatic conditions (Mayr, 1969), and any taxonomic division along such a dine would be subjective (Endler,
1977). This gentle dine could be interpreted as a zone of intergradation (primary or secondary) between flabellare and lineolatum. But so defined (lower Cumberland to the
Muskingum River; ca 650 km), it includes an area as wide as the geographic range of either proposed parental. This interpretation also presumes an adequate definition for flabellare and lineolatum. If LINES are used as the diagnostic character for the recognition of lineolatum, then the seemingly mosaic occurrence of LINES (Fig. 15) would requ^ire recognition of a polytopic subspecies, i.e., a geographically heterogeneous subspecies. There are no 75
distinct morphologie disjunctions expressed by the canonical
variate ordinations of populations, which would coincide with the geographic variation in LINES. However, there are morphologically dissimilar populations, as indicated by
their canonical variate ordinations, which share conspicuous
LINES (e.g., North Fork White and Guyandot Rivers). The subjectivity in counting LINES is also a concern.
In light of the taxonomic weakness of the pigmentation character LINES, and with respect to the multivariate dine
joining lined and unlined populations, E. jE. lineolatum (Agassiz) should be suppressed and held in synonymy with E. f. flabellare Rafinesque.
Pigmentation of the Ozark-Tennessee Group
The White and Lower Tennessee River samples, along with
the Interior samples, which have been traditionally identified as lineolatum (primarily populations west of the Green River), ordinate in the negative region of CVl
(Fig. 4), and exhibit conspicuous horizontal dark lines. An
exception is the Upper Current and Black River samples.
Conversely, specimens from the White and Lower Tennessee
Rivers often exhibit dark vertical bars (x=10.2), lacking in
Lower Current River samples. This is in contrast to the
Arkansas Cx=1.6) and Missouri (x=5.2) River samples, which generally lack these bars or have them reduced to a series 76
of small blotches along the lateral line series. Specimens
collected (OSUM 42037) in cool spring tributaries of the
North Fork White River (Pine Creek, Ozark County, Missouri) are strikingly colored, with dark bars and lines, dark
spotting in the soft dorsal, and darkened interradiais on the distal one third of the pelvic and pectoral fins.
MERISTIC VARIATION
Scalation
A general scalation pattern is evident in the following characters; LL, SAL, SBL, SACP, TRS, and VSQ (Table 13).
Etheostoma f . flabellare has the greatest mean number of scales in all scalation characters and E. f. brevispina has the least. The most influential characters in the discriminant function analysis differentiating the five groups (Fig. 10) are LL and PL.
Lateral Line Scales
Etheostoma f . flabellare has the greatest mean number of lateral line scales (Table 13) relative to the
Ozark-Tennessee and Montane-East Slope groups. Etheostoma f. robustum (Fig. 17) has the greatest average number of lateral line scales within the Montane-East Slope group.
Within humeralis, there is a general north-south clinal Table 13. Mean differences in meristic and pigmentation characters for subspecies of E^. flabellare. Standard deviations are in parentheses. The last two columns are results of paired t-tests among three forms.
Montane-East Slope Different Group (t-test; p=.05)
Among West plus robust.
Character flabellare robustum humeralis brevispina -Tenna.
N 711 52 117 134 49
LL 52.4 (3.1) 48.8 (3.3) 51.2 (3.2) 48.3 (3.2) 45.6 (3.1) Each PL 28.2 (5.8) 29.6 (3.7) 38.1 (5.1) 32.8 (4.0) 35.1 (4.0) Each robust. PL/LL .54 (0.1) .61 (0.1) .74 (0.1) .68 (0.1) .77 (0.05) SAL 7.7 (0.7) 7.1 (0.8) 7.4 (0.7) 7.0 (0.6) 6.5 (0.5) Each flabell. SBL 8.9 (1.0) 8.6 (1.0) 8.5 (0.8) 8.4 (0.9) 8.0 (0.7) brevisp. SACP 24.0 (1.8) 23.1 (1.5) 22.8 (1.4) 23.2 (1.5) 22.1 (1.4) brevisp. flabell. TRS 31.9 (2.5) 30.0 (2.0) 30.9 (2.0) 29.8 (2.3) 28.0 (1.6) robust. flabell. VSQ 22.3 (2.2) 21.3 (1.6) 21.3 (1.6) 20.3 (2.0) 19.3 (1.7) flabell. D1 7.6 (0.5) 6.8 (0.5) 7.9 (0.5) 7.0 (0.7) 6.8 (0.5) robust. Each D2 13.4 (0.7) 14.6 (0.7) 13.1 (0.6) 13.4 (0.7) 12.5 (0.5) PI 12.5 (0.6) 12.8 (0.4) 12.8 (0.4) 12.2 (0.5) 12.0 (0.7) robust. flabell. AN 8.1 (0.5) 8.2 (0.7) 8.2 (0.5) 7.5 (0.6) 7.8 (0.5) robust. LB NA 10.9 (1.6) 8.5 (1.9) 10.3 (1.6) 6.1 (1.1) 7.1 (1.0) 6.1 (0.7) 6.9 (1.5) 7.0 (1.2) 6.9 (1.2) AN Color blotched PI Spots present present present
(NA = not applicable) LATERAL LINE SCALES (LL)
Fig. 17. Variation in lateral line scales. 79 decrease in LL, extending from the Potomac to the Neuse
River (44.1). However, the Roanoke samples (western and eastern) disrupt this pattern by having greater mean LL than their adjacent neighbors, either to the north or south.
Pored Lateral Line Scales
The mean number of pored lateral line scales is greater in the Montane-East Slope samples, compared to Interior forms (Table 13). The overall pattern of variation in PL, for the Montane-East Slope samples (Fig. 18), is similar to that for LL. Etheostoma f . robustum samples typically have the greatest mean PL (36-42), with humeralis samples consistently ranging from 32 to 34, excepting the Neuse
River samples (x= 26.0), which have the fewest pored lateral line scales.
There is considerable variation in the number of pored lateral line scales (Fig. 19) throughout the range of E. f . flabellare. Most samples overlap in range, with the noted exception of the Coal, Elk, and Guyandot River populations, which have greater average pored lateral line scales, relative to the adjacent Ohio River basin samples. A general northeast-southwest pattern is evident. The upper
Ohio River basin samples tend to have more pored lateral line scales than samples from the Missouri and Mississippi
Rivers, and lower Ohio tributaries. The Kentucky and Salt PORED LATERAL LINE SCALES (PL)
DRAINAÊE
Cheat N .N ew S. New U. Tennessee
L. Susquehanna Potomac
W. Roanoke E. Roanoke ■ m Pee Dee Catawba
Fig. 18. Variation in pored lateral line scales. Variation in pored lateral line scales in Interior populations of fantails . 82
River fantails typically have the least number of pored
lateral line scales. The Arkansas River populations are atypicalr in comparison to adjacent populations, by having
increased scalation, including pored lateral line counts.
Index of Pored To Total Lateral Line Scales
The ratio of pored lateral line to total lateral line
scales is lowest in the Interior forms, and highest in the
Montane-East Slope forms (Table 13).
Etheostoma robustum and E. f. brevispina
samples have greater means in their ratios of pored to total
lateral line scales, relative to E. f. humeralis.
Within this group, the Neuse River samples have the least
number of lateral line scales being pored (Fig. 20).
Fin Ray Elements
Dorsal Spines
Etheostoma f . flabellare and E. f. robustum
have the greatest mean number of dorsal spines, 7.9 and 7.6
respectively (Table 13). Etheostoma f^. humeralis
usually exhibits fewer dorsal spines (x=7.0), with the
Ozark-Tennessee and brevispina samples typically having ,
the least (x=6.0). Etheostoma f. robustum typically
has the,most number of dorsal spines within the Montane-East
Slope group (Fig. 21), averaging between 7.6 and 8.0. The PL/LL
L. Susquehanna Potomac
Roanoke E. Roanoke
Poe Deo Catawba
Fig. 20, Variation in the ratio of pored lateral line scales to lateral line scales. (D1)
Fig. 21, Variation in dorsal spines. 85 northern humeralis samples consistently average between
7.0 and 7.3 dorsal spines, with a clinal decrease from the
Eastern Roanoke (7.0), through the Dan River (6.5), and ending in the Neuse River (6.1).
Dorsal Soft Rays
The Ozark-Tennessee samples tend to have more soft dorsal fin rays (x=14.6), than E. f. flabellare
(x=13.4), robustum (x=13.1), and humeralis (x=13.4)
(Table 13). Etheostoma f. brevispina generally has the fewest number of dorsal soft rays (x=12.5).
Pectoral Rays
The Ozark-Tennessee, flabellare, and robustum samples average more pectoral fin rays (12.8, 12.5, and
12.8, respectively) than humeralis (12.2) and brevispina
(12.0) (Table 13).
Etheostoma f. robustum samples, in general, average greater than 12.5 pectoral rays, in contrast to
Atlantic Slope subspecies, which average less than 12.5 rays
(Fig. 22). Similar to the pattern of variation in LL,
Roanoke samples tend to have a higher mean of pectoral rays than their adjacqnt drainage neighbors, either to the north or s o u t h . 86
PECTORAL RAYS DRAINAGE 1 4
N. New
L. Susquehanna
Catawba'
Fig, 22. Variation in pectoral rays. Anal Rays
The geographic pattern of variation of anal rays is similar to the pattern of variation in pectoral rays (Table
13). The Ozark-Tennessee, flabellare, and robustum samples average more anal rays (8.2, 8.1, and 8.2, respectively) than humeralis (7.5) and brevispina
(7.8). Within the Atlantic slope drainages, there is a clinal increase (Fig. 23) from the lower Susquehanna and
Potomac River samples (x=7.2) to the Dan and Neuse Rivers
(x=7.9).
MORPHOMETRIC VARIATIONS
Dice-Leraas diagrams (Figs. 24,25,26,27) for Interior populations indicate similar clinal patterns of geographic variation in GW, BD, CPD, and D2L, both north and south of the glacial maximum. Northeastern (upper Ohio basin) samples tend to have narrower gapes (Fig. 24), thinner bodies, caudal peduncles (Figs. 25, 26), and shorter soft dorsal fins (Fig. 27). Within E. f^. humeralis (Fig. 28), the
D2L clinally increases to the south.
Spinous Dorsal Fin Base Length
Table 14 indicates flabellare and robustum, in general, have longer spinous dorsal fins in comparison to DRAINAGE
C h e a t
S.
P o t o m a c
Fig. 23. Variation in anal rays. GW/STDL
0.04 0.05 0.06 0.07 0.08 0.00 0.10 DRAINAGEN
Missouri 2 6
Green 32
Salt 1 5
Kentucky 58
Licking 33
L. Sandy 1 8
B. Sandy 1 1
L. Kanawha 2 2
Illinois 1 2
Wabash 2 9
Miami 28
78
Muskingum 38
Mahoning 1 7
Allegheny 1 9
Fig. 24, Variation.in the ratio of gape width to standard length. Samples are arranged in two transects, south and north of the glacial border, * 32
K entucky
Licking
B. Sandy
Musklngut
Mahoning
Allegheny
Fig. 25. Variation in proportion of body depth to standard length. 91
CPD/STDL
0.10 0.11 0.12 0.13 0.14 0.1 5 0.18 DRAINAGE N
Missouri 26
Green 3 2
Salt 1 6
Kentucky 58
Licking 33
L. Sandy 1 8
B. Sandy 1 1
L. Kanawha 22
Illinois 1 2
Wabash 28
Miami 2 8
Scioto 78
Muskingum 38
Mahoning 1 7
Allegheny 1 9
Fig, 26, Variation in caudal peduncle depth to standard length. D2L/STDL
DRAINAGE
U. Current N.Fk. White Arkansas Missouri Green Salt Kentucky Licking L. Sandy B. Sandy L. Kanawha Illinois Wabash Miami Scioto Muskingum Mahoning Allegheny N. New
Fig. 27. Variation in soft dorsal fin length (expressed in thousandths of standard length), Fig. 28. Variation in proportion of soft dorsal fin length to standard, length. Table 14. Mean differences in proportional measurements of mensural characters for subspecies of E. flabellare Expressed in thousandths of SL (standard deviations are in parentheses).
