mi liiiiiiii ■iiiiii^ '": m»»CT;i;' HARVARD UNIVERSITY. LIBRARY OF THE MUSEUM OF COMPARATIVE ZOOLOGY Yr|,v(.^ta/n^ CWvJt r^.T, i^n- S-eii^vLu 30, mx SEP 3 01929 MEMOIRS OF THE MUSEUM OF COMPARATIVE ZOOLOGY AT HARVARD COLLEGE. VOL. XLIII. CAMBRIDGE, MASS., U. S. A. printeJ) tor tbe flDuseum. 1917-1929 CONTENTS. THE AMERICAN CHARACIDAE Part I. By Carl H. Eigenaiann. pp. 1-102. 16 plates. August, 1917. Part II. ByCarl H. Eigenmann. pp. 103-208. 22 plates. January, 1918. Part III. By Carl H. Eigenmann. pp. 209-310. 28 plates. July, 1921. Part IV. By C.\rl H. Eigenmann. pp. 311^28. 24 plates. May, 1927. Part V. By Carl H. Eigenmann and George S. Myers, pp. 429-558. 11 plates. Sep- tember, 1929. Page Introduction 3 Thayer Brazilian Expedition ... 5 Carnegie Museum Expedition to Central South America .... 10 Guiana Expedition 12 Gimbel Expedition 12 Colombian Reconnaissance ... 13 Landon-Fisher Expedition to Co- lombia 14 Landon Expedition to Colombia and Ecuador 14 Central American Expeditions . . 16 Expedition to Guatemala .... 16 The Characidae 17 Zoological Position 17 Radial Adaptation 19 Literature 23 Geographical Distribution ... 30 Chronological List of Generic Names 32 The American Tetragonopterinae . . 38 Contrasted Generic Characters . . 42 Polyphyletic Characters .... 43 Selective Grouping of Characters . 46 Polyphyletic Genera 46 Key to the Genera 50 1. Tetragonopterus 54 2. Entomolepis 63 3. Moenkhausia 65 4. Knodus . . 114 5. Markiana 123 6. Gymnocorymbus 124 7. Thayeria 128 8. Pristella 130 Page 9. Hemigrammus 135 10. Hyphessobrycon 172 11. Hasemania 221 12. HoUandichthys 225 13. Pseudochalceus 226 14. Astyanax 227 Subgenera: Poecilurichthys and Zygogaster 230 Subgenus: Astyanax .... 271 15. Ctenobrycon 330 16. Psellogrammus 336 17. Astyanacinus 338 18. Deuterodon 341 19. Landonia 349 20. Nematobrycon 351 21. Microgenys 353 22. Cerotobranchia 356 23. Bryconamericus 358 23a. Argopleura 395 23b. Phenacobrycon 400 24. Hemibrycon 401 24a. Acrobrycon 416 25. Creagrutus 417 26. Piabina 429 26a. Piabarchus 432 27. Brycochandus 433 28. Creatochanes 434 29. Bryconops 440 30. Phenaeogaster 442 31. Vesicatrus 450 32. Genycharax 451 33. Scissor 453 34. Henochilus 454 35. Psalidodon 455 CONTENTS. Page The African Tetragonopterinae . . . 456 Subfamily Rhoadsiinae 457 36. Parastremma 457 37. Rhoadsia 459 Subfamily Glandulocaudinae .... 463 38. Corynopoma 468 39. Diapoma 471 40. Pterobrycon 472 41. Pseudocorynopoma .... 473 42. Gephyrocharax 477 43. Microbrycon 484 44. Hysteronotus 485 45. Glandulocauda 487 46. Mimagoniates 490 Subfamily Iguanodectinae .... 493 47. Iguanodectes 494 48. Piabucus 496 Page Subfamily Stethaprioninae .... 499 49. Stethaprion 500 50. Epphipicharax 503 51. Brachychalcinus 507 52. Popteila 509 Subfamily Stichonodontinae .... 511 53. Stichonodon 511 The Fossil Characins 512 54. Lignobrycon 513 55. Eobrycon 513 56. Characilepis 515 Supplement 516 Later Expeditions 517 Additional Generic Names . . . 519 Addenda et Corrigenda .... 520 4a. Bertoniolus 527 23c. Bryconacidnus 545 24b. Creagrudite 546 LIST OF ABBREVIATIONS. M. C. Z. or numbers without any reference means Museum of Comparative Zoology, Cambridge, Mass. I. and I. U. M. means Indiana University Museum, Bloomington, Ind. A. N. S. P. means Academy of Natural Sciences, Philadelphia, Pa. C. and C. M. means Carnegie Museum, Pittsburgh, Pa. "Jf Itl^J ^L.'f36' /IDemotre of tbe flDuseum ot Comparative 2;ooloop AT HARVARD COLLE&E. Vol. XLIII. Part 1. THE AMERICAN CHARACIJ^^If^ I I I' U li U I BY CARL H. EIGENMANN. WITH SIXTEEN PLATES. CAMBRIDGE, U. S. A.: IPrinteb tor tbe flDuseum. August, 1917. (/(Oy,i /^yo y^ \\i{/ /iDemotre of tbe nDuaeum ot domparatlve Zooloae AT HARVARD COLLEGE. Vol. XLIII. Part 1. THE AMERICAN CHARACIDAE. BY CARL H. EIGENMANN WITH SIXTEEN PLATES. CAMBRIDGE, U. S. A.: IPrtnteJ) tor tbe /IDuseum. August, 1917. THE AMERICAN CHAMCIDAE/ INTRODUCTION. The revision of the American Characidae, based on the collections of the Nathaniel Thayer Brazilian Expedition in the Museum of Comparative Zoology was commenced by Mrs. Rosa Smith Eigenmann and myself in 1888. A little more than a year later the work was temporarily suspended. The published results, so far as obtained, are enumerated on p. 8-10. Work was resumed in 1903 with the collections of Indiana University and of the United States National Museum, and a monograph completed in 1906. Types in several museums in London, Paris, and Vienna were examined in 1906 and 1907. In 1907 the study of the collections made by the Thayer Expe- dition was renewed. This material made it necessary to rewrite the entire monograph, for, forty years after it was gathered, it was still the most impor- tant collection of South American fresh-water fishes, and contained a large number of undescribed species and genera. The revised monograph, through the Tetragonopterinae, was finished in the spring of 1908. During the preparation of this revision it became apparent that the material at my command, from several regions of South America, was deficient. To obtain material from one of these regions, I spent the autumn of 1908 in British Guiana. Volume 5 of the Memoirs of the Carnegie Museum contains a detailed account of this Guiana Expedition. In May, 1909, my position of Curator of Ichthyology in the Carnegie Mu- seum placed me in charge of another large series of South American fresh-water fishes. This collection was made under my general direction, between November 6, 1907 and January 10, 1910, by Mr. John D. Haseman, one of my former students. For the most part it came from areas not covered by the Thayer Expedition. The Director of the Carnegie Museum, Dr. W. J. Holland, has kindly allowed me to incorporate in the present monograph the results obtained from a study of the Carnegie collections. ' Contribution from the Zoological Laboratory of Indiana University, No. 98. 4 THE AMERICAN CHARACIDAE. Still later (December 1911-April 1912) I made extensive collections in Colombia; and in the early part of 1913 two of my students, Mr. Arthur M. Henn and Mr. Charles Wilson, gathered many fishes in the streams of western Colombia. Mr. Henn remained in South America till 1914 and extended the explorations to western Ecuador. Smaller collections were made by Miss Lola Vance in the streams about Tarma, Peru, and by Dr. Ellis and Dr. William Tucker in British Guiana. The collections of the Yale-National Geographic Society Expedition to the Urubamba River have also been available in the preparation of this monograph. These different collections admirably supplement each other. They are the most extensive collections that have ever been brought together from South America. In the number of specimens at my disposal they far exceed the com- bined collections recorded in all other museums of the world. The number of American species of Characins exceeds six hundred; and as they offer some of the best material to demonstrate several facts of evolution it is hoped to illustrate the species quite thoroughly. To provide for the illus- trations of the Tetragonopterinae I am indebted to the Trustees of the Eliza- beth Thompson Science Fund for an appropriation of $250.00, and to the Trustees of the Bache Fund for a similar amount. The Carnegie Museum and Mr. H. W. Fowler have granted the reproduction of a number of figures. The portion of the monograph dealing with Hemigrammus, Hyphessobrycon, and Hasemania was prepared with the collaboration of Marion Lee Durbin, now Marion Durbin Ellis. This Memoir deals with the Tetragonopterinae,' Rhoadsinae, Glandulo- caudinae, Iguanodectinae, Stethaprioninae, and Stichanodontinae. All of the then known species of these subfamilies, except the species of Stevardia, were included by Giinther (Cat. fishes Brit. Mus., 1864, 5) in the Tetragonopterinae. There are recognized in this Memoir fifty-two genera, and three hundred and twelve species. Over half of these were described during the course of the prepa- ration of the Monograph. The subfamily Tetragonopterinae, at present the dominant group of tropi- cal American fishes, was defined by Giinther {loc. cit.) to include Piabucina, Alestes, Brachyalestes, Chirodon, Chalceus, Brycon, Chalcinopsis, Chalcinus, Gasteropelecus, Piabuca, and Agoniates, besides the genera hsted below. Elimi- nating these genera, which are now relegated to various separate subfamilies ' The portion of the Monograph dealing with the Cheirodontinae appeared recently in the Memoirs Carnegie Museum, 1916, 7, p. 1-99, pi. 1-17. SOURCES OF THE MATERIAL. 