UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN BIOLOGY MAR 2 6 1985 A. ? FIELDIANA Zoology Published by Field Museum ot Natural History No. 12 FISHES OF THE FAMILIES EVERMANNELLIDAE AND SCOPELARCHIDAE: SYSTEMATICS, MORPHOLOGY, INTERRELATIONSHIPS, AND ZOOGEOGRAPHY ROBERT KARL JOHNSON ''■'■' , m Hie library of the 1982 University ot Illinois at Urtooe-Chnnpeign August 18, 1982 Publication 1334 FISHES OF THE FAMILIES EVERMANNELLIDAE AND SCOPELARCHIDAE: SYSTEMATICS, MORPHOLOGY, INTERRELATIONSHIPS, AND ZOOGEOGRAPHY FIELDIANA Zoology Published by Field Museum of Natural History New Series, No. 12 FISHES OF THE FAMILIES EVERMANNELLIDAE AND SCOPELARCHIDAE: SYSTEMATICS, MORPHOLOGY, INTERRELATIONSHIPS, AND ZOOGEOGRAPHY ROBERT KARL JOHNSON Curator Division of Fishes Department of Zoology Field Museum of Natural History Accepted for publication July 24, 1979 August 18, 1982 Publication 1334 Library of Congress Catalog Card No.: 81-65829 ISSN 0015-0754 PRINTED IN THE UNITED STATES OF AMERICA For Pat CONTENTS Abstract 1 Introduction 2 Acknowledgments 3 Methods 4 Descriptions 4 Material Examined 4 Counts and Measurements 6 Cephalic Laterosensory Pores 7 Snout-Pad Series 7 Mandibular Series 7 Preopercular Series 9 Temporal Series 9 Frontal Series 9 Infraorbital Series 9 Osteology 11 Systematics 11 Systematic Position of the Evermannellidae 11 Family Evermannellidae 12 EHagnosis 12 Description 13 Content and Range 14 Diagnostic Characters 16 Meristic Characters 16 Morphometric Characters 16 Osteological Characters 18 Laterosensory System 18 Eye Morphology 19 Gut Morphology 19 Luminous Tissue 19 Pigmentation 20 Larval Characters 20 (1) Peritoneal Pigment Sections 20 (2) Meristic Characters 23 (3) Gut Morphology 23 (4) Pigmentation 23 (5) Metamorphosis 27 Aspects of the Biology of Evermannellids 27 Sampling Difficulties 27 Size and Habits 28 Reproduction 32 Luminescence 32 Vision 33 Osteology 36 Cranium 37 Superficial Dermal Bones 41 Mandibular Arch 43 Palatine Arch 45 Opercular Apparatus 47 viii CONTENTS Hyoid Arch 48 Branchial Arches 50 Vertebrae, Supraneurals, Intermuscular Bones and Caudal Skeleton 54 Dorsal Fin 57 Anal Fin 57 Pectoral Girdle 57 Pelvic Girdle 61 Interrelationships 62 Catalogue of Characters and Character States 62 1. Eye Morphology 67 2. Lateral Division of Musculature 68 3. Swimbladder 68 4. Squammation 69 5. Lateral Line Scales 69 6. Mode of Reproduction 70 7. Pyloric Caecum 71 8. Peritoneal Pigment Sections 71 9. Accessory Pigment Spots or Areas 72 10. Juvenile-Phase Pigmentation 73 11. Stomach Pigmentation in Juveniles 73 12. Number and Distribution of Branchiostegal Rays 74 13. Number of Vertebrae 74 14. Frontal/Dermethmoid Contact 77 15. Parietals 77 16. Attachment of Dermosphenotic 77 17. Basisphenoid 77 18. Orbitosphenoid and Ethmoid Cartilage 78 19. Sclerotic Bones 78 20. Subocular Shelf 78 21. Antorbitals 78 22. Supraorbitals 79 23. Infraorbital Series 79 24. Premaxillary Fenestra 79 25. Modification of Maxilla 79 26. Dentary Fossa 80 27. Jaw and Palatine Teeth 81 28. Basihyal 81 29. Gill Rakers 82 30. Distribution of Gill Teeth 82 31. Fourth and Fifth Upper Pharyngeal Toothplates 83 32. Second Upper Pharyngeal Toothplate 84 33. Third Upper Pharyngeal Toothplate 85 34. Third Epibranchial Toothplate 85 35. M. Retractor Arcuum Branchialum 85 36. Fifth Ceratobranchial Toothplate 86 37. Basibranchial Dentition 86 38. First Pharyngobranchial 86 39. Uncinate Process of Second Epibranchial 86 40. Attachment of First Centrum 88 41. Supraneurals 88 42. Intermuscular Bones 89 43. Number of Hypurals 89 44. Second Ural Centrum 89 45. Number of Epurals 90 46. Fleshy Midlateral Keel 90 47. Posttemporal 90 48. Pectoral Girdle 91 49. Luminous Tissue 91 CONTENTS ix 50. PSM Cephalic Laterosensory Pores 92 51. Number of Dorsal-Fin Rays in Evermannellid Species 92 Interrelationships Among Iniomous Fishes 93 Interrelationships Among Evermannellid Species 100 Evermannellidae: Species Accounts 101 Artificial Key to the Species of Evermannellidae 101 Coccorella Roule 1929 103 Coccorella atlantica (Parr, 1928) 117 Coccorella atrala (Alcock, 1893) 120 Evermannella Fowler 1901 123 Evermannella ahlstromi Johnson & Glodek 1975 126 Evermannella balbo (Risso, 1820) 127 Evermannella indka Brauer 1906 136 Evermannella megalops Johnson & Glodek, 1975 146 Odontostomops Fowler 1934 147 Odontostomops normalops (Parr, 1928) 148 Scopelarchidae: Species Accounts 152 Benthalbella Zugmayer 1911 154 Benthalbella dentata (Chapman, 1939) 154 Benthalbella elongata (Norman, 1937) 154 Benthalbella infans Zugmayer 1911 154 Benthalbella linguidens (Mead & Bohlke, 1953) 157 Benthalbella macropinna Bussing & Bussing 1966 157 Rosenblattichthys Johnson 1974 157 Rosenblattichthys alatus (Fourmanoir, 1970) 157 Rosenblattichthys hubbsi Johnson 1974 158 Rosenblattichthys volucris (Rofen, 1966) 163 Scopelarchoides Parr 1929 164 Scopelarchoides climax Johnson 1974 164 Scopelarchoides danae Johnson 1974 165 Scopelarchoides kreffti Johnson 1972 165 Scopelarchoides nicholsi Parr 1929 165 Scopelarchoides signifer Johnson 1974 167 Scopelarchus Alcock 1896 169 Scopelarchus analis (Brauer, 1902) 169 Scopelarchus guentheri Alcock 1896 171 Scopelarchus michaelsarsi Koefoed 1955 171 Scopelarchus stephensi Johnson 1974 174 Zoogeography and Evolution 174 Distribution-Pattern Categories 179 (1) Division by Inshore/Offshore Association 181 (2) Division into "Cold-Water" vs. "Warm-Water" Areas 181 (3) Division by Ocean Basin 183 (4) Division with Respect to Water-Mass Regions 185 Distribution of Evermannellid and Scopelarchid Species: Regional Accounts 187 Atlantic Ocean 187 Subtropical Species 193 Tropical-Subtropical Species 193 Tropical Species 194 Eastern Species 195 Indian Ocean 197 Transition Region Species 199 Subtropical Species 199 Tropical-Subtropical Species 199 Tropical Species 200 Species Occurring North of 10° N 200 Pacific Ocean 202 Species Occurring in Austral-Asian Seas 203 x CONTENTS Transition Region Species 203 Subtropical Species 205 Tropical-Subtropical Species 205 Tropical Species 206 Species Occurring in the Eastern Tropical Pacific 207 The Pacific Central-Gyral Species 226 Literature Cited 237 LIST OF ILLUSTRATIONS 1. Cephalic laterosensory pores in three species of evermannellids 8 2. Distribution of the Family Evermannellidae 15 3. Eye morphology in evermannellids 21 4. Peritoneal pigment sections in larval specimens of two evermannellid species .... 22 5. Larvae of three evermannellid species 24 6. Larvae of four evermannellid species 25 7. Larvae and juvenile of three evermannellid species 26 8. Body musculature of tail region in evermannellids, other alepisauroids, scopelarchids, and chlorophthalmids 30 9. Cranium of Evermannellidae 37 10. Superficial dermal bones of the snout and orbital regions in evermannellids 42 11. Mandibular arch in evermannellids 44 12. Palatine arch, opercular apparatus, and part of hyoid arch in evermannellids .... 46 13. Hyoid arch (partial) in Evermannellidae 49 14. Branchial arch elements in Evermannellidae 51 15. Branchial arch elements in Evermannellidae 52 16. Vertebrae, supraneurals, intermuscular bones, and caudal skeleton in evermannellids 56 17. Dorsal and anal fin supports in evermannellids 58 18. Pectoral girdle in Evermannella balbo 59 19. Pelvic girdle of evermannellids 61 20. Possible interrelationships among iniomous taxa 95 21. Proposed relationships among evermannellid species 101 22. The species of Coccorella 105 23. Frontal canal commissure in Coccorella 107 24. Distribution of Coccorella atlantica and Coccorella atrata 108 25. Interorbital width vs. SL for species of Coccorella Ill 26. Geographic distribution of specimens of Coccorella used in morphometric analyses 113 27. Combined character index vs. SL plotted for specimens of Coccorella 114 28. Canonical variate analysis of morphometric characters plotted for Coccorella spp. . 115 29. The species of Evermannella 125 30. Distribution of Evermannella ahlstromi 128 31. Distribution of Evermannella balbo compared with all other evermannellid species . 133 32. Distribution of Evermannella balbo 134 33. Geographic subareas chosen for study of variation in numbers of anal-fin rays in Evermannella indica 141 34. Distribution of Evermannella indica 142 35. Distribution of Evermannella megalops 145 36. Odontostomops normalops 149 37. Distribution of Odontostomops normalops 151 38. Distribution of the Family Scopelarchidae 153 39. Distribution of Benthalbella infans 156 40. Distribution of two species of Rosenblattichthys 159 41. Larvae of Rosenblattichthys hubbsi 161 42. Distribution of Rosenblattichthys volucris 164 43. Distribution of Scopelarchoides danae 166 44. Distribution of Scopelarchoides nicholsi 167 xii LIST OF ILLUSTRATIONS 45. Distribution of Scopelarchoides signifer 168 46. Distribution of Scopelarchus analis 170 47. Distribution of Scopelarchus guentheri 172 48. Distribution of two species of Scopelarchus 173 49. "Cold-water" vs. "warm-water" faunal areas 182 50. Distribution of evermannellid and scopelarchid species exhibiting subtropical distribution patterns 184 51. Distribution of evermannellid and scopelarchid species exhibiting tropical distribution patterns 186 52. Distribution of evermannellid and scopelarchid species exhibiting tropical- subtropical distribution patterns 188 53. Distribution of Evermannella balbo relative to proposed faunal and water-mass regions 189 54. Comparison of Atlantic and Indian Ocean distributions of Coccorella atlantica and Scopelarchus guentheri 196 55. Distribution of Evermannella ahlstromi, Rosenblattichthys volucris, and Scopelarchoides nicholsi in the eastern tropical Pacific 208 56. Extent of the oxygen minimum layer in the eastern tropical Pacific 210 57. Distribution of two myctophid species endemic to the eastern tropical Pacific .... 212 58. Distribution of two species of Vinciguerria in the eastern tropical Pacific 213 59. Distribution of three ETP-endemic species relative to those areas delineated by Brandhorst (1959) as showing the greatest development of an oxygen minimum layer 214 60. Distribution of three ETP-endemic species relative to those areas delineated by Austin (1960) as showing the greatest development of an oxygen minimum layer 215 61. Distribution of three ETP-endemic species relative to those areas of the ETP in which the 1.00 ml/1 dissolved-oxygen isopleth lies at or shallower than 100 m 216 62. Vertical distribution of dissolved oxygen along 126° W 217 63. Vertical distribution of dissolved oxygen along 119° W 218 64. Vertical distribution of dissolved oxygen along 112° W 219 65. Vertical distribution of dissolved oxygen along 98° W 220 66. Vertical distribution of dissolved oxygen along 85° W 221 67. Vertical distribution of dissolved oxygen along 82° W 222 68. Vertical distribution of dissolved oxygen in the vicinity of 119° W: January to February, 1967; April, 1967; August, 1967; October, 1967 223 69. Vertical distribution of dissolved oxygen along 98° W: February to March, 1967; August to September, 1967 224 70. Distribution of five evermannellid and scopelarchid species in the vicinity of 15° S to 15° N, ca. 120° W 225 71. Pacific central- water species assemblage areas 228 72. Correlates of distribution of Pacific central-gyral species 230 73. Comparison of broadly distributed vs. central-gyral euphausiid species in the Pacific Ocean 233 74. Sister-group relationships among the scopelarchid species comprising the lineage containing Scopelarchus 236 LIST OF TABLES 1. Number of pores in six series of the cephalic laterosensory system in evermannellids 10 2. Comparison of meristic characters in evermannellids: dorsal fin, pectoral fin, pelvic fin, branchiostegal rays 16 3. Comparison of meristic characters in evermannellids: anal fin, vertebrae 17 4. Comparison of selected morphometry characters among evermannellid species and genera 18 5. Abbreviations used in osteological descriptions 38 6. Distribution of branchial tooth-bearing elements in evermannellids 53 7. Listing of cleared and stained material 64 8. Character state by OTU matrix for iniomous taxa 65 9. Comparison of meristic characters among iniomous taxa 75 10. Number of derived character states shared by iniomous taxa 96 11. Character state by OTU matrix for evermannellid species 102 12. Number of vertebrae in species of Coccorella 109 13. Comparison of selected morphometric characters for two "populations" of Coccorella spp 110 14. Distribution by size class of specimens of Coccorella spp. used in morphometric analyses 116 15. Geographic variation in number of anal-fin rays in Coccorella atlantica 119 16. Geographic variation in certain meristic characters in Evermannella balbo 121 17. Anal-fin ray counts in Evermannella indica and Odontostomops normalops 139 18. Vertebral counts in Evermannella indica and Odontostomops normalops 140 19. Geographic variation in anal-fin ray counts in Evermannella indica 143 20. Geographic variation in anal-fin ray counts in Benthalbella infans 155 21. Comparison of values for meristic characters in three species of Rosenblattichthys 158 22. Geographic variation in meristic characters in Rosenblattichthys hubbsi 160 23. Classification of distributional patterns exhibited by evermannellid and scopelarchid species 180 24. Comparison of systems of distribution patterns recognized for Atlantic mesopelagic fishes 191 ABSTRACT The alepisauroid family Evermannellidae contains seven species arranged in three genera. Diagnostic characters for the recognition of species and genera include meristic and morphometric characters, conformation of the eye, number and arrangement of laterosensory pores and associated structures, osteological features, pigment patterns, gut morphology, larval morphology and pigmenta- tion, and the presence or absence of luminous tissue. Evermannellids are rela- tively large-bodied, mesopelagic predators, not known to exhibt diel vertical migration, and, based on rarity of representation (of adults) in collections, are probably often successful at net avoidance. Although the principal prey of most evermannellid species appears to be fish, species in the genus Coccorella may frequently concentrate on squid. Evermannellids exhibit a highly peculiar tripar- tite division of the tail musculature that may relate to the need for maintaining position in the water column and for achieving short but rapid bursts of speed during prey capture. Luminous tissue occurs in the species of Coccorella. The tubular eyes of Evermannella spp. and scopelarchid species, although very similar in morphology and no doubt in function, have probably evolved independently. Examination of a large number of characters based on external morphology, larval morphology, osteology, and myology fails to confirm sister-group relationship for the Evermannellidae and Scopelarchidae. Although everman- nellids appear to be most closely related to other alepisauroids, particularly to the Alepisauridae and Omosudidae, available evidence suggests that scopelar- chids are members of a chlorophthalmoid lineage, not closely related to the alepisauroids. Within the Evermannellidae, Odontostomops is regarded as the sister group of Coccorella + Evermannella. Each evermannellid genus can be well defined, largely in terms of autapomorphous features. Evermannellids occur throughout the tropical and subtropical Atlantic, Indian, and Pacific Oceans, including (one species) the Mediterranean Sea, but no evermannellid occurs in polar seas. Coccorella contains two species: C. atlantica, a subtropical species occurring in all three oceans, and C. atrata, a tropical species limited to the Indian and Pacific Oceans. Evermannella contains four species: E. ahlstromi, endemic to ecotonal areas in the eastern tropical Pacific; E. balbo, largely restricted to rela- tively cool and/or productive areas in the Atlantic, Indian, and Pacific Oceans; E. indica, a tropical-subtropical species that is widely distributed in all three oceans; and E. megalops, endemic to central-water areas in the South Pacific. Odonto- stomops is monotypic, O. normalops is a tropical-subtropical species that is widely distributed in all three oceans. New material is reported for 15 (of 17) scopelar- chid species, adding substantially to our knowledge of the distribution of several of these species. The distribution of evermannellid and scopelarchid species, taken singly and in combination, is examined for agreement with patterns ex- 2 FIELDIANA: ZOOLOGY hibited by other midwater organisms, with patterns of possible environmental correlates of distribution, and, to a lesser extent, with data bearing on the his- torical components of open-ocean species assemblages. INTRODUCTION The family Evermannellidae, the saber-toothed fishes, occurs in midwater throughout the tropical and subtropical ocean. Despite the widespread distribu- tion of evermannellids, the family has until now remained poorly known, largely a result of lack of material. I recognize seven species of evermannellids arranged in three genera: Coccorella, Evermannella, and Odontostomops. It has long been supposed that the closest relatives of the evermannellids are to be found among the Scopelarchidae, the pearl-eyed fishes, a suggestion based largely on the occurrence of tubular-eyed forms in both groups. The present study was prompted by the supposed sister-group status of these two families; it seemed an ideal opportunity to compare patterns of distribution and variation between two morphologically and perhaps ecologically similar open-ocean families. This study has the following objectives: (1) recognition and definition of extant evermannellid species and genera, (2) description of patterns of variation in widely distributed evermannellid species, (3) assemblage of all available mor- phological and ecological data bearing on the natural history of evermannellid species, (4) based on examination of considerable additional material, to update my revision of the Scopelarchidae (Johnson, 1974c), (5) to make detailed com- parisons of the distributions of evermannellid and scopelarchid species with possible physical, chemical, and biological correlates of distribution and to com- pare observed patterns with those exhibited by other midwater organisms, (6) on the basis of study of all available characters, external morphology, larval morphology, osteology, etc., to produce the best-possible hypothesis of inter- relationships among evermannellid species, (7) to attempt a synthesis of phylogenetic and zoogeographic information leading to accounts of the evolu- tionary zoogeography of the two groups, (8) by examining other iniomous (sensu Gosline et al., 1966) taxa, to assemble and assess evidence bearing on the hypothesis of sister-group relationship for the Evermannellidae and Scopelar- chidae. Species of scopelarchids recognized are those included in Johnson (1974c). That work was based on some 2,102 specimens from 1,122 lots. Since that time I have had the opportunity to examine an additional 1,549 specimens from 717 lots. My account of the Evermannellidae is based on 2,763 specimens from 1,270 lots. The listing of nominal species and genera below summarizes the taxonomic conclusions of this study. Names are listed in chronological order of their ap- pearance in the literature, with the original combination on the left and the currently recognized combination (if different from the original) on the right. Coccorella Roule 1929 Type species: Coccorella atrata (Alcock 1893) Odontostomus atratus Alcock 1893 = Coccorella atrata (Alcock) Evermannella atrata atlantica Parr 1928 = Coccorella atlantica (Parr) Odontostomops braueri Rofen 1963 = Coccorella atrata (Alcock) JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 3 Evermannella Fowler 1901 Type species: Evermannella balbo (Risso 1820) Scopelus balbo Risso 1820 = Evermannella balbo (Risso) Odontostomus hyalinus Cocco 1838 = Evermannella balbo (Risso) Evermannella indica Brauer 1906 Evermannella indica melanoderma Parr 1928 = Evermannella indica Brauer Odontostomus balbo atlanticns Borodin 1931 = Evermannella indica Brauer Evermannella borodini Whitley 1958 = Evermannella indica Brauer Evermannella sicaria Rofen 1963 = Evermannella balbo (Risso) Evermannella ahlstromi Johnson & Glodek 1975 Evermannella megalops Johnson & Glodek 1975 Odontostomops Fowler 1934 Type species: Odontostomops normalops (Parr 1928) Evermannella normalops Parr 1928 = Odontostomops normalops (Parr) ACKNOWLEDGMENTS I am grateful to the following individuals for making available to me valuable specimens from collections in their care: E. Ahlstrom, R. Backus, E. Bertelsen, P. Bourret, J. Butler, T. Clarke, J. Copp, J. Craddock, W. Eschmeyer, P. Four- manoir, R. Gibbs, R. Haedrich, P. Herring, P. Hulley, S. Kannemeyer, L. Knapp, A. Kotthaus, G. Krefft, E. Lachner, R. Lavenberg, K. Liem, N. Merrett, B. Nafpaktitis, N. Parin, J. Paxton, W. Pearcy, D. Powell, J. Rivaton, C. Robins, R. Rosenblatt, R. Schoknecht, P. Sonoda, P. Struhsaker, C. Swift, and S. Weitzman. Thanks to M. Barnett, R. Gibbs, G. Krefft, and R. Rosenblatt for valuable information from their own research. Special thanks to G. Glodek, D. Ingle, J. Pizzimenti, and R. Wassersug, who separately have worked with me on various aspects of this project. My thanks to P. Herring and J. Paxton, who informed me of the occurrence of luminous tissue in Coccorella, and, in the case of Herring, provided me with a copy of a manuscript describing this tissue. My thanks to N. A. Locket for providing useful information on the structure and function of tubular eyes in midwater fishes. My thanks to R. Rosenblatt, R. Backus, and J. Craddock for their hospitality during my visits to their respective institutions. I am especially grateful to R. H. Rosenblatt, who first started me working on alepisauroid fishes. My thanks to the Division of Photography, Field Museum of Natural History, for aid in preparation of the figures. My thanks to J. VanStone, T. Bushman, and J. Nedrow for aid in processing the completed manuscript. My thanks to B. Scott for typing the section on Interrelationships. Special thanks to B. Peyton, who typed earlier versions of several sections and who listened to innumerable progress reports. I am also grateful to M. Barnett, A. Ebeling, R. Gibbs, N. Parin, and R. Rosenblatt, who read the manuscript and offered valuable and constructive criticism. FIELDIANA: ZOOLOGY I am most grateful to the late Loren P. Woods for his encouragement, aid, advice, and friendship. Last, but by no means least, thanks to my wife, Pat, who aided me in prepara- tion of the manuscript, typed the final draft, helped read various stages of proof, and who puts up with me. METHODS Descriptions. — In the descriptions of the various taxa of evermannellids I have attempted to avoid redundancy by including characters common to all members of a taxon only in the description of that taxon. Except where otherwise noted, all drawings are by the author. Material Examined. — The following abbreviations are used in reference to material examined: AMS CAS FMNH FSBC IOAN ISH LACM MCZ NIO NMFS(H) NMFS(LJ) Australian Museum, Sydney; material listed by ship, cruise, and station number. California Academy of Sciences, San Francisco; material listed either by CAS or SU (Stanford University) catalogue number. Field Museum of Natural History, Chicago; material listed by FMNH catalogue number. Florida State Department of Natural Resources, St. Peters- burg; material listed by FSBC catalogue number. Institute of Oceanography, Academy of Sciences of the U.S.S.R., Moscow; material listed by ship and station number: AK B V R/V Akademik Kurchatov R/V Baikal R/V Vityaz Institut fur Seefischerei, Bundesforschungsanstalt fur Fisch- erei, Hamburg; material listed by ISH catalogue number or by ship and station number: WH FFS Walther Herwig Natural History Museum of Los Angeles County, Los Angeles; material listed by LACM catalogue number. Museum of Comparative Zoology, Harvard University, Cambridge; material listed by MCZ catalogue number or by ship, cruise, and station number: AB R/V Anton Bruun Institute of Oceanographic Sciences, Wormley, Godalming, Surrey, U.K.; material listed by ship and station number: DY R/S Discovery National Marine Fisheries Service, Honolulu; material listed by ship and station number: HMS R/V Hugh M. Smith JRM R/V John R. Manning National Marine Fisheries Service, La Jolla; material listed by ship and station number: J R/V David Starr Jordan TC R/V Townsend Cromwell JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 5 ORSTOM Office de la Recherche Scientifique et Technique Outre-mer, Noumea, New Caledonia; material listed by name of ship, cruise, and station number or by date of capture. SAM South African Museum, Capetown; material listed by SAM catalogue number. SIO Scripps Institution of Oceanography, University of California at San Diego, La Jolla; material listed by SIO catalogue number. SOSC Smithsonian Oceanographic Sorting Center, Smithsonian Institution, Washington, D.C.: all SOSC material will be permanently deposited at either Field Museum of Natural History or the National Museum of Natural History; mate- rial listed by ship, cruise, and station number: AB R/V Anton Bruun DES USNS De Steiguer ELT USNS Eltanin J R/V David Starr Jordan TV R/V Te Vega UND R/V Undaunted UH Hawaii Institute of Marine Biology, University of Hawaii, Kaneohe; all UH material is or will be deposited in several permanent institutional collections; material listed by date of capture (year/month/station number) or by ship, cruise, and station number: TC R/V Townsend Cromwell UMML Rosenstiel School of Marine and Atmospheric Sciences, Uni- versity of Miami, Miami; material listed by UMML catalogue number. USC University of Southern California, Los Angeles; material listed by ship, cruise, and station number: ELT USNS Eltanin USNM National Museum of Natural History, Smithsonian Institu- tion, Washington, D.C.; material listed as follows: (1) USNM catalogue number; (2) ACRE, Ocean Acre Expedi- tions, material listed by cruise and station number (see Gibbs, Roper, et al., 1971; Gibbs & Roper, 1971); (3) MED, Mediterranean Biological Studies Expeditions, material listed by cruise and station number (see Goodyear et al., 1972); (4) uncatalogued material listed by ship, cruise, and station number: AB R/V Anton Bruun ELT USNS Eltanin ORE R/V Oregon TV R/V Te Vega UND R/V Undaunted WHOI Woods Hole Oceanographic Institution, Woods Hole, Mass.; most of this material has been or will be deposited in the Museum of Comparative Zoology, Harvard University; material listed by Richard H. Backus (RHB) station number. 6 FIELDIANA: ZOOLOGY ZIZM Zoological Institute and Zoological Museum, Hamburg; material listed by IOES or RBF catalogue number. ZMUC Zoological Museum of the University of Copenhagen, Copenhagen; material listed by Dana (D) station numbers. A number of vessels were involved in the collection of this material, notably the M/S Dana I and R/S Dana II (see Schmidt, 1929, Jesperson & Taning, 1934). The list of material examined includes only the institutional catalogue or sta- tion number and the number of specimens examined corresponding to each lot. Complete capture and locality data may be obtained upon motivated request from the respective institutions or from the author. Material listed under the heading "additional larval and juvenile material" represents specimens for which standard lengths were not recorded. Most specimens were taken in non-closing net hauls with the following types of gear: Isaacs-Kidd Midwater Trawl (IKMT, 3-m diameter unless otherwise specified); CMBT 1600 Midwater Trawl (ISH Walther Herwig collections); and the conical nets used by the Dana Expeditions (S150, S200, E300, etc., see Dana, Rep. No. 1 [1934] p. 18). Counts and Measurements. — Unless specified below, methods of taking counts and measurements follow those given by Hubbs & Lagler (1958, pp. 19-26) and Johnson (1974c, pp. 6-7). The last rays of the dorsal and anal fins are divided completely to the base and in each case were counted as one. All vertebral centra were counted, including the compound stegural element. A free second ural centrum is present in all evermannellids (presumed for Evermannella megalops) but could not be distinguished in radiographs and is not included in vertebral counts. Vertebral counts were made from radiographs and from cleared and stained specimens. Lengths are given as the standard length (SL) in millimeters. Measurements were made to 0.1 mm with 180- mm dial calipers or needle-point dividers except for those measurements less than 3 mm, which were taken to 0.01 mm using an ocular micrometer on a Wild M5 microscope. All measurements were taken in a straight-line point-to-point fashion. In all, 29 measurements were made on each selected specimen. Those mea- surements listed below are either undefined in or taken differently than methods used in Hubbs & Lagler (1958) or Johnson (1974c). Body depth at anal-fin origin = vertical distance between base of anteriormost anal-fin ray and dorsal margin of body. Adipose fin: distance from dorsal-fin base = distance between base of pos- teriormost dorsal-fin ray and origin of adipose fin; distance from snout = dis- tance from tip of snout to origin of adipose fin. Head length = distance from tip of snout to posteroventral margin of opercle. Postorbital head length = distance from posteriormost margin of orbit to pos- teroventral margin of opercle. Eye diameter, horizontal and vertical = in each case the greatest fleshy diame- ter of eye. Longest tooth, dentary and palatine = in each case the longest straight-line distance from base to distal tip. Interorbital width = least transverse distance between dorsolateral ridge of JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 7 each frontal bone; dorsolateral ridge of each frontal bone forming lateral margin of frontal bone in interorbital region. Cephalic Laterosensory Pores. — The cephalic laterosensory system is well developed in evermannellids and is composed of characteristic series of large to very large pores and numerous sensory papillae (especially in Coccorella and Odontostomops). The number and arrangement of pores in certain of these series are of diagnostic value in separating certain species. I have (more or less arbi- trarily) divided the cephalic laterosensory system as found in evermannellids into six major series. SNOUT-PAD SERIES. — A group of seven to 11 pores piercing the snout-pad. The snout-pad is defined as the flattened dorsoanterior surface of the head lying between the anterior nostril and orbit and medial to the posterior nostrils. The snout-pad encloses the nasal bones and overlies the dermethmoid. The snout- pad forms a membranous roof extending from the interorbital area of the frontals to the anterior margins of the nasal bones. An extensive laterosensory canal system underlies the snout-pad. This system is divided anteriorly by the nasal bones and a median membranous septum, but the two canals thus formed share a common pore posteriorly (the posterior snout-pad pore) at about the point each canal enters the respective (left and right) frontal bone. The snout-pad series includes up to five separately identifiable pores or pore series (fig. 1). Nasal pores (PSN): two pores, one on each side, piercing the moderately to strongly truncate anterior wall of the snout-pad, and each lying dorsomedial to the respective tubular anterior nostril. Each nasal pore providing direct access to the hollow, trough-like lumen of the respective nasal bone. Anterior snout-pad pores (PSA): four pores arranged transversely across snout-pad at about posterior borders of nasal bones, lying on or slightly pos- terior to a transverse line through the center of each posterior nostril. PSA pores subequal in size, or the two lateral pores slightly larger than the two medial pores. Medial snout-pad pores (PSM): if present, two pores each slightly lateral to dorsal midline, one on each side, about midway between anterior snout-pad pores and posterior snout-pad pore. Posterior snout-pad pore (PSP): a single pore centered at dorsal midline be- tween or just behind anterior terminus of dorsomedial ridge of each frontal bone (fig. 1), just anterior to or over anterior portion of interorbital region of frontals. Lateral snout-pad pores (PSL): if present, two pores, one on each side, an- terolateral to PSP pore. Only three species, Coccorella atlantica, C. atrata, and Evermannella balbo, have the full complement of 11 snout-pad pores (table 1). MANDIBULAR SERIES (PMD).— A series of 12 to 14 pores arranged in four groups on the ventrolateral and anterolateral surfaces of the articular and den- tary of each side (fig. 1). Five pores along ventrolateral margin of articular, subequal in size, posteriormost pore of series at posteroventral corner of lower jaw. Four to six pores (table 1) on lateral face of dentary, the posteriormost directly over anterior pore of articular series. Counts for this group are variable, sometimes differing between left and right sides of a single specimen. The counts given are the maximum number of large pores observed. A number of much smaller pores (especially common in larger adults) may or may not be present in this area. Two pores ventrally on anterolateral face of dentary. One JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 9 pore closely adjacent and lateral to protruding vertical ridge marking dentary symphysis (Coccorella, Odontostomops) or at ventrolateral margin of a vertically elongate fossa centered on dentary symphysis (Evermannella) . PREOPERCULAR SERIES (PPOP).— Typically a series of six pores over each preopercle, three pores arranged vertically over ventral portion of troughlike main shaft of preopercle and three pores arranged in the rough form of a triangle over ventral plate of preopercle. Membrane over preopercle in Ei'ermannella ex- tremely delicate, and counts of the preopercular series in species of Evermannella difficult or impossible to obtain in the material available (table 1). TEMPORAL SERIES. — A series of five pores (fig. 1) connecting the infraorbi- tal, preopercular, and lateral line laterosensory canals. Pterotic pore (PPTO): a single pore centered over canal-like portion of pterotic and anterolateral to posterior frontal pore. Extrascapular pores (PESC): four pores, one at each corner and one at center of lateral edge of the roughly triangle-shaped extrascapular bone. Anteriormost pore marking junction of temporal canal (from pterotic) and preoperculoman- dibular canal (from preopercle). Medialmost pore marking terminus of the short supratemporal canal. Posteriormost pore at the point where temporal canal passes to the lateral line system via posttemporal and supracleithrum. FRONTAL SERIES. — Two well-defined pore series (fig. 1) over postorbital region of frontal bones. Posterofrontal pore (PPF): two pores, one on each side marking posterior terminus of the short posterofrontal canal that runs along the posterior continua- tion of the dorsomedial ridge of each frontal. Posterofrontal pore of each side almost directly posterior to the respective lateralmost pore of frontal canal com- missure and just anterior to articulation of frontal and parietal bones. Frontal canal commissure (FCC): a series of four [two on each side (Coccorella atrata)] or six [three on each side (all other evermannellid species)] pores directly posterior to interorbital region of each frontal and forming a transverse connec- tion between infraorbital series of each side (see Gosline et al., 1966, pp. 3-4, fig. 2). A very small percentage of the specimens of Coccorella have an asymmetric 2 -I- 3 or 3 + 2 (left + right) FCC pore count, and a detailed discussion of these abnormal specimens is provided following the description of Coccorella. INFRAORBITAL SERIES (PIFO).— A series of small pores arranged in two limbs corresponding to the first and second and to the third through eighth infraorbitals, respectively (fig. 1), with pores of the two series separated by a slight but noticeable gap at the ventroposterior corner of the infraorbital series Opposite: Fig. 1. Cephalic laterosensory pores in three species of evermannellids. A, B, Odonto- stomops normalops, SIO 72-316, 76.5. C, Coccorella atrata, SIO 70-346, 85.4. D, Evermannella balbo, ZMUC uncatalogued, from Mediterranean Sea, 75.2. Pore abbreviations: Snout-pad series: PSN=nasaI pore; PSA=anterior snout-pad pores; PSM=medial snout-pad pores; PSP=posterior snout-pad pore; PSL=lateral snout-pad pores. Mandibular series: PMD. Preopercular series: PPOP. Temporal series: PPTO=pterotic pore; PESC=extrascapular pores. Frontal series: PPF=postero-frontal pore; FCC=frontal canal commissure. Infraor- bital series: PIFO. Other abbreviations: AN=anterior nostril; PN=posterior nostril; LFR=lateral frontal ridge (interorbital region); MFR=medial frontal ridge (interorbital re- gion). c « g •■a ■S (N ^i tN r-i O N <*OHO rn + + rH + E 73 ,.5 en I 42 « -~ (N Tf (N H N g + o «J o-.E a 8 'g 60S. o tn « O '2 5 t« .5. a! c j= o ■ > ■5 I c p E 1 m> < sc C I E •8 jj H c _ (Q (0 o> — •E< Si sc a '8 m D- •5^ tn Q c y E c e O at o c ■£ □ o Ov "* .i-.'. r • •■• • • . 1* *••'■. ■w3R 21 e "o 45 2 ma |)§ g a, i c "j'fl jS w OJ c e ^P c «i 01 o _ > • c « ' ..1 ". IT ,L "Q O) O (/> i- S ^ II rati j5 T! II ■■ - f. -^ — ! o o in *« 01 (0 o o v * - S 13 ».a c 2 £ O « g S (N ,1/5 TJ £ XU-3 O 8 .& 13 J/5 £«;- w to m .£ " ^■5D 035 S .5 a, u b u « 3 •£ •- 01 .22 a. o ex? <-T> (N O CD O ID (N 00 00 S ix Z £ 000 .2 o h U CT\ GO 0h ±: S 00 a O 00 a 30 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 31 tinguished from the underlying epaxialis and hypaxialis in both texture (the fibers are narrower in diameter) and color. In alcohol-preservation the fibers of the lateralis superficialis appear lighter in color than those of the epaxialis and hypaxialis (in species I have examined), but in life they tend to be distinctly red. The large lateralis superficialis of evermannellids is apparently bright red in life (R. McGinnis, pers. comm., based on observation of freshly captured material of £. ahlstromi). The lateralis superficialis is enormously well developed in many mesopelagic fishes (Marshall, 1971, p. 78), and this is apparently true for all alepisauroids (fig. 8). In all examined alepisauroids the lateralis superficialis tapers anteriorly, resembling the condition illustrated by Winterbottom (1974, figs. 53, 54). Sub- sequent discussion is restricted to the body musculature posterior to the anal-fin origin. The lateralis superficialis in Parasudis truculentus (Goode & Bean) (Chloroph- thalmidae), a bottom-dwelling species that feeds in "midwater" (Mead, 1966e, p. 183), is similar in position and size to that normal for teleosts (Barets, 1961; Winterbottom, 1974) — a narrow, midlateral sheath, considerably less massive than the underlying epaxialis and hypaxialis (fig. 8B). In the scopelarchids (species examined: Scopelarchus analis (Brauer), FMNH 88146, 90.2 mm; Ben- thabella infans Zugmayer, FMNH 79702, 134.0 mm) the lateralis superficialis is proportionately somewhat larger than that of Parasudis but otherwise is similar in arrangement (fig. 8C). In alepisauroids other than evermannellids (species examined: Alepisaurus sp. [Alepisauridae], FMNH 83657, 35.5 mm; Anotopterus pharao Zugmayer [Anotopteridae], SIO uncat., 350 mm; Omosudis lowei Guenther [Omosudidae], FMNH 84778, 200.0 mm; Paralepis sp. [Paralepididae], FMNH 85322, 120.5 mm) the lateralis superficialis is enormously well developed, occupying most of the lateral surfaces of the body in the tail region (fig. 8D, E; see also Marshall's [1971, fig. 34a] illustration of the red muscle system in the Antarctic paralepidid Notolepis coatsi Dollo). In the evermannellids (species examined: Coccorella atrata, FMNH 85321, 85.4 mm; Evermannella indica, FMNH 82736, 96.5 mm; Odontostomops normalops, FMNH 88171, 106.8 mm) the arrangement of the well-developed lateralis super- ficialis is quite different from that seen in other alepisauroids in that in the tail region the epaxialis and hypaxialis muscles do not extend beneath the lateralis superficialis (fig. 8F, G). Thus, the apparent tripartite division of the tail muscu- lature visible on external view (fig. 8A) actually reflects the underlying structure. Marshall (1971, pp. 77-79) summarizes the available information on the distri- bution of red and white muscle tissue in midwater fishes, noting that for the mesopelagic species examined by him the proportion of red to white is signifi- cantly larger than normal in teleosts. Apparently the greater development of red muscle tissue in midwater species is associated with the problem of maintaining position in the water column faced by many mesopelagic species. Many mid- water fishes (including all alepisauroids) have no swimbladder and, despite reductions in the degree of bone ossification, loss or reduction of scales, and other adaptations directed toward decreasing overall specific gravity, remain negatively buoyant (Marshall, 1955; Denton & Marshall, 1958). The result is a need for continuous, if low-velocity, swimming if position in the water column is to be maintained. It is this sustained but relatively low-level activity for which red muscle tissue is well adapted (Barets, 1961; Rayner & Keenan, 1967; Alexan- 32 FIELDIANA: ZOOLOGY der, 1969; Marshall, 1971; Prosser, 1973). Although no histological examination of the lateralis superficialis in evermannellids has been made, it seems very likely (based on work on other teleosts, see Barets, 1961; Winterbottom, 1974) that this muscle is composed primarily of red muscle fibers. The great develop- ment of this muscle in evermannellids provides additional support (for ever- mannellids) for the contention of Gosline et al. (1966, p. 10) that ". . . alepisauroids . . . have a hovering and darting, pike-like way of securing their food." If true, it seems likely that the red muscle lateralis superficialis is largely used to maintain position in the water column, the white muscle hypaxialis and epaxialis to provide short but powerful bursts of speed needed in the capture of food and avoidance of predators (see Hudson, 1973). According to Marshall (1954, pp. 325-328), alepisauroids are partly characterized by a hydroplane- like configuration of the pectoral fins. A key component in this configuration is the low angle (said to be 15°-20° in E. balbo) between the horizontal axis of the body and the axis of the pectoral-fin insertion. This low setting and relatively large size of the pectoral fins combined with the joint action of the caudal and anal fins provide, in Marshall's (1954) view, requisite lift. The rich development of red muscle tissue might provide the basis for continuous power output. Both lift and continuous power output are required components of the mechanism envisioned as allowing maintenance of position in the water column. Reproduction. — The evermannellids (and all other alepisauroid fishes) are syn- chronous hermaphrodites with a functional ovotestis similar to that described for certain alepisauroid and other iniomous fishes (Mead, 1960; Mead et al., 1964; Nielsen, 1966, Gosline et al., 1966; Johnson, 1974c; Bertelsen et al., 1976; Herring, 1977). Luminescence. — The first description of luminous tissue in an evermannellid is provided by Herring (1977). Herring discovered and described luminous tissue in Coccorella atrata, based on living and preserved material. The following ac- count (for C. atrata) is abstracted from Herring's paper. The light produced is a weak blue glow along the ventral midline. All luminous areas are part of the ventral wall of the intestine (intestinal organs) or part of the ventral wall of the entire length of the pyloric caecum (isthmus organ). The three intestinal areas are as follows: median ventral organ, extending anteriad from just anterior to the pelvic-fin base to a point about three-fourths the distance to the pectoral-fin base; post-pelvic organ, just behind the pelvic-fin base; anal organ, just anterior to the anus. Each of these organs is indicated by a reflective, silvery-colored (in alcohol) streak or patch in ventral midline situated ventrally to a translucent line or patch in the otherwise heavily pigmented peritoneum. The dense pigmenta- tion of the skin of the ventral abdominal body wall is interrupted in the region of these reflective streaks such that the skin is translucent except for a number of large but scattered melanophores. An elaborate concentric reflector system (with a ventral aperture permitting exit of light) is associated with the isthmial organ. There is no reflector system associated with the intestinal organs. The down- welling light shines through and must be diffused by the ventral musculature, particularly in the case of the isthmus organ. The light-producing system is intrinsic, the light is nonbacterial in orgin. I have reexamined material representing all evermannellid species in an at- tempt to determine the presence or absence of possible luminous tissue. Coc- corella atlantica possesses midventral reflective streaks and patches in exactly the JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 33 same positions as those of C. atrata, and, with C. atrata, shares the unique possession of a pyloric caecum extending into the head. It seems very safe to assume that C. atlantica is also luminous. Herring (1977) tested freshly caught species of Odontostomops normalops and Evermannella sp. (almost certainly E. indica) and failed to demonstrate any lumi- nescent capability, either on simple exposure of the ventral gut wall or on appli- cation of dilute hydrogen peroxide. Herring reports that O. normalops lacks any reflective streaks, but that Evermannella sp. does show an ill-defined reflecting organ ventrally. Herring's examination of alcohol-preserved material of O. nor- malops and £. indica failed to demonstrate the presence of luminous tissue. My reexamination of material of all evermannellid species confirmed Herring's results for O. normalops — there are no reflective streaks nor any other indication on gross examination that luminous tissue is present. However, in all four species of Evermannella there are narrow but well-defined reflective streaks, one on each side of the ventral midline, extending from the pelvic-fin base to just before the anus. These streaks are most visible in my material of E. megalops. The streaks diverge under the pelvic-fin base and disappear or become much less distinct anterior to the pelvic-fin base. The arrangement is vaguely reminiscent of the luminous organs of the paralepidid genus Lestrolepis (Rofen, 1966a, p. 371), although no discrete ducts are involved and the reflective streaks in Ever- mannella are best-developed posterior to the pelvic-fin base. On cross examina- tion these reflective streaks in Evermannella appear similar to the unpaired and discretely distributed intestinal organs of Coccorella. The question of whether they are or are not indicative of luminous tissue in Evermannella awaits histologi- cal examination. Although Herring (1977) did not propose any functions for the ventral lumin- ous tissue of Coccorella, it seems that the best available explanation is that widely put forward for midwater fishes with ventral concentrations of luminous organs, viz., that the light produced matches the background of downwelling sunlight, breaking up the silhouette and tending to render the fish less visible to predators lower in the water column (Clarke, 1963; Babcock, 1970; Lawry, 1974). This function has been all but proved for mesopelagic squid (Young & Roper, 1976, 1977; Young, 1977), for Sergestes similis (Crustacea, Sergestidae, see Warner et al., 1979), and, by analogy, for midwater fishes. Marshall (1971) offers possible objections to this theory in the case of certain species. Vision. — Of all the remarkable characters of evermannellids and scopelar- chids, that which has provoked the widest interest in these fishes is the development of tubular eyes. All scopelarchid species possess fully developed tubular eyes directed straight upward or dorsoanteriad (three species). The gross external features of eye morphology in evermannellids have already been des- cribed (fig. 3). The eyes and details of eye morphology of Coccorella atrata and Evermannella indica are illustrated in Brauer (1908) and Munk (1966), and for the scopelarchids Scopelarchus analis and Benthalbella infans (Neoscopelarchoides sp. of Munk, 1966, p. 32) by Brauer (1908), Munk (1966), and Locket (1970, 1971). A number of the main features of the tubular eyes of both Evermannella indica and Scopelarchus analis are at least superficially similar. The visual axes are parallel and directed dorsad (and slightly anteriad in Evermannella). The retinal cup is roughly tube- shaped (in S. analis the retinal cups flare out ventrally and hence increase in 34 FIELDIANA: ZOOLOGY diameter from the pupillary margin to the floor of the eye). The bottom of the tube is formed by the main retina, whereas an accessory retina forms the medial wall of the tube. Locket (1971) has shown that the main retina in S. analis is actually divided into anterior and posterior portions in terms of histological structure, containing, respectively, non-grouped and grouped rods. The pupil is tilted such that the medial margin is distinctly more dorsal to the main retina than the lateral margin. The lens pad (= "pearl organ" of Johnson, 1974c, p. 28) is nearly centered on the lateral dorsal margin of the pupil. Locket (pers. comm.) has objected to my use of the term pearl organ to describe the glistening white oval-shaped tissue centered at the lateral pupillary margin and characteristic of the tubular-eyed species of evermannellids and scopelarchids. Locket points out that the term lens pad (based on Brauer's [1908, p. 218] "Linsenpolster oder Linsenkissen") was coined by Munk (1966, p. 32) and used by Merrett et al. (1973) as well as Locket (1970, 1971). Locket's objections are based not only on priority of usage but also on function. The lens pad in scopelarchids (confirmed for B. infans by Locket, for B. dentata by myself) is transparent in life, possibly as Locket (1970) has suggested, serving as a light guide, picking up light from almost beneath the fish (within 20° of the vertical below the fish) and guiding the light to the lens and thence to the dorsalmost part of the accessory retina. Thus the use of the term pearl organ, based on an artifact of preservation, implying as it does opacity, obscures the probable function of this organ. I follow Locket's suggestion in this paper, abandoning pearl organ in favor of lens pad. The first detailed anatomical description of the eye in evermannellids is provided by Brauer (1908). Brauer was aware of three evermannellid species, Coccorella atrata, Evermannella balbo, and E. indica, and was able to make detailed histological studies of the eyes of C. atrata and E. indica. Brauer (1908, p. 192) expresses surprise at the difference in eye structure between C. atrata and the two species of Evermannella, particularly that in C. atrata the visual axis is more lateral than dorsal, and the aperture in the palpebral fold is much smaller (the palpebral fold [Munk, 1966] is termed the adipose eyelid by Rofen [1966d] and elsewhere in the present paper). Brauer's figures (cf. plate 35 fig. 15 vs. plate 38 fig. 4) clearly show the difference in structure between the semitubular eye of Coccorella and the fully tubular eye of Evermannella. Munk (1966) has provided additional information on the eye of Evermannella indica. It is unfortunate that no one has studied in detail the eye morphology of Odontostomops normalops. Such a study is sorely needed if the sequence suggested below is to be confirmed. Munk's (1966, p. 44) description of the ontogenetic development of tubular eyes in conjunction with Brauer's (1908) detailed descriptions and illustrations of the eyes of Coccorella and Evermannella allow further comment on the possible phylogenetic development of the tubular eyes in Evermannella. The typical mor- phology of the tubular eye develops gradually from a laterally directed eye in the larvae of tubular-eyed species. The sequence of events is illustrated by Contino (1931) for the sternoptychid Argyropelecus hemigymnus; Merrett et al. (1973), for the scopelarchid Benthalbella infans; and Brauer (1908), for the scopelarchid Scopelarchus analis. In the course of development of the tubular eye the following events occur: (1) the lens and pupil are displaced dorsally, and (2) the main retina is developed from that part of the larval retina located ventrad to the optic papilla and mainly from the lower temporal quarter. The latter is brought about through a rotation around a vertical axis of the retinal cup during growth. JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 35 Munk (1966, p. 45) viewed the non-tubular eye of Omosudis lozvei Guenther (Omosudidae) as nearly ideal in structure for a