Montane-East Slope
brevispina
N 52 711 117 134 4S Character XL 210 (18) 201 (13) 193 (15) 196 (11) 195 (12) HL 293 (13) 292 (13) 287 (12) 282 (12) 291 (13) HW 136 (10) 143 (13) 143 (11) 148 (11) 150 (11) BD 178 (10) 175 (14) 177 (11) 183 (11) 178 ( 7) 130 ( 9) 125 (11) 125 (10) 137 ( 9) 132 ( 7) PIL 225 (15) 216 (18) 218 (15) 226 (17) 243 (20) P2L 178 (14) 171 (15) 172 (15) 173 (17) 181 (17) AL 278 (14) 263 (14) 260 (15) 254 (15) 270 (15) DlL 174 (16) 197 (16) 192 (16) 179 (19) 175 (15) DIH 62 (10) 60 ( 9) 62 ( 8) 59 ( 9) 60 ( 8) D2L 369 (16) 335 (18) 315 (19) 338 (18) 323 (15) PRDL 356 (13) 355 (15) 348 (12) 348 (12) 356 (10) PROL 60 ( 6) 64 ( 7) 66 ( 5) 65 ( 5) 64 ( 6) low 45 ( 4) 46 ( 6) 48 ( 5) 47 ( 5) 45 ( 5) ED 57 ( 4) 55 ( 6) 56 ( 4) 54 ( 4) 60 ( 6) GW 61 ( 7) 67 ( 9) 68 ( 9) 70 ( 8) 73 ( 8) STDL (avg.) 37.2 (3.8) 47.4 (6.8) 52.7 (8.1) 48.2 (7.3) 42.9 (7.2) 95 humeralis, brevispina, and the Ozark-Tennessee group.
This longer spinous dorsal fin is presumably due to the addition of dorsal spine elements (see Dorsal Spines).
Spinous Dorsal Height
Etheostoma f. robustum generally has the tallest spinous dorsal fin (Fig. 29). Within this form, the Southern
New River samples have the shortest. Atlantic Slope samples generally average shorter spinous dorsals, relative to robustum (e.g.. Northern New and Cheat Rivers), with the
Neuse River specimens exhibiting the lowest average spinous dorsal fin height. The Western Roanoke samples tend to have the tallest dorsal fins within the Atlantic Slope samples, similar to the Northern New River specimens.
Pectoral Ray Length
The Montane-East Slope forms, i.e., robustum, humeralis, and brevispina, tend to have longer pectoral rays than Interior forms (Table, 14).
Caudal Peduncle Depth
Etheostoma f . flabellare and E. f . robustum generally have narrower caudal peduncles (Tables 14) than
humeralis, brevispina, and the Ozark-Tennessee
specimens. In the Atlantic slope drainages, the caudal D1H/STDL
DRAINAGE
Cheat
N. New 24
S. New 55
U. Tennessee
L. Susquehanna
Potomac 4 0
W. Roanoke
E. Roanoke
Neuse
Pee Dee 33
Catawba
Fig. 29. Variation in proportion of dorsal fin height to standard length. peduncle depth decreases clinally (north to south), starting in the Dan River of the Roanoke drainage (Fig. 30).
Eye Diameter
Eye diameter (Fig. 31) decreases clinally from north to south in E. f. robustum samples. The variation in eye diameter does not follow a clear geographic trend within the
Atlantic Slope samples. Etheostma f. brevispina tends to have a larger eye diameter, with the Catawba River samples having the greatest mean eye diameter of all fantails studied (Table 14).
Head Width
The Ozark-Tennessee form tends to have the narrowest head (Table 14). Etheostoma f. robustum has a moderate head width, while E. f. brevispina and E. f. humeralis have wide heads. There is clinal variation in head width within humeralis, southern populations having slightly greater mean head widths (Fig. 32).
Head Length
Although the means for head length do not vary greatly, it is apparent that humeralis exhibits the smallest average head length (Table 14). CPD/STDL
DRAINAGE
y. Tennessee
L. Susquehanna
Fig. 30 30. Variat Variation in proportion of caudal peduncle depth to standard length. . ED/STDL
04 0.05 0.06 0.07 DRAINAGE N °
Cheat 10
N. New 24
S. New 55
U. Tennessee 2 8
L. Susquehanna 1 2
Potomac 40
James 1 7
W. Roan.oke 1 5
E. Roanoke 1 3
2 1
1 6
Pee Dee 33
Catawba 1 6
Fig, 31. Variation in proportion of eye diameter to standard length. h w / s t d l
DRAINAGE
L. Susquehanna
Potomac
Pee Dee
C a t a w b a ,
Fig. 32. Variation in proportion of head width to standard length. 101
Morphologie Convergence: Southwestern and Atlantic
Slope Groups
Ordination of the five groups in the canonical graph
(Fig. 10) indicates parallel patterns in character variation between Ozark-Tennessee, humeralis, and brevispina
(along CV 2). These three forms tend to average less scales
(including LL, SAL, SBL, SACP, TRS, and VSQ), fewer spines in the dorsal fin, greater body depth, caudal peduncle depth, and pectoral and pelvic fin lengths (Tables 10, 13,
14). Since these three forms are geographically distant and separated by drainage divides, this morphologic similarity is presumably not due to gene flow, but possibly to convergence or retention of primitive character states. TAXONOMIC DISCUSSION
Etheostoma flabellare Rafinesque
Barred Fantail Darter
Diagnosis.— This species is distinguished from other members of Catonotus by a combination of characters considered as suite. Branchiostegal gill membranes are broadly joined, spines of the first dorsal fin are tipped with enlarged fleshy bulbs, there is no distinct submarginal band in the spiny dorsal fin, and no bicolored bar on the
Description.— The infraorbital canal is interrupted with 4 pores anteriorly and 2 pores posteriorly, the supratemporal canal can be complete or incomplete (usually incomplete), and the preoperculomandibular canal usually has
10 pores. The nape, cheek, opercle, breast, and prepectoral areas are usually unsealed; the belly is fully scaled or missing a few scales anteriorly.
102 m S M '
Ozark-Tenna. form ■ E. f. (labellare # E. f. robust urn * E. I. humeralis O E. f. brevispina Q robustum xflabeilare^
Fig, 33, Distribution of E, flabellare subspecies and an Ozark-Tennessee form. Fig. 34, Adult males of Etheostoma flabellare;
A. E. f. flabellare L. Kanawha R. CU 13474 59.5 m m SL.
B. E. f, robustum Ne w R. CU 52224 58.6 mm SL.
C. E. humeralis Roanoke R. CU 20440 60.5 mm SL.
D. E. f. brevispina Catawba R. OSUM 41851 47.8 mm SL. 105 Etheostoma flabellare flabellare Rafinesque
Figure 33, 34A; Tables 13, 14
Etheostoma flabellaris Rafinesque, 1819 (original description). Etheostoma flabellata Rafinesque, 1820:36. Etheostoma fontinalis Rafinesque, 1820:86. Etheostoma linsleyi Storer, 1851. Catonotus lineolatus Agassiz, 1854. Catonotus fasciatus Girard, 1859. Poecilichthys flabellatus Cope, 1868. Catonotus lineolatum Vaillant, 1873. Etheostoma flabellare lineolatum Jordan, 1878. Etheostoma lineolatum (Agassiz) Jordan and Gilbert, 1883. Etheostoma flabellare lineolatum (Agassiz) Jordan and Evermann, 1896. Catonotus flabellaris lineolatus (Agassiz) Hubbs, 1926 Catonotus flabellare lineolatum Jordan, Evermann and Clark, 1930. Etheostoma flabellare lineolatum Bailey and Gosline, 1955.
Diagnosis.— The nominate subspecies can be distinguished from its consubspecifics by its protruding lower jaw (especially prominant in adult males). Usually less than 60% of the lateral line scales are pored, anal and pelvic fins are not completely black in nuptial males, and adult males lack distinct spots on their pectoral fins.
Description.— Lateral line scales 41-63 (50-54); pored lateral line scales 13-45 (24-32); scales above the lateral line 5-10 (7-8); scales below the lateral line 6-14
(8-9); scales around the caudal peduncle 17-35 (23-25); transverse scales 24-43 (30-33); ventral scales 13-29
(21-24); dorsal spines 6-9 (mode 8); soft dorsal rays 11-16
(13); pectoral rays 11-14 (12); anal rays 6-10 (8); vertebrae 33-37. 107
The body is dark brown, tan or olive dorsally, grading
to lighter yellow to tan ventrally; depending on geographic
area, the body has black longitudinal lines, especially in western populations (Fig. 35D), or 10-16 dark brown vertical bars (eastern populations) or combinations of both. Dark brown vermiculations are commonly distributed over the dorsal surface, especially the nape. There are typically 8 dark brown saddles, which may or may not connect to the vertical bars. The spinous dorsal fin is black basally,
transparent medially, and dusky distally, with yellow or gold knobs on the tips of the spines. These knobs are particularly apparent on nuptial males. The caudal and
second dorsal fins are tan to light yellow-orange with black
bands. Pectoral fins are gradations of orange-yellow to
transparent. Pectoral fins of adult males do not have spots
(although females from all populations, and males from the
Middle Cumberland River populations, may show faint spots).
A dark basal band is often noticeable in the pectoral fins
of western populations. Anal and pelvic fins are often milky
white (darkened in nuptial adults, but not totally black).
The anal fin may have dark blotches, primarily in the
posterior interradial membranes. A distinct humeral spot is
present. Pre- and postorbital bars are present; a suborbital
bar (teardrop) is faint or absent. Nuptial males become
dark, especially the head and nape areas which become Fig, 35. Adult males of Etheostoma flahellare:
A, Intergrade CE. f. flahellare x E. f. robustum) Guyandot R. AEL 3 54.3 mm SÏÏ. ~ “ ------
B. E. f. flahellare Middle Cumberland R, OSUM 37800 42,3 mm SL,
C. E, f, flahellare? North Fork White R, OSUM 42037 42,8 mm SL,
D, E, flahellare lined specimen, Missouri R, CU 42234 43,8 mm SL,
110 swollen. The body bars and bands on the soft dorsal and caudal fins become more intense.
Type material.— Rafinesque did not preserve type material for E. flahellare. Call (1899), in his reprint of the original Ichthyoloqia Ohiensis (Rafinesque,
1820), noted E. flabellata (= E. f l a h e l l a r e ) was described from the Falls of the Ohio, near the present site of Louisville, Kentucky. However, in the original description of E. flabellaris (Rafinesque, 1819 ; published in French), the Falls of the Ohio was not mentioned. Collette and Knapp (1967) simply indicated the type locality for E. flahellare as the Ohio River and
Jordan et al. (1930) cite tributaries of the Ohio River.
Fantails from near the Falls of the Ohio (Louisville,
Kentucky) usually exhibit a moderate number of horizontal lines. Collette and Knapp (1967) published a sketch of E. flahellare from Rafinesque's note book (Fig.2), which can be considered an iconotype. In this illustration there is no indication of horizontal lines but there are clearly figured
12 vertical bars. The illustration resembles the upper Ohio
River form, which traditionally has been considered E. flahellare flahellare.
Unlined populations of E. flahellare can be found in all major tributaries of the Ohio River, between
Pittsburgh, Pennsylvania (the origin of Rafinesque's collecting trip) and the confluence of the Scioto River.
Since Rafinesque (1819) did not mention lines or have them illustrated in his rather accurate sketch, and also to stabilize the taxonomic concept of E. flahellare, specimens for neotypic designation should be chosen from an unlined population inhabiting the Ohio River drainage, between the confluence of the Scioto River and Pittsburgh,
Pennsylvania.
Distribution.— E. flahellare flahellare is distributed from the Mohawk-Hudson, upper Susquehanna, and
St. Lawrence River drainages on the east; west through the
Great Lakes basin; south in the interior drainages from
Minnesota to the Arkansas River; and east through the
Missouri and Ohio River basins (excluding the Tennessee and
New River drainages)(Fig. 33).
Etheostoma flahellare humeralis (Girard)
Figure 33, 34C; Tables 13, 14
Oliqocephalus humeralis Girard, 1859. Poecilichthys flabellatus Cope, 1868. Etheostoma flahellare flahellare Ross, 1958. Etheostoma flahellare flahellare Jenkins, Lachner and Schwartz, 1972.
Diagnosis.— This subspecies is distinguished from its consubspecifics by a suite of characters including : terminal 112 to subterminal mouth, black anal and pelvic fins on nuptial males, usually 7 dorsal spines and 12 pectoral rays, a relatively small head and eye, and deep body and caudal p e d u n c l e .
Description.— Lateral line scales 39-56 (usually
46-50), scales above the lateral line 6-8 (7), scales below the lateral line 6-10 (8-9), scales around the caudal peduncle 19-27 (22-24), transverse scales 23-34 (28-32), ventral scales 14-24 (19-22), dorsal spines 5-8 (mode 7), soft dorsal rays 12-15 (13), pectoral rays 11-14 (12), anal rays 6-9 (8), vertebrae 34-36.
Overall coloration is similar to the nominate subspecies, with the following exceptions: lateral bars
(mode=10) often reduced to blotches, especially in northern populations; crescent shaped pigmentation patterns on individual scales often forming a cross-hatched pattern, especially in southern populations; lines absent; spotting in the pectoral fin; and black pigmentation in anal and pelvic fins of nuptial males.