5 or which (Alestes and Brachyalestes) are geographically extra-limital, there remain: — Tetragonopterus, with thirty-two species; Scissor, with one species; P.seudochalceus, with one species; Bryconops, with two species; Creagrutus, with one species; making a total of five genera, with thirty-seven species. Of the Iguanodectinae, which were included in Giinther's Tetragonopterina, he recorded two species belonging to the genus Piabuca. He recorded four species of Corynopoma and placed them in his Erythrinina. They are the Stevardia albipinnis considered below in the discussion of the new subfamily, Glandulocaudinae. In the last general review of Gixnther, there were a grand total of seven genera and forty-three species, as compared with the fifty-two genera and over three hundred species, known today. SOURCES OF THE MATERIAL. The Nathaniel Thayer Brazilian Expedition. In January 1807, Karolina, Archduchess of Austria, was married to the Crown Prince of Brazil. In the retinue accompanying the young couple to Brazil went Johann Natterer of Wien. Natterer remained in Brazil for eighteen years and collected extensively from Rio de Janeiro to Cuyaba, thence down the Madeira to the Amazon and up the Rio Negro and Rio Branco. Taking advan- tage of the opportunity, the King of Bavaria sent two naturahsts, Johann Bap- tist von Spix and Carl Friedrich von Martins with the bridal party. They also collected natural history specimens. The routes of travel of Natterer and of Spix and Martins are shown on Plate 1. The fishes collected by Natterer were described by Heckel, Kner, and Steindachner. Those collected by Spix and Martins were being described by Spix when he died. Louis Agassiz, then a student at the University of Munich, w:as selected to complete the work of Spix. Agassiz's work resulted in a sumptu- ous folio, and a desire to personally inspect the fauna of Brazil, a desire fulfilled forty years later. In 1865, the generosity of Mr. Nathaniel Thayer made it possible for Agassiz to undertake his journey to Brazil. The assistants of the Thayer Expedition were James Burkhardt, artist, J. G. Anthony, conchologist, C. F. Hartt, and Orestes St. John, geologists, J. A. Allen, ornithologist, and George Sceva, preparator. Besides these assistants 6 THE AMERICAN CHARACIDAE. several volunteers accompanied the Expedition: these were Edward Copeland, Newton Dexter, Walter Hunnewell, William James, Stephen Van Rensselaer Thayer, and Thomas Ward. Agassiz had the further assistance of the Brazilian Government through the Emperor, Dom Pedro II. Among Brazilians who joined the Expedition or aided in other ways were Major Coutinho, Messrs. Vinhas, Bourget, Talisman, Dr. Justa, Couto de Magalhaes, and others. The Expedition started late in March, 1865, and landed at Rio de Janeiro on April 22nd. Three months were spent about Rio. On July 25th Agassiz with Coutinho, Burkhardt, Bourget, Hunnewell, and James, joined later by Dexter and Thayer, went along the coast to Bahia (July 28th), Pernambuco (31st), Parahyba do Norte (August 2nd), Ceard (4th), Maranham (6th) and Para (11th). On August 20th the party started up the Amazon. It was divided in various ways. One or two of the assistants were left, or sent to some place to collect for longer or shorter periods to rejoin the main party at times. The itinerary up the Amazon was: — August 20,1 865. Start up the Amazon. 20. Breves. 21. Tajapuru. 32. Giirupa. 23. Porto do Moz. 25. Montalegre. (Monte Alegre). 26. Santarem (Dexter, .James, and Talisman, a young Brazilian start up the Tapajos from here; Bourget and Hunnewell remain at Santarem. They rejoin the Expedition at Manaos). September 1. On an arm of the Rio Ramos, connecting the Amazons, through the Mauhes, with the Madeira. 3. Return to steamer at Villa Bella. 4- Manaos. 10. Leave Manaos. 13. Coari. i4- Teffe. 15. Fonte Boa. 17. San Paolo. 18. Tabatinga. Bourget remains at Taba- tinga. 20. Left Tabatinga. James and Talisman remain at San Paolo to ascend the 19a. 20. Left Fonte Boa again. 21. Teffe again — take up residence on Lake Teffe. Sitio. most remarkable catch in Forest Lake. October 17. James and Talisman returned from I^a, and Jutahy. About Oct. 21, left Teffe. Verge of rainy season. Bourget returns from Tabatinga. 23. Reach Manaos. 27. Go to Lake Hyanuary. 29. Return to Manaos for a six weeks' stay. December 6. Thayer returns from Lago Alexo, Bourget and Thayer from Cudajas, James from Manacapuru, Coutinho from Lake Hyanuary, Jose-Fernandez, Curupira, etc. 10. Dexter and Talisman returned from Rio Branco. Water too high. Leave for Mauhes on war vessel. 14- Mucaja-Tuba. 16. Mauhes. 20. Leave Mauhes. 25. Manaos. 27. Leave Manaos to ascend the Rio Negro. 29. Pedreira. 31. Manaos. January 16, 1866. Villa Bella and Mauhes again. 18. Lago Maximo. 20. Leave Villa Bella. 21. Obidos. 22. Santarem at the mouth of the Tapajos. 26. Monte Alegre. Trip to .Serra Erere. 29. Leave Monte Alegre. About 30. Porto do Moz on the Xingu: \'inhas brings collections from above and below the cascades of the Xingu. February I. Gurupa. 2. Tajapuru. 4- Reached Para. March 26. Leave Para for Rio de Janeiro. April 2. An inland excursion at Ceara. 6. Pacatuba and the region about it. //. Cearii. 16. Leave Ceara for Rio. July 2. Sail for home. During Agassiz's trip up the Amazon with his immediate assistants, the rest of his party was divided to collect in the rivers of eastern Brazil from Rio de Janeiro to Para. "One object was kept constantly in view throughout this SOURCES OF THE MATERIAL. 7 Expedition, — namely, that of ascertaining how the freshwater fishes are dis- tributed throughout the great river systems of Brazil." The routes of the separate parties are indicated on Plate 1. Hartt and Copeland collected in the lower courses of the rivers from the San Francisco to Rio de Janeiro. They left Rio de Janeiro on June 19. They visited the lower courses of the Parahyba collecting at Campos and San Fidelis and ascended the tributary Muriahy. From San Fidelis they crossed the divide to the Itabapuana and descended that stream to its mouth going thence via the Rio Itapemerim, Rio Novo,' and Guarapary to Victoria. On a subsequent journey from Rio they ascended the Rio Doce for ninety miles to the first fall at Porto de Souza and collected also at Linhares in the Rio Doce and the lake and river Juparana. From Linhares they went to San Matheus and the Rio Mucury in which they collected at its mouth, Porto Alegre " and some distance inland at Santa Clara. From Santa Clara they went to Philadelphia collecting on the way in the Rio Urucu and thence to Calhoa and Minas Novas both on the Arassuahy. They separately descended the Rio Jequitinhonha, three hundred and sixty miles to the sea. They collected next at Cannavierias and ascended the Rio Pardo to its fall. Collections were also made at Belmonte, Santa Cruz and Porto Seguro on the coast south of the Rio Jequitinhonha. They then visited Bahia and retiu-ned to Rio de Janeiro.^ St. John, Allen, Ward, and Sceva went from Rio de Janeiro to the Rio Parahyba and Juiz de Fora, across the Serra do Mantequeira to Barbacena. They collected at Rodeio, Rio Macacos, a tributary of the Santa Anna, State Rio de Janeiro. Ward left the party at Barbacena and went over Ouro Preto and Santa Barbara to Diamantina "passing from one river-basin to another in order to examine as many of the tributaries of the Rio Doce and the Jequitin- honha as possible." Ward then crossed the San Francisco at Januaria and went to the Tocantins which he followed to its junction with the Amazon. St. John, Ward, and Allen continued via Logoa Dourada and Pradas across the Rio Carandahy and Rio Paraopeba to Sabura (Rio Macacos, into Rio San Fran- cisco), Santa Luiza, Lagoa Santa, Sete Lagoas and Gequitiba. Sceva remained about Lagoa Santa and later went via Rio de Janeiro to Canto-Gallo. St. John and Allen went down the San Francisco to Januaria. Here ■ Note also another Rio Novo in Haseman's itinerary, p. 10. ' Not the Porto Alegre, Rio Grande do Sul where von Ihering and Haseman collected. 'Collections were made by Hartt in the Rio San Francisco below the fall either at this time or on a later trip. Hartt and Copeland sent in specimens recorded as from Jacurpe and Posuca, localities 1 have not been able to locate. 8 THE AMERICAN CHARACIDAE. Allen, on account of his health, was compelled to leave the Expedition. St. John followed the San Francisco to Villa do Barra then went by land through the valley of the Rio Grande, a tributary of the San Francisco to Santa Rita, "thence to Mocambo and across the table-land separating the basin of the Rio San Francisco from that of the Rio Paranahyba." He collected in the Basin of the Rio Paranahyba at Paranagua, Manga, San Gongallo, and Therezina. Near the latter place he collected in the Rio Poty or Puty a tributary of the Rio Paranahyba. From Therezina he crossed the divide to Caxias on the Itapicuru which empties into the ocean at Maranham.' He reached the latter place on January 8, 1866. The collection of fresh-water fishes made by Louis Agassiz and his assist- ants is the largest ever brought together by a single expedition. The hope of Agassiz to complete the work on the fishes was not realized, the very wealth of material making a comprehensive report almost herculean. Dr. Franz Steindachner, Hofrat and Intendant of the K. K. Naturhistorisches Hofnmseum, Wien, who has written extensively concerning the collections made by Natterer, obtained leave of absence from May 1870 until June 1873. During this time he accompanied Professor Agassiz on the Hassler Expedition, December 1871 to August 1872, and devoted more than two years to the study of the fishes secured during the Thayer Expedition which he considered "without any exaggeration the richest and most complete in the world." The following papers are based wholly or in part on the Thayer Brazilian collection: Agassiz, Louis and Mrs. E. C. A journey in Brazil. Boston, 1868. EiGENMANN, C. H. The evolution of the catfishes. Zoe, 1890, 1, p. 10-1.5. On the presence of an operculum in the Aspredinidae. Amer. nat., 1892, 26, p. 71. Steindachneria. Amer. nat., 1897, 31, p. 158-159. New genera of South American freshwater fishes, and new names for some old genera. Smith, misc. coll., 1903, 45, p. 144-148. EiGENMANN, C. H. and R. S. A list of the American species of Gobiidae and CallionjTnidae, with notes on the specimens contained in the Museum of Comparative Zoology, at Cambridge, Massachusetts. Proc. Cal. acad. of sci., 18S8, ser. 2, 1, p. 51-78. American Nematognathi. Amer. nat., 1888, 22, p. 647-649. Preliminary notes on South American Nematognathi, I. Proc. Cal. acad. sci., 1888, ser. 2, 1, p. 119-172. Preliminary descriptions of new species and genera of Characinidae. West Amer. sci., 1889, 6, p. 7-8. ' There is anotlier river Itapicuru emptying into the ocean between the bay of Bahia and the mouth of the San Francisco, in which Dr. Haseman made extensive collections, (p. 10). SOURCES OF THE MATERIAL. 