QL 638 .P4 S53 1999 Scientific Papers Natural History Museum The University of Kansas 28 July 1999 Number 12:1-16 A Reexamination of the Phylogenetic Relationships of the Sand Darters (Teleostei: Percidae) By Kate A. Shaw', Andrew M. Simons^ and E. O. Wiley' 'Division oflchthyologi/, Natural Histonj Museum, and Department of Ecology and Evolutionary Biology, The University of Kansas, Lawrence, Kansas 66045-2454, USA ami ^Bell Museum of Natural History and Depmrtment of Fisheries and Wildlife, The University of Minnesota, 1980 Fohvell Avenue, St. Paul, Minnesota 55108-6124, USA CONTENTS ABSTRACT 1 INTRODUCTION 2 Acknowledgments 3 MATERIALS AND METHODS 3 RESULTS 6 DISCUSSION 10 LITERATURE CITED 12 APPENDIX 1. Specimens Examined 13 APPENDIX 2. Biochemical Data 14 APPENDIX 3. Character Matrix 15 ABSTRACT Phylogenetic relationships among the Boleosoma group of darters were examined for allozymic variation at 25 presumptive gene loci alone and in combination with 29 morphological char- acters. Qualitative analyses of allozymic and morphological variation resulted in two most parsimoni- ous trees showing the following relationships: Tree 1. — (Etheostoma davisoni ((£. longimanum, E. nigrum) (E. vitreum (£. clarum (£. pellucidum (E. meridianuni (E. vivax (E. bifascia, E. beanii)))))))); Tree 2. — (E. davisoni (E. nigrum (E. longimanum (E. vitreum (E. clarum (E. pellucidum (E. meridianuni (E. z'iz'ax (E. bifascia, E. beanii))))))))). Quantitative analyses that incorporate an estimate of allelic frequencies were © Natural History Museum, The University of Kansas ISSN No. 1094-0782 QL 638 .P4 S53 1999 Scientific Papers Natural History Museum The University of Kansas 28 July 1999 Number 12:1-16 A Reexamination of the Phylogenetic Relationships of the Sand Darters (Teleostei: Percidae) By Kate A. Shaw', Andrew M. Simons^ and E. O. Wiley' ^Division of Ichthyology , Natural History Musciiin, and Department of Ecology and Evolutionary Biology, The University of Kansas, Lawrence, Kansas 66045-2454, USA and 'Be// Museum of Natural History and Department of Fisheries and Wildlife, The University of Minnesota, 1980 Fohvell Avenue, St. Paul, Minnesota 55108-6124, USA CONTENTS ABSTRACT 1 INTRODUCTION 2 Acknowledgments 3 MATERIALS AND METHODS 3 RESULTS 6 DISCUSSION 10 LITERATURE CITED 12 APPENDIX 1. SPEaMENS Examined 13 APPENDIX 2. Biochemical Data 14 APPENDIX 3. Character Matrix 15 ABSTRACT Phylogenetic relationships among the Boleosoma group of darters were examined for allozymic variation at 25 presumptive gene loci alone and in combination with 29 morphological char- acters. Qualitative analyses of allozymic and morphological variation resulted in two most parsimoni- ous trees showing the following relationships: Tree 1 . — (Etheostoma davisoni ((£. lon^imanum, E. nigrum) (E. vitreum (£. claruin (£. peltucidum (£. meridiainim (£. znz'ax (£. hifascia, E. beanii)))))))); Tree 2. — (£. davisoni (E. nigrum (£. longimanum (£. vitreum (£. clarum {£. pellucidum (£. meridianum {£. vivax (E. hifascia, E. beanii))))))))). Quantitative analyses that incorporate an estimate of allelic frequencies were © Natural History Museum, The University of Kansas ISSN No. 1094-0782 ScFENTiFic Papers, Natural History Museum, The University of Kansas used to discriminate among trees and showed that Tree 1 was 0.598 FREQPARS units shorter than Tree 2. There was both congruence and complementarity in the support provided by allozymic and mor- phological datasets. The sister-group relationship and allopatric distributions of £. beanii and £. bifascia are consistent with their origin resulting from vicariant speciation associated with the origin of the Mobile Basin. Key words: Etheostoma, Ammociypta, allozymes, historical biogeography, phylogeny INTRODUCTION The sand darters are slender, elongate, translucent darters known for their habit of burying themselves in sandy substrates. Six species of sand darters are currently classified in the subgenus Ammocri/pta: Etheostoma peUucidum, E. beanii, E. vivax, E. claruni, E. bifascia, and £. meridianum. The taxonomic history of Ammocrypta began with its use as a generic name by Jordan (1877) with his descrip- tion of y4. beanii. Three additional species now included in the subgenus Ammocrypta (originally described as Pleurolepis pellucida, A. vivax, and A. clam) were described by 1886 (Agassiz, in Putnam, 1863; Hay, 1883; and Jordan and Meek, 1885, respectively). Bailey and Gosline (1955) added Crystallaria asprelln to the genus Ammocrypta, but placed it in a monotypic subgenus Crystallaria. Although Bailey and Gosline (1955) argued for a classification of darters containing three genera (Percina, Anunocn/pta, and Etheostoma) that was generally accepted, a few authors (e.g., Moore, 1968; Miller and Robison, 1973) continued to rec- ognize Crystallaria as a monotypic genus. Williams (1975) described two additional species within the subgenus Ammocrypta. Within this subgenus, his A. beanii group con- sisted of the species with a few scale rows (£. beanii, E. bifascia, and £. clariim) and his A. pellucida group contained the partially to almost completely scaled species (£. peUucidum, E. meridianum, and £. vivax). Williams (1975) considered the genus Ammocrypta more closely allied to the subgenus hnostoma of the genus Percina than to Etiieostoma. However, Page and Whitt (1973a) argued that Percina was monophyletic and sug- gested that Ammocrypta was related to Etheostoma; this con- clusion was based on a unique LDHB4 isozyme found only in Perciim. (Their assessment of the polarity of this charac- ter is unclear — in their Fig. 4, p. 6, Etheostoma and Ammocrypta are united by a derived LDH B4 isozyme.) Page and Whitt (1973b) stated that the TO isozyme shared by £. (Vaillantia) chlorosoma and A. pellucida indicated that Ammocrypta was more closely related to Etheostoma than to Percina. Simons (1989, 1991) removed Crystallaria from Ammocrypita and hypothesized that the phylogenetic posi- tion of Crystallaria was basal to Etheostoma and Percina. Simons (1989, 1992) subsumed the remaining six species oi Ammocrypta within the genus Etheostoma and suggested that Etheostoma (loa) vitreum was the sister group to the subgenus Ammocrypta. This change resulted in four sub- genera being included in the Boleosoma group of Etheostoma: Boleosoma, loa, Vaillantia, and Ammocrypta. Simons' hypoth- esis of relationships within the subgenus Ammocrypta dif- fered somewhat from those postulated by Williams (1975). Simons (1992) did find support for the £. peUucidum group; £. meridianum and £. peUucidum were sister taxa, and £. vivax was the sister group to this pair. However, Simons (1992) did not find support for the £. beanii group; instead, the species pair £. beanii-E. bifascia was the sister group to the £. peUucidum group and £. clarum was the sister group to all other members of the subgenus Ammocrypta. The relationships hypothesized by Simons (1989, 1991, 1992) and based on phylogenetic analyses, have not been widely accepted. For example, Etnier and Starnes (1993), Jenkins and Burkhead (1994), Mettee et al. (1996), and Pflieger (1997) all continued to recognize the genus Ammociypta in their state guides (Tennessee, Virginia, Ala- bama, and Missouri, respectively). Although Jenkins and Burkliead (1993) alone justified their retention of the ge- nus Ammocrypta, in no case is an explicit alternative phy- logenetic analysis or hypothesis presented. Wood and Mayden (1997) however, found support for a sister-group relationship between Etheostoma beanii and Crystallaria in maximum parsimony analyses and in their most parsi- monious FREQPARS tree, but they did not make any taxo- nomic recommendations based on this result. In the absence of an explicit alternative phylogenetic analysis of more than two taxa, we conrinue to consider members of Ammocrypta, members of the genus Etheostoma, subgenus Ammocn/pta. The particular relationships of these species are of in- terest to biogeographers. Wiley and Mayden (1985) fol- lowed the relationships suggested by Williams (1975) and postulated that the distribution of members of the Etheostoma beanii group was the result of a western vicariance event loosely associated with the Mississippi River. Etheostoma clarum is found in the Mississippi River Basin and drainages to the west, and the presumed sister taxon (£. beanii-E. bifascia) is found east of the Mississippi River mainstem. Wiley and Mayden (1985) also suggested that the distributions of two other groups could be attrib- Phylogenetic Relationships of Sand Darters uted to vicariance events involving the Mobile Bay Basin. Ethcostoina beaiiii is found in the Mobile Basin and west- ward to the Mississippi River mainstem, whereas E. bifascia is only found east of Mobile Bay. A similar situation is found in the £. pelliicidiim group. Etheostoma meridiaiiiim is a Mobile Basin endemic, whereas