Type material.— The Mid Atlantic Slope form is referable to E. f. humeralis (Girard). Oliqocephalus humeralis was described by Girard (1859) from specimens collected in the James River drainage of Virginia, and subsequently synonymized with E. flahellare by Jordon and Evermann (1896). Type specimens were designated by Collette and Knapp (1967), as follows: Lectotype— USNM 1203
(male, 42 mm SL), James R., Virginia; S.F. Baird, 1850; and
Paralectotypes— USNM 198001 (45, 23-49), removed from USNM
1203; MCZ 24634 (10, 25-54), originally out of USNM 1203, then MCZ 16; MCZ 35975 (2, 22-33), originally out of USNM
1203.
Distribution.— This subspecies is herein limited to the Lower Susquehanna, Potomac, James, Roanoke, and Neuse
Rivers (Fig. .33).
Although specimens were not examined from the Cape Fear
River system. North Carolina, a photograph (Page, 1983:
Plate 330 of a fantail darter from Little Alamance Creek
(Haw River, Cape Fear drainage) appears identical to E. f . humeralis specimens I have photographed from the
Neuse River drainage. The Neuse River samples are the most aberrant within humeralis, and are possibly worthy of separate taxonomic recognition, as implied by Buhan (1966).
Further study, however, including additional material from the Cape Fear drainage, is advisable.
Etheostoma flahellare brevispina (Coker)
Figure 33, 34D, Tables 13, 14
Poecilichthys flabellatus Cope, 1870. Etheostoma flahellare Jordan, 1890. Richia brevispina Coker, 1926 (original description). 114
Richiella brevispina Coker, 1927 (Richia preoccupied). Poecilichthys flabellaris brevispinus (Coker) Myers, 1927. Catonotus flabellaris brevispina Hubbs, 1927. Etheostoma flahellare brevispinna Ross and Cari C O , 1963. Etheostoma flahellare brevispinna Jenkins, Lachner and Schwartz, 1972. Etheostoma flahellare brevispina Page, 1983. Etheostoma flahellare brevispina Kuenhe and Barbour, 1983.
Diagnosis.— This subspecies is distinguished from its
consubspecifics by the following combination of characters :
terminal to subterminal mouth, greater ratio of pored to
total lateral line scales (x= 0.77), black anal and pelvic fins on nuptial males, usually less than 7 lateral bars, usually less than 47 lateral line scales, and a relatively wide head and gape, large eye, and long pectoral fins.
Description.— Lateral line scales 38-53 (usually
43-48), pored lateral line scales 25-43 (32-38), scales above the lateral line 5-7 (6-7), scales below the lateral
line 7-9 (8), scales around the caudal peduncle 19-25
(21-23), transverse scales 25-31 (27-29), ventral scales
15-22 (18-20), dorsal spines 6-8 (mode 7), soft dorsal rays
11-13 (13), pectoral rays 10-13 (12), anal rays 7-9 (8), vertebrae 34.
Overall coloration is similar to the nominate
subspecies with the following exceptions: fewer lateral bars (mode=7); a prominant dark dorsal saddle anterior to the
dorsal fin is present; ground coloration is lighter, thereby
accentuating the dark bars; absence of lines, chevron
pattern indistinct if present; spotting on the pectoral
fins; black pigmentation on anal and pelvic fins of nuptial
Type material.— Coker (1926) described Richia brevispina from specimens collected in Paddys Creek
(Catawba drainage). This form was synonymized with
flahellare (Hubbs, 1927; also see Myers, 1927) and
.subsequently regarded as a subspecies by Ross and Carico
(1963).
Types.— Specimens designated by Coker as types
include: Holotype.— USNM 87411 (male, 36 mm SL), North
Carolina, Burke County, Paddys Creek, just above head of
Paddys Creek Lake, part of artificial Lake James system near
Bridgewater, Catawaba drainage, collected by R. E. Coker,
August 1922, figure in Coker (1926:107); and Paratopotypes—
USNM 87412 (2, 27-36), same data as types.
Distribution.— This distinctive form is herein limited
to the Pee Dee and Catawba River drainages (including the
Broad River system) of North and South Carolina (Fig. 33). 116
Etheostoma flahellare robustum n. ssp.
Figs 33, 34B; Tables 13, 14, 15
Poecilichthys flabellatus Cope, 1870. Poecilichthys flabellatus Cope, 1868. Etheostoma flahellare flahellare Ross and Carico, 1963. Etheostoma flahellare brevispinna Ross and Carico, 1963. Etheostoma flahellare brevispinna Jenkins, Lachner and Schwartz, 1972.
Type material.— Holotype (OSUM 51624), adult male 57.2 mm SL.; coll. by L. McGeehan and W. LeGrande, 10 April 1976.
West Virginia, Pocahontas County, Edray Twp., Stoney Creek, at U.S. 219 bridge, Campbelltown. Allotopotype (OSUM
51624), adult female, 50.2 mm SL. Paratopotypes (OSUM
51624), 7 specimens, 40.2-48.4 mm SL.
Diagnosis.— This subspecies is distinguished from its consubspecifics by the following combination of characters, considered as a suite: a greater ratio of pored to total lateral line scales, terminal to subterminal mouth and declivous head profile, diffuse black pigmentation totally covering pelvic and anal fins of nuptial males, spotting in the pectoral fins, usually more dorsal spines (mode 8); and a relatively large standard length (maximum SL=78.5 mm), tall spinous dorsal fin, and short soft dorsal fin. Table 15. Proportional measurements of Etheostoma flahellare robustum Expressed as thousandths of standard length.
Holotype Allotopotype Paratopotypes OSUM 51624 OSUM 51624 OSUM 51624 Male Females Character N=2 N=5
SL mm 57.2 mm 50.2 mm 45.7-47.1 40.2-48.4 HL 281 289 293-295 280-302 HW 131 137 133-144 123-149 GW 58 68 61-64 55-60 ED 58 68 61-64 55-67 lOW 35 32 36-42 35-47 PROL 63 68 68-70 56-70 PRDL 351 360 363-272 360-371 BD 180 191 171-197 185-205 CPD 133 123 125-129 113-125 PIL 217 215 214-225 207-241 P2L 182 185 173-180 165-202 AL 274 235 254-255 233-257 DIL 203 201 189-197 161-194 DlH 66 76 70-77 74-100 D2L 302 297 293-301 264-320 Description.— Lateral line scales 46-60 (usually 49-53),
pored lateral line scales 21-53 (35-42), scales above the
lateral line 5-9 (7-8), scales below the lateral line 7-11
(8-9), scales around the caudal peduncle 20-26 (22-24),
transverse scales 26-36 (29-32), ventral scales 18-26
(20-22), dorsal spines 6-10 (mode 8), soft dorsal rays 11-15
(13), pectoral rays 12-14 (13), anal rays 7-9 (8), vertebrae
35-37.
Overall pigmentation is similar to the nominate
subspecies, with the following exceptions: lateral bar
number is variable, decreasing from 10 in the Greenbrier
River to 8 in the upper Tennessee R.; lines are absent except on specimens from Walker Creek, Pulaski County,
Virginia; pectoral fins are spotted; pelvic and anal fins are black ori nuptial males.
Etymology.— The name robustum means large and is in reference to this subspecies adult size.
Distribution.— Etheostoma flahellare robustum inhabits the upper Tennessee River drainage (upstream of the confluence of the Little Tennessee River), the New River drainage, and as a localized population in the headwaters of the Shavers Fork Cheat River, a Monongahela-Ohio tributary
(Fig. 33). ZOOGEOGRAPHIC DISCUSSION
Although formal subspecific taxa are not recognized within the Interior group, the interdemic geographic variation (expressed by the canonical variate ordination of population centroids), in conjunction with variation in
LINES, allows zoogeographic interpretations.
According to Bailey and Smith (1981:1555) "...two different stocks of fantails apparently with discrete glacial réfugia in the Mississippi basin have invaded the
Great Lakes basin." A moderately lined population (= E.
f^. flahellare of Bailey and Smith, 1981) entered through
the Wabash-Maumee portal (Ohio River basin to Lake Erie basin), dispersed north to occupy the southern half of the
lower peninsula of Michigan (including Lake Michigan
tributaries), and east through the lower Great Lakes to
Quebec (Bailey and Smith, 1981; Lee et , 1980).
A strongly lined population (lineolatum of Bailey and
Smith, 1981), possibly surviving glaciation in an Ozark
uplands réfugia, dispersed up the Mississippi River and
utilized the Wisconsin-Fox outlet (Greene, 1935; Bailey and
Smith, 1981). It now inhabits western tributaries of Lake
119 Michigan in Wisconsin (Becker, 1976), and the western Upper
Peninsula of Michigan. Greene (1935) noted the occurrence of
fantails above falls in the Chippewa and Wisconsin Rivers,
and suggested the probable use of glacial waters during
their dispersal. The strong affinity (in the canonical
variate graphs), between the Illinois and Wisconsin River
samples with those from the Missouri River, is probably due
to an ancestor-descendant relationship, supporting an Ozark
Upland glacial réfugia hypothesis.
The sharp west-east clinal decrease in LINES (Fig. 15),
across the southern tributaries to Lake Erie, indicates
possible headwater stream transfer of upper Ohio River
fantails into the Great Lakes basin. Two modes of fish
dispersal (including flahellare), from the Allegheny River
(Ohio basin) into the Upper Genesee River (Lake Ontario
drainage), have been postulated. Bailey and Smith (1981)
suggested stream capture, whereas Miller (1968) indicated
glacial lake outlets facilitated transgression of the
divide. The direct transfer of Ohio basin fishes into Lake
Erie (circumnavigating the Niagara Falls barrier) may have
also occurred. There are low headwater divides separating
both the Mahoning (Ohio River basin) and Grand Rivers (Lake
Erie drainage), and Little Valley (Allegheny River drainage)
and Cattaraugus Creeks (Lake Erie drainage). 121
An alternative hypothesis is the clinal decrease in
LINES is associated with an environmental parameter,
although there is no direct supporting evidence.
In the northern Appalachians, stocks of fantails have dispersed eastward across drainage divides, presumably utilizing glacial lake outlets. These glacial lake outlets opened avenues of dispersal from Lake Ontario into the
Mohawk-Hudson and upper Susquehanna River drainages
(Fairchild, 1908). The curious disjunct distribution of two
localized populations of fantails, inhabiting different
headwater drainages within the upper Susquehanna system
(Cooper, 1983), suggests the possibility of two distinct dispersals of fantails into this drainage. Populations
inhabiting the headwaters of the North Branch Susquehanna
River presumably utilized the Canisteo glacial outlet as a dispersal route from the Genesee River (Lake Ontario drainage), as suggested by Ross (1958). A direct faunal
transfer between the Allegheny and Susquehanna drainages was postulated by Lachner and Jenkins (1971) to explain the distribution of Nocomis micropoqon, which is not present
in the Genesee River. Denoncourt et al. (1977)
speculated that the localized population of Etheostoma
blennioides, inhabiting Sinnemahoning Creek (a tributary
of the West Branch Susquehanna River), gained entrance 122
directly to the system from the Allegheny River. However,
they were unsure as to the locality of faunal interchange.
An examination of the Appalachian divide, in the area
of possible interchange, indicates a natural gap at an
elevation of 573 m (1,880 feet) at Keating Summit, Potter
County, Pennsylvania. This gap is approximately 152 m (500
feet) below the average Crestline of the adjacent ridges,
and is anomalous for this region. The divide is generally
high (greater than 640 m or 2,100 feet) and even.
Geologic evidence suggests this gap is a col developed by overflow waters from a glacially impounded Allegheny
River. MacClintock and Apfel (1944) noted a large, but
short-lived, ponding of the Allegheny River took place when
an ice tongue crossed the river at Olean, New York, and
impinged against the south valley wall. Glacially deposited
sands and gravel were found near Birch Run (a southern
tributary of the Allegheny River), at 512 m (1,680 feet)
(MaClintock and Apfel, 1944). Glacial melt water drained
across a saddle (537 m or 1,760 feet) into Birch Run from
the impounded Allegheny River (Muller, 1960). The Allegheny
River valley, upstream of Olean, shows géomorphologie
characters, which are typically associated with a glacially
ponded valley. Clay deposits fill the valley, causing the
river to erode a sinusoidal channel, thereby allowing
numerous oxbows to form. 123
In light of the geologic evidence, I hypothesize a proglacial lake (=Lake Olean) breached a col in the
Appalachian divide, at the present site of Keating Summit,
Pennsylvania, and eroded the gap. This would have allowed the direct transfer of Ohio River basin fishes into the upper Susquehanna system, including : Clinostomus elonqatus, Nocomis micropoqon, Etheostoma blennioides, and Etheostoma flahellare (for distributions see Cooper, 1983).