9 Description of new nematognathoid fishes from Brazil. West Amer. sci., 1889, 6, p. 8-10. Preliminary notes on South American Nematognathi, II. Proc. Cal. acad. sci., 1889, ser. 2, 2, p. 28-56. A review of the Erythrininae. Proc. Cal. acad. sci., 1889, ser. 2, 2, p. 100-116, Pl- 1- A revision of the edentulous genera of Curimatinae. Ann. N. Y. acad. sci., 1889, 4, p. 409-440. A revision of the South American Nematognathi or catfishes. Occ. papers, Cal. acad. sci., 1890, 1, p. 1-508, pl. 1. A catalogue of the freshwater fishes of South America. Proc. U. S. N. M., 1891, 14, p. 1-82. Carman, Samuel. On the species of the genus Chalcinus in the Museum of Comparative Zoology at Cambridge, Mass., U. S. A. Proc. Essex inst., 1890, 22, p. 1-7. On the species of Gasteropelecus. Proc. Essex inst., 1890, 22, p. 8-10. On the species of Cynopotamus. Proc. Essex inst., 1890, 22, p. 11-14. On the species of the genus Anostomus. Proc. Essex inst., 1890, 22, p. 15-23. On a genus and species of the Characines (Henochilus wheatlandii, gen. n. et sp. n.). Proc. Essex inst., 1891, 22, p. 49-52, pl. 1. The cyprinodonts. Mem. M. C. Z., 1895, 19, p. 1-180, pl. 1-12. Jordan, D. S. and Eigenmann, C. H. A review of the Sciaenidae of America and Europe. Rept. U. S. comm. fish., 1889, 14, p. 343^52, pl. 1-4. A review of the genera and species of Serranidae found in the waters of America and Europe. Bull. U. S. fish comm., 1888, 8, p. 329-442. Steindachner, Franz. Die siisswasserfische des siidostlichen Brasiliens. Sitz. Akad. wis- sensch. Wien, 1875, 70, abth. I, p. 499-538, pl. 1-6. Separate, p. 1-40, pl. 1-6. Beitrage zur kenntniss der chromiden des Amazonenstromes. Sitz. Akad. wissensch. Wien, 1875, 71, abth. I, p. 61-137, pl. 1-8. Separate, p. 1-77, pl. 1-8. Ueber einige neue Brasilianische siluroiden aus der gruppe der doradinen. Sitz. Akad. wissensch. Wien, 1875, 71, abth. I, p. 138-151, pl. 1^. Separate, p. 1-14, pl. 1-4. Die siisswasserfische des siidostlichen, Brasiliens. II. Sitz. Akad. wissensch. Wien, 1875, 71, abth. I, p. 211-245, pl. 1-6. Separate, p. 1-35, pl. 1-6. Beitrage zur kenntniss der der characinen des Amazonenstromes. Sitz. Akad. wissensch. Wien, 1875, 72, abth. I, p. 6-24, pl. 1-2. Separate, p. 1-18, pl. 1-2. Ichthyologische beitrage. IV. Sitz. Akad. wissensch. Wien, 1876, 72, abth. I, p. 551-616, pl. 1-13. Separate, p. 1-65, pl. 1-13. Ichthyologische beitrage. V. Sitz. Akad. wissensch. Wien, 1876, 74, abth. I, p. 49-240, pl. 1-15. Separate, p. 1-190, pl. 1-15. Die siisswasserfische des siidostlichen Brasiliens. III. Sitz. Akad. wissensch. Wien, 1876, 74, abth., I, p. 559-694, pl. 1-13. Separate, p. 1-136, pl. 1-13. Die siisswasserfische des siidostlichen Brasiliens. IV. Sitz. Akad. wissensch. Wien, 1877, 76, abth. I, p. 217-230, pl. 1-2. Separate, p. 1-14, pl. 1-2. Uber einige neue und seltene fischarten, etc. Denk. Akad. wissensch. Wien, 1878, 41, p. 1-52, pl. 1-9. Beitrage zur kenntniss der flussfische Siidamerika's, I. Denk. Akad. wissensch. Wien, 1879, 41, p. 151-172, pl. 1-3. Separate p. 1-24^ pl. 1-1. Ichthyologische beitrage. VIII. Sitz. Akad. wissensch. Wien, 1880, 80, abth. I, p. 119-191, pl. 1-3. Separate, p. 1-73, pl. 1-3. 10 THE AMERICAN CHARACIDAE. BeitragezurkenntnissderflussfischeSudamerika's. II. Denk. Akad. wissensch. Wien, 1881, 43, p. 103-146, pi. 1-7. Separate p. l-i6. Beitrage zur kenntniss der flussfische Siidamerika's. III. Denk. Akad. wis- sensch. Wien, 1881, 44, p. 1-18, pi. 1-5. Separate, p. 1-18, pi. 1-5. Beitrage zur kenntniss der fliLssfische Siidamerika's. IV. Denk. Akad. wissen- sch. Wien, 1882, 46, p. 1-44, pi. 1-7. Separate, p. 1-44, pi. 1-7. The Carnegie Museum Expedition to Central South Amreica. A detailed account of this Expedition is given in volume 9 of the Annals of the Carnegie Museum. During this Expedition Mr. Haseman collected at the following localities: — November 6, 1907. Rio Coite, into the Rio Salitre, into Rio San Francisco. 6. Rio Aqua Briinca, into Rio Itapicurii. 7. Rio Ipema, into Rio Itapicuru. Rio Lamaras, small creek into Rio Itapicuru. Rio Zinga, small creek into Rio Itapicuru. 8. Rio Itapicuru Grande, headwater of Rio Itapicurii. Rio Paiaia, into Rio Itapicurii. Rio de Jacobina, into Rio Itapicurii. 10. Lagoa Salgado — Rio Salitre into Rio San Francisco. 11. Bom Fim, Rio Amaratii, into Rio Itapicuru. 12. Saa Thome, Rio Salitre, into Rio San Francisco. IJf. Rio Salitre, into Rio San Francisco. Baixa Grande, Rio Paqui into Rio Salitre, into Rio San Francisco. Rio Paqui, into Rio Salitre near Baixa Grande. 21. Finca Amaratii, Rio Itapicuru. Creek on farm emptying into Rio Itapicurii Mirim. 27, 28. Joazeiro, Rio San Francisco. December 6. Barra, fork of Rio San Francisco and Rio Grande. December 12, 18. Januaria, Rio San Francisco. 15. Cachoeira de Pirapora, Rio San Francisco. 23. Lagoa de Joao Pereira Barra, Rio San Francisco. 24- Lagoa de Porto, near Barra, Rio San Francisco. Januanj 4, 1908. Lagoa Barreiras, Rio San Francisco. 6-9. Boqueirao, Rio Grande of Rio San Francisco Basin. 16. Lagoa Parnagua or Paranagua, Paranahyba Basin. 24. Santa Rita de Rio Preto, into Rio Grande, into Rio San Francisco. 27. Rio Preto, ten miles below fork of Rio Sapao. February 4- Cachoeira da Velha, Rio Novo, into Rio Somno, into Tocantins. 6. Strome, Rio Somno. Headwaters. 11. Rio Sapao, into Rio Preto, into Rio San Francisco. Near Prazcr. 15. Rio Preto, into Rio Grande, into Rio San Francisco. 23, 24- Barra, Rio San Francisco. March 2. Queimadas, Rio Itapi- curii. 4- Alagoinhas, Rio Catu. 5. Rio Itapicurii, 12 miles from Timbo. 11, 13. About Bahia; 22, Penedo, mouth of Rio San Francisco. SO. Propria, Rio San Francisco. April 2. Penedo. 6. Maceio, on the coast. 7. Penedo. 10. Acaraju. I4. Cachoeiro, Rio Paraguassu. May 4- Sete Lagoas, into Rio das Velhas, into Rio San Francisco. 10, 11, IS. Creek, ponds, mountain rills near Rio das Velhas. 14- Miguel Burnier, headwaters of Rio das Velhas and tributary of Paraopeba. 19. Sao Joao del Rey, Rio das Mortes, into Rio Grande, into Rio Parana. 21. Sitio, Rio das Mortfes. 22. Serraria, Rio Parahybuna, into Rio Parahyba. 24-28. Creeks, pools, river at Rio Doce. June 1-3. Entre Rios, Rio Para- hyba. 13-16. Campos, Rio Parahyba. 16. Lagoa Feia, south of mouth of Rio Parahyba. IS, 19. Rio Itapemerim and swamp near Munez Freire. 23, 24. Sao Joao da Barra, Rio Parahyba. July 5. Barra da Pirahy. 7, 8. Bom Jardin, above and below falls, Rio Grande, into Rio Parana. .9, 10. Santa Rita de Jacutinga, Rio Preto, into Rio Parahyba. 12,13. Barra da Pirahy, Rio Parahyba. 14<15. Jacarehy, Rio Parahyba. 17,20. Mogy das Cruzes, Rio Tiete, into Rio Parana. 23. Piracicaba, Rio Tiete, into Rio Parana. 23. SOURCES OF THE MATERIAL. 1 1 Sapina, Rio Tiete, into Rio Parand. Santos, coast of Sao Paulo. 25. Alto da Serra, Rio Tiete, into Rio Parana. £6, 28. Mogy, into Santos Bay. Creek at base of mountains, ten miles from Santos. 29. Piassaguera, near Santos. 29. Santos. 31. Rio Pilao, fifteen miles southwest of Santos and Cubatiio, Rio Cubatao. Clear, swift and rock^' creeks seven miles west of Santos. Auguat 7. Mogy Mirim, into Rio Mogy Guassu, into Rio Grande, into Rio Parana. Corrego de Joao de Deus. Twelve miles from Mogy Mirim. H. Rio Paranahyba, into Rio Parana. 18, 19. Jaguara, Rio Grande, into Rio Parana. 25, 26. Mogy Guassu, Rio Mogy Guassu, into Rio Grande, into Rio Parana. September 1. Bebedouro, near Rio Grande, and Rio Parana. 5-9. Piracicaba, into Rio Tiete, into Rio Parana. Above and below big falls. 14-. Salto de Avanhandava, Rio Tiete. 22. Salto das Cruzes, Rio Tiete. 27. Itapura, Rio Tiete; Riberao Azul, twelve miles from Tiete. October 8. Riberao Azul. 11. Miguel Calmon; twenty miles east of Miguel Calmon. Bauliru, Rio Tiete. Salto Grande de Paranapanema, into Rio Parana. November 27. Agua Quente, into Rio Ribeira da Iguape. 28. Cavernas das Areas. Si.xteen miles south- west of Iporanga in Serra do Mar. December 1. Iporanga, Rio Ribeira da Iguape. 5, 8. Xiririca, Rio Ribeira da Iguape. 15, 16. Iguape, Rio Ribeira da Iguape. 22. Serrinha Parana, Rio Iguassii, into Rio Parana. Rio das Mortes, into Rio Iguassu. Creek six miles west of Serrinha, with numerous falls. 27-29. Porto Uniao da Victoria, Rio Iguassu. January 2-4, 1909. Morretes, on Marunby, into Rio Nhundiaquara, into ocean at Pranagua. 17-24. Porto Alegre, Rio Grande do Sul, Rio Guahyba. 26-27. Cachoeira Rio Jacuhy, into Lago dos Patos at Rio Grande do Sul. 29. Santa Maria, Rio Vaccacahy- Mirim, into Rio Guahyba into Rio Jacuhy. 31. Cacequy, Rio Ibacuhy, into Rio Uruguay. February 1. Cacequy. 