hypothetical precursor of the tubular eye of Evermannella. He shows (Munk, 1966, fig. 11) that the course of the choroid fissure in Omosudis indicates that the broad ventral portion of the retina in that species corresponds to the lower temporal quarter of the retina in other teleosts, with the displacement of the choroid fissure possibly accomplished by rotation of the optic cup around the anterior-posterior axis. Thus the broad ventral portion of the retina in Omosudis is located in that exact part of the retina from which the main retina of Evermannella indica (and certain other tubular-eyed species listed by Munk, 1966, p. 44) develops. Munk (1966) points out additional similarities between the eye morphology of Omosudis loivei and Evermannella indica that suggest that the eye structure of O. loivei may represent more than an idealized precursor of the eye structure of evermannellids. Munk (1965) showed that Omosudis loivei possesses an almost pure cone retina, with a very small number of irregularly distributed typical teleost rods and a broad, ventrally located, rod-free area. Virtually all other midwater fishes have pure rod retinae (Munk, 1966; Marshall, 1971) — the shift to scotopic vision might be expected in the dimly lit mesopelagic environment (Goldsmith, 1973). The only other known exception is Evermannella indica, which (Munk, 1966) believes to have a retina composed of rods that are in fact highly modified cones. Munk (1966, p. 50) lists four characters of the rodlike cells of E. indica that he believes indicate the derivation of these cells from cones — normal teleost rods (present in small numbers in Omosudis) are lacking in Evermannella. A final point of similarity in eye structure between Omosudis and the everman- nellids is the development of the adipose eyelid (palpebral fold). In large speci- mens of O. lozvei the adipose eyelid encloses the entire eye except for a round aperture dorsolaterally (Rofen, 1966b, fig. 166), in shape and position similar to that found in Coccorella spp. (fig. 3). All three genera of evermannellids have such an adipose eyelid with the aperture very small in Odontostomops, very large in Evermannella, and intermediate in size in Coccorella (fig. 3). If, as Munk (1966) suggests, the eye of Omosudis loivei represents an ideal precursor for the tubular eye of Evermannella, the nontubular eye of Odontostomops and the semitubular eye of Coccorella represent logically ideal stages in the development of tubular eyes in this family. The tubular eye represents an adaptation to the dim lighting of the mesopelagic and has been independently evolved in a number of midwater groups (Marshall, 1971, pp. 42-43). Thoughts on the functional significance of tubular eyes are summarized in Munk (1966), Marshall (1971), and Locket (1970, 1971). Tubular eyes are modified such that a relatively large lens is associated with a proportionally small area of main retina. This arrangement allows an enlarged (but not brighter) image on the main retina. The parallel visual axes enlarge the binocular field, resulting in greater sensitivity and a better judgment of distance. The advantages of these improvements to an inhabitant of the poorly lit mesopelagic, particularly in the case of active predators, are clear enough. Munk (1966) discards the suggestion (e.g., Walls, 1942) that the division of the retina into main and accessory regions implies that tubular eyes are bifocal, the accessory retina used for the perception of distant objects, the main retina used for the perception of nearby objects. Munk goes to some length to show that no 36 FIELDIANA: ZOOLOGY sharp image can be formed on the accessory retina and concludes that the advantages of the tubular eye (in terms of enlarged image and binocularity) have been at the expense of optical adjustment of the accessory retina. Locket (1970), in discussing the tubular eye of scopelarchids, has pointed out that the most important difference in probable function between the accessory and main retinae lies not in optical adjustment but in field of view — the two main retinae covering a rather restricted binocular field dorsad and slightly anteriad, the accessory retinae covering two much wider monocular fields at the sides (made wider by the presence of the lens pad, if Locket's suggested function for this organ proves true). Locket (1970) goes on to suggest that an object to one side of the fish forms a blurred image on the accessory retina (which may be constructed such that its best response is to movement), stimulating the fish to turn toward the object and swim beneath it. The image would thereby be transferred to the sharply focused binocular field. The division of the main retina into anterior and posterior regions with non-grouped and grouped rods, respectively (see Locket, 1971, for a detailed discussion), represents a possible compromise between sen- sitivity (possibly higher in the posterior portion) and acuity (higher in the an- terior portion). Thus the sequence implied by Locket (1970) is that the accessory retina is used for first location of an object (and in this is aided by the lens pad), that the function of the grouped-rod portion of the main retina is to allow homing-in on the object, and that the visual acuity provided by the anterior non-grouped main retina allows an accurate strike. Locket admits this to be conjectural but in accord with available morphological evidence (see also Merrett etal., 1973, p. 44). The limiting light threshold for vision in midwater fishes with well-developed eyes is thought to be on the order of 3 x 10" 10 /xWlcm 2 (Clarke & Denton, 1962), some 10 to 100 times more sensitive than the eyes of epipelagic fishes (see Brett, 1957; Denton & Warren, 1957; Munz, 1971). A number of authors have used this value to estimate the greatest depth at which midwater fishes should be able to d?tect diel changes in light intensity (among the estimates: 1,000 to 1,100 m [Sargasso Sea, Clarke & Denton, 1962]; 700 to 1,300 m [Indian Ocean, Clarke & Kelly, 1964]). Marshall (1971) places the average of the various available esti- mates at about 1,000 m for very clear oceanic waters. Virtually all estimates are well below the probable depth range of maximum abundance for all everman- nellid species. OSTEOLOGY Only five papers have dealt with osteological characters of the Evermannel- lidae. Regan (1911) presented a very brief description of some of the skeletal features of Evermannella balbo. Parr (1929) studied the osteology of Coccorella atlantica and Evermannella indica. The works of Rofen (1966d), McAllister (1968), and Rosen (1973) considered a limited number of characters. My studies of the osteology of evermannellids are based on six of the seven species {Evermannella megalops had to be excluded due to lack of material) and all three genera. When compared with the considerable variation in skeletal morphology exhibited by scopelarchid species (Johnson, 1974c), evermannellids are very similar to one another in osteological features, and the description below is based on Everman- nella balbo except as indicated. Following the account of skeletal features in the Evermannellidae is a limited comparison of the Evermannellidae with its sup- JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 37 posed sister group, the Scopelarchidae, and with other alepisauroid and myc- tophoid fishes. The terminology I have employed for the various skeletal structures closely follows that used for the Scopelarchidae (Johnson, 1974c) and for the Chias- modontidae (Johnson & Cohen, 1974). A listing of abbreviations used in the figures and tables is presented as Table 5, and a listing of cleared and stained material examined is presented as Table 7. Cranium In evermannellids the ethmoid cartilage and 16 bones, 10 paired and six me- dian, form the cranium (fig. 9). The bones include the following: basioccipital (BOC), basisphenoid (BAS), dermethmoid (DEM), epiotics (EPO), exoccipitals (EXO), frontals (FR), lateral ethmoids (LEM), opisthotics (OPO), parietals (P), pal BOC Fig. 9. Cranium of Evermannellidae. A, Dorsal view of cranium of Evermannella balbo, ISH 546/73, 94.6 mm. Left palatine bone left in place. B, Dorsal view of cranium of Odon- tostomops normalops, ISH 2220/71, 111.5 mm. C, Dorsal view of cranium of Coccorella atrata, SIO 68-533, 99.6 mm. D, Ventral view of cranium of E. balbo, ISH 546/73, 94.6 mm. E, Posterior view (diagrammatic) of cranium of Coccorella atrata, 99.6 mm. » M 1 i 1 1 *§! -| - 1 1 - 1 3 "2 Si § ex ^ 01 ±3 — . X C „_, O re o o XQ 3 3 3 3 3 3 EX-S H fa fa (/i!/5333 ■5 g & cx cx cx cx exx x §fo cx cn r U "g en en O O ex <" H p d ^tn encnen > 3 i3 € >. X c x CX re (Q g fa C e o 9 > > s U r^ - ■** > '5 - 5 kJ — u n3 1 re C :s£ e "S 2^?«S C? „ re C «S H £ — ' X re -i Oh , *aj "5j »a ^a Ssb^ iS'OiS reO-§ | fax o o«^^.2.2 .£ 3 .2 .9- fa o S £ - fx o cx a; J2 v s g -a ^ U CX CX CX CX x x S.Stv?^ C? C (2 C re s 2 c c aia;a;ajaia»-faa box X X .fa .fa .fa 43 ^ fa fa fa ChJK 5 Cx s g rerS "2 .fi 3 1 -O 3 "5 2 .2 -73 re 'C O QQq w u u 3o JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 39 parasphenoid (PAS), prootics (PRO), pterosphenoids (PTS), pterotics (PTO), supraoccipital (SOC), sphenotics (SPO), and vomer (V). Ethmoid Region. — Ethmoid region consisting of central mass of ethmoid carti- lage, paired lateral ethmoid bones, and dermethmoid. Ethmoid cartilage single, covered dorsally over almost entire length by dermethmoid and by frontals (which lie between ethmoid cartilage and dermethmoid over posterior two- thirds of length of ethmoid cartilage). Ethmoid cartilage projecting anteriorly slightly beyond anterior terminus of dermethmoid. Ethmoid cartilage separating dermethmoid from vomer and parasphenoid. Ethmoid cartilage in both species of Coccorella considerably expanded posteriorly, forming an orbital septum be- tween the eyes, and extending to or nearly to midline in posterior wall of orbit. Ethmoid cartilage in Evermannella and Odontostomops not extending into orbit and not forming an orbital septum. Dermethmoid thin in cross-section but elongate, with two posteriorly directed projections overlying respective (right and left) anterior sheetlike portions of frontal bones (fig. 9). Ascending head of each premaxillary bone bound to dermethmoid by a mass of connective tissue. Lateral ethmoids prominent, thin, sheetlike bones completely enclosing a cartilaginous core, forming a major point of abutment of dorsal margins of palatine. Vomer. — Vomer an elongate sheathlike bone. Head of vomer lying ventral to ethmoid cartilage and bearing one tooth on each side, although in most speci- mens only one tooth (right or left) is present — the other tooth is apparently lost. Palatine bone of each side articulating with and strongly connected to vomer. Vomer with a posteriorly tapering shaft resting in an elongate shallow fossa on anteroventral surface of parasphenoid. Frontals. — Frontals the largest, most complex bones of dorsal skull roof. Fron- tals meeting tightly in midlongitudinal line posteriorly but not fusing, frontals closely adjacent but not directly articulating anteriorly. In Evermannella (fig. 9A) each frontal is divided among three main portions: an anterior troughlike sheet, a tubular interorbital area, and an expanded posterior plate. Anterior troughlike area overlying posterior two-thirds of ethmoid cartilage and supporting (dor- sally) posterior projection of dermethmoid on each side. Tubelike interorbital area markedly constricted and narrow in the tubular-eyed species belonging to Evermannella. Tubelike area opening anteriorly on troughlike frontal plate, pos- teriorly through pores of frontal commissural canal, and posterolaterally at junc- tion of supraorbital, pterotic, and infraorbital cephalic laterosensory canals. Interorbital area of frontals in Odontostomops and Coccorella somewhat but not markedly narrower than posterior plate (fig. 9B, C). Posteriorly, frontals form- ing a thin sheet of bone occupying most of dorsal surface of cranial vault and partly overlying parietals posteriorly. Posterior frontal plate articulating pos- terolaterally with pterotics, anterolaterally with sphenotics, and anteriorly with pterosphenoids. Parietals — Although most definitions of the Evermannellidae (Parr, 1929, Gos- line et al., 1966; Rofen, 1966d; Johnson, 1974c [following Parr, 1929]) have stated that the parietals are indistinguishably fused with the frontals, this is not true for any evermannellid species examined by me. Parietals present as thin platelike bones, which rather than being separated by supraoccipital in midline in fact overlap anterior and anterolateral margins of supraoccipital and thus meet in midline. Parietals overlain by frontals anteriorly and overlay anteromedial mar- gin of epiotics posterolaterally. Posteriorly, parietals with a dorsally directed 40 FIELDIANA: ZOOLOGY ridge that articulates with and serves as a continuation of a similar ridge on epiotics — and the anterior somatic musculature thus inserts on both epiotics and parietals, although the major insertion is on the epiotics. Supraoccipital. — Supraoccipital a large oblong bone centered at posterodorsal margin of cranium. There is no supraoccipital spine, but there is a strongly developed, short, blunt knob of bone at center of supraoccipital. Supraoccipital overlain anteriorly and anterolaterally by parietals, articulating with epiotics posterolaterally and separated from exoccipitals posteriorly and ventrally by an area of cartilage. Epiotics. — Epiotics relatively large, rounded bones contributing significantly to posterior wall of cranial vault (fig. 9) but largely excluded from dorsal wall of cranial vault. Epiotics overlain by parietals anteriorly, articulating with supraoc- cipital medially, with exoccipitals posteriorly, and with pterotics laterally within the posttemporal fossae. Dorsal projection of posttemporal connected to epiotics via a strong ligament. Exoccipitals. — Exoccipitals forming the major portion of posterior wall of cranium. Exoccipitals articulating with each other above foramen magnum, which they completely enclose. Exoccipitals articulating dorsolaterally with epi- otics, laterally with pterotics, and ventrally with basioccipital. Exoccipital partially overlain by opisthotic at common junction of exoccipital, epiotic, and pterotic on each side. Joint between exoccipitals and supraoccipital interrupted by an area of cartilage. Exoccipitals and basioccipital combining to form a bowl-shaped, centrum-like posterior face that serves as a point of articulation (through dense fibrous connective tissue) for first vertebral centrum. Each exoccipital extending anteriad on ventral surface of cranium between pterotic and basioccipital and meeting the prootic. Opisthotics ( =Intercalars) . — Opisthotics lying astride common junction be- tween exoccipital, epiotic, and pterotic on each side. A ligament connecting opisthotic to posttemporal is present on each side. An anteroventral projection of opisthotic lies astride pterotic-exoccipital joint but does not reach prootic. Basioccipital. — Basioccipital forming posterior portion of cranial floor. Basioc- cipital meeting prootics and parasphenoid anteriorly and exoccipitals laterally and dorsally, forming, with exoccipitals, articulatory surface for attachment of vertebral column. Pterotics. — Pterotics forming posterolateral corner of skull roof, articulating anteriorly with sphenotics and frontals, medially with epiotics, posteriorly and ventrally with exoccipitals and opisthotics, and anteroventrally with prootics. Junction of epiotics and pterotics within posttemporal fossae on each side. Dor- sal troughlike surface of pterotic carrying temporal cephalic laterosensory canal from frontal to extrascapular. A shallow troughlike depression in pterotic receiving posterior head of hyomandibular. Sphenotics (Autosphenotics). — Sphenotics forming lateral margin of posterodor- sal orbit at anterolateral corner of cranial vault and largely overlain by lateral margin of frontals. Sphenotics meeting pterotic posteriorly, prootic ventrally, and pterosphenoids anteromedially within the orbit. Dermosphenotic (eighth infraorbital) connected to anterior margin of sphenotic and frontal, not overlying sphenotic. Sphenotic receiving and supporting anterior head of hyomandibular. Prootics. — Prootics extensive, forming most of posteroventral wall of orbit. JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 41 Each prootic meeting its fellow in midline, dorsal to the myodome. Each prootic meeting parasphenoid ventrally, basioccipital and exoccipital posteriorly, and pterosphenoid, sphenotic, and pterotic dorsally. Basisphenoid. — Basisphenoid (not illustrated) a narrow splintlike bone overly- ing articulation of prootics in midline of ventroposterior wall of orbit and with basisphenoid pedicel articulating with parasphenoid in ventral midline. Basi- sphenoid lacking in Odontostomops. Parasphenoid. — Parasphenoid elongate, forming much of ventral contour of cranium. Parasphenoid lying between ethmoid cartilage and vomer anteriorly. Posteriorly two dorsolateral wings of parasphenoid meet with prootics. Para- sphenoid articulates with and partly overlies basioccipital in ventral midline of skull. Pterosphenoids. — Pterosphenoids forming bony posterior wall of orbit dorsal to prootics and meeting frontals dorsally, sphenotic laterally, and prootics ven- trally. Pterosphenoids expanded medially in Coccorella atrata, nearly meeting in midline of posterior wall of orbit. Pterosphenoids not extended medially and widely separated in other evermannellid species. Otoliths. — Due to a lack of fresh material, no detailed study of otoliths in evermannellids was possible. In cleared and stained material examined by me the sagittae are large and quite evident. An otolith of Evermannella indica is pictured in Kotthaus (1967). Superficial Dermal Bones This section includes descriptions of the superficial dermal bones of the snout and orbital region. Included are the following bones: infraorbitals (IO, IO-l to 8), nasals (NA), and supraorbitals (SO). The tubular-eyed Scopelarchidae invari- ably possess two sclerotic bones on each side, but no evermannellid possesses sclerotic bones. Nasals. — Nasal bones thin, troughlike elements, one on each side of snout above dermethmoid and just anterior to anterior margin of frontals. Two pores in posterolateral wall of each nasal marking presence of supraorbital laterosen- sory canal, received by nasals from frontals. Supraorbitals. — Supraorbitals elongate, strutlike, slightly expanded ventrally, and noticeably expanded dorsally (fig. 10). Supraorbitals connected via loose connective tissue to anterodorsal margin of first infraorbital ventrally. Supraor- bital abutting on and connected to dorsolateral frontal ridge dorsally and over- lying anterior margin of lateral ethmoid medially. Supraorbital apparently lack- ing in Odontostomops. Infraorbitals. — Evermannellids possess eight infraorbital bones on each side, including the lachrymal (IO-l) and dermosphenotic (IO-8). All carry in turn a segment of the infraorbital laterosensory canal. Infraorbital-1, an elongate ob- long platelike element with a raised shelf of bone along dorsolateral margin (fig. 10). One to three pores, varying by species, piercing IO-l. Infraorbital-1 the largest and IO-2 the next largest infraorbital elements; IO-2 basically triangular in shape, with apex directed posteroventrally (fig. 10), except in Odontostomops normalops in which all members of infraorbital series are only partially ossified and are irregular in outline. Infraorbital-2 forming posteroventral corner of in- ^\Jk? 42 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 43 fraorbital series. Infraorbital-3 to -8 basically similar in outline, with a central plate pierced by a single pore and with anterior and/or posterior margins curved laterally into bony ridges. Mandibular Arch The mandibular arch consists of the upper and lower jaws. Paired elements of the mandibular arch include the following bones: premaxillaries (PMX), maxil- laries (MX), supramaxillaries (SMX), dentaries (D), articulars (AR), angulars (AN), Meckel's cartilages (MC), and coronomeckelian bones (CMN). Upper Jaw — Premaxillaries. — Each premaxillary an elongate, very narrow and thin dentigerous bone, tapering to a point posteriorly (fig. 11 A). Premaxillary and maxillary largely separated anteriorly, but premaxillary largely overlain by maxillary posteriorly. Premaxillary and maxillary sharply curving medially and dorsally at anterolateral corner of snout. Premaxillary expanded anteriorly into a bladelike ascending process. A triangular cartilaginous element (here termed the rostral cartilage) attached to but not fused with each premaxillary ascending process. The stout interpremaxillary ligament is largely attached to the rostral cartilage of each side. A short but stout ligament connects the rostral cartilage and maxillary on each side. Premaxillary teeth small, uniserial, retrorse an- teriorly but typically straight posteriorly. All evermannellid species with an edentulous area on anterior premaxillary centered at the point where premaxil- lary turns sharply mediad. Upper Jaw — Maxillaries. — Each maxillary an elongate, thin, and narrow bone similar in shape to the premaxillary over most of its length. Anteriorly the maxillary ends in four articulating processes (fig. 11): a dorsoposterior process connected via a strong ligament to the palatine, a medially directed dorsoan- terior process connected via a ligament with the rostral cartilage, and two ven- trally directed processes, one lying lateral and the other medial to the base of the ascending process of the premaxillary. Upper Jaw — Supramaxillaries. — One supramaxillary. Supramaxillary rather large, one-third to one-fourth of maxillary length, deep and bladelike pos- teriorly, narrow and spikelike anteriorly. Lower Jaw — Dentaries. — Dentaries the largest and most complex bones of the lower jaw. Dentaries consisting of a dorsal dentigerous ridge and posterolateral and anteroventral sheets of bone (fig. 11). When viewed medially, the antero- ventral sheet curves abruptly laterad at its dorsal margin, forming a narrow, slightly oblique platform on which rests the anterodorsal process of the articular and the anterior portion of Meckel's cartilage. This platform ends at the ventral margin of the posterolateral sheet of bone. In Evermannella and Odontostomops dentary teeth partially biserial with four or fewer smaller fangs anteriorly, and these followed by 10 or fewer large barbed fangs (largest in length anteriorly and decreasing in length posteriorly), with the larger fangs bordered anterolateral^ by six or fewer smaller teeth. In Coccorella dentary teeth uniserial, and none of the fangs are barbed (fig. HE). A row of partially ossified replacement teeth occurring medially to row of fangs. Preoperculomandibular sensory canal par- tially encased in bone on anterolateral face of dentary. A distinct oval fossa (fig. 11F) at dentary symphysis in all species of Evermannella and lacking in Coccorella and Odontostomops. «5 J 5 T" 1 uj 53 01 i—l CO | 6 5 "§ Q "^ii ON (O^ R . R ON o o G E a V . — ih ^ e o | Bcrio" no JB § W « «. bo™ & S£ ^ j s E>- .» LiJ SI SO £.5 1 C wig *u <3 X, CO w 8 | * g I fr =§.£ £ .2* >,« c « -C R a; •? 13 ■§ D J JS o 44 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 45 Lower Jaw — Articulars. — Articulars forming posterior two-fifths of lower jaw. Each articular divided into dorsal and ventral portions by the elongate bar of Meckel's cartilage (fig. 11B). Ventral portion tubular posteriorly, completely en- casing the articular portion of the preoperculomandibular sensory canal in bone, and bladelike anteriorly. A dorsoposterior projection, the retroarticular process, supporting the articulation of the quadrate and receiving the preoperculoman- dibular sensory canal from the preopercle. Dorsal to the articulation of the quad- rate, a dorsal process of the articular is connected via strong ligaments to the distal end of the maxillary and posterior terminus of the dentary. Articular and dentary articulating anteriorly in a complex pattern of overlapping projections (fig. 11B). Lower Jaw — Angulars (Retroarticulars) . — Each angular a small, irregularly shaped bone on ventromedial surface of articular and at posteroventral corner of lower jaw. Angular connected via a ligament to interopercle. Lower Jaw — Coronomeckelian Bones (Sesamoid Articulars). — Each coronomecke- lian bone a small, flattened, nodular element lying lateral to Meckel's cartilage, just anterior to retroarticular region of articular, and on the approximately hori- zontal line separating the dorsal and ventral portions of the articular. A strong ligament connecting coronomeckelian bone with pars mandibularis of adductor mandibulae muscle. Palatine Arch The palatine arch includes the ectopterygoid (ECP), mesopterygoid (MSP), metapterygoid (MTP), and palatine (PAL) (fig. 12A). Elements of the palatine arch are essentially identical in size and shape in all evermannellid species. Only the palatines bear teeth. Metapterygoids. — Each metapterygoid inserted between ventral shaft of hyomandibular (which is partially overlapped by the metapterygoid), dorsal border of quadrate, and posterior border of mesopterygoid. Mesopterygoid and metapterygoid not overlapping in Evermannella and Odontostomops, but in Coc- corella, metapterygoid overlapping medial surface of mesopterygoid. Meta- pterygoid large and well ossified, except in Odontostomops in which the metapterygoid is membranous and only partly ossified, forming a major brace between the hyoid and palatine arches. Mesopterygoids. — Each mesopterygoid bladelike, membranous, poorly ossified at dorsal and posterior margins, tapering to a splintlike process anteriorly, and entirely supported by dorsal surface of ectopterygoid. Ectopterygoids. — Each ectopterygoid an elongate, strong, strutlike bone par- tially hollowed to form a troughlike surface on posterior and dorsal surface. This concave surface receiving anterior margin of quadrate ventrally and supporting the mesopterygoid dorsally. Splintlike posterior portion of palatine attached to anteroventral surface of ectopterygoid. Ectopterygoid splintlike anteriorly, lying along dorsal margin of palatine for nearly one-half the length of palatine. Palatines. — Each palatine an elongate, massive, dentigerous bone, forming (with the dentary) the principal bite of evermannellid species. Palatine teeth uniserial, decreasing in length from anterior to posterior, with replacement teeth forming medially to the active row. Palatine teeth rather loosely attached, all but the posteriormost teeth depressible. Anteriormost palatine teeth remarkably u") ~ fiotf 6c ; a. -tI^cj S3 a; tn v ^^ I—, 46 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 47 enormous barbed (except Coccorella) fangs and are by far the largest teeth occur- ring in evermannellids. In Coccorella the anteriormost palatine fangs are enor- mously large (the palatine fangs in Coccorella are the largest teeth relative to body length found in evermannellids: longest palatine tooth 7.1% to 10.0% SL in Coccorella cf . 4.6% to 7.3% in other evermannellid species) but are unbarbed (fig. 12B). A dorsal cartilaginous head of the palatine articulates with and is con- nected to the lateral ethmoid. Anterior to this head, connecting with the lateral ethmoid, is an ascending process articulating directly with the ethmoid cartilage. A concave anteroventral articular surface connecting with and attached to a convex dorsolateral articular surface of vomer. Contrary to the statement of Gosline et al. (1966, p. 3), I find no anterior process of the palatine directed upward and laterally to overlap the proximal end of the maxillary in any evermannellid species. Gosline et al. (1966) supposed that such a connection was typical for myctophiform fishes. Johnson (1974c, p. 42) reports a modified version of this condition in scopelarchids. In evermannellids, however, although the posterodorsal process of the maxillary is connected (via a ligament) to a short anterolateral process of the palatine, there is no direct articulation. Opercular Apparatus Bones of the opercular apparatus include the interopercle (IOP), preopercle (POP), opercle (OP), and subopercle (SOP) of each side (fig. 12C-E). Opercles. — Each opercle a thin, membranous bone, incompletely ossified at ventroposterior margin, typically resulting in a serrate margin. A disc-shaped articulatory apparatus at about center of anterior margin at the point where opercle is connected to posterior arm of hyomandibular. A bony ridge directed ventroposteriad from articulatory disc. Opercle considerably larger than subopercle in all evermannellid species. Opercle essentially rectangular in out- line, except at anterodorsal margin where anterodorsal border of opercle is directed at a sharp diagonal from posterodorsal corner of opercle to just above articulatory disc. This arrangement results in a large, roughly triangular area bordered but not covered by preopercle, extrascapular, supracleithrum, and opercle. Directly beneath the skin covering this area is the large levator operculi muscle (see Winterbottom, 1974) that inserts on and covers most of the dor- somedial surface of the opercle. Dorsoposterior corner of opercle overlying and supported by lateral face of supracleithrum. Anterior margins of opercle and subopercle closely adjacent to posterior margin of preopercle except for dorsoan- terior margin of opercle. Subopercles. — Each subopercle a thin, membranous bone, incompletely os- sified posteriorly, typically resulting in a serrate posterior margin. Dorsal portion of subopercle overlain by and attached to opercle. Subopercle partly overlying and connected to posterior border of interopercle. Pectoral insertion in all ever- mannellids at or below a horizontal line through center of subopercle. Interopercles. — Each interopercle bladelike and membranous, roughly V- shaped, with one limb directed dorsoventrally, the other limb anteriorly. A bony thickening at center of interopercle marking origin of a very strong ligament connecting interopercle to angular. A strong ligament connecting medial face of interopercle to interhyal. Lateral face of interopercle connected via loose connec- tive tissue to ventral portions of preopercle and quadrate. 48 FIELDIANA: ZOOLOGY Preopercles. — Each preopercle vertically elongate, somewhat expanded ven- trally, carrying preopercular section of preoperculomandibular laterosensory canal. Preopercle supported by hyomandibular dorsally and posteroventral process of quadrate ventrally. Sensory canal carried in a troughlike structure made up of dorsal three-fourths of preopercle, entering into a short, bony canal ventrally and emerging onto a slightly expanded ventral plate from which the canal passes to retroarticular portion of articular. Hyoid Arch The hyoid arch includes the following elements: basihyal (BH), basihyal toothplate (BHTP), branchiostegal rays (BRR), ceratohyal (CH), dorsal hypohyal (DHH), epihyal (EH), hyomandibular (HYOM), interhyal (IH), quadrate (Q), symplectic (S), urohyal (UH), and ventral hypohyal (VHH). All but the basihyal, basihyal toothplate, and urohyal are paired. Hyomandibulars. — Each hyomandibular with two stiff, flattened rods of bone dorsally, anterior rod articulating with sphenotic, posterior rod with pterotic (fig. 13E — G). A thin sheet of bone connecting ventral portions of these two rods. Hyomandibular with two stiff, flattened rods ventrally, a ventral shaft articulating (through cartilage) with interhyal and symplectic, and a postero- ventral shaft articulating with and supporting opercle. A stout ridge of bone extending vertically along lateral face of main dorsoventral axis closely connect- ing with and supporting preopercle. A thin sheet of bone extending anteriorly, underlying and supporting dorsoposterior corner of metapterygoid. Symplectics. — Each symplectic an elongate, well-ossified splint, articulating with medial face of posteroventral process of quadrate. Level of hyoman- dibular-symplectic joint about the same as the level of quadrate-metapterygoid joint. Quadrate. — Each quadrate lying ventral to metapterygoid and composed of the usual three main parts: body, posteroventral process, and articular head. Body delta-shaped, articulating dorsally through a synchondral joint with meta- pterygoid and fitting anteriorly into a shallow, troughlike depression formed by posterior surface of ectopterygoid. Posteroventral process large, well ossified, receiving preopercle on posteroventral face and symplectic on anterodorsal face. Articular head with two strong condyles, one lateral and one medial, separated by a concave surface and fitting convex articulatory surface of retroarticular process of articular. Interhyals. — Each interhyal rodlike dorsally and flattened, rounded, and somewhat expanded ventrally (except Coccorella in which symplectic is rodlike over entire length, with no portion distinctly expanded). Interhyal bound to hyomandibular-symplectic joint dorsally and to posterior articular facet of epihyal ventrally. A bony protuberance at about midlength of interhyal marks interhyal insertion of interoperculo-interhyal ligament. Epihyals. — Epihyal stout, wedge-shaped, articulating via synchondral joint with ceratohyal. Epihyal equal to or greater than ceratohyal in length in Ever- mannella but less than ceratohyal in length in Coccorella and Odontostomops. Ceratohyals. — Ceratohyal similar to epihyal in shape, articulating anteriorly with hypohyals. There is no ceratohyal foramen. Hypohyals. — A dorsal hypohyal and slightly larger ventral hypohyal im- mediately anterior to each ceratohyal. A ventroanterior knob on ventral E E •5 « «j — O 6 fO T3 . to £ tf. s» £3: .en CO '-' •*. *« O^ Si «5^ • § ^ I .§ •a tn "« •5; yso-s w a to c Iw E 1 »*>. § .£ & E « co o Hrto sa vo Sn « I— I ^ en c <£ ^) 49 50 FIELDIANA: ZOOLOGY hypohyal connected via a strong ligament to corresponding (right or left) fork of anterior projection of urohyal. Branchiostegal Rays. — Eight branchiostegal rays on each side. Branchiostegal rays acinaciform and distributed in the 4 + 4 pattern described by McAllister (1968), with four external epihyal branchiostegal rays and four ventral or internal ceratohyal branchiostegal rays. Epihyal branchiostegal rays with a flattened, much expanded head (lacking on ceratohyal branchiostegal rays, which are rodlike dorsally). The anteriormost epihyal branchiostegal ray inserts at the ceratohyal-epihyal joint in Coccorella and Odontostomops and posterior to this joint in Evermannella. Urohyal. — Urohyal short (Coccorella atrata) to elongate (Evermannella indica), much less than ceratohyal in length in C. atrata, about equal to ceratohyal in length in C. atlantica, and distinctly greater than ceratohyal in length in Odonto- stomops and Evermannella. Urohyal bladelike posteriorly, with posterior margin irregularly ossified or not ossified (O. normalops). A rodlike structure anteriorly, divided into right and left projections, as in a yoke, each projection connected via a ligament to corresponding ventral hypohyal. Basihyal. — Basihyal much reduced in evermannellids, a rodlike structure con- nected via a hingelike joint to anterior margin of first basibranchial and via ligaments on each side to each dorsal hypohyal. Basihyal only half ossified, cartilaginous anteriorly. A distinct basihyal toothplate covering dorsal and dor- solateral margins of basihyal except in Coccorella atlantica (basihyal toothplate over only posterior two-thirds of basihyal) and C. atrata (basihyal toothplate apparently lost). No basihyal teeth. Branchial Arches Endoskeletal components of the branchial arches include the following: basi- branchials (BB), ceratobranchials (CB), epibranchials (EB), hypobranchials (HB), and infrapharyngobranchials herein referred to as pharyngobranchials (PB) (fig. 14). Dermal elements associated with these bones include nodules or plates supporting two types of dentition: gill teeth (GT) vs. conical teeth (CT). Gill toothplates, developed as small nodules of bone supporting one to several small teeth in a uniserial row, occur on the second ceratobranchial and are limited to this bone, except in Evermannella balbo and E. indica in which the largest gill toothplate lies astride the articulation of the ceratobranchial and hypobranchial of the second arch (fig. 15E). Conical teeth, so-called to distinguish the larger single-based teeth of the gill arches from the gill teeth, are invariably present on pharyngobranchials 3 and 5 and are present or absent (table 6) on ceratobran- chial 5, epibranchial 3, and pharyngobranchial 4. A toothplate of the fourth basibranchial (i.e., over the cartilaginous third basibranchial copula, see Nelson, 1969a) is unique to Evermannella balbo among evermannellids (fig. 14C). All coni- cal teeth are no doubt associated with dermal toothplates that may (CB5, EB3, PB3) or may not (PB4, PB5) be fused with their respective endoskeletal elements. Basibranchial Series. — Lingual teeth lacking in all evermannellid species. A single dermal toothplate overlying first basibranchial, indistinguishably fused with second basibranchial, and overlying anterior one-half or more of third basibranchial (fig. 14). Compound second basibranchial element easily the longest member of basibranchial series, with the result that the distance between JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE TP 51 Fig. 14. Branchial arch elements in Evermannellidae. A, Branchial arch elements in Coccorella atrata, SIO 68-533, 99.6 mm. B, Basihyal and first basibranchial in C. atrata as above. C, Tooth plate of third basibranchial copula in Evertnannella balbo, ISH 546/73, 102.0 mm. D, Posteroventral branchial arch elements in E. balbo as above. ventral portions of first and second arches is noticeably greater than the distance between ventral portions of any succeeding pair of gill arches. A cartilaginous area, apparently single and median, lying posterior to third basibranchial and representing third copula of basibranchial series. This element receiving car- tilaginous articular heads of fourth and fifth ceratobranchials. In Evertnannella balbo a fourth basibranchial toothplate, bearing a patch of six to nine conical teeth, lies embedded in the skin over this cartilaginous element (fig. 14C). Hypobranchials. — There are three paired hypobranchials corresponding to the first three gill arches, decreasing in length from the anterior pair to the posterior pair. Hypobranchials 1 and 2 articulating through cartilage with cartilaginous O J Q LLi 52 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 53 Table 6. Distribution of branchial teeth in evermannellids. Abbreviations for branchial arch elements given in text. Species Mrn.-hi.il arch element 1 2 3 4 5 6 7 GT/CB2 + + + + + + + GT/HB2 + + ? CT/PB5 + + + + + + + CT/PB4 + + + + ? + CT/PB3 + + + + + + + CT/EB3 + + + 7 CT/CB5 + + CT/BB4 tooth plate + KEY: +=present, 0=absent, ?=unknown, \=Coccorella atlantica, 2=C. atrata, 3=Ever- mannella ahlstromi, 4=E. balbo, 5=E. indica, 6=E. megalops, 7=0dontostomops normalops, GT=gill teeth, CT=conical teeth. areas between basibranchials 1 and 2 and 2 and 3, respectively, beneath the single basibranchial toothplate. Articulation of first two pairs of hypobranchials complete in that hypobranchial of each side articulates (through cartilage) with its fellow of the opposite side medially and with corresponding basibranchial element anteriorly. Third hypobranchials closely attached to third basibranchial and parallel it along half or more of its length. Third hypobranchial converging anteroventrally with its fellow, and the two tightly bound together beneath third basibranchial. Posteriorly, third hypobranchial of each side articulating (through cartilage) with third ceratobranchial laterally and third basibranchial copula medially. Ceratobranchials. — There are five paired ceratobranchials, with the first four essentially identical in shape and articulating dorsally through a considerable length of cartilage with corresponding epibranchials. Fifth ceratobranchial (lower pharyngeal), no doubt a compound bone, a rod-shaped element bearing a distinct medially directed flange at about midlength. Fifth ceratobranchial edentulous, except in Odontostomops normalops and Evermannella balbo where one (O. normalops) to several (£. balbo) conical teeth are present on lateral flange. Epibranchials. — There are four paired epibranchials. Epibranchials of first gill arch an elongate Y-shaped bone connecting dorsally through a ligament to an- terior end of third pharyngobranchial. A rodlike projection (uncinate process of Rosen, 1973) at midlength connects posteriorly through a ligament to ventroan- terior fork of second pharyngobranchial. Epibranchial of second arch elongate and with a distinct keel over ventral one-half of length in Evermannella (keel poorly developed in Coccorella and Odontostomops). A short, forklike anterior projection articulating with ventroposterior limb of second pharyngobranchial. A markedly elongate projection (= uncinate process) extending dorsomedially and attaching second epibranchial to third pharyngobranchial at about the mid- dle of the latter bone. Epibranchial of third arch elongate, unkeeled, rodlike, connecting dorsally to posterior edge of third pharyngobranchial, between the third and fourth pharyngobranchial toothplates. Several evermannellid species (table 6) possess conical teeth occurring at about midlength on third epibranchial. 54 FIELDIANA: ZOOLOGY Epibranchial of fourth arch a stout, Y-shaped bone. A dorsoposterior projec- tion at about midlength of fourth epibranchial forming short leg of the "Y," and the longer limbs of the "Y" connect respectively with fourth ceratobranchial ventrally and fourth pharyngobranchial dorsally. A distinct keel present on ventral limb. Fourth epibranchial connecting (through cartilage) with cartilagin- ous fourth pharyngobranchial and for the most part not closely associated with the large fifth pharyngobranchial toothplate. Pharyngobranchials. — There are three (or questionably four) paired pharyngo- branchials. First (suspensory) pharyngobranchial absent or, if present, reduced to an exceedingly minute nodule of cartilage. This element is both difficult to see and to distinguish from cartilaginous dorsal terminus of first epibranchial. Rosen (1973, fig, *12) illustrates a first pharyngobranchial in a specimen identified as Evermannella sp. Although I have seen what appears to be a separate cartilagin- ous element dorsal to first epibranchial in several evermannellid species (Ever- mannella balbo, E. indica, Odontostomops normalops), such an element is not dis- cernible in the other species examined, nor is it ever ossified, nor is it as large as Rosen's figure indicates. Second pharyngobranchial an edentate, Y-shaped bone, with longest limb connecting dorsally with third pharyngobranchial, an anteroventral limb con- necting through a ligament to first epibranchial, and a posteroventral limb ar- ticulating with second epibranchial. Third pharyngobranchial easily the largest pharyngobranchial bone and providing the principal dorsal support for the gill arches. Conical teeth invari- ably present on third pharyngobranchial but in no evermannellid species are they as large or as numerous as conical teeth on fifth pharyngobranchial toothplate. Conical teeth of third pharyngobranchial limited to posterior one- fourth of ventral surface of that bone. A cartilaginous fourth pharyngobranchial and an associated but separate fourth pharyngobranchial toothplate present in all evermannellid species. Fourth pharyngobranchial articulating posteriorly with fourth epibranchial and anteriorly with third pharyngobranchial. Toothplate of fourth pharyngobran- chial narrow but well ossified in all evermannellids, roughly trough-shaped, enclosing (ventral) surface of fourth pharyngobranchial. Fourth pharyngobran- chial toothplate provided with strong conical teeth except in Coccorella atrata in which the toothplate is present but edentate. There exists no identifiable fifth pharyngobranchial, but the toothplate of the fifth pharyngobranchial (see Nelson, 1969a) is by far the largest pharyngo- branchial toothplate and bears the largest and most numerous conical teeth of any gill arch element. Rosen (1973, p. 407) apparently did not see the fourth pharyngobranchial toothplate in his evermannellid material and thereby misidentified the fifth toothplate as the fourth (Rosen, 1973, pp. 407, 435). A dorsomedially directed flange from anterior terminus of fifth toothplate closely paralleling a similar projection on fourth toothplate and thus partially encloses (ventrally) anterior end of fourth epibranchial and posterior portion of fourth pharyngobranchial. Vertebrae, Supraneurals, Intermuscular Bones, and Caudal Skeleton Vertebrae. — The number of vertebrae varies between 45 and 54 (table 3). Point of separation between precaudal and caudal vertebrae taken as the first centrum, JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 55 with a complete haemal arch being termed first caudal vertebra. This point could not be determined in radiographs. For cleared and stained material the following precaudal + caudal vertebral counts were obtained: Coccorella atlantica (16 + 34), C. atrata (16 + 31), Evermannella balbo (17 + 36), E. indica (17 + 33), Odontostomops normalops (16 + 33). These counts include the fused first preural plus first ural element (PUi + U,) but do not include the second ural element (U 2 ). All centra large, amphicoelous, pierced centrally by a large notochord canal. First precaudal centrum complete. Neural arch elements on all centra as deep as or deeper than spool-shaped centrum. Neural arch prezygapophyses and postzygapophyses well developed, rigidly interlocking, and strengthening the vertebral column. Neural spines present on all centra, but those of first three centra unfused distally. Each precaudal centrum bearing bilateral large ventrolaterally directed parapophyses. Most precaudal parapophyses articulating with and providing support for an epipleural and a pleural (ventral) rib. First (anteriormost) precaudal centrum lacking both pleural and epipleural ribs. Second precaudal centrum bearing an epipleural rib but lacking a pleural rib. Parapophyses of eleventh precaudal centrum prolonged ventrally as spinelike processes, with parapophyses of succeeding precaudal centra sequentially longer than those on preceding centra. Sequence terminated with distal fusion of these processes on first caudal vertebra. Intermuscular Bones and Ribs. — First precaudal centrum bearing epineural (EN) and epicentral (EC) intermuscular bones (fig. 16). Second centrum bearing only an epipleural (EPR) rib. Succeeding precaudal centra bearing an epipleural and a pleural (PLR) rib. Additional intermuscular bones absent. Caudal vertebrae lacking intermuscular bones, except that first caudal vertebra may or may not bear an epipleural rib. Supraneurals. — Supraneurals (SN) invariably two, the first inserted between neural arch elements of first precaudal vertebra, the second between neural arch elements of third precaudal vertebra. Caudal Skeleton. — The six autogenous hypurals (HYP 1 to 6) and the autogen- ous parhypural (PH) support the 1+9 + 8 + 1 (counting from dorsalmost ray) principal caudal rays. The sixth (dorsalmost) hypural is greatly reduced to a very small wedge-shaped element (fig. 16E). First preural centrum (PUO, first ural centrum (UO, and first pair of uroneurals (UN!) fused into a single bone — the stegural (ST) of Monod (1968). The first uroneurals are fused with the stegural ventrally but are separate distally (fig. 16G). Thin bladelike second uroneurals (UN 2 ) attached to but not fused with dorsoposterior lateral faces of respective (right or left) first uroneurals. Stegural supporting the parhypural and first two hypurals. Third hypural inserted between second hypural, stegural, and ventral margin of second ural centrum. Second ural centrum (U 2 ) invariably present and free in all evermannellids, inserted behind and partly between the separate uroneural wings of stegural. Second ural centrum expanded and centrum-like ventrally but rodlike dorsally (fig. 161). A marked notch in posterior border of second ural centrum receiving and supporting fourth hypural. Fifth hypural inserted between fourth hypural and uroneural portion of stegural. Sixth hypural much reduced, a wedge-shaped bony nodule attached to posterodorsal margin of fifth hypural. Only one epural (EP) present in evermannellids — an oblong platelike element situated dorsally between stegural and neural spine of third preural vertebra. Neural spine of second preural vertebra represented by a A c S o 5 | 5*2 2 3 en T3 . g 6 * a ~ J 1 «S c „ o > SO t c 0> w «*? > CM oi3 o H jri 01 «3 01 nun B 01 B S wo ► e? •« ro 3 oi II S*oJ a &.