An alternative explanation for the distribution patterns of these fishes is via bait bucket introductions of
Allegheny River fishes into the upper Susquehanna drainage, as postulated for Etheostoma zonale (Kneib, 1972).
However, the differences in the sample means for E. flahellare and E. blennioides (Denoncourt et al.,
1977) suggest these populations have inhabited the upper
Susquehanna basin for some time; thereby supporting the proglacial lake dispersal hypothesis.
Growl (1980) concluded the last glacial ice sheet probably began to retreat from this area approximately
15,000 years B.P. This represents a maximal estimate of time for the existence of the Sinnemahoning River population of fantails. The Sinnemahoning samples do show a different ordination .relative to the Allegheny River samples. 124
presumably due to their isolation and differing responses to
environmental conditions.
Another example of headwater transfer of fantail stocks
is indicated by the canonical ordination of the Middle
Cumberland River samples (below the Cumberland Falls). These
specimens ordinate nearer to the Kentucky River samples than
to the Lower Cumberland centroid; contrary to what would be
expected for intradrainage demes. This result indicates the
Middle Cumberland River population is a derivative of a
Kentucky River stock, presumably due to headwater transfer via stream capture. A discriminant analysis separates the two Cumberland River populations with 100% accuracy, based on SBL, ED, 01, and PL. The best character for discrimination is SBL. The Middle Cumberland River specimens average more scales below the lateral line (x=9.5) than the
Lower Cumberland River samples (x=7.2). They also have the highest spinous dorsal fins, within the Interior group (Fig.
35B). Faunal transfers, between the Kentucky and Cumberland
River drainages, have been reported by Jenkins et al.
(1972), for Etheostoma sagitta and Etheostoma baileyi, and by Miller (1968) for E. blennioides. In spite of morphologic similarities between these populations, the Middle Cumberland River population has a unique LDH
(lactate dehydrogenase) banding pattern, not shared with the Kentucky or Lower Cumberland populations (Wolfe and Branson,
1979; Wolfe et al., 1979) .
Headwater transfer of fantails from the Missouri River basin into the Arkansas River drainage is also indicated by the distribution patterns of fantails in western Missouri,
Arkansas, and eastern Kansas and Oklahoma (Fig. 33). The distribution pattern of Notropis spilopterus (Lee et al., 1980) is very similar to that of the fantail darter in the same region.
Arkansas River specimens tend to have more lateral line
(Table 16) and pored lateral line scales, relative to other
Ozark populations of fantails. This population is isolated on the periphery of the subspecies range; thus, this atypical pattern may be due to founder effects (Mayr, 1969).
The White I^iver, a lower Arkansas River tributary, is inhabited by the Ozark-Tennessee form.
The presence of localized populations of E. f. humeralis, in a few southern tributaries of the lower
Susquehanna River, indicates a dispersal of humeralis from the adjacent Potomac River. An area of possible interchange exists between Conodoquinet Creek, of the Susquehanna drainage, and Conococheague Creek, of the Potomac drainage.
The headwaters of these two streams interdigitate in a broad limestone 'valley. Table 16. Average differences in meristic characteirs for the Ozark-Tennessee form and adjacent populations. Standard deviations are given in parentheses.
Peculation L.Tenna. N.Fk.White U.White U.Qirrent L.Current Arkansas Missouri N 19 15 7 7 4 38 27 Character D2 14.3 (0.8) 14.9 (0.5) 14.3 (0.5) 14.7 (0) 14.8 (0.5) 14.0 (0.5) 13.7 (0.7) D1 6.9 (0.2) 7.1 (0.3) 6.1 (0.4) 6.2 (0.5) 7.1 (0.4) 7.6 (0.5) 7.9 (0.3) SPC 6.0 (0.8) 6.1 (0.7) 5.9 (0.4) 6.0 (2.4) 6.6 (1.1) 6.7 (1.0) 7.1 (1.1) LL 46.6 (3.7) 50.3 (1.6) 47.8 (1.9) 53.0 (3.2) 50.0 (2.1) 55.2 (2.8) 52.7 (3.0) 127
In this study, the Roanoke River samples were initially divided into eastern and western groups to test possible phenetic relationships between Western Roanoke and New River populations. Ichthyologists have often cited examples of transferals of New River fishes into Atlantic slope drainages (Jenkins et al., 1972). In the canonical variate graph (Fig. 9), the Western Roanoke samples ordinate closer to the Eastern Roanoke and other humeralis populations, than to New River specimens.
The Cheat River population of robustum presumably gained entrance to this system via a stream capture between the Shavers Fork Cheat R. (Monongahela River drainage) and
Greenbrier R. (New River drainage). An exact area of transfer is speculative, but Schwartz (1955) felt a transfer could have occurred via a Blister Swamp-Laurel Creek connection. Fishes believed to have crossed from the
Greenbrier River to the headwaters of the Monongahela River, are; Etheostoma blennioides, Exoqlossum laurae,
Percina oxyrhyncha, Clinostomus funduloides,
Ericymba buccata, and Nocomis platyrhynchus
(Schwartz, 1965; Jenkins et al., 1972; Lachner and
Jenkins, 1971; Wallace, 1973; Hocutt et a 2 . , 1978; and
Hocutt, 1979).
Collections of fantails from Shavers Fork at Cheat
Bridge (CU 32333 and AEL 60) are all good robustum. The 128
Cornell collections were made in 1951, thereby establishing this form in the upper Cheat R. system for at least 33 years. Downstream from the Cheat Bridge is the High Falls of the Cheat, a 5 m high waterfall, which could act as a barrier for upstream movement of fishes (Dan Cincotta-WVDNR,
1982, personal communication). Three specimens (AEL 68), collected at Stuart Recreation Area (downstream of the falls), appear to be typical robustum, although they are all small females. Approximately 16 km downstream, typical flahellare specimens (OSUM 52034, 1982) were collected from the mouth of Haddix Run, a small tributary of the
Shavers Fork, at Forterwood, West Virginia. Although an attempt was made to collect specimens from the main river at
Porterwood, no fantails were found. A snorkling observation of the riffles in the main river indicated a depauperate fish fauna. Dan Cincotta (1982, personal communication) indicated the Shavers Fork was environmentally stressed due to runoff from active coal mines.
Although no syntopic or intermediate specimens (between flahellare and robustum) were located in the Cheat River system, the intermediacy of pored to total lateral line scales (Fig. 36) in the Coal, Elk, and Guyandot drainages, indicates probable gene flow between Ohio basin and New
River populations. Also, head profiles of Guyandot River specimens (Fig. 37, 35A) are intermediate to specimens from DRAINAGE 0.3 0.4 0.5 0.7
Monongahela
B. Sandy
C. Ohio
L. Sandy
Guyandot
Elk
Cheat
Fig, 36. Variation in the ratio of pored lateral line scales to total lateral line scales. Fig. 37, Head profiles of adult males (drawn from photo projections) :
A. Etheostoma £. robustum.
B. E. f. robustum x E. £. £l a b e l l a r e .
C. E. £. £labellare. Fig. 37. the New and Scioto River drainages. The intermediate
populations also exhibit pigmentation characters typical of
robustum, including conspicuous lines (similar to the
lined population inhabiting Walker Creek of the New River drainage), and darkened anal and pelvic fins. Specimens from the Big Sandy River drainage also show a slight indication of LINES, as well as increased IPL (possibly due to gene f l o w ) .
The Coal and Elk Rivers are tributaries of the Kanawha
River, entering below the Kanawha Falls (located at Glen
Ferris, West Virginia). The Kanawha Falls demarks the arbitrary boundary between the New and Kanawha Rivers. The
Falls, in conjunction with the New River Gorge, have been postulated as barriers to gene flow for numerous aquatic organisms (Jenkins et al., 1972).
The coincidence of barrier falls (i.e., Kanawha and
High Falls of the Cheat) and phenetic differentiation in fantails is presumably due to reduced or asymmetric gene flow. As Endler (1977) has tested with simulated genetic drift and flow models, barriers need not restrict gene flow from all directions to be effective as differentiating agents; a restriction in only one direction accelerates the p r o c e s s . DISCUSSION
EVOLUTION WITHIN ETHEOSTOMA FLABELLARE
ZOOGEOGRAPHIC AND TAXONOMIC CONCLUSIONS
Collette (1965) recognized two groups within
Catonotus, based on the presence or absence of fleshy bulbs on the tips of the dorsal spines. Page (1975) further subdivided the subgenus into three groups. These groups, as currently recognized, are: the E. squamiceps group (E. squamiceps, E. neopterum, and E. olivaceum); the
E. virqatum group (E. virqatum, E. barbouri,
E. s m i t h i , E. s t r i a t u l u m , and E. o b e y e n s e ) ; and the E. flabellare group (E. flabellare and E. kennicotti).
Page (1983), and Bailey and Gosline (1955) have noted a general evolutionary trend within darter genera and subgenera; a reduction in size with evolutionary advancement, although they do not define evolutionary advancement. Bailey and Gosline felt that decreased size is probably causally correlated with reduction in vertebral n u m b e r .
Page (J.983 ) described what he considered evolutionary trends in meristic counts and degree of development of the
133 ■ 134 acoustico-lateralis system among darters; primitive species have completely pored lateral lines, uninterrupted head canals, and high meristic counts. He indicated advanced species have incomplete lateral lines, interrupted head canals, and high meristic counts. Page cited ecological reasons, eg., an increased food resource and protection from predators while inhabiting riffles, as impetus for the evolutionary diminution of darters (advanced darters are riffle inhabitants).
Page (1975) believed that E. squamiceps is the most primitive member of the subgenus, due to its possession of the following character states, which he considerd primitive; large size (85 mm SL maximum), greater amount of scalation (E. squamiceps has a fully scaled nape and scales on the head, which are lacking in other members of
Catonotus except for members of the E. squamiceps group); the lack of large gold-colored knobs on the tips of the first dorsal fin spines (as found in E. flabellare and E. kennicotti); lack of a cheek bar and red and blue pigments (as in the E. virqatum group); the brief period of female inversion during egg deposition (females of E. flabellare, E. kennicotti, and E. obeyense are known to invert under slab rocks for up to two hours during courtship and egg deposition). 135
Pigmentation patterns shared between two of the three groups within Catonotus , and therefore presumably plesiomorphic (primitive), are: diffuse black pigment covering the pelvic and anal fins of nuptial males, vertical bars and horizontal lines along the body, and spotting in the pectoral rays.
If these evolutionary trends are consistent within the
E. flabellare complex and the above characters are assumed to be plesiomorphic, then E. f. robustum retains numerous primitive features including: dark pigmentation of the pelvic and anal fin; 8-10 dark vertical bars as well as horizontal lines (at least in the Walker
Creek population); large standard length (78.5 mm maximum); a large number of vertebrae (modally 36); a greater number of dorsal spines and pectoral rays ; and increased scalation and development of the acoustico-lateralis system.
The protruding lower jaw of E. f . flabellare is presumably a derived character (synapomorphic) within
Catonotus, since it is not found in any of the other species. Etheostoma f. robustum has a subterminal to terminal mouth with a more declivous head profile. The shortened dorsal spines with enlarged bulbs on their tips are also derived characters, within Catonotus.
Etheostoma. f . robustum has the tallest spinous dorsal fin with the smallest bulbs. Under these assumptions, morphologic evolution within
E. flabellare has been primarily due to reductions in
the following: overall size, number of vertebrae, dorsal
spines, pectoral rays, scalation, development of the
acoustico-lateralis system (loss of pored lateral line
scales), lateral bars, lines, and spots in the pectoral fins
(at least in adult males). Derived characters include:
oblique angle of the mouth, enlargement of spinous dorsal
bulbs, and increase in number of soft dorsal rays and
consequently dorsal fin length.
The geographic range of E. f. robustum is
centrally located within the total range of E.
flabellare and adjacent to the Cumberland and Tennessee
River drainages, where all of the other darters in
Catonotus are located (Fig. 38). Avenues for fish dispersal from the New River, into all neighboring drainages
have been postulated (Jenkins et a]^., 1972).
Intraspecific geographic variation within E. flabellare could be due to adaptive responses to
environmental conditions associated with altitude. Barlow
(1961) suggested that slow embryonic development and/or low
temperatures during critical formative periods could
increase meristic counts.
Under 'this hypothesis a widespread ancestral population of fantails, which had reduced meristics and ma#
Fig, 38. Geographic range of E. flabellare (enclosed by dark line) and the ranges of the other nine members of the subgehus Catonotus (stippled areas). Ranges are from Page (1983). acoustico-lateralis systems, could have differentiated into
the differing forms, through selection in accord with localized environmental constraints. The similarity in body proportions of the Atlantic Slope samples and western populations of E. flabellare could thereby be explained via retention of common ancestral character states.