6. Uruguayana, Rio Uruguay. Rio Negro, Uruguay or Paso de los Torros, into Rio Uruguay. 17. Arroyo Miguelete, Montevideo. 25, 27. San Juan, Argentina. March 4- Rio Colorado, Argentina. Choel-choel, tributary of Rio Negro. 5. Rio Negro. 6. Muddy ponds twenty miles east of town of Colorado. 11. Buenos Aires, Rio de Prata. 23. Asuncion, Paraguay. 30. Cerro de Lambare, five miles below Asuncion in saline swamp. 31. Bays in front of and near Asuncion. April 2. Sapucay, Paraguay. In mountain rills. 5. Arroyo Pona near Sapucay. 7, 8. Arequa, Laguna Ipacary. 11, 13, 14- Villa Hays. 27. Urucum Mountains, 25 miles back of Corumba. 28. Corumba. May 2. Urucum Mountains. 6, 7. Puerto Suarez, 15 km. from Corumba. 9. Corumba. 23-27. Sao Luiz de Caceres, Matto Grosso. June 2, 3. Cam- pos Alegre, Rio Jauru, into Rio Paraguay. San Matias, Bolivia, into Rio Paraguay. 10. Rio Sao Francisco, into Rio Paraguay. 12. Rio Santa Rita, into Rio Paraguay. IS. Rio Petas, Bolivia, into Rio Paraguay. 16. Rio Boa Ventura, into Rio Guapore. 21, 27. Posada, into Rio Guapore, about 40 miles south of Villa de Matto Grosso. 26, 28. Bastos, Rio Alegre, into Rio Guapore. July 8. Below Rio Paragahu, in Rio Guapore. 23. Sixty miles above San Antonio de Guapore. July 29-August 13. San Antonio de Guapore. Augxist 28. Rio Machupo, Bolivia, into Rio Guapore. Twenty miles below San Joaquin. September 4-7. San Joaquin, Bolivia. Rio Machupo. Lake one mile west of town and mud-hole near town. I4, 15. Berlin, Rio Mamore. 10. Rio Mamore, below mouth of Rio Guapore. 28. Guaja-ussu, Rio Madeira. 30. Palo Grande, Rio Mamore. In rapids under stones. October 5. Villa Bella, Bolivia, Rio Beni, into Rio Madeira. 13. Cachoele de Theotonio, Rio Madeira. Whirlpools. 17. Cachoele de Ribeirao, Rio Madeira. 26. Cachoele de Girao, Rio Madeira. Whirlpools. November 2, 3. Sao Antonio de Rio Madeira. 15, 19, 25, 27-29. Manaos, mouth of Rio Negro. 30. Igarape de Cachoeira Grande, two miles out of Manaos. December 2. Manaos. 4- Ten miles above Manaos on Rio Negro. 6, 8. Santarem, Rio Tapajos. 7. Swampy pools of Amazon and Rio Tapajos opposite Santarem. 9. Upper end of island, Amazon, four miles above Santarem. 10. Tapajos, in 12 THE AMERICAN CHARACIDAE. hollow logs — Amazon, one mile above San tarem. 11, 19. Igarape de Jrura entering Rio Tapajos, two miles above Santarem. 12. Igarape de Maica, four miles below Siintarem. 7^. Tapajos in front of Santarem. 15. Island in Amazon, three miles above Santarem. 20. Rio Tapajos at Santarem. 21. Amazon. Half-way between Santarem and Para. 24. Para market. 27. From Gran Para between Belem and Salinas. 29. Braganoa, Rio Caete. 16 kilometers from ocean, 162 from Para. January 1, 1910. Salt water, mouth of Rio Caete. 10. Alcobaca, Tocantins. Below first falls. 15-22. Para market. The Guiana Expedition. The joint Expedition of the Indiana University and the Carnegie Museum, led by myself, collected in British Guiana, between September 9 and December 1, 1908. A detailed account of the results of this Expedition is published as volume 5 of the Memoirs of the Carnegie Museum. The locaUties, enumer- ated from east to west are : — Maduni Stop-Off, Lama Stop-Off , Cane Grove Corner, the Georgetown Trenches, Morawhanna, Mora Passage, and Koriabo and Issorora Rubber Plantations, all in low tidal land; Christianburg, Wismar, and Malah, the latter about one hundred miles from its mouth, on the Demerara River and Bartica, Rockstone, Gluck Island, Crab Falls, Konawaruk, Warra- puta Cataract, and Packeoo Fall, all on the middle course of the Essequibo River. A special effort was made to get a complete series from the Potaro River both above and below the seven hundred and forty-one foot Kaieteur Fall. Collections were made between October 6 and November 4, 1908, at Aruataima, Holmia, and Savannah Landing above the Kaieteur and at Shrimp Creek, Tukeit, Waratuk, Amatuk, Erukin, Kangaruma, Potaro Landing, and Tumatumari below the Kaieteur. Mr. WiUiam Grant my Indian guide sent in additional collections from the Rupununi and the Ireng Rivers. The first series of specimens of this Expedition is in the Carnegie Museum, the second series in Indiana University. Other series are in the Museum of Comparative Zoology, the U. S. National Museum, the Field Museum, Stanford University, the British Museum, the Museums at Amsterdam, BerUn, Vienna, and George- town, British Guiana. The Gimbel Expedition. Through the generosity of Mr. Jake Gimbel of Vincennes, Indiana, Dr. Max Mapes ElUs and Dr. William M. Tucker were enabled to go to British Guiana primarily to gather material for a monograph of the Gymnotidae.^ ' The gymnotid eels of Tropical America. Memoirs Carnegie museum, 1913, 6, p. 109-195, pi. 15-23. SOURCES OF THE MATERIAL. 13 They started from New York in August 1910 and went direct to Georgetown. After collecting in the Demerara and along the coast at Georgetown and con- ducting experiments in the regeneration on various species of the Gymnotidae they ascended the Demerara to canal Number Two. They went through the canal to a tributary of Hubabu Creek, descended to the Demerara and returned to Georgetown. They also ascended the Demerara to Wismar and crossed over to the Essequibo at Rockstone. After collecting on Gluck Island in the Esse- quibo and in a tributary of the Essequibo at the railway crossing between Rockstone and Wismar they returned to the coast. The Colombian Reconnaissance. Preliminary notes of my reconnaissance in Colombia have been published in Indiana University Studies 16 and 18. A detailed account will appear in the Memoirs of the Carnegie Museum. I entered Colombia at Cartagena in December 1911, and left from the same port in April 1912. From Cartagena I went to Soplaviento on the Dique and to Calamar on the Magdalena; from Calamar up the Magdalena River by steamer to La Dorado, collecting at various stopping places, Barbosa, El Blanco, Canaletal, Puerto Wilches, Peiias Blancas, Puerto Berrio. From La Dorado, the route was by rail to the upper part of the Magdalena, collections being made on the way at Honda, especially in Bernal Creek. The upper part of the Magdalena was followed to Girardot, where exten- sive collections were made. From Girardot, the route led first over the western rim of the plain at an elevation of about 8800 feet to Bogota, on an elevated plain among the eastern Cordilleras; collections were made on the plain near Chapinero, north of Bogota, and at Madrid, near the western margin of the plains of Bogota. A return was made to Girardot, from which via Chicoral to Quatro Esquinas, Ibague, Toche, across the Quindio Pass of the central Cordilleras, at an elevation of 11,200 feet, to Boquia, Piedra Moler, and Cartago near the Cauca River. Up the Cauca Valley via Paila, Buga La Grande, Buga to Cali, collections being made at Paila and at Cali, and in the Cauca near Cali. After collecting at Caldas (elevation of 3722 feet), the valley of the Dagua was descended, collections being made at Cisnero (1046 feet), at Cordova (120 feet), and in tide- water. From Buenaventura, on the Pacific coast of Colombia, a steamer was taken up the San Juan River to Puerto Negria; thence a dugout carried the Expedition as far as Istmina; collections were made in both the latter places. From Istmina, after a ride of two hours up a Uttle stream, and across the low 14 THE AMERICAN CHARACIDAE. continental di\'ide (elevation 300 feet above sea-level) the valley of the Atrato was entered near Tambo. By dugout to the settlement of Boca de Raspadura; thence the Raspadura River was followed into the Quito River, then the Quito River. Collections were made at Boca de Certegui and near the town of Quibdo, at the junction of the Quito River with the Atrato. From Quibdo, a steamer was taken to Rio Sucio, where additional collections were made. From Sucio, a steamer carried the Expedition back to the starting point at Cartagena. These collections have been supplemented since my return by material collected by Manuel Gonzales, near Puerto Wilches, at Las Juntas on the Rio Bogota, in the province of Santander, and along the way from Bogota to Villa- vicencio. The first series of the specimens and the duplicates are in the Carnegie Museum, the second series in Indiana University. The Landon-Fisher Expedition to Colombia. A second Expedition into Colombia was made possible by Mr. Hugh McK. Landon and Mr. Carl G. Fisher, of Indianapolis. Mr. Arthur W. Henn and Mr. Charles Wilson, undergraduates in Indiana University, left in December, 1912. They landed at Tumaco, near the south- western corner of Colombia. After devoting about a month's time to the Telembi River, a tributary of the Patia, they separated. Mr. Wilson went to the San Juan River, collecting in the Upper San Juan Basin, the Condoto River at Condoto, and in the San Juan River at Istmina and Tado of the Pacific side, and later on the Atlantic slope at Tambo, Raspadura, Boco de Raspadura, Managru, Quibdo, in the Atrato between Quibdo and Rio Sucio, and especially in the Truando River emptying into the Atrato near Rio Sucio. The Landon Expedition to Colombia and Ecuador. Through the continued liberality of Mr. Hugh McK. Landon of Indianapolis, Arthur W. Henn was able to remain in South America and spend the time between February 15, 1913 and March, 1914 in collecting fresh-water fishes in Colombia and Ecuador. He sailed from Tumaco, Colombia for Barbacoas on February 15th. From Barbacoas he went by packtrain to Tuquerres (10,090 feet), Ancuya (5000 feet), Los Llanos de Sandona (5000 feet), Tambo, Peiiol, and to Guayabillo on the brink of the Canon of the upper Patia. Descending to the Patia River, about SOURCES OF THE MATERIAL. 