E. pellucidum and E. vivax are found in drainages north and west of the Mobile Ba- sin. Simons' (1992) results did not substantiate the vicariance event associated with the Mississippi River or a vicariance event between E. meridiniiuni and the species pair E. vivax-E. pellucidum. The main purpose of this study is to reexamine the rela- tionships of members within the subgenus Aiumocrypta and the controversial relationship of the subgenera Ammocnjpta and loa by including new data from electrophoretic studies. A secondary purpose of this study is to reexamine the bio- geographic liistory of the subgenus Ammocnjpta. Acknowledgments We thank G. Harp (Arkansas State University), L. Page, P. Ceas, and C. Johnston (Illinois Natural History Survey), F. Pezold (Northeast Louisiana University), R. Jenkins (Roanoke College), H. Bart and M. Taylor (Tulane Univer- sity), R. Mayden (University of Alabama), R. Cashner and J. Grady (University of New Orleans), and F. Cross, T. Schmidt, D. Siegel-Causey, and K. Toal, III (University of Kansas) for their help collecting specimens used in this project. We also thank S. Layman for allowing us to use his unpublished results on tuberculation in darters and R. Wood for advice with the data analyses. This project was supported by grants from the General Research Fund, University of Kansas (KU 3365) to E. O. Wiley, and the National Science Foundation (BSR 8722562) to E. O. Wiley and D. Siegel-Causey. MATERIALS AND METHODS Taxa examined. — All six members of the subgenus Ammocnjpta were included in the ingroup. Outgroups were chosen using the hypothesis of Simons (1992) as a guide. Etheostoma nigrum and E. lo}igimaiutin were included as representatives of subgenus Boleosoma. These species have the potential to act as the first (close) taxonomic outgroup. Within Boleosoma, his clade consisting of £. nigrum, E. olmstedi, and E. perlongum is represented here by E. nigrum. The species pair E. longimanum-E. podostcmone is repre- sented by E. longimaiium. Although E. )iigrum and £. longimanum were included in the analysis as taxa outside the taxonomic group of interest {Ammocri/pta and E. vit- reum), they were not designated as outgroups in the analy- ses. The second (distant) taxonomic outgroup was £. davisoni. This species represents the species pair that make up the subgenus Vaillantia and was designated as the sole outgroup in the analyses. Enzyme electrophoresis. — Fishes were collected by seining, frozen immediately in liquid nitrogen, transported to the laboratory, and stored at -70° C for up to 5 yr (see Appendix 1 — Specimens Examined). Skeletal muscle, liver, and brain/eye tissues were dissected and homogenized separately in a 1:1 (v:v) mixture of tissue and 0.01 M Tris, 0.001 M EDTA, and 0.001 M mercaptoethanol, pH 6.8. Homogenates were centrifuged at 15,000x ^^ for 10 min at 5° C. Within 72 h, the supernatant fractions were electro- phoresed at 5° C on horizontal starch gels composed of 12% hydrolyzed potato starch (Starch Art Corp.). His- tochemical staining protocols did not differ substantially from those of Murphy et al. (1990). Enzyme nomenclature follows the recommendations of the International Union of Biochemistry Nomenclature Committee (1984), and locus nomenclature follows the rec- ommendations of Buth (1983). Enzymes, loci, tissue sources, and electrophoretic conditions are listed in Ap- pendix 2. Twenty-seven presumptive gene loci were visu- alized by histochemical staining. Electromorphs for each locus were coded a, b, c, etc., in order of increasing anodal mobility. These designations are relevant to this study only. Two loci were fixed (M-Icdh-A and Ldh-A) for all taxa sampled and not included in the analyses. Two loci were variable only within single species (Ldh-B within Etheostoma meridianum and Tpi-B within E. vivax), and these autapomorphies were not considered further. The remain- ing 23 variable loci were considered independent trans- formation series (TS), and each allele was considered a unique state (TS 1-TS 23, Appendix 3). Morphological characters. — Twenty-nine morpho- logical transformation series originally described by Simons (1992) were included in this study. Many of the descriptions of these transformation series have been slightly revised, and for this reason they are presented below. Transformation series and state numbers refer to those presented in Appendix 3 (TS 24-TS 52). TS 24. — Ascending process