- ill ' ' T3 01 o 3 " •,-o JS =3 v© 3 ■a "* is 8 £ 8 0) >-. >H 01 >sO CL, > ^ a S u o oi ra is C P 56 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 57 broad spatulate projection, tapering to a point distally, and about one-half the length of preceding neural arch and spine elements. Dorsal Fin Evermannellids exhibit a range of 10 to 13 dorsal-fin rays (table 2). Last dorsal-fin ray divided completely to base and counted as one element. Number of pterygiophores equal to number of fin rays (FR), but whereas the greatly expanded first (anteriormost) pterygiophore supports the first two dorsal-fin rays (fig. 17), the last two pterygiophores support the posteriormost divided ray. First pterygiophore consisting of an elongate bladelike proximal radial (PR) that is enormously expanded ventrally and two distal radials (DR) supporting the first two dorsal-fin rays. First pterygiophore inserted over seventh through tenth precaudal vertebrae in Coccorella atlantica. All distal radials consisting of a spheri- cal central cartilage, with two hemispherical bony capsules providing the ar- ticulatory surface. Second pterygiophore consisting of an elongate proximal radial and a single distal radial, providing support for third dorsal-fin ray. Suc- ceeding pterygiophores composed of proximal, medial (MR), and distal radials except for the terminal pterygiophore that consists of a single proximal radial. A single distal radial and the last two pterygiophores supporting the posteriormost dorsal-fin ray. Anal Fin Anal-fin rays 26 to 37 (table 3). Posteriormost anal-fin ray divided completely to base and counted as one element. Number of pterygiophores one less than the number of anal-fin rays — first (anteriormost) pterygiophore supporting first two anal-fin rays, final pterygiophore supporting terminal divided anal-fin ray. Anteriormost pterygiophore consisting of an elongate, rodlike proximal radial, inserting between haemal spines of 19th to 20th to 21st to 22nd vertebrae, depending upon the species, and two distal radials supporting first two anal-fin rays. Second pterygiophore consisting of proximal and distal radials only, sup- porting third anal-fin ray. Third and successive (fourth and successive in Coc- corella atlantica) pterygiophores consisting of proximal, medial, and distal radials. Final two or three distal radials lacking ossified bony caps. Pectoral Girdle The dermal elements of the pectoral girdle include the posttemporal (PT), supracleithrum (SCL), cleithrum (CL), and three postcleithra (PCL) — dorsal (DPCL), medial (MPCL), and ventral (VPCL). Endochondral elements include the scapula (SC) and coracoid (COR). One extrascapular (ESC) is associated with but not part of the pectoral girdle. In evermannellids the insertion of the pectoral fin lies just above the ventral contour of the body, and the line of insertion of the pectoral-fin rays is rather more horizontal than vertical. Marshall (1955) gives 15° to 20° as the value of the angle between the horizontal axis of the body and the axis of the pectoral-fin insertion in Evermannella balbo. Posttemporal. — Posttemporal unforked, consisting of two portions: (1) a rod- like dorsal articulating process and (2) a slightly expanded ventral bladelike area (fig. 18). Dorsal process of posttemporal converging medially with its fellow of DC CE DC DC Q- 5 Q LL 58 JJ.S X « _ £ «J 6 "5 = 1*1215 g .in c 2 ™^ , "J u u P a; c ^5 § :i — in > "0 n 59 60 FIELDIANA: ZOOLOGY the opposite side and loosely connected to the epiotic. Ventral bladelike area partially covering and loosely connected to dorsal tip of supracleithrum. Typi- cally a raised tubelike bony process present on posttemporal and encasing laterosensory canal which posttemporal receives from extrascapular and passes to supracleithrum. A strongly developed nodular bony process near anterome- dial margin of posttemporal marking insertion of strong posttemporal-opisthotic ligament. Extrascapular. — Only one extrascapular on each side present in evermannellid species. Extrascapular Y-shaped, partly troughlike, pierced by three pores, one pore in each limb of the Y. Three branches of the laterosensory system meet at the extrascapular, the temporal branch from the pterotic, a supratemporal branch from over the posterior portion of the cranium, and the main lateral line canal. Supracleithrum. — Supracleithrum an elongate, nearly straight, flattened, ob- long bone, slightly tapering dorsally and expanded and rounded ventrally. Supracleithrum overlying and loosely attached to rodlike dorsal process of clei- thrum. Dorsal postcleithrum attached to supratemporal medially. A raised tubelike bony process passes the laterosensory canal from the posttemporal to the lateral line. Rofen (1966d, p. 512) states that no bony processes on the posttemporal or supracleithrum are associated with the lateral line in everman- nellids. This statement is in error. There is invariably just such a process on the supracleithrum and typically such a process on the posttemporal (apparently absent in Coccorella atrata). Cleithrum. — Clei thrum the largest and most complex element of the pectoral girdle. Main shaft of cleithrum strongly arched, ending dorsally in a rodlike shaft lying medial to and attached loosely to the supracleithrum. A posteriorly directed process, the posterior lamina (PL) may (Evermannella, Odontostomops) or may not {Coccorella) overlap the lateral face of the medial postcleithrum. Ven- trally, the main shaft of the cleithrum curving sharply anteriad and slightly mediad, converging with its fellow of the opposite side. A medial shelf, the midcleithral process (MCP, fig. 18B), projecting posteriorly from main shaft and articulating dorsally with scapula and ventrally with coracoid. Scapula. — Scapula roughly oval in outline, articulating dorsally, anteriorly, and laterally with cleithrum, and ventrally with coracoid. A saddle-shaped raised bony process supporting the enlarged dorsalmost pectoral-fin ray. Scapula providing support for first (dorsalmost) proximal pectoral radial. Scapular foramen large, an elongate oval lying nearly in center of scapula. Coracoid. — Coracoid an expanded bony sheet, rounded concavely at anterior margin and convexly at posterior margin, narrowing to a bony rodlike process anteroventrally and connected to ventral tip of cleithrum. Coracoid overlying and strongly connected to posterior margin of midcleithral process. Coraco- cleithral fenestra moderately large but not readily visible in lateral view. A dis- tinct coracoid foramen present. A cartilaginous ridge along posterodorsal margin of coracoid supporting second, third, and fourth proximal pectoral radials. Postcleithra. — Three postcleithra — here termed dorsal, medial, and ventral. Each postcleithrum membranous, bladelike, ovoid, distinctly longer than wide. Ventral postcleithrum especially narrow and splintlike. Dorsal postcleithrum attached to medial face of supracleithrum. Each succeeding (medial, ventral) postcleithrum attached to medial face of preceding (from dorsal) postcleithrum. JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 61 Rofen (1966d, p. 512) illustrates only two postcleithra for a specimen of Everman- nella indica, but I find three postcleithra in my material of E. indica, as in all evermannellid species. Pectoral Radials. — The four proximal pectoral radials (PR) are roughly hourglass-shaped. Dorsalmost radial the smallest, with successively ventral ra- dials sequentially larger. Distal radials unossified, one such distal radial associ- ated with each pectoral-fin ray except no distal radial is associated with the first pectoral-fin ray. It seems likely that the distal radial of the first pectoral-fin ray is both ossified and fused with the medial half of the enlarged first ray. Pectoral-fin rays 11 to 13 (table 2). Pelvic Girdle Pelvic bones paired, flattened, elongate, lying in ventral plane of abdominal body wall, neither reaching to nor connected with pectoral girdle (fig. 19). An- terior region of each pelvic bone a thick lateral strut supporting a thin, narrow, medial bony shelf. Pelvic bones converging in midline anteriorly but are only Fig. 19. Pelvic girdle of evermannellids. A, Pelvic girdle, ventral view, Coccorella atlantica, UH 71/3/8, 93.3 mm SL. B, Pelvic girdle, ventral view, Evermannella indica, UH 70/12/31, 108.8 mm SL. 62 FIELDIANA: ZOOLOGY loosely connected. Posteriorly, pelvic bones consisting of two regions: (1) a lateral expanded ridge of bone providing support for the nine movable pelvic-fin rays and (2) a medial area of two very thin, bony lamellae encasing a cartilagin- ous plate. Cartilaginous plate of each side coalescing indistinguishably with its fellow in midline, forming a single common plate. Pelvic cartilaginous plate extending posteriad far behind ossified region of pelvic girdle — posterior exten- sion up to two- thirds the length of ossified region of pelvic girdle (fig. 19). A thickened area of bone, forming a bony strut extending medially from the region of insertion of first pelvic-fin ray on each side, with strut from each side nearly meeting in midline. Pelvic-fin rays numbered in order from lateral (anterior) to medial (posterior). Two bony or cartilaginous radials (R, fig. 19A) associated with pelvic-fin rays, the first between expanded bases of stout first pelvic-fin ray, the second (which may be partly ossified or entirely cartilaginous) between bases of fourth to sixth pelvic-fin rays. Seventh and eighth pelvic-fin rays supported ventrally by ex- panded base of ninth pelvic-fin ray. Gosline (1961, pp. 20, 21) points out that this expanded base is a compound element formed by fusion of the fin-ray base with the innermost radial. There is no curved splint of bone along the outer surface of the upper half of the first pelvic-fin ray as in Solivomer (Neoscopelidae, Gosline, 1961, p. 18) and most myctophids (Paxton, 1972, p. 32). INTERRELATIONSHIPS In attempting to infer relationships among evermannellid species and be- tween the family Evermannellidae and other iniomous families, I have applied the same methods of phylogenetic reasoning used in my earlier (Johnson, 1974c, pp. 199-220) study of the scopelarchids. Two fundamental difficulties char- acterize the following discussion. The first is that evermannellid genera as defined herein are distinctly monothetic (Mayr, 1969), for the most part defined by autapomorphous (Brundin, 1966) character states. Such uniquely diagnostic, presumably derivative features reinforce the concept of monophyly in the case of each evermannellid genus but do not aid the attempt to infer relationships among all evermannellid species. The second difficulty pertains to the attempt to determine relationships among iniomous families — it is that the survey of in- iomous taxa for many (seemingly) critical characters is woefully incomplete. Catalogue of Characters and Character States The list of characters compiled below is used for two purposes: (1) to sum- marize all available information relevant to the assessment of interrelationships among evermannellid species; and (2) as a basis for my discussion of the hypothesis (e.g., Gosline et al., 1966, p. 17) of the sister-group relationship between evermannellids and scopelarchids. The complete character state by OTU matrix is presented as Table 8, excluding characters for which states could not be adequately defined and those characters used solely in the attempt to infer relationships among evermannellid species. Characters catalogued below were selected specifically for the attempt to study scopelarchid/evermannellid relationships. Many other characters could have been included and will have to be included in any thoroughgoing study of interrelationships among iniomous taxa. JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 63 To avoid confusion in nomenclature and terminology, I list below the working classification of iniomous fishes adopted in this paper. The groupings adopted herein are for the most part those traditional among ichthyologists, but also reflect results of recent work on iniomous taxa, including the present paper: Aulopoidei: Aulopidae (Aulopus + Hime) Myctophoidei: Myctophidae, Neoscopelidae Chlorophthalmoidei: Chlorophthalmidae (Bathysauropsis + Chloropthalmus + Parasudis), Ipnopidae (including Bathypteroidae), Notosudidae, Scopelarchidae Synodontoidei: Bathysauridae, Harpadontidae (Harpadon + Saurida), Synodontidae (Synodus + Trachinocephalus) Alepisauroidei: Alepisauridae, Anotopteridae, Evermannellidae, Omosudi- dae, Paralepididae. This working classification assumes the monophyly not only of the 15 family-level OTU's separately listed but also each of the five groups accorded subordinal status and of the Iniomi itself. The character catalogue presented below is the basis for my discussion of the evidence for and against these hypotheses of monophyly and intergroup relationships implicit in this classifica- tion. Groupings proposed by other authors are referenced in the text by the author's (authors') initial in parentheses, e.g., Aulopiformes(R) refers to the order Aulopiformes as proposed by Rosen (1973). Only two papers, Gosline et al., 1966 ("G"), and Rosen, 1973 ("R"), are cited in this fashion. In addition to material examined personally (table 7), statements regarding character states exhibited by various iniomous taxa are based (1) on previous studies of interrelationship among some or all of these taxa (Harry, 1952; Mar- shall, 1955; Gosline et al., 1966; Goody, 1969; Rosen & Patterson, 1969; Paxton, 1972; Rosen, 1973; Johnson, 1974c; Marshall & Staiger, 1975; Zehren, 1975; Sulak, 1977); (2) on morphological data (Gregory, 1933; Marshall, 1960; Monod, 1968; Nelson, 1969a; Okiyama, 1974); and, in addition to the above, (3) on data presented for particular taxa, such as Alepisauridae (Regan, 1911; Gibbs & Wilimovsky, 1966; Rofen, 1966b); Anotopteridae (Rofen, 1966c); Aulopidae (Mead, 1966a); Bathysauridae (Mead, 1966b); Chlorophthalmidae (Taning, 1918; Mead, 1966e; Ahlstrom, 1971); Evermannellidae (Herring, 1977); Harpadontidae (Gibbs, 1959; Anderson et al., 1966); Ipnopidae (Regan, 1911; Mead, 1966c,d; Nielsen, 1966; Theisen, 1966); Myctophidae (Jollie, 1954; Moser & Ahlstrom, 1970, 1972; Nafpakritis et al., 1977); Neoscopelidae (Butler & Ahlstrom, 1976; Nafpaktitis, 1977); Notosudidae (Bertelsen et al., 1976); Omosudidae (Parr, 1929; Rofen, 1966b); Paralepididae (Harry, 1953; Rofen, 1966a); Synodontidae (Gibbs, 1959; Anderson et al., 1966). In the following catalogue of characters, states believed to be primitive are all identified with a "0"; states believed to be derived are designated by positive integers. Derived states recognized exclusively for evermannellid species are identified by the prefix "E." In attempting to determine "derivativeness" I fol- low criteria established by Marx & Rabb (1972, pp. 5, 6), relying most heavily on the following criteria: (1) uniqueness and (2) relative abundance. Assignment of character states to OTU's in Table 8 was based in part on the assumption that possession by one or more representatives of a particular OTU of a state consid- ered primitive indicates (except where contrary evidence can be cited) the primi- t^ >>c vO «— ' oo co rH ^^ i— I ^ " w rH ^ no Sn« oo d r* y S On K (N On v a\ S ffl £ w 6 w 3 3 . *5 r-t On m^l^ i[-^ > f <5 vD »C 92«s«2 2 <92 t/nnDUDtfl _) on en NO s rH CM *-^ (103.2) V-2, 1 (70.0) (95.0) 112.4) (68.0) 82 rH_00 rH rH h>-h h ^_ On"(N SIO 65-243, ORSTOM C SIO 68-534, UH 70-9-12, SIO 69-21, 1 UH 70-9-9, CM On O CO r\i in ^o ■* U. Uh •c ;a a nO^ I co^gin^^. KB* f\| ON r- I - LUU.LL.L-. ,N .N .§ »1 « 18 SP Sp fc -2 -5 2 ") to « ui S S -s: (0 tfi 2 lmid almu almu almu rucul rucul 5 2 <**«•** ** u s 55. ■S £ ^ 5 -a .» ft re fLg-frfrll 1* < B oSSS « « 9 < £ 2 uu uo< d_ 64 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 65 tiveness of that state for that OTU. This simplifying assumption was necessary due to inadequacy of survey for many characters for certain speciose families (e.g., Myctophidae, Paralepididae) and also to lack of material of seemingly critical taxa (e.g., Bathysauropsis, Bathysaurus). 1. Eye Morphology. — Among iniomous fishes the evermannellids (except Odontostomops) and scopelarchids are unique in possessing dorsally directed tubular eyes. Anteriorly directed tubular eyes are present in giganturids, a group included by Rosen (1973) among aulopiform(R) fishes but not treated in this paper. Dorsally directed semitubular eyes are present in Protomyctophum (Hierops) spp. (Myctophidae, see Wisner, 1976, p. 18). (A) States Recognized for Iniomous Fishes (0) = Eyes not tubular. (1) = Eyes tubular, directed dorsad, large to extremely large, provided with a transparent (in life) lens pad, with division of the retina into an optically adjusted main retina and a nonadjusted accessory retina. It should be emphasized that state (0) incorporates (artifically) an enormous range of variation in eye morphology, from Aulopus to Chlorophthalmus to lpnops. It should also be noted that the family Scopelarchidae is unique in that all species exhibit state (1). Fundamental to my interpretation of evermannellid interrelationships, whether among evermannellid species or between iniomous families, is my belief that the semitubular or tubular eyes of Coccorella and Evermannella , respec- tively, represent synapomorphous features, indicating sister-group relationship for these two genera and implying convergent and independent acquisition of tubular eyes by the Evermannellidae and Scopelarchidae. Marshall (1971, p. 42) lists 11 mesopelagic families (including the scopelarchids and evermannellids) that have ". . . independently acquired . . . tubular eyes, each fully stopped by a relatively large lens, and with the two main axes virtually parallel." My belief that the "normal" laterally directed eyes of Odontostomops normalops represent the primitive state for this character in the family Evermannellidae is based on the intuitively appealing notion that reversal from the tubular-eyed state 2 to the "normal" laterally directed state is less likely than the presumably sequential acquisition of binocularity implied by relationships postulated herein. Evidence is needed to support this notion. Such evidence as exists is found in Munk's (1965, 1966) work on the ocular morphology of Omosudis and certain evermannellids and scopelarchids. Munk (1966) lists the following points of similarity between the nontubular eyes of Omosudis and the fully tubular eyes of Evermannella: (1) retina composed almost entirely of cones (Omosudis) or highly modified cones (Evermannella)*; (2) presence in Omosudis (as in all evermannel- lids) of an adipose eyelid pierced by a small (Odontostomops) to moderately sized (Omosudis, Coccorella) to large (Evermannella) pore; (3) correspondence in position between ventral portion of retina in Omosudis vs. main retina of Evermannella. As 2 With concomitant changes in skull morphology (fig. 9); number and arrangement of cephalic laterosensory pores (fig. 1, table 1); development of the lens pad; increased size of aperture in the adipose eyelid; and, presumably (see discussion of "Vision" in preceding section), in behavior relating to detection and capture of prey (see Marshall, 1971, p. 44). 3 Scopelarchids possess the pure-rod retina typical for midwater fishes (see Munk, 1966; Marshall, 1971). x 01 01 )H r: ja a CD X _2 c at X 5 +"* ,3 C >> 73 Ol in t; X 2 5 x 73 73 s -a c In M 73 U c 73 E 0) 3 73 X o 01 u hi a. X X o ~3 2 X X j-i X 01 I. o 60 01 -5 01 73 > C XI , at -O ia ~o i: _o U o c ui o 01 73 -C X 01 s- «J 3 E o > •c c 01 ;s "0 73 "o 01 v- X) 01 O.T3 3 X C C H G a 3 O u t^ OfNOOvOOOrHOOOOOOOOOOO ■>D 0(NOO^DOOt-iOOOOOOOOOOO U1 rHfNOO\COOi-HOv£>OOOOOOOO00 ** OCNOO^Ot-vOt— iOOOOOOOCNOOO X .. 'R i. *- o ra ; 6 >. Si 01 73 ns 0> *- J- (A) (0 5 > U -43 — 3 TO >- X - c U a. 5 • 73 00 H C 01 tr> oooooixo(Nrooooa\Or-ioooo «N OOOOOtXOrMTTOOOOOOrHOOOO OtNOOOOOOOOOOOOOOOOO 66 CO ° II a £ « CI. 73 •£ ° £ 2 g II >, ° II 01 o re T3 'S- J*" ° 43 3 '^ • < c H ii 42 •§ <-• 31 •= .. B-g x X re 2 II £ oo i- .5 > UJ TJ 0? II oi re in C TJ <-i CC K - 0» »- O" "3 II 3 s .o o Si II •1*: n to nj ^* ^ ^ ^t* o 2 .22 - o a ^ o> . J< * II II so El — > E2 2. Lateral Division of Musculature — Evermannellids are apparently unique among iniomous fishes in possessing a complete, externally visible, tripartite division of the tail musculature such that the epaxialis and hypaxialis muscle masses are completely separated by an enlarged lateralis superficialis muscle (fig. 8; see discussion under Size and Habits). Due to inadequate survey for states of this character among iniomous fishes, I choose not to formally define states of character 2 in this paper. 3. Swimbladder — According to Marshall (1954, p. 323; 1960, p. 54) and Mar- shall & Staiger (1975, p. 110), the only iniomous fishes retaining a swimbladder are the neoscopelids Neoscopelus and Solivomer and the lantern fishes (Myc- tophidae). 4 The absence of a swimbladder, clearly a derivative feature, is thus characteristic of aulopiform(R) fishes. Rosen (1973, p. 452) states that the swimbladder and gas glands of neo- scopelids and myctophids are of the advanced "acanthopterygian type," thus supposedly bolstering his argument for inclusion of his Myctophiformes(R) with 4 Anderson et al. (1966, p. 31) state that the swimbladder is "small or absent" in synodontids(G) but provide neither details nor documentation. JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 69 "higher" neoteleosts (Paracanthopterygii + Acanthopterygii). Information presented by Marshall (1960, pp. 54, 55) and Marshall & Staiger (1975, p. 109) raises doubts about the "advanced" nature of the swimbladder of neoscopelids. For the time being two states of character 3 are recognized for iniomous fishes. There is no variation in this character among evermannellids (which lack a swimbladder). (A) States Recognized for Iniomous Fishes (0) = Swimbladder present. (2) = Swimbladder absent. 4. Squammation. — Four iniomous families (Alepisauridae, Anotopteridae, Evermannellidae, and Omosudidae) have completely lost all body scales. Among paralepidids, only members of the Tribe Paralepidini possess body scales. Primitively, at least, in all other iniomous families the body and postor- bital regions of the head are covered with scales. Rosen (1973, p. 506) argues that the presence of ctenoid scales in some myctophids (four species of Myctophum [Nafpaktitis et al., 1977], Notoscopelus japonicus [Tanaka, 1908] [see Nafpaktitis, 1975]) distinguishes the Myctophiformes(R) from the Aulopiformes(R) and allies the former with higher neoteleosts (Paracanthopterygii + Acanthopterygii). The possession of ctenoid scales by species in other iniomous families (including the Aulopidae [Mead, 1966a, p. 20], Chlorophthalmidae [Mead, 1966e, p. 162], and the Cretaceous Sardinioididae [Goody, 1969, p. 160]) offers little support for Rosen's conclusion. Given the strong possibility for convergent reduction or loss of body scales among the five alepisauroid families listed, the possession of "normal" cycloid scales distributed over the entire body in scopelarchids (not to mention myc- tophids and notosudids) as in benthonic inioms leads me to recognize two states of character 4 among iniomous fishes. (A) States Recognized for Iniomous Fishes (0) = Body and (typically) postorbital regions of head bearing scales. (3) = Scales or scalelike structures absent or, if present, limited entirely to association with the lateral line. 5. Lateral Line Scales — Distinctive lateral line scales that are highly modified relative to adjacent body scales are found in representatives of many iniomous families. Typically, such scales are larger than the adjacent body scales and each (except the posteriormost) consists of a bony plate, pierced by a central pore, the latter covered or partly so by a bony shelf or tympanum (see Johnson, 1974c, p. 15). Among scopelarchids, the shape and size of these components differs be- tween species such that in combination the lateral-line-scale conformation is distinctive for each known scopelarchid species (Johnson, 1974c, p. 21). Marshall & Staiger (1975) conducted a limited survey of lateral line scale and associated laterosensory organ morphology among iniomous families. They conclude that an "aulopoid" lineage, comprised of the families Aulopidae, Chlorophthal- midae, Neoscopelidae, and Ipnopidae (including Bathypteroidae), can be distin- guished by the presence of free-ending laterosensory organs on the lateral line scales. These organs are innervated by bilateral nerves emerging from the main lateral line nerve (see Marshall & Staiger, 1975, p. 109). Such a system is proba- bly also present in the Scopelarchidae and Paralepididae and may be present, in 70 FIELDIANA: ZOOLOGY highly modified form, in the Evermannellidae. It is certainly lacking in the Alepisauridae and Omosudidae (which lack lateral line scales), is probably lack- ing in the Anotopteridae (see Rofen, 1966c, p. 506), and, according to Marshall & Staiger (1975, p. 109), free-ending organs on the trunk and tail are absent in Bathysaurus, synodontids, harpadontids, and myctophids. Based on the taxonomic distribution of such organs, I suspect that the presence of these free- ending, lateral-line-scale-associated organs is primitive for iniomous fishes. This is the reverse of the character-state sequence implied by the discussion of Mar- shall & Staiger. I have not, however, conducted an adequate survey for the presence or absence of such organs among iniomous fishes and choose not to distinguish character states relating to their presence at this time. The total lack of scales or ossified, scalelike structures associated with the lateral line (in all included species) is apparently unique to three iniomous families (Alepisauridae, Evermannellidae, Omosudidae). State (4) represents a possible morphological intermediate between state (0) and state (5), and thus three states of character 5 are tentatively recognized among iniomous fishes. (A) States Recognized for Iniomous Fishes (0) = Scales or ossified scalelike structures associated with the lateral line. (4) = Scales absent; a series of membranous, nonossified, shieldlike structures (see Rofen, 1966d, p. 523) segmentally arranged along the lateral line. (5) = No scales or scalelike structures are associated with the lateral line (which is present, at least in young stages; Rofen, 1966b, p. 467; Gibbs & Wilimovsky, 1966, p. 483). Hypothesized character-state sequence: — > 4 — » 5 Evermannellids lack normal scales, but most evermannellid species have a series of membranous, nonossified shieldlike structures segmentally arranged along the lateral line. The three genera of evermannellids differ in the maximum observed posterior extent of lateral line segments and in the maximum observed number of lateral line segments (both features vary ontogenetically). State (E3) represents a possible morphological intermediate between state (0) and (E4). (B) States Recognized for Evermannellid Species (0) = Lateral line present, extending (maximally) to a vertical through middle of anal-fin base and composed of up to 43 segments. (E3) = Lateral line present, extending (maximally) to a vertical through anterior one-third of anal-fin base and composed of 34 or fewer segments (34 or fewer in Coccorella atlantica; 30 or fewer in C. atrata). (E4) = Lateral line present or (possibly) absent (Evermannella ahlstromi), ex- tending (maximally) to a vertical through a point just posterior to pelvic-fin base and containing no more than 18 segments. Hypothesized character-state sequence: — > E3 — > E4 6. Mode of Reproduction. — Gonochorism is presumably the primitive mode of reproduction among iniomous fishes and is found in aulopids (Mead, 1966a), synodontids and harpadontids (Sulak, 1977), neoscopelids (Nafpaktitis, 1977), and myctophids (Nafpaktitis et al., 1977). Synchronous hermaphroditism is uni- versal among the abyssal-benthic families Bathysauridae (Gosline et al., 1966), Ipnopidae (Mead et al., 1964; Mead, 1966b; Nielsen, 1966; Sulak, 1977) and JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 71 among the Chlorophthalmidae (Mead, l%6e), Scopelarchidae, and the alepisauroids (Johnson, 1974c). If Sulak's (1977) interpretation of the relation- ships of Bathysaunis is correct, the monoecious state characterizing this genus may have been acquired independently. Two states of character 6 are recognized among iniomous fishes. All evermannellids are synchronous hermaphrodites (Herring, 1977). (A) States Recognized for Iniomous Fishes (0) = Species are dioecious. (6) = Species are monoecious. 7. Pyloric Caecum. — Pyloric caeca are present in chlorophthalmids (includ- ing Bathysauropsis) , ipnopids, and notosudids (Nielsen, 1966; Bertelsen et al., 1976; Sulak, 1977). Most other iniomous families have been inadequately sur- veyed for this character. As far as is known (Wassersug & Johnson, 1976), pyloric caeca are absent in scopelarchids and in all alepisauroids except species in the genus Coccorella. Thus, although the presence of pyloric caeca may (or may not, see Qureshi, 1966, for comments on the "plasticity" of pyloric caeca among teleosts) be primitive for iniomous fishes, the presence of a pyloric caecum is here presumed to be derivative for evermannellids. This conclusion is reinforced by the fact that the pyloric caecum in Coccorella (Wassersug & Johnson, 1976; Herring, 1977) exhibits a conformation that may be unique among teleosts. (B) States Recognized for Evermannellid Species (0) = Pyloric caeca absent. (E5) = A single pyloric caecum present as a narrow elongate structure ex- tending into the head and easily visible in the floor of the orobran- chial cavity as a distinct tubelike structure beneath the basibranchial series. 8. Peritoneal Pigment Sections. — I believe it likely that the development of discrete peritoneal pigment sections (see section on Larval Characters, fig. 4; Johnson, 1974c, pp. 23, 24; Okiyama, 1974, pp. 618-620) in larval inioms will prove of great importance to the satisfactory definition of myctophiform or aulopiform (sensu Rosen, 1973) fishes. The pigment sections are striking features of the larval morphology of most aulopiform(R) fishes, and this is certainly true for both evermannellids and scopelarchids. The function of these pigment sec- tions is unknown — Okiyama (1974, p. 619) suggests a possible correlation with the loss of the swimbladder. A single dorsomedial section is found in larval aulopids (Hime only, see Okiyama, 1974, p. 610), chlorophthalmids (Taning, 1918; Ahlstrom, 1971), ip- nopids (Bathytyphlops, see Okiyama, 1972), and (primitively) scopelarchids (Johnson, 1974c, pp. 206, 207). Multiple (three or more, serially arranged, paired or unpaired) peritoneal pigment sections occur in aulopids (Aulopus, see Okiyama, 1974, p. 611), bathysaurids (Johnson, 1974b), synodonrids and har- padontids (Gibbs, 1959; Anderson et al., 1966; Okiyama, 1974), paralepidids (Rofen, 1966a), Omosudis (Rofen, 1966b), and evermannellids (state unknown in Anotopterus) . Larvae of certain scopelarchids (Scopelarchoides danae, S. nicholsi, Scopelarchus spp.) exhibit three peritoneal pigment sections, but it seems certain that the unique unpaired and paired conformation of these sections is autapomorphous for this scopelarchid lineage only (see Johnson, 1974c, pp. 23, 72 FIELDIANA: ZOOLOGY 206). The peritoneal pigment sections are paired in the harpadontids and synodontids (Anderson et al., 1966; Okiyama, 1974, pers. obs.) but single (con- nected dorsomedially over the gut, see fig. 4) in the others. If, as Rosen (1973, pp. 438-441) suggests, giganturids are allied with synodontoids, they (gigan- turids) are unique among synodontoids in apparently lacking any peritoneal pigment sections (Tucker, 1954). Discrete peritoneal pigment sections are lacking in notosudids (Bertelsen et al., 1976), at least some ipnopids (Okiyama, 1974, p. 614), and in all known neoscopelid and myctophid larvae (Moser & Ahlstrom, 1970, 1972; Okiyama, 1974; Yefremenko, 1977). Whether the ". . . elliptical shield of melanophores [lying] above the developing gas bladder ..." in such forms as Scopelopsis multipunctatus (quotation from Moser & Ahlstrom, 1972, p. 547) is homologous to aulopiform(R) peritoneal pigment sections is unknown, but here the presumption is that it is not. According to Rofen (1966b), peritoneal pigment sections in larvae of Alepisaurus are lacking or indistinct. I examined larvae of Alepisaurus taken off Hawaii and found no discrete peritoneal pigment sections, although I found ample pigmentation completely enclosing the gut in larger larvae and junveniles (as in adults, see Gibbs 1960). Based on the above information and following Okiyama (1974), I regard the presence of a single, dorsomedial peritoneal pigment section as primitive for iniomous fishes. (A) States Recognized for Iniomous Fishes (7) = Peritoneal pigment sections absent. (0) = A single, dorsomedial peritoneal pigment section. (8) = Multiple (3 or more), serially arranged, unpaired peritoneal pigment sections. (9) = Multiple (3 or more), serially arranged, paired peritoneal pigment sec- tions. I (Alepisaurus) i T Hypothesized character-state sequence: 7 «— — » 8 — » 9 Three peritoneal pigment sections are found in larvae of Coccorella and Ever- mannella (figs. 4-7), Omosudis (Rofen, 1966b, p. 472), and the paralepidine bar- racudina Paralepis atlantica, characterized by Rofen (1966a, p. 238) as ". . . the most primitive species in the Paralepididae." Larvae of Odontostomops normalops exhibit 12 or more unpaired peritoneal pigment sections — a state regarded herein as autapomorphous. (B) States Recognized for Evermannellid Species (0) = Larvae with three unpaired peritoneal pigment sections. (E6) = Larvae with 12 or more unpaired peritoneal pigment sections. 9. Accessory Pigment Spots or Areas. — Johnson (1974c, p. 207) notes the presence of deep-lying pigment spots or areas in larval chlorophthalmids, paralepidids, and scopelarchids. Such accessory pigment spots or areas occur widely in larval inioms (aulopids, alepisaurids, myctophids, ?neoscopelids, notosudids, and synodontoids) and are of great diagnostic value at the species level for some groups (e.g., Gibbs, 1959; Moser & Ahlstrom, 1972; Anderson et al., 1966; Johnson, 1974c). Discrete accessory pigment spots or areas, such as JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 73 described by Johnson (1974c) for some scopelarchid species are lacking in ever- mannellids, Omosudis, and at least some ipnopids (Bathytyphlops, see Okiyama, 1972). Evermannellids have a distinctive larval-phase pigmentation described above (see Larval Characters). The survey of iniomous larvae is by no means complete enough to justify recognition of character states at this time, but I believe that the presence and conformation of accessory pigment spots and areas will (when known) provide valuable systematic information not only at the species but also at higher-category levels. 10. Juvenile-Phase Pigmentation. — The major pattern of body pigmentation in evermannellid larvae occurs in two phases — a larval phase (fig. 5) and a juvenile phase (figs. 6, 7) — with a gradual transition between the two phases (see Larval Characters). The juvenile-phase pigmentation of Evermannella larvae is characterized by the development of three rows of very large melanophores, each row associated with one of the three main divisions of the tail musculature. I regard the fixation of this pattern in Evermannella as indicative of relationship between the four species comprising this genus. The juvenile-phase pigmenta- tion in Odontostomops (figs. 6, 7) is characterized by the development of numer- ous fine melanophores generally distributed over the head and body. The juvenile-phase pigmentation in Coccorella (figs, 6, 7) is intermediate in state, with larger, more prominent melanophores than seen in Odontostomops, but the developing melanophores are smaller, much more numerous, and not arranged in rows as in Evermannella. For the present I recognize two states of character 10 among evermannellids. (B) States Recognized for Evermannellid Species (E7) = Juvenile-phase pigmentation characterized by the development of three distinct rows of very large melanophores, each row associated with one of the three main divisions of the tail musculature (figs. 6-8). (0) = Juvenile-phase pigmentation not as described above. 11. Stomach Pigmentation in Juveniles. — According to Wassersug & Johnson (1976, p. 280), juveniles of Alepisaurus and Omosudis differ from scopelarchids and other alepisauroids in two aspects of gut morphology: (1) the stomach is thin-walled and balloon-like, not the heavily muscularized sac char- acteristic of scopelarchids and other alepisauroids, (2) the stomach wall is densely pigmented, and this pigment appears before or coincident with but independent from the development of peritoneal pigment sections (such sec- tions are present in Omosudis, apparently absent in Alepisaurus). Wassersug & Johnson examined only juvenile (less than 100 mm SL) specimens of Alepisaurus and Omosudis. Gibbs (1960, p. 4) describes the stomach in adult Alepisaurus brevirostris as follows, "Stomach black, highly distensible, forming a long blind sac." According to Gibbs' description, the pigmentation of the stomach is dis- tinct from the well-developed peritoneal pigmentation. Wassersug & Johnson (1976) conducted a limited survey of other iniomous fishes (Chlorophthalmus, Parasudis, Harpadon) and report a stomach/pigment conformation similar to that seen in scopelarchids and alepisauroids other than Alepisaurus and Omosudis (i.e., thick-walled, heavily-muscularized, unpigmented). State (10), described below, is apparently unique to Alepisaurus and Omosudis, although my survey of iniomous taxa for this character is far from thorough. 74 FIELDIANA: ZOOLOGY (A) States Recognized for Iniomous Fishes (10) = Stomach large, a highly distensible blind sac, thin-walled, not heav- ily muscularized; stomach wall densely pigmented, the pigmenta- tion appearing (in ontogeny) independent of any peritoneal pig- mentation. (0) = Stomach/pigment conformation not as described above. MERISTIC CHARACTERS: VARIATION AMONG INIOMOUS FISHES Values for meristic characters differ strikingly between various iniomous families (table 9) and are of great value in writing diagnoses for individual families. For the most part, however, intrafamilial variation results in extensive overlap between families, making difficult or impossible the meaningful defini- tion of separate character states and the determination of character-state se- quence. In certain cases, where both state and probable sequence can be deter- mined (e.g., the unique lack of a dorsal fin in Anotopterus, the uniquely high dorsal-fin ray count in alepisaurids), the derived states are evident autapomor- phies, of no readily discernible value to the attempt to infer relationship among iniomous families. The same statement can probably be made in the case of the very low dorsal-fin ray count exhibited by scopelarchids (table 9), although this distinction is masked to some extent by overlap. 12. Number and Distribution of Branchiostegal Rays. — McAllister (1968, pp. 89, 90) recognizes three character states among iniomous fishes with respect to number and distribution of branchiostegal rays. My redefinition of these states includes data presented in Table 9. (A) States Recognized for Iniomous Fishes (0) = Branchiostegal rays numerous (12 or more), lacking the 4 + X pat- tern (McAllister, 1968), with 6 to 9 branchiostegal rays on the epihyal (see Rosen & Patterson, 1969, p. 452). (11) = Branchiostegal rays fewer (7 to 14, usually 8 to 10), with the 4 + X pattern, with 3 to 5 (usually 4) branchiostegal rays on the epihyal (see Rosen & Patterson, 1969, p. 452). (12) = Branchiostegal rays fewer (6 to 12, usually 7 to 11), with the 4 -I- X pattern, but with only two branchiostegal rays on the epihyal (see Paxton, 1972, p. 25). Hypothesized character-state sequence: — » 11 — » 12 Sulak (1977, pp. 53, 54) states that a high number of branchiostegal rays is primitive for iniomous fishes. Paxton (1972, p. 56) notes that the presence of only two branchiostegal rays on the epihyal is an advanced neoteleost characteristic distinguishing the myctophids and neoscopelids from all other inioms. AH evermannellids have eight branchiostegal rays. The anteriormost epihyal branchiostegal inserts at the ceratohyal-epihyal joint in Coccorella and Odonto- stomops and posterior to this joint in Evermannella (fig. 13). 13. Number of Vertebrae. — Vertebral number is quite variable among in- iomous fishes (reported range 28 to 121, table 9). s In the definition of character s The reported extreme is ca. 186 vertebrae in Polymerichthys nagurai Uyeno, 1967 (Polymerichthyidae), a species known from one specimen from the Middle Miocene Tubozawa Formation of Japan. Uyeno (1967) allied Polymerichthys with Anotopterus princi- o E § I I oc c o £ (8 12 t; a. E o U co i/"i iri r^ m on in **■ no vp -" T T (jvo6cCTNvoooinpr>oooovin O ^ r-< i-H rH J- H tj< r^ oo cm OnOnOnOnOnOnOnOnOnOnOnOnOnOnOnOnOnOnOn u Tj«a\(NmoooosOTti— iTtcot^Kmininvomro O nriNnNririN(v|rHrHririnpinHr^riOO*HTfn'f* ^ i— i t-h t-i i— 1 r-i iHriNHNrHHnN i— I i— i ri i— i i— i i— i i— • i*> i— i t— i III III a 3 O u (Nrorjiinvotv.ooavOr-i r ? r r vG? m O — ,»NN TOi— I 8 |5t32£;o~ b <8 U J2 _ « 73 is ns o; « O O 2 ^ o ^2 on A3 . ^ . ^ c_ r. g 2 «} O r- OS r-^K e U Ot C ^t-h w e >^ o._r _ .. i ocS i ~ «5T3 ^ q — m £ S. ^ — o :g ft-TC § c ^ '"' 3 SN5k x O ■> v ^3 .E ■? « x f § « £ £ coo I s =o 13 ^.g T-S -'rx ft, (« o t> J= UN *, ON C CD • - in t- -5: On (8 ->- r- 1 . « ii D." 3 ^ JS On TJ O ,_^ (8 3 i- 1 C >!» "E^ - 3 _^ CD •_-. C (/) ^$£ft- *$£2ft^ O* On n£> _ X> t-i tv. On E ? S rH c " J -a ft- "5 ^ S §-§ ^2 So 3 S ir, 73 X (0 <8 * *S 01 ^ V O ft" 1/5 _. QJ ■ — ,3 u) «S O ^^ i< c at «i (8 QJ -c "q3 E § ^E E ^ o > o ^s 3 U + ft, S^JS " ?£ ■£•§•*" — On 5 O (8 > 01 <8 1-1 0> c -£ 5 (8 *. S =*> 3 _o 73 at v'i'Uji/) T3 "3 -J= O (8 -^ ^ n 0; 3 J< OS 73 .- as 1 c 1- 73 (8 01 £ s o ft- • fi II Itl _ u -° O Qj O '-' 60 • -*, c "> o « S " 6? & 6Ct-i Si DQ CD .XCD UU i^i-i 75 76 FIELDIANA: ZOOLOGY states below I have followed techniques devised by Marx & Rabb (1972, pp. 54-62) for the analysis of continuous meristic characters. Due to enormous varia- tion in number of species per taxon for the groups as listed in Table 9, 1 chose to apply the techniques of Marx & Rabb first to Paxton's (1972, p. 33) data for myctophid genera (reported range: 28-45). "Spans" of seven vertebrae or less characterized 90% of the myctophid genera, and 22 of 31 myctophid genera fall into the interval 35 to 41 vertebrae defined through a lower limit of 28 and a span of seven. Using this basis (span = 7, lower limit = 28), all iniomous taxa were assigned to one of four character states. (A) States Recognized for Iniomous Fishes Number of vertebrae: (14) = 28 to 34 (13) = 35 to 41 (0) = 42 to 62 (15) = 63 to 121 Hypothesized character-state sequence: 14 «— 13 <— -* 15 Most iniomous families (all but two: Anotopteridae, Neoscopelidae) have some or all members that exhibit state (0) (see below and table 9), and on this basis I believe 42 to 62 vertebrae to be the primitive state. Due to large intrafamil- ial variation in some groups, e.g., Notosudidae (reported range: 43-67), Scopelarchidae (reported range: 40-65), Paralepididae (reported range: 53-121), there exists little basis for subdivision of states (0) or (15) at this time. The following notes explain the assignment of states to certain taxa in Table 8. The character state (0) assigned to the Chlorophthalmidae is based on the total reported range of variation (38 to 58) for this family (including Bathysaurop- sis, Chlorophthalmus, and Parasudis). The high number of vertebrae (58) in Bathysauropsis parallels an apparent trend toward a higher number of vertebrae in ipnopids (49 to 80). The character state (13) assigned to the Myctophidae is based on the fact that 26 of 31 myctophid genera exhibit this state (22 genera) or a state intermediate between states (13) and (14) (four genera). None of the four genera (Diogenichthys, Lepidophanes, Notolychnus, Triphotums) exhibiting state (14) or the single genus (Gymnoscopelus + Nasolychnus — regarded by Paxton, 1972, as only subgenerically distinct) exhibiting state (0) is noted by Paxton (1972) as basal or primitive within the family. Because only one genus, Gymno- scopelus, exhibits state (0), this is regarded as a reversal. Within the Ipnopidae, Bathypterois (reported range = 49-61) and Ipnops (reported range = 55-61) ex- hibit state (0), the genera Bathymicrops (reported range = 65-80, 65 to 69 in B. regis, 76 to 80 in B. brevianalis) and Bathytyphlops (62-67) exhibit state (15). It may well be that state (15) should be assigned to the Paralepididae — the low number of vertebrae in Sudis (reported range: 53-60) plus knowledge of fossil paralepidids with as few as 45 vertebrae (Rofen, 1966a, p. 208) resulted in the assignment of state (0) to this family. pally on the basis of similarities in the shape of the body and head and in the morphology of the palatine teeth. Polymerichthys differs from Anotopterus in having a well-developed, enormously elongate dorsal fin (300-350 dorsal-fin rays) and a much restricted, "almost vestigial" caudal fin. The apparently extinct Polymerichthyidae is not treated elsewhere in this paper. JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 77 There is an undoubted trend toward a greater number of vertebrae among some benthonic (ipnopids, synodontoids) and pelagic (e.g., Anotopteridae, Paralepididae) lineages. Evolution in the Scopelarchidae has apparently featured trends toward a higher number of vertebrae in the line leading to Benthalbella and a lower number of vertebrae in the line leading to Scopelarchus (Johnson, 1974c). The neoscopelids and myctophids are largely distinct from other iniomous fishes in having fewer vertebrae. 14. Frontal/Dermethmoid Contact. — Evermannellids differ strikingly from scopelarchids in the configuration of the frontal/dermethmoid contact zone. In evermannellids the dermethmoid exhibits two posteriorly directed projections (fig. 9) overlying the respective (right and left) anterior sheetlike portions of the frontals. In scopelarchids the anterior end of the frontals overlies the der- methmoid (Johnson, 1974c, p. 35). The state in Omosudis is like that in everman- nellids. The state in Alepisaurus is unknown. Apparently other inioms are similar to scopelarchids with respect to this character (see Goody, 1969, p. 205; Bertelsen et al., 1976, p. 8), but the survey is too incomplete to justify formal recognition of character states and assignment of states to taxa at this time. 15. Parietals. — Evermannellids were said to be closely related to scopelar- chids because parietal bones were said to be lacking in the two families ("pari- etals fused with frontals," see Gosline et al., 1966, p. 17). Actually this state is true for neither group. The parietals are present and do meet in midline (or very nearly so) in all evermannellids (fig. 9). Parietals are present but widely sepa- rated in most scopelarchids (Johnson, 1974c, p. 31). The presence of parietal bones that are in contact in dorsal midline is primitive for iniomous fishes (Goody, 1969; Sulak, 1977). (A) States Recognized for Iniomous Fishes (0) = Parietals present and in contact (or nearly so) in dorsal midline. (16) = Parietals absent or when present are widely separated, not in con- tact in dorsal midline. 16. Attachment of Dermosphenotic. — In evermannellids the dermosphen- otic (IO-8, fig. 10) is attached to the lateral face of the autosphenotic. In scopelar- chids, as (apparently) in most iniomous fishes, the dermosphenotic overlies (is dorsal to) the autosphenotic. (A) States Recognized for Iniomous Fishes (0) = Dermosphenotic overlies (is dorsal to) the autosphenotic. (17) = Dermosphenotic is attached to lateral or anterolateral face of autosphenotic. 17. Basisphenoid. — A basisphenoid bone is present in Aulopus, the Creta- ceous Sardinioides, Chlorophthalmus, Saurida, Omosudis, and the presence of this bone is primitive for iniomous fishes (Goody, 1969). The bone has been lost, apparently independently, in a number of iniomous families (including all scopelarchids). Among evermannellids a narrow, splintlike basisphenoid is present in Coccorella and Evermannella but apparently absent in Odontostomops. (B) States Recognized for Evermannellid Species (0) = Basisphenoid present. (E8) = Basisphenoid absent. 78 FIELDIANA: ZOOLOGY 18. Orbitosphenoid and Ethmoid Cartilage — An orbitosphenoid bone is present in Aulopus and the Cretaceous Sardinioides and is apparently a primitive feature for iniomous fishes (Goody, 1969). I am unable to confirm the presence of an orbitosphenoid in any other extant iniom (neither scopelarchids nor ever- mannellids have it, but Regan [1911, p. 124] reports its presence in some synodontoids). Due to incompleteness of the survey of iniomous taxa for this character, I choose not to define formal states at this time. A presumably neomorphous feature characteristic of the evermannellid genus Coccorella is a considerable rearward expansion of the ethmoid cartilage des- cribed in the states recognized below. (B) States Recognized for Evermannellid Species (0) = Ethmoid cartilage not entering orbit, not forming an orbital septum. (E9) = Ethmoid cartilage considerably expanded posteriorly, forming an orbital septum between the eyes, extending to or nearly to midline in posterior wall of orbit. 19. Sclerotic Bones. — All scopelarchids possess two sclerotic bones (anterior and posterior), which are embedded in and which strengthen the tubular eye. Sclerotic bones are also present in aulopids (Hime), Chlorophthalmus, Parasudis, notosudids, and at least certain synodontoids (e.g., Saurida). Sclerotic bones are absent in evermannellids, and, as far as is known, in all other iniomous fishes. The presence of such elements in all three benthonic "lineages" (Sulak, 1977) suggests the presence of sclerotics may be primitive for iniomous fishes. Harder (1975, p. 345) notes the widespread (if sporadic) occurrence of scleral ossifica- tions and suggests that presence or absence is a function of eye size. It may be worthwhile to note that the eyes of Evermannella spp. are (relative to body size) no smaller than those of scopelarchids, yet Evermannella spp. lack scleral ossifica- tions. (A) States Recognized for Iniomous Fishes (0) = Sclerotic bones present. (18) = Sclerotic bones absent. 20. Subocular Shelf. — According to Paxton (1972, p. 10), a subocular shelf extends medially from IO-3, IO-4, and usually IO-5 in all myctophids. The presence of a subocular shelf is an advanced neoteleost (acanthopterygian not paracanthopterygian) character (Smith & Bailey, 1962; Rosen & Patterson, 1969; Rosen, 1973; Zehren, 1975). A subocular shelf is not found in any other iniomous fish (including neoscopelids). Among other teleosts, a subocular shelf is known only from the osteoglossomorph genus Notopteris (Rosen & Patterson, 1969, p. 379). (A) States Recognized for Iniomous Fishes (0) = Subocular shelf absent. (19) = Subocular shelf present. 21. Antorbitals — Reviews of the morphology of infraorbital bones in teleosts are provided by Smith & Bailey (1962), Gosline (1965), and Nelson (1969b). Nelson (1969b, pp. 2, 3) regards the presence of an antorbital followed by six infraorbitals as the primitive pattern for most or all major teleost groups, in- cluding iniomous fishes. Antorbitals are present in representatives of most in- iomous families, including the basal scopelarchid genera Scopelarchoides and JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 79 Rosenblattichthys, as well as Benthalbella macropinna (Johnson, 1974c), but antorbi- tals apparently are lacking in all alepisauroid families. Paxton (1972, p. 9) de- scribes an antorbital/nasal configuration unique to myctophids among iniomous fishes. (A) States Recognized for Iniomous Fishes (0) — Antorbitals present. (20) = Antorbitals absent. 22. Supraorbitals. — In evermannellids the elements herein termed supraor- bitals are elongate, strutlike bones (fig. 10), slightly expanded ventrally and noticeably expanded dorsally. These elements are termed supraorbitals (rather than antorbitals) because of their close abutment on and attachment to the dor- solateral ridge of the frontal (on each side). Supraorbitals have been lost, appar- ently in most cases independently, in many iniomous lineages, including the evermannellid genus Odontostomops and all scopelarchid genera except Rosenblattichthys (Johnson, 1974c, p. 36). (A) States Recognized for Iniomous Fishes (0) = Supraorbitals present. (21) = Supraorbitals absent. 23. Infraorbital Series. — Most iniomous fishes, including aulopids, chlorophthalmids, myctophids, neoscopelids, and scopelarchids, have six infraorbitals — the primitive state for most or all major teleost groups (Nelson, 1969b). Notosudids apparently have seven infraorbitals (Bertelsen et al., 1976). Evermannellids, Omosudis, and paralepidids have eight infraorbitals. (A) States Recognized for Iniomous Fishes Number of infraorbitals: (0) = 6 (22) = 7 (23) = 8 Hypothesized character-state sequence: (22) <— (0) — * (23) 24. Premaxillary Fenestra. — Rosen (1973, p. 450) notes the presence of a peculiar fenestrated premaxilla in the Cretaceous fossil genera Enchodus, Palaeolychus, and Eurypholis (see Goody, 1969, p. 104). Rosen states that a premaxillary fenestra is present in and characteristic of the extant alepisauroid families Paralepididae (see Rofen, 1966a, p. 232), Omosudidae, and Alepisauridae. The fenestra is unquestionably present in paralepidids 6 and in Anotopterus (Goody, 1969, p. 171). According to Goody (1969, p. 172), the fenestra is not present in Alepisaurus, and I am unable to confirm its presence in Omosudis. It is absent in evermannellids and other iniomous fishes. (A) States Recognized for Iniomous Fishes (0) = Premaxillary unfenestrated. (24) = Premaxillary fenestrated. 25. Modification of Maxilla. — According to Goody (1969), the following fea- tures of the upper jaw, as seen in Aulopus and the Cretaceous Sardinioides, are primitive for myctophiform fishes: (1) presence of two supramaxillae (in extant 6 According to Harry (1953, p. 225), Sudis is unique among paralepidids in lacking a fenestrated premaxilla. 80 FIELDIANA: ZOOLOGY forms true only of aulopids and Saurida, in the latter the two supramaxillae are extremely reduced, see Sulak, 1977, p. 55), (2) maxilla long and narrow except posteriorly where it is dilated, (3) premaxilla with ascending and articular pro- cesses (see Rosen & Patterson, 1969, p. 457) and a very long alveolar arm. Sulak (1977) recognizes three "natural" groupings of benthonic inioms — the Aulopidae, a synodontoid lineage, and a chlorophthalmoid lineage. The chlorophthalmoid lineage (Chlorophthalmidae + Ipnopidae) is said (Sulak, 1977, p. 54) to have retained a more primitive configuration of the upper jaw, with a prominent maxilla that is dilated (deepened) and free posteriorly and a single elongate supramaxilla. According to Sulak, the synodontoid lineage (Saurida + Harpodon, Bathysaurus, Synodus + Trachinocephalus) is characterized by an upper jaw dominated by a strong premaxilla, with the maxilla reduced and variously modified (usually partly or wholly adherent to the premaxilla). These modifications were recognized by Sulak (1977) and in the present paper as three separate and presumably autapomorphic character states respectively defining the three main synodontoid lineages. (A) States Recognized for Iniomous Fishes (25) = Maxilla separated into anterior and posterior portions; the posterior portion reduced to a thin, adherent lamina along the posterior half of the premaxilla; the anterior portion consisting of the isolated head of the maxilla, which lies between the palatine and premaxilla and retains its original articulating functions (Saurida, Harpadon — see Sulak, 1977, p. 55). (26) = Maxilla undivided, present only as a short (less than 10% of premaxillary length in length), slender rudiment lying between the palatine and premaxillary heads (Bathysaurus — see Sulak, 1977, p. 58). (27) = Maxilla undivided, about equal to premaxilla in length but reduced to a simple lamina that is closely adherent to the premaxilla and lacks a free articulating head anteriorly or a free and dilated expan- sion posteriorly (Synodus + Trachinocephalus — see Sulak, 1977, p. 58). (0) = Maxilla not as described above. (25) «- (0) -> (26) Hypothesized character-state sequence: * (27) It should be noted that state (0) includes additional states to be recognized in some future study of iniomous relationships, these states ranging from the basal (for inioms) configuration of upper jaw bones seen in aulopids, chlorophthal- mids, neoscopelids, and scopelarchids (Rosen & Patterson, 1969; Johnson, 1974c) to highly derived states such as seen in Anotopterus (see Goody, 1969, p. 171) and Omosudis, in which the anterior maxilla is present as an extremely narrow, threadlike structure, lacking articulating processes. Incompleteness of information precludes subdivision of state (0) at this time. 26. Dentary Fossa. — The genus Evermannella is uniquely characterized among evermannellids, and, apparently, among iniomous fishes, by the pres- ence of a vertically elongate fossa at the dentary symphysis. This fossa (fig. 11) JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 81 contains the anteriormost mandibular cephalic laterosensory pore of each side and also contains two to four vertically oriented rows of laterosensory papillae. The exact function — presumably sensory — of this structure is unknown. In Odontostomops and Coccorella an externally visible ridge marks the line of the dentary symphysis. (B) States Recognized for Evermannellid Species Dentary fossa (as described above): (0) = Absent. (E10) = Present. 