15 3000 feet collections were made above the mouth of the Guaitara. He returned via Pasto to Tuquerres and Barbacoas. He next descended the Rio Telembi to its mouth and ascended the Rio Patia and the Rio Magui, the first large tributary of the Patia above the Telembi to the village Payan. Returning to Tumaco on the coast he went via Buenaventura to Puerto Negria, the head of steam navigation of the San Juan River. Drifting down the San Juan to the Rio Calima, the last large tributary of the San Juan from the east, he ascended the Calima, collecting in a small creek near its mouth, and at Boca de Guineo about thirty miles from the mouth of the Calima. After ascending the Rio El Guineo to a portage, he crossed over to the San Joaquin and descended it to Buenaventura. He next went to Guayaquil from where several shorter excursions were made. The first of these lead to Naranjito. Collections were made south of Naranjito in a small creek Estero Verdes, a tributary of the Rio Chan Chan, and in a deep river, the Rio Barranca Alta. The second shorter trip from Guayaquil was to the small rivers at Chone and Portoviejo. The third trip took him to Daule, Santa Lucia, and Colimes on the Rio Daule. Returning to Daule he crossed over along a winding cut-off from the Rio Palenque to Vinces. Returning to Guayaquil from Vinces, he went by rail to Quito, collecting en route at Huigra (4000 feet), Rio Bamba (9020 feet), Latacunga (9055 feet) and Quito (9375 feet). A short trip was made from Quito to Mindo (4108 feet). Later he went to El Angel (10,000 feet) and down the Rio Chota or Mira to Maria Luisa, the property of Sn. C6sar Mena. Collections were also made in the Rio Chota at Guallupi (5,000 feet). He returned to Quito and later went via Angel by packtrain to Barbacoas and steamer to Tumaco. By the various expeditions outlined above, collections have been obtained from most of the Pacific slope streams between Panama and Peru. The first series of the specimens collected in the Landon Expedition is in Indiana University. Other series are in the Carnegie Museum, the Museum of Comparative Zoology, the U. S. National Museum, the Field Museum, and the Museum of Stanford University. 16 THE AMERICAN CHARACIDAE. Central American Expeditions. From time to time specimens have been secured by exchange with the Field Musemii of Chicago. These had been collected by the late Dr. S. E. Meek in various parts of Mexico and Central America. Accounts of this material have appeared in various pubhcations of the Field Museum. In collaboration with Mr. S. F. Hildebrand, Dr. Meek was engaged at the time of his death on a report of the fishes of Panama. Mr. Hildebrand has com- pleted the report in the Laboratory of Indiana University and has given me the opportunity to examine many of the specimens collected in Panama. Dr. Meek as the representative of the Field Museum, and Mr. Hildebrand, Director of the Beaufort Laboratory of the Bureau of Fisheries, as the representative of the Smithsonian Institution, spent two seasons in Panama to collect fishes. A report on their itinerary and on their collections has been published by the Field Museum (Field museum publication 191, zool. ser., 10, 1910, p. 217-374, pi. 6-32). They collected in various localities of the east-slope Chagres Basin, many of which are now covered by Gatun Lake, and in others of the Pacific slope Grande basin of the Canal Zone. In addition they collected in the basins of the Rio Chepo, and Tuyra of the Pacific slope, south of the Canal Zone, and in a few of the smaller rivers both north and south of the Canal Zone on both the east and west slopes. The Expedition to Guatemala. During January, February, and March of 1915, Mr. Newton Miller assisted by Messrs. E. B. Williamson, C. C. Deam, and Prof. J. Hines collected fishes at Tenedores, Los Amates, Algeria, Gualan, Zacapa, and El Rancho in the Motagua Basin of Guatemala. Additional collections were secured at Puerto Barrios and Santa Lucia. The first series of specimen is in Indiana University. THE CHARACIDAE. 17 THE CHARACIDAE. KDermopteres DumSril, Zool. analytique, 1806, p. 146. Characidae Gill, Mem. Nat. aead. sci., 1893, 6, p. 131. Proc. U. S. N. M., 1895, 18, p. 206. = CAarociretyac Boulengbr, Poissons bassin Congo, 1901, p. 132. Cambridge natural history. Fishes 1904, p. 575. Cat. freshwater fishes Africa, 1909, 1, p. 174. = Characidae Eiqenmann, Rept. Princeton univ. e.xped. Patagonia, 1909, 3, p. 252; 1910, p. 420. = Characiformes Reoan, Ann. mag. nat. hist., 1911, ser. 8, 8, p. 15. > Characidae Regan, Loc. cit. < Erythroides Valenciennes, Hist. nat. poissons, 1846, 19, p. 480. = Erythrinidae Richardson, Encyol. Brit., 1856, ed. 8, 12, p. 250. > Erythrinmdei Bleeker, Enum. sp. piseium Archipel. Indico, 1859, p. 31. > Erythrinidae Gill, Ann. Lye. nat. hist. N. Y., 1858, 6, p. 410. Mem. Nat. acad. sci., 1893, 6, p. 131. Proc. U. S. N. M., 1895, 18, p. 206. > Erythrinidae Cope, Proc. Amer. assoc. adv. sci., 1872, p. 333. > Erythrini Fitzinger, Sitzungsb. Akad. wiss. Wien, 1873, 67, abth., p. 37. > Myletidae Adams, Man. nat. hist., 1854, p. 108. > Gastropelecidae Regan, Ann. mag. nat. hist. 1911, ser. 8, 8, p. 19. > Xiphosiomalidae Regan, Loc. cit., p. 20. > Anastomidae Regan, Loc. cit., p. 20. > Hemiodontidae Regan, Loc. cit., p. 21. > Citharinidae Regan, Loc. cit., p. 21. Zoological Position. The Characins are a family of Ostariophysi which, with the other famiUes of this superorder, are now at their prime. They are the dominant family of fresh-water fishes in Tropical America and they play a prominent role in Africa. The Ostariophysi, which include most of the fresh-water fishes of the world, are distinguished from all other fishes by the pecuUar arrangment of a series of 18 THE AMERICAN CHARACIDAE. ossicles for placing the air-bladder in communication with the auditory appa- ratus. The first four vertebrae are modified. The first vertebra lacks the superior arch which is replaced by the "claustrum" and "scaphium" of the Weberian apparatus; the principal ossicle of the series, the " tripus," is associated with the third vertebra consisting of the rib and parapophysis of the third vertebra. The " iritercalarium," representing the neural arch of the second vertebra, is imbedded in the ligament extending from the tripus to the scaphium. Very frequently the air-bladder comes in close contact with the skin, forming a pseudotympanum above the pectorals. In the Characins the area is not unfrequently marked by a humeral spot, a gathering of pigment cells from contiguous areas. This spot may become shifted away from the tympanum. The air-bladder is usually connected with the intestine by a duct. The pectoral girdle is suspended from the skull by a long posttemporal ; the mesocoracoid is present and the ventral fins are abdominal. The orders and suborders of the Ostariophysi may be separated by the following key: — a. Maxillary bone usually a toothless vestige carrying a barbel; no subopercle or symplectic; no scales; supraoccipital and parietals coossified; mouth usually with teeth; ribs attached to trans- verse apophyses; skin naked or covered with bony plates. Usually an adipose dorsal. Nematognathi or Sildroidea. aa. Maxillary usually well developed, not forming the base of a barbel, but sometimes one or more small barbels at or near its end; subopercle and symplectic present; parietals distinct from the supra- occipital; thoracic vertebrae without parapophyses; ribs mostly sessile; naked or with scales. Plectospondyli or Cyprinoidea. ft. Lower pharyngeals falciform, parallel with the gill-arches; jaws toothless; brain-case produced between the orbits; basis cranii simple; two superior pharyngeals; mouth without teeth, more or less protractile; no adipose fin (Eventocnathi or Cypriniformes). bb. Lower pharyngeals not falciform; three basal branchihyals; basis cranii double, sometimes with myodome; one to four superior pharyngeals; mouth usually not protractile, usually with teeth. c. Anus submedian; body variously shaped, never eel-shaped; dorsal and usually an adipose fin present; ventrals abdominal (IIeterognathi or Characiformes). cc. Anus at throat; body eel-shaped; dorsal absent, or reduced; ventrals absent. .. (Gymnonoti). The Eventognathi do not enter the Neotropical realm. The Gymnonoti are confined to it, the Heterognathi, as stated, occur in Africa and South America. The Nematognathi have an all but universal distribution. Sagemehl (Morphologische jahrbiich, 1884, 10) pointed out the similarity of the Erythrinoids to Amia, and suggested the derivation of the family from the Holostci and more particularly from the Cycloganoidea. Boulcnger (Poisson bassin Congo, 1901) considers the Characidae, more particularly the Erythrins, THE CHARACIDAE. 