of the premaxilla: (0) per- pendicular to the alveolar process or (1 ) reclined posteriorly. TS 25. — Maxillary process of the premaxilla: (0) not elongate or (1) elongate and enlarged. TS 26. — Premaxillary socket of the maxilla: (0) V- shaped, with the lateral and medial walls approximately equal in length or (1) U-shaped, with the lateral wall curv- ing medially and longer than the medial wall. TS 27. — Palatine teeth: (0) present or (1) absent. TS 28. — Notch posteroventral to the articular process of the quadrate: (0) shallow to absent or (1 ) cut deeply into the body of the quadrate. TS 29. — Body of the quadrate: (0) rounded, with a SciENTinc Papers, Natural History Museum, The University of Kansas notch between the body and the posterior process of the quadrate or (1) rectangular, without a notch. TS 30. — Hyomandibular struts: (0) present as cruci- form thickenings within the hyomandibula or (1) extremely reduced to absent. TS 31. — Descending process of the hyomandibula: (0) long, extending beyond the preopercular groove or (1) short, terminating at the end of the preopercular groove. TS 32. — Hyomandibular spur: (0) absent or (1) present. TS 33.— Ventral plate of the urohyal: (0) flattened, with the insertion for the urohyal-hypohyal ligaments directed anteriorly from the anterior surface or (1 ) curved, with the insertion sites directed anteroventrally from the ventral surface. Simons (1992) considered this character homoplas- tic, because in his analysis it supported the monophyly of the subgenera Boleosoma and loa. TS 34. — Articular process for the interhyal on the pos- terior ceratohyal: (0) present or (1) absent. TS 35. — Posterior margin of the preopercle: (0) smooth or (1) serrate with a few points projecting beyond the mar- gin of the bone. TS 36. — Notch in the anterior angle of the preopercle: (0) present, roofing the articulation for the interhyal or (1) absent. TS 37. — Opercular spine: (0) present or (1) absent. TS 38. — Opercular strut: (0) strong, extending from the hyomandibular articulation to or almost to the posterior margin of the opercle or (1 ) greatly reduced, extending less than half the distance to the margin. TS 39. — Posterodorsal extension of the subopercle: (0) elongate and filamentous or (1) truncated near the dorsal margin of the opercle. TS 40. — Mesethmoid: (0) thick and expanded anteri- orly, extending anteriorly beyond the lateral ethmoids or (1) thin and concave anteriorly, not extending beyond lat- eral ethmoids. This character was considered homoplastic by Simons (1992), because in his analysis it supported group- ing together the subgenera Vailhmtin, Boleosoma, and Ion. TS 41. — Maxillary ligament insertion: (0) on two dorso- lateral projections of the mesethmoid, (1) on a single dorsomedian knob of the mesethmoid, or (2) on two dorsomedian ridges of the mesethmoid. Simons (1992) con- sidered this character homoplastic because In his analysis it supported grouping together the subgenera Boleosoma and Ion. TS 42. — Vomerine teeth: (0) always present or (1) usu- ally absent. TS 43. — Membrane bone on the lateral margin of the nasal: (0) extensive, overlying the olfactory capsule or (1) reduced to a thin slip along the canal. TS 44. — Remnant of the lateral line canal of the supracleithrum: (0) present or (1) absent. TS 45. — Postcleithrum 2: (0) present or (1) absent. TS 46. — Longitudinal struts on the anal proximal pterygiophores: (0) present or (1) absent. TS 47. — Process for the insertion of the m. infracarijinlis medius on the anterior face of the first anal pterygiophore: (0) present or (1) absent. This character was considered homoplastic by Simons (1992), because in his analysis it supported a group consisting of Etheostoma bennii, E. chlorosoma, E. davisoiii, E. stigmaeum, and E. jessiae. TS 48. — Swollen, thickened tips of fin spines and rays of breeding males: (0) absent or (1) present, such that the tip of the pelvic spine is covered by a large fleshy knob and the ventralmost pectoral rays and pelvic rays are swol- len and thickened at the tips. TS 49. — Body squama tion: (0) almost complete or (1) reduced laterally to a few rows of scales. TS 50. — Tubercles on pelvic fins of breeding males: (0) present or (1) absent. This transformation series and the next one were both part of a single transformation series in Simons' (1992) analysis. Although tuberculation is usu- ally considered a single multistate variable, the