Bailey and Gosline (1955) believed there was a real but erratic tendancy for higher vertebral numbers with increase in altitude, and/or percid egg development at colder temperatures. They were unable to discern how much of the geographic variation was due to environment and/or inherited causes. In reference to the geographic variation of vertebral numbers in E. flabellare populations from the
New River drainage (=robustum), Bailey and Gosline
(1955:27) stated "...there are strong indications of an increase in vertebral number with increase in altitude among the seven samples taJcen in the Upper Kanawha system. "
However, a correlation analysis of their samples indicates a positive correlation between altitude and vertebral number
(r=0.64), but it is not significant at P=0.05.
Also, no significant correlation was found between altitude and lateral line scales, pored lateral line scales, and the ratio of pored to total lateral line scales within the New River basin (utilizing 53 specimens from ten collections, ranging in altitude from 506 m [1,662 feet] to
762 m [2,500 feet] above mean sea level).
SUMMARY
A series of multivariate discriminant function analyses, utilizing 31 meristic, morphometric, and pigmentation characters, indicated Etheostoma flabellare is a widespread polytypic species. There are five recognizable allopatric forms within ^ flabellare. When compared to adjacent geographic populations, the five forms are recognizable via discriminant function analyses, 92 to
96 percent of the time.
The levels of accuracy in identification 0 95%) for four of the five groups (as defined by discriminant function analyses), in conjunction with their allopatric distributions, are consistent with the subspecific concept visualized by Mayr, Simpson, and Hubbs and Hubbs.
The five allopatric forms (Fig. 33) are combined via multivariate discriminant function ordinations and cluster anlysis into two phenetic groups : a Montane-East Slope group, including E. f. robustum (new subspecies),
E. f . h u m e r a l i s , and E. b r e v i s p i n a ; and an
Interior group, including an Ozar)c-Lower Tennessee River assemblage and E. f. flabellare.
The range of the nominal form, E. f. flabellare
(Rafinesque), is herein limited to the Mississippi, Ohio, 140
and Great Lakes basins, and isolated localities in the
Mohawk, Hudson, and upper Susquehanna Rivers. This indicates
the subspecies ability to transgress drainage divides,
especially in terrain formerly covered by Wisconsinan
glacial ice.
Etheostoma f. lineolatum (Agassiz) is reduced to
a synonym of E. f . flabellare; since there is clinal
variation between the two forms, any division along this
Cline would be arbitrary.
Another questionable form (recognized via discriminant
analysis 92% of the time from adjacent populations), herein
termed the Ozark-Tennessee group, inhabits the White River
system of the Ozark Uplands in Missouri and Arkansas, and tributaries of the lower Tennessee River in the southern
Appalachians. Further analyses (including multi- and univariate analyses of standard characters, and possibly electrophoretic data) of this form are needed to clearly define its geographic distribution and relationships to the nominal form. Samples from the geographically intermediate populations, i.e., the Duck, Clarks, and Buffalo Rivers
(Tennessee River drainage), should be examined to clarify
this problem.
Etheostoma f. robustum (n. ssp.) inhabits the upper Tennessee River system (upstream of the confluence of
the Little Tennessee River), the New River drainage, and as 141
a localized population in the headwaters of Shavers Fork
Cheat River, a Monongahela-Ohio R. tributary. Etheostoma
f . flabellare and E. f . robustum are both present
in the Shavers Fork drainage, but are not syntopic.
Intergrades between these two subspecies are identified from the headwaters of the Guyandot, Coal, and Elk Rivers in West
V i r g i n i a .
Etheostoma f . humeralis inhabits the Atlantic slope drainages, including the lower Susquehanna, Potomac,
James, Roanoke, Neuse, and presumably Cape Fear Rivers.
Etheostoma f . brevispina is the most southern representative of the barred fantail group on the Atlantic slope, inhabiting the Catawba, Broad, and Pee Dee River drainages in North and South Carolina. LITERATURE CITED
Agassiz, L. 1854. Notice of a collection of fishes from the southern bend of the Tennessee River, Alabama. Amer. J. Sci. and Arts. (Series 2) 17:297-308, 353-369.
Bailey, R. M., and W. A. Gosline. 1955. Variation and systematic significance of the vertebral counts in the American fishes of the family Percidae. Misc. Publ. Mus. Zool. Univ. Mich. No. 93. 44 p.
and G. R. Smith. 1981. Origin and geography of the fish fauna of the Laurentian Great Lakes basin. Can. J. Fish Aquat. Sci. 38 (12 ) .-1539-1561.
Barlow, G. W. 1961. Causes and significance of morphological variation in fishes. Syst. Zool. 10:105-117.
Becker, G. C. 1976. Environmental status of the Lake ■ Michigan Region. Vol. 17. Inland fishes of the Lake Michigan drainage basin. Argonne National Laboratory. ANL/ES-40. Vol. 17, 237 p.
Braasch, M. E., and L. M. Page. 1979. Systematic status of darters of the subgenus Catonotus (Percidae), with the description of a new species from Caney Fork, Tennessee. Occ. Pap. Mus. Nat. Hist., Univ. of Kansas. No 78, 10 p.
Buhan, P. J. 1966. Two Races of Etheostoma flabellare flabellare Rafinesque from the Roanoke River of Virginia and the Neuse River of North Carolina. Unpubl. MS Thesis, Virginia Polytechnic Institute, Blacksburg, Virginia. 87 p.
Call, R. E. 1899. Ichthyologia Ohiensis and the ichthyological bibliography of Rafinesque, 175 p.
Clay, W. M. 1975. The fishes of Kentucky. Kentucky Dept, of Fish and Wildlife Resources, Frankfort, Kentucky, 416 p.
Cocker, R. E. 1926. New genus of darter from western North Carolina. Bull. U.S. Bur. Fish. 42:105-108.
Cocker, R. E. 1927. Richiella to replace Richia as name for genus.of darter. Copeia 1927(162);17-18. Collette, B. B. 1965. Systematic significance of breeding tubercles in fishes of the family Percidae. Proc. of U.S. Nat. Mus. 117:567-614.
and L. W. Knapp. 1967. Catalog of type specimens of the darters (Pisces, Percidae, Etheostomatini). Proc. U.S. Nat. Mus. 119:1-88.
Cooper, E. L. 1983. Fishes of Pennsylvania and the northeastern United States. Pennsylvania State University Press, University Park, Pennsylvania, 243 p.
Cope, E. D. 1868. On the distribution of freshwater fishes in the Allegheny region of southwestern Virginia. J. Acad. Nat. Sci. Phila. (Series 2) 6:207-247.
Cope,. E. D. 1870. On some etheostomine perch from Tennessee and North Carolina. Proc. Amer. Philos. Soc. 11:261-270.
Crowl, G. H. 1980. Woodfordian age of the Wisconsin glacial border in northeastern Pennsylvania. Geology 8:51-55.
Denoncourt, R. P., W. A. Potter and J. R. Stauffer, Jr. 1977. Records of the greenside darter Etheostoma blennioides from the Susquehanna River drainage in Pennsylvania. Ohio J. Sci. 77(4):38-42.
Dixon, W. J. ed. 1983. BMDP Statistical Software, 1983 Printing with additions. University of California Press, Berkeley. '734 p.
Endler, J. A. 1977. Geographic variation, spéciation and . dines. Princeton Univ. Press, Princeton. 246 p.
Fairchild, H. L. 1908. Pleistocene history of the Genesee valley. In: Geologic map and description of the Portage and Nunda quadrangles including a map of Letchworth Park. J. N. Clarke and D. Luther, eds., N.Y. State Mus. Bull. No. 118, 88 p.
Gerking, S. D. 1945. The distribution of the fishes of Indiana. Investigations of Indiana lakes and streams. 3(1):1-137.
Girard, C. 1859. Ichthyological notices, 5-27. Proc. Acad. Nat. Sci. Philadelphia 11:56-68.
Goldthwait, R. P., A. Dreimanis, J. L. Forsyth, P. F. Narrow, and G. W. White. 1965. Pleistocene deposits of the Erie lobe. In: The Quaternary of the United States. H. E. Wright, Jr. and D. G. Frey, ed. Princeton Univ. Press. Princeton. 922 p.
Greene, C. W. 1935. The distributions of Wisconsin fishes. Wise. Dep. Cons., Madison, Wisconsin. 235 p.
Gould, S. J., and R. F. Johnston. 1972. Geographic variation. Ann. Rev. Ecol. Syst. 3:457-498.
Hubbs, C. L. 1926. Checklist of the fishes of the Great Lakes and tributary waters with nomenclatorial notes and analytical keys. Univ. Mich. Mus. Zool. Misc. Publ. 15: 77 p. Hubbs, C. L. 1927. Richiella brevispina (Coker) considered a synonym of Catononotus flabellaris humeralis (Girard). Copeia (163):43-45.
______, and C. Hubbs. 1953. An improved graphical analysis and comparison of series of samples. Syst. Zool. 2 ( 2 ) :49-59, 92.
______, and K. F. Lagler. 1941. Guide to the fishes of the Great Lakes and tributary waters. Bull. Cranbrook Inst. Sci. 18: 100 p.
and K. F. Lagler. 1958. Fishes of the Great Lakes region. Univ. Mich. Press, Ann Arbor. 213 p.
Hocutt, C. H. 1979. Drainage evolution and fish dispersal in the central Appalachians. Geol. Soc. Amer. Bull, part II, 90:197-234.
Denoncourt and J. R. Stauffer, Jr. 1978. Fishes of the Greenbrier River, West Virginia, with drainage history of the Central Appalachians. Jour. Biogeog. 5:59-80.
H u m p h r i e s , J. M . , F. L. B o o k s t e i n , B. C h e r n o f f , G. R. Smith, R. L. Elder and S. G. Poss. 1981. Multivariate discrimination by shape in relation to size. Syst. Zool. 30:291-308.
Jenkins, R. F . , E. A. Lachner and F. J. Schwartz. 1972. Fishes of the Central Appalachian drainages; their distribution and dispersal, p. 43-118. In: The distributional history of the biota of the Southern Appalachians. Part III. Vertebrates. P. C. Holt, ed. Res. Div. Monogf. 4, Virginia Poly-Technical Institute, Blacksburg. Jordan, D. S. 1878. A catalogue of the fishes of Illinois. Illinois State Laboratory of Natural History Bull. 1(2):37-70.
Jordan, D. S. 1890. Report of explorations made during 1888 in the Allegheny region of Virginia, North Carolina, and Tennessee, and in western Indiana, with an account of the fishes found in each of the river basins of those regions. Bull. U.S. Fish Comm. (1889) 8:97-173.
Jordan, D. S. and B. W. Evermann. 1896. The fishes of North and Middle America. U.S. Nat. Mus. Bull. 47. part 1. 1240 p. Clark. 1930. Checklist of the fishes and fishlike vertebrates of North and Middle America, north of the northern boundary of Venezuela and Columbia. Kept. U. S. Comm. Fish. (1928):l-670 p.
Jordan, D. S., and C. H. Gilbert. 1883. Synopsis of the fishes of North America. U.S. Nat. Mus. Bull. 16. 1018 p.
Klecka, W. R. 1980. Discriminant analysis. Sage University series: Quantitative Applications in the social sciences, series no. 07-001. Beverly Hills, CA. 71 p.
Kneib, R. T. 1972. The effects of man's activity on the distribution of five stream fishes in Little Pine Creek, Pennsylvania. Proc. Penna. Acad. Sci. 46:49-51.
Kuehne, R. A., and R. W. Barbour. 1983. The American darters. Univ. Press of Kentucky. 177 p.
and J. W. Small, Jr. 1971. Etheostoma barbouri, a new darter (Percidae, Etheostomatini) from the Green River with notes on the subgenus Catonotus. Copeia 1971(1):18-26.
Lachenbruch, P. A. 1975. Discriminant analysis. Hafner, New York. 128 p.
Lachner, E. A., and R. E. Jenkins. 1971. Systematics, distribution, and evolution of the chub genus Nocomis Girard (Pisces, Cyprinidae) of eastern United States, with descriptions of new species. Smithson. Contrib. Zool. No. 85.
Lee, D. S., C. R. Gilbert, C. H. H o c u t t , R. E. Jenkins, D. E. McAllister and J. R. Stauffer. 1980. Atlas of North American freshwater fishes. Pileigh: North Carolina State Museum of Natural History, N rth Carolina Biol. Surv. Publ. No. 1982-12, 854 p.
Lutterbie, G. W. 1979. Reproduction and age growth in Wisconsin darters (Osteichthyes: Percidae). Rep. Fauna Flora Wis. 15:44 p.