19 as the primitive Ostariophysi from which the Cyprinidae (Gymnonoti) and Nematognathi have been derived. These famihes seem to him to represent three states in the evolution of Teleosteans, the Cyprinidae and Siluridae being derived apparently from a common ancestor, very near the Characins, these latter ranging themselves to the holostean Ganoids through the Erythrins. Later Boulenger (Cambridge natural history. Fishes, 1904, p. 575) refers to them as "a very generalized type, although perhaps not directly derived from the bony Ganoids." The relationship recognized by him he represented as follows : Eventognathi Nematognathi Gymnonoti Heterognathi Rowntree (Trans. Linn. soc. Lond., 1903, ser. 2, 9, p. 78) concludes his extensive account of the visceral anatomy of Characins with the following words: — "With the possible exception of the indications of a cellular air-bladder, there appears to be nothing in the visceral anatomy of the Characinidae which strengthens the deductions made from the skull as to the Amioid affinities of the group. In opposition to such deductions are especially the cystoarian ovaries, the asymmetric ductus pneumaticus, the presence of pyloric appendages and the absence of all trace of a valvular conus and of an intestinal spiral valve." In view of the observations of every naturalist who has studied the Char- acins, notably Boulenger with the African species, and the author with the American forms, it is doubtful whether the similarity of the Erythrins with the Ganoids is anything more than one of the innumerable radial adaptations this plastic family has undergone. Radial Adaptation. Boulenger's remarks (Poissons bassin Congo, 1901, p. 132-135) concerning the African representatives may be translated as follows: — • This is a very natural family whose internal organization shows a great affinity with the Cyprinidae but whose exterior appearance, due to adaptation to various modes of life, varies so much that the beginner in African ichthyology will mistake its various representa- tives as near the Salmonids, pikes, and roaches of Europe. It is also impossible to state in a 20 THE AMERICAN CHARACIDAE. few words the marks distinguishing the members of this family from such other fishes with abdominal ventrals and soft rayed fins as the Siluridae, Mormyridae, and Nototeridae. The presence of an adipose dorsal behind the rayed dorsal, although absent in some non-African genera, is a common character of the African fauna which it shares only with the Siluridae, which differ by their naked skin [one of the South American Characins is naked]. The Salmonids which also possess this adipose fin belong, on account of their general structure, to the same great group as the Clupeids whicli lack it and are strangers to Africa south of the Atlas. One of the prime characters used in separating the Characinidae from its nearest related families is the structure of the mouth described as bordered by the premaxillaries in the middle and the maxillaries on the sides, while in the Cyprinidae and Siluridae the upper jaw is said to be bordered by the premaxillaries only. While this definition is valid for the most of the representatives of this family we must not forget, as Segemehl has impressed on us, that there are numerous exceptions which destroy the diagnostic value of this character. Thus the maxillary is so reduced in Serrasalmo, Citharinus, Eugnatichthys, Phago, being confined to the angle of the mouth that it is proper to question if this bone really forms part of the border of the mouth and very recently I have been able to determine that it is absolutely excluded in Ichthyoborus and Neoborus. On the other hand, among Siluridae (Nematognathi), whose premaxillaries are greatly reduced the mouth is bordered laterally by the maxillaries. Chaca has a large maxillary bordering the mouth, Diplomystes and Eutropichthys have a toothed maxillary, and among the Cyprinidae Catostomus shows us the premaxillary and maxillary together forming the mouth border. There are otherwise very great differences in the structure of the jaws among the fishes which are justly united in one family, the Chara- cinidae. Thus the premaxillary, while never protractile, [I have since found Biviljranchia, a new genus, with protractile premaxillaries] is often vertically mobile (like the lower jaw), Ichthyoborinae; the maxillaries are either ankylosed to the premaxillaries or mobile on them, and in the majority of genera they do not directly articulate with the cranium, the only African exception being Sarcodaces. The character given by other authors to distinguish the Characins from the Cyprinids, i. e. the non prolongation of the brain capsule between the orbits in the former is not more constant, for in Citharinus and Xenocharax, for example, it extends forward to the nasal region. There are all degrees between this and the more usual one from which the objection- able definition is drawn. The orbitosphenoid, undivided and often membranaceous, forms a large interorbital septum in front of the cerebral cavity or below its anterior part. We have here an important character although it may not be constant, a specialization in this reduc- tion of the anterior part of the basis cranii and in its replacement by a thin interorbital parti- tion. The presence of teeth in the jaws does not distinguish the Characins from the Cyprinids because both African and American genera lack them. The teeth vary enormously in struc- ture and furnish valuable characters to distinguish numerous genera of this family. They are found on the premaxillary, the lower jaw, sometimes they are also present on the maxil- laries. They are but rarely found on the palate and none of the African genera offer any examples. The branchiostegal rays number only 3 to 5. There are four branchial arches. Pseudo- l>rancliia are lacking or are rudimentary and glandular. The coracoid often forms a \entral ridge. The scapulars, epicoracoid, and postclavicle are distinct. The ventral rays number 10 to 13. The first pectoral, the dorsal, and anal rays are but rarely ossified and never form a formidable spine as in the Siluridae. Vertebrae in the African species vary from 17-33 -|- 13-24 = 33-57. THE CHARACIDAE. 21 The scales, large or small, are ciliated in certain genera — a character very rare among tjie Physostomes. As Valenciennes says of Distichorlus, the free part of the scales is truly ctenoid, the rest cycloid. The head always lacks scales, a rare thing in fishes. The lateral line organ is represented on the opercle by a branch from the suborbital. The air-bladiler is ahvay.s divided into two parts Ity a constriction. The anterior part is much smaller than the posterior. In some, Alestes, the air-bladfler may be prolonged along the right side of the hemapophyses and interneurals as far as the posterior end of the anal. There are generally 10-25 coeca on the stomach and this number may be raised to 35 or 40 (Hydrocj'on, Citharinus). The intestine in tlie carnivorous genera is short and makes but one loop; on the other hand, it is excessively long and with many windings in the herbivorous t.\'pes. In looking o\ er this review of the characters of the family of Characins it is seen that there is none that taken alone and allowing for the exceptions, justifies its separation from the Cyprinids. We are forced to content ourselves with a combination of characters, any one of which taken by itself is insufficient. These words apply almost verbatim to the American members of the family. The American Characins range from the border of the United States to some distance south of Buenos Aires. They form about one third of the entire South American fresh-water fauna and have diverged in adaptation to diverse food, diverse habitat, and diverse enemies to fill nearly every niche open to fishes. The ends of the three hues of adaptation to different food give us mud- eating forms, with long intestinal tract and no teeth/ flesh-eaters with shear- Uke teeth that are able to cut their way out of nets, attack large fishes, horses, and bathers, and conical-toothed forms with sharp, needle-like teeth and com- paratively huge fangs. Greater diversity could scarcely be imagined, and one is led to suspect that some of the forms are over-adapted. In their divergence in form they have reached almost every conceivable shape, and have approached or paralleled many members of the diverse families of North American fresh- water fishes. Our shads and fresh-water herrings have their counterparts in Elopomorphus, Potamorhina, and Psectrogaster, our salmons are paralleled by Salminus and Catabasis, and our minnows by Astyanax and its relatives. It takes but a slight flight of the imagination to detect the striking similarity of Luciocharax to our gar pikes; our mullets are duplicated by Prochilodus; our top minnows are mimicked by Nannostomus. Bivibranchia, a recent dis- covery, shows a close similarity to Albula, and even our festive darters are duplicated by members Characidium of this most remarkable family. This plasticity of the family in both America and Africa, and the apparent if not real duplication of forms in the two continents, is the more remarkable ' The toothless forms are not represented in Africa where members of the Cyprinidae replace them. 22 THE AMERICAN CHARACIDAE. when we consider the very probable long separation of the African and Ameri- can sections of the family, and that the present forms in Africa and America have probably independently evolved in the two continents from a common rudiment. As von Ihering first pointed out for animals in general, and as I have more particularly shown for the