presence of tubercles on many areas of the body occur independently across all taxa of darters and are considered different trans- formation series in this study. That is, the presence of tu- bercles on the pelvic fins does not show a one-to-one cor- respondence with the presence of tubercles on the anal fin. Tubercles on the pelvic fins of breeding males have been observed for many darters, including: Percina evides, P. pal- mnris, P. shumardi, P. vigil, P. auraiitiaca, P. copelandi, Crystallaria asprella and, within the genus Etheostoma, all members of the subgenera Allohistium, Ammocrypta, Doration, and loa, and E, chlorosoma, E. punctulatum, E. boschtiiigi, E. cragini, E. pallididorsum , E. aiistrale, E. hopkinsi, E. spectabile, E. luteovinctum, E. serrifer, E. gracile, E. zonifer, E. fiisiforme, E. sahidae, E. collis, E. pweliare, E. microperca, and £. fonticola (Collette, 1965; Bailey and Etnier, 1988; Jenkins, 1971; and S. R. Layman, pers. comm.). TS 51. — Tubercles on the anal fins of breeding males: (0) absent or (1) present (Collette, 1965; Bailey and Etnier, 1988; Jenkins, 1971; and S. R. Layman, pers. comm.). Anal fin tubercles have been observed in Etheostoma parvipinne, E. fricksium, E. radiosiim, E. whipplei, and E. trisella; these five taxa lack pelvic fin tubercles. Of the taxa that exhibit pelvic fin tubercles (listed above), only E. pellucidum, E. meridinniim, and E. vitreum lack anal fin tubercles. All mem- bers of the subgenera Etheostoma, Ulocentra (sensu Bailey and Etnier, 1988), and Boleosoma lack anal (and pelvic) fin tubercles, as does E. davisoiii. TS 52. — Shape of the female genital papilla: (0) coni- cal, (1) cupped, or (2) flat and bilobed. Variation in the shape of the female genital papilla is widespread among members of Etheostoma; however, most species have short. Phylogenetic Relationships of Sand Darters conical papillae. Of the taxa included in this analysis, fe- male £. davisoni have cup-like papillae (Howell, 1968), fe- male £. longimanum and £. nigrum have bilobed, flattened papillae (Cole, 1967), and all females of the subgenera to? and Atiiiuocrifpta have conical papillae. Some of the transformation series used by Simons (1992) are not included here, because they are uninforma- tive in this study. They were either synapomorphic for the entire Boleosotna group (his characters 3 and 4), provided support for groups within the subgenus Boleosouia (his characters 7, 24, 25, and 32), supported the subgenus Vaillantia (his characters 10, 35, 36, and 38 [in part]), or were autapomorphic for Etheostoma vitreiiin (his characters 29 and 37 [in part]). Phylogenetic analyses. — The underlying methodol- ogy for this study is phylogenetic parsimony analysis (Hennig, 1966; Kluge and Farris, 1969; Wiley, 1981; Farris, 1983; Churchill et al., 1985; Farris and Kluge, 1985, 1986). The precept that all available evidence must be brought to bear on any statement about relationships (Kluge, 1989) is also adhered to. The outgroup comparison method has been shown to be the most comprehensive for polarizing hypotheses of transformation (Stevens, 1980; Farris, 1982; Kluge, 1984, 1985; Brooks and Wiley, 1985), and conse- quently, it was used in this study. A variety of analyses were performed on the allozyme data alone and on the allozyme data in conjunction with the morphological data: (1) analyses of allozyme data us- ing BIOSYS-1 (Swofford and Selander, 1981), (2) qualita- tive analyses of allozyme and combined allozyme and morphological data using PAUP 3.1.1 (Swofford, 1993), and (3) quantitative analyses of the topologies produced from the preceding BIOSYS-1 and PAUP 3.1.1 analyses utilizing FREQPARS (Swofford and Berlocher, 1987; Swofford, 1988). For BlOSYS-1 analyses, genotype frequencies from the allozyme data were used to construct topologies follow- ing a variety of methods: Rogers (1972) genetic distance, modified Rogers distance (Wright, 1978), Prevosti distance (Wright, 1978), Cavalli-Sforza and Edwards (1967) chord distance, Cavalli-Sforza and Edwards (1967) arc distance, and Edwards (1971, 1974) "E" distance. These genetic dis- tances were summarized in distance- Wagner trees (Farris, 1972) using Swofford's (1981) Multiple Addition Criterion Procedure and were rooted with the single outgroup, Etheostoma davisoni. The resultant topologies were then used as input trees for the third group of analyses (FREQPARS). Two groups of PAUP analyses were