MacClintock, P., and E. T. Apfel. 1944. Correlation of the drifts of the Salamanca re-entrant. New York. Geol. Soc. Amer. Bull. 55:1143-1164.
Mayr, E. 1969. Principals of systematic zoology. McGraw Hill Book Co., New York.
1970. Populations, species and evolution. Belknap Press of Harvard Univ. Press, Cambridge.
Miller, R. V. 1968. A systematic study of the greenside darter, Etheostoma blennioides Rafinesque (Pisces: , Percidae). Copeia. 1968:1-40.
Muller, E. H. 1960. Glacial geology of Cattaragus County, New York: Friends of Pliestocene Geol. 23rd Ann. Meeting Guidebook, Syracuse Univ. 33 p.
Myers, G. S. 1927. The status of the darter Richiella brevispina (Coker). Copeia 1927(163):39-43.
Page, L. M. 1975. Relations among the darters of the subgenus Catonotus of Etheostoma. Copeia 1975:782-784.
Page, L. M. 1981. The genera and subgenera of darters (Percidae, Etheostomatini). Occ. Pap. Mus. Nat. Hist., Univ. Kansas. No. 90, 69 p.
. 1983. Handbook of darters. TFH Publications. 271 p.
and L. E. Cordes. 1983. Variation and systematics of Etheostoma euzonum the Arkansas saddled darter (Pisces; Percidae). Copeia 1983:1042-1050.
Pimentel, R. A. 1959. Mendelian infraspecific divergence levels and their analysis. Syst. Zool. 8:139-159.
1979. Morphometries the multivariate analysis of biological data. Kendall/Hunt Publishing Co., Dubuque. 276 p. and D. S. Frey. 1978. Multivariate analysis of variance and discriminant analysis, p. 247-274. In; Quantitative ethology. P. W. Colgan, ed. John Wiley, New York.
Pflieger, W. L. 1971. A distributional study of the Missouri fishes. Mus. Nat. Hist. Univ. Kan. Publ. 20:225-570.
Rafinesque, C. S. 1819. Prodrome de 70 nouveax genres d 'animaux découverts dans 1'intérieur des Etats-Unis d'Amérique durant l'annee 1818. Journ. Phys., Chim., Hist. Nat., Arts 88:417-429.
1820. Ichthyologia Ohiensis or natural history of the fishes inhabiting the river Ohio and its tributary streams. 90 p. [not seen, see Call, 1899].
Ross, R. D. 1958. Some taxonomic problems of Shenandoah River fishes. Va. Ag. Exp. Sta. Tech. Bull. 137:1-10.
Carico. 1963. Records and distribution of • fishes of the North, Middle, and South Forks of the Holston River, Virginia. Va. Ag. Exp. Sta. Tech. Bull. 161:1-24.
Schwartz, F. J. 1965. The distribution and probable postglacial dispersal of the Percid fish Etheostoma b. blennioides, in the Potomac River. Copeia 1965:285-290.
Seifert, M. F. 1963. Some aspects of territorial behavior of the fantail darter. Iowa Acad. Sci. 70:224-227.
Simpson, G. G. 1961. Principals of animal taxonomy. Columbia Univ. Press, New York.
Sneath, P. H. A., and R. R. Sokal. 1973. Numerical taxonomy. W. H. Freeman and Co., San Francisco.
Storer H. R. 1851. A description of a new species of Etheostoma, under the name of Etheostoma Linsleyi. Pro. Boston Soc. Nat. Hist. (4):37-39.
Thorpe, R. S. 1976. Biometric analysis of geographic variation and racial affinities. Biol. Rev. 407-452.
Todd, T. N., G. R. Smith and L. E. Cable. 1981. Environmental and genetic contributions to morphological differentiation in ciscoes (Coregoninae) of the Great Lakes. Can J. Fish. Aquat. Sci. 38:59-67. Trautman, M. B. 1957. The fishes of Ohio. Ohio State Univ. Press. 683 p.
. 1981. The Fishes of Ohio. Ohio State Univ. Press. 782
Vaillant, L. 1873. Recherches sur les poissons des eaux douces de l"Aerique septentrionale. Nouv. Arch. Mus. Hist. Nat. Paris (l):5-54.
Wallace, D. C. 1973. The distribution and dispersai of the silverjaw minnow Ericymba buccata Cope. Amer. Mid. Nat. 89:145-155.
, and B. A, Branson. 1979. Lactate dehydrogenase isozymes in six species of darters in the subgenus Catonotus Agassiz (Percidae: Etheostomatini). Amer. Mid. Nat. 102(2):392-394.
Wolfe, G. W., B. A. Branson, and S. L. Jones. 1979. An electrophoretic investigation of six species of darters in • the subgenus Catonotus. Biochem. Syst. and Ecol. 7:81-85. APPENDIX A
Material Examined Samples are arranged by drainage. Atlantic slope samples are listed from north to south, followed by the Gulf of Mexico drainages, arranged from northeast to southwest, and finally the Great Lakes samples listed from west to east. Samples within drainages are listed alphabetically.
Etheostoma flabellare humeralis
Lower Susquehanna River Drainage. Pennsylvania: Franklin County: PSÜ 219 (12) Conodoquinet Creek, 60 m below upper edge of State Game Lands, Letterkenny Township. Potomac River Drainage. Pennsylvania: Bedford County; OSUM 37586 (9) Little Wills Creek, 1.6 km S of Madley. Franklin County: PSU 187 (9) East Branch Antietam Creek, 27 m downstream from first bridge above Glen Forney. AEL 213 (5) Licking Creek at St. Rt. 416. AEL 206 (6) Back Creek at St. Rt. 30. AEL 223 (4) West Branch Conococheague Creek at Twp. Rt. 28037 E of Mercersburg. Virginia: Rockingham County: CU 8358 (7) North River at Mt. Crawford. James River Drainage. Virginia: Allegheny County; OSUM 41766 (.6) Potts Creek at St. Rt. 18. Greene County; CU 20948 (3) Swift Run, 9.0 km NW of Stanardsville on U.S. Rt. 33. Nelson C o u n t y ; CU 8424 (6) Tye River, 0.3 km NW of Roseland. Roanoke County : OSUM 38867 (12) Catawba Creek at St. Rt. 311, 0.72 km E of Catawba. Roanoke River Drainage. (Western Roanoke). Virginia: Montgomery County; CU 20440 (8) trib. of Elliott Creek, 1.0 km S of Rogers on Co. Rt. 615. CU 20330 (7) North Fork Roanoke River at Bennetts Mills, 8.0 km NE Blacksburg on Co. Rt. 785. Roanoke River Drainage. (Eastern Roanoke). Virginia: Brunswick County; CU 11930 (1) Great Creek trib. of Meherrin River (Chowan River Drainage), 6.4 km SW of Alberta on U.S. Rt. 1. Cambell County: CU 8523 (2) trib. of Otter River, 7.2 km N of Lynch on St. Rt. 626. Roanoke County; OSUM 34632 (10) Roanoke River at Dixie Caverns, 12.8 km W of Salem on U.S. Rt. 460. Dan River Drainage (trib. of Roanoke River). Virginia; Pktrick County: OSUM 41730 (8) Rock Castle Creek on St. Rt. 40, upstream of Philpott Reservoir. North Carolina: Forsythe County: OSUM 41841 (4) Town Fork Creek, just N of jet. St. Rt. 8 and *65, 0.8 km E of
149 Germantown. Rockingham County; CU 168 49 (7) trib. of Dan River, 2.7 km W. of Leakesville. Stokes County; CU 13877 (2) trib. of Dan River, 1.9 km E of Germantown. Neuse River Drainage. North Carolina: Durham County: CU 9680 (4) Mountain Creek, just W of Bahama. Orange County: CU 18583 Eno River, 3.2 km E of Efland. CU 31683 (2) Eno River, 3.2 km W of Hillsborough on U.S. Rt. 70. Wake County: CU 34489 (1) Upper Barton Creek, 4.8 km S of jet St. Rts. 50 and 98 on St. Rt. 50. CU 9728 (4) small trib. NW of Raleigh on Rt. 7A. OSUM 34677 (4) Horse Creek at St. Rt. 98, 6.4 km W of Wake Forest.
Etheostoma flabellare brevispina
Pee Dee River Drainage. North Carolina: Randolph County: OSUM 42970 (2) Back Creek at U.S. Rt. 64, 8.0 km W of Asheboro. Surry County: CU 10611 (2) trib. of Ararat River, 4.8 km NE of Mt. Airy on Rt. 104. OSUM 41798 (7) Mitchell River at St. Rt. 268, 8.0 km W of jet. with U.S. Rt. 601. Wilkes County: CU 11424 (6) trib. of Yadkin . [Mulberry Creek?], 8.8 km NE of Wilkesboro. CU 10992 (1) Moravian Creek, 1.9 km W of Moravian Falls. CU 38175 (2) off St. Rt. 18, 0.8 km on Co. Rt. 1128, 24.0 km SW of Wilkesboro. South Carolina: Chesterfield County; CU 29835 (13) Thompson Creek, 4.3 km NE of Mt. Croghan on St. Rt. 109. Catawba-Saluda River Drainage. North Carolina: Burke County: OSUM 34687 TOPOTYPES for E. f. brevispina. (10) Paddys Creek at St. Rt. 126, 0.2 km W of jet. with Kistler Memorial Highway (dirt road). Caldwell County: OSUM 41851 (2) Mulberry Creek, 5.6 km E of Collettsville on St. Rt. 90. McDowell County: CU 54938 (2) Curtis Creek, 1.9 km E of Old Fort on U.S. Rt. 70. Rutherford County: OSUM 41807 (2) Second Broad River secondary road east of Thermal City and jet. with U.S. Rt. 221.
Etheostoma flabellare robustum
Upper Tennessee River Drainage. North Carolina: Buncombe County : CU 63433 (8) Cane Creek at Co. Rt. 3116, 4.8 km E of Fletcher. Madison County: CU 29258 (1) Paint Fork above Little Ivy Creek and Beech Glen, 9.1 km SE of Mars Hill. Swain County: OSUM 46789 (5) Little Tennessee River at confl. with Sawmill Creek. Transylvania County; OSUM 41717 (5) French Broad River at St. Rt. 64, at jet. with Co. Rt. 215. Virginia: Scott County: OSUM 38818 (1) Copper Creek, 1.9 km E of Nicklesville, at jet. of Co. Rts. 611 and 612. Smythe County: OSUM 34637 (1) Comers Creek, along Co. Rt. 650, 4.8 km ESE of Teas. PSU 1699 (6) Bear Creek, trib. of Middle Fork Holston River. OSUM (3) Bear Creek above Marion. Tazewell County; OSUM 38755 (1) North Fork Clinch River at St. Rt. 651. Tennessee: Blount County: OSUM 35818 (1) Millstone Creek, 2.4 km E of confl. with Ellejoy Creek. Seiver County: CU 41483 (2) trib. of Little Pigeon River, 3.2 km S Pitman Center on dirt rd. off St. Rt. 73. New River Drainage (Southern section). West Virginia: Monroe County: AEL 107 (2) Indian Creek at U.S. Rt. 219. Virginia: Carroll County : CU 9514 (2) Glade Creek, 4.8 km SW Hillsville. Floyd County : CU 20367 (8) Little River, 13.7 km S of Copper Hill on Co. Rt. 664. CU 19464 (2) Dodds Creek, 2.7 km S of Floyd on U.S. Rt. 8. Grayson C o u n t y : AEL 31 (15) New River, just upstream from Co. Rt. 601, 8.6 km SW of Indépendance. CU 52224 (5) Fox Creek at mouth to 0.4 km upstream, U.S. Rt. 58, 3.2 air 1cm NE of Mouth Of Wilson. CU 9471 (2) Chestnut Creek, 1.6 km S of Galax. OSUM 47276 (1) Wilson Creek, 3.7 km W of Volney on Rt. 58. Giles County: CU 25308 (5) Spruce Run at confl. with New River at Goodwins Ferry. Montgomery County: CU .25375 (5) Norris Run at confl. with New R. [jet. Co. Rts. 625 & 708)]. CU 20448 (2) Toms Creek, 3.4 km NW of Blacksburg. Tazewell County: CU 21653 (1) Bluestone River, 10.4 km SW of Bluefield on U.S. 460. Pulaski County : CU 42137 (2) Little Walker Creek at Rt. 100, N of Dublin. Wythe County: CU 8044 (1) Reed Creek, 1.6 km SW of Wythville on Rt. 21. AEL 34 (1) New River, Dickensons Ferry at end of Co. Rt. 638. CU 24175 (3) Middle [South] Fork Reed Creek, 19.p km SW of Wythville on Co. Rt. 667. CU 209 42 (7) trib. to New River,12.3 km NE of Fort Chiswell on U.S. 21. North Carolina: Ashe County; OSUM 18159 (1) South Fork New River at St. Rt. 163, 12.3 km SE of West Jefferson, 3.5 km SW of Glendale Springs. CU 7681 (2) South Fork New River, 2.4 km NE of Baldwin on Rt. 221. CU 7654 (1) North Fork New River at Crumpter, on Rt. 16. New River Drainage (Northern section) West Virginia: Monroe County: AEL 111 (4) Second Creek, county road bridge upstream of U.S. Rt. 219 [219/1]. Nicholas County: AEL 190 (2) Muddlety Creek. CU 32620 (4) North Fork Cherry Creek [River?] above Richmond [Richwood]. Pocahontas County: OSUM 51624 (3) Type for robustum Stoney Creek at U.S. Rt. 219 at Cambelltown. CU 20602 (1) trib. of Greenbrier River, just E of B artow at St. Rt. 28. AEL 182 (5) Williams River, Rt. 150. OSUM 51950 (1) Williams River, 1.2 km downstream from USFWS Rt. 86, Three Forks of W i l l i a m s . Cheat River (Monongahela drainage) West Virginia: Randolph County: CU 32333 (1) Shavers Fork Cheat River at confl. with Red Run, downstream of Cheat Bridge. AEL 60 (9) Shavers Fork Cheat River on Mower logging Rd., approximately 180 m below Lambert Run.