fishes, there has probably been no inter- migration between the two continents since or previous to the early Tertiary. Before the Tertiary von Ihering has postulated an Archhelenic continent between Africa and South America from which the two continents probably derived the ancestors of the Characins among other elements of the present fauna. It is known that the Characins have inhabited South America since the early Tertiary at a time when the tropical part of the continents probably con- sisted of two islands, one occupying the Guianas, the other the plateau of eastern Brazil. From this small area and ffbm the primitive group of Characins inhabiting it, the continent and family grew contemporaneously to their present proportions. They spread from these centers over the developing continent till they met unfavorable cUmatic conditions in the south, and high mountains or contestants in the north. They spread southward in decreasing numbers to and beyond the barren areas of Argentine to the edge of Patagonia, and they spread northward decreasing rapidly in numbers at Panama, only one of them peaching the United States, three southern Mexico, and three the Motagua River. In South America they have always flourished, their only competitors in the rivers of the growing continent being the indigenous fishes contemporaneously developing from similar small rudiments. They have never been hampered or affected by intrusive elements. Their territory has from the first been unap- proachable to foreign fresh-water fishes, and the marine species that have accli- mated themselves have not diverged to any extent from their marine relatives. The African section, on the other hand, came into competition with emi- grants from India, and, according to Boulenger, the affinity of the entire Afri- can fauna with Asia, "is much greater than with America which is emphasized by the fact that the genera in the two first named regions are identical, while America possesses genera very closely related but not identical with those of Africa." The American Characins have diverged, as stated before, in the most amaz- ing manner in almost all possible ways towards diverse forms assumed by fishes. Some are long and slender, some nearly as deep as long, some spindle-shaped, others compressed. It is, however, in the teeth and alimentary canal that we find the greatest diversity, as has already been pointed out, and as will be THE CHARACIDAE. 23 described in detail. The fact that different authors have associated different members of the Characidae with the herring, trout, cyprinids, and poecilids indicates in a measure their versatility. Literature. The earliest observations on Characins are recorded on some ancient monuments of Egypt. Meckel (Die fische Aegyptens chronologisch dcr zeitfolge ilirer ersten und spiiteren wissenschaftlichen kenntnissnahme nach geordnet, p. 213 of Abbildungen und beschreibungen der fische Syriens, 1843) identifies Distichodus niloticus and Citharinus geoffroyi from mural decorations of graves near the Pyramids. Boulenger (Fishes of the Nile, 1907, p. 150) finds that Citharinus citharus Geoffrey is represented on the mural paintings of the tombs at Giza and Sapara, at Deir el Gebrawi, and on the tomb of Ti at Sakkara. The American Characins were first brought to the notice of naturalists by Marcgrav, who, in 1648 in his Historiae rerum natm-alium Brazihae, 4, described the following : — Curimata, p. 156 = ProchUodus argrniciis (Agassiz). Tareira do Rio, p. 157 = llopUas malabaricus (Bloch). Piraya, p. 164 = Pygocentrus piraya (Cuvier). Maturaque, p. 169 = Iloplias malaharicns (Bloch). Piabucu, p. 170 = Piahucus dcntatus (Koclreuter). Piaba, p. 170 = An Anostomus (?) Ten years later, in 1658, Piso followed this account in his Historiae naturalis et medicae Indiae Occidentalis libri quinque, which is but a second edition of the former work, with figures and descriptions of the same species: — Piabucu, p. 66, Piaba, p. 67, Maturaque, p. 67, Tareira II do Rio, p. 68, Piranha, p. 69, Curimata, p. 70. The detailed history of the African section of the family will not be given here.' The number of known African species is far smaller than the number of American species which exceeds six hundred. In Seba's Locupletissimi verum naturalium thesauri accurata descriptis, ' Boulenger (Fishes of the Nile, 1907, p. 117) indicates that the Salmo niloticus Linne, Syst. nat., ed. 12 is not the Salmo niloticus of Hasselquist. The date, 1757, of Hasselquist, however, being pre- Linnean, the name Salmo niloticus must be applied to the species of Linnd ed. 10. The 5. niloticus of the Systema ed. 12, which is the same as that of ed. 10, is, according to Boulenger, identical with Myletes haremose Joannis, and the latter = Alestes baremose of Boulenger, p. 117, should stand as Myletes nilo- ticus. Further according to Boulenger, p. 141, the Salmo niloticus of Hasselquist is Salmo aegyptiacus of GmeUn. The name Salmo niloticus being a synonym of Myletes niloticus cannot be used for any other species and the oldest name after Hasselquist should be applied to his species. This oldest na me is Salmo aegyptiacus and since this is a Distichodus the species should be Distichodus aegyptiacus inste ad of Distichodus niloticus as given by Boulenger, p. 141, 24 THE AMERICAN CHARACIDAE. 1758, over the signature of Artedi, but according to Gill ^ by another author, the name Tetragonoptrus, coined by Klein and originally intended for entirely dif- ferent fishes, was applied to the species now known as Tetragonopierus argenteus. In 1754 and 1756 Gronovius in his Museum ichthyologicum defined the following genera : — Charax, 1754, p. 19. Gasteropelecus, 1756, p. 7, pi. 7, fig. 5. Anostomus, 1756, p. 13, pi. 7, fig. 2. Erythriinus, 1756, p. 6, pi. 7, fig. 6. The two species referred by Gronovius to Charax were subsequently in- corporated in the genera Charax (gibbosus) and Astyanax (fimaculatus) . The Gasteropelecus is the Gasteropelecus sternicla of present authors. His Anos- tomus is Anostomus anostomus and Erythrinus is Erythrinus salmoneus. In 1777 Scopoli (Introductio ad historiam naturalem * * *) adopted these genera into the binomial system with Gronovius species as the types. Linn^ abandoned all of these genera and distributed the species known to him in the genera Clupea, Cyprinus, and Salmo. The species described or recognized by Linn6 in his tenth and twelfth editions and in Gmelin's the thir- teenth edition of the Systema Naturae are given in the following table modified from that of Gill (Proc. U. S. N. M., 18, p. 213) : 1758 1766 \ 1788 Page No. Page No. Page No. Sdmo argentimts - 2 511 12 1372 12 Piabunts dcntatus gibboDus 311 19 513 20 1384 20 Charax gibbosus. notatus 513 21 1385 21 Astyanax fasdatus bimaculutus 311 20 513 22 1385 22 Astyanax biinacvlatus immaculatus 312 21 513 23 1385 23 (?) cyprinoidcs 514 25 1385 25 Curimatus cyprinoides niloticns 312 22 514 26 1386 26 Mylctcs nihticAis acgyptiacus 1386 49 Distichodus aegyptia- cus & rostratus pxdverulmtus 312 23 514 27 1386 27 Astyanax ? rhombcus 514 28 1386 28 Scrrasal mo rhombcus anostomus 312 24 514 29 1387 29 Anostomus anostomus f Clupea sima ^ 319 6 524 7 y '? sternicla 319 7 524 3 1384 48 Gastc ropclcc us stern id a Cyprinus cephalus part! '322 7 527 7 1417 6 Erythrinus ccphcdus dentex 325 531 26 1383 47 Myletes dentex & bare- mose ■ I'roc. U. S. N. M., 1895, 18, p. 225-227. * The generic names arc those used in the tenth edition. ' Clupa siynn is credited to Asia by Linn6. It is placed in the synonymy of sternicla by Bloch, p. 418, I know nothing further about it. THE CHARACIDAE. 25 The total number of species known to Linn6 is thus seen to have been from eight to ten in 1758 and from twelve to fourteen in 1766, to which but a single species had been added at the time of the publication of the 13th edition in 1788. Lhm6 apparently knew none of the species described by Marcgrav except Marcgrav's piabucu which is Salmo argeniinus Linn6. In 1794 Bloch in his Auslandische fische added a number of species without, however, recognizing any distinct Characinid genera. In Schneider's edition of Bloch's Systema ichthyologiae published in 1801 the following species were recognized: — Sy7iodus malabricm Bloch, p. 397. (tafel 392 of Bloch). Synodiis erythrinus Bloch & Schneider, p. 397. Synodus tareira Bloch & Schneider, p. 398, pi. 79. Synodus palustris Bloch & Schneider, p. 398. Sabno friderici Bloch, p. 403. (tafel 378 of Bloch). Salmo fasciatus Bloch, p. 403. (tafel 379 of Bloch). Salmo argcntinus Linne, p. 403. (tafel 382, fig. 1 of Bloch). Salmo rhomheus Linn^, p. 404. (tafel 383 of Bloch). Salmo falcatus Bloch, p. 404. (tafel 38.5 of Bloch). Salmo odoe Bloch, p. 405. (tafel 386 of Bloch). Sabno pulverulentus Linne, p. 406. Salvia dcntex Linne, p. 407. Salmo cdcntulus Bloch, p. 412. (tafel 380 of Bloch). Salmo melanurus Bloch, p. 412. (tafel 381, fig. 2 of Bloch). Salmo unimacuhtus Bloch, p. 412. (tafel 381, fig. 3 of Bloch). Salmo himacidatus Linne, p. 413. (tafel 382, fig. 2 of Bloch). Salmo anastomns Linne, p. 414. Salmo Jiiloticus Linne, p. 414. Salmo cyprinoides Linne, p. 414. Salmo curimata Bloch & Schneider, p. 417. Salmo aegyptius Gmelin, p. 418. (aegyptiacus of Gmelin). Salmo gasteropelecvs Gmelin, p. 418. Salmo immaculatus Linne, p. 419. Salmo gibhosus Linne, p. 419. In 1802 Lac^p^de (Histoire