performed. Any transformation series with three or more character states was polarized if possible, but unordered. Only Etheostoma davisoni was designated as an outgroup, because then the most parsimonious arrangement of all the data will deter- mine whether E. vitreum is the sister group to one of the following: the subgenus Annnocrypta, any species or com- bination of species of Ammociypta, the subgenus Boleosoma, a clade consisting of the subgenera Boleosoma and Ammocn/pta, the species pair £. nigrum-E. longimanum, the species E. nigrum, or the species £. longimanum. This tests Simons' (1992) assertion that E. vitreum is the sister group to the subgenus Ammocrypta. Allozyme data were coded using each locus as the transformation series and the allelic arrays as the charac- ter states. The order or cost of transformation from one character state to another was directed by step matrices (Mabee and Humphries, 1993). Two kinds of step-matrix files (available from the authors upon request) were con- structed. In the first, allozyme transformation series were directed by a series of step matrices that contained only character states observed in the study taxa. This allows coding of all observed allelic combinations; hypothetical ancestral character states at interior nodes of resulting phylogenetic trees were limited to character states observed in study taxa. The second step-matrix file contained a single, large step matrix of character states observed in extant taxa as well as all other possible allelic arrays. Be- cause of restrictions on the number of characters allowed by PAUP, this approach can only accommodate five or fewer alleles (31 states) per locus. Therefore the original data matrix had to be reduced in size. This reduction was accomplished by deleting autapomorphic alleles from the data matrix for those taxa in which they occurred. For PAUP analyses the most parsimonious and several near most parsimonious topologies were saved for FREQPARS analysis. Two groups of FREQPARS analyses were performed. The first included only the allozyme frequencies. The sec- ond included both allozyme frequencies and morphologi- cal characters. Morphological characters were treated as if they were fixed electromorphs for the taxa in which they occur. That is, the presence of a particular morphological character state in a particular taxon was considered ho- mozygous and present in 100% of the population sampled. All topologies produced using BIOSYS-1 and PAUP analy- ses were tested using FREQPARS. Simons' (1992) tree was also tested in this fashion. Tree lengths resulting from these FREQPARS analyses were used as the final arbiter between competing hypotheses of phylogenetic relationships of the ingroup taxa. This is because these analyses include all available data pertaining to these species (i.e., both allozyme frequency data and morphological character states) and they provided greater discrimination among competing hypotheses. Scientific Papers, Natural History Museum, The University of Kansas RESULTS Observed genotypic frequencies are provided in Table 1. All taxa were polymorphic for at least two loci, most taxa were polymorphic at 6-8 loci, and Etheostoma beanii was polymorphic at 10 loci. Analysis of the allozyme frequency data using BIOSYS- 1 produced two trees. These differed from each other only in the placement of Etheostoitm nigrum. Rogers genetic dis- tance, modified Rogers distance, Prevosti distance, and Edwards "E" distance methods all resulted in a topology wherein £. nigrum was the sister taxon to the subgenus Ammocrypta (Fig. 1). The Cavalli-Sforza and Edwards chord distance and Cavalli-Sforza and Edwards arc distance methods resulted in identical topologies, with £. nigrum the sister taxon to the species pair £. longimnnum-E. vit- reum (Fig. 2). PAUP analysis of the allozyme data alone, using a se- ries of step matrices that only allowed character states ob- served in the study taxa, produced two equally parsimo- nious trees (TL = 166 steps). Each contained a monophyl- etic subgenus Ammocrypita that was the sister group to Etheostoma nigrum, E. longimanum, and E. vitreum. In one tree (Fig. 3), £. vitreum was the sister group to the species pair £. longimanum-E. nigrmn, and in the other (Fig. 4), £. nigrum was the sister group to the species pair £. longimanum-E. vitreum. PAUP analysis of the allozyme data alone, with the changes in transformation series directed