Etheostoma f . robustum X Etheostoma f. f labellare
Kanawha River Tributaries. Elk River. West Virginia. Webster County; OSUM 537 55 (5) Elk River, 185.0 km above mouth (3.2 km above Bear Run). AEL 711 (3) Laurel Creek at Co. Rt. 9, 3.2 km SW of Waynesville [Wainville, LTM]. Coal River. West Virginia. Boone County; OSUM 53757 (2) Pond Fork, 25.3 km from mouth, upstream of Bim, 0.2 km upstream from Dry Branch. Fayette County; CU 14872 (4) Armstrong Creek at Kimberly. Logan County; CU 2080 (2) Spruce Fork, 3.2 km below Sharpies on Rt.119. Raleigh County: CU 20875 (3) Clear Fork Coal River. 1.6 km above D o r o t h y . Guyandot River. West Virginia. Logan County: WVWR 320 (2) Island Creek, 22.0 km upstream from mouth, 13.5 m above right fork. WVWR 321 (5) Island Creek 21.6 km upstream .from mouth. Wyoming County: CU 13394 (12) Barkers Creek. AEL (3) Huff Creek, 0.6 km W of Cyclone, along Rt. 10. WVWR (10) Huff Creek, 20.2 km upstream from mouth, 38.4 m below Road Branch.
Etheostoma flabellare flabellare
Mohawk River. New York: Herkimer County: CU 18806 (10) Moyer.Creek. Genesee River. New York: Wyoming County: CU 9209 (12) East Koy Creek at Weathersfield Springs. Pennsylvania: Potter County: OSUM 37028 (4) Middle Branch Genesee River, 3.2 km N of Gold. Lake Ontario. New York: Franklin County: CU 9012 (2) Little Trout River, 2.8 km NNW of Burke and 8.0 km W of Chateaugay. Monroe County: CU 22357 (10) Allens Creek off Monroe Avenue, Brighton, from Westfall Rd. S to bridge S of Monroe Avenue [Rochester Suburbs]. Tompkins County; CU 13632 (1) Fall Creek at Varna. CU 38214 (1) Cayuga L. Taughannock Point, North Bay. CU 4219 (2) Cayuga Inlet, near Buttermilk Falls. Central Lake Erie. New York: Cattaraugus County: CU 6642 (2) Clear Creek. Erie County: CU 9143 (1) Clear Creek at Bagdad, 2.4 km above Gowanda. Ohio: Grand River Ashtabula-Lake Counties: OSUM 51551 (1) Mill Creek at Ross Rd. br. Lake County: OSUM 51549 (1) Grand River at Painsville Park. OSUM 51548 (4) Grand River at confl. with Paine Creek. Trumbell County: OSUM 51550 (1) Grand River at St. Rt. 87. OSUM 21684 (2) Coffee Creek at St. Rt. 534 just N of Peck-Leach Rd., 1.0 km N of Farmington. Chagrin River Geauga County; OSUM 51544 (3) Chagrin River at Fowlers Mills. OSUM 51545 (2) Silver Creek at Novelty. Lake County; OSUM 51547 (1) Chagrin River, 0.8 km below Daniels Park Dam. Cuyahoga River Portage County: OSUM 51541 (1) Cuyahoga River at Shalersville, St. Rt. 303. OSUM 48739 (5) Cuyahoga River at St. Rt. 303. Summit County; OSUM 51540 (2) Yellow Creek at W. Bath Rd. (Co. Rt. 48). OSUM 51539 (4) Furnace Run at St. Rt. 303. Rocky River Medina County; OSUM 51553 (2) East Branch Rocky River at Bellus Rd. OSUM 51552 (1) West Branch Rocky River at St. Rt. 162. OSUM 51556 (1) East Fork Rocky River at Walker Rd., E of Metro Lake. Black River Medina County: OSUM 51555 (1) West Fork of East Branch Black River at Co. Rt. 29. Western Lake Erie. Michigan; Macombe County; PSU 1727 (6) Stoney Creek [Clinton R.- Lake St. Claire drainage]. Washtenaw County; PSU 1720 (6) City Park below dam [Ann Arbor, Huron River]. Ohio; Auglaize County; OSUM 4112 (1) Auglaize River, Duchouquet Twp., Sec. 14. Putnam County; OSUM 4204 (5) Auglaize River, Perry Twp., Sec. 21. Williams County; OSUM 1068 (2) West Branch Saint • Joseph River, Bridgewater Twp. Wood County; OSUM 4167 (2) Maumee River, Washington Twp. Lake Erie Ottawa County; Lake Erie, Peach Point South Bass Island. Upper Susquehanna River. New York; Steuben County; CU 25367 (1) Chemung River, 0.6 km SW of Rexville. Pennsylvania; Cameron County ; OSUM 51538 ( 5) First Fork Sinnemahoning Creek, 4.8 km downstream from G.B. Stevenson Dam, along St. Rt. 872. OSUM 37569 (2) First Fork Sinnemahoning Creek, just above confl. with Bailey Run. Potter County; OSUM 37335 (2) First Fork Sinnemahoning Creek, 4.8 km S of Wharton. OSUM 37640 (1) East Fork of First Fork Sinnemahoning Creek, near Logue. Allegheny River. Pennsylvania; McKean County: CU 24012 (2) Kinzua Creek, 14.7 km NW of Kane on R t . 68. Potter County; OSUM (6) Allegheny River at Coudersport. OSUM 377 46 (11) Oswayo Creek at Sharon Center. Monongahela River. West Virginia; Harrison County; OSUM 53759 (2) Tenmile Creek at confl. with West Fork River. Lewis County ; CU 32652 ( 5) West Fork River at Jacksons Mills. Monongalia County; OSUM 53761 (3) Whiteday Creek, 4.8 km above mouth at St. Rt. 73. Randolph County; CU 148 42 (2) Middle Fork River at Ellamore. Preston County; OSUM 53762 (3) Little Sandy Creek, 2.4 km above mouth (Clifton Mills). Taylor County: OSUM 53760 (5) Wickwire Creek, 0.8 km above confl. with Tygart Valley River. Tucker County; OSUM 52034 (3) Haddix Run at Porterwood. Youghiogheny River. Pennsylvania; Fayette County; AEL 258 (5) Youghiogheny River at first riffle above old distillary at Broadford. Somerset County; AEL 253 (11) Laurel Hill Creek at St. Rt. 653. Mahoning River. Ohio: Belmont County: OSUM 6861 (2) Bend Fork, Washington Twp., Sec. 23. Jefferson County: OSUM 6848 (4) Clay Lick Creek, N. Wayne Twp., Sec. 12. Mahoning County: OSUM 9954 (4) Mill Creek, Beaver Twp. OSUM 4918 (1) Burgess Run, Boardman Twp. Starke County: OSUM 3242 (1) Deer Creek, Lexington Twp., Sec. 4. OSUM 4899 (5) [illustrated in Fishes of Ohio; (Trautman, 1957)] Deer Creek, Lexington Twp., Sec. 4. Muskingum River. Ohio: Ashland County: OSUM 4-150 (1) Honey Creek, Green Twp., Auten. Coshocton County: OSUM 1254 (4) Tuscarawas River, Lafayette Twp. OSUM 984 Walhonding River at Bethlehem, 1.6 km above Six Mile Dam. OSUM 12774 (2) Walhonding River, Bethleham Twp. Holmes County: OSUM 21020 (1) Doughty Creek, below confl. with Charm Run. Licking County: OSUM 35783 (4) South Fork Licking River at Co. Rt. 25. OSUM 47357 (1) Sharon Valley Run at OSU-Newark Campus. Monroe County: OSUM 6290 (1) South Fork Wills Creek. Morrow County: OSUM 19268 (3) Kokosing River at U.S. 42, 5.1 km E of Mt. Gilead. Muskingum County: OSUM 951 (6) Muskingum River, Washington Twp. Tuskarawas County: OSUM (5) Tuskarawas River, Goshen Twp. Wayne County: OSUM 2177 (5) Killbuck Creek, Congress Twp. Scioto River. Ohio: Delaware County: OSUM 9953 (3) Scioto River Scioto Twp. OSUM 23862 (9) Big Walnut Creek upstream of St. Rt. 36. Franklin County: OSUM 21266 (2) Olentangy River at Columbus Zoo. OSUM 13003 (3) Rocky Fork Creek at P‘lain-Jefferson Twp. line. OSUM 19345 (1) Alum Creek on Watt Rd. (Twp. Rt. 182) on W border of Westerville, 4.6 km NE of Worthington. OSUM 2387 (4) Big Darby Creek, Pleasant Twp. Hardin County: OSUM 6256 (3) Taylor Creek, Reservoir site, S Buck Twp., 4.8 km S of Kenton. Highland County: OSUM 21370 (1) Farmers Run, from abandoned bridge to confl. with Paint Creek, 2.7 km S of Greenfield. OSUM 21320 (1) Rattlesnake Creek W of Rt. 138, Madison-Paint Twp. OSUM 21355 (1) Fall Creek, 0.8 km above confl. with Paint Creek, 9.6 km SW of Greenfield. Hocking County; OSUM 22564 (4) Laurel Run at Laurelville. OSUM 7215 (15) Salt Creek, Salt Creek Twp. Sec. 1 and 36. Jackson County: OSUM 4734 (2) Little Scioto River, SW Scioto Twp. Pickaway County: OSUM 6224 (6) Big Darby Creek, Darby-Scioto Twp. OSUM 21240 (3) Big Darby Creek at Fox. OSUM 5995 (4) Big Darby Creek at Fox. Pike County; OSUM 1390 (5) Sunfish Creek, SW Newton Twp. Ross County : OSUM 13947 (2) Scioto River above confl. with Deer Creek, above St. Rt. 104, E Union Twp. Scioto County: OSUM 43149 (2) Scioto Brush Creek at St. Rt. 104, 0.3 km N of Rushtown. OSUM 2575 (2) Bear Creek at mouth. Morgan Twp. Union County; OSUM 35703 (5) Hay Run, Union T w p . Miami River. Ohio. Auglaize County; OSUM 3941 (2) Muchinippi Creek, Sec. 8, Goshen Twp. Butler County; EKU 506 (9) Paddy's Run Creek, just below Rt. 126. Champaign County; OSUM 9952 (6) Cedar Run. Clarke County: OSUM 35171 (1) Beaver Creek, upstream of Bird Rd., 1.6 km E Springfield. Darke County: OSUM 4024 (3) Greenville Creek, Sec. 36, Greenville Twp. Logan County : OSUM 35141 (4) Mad River, downstream from old NYC RR bridge in West Liberty. Indiana. Franklin County : OSUM 28219 (3) Big Salt Creek, 3.2 km W of Pepperton. Ohio County: OSUM 28092 (1) Laughery Creek, Union Twp., Milton. Union County; OSUM 28270 (2) Silver Creek 3.2 km SW of Liberty. Little Kanawha River. West Virginia. Ritchie County: [AEL uncat. CHH 81-017] (2) South Pork River at St. Rt. 53, 1.6 km N of Girta. Upshur County: CU 13474 (7) Little Kanawha River at Fishers Camp. Wirt County; OSUM 53750 (1) Little Kanawha River, 30.9 km above mouth, at Newark Riffle. OSUM 53751 (3) Little Kanawha River, 59.0 km . above mouth, at confl. with Nettle Run, below Burning Springs. OSUM 53752 (6) Hughes River, 14.1 km above mouth, at Deems Ford at end of WV Sec. Rd. 47/5. OSUM 53753 (2) Hughes River, 9.9 km above mouth at confl. with Goose Creek at Freeport. Wood County: AEL [CHH 81-83, uncatol.] Walker Creek at Co. Rt. 7 in Hanna. Big Sandy River. Kentucky. Martin County; OSUM 52249 (1) Tug Fork Big Sandy River at br. on Little Laurel Creek Rd.,. 