naturelle des poissons) created for Salmo rhom- heus Linn6 the genus Serrasalmo and adopted the Charax of Gronovius as Characinus for piabuca, dentex, gibbosus, notatus, bimaculatus, immaculatus, cyprinoides, niloticus, nefasch, pulverulentus, anostomus, friderici, fasciatus, melanurus, and odoe. The real advance in our knowledge of the relationships of the Characins did not begin until several years later when Cuvier (1817) published his R6gne animal, and a series of articles in the Memoires Museum d'histoire naturelle. In the Regne animal (p. 174) he recognized Erythrinus which he placed in his Clupes and the following genera which were included in 26 THE AMERICAN CHARACIDAE. the family of Salmones: — Characinus, Curimatus, Anostomus, Serrasalmo, Piabucus, Tetragonopterus, Myletes, Hydrocynus, Citharinus, and Gastero- pelecus.' In 1829, Memoires Museum d'histoire naturelle, 4, he further defined Chalceus and in 1819, 6, apparently substituted Hydrocyon for Hydrocynus. A great advance towards a knowledge of the South American Characinid fauna was made by Spix and Agassiz in the Selecta genera et species piscium Brasiliensis, 1829. They defined Prochilodus Agassiz (= Pacu Spix), Anodus Spix, Lepori- nus Spix, Schizodon Agassiz, Sahninus Agassiz, Hiphorhynchus Agassiz ( = Acestrorhynchus Eigenmann), Rhaphiodon Agassiz (= Cynodon Spix), and Xiphostoma; they also described many new species. Up to tliis time and for several years later the Characins were distributed among the Salmonids and Clupeids and the peculiar parallelism between some of the genera of these families and the genera of the Characidae made such an association seem natural. In 1842 Johannes Miiller in his treatise on the air- bladder of fishes (Monatsb. Acad, wiss., Berlin, June 1842 and Arch. anat. u. phys., 1842, p. 307) described the genera Macrodon ( = Hoplias) and Hemiodus and united all of the Characins in his new family Characinidae. In 1844 Miiller and Troschel published a synopsis of the known genera (Wiegmann's archiv, 1844, 1, p. 81) and defined the new genera Chilodus, Distichodus, Alestes, Brycon, Exodon, Epicyrtus, Hydrolycus, Pygocentrus, Pygopristis, Catoprion, and Myleus. They followed this in 1845 by the first monograph on the Characinidae, Die familie der Characinen (Horae Ichthyo- logicae, 1, 2). Here all of the then known genera, including the new genus Agoniates are described and the known species enumerated. This work by Miiller and Troschel was up to that time by far the most important as well as the most comprehensive work on the Characins. It is the first of three general accounts that have appeared. In it were recognized thirty-one genera and eighty-eight species. Of these twenty-seven genera and eighty species were American, the remainder African. Miiller and Troschel's work was closely followed by the second revision of the group. Cuvier and Valenciennes in the 19th (1846) and 22nd (1848) volumes of their Histoire naturelle des poissons described many species and the genera Lebiasina and Pyrrhulina, 19 and Parodon, Piabucina, Tometes, Mylesinus, Chalcinus and Cynopotamus, 22. A retrograde step was taken in rejecting the Characinidae and including the genera in the Salmonidae. • Gasteropelecus is attributed to Bloch. THE CHAKACIDAE. 27 In 1854 Girard (Proc. Acad. nat. sci. Phil., 6, p. 199) described the genus Cheirodon, the first Characin reported from the Pacific slope of America. This genus was afterward found to be widely distributed on the eastern slope. In the same year Baird & Girard (Proc. Acad. nat. sci., Phil., 7, p. 27) described Astyanax argentatus, the only species that reaches the United States and the first to be recorded of the overflow from South America northward. The notable work by Castelnau, (1855) Exped. Amerique Sud. Poissons, while containing many figures and description of species added nothing to our understanding of the relations of the various members of the family. In 1858 Gill published (Ann. Lye. nat. hist., N. Y., 6) a short paper on the fresh-water fishes of Trinidad. In it he recognized Erythinus and Macrodon as forming a distinct family and described a new subfamily, Stevardiinae. He also established the genera Poecilurichthys and Hemigrammus. In the following year (1859) appeared a very important contribution to the knowledge of the Characins, Zur familie der Characinen, by Kner. He defined the new genera Microdus, Rhytiodus, and Bryconops and described and figured a large number of new species from various parts of South America. Erythrinus and Macrodon are excluded from the Characidae on account of the absence of an adipose fin. The same author (1863) defined the genera Pseudochalceus, Chalcinopsis, and Saccodon, the latter in connection with Steindachner who has since contributed so much to the knowledge of tropical American fishes. In the same year Giinther defined the genus Crenuchus. In 1864 we have the culmination of an epoch in the history of the Characins. Before 1802, at least during the binomial period, naturalists concerned them- selves altogether with the description of new species. Beginning with Lac^pede, in 1802, we have a series of descriptions of new genera by Cuvier, Spix, Agassiz, and Miiller. To this period belongs the work of Cuvier and Valenciennes although it appeared later. In 1844 and 1845 we have the first attempts by Miiller and Troschel of a philosophic review of the material that had been accumulated. All subsequent work was tinged by Miiller and Troschel's Die famiUe der Characinen. In 1864 was published the second revision of the family. Cuvier and Valenciennes's work being an enumeration, or descriptive catalogue, rather than a revision, and Kner's works being confined to American species. Giinther, in the Catalogue of the fishes of the British Museum, 1864, 6, reunited the Erythrinidae with the Characinidae. He defined or used for the first time the generic names Caenotropus, Brachyalestes, Creatochanes, Hemibrycon, Scissor, Creagrutus, Anacyrtus, Roestes, Roeboides, Hystricodon, 28 THE AMERICAN CHARACIDAE. Sarcodaces, Oligosargus, and Ichthyborus. Gtinther recognized in all forty- seven genera and two hundred and five species. Of these forty genera and one hundred and eighty-one species are American. Giinther groups the genera in a number of subfamilies whose characters are largely the presence or absence of an adipose fin, perfection or imperfection of dentition, the length of the dorsal fin, and the character of the gill-openings. The following synopsis will indicate his subfamiUes and the characters on which they are based : — a. Adipose fin absent Erythrinina,^ American. aa. Adipose fin present b. Dentition imperfect c. Dorsal fin short Curimatina, American. cc. Dorsal fin rather long Citharinina, African. 66. Dentition well developed d. Dorsal fin short e. Gill-openings narrow, the giU-membrane grown to the isthmus /. Elements of the jaws separate g. Nasal openings remote from each other Anoslomatitia, American. (gg. Nasal openings close together Nannocharacina, African). ee. Gill-openings wide, the gill-membrane not grown to the isthmus. h. Teeth compressed, notched, or denticulated. Tetragonopterina, American and African. hh. Teeth all conical Hydrocyonina, American and African. (//. Both jaws very movable, their lateral halves being united into one piece. Phagonina, African). dd. Dorsal fin rather long i. Gill-openings of moderate width, the gill-membrane being attached to the isthmus. Distichodonlina, African. ii. Gill-openings wide, the gill-membrane not being attached to the isthmus. j. Belly rounded; jaws with conical teeth Ichthyborina, African. jj. Belly rounded; canine teeth none. . . .Crenuchina, American and African. jjj. Belly with a spinous serrature Serrasahnonia, American. Giinther immediately (1865) added the Phagonina and shortly afterwards (1867) the Nannocharacina from Africa, assigning the latter to a place between the Anostomatina and the Tetragonopterina. Both are included in parentheses in the foregoing synopsis. Giinther's work stimulated exploration and description to a great extent, and following the publication of his Catalogue we have contributions by Giinther himself between 1864 and 1900; by Gill 1864-1903; Reinhard 1866; Cope between 1870 and 1894; Lutken 1874-1890; Steindachner 1875 to the present; Boulenger from 1887 to the present; Eigenmann and Eigenmann from 1889 to the present; Garman 1890-1895; Perugia 1891-1897; von Ihering 1893 to ' Including the Stevardiinae of Gill. THE CHARACIDAE. 29 the present; Berg 1895-1901; Lahille 1895 to the present; Ulrey 1895; Regan 1900 to the present; Meek 1885-1916; Fowler; Nichols; Hildebrand and Mrs. Marion Durbin Ellis. The discovery of the numerous species and genera since 1864 are too com- plicated for detailed enumeration. Reinhard and Liitken considered chiefly the species inhabiting the basin of the Rio das Velhas a tributary of the Rio San Francisco, Central America and Trinidad. Cope, 1870-1878, dealt largely with species of the Upper Amazons, and later with species from Rio Grande do Sul. Steindachner has published descriptions of numerous new species, col- lected by Agassiz and his associates during the Thayer Expedition and in part by liimself and correspondents. Boulenger has reported on numerous collec- tions received by the British Museum from different parts of South America