27. Jaw and Palatine Teeth. — A character supposedly separating alepisauroids(G) from myctophoids(G) are uniserial premaxillary teeth in alepisauroids (true in all cases) vs. arranged in more than one row in mycto- phoids. The problem — as pointed out by Gosline et al. (1966, p. 8) — is separa- tion of similarity due to relationship from similarity due to convergence. In some families in which premaxillary teeth usually occur in bands, there exist repre- sentatives with uniserial premaxillary teeth, e.g., Parasudis among the chlorophthalmids, the genera Centrobranchus, Diogenichthys, and Gonichthys among the myctophids (Gosline et al., 1966; Paxton, 1972). In Odontostomops and Evermannella, as in all scopelarchids, the dentary teeth are arranged in two series, with an outer series of smaller teeth and an inner series of large, barbed fangs (Johnson, 1974c, p. 39). The arrangement of dentary teeth in two or more series occurs widely among iniomous fishes and is therefore taken as primitive. In Odontostomops and Evermannella, as in all scopelarchids, as well as many other inioms, at least the largest dentary and palatine teeth are barbed (fig. 11). Barbed teeth are present in some or all representatives of the following: Aulopidae, Myctophidae, Notosudidae, Paralepididae, synodon- toids. Barbed teeth are absent in Anotopterus, Alepisaurus, and Omosudis. In Coccorella neither dentary nor palatine teeth are barbed, the saber-like palatine fangs are greatly enlarged, and the dentary teeth are uniserial. I believe that in combination this suite of characters defines a derived state unique to Coccorella among evermannellids — a state possibly related to differences in prey and prey-capture style between Coccorella and other evermannellids (see Size and Habits). (B) States Recognized for Evermannellid Species (0) = Dentary teeth arranged in two series; at least some dentary and palatine fangs barbed; longest palatine tooth = 4.6% to 7.3% SL. (Ell) = Dentary teeth uniserial; dentary and palatine fangs not barbed but saber-like; longest palatine tooth = 7.1% to 10.0% SL. 28. Basihyal. — Scopelarchids and evermannellids differ strikingly in the development and configuration of the basihyal, basihyal toothplate, and basihyal dentition. In scopelarchids the basihyal is well ossified, the basihyal toothplate is (apparently) indistinguishably fused with the basihyal, and large, hooked basihyal teeth are present in all species (Johnson, 1974c, p. 48). In ever- mannellids the basihyal is much reduced (fig. 14), consisting of a rodlike structure that is only half-ossified (cartilaginous anteriorly) and is connected via a hingelike joint to the anterior margin of the first basibranchial. A distinct but edentate basihyal toothplate covers the basihyal except in Coccorella (see below). 82 FIELDIANA: ZOOLOGY A toothed basihyal is presumably primitive for iniomous fishes (Nelson, 1969a; Zehren, 1975) and occurs in aulopids, paralepidids, and all scopelarchids. More typical (apparently) for extant inioms is the presence of a basihyal (that is often cartilaginous or at least partly so) and an edentate basihyal toothplate (true for chlorophthalmids, myctophids, neoscopelids, notosudids, evermannellids). Both basihyal and basihyal toothplate are apparently lacking in Anotopterus and Omosudis (state unknown for Alepisaurus) . I choose not to define formal states of character 28 for iniomous fishes at this time for two reasons: (1) the survey of iniomous taxa is insufficient for this character, and (2) the problem of how to treat groups such as chlorophthalmids (Chlorophthalmus, Parasudis) in which partly hooked and strongly developed basihyal teeth are present in larvae and juveniles (Rosen, 1971; pers. obs.) but are lost in adults. The species of Coccorella differ from other evermannellids in the reduction (C. atlantica) or loss (C. atrata) of the basihyal toothplate as described in character states defined below. State (E12) is regarded as logically intermediate between states (0) and (E13). (B) States Recognized for Evermannellid Species (0) = Basihyal toothplate covers dorsal and dorsolateral margins of basihyal. (E12) = Basihyal toothplate reduced, covering only posterior two-thirds of dorsal margin of basihyal. (E13) = Basihyal toothplate absent. Hypothesized character-state sequence: (0) — » (E12) — > (E13) 29. Gill Rakers. — Normal lath- or bladelike gill rakers occur in aulopids, chlorophthalmids, ipnopids, neoscopelids, notosudids, and most myctophids (all but Centrobranchus, Paxton, 1972, p. 24). Said by Sulak (1977, p. 53) to partly characterize the synodontoid lineage is modification of gill rakers into clusters of short gill teeth — a state that also characterizes the Paralepididae (Rofen, 1966a, p. 218) and Alepisauridae (Gibbs & Wilimovsky, 1966, p. 485: "Gill rakers with tufts of depressible filaments."). Anotopterus lacks both gill rakers and gill teeth (except for conical teeth limited to the fifth pharyngobranchial toothplate). In Omosudis the gill rakers are reduced to fixed individual short gill teeth (Rofen, 1966b, p. 463). In scopelarchids gill tooth plates replace gill rakers — the plates expanding with growth to form flattened plates of bone bearing one to many small teeth arranged in one to three rows on the dorsal margin of the plate. Similar gill tooth plates in evermannellids support one to several small teeth arranged uniserially, with gill teeth limited to the second gill arch. The possibil- ity of convergent loss of lathlike gill rakers is discussed by Gosline et al. (1966, p. 12). (A) States Recognized for Iniomous Fishes (0) = "Normal" lath- or bladelike gill rakers present. (28) = "Normal" lath- or bladelike gill rakers absent. 30. Distribution of Gill Teeth. — Evermannellids are apparently unique among inioms in the restriction of gill teeth (for the distinction between "gill teeth" vs. "conical teeth" see section on Branchial Arches above) to the cerato- branchial of the second gill arch. Other iniomous groups exhibiting a restricted distribution of gill teeth include Anotopterus (which lacks gill teeth) and Omosudis (in which gill teeth are restricted to the first and second arches). In at least some JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 83 representatives of all other iniomous families either gill rakers or gill teeth (or both) are found on all four gill arches (HB1 to 3, CB1 to 4 [in most, 1 to 5], EB1 to 4 [1 to 3 in scopelarchids and notosudids]). (A) States Recognized for Iniomous Fishes (0) = Gill rakers or gill teeth (or both) present on gill arches one to three (or on all four gill arches). (29) = Gill rakers absent; gill teeth (if present) lacking on third and fourth gill arches. UPPER BRANCHIAL TOOTHPLATES Rosen (1973, pp. 406, 435) distinguishes (in part) his "alepisauroid" ( = synodontoids + giganturids + alepisauroids) lineage from aulopoids(R) by list- ing the following characters as supposedly advanced features shared by alepisauroids(R): (A) Loss of an independent fifth upper pharyngeal toothplate (PB5TP) with concomitant great enlargement of the fourth upper pharyngeal toothplate (PB4TP; the fourth pharyngobranchial [PB4] is cartilaginous in all iniomous fishes). (B) Loss of the second upper pharyngeal toothplate (PB2TP). (C) Loss of the third epibranchial toothplate (EB3TP). (D) Frequent loss or reduction of the third upper pharyngeal toothplate (PB3TP). (E) Confinement of the muscle retractor arcuum branchialum (RAB) to the PB4, the PB4TP, the PB5TP (see discussion of character 31 below), and (in some cases) to the distal half of the fourth epibranchial (EB4). (F) Presence of relatively few fanglike teeth on the remaining (branchial) toothplates. (G) "Alepisauroids exhibit an exaggeration of the principal epibranchial and pharyngobranchial specializations of aulopoids. All of the elements are greatly attenuated, seemingly in relation to the development of a very long jaw with an oblique suspensorium" (Rosen, 1973, p. 435). (H) "The teeth of alepisauroids, both on the pharyngobranchials and on the jaws and palate, are much larger and fewer than those of aulopoids. They are often notched distally or bear scalpel-like tips" (Rosen, 1973, pp. 435, 436). Feature (F) is actually true for virtually all inioms (one prominent exception being the scopelarchids in which some species have remarkably large teeth over the basibranchials). Feature (G) apparently relates to the marked prolongation of the uncinate process of the second epibranchial (EB2) discussed below (character 39). Feature (H) is only partly true, fewer and larger teeth are found in alepisauroids but not synodontoids (where, as Rosen, 1973, p. 436, points out, the state tends to be intermediate), and the dental configuration of alepisauroids is likely related to the common predaceous feeding mode of these fishes. Barbed teeth are neither unique to nor characteristic of all alepisauroids (see character 27). Features (A) through (E) are discussed in the following paragraphs. 31. Fourth and Fifth Upper Pharyngeal Toothplates. — Rosen (1973) consid- ers the loss of an independent PB5TP (with concomitant enlargement of the PB4TP) to be characteristic of alepisauroid(R) fishes. In fact this supposed ad- vanced feature is probably true of no iniomous fish. In scopelarchids, everman- nellids, and most other inioms the fifth toothplate (see Nelson, 1969a, pp. 488- 84 FIELDIANA: ZOOLOGY 490, for a discussion of the possible homologies of this element) is the most prominent (in terms of size of the toothplate, number of teeth, and size of teeth) and in some cases is the largest upper pharyngeal element (although in most inioms the PB3 is the largest upper pharyngeal element). Rosen (1973, p. 407) obviously did not see the small and easily missed PB4TP in his specimen of Evermannella sp. (a PB4TP is present in all evermannellids) and, as a result, identified the large PB5TP as the fourth. In evermannellids the PB4TP is quite small (fig. 15) and in one species (Coccorella atrata) is edentate (also true for Notolepis among the paralepidids). Although the PB4TP is somewhat larger (rela- tive to the PB5TP) in scopelarchids (Johnson, 1974c, p. 49), the PB5TP is the largest and most prominent upper pharyngeal toothplate. Only one upper pharyngeal toothplate (either PB4TP or PB5TP or possibly a compound element) is present in this position in neoscopelids, myctophids, some synodontoids (Synodus and Trachinocephalus), Anotopterus, and Alepisaurus. In all of these cases I suspect that the single element present is the PB5TP — this based (a) on the trend toward reduction of the PB4TP as exhibited by evermannellids and certain paralepidids (e.g., Paralepis, Notolepis), and (b) on the relative size of the PB4TP vs. PB5TP in iniomous families that have both. Rosen (1973, pp. 453, 454, 506) argues that reduction of the posterior upper pharyngeal dentition (relative to the PB3TP) is an advanced feature separating myctophids and neoscopelids from other inioms and allying the Myctophiformes with higher neoteleosts. Yet Pax- ton (1972, p. 26), while noting that the PB3TP is larger than the "PB4TP" in most myctophids, states "However in Protomyctophum, Hierops, aWElectrona (except E. rissoi), most Diaphus, and Notolychnus, the fourth [here interpreted as the fifth toothplate] is equal in size or larger than the third." Paxton (1972, p. 28) also notes that the ". . . fourth [here called fifth] . . . bears the largest and most posterior teeth in the oral cavity." I interpret these features as retention of the basal iniom state by the myctophid taxa listed and note that Protomyctophum is said (Paxton 1972, p. 63) ". . . to approach the most primitive adult condition in the tribe Myctophini and family." (A) States Recognized for Iniomous Fishes (0) = Both PB4TP and PB5TP are present. (30) = Only one upper pharyngeal toothplate present in this position (here assumed to be the PB5TP). Coccorella atrata is unique among evermannellids in that the PB4TP is edentate. (B) States Recognized for Evermannellid Species (0) = The PB4TP bears teeth. (E14) = PB4TP is edentate. 32. Second Upper Pharyngeal Toothplate. — A PB2TP occurs in aulopids, chlorophthalmids, myctophids, neoscopelids, notosudids, some synodontoids (Synodus, see Rosen, 1973, p. 404), and scopelarchids (Benthalbella elongata, B. infans, B. macropinna, Scopelarchoides signifer). No alepisauroid is known to have a PB2TP (several gill toothplates are present on the PB2 of the specimen of Notolepis examined by me). (A) States Recognized for Iniomous Fishes Conical teeth on PB2 are (0) = Present. (31) = Absent. JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 85 33. Third Upper Pharyngeal Toothplate. — An advanced feature supposedly characteristic of alepisauroids (R) is, according to Rosen 1973 (p. 435), the reduc- tion or loss of the PB3TP. With respect to loss, the actual distribution of this state is rather monotonous among inioms (true only for Anotopterus; and certain paralepids [Rosen, 1973, p. 408]; among paralepidid genera I have examined a PB3TP is present in Paralepis and Notolopis but not Macroparalepis) . (A) States Recognized for Iniomous Fishes Conical teeth on PB3 are (0) = Present. (32) - Absent. 34. Third Epibranchial Toothplate.— Rosen (1973, p. 435) states that an ad- vanced feature characteristic of alepisauroids is reduction or loss of the EB3TP. The EB3TP is absent in myctophids, some synodontoids {Synodus, see Rosen, 1973, p. 404), some paralepidids (Notolepis, Macroparalepis but not Paralepis), Alepisaurus (Rosen, 1973, p. 405), Anotopterus, and Omosudis. (A) States Recognized for Iniomous Fishes Conical teeth on EB3 are (0) = Present. (33) = Absent. 35. M. Retractor Arcuum Branchialum. — Rosen (1973, pp. 399, 400) argues that the presence of the muscle retractor arcuum branchialum (RAB) divides the Euteleostei into two major groups. Those with the RAB (Stomiatiformes + Aulopiformes + Myctophiformes + Paracanthopterygii + Acanthopterygii) are collectively termed neoteleosts. Winterbottom (1974, pp. 256-258) reviewed the literature on this muscle, termed by Winterbottom the m. retractor dorsalis, and noted that there exists some question concerning the uniqueness of this feature to neoteleosts, but he tentatively accepted Rosen's conclusion. According to Rosen (1973), the RAB is always paired, the anterior end inserts on dorsal gill arch elements, and, in most cases, the posterior end originates directly or via a tendon from the first to sixteenth vertebrae. The RAB's are believed to assist in swallowing. I have not surveyed iniomous fishes with re- spect to configuration of the RAB, and the following account follows Rosen (1973). (A) States Recognized for Iniomous Fishes (0) = RAB undivided, a flat sheet of muscle inserting on distal half of EB4, the cartilaginous PB4, and along the posterior ventromedial edge of the PB3 (Rosen, 1973, p. 400). (34) 7 = RAB undivided, confined to the PB4, associated toothplates (PB4TP if present, PB5TP), and, in some, distal half of EB4 (Rosen, 1973, p. 406). Tor most of the species to which Rosen assigns state (34), Rosen terms the element herein called PB5TP as the PB4TP. Rosen's illustrations suggest that state (34) occurs in Synodus (p. 404), Harpadon (p. 405), Alepisaurus (p. 405), Anotopterus (p. 406), Evermannella (p. 407), Lestrolepis (p. 408), and Paralepis (p. 408). State (34) also occurs in giganturids (p. 409) and stomiatoids (pp. 409, 410). Rosen's illustration for Scopelarchoides (p. 407) suggests that state (0) occurs in scopelarchids — a conclusion in agreement with other results presented in this paper (see below). The state for Omosudis (p. 406) appears to be somewhat intermediate between states (0) and (34), but closer to the latter. 86 FIELDIANA: ZOOLOGY (35) = RAB divided, with distinct medial (the smaller, inserting on PB3) and lateral (the larger, inserting on PB4 and associated toothplate, here assumed to be PB5TP) bundles (see Rosen, 1973, pp. 453-455). Hypothesized character-state sequence: (34) <— (0) — » (35) 36. Fifth Ceratobranchial Toothplate. — Conical teeth occur on the fifth ceratobranchial (fig. 14D) in all iniomous families except the Anotopteridae. Among evermannellids, however, conical teeth occur on the fifth ceratobran- chial only in Odontostomops normalops and Evermannella balbo. (B) States Recognized for Evermannellid Species Conical teeth on fifth ceratobranchial are (0) = Present. (E15) = Absent. 37. Basibranchial Dentition. — The presence of teeth over the ossified basi- branchials (BB 1 to 3, fig. 14) is widespread in primitive teleosts (Nelson, 1969a) and regarded as primitive for iniomous fishes. A number of iniomous families have lost basibranchial dentition (although not the toothplate), and this includes all alepisauroids. 8 Among scopelarchids, Rosenblattichthys, Scopelarchus, and two species of Scopelarchoides (S. danae, S. nicolsi) exhibit strongly developed, hooked teeth arranged uniserially above at least the first two basibranchials (Johnson, 1974c, p. 48). (A) States Recognized for Iniomous Fishes A tooth-bearing toothplate dorsal to the ossified basibranchials (BB 1 to 3) is (0) = Present. (36) = Absent. 38. First Pharyngobranchial. — The dorsal support of the first epibranchial (EB1) is through a first or suspensory pharyngobranchial (PB1) in most iniomous families (Johnson, 1974c, pp. 49, 201). In three families (Anotopteridae, Ever- mannellidae, Omosudidae) there is no PB1, and the EB1 attaches directly (through a ligament) to the anterior end of the third pharyngobranchial (PB3, fig. 14). In Alepisaurus (Rosen, 1973, p. 405) as in one scopelarchid lineage (that including the genus Scopelarchus, Johnson, 1974c, p. 51) there is no PB1, and support for the EB1 is provided by the second pharyngobranchial (PB2), which in turn attaches (ligamentously) to the anterior end of PB3. (A) States Recognized for Iniomous Fishes A suspensory pharyngobranchial (PB1) is (0) = Present. (37) = Absent. Myctophids are to my knowledge unique among iniomous fishes in posses- sion (= ? retention) of a PB1TP (Paxton, 1972, p. 25). 39. Uncinate Process of Second Epibranchial. — Rosen (1972, 1973) discusses the apparent diagnostic value of a markedly elongate uncinate process on the second epibranchial in the recognition of aulopiform(R) fishes. According to 8 In Evermannella balbo (fig. 14) and certain paralepidids (Paralepis, Notolepis, but not Macroparalepis) a fourth basibranchial toothplate (or toothplates) lies embedded in the skin over the cartilaginous element representing the third copula (Nelson, 1969a) of the basi- branchial series. JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 87 Rosen, all iniomous fishes (excepting the Myctophidae, Neoscopelidae, and Paralepididae) have such a "notably elongate" uncinate process, except (pre- sumably rare) cases where secondarily reduced, and this modification of the second epibranchial is unique to aulopiform(R) fishes. Apparently primitive for euteleosts (Rosen, 1973, pp. 401, 402) is the presence of two dorsal articulating heads or processes on the first three epibranchials (EB1 to 3). The anterior process articulates with its respective pharyngobranchial (PB1 to 3), and the posterior (uncinate) process with its respective succeeding pharyngobranchial (PB2 to 4). Whereas the EB1 typically retains both processes, according to Rosen (1973, pp. 402, 403), the uncinate processes of EB2 and EB3 are "... lost, greatly modified or reoriented ... in the more advanced euteleosteans." In inioms the marked elongation of the EB2 uncinate process is apparently associated with reorientation of the pharyngobranchials and enlargement of the PB3 (both ad- vanced characters helping to define Rosen's "Section Eurypterygii") and espe- cially with an elongation and lateral extension of the PB2 (Rosen, 1973, p. 402). Although the EB2 uncinate process is indeed markedly elongate in most in- ioms (fig. 14; see also Rosen, 1973, pp. 403-408), there is variation among inio- mous fishes in relative length of this process. The question remains — how does one quantitatively define "notably elongate"? Rosen (1973, p. 404) provides a possible test in his figure caption for Bathypterois atricolor, a species in which the ". . . uncinate process on [EB2 is] also apparently secondarily reduced judging from its failure to contact PB3." This suggests that states for character 39 can be defined as follows: (A) States Recognized for Iniomous Fishes (0) = EB2 uncinate process extremely elongate, extending to a point dor- somedial to dorsolateral border of the PB3 (e.g., Rosen, 1973, text- fig- 7) (38) = EB2 uncinate process not as elongate, not extending to a point dor- somedial to dorsolateral border of PB3 (e.g., Rosen, 1973, text-fig. 6). Application of this test results in assignment of state (0) to all iniomous families except the Myctophidae. I find, for my material, that certain paralepidids {Paralepis, Notolepis) exhibit state (0), in that the (cartilaginous) dor- sal terminus of EB2 extends to a point over the PB3. Other paralepidids (Mac- roparalepis) may exhibit state (38), the EB2 uncinate process not reaching a point dorsomedial to the dorsolateral border of the PB3. Both the specimen of Neo- scopelus microchir in my material and N. macrolepidotus as illustrated in Rosen (1973, p. 454) apparently exhibit state (0) — contrary to Rosen's comments (p. 455). My interpretation of the EB2 morphology in Neoscopelus is that the anterior articulating process has been reduced, and the uncinate process has retained the configuration typical for inioms. Oiagrams presented by Paxton (1972, p. 25) and Rosen (1973, p. 453) suggest that myctophids may have taken the opposite tack, so to speak, retaining the anterior process (which articulates with the PB2) but losing (or showing reduction of) the uncinate process. Note that Rosen's (1973) discussion is concerned with relationships among all neoteleosts, whereas my discussion is limited to inioms. Thus, the markedly elongate EB2 uncinate process is regarded by Rosen as an advanced feature of aulopiform fishes (relative to other euteleosts) and is regarded by me as a state 88 FIELDIANA: ZOOLOGY primitive for inioms (relative to iniomous taxa in which the process is secondar- ily reduced). Whether the state typifying myctophids represents retention of the basal euteleost configuration (as suggested by Rosen) or secondary reduction of the configuration typical for inioms is unknown — I argue that the latter hypothesis cannot be rejected on the basis of information presented by Rosen. 40. Attachment of First Centrum. — Gosline et al. (1966, p. 8) note that in neoscopelids (Neoscopelus, Scopelengys, Solivomer) as well as chlorophthalmids, there is a gap in ossification between the skull and the first vertebral centrum. This gap is said to attain its greatest extent in Solivomer — where it is nearly equal to the width of the two centra following the gap. The gap is filled by a rubbery membrane (= ? coat of notochord), and the whole structure may represent development of the intervertebral disk between the skull and first centrum. Such a gap also occurs in ipnopids and is said by Sulak (1977, p. 54) to characterize a "chlorophthalmid lineage" among iniomous fishes. Rosen & Patterson (1969) argue that the peculiar cervical joint (i.e., the gap in ossification) of chlorophthalmids and neoscopelids indicates relationship between the two groups and state (contrary to Gosline et al., 1966; and Sulak, 1977) that this peculiar joint is ". . . present in a somewhat simpler form in aulopids" (it is not evident in Hime). Johnson (1974c, p. 206) reports that scopelarchids possess the unossified gap of chlorophthalmids and neoscopelids — the first vertebra is a half-centrum attached to the rear of the skull through a tube of fibrous tissue, the length of this tube equal to or greater than the length of the first centrum. An extremely clear illustration (x-radiograph) of this gap in an adult Benthalbella infans is provided by Merrett et al. (1973, p. 17). They (Merrett et al., 1973, pp. 42-44) suggest that in scopelarchids the presence of the gap is related to feeding style, "... during active feeding, prey located above is struck at by a rapid, upwards arching of the predorsal region, with the flexible unossified section of the vertebral column allowing an extra backward bending of the head and hence a widening of the gape." In notosudids (Bertelsen et al., 1976, p. 8) there is a very slight gap in ossification — the gap filled by a short, rubbery tube — but the first centrum is not shorter than succeeding centra. In evermannellids as in all other inioms (including myctophids) there is no gap in ossification between the first centrum and the skull, and the first vertebra includes a full amphicoelous centrum (fig. 16A). (A) States Recognized for Iniomous Fishes Gap in ossification between skull and first centrum with reduction in size of first centrum relative to succeeding centra as described above is (0) = Absent. (39) = Present. 41. Supraneurals. — Rosen (1973, p. 450) states that a trend toward the reduc- tion or loss of supraneurals is characteristic of alepisauroids(R). Goody (1969, p. 220) regards the presence of three or four supraneurals, as seen in Aulopus and the Cretaceous Sardinioides, as primitive for iniomous fishes. All scopelarchids have three supraneurals (also true for chlorophthalmids, including Bathysaurop- sis; Bathysaurus and certain other synodontoids [Saurida, Harpadon, Tra- chinocephalus], and apparently also for myctophids [Jollie, 1954, p. 92]). Four supraneurals are present in the Neoscopelus microchir specimen in my material. Ipnopids have one or two supraneurals (Sulak, 1977). Evermannellids have two JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 89 supraneurals (fig. 16A), and this is apparently true for most paralepidids. Only a single supraneural is present in notosudids and Omosudis. Rofen (1966c, p. 499) indicates three supraneurals in Anotopterus, but I believe these to be expanded neural arch elements and find no supraneurals in my material. The state in Alepisaurus is unknown. (A) States Recognized for Iniomous Fishes The number of supraneurals is (0) = 3or4 (40) = 0, 1, or 2 42. Intermuscular Bones. — Intermuscular bones are well developed in all evermannellid and scopelarchid species (contrary to Marshall, 1954, p. 331; Gos- line et al., 1966, p. 11), and, apparently, in most (or all) inioms, including aulopids and the Cretaceous Sardinioides (Goody, 1969). The alepisauroid families Alepisauridae, Anotopteridae, Omosudidae, and Paralepididae (some genera) are characterized by an enormous proliferation (in both number of ele- ments and length of individual elements) of intermuscular bones (see Marshall, 1954, pp. 329, 330). The survey of iniomous groups with respect to configuration of intermuscular bones is too incomplete to allow formal recognition of character states at this time. 43. Number of Hypurals. — Rosen (1973, pp. 422-432) provides a detailed description of major patterns in the variation of caudal-skeleton morphology among euteleosts and characterizes (pp. 423, 424) the basal configuration for iniomous fishes. Typical for inioms (and many other euteleost groups) is the presence of six hypurals, and this is true for all extant inioms except certain synodontoids (Synodus foetens [HYP=5], Rosen, 1973, p. 427; Trachinocephalus tnyops [HYP=5], Rosen, 1973, p. 428; Harpadon nehereus [HYP=5], Rosen, 1973, p. 428; but not Saurida brasiliensis [HYP=6] or Synodus synodus [HYP=6], see Sulak, 1977, p. 56), some paralepidids (HYP=5, e.g., Paralepis, see Rosen, 1973, p. 429), and the families Anotopteridae, Alepisauridae, and Omosudidae (HYP = 5 in each case). In most inioms the dorsalmost hypural, HYP-6, is by far the smallest hypural element. Based on this fact plus the size of hypural elements remaining in those inioms with HYP =5, it appears that the typical pathway of reduction has been loss of HYP-6. The only exception appears to be certain paralepidids (Paralepis, Notolepis) in which it seems clear that reduction (in number of hypurals) has been through fusion of HYP-1 and HYP-2 (the articular heads apparently remain distinct, see Rosen, 1973, p. 429). This fusion is not present in all paralepidids (e.g., Macroparalepis, HYP=6). (A) States Recognized for Iniomous Fishes Number of hypurals is (0) = 6. (41) = 5, resulting from fusion of HYP 1+2. (42) = 5, resulting from loss of HYP 6. Hypothesized character-state sequence: (42) <— (0) -» (41) 44. Second Ural Centrum. — According to Goody (1969) and Rosen (1973), the basal configuration of the caudal skeleton in iniomous fishes includes fusion of the first preural (PU-1) and ural (U-l) centra but retention (as a terminal 90 FIELDIANA: ZOOLOGY half-centrum) of a free second ural centrum (U-2). Only three iniomous families lack representatives with a free U-2 centrum (Alepisauridae, Myctophidae, Scopelarchidae: see Paxton, 1972; Rosen, 1973; Johnson, 1974c). (A) States Recognized for Iniomous Fishes A free U-2 centrum is (0) = Present. (43) = Absent. Rosen (1973, p. 438) states, "Although in synodontids the second ural cen- trum is free as in aulopoids, this centrum is consolidated with the preceding compound centrum in other alepisauroids." This statement is true only for Alepisaurus and the scopelarchids among alepisauroid(R) fishes. 45. Number of Epurals. — In Aulopus and the Cretaceous Nematonotus there are three epurals (Rosen & Patterson, 1969), and the distribution of this state (table 8) suggests that possession of three epurals is primitive for iniomous fishes. In scopelarchids there are three epurals in all species except Benthalbella dentata, which has two. All evermannellids have only a single epural (fig. 16E). In representatives of a number of lineages (ipnopids, synodontoids, alepisauroids) the number of epurals has been reduced (EP=2 in Alepisaurus, Harpadon, Ipnops, Omosudis [the posterior epurals are fused distally], Saurida; EP=1 in Anotopterus, Bathymicrops, Synodus, Trachinocephalus) . (A) States Recognized for Iniomous Fishes Number of epurals is (0) = 3. (44) = 1 or 2. 46. Fleshy Midlateral Keel. — An adipose lateral keel, a raised, midlateral, dermal ridge arranged longitudinally, is apparently unique among inioms to two families. In Omosudis (Rofen, 1966b, p. 468) the keel is essentially limited to the caudal peduncle. In Alepisaurus (Gibbs & Wilimovsky, 1966, pp. 490, 492), the keel is present midlaterally along the caudal one-third to one-half of the body. (A) States Recognized for Iniomous Fishes An adipose midlateral keel as described above is (0) = Absent. (45) = Present. 47. Posttemporal. — Scopelarchids (Johnson, 1974c, pp. 57, 58) and ever- mannellids (fig. 18) are unique among inioms in that the posttemporal is un- forked. In both families the posttemporal of each side consists of a rodlike dorsal articulating process, which is connected to the epiotic, and a ventral bladelike area from which a strong ligament extends to the opisthotic. In all other inioms (e.g., Paxton, 1972, p. 28) there is also a ventral rod or spikelike process (of the posttemporal) directed toward and strongly connected to the opisthotic. (A) States Recognized for Iniomous Fishes The posttemporal is (0) = Forked. (46) = Unforked. JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 91 48. Pectoral Girdle.— According to Marshall (1954, pp. 325-328), a key char- acter tending to separate alepisauroid(G) from myctophoid(G) fishes involves the setting of the pectoral fin as measured by the angle between the horizontal axis of the body and the axis of the pectoral fin (bases). In most alepisauroids(G) this angle is less than 45°, and the fin tends to be low on the body and is rather hydroplane-like in configuration. In myctophoids(G) the fin is set higher on the body, and the angle of inclination tends to be steeper, generally greater than 45°. Marshall relates the hydroplane-like conformation in alepisauroids(G), which lack a swimbladder, to problems involved in maintaining position in the water column (and in part to problems involved in maneuvering). Marshall relates the relatively steep pectoral angle in myctophids to the presence of a swimbladder, tending to free the pectorals for use in braking and maneuvering. I am unwilling to use the angle of pectoral-fin insertion as a systematic character because, as Marshall's data (p. 327) on Gonostoma denudation (has functional swimbladder, pectoral angle = 45° to 50°) vs. Gonostoma bathyphilum (no swimbladder, pectoral angle = 20° to 25°) shows, the possibility for convergence (among mesopelagics with no swimbladder) is great. Scopelarchids and evermannellids differ in the number of extrascapulars (scopelarchids have two [or, possibly, in the case of Rosenblattichthys volucris, three]; evermannellids have one [fig. 18]) and postcleithra (scopelarchids have two widely separated postcleithra; evermannellids have three postcleithra con- nected in sequence [fig. 18]). The number of extrascapulars in other inioms varies from one (Aulopus, Chlorophthalmus, neoscopelids, notosudids, paralepidids, Anotopterus, Omosudis) to two (Parasudis, some synodontoids [Synodus], myctophids). The number of postcleithra in other iniomous fishes varies from one (some ipnopids [Ipnops, Bathymicrops], Anotopterus) to two (some ipnopids [Bathytyphlops], myctophids, neoscopelids, notosudids, Omosudis), to three (aulopids, chlorophthalmids, Bathysauropsis, certain synodontoids [Bathysaurus, Saurida], paralepidids). In both cases the available data is far too meager relative to the observed variation to allow formal recognition of character states at this time. 49. Luminous tissue. — Rosen (1973, p. 505) lists two derived character states as unique (autapomorphic) to his Myctophiformes: (1) presence of an ethmoid crest and (2) presence of photophores. Rosen does not elaborate on the "ethmoid crest," nor can I. Photophores are of course not unique to the Myc- tophiformes(R), occurring widely in elasmobranch and teleost as well as inver- tebrate groups. More to the point, I note (1) the development of a myctophiform(R)-like photophore (isthmial organ) in the paralepidid Lestidium bigelowi (Graae, 1967), and (2) Bolin (1966, pp. 192, 193) notes that the arrange- ment and (more importantly) structure of photophores in Neoscopelus (the only neoscopelid genus whose members possess photophores) are ". . . markedly different [in] character [from those of myctophids]." In Neoscopelus the photogenic tissue and enveloping black pigment is largely restricted to the pos- terior end of the organ, the anterior portion is a simple, flaring sheet of reflective guanine (overlain by translucent tissue) and lacking definite anterior limits, and the overlying scales are not modified into lenses (cf., Myctophidae, see Nafpak- titis et al., 1977, pp. 15-18). Despite these differences, the consensus of opinion, based on external morphology, osteology, and larval morphology (Fraser- 92 FIELDIANA: ZOOLOGY Brunner, 1949; Moser & Ahlstrom, 1970; Paxton, 1972; Rosen, 1973; but not Marshall & Staiger, 1975), indicates sister-group relationship for the myctophids and neoscopelids. Following this consensus, I recognize two states of character 49 for iniomous fishes: (A) States Recognized for Iniomous Fishes (0) = Photophores absent. (47) = Photophores present. Note that in assigning state (47) to the neoscopelids (table 8), I am following the preliminary discussion of interrelationships among neoscopelids provided by Nafpaktitis (1977, p. 3). Thus the presence of photophores is here hypothesized to be primitive for this family. Among other iniomous families, only certain paralepidids (Rofen, 1966a; Graae, 1967), scopelarchids (Merrett et al., 1973; Johnson, 1974c), and the ever- mannellid genus Coccorella (Herring, 1977) are known to be luminous. Evidence suggests that the presence of luminous organs in Coccorella is synapomorphous (a state which may be shared with species of Evermannella, see section on Luminescence). (B) States Recognized for Evermannellid Species (0) = Luminous tissue absent. (E16) = Luminous tissue present, associated with ventral wall of intestine ("intestinal organs") or pyloric caecum ("isthmus organ," see Her- ring, 1977). 50. PSM Cephalic Laterosensory Pores. — Evermannella megalops is unique among evermannellid species in lacking the medial snout-pad pores (PSM, fig. 1; see Johnson & Glodek, 1975, p. 721). This feature is probably related to the tremendous expansion of the eyes and foreshortening of the snout in this species (Johnson & Glodek, 1975). (B) States Recognized for Evermannellid Species PSM cephalic laterosensory pores are (0) = Present. (E17) = Absent. 51. Number of Dorsal-Fin Rays in Evermannellid Species. — Almost with- out overlap the genus Evermannella can be divided on the basis of number of dorsal-fin rays (table 2): 10 to 11 in E. ahlstromi (one specimen of 78 with 12 rays) and E. megalops (one specimen of 10 with 12 rays) vs. 12 to 13 in E. balbo and E. indica (no exceptions). I regard the state in E. ahlstromi and E. megalops as derived for the following reasons: (1) In Odontostomops normalops and Coccorella atlantica the great majority of specimens (92% and 95%, respectively) have 12 or 13 dorsal-fin rays; only in Coccorella atrata is there substantial overlap (40% with 10 or 11, 60% with 12 or 13 rays, table 2); (2) within Evermannella, the state ex- hibited by E. ahlstromi and E. megalops apparently correlates with derived states in other characters (lateral line segments lacking [character 5], lack of PSM pores in E. megalops [character 51], and hypertrophy of gill filaments in E. ahlstromi [fig. 29]). JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 93 (B) States Recognized for Evermannellid Species Modal number of dorsal-fin rays is (0) = 12 or 13. (E18) = 10 or 11. Interrelationships Among Iniomous Fishes Most attempts at finding derivative character states monothetically defining the Iniomi have failed. The problem is well summarized in Gosline et al. (1966, p. 5): "Though there is apparently no one feature that will separate all iniomous from all isospondylous fishes, this is hardly surprising in two such large and varied groups, groups which are ... an expression of very early teleost ("explo- sive") evolution. . . ." Again, as stated by Rosen & Patterson (1969, pp. 452, 453): "What we find is a mosaic of primitive and advanced features, the first lost and the second acquired in different lineages in a pattern that can be disentangled in a number of ways depending upon the weight attached to various charac- ters. . . ." Most "definitions" of the Myctophiformes (e.g., Gosline et al., 1966; Goody, 1969; Gosline, 1971) have involved listings of character states that are shared with certain salmoniform groups and/or are primitive for neoteleosts. Such listings are not useful in demonstrating the monophyletic origin of the Iniomi. Rosen (1973, pp. 505-510) argues for the monophyly of an aulopiform(R) group containing all iniomous families except the Myctophidae and Neo- scopelidae. The latter families constitute in Rosen's classification a monophyletic and much restricted order Myctophiformes(R), the supposed sister-group of "higher" neoteleosts (Paracanthopterygii + Acanthopterygii). As noted by Zehren (1975, p. 213), the aulopiforms(R) exhibit a number of specialized char- acters also occurring in the myctophiformes(R), Paracanthopterygii and Acan- thopterygii, arguing for the monophyletic origin of the entire assemblage but not adding to Rosen's arguments for the monophyly of his Aulopiformes. Rosen (1973, p. 403) regards the presence of a notably elongate EB2 uncinate process as uniquely diagnostic of his aulopiform group, but, as indicated in my discussion of character 39, I regard as open to question the exclusion of the Myctophidae + Neoscopelidae from the aulopiform(R) group on the basis of this character. An additional character, not considered by Rosen (1973), involves the presence and conformation of peritoneal pigment sections in larvae (character 8). Discrete peritoneal pigment sections are characteristic of the larvae of iniomous fishes (excepting apparently the families Myctophidae, Neoscopelidae, Notosudidae) and may prove diagnostic for the group — unfortunately adequate knowledge of larval morphology in other major neoteleost groups is lacking. Rosen (1973, p. 505) lists two features as autapomorphic for his Myc- tophiformes: (1) presence of photophores, (2) presence of an ethmoid crest, but elaborates on neither feature (see discussion of character 49). Other character states listed by Rosen (1973, p. 506) are said to separate the Aulopiformes(R) vs. Myctophiformes(R) and are said to be shared by the latter with the Paracantho- pterygii + Acanthopterygii. Zehren (1975, p. 214) agrees with Rosen's conclu- sions but notes that the evidence provided by Rosen's listing of characters might 94 FIELDIANA: ZOOLOGY well be taken to suggest ". . . that myctophids are more closely related to paracanthopterygians and acanthopterygians than to neoscopelids." For purposes of exploring relationships between the evermannellids and scopelarchids, I herein adopt the view traditional among ichthyologists (e.g., Gosline et al., 1966; Goody, 1969; Marshall & Staiger, 1975), viz., that the Iniomi constitutes a monophyletic assemblage. I note, however, that myctophids and neoscopelids differ strikingly from other iniomous fishes in several seemingly trenchant characters. It may well be that thoroughgoing study of this group of fishes will corroborate Rosen's classification. Based on the assumption of monophyly of the Iniomi, I have prepared a dendrogram (fig. 20) illustrating the distribution of derived character states among five possible major groups of iniomous fishes distributed among three perceived lineages. Evidence presented in support of these groupings is based on the foregoing character catalogue. Evidence for two of the groups (myc- tophoids, alepisauroids) appears substantial, but the remaining groups are not well supported. This study makes no pretense at producing a formal classifica- tion of iniomous fishes, an effort awaiting a more comprehensive survey than presented here. My object in the discussion to follow is twofold: (1) to sum- marize the preceding catalogue of characters and character states in terms of distribution of derived character states among iniomous taxa and (2) to examine the available evidence in support of (or against) the supposed sister-group relationship between evermannellids and scopelarchids. Of the three perceived "lineages" — aulopoids, myctophoids + chlorophthal- moids, synodontoids + alepisauroids — there exists, to my knowledge, no basis for asserting closer relationship among any two relative to the third. A central problem is illustrated by the fact that of 36 characters for which states were defined for iniomous taxa, the aulopids exhibit a derived state only in character 3 (2, absence of a swimbladder — a state shared with all iniomous taxa except the myctophoids). Rearrangement of the tree to include the aulopids but to exclude the myctophoids conflicts with character 12 (number of branchiostegal rays) — the result, for the time being, is an impasse. Adding to the difficulty is evident diversity within the Aulopidae, illustrated by the divergence between Aulopus and Hime in character 8 (peritoneal pigment section single in Hime; multiple sections present in Aulopus), diversity that can only be interpreted after a much-needed revisionary study of this family is completed. As noted above, of the remaining groups and lineages, the best-supported groups, in terms of number of synapomorphic features (fig. 20, table 10), are the alepisauroids and myctophoids. The five families herein postulated to represent a chlorophthalmoid + myctophoid lineage group together primarily on the basis of symplesiomorphy. Only a single derived state in character 12 (11, number of branchiostegal rays) provides evidence for the grouping suggested, and that state is shared with the alepisauroids. The myctophoids (Myctophidae, Neoscopelidae) share derived states of char- acters 8 (7, lack of peritoneal pigment sections), 12 (12, number and arrangement of branchiostegal rays), 13 (13, number of vertebrae), 22 (21, lack of supraorbital bones), 31 (30, loss of PB4TP), 35 (35, division and insertion of RAB muscle), and 49 (47, possession of photophores). The myctophoids differ from all other in- iomous fishes in the symplesiomorphic state of character 3 (possession of a swimbladder). Additional derived states shared by myctophoids as listed by 38" JJ_L Miansow SSSS22" I I I I I I 1 3WITUKWNW 1 I I 555 8*" I'll I mTTITiii 1 1 1 1 1 1 1 "ill I JJJ _LL _LLL -O o I I ■ •8 Ss MldONdl I I I wdiinvHiHdoaonio S382' wairanraxns I I I I I I Ji l_L -LL 2 • 0 I I o c u 00 rH o O rn CN i— i (N tv H (N ri^t ^ 't »C H (N| H Tt ,2 O T3 Jo JS II ■5 ^ a, . O 01 fe « O -0 si II s. £ 'EL ~> (X) II o c^ I I I I I 1 3 HMms«in»oNooa>oriMm^ O OOOOOOOOOrHrHr-lrHr-i M o -go EL 411.94±4.96 144.25±2.76 <166.24±4.44 35.95±1.07 < 52.59±1.61 Table 13. Comparison of values for five morphometric characters (and range of SL) between North Atlantic (NA) specimens of Coccorella atlantica and northern Indian Ocean (ION) specimens of C. atrata. Character SL (mm) BDAO ADC PD POHL IO BDAO ADC PD POHL IO KEY: BDAO=body depth at anal-fin origin; ADC=distance from adipose-fin base to bases of midcaudal rays; PD=distance from snout to dorsal-fin origin; POHL=postorbital head length; IO=interorbital width. In my attempt to find morphometric characters useful in separating C. atlantica and C. atrata, I first compared results for 20 specimens assigned to C. atlantica from the North Atlantic with results for 17 specimens assigned to C. atrata from the northern Indian Ocean. Of the 29 measurements per specimen (see Methods), values for five characters (table 13) sufficiently differed between the two species to be of possible value in separating the two forms. Only one of these characters, interorbital width, showed separation without overlap (table 13), and this separation held throughout subsequent study (fig. 25). The interor- bital width is 3.2% to 4.7% SL in C. atlantica vs. 4.7% to 6.1% SL in C. atrata. If the intermediate specimens are hybrids, they might be expected to show intermediate values of these five morphometric characters. If true, then each of the six intermediates should exhibit an intermediate value of a combined char- acter index based on all five characters. The combined character index for each specimen was computed as the sum of a standard score for each character for each specimen. The standard score was constructed by adjusting the value for each character for each specimen relative to a mean value of 100 for each char- acter for all specimens of C. atlantica from the North Atlantic Ocean. The stan- dard score and the combined character index were computed as follows: (1) Y u = (Xij/SLj) x 1,000 where Xjj represents the original measurement (in mm) for character i on speci- men j, SLj is the standard length (in mm) of specimen j, and Y u is the raw measurement expressed in thousandths of the standard length. (2) Mi = (2 Y,j) / N, where N is the number (=20) of North Atlantic specimens of C. atlantica mea- sured for each character, and Mj is the mean value for each character for these specimens. .2 M> ■5*8 £ c O « c S •Si *5 ^? £? s.y£ B (M _ C/) ^ S3 o> u«vf '.e J2 en ■ a < ri ^ •c ^ tn ^ bbo;- 3 I o g ok us * ^T3 _ (fl (0 Jil|l .3. 8 B w iJ II E^^w 1 1 1 s & i 3 -S3 frSS E_i 111 112 FIELDIANA: ZOOLOGY (3) Z,j = (Yy/M,) x 100 where Zjj is the standard score for each character for each specimen. (4) CIj = Z u + Z 2J + ( - Z 3J ) + Z 4J + Z»j where Z^ is the standard score for each specimen for body depth at anal-fin origin, Z 2 corresponds to distance from adipose-fin base to bases of midcaudal rays, Z 3 corresponds to distance from snout to dorsal-fin origin, Z 4 corresponds to postorbital head length, Z 5 corresponds to interorbital width, and CIj is the combined character index for each specimen. Combined character indices were determined for 62 specimens assigned to C. atlantica, 44 specimens assigned to C. atrata, and the six intermediate specimens. The specimens were chosen from throughout the range of each species (figs. 24, 26) and represented the full size range (by 10-mm increments) of post- metamorphic juveniles and adults available from each of the eight subareas of the total range of Coccorella (C. atlantica: North Atlantic, 20 [34.0-142.7]; South Atlantic, 12 [40.8-184.5]; southern Indian Ocean, 2 [56.2-146.2]; North Pacific, 11 [37.2-79.6]; South Pacific, 17 [35.5-152.5]; C. atrata: northern Indian Ocean, 17 [36.2-104.6]; insular western Pacific, 11 [43.2-98.6]; equatorial Pacific, 16 [38.7-100.9]). A plot (fig. 27) of the combined character indices for the 68 specimens included in the analysis shows a clear separation between the two species and also shows that the six intermediates do not tend to group between the two modes but rather agree with the results for number of vertebrae and the distributional data in allowing the clear assignment of specimen (1) to C. atrata and specimens (2 to 6) to C. atlantica. Application of multivariate techniques to this problem yielded virtually identi- cal results. Canonical analysis of the data was performed using the program BMD 07M (Dixon, 1968). In each case the previously determined groups were identical to those used for the character index comparison (fig. 27), with the following exceptions: (1) data for the two Indian Ocean specimens of C. atlantica were combined with data for the South Atlantic specimens of this species in one group; (2) in the first run of the data, values for the intermediate specimens were not included (fig. 28A); (3) in the second run, values for the intermediate speci- mens were included with data for the geographically most adjacent group (i.e., specimen (1) was included with IWP [fig. 28B], specimens (2, 3, 4, 5) with SA, specimen (6) with SP); (4) in the third run, values for the intermediate specimens were included as one separate group — the purpose of this being to test whether or not the analysis accepted the six intermediates as one group. Although the data for each character for each specimen were entered as ratios, i.e., expressed as thousandths of the standard length, at least part of the objection to this procedure is countered by my effort to equally represent in the data all size classes (by 10-mm increments) for each area for each species (table 14). Obvi- ously this was not possible for specimens exceeding 104.6 mm SL (the length of the largest known individual of C. atrata). The results of this analysis, based on the five characters (table 13) included in the combined character index analysis, compare favorably with results for vertebral number and for the character index analysis, i.e., intermediate specimen (1) clusters with C. atrata, intermediate specimens (2 to 6) cluster with C. atlantica (fig. 28C). I have run the same analysis using 20 of the 29 morphometric characters (excluding those showing obvious 113 o to x re oi C " — ' CD vo F, II "8 Cn be JS 5 a bo * a c Oi TO I JS * X • F, en TJ I 01 o> I "™ ~ c ' ^i -S -S I 1? ~ X> ■ ti n't! l <° is E IP S 2 & <1 -S 3 « < J3 II - ~° -2 rr, 8 <° 8< • oZ fc cn a a 32 111 D U •*- TO is I * J A TO-O-gCD bob x Ch o w T3 U c o» II • | II >, ©•° £* 0) Qo II | 1 - ill X «j « 111 .' — c a <4 'T • • « e «f« CD OOr •4 an o° ® I □ m o (D D o a 114 3 > 5 UBX.O & c o.