by a single, large step matrix produced a single most par- simonious tree (TL =132 steps) with a monophyletic sub- genus Ammocrypta that was the sister group to a group consisting of £. vitreum and the species pair £. longimanum- E. nigrum. This tree has the same topology as that in Fig- ure 3. Total evidence PAUP analysis using a series of step matrices found a single most parsimonious tree (TL = 207) with a monophyletic subgenus Ammoeri/pta as the sister taxon to Etheostoma vitreum; these taxa formed the sister taxon to the species pair, £. longimanum-E. nigrum (Fig. 5). Total evidence PAUP analysis using the single, large step matrix resulted in two most parsimonious trees (TL = 174) with a monophyletic subgenus Ammocrypta that was the sister group to £. vitreum. In one tree (Fig. 6), £. nigrum is the sister group to a monophyletic group containing £. longimainim, E. vitreum, and the subgenus Amnwcn/pta. In the other tree (Fig. 5), the species pair £. longimanum-E. nigrum form a monophyletic sister group to £. vitreum and the subgenus Anunocrypta. The most parsimonious tree examined in the FREQPARS analyses based on the allozyme data alone was 113.442 FREQPARS units long. This tree is isomorphic with one of the two most parsimonious trees based on the PAUP .# r.0^ A<^ .<>^ ^ A<^ Fig. 1 . Tree topeilogy for ten species of Etheostoma resulting from BIOSYS-1 analysis of the allozyme data alone using Rogers genetic dis- tance, modified Rogers genetic distance, Prevosti distance and Edwards "E" distance methods. This topology is 116.006 freqpars units when only the allozyme data are considered and 21 6.004 freqpars units for the total- evidence data. /, <'"^-"^^^*°' .// /„. <- ■ <<.- ■ t-^ <<>■ <<>■ <<;■ <<-■ <> /.//^%/z/.<.*° K)®'' x^"^" 4^^^ <^^■ /" c>^\ ^^^\ <>^^: \o^: 6^ Fig. 3. The first of two most parsimonious trees (166 steps) for ten species of Etheostoma resulting from PAUP analyses of the allozyme data alone using a series of step matrices. This is also the single most parsi- monious tree (132 steps) resulting from PAUP analyses of the allozyme data alone using a single large step matrix. This topology is 113.442 freqpars units when only the allozyme data are considered and 205.441 freqpars units for the total-evidence data. p.^' .<^ .<^ c>y .<^ /^^ Fig. 4. The second of two most parsimonious trees (166 steps) for ten species of Etheostoma resulting from PAUP analyses of the allozyme data alone using a series of step matrices. This topology is 1 1 3.840 freqpars units when only the allozyme data are considered and 209.839 freqpars units for the total-evidence data. Fig. 5. The single most parsimonious tree (207 steps) for ten species of Etheostoma resulting from PAUP analyses of the total-evidence data us- ing a series of step matrices. This is also the first of two most parsimonious trees (174 steps, the second is Fig. 6) resulting from PAUP analyses of the total-evidence data using a single large step matrix. This topology is 117.376 freqpars units when only the allozyme data are considered and 193.375 freqpars units for the total evidence. Unambiguous character-state support (see text) resulting from PAUP analysis of the total-evidence data using a series of step matrices for each of the labeled nodes is as follows: 1. — TS 1 (presence of c and loss of b), TS 13 (loss of d), TS 33:1 (curved ventral plate of the urohyal), TS 48:1 (thick- ened fin tips of breeding males), and TS 52:2 (female genital papillae flat and bilobed). 2.— TS 15 (loss of a), TS 19 (presence of c and loss of b), TS 22 (loss of d or f, depending on which step matrix was used), TS 24:1 (poste- riorly reclined ascending process of the premaxUla), TS 27:1 (palatine teeth absent), TS 29:1 (rectangular body of the quadrate), TS 30:1 (hyomandibular struts extremely reduced to absent), TS 34:1 (articular process for the interhyal on the posterior ceratohyal absent), TS 42:1 (vomerine teeth usu- ally absent), TS 44:1 (remnant of the lateral-line canal on the supracleithrum absent), and TS 50:0 (tubercles on pelvic fins of breeding males present). 3. — TS 7 (presence of b and loss of a), TS 1 1 (presence of d and loss of a), TS 13 (presence of e and loss of b), TS 14 (loss of c and presence of d or f, depending on which step matrix was used), TS 22 (presence of b and loss of e), TS 31:1 (descencling process of hyomandibula short), TS 32:1 (hyomandibular spur present), TS 36:1 (notch in the anterior angle of the preopercle absent), TS 40:0 (mesethmoid thick and expanded anteriorly), and TS 43:1 (membrane bone on the lateral margin of the nasal reduced). 