0.2 km W of St. Rt. 3. Virginia. Tazewell County; OSUM 51965 (5) Dry Fork at Camp Tazewell at Co. Rt. 637. West Virginia. McDowell County: CU 23803 (2) Horsepen Creek, 1.0 km NE of Bishop. CU 13423 (1) Panther Creek of Tug Fork. WVWR 137 (1) Dry Fork, 8.0 km above mouth. Wayne County: CU 32596 (1) Twelvepole Creek, 1.6 km N of Wayne. Little Sandy River. Kentucky. Boyd County: OSUM 52247 (3) East Fork Little Sandy River, E of St. Rt. 3, 1.5 km N of jet. of KY 3 and K966. Carter County: CU 25587 (1) Little Sandy River at jet. U.S. 60Y and KY 1. OSUM 52245 (4) Big Sandy Creek , 1.7 km upstream from confl. with Little Sandy River. Elliot County: OSUM 52244 (9) Little Fork Little Sandy River at jet. of Ky 486 and Wallow Hole Creek Rd., 1.0 km NE of Culver. Greenup County; OSUM 18298 (1) Little Sandy River at Rt. 1, 9.6 km S of Greenup, 16.0 km WNW of Ashland. Licking River. Kentucky. Bath County; OSUM 47229 (7) Licking River at Rt 826, 7.8 km SE of Saltlick and 6.4 km S of U.S. 60. EKU 132 (1) Licking River at U.S. 60. Flemming County; EKU 543 (5) Licking River above St. Rt. 111. EKU 539 (3) Licking River above St. Rt. 11. EKU 538 (1) Fox Creek off St. Rt. 111. Magoffin County; OSUM 52 241 (4) Licking River, 17.0 km SE Bert T. Combs Pkwy. on St. Rt. 7. Menifee County; OSUM 52240 (7) Beaver Creek, 0.2 km S of St. Rt. 1274 near confl. with Clifton Creek. Robertson County; EKU 139 (2) Licking River, 4.8 km N of Mt. Olive, Bridgeville on St. Rt. 165. Rowan County; CU 45379 (2) Triplett Creek near RR in Morehead. OSUM 52243 (1) North Fork Triplett Creek, 9.5km NNE from jet. St. Rt. 32 and St. Rt. 377. Central Ohio River tributaries. Ohio. Lawrence County: OSUM 7376 (3) Symmes Creek. Meiges County; OSUM 4737 (1) Leading Creek, Sec. 26, Rutland Twp. West Virginia. Cabell County: AEL 8 (2) Mud River, 1.6 km above mouth on Rt. 60. Lincoln County; AEL 6 (1) Trace Fork of Mud River, on Trace Fork Road, 1.6 km from jet. Rt. 34. Kentucky. Greenup County : EKU 163 (1) Tygarts Creek at St. Rt. 7, near Carter County line. Kentucky River. Kentucky. Carrol County: OSUM 42292 (5) Georges Creek, E of U.S. Rt. 42. Clay County; INKS 79022 (8) mouth of Jacks Creek at confl. with Red Bird River. INKS 86869 (2) Little Sexton Creek, 1.6 km E of Sextons Creek. Breathitt County: EKU 133 (5) Bear Branch Creek, upstream from confl. with Quicksand Creek. EKU 130 (3) Leatherwood Creek, 0.8 km downstream from head. EKU 131 (1) same as EKU 130. Garrard County; EKU 409 Indian Branch, 5 km NE of Crab Orchard. Estill County; INKS 79082 (2) Station Camp Creek, 4.8 km SE of Wagersville on St. Rt. 1209. Jessamine County ; OSUM (2) Hickman Creek at bridge 2.1km NW of Little Hickman, 10.7 km S of Nicholasville. Knox Counl-y: CU 41373 (8) Goose Creek, 9.8 km N of Girdler, on St. Rt. 11. Letcher County; EKU 853 (1) Whitaker Branch Creek trib. to Line Fork at Lilly Cornett Woods. Madison County: EKU 274 (3) Muddy Creek, Rt. 52 at Waco. EKU 225 (3) Tates Creek. Owsley County; INKS 79211 (3) Sexton Creek, 3.2 km W of St. Rt. 11. Rockcastle County; OSUM 37772 (4) Negro Creek at U.S. Rt. 150, 1.6 km S of Broadhead. EKU 102 (2) Red River Station 13. EKU 98 (2) Red River Station 21. Salt River. Kentucky. Anderson County; EKU 295 (5) Salt River, 8.0 km SW of U.S. 62 (Lawrenceberg). Casey County; OSUM 43436 (3) North Rolling Fork at St. Rt. 37. Jefferson County : OSUM 45337 (1) trib. of Ohio River, 3.2 km E of Louisville. Marion County; OSUM 37899 (4) Big South Fork, 9.6 km SE of Bradfordsville on St. Rt. 49. Indiana. Harrison County: OSUM 28679 (1) Buck Creek, 3.2 km S New Middleton. Washington County: OSUM 28647 (1) South Fork Blue River, 4.8 km E of Pekin. Green River. Kentucky. Allen County; EKU 372 (1) Bays Fork of Barren River, 6.4 km N of Scottsville. Barren County; EKU 1057 (1) Fallen Timbers Creek, Rt. 90. Casey County: OSUM 43488 (6) Green River at St. Rt. 198 (Yosemite). OSUM (1) Knoblick Creek near Yosemite. OSUM (3) Brush Creek, 3.2 km NW of Liberty. Edmonson County: OSUM (1) Dog Creek at St. Rt. 728. Green County: OSUM 19232 (1) Little Brush Creek, 3.6 km NW of Summerville at St. Rt. 61. Hart County; OSUM (1) Green River. Logan County : OSUM 19024 (1) Wiggington Creek, 1.7 km from mouth. Warren County: OSUM (1) Gasper River. Tennessee. Clay County: CU 37492 (3) Barren River at Hermitage Springs [Trace Creek at St. Rt. 52, LTM]. EKU (10) Hurricane Creek near Oak Grove, Rt. 52. Macon County: EKU 1093 (1) Long Fork Creek, Rt. 52. EKU (1) Salt Lick Creek at Rt. 52. Wabash River. Illinois. Coles County: INHS 18852 (2) Embarrass River, 7.2 km S of Charleston. Vermillion County: INHS 11739 (2) Salt Fork, 4.8 km N of Fairmont. INHS 11899 (3) Middle Fork River, 5.6 km NW of Hillery. Indiana. Fountain County: INHS 74189 (10) East Fork Coal Creek, 2.4 km SE of Veedersburg. Montgomery County: CU 51001 (4) Honey Creek at Rt. 47, E of Darlington. CU 51142 (1) Sugar Creek, 1.6 km W of Darlington. OSUM 27137 (1) Sugar Creek, 1.6 km N of Darlington. Putnam County: OSUM 27 430 (2) Plumb Creek, 12.8 km NE of Greencastle. OSUM 27192 (1) Raccoon Creek, 3.2 km E of Portland Mills. Randolph County: OSUM 29501 (2) Little White River, 19.2 km W of Winchester. Ohio. Mercer County : OSUM 19470 (1) Wabash River, Washington Twp., 15.5 km SW of Celina. Middle Cumberland River Kentucky. McCreary County: EKU 1124 (1) Little South Fork Cumberland River, 4.8 km downstream of Rt. 92. Pulaski County: OSUM 37800 (3) Fishing Creek, 8.0 km S of Eubank at St. Rt. 70. OSUM 43469 (5) Fishing Creek at St. Rt. 70, 0.4 km E of St. Rt. 635. Rockcastle County: EKU (2) Rockcastle River at St. Rt. 490. EKU 216 (3) Todd Branch Clear Creek. OSUM 47225 (3) Clear Creek, 3.2 km S of Disputania, 14.7 km NW of Mt. V e r n o n . Lower Cumberland River. Kentucky. Christian County: CU 23024 (2) West Fork Red River, 1.7 km E of Hensleytown, St. Rt. 115. Tennessee. Cannon County : INHS 75953 (1) Saunders Fork [Caney Fork], 4.8 km S of Aubornton. Montgomery County: OSUM 48831 (6) Spring Creek at Spring Creek Campground, 12.8 km NNE of Clarksville. Stewart County: CU 47399 (5) South Cross Creek, Crittendon. Sumner County: OSUM 19032 (3) Red River, 5.4 km SW of Portland. Warren County: CU 51764 (2) Charles Creek, 1.3 km N of jet. 50 and 70, S of McMinnville. Wilson County; INHS 75586 (4) trib. to Helton Creek, 0.4 km W of Cottage Home.
Missouri River. Kansas. Bourbon County; CU 42234 (4) Little Osage River at St. Rt. 31, 8.0 km E of Mapelton. Dade County: OSUM 39249 (8) Limestone Creek at South Greenfield. Miller County; OSUM 44010 (12) Big Tavern Creek, 4.0 km SE of Iberia. Pike County: CU 42187 (2) Peno Creek on Hwy. 61, 18.4 km N jet. 61 and 54. Arkansas River. Arkansas. Crawford County; WJC-TMB 74-14 (13) Clear Creek at St. Rt. 162. WJC-TMB 71-22 (10) Lees Creek at St. Rt. 59. Scott C o u n t y ; WJC-TMB 74-32 (2) Fourche La Fave River, 1.6 km E of Boles. Washington County; CU 35569 (1) Clear Creek at Savoy, 1.0 km NE confl. with Illinois River. Missouri. Barry County; CU 37397 Shoal Creek on Rt. B, 12.8 km W of Purdey. CU 38664 (1) Shoal Creek at Rt. 97, 12.8 km W of Purdey. CU 32922 (4) George Branch of Shoal Creek SW of Purdey. Jasper County; CU 55242 (1) at Sarcoxie [Spring River drainage]. Oklahoma. Cherokee County; CU 17899 (1) Fourteen Mile Creek [Neosho drainage]. Kansas. Cherokee County; EKU 245 (4) Spring River, 4.8 km W of U.S. 171. Illinois River. Illinois. Bureau County; INHS 12734 (8) Spring Creek, 1.6 km E of Ladd. Marshall County ; INHS 10539 (4) Sandy Creek St. Rt. 89, 2.4 km S of Magnolia. Rock River. Illinois. Winnebago County; OSUM 38979 (5) North Branch Otter Creek, 6.4 km NW of Durand. Wisconsin River. Wisconsin. Marathon County; OSUM 516477 (11) Plover River at Co. Rt. C-Y. Vernon County; UWSP 5870 (17) Billings Creek at Co. Rt. F.
"Ozark-Tennessee Group"
Lower Tennessee River. Tennessee. Coffee County; CU 42034 (1) Duck River, 2.2 km NW of Manchester on Rt. 41. CU 42152 (*2) Garrison Fork of Duck River at dirt road, crossing, 1.0 km N of Beech Grove. Giles County; CU 4653 (4) trib. of Elk River, 17.6 k m SW of Pulaski, Rt. 11. Henery County; OSUM 34881 (1) Eagle Creek upstream from U.S. 79. Stewart County; OSUM 34880 (4) Standing Rock Creek, 1.3 km S of U.S. 79. Alabama. Lauderdale County; OSUM 43020 (4) Cypress Creek at St. Rt. 16. OSUM 42998 (1) Middle Cypress Creek, 0.3 km S of Cloverdale on St. Rt. 157. OSUM 34387 (1) Little Cypress Creek, 9.6 km off St. Rt. 20 on Jackson Rd. OSUM 43013 (1) Little Cypress Creek downstream of Co. Rt. 8. Upper Current River. Missouri. Shannon County; P-64-123 (7) Current River at mouth of Sinking Creek. Lower Current River. Missouri. Ripley County; P-64-119 (4) Current River at Doniphan. North Fork White River. Missouri. Ozark County; OSUM 42037 (15) Pine Creek at St. Rt. 181 at Zanoni, in spring op W side of Pine Creek downstreamm from bridge. Upper White River. Missouri Webster County; CU 38416 (5) W. Finley River at Hwy. Z, about 12.8 km S of F o r d l a n d . Arkansas. Madison County; CU 36347 (2) Bice Creek, trib. of Kings River, 2.0 km NE of Forum.