^ c ° 3 II a. "-a 1 " cxxi £ -.3 =3 ISO £ 5 DcC2 + " r" 01 W a» c c Q.5 « j= . E c * a. >^ £Jg>J*IB&8. y«.Sg.ESs3 S lie SS^S* £'§t-3*-3:H as 1*1 l- o £ n " &8 2 _ o> « o. c • m o U ■n oi to o C 5 - _i c > o — x II c _ Si n » r i" fcV C — *S * U e I « 'C r^ e •-" C P ra | £ °J'05> a ii oa o 1 ^ 'S oi • - « Sec " ^^2 g u. . < c £ g-oy o .S & .5 l ™ g\S ii ~ 1 E "5 ^£-9^ E -a e: ms^ .JnslJ-SE.g ^S^ESS.Sfr 115 116 FIELDIANA: ZOOLOGY Table 14. Representation by size class (10-mm increments) of specimens of Coccorella in canonical analysis of five morphometric characters. Geographic areas as defined in Figure 26. See text for additional explanation. Geographic areas Size C. atlantica C. atrata class SA + Intermediates (mm) NA IOS NP SP ION IWP CEP (all specimens) No of Specimens 30.0-39.9 3 2 3 2 1 40.0-49.9 3 4 1 3 5 2 3 50.0-59.9 3 1 3 2 3 4 2 60.0-69.9 4 1 2 1 2 2 1 70.0-79.9 1 1 3 2 1 1 80.0-89.9 2 1 2 4 3 2 90.0-99.9 1 1 3 4 100.0-109.9 2 1 1 2 - > 110.0 1 5 4 1 Totals 20 14 11 17 17 11 16 6 allometric growth) for 37 specimens of C. atlantica, 24 specimens of C. atrata, five of the six intermediates (it was impossible to determine values for certain char- acters in specimen (3) due to damage to the specimen), and from throughout the geographic range and over the post-metamorphic size range of Coccorella. Results for this analysis (fig. 28D) are virtually identical with results based on the five selected characters. I believe that the conclusions to be drawn from these results are (1) that there exist two discretely recognizable species of Coccorella, C. atlantica and C. atrata, and (2) that the six intermediate specimens do not represent hybrids and that the abnormal state of the frontal canal commissure in each is probably the result of a developmental accident. Two additional points derived from this study of Coccorella seem worthy of mention at this point: (1) Coccorella atlantica, the central water species, apparently grows to nearly twice the size of C. atrata, the species found in more productive equatorial waters (figs. 24, 25). The largest known specimen of C. atlantica is 184.5 mm SL, whereas the largest known specimen of C. atrata is 104.6 mm SL. Ebeling (1962, pp. 145-147) discusses at some length the apparent trend among species of Melamphaes for areas of lower biological productivity to be occupied by smaller-bodied species (viz., "dwarf" species of the M. simus group). Certainly no evidence for such "dwarfing" of central water species is to be found in comparing C. atlantica with C. atrata; in fact the reverse appears to be true. Dr. Gerhard Krefft (in a letter) cites evidence that in the circumtropical, mesopelagic gonostomatid Diplophos taenia (see Johnson & Barnett, 1972, 1975) populations in areas of higher productivity are smaller-bodied than those in areas of lower productivity. Krefft found that specimens from the relatively productive equatorial Atlantic area (ca. 20° N to 07° S at about 20° W) represented a "dwarf" form, sexually mature at less than 100 mm SL. Specimens from the central South Atlantic (ca. 20° to 40° S), an area of relatively low biological productivity, included JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 117 "giants," reaching 265 mm in SL and not attaining sexual maturity until a size of 150 mm SL or larger. Thus, what appears to be true between species in Coccorella may be true between populations in Diplophos, i.e., popula- tions or species found in the areas of lowest productivity attain the largest adult body sizes. In general this trend, if true, parallels the conclusion offered by Ebeling & Cailliet (1974) that forms inhabiting zones of lower food availability tend to have larger bodies with consequently larger mouths, allowing the capture and ingestion of a wide range of food parti- cle sizes. (2) The difference in vertebral counts between C. atlantica (48 to 50) vs. C. atrata (45 to 47) parallels the results of Johnson & Barnett (1972, 1975), who showed that populations of widely ranging species inhabiting areas of lower productivity tend to exhibit higher values of longitudinal meristic characters than populations inhabiting areas of higher productivity. The results for C. atlantica and C. atrata suggest, at least, that this trend may also prove true for comparisons between closely related species. Coccorella atlantica (Parr 1928), Figure 22 Evermannella atrata atlantica Parr 1928, p. 166 (original description, seven syntypes from off the Bahamas); Parr, 1929, p. 21 (osteology); Parr, 1930, p. 154 (synonymy); Grey, 1955, p. 284 (report of two specimens taken off Bermuda, one of these specimens later reidentified as Ei'ermannella indica by Rofen, 1966d, p. 536). Coccorella atlantica Wassersug & Johnson 1976, p. 273 (gut morphology, illustration of larval and juvenile specimens, records from Atlantic, Indian, and Pacific oceans); Herring, 1977, p. 306 (name only). Coccorella atrata Rofen 1966d, p. 528 (not Coccorella atrata Alcock 1893, description, records from western North Atlantic); Johnson, 1974c, p. 30 (not Coccorella atrata Alcock 1893, record from central North Pacific). Coccorella atrata atlantica Roule 1929, p. 11 (original description, not based on Everman- nella atrata atlantica Parr 1928, type from eastern North Atlantic). Evermannella atlantica Beebe 1937, p. 205 (record from off Bermuda). Odontostomops sp. A, Schmidt 1918, p. 33 (illustration of 9-mm specimen from western North Atlantic, records from North Atlantic Ocean). Lectotype. — Bingham Oceanographic Collection No. 2141, 1 (40.6). Western North Atlantic, 23° 58' N, 77° 26' W (7,000 ft wire), Nov. 2-3, 1927. Parr (1928, p. 166) based his description of Evermannella atrata atlantica on seven specimens from five collections. Of this series of syntypes the specimen herein designated as the lectotype was the largest and the specimen labeled as the holotype by Parr (see Rofen, 1966d, p. 528). Diagnosis. — A species of Coccorella with 48 to 50 vertebrae, typically (238 of 243 specimens examined for this character) with six pores, 3 + 3, in the frontal canal commissure of the cephalic laterosensory system (fig. 23) and interorbital width = 3.2% to 4.7% SL. Detailed comparisons of the two species of Coccorella are given above, following the description of the genus. Description. — Values for meristic characters are presented in Tables 2 and 3. PROPORTIONAL DIMENSIONS: Based on 37 (36.0-184.5 mm SL) specimens from throughout the range of the species. Expressed as thousandths of the SL and given as the mean and range (values in parentheses). Body: depth at anal-fin origin, 163 (144-191). Caudal peduncle: least depth, 74 (60-87); length, 94 (77-105). Adipose fin: distance to midcaudal rays, 122 (108- 141); distance to dorsal-fin base, 338 (306-365). Anal fin: length of base, 256 118 FIELDIANA: ZOOLOGY (233-283). Dorsal fin: length of base, 113 (99-132); dorsal-fin origin to anal-fin origin (distance between verticals), 260 (232-287); end of dorsal-fin base to base of midcaudal rays, 501 (472-527). Pelvic-fin insertion to anal-fin origin: 217 (193-239). Pectoral-fin insertion to pelvic- fin insertion: 204 (164-247). Anus to anal-fin origin: 63 (27-104). Distance from snout to: anus, 610 (558-673); dorsal- fin origin, 432 (414-459); adipose fin, 839 (800-864); anal-fin origin, 665 (631- 707); pectoral-fin insertion, 258 (234-287); pelvic-fin insertion, 451 (418-501); anterior margin of eye (=snout length), 59 (51-69). Head length: 222 (196-243). Postorbital head length: 142 (122-157). Eye: horizontal diameter, 53 (40-65); vertical diameter, 56 (42-76). Upper jaw length: 172 (157-184). Lower jaw length: 171 (151-188). Longest dentary tooth: 52 (41-63). Longest palatine tooth: 84 (71-96). Interorbital width (based on 62 [34.0-184.5 mm SL] specimens): 38 (32-47). BODY: Body moderately elongate, largest known specimen 184.5 mm SL (At- lantic Ocean, ISH 1436/68). Body moderately deep, body depth at anal-fin origin, 14.4% to 19.1% SL. Anus distinctly posterior to a point midway between pelvic-fin insertion and anal-fin origin. Lateral line extending to a point over anterior one- third of anal-fin base and composed of 34 or fewer segments. About eight pairs of sensory (presumably) papillae along dorsal margin of body an- terior to dorsal-fin origin. CEPHALIC LATEROSENSORY PORES: Snout-pad pore formula: 2 + 4 + 2 + 1+2. Mandibular pore formula: 5 + 6 + 2 + 1. Preopercular pores: ? 3 + 3. Temporal pores: PPTO = 1, PESC = 4. Frontal pores: PPF = 1. Frontal canal commissure: 3 + 3 (in 238 of 243 specimens examined, a discussion of the five specimens assigned to C. atlantica and not agreeing in this character is given above). Infraorbital pores: 13 or 14 + 9. Numerous sensory (presumably) papil- lae distributed over occiput, interorbital region, snout, cheeks, and anterior lower jaw. MOUTH: Upper jaw extending to or nearly to anterior margin of preopercle, well past a vertical through posterior margin of eye. Lower jaw projecting an- teriorly very slightly beyond snout. Tooth counts given below are based on 11 (48.7-146.2 mm SL) specimens. Premaxillary teeth small, retrorse, uniserial, numbering 29 to 55 in the 11 speci- mens counted. Dentary with two smaller fangs anteriorly near symphysis fol- lowed by nine to 17 unbarbed larger fangs arranged uniserially. The largest fangs occur anteriorly. Vomer with one small tooth on each side, but in numer- ous specimens only a single, laterally positioned tooth is present, its counterpart probably being lost. Anteriormost palatine tooth an enormous, unbarbed, saber-like fang, easily the largest tooth on each side, noticeably exceeding the snout length in length. Anteriormost palatine fang separated from posterior palatine teeth by a distinct gap. Posterior palatine teeth numbering six to nine in the 11 specimens counted. Number of premaxillary, dentary, and palatine teeth higher in larger specimens. COLOR: Color in alcohol a dark brown over head, body, and fins, without any distinct concentration into stripes, bars, or markings. In well-preserved speci- mens a brassy green, iridescent layer is present along flanks, beneath eyes, and on cheeks. In fully metamorphosed specimens less than 60 mm SL a distinct and typically elliptical lens pad is present. In larger specimens there is no distinct lens pad as in Evermannella, but an opaque (presumably transparent in life), JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 119 glistening tissue centered beneath the lens is probably similar in function. Mid- ventral areas of body anterior to anal-fin origin with especially dense pigmenta- tion except for areas of luminous tissue (see above). Peritoneum dense black. Discussion. — Rofen (1963, p. 2; 1966d, p. 536) synonymized Evermannclla atrata atlantica Parr 1928 with Coccorclla atrata (Alcock); stated that the range of C. atrata included the North Atlantic Ocean, Bay of Bengal, and Andaman Sea; and described Odontostomops braueri based on Brauer's (1906, pp. 136, 137, plate 10, figs. 3, 4) description and illustrations of a specimen Brauer identified as Ever- mannella atrata (Alcock). As I have shown, there are two clearly separable species of Coccorella, and the species occurring in the North Atlantic (C. atlantica) is not conspecific with the species occurring in the northern Indian Ocean (C. atrata). Because Rofen's description of C. atrata is based on nine specimens from the North Atlantic, including four syntypes of E. atrata atlantica and five specimens from off Bermuda, it is clear that these specimens in fact belong to C. atlantica. This is confirmed by reading Rofen's (1966d, pp. 529-534, figs. 194, 195) excel- lent description of his specimens and by my reexamination of the Bermuda specimens now deposited at Field Museum of Natural History. Further, because Brauer's (1906) specimen possessed a semitubular eye, huge but unbarbed fangs on the palatine bone, and other characters diagnostic for Coccorella (Brauer 1906, pp. 136, 137, plate 10, figs. 3, 4); was taken well within the known Indian Ocean range of C. atrata (it was taken at 04° 5.1' S, 73° 24.1' E, cf. fig. 24), and exhibits 2 + 2 pores in the frontal canal commissure (Brauer, 1906, plate 10, fig. 4). I assign Odontostomops braueri Rofen to the synonymy of Coccorella atrata (Alcock). Geographic Variation. — Although mean values for number of anal-fin rays are higher (table 15) in Pacific samples of C. atlantica than in Atlantic samples, the differences between mean values for Pacific vs. Atlantic specimens are not statis- tically significant (p > .05). Thus, the question of whether there exists geo- graphic variation in this character awaits resolution based on additional material. Distribution. — Coccorella atlantica is known from central water areas of the At- lantic, Indian, and Pacific oceans (fig. 24). Only two specimens from two lots are known from the Indian Ocean (IOAN, V 4603, 16° 05' S, 76° 16' E, 1 (56.0); SOSC, ELT 35-2271, 38° 12'-06.2' S, 128° 01.0' to 127° 57.5' E, 1 [146.5]). The distribution of C. atlantica and C. atrata is compared above, and the distribution of C. atlantica is compared with that of other central water species in a sub- sequent section of this paper. Larvae and small juveniles (to 30 mm SL) of C. atlantica have been taken commonly in the upper 125 m, but very few records are from hauls to depths less than 50 m. Most adults (greater than 50 mm SL) were taken in hauls to depths exceeding 500 m, but both juveniles and adults have been taken on numerous Table 15. Geographic variation in number of anal-fin rays in Coccorella atlantica. No. of anal-fin rays Area 26 27 28 29 30 N Mean ± 95% limits North Atlantic — 9 44 2 — 55 27.87±.117 South Atlantic 1 5 11 — — 17 27. 59 ±.318 Indian — 1 1 — — 2 27.50 North Pacific — 1 13 4 1 19 28.26±.315 South Pacific — 2 10 5 — 17 28.181.327 120 FIELDIANA: ZOOLOGY occasions in hauls to between 100 and 400 m. Larvae and small juveniles have been taken throughout the year. Material Examined. — A total of 403 (6.2-184.5 mm SL) specimens from 269 collections. ATLANTIC OCEAN. A total of 344 (6.2-184.5 mm SL) specimens from 218 collections. FMNH: 49985 (1), 49987 (1), 66096 (2), 78579 (1). IOAN: AK 826 (1). ISH: 86/66 (1), 186/66 (2), 230/66 (2), 676/66 (1), 702/66 (1), 722/66 (1), 766/66 (1), 786/66 (1), 824/66 (1), 1168/68 (1), 1281/68 (2), 1318/68 (1), 1350/68 (1), 1369/68 (1), 1436/68 (1), 2127/68 (1), 628/71 (1), 802/71 (1), 847/71 (1), 919/71 (2), 1106/71 (2), 1135/71 (1), 1173/71 (1), 1227/71 (1), 1362/71 (1), 1431/71 (1), 1436/68 (1), 1521/71 (1), 1562/71 (2), 1613/71 (1), 1632/71 (1), 1986/71 (2), 2034/71 (1), 2884/71 (1), 2939/71 (3). SIO: 63-552 (1). UMML: 9047 (2), 11832 (2), 11875 (1), 14897 (1), 15903 (1), 15906 (1), 17784 (1), 18355 (1), 22987 (1), 24178 (2), 26439 (1), 26452 (1), 27278 (1), 27728 (1), 27906 (1), 29435 (1), 29456 (1). USNM, UNCAT.: ORE 3219 (1), ORE 4569 (1), UND 1966-3, 17 (1); USNM, ACRE: 1-4A (1), 1-31 (1), 4-30 A-D (1), 7-13 (1), 8-2 (1), 8-3 (2), 10-15m (1), 11-lOc (1), 12-18A (1), 12-53 (1), 12-56 (1), 12-63 (2), 12-64 (1), 12-67 (1), 12-69 (3), 12-70 (3), 12-72 (2), 12-79 (1), 12-81 (2), 12-83 (1). WHOI, RHB: 867 (2), 873 (1), 910 (1), 1101 (2), 1112 (1), 1261 (2), 1263 (7), 1264 (2), 1271 (1), 1274 (7), 1281 (5), 1282 (3), 1287 (1), 1289 (1), 1290 (23), 1291 (2), 1294 (2), 1297 (2), 1300 (1), 1302 (1), 1307 (2), 1308 (1), 1309 (1), 1310 (3), 1313 (2), 1428 (1), 1441 (1), 1505 (12), 1509 (7), 1713 (3), 1716 (1), 1718 (1), 1728 (1), 1731 (1), 1737 (4), 2024 (1), 2093 (1), 2100 (1), 2111 (1), 2904 (1), 2908 (1), 2913 (1), 2917 (1), 2926 (1), 2945 (1), 2946 (2), 2948 (1), 2951 (2), 2956 (1), 2957 (2), 2960 (2), 2962 (1), 2965 (2), 2976 (1), 2985 (2), 2990 (1), 2993 (1), 3003 (2), 3014 (1), 3015 (1), 3017 (1), 3018 (2), 3021 (1), 3104 (2), 3105 (1), AEJ 008 (1). ZIZM: RBF #32 (1). ZMUC: D 859 (1), D 1041 (1), D 1043 (1), D 1180 (1), D 1182 II (1), D 1183 VIII (3), D 1190 I (1), D 1238 I (1), D 1256 II (1), D 1339 II (1), D 1342 VIII (1), D 4014 IV (1). Additional larvae and juvenile material from the Atlantic Ocean. CAS: CAS 14857 (4), SU 57717 (1). USNM, ACRE: 3-13 (1), 3-14 (1), 10-38N (2), 12-1B (1), 12-13B (2), 12-13C (4), 12-13M (1), 12-14B (1), 12-14M (1), 12-34M (2), 12-36B (1). ZMUC: D 839 (1), D 842 (1), D 845 (3), D 855 XVII (3), D 855 XVIII (1), D 856 VII (1), D 857,150 mwo (2), D 857,200 mwo (1), D 863 (2), D 864 (1), D 865 (1), D 944 (2), D947,300 mwo (1), D 947,400 mwo (1), D 947,1000 mwo (1), D 948 (1), D 949 (1), D 1157 V (1), D 1185 IX (1), D 1186 VII (1), D 1189 VII (1), D 1195 II (3), D 1217 V (1), D 1218 II (1), D 1228 II (4), D 1229 II (1), D 1230 IV (1), D 1231 II (1), D 1239 II (1), D 1239 IV (1), D 1239 VI (1), D 1241 VII (2), D 1242 IX (2), D 1242 XV (1), D 1243 III (4), D 1253 II (1), D 1261 II (1), D 1267 IV (1), D 1283 X (1), D 1285 III (1), D 1322 IX (1), D 1323 VIII (2), D 1323 XIV (1), D 1327 II (1), D 1332 XV (1), D 1335 V (1). INDIAN OCEAN. A total of two (56.0-146.5 mm SL) specimens from two collections. IOAN: V 4603 (1). SOSC: ELT 35-2271 (1). PACIFIC OCEAN. A total of 57 (15.7-175.0 mm SL) specimens from 49 collections. IOAN: AK 236 (1), V 6033 (1), V 6493 (3). NMFS (LJ): J 24.133 (1), J 24.145 (1), J 31.145 (1). ORSTOM: CY 111-18 (1), CY VI-16 (2). SIO: 61-47 (1), 64-482 (1), 68-442 (1), 68-490 (1), 69-341 (1), 69-354 (1), 70-102 (1), 70-110 (1), 70-121 (1), 70-311 (1), 70-331 (1), 70-336 (1), 71-297 (1), 71-301 (1), 71-310 (1), 72-9 (1), 72-24 (2), 72-304 (1), 72-305 (1), 72-308 (1), 72-316 (1), 72-321 (1), 73-322 (1), 76-121 (1), 76-133 (2), 76-134 (1), 76-147 (1). UH: 70-7-25 (2), 71-3-8 (1), 71-6-4 (1), 71-6-17 (1), 71-6-22 (1), 71-9-4 (1). USNM: 201187 (1), 208109 (1). WHOI: AB 13-23 (1), AB 13-28 (2), AB 13-30 (2). ZMUC: D 3588 II (1). Additional larvae and juvenile material from the Pacific Ocean. ZMUC: D 3570 IV (1), D 3585 III (1). Coccorrella atrata (Alcock 1893), Figure 22 Odontostomus atratus Alcock, 1893, p. 182 (original description, type from Bay of Bengal); Alcock, 1896, p. 333 (name only, after Alcock, 1893); Alcock, 1899, p. 167 (description of specimens from Bay of Bengal and Andaman Sea); Alcock & McArdle, 1900, plate 33, fig. 3 (illustration of specimen from Bay of Bengal); Alcock & MacGilchrist, 1905, plate 37, fig. 2 (illustration of specimen from Andaman Sea); Garman, 1899, p. 402 (name only, after Alcock, 1893); Schmidt, 1918, p. 33 (name only, after Alcock, 1893). Coccorella atrata Wassersug & Johnson 1976, p. 273 (description of gut morphology, records from Indian and Pacific oceans); Herring, 1977, p. 297 (first report of luminous tissue in C. atrata, records from Banda and Halmahera seas). JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 121 Evermannella atrata Brauer 1906, p. 136 (record from southern Indian Ocean); Brauer, 1908, p. 192 (description of morphology of eye); Fowler, 1901, p. 212 (synonymy); Marshall, 1954, p. 142 (illustration of specimen from Andaman Sea, after Alcock & MacGilchrist, 1905); Marshall, 1955, p. 323 (swimbladder absent). Ei'ermannella atrata atrata Parr 1928, p. 163 (name only, after Alcock, 1893). Odontostomops braueri Rofen 1963, p. 2 (original description, based on Brauer's [1906, p. 136, plate 10, figs. 3, 4] description of a specimen from 04° 05' 08" S, 73° 24' 08" E); Rofen, 1966d, p. 520 (listed in key to species of Odontostomops). Holotype. — Ca. 89 mm (3.5 inches). RIMS Investigator, Bay of Bengal, 573 fathoms. Deposited in Indian Museum, Calcutta. Diagnosis. — A species of Coccorella with 45 to 47 vertebrae; typically (122 of 123 specimens examined for this character) four pores, 2 + 2, in the frontal canal commissure of the cephalic laterosensory system (fig. 23); and interorbital width = 4.7% to 6.1% SL. Detailed comparisons of the two species of Coccorella are given above, following the description of the genus. Description. — Values for meristic characters are presented in Tables 2 and 3. PROPORTIONAL DIMENSIONS: Based on 24 (38.3-104.6 mm SL) specimens from throughout the range of the species. Expressed as thousandths of the SL and given as the mean and range (values in parentheses). Body: depth at dorsal origin, 186 (171-210). Caudal peduncle: least depth, 85 (76-99); length, 96 (85-108). Adipose fin: distance to midcaudal rays, 134 (122- 151); distance to dorsal-fin base, 331 (309-358). Anal fin: length of base, 268 (247-313). Dorsal fin: length of base, 122 (107-131); dorsal-fin origin to anal-fin origin (distance between verticals), 268 (244-290); end of dorsal-fin base to bases of midcaudal rays, 513 (480-551). Pelvic-fin insertion to anal-fin origin: 218 (190-260). Pectoral-fin insertion to pelvic-fin insertion: 187 (151-230). Anus to anal-fin origin: 57 (41-79). Distance from snout to: anus, 612 (556-655); dorsal- fin origin, 418 (392-445); adipose-fin, 827 (805-862); anal-fin origin, 661 (596- 694); pectoral-fin insertion, 275 (240-309); pelvic-fin insertion, 450 (403-493); anterior margin of eye (=snout length), 62 (56-70). Head length: 242 (225-261). Postorbital head length: 166 (147-183). Eye: horizontal diameter, 55 (47-65); vertical diameter, 61 (54-70). Upper jaw length, 176 (164-193). Lower jaw length: 173 (163-185). Longest dentary tooth: 57 (50-63). Longest palatine tooth: 90 (80-100). Interorbital width (based on 44 [36.2-104.6] specimens): 54 (47-61). BODY: Body moderately elongate, largest known specimen 104.6 mm SL (In- dian Ocean, ZMUC D 3951 1). Body deep, body depth at dorsal origin 17.1% to 21.0% SL, the deepest of any evermannellid species. Anus at or slightly pos- terior to a point midway between pelvic-fin insertion and anal-fin origin. Lateral line extending to a point over anterior one-third of anal-fin base and composed of 30 or fewer segments. About eight pairs of sensory (presumably) papillae along dorsal margin of body, lateral (left and right) to middorsal contour be- tween occiput and dorsal-fin origin. These palps are often difficult to see or missing, presumably due to damage during capture. CEPHALIC LATEROSENSORY PORES: Snout-pad pore formula: 2 + 4 + 2 + 1+2. Mandibular pore formula: 5 + 6 + 2 + 1. Preopercular pores: 3 + 3. Temporal pores: PPTO = 1, PESC = 4. Frontal pores: PPF = 1; frontal canal commissure typically 2 + 2 (in 122 of 123 specimens examined for this character, a discussion of the single specimen assigned to C. atrata and not agreeing in this character is given above). Infraorbital pores: 13 or 14 + 9 or 10. Numerous 122 FIELDIANA: ZOOLOGY sensory (presumably) papillae distributed over occiput, interorbital region, snout, cheeks, and anterior lower jaw. MOUTH: Upper jaw extending to or nearly to anterior margin of preopercle, well past a vertical through posterior margin of eye. Lower jaw projecting an- teriorly very slightly beyond snout. Tooth counts given below are based on 10 (46.3-104.6 mm SL) specimens. Premaxillary teeth small, retrorse, uniserial, numbering 25 to 37 in the 11 speci- mens counted. Dentary with two smaller fangs anteriorly near symphysis followed by nine to 16 unbarbed fangs arranged uniserially. Largest fangs posi- tioned anteriorly. Vomer with one small tooth per side, but in numerous speci- mens only a single, laterally positioned tooth is present, its counterpart probably being lost. Anteriormost palatine tooth an enormous, unbarbed, saber-like fang, easily the largest tooth on each side, noticeably exceeding the snout length in length. Anteriormost palatine fang in C. atrata, 8.0% to 10.0% SL, on a propor- tional basis the largest fang occurring in any evermannellid species. Anterior- most palatine fang separated from posterior palatine teeth by a distinct gap. Posterior palatine teeth numbering four to eight in the 11 specimens counted. Number of premaxillary and dentary teeth higher in larger specimens. COLOR: Color in alcohol a dark brown over head and body, without any distinct concentration into stripes, bars, or markings. In many (but not all) larger adult specimens median and paired fins with numerous moderately sized, dis- cretely separated melanophores resulting in a spotted appearance. This pattern is shown quite clearly in Brauer's (1906, plate 10, fig. 3) illustration of C. atrata. In well-preserved specimens a brassy green, iridescent layer along flanks, beneath eyes, and on cheeks. In fully metamorphosed specimens less than 70 mm SL a distinct and typically elliptical lens pad is present. In larger specimens the lens pad is indistinct, but an opaque (presumably transparent in life) glistening tissue centered beneath the lens is probably similar in function. Midventral areas of body anterior to anal-fin origin with especially dense pigmentation except for areas of luminous tissue (see above). Peritoneum dense black. Distribution. — Coccorella atrata is limited to equatorial areas of the Indian and Pacific oceans (fig. 24). In the Indian Ocean C. atrata is limited to the area of Indian Ocean Equatorial Water. In the Pacific Ocean C. atrata occurs throughout the semi-isolated seas of the Indo-Malayan Archipelago but in the central Pacific is limited to a relatively narrow band along the equator. Only one specimen (NMFS(LJ) J 20.145, 20° N, 145° W, 1 [93.6]) has been taken well poleward from the equator in the central Pacific. This specimen is in all respects a typical exam- ple of C. atrata, and I have no explanation for this apparently anomalous record. Although the range of C. atrata extends well into the eastern tropical Pacific (easternmost Pacific records: NMFS(LJ) J 65-76, 11° 27' S, 97° 59' W, 1 [97.0]; SIO 52-338, 00° 17.7 to 42.0' N, 110° 26.0 to 12.1' W, 1 [40.5]), it does not extend to the American mainland nor does C. atrata occur in the main areas of the oxygen- minimum layer in the eastern Pacific. The distributions of C. atlantica and C. atrata are compared above, and the distribution of C. atrata is compared with other equatorial midwater species in a subsequent section of this paper. Larvae and small juveniles (to 30 mm SL) have most commonly been taken in hauls to between 100 and 300 m. Larvae, small juveniles, and adults have been taken in the upper 80 m, but I have found no records for this species from hauls limited to the upper 50 m. Most large adults (greater than 50 mm SL) were taken JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 123 in hauls to depths exceeding 300 m, but both juveniles and adults have been taken on numerous occasions in hauls to between 100 and 300 m. Larvae and small juveniles have been taken throughout the year. Material Examined. — A total of 149 (10.5-104.6 mm SL) specimens from 91 collections. INDIAN OCEAN. A total of 71 (16.0-104.6 mm SL) specimens from 42 collections. IOAN: B-6 (3), B-7 (5), V 4631 (1), V 4953 (1), V 4957 (1), V 5207 (1), V 5255 (1), V 5277 (1), V 5278 (1). NIO: DY 5353 (1), DY 5359 (2), DY 5395 (2), DY 5399 (5), DY 5413 (4), DY 5420 (2). MCZ: AB 6-332B (1), AB 6-333B (1). SOSC: TV Cr 5, stn. 193 (1). SIO: 71-71 (1). ZMUC: D 3815 VI (1), D 3817 III (2), D 3827 1 (1), D 3856 1 (1), D 3902 II (2), D 3904 III (4), D 3907 II (1), D 3908 I (1), D 3908 II (1), D 3908 III (1), D 3909 III (1), D 3912 1 (1), D 3916 1 (1), D 3925 1 (1), D 3944 I (1), D 3951 I (1). Additional larvae and juvenile material from the Indian Ocean. ZMUC: D 3821 III (3), D 3831 I (1), D 3903 III (3), D 3906 IV (1), D 3907 III (3), D 3910 II (2), D 3914 III (2). PACIFIC OCEAN. A total of 78 (10.5-100.9 mm SL) specimens from 49 collections. AMS: Alpha Helix 1974 stn. 24 (JP75-24) (1). IOAN: V 5117 (3), V 5139 (1), V 6429 (8). NMFS (LJ): J 20-145 (1), J 57-112 (1), J 65-76 (1), J 77-28 (1), J 77-115 (1). ORSTOM: CA II-103 (1), CA III-135 (1), CA III-145 (1), CY IV-10 (1), CY V-6 (1). SIO: 52-338 (1), 60-219 (1), 60-225 (1), 61-540 (1), 61-584 (1), 61-588 (1), 68-533 (2), 68-534 (1), 68-535 (1), 69-19 (1), 70-346 (2), 73-104 (1), 73-120 (1), 73-108 (4), 73-169 (1). UH: TC 47-57 (1), TC 47-58 (10), TC 47-60 (1), TC 47-61 (1), TC 47-68 (3), TC 47-69 (3). USNM: ELT 31-11A (RHG 67-43) (1). ZMUC: D 3676 VI (1), D 3683 I (1), D 3683 VII (2), D 3714 I (1), D 3738 I (1), D 3788 I (1), D 3800 I (1). Additional larvae and juvenile material from the Pacific Ocean. ZMUC: D 3676 VIII (1), D 3753 II (2), D 3755 II (1), D 3782 III (1), D 3789 VIII (1), D 3800 IV (1). Evermannella Fowler 1901 Evermannella Fowler 1901, p. 211 (original description; replacement name for Odonto- stomus Cocco 1838, preoccupied by Odontostomus Beck 1837 (Mollusca); type-species by original designation Odontostomus hyalinus Cocco 1838, a junior synonym of Scopelus balbo Risso 1820). Odontostomus Cocco 1838, p. 192 (original description; type-species by original designa- tion Odontostomus hyalinus Cocco 1838). Type Species. — Ei'ermannella balbo (Risso 1820). Diagnosis. — Evermannellids with tubular eyes directed dorsad and slightly anteriad. Horizontal eye diameter much broader than interorbital width, ratio of horizontal eye diameter to interorbital width exceeding 3.00. Aperture in adipose eyelid distinctly broader than lens of eye. A distinct, elliptical lens pad present in all post-metamorphic specimens. At least some dentary and palatine teeth barbed. Anteriormost palatine tooth a large, barbed fang, 4.6% to 7.3% SL. Dentary teeth biserial. A large vertically elongate fossa centered on dentary symphysis. Body relatively shallow to extremely deep, body depth at anal-fin origin 13.6% to 20.0% SL. No pyloric caecum. Ethmoid cartilage not expanded posteriorly into orbit, not forming an orbital septum. Basisphenoid present. Luminous tissue may be present. Larvae with three peritoneal pigment sections. Description. — Dorsal-fin rays 10 to 13. Anal-fin rays 27 to 37. Pectoral-fin rays 11 to 12. Vertebrae 47 to 54. Body short to relatively elongate, relatively shallow to quite deep, strongly compressed. Anus slightly anterior to, at, or slightly posterior to a vertical mid- way between pelvic-fin insertion and anal-fin origin. Lateral line either lacking (in available material) or relatively short, not extending posterior to a vertical just behind pelvic-fin base and composed of 18 or fewer segments. 124 FIELDIANA: ZOOLOGY Head moderately large, head depth and width subequal to body depth and width. Snout relatively low, rounded, or moderately truncate. Eyes large to extremely large, distinctly tubular, typically directed somewhat dorsoanteriad. Horizontal diameter of eye 5.2% to 9.3% SL, vertical diameter 5.9% to 11.0% SL. Fleshy eye diameter varying from being subequal to considerably exceeding snout length. Diameter of aperture in adipose eyelid considerably exceeding diameter of lens. Pupil distinctly broader than lens. A roughly elliptical lens pad centered on dorsal margin of lateral face of pigmented eye cup, with major axis normal to visual axis of eye. Dentary symphysis with a well-marked, vertically elongate fossa containing anteriormost mandibular cephalic laterosensory pore on each side and also con- taining two to four vertically oriented rows of laterosensory papillae. Branchio- stegal membranes free from isthmus, united by a small membrane anteriorly, at, or slightly anterior to a vertical through anterior margin of eye. Gill filaments elongate and narrow but not extending beyond posterior and ventral margins of gill covers except in E. ahlstromi, where gill filaments are exceptionally dense and elongate, forming a matlike surface that extends beyond posterior and posteroventral margins of gill covers and that nearly reaches the pectoral-fin insertion. Pseudobranchiae with filaments nearly as long as longest gill filaments. Number of pseudobranch elements: E. ahlstromi (N=5, 38.5 to 67.9) 7 to 10; E. balbo (N=8, 62.5 to 139.1) 10 to 16; E. indica (N=12, 46.0 to 102.0) 9 to 13; E. megalops (N=2, 65.6 to 66.0) 8 to 9. Pseudobranch counts tending to be higher in larger specimens (presumed for E. megalops). Dorsal fin relatively short based, 9.1% to 12.0% SL. Middle of dorsal-fin base distinctly anterior to a vertical at middle of standard length. Pelvic-fin insertion anterior to a vertical through middle of dorsal-fin base but not anterior to a vertical through dorsal-fin origin. Appressed pelvic fins reaching to or slightly past anus in best-preserved specimens but not reaching anal-fin origin. Pectoral fins distinctly exceeding pelvic fins in length. Appressed pectoral fins reaching to or distinctly past a vertical through a point midway between pectoral-fin insertion and pelvic-fin insertion but not reaching pelvic-fin insertion. Anal-fin base relatively short to quite elongate, 25.0% to 34.3% SL. Content. — I recognize four species of Evermannella (fig. 29): E. ahlstromi, re- stricted to the eastern Pacific Ocean; E. balbo, for the most part restricted to relatively cool and productive areas of the Atlantic (including the Mediterranean Sea), Indian, and Pacific oceans; E. indica, nearly circumtropical in distribution; and E. megalops, restricted to central water areas of the South Pacific Ocean. The four species may be distinguished on the basis of characters provided in the Key on page 101 and by additional characters cited below. Evermannella balbo differs from its three congeners in the following characters: number of anal-fin rays 33 to 37 vs. 27 to 32; number of vertebrae 52 to 54 vs. 48 to 52; fourth basibranchial toothplate present vs. absent; snout-pad pore formula = 2+4 + 2 + 1+2 vs. 2 + 4 + 2 + 1+0 (E. ahlstromi, E. indica) or 2 + 4 + + 1 + (E. megalops). Evermannella balbo differs from E. ahlstromi and E. megalops in having a modally higher number of dorsal-fin rays, 12 to 13 (83 of 85 counted with 12 rays) vs. 10 to 12 in E. ahlstromi (one of 78 counted with 12 rays) and 10 to 12 in E. megalops (one of 10 counted with 12 rays). Evermannella balbo differs from E. ahlstromi and E. megalops in typical values for the following morphometric characters: caudal peduncle depth, 6.3% to 8.3% SL (x = 7.1% ± .18) vs. 8.5% Fig. 29. The species of Evermannella. A, Evermannella ahlstromi, holotype, USNM 211302, 63.1 mm SL; B, Evermannella balbo, MNHN 98-1105, 171 mm SL; C, Evermannella indica, FMNH 49864, 77.3 mm SL; D, Evermannella megalops, holotype, SIO 72-305, 65.6 mm SL. (A and D from Johnson & Glodek, 1975, drawings by R. K. Johnson; B and C from Rofen, 1966d, drawings by E. M. Soule.) 125 126 FIELDIANA: ZOOLOGY to 10.5% SL (x = 9.5% ± .40) in E. ahlstromi and 8.2% to 9.5% SL (x = 8.7% ± .64) in E. megalops; distance from adipose-fin base to bases of midcaudal-fin rays, 10.6% to 12.5% SL (x = 11.7% ± .19) vs. 13.8% to 15.5% SL (x = 14.6% ± .39) in E. ahlstromi and 13.9% to 15.7% SL (x = 14.6% ± .93) in E. megalops. Everman- nella balbo differs from E. megalops in typical values for the following morphomet- ric characters: caudal peduncle length, 8.4% to 10.1% SL vs. 10.2% to 12.3% SL; distance from pelvic-fin insertion to anal-fin origin, 14.6% to 19.3% SL (x = 17.4% ± .40) vs. 18.7% to 21.5% SL (x = 20.1 ± 1.53). Evermannella ahlstromi differs from its three congeners in having very elongate gill filaments that project beyond the margins of the gill covers and nearly reach the pectoral-fin insertion. Evermannella ahlstromi may be distinguished from E. indica and E. megalops on the basis of additional characters discussed in detail by Johnson & Glodek (1975, pp. 724, 725). Evermannella indica is the most generalized species of Evermannella in the sense that it lacks the distinctive features uniquely and respectively defining its three congeners. It is thus not possible to provide a monothetic diagnosis of E. indica. Evermannella megalops differs from its three congeners, and indeed from all other evermannellids, in having quite enormously enlarged eyes: horizontal eye diameter 7.4% to 8.5% SL (x = 8.1) vs. 6.7% to 8.1% SL (x - 7.2) in E. ahlstromi, 5.2% to 7.2% SL (x = 6.2) in E. balbo, and 4.9% to 9.3% SL (x = 6.9) in E. indica; vertical eye diameter 8.6% to 11.0% SL (x = 9.5) vs. 6.9% to 8.7% SL (x = 7.9) in E. ahlstromi, 5.9% to 8.1% SL (x = 7.0) in E. balbo, and 6.0% to 9.7% SL (x = 7.9) in E. indica. In E. megalops the interorbital width actually decreases with growth over the known size range (to 66.0 mm, see Johnson & Glodek, 1975, p. 724) and for juvenile and young adult specimens (based on five specimens, 32.5-66.0 mm SL) is equal to or less than 0.55 mm, both proportionately and absolutely the least of any evermannellid species. Evermannella megalops further differs from its three congeners in having a snout-pad pore formula = 2 + 4 + + 1+0. Evermannella megalops may be distinguished from E. ahlstromi and E. indica on the basis of additional characters discussed in detail by Johnson & Glodek (1975, pp. 724-725). Evermannella ahlstromi Johnson & Glodek 1975, Figure 29 Evermannella ahlstromi Johnson & Glodek 1975, pp. 716-721 (original description based on 84 specimens from eastern Pacific Ocean). Holotype.— 63.1 mm SL. USNM 211302. Eastern equatorial Pacific, 00° 41' S, 91° 36 to 38' W, IKMT, 0-390 m, 26 May 1966. Diagnosis. — A species of Evermannella with 10 to 12 dorsal-fin rays (only one specimen of 78 counted with 12 dorsal-fin rays), 29 to 32 anal-fin rays, and 47 to 49 vertebrae. Snout-pad pore formula = 2 + 4 + 2 + 1 + 0. Gill filaments notably elongate, projecting beyond gill covers both posteriorly and ventrally. Detailed comparisons of E. ahlstromi with other species of Evermannella are given above, following the description of the genus. Description. — Values for meristic characters are presented in Tables 2 and 3 and in the original description (Johnson & Glodek, 1975). Although substantial additional material of E. ahlstromi has come to hand, my study of this material has not resulted in data requiring meaningful alteration of the original descrip- tion. Therefore, a full description of E. ahlstromi is not presented here. Only those characters not listed in the original description are given below. JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 127 PROPORTIONAL DIMENSIONS.— Based on 10 specimens (38.2-67.9 mm SL) from throughout the range of the species. Expressed as thousandths of the SL and given as the mean and range (values in parentheses). Interorbital width, 21 (17 to 26). CEPHALIC LATEROSENSORY PORES.— Snout-pad pore formula: 2 + 4 + 2 + 1+0. Mandibular pore formula: 5+4+2 + 1. Preopercular pores: ? 3 + 3, about 6. Temporal pores: PPTO = 1, PESC = 4. Frontal pores: PPF = 1. Frontal canal commissure = 3+3. Infraorbital pores: about 9 + 5. Distribution. — Evermannella ahlstromi is limited to the eastern Pacific Ocean (fig. 30). It is known from the Transition Region off Baja California, the regions of transition between Pacific Equatorial Water and central waters of the North and South Pacific, and from a constricted zone along the equator from 155° W to near the American mainland. The distribution of £. ahlstromi as well as other eastern Pacific endemics is discussed in considerably greater detail in a sub- sequent section of this paper. No specimens of E. ahlstromi have been taken in discrete-depth sampling devices. Larvae and small juveniles (to 30 mm SL) have commonly been taken in the upper 100 m, but most adults have been taken in hauls to depths exceeding 400 m. Larvae and small juveniles have been taken throughout the year, specifi- cally in September, November, January, February, March, and May. Material Examined. — The original description of E. ahlstromi (Johnson & Glodek, 1975) was based on 84 (17.7-70.0 mm SL) specimens from 34 collec- tions. These are not listed here. An additional 37 (20.7-65.1 mm SL) specimens from 19 collections have come to hand. PACIFIC OCEAN. IOAN: V 5090 (2). NMFS (LJ): J 57-123 (1), J 60-42 (1), J 77-38 (3), J 77-86 (1), J 77-92 (1), J 77-109 (2), J 77-138 (1), TC 51-37 (3), TC 51-70 (1), TC 51-78 (1), TC 51-87 (1). SIO: 60-229 (1), 73-20 (2), 73-171 (1). ZMUC: D 3556 II (1), D 3556 IV (2), D 3561 IV (3), D 3561 IX (9). Evermannella balbo (Risso 1820), Figure 29 Scopelus balbo Risso 1820, p. 268 (original description, from Mediterranean Sea). Evermannella balbo, Fowler 1901, p. 211 (synonymy); Rofen 1966d, p. 553 (description, references not given here, records from Atlantic Ocean and Mediterranean Sea); Goodyear, Zahuranec, et al. 1972, p. 149 (records from Mediterranean Sea, discus- sion of vertical distribution and seasonal distribution of young stages); Karrer 1973, p. 149 (record from South Atlantic Ocean); Johnson & Glodek 1975, p. 274 (comparison with Evermannella ahlstromi and E. megalops, record from Mediterranean Sea); Wasser- sug & Johnson 1976, p. 276 (gut morphology, record from North Atlantic Ocean). Evermannella hyalina, Regan 1911, p. 130 (osteology, figure of pectoral girdle). Evermannella sicaria Rofen 1963, p. 1 (original description, eight types from off Bermuda); Johnson & Glodek 1975, p. 729 (list two paratypes from off Bermuda). Evermannella spp., Johnson & Glodek 1975, p. 729 (records from Atlantic and Pacific oceans). Odontostomus balbo, Muller 1844, p. 185 (Scopelus balbo Risso placed in Odontostomus, cf. O. hyalinus Cocco). Odontostoma hyalinus Doderlein 1878-79, p. 54 (name only). Odontostomus hyalinus Cocco 1838, p. 192 (original description from Mediterranean Sea). Holotype.— MNHN nr. B 1034, Paris. Diagnosis. — A species of Evermannella with 12 or 13 dorsal-fin rays (97.6% of 85 specimens counted had 12 dorsal-fin rays), 33 to 37 anal-fin rays, and 52 to 54 vertebrae. Snout-pad pore formula =2 + 4 + 2 + 1+2. Gill filaments not projecting beyond gill covers. Detailed comparisons of E. balbo with other species of Eiiermannella are given above, following the description of the genus. 5 128 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 129 Description. — Values for meristic characters are presented in Tables 2 and 3. PROPORTIONAL DIMENSIONS.— Based on 39 (39.6-149.2 mm SL) speci- mens from throughout the range of the species. Expressed in thousandths of the SL and given as the mean and range (values in parentheses). Body depth at anal origin: 158 (145-181). Caudal peduncle: least depth, 71 (63-83); length, 92 (84-101). Adipose fin: distance to midcaudal rays, 117 (106- 125); distance to dorsal-fin base, 338 (321-363). Anal fin: length of base, 295 (266-333). Dorsal fin: length of base, 103 (93-110); dorsal-fin origin to anal-fin origin (distance between verticals), 199 (170-218); end of dorsal-fin base to base of midcaudal rays, 488 (468-502). Pelvic-fin insertion to anal-fin origin: 174 (146-193). Pectoral-fin insertion to pelvic- fin insertion: 210 (169-252). Anus to anal-fin origin: 82 (59-101). Distance from snout to anus: 550 (519-601); dorsal- fin origin, 438 (406-460); adipose fin, 849 (825-874); anal-fin origin, 629 (585- 657); pectoral-fin insertion, 254 (231-286); pelvic-fin insertion, 457 (431-481); anterior margin of eye (= snout length), 58 (51-65). Head length: 229 (209-250). Postorbital head length: 127 (117-139). Eye: horizontal diameter, 62 (52-72); vertical diameter, 70 (59-81). Upper jaw length: 175 (161-195). Lower jaw length: 174 (163-201). Longest dentary tooth: 54 (46-63). Longest palatine tooth: 61 (54-69). Interorbital width: 13 (9-19). BODY. — Body moderately elongate, largest known specimen 168.5 mm SL (Atlantic Ocean: ISH 1750/68). Body moderately deep, body depth at anal origin 14.5% to 18.1% SL. Anus at or slightly posterior to a point midway between pelvic-fin insertion and anal-fin origin. Lateral line not extending beyond a vertical just posterior to pelvic-fin base and composed of 18 or fewer segments. CEPHALIC LATEROSENSORY PORES.— Snout-pad pore formula: 2 + 4 + 2 + 1+2. Mandibular pore formula: 5 + 6 + 2 + 1. Preopercular pores: not countable in any specimen examined by me. Temporal pores: PPTO = 1, PESC = 4. Frontal pores: PPF = 1; frontal canal commissure = 3 + 3. Interorbital pores: not countable in any specimen examined by me but approximately 14 + 10. MOUTH. — Upper jaw extending to or nearly to anterior margin of preopercle, well past a vertical through posterior margin of eye. Lower jaw projecting an- teriorly very slightly beyond snout. Premaxillary teeth small, retrorse, uniserial, numbering 33 to 62 in seven (62.5-139.1 mm SL) specimens counted. Dentary with two smaller fangs an- teriorly near symphysis, followed by a row of five to nine large barbed fangs, with these bordered anterolateral^ by a row of two to four smaller teeth. Den- tary tooth counts based on eight (62.5-139.1 mm SL) specimens. Large, dentary fangs the longest anteriorly and decreasing in length posteriorly. Vomer proba- bly with one small tooth on each side, but in numerous specimens only a single, laterally postitioned tooth is present, its counterpart either failing to develop or lost. Anteriormost palatine tooth an enormous, barbed fang, easily the largest tooth on each side. Each such fang with a peculiar elbow-shaped bend near distal terminus, at which point tooth angles forward and ventrally and ends in a triangular point. Palatine teeth numbering six to nine. Palatine tooth counts based on eight (62.5-139.1 mm SL) specimens. Number of premaxillary and palatine teeth higher in larger specimens. Teeth lacking on basihyal and over first three basibranchials, but a fourth basibranchial toothplate, bearing a patch of six to nine small, conical teeth, overlies the third (cartilaginous) copula of the basibranchial series (fig. 14c). 130 FIELDIANA: ZOOLOGY COLOR. — Color in alcohol typically a light brown, with numerous, variably sized melanophores distributed over body and head. Melanophores on body arranged in seven to 13 irregular rows, with arrangement in recognizable rows most prominent on body posterior to dorsal-fin base. Variation in pigmentation ranges from nearly unpigmented specimens, with no melanophores or very few and extremely punctate melanophores visible, to highly melanistic specimens essentially covered with brownish black pigment. Individuals at either extreme of pigment development apparently occur throughout the range of E. balbo. Head lightly pigmented, with pigment concentrations on occiput, cheek, gill covers, snout, and anterior lower jaw. All fins with pigment present at fin-ray bases and finely scattered on rays and membranes. Midventral region between pectoral- and pelvic-fin bases with especially dense pigmentation in most specimens. Peritoneum black. Discussion. — Rofen (1963, 1966d) based his description of Evermannella sicaria on eight (19.6-32.7 mm SL) specimens, all from the vicinity of Bermuda. Rofen's description leaves no doubt that the eight specimens in question belong to Evermannella, and the very high anal-fin ray counts (35 to 36) limit the need for comparison to E. balbo. The key characters supposedly distinguishing E. sicaria from E. balbo were said to be as follows (Rofen 1966d, pp. 537, 538): "lens very large, directed dorsally, appreciably larger than width of interorbital; no lon- gitudinal rows of large dark spots on sides of head and body," (E. balbo) vs. "lens moderate in size, directed dorsolaterally, smaller than width of interorbital; 4-7 rows of large dark spots on sides of head and body," (E. sicaria). Additional characters said to separate the two species included (Rofen, 1966d, p. 554): eye diameter into head length, 3.7 to 4.5 (E. balbo) vs. 5.3 to 5.8 (E. sicaria); lens diameter into eye diameter, 1.4 to 1.5 (E. balbo) vs. 1.6 to 1.9 (E. sicaria); interor- bital width into head length, 11.8 to 18.4 (E. balbo) vs. 5.7 to 9.1 (E. sicaria); peritoneum black (E. balbo) vs. light (E. sicaria). All but one of these supposed differences are in fact ontogenetic in origin. The states supposedly defining E. sicaria, viz. smaller eye diameter, smaller lens diameter, relatively wider interor- bital diameter, are typical of smaller juveniles not only of E. balbo but of all species of Evermannella. The states supposedly defining E. balbo are typical of larger juveniles not only of E. balbo but of all species of Evermannella. Rofen's (1966d) study material of E. sicaria consisted of eight small juveniles (19.6-32.7 mm SL) with values reported for three (28.9, 30.1, and 32.7 mm SL) specimens. Rofen's study material of E. balbo consisted of four (21.4, 28.6, 37.1, and 171 mm SL) specimens with values reported only for the latter three. The one character purportedly separating the two forms that is not strictly explainable in terms of ontogenetic variation is the presence or absence of rows of melanophores along the sides of the head and body. The presence or absence of such melanophores and particularly the tendency for melanophores on the body to be arranged in rows is variable in juvenile and adult specimens of E. balbo, as discussed above. I have examined 837 specimens of E. balbo from 178 collections, including six paratypes of E. sicaria. I can see no basis for recognizing two species on the basis of characters used by Rofen or indeed on the basis of any other characters, and I conclude that E. sicaria and E. balbo are conspecific. It might be mentioned at this point that the key to known western North Atlantic postlarval evermannellids provided by Rofen (1966d, p. 515) suffers from much the same problem as Rofen's description of E. sicaria as distinct from JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 131 E. balbo, i.e., most of the supposed differences are in fact ontogenetically vari- able. Thus this key, at least in part, is more useful in separating growth stages than in separating species. Geographic Variation. — Schmidt (1918) recorded 72 specimens of E. balbo taken by Danish collecting efforts in the North Atlantic (33 specimens) and Mediterra- nean Sea (39 specimens). Schmidt found no discernible geographic variation in counts of dorsal-fin rays or anal-fin rays but reports (1918, p. 31) apparent geographic variation in numbers of vertebrae. Schmidt found that specimens from the Mediterranean tended to have more vertebrae (x = 52.4, range = 52 to 53, based on five specimens) than specimens from the North Atlantic (x = 51.7, range = 50 to 53, based on nine specimens). Examples of species exhibiting meristic (and morphometric) differences between North Atlantic and Mediterra- nean "populations" are cited by Nafpaktitis (1975, p. 83). I have examined values of meristic characters for material of E. balbo from throughout the range of the species. For purposes of comparison I have (somewhat arbitrarily) divided the range of the species (fig. 32) into five subareas: CNA, includes all North Atlantic specimens taken north of 40° N and also includes all North Atlantic specimens taken north of 30° N and west of 25° W; MED, includes all specimens taken in the Mediterranean Sea; ENA, includes all North Atlantic specimens taken south of 30° N, all North Atlantic specimens taken south of 40° N and east of 25° W, and all South Atlantic specimens taken north of 05° S; CSA, includes all South Atlantic specimens taken south of 05° S; SP, includes all specimens taken in the South Pacific Ocean. I find geographic variation in values for two meristic characters: number of anal-fin rays and number of vertebrae (table 16). My results for numbers of vertebrae disagree with the results of Schmidt (1918) in two respects: (1) Schmidt found a range in vertebral counts of 50 to 53 for specimens of E. balbo, I find a range of 52 to 54; (2) I find no evidence that specimens from the Mediterranean tend to exhibit higher vertebral counts (table 16). The apparent difference between CSA specimens (six of six specimens with 54 vertebrae) and specimens from other areas (49 of 49 specimens with 52 or 53 Table 16. Geographic variation in certain meristic characters in Evermannella balbo. A. No. of anal-fin rays Geographic area 33 34 35 36 37 N x ± 95% limits CNA — 13 28 12 — 53 35.0±0.19 MED 4 9 2 1 — 16 34.0±0.43 ENA 3 17 3 — — 23 34.0±0.23 CSA — 4 10 15 2 31 35.5±0.30 SP 3 21 10 — — 34 34.2±0.21 B. No. of vertebrae Geographic area 52 53 54 N x ± 95% limits CNA, ENA 11 18 — 29 52.6±0.19 MED 4 2 — 6 52.3±0.54 CSA — — 6 6 54.0 SP 6 8 — 14 52.6±0.30 KEY to geographic areas (defined in the text): CNA=central North Atlantic, MED=Mediterranean Sea, ENA=eastern North Atlantic and equatorial Atlantic, CSA=central South Atlantic, SP=South Pacific. 