4. — TS 35:1 (posterior margin of the preopercle serrate) and TS 39:1 (posterodorsal extension of the subopercle truncated). 5. — TS 1 4 (presence of d and loss of e or f, depending on which step matrix was usecf) and TS 23 (presence of c and/or the loss of b, depending on which step matrix was used). 6. — TS 22 (presence of a and/or loss ofb, depending on which step-matrix was used) and TS 51:1 (tubercles on the anal fins of breeding males present). 7. — TS 1 (presence of c and loss of b), TS 2 (presence of f and g and loss of d), TS 4 (presence of b and loss of c), TS 21 (presence of c), TS 26:1 (premaxillary socket of the maxilla U-shaped), TS 28:1 (notch posteroventral to the articular process of the quadrate deeply cut), TS 35:0 (posterior margin of the preopercle smooth), TS 38:1 (opercular strut greatly reduced), TS 45:1 (postcleithrum 2 absent), TS 46:1 (longitudinal struts on the anal proximal pterygiophores absent), and TS 49:1 (body squamation reduced laterally). A. — allozymic character states that support the node, M. — morphological character states that support the node. Scientific Papers, Natural History Museum, The University of Kansas Xi to X3 Xi XI 01 QJ T3 OJ ^o rj 00 00 00 ro (N (Ti vo r^ 00 XI XI XI X2 ^ ^ XI Xi Xl u Xi X) n; (l; OJ LTj ^- '^f X — x: X :=; Xl Xl n; ra u Xl Xl Xl Xl QJ 01 T3 HI a\ m r— LDf^tN'— ' iri LT) 1-H Tf LTl Tf ^ !S >£i^ bc 6C-^ ^ bO Xl Xl Xl Xi Xl T3 --^H 01 OJ T3 01 ^rjfN CNim fj,— ^ in in in in in in-*'— in ^'* S S ^^ ^^ T3 6C Xi Xi re Xi ^ Cj Xl re Ti 01 Xi u u re T3 73 X) XJ re re XJ u Xl re T3 Ol Xl Xi u re \?) in in in in in in , — 1 1 — 1 ro s ^ nl f^ in in C C ' — ' ' ^ C^ ^^ ^ !3^ •o Xi Xi Xl u Xl u Oi Oi OJ 01 01 re Xi u re T3 XI Xi Xi u Xi u Xl u 01 U 01 re re u re s ^ in in in in in in , — 1 ni s s ^ in ^« ^ in ^ ' — c; ' — c ■a -^ Xi Xl Xl u SJ o Xi Xi 73 01 Ol re u u re Xi "0 Xi Xi Xi u Xl y re Xl Xl ii 01 re Xl u re in in in in in . — 1 ^ in in in in in 5^ ^ ^ — — ^^ ■ — ' ' — "^ ' — ^ — ^^ ^^ ^ — ^^ T3 Xi Xi Xi u Xl Xi u T3 Oi re u re Xl T3 Xi Xl ^ u re .Xl u T3 01 re u re re LO in ^ ?■ in in 5^ in in ^ So 3 5' S ^ IT) (—1 ^^ T— 1 ■^^ ^-' ^-' ■ — ■ — ' — " — ^^ ^^ Xi Xi XI Xl u Xi Xi u -a "Q ■-" ^ Xl u u u re Xi re Xl Xl u re Xi u •T3 73 73 sr; Xi Xl Xi u re en 01 C/i 01 x: LT; in in m, s t^ s LTi in I — 1 m T-H in in in o in c 1—1 T-H (—1 P ^ ■ — - ^ ^^ ^-' ^-' ' — ^^ ^^ re O Xi Xl Xl Xl u u re u 73 73 iK re OJ XI X re U Xi Xl Xl Xi Xl u re u -a re 01 re Xl re c 01 vh re 1 o u , , ^ ^ , , . , , , , ^ , ^ ^ , , . , — , , — , , — , , — , , — , ITj in IT, in (N ^o t^ in in in in in ITj LTj ^^ — ' — ' ^-' ^-' ^^ ^^ ^-^ ■~-' ^~^ ^ — n: Xi Xj Xi U CJ xi Xl Xi T3 Xl 73 re re ^ fO Xi Xl Xl re u u Xl Xl ■a Xi 73 re re re 3 • 73 ■? > in iS 2i in LT) ITj LT; IT) in in ^ ^ ^H -* in o ■- T— I E o u XI Xi Xi Xl U Xi u T3 re Oi M- X Xl re g £ u re XI Xl Xl O Xl u TJ re Oi 01 re 01 c _ _ _ ^^ ^ ^ ^.^ ^^ ^.^ 5 « in in in in in in in in in rri ^ ^ in in C O ^ O ^^ ^^ C^ O x^ QJ 03 re Xi Cj Xi Xl ■n re 01 ■J- ^ Xl re -J o T3 r oi re re Xl u Xl Xl 73 re Ol Oi 3:; Xl re g^ < ■a J: ■o Ci. < op J: "0 < x: T3 2 1 2 < J: -a 2 < 2 < < CQ a, Oi < CQ -*■ 01 < >- i: £ 73 Oi ^ i! V Oi * • U O 53 J en Ch Gh cu H H 10 Scientific Papers, Natural History Museum, The University of Kansas ^ ^ ./..<^ ^o^ .<^ <<> ^ <(>■ <0- <0^ <<£ 3 m t^ <" & -g -a; < 3 =3 £ o; >- s o o 3 O ro c Cu ^ > > i m o 3 fD cr 01 ^-^ ^1 Q> _ Ul Ij re X ej-o 0/ re u>— »— o^^o-— ^^^-'— o--^^-^- o<— oo^^^o a. o u-^g u re ' j::* u xi 03 ti "^ i: '2 ' T3 -a I u X -^^ '^ ^•O^^^ - - ;— ■— Ot— O^OOOOO ' X X ^ re o 3 re tu -g ^ ^ o ^ o o o o o o ^ ts oj u X re X X X 73 re o -— o OOOOOi—OOOOOO^tNOOOO uojreureoreure X o X u 73 re o> o^ooooooo^oooooo o o o o ore o 73 re Xi X ej ^*- X < £ re re ?> TT ^-- t- re 2 cu c- r- r" O QJ 1 c CI, : ■^ 73 = 73 re — O ■^ 1^ — ^ 3 e 2 ^ a ^. S r re Q I ;> < 2 u I- ly-J , CI- u . <*- re o 3 3 -S 2 O O w 2 yl X u e G s QJ 2 >■ r; ■£ re re 2 c E re X S 1 I— 'CNtr)^Lr)\Di>Nooo\0'— 'rJr^l•*lCs£)^^oco^ 16 SciENTinc Papers, Natural History Museum, The University of Kansas o o o o o o o o o o o o o o — o o •— o ^ o ^ o — o ^ o ■a c c o •a c a. a. < u o o o o o -— o o o o --^ o ^ o o o o o o o o o o -— o --^ o rj O O .-^ O t— O CN o ^ o o ■— o ^ 2; 3 a. = o -■ £^0 ^i^iz C ra o -c 3 3 IS < C H O 01 CD C 2 c < o