132 FIELDIANA: ZOOLOGY vertebrae) needs confirmation on the basis of additional material. Only three specimens from the ENA area were suitable for radiography — the counts for these three (52, 52, 53) suggested no difference between CNA and ENA speci- mens, but this too needs to be tested on the basis of additional material. The most interesting geographic variation in meristic character values seen in my study of E. balbo was in numbers of anal-fin rays (table 16). Values for specimens from two of the areas (CNA, CSA) are significantly higher than values for specimens from each of the three remaining areas (MED, ENA, SP). Among the environmental features known to affect meristic characters in fishes (see Barlow, 1961; Fowler, 1970), the most obvious feature to invoke in attempt- ing to explain this pattern would be temperature were it not for values for specimens from the South Pacific, a cold water area (Sverdrup et al., 1942, chart III). Johnson & Barnett (1975) report an inverse correlation between central values of meristic characters and three measures of biological productivity for five species of midwater fishes. It seems certain that average productivity condi- tions throughout most of the area defined by the records of E. balbo categorized as CNA or CSA are lower than average productivity conditions for the other areas (especially the areas ENA and SP, see Cushing, 1971). Thus, it might be suggested that the pattern of meristic variation exhibited by E. balbo parallels that for the species discussed by Johnson & Barnett (1975). This certainly needs confirmation on the basis of study of additional specimens. I know of no other character in E. balbo exhibiting discernible geographic variation. Distribution. — Evermannella balbo is known from the Atlantic, Indian, and Pacific oceans (figs. 31, 32), but only one specimen is known from the Indian Ocean (SAM 23612, 29° 52' S, 31° 36' E; IKMT: 0-500 m; 25-26 February 1963; 1 [57.6]). Evermannella balbo is the only evermannellid known to occur in the Mediterranean Sea. It occurs throughout the Mediterranean Sea (fig. 32) and has been recorded from the Aegean Sea (Schmidt, 1918, p. 35). Evermannella balbo occurs throughout the North Atlantic from 59° 49.6' N (at 20° 22.9' W, DY 7709, 1 [109.0]) to about 30° N. South of 30° N, E. balbo is limited to the eastern North Atlantic. Evermannella balbo occurs in the equatorial Atlantic and Gulf of Guinea. Evermannella balbo is known from throughout the South Atlantic between (roughly) 25° S and 40° S. Evermannella balbo is unknown from the South Atlantic between roughly 05° S and 25° S. The distribution of South Atlantic capture records for evermannellid species (fig. 31) suggests that E. balbo may well be absent from the western and central South Atlantic between these latitudes, but the absence of E. balbo from the Benguela Current area of the eastern South Atlantic is, I believe, questionable, and can only be resolved with additional sampling effort and study. In the central and eastern South Pacific, the distribu- tion of E. balbo agrees very well with the distribution of the Transition Zone fauna discussed by McGowan (1971). Schmidt (1918) discusses the North Atlantic distribution of E. balbo based on material obtained by Danish collecting efforts. Schmidt (1918, p. 35) discusses the apparent "replacement" of E. balbo by other evermannellid species in the western subtropical North Atlantic. My data (fig. 31) confirm the basic pattern discussed by Schmidt except that north of 30° N, E. balbo occurs throughout the North Atlantic (to roughly 60° N). A chart (fig. 31) comparing the distribution of E. balbo with all other evermannellid species suggests that throughout a large part of its range, E. balbo co-occurs with no other evermannellid species. 133 134 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 135 Schmidt (1918) notes two points concerning the capture of postlarval E. balbo represented in his material: (1) a great majority of the smallest postlarval speci- mens were taken with only 25 mwo, i.e., very near the surface, whereas the larger postlarvae were taken at greater depths, (2) young stages of E. balbo were taken (in the Mediterranean) almost exclusively in July, August, and September. Schmidt tentatively concludes that £. balbo is a summer-spawning species in the Mediterranean. He (Schmidt, 1918, p. 36) contrasts his results for E. balbo with the results for two other species occurring in the Mediterranean — Argentina sphyraena Linnaeus and Nansenia oblita (Facciola) — both of which he found to be winter-spawners in the Mediterranean. Schmidt's data suggest that £. balbo is a more southerly form in the North Atlantic than either A. sphyraena or N. oblita, and he thus explains the co-occurrence of these three species in the Mediterra- nean on the basis of timing of spawning to seasonal changes in surface tempera- tures in the Mediterranean. My data tend to blur this neat picture — E. balbo occurs much farther north in the North Atlantic than Schmidt had supposed (to 59° 49.6' N, fig. 32), and in the North Atlantic larvae and small juveniles of £. balbo have been taken in every month except December, January, and February. Larvae and small juveniles of £. balbo have been taken in January and February in the South Pacific and South Atlantic, respectively. In line with Schmidt's suggestion, however, it may be of interest to note that whereas E. balbo is known as far north as 59° 49.6' N, the most northerly records (north of 45° N) are only of large (greater than 87 mm SL) adults. The most northerly record for a young specimen is from 44° 07 to 13' N, 44° 09 to 11' W (RHB 1024, 1 (20.8), 10 Sep- tember 1964). If it is indeed true that adults occur far to the north of the northerly limit for occurrences of larval and postlarval stages (and thus presumably far to the north of the northern limit of spawning), this differential distribution of young and adult stages would be unique to £. balbo among evermannellids and scopelarchids (with the possible exception of the Antarctic scopelarchid species Benthalbella elongata and B. macropinna, see Johnson, 1974c, p. 228). Additional data will be required to test this suggestion. Goodyear, Zahuranec, et al. (1972) discuss the capture of six specimens taken with discrete-depth sampling devices in the Mediterranean in 1970. Specimens of E. balbo were taken at four of the five localities sampled, excluding only the westernmost sampling area (Goodyear, Gibbs, et al., 1972). The depth records for the six specimens, arranged by size, follow: 15 mm (200 m, night), 21 mm (600 m, day), 24 mm (700 to 500 m, day), 32 mm (200 m, day), 44 mm (325 to 250 m, night), 87 mm (200 m, night). The authors conclude "these catches suggest, if anything, that this species may occupy a wide vertical range during the daytime, and, perhaps, also at night." Larvae and small juveniles (to 30 mm SL) of £. balbo have commonly been taken in the upper 100 m and, in numerous cases, in the upper 50 m. Most adults have been taken in hauls to depths greater than 400 m, but adults have been taken on a number of occasions in hauls to between 100 and 300 m. Most of the shallower records (less than 300 m) for larger adults (greater than 55 mm SL) are from the eastern South Pacific. Material Examined. — A total of 834 (10.6-168.5 mm SL) specimens from 177 collections. ATLANTIC OCEAN: A total of 796 (10.6-168.5 mm SL) specimens from 152 collections. FMNH: 49850 (2) paratypes of E. sicaria Rofen, 63114 (2). ISH: 371/66 (1), 914/66 (1), 1628/68 136 FIELDIANA: ZOOLOGY (1), 1750/68 (1), 627/71 (2), 665/71 (1), 801/71 (2), 846/71 (1), 918/71 (1), 947/71 (1), 1135/71 (1), 1520/71 (1), 2220/71 (1), 2285/71 (1), 2442/71 (1), 454/73 (5), 485/73 (2), 546/73 (32), 699/73 (2). MCZ: 42380 (2). NIO: DY 5797 (3), DY 5798 (1), DY 5801 (1), DY 7036 (1), DY 7709 (1), DY 7824 (1). UMML: 19986 (1). USNM: 40054 (1). USNM, ACRE: 3-4 (2), 3-5 (2), 3-6 (2), 6-16 B (1), 7-19 (1), 10-36 A (1), 10-38 N (2), 12-6 A (1), 12-15 B (1), 12-53 (1), 12-71 (1), 12-83 (3). USNM, MED: 1-14 M (2), 2-2 P (3), 2-5 B (1), 2-18 Z (1), 2-19 A (1), 3-4 M (1), 3-9 P (2), 3-18 P (2), 4-18 (1), 5-2 B (1), 5-2 M (2). WHOI, RHB: 473 (1), 474 (1), 482 (2), 486 (1), 1018 (2), 1024 (1), 1028 (2), 1041 (1), 1045 (5), 1046 (4), 1438 (3), 1503 (1), 1940 (2), 2006 (1), 2059 (1), 2066 (1), 2070 (1), 2217 (3), 2234 (1), 2263 (1), 2406 (1), 2407 (1), 2416 (1), 2417 (1), 2418 (1), 2621 (1). ZMUC: Uncat, no data, from Mediterranean Sea (1), 1107 I (3), 1163 III (3), 1342 HI (1). Additional larvae and juvenile material from the Atlantic Ocean. CAS, SU: 57689 (1) paratype of E. sicaria ( = P), 57690 (1) P, 57691 (1) P, 57692 (1), 57693 (1) P, 57718 (1), 57720 (1). USNM, ACRE: 3-2 (1), 3-8 (1), 10-4 B (1). ZMUC: Texas 528 (5), Dana 830 (2), D 857 (1), D 1107 1 (2), D 1107 V (4), 1107 VI (2), D 1107 VII (1), D 1107 XI (2), D 1108 1 (1), D 1120 II (1), D 1122 I (40), D 1122 II (31), D 1122 III (21), D 1122 IV (7), D 1122 V (22), D 1122 VI (2), D 1122 VII (58), D 1123 1 (4), D 1123 II (51), D 1123 III (34), D 1123 IV (13), D 1123 V (2), D 1124 I (9), D 1124 II (48), D 1124 III (44), D 1124 IV (46), D 1126 III (1), D 1130 III (4), D 1141 1 (1), D 1141 VI (1), D 1141 XI (1), D 1141 XVI (8), D 1141 XVII (6), D 1229 1 (18), D 1363 II (1), D 3523 III (1), D 3527 1 (1), D 3528 II (1), D 3533 I (1), D 3533 II (1), D 3533 III (25), D 3534 I (6), D 3535 I (1), D 3535 III (26), D 3536 I (16), D 3979 II (4), D 3979 III (1), D 4008 I (1), D 4009 I (1), D 4009 III (13), D 4009 VIII (1), D 4017 VIII (1), D 4066 III (14), D 4066 IV (6), D 4070 VII + XI (2), D 4070 VIII + XII (2), D 4070 X + XIV (8), D 4070 XVI (1), D 4070 XX (2). INDIAN OCEAN: A total of one (57.6 mm SL) specimen from one collection. SAM: 23612 (1). PACIFIC OCEAN: A total of 37 (26.4-115.0 mm SL) specimens from 24 collections. IOAN: AK 229 (1). LACM: 11078 (1), 11243 (1), 11250 (1), 11284 (1), 11297 (1). SIO: 65-665 (1), 65-667 (1). SOSC: ELT 25-303 (1), ELT 25-338 (1). USNM: 208085 (1). WHOI: AB 13-2 (1), AB 13-3 (1), AB 13-4 (2), AB 13-7 (1), AB 13-17 (1), AB 13-19 (6), AB 13-23 (1), AB 13-24 (1), AB 13-30 (2), AB 13-48 (1), AB 13-50 (7), AB 13-51 (1), AB 13-52 (1). Evermannella indica Brauer 1906, Figure 29 Evermannella indica Brauer 1906, p. 135 (original description, three syntypes from Indian Ocean); Rofen 1966d, p. 544 (description, references not given here, records from Atlantic Ocean and Banda Sea); Johnson 1974c, p. 30 (osteology, records from Atlantic and Pacific oceans); Johnson & Glodek 1975; p. 724 (comparison with Evermannella ahlstromi and E. megalops, records from Atlantic, Indian, and Pacific oceans); Wasser- sug & Johnson 1976, p. 276 (gut morphology, record from Pacific Ocean). Herring 1977, p. 306 (name only). Evermannella indica indica, Parr 1928, p. 164 (name only, in key to species of Evermannella). Evermannella indica melanoderma Parr 1928, p. 170 (original description, two syntypes from off Bermuda). Evermannella borodini Whitley 1958, p. 32 (original description, new name proposed for Odontostomus balbo atlanticus Borodin 1931, preoccupied by Evermannella atrata atlantica Parr 1928). Evermannella melanoderma, Beebe, 1937, p. 205 (name after Parr 1928, records from off Bermuda). Odontostomus balbo atlanticus Borodin 1931, p. 78 (original description, holotype from off Bermuda). Syntypes. — A total of three, the largest 32.5 mm (Brauer 1906, p. 136). VAL- DIVIA station 182, Indian Ocean, 10° 08.0' S, 97° 14.2' E; VALDIVIA station 231, Indian Ocean, 03° 24.1' S, 58° 38.0' E; VALDIVIA station 239, Indian Ocean, 05° 42.1' S, 43° 36.1' E. Syntypes presumably deposited in Museum fur Natur- kunde, East Berlin, D. D. R. Diagnosis. — A species of Evermannella with 12 or 13 dorsal-fin rays (97.2% of 215 specimens counted had 12 dorsal-fin rays), 27 to 31 anal-fin rays, and 48 to JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 137 52 vertebrae. Snout-pad pore formula =2 + 4 + 2 + 1+0. Gill filaments not projecting beyond gill covers. Detailed comparisons of £. indica with other species of Ez>ermannella are given above, following the description of the genus. Description. — Values for meristic characters are presented in Tables 2 and 3. PROPORTIONAL DIMENSIONS.— Based on 82 (30.0-119.0 mm SL) speci- mens from throughout the range of the species. Expressed as thousandths of the SL and given as the mean and range (values in parentheses). Body: depth at anal-fin origin, 155 (136-173). Caudal peduncle: least depth, 83 (69-97); length, 104 (87-121). Adipose fin: distance to midcaudal rays, 138 (109-162); distance to dorsal-fin base, 324 (291-387). Anal fin: length of base, 282 (250-309). Dorsal fin: length of base 109 (97-120); dorsal-fin origin to anal-fin origin (distance between verticals), 226 (201-263); end of dorsal-fin base to base of midcaudal rays, 493 (451-536). Pelvic-fin insertion to anal-fin origin: 192 (151-224). Pectoral-fin insertion to pelvic-fin insertion: 207 (162-263). Anus to anal-fin origin: 88 (42-136). Distance from snout to: anus, 558 (469-607); dorsal- fin origin, 421 (383-464); adipose fin, 831 (800-880); anal-fin origin, 641 (601- 672); pectoral-fin insertion, 269 (233-317); pelvic-fin insertion, 455 (408-511); anterior margin of eye (= snout length), 70 (57-85). Head length: 249 (216-265). Postorbital head length: 118 (91-140). Eye: horizontal diameter, 69 (49-93); verti- cal diameter, 79 (60-97). Upper jaw length: 186 (170-209). Lower jaw length: 191 (169-225). Longest dentary tooth: 58 (43-70). Longest palatine tooth: 62 (48-73). Interorbital width: 11 (8-20). BODY. — Body moderately elongate, largest known specimen 127.2 mm SL (Pacific Ocean: SIO 70-314). Body moderately deep, body depth at anal origin 13.6% to 17.3% SL. Anus at or slightly posterior to a point midway between pelvic-fin insertion and anal-fin origin. Lateral line not extending beyond a point above pelvic-fin base and composed of 16 or fewer segments. CEPHALIC LATEROSENSORY PORES.— Snout-pad pore formula: 2 + 4+2 + 1+0. Mandibular pore formula: 5 + 5 + 2 + 1. Preopercular pores: not countable in any specimen examined by me. Temporal pores: PPTO = 1, PESC = 4. Frontal pores: PPF = 1; frontal canal commissure = 3 + 3. Infraorbital pores: about 11 or 12 + 8. MOUTH. — Upper jaw extending to or nearly to anterior margin of preopercle, well past a vertical through posterior margin of eye. Lower jaw projecting an- teriorly very slightly beyond snout. Premaxillary teeth small, retrorse, uniserial, numbering 34 to 52 in 12 (46.0- 102.0 mm SL) specimens counted. Dentary with one or two smaller fangs an- teriorly near symphysis, followed by a row of five to 10 large barbed fangs, with these bordered anterolateral^ by a row of three to four smaller teeth. Dentary tooth counts based on 13 (46.0-102.0 mm SL) specimens. Largest dentary fangs the longest anteriorly and decreasing in length posteriorly. Vomer probably with one small tooth on each side, but in numerous specimens only one laterally positioned tooth is present, its counterpart either failing to develop or lost. Anteriormost palatine tooth an enormous barbed fang, easily the largest tooth on each side. Each such fang with a peculiar elbow-shaped bend near distal terminus, at which point tooth angles forward and downward, ending in a triangular point. Palatine teeth numbering six to 10 in 13 specimens (46.0 to 102.0) counted. Number of premaxillary, dentary, and palatine teeth distinctly higher in larger specimens. Lingual teeth lacking. 138 FIELDIANA: ZOOLOGY COLOR. — Color in alcohol a light brown with numerous, variably sized melanophores irregularly distributed over head and body. A brassy iridescent coloration overlying dermal pigmentation on cheeks and flanks, evident in best-preserved specimens. No marked concentration of pigment into bars, stripes, patches, or the like on either body or head. Evermannella indica exhibits marked individual variation in intensity of pigmentation, with individuals varying from pale-colored specimens with very sparse pigmentation to speci- mens essentially covered with brownish black pigmentation. Individuals at both extremes of color variation occur throughout the range of the species. Head lightly pigmented, with pigment concentrations on occiput, cheek, gill covers, snout, and anterior lower jaw. All fins with pigment present at fin-ray bases and finely scattered on rays and membranes. Peritoneum dense black. Geographic Variation. — In both Evermannella indica and Odontostomops normalops there is significant and interesting geographic variation in number of anal-fin rays. The pattern of variation in both species appears to be similar, and therefore data for both species is presented here. Evermannella indica and O. normalops have the broadest distributions of any evermannellid species. Both species occur in central and equatorial waters of the Atlantic, Indian, and Pacific oceans (figs. 34, 37). Neither species occurs in the Mediterranean Sea, and both are largely or entirely excluded from most of the area of eastern Pacific Equatorial Water. In both species counts of anal-fin rays are lowest in the Atlantic, intermediate in the Indian, and highest in the Pacific Ocean (table 17). In the case of E. indica differences between mean values of anal-fin ray counts for specimens from each of the three oceans are statistically significant (p < .05) for all three possible comparisons. In the case of O. normalops the mean value for Atlantic specimens is significantly lower than those for specimens from the Indian Ocean and Pacific Ocean, but the difference between mean values for Indian and Pacific Ocean specimens is not statistically significant. Values for vertebral counts in O. normalops apparently parallel the results for numbers of anal-fin rays, with lowest values in the Atlantic, highest values in the Pacific, and intermediate values in the Indian Ocean (table 18). This trend is not seen in results for E. indica (table 18), although values for Pacific Ocean specimens are again the highest. No other meristic character in either species was found to exhibit discernible geographic variation. In the case of E. indica enough material was available that I was able to exam- ine geographic variation in anal-fin ray counts on a within-ocean as well as a between-ocean basis. It was necessary to divide the range of E. indica into a number of subareas such that comparisons between areas could be made. I chose nine subareas for detailed study. These subareas are indicated in Figure 33 and were given the following designations: western North Atlantic (WNA), eastern North Atlantic (ENA), equatorial Atlantic (EQA), central South Atlantic (CSA), Indian Ocean (IOC), western equatorial Pacific (WEP), central equatorial Pacific (CEP), central North Pacific (CNP), and central South Pacific (CSP). These areas were chosen to meet the following criteria: (1) enough material of E. indica was available from each subarea to make possible meaningful between-area comparisons of anal-fin ray counts; (2) the subareas chosen taken as a whole covered most of the range of E. indica (compare figs. 33 and 34); (3) boundaries between the subareas corresponded more or less closely with distri- a +i S a +i +i +i Tf (N \D +1 +1 + o t^ o I I I I 1 a I c » I I I — - < On CM NNvO IT) (N ^* I I I I I I I I I c c 45 S =3 'o 5^-0 < £ a. < — cu 139 140 FIELDIANA: ZOOLOGY Table 18. Vertebral counts in Evermannella indica and Odontostomops normalops. Area 48 49 50 51 52 N Mean ± 95% 1 Atlantic — 6 21 4 — 31 49.9±.21 Indian 5 6 3 — — 14 48.9±.45 Pacific 1 12 20 7 0. normalops 40 50.8±.24 Atlantic 1 2 6 — — 9 49.6±.56 Indian — — 4 — — 4 50.0 Pacific — — 4 7 1 12 50.8±.40 butional boundaries for one or more species of midwater fish as reported in the literature; and (4) the subareas chosen represented a broad range of open ocean habitats with respect to physical and biological features and particularly with respect to measures of biological productivity. The fit of the subareas chosen with respect to criterion (3) is variable. The faunal distinctiveness of the central gyral areas in the North and South Pacific is reasonably well documented (e.g., McGowan, 1971; Barnett, 1975; Johnson & Glodek, 1975). Other areas were delineated more arbitrarily, e.g., it seems likely that the Indian Ocean should be divided into at least two subareas, but lack of material of E. indica from the Indian Ocean, particularly from south of 10° S, precluded such a division. Criterion (4) was included because of the findings of Johnson & Barnett (1972, 1975) that at least some midwater species exhibit geographic variation in values for meristic characters that may best be related to variation in measures of biological productivity. I expected that the results for E. indica would parallel the results for Diplophos taenia Guenther, Vinciguerria nimbaria (Jordan & Williams), and the other species studied by Johnson & Barnett (1975). That is, I expected values for anal-fin ray counts in E. indica to be lowest in areas of highest produc- tivity, highest in areas of lowest productivity, and intermediate in areas of in- termediate productivity. This did not prove to be the case. In E. indica anal-fin ray counts (table 19) are lowest in the western North Atlantic, highest in the central North Pacific, and the highly significant tau value ([table 19], tau 8 = +0.893, p < .01, tau is Kendall's rank- correlation coefficient [see Tate & Clelland, 1957]) indicates that mean values for anal-fin ray counts increase sequentially, i.e., clinally, around the world from the western North Atlantic to the central North Pacific. Two additional points should be noted. The result obtained would not be different if the value for specimens from the central South Pacific were used in place of the value for specimens from the central North Pacific (table 19). One might question the order of geographic proximity used in Table 19, viz., 6— WEP, 7— CEP, 8— CNP. At least part of the WEP subarea (viz., the South China Sea) is geographically closer to part of the CNP subarea (viz., the Philip- pine Sea) than to any portion of the CEP subarea (fig. 33). It might therefore be argued that the order of geographic proximity should read: 6 — WEP, 7 — CNP, 8— CEP. However, the results of Johnson & Barnett (1975), McGowan (1971), Brinton (1975), and the distribution of such forms as the two species of Coccorella (see above) suggest that the ordering used (6 — WEP, 7 — CEP, 8 — CNP) is ^g±--£<% 42Zr^>--^» — tH^j^TnI / ^5pr"«~ k — / ^il3p*-&-c / /f"t ** / tr&~~* 1 ' 7 Xs? * vn i kr^==r---l--jr7 — OxStSlxw a 1 * /- OP^ ^ >s! • P~~hHr~--Cl P~i?>CKS!So \ Jflllllk €r J^ \^OT5ri (- \^v^Sk N ^^Cjr~*-J5j *• • \ AXNy^r^CS'-Xji * — ^vrC^K*^ '^~^S^-— i <^\^K^\/x/^3/'° c ^ilis ^<^6r »- >X» < r- b ^ V^ 5 "-^^ /Tl r /^ Y/^/t-T~"~*~ / /fL-. 6 7 ^r-t — *~ "^r^^ T^r^-x>oCSiSS^v^ *A ^A^zr -j -Mi +^-^~* ^ o ~^335kn^ L H3 Jr"" ' * *l~ .»_ ( , — — . j^rS^\ ■s*£e3 v£"~ •-.•■ — <» — gJs- 3 4 » -^9.4^ BISk^St"" ^~ ^ "Bra* ^fcr--^^ — * y ' • — — '"^H^^^ Wp Sfc^-i" ~~S — |fcs it ___^-- ^/Wm / *' — ^^KT^^n^xSi^ / ^mu%^^ -/ — ^^QtCL<^J0^X^/ 8 b^rrr /^y9w/^ ^ \ v|7N^r4-C ^hkl ^%xw J 5^ Q0$7 l - 'n, ^ — *-!< 2a "rt£" j£v* ? >^vi^^i^4^>b^^ = <&! 5 eg &--£ u O « £Z§- 5 c ^ x o> 1" ? gWU ■3 l?U c •« I 2 (A I- £ £5 u C n < ■s u ■o a c c 3 5 u 6 en 6 -5 s -a C/1 1 1 B 3 - c u 9 B 'J II 1 g II C- _> T3 < u u c tn U ■J S3"*- . 1 a V. C B B 3 < Pu fi _, B G TO 2 x 'j n B u. S c c a o 141 £r3^~-f^^£- ■%£ ♦h^SS 6 — r~-r^ir ^K^f — t— ~~~^T~~3~r^A '"* s- l||5*c4v ^"^ SsL " i-^f^ | !> > o Toi-- i_ 7 i K s &s > *&^ryu"/LX--E i»Cj ^^^ 1- j» s 7^N1 stf — f~— -Tp^tr~/ ^tC^C^^v 6 r~r^T 7 t— —4-Jlj" fe JJT O s r~i ii-^ ir XX !£> n r0 ^ ^ A > o fc ° ^ - O ;Q =: 1 ffioAJZiL^j 3 I 142 ^ M O u ■a | OS T3 Nnt m vonoo rn O g in in 1 OS t-h X •a 1 r— 1 oo in in ri en r-i \o en s iH 0) 0* ~" +i +i +i +i +i +i +i +i c S o" fN en o o cm rH in oo <0 3 in 00 00 On ON ON ^ O^ O^ O^ n (N , 2 c Tj< I— < g g j 9 « C < J2 H Cm B I I I I I W5 O V U 143 144 FIELDIANA: ZOOLOGY biologically more meaningful in that interchange between populations in the two equatorial areas is more likely than interchange between populations in either equatorial area with the population(s) in the North Pacific central gyral area. These results for E. indica remain unexplained. I know of no physical, chemi- cal, or biological feature of the open ocean environment to which this sequential around-the-world clinal variation in E. indica might be related. I am unaware of any other midwater species that exhibits variation in meristic counts that paral- lels the variation observed in E. indica (and O. normalops). I believe that addi- tional study of variation in E. indica could provide us with clues to the population structure of wide-ranging midwater species and possibly to the level of inter- change between such populations. The problem that remains is lack of material. Distribution. — Evermannella indica is a nearly circumglobal warm-water species found in central and equatorial waters of all three oceans (fig. 34). With the possible exception of Odontostomops normalops (fig. 37), £. indica has the broadest geographic distribution of any evermannellid species. Evermannella indica ap- pears to be replaced by its closely related congeners, E. ahlstromi and E. megalops, in portions of Pacific Equatorial Water and in much of the central gyral area of the South Pacific, respectively (figs. 30, 35). Larvae and small juveniles (to 30 mm SL) of E. indica have commonly been taken in the upper 100 m and, on a number of occasions, in the upper 50 m. Gibbs & Roper (1971, p. 127) report the capture of juvenile specimens of E. indica between 200 and 400 m at the Ocean Acre site near 32° N, 64° W. Most adults were taken in hauls to depths exceeding 400 m and for the most part in hauls to depths between 500 and 800 m. However, adults have been taken on a number of occasions in the upper 200 m and on several occasions in the upper 100 m. Larvae and small junveniles have been taken in all months of the year. Material Examined. — A total of 986 (9.0-127.2 mm SL) specimens from 483 collections. ATLANTIC OCEAN: A total of 564 (9.0-116.9 mm SL) specimens from 261 collections. FMNH: 49846 (1), 49847 (1), 49864 (1), 49873 (1), 49876 (1), 49883 (1), 49984 (1), 49986 (1), 66088 (1). ISH: 152/66 (1), 476/66 (3), 523/66 (2), 524/66 (1), 589/66 (1), 202/67 (1), 330/68 (1), 531/68 (1), 1168/68 (1), 1349/68 (1), 1484/71 (1), 2286/71 (2), 2836/71 (3), 2981/71 (1), W.H. 506/71 (1). MCZ: 32280 (1). NIO: DY 3700 (1), DY 4258 (1), DY 4947 (1), DY 4949 (1), DY 6411 (1), DY 7036 (2), DY 7089-13 (1), DY 7089-53 (1), DY 7802 (1), DY 7836-2 (1). SIO: 64-443 (1). UMML: 16557 (1), 18628 (1), 23638 (1), 24161 (1), 24329 (1), 24360 (2), 27581 (2). USNM, ACRE: 1-4B (1), 1-4C (1), 1-8 C+D (1), 1-11B (2), 1-16C (1), 1-18B (1), 1-19A (3), 3-3 (1), 3-10 (1), 3-13 (3), 4-3C (1), 4-5A (1), 4-5B (2), 4-10B (1), 4-10C (2), 4-11B (1), 4-21D (1), 7-12 (2), 7-15 (2), 7-16 (2), 7-17 (1), 7-18 (1), 7-19 (4), 8-2 (1), 8-3 (1), 9-27 (1), 10-1N (1), 11-13C (1), 12-2M (1), 12-5M (1), 12-9C (1), 12-12B (1), 12-12C (1), 12-12M (1), 12-15C (1), 12-18A (3), 12-20M (1), 12-22B (1), 12-22M (1), 12-26A (2), 12-26C (2), 12-32A (9),12-32B (18), 12-32C (4), 12-33M (1), 12-34C (2), 12-35B (1), 12-36A (3), 12-36B (1), 12-55 (4), 12-61 (3), 12-61N (1), 12-70 (6), 12-72 (3), 12-72N (1), 12-74 (3), 12-80 (2), 12-83 (1), 12-84 (1), 12-88 (1), 13-05B (1), 13-08B (1), 13-15B (1). WHOI, RHB collection numbers: 1013 (1), 1054 (1), 1108 (1), 1119 (1), 1127 (1), 1257 (1), 1258 (1), 1263 (2), 1266 (1), 1267 (1), 1277 (17), 1289 (4), 1297 (3), 1298 (1), 1302 (1), 1423 (1), 1435 (2), 1505 (2), 1509 (7), 1510 (2), 1515 (1), 1733 (1), 2005 (1), 2006 (2), 2018 (2), 2020 (1), 2027 (1), 2031 (2), 2066 (2), 2067 (1), 2069 (1), 2082 (1), 2090 (1), 2109 (1), 2118 (1), 2218 (1), 2265 (1), 2295 (1), 2906 (3), 2908 (1), 2909 (1), 2910 (1), 2912 (2), 2923 (2), 2924 (3), 2925 (1), 2927 (2), 2928 (3), 2929 (1), 2930 (3), 2931 (2), 2938 (1), 2939 (1), 2944 (1), 2966 (1), 2967 (1), 2972 (1), 2973 (1), 2976 (1), 2980 (1), 2987 (1), 2988 (1), 2990 (2), 2993 (1), 3000 (1), 3015 (1), 3102 (3), 3104 (1), A & J 015 (1). ZIZM: RBF 122 (1). ZMUC: D 855 III (1), D 855 XXI (1), D 858 (1), D 883 (1), D 891 (3), D 1016 IV (2), D 1148 I (3), D 1153 (2), D 1153 1 (3), D 1153 VI (2), D 1161 IV (2), D 1162 II (2), D 1163 III (2), D 1166 IV (1), D 1185 VIII (1), D 1185 XI (1), D 1240 I (2), D 1365 IX (2), D 4180 II (1). i B B SB 9 145 146 FIELDIANA: ZOOLOGY Additional larval and juvenile material from Atlantic Ocean: CAS, SU catalogue num- bers: 57694 (1), 57696 (1), 57698 (2), 57699 (2), 57700 (1), 57701 (1), 57703 (2), 57704 (1), 57705 (1), 57706 (1), 57707 (1), 57708 (1), 57709 (1), 57710 (1), 57711 (1), 57712 (1), 57713 (1), 57714 (2), 57715 (1), 57716 (1), 57719 (1), 57721 (1), 57725 (1), 57726 (1), 57727 (1), 57892 (1), 57893 (1). USNM, ACRE: 3-5 (1), 3-13 (9)?, 4-25 (1)?, 7-14 (1), 12-24M (1), 12-34C (1)?, 12-35C (1)?. ZMUC: D 837 (1), D 842 (1), D 844 (7), D 849 (3), D 855 II (1), D 855 XVIII (1), D 856 VIII (2), D 857 (3), D 862 (7), D 864 (8), D 865 (4), D 882 (14), D 926 (7), D 934 (3), D 935 (1), D 941 (1), D 944 (2), D 945 (3), D 947 (3), D 1053 V (1), D 1142 IX (1), D 1155 I (2), D 1155 II (3), D 1157 II (1), D 1183 XIV (1), D 1216 II (1), D 1218 III (1), D 1242 III (1), D 1269 VII (1), D 1320 III (1), D 1362 IV (1), D 3538 I (52), D 3541 1 (49). INDIAN OCEAN: A total of 142 (12.8-97.1 mm SL) specimens from 62 collections. IOAN: B 6 (1), V 5177 (1), V 5207 (1), V 5290 (1). NIO: DY 5413 (1), DY 5420 (1). MCZ: MCZ 49182 (1), MCZ 50993 (3), MCZ 50996 (1), AB 3-147 (1), AB 6-335A (1), AB 6-337A (1). ZIZM: IOES No. 39 (1). ZMUC: D 3821 1 (1), D 3828 V (1), D 3850 1 (1), D 3851 1 (2), D 3902 II (2), D 3904 III (2), D 3906 III (4), D 3906 IV (3), D 3907 II (1), D 3907 III (3), D 3908 I (1), D 3908 II (7), D 3908 III (1), D 3915 III (2), D 3917 VIII (1), D 3920 VIII (1), D 3931 II (1), D 3933 III (2), D 3933 IV (1), D 3935 I (1), D 3947 I (3), D 3949 I (1), D 3949 II (1). Additional larvae and juvenile material from the Indian Ocean. ZMUC: D 3828 XI (1), D 3849 1 (2), D 3851 II (4), D 3851 III (6), D 3852 1 (2), D 3852 II (5), D 3852 III (1), D 3853 1 (3), D 3856 II (8), D 3856 III (6), D 3872 1 +11 (1), D 3906 IV (3), D 3907 1 (1), D 3907 III (2), D 3910 II (1), D 3912 III (4), D 3914 III (3), D 3934 II+VII+XII (1), D 3935 I (1), D 3937 I (4), D 3937 II (2), D 3941 1 (1), D 3943 I (1), D 3951 1 (1), D 3953 I (2), D 3955 I (1). PACIFIC OCEAN: A total of 280 (9.6-127.2 mm SL) specimens from 160 collections. FMNH: 80405 (1), 80406 (2). IOAN: V 3786 (1), V 4494 (1), V 5117 (1), V 6429 (6), V 6469 (2), V 6490 (1), V 6493 (1). NMFS: J 20.135 (1). J 22.143 (2), J 27.135 (1), J 31.127 (1), J 31.145 (1). ORSTOM: CA 11-70 (1), CA III-121 (1), CA IV-3 (1), CY 11-15 (5), CY H-14 (1), CY IV-10 (1), CY VI-12 (1). SIO: 60-206 (2), 60-239 (1), 60-249 (1), 69-341 (2), 69-344 (3), 70-103 (1), 70-314 (1), 70-345 (1), 70-346 (1), 71-301 (1), 71-305 (2), 71-310 (1), 72-14 (1), 72-16 (2), 72-17 (1), 72-18 (2), 72-23 (1), 72-314 (1), 73-147 (1), 73-148 (2), 73-149 (1), 73-159 (1), 73-161 (1), 73-170 (2), 73-171 (1), 73-325 (1), 73-326 (1), 73-328 (1), 73-330 (1), 73-334 (1), 73-336 (1), 73-338 (1), 75-631 (1), 75-632 (1), 76-14 (1), 76-144 (1). UH: 69/9/11 (1), 69/11/1 (3), 69/11/3 (4), 69/11/6 (1), 70/7/14 (1), 70/7/16 (1), 70/7/22 (1), 70/7/26 (1), 70/7/28 (1), 70/9/4 (2), 70/9/9 (1), 70/9/11 (1), 70/9/12 (2), 70/9/13 (2), 70/9/14 (2), 70/9/20 (1), 70/9/24 (1), 70/9/28 (1), 70/12/7 (1), 70/12/9 (2), 70/12/13 (1), 70/12/31 (1), 71/2/7 (1), 71/3/3 (1), 71/3/4 (1), 71/3/9 (1), 71/6/9 (1), 71/6/10 (5), 71/6/11 (1), 71/6/14 (2), 71/6/14 (1), 71/6/20 (1), 71/6/22 (2), 71/6/28 (2), 71/6/31 (1), 71/6/34 (1), 71/9/8 (1), 71/10/7 (1), 71/10/8 (1), 73/8/15 (1), 73/8/19 (1), 73/8/21 (1), 73/8/27 (1), 73/8/30 (1), 73/8/31 (2), 73/9/13 (1), TC 47-57 (1), TC 47-68 (4), TC 47-69 (3), TC 52-52 (1). USNM: USNM 201694 (1). ELT: 31-11A (2), 31-21A (3), 31-22A (5). ZMUC: D 3576 I (1), D 3578 II (1), D 3581 1 (1), D 3585 VIII (2), D 3602 VIII (1), D 3676 II (2), D 3677 1 (1), D 3678 III (1), D 3678 V (1), D 3681 II (1), D 3683 IV (1), D 3684 III (1), D 3689 I (2), D 3689 VIII (1), D 3731 XIII (1), D 3734 I (1), D 3738 III (2), D 3739 VI (2), D 3746 II (1), D 3751 II (4), D 3755 II (1), D 3768 IV (1), D 3768 V (1), D 3789 II (1), D 3789 VIII (2), D 3795 HI (1), D 3797 III (1), D 3800 II (1). Additional larval and juvenile material from the Pacific Ocean. ZMUC: D 3563 III (2), D 3580 IX (1), D 3581 II (1), D 3613 II (1), D 3678 VIII (1), D 3683 VIII (2), D 3689 V (1), D 3715 III (1), D 3738 II (1), D 3745 III (2), D 3746 II (1), D 3749 III (8), D 3752 II (6), D 3753 II (2), D 3768 VI (37), D 3768 VII (4), D 3768 IX (1), D 3775 I (2), D 3795 III (1), D 3797 III (5), D 4799 (1). Evermannella megalops Johnson & Glodek 1975, Figure 29 Evermannella megalops Johnson & Glodek 1975, pp. 721-723 (original description based on 10 specimens from the South Pacific Ocean). Evermannella indica, Craddock & Mead 1970, p. 3.26 (not of Brauer, 1906; in part, speci- mens from offshore stations, R/V Anton Bruun stations 13-26, 13-28, 13-30). Holotype.— 65.6 mm SL, SIO 72-305, central South Pacific, 25° 05.3' to 08.5' S, 154° 54.5' to 155° 12.5' W, IKMT, 3,000 mwo, 27 July 1972. Diagnosis. — A species of Evermannella with 10 to 12 dorsal-fin rays (eight of 10 specimens counted with 11 dorsal-fin rays), 29 to 31 anal-fin rays, and 48 to 50 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 147 vertebrae. Snout-pad pore formula = 2 + 4 + + 1+0. Gill filaments not projecting beyond gill covers. Detailed comparisons of £. megalops with other species of Evermannella are given above, following the description of the genus. Description. — Values for meristic characters are presented in Tables 2 and 3 and in the original description (Johnson & Glodek, 1975). Only six additional specimens of E. megalops have come to hand, and the description of this species is not repeated here. The following constitutes an addition to the original de- scription. PROPORTIONAL DIMENSIONS.— Expressed as thousandths of the SL and based on five specimens, 32.5, 58.0, 61.0, 65.6, 66.0 with values listed in that order. Interorbital width, 17, 7, 6, 4, 5. CEPHALIC LATEROSENSORY PORES.— Snout-pad pore formula: 2 + 4 + + 1+0. Mandibular pore formula: 5 + 4 + 2 + 1. Preopercular pores: about 4. Temporal pores: PPTO = 1, PESC = 4. Frontal pores: PPF = 1; frontal canal commissure = 3 + 3. Infraorbital pores: 8 or 9 + 5. Distribution. — Evermannella megalops is restricted to the central South Pacific (fig. 35). The distribution of E. megalops as well as other central gyral endemic species in the Pacific Ocean is discussed in considerable detail in a subsequent section of this paper. No specimens of E. megalops have been taken in discrete-depth sampling devices. Only four larvae and small juveniles (to 30 mm SL) from four collections are known. These were taken in March (1), August (1), and October (2). Material Examined. — The original description of E. megalops (Johnson & Glodek, 1975) was based on 10 (22.1-66.0 mm SL) specimens from nine collec- tions. These are not listed here. An additional six (29.8-68.1 mm SL) specimens from four collections have come to hand. PACIFIC OCEAN: SIO: 72-305 (1), 76-122 (1). ZMUC: D 3577 IX (1), D 3602 IV (3). Odontostomops Fowler 1934 Odontostomops Fowler 1934, p. 322 (original diagnosis as a subgenus of Evermanella, type-species by original designation Evermannella [Odontostomops] normalops [Parr 1928]). Type-Species. — Odontostomops normalops (Parr 1928). Diagnosis. — Evermannellids with "normal," nontubular eyes, directed laterad. Horizontal eye diameter distinctly less than interorbital width. Aperture in adipose eyelid much smaller than lens of eye in diameter. No lens pad. At least some dentary and palatine teeth barbed. Anteriormost palatine tooth a large fang, 5.3% to 6.9% SL. Dentary teeth biserial. No fossa centered on den- tary symphysis. Body relatively shallow, body depth at anal-fin origin 13.5% to 17.0% SL. No pyloric caecum. Ethmoid cartilage not expanded posteriorly into orbit, not forming an orbital septum. Basisphenoid absent. Luminous tissue absent. Larvae with 12 or more peritoneal pigment sections. Description. — Dorsal-fin rays 11 to 13. Anal-fin rays 30 to 35. Pectoral-fin rays 11 to 13. Vertebrae 48 to 52. Body moderately elongate, relatively shallow, strongly compressed, Anus dis- tinctly posterior to a point midway between pelvic-fin insertion and anal-fin origin. Lateral line extending to a vertical through middle of anal-fin base and composed of up to 43 segments. 148 FIELDIANA: ZOOLOGY Head moderately large, head depth and width subequal to body depth and width. Snout relatively high and distinctly truncate. Eyes small, rounded, lateral in position, not tubular, and directed laterad. Horizontal diameter of eye 2.7% to 4.2% SL, vertical diameter 2.8% to 4.0% SL. Fleshy eye diameter distinctly less than snout length. Diameter of aperture in adipose eyelid much less than diameter of lens. Pupil broader than lens. Lens pad absent. Dentary symphysis marked by a distinctly protruding ridge and lacking any vertically elongate fossa. Branchiostegal membranes free from isthmus, united by a small membrane anteriorly, at or just posterior to a vertical through anterior margin of eye. Gill filaments elongate and narrow, typically extending to or nearly to margin of gill covers but not projecting beyond margin of gill covers. Pseudobranchiae with filaments distinctly shorter than longest gill filaments, numbering four to seven in seven (37.2-118.7 mm SL) specimens counted. Number of pseudobranch elements tending to be higher in larger specimens. Dorsal fin relatively short based, 9.0% to 11.5% SL. Middle of dorsal-fin base just anterior to a vertical through middle of standard length. Pelvic-fin insertion under anterior one-third of dorsal-fin base. Appressed pelvic fins reaching to or nearly to anus in best-preserved specimens but not reaching anal-fin origin. Pectoral fins distinctly exceeding pelvic fins in length. Appressed pectoral fins reaching to or nearly to a vertical through a point midway between pectoral-fin insertion and pelvic-fin insertion but not reaching pelvic-fin insertion. Anal-fin base relatively elongate, 27.3% to 30.4% SL. Content. — Odontostomops is monotypic, and the single species, O. normalops (fig. 36), is nearly circumtropical in distribution. Odontostomops normalops (Parr 1928), Figure 36 Evermannella normalops Parr 1928, p. 164 (original description, holotype from western North Atlantic); Parr 1930, p. 154 (name only). Evermannella (Odontostomops) normalops Fowler 1934, p. 322 (diagnosis of subgenus Odontostomops Fowler, type-species = Evermannella normalops Parr). Odontostomops normalops Rofen 1966d, p. 520 (description, records from western North Atlantic); Johnson 1974c, p. 30 (record from North Pacific Ocean); Wassersug & Johnson 1976, p. 276 (gut morphology, record from North Pacific Ocean); Herring 1977, p. 306 (lack of luminous tissue, cf., Coccorella atrata Alcock). Holotype. — Bingham Oceanographic Collection No. 2143; 51.6 mm SL. PAWNEE station 23 (third expedition), western North Atlantic, 24° 29' N, 77° 29' W (8,000 ft wire out). 14 March 1927. Diagnosis. — As for the genus. Description. — Values for meristic characters are presented in Tables 2 and 3. PROPORTIONAL DIMENSIONS: Based on 25 (36.5-122.0 mm SL) specimens from throughout the range of the species. Expressed as thousandths of the standard length and given as the mean and range (values in parentheses). Body: depth at anal-fin origin, 145 (135-170). Caudal peduncle: least depth, 72 (63-82); length, 95 (84-104). Adipose fin: distance to midcaudal rays, 122 (113- 134); distance to dorsal-fin base, 315 (291-356). Anal fin: length of base, 287 (273-304). Dorsal fin: length of base, 99 (90-115); dorsal-fin origin to anal-fin origin (distance between verticals), 193 (164-216); end of dorsal-fin base to base of midcaudal rays, 480 (464-505). Pelvic-fin insertion to anal-fin origin: 180 (151-206). Pectoral-fin insertion to pelvic- fin insertion: 209 (179-267). Anus to 149 150 FIELDIANA: ZOOLOGY anal-fin origin: 52 (41-68). Distance from snout to: anus, 586 (554-614); dorsal- fin origin, 451 (433-491); adipose fin, 844 (817-878); anal-fin origin, 629 (576- 659); pectoral-fin insertion, 259 (231-285); pelvic-fin insertion, 455 (409-490); anterior margin of eye (=snout length), 60 (52-71). Head length: 221 (197-247). Postorbital head length: 141 (113-167). Eye: horizontal diameter, 33 (27-42); vertical diameter, 34 (28-40). Upper jaw length: 166 (149-182). Lower jaw length: 161 (144-178). Longest dentary tooth: 52 (45-60). Longest palatine tooth: 61 (53-69). Interorbital width: 41 (36-52). BODY: Body moderately elongate, largest known specimen 123.1 mm SL (At- lantic Ocean: IOAN AK 998). Body relatively shallow, body depth at anal-fin origin 13.5% to 17.0% SL. Anus posterior to a point midway between pelvic-fin insertion and anal-fin origin. Ratio of distance from pelvic-fin insertion to anus over distance from pelvic-fin insertion to anal-fin origin = 0.73 to 0.80 in 12 (37.2-118.7 mm SL) specimens measured (from Atlantic [n=8], Indian [n=2], and Pacific [n=2] oceans). Lateral line extending to a vertical through middle of anal-fin base and composed of up to 43 segments. CEPHALIC LATEROSENSORY PORES: Snout-pad pore formula: 2 + 4 + 2 + 1+0. Mandibular pore formula: 5 + 5 + 2 + 1. Preopercular pores: 3+3. Temporal pores: PPTO = 1, PESC = 4. Frontal pores: PPF = 1; frontal canal commissure = 3+3. Infraorbital pores: about 11 + 9. MOUTH: Upper jaw extending nearly to anterior margin of preopercle, well past a vertical through posterior margin of eye. Lower jaw projecting anteriorly very slightly beyond snout. Premaxillary teeth small, retrorse, uniserial, numbering 34 to 51 in eight (37.2-118.7 mm SL) specimens counted. Dentary with one or two smaller fangs anteriorly near symphysis, followed by a row of five to seven large, barbed fangs, with these bordered laterally by a row of four to nine smaller teeth. Dentary tooth counts based on eight (37.2-118.7 mm SL) specimens. Largest dentary fangs the longest anteriorly and decreasing in length posteriorly. Vomer probably with one small tooth on each side, but in the majority of specimens only one laterally positioned tooth is present, its counterpart either failing to develop or lost. Anteriormost palatine tooth an enormous, barbed fang, easily the largest tooth on each side. Each such fang with a peculiar elbow-shaped bend near distal terminus, at which point tooth angles forward and downward, ending in a triangular point. Palatine teeth numbering five to seven in eight (37.2-118.7 mm SL) specimens counted. Number of premaxillary teeth distinctly higher in larger specimens. Lingual teeth lacking. COLOR: Color in alcohol a solid dark brown over head, body, and fins. A brassy, iridescent coloration overlying head and flanks, evident in best- preserved specimens. No marked concentration of pigment into bars, stripes, patches, or the like on either head or body. Peritoneum black. Geographic Variation. — Odontostomops normalops exhibits geographic variation in number of anal-fin rays and number of vertebrae (tables 17, 18). The pattern of variation observed — with counts lowest in the Atlantic, intermediate in the In- dian, and highest in the Pacific Ocean — apparently parallels that seen in results for Evermannella indica. Results for both species are discussed under E. indica. Distribution. — Odontostomops normalops is nearly circumglobal in tropical and subtropical waters, occurring in central and equatorial waters in the Atlantic, Indian, and Pacific oceans (fig. 37). Odontostomops normalops, however, is appar- 151 152 FIELDIANA: ZOOLOGY ently excluded from much of the North Atlantic range of Evermannella balbo (compare figs. 32 and 37), including the Mediterranean Sea, and is also appar- ently excluded from the eastern Pacific. Larvae and small juveniles (to 30 mm SL) of O. normalops have commonly been taken in the upper 100 m, and, on a number of occasions, in the upper 50 m. Most adults (greater than 50 mm SL) were taken in hauls to depths exceeding 400 m, but adults have been taken on a number of occasions in hauls to between 100 and 400 m. Larvae and small juveniles have been taken throughout the year. Material Examined. — A total of 254 (9.0-123.1 mm SL) specimens from 184 collections. ATLANTIC OCEAN. A total of 115 (9.0-123.1 mm SL) specimens from 78 collections. FMNH: 66087 (4), 71704 (1). FSBC: 2985 (1). IOAN: AK 820 (1), AK 991 (1), AK 998 (1). ISH: 371/66 (1), 523/66 (3), 722/66 (1), 286/68 (1), 825/68 (3), 895/68 (1), 947/68 (1), 1024/68 (1), 1139/68 (1), 1219/68 (2), 1282/68 (1), 1318/68 (3), 1684/71 (1), 2087/71 (2), 2220/71 (2), 2286/71 (3), 2390/71 (1). NIO: DY 7089-53 (4), DY 7836-3 (1). UMML: 14811 (1), 15960 (1), 17333 (2), 17584 (1), 20002 (1), 20006 (1), 22999 (1), 23903 (1), 26442 (1). USNM: 108305 (1). ACRE: 3-4 (1), 4-10D (1), 9-31 (1), 12-14M (1), 12-22C (1). WHOI, RHB: 970 (1), 971 (1), 975 (1), 1100 (1), 1107 (1), 1252 (1), 1253 (8), 1255 (1), 1258 (1), 1263 (3), 1277 (3), 1281 (1), 1290 (1), 1308 (1), 1315 (3), 1506 (3), 2294 (1), 2295 (4), 2296 (2), 2300 (1), 2918 (1), 2941 (1), 2958 (1), 2965 (1), 2979 (1), 2992 (1), 3010 (1), 3056 (1). ZMUC: D 1171 XII (1), D 1189 1 (1), D 1225 II (1), D 1230 VI (1), D 3999 11(1). Additional larvae and juvenile material from the Atlantic Ocean. ZMUC: D 1189 III (1), D 1231 II (1), D 1260 II (1), D 1269 IX (1), D 1270 II (1). INDIAN OCEAN. A total of 19 (21.9-118.7 mm SL) specimens from 15 collections. IOAN: V 4953 (1). MCZ: AB 3 AE-17d (1), AB 6-332B (2), AB 6-337A (1), AB 6-339A (1). SIO: 69-22 (1). ZMUC: D 3828 V (2), D 3847 V (1), D 3904 III (2), D 3908 III (1), D 3917 VIII (1), D 3951 1 (1). Additional larvae and juvenile material from the Indian Ocean. ZMUC: D 3907 III (2), D 3916111(1), D 3951 11(1). PACIFIC OCEAN. A total of 120 (10.5-112.5 mm SL) specimens from 91 collections. FMNH: 80403 (1), 80404 (1). IOAN: V 23 (1), V 3700 (1), V 6429 (3 May 1971) (5), V 6429 (5 May 1971) (3), V 6429 (7 May 1971) (1), V 6437 (2), V 6469 (1). NMFS(LJ): J 20.145 (2), J 24.145 (1), J 31.139 (1). ORSTOM: CY 11-14 (1), CY V-21 (1), CY VI-ll(l). SIO: 51-375 (1), 60-134 (1), 60-236 (1), 60-239 (2), 61-576 (2), 61-588 (1), 68-535 (1), 69-341 (1), 69-345 (1), 70-311 (1), 70-336 (1), 70-340 (1), 70-343 (1), 70-345 (1), 71-294 (1), 71-296 (1), 71-297 (2), 71-309 (1), 71-310 (1), 72-9 (2), 72-15 (1), 72-22 (1), 72-23 (1), 72-316 (1), 73-147 (1), 73-149 (2), 73-151 (1), 73-165 (3), 73-166 (1), 73-170 (1), 73-325 (1), 73-329 (1), 73-331 (1), 73-336 (1); SIO uncat., 76-6 (1). UH: 69/11/1 (5), 61/11/2 (1), 69/11/6 (1), 70/7/1 (1), 70/12/9 (1), 70/12/31 (1), 70/12/34 (1), 71/3/10 (1), 71/6/1 (1), 71/6/14 (1), 71/6/20 (1), 71/6/21 (2), 71/6/22 (3), 71/6/31 (1), 73/8/38 (1), TC 47-68 (1), TC 47-69 (2), TC 52-63 (1). USNM: 201691 (1). ZMUC: D 3676 1 (1), D 3682 II (1), D 3689 II (1), D 3689 VII (1), D 3714 III (1), D 3714 XI (1), D 3740 II (6), D 3751 VI(1), D 3788 1(1). Additional larvae and juvenile material from the Pacific Ocean. ZMUC: D 3676 III (1), D 3745 II (1), D 3752 II (1), D 3753 I (1), D 3753 II (1), D 3766 XIII (2), D 3773 I (1), D 3789 VIII (1), D 3789 IX (1), D 3791 II (1), D 3792 II (1), D 3800 II (1). SCOPELARCHIDAE: Species Accounts The alepisauroid family Scopelarchidae contains 17 species arranged in four genera. The family is worldwide in distribution (fig. 38) except that no scopelar- chid occurs in the Arctic Ocean or in the Mediterranean Sea. My revision of the family (Johnson, 1974c) was based on 2,102 specimens from 1,122 collections. Since publication of that revision, I have examined an additional 1,557 scopelar- chid specimens from 714 collections and from throughout the geographic range of the family. This new material represents all known scopelarchid species ex- cept Scopelarchoides kreffti and Scopelarchus stephensi. The new material adds sub- 153 154 FIELDIANA: ZOOLOGY stantially to our knowledge of several scopelarchid species and confirms distri- butional patterns discussed in Johnson (1974c). In this section I present additions to the revision of the Scopelarchidae necessitated by study of this new material. Benthalbella Zugmayer 1911 Benthalbella dentata (Chapman, 1939) Distribution. — Benthalbella dentata is limited to the Pacific Subarctic and Transi- tion Region areas of the North Pacific Ocean (Johnson, 1974c, p. 70, fig. 19). Material Examined.— Johnson (1974c, pp. 70, 235) lists 127 (20.0-203.0 mm SL) specimens from 108 collections. An additional 17 (29.0-193.5 mm SL) specimens from 11 collections are listed here. PACIFIC OCEAN: FMNH: 71698 (1), 71699 (1). SOSC: DES 5-T3-B (1), DES 5-T3-C (1), DES 5-T6-A (3), DES 5-T6-B (4), DES 5-T7-D (2), DES 5-T8-B (1), DES 5-T8-D (1), DES 6-T2-A (1), DES 6-T13-A (1). Benthalbella elongata (Norman) 1937 Distribution. — Benthalbella elongata is an Antarctic species known only from south of the Subtropical Convergence (Johnson, 1974c, p. 76, fig. 21). Material Examined. — Johnson (1974c, p. 76) lists 68 (33.0-234.0 mm SL) speci- mens from 48 collections. An additional four (98.5-192.5 mm SL) specimens from four collections are listed here. ATLANTIC OCEAN: IOAN: AK 883 (1), AK 907 (1), AK 923 (1), AK 942 (1). Benthalbella infans Zugmayer 1911 Johnson (1974c, p. 82) reports that specimens of B. infans from the central North Pacific exhibit higher anal-fin ray counts than specimens from other geo- graphic areas. This report is corroborated by results for the additional material reported here. Anal-fin ray counts for specimens from the central North Pacific are significantly higher than for specimens from any other area (table 20). The lack of significant difference between counts for areas A, B, and C (table 20) suggests that the pattern of variation in B. infans is different from the pattern in Evermannella indica and Odontostomops normalops (see above). Johnson & Barnett (1975, pp. 293, 294) present limited evidence for the exis- tence of two separable populations of the photichthyid Vinciguerria nimbaria (Jor- dan & Williams) in the insular west Pacific (specifically the South China Sea) vs. the central North Pacific. The data for B. infans presented here partially parallel those for V. nimbaria to the extent of suggesting the distinctness of the North Pacific central gyral population of B. infans. This suggestion needs further con- firmation based on additional material and characters and particularly requires the examination of additional adult material of B. infans from the central North Pacific. Distribution. — Benthalbella infans is a nearly cosmopolitan warm-water species (fig. 39) inhabiting central and equatorial water mass areas of all three oceans. Benthalbella infans does not occur in the Transition Region areas of the North or South Pacific nor does it occur in the eastern portion of Pacific Equatorial Water. Benthalbella infans occurs almost throughout the area of North Atlantic Central Water, is known from the Caribbean Sea, but has apparently not been taken in the Gulf of Mexico. in j +1 cC £ s £33 vC ^ K +i +i +i iritn n Ov Tj< vO Pi I I a i 4) o < C t '•6 t? c«= '0 re 39 a. B n i — i en .5 ^- *v^x\ScS^NvS^ir^-^i -s ^^^^^^t5E ^ 5 - "~~"~^=sL \^&7%>- Jr-^Zr^^?i •*/Os^^ | ^Y\2z' -4— k JTr^fc^r^^^ -6 '. i iT~-~i^T\7\X\)\\AV WBfO m£-^\~\~- o v J^j-^f^ c > ? g Rsk$Scs> TjH J - ' 1 via iw^irr-tz < l> '•■ - \ ?* 1 l*§ap» 7a^a/x/^ .1 f - o 1/ 156 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 157 Material Examined. —Johnson (1974c, pp. 84, 235) lists 93 (6.6-137.4 mm SL) specimens from 64 collections. An additional 90 (12.2-124.4 mm SL) specimens from 54 collections are listed here. ATLANTIC OCEAN: WHOI, RHB: 807 (1), 1037 (1), 1038 (12), 1046 (1), 1411 (3), 1721 (1), 1730 (1), 1886 (1), 1911 (1), 2025 (2), 2026 (1), 2032 (1), 2058 (1), 2070 (1), 2080 (6), 2087 (1), 2097 (1), 2116 (1), 2216 (1), 2229 (1), 2245 (1), 2264 (1), 2265 (1), 2274 (11), 2278 (1), 2531 (1), 2555 (2), 2565 (1), 2601 (1), 2900 (1), 2915 (1), 2917 (2), 2918 (2), 2923 (1), 2926 (1), 2979 (1), 2988 (1), 2998 (1), 3004 (2), 3005 (2), 3010 (3), 3014 (1), 3015 (1), 3017 (1), 3018 (1), 3020 (1). INDIAN OCEAN: SOSC: AB 3-6 (1), AB 3-13 (1), AB 6-348A (1), AB 6-349A (1), AB 6-351C + 352A (1). PACIFIC OCEAN: IOAN: V 25 (1). UH: 70/7/24 (1), TC 52-52 (1). Benthalbella linguidens (Mead & Bohlke) 1953 Distribution. — Benthalbella linguidens is known only from the subarctic North Pacific, north of 39° 21' N, from off northern Japan to off Oregon (Johnson, 1974c, p. 87, fig. 19). Material Examined. — Johnson (1974c, pp. 88, 235) lists 22 specimens (40.6- 221.0 mm SL) from 12 collections. One additional specimen (82.1 mm SL) is reported here. PACIFIC OCEAN: IOAN: V 3605 (1). Benthalbella macropinna Bussing & Bussing 1966 Distribution. — Bethalbella macropinna is an Antarctic species occurring through- out subantarctic and antarctic waters in the Antarctic Circumpolar Current (Johnson, 1974c, p. 94, fig. 28). Material Examined.— Johnson (1974c, p. 94) lists 63 (29.1-233.5 mm SL) speci- mens from 47 collections. An additional 15 (52.8-240.0 mm SL) specimens from 11 collections are listed here. ATLANTIC OCEAN: IOAN: AK 844 (3), AK 845 (1), AK 850 (1), AK 855 (2), AK 856 (2), AK 886 (1), AK 936 (1), AK 942 (1), AK 949 (1), AK 960 (1). WHOI, RHB: 2250 (1). Rosenblattichthys Johnson 1974 Rosenblattichthys alatus (Fourmanoir) 1970 The account of R. alatus in Johnson (1974c, pp. 97-103) was based on 26 specimens from 24 collections. The discovery of substantial additional material of/?, hubbsi Johnson 1974 (see below), allowed elucidation of characters useful in separating larval material of R. alatus from that of R. hubbsi. Using this new information, I reexamined specimens and/or recorded character data for material of R. alatus. A number of the specimens previously identified as R. alatus are in fact R. hubbsi— SIO: 70-310 (1), 71-295 (1), 72-9 (1); ZMUC: D 3932 VII (1), D 3932 VIII (1), D 3964 II (2). The following specimens are now tentatively identified as R. hubbsi— ZMUC: D 3927 II (1), D 3928 I (1), D 3929 I (1). I have also examined new material of R. alatus, six specimens from six collections. This new material and the changes in identification of the above specimens force me to partially modify my previous description of R. alatus and to substantially modify the zoogeographic account for this species. Description. — Only necessary changes in my previous description (Johnson, 1974c) of R. alatus are included in the following account. Meristic characters: anal-fin rays 20 to 22, pectoral-fin rays 23 to 26 (table 21). Development: smallest known larva (D 3893 1, 11.1 mm SL) with only pectoral- and caudal-fin rays fully 158 FIELDIANA: ZOOLOGY Table 21. Comparison of values for meristic characters in three species of Rosenblat- tichthys. Mean ± 95% limits A. Anal-fin i rays Species 20 21 22 23 24 25 N alatus 1 3 7 — — — 11 hubbsi — — — 4 13 2 19 volucris — 2 8 22 8 — 40 B. Pectoral-fir i rays Species 21 22 23 24 25 26 N alatus — — 3 1 3 4 11 hubbsi 4 16 3 — — — 23 volucris — — 2 11 19 8 40 21.5±.46 23.9±.27 22.9±.25 Mean ± 95% limits 24.7±.85 22.0±.24 24.8±.26 differentiated and with the following accessory pigment spots or areas de- veloped: DA, CA, PA, IA (see Johnson, 1974c, pp, 96, 101, for explanation of accessory pigment spots). The following accessory pigment spots or areas are present in all known larger larvae (14.6 mm SL and larger): DA, CA, PA, AA. Distribution. — Rosenblattichthys alatus is known from a total of 25 (11.1-93.9 mm SL) specimens (including the holotype and two additional specimens re- ported by Fourmanoir [1971] but not examined by me) from the Indian and Pacific oceans (fig. 40). Rosenblattichthys alatus is not known from the Atlantic Ocean. The distributions of R. alatus and R. hubbsi (fig. 40) as now known are mutually exclusive: R. hubbsi in central waters of the North and South Atlantic, Indian, and North Pacific oceans; R. alatus in equatorial waters of the Indian and Pacific oceans, insular west Pacific, and (based on one [80.1 mm SL] specimen) central water in the South Pacific Ocean. Except for the absence of R. hubbsi from the central gyral area of the South Pacific, the distribution of these two species of Rosenblattichthys is quite reminiscent of the distributions of Coccorella atlantica and C. atrata (see above). Material Examined. — The following listing includes all valid records of R. alatus for specimens I have examined. A total of 22 (11.1-93.9 mm SL) specimens from 21 collections. INDIAN OCEAN: A total of 14 (11.1-53.5 mm SL) specimens from 13 collections. MCZ: AB 6-339A (1), AB 6-340B (1). USNM: AB 6-340A (1). WHOI: AB 6-335 (1). ZMUC: D 3814 1 (1), D 3893 1 (1), D 3902 III (1), D 3903 II (1), D 3906 III (1), D 3921 III (2), D 3921 V (1), D 3921 VIII (1), D 3925 IV (1). PACIFIC OCEAN: A total of eight (19.5-93.9 mm SL) specimens from eight collections. IOAN: V 4494 (1), V 6429 (3 May 1971) (1), V 6429 (5 May 1971) (1). ORSTOM: MARURU 18A (1). SIO: 60-130 (1), 70-343 (1), 70-344 (1), 72-317 (1). Rosenblattichthys hubbsi Johnson 1974 Rosenblattichthys hubbsi was described from one adult specimen (144.5 mm SL) from the equatorial South Atlantic (Johnson 1974a, 1974c). Three additional lar- val specimens (Dana 1288 II, 1 [23.1 mm SL]; SIO 69-26, 2 [18.8—22.1 mm SL]) from the Atlantic and Indian oceans, respectively, were tentatively identified as this species by Johnson (1974c, p. 106). Included in the large number of addi- tional scopelarchid specimens examined since publication of my revision of this family are 42 larval and juvenile specimens of R. hubbsi from all three oceans. This material resulted in confirmation of my identification of the three larval 159 160 FIELDIANA: ZOOLOGY Table 22. Comparison of values for meristic characters for specimens of Rosenblattichthys hubbsi from the Atlantic, Indian, and Pacific oceans. A. Anal-fin rays Area 23 24 25 N Mean ± 95% limits Atlantic Indian Pacific 3 1 6 2 5 2 9 3 7 23.7±.38 23.7 24.3±.45 B. Pectoral-fin rays Area 21 22 23 N Mean ± 95% limits Atlantic Indian Pacific 4 7 2 7 1 1 1 12 3 8 21.8±.39 22.3 22.1 ±.30 specimens listed above and allows me to expand the description of R. hubbsi to include larval and juvenile specimens. Meristic Characters. — Counts based on a total of 23 specimens from all three oceans: dorsal-fin rays 8 or 9 (usually 9); anal-fin rays 23 to 25; pectoral-fin rays 21 to 23. Rosenblattichthys hubbsi has more anal-fin rays and fewer pectoral-fin rays than either R. alatus or R. volucris (table 21), but there is significant overlap in the case of R. hubbsi and R. volucris. Specimens of R. hubbsi from the Atlantic Ocean appear to have fewer anal-fin rays (and possibly fewer pectoral-fin rays) than specimens from the Pacific Ocean (table 22). This is similar to the pattern described above for Evermannella indica and Odontostomops normalops. However, the differences between the mean values for anal-fin ray counts and pectoral-fin ray counts given in Table 22 are not statistically significant, and the possibility of inter-ocean differences in these counts for R. hubbsi needs to be confirmed on the basis of additional material. Description. — Description is based on larval and juvenile specimens. RECOGNITION.— Larvae of R. hubbsi (fig. 41) are distinguished by the fol- lowing combination of characters: dorsal-fin rays 8 or 9, anal-fin rays 23 to 25, pectoral-fin rays 21 to 23; head remarkably large in smaller larvae, head length in larvae up to 30 mm exceeding 30% SL; pectoral fins precocious, very prominent, and elongate in smaller larvae, ossification of pectoral-fin rays preceding ossifi- cation of rays of all other fins; unique combination of two accessory pigment spots or areas, one middorsal appearing over posterior one-third of anal-fin base (DA) and one appearing as an oblong dash at fork of caudal fin (CA) on each side of body. Larvae of R. hubbsi larger than 12 to 15 mm SL may easily be distin- guished from those of R. alatus and R. volucris in having only two accessory pigment spots or areas rather than five or seven, respectively, accessory pigment spots or areas. Larvae and juveniles of R. hubbsi larger than 20 to 25 mm SL possess dermal pigmentation on the body (described below); such pigmentation is lacking in larvae and juveniles of R. alatus. FINS. — In smallest known specimen (RHB 3011, 9.2 mm), pectoral-fin rays essentially fully differentiated, caudal-fin rays partially differentiated, rays of remaining fins unformed. Pectoral fins precocious, appearing very early in de- velopment, prolonged, and remarkably prominent in smaller larvae. Light pig- mentation on pectoral-fin rays in larvae 12 to 15 mm in length and larger. Pigment at bases and/or on rays of all fins in juveniles. Pelvic fins appearing 161 162 FIELDIANA: ZOOLOGY midlaterally, dorsal to level of gut, and beneath developing dorsal-fin base. Pelvic-fin insertion beneath dorsal-fin base in larger larvae and slightly in ad- vance of dorsal-fin origin in juveniles larger than 30 to 35 mm SL. Ventral adipose fin never exceeding one-third of pelvic-anal distance, reduced to a thin triangular flap in larvae larger than 18 mm, absent in juveniles. Dorsal adipose fin remaining elongate, to over anterior anal-fin rays, in largest larvae. All fin rays ossified in specimens exceeding 18 to 20 mm. Order of fin ray ossification: pectoral, caudal, dorsal, anal, pelvic. PERITONEAL PIGMENT SECTIONS.— Only one peritoneal pigment section. Peritoneal pigment section present in smallest known larvae (9.2 mm) as a thin, dark brown, transverse sheet above gut and medial to pectoral-fin base. Peritoneal pigment section expanding posteriad during subsequent growth and forming a canopy above gut prior to laterad expansion, forming a complete tube of pigmentation around gut. Posteriad expansion of and enclosure of gut by peritoneal pigment section complete in specimens 30 to 35 mm SL and larger. Completion of peritoneal pigment tube around gut taken arbitrarily as basis for classifying specimens as larvae or as juveniles. ACCESSORY PIGMENT SPOTS OR AREAS.— A maximum of two accessory pigment spots or areas (fig. 41): DA and CA. Both pigment spots absent in smallest known larvae (9.2 mm). All known larger larvae, to about 28 mm SL maximum, have both DA and CA pigment spots. The DA appearing over pos- terior one-third of anal-fin base, middorsally, well anterior to posterior terminus of dorsal adipose-fin base. The CA appearing as a single spot, group of spots, or oblong dash of pigment over bases of middle caudal-fin rays. Appearance of dermal pigmentation (see below) results in masking DA, making DA indiscerni- ble in specimens larger than 28 to 30 mm SL. The CA becoming indiscernible in specimens 35 to 40 mm SL and larger. DERMAL PIGMENTATION.— Dermal pigmentation (see Johnson, 1974c, pp. 20-23) present in larvae larger than 20 to 25 mm SL. Two areas of dermal pigmentation appear at about the same time, one middorsally above pectoral-fin base, the other associated with and just posterior to DA pigment spot. Dermal pigmentation on epaxial portion of body appearing to spread posteriorly and anteriorly, respectively, from these two initial areas, with the last epaxial area of body to develop dermal pigmentation being that area immediately ventral to dorsal-fin base. Epaxial dermal pigmentation essentially complete in juveniles 30 to 35 mm and larger. Dermal pigmentation on hypaxial portion of body first appearing on caudal peduncle, immediately ventral to horizontal septum and centered on a vertical through a point immediately posterior to adipose-fin base. Hypaxial pigmentation on body appearing (in successively larger larvae) to spread anteriorly from this area and still incompletely developed in largest known juvenile (SIO 72-9, 63.9 mm SL). Juveniles with a distinctive band of dark pigment along dorsolateral margin of lower jaw. GUT. — Post-pelvic gut length exceeding pelvic-anal distance in smaller lar- vae, forming a partial loop outside of normal limits of abdominal cavity (fig. 41), and attached to ventral adipose fin. Gut with essentially adult proportions — anus immediately in advance of anal-fin origin — in larger larvae and juveniles. In smaller larvae stomach visible only as a short blind sac protruding from posterodorsal margin of gut between pectoral-fin bases. Stomach expanding posteriad with growth, reaching to just before anus in largest juvenile. JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 163 METAMORPHOSIS. — As is true for other species of Rosenblattichthys, R. hubbsi exhibits gradual metamorphosis, with changes leading to adult morphol- ogy occurring over a wide size range. These changes include at least the follow- ing: resorption of ventral adipose fin, decrease in length of base of dorsal adipose fin, extensive posteriad expansion of peritoneal pigment section and stomach, development of distinctive dermal pigmentation of juveniles, and in- vasion of abdominal body wall by musculature. Based on material I have exam- ined, these processes begin in specimens 18 to 20 mm in size and are not com- plete in any of the juvenile specimens. Distribution. — Rosenblattichthys hubbsi is now known from 47 specimens from the Atlantic, Indian, and Pacific oceans (fig. 40). Rosenblattichthys hubbsi occurs in the North and South Atlantic central, Indian Ocean central, and North Pacific central- water- mass areas. Larvae of R. hubbsi have been taken throughout the year in hauls to depths as shallow as 60 m. The holotype remains the only known adult specimen. Material Examined. — A total of 46 (9.2-63.9 mm SL) specimens from 38 collec- tions plus the 144.5 mm SL holotype, ISH 2219/71. ATLANTIC OCEAN: A total of 28 (9.2-41.3 mm SL) specimens from 23 collections. WHOI, RHB collection numbers: 954 (1), 1423 (2), 1431 (1), 2028 (1), 2084 (2), 2103 (1), 2903 (1), 2904 (1), 2906 (1), 2910 (1), 2913 (1), 2914 (1), 2917 (1), 2918 (2), 2922 (1), 2981 (1), 2983 (1), 2989 (1), 2997 (1), 3011 (2). ZMUC: D 1243 III (1), D 1285 HI (2), D 1288 II (1). INDIAN OCEAN: A total of seven (10.0-27.2 mm SL) specimens from five collections. SIO: 69-26 (2). SOSC: AB 6-343A (1). ZMUC: D 3932 VII (1), D 3932 VIII (1), D 3964 II (2). PACIFIC OCEAN: A total of 11 (12.6-63.9 mm SL) specimens from 10 collections. NMFS (LJ): cruise 7205, station 24.145 (1). SIO: 70-310 (1), 71-295 (1), 72-9 (2), 72-14 (1), 73-329 (1), UH: 71/2/9 (1), 71/6/17 (1), 71/6/23 (1), 71/6/27 (1). The following specimens are tentatively assigned to R. hubbsi on the basis of capture location. They are too small for positive identification. Localities for these specimens are not plotted on Figure 40. INDIAN OCEAN: ZMUC, D 3927 II, 1 (9.0), 10° 55' S, 50° 15' E; D 3928 I, 1 (12.1), 11° 20' S, 50° 10' E; D 3929 I, 1 (10.5), 12° 11' S, 50° 18' E. Rosenblattichthys volucris (Rofen, 1966) On the basis of counts and other information presented, the three specimens recorded as Scopelarchus sp. by Parin et al. (1973)— IOAN: B 92 (1), B 124 (1), B 125 (1) — are assigned to R. volucris. Distribution. — Rosenblattichthys volucris is confined to the eastern half of the Pacific Ocean (fig. 42). It is known from the Transition Region off California and off Chile and from a relatively narrow zone along the equator from near the American mainland to 161° 52.5' to 51.5' W. A detailed discussion of the zoogeography of scopelarchids and evermannellids inhabiting the eastern por- tion of the Pacific Equatorial Water Mass region is presented in a subsequent section of this paper. Material Examined.— Johnson (1974c, p. 116) lists 108 (5.9-103.5 mm SL) specimens from 71 collections. An additional 64 (4.7-81.3 mm SL) specimens from 35 collections are listed here. PACIFIC OCEAN: CAS: 63778 (1). IOAN: B 74 (2). NMFS (LJ): J 60-71 (1), J 65-41 (1), J 65-118 (1), J 77-86 (1), J 77-92 (1), J 77-127 (1), J 77-138 (1), J 77-142 (1), TC 51-31 (1), TC 51-37 (1), TC 51-55 (1), TC 51-65 (1), TC 51-66 (1). SIO: 72-171 (1), 74-28 (2), 74-40 (1), 74-41 (2). SOSC: AB 16-618D (1), AB 16-622 A (1), DES 1-T7-C (1), DES 1-T14-D (1), DES 2-T1-C (1), DES 2-T1-D (1), DES 4-T13-C (1), DES 4-T13-D (1), DES 4-T14-B (1), DES 4-T14-C (1), DES 4-T16-D (1), DES 5-T7-D (1), UND 46-50 (1). ZMUC: D 3556 II (1), D 3556 III (3), D 3556 VI (25). 164 FIELDIANA: ZOOLOGY Fig. 42. Distribution of Rosenblattichthys volucris. Scopelarchoides Parr 1929 Scopelarchoides climax Johnson 1974 Scopelarchoides climax was described from a total of nine (18.6-99.3 mm SL) specimens from seven collections, all from the vicinity of 24.5° to 25° S, ca. 155° W, in the central South Pacific. Only one additional specimen has come to hand: ORSTOM, Sillage sta. No. 6, south Tuamotu Islands, 22° S, 139° W, IKMT, 220 m, December 1977, 97.8 mm SL. This specimen is the second known adult of S. climax, and meristic and morphometric data for it are presented below. Meristic characters: dorsal- 'in rays 8, anal-fin rays 27, pectoral-fin rays 25, lateral line scales ca. 52. Proportional dimensions, expressed as thousandths of the SL: Body: depth at dorsal-fin origin, 148. Caudal peduncle: least depth, 74; length, 102. Adipose fin: distance to midcaudal rays, 204; length of base, 49. Anal fin: length of base, 286. Dorsal fin: length of base, 44; dorsal-fin origin to anal-fin origin (distance between verticals), 254; end of dorsal-fin base to bases of mid- caudal rays, 602. Pectoral-fin insertion to pelvic-fin insertion, 134. Anus to anal- fin origin, 52. Distance from snout to: dorsal-fin origin, 367; anal-fin origin, 627; pectoral insertion, 242; pelvic insertion, 373; orbit, 69. Head length, 231. Postor- JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 165 bital head length, 80. Orbit: horizontal diameter, 94; vertical diameter, 97. In- terorbital width, 10. Upper jaw length, 176. Lower jaw length, 195. Longest dentary tooth, 29. In meristics, morphometries, and all other examined char- acters, including the presence of a distinctive black cap of melanophores at the dorsal margin of the lens pad and the distribution of pigment on the head, body, and fins, the ORSTOM specimen agrees well with Johnson's (1974c) description of the holotype. Barnett (1975) lists S. climax as endemic to the central-gyral area of the South Pacific. Scopelarchoides danae Johnson 1974 Distribution. — Scopelarchoides danae is a wide-ranging tropical species occurring in all three oceans (fig. 43). Scopelarchoides danae is known from both sides of the Atlantic, as far as 41° 31 ' to 33' N (RHB 1010, 1 [21.0], specimen taken in a station in the direct path of the Gulf Stream) and from the Caribbean Sea, Gulf of Mexico, and Gulf of Guinea. The only South Atlantic records for S. danae are within 06° of the equator, near the African mainland. Scopelarchoides danae occurs throughout the Indian Ocean and insular western Pacific. In the Pacific Ocean S. danae is known from only two adult specimens and one larval specimen: ORSTOM, Coriolis P 1-6, 1 (113.6), 22° 03' S, 165° 58.0' E, near New Caledonia; ORSTOM, Caride V-20A, 1 (86.1), 09° 54' S, 141° 53' W (not 141° 33' W as listed by Johnson, 1974c, p. 133), near the Marquesas; ZMUC, D 3567 1, 1 (17.4), 09° 06' S, 140° 21.5' W, near the Marquesas. Most captures of S. danae have been near continental or insular land masses. Scopelarchoides danae has not been captured and probably does not occur in the central gyral areas of the Pacific away from island chains, nor does S. danae occur in the eastern Pacific. Material Examined.— Johnson (1974c, pp. 133, 235) lists 230 (6.5-121.2 mm SL) specimens from 104 collections. An additional 119 (10.0-107.5 mm SL) speci- mens from 57 collections are listed here. ATLANTIC OCEAN: WHOI, RHB: 1010 (1), 1104 (2), 1251 (3), 1253 (1), 1259 (2), 1262 (1), 1266 (1), 1274 (20), 1275 (2), 1278 (1), 1282 (4), 1283 (1), 1289 (1), 1312 (1), 1315 (2), 1506 (3), 2288 (2), 2289 (6), 2290 (1), 2291 (2), 2929 (2), 2938 (1), 2942 (2), 2943 (1), 2945 (1), 2947 (2), 2948 (2), 2949 (3), 2950 (1), 2952 (2), 2953 (2), 2955 (5), 2956 (5), 2957 (1), 2960 (1), 2961 (1), 2962 (2), 2963 (2), 2964 (1), 2966 (2), 2972 (1), 2979 (2), 2984 (1), 3050 (1), 3053 (1). ZMUC: D 1188 III (1), D 1228 II (1), D 1230 III (1), D 1285 III (1), D 3545 V (1), D 3546 II (2), D 3547 1 (3). INDIAN OCEAN: IOAN: V 4618 (1). MCZ: AB 6-340B (2); O. Nographer, station J 20-12 (1). PACIFIC OCEAN: ZMUC: D 3687 III (1), D 3712 III (1). Scopelarchoides kreffti Johnson 1972 Distribution. — Scopelarchoides kreffti is known only from the 10 (52.8-187.5 mm SL) specimens from five collections listed in Johnson (1974c, p. 136). All speci- mens were taken in the South Atlantic Ocean between 34° and 41° S, and 48° to 07° W (Johnson, 1974c, p. 136, fig. 41). Scopelarchoides nicholsi Parr 1929 Parin et al. (1973, p. 125) list 14 (27-119 mm SL) specimens from the eastern tropical Pacific. Brewer (1973, p. 19) lists an additional 18 (25-106 mm SL) specimens from the Gulf of California and eastern tropical Pacific. Distribution. — Scopelarchoides nicholsi is restricted to the eastern Pacific Ocean in those areas of Pacific Equatorial Water exhibiting the greatest development of a subsurface layer of poorly oxygenated water (fig. 44). A detailed account of the 166 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 167 Fig. 44. Distribution of Scopelarchoides nicholsi. distribution of S. nicholsi is included in the discussion of the zoogeography of scopelarchids and evermannellids inhabiting the eastern portion of Pacific Equatorial Water in a subsequent section of this paper. Material Examined.— Johnson (1974c, pp. 144, 145) lists 241 (8.1-115.5 mm SL) specimens from 100 collections. An additional 106 (5.2-111.0 mm SL) specimens from 51 collections are listed here. PACIFIC OCEAN: IOAN: AK 229 (1), AK 276 (1), AK 282A (1), AK 291 (1), AK 300 (1). NMFS(LJ): J 57-1 (2), J 57-6 (1), J 57-12 (1), J 57-18 (4), J 57-24 (3), J 57-41 (1), J 57-47 (7), J 57-50 (2), J 57-52 (2), J 57-132 (1), J 57-134 (2), J 57-136 (1), J 60-170 (4), J 77-3 (2), J 77-142 (1), J 77-144 (3), J 77-157 (4), J 77-161 (2), J 77-168 (1), J 77-176 (1), J 77-180 (1). SOSC: AB 16-650R (5), AB 16-655D (2), AB 16-655F (1), AB 16-656A (4), AB 16-656F (4), AB 16-6560 (1), AB 16-656Q (1); ARGO 11-48 (1); ALAMINOS 14-213 (1), J 12-260 (1), J 12-268 (2), J 30-128 (1), J 30-167 (1), J 50-146 (1), J 50-162 (1), J 60-183 (2), UND 46-118 (1); WASHINGTON 45-23 (1), 45-46 (1), 46-11 (1). UMML: 24511 (1), 24530 (1), 28005 (1). ZMUC: D 3548 V (4), D 3548 VII (15). Scopelarchoides signifer Johnson 1974 Distribution. — Scopelarchoides signifer is limited to the Indian and Pacific oceans (fig. 45). Scopelarchoides signifer occurs throughout the Indian Ocean and the 168 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 169 semi-isolated seas of the Indo-Malayan Archipelago. Most records in the Pacific are from a relatively narrow band along the equator. Scopclarchoides signifer does not occur in the central North Pacific but is known from 10 larvae taken near 24.5° to 25° S, 155° W, in the central South Pacific. Material Examined.— Johnson (1974c, p. 152) lists 208 (4.8-104.6 mm SL) specimens from 76 collections. An additional 27 (10.0-78.0 mm SL) specimens from 15 collections are listed here. INDIAN OCEAN: SOSC: AB 6-336B (1), AB 6-337A (1), AB 6-338A (1), TV 4-171 (2). PACIFIC OCEAN: SOSC: ELT 30-2102 (2). UH: TC 47-69 (1). IOAN: V 5117, 21-22 October 1961 (2), V 5117, 22 October 1961 (1), V 5139 (2), V 6033 (1), V 6429, 3 May 1971 (5), V 6429, 5 May 1971 (4), V 6469 (1), V 6490, 25 June 1971 (2), V 6490, 26 June 1971 (1). Scopelarchus Alcock 1896 Scopelarchus analis (Brauer, 1902) Distribution. — Scopelarchus analis is a circumglobal warm- water species occur- ring throughout the North and South Atlantic, Indian Ocean, and insular west- ern Pacific. In the Pacific Ocean virtually all records of S. analis are from central water areas of the North and South Pacific. Scopelarchus analis is known from the Transition Region off California and Baja California, but it is not known from the Transition Region off Chile. Scopelarchus analis appears to be largely excluded from the area of Pacific Equatorial Water (fig. 46). Material Examined.— Johnson (1974c, pp. 170 to 173) lists 602 (6.0-126.3 mm SL) specimens from 275 collections. An additional 815 (6.2-98.5 mm SL) speci- mens from 290 collections are listed here. ATLANTIC OCEAN: IOAN: AK 820 (1), 826 (1), 829 (1), 1009 (1). UMML: 19993 (1), 20003 (1), 20010 (1), 22843 (1), 22971 (1), 24179 (3), 24300 (1), 24575 (1), 24720 (1), 26326 (6). USNM: 113298 (2), 196068 (1); ACRE4-5B (1). WHOI, RHB: 801 (3), 803 (1), 805 (1), 970 (1), 1013 (1), 1018 (1), 1023 (1), 1046 (1), 1047 (10), 1107 (1), 1200 (1), 1254 (3), 1257 (2), 1258 (1), 1267 (1), 1281 (2), 1297 (13), 1305 (2), 1314 (2), 1315 (1), 1423 (1), 1427 (2), 1431 (2), 1438 (5), 1441 (1), 1505 (2), 1509 (5), 1511 (1), 1520 (1), 1718 (2), 1721 (1), 1728 (7), 1729 (2), 1730 (1), 1733 (1), 1735 (1), 1736 (3), 1737 (9), 1869 (1), 1873 (2), 1884 (1), 1888 (8), 1889 (6), 1890 (10), 1891 (2), 1892 (3), 1893 (12), 1894 (5), 1895 (12), 1896 (10), 1897 (3), 1898 (1), 1899 (3), 1900 (4), 1901 (2), 1902 (9), 1903 (2), 1904 (2), 1906 (2), 1907 (2), 1908 (5), 1911 (1), 1914 (3), 1921 (1), 1923 (1), 1928 (1), 1929 (2), 1934 (3), 1935 (2), 2000 (20), 2001 (27), 2002 (6),2003 (3), 2004 (26), 2005 (3), 2006 (6), 2007 (44), 2008 (8), 2011 (8), 2012 (1), 2013 (3), 2014 (4), 2015 (8), 2016 (3), 2017 (1), 2020 (1), 2021 (8), 2027 (2), 2029 (2), 2034 (1), 2037 (5), 2042 (1), 2043 (1), 2045 (1), 2046 (1), 2050 (2), 2055 (1), 2059 (1), 2061 (2), 2065 (1), 2066 (3), 2071 (6), 2082 (1), 2086 (1), 2090 (3), 2095 (6), 2100 (2), 2101 (1), 2105 (2), 2109 (1), 2111 (8), 2112 (9), 2114 (1), 2117 (5), 2118 (15), 2213 (2), 2216 (1), 2218 (1), 2225 (1), 2226 (2), 2230 (1), 2231 (1), 2234 (4), 2235 (6), 2236 (2), 2245 (1), 2248 (1), 2255 (5), 2262 (1), 2263 (1), 2265 (1), 2266 (4), 2776 (2), 2278 (1), 2281 (4), 2282 (3), 2285 (2), 2286 (6), 2288 (1), 2289 (2), 2295 (2), 2546 (1), 2549 (2), 2551 (1), 2553 (3), 2556 (2), 2560 (1), 2562 (1), 2565 (2), 2901 (5), 2904 (1), 2906 (4), 2907 (1), 2908 (1), 2909 (1), 2912 (1), 2914 (2), 2917 (2), 2918 (1), 2919 (3), 2921 (2), 2922 (3), 2923 (3), 2924 (3), 2925 (11), 2928 (2), 2935 (1), 2948 (1), 2950 (1), 2961 (1), 2969 (1), 2971 (1), 2973 (1), 2978 (1), 2979 (1), 2986 (1), 2988 (2), 2991 (1), 2992 (1), 2993 (4), 2995 (2), 2996 (2), 2997 (2), 2999 (2), 3000 (1), 3002 (1), 3005 (1), 3008 (2), 3009 (1), 3012 (2), 3014 (1), 3015 (2), 3018 (1), 3019 (2), 3102 (1), 3103 (2), 3106 (1), 3110 (1), 3112 (1), 3117 (1), 3120 (2), 3124 (1). ZMUC: D 1228 II (4), D 1230 III (2), D 1324 V (1), D 1330 (1), D 3515 V (2), D 3515 VIII (1), D 3534 II (5), D 3536 I (1). INDIAN OCEAN: IOAN: V 5220 (3). SOSC: AB 3-14 (9/9/63) (1), AB 3-14 (9-10/9/63) (1), AB 6-344A (1), AB 6-351D (2). PACIFIC OCEAN: IOAN: B 74 (1), V 4261 (1), V 6429 (1971): 3 May (6), 5 May (2), 6 May (1); V 6469 (1971): 2 June (1), 5 June (1); V 6490 (1); V 6493 (1971): 5 July (1), 5-6 July (1). SIO: 70-314 (1), 70-329 (1). SOSC: RV WASHINGTON: 75-137 (2). UH: 70/7/24 (1), 70/12/7 (2), 70/12/9 (1), 71/2/2 (8), 71/2/3 (1), 71/2/6 (3), 711218 (1), 71/2/9 (4), 71/2/12 (1), 71/3/1 (3), 71/3/8 (2), 71/3/11 (1), 71/6/2 (1), 71/6/5 (3), 170 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 171 71/6/11 (1), 71/6/13 (2), 71/6/16 (5), 71/6/17 (6), 71/6/20 (1), 71/6/21 (1), 71/6/28 (1), 71/6/31 (3), 71/6/34 (1), 71/9/4 (1), 71/9/6 (4), 71/9/8 (3), 71/10/1 (1), 71/10/2 (2), 71/10/3 (2), 71/10/4 (3), 71/10/5 (8), 71/10/6 (10), 71/10/7 (1), 71/10/8 (8), TC 52-59 (1). ZMUC: D 3659 III (2), D 3683 III (1), D 3718 II (1). SOSC: DES 4-T3-B (1), DES 4-T4-D (1), DES 4-T13-D (1), DES 4-T14-A (1), DES 4-T14-D (1), DES 4-T15-D (1), DES 4-T16-B (1) Scopelarchus guentheri Alcock 1896 Distribution. — Scopelarchus guentheri is a circumglobal warm- water species occurring in all three oceans (fig. 47). In the North Atlantic S. guentheri is known only from the Caribbean Sea and from in or south of the boundary area between North Atlantic Central Water and South Atlantic Central Water. Scopelarchus guentheri probably occurs throughout the South Atlantic, Indian Ocean, and insular western Pacific. Scopelarchus guentheri apparently occurs throughout the warm-water Pacific, but there exists evidence suggesting that it is more abun- dant in areas peripheral to the central gyral areas and along the equator and less abundant within the central portions of the Pacific central gyral areas (see dis- cussion below and Johnson, 1974c, pp. 229-231). Material Examined.— Johnson (1974c, pp. 182, 183) lists 205 (7.0-119.0 mm SL) specimens from 121 collections. An additional 143 (10.5-115.0 mm SL) speci- mens from 79 collections are listed here. ATLANTIC OCEAN: IOAN: AK 991 (1). UMML: 29326 (1). WHOI, RHB: 801 (1), 2056 (1), 2075 (1), 2076 (1), 2077 (6), 2078 (1), 2081 (1), 2920 (1), 2923 (1), 2924 (2), 2925 (5), 2927 (2), 2928 (2), 2929 (10), 2930 (5), 2931 (2), 2932 (1), 2933 (1), 2934 (4), 2935 (1), 2936 (1), 2938 (1), 2945 (1), 2946 (1), 2949 (1), 2951 (3), 2952 (2), 2954 (1), 2956 (1), 2996 (7). INDIAN OCEAN: IOAN: B 7 (1), V 4562 (1), V 4618 (1), V 4623 (1), V 4638 (1), V 4796 (2), V 4940 (1), V 5247 (1). MCZ: AB 6-333B (1), AB 6-340B (1). SIO: uncat, 24 February 1971 (1). SOSC: AB 3-5 (1), AB 3-6 (1), AB 6-333A (1), AB 6-340B (8), AB 6-341B (1), AB 6-342A (3), AB 6-344A (1). PACIFIC OCEAN: IOAN: V 3658 (1), V 4291 (1), V 5086 (1), V 5094 (1), V 5139 (1), V 5153 (1), V 6493 (2). NMFS (LJ): J 60-42 (1), J 77-38 (1), J 77-129 (1), J 20.145 (2), J 31.135 (1), J 7205 (130.90) (2), TC 51-53 (3), TC 51-37 (2), TC 51-66 (1), TC 51-74 (1), TC 51-81 (1). SIO: 73-139 (9), 73-142 (2), 73-166 (1), 74-40 (1). SOSC: J 60-56 (1); R/V WASHINGTON: 45-133 (1), 75-73 (1). UH: 70/6/4 (1), 71/2/14 (1), 71/6/22 (2), 71/10/6 (1). Scopelarchus michaelsarsi Koefbed 1955 Distribution. — Scopelarchus michaelsarsi is a tropical-subtropical species occur- ring in all three oceans (fig. 48). Scopelarchus michaelsarsi is now known from the eastern North Atlantic, but most Atlantic records for this species are from the western or central North Atlantic. Scopelarchus michaelsarsi apparently does not occur in the eastern Pacific — all Pacific records for this species are west of 150° W. Material Examined.— Johnson (1974c, p. 192) lists 69 (12.0-101.5 mm SL) specimens from 53 collections. An additional 106 (8.5-70.0 mm SL) specimens from 64 collections are listed here. ATLANTIC OCEAN: WHOI, RHB: 801 (1), 1051 (1), 1101 (1), 1293 (1), 1294 (9), 1297 (4), 1307 (-1), 1309 (1), 1728 (2), 1729 (1), 1733 (2), 1737 (10), 2015 (2), 2076 (1), 2082 (1), 2111 (2), 2113 (1), 2903 (1), 2906 (1), 2909 (1), 2911 (1), 2913 (1), 2919 (1), 2921 (1), 2924 (1), 2925 (2), 2926 (2), 2929 (1), 2932 (2), 2933 (2), 2936 (1), 2938 (2), 2939 (1), 2943 (1), 2944 (1), 2945 (1), 2949 (1), 2989 (1), 2991 (1), 2994 (1), 29% (1), 3002 (1), 3007 (1), 3009 (1), 3010 (1), 3014 (1), 3018 (2), 3019 (1). INDIAN OCEAN: SOSC: AB 6-340B (4), TV 5-190 (2). PACIFIC OCEAN: IOAN: V 5117 (1), V 6469 (1). UH: 70/6/4 (1), 711218 (2), 71/6/1 (1), 71/6/18 (1), 71/6/22 (1), 71/6/28 (1), 71/6/30 (1), 71/6/31 (2). ZMUC: D 3680 VII (1), D 3712 III (2), D 3788 I (1). 172 173 174 FIELDIANA: ZOOLOGY Scopelarchus stephensi Johnson 1974 Distribution. — Scopelarchus stephensi is endemic to the central gyral area of the North Pacific Ocean (Johnson, 1974c, p. 198, fig. 58). Johnson (1974c, pp. 198, 235) lists 27 (14.8-62.0 mm SL) specimens from 13 collections. No additional material has come to hand. For purposes of comparison with S. michaelsarsi, the distribution of S. stephensi, as known, is replotted on Figure 48. ZOOGEOGRAPHY AND EVOLUTION Zoogeography is and has been plagued by two major problems: (1) an over- whelmingly large but frequently spotty, discontinuous, and questionable data base and (2) an all too common lack of agreement among zoogeographers on the general concept, purpose, and best methodology for distributional studies (e.g., Briggs, 1974a; Brundin, 1972a; Croizatetal., 1974; Darlington, 1970). Common to all zoogeographic studies is a basic question: Why does this animal species occur here and not elsewhere? Cohen (1973) describes two distinct approaches in attempts to answer this question: ecological zoogeography vs. historical zoogeography. Ecological zoogeography is concerned with fitting animal distribution patterns to the dis- tribution of environmental parameters. The parameters studied may be physical, chemical, or biological in nature, but the underlying assumption is one of ecological association. The scale of study may range from populations and single species to communities and faunas, but the underlying belief is that individual and recurrent distributional patterns exhibited by the organisms parallel and are causally related to the distribution of important environmental parameters. Historical zoogeography is concerned with fitting the present-day distribution of species in a monophyletic lineage to the presumed evolutionary history of that lineage. This is at base a group-oriented approach that depends upon prior systematic and phylogenetic study of the taxon in question. The unifying factor is the deduced interrelationship of the constituent species of a taxon. The species themselves may occupy quite different kinds of habitats and belong to strikingly distinct ecological assemblages. Comparison of patterns exhibited by similarly studied taxa may lead the historical zoogeographer to put forward broad prop- ositions concerning the history of large faunal units. Differences between the two approaches might partially be characterized as a difference in the scale of study — the scale of the here and the elsewhere (Ball, 1975, p. 408). Ecological zoogeography is largely concerned with studies at the scale of local species assemblages in the here and now. Historical zoogeography typi- cally has a broader spatial and temporal scale and is concerned with the world distribution of single taxa and the deduced history of recognizable general dis- tribution patterns. In practice the results of the two approaches appear very different (see Banarescu, 1975, pp. 6-9, for additional comments on "causal" zoogeography), and these differences have led one prominent historical zoogeographer to exclude ecological zoogeography from the discipline of zoogeography (Briggs, 1974a, p. 5). In recent years there have been at least three distinct and competitive ap- proaches to historical zoogeography. The "traditional" school, typified by Mat- thew (1915), Darlington (1957, 1965, 1970), and Briggs (1974a, 1974b), can be characterized as the center-of-origin/dispersal approach to historical zoogeo- JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 175 graphic analysis. The essence of this approach is the search for evolutionary centers, the centers of origin, in which competitively dominant species evolve. Rules for finding such centers are listed in Ebeling (1962, p. 148). Competitively superior species (equivalent to "dominant" species and usually equivalent to most recently evolved species) are seen by this school as arising in the centers of origin and, by virtue of their competitive superiority, spreading out (via disper- sal) from the centers, replacing and extinguishing or forcing into relictual distri- butions earlier-evolved species. A very clear example of this approach to zoogeographic explanation is Gibbs' (1969) account of the distributional history of Stomias, discussed below. A corollary of this approach is the hypothesis that within a monophyletic lineage the more primitive species (roughly combining the notions of earlier-evolved and plesiomorph [sensu Hennig, 1966a] species) will be found in areas geographically (or ecologically) peripheral to the evolu- tionary centers; the more advanced species will be found in the areas of the evolutionary centers. A second school, typified by Hennig (1966a, 1966b), Brundin (1966, 1972a, 1972b), and Nelson (1969b), can be characterized as the cladistic approach to historical zoogeographic analysis. This approach is dependent upon a prior phylogenetic analysis of the group being studied, and that analysis must be performed utilizing cladistic methodologies (Hennig, 1966a; Brundin, 1966). Given the results of this analysis and the knowledge of the present-day distribu- tion of species in the group, the method in essence is an effort to determine the simplest possible (fewest dispersal events) direct correspondence between present-day distribution of species and the sister-group relationships of species within the group. That is, the inferred distributional history is based on the deduced interrelationships of members of the group studied. An obvious corol- lary of this approach is stated by Brundin (1972b, p. 2): ". . . [a] fundamental biogeographical principle [is] that the primitive group at least primarily is closer to the area once occupied by the ancestral species than the comparatively deriva- tive sister group." A third school, typified by Croizat et al. (1974), Nelson (1974), and Rosen (1975), can be characterized as the generalized-track/vicariance approach to his- torical zoogeographic analysis. The essence of this method was described by Croizat et al. (1974, pp. 265, 266): "If a given type of geographical distribution (individual track) recurs in group after group of organisms, the region delineated by the coincident distributions (generalized track) becomes statistically and, therefore, geographically significant . . . ." In the view of this school complexity in distributional patterns has been introduced by a continuing temporal se- quence of vicariance events (i.e., subdivision of ancestral biotas as a result of changing geography) and by subsequent dispersal modifying (in the case of some groups) earlier vicariant patterns. Thus, in general, vicariant events have led to geographic division and differentiation of biotas and the multiplication of numbers of species, whereas dispersal events have resulted in sympatry and the possibility of interspecific interactions (with the possible concomitants of com- petitive exclusion, ecological differentiation, and extinction). Dispersal events are recognized as having occurred and indeed may be (hypothetically) called upon when all else fails in explaining the distribution of exceptional species, but dispersal is viewed as difficult to interpret and not of overriding importance in explaining the distribution of complex biotas (see McDowall, 1978, for an exten- 176 FIELDIANA: ZOOLOGY sive criticism of this viewpoint). This school rejects the notion of a center of origin as an essential initial premise and therefore takes no firm position on the question of where we should expect to find earlier-evolved and later-evolved species relative to such centers (cf., Nelson, 1969b vs. Nelson, 1974). Ball (1975), Platnick & Nelson (1978), and Rosen (1978) explore the possibilities for combin- ing the cladisric and generalized-track/vicariance approaches to zoogeography (for a recent overview, see Ferris, 1980). Due to seemingly mutually exclusive initial assumptions and hypotheses, no one has yet suggested an acceptable combination of the center-of-origin/dispersal school with either of the other two schools. Most attempts at explaining distribution patterns exhibited by open ocean organisms have been couched in ecological, not historical, terms. Attempts at explaining the distribution of species of a given midwater group with respect to the deduced evolutionary history of that group for the most part suffer from an incomplete or inadequately documented knowledge of relationships between the species studied and/or an incomplete knowledge of the distribution of vari- ous included species. An additional difficulty is that some of the most exhaustive studies of the relationship between groups of midwater fishes, e.g., Paxton (1972) on myctophids, Baird & Eckhardt (1972) on sternoptychids, Weitzman (1967, 1974) on various stomiatoids, Pietsch (1974) on oneirodids, among others, deal with interrelationships of higher taxa (rank of genus or higher) and are for the most part inapplicable to open ocean zoogeographic studies. This inapplica- bility stems from the fact that higher taxa of midwater organisms occurring in the warm-water ocean are typically cosmopolitan or nearly so within the approxi- mate limits of 40° N to 40° S. Thus, as McGowan (1971, p. 11) has emphasized, it is studies at the species and subspecific level that are of greatest importance for interpreting distributional patterns exhibited by open-ocean organisms. Of those attempts that have been made to interpret present-day distributions of midwater fish species in terms of presumed evolutionary history, most have been phrased more or less vaguely in terms of the center-of-origin/dispersal approach to zoogeographic explanation (e.g., Fraser-Brunner 1949; Bolin, 1959; Ebeling, 1962; Gibbs, 1969; Johnson, 1974c; Bertelsen et al., 1976). Typifying this approach is Gibbs' (1969) account of the evolutionary history of Stomias. In this view the history of Stomias is a running competitive battle for possession of the more productive areas of the world ocean. The most recently evolved species of Stomias, those in the S. boa group, are regarded as having proven competitively superior ("dominant" sensu Briggs, 1974a, p. 254) to species in the earlier- evolved S. brevibarbatus, S. colubrinus, and S. nebulosus species groups. Thus the competitive superiority of species in the S. boa group has forced species in the older groups into relictual distributions in unfavorable or specialized environments — examples given include areas with marked oxygen-minimum layers in the Atlantic and Pacific, central gyral areas, and near-shore areas. Gibbs (1969, p. 20) believes that the competitive pressure on S. danae may have been (and is) so severe that ". . . S. danae may even be on the verge of extinc- tion." To my knowledge no one has yet attempted direct application of the cladistic methodology to a distributional study of any midwater fish group. Application of an essential part of the generalized-track/vicariance approach, viz., the deter- mination of generalized tracks (although not referred to in existing literature by JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 177 that name), has become increasingly common in studies of distribution patterns of open-ocean organisms. In fact the recognition of concordant restriction ( = generalized tracks) of species to a given subarea of the world ocean is at present the only widely available method for recognizing distinct assemblages of open- ocean species. Such assemblages are fully the open-ocean equivalent of the "biotas" of Croizat et al. (1974), but the study of such assemblages by workers on the open-ocean fauna has been entirely ecological in orientation. Aside from more or less vague references to the possible effects of closure of the Tethys Seaway on the distribution of certain midwater organisms (e.g., Andriashev, 1962; Brinton, 1962; Crane, 1966; Gibbs, 1969; Goodyear, 1970; Judkins, 1972), there has been, for fairly obvious reasons, no attempt to explain the present-day distribution of such assemblages in terms of subdivision (vicariation) of ancestral assemblages (see Croizat et al., 1974). Most attempts to explain distribution patterns exhibited by open-ocean or- ganisms have involved the ecological approach to zoogeographic analysis. Even the few attempts to explain present-day distribution patterns in terms of an inferred history of those patterns commonly have been based on presumed historical changes in the distribution of important environmental parameters, e.g., Brinton's (1962, pp. 245-252) explanation of the origin of biantitropical distributions or McGowan's (1971, p. 53) belief that the development of present-day distributional patterns must be explained in part in terms of ". . . ancient circulation systems and water-mass structures." A large number of physical, chemical, and biological parameters have been used, singly or in combination, to explain the observed distributions of various species or groups of species (summary outlines of such attempts are provided by Brinton, 1962; Ebeling, 1962; Johnson & Brinton, 1963; Ebeling, 1967; Ebeling et al., 1970; Parin, 1970; Baird, 1971; McGowan, 1971, 1974; Johnson, 1974c; Bad- cock & Merrett, 1976, 1977, among others). Included among these factors are the following: CURRENTS: Including confinement of indigenous populations by gyral sys- tems and the boundary effects of convergences and divergences: Bruun, 1958; Wisner, 1959; Ebeling, 1962; Reid, 1962; Wickett, 1967; Frost, 1969; McGinnis, 1974; and Reid, 1977. TEMPERATURE: Brinton, 1962; Nafpaktitis, 1968; Backus et al., 1969; Parin, 1970; and Briggs, 1974a. DENSITY: Pickford, 1946. OXYGEN: Gibbs & Hurwitz, 1967; Longhurst, 1967; and Johnson & Glodek, 1975. BIOLOGICAL PRODUCTIVITY: Bogorov, 1958; Ebeling, 1962; Backus et al., 1965; Roper, 1969; Baird et al., 1973; Briggs, 1974a; Fleminger & Hulsemann, 1974; and Pietsch, 1974. WATER MASSES: Brinton, 1962; Johnson & Brinton, 1963; Ebeling, 1962, 1967; Backus et al., 1965; Ebeling etal., 1970; McGowan, 1971, 1974; Wormuth, 1971; Kobayashi, 1973; McGowan & Williams, 1973; and Bertelsen et al., 1976. Although the water-mass hypothesis (see below) is implicitly a multifactorial approach (Ebeling, 1962), there have also been explicit attempts at using mul- tivariate techniques to assess the relative importance of various environmental features on the distribution of midwater fish species (Ebeling etal., 1970; Ebeling etal., 1971). 178 FIELDIANA: ZOOLOGY Since the early 1960's, the leading paradigm in open ocean zoogeographic studies has been the so-called "water-mass hypothesis" — the association of the distributional boundaries of oceanic fish species with water-mass boundaries (see Sverdrup et al., 1942, and Ebeling, 1962, for discussions of water-mass identification, properties, and distribution). The distributional limits and/or areas of maximum abundance (e.g., Craddock & Mead, 1970; McGowan, 1971; Barnett, 1975) have been shown to conform closely with the area underlain by a given water mass or water masses. The result has been that in recent years vir- tually all open-ocean zoogeographers have compared the distributions of their organisms to water-mass boundaries and have discussed oceanic zoogeography in terms of water-mass regions (see Johnson, 1974c, pp. 221, 222, for a represen- tative list of papers utilizing, in part, the water-mass hypothesis). Most compari- sons have been made with respect to the upper water masses (Ebeling, 1962, p. 145), but the distributions of certain species have been associated with the distri- bution of the deeper intermediate or even deep water masses (e.g. , Ebeling, 1975) . Opposition has arisen to the water-mass hypothesis. For example, Briggs (1974a, pp. 335-338), in reviewing a limited (and highly selective) sample of the open-ocean literature, rejects (p. 338) the water-mass hypothesis and opts for temperature as the single factor most critically influencing the shape of open- ocean distribution patterns. Briggs (1974a, p. 337) bases an important part of his argument on the work of Parin (1970) but fails to note Parin's (1970, p. 128) emphasis on the important distributional distinctions between what Parin refers to as "planktonic" vs. "nektonic" fishes. Most opposition to the water-mass hypothesis stems from the belief that it is too simplistic (Johnson, 1974c, p. 222), that patterns of midwater fish distribution are too numerous and too complex to be explainable solely in terms of water-mass distribution (e.g., Backus et al., 1970; Baird, 1971; Hartman & Clarke, 1975; Jahn & Backus, 1976). Fit with respect to the water-mass hypothesis unquestionably varies by ocean, being worst for the Atlantic (Backus et al., 1970; Krefft, 1974; Fasham & Angel, 1975; Backus et al., 1977) and best for the Pacific (e.g., McGowan, 1971, 1974; McGowan & Williams, 1973). There is growing evidence that other factors, e.g., productivity (Roper, 1969; Pietsch, 1974), dissolved oxygen (Johnson & Glodek, 1975), temperature/light effects (e.g., Paxton, 1967; Badcock & Merrett, 1977), may be associated with substantial departures of the distributions of individual species or groups of species from concordance with water-mass regions. There exists in the open-ocean literature a growing emphasis on comparing the distribution of individual species with the distribution of other oceanic species and then comparing the distribution of concordantly restricted as- semblages of species with the distribution of selected environmental features (e.g., Parin, 1970; McGowan, 1971, 1974; Johnson & Glodek, 1975). The recogni- tion of the species assemblages uses the same methods as the determination of generalized tracks as described by Croizat et al. (1974). The generalized tracks or "recurrent" patterns (Fager & McGowan, 1963) thus determined are apparently both relatively few in number, are repeated from one taxonomic group to another, and are repeated virtually irrespective of trophic level or guild. Mc- Gowan (1971, 1974, 1977) has summarized the available information on discerni- ble generalized tracks, which he refers to as "ecosystems," in the Pacific Ocean. His suggestion is that the concordance in distribution of oceanic species from many taxonomically distinct groups, the recurrent assemblages of oceanic JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 179 species, indicates the great importance of biological interaction in determining the major patterns of distribution of oceanic species. Maintenance of these recurrent distributional patterns is seen not only in terms of those processes that maintain the shape of water-mass patterns (McGowan, 1971, p. 50) but also in terms of evolutionary co-adaptation (Mc- Gowan, 1974, p. 16). The perceived character and structure of these "community-ecosystems" is summarized by McGowan (1972, pp. 16-23; 1977, p. 425). The definition, identification, and study of communities in the open ocean is an exceedingly complex problem (see Angel, 1977). The scale of study rep- resented in the literature varies from fine, e.g., the "community" of organisms represented in Alepisaurns stomachs described by Haedrich & Nielsen (1966), to medium, e.g., the offshore middepth vs. offshore deep "communities" of Ebel- ing et al., 1971, to the very broadscale "community-ecosystems" discussed by McGowan (1971, 1974) and his students (e.g., Barnett, 1975; Shulenberger, 1977). Adding to the complexity are the very real quantitative and qualitative differences in basic processes between ecosystems (McGowan, 1974; McGowan & Williams, 1973) and between "ecosystem" assemblages vs. "ecotonal" as- semblages (McGowan, 1974, 1977). There exist well-reasoned arguments that deny the reality of communities as discrete and coherent natural units formed by species bound together by co-adaptation (Whittaker & Woodwell, 1972). Krefft (1974, pp. 233, 234) argues ". . . actually each species develops its own specific [distribution] pattern, depending on biotic and abiotic environmental factors as well as its distributional history . . . however, in spite of the diversity of specific patterns, the threads of the web join to form larger patterns, which characterize well-defined faunal communities." No attempt is made in this paper to resolve conflicts in concept and definition of oceanic communities, but an attempt is made to discuss observed distribution patterns in terms relevant to these conflicts. In the remainder of this chapter I have three major purposes: 1) To discuss, where possible, the distributions of evermannellid and scopelarchid species with reference to recognizable recurrent distributional pat- terns of open-ocean species. Because of the relative abundance of evidence, particular attention is paid to discussion of species occurring in either the eastern tropical Pacific or in central gyral areas of the North and South Pacific. 2) To discuss the categories of evidence that, as Barnett (1975) has argued, at least some of these generalized tracks, these oceanic "ecosystems" (sensu McGowan, 1977), broadly correspond to the concept of biologically accommo- dated communities in which interspecific interactions are more important in the regulation of community structure than variation in physical and chemical parameters of the environment. 3) To support my belief that study of open-ocean systems may lead to partial resolution, at least, of the seemingly wide gap in purpose, methodology, and results separating historical from ecological approaches to zoogeography. Distribution-Pattern Categories There exist numerous schemes for describing and categorizing distributional patterns exhibited by midwater fishes. Four commonly used bases for dividing 180 FIELDIANA: ZOOLOGY observed distributional patterns are discussed in this paper: (1) division by in- shore vs. offshore association; (2) division into "cold-water" vs. "warm-water" areas; (3) division by ocean basin (i.e., Atlantic, Indian, Pacific); and (4) division with respect to water mass regions. Table 23 summarizes the distribution of evermannellid and scopelarchid species with respect to these broad distri- butional categories. Table 23. Classification of distributional patterns exhibited by evermannellid and scopelarchid species in terms of broad distributional patterns discussed in text. Informa- tion given in parentheses gives geographic and water mass region information. I. "Cold-water" species Evermannellidae (0/7) Scopelarchidae (5/17) Benthalbella dentata (North Pacific: Pacific Subarctic and Transition Region) Benthalbella elongata (Southern Ocean: Subantarctic and Antarctic) Benthalbella linguidens (North Pacific: Transition Region) Benthalbella macropinna (Southern Ocean: Subantarctic and Antarctic) Scopelarchoides kreffti (Southern Ocean: known only from South Atlantic but presumed circumglobal in southern Transition Region) II. "Warm- water" species A. Species limited to a single ocean basin Evermannellidae (2/7) Evermannella ahlstromi (Pacific: most records from eastern Pacific Equatorial) Evermannella megalops (Pacific: Eastern and Western South Pacific Central) Scopelarchidae (4/17) Rosenblattichthys volucris (Pacific: most records from eastern Pacific Equatorial) Scopelarchoides climax (Pacific: Eastern and [presumably] Western South Pacific Cen- tral) Scopelarchoides nicholsi (Pacific: most records from eastern Pacific Equatorial) Scopelarchus stephensi (Pacific: Eastern and Western North Pacific Central) B. Species limited to two ocean basins Evermannellidae (1/7) Coccorella atrata (Indian and Pacific: most records from Indian Equatorial or Pacific Equatorial) Scopelarchidae (2/17) Rosenblattichthys alatus (Indian and Pacific: most records from Indian Equatorial or Pacific Equatorial) Scopelarchoides signifer (Indian and Pacific: most records from Indian Equatorial or Pacific Equatorial) C. Species occurring in three ocean basins Evermannellidae (4/7) Coccorella atlantica (subtropical in all three oceans) Evermannella balbo (cooler more productive parts of North and South Atlantic; Mediterranean Sea; presumed to be circumglobal in southern Transition Region) Evermannella indica (tropical-subtropical in all three oceans) Odontostomops normalops (tropical-subtropical in all three oceans) Scopelarchidae (6/17) Benthalbella infans (tropical-subtropical in all three oceans) Rosenblattichthys hubbsi (subtropical in all three oceans, not known from South Pacific) Scopelarchoides danae (tropical in all three oceans) Scopelarchus analis (tropical-subtropical in all three oceans) Scopelarchus guentheri (largely tropical in Atlantic, tropical-subtropical in Indian and Pacific oceans) Scopelarchus michaelsarsi (tropical-subtropical in all three oceans) JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 181 A major problem in discussing distributional patterns exhibited by everman- nellid and scopelarchid species is that there exists no adequate picture of vertical distribution for any of the 24 species. Although available information, largely from open-net hauls, suggests the expected (e.g., Vinogradov, 1970, p. 93) verti- cal segregation of life history stages, with small larvae occurring high in the water column (except, possibly, Benthalbella dentata, see Johnson, 1974c, p. 70), for the adults of most species the available information suggests only a wide- spread vertical distribution in the upper 1,000 m. (1) Division by InshorelOffsfwre Association. — A number of groups of midwater organisms contain species or groups of species that are largely neritic, closely associated with inshore waters near continental or insular land masses. Such species may spend their entire life cycles in inshore waters or may be pelagic as larvae and juveniles and "pseudo-oceanic" (Bertelsen et al., 1976, p. 101) or bottom-associated as adults. Apparently representative of such land-associated forms are the myctophids Benthosema panamense (Wisner, 1976); Diaphus taaningi (Nafpaktitis, 1968); Diaphus adenomus (Nafpaktitis et al., 1977); Diaphus coeruleus, D. garmani, and D. watasei (Nafpaktitis, 1978); the neoscopelids Neoscopelus and probably Solivomer (Nafpaktitis, 1977); the melamphaid Melamphaes acanthomus (Ebeling, 1962); the sternoptychid (sensu Weitzman, 1974) genera Argyripnus (Struhsaker, 1973) and Polyipnus (Baird, 1971); a substantial number of notosudid species (Bertelsen et al., 1976); and a number of euphausiid species (Brinton, 1962). All evermannellid and scopelarchid species are pelagic and oceanic. Despite an apparent association of Scopelarchoides danae with insular and continental land masses (fig. 43, and see account of S. danae above), all speci- mens, larvae, juveniles, and adults, have been taken with pelagic gear. (2) Division into "Cold-Water" vs. "Warm-Water" Areas. — A major division in the distribution of open-ocean organisms is between the "cold-water" vs. "warm-water" faunal areas (Ekman, 1967, pp. 324, 325). Dividing lines corre- spond with the North (Pacific only) and South Subtropical Convergences (Dea- con, 1963; Alverson et al., 1964; McGowan & Williams, 1973), ca. 40° N and 40° S except in eastern-boundary-current areas (fig. 49). Included in the southern cold-water area are the regions of Subantarctic and Antarctic Water in the Southern Ocean. Included in the northern cold-water area is the region of Pacific Subarctic Water (Sverdrup et al., 1942, p. 740). A number of "cold-water" species enter hydrologically intermediate but biologically distinctive "Transition Regions" in the eastern-boundary-current areas of the North and South Pacific (e.g., Brinton, 1962; Lavenberg & Ebeling, 1967; Craddock & Mead, 1970; Ebel- ing et al., 1970; McGowan, 1971). The "Transition Region" fauna of the eastern South Pacific (e.g., Craddock & Mead, 1970) at least in part is included in or coincides with the "Subtropical Convergence" or "Transition-Zone" faunas of Brinton (1962, p. 202), Gibbs (1968, p. 3), and McGowan (1971, p. 44). The distinctiveness of the southern cold-water fauna is documented in Andriashev (1962), Brinton (1962), David (1962), Bussing (1965), Gibbs (1968), Parin (1970), Briggs (1974a), and Backus et al. (1977), among others. The distinctive North Pacific boreal fauna is documented in Brinton (1962), Ebeling (1967), McGowan (1971, 1974), and McGowan & Williams (1973). There is no distinct boreal fauna in the North Atlantic (e.g., Backus et al., 1977). In this paper a species is assigned to the category "cold-water species" if the known range of adults does not enter the region of any central or equatorial 182 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 183 water mass. Five scopelarchid spedes apparently meet this definition (table 23), and I have nothing to add to the discussion of these species presented by Johnson (1974c). Three of the 19 warm-water species (Coccorella atlantica, fig. 24, Evermannella balbo, fig. 32, and Scopelarchus guentheri, fig. 47) occur in the Transi- tion Region of the Southern Ocean but, in each case, most records are from warm-water areas. The possible existence of a separable Subtropical- Convergence-region population of Scopelarchus guentheri is discussed by Kashkin (1977) in relationship to McGinnis' (1974) study of circulation in the Pacific Subantarctic sector of the Southern Ocean. (3) Division by Ocean Basin. — Despite the continuity of the world ocean, there are geographic barriers, more or less effective, to panmixis among widely dis- tributed warm-water oceanic species. Most notably these barriers include the American, African, and Indonesian-Australian land masses. Brinton (1975) pro- vides extensive discussion of the four possible "barrier/pathways" in the tropical and subtropical ocean: (1) the Indo-Australian Archipelago, (2) the tip of South Africa, (3) southern Australia, and (4) the tip of South America. Briggs (1960, 1974a) has argued that the so-called East-Pacific Barrier may be effective for many surface and midwater pelagic species. Isolation of populations among ocean basins is known for a number of midwater species, both invertebrates and fishes (e.g., Gibbs, 1968; Mauchline & Fisher, 1969; and see data for Coccorella atlantica, above, fig. 24). Fleminger & Hulsemann (1973), in discussing the distributions of a number of warm-water copepod species, argue that one might expect greater provincialism (separation of populations and species) in equatorial rather than central water forms. Their evidence (Fleminger & Hulsemann, 1973, 1974) indicates that tropi- cal (=equatorial) copepods tend to be restricted to either (1) the Atlantic, (2) Indian Ocean and (at least) western portion of Pacific Ocean, and (3) eastern tropical Pacific. Judkins (1978) shows that the largely equatorial distributions of pelagic decapod shrimp in the Sergestes edwardsii group agree well with Fleminger & Hulsemann's predictions. Among the species of evermannellids and scopelarchids largely restricted to equatorial waters (table 23), one species, Scopelarchoides danae, occurs in all three ocean basins. Three species {Coccorella atrata, Rosenblattichthys alatus, and Scopelarchoides signifer) occur only in the In- dian and Pacific oceans, and only one of these, C. atrata, reaches the eastern tropical Pacific. Among the 24 species in the Evermannellidae and Scopelar- chidae, only three warm-water species are limited to two of the three ocean basins, all three are largely equatorial in distribution, and none occurs in the Atlantic. This agrees with the predictions of Fleminger & Hulsemann. There appear to be very few warm-water oceanic species limited to two ocean basins in which one member of the pair is the Atlantic, but there exist a fair number of oceanic species limited to the Indo-Pacific area. 9 Gibbs & Craddock (1973, p. 159) 9 A number of warm-water species known from the Atlantic and only one other ocean are known from very few specimens from one or both members of the pair, e.g., Rondeletia tricolor (Paxton, 1974), Bolinichthys distofax, Diaphus adenomus, D. dumerilii, and Lampanyctus photonotus (Nafpaktitis et al., 1977). Others are poorly understood taxonomically, e.g., Lampadena urophaos atlanticus vs. L. u. urophaos (Nafpaktitis et al., 1977). Other species, particularly those that are subtropical (sensu Backus et al., 1977) in the Atlantic and Pacific oceans, may well occur but yet have not been captured in the poorly sampled central water area of the southern Indian Ocean, e.g., Hygophum reinhardti and Diaphus anderseni (Naf- paktitis et al., 1977). There do, however, appear to be some warm- water species that are 184 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 185 Opposite: Fig. 50. Distribution of evermannellid and scopelarchid species exhibiting subtropical distribution patterns. SYMBOLS: O = Coccorella atlantica; 3 = Evermannella megalops; A = RosenblatHchthys hubbsi; © = Scopelarchoides climax; © = Scopelarchus stephensi; # = overlap- ping records for C. atlantica, S. climax, and E. megalops. Shaded bands denote areas transi- tional between water-mass regions (after Sverdrup et al., 1942): 1 = North Atlantic central; 2 = South Atlantic central; 3 = Indian equatorial; 4 = Indian central; 5 = western North Pacific central; 6 = eastern North Pacific central; 7 = western South Pacific central; 8 = eastern South Pacific central; 9 = Pacific equatorial. Dashed lines indicate boundaries (approx.) between areas recognized as mesopelagic faunal regions by Backus et al., 1977: A = Atlantic Subarctic Region; B = North Atlantic Temperate Region; C = North Atlantic Subtropical Region; D = Gulf of Mexico Region; E = Mauritanian Upwelling Region; F = Atlantic Tropical Region; G = South Atlantic Subtropical Region. argue for isolation of the Pacific Ocean population of Eustomias trewavasae from populations in the Atlantic and Indian oceans. Because E. trewavasae is limited to the zone of the southern Subtropical Convergence, these results are in agree- ment with those of Fleminger & Hulsemann. I believe that the possible round- the-world dine discussed above for Evermannella indica is similarly related, at least in part, to geographic barriers to dispersal and gene flow. Further discus- sion of distribution patterns with respect to ocean basins is combined with an account of distribution with respect to water-mass regions. (4) Division with Respect to Water-Mass Regions. — For the warm- water areas of the ocean, it is heuristically useful to distinguish two major categories of distri- bution exhibited by mid-depth oceanic species (e.g., Brinton, 1962; Ebeling, 1962, 1967; Johnson & Brinton, 1963; Johnson, 1974c): 1. Species relatively restricted in distribution, limited to one ocean basin, and limited to all or part of one water-mass region 10 or limited to the region of two adjacent and physically and biologically similar water masses. Among the 19 warm- water species of evermannellids and scopelarchids, six species (table 23), all limited to the Pacific, fit in this category. 2. Species more widespread, with distributions crossing water-mass bound- aries, typically in two or more ocean basins, and exhibiting varying approaches toward warm-water cosmopolitanism (no evermannellid or scopelarchid is cos- mopolitan in the warm- water ocean). Ebeling (1967) discusses two subcategories among widely distributed species to which I add a third. a. Subtropical or central-water species 11 (fig. 50) — species for the most part associated with (or restricted to) the central water-mass regions within the large, subtropical anticyclones of the Atlantic, Indian, and Pacific oceans. Two species treated in this paper, Coccorella atlantica (fig. 24) and RosenblatHchthys hubbsi (fig. 40), fit this subcategory. b. Tropical or equatorial species (fig. 51) — species for the most part associ- ated with (or restricted to) the equatorial water-mass regions of the Indian and Pacific oceans and/or to a relatively productive equatorial zone in the Atlantic exceptional in occurring in the Atlantic and in either the Indian or Pacific Ocean (but not both), e.g., Bolinichthys indicus, Diaphus termophilus, and Symbolophorus rufinus (Nafpaktitis etal., 1977). ,0 Refers to geographic area underlain by a principal upper water mass as depicted by Sverdrup et al., 1942, p. 740. "Termed "central-tropical" species by Ebeling, 1967. 186 JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 187 Ocean (more or less congruent with the "Atlantic Tropical Region" of Backus et al., 1977). Equatorial species can occur in one, two, or all three ocean basins. Four species treated herein, Coccorella atrata (fig. 24), Rosenblattichthys alatus (fig. 40), Scopelarchoides danae (fig. 43), and Scopelarchoides signifer (fig. 45), apparently fit this subcategory. Note that this listing does not include the three equatorial species endemic to the eastern tropical Pacific (table 23). c. Tropical-subtropical species (fig. 52) — species broadly distributed in tropical and subtropical waters, occurring in both central and equatorial water- mass regions (not necessarily throughout a given water- mass region nor in equal abundance over all regions), and (typically) in all three ocean basins. Six species treated herein fit this subcategory: Evermannella indica (fig. 34), Odontostomops normalops (fig. 37), Benthalbella infans (fig. 39), Scopelarchus analis (fig. 46), S. guentheri (fig. 47), and S. michaelsarsi (fig. 48). It is possible the availability of more complete distributional information would increase the number of categories deemed useful for descriptive purposes — Backus et al. (1977) recognize nine major distributional categories to describe the zoogeography of Atlantic myctophid species. However, none of the 19 warm-water species treated herein exhibits a known distribution exactly coin- cident with any other species (some approach this, e.g., Evermannella ahlstromi cf. Rosenblattichthys volucris), and I believe that the number of descriptively use- ful categories is both finite and relatively small. It is certain that one "warm- water" species, Evermannella balbo (fig. 53), fits none of the above categories. Peculiarities in the distribution of £. balbo are discussed in the regional accounts presented below. Distribution of Evermannellid and Scopelarchid Species: Regional Accounts Atlantic Ocean Ten of the 19 warm-water species of evermannellids and scopelarchids occur in the Atlantic Ocean (table 23). None of these species is restricted to the Atlan- tic. Recent papers contributing to our knowledge of distribution patterns ex- hibited by midwater fishes in the Atlantic Ocean include Tortonese (1960), Ebeling (1962), Backus et al. (1965, 1969, 1970, 1977), Harrison (1967), Nafpaktitis (1968), Briggs (1970), Gibbs & Roper (1971), Gibbs, et al. (1971), Rass (1971), Goodyear, Gibbs, et al. (1972), Krefft (1974), Badcock & Merrett (1976, 1977), Parin &Golovan (1976), Backus & Craddock (1977), and Nafpaktitis etal. (1977). A number of authors have proposed systems for division of the Atlantic Ocean into open-ocean zoogeographic regions or provinces (e.g., Ebeling, 1967; Ekman, 1967; Backus et al., 1970, 1977; Parin, 1970; Baird, 1971; Briggs, 1974a; Krefft, 1974). The following discussion is largely in terms of zoogeographic sys- tems proposed by Krefft (1974) and Backus et al. (1977, hereafter abbreviated to B) — the most recent and by far the most synoptic studies of Atlantic Ocean midwater fish distributions. Krefft (1974) divides the Atlantic midwater fish fauna among four major geo- graphic assemblages or distribution patterns. Each of these major groups is further divided into two or more subgroups. The main features of Krefft's clas- sification parallel those in the system proposed by B, and I have attempted a 22 » u ro . it •§ -a ? £&& 188 189 190 FIELDIANA: ZOOLOGY rough listing of corresponding components (table 24). The listing does not in- clude "bathypelagic" or "pseu do- oceanic" groups. The system of Atlantic Ocean zoogeographic regions and provinces proposed by B, based on the study of more than 280,000 myctophid specimens (and an unstated number of specimens of other midwater fish groups) representing more than 105 myctophid species from more than 1,500 stations, represents by far the most thorough, synoptic distributional study of any midwater fish group in any ocean basin. The system includes seven zoogeographic regions (Atlantic Subarctic, North Atlantic Temperate, North Atlantic Subtropical, Gulf of Mexico, Mauritanian Upwelling, Atlantic Tropical, and South Atlantic Subtropi- cal) and incorporates all but the southernmost warm-water Atlantic. Four of the regions (B, pp. 270, 271) are subdivided into two or more provinces (a total of 23 provinces). Nine "distribution patterns" (B, pp. 272, 273) exhibited by Atlantic myctophids are described in terms of these regions and provinces or parts or combinations thereof. The system provides a powerful tool for describing distri- bution patterns exhibited by Atlantic midwater organisms, and I have attempted below to describe the Atlantic distributions of 10 evermannellid and scopelarchid species largely in terms of this system. Despite its usefulness, I believe that a careful reading of B with its companion paper by Nafpaktitis et al. (1977, hereafter referred to as N) reveals a number of inconsistent applications of the system and perhaps failings. These are briefly outlined in the following paragraphs. To be noted first is the fact that neither restriction nor endemism of any species is prerequisite to the recognition of a zoogeographic region, province, or, for that matter, distribution pattern. Of the 82 myctophid species listed (B, p. 267), only one, Lampadena pontifex, is said to be limited to one region, the Mauritanian Upwelling, and (N, p. 182) "apparently limited to the southern province of that region" (this despite the fact that the distribution chart pre- sented for this species [N, p. 181] shows records from the northern province and the adjacent Guinean Province of the Atlantic Tropical Region). No other myc- tophid species listed is endemic or restricted to any zoogeographic region or province. 12 Very few species are restricted to the indicated geographic area of the distribution pattern to which they are assigned. It is not my belief that absolute restriction is the sine qua non for inclusion or exclusion of any species in a given part of any system of zoogeographic regions, provinces, or distribution patterns. The case for relative abundance as a compo- nent of faunal description and differentiation has been made (e.g., Craddock & Mead, 1970, p. 342) and needs no elaboration here. But it seems to me that B have gone too far, that their system, at least as presented, comes dangerously close to ignoring the value of restriction and endemism to our understanding of open-ocean species assemblages (see below under Pacific central- water species). There are several problems thereby introduced. First, there is the real difficulty in fitting distribution patterns exhibited by groups such as the evermannellids and scopelarchids (and indeed most mid- water fish groups) into a zoogeographic system based largely on geographic differences in relative abundance. The samples do not exist that would allow ,2 Mention is also made (N, p. 78; B, p. 284) of an undescribed Symbolophorus sp. also endemic to the Mauritanian Upwelling Region. •a 5 c 5> o OS 1.1 si ss I I 5 o 9 u 6 a u E o £ x> ? 3 z i H 2 01 d. K 01 O fe e x. £ 6 3 0) 0) iflhh _ 3 S 3'R-R- £ X> Xt ■» 3 3 X 3 o t. o <-< ex £> a .3 o c •o a a. (A ts £ o o 60*" eg ( ca ot at ns £ b > 3 "> a £ ~ .3 £ 3 £ 00 5 7 £"w a -3 m : ±: U ra 1 IS"?*! a « -3 ^ « O n "ay q ° S ° *2 $ |~x S s "? 3 '"^ ^J *3 00 Ci c <0 • 5 3 a -3 m S c^ it o *ti ^*^ ii &.<»<% S ° i,t3 air c 73 -X) ?y '«! fe 8 agL-S S « •£ |« £ °; a £P ■n E 13 3 c U 5 S ^ C £ S£ J! 5 sx fc nj a u SC^| z| sc| o;"S § ^s s at x (« ■CC.C «C £■" .. at J5 Ss . .£ 191 192 FIELDIANA: ZOOLOGY meaningful abundance comparisons on a geographic basis for most midwater fish groups for most areas of the world ocean. Second, there is the possibility that exclusion (from consideration) of some distributional records as "waifs" or "expatriates," a course invited and practiced in B and N, may prove seriously misleading (e.g., compare O'Day & Nafpaktitis, 1967, with Karnella & Gibbs, 1977). Third, the system as applied by B has resulted in what seems to me to be remarkable inconsistencies. For example, Lepidophanes guentheri (N, pp. 226- 228), said to be a "tropical" species (N, p. 228; B, p. 274), was in fact taken in abundance not only throughout the Atlantic Tropical Region but also in the North Atlantic Subtropical, South Atlantic Subtropical, Gulf of Mexico, Mauritanian Upwelling, and even the western North Atlantic Temperate Region (N, p. 227; also see authors' comments [N, p. 228] on the distribution of this species). It seems to me that a system that includes Lepidophanes guentheri (N, p. 227) and Diaphus vanhoeffeni (N, p. 156) as examples of the same distribution pattern ("tropical") requires substantial modification. The same objection, i.e., apparent nonobjectivity in the assignment of a given species to one pattern or another, seems more or less appropriate in the cases of Notolychnus valdiviae (N, p. 94), Lobianchia dofleini (N, p. 98), Diaphus dumerilii (N, p. 115), D. lucidus (N, p. 139), D. perspicillatus (N, p. 145), Lampanyctus alatus (N, p. 225), and a number of other species. In fairness, there are also species showing moderate to strong restriction to the distribution patterns assigned them, e.g., Protomyctophum arcticum (N, p. 32), Lampadena urophaos atlantica (N, p. 177), L. speculigera (N, p. 179), and Notoscopelus elongatus (N, p. 254). Fourth, but possibly first in importance, is a conceptual problem implicit in the zoogeographic system proposed by B. Two core elements of this system include the authors' concept of a pelagic region: "From the variety of overlapping distribution patterns, it follows that each pelagic region is faunally distinct with its characteristic assemblage of species whose numbers are in characteristic proportion, its characteristic diversity, and so on" (Backus et al., 1970, p. 196); and a faunal boundary: "By faunal boundary we mean a narrow zone across which there is a relatively rapid change from one constituency of species, or one fauna, to another. ... A principal goal of this zoogeographic system has been to describe the way in which the ocean changes physically at those places in which it changes faunally" (B, p. 269; italics theirs). It appears that the authors are so preoccupied with describing oceanographically separable regions and provinces that the primary purpose of the exercise, i.e., the recognition of different faunas, is sometimes lost. This seems to me to be implicit in the authors' need to set up a dual system of zoogeographic regions and provinces vs. distribution-pattern areas. It is explicit in the authors' own description of how they came to propose the "Azores-Britain Province" — a province whose "boundaries are determined by the boundaries of its neighbors" (B, p. 280). Where is the faunal evidence for the distinctness of this province? I believe that the only widely applicable method for the recognition of oceanic species assemblages lies in the discovery of species (or populations) exhibiting concordantly restricted distributions — a method employed to good advantage in studies of the Pacific (see below). The first step is the discovery of "recurrent" (sensu Fager & McGowan, 1963) distribution patterns. The second step is the study of hydrographic and biological parameters that might define faunal JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 193 boundaries separating such patterns. I believe that, to some extent, B have reversed these steps, and, to that extent, their zoogeographic system is lacking. I conclude by affirming the great utility of their system for describing the distribu- tion of individual species — their system provides a precisely defined set of geo- graphic and oceanographic adjectives. It may well be that the main features of the system of Atlantic mesopelagic zoogeography proposed by B wall be con- firmed, but, in my opinion, that confirmation is not to be found in B. The 10 evermannellid and scopelarchid species occurring in the Atlantic ex- hibit distributions comparable to the following patterns described by B for myc- tophids: subtropical (2/10), tropical-subtropical (5/10), tropical (2/10), ? eastern (1/10). Subtropical Species. — Coccorella atlantica (fig. 24; 218 Atlantic Ocean records) closely fits the definition of subtropical or central-water species given above, occurring in and essentially restricted to central water-mass areas of the Atlantic, Indian, and Pacific oceans. In the Atlantic, except for a considerable number of records from the Lesser Antillean and Caribbean provinces of the Atlantic Tropi- cal Region and from the Gulf of Mexico Region, C. atlantica fits well the subtropi- cal distribution pattern as depicted and described by B (pp. 272, 280-283) for 13 mesopelagic, myctophid species. The distribution of C. atlantica is similar to the distributions of Hygophum reinhardtii (N, p. 41), H. taaningi (N, p. 49, North Atlantic only), Lampanyctus cuprarius (N, p. 198), and L. lineatus (N, p. 200, North Atlantic only). All four myctophid species also occur in the western Tropical provinces and the Gulf of Mexico. Rosenblattichthys hubbsi is known from 22 Atlantic Ocean collections (fig. 40). Except for the site of capture of the holotype (02° 27' S, 19° 00' W) and a number of records from the Lesser Antillean Province, the known distribution of R. hubbsi appears to best agree with the subtropical pattern. Most Atlantic records for R. hubbsi are from the North and South North African Subtropical Sea prov- inces of the North Atlantic Subtropical Region (B, p. 270), with only a single record from either Sargasso Sea Province. I am unable to find another example of a subtropical midwater species similar to R. hubbsi in apparent relative restric- tion to the eastern subtropics in the North Atlantic. B (p. 287) cite Pollichthys mauli (a photichthyid) and Melamphaes pumilus (a melamphaid) as examples of species ". . . common in the Sargasso Sea and rare in the North African Sub- tropical Sea." In the Indian and North Pacific oceans, R. hubbsi is, as far as is known, restricted to central-water areas. Tropical-Subtropical Species. — Five evermannellid and scopelarchid species ex- hibit distributions comparable to the tropical-subtropical pattern as described and depicted for 18 mesopelagic myctophid species (B, pp. 273, 283, 284). All five are also known from central and equatorial water-mass areas in the Indian and Pacific oceans. Scopelarchus michaelsarsi (fig. 48; 74 Atlantic Ocean records), Evermannella indica (fig. 34; 261 Atlantic Ocean records), and Odontostomops nor- malops (fig. 37; 78 Atlantic Ocean records) show fairly close restriction to the Tropical and North and South Subtropical regions as depicted by B (p. 270). All three species occur in the Gulf of Mexico, but S. michaelsarsi has not been taken in the Caribbean Sea. Of these three, only Odontostomops normalops has been taken in the Gulf of Guinea, and, although all three are known from the eastern subtropical North Atlantic, none of them has been taken in the region of the Mauritanian Upwelling. Odontostomops normalops is known from very few rec- 194 FIELDIANA: ZOOLOGY ords in the eastern subtropical North Atlantic, and, as noted above, the range of O. normalops is essentially complementary to that of Evermannella balbo (fig. 32). Examples of somewhat similar pairings are provided by Diaphus mollis (N, p. 160) vs. D. rafinesquii (N, p. 158) and by Scopelogadus mizolepis mizolepis vs. S. beanii (Ebeling & Weed, 1963, p. 41). In the Atlantic Ocean, Scopelarchus analis (fig. 46; 416 Atlantic Ocean records) and Benthalbella infans (fig. 39; 72 Atlantic Ocean records) are widely distributed in tropical and subtropical waters (although there are few records for B. infans from the Caribbean Sea and none from the Gulf of Mexico). Both species have also been taken commonly in temperate waters, especially in the North Atlantic Temperate region (B, p. 270). The myctophids Benthosema suborbital (N, p. 55), Diogenichthys atlanticus (N, p. 58), Myctophum nitidulum (N, p. 67), Gonichthys cocco (N, p. 88), Notolychnus valdiviae (N, p. 94), Lobianchia gemellari (N, p. 101), Diaphus mollis (N, p. 160), Lampanydus photonotus (N, p. 213), Ceratoscopelus war- mingii (N, p. 246), and Notoscopelus resplendens (N, p. 252) are all examples of species classed by B as tropical-subtropical which, similar to Scopelarchus analis and Benthalbella infans, were commonly taken in at least parts of the south and/or north Temperate regions in the Atlantic in addition to the tropics and subtropics. A similar distribution pattern is exhibited by Lepidophanes guentheri (N, p. 227), a species classed by B (p. 274) as tropical. Odontostomops normalops, Scopelarchus analis, and Benthalbella infans have all been taken in the Mediterranean Outflow Province (B, p. 270), but none of them is known from the Mediterranean. As Merrett et al. (1973) have pointed out, the overwhelming majority of captures of Benthalbella infans in the North Atlantic are from the eastern sector (temperate and subtropical). Both Scopelarchus analis and Benthalbella infans have been taken in the Mauritanian Upwelling Region. Benthalbella infans holds the northernmost distributional record (61° 04' N, 14° 39' W; Petr Lebedev, sta. 98) of any scopelarchid or evermannellid (Evermannella balbo is second — the northernmost record for this species is 59° 49.6' N, 20° 22.9' W). B (p. 274) consider the mesopelagic habitat to extend from near the surface to about 700 or 800 m. The limited data for B. infans presented by Merrett et al. (1973) and Johnson (1974c) suggest that the daytime depth stratum occupied by most juveniles and adults may be deeper than 800 m, making B. infans a bathypelagic species as the term is used by B. However, there are nighttime records of large adults in the upper 200 m (possibly suggesting diel vertical migration), and Merrett et al. (1973) give the overall depth range (day) as 90 to 1,500 m (night = upper 100 to 900 m). Distributional records for Scopelarchus analis and Benthalbella infans provide good evidence for continuity of distribution around the Cape of Good Hope (cf. figs. 39, 46), suggesting that, for these species, Africa does not completely isolate the Atlantic from Indian Ocean populations (for additional discussion of "the South African Barrier" see Brin- ton, 1975, pp. 146, 147). Tropical Species. — Two scopelarchid species, Scopelarchoides danae and Scopelar- chus guentheri, exhibit Atlantic distributions comparable to the "tropical pattern" described by B (pp. 273, 283) for 18 mesopelagic myctophid species. Virtually all Atlantic records for S. danae (fig. 43; 109 Atlantic Ocean records) are from the Atlantic Tropical Region as defined by B (pp. 270, 271), with the majority of records from either the Gulf of Guinea or the Lesser Antilles/ Caribbean Provinces. Exceptions include one record from the Gulf of Mexico, JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 195 one record from Slope Water, one record from the Mauritanian Upwelling, and a number of records from the Sargasso Sea and Straits of Florida provinces of the North Atlantic Subtropical Region. In the Indian and Pacific oceans, S. danae has been taken in both equatorial and central water-mass areas, but the distribution of S. danae is largely equatorial, with most records from near continental and insular land masses. Aside from the apparent association with relatively near- shore areas, the Atlantic distribution of S. danae is similar to that of the myc- tophids Myctophum asperum (N, p. 69), M. obtusirostre (N, p. 72), Diaphus luetkeni (N, p. 133), and Lampanyctus tenuiformes (N, p. 220) as well as the decapod crustacean Sergestes edwardsii (Judkins, 1978, p. 13). Scopelarchus guentheri (fig. 47; 45 Atlantic Ocean records) is widespread in tropical, subtropical, and even temperate waters in the Indian and especially in the Pacific Ocean, but it is essentially limited to the Atlantic Tropical Region in the Atlantic. Scopelarchus guentheri is all but excluded from the region of the North Atlantic Central Water Mass. The Atlantic distributions of Scopelarchus guentheri and Coccorella atlantica are largely complementary, providing a nearly perfect example of the contrast between central (=subtropical) vs. equatorial (=tropical) distribution patterns (fig. 54). The most glaring discrepancy in this contrast is the large number of records of C. atlantica from the Gulf of Mexico and Caribbean and Lesser Antilles provinces — S. guentheri is virtually unknown (two Caribbean Sea records only) from these areas. At least two myctophid species, classed as subtropical by B (p. 274), resemble C. atlantica in distribution, having been commonly taken in the Gulf of Mexico, Caribbean, and Lesser Antilles provinces but not in the Amazonian or Guinean provinces: Hygophum taaningi (N, p. 49, North Atlantic only) and Lampanyctus cuprarius (N, p. 198). Pietsch (1974, p. 93) notes the value of the 14° C isotherm at 200 m (as depicted by Schroeder, 1963) in indicating the boundary between North Atlantic vs. South Atlantic Central Water. Addition of a line indicating the position of this isotherm (fig. 54) gives good separation between the North Atlantic distributions of C. atlantica and S. guentheri. This, in connection with other distributional evidence cited above, might suggest that part or all of the "Lesser Antilles Province" should be placed in the North Atlantic Subtropical rather than Atlantic Tropical Region. I have discussed limited evidence suggesting that in the Pacific S. guen- theri is, relative to its congeners, partly excluded from central water-mass areas (Johnson, 1974c; see discussion of Pacific central-water species below). Scopelar- chus guentheri is apparently more abundant (in the Pacific) in the more produc- tive waters peripheral to the central regions of the subtropical anticyclones. Thus, it is surprising to note that in the Atlantic S. guentheri has yet to be taken in the Gulf of Guinea and far eastern South Atlantic, areas of relatively high pro- ductivity, and that S. analis (fig. 46) has been taken in those areas. Eastern Species. — The distribution of Evermannella balbo (fig. 53; 152 Atlantic Ocean records) may be comparable with the "eastern pattern" described by B (pp. 273, 274, 284, 285) for two mesopelagic myctophid species: Electrona risso (N, p. 34, known from widely scattered records in the Indian and Pacific oceans) and Diaphus holti (N, p. 163, known only from the Atlantic). Evermannella balbo is known from both the eastern and western Mediterranean and is the only ever- mannellid or scopelarchid occurring in that sea. Evermannella balbo has been taken in the Atlantic Subarctic (to 59° 49.6' N), North Atlantic Temperate, and the cooler and more productive portions of the North Atlantic Subtropical, At- Fig. 54. Atlantic and Indian Ocean distributions of Coccorella atlantica (open symbols) and Scopelarchus guentheri (closed symbols). Shown for the Atlantic are the approximate boun- daries of seven pelagic faunal regions recognized by Backus et al., 1977: A = Atlantic Subarctic Region; B = North Atlantic Temperate Region; C = North Atlantic Subtropical Region; D = Gulf of Mexico Region; E = Mauritanian Upwelling; F = Atlantic Tropical Region; G = South Atlantic Subtropical Region. Open bands show approximate bound- aries (after Sverdrup et al., 1942) between the major upper water masses: 1 = North Atlantic Central; 2 = South Atlantic Central, 3 = Indian Ocean Equatorial; 4 = Indian Ocean Central. Heavy black band denotes position of 14° C isotherm at 200 m (after Schroeder, 1963). lantic Tropical, and South Atlantic Subtropical regions. Evermannella balbo is known from only one record in the Indian Ocean but has been widely taken in the southern Transition Region (see McGowan, 1971, p. 44) of the Pacific Ocean, and £. balbo is probably circumglobal in the Subtropical Convergence Region of the Southern Ocean. Evermannella balbo thus illustrates the "east- west effect" described by B (p. 285), who characterize the waters on the eastern side of the Atlantic as generally ". . . cooler, . . . fresher, denser, lower in dissolved oxygen, higher in dissolved inorganic phosphorus, and more productive than waters to the west." In the eastern North Atlantic E. balbo is known from the Mediterranean, Mediterranean Outflow, far eastern North African Subtropical Sea, Mauritanian Upwelling, and Guinea provinces (B, p. 270, 271). In the western North Atlantic E. balbo is known from the western Temperate provinces (including Slope Water) and the northern Sargasso Sea, a pattern recognized as distinctive by Backus et al. (1970) JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 197 and Jahn & Backus (1976). Backus et al. (1969) described the faunal differences between the northern and southern Sargasso Sea at the apparent boundary region (ca. 27° N at 70° 20' W) largely in terms of primary productivity and vertical distribution of temperature. Species agreeing with £. balbo in western North Atlantic distribution include Lampanyctus crocodilus (N, p. 208, a "temperate-semisubtropical" species limited to the North Atlantic), Lampanyctus pusillus (N, p. 222), Lobianchia dofleini (N, p. 98), Stomias boaferox (Gibbs, 1969, p. 15), and the decapod crustacean Sergestes arcticus (Judkins, 1972, p. 166). Lam- panyctus pusillus, S. boaferox, and S. arcticus agree in distribution with E. balbo in occurring (or presumably so) circumglobally in the Southern Ocean but differ in apparent exclusion from the Atlantic Tropical Region. The distribution of E. balbo is similar to that depicted by Ebeling & Weed, 1963, p. 39, for Scopelogadus beanii (except that the latter is apparently unrecorded from the Mediterranean or east- ern South Pacific). The distribution of E. balbo and Lobianchia dofleini is virtually congruent. Lobianchia dofleini is said to occur circumglobally in the neighborhood of the southern Subtropical Convergence (N, p. 98), and it is likely that this is true of E. balbo. Lobianchia dofleini is classed by B (p. 274) as a "temperate- semisubtropical" species, a distribution pattern described for eight mesopelagic, myctophid species. In the main features of its North Atlantic distribution E. balbo agrees moderately to very well with the myctophid species assigned to this pattern. Nonetheless, I have retained my assignment of E. balbo to the "eastern" pattern because I believe that E. balbo well illustrates the east-west polarity discussed by B (pp. 284, 285) and particularly so when the distribution of E. balbo is compared with that of other evermannellid species (fig. 31). Thus, the distri- bution of E balbo in boreal, temperate, subtropical, and tropical waters can be associated largely with the cooler and more productive areas of the Atlantic and southern Indian and Pacific oceans. Indian Ocean Thirteen of the 19 warm-water species of evermannellids and scopelarchids occur in the Indian Ocean (table 23). None of these species is restricted to the Indian Ocean. Recent papers contributing to our knowledge of distribution pat- terns exhibited by midwater organisms in the Indian Ocean include Ebeling (1962, 1967), Ebeling & Weed (1963), Baker (1965), Gibbs & Hurwitz (1967), Kotthaus (1967), Aron & Goodyear (1969), Mauchline & Fisher (1969), Nafpak- titis & Nafpaktitis (1969), Parin (1970), Bradbury et al. (1971), Brinton & Gopala- krishnan (1973), Cohen (1973), Fleminger & Hulsemann (1973), and Nafpaktitis (1978). Survey efforts in the Indian Ocean have resulted in an extremely uneven distribution of sampling effort (e.g., Dietrich, 1973), with most attention having been given to the northern and far-western regions. A huge gap in sampling effort for midwater fishes is present in the southern and southeastern Indian Ocean as reflected by Figure 49 and clearly shown in Nafpaktitis (1978, p. 3). The gap in sampling effort results in considerable uncertainty concerning subtropical (=central- water area) distribution patterns in the Indian Ocean — an uncertainty reflected in the following distribution accounts. A valuable summary of the oceanography of the Indian Ocean is provided by Wyrtki (1973). In the tropical and subtropical Indian Ocean two major and dis- tinct circulation systems are recognizable: the unique monsoon gyre that 198 FIELDIANA: ZOOLOGY changes seasonally and the southern hemisphere subtropical, anticyclonic gyre. Dividing the monsoon gyre (= "tropical" or "equatorial- water" area) from the subtropical gyre (= "subtropical" or "central- water" area) is a prominent "hydro-chemical front" (Wyrtki, 1973, pp. 23-25) at ca. 10° S. Identified by a horizontal salinity minimum that stretches from Sumatra to Africa, this front separates the low-nutrient, high-oxygen water of the south from the nutrient- rich, oxygen-poor water of the north. Although identifiable in the vertical and horizontal distribution of a number of parameters, the front is perhaps most dramatically delineated by the distribution of dissolved oxygen (Wyrtki, 1973, p. 24). Most authors who have dealt with open-ocean distribution patterns in the Indian Ocean have explicitly or implicitly recognized the importance of this frontal zone, marking as it does the division between Indian Equatorial and Indian Central Water-Mass areas (e.g., Ebeling, 1962, 1967; Cohen, 1973; Brinton & Gopalakrishnan, 1973; Nafpaktitis, 1978). It is clear, however, that distribution patterns exhibited by Indian Ocean mid- water organisms are more numerous than a simple "equatorial" vs. "central." Brinton & Gopalakrishnan (1973) recognize (for euphausiids) five major, zonally distributed boundary regions, at 10° N, 0°, 10° S, 25° to 30° S, and 40° to 45° S. The boundary at 10° N corresponds roughly to the extent of the main subsurface low-oxygen waters of the northern Indian Ocean (see below). The boundary at the equator is said (Brinton & Gopalakrishnan, 1973, p. 379) to correspond with the southern edge of the North Equatorial Current (NE Monsoon). The bound- ary at 10° S is the prominent hydrochemical front discussed above. There is no specific hydrographical feature identified by Brinton & Gopalakrishnan (1973, p. 381) with the boundary at 25° to 30° S, but a number of euphausiid species apparently find the northern or southern end-points of their range within this latitudinal zone. The boundary at 40° to 45° S is identifiable with the Subtropical Convergence zone — the boundary between subtropical and subantarctic waters in all oceans. There is sufficient information to allow discussion of the Indian Ocean distributions of evermannellid and scopelarchid species with respect to three of these boundary regions — those at 10° N, 10° S, and at the Subtropical Convergence. Brinton & Gopalakrishnan (1973) also report on meridional features of euphausiid distribution in the Indian Ocean, particularly the extensive north to south ranges of coastal forms and the far southerly occurrences of tropical and subtropical species in the region of the Agulhas Current system (tropical species range as far as 33° S, tropical-subtropical species as far as 38° S). Wyrtki (1973) notes the lack of a well-developed eastern boundary current (see Wooster & Reid, 1963) in the southern Indian Ocean. A western boundary current, the Agulhas Current, is by far the strongest component of the subtropical gyral, with core velocities up to 2 m/sec and large transport values south of 10° to 12° S (Wyrtki, 1973; Pearce, 1977). Nafpaktitis (1978) notes the southward extension (to 20° to 25° S and beyond) of equatorial (=tropical) myctophid species in the Agulhas Current. Whether directly (Nafpaktitis, 1978, p. 84) or indirectly (Wyrtki, 1973, p. 30), the Agulhas Current system provides a basis for transport of Indian Ocean water and midwater species into the eastern South Atlantic (Krefft, 1974, pp. 231, 232). A very clear example of the "Agulhas Pattern" can be seen in Brinton's (1975, p. 98) depiction of the distribution of Euphausia diomedeae. In the far-western Indian Ocean the southern distributional end- JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 199 points of six evermannellid and scopelarchid species are possibly or probably related to the Agulhas system. The six species are: Coccorella atrata (fig. 24), Evermannella indica (fig. 34), Scopelarchoides danae (fig. 43), S. signifer (fig. 45), Scopelarchus guentheri (fig. 47), and S. michaelsarsi (fig. 48). This conclusion seems particularly well supported for three species that are largely equatorial in distri- bution, viz., Coccorella atrata, Scopelarchoides danae, and S. signifer. Bradbury et al. (1971, pp. 423-436) suggest a possible east-west polarity in the distribution of certain Indian Ocean species, i.e., some species limited to the east, others occurring only in the west. The cruise track and sampling regime (Bradbury et al., 1971, pp. 411, 436) providing the samples on which this conclu- sion was based seem to me to be largely inappropriate to detecting such a polarity even if the suggestion is valid. Nonetheless, east-west polarity in the Indian Ocean may well occur as Nafpaktitis (1978, p. 84) has suggested for a number of myctophid species. No evermannellid or scopelarchid species shows any evidence of such eastern or western limitation in the Indian Ocean. Transition Region Species. — No warm- water species of evermannellid or scopelarchid is actually recorded from the Subtropical Convergence Region of the southern Indian Ocean. The most probable candidate, Evermannella balbo, is known from a single Indian Ocean specimen taken in the Agulhas Current system at 29° 52' S, 31° 36' E. Based on the distribution of E. balbo in the South Atlantic and South Pacific (fig. 53), it seems likely that E. balbo will be found to occur throughout the southern Indian Ocean in the Subtropical Convergence Region. I also predict the occurrence in this region of two additional species: Coccorella atlantica (fig. 24) and Scopelarchus guentheri (fig. 47). Subtropical ( =Central-Water) Species. — Two species, Coccorella atlantica (fig. 24) and Rosenblattichthys hubbsi (fig. 40), are known in the Indian Ocean only from the region of the Central Water Mass (Sverdrup et al., 1942, p. 740), south of 10° S. Coccorella atlantica is known from only two specimens from two collections in the Indian Ocean — one record from 16° 05' S at 76° 16' E; the other from the South Australian Basin (ca. 38° S, 128° E). Rosenblattichthys hubbsi is known from only seven Indian Ocean specimens, all larvae or small juveniles, representing five Indian Ocean records. Both C. atlantica and R. hubbsi are replaced by closely related congeners, C. atrata and R. alatus, respectively, in the Equatorial Water-Mass area of the Indian Ocean, north of 10° S. Coccorella atlantica and R. hubbsi are essentially limited to central-water areas in the Atlantic and Pacific oceans. A similar equatorial vs. central distributional contrast is exhibited by the lantern fishes Diogenichthys panurgus vs. D. atlanticus (Nafpaktitis & Nafpaktitis, 1969, p. 15). Cohen (1973, p. 462) tallied distributional information for a number of mid- water fish groups occurring in the Indian Ocean. He (p. 462) reports that for 50 myctophid species surveyed, 18 species appear to be restricted to the region of the Equatorial Water Mass, six are restricted to the region of the Central Water Mass, eight occur in (at least parts of) both areas, 11 occur in central and sub- antarctic water, five are restricted to the subantarctic, and two are broadly dis- tributed, ranging from the Arabian Sea to the subantarctic. Tropical-Subtropical Species. — Three species are broadly distributed in Equato- rial and Central Water-Mass areas of the Indian Ocean: Odontostomops normalops (fig. 37; 15 Indian Ocean records: 09° 36' N to 20° 47.9' S), Benthalbella infans (fig. 39; 24 Indian Ocean records: 00° 14' S to 34° S), and Scopelarchus analis (fig. 46; 16 200 FIELDIANA: ZOOLOGY Indian Ocean records: 00° 01' N to 38° 53.2' to 58.5' S). All three species are tropical-subtropical in the Atlantic and Pacific Oceans. A fourth species, Scopelarchus guentheri, is also broadly distributed in the warm-water Indian Ocean (fig. 47), but the account of this species is reserved for the section dealing with species occurring north of 10° N (see below). Extremely limited evidence suggests that B. infans and S. analis may agree with the tropical-subtropical euphausiids Euphausia brevis and Thysanopoda subaequalis (Brinton & Gopala- krishnan, 1973, pp. 366, 376) in sharing the equator as the (approximate) north- ern limit of distribution in the Indian Ocean. Midwater fish species showing a similar distribution in the Indian Ocean include the lantern fishes Benthosema suborbitale and Lampanyctus alatus (Nafpaktitis & Nafpaktitis, 1969, pp. 11, 56). Odontostomops normalops agrees with many tropical-subtropical and tropical euphausiid species in not occurring north of ca. 10° N (Brinton & Gopala- krishnan, 1973). Tropical ( =Equatorial) Species. — Five species are largely or entirely restricted to the region of the Equatorial Water Mass in the Indian Ocean: Evermannella indica (fig. 34; 62 Indian Ocean records), Rosenblattichthys alatus (fig. 40; 13 Indian Ocean records), Scopelarchoides danae (fig. 43; 34 Indian Ocean records), S. signifer (fig. 45; 48 Indian Ocean records), and Scopelarchus michaelsarsi (fig. 48; 8 Indian Ocean records). Rosenblattichthys alatus and Scopelarchoides signifer are limited to the Indian and Pacific Oceans, the other three species occur as well in the Atlantic Ocean. All five species agree with most Indian Ocean equatorial mid- water species, including the majority of species of Diaphus (Nafpaktitis, 1978), in showing restriction (or near restriction) to the zone between 10° N and 10° S except for southerly range extensions in the Agulhas Current system. One addi- tional species, Coccorella atrata, is restricted to the Equatorial Water-Mass Region of the Indian Ocean but also occurs north of 10° N (see below). Rosenblattichthys alatus, Scopelarchoides danae, and S. signifer are (on the basis of known captures) essentially restricted to equatorial or tropical waters through- out their respective ranges. Evermannella indica and Scopelarchus michaelsarsi are tropical-subtropical species in both the Atlantic and Pacific oceans. It is quite possible that the apparent restriction of the latter two species to the Equatorial Water-Mass Region in the Indian Ocean is an artifact of sampling effort. Species Occurring North of 10° N. — Two species are known from north of 10° N in the Indian Ocean: Coccorella atrata (fig. 24; 42 Indian Ocean records) and Scopelarchus guentheri (fig. 47; 54 Indian Ocean records). Coccorella atrata, known only from the Indian and Pacific oceans, is essentially restricted to tropical or equatorial waters throughout its range. In the Atlantic S. guentheri is essentially restricted to the Atlantic Tropical Region (Backus et al., 1977, p. 270), but S. guentheri is widely distributed in equatorial, central, and Transition Region wa- ters in the Pacific and (probably) Indian oceans. The majority of Indian Ocean records for both species are from the zone between 10° N and 10° S, a result that may partly reflect sampling effort. The northern Indian Ocean including the semi-enclosed basins of the Arabian Sea and Bay of Bengal, an area strongly influenced by the seasonal monsoon systems, is characterized by strong regional upwelling systems, high nutrient levels, high productivity, and a marked oxygen minimum layer (Vinogradov & Voronina, 1962; Gibbs & Hurwitz, 1967; Kinzer, 1969; Wyrtki, 1971; Brinton & Gopalakrishnan, 1973; Currie et al., 1973; McGill, 1973). The horizontal and JOHNSON: EVERMANNELLIDAE AND SCOPELARCHIDAE 201 vertical extent of the oxygen minimum layer is depicted in a number of charts presented by Wyrtki (1971). If the oxygen minimum layer is defined as subsur- face waters in which dissolved oxygen values are ^=1.0 ml/L, the layer is more than 1,000 m thick in parts of the Arabian Sea, with values less than 0.1 ml/L occurring vertically over an extent of several hundred meters in some areas (e.g., Wyrtki, 1971, p. 441). A similarly striking if somewhat less pronounced oxygen minimum layer is present throughout the Bay of Bengal (e.g., Wyrtki, 1971, p. 413). Gibbs & Hurwitz (1967) found the southern limit of occurrence of Chauliodus pammelas, a species endemic to the low-oxygen area, to be at about 05° S in the Arabian Sea. This latitude was stated (Gibbs & Hurwitz, 1967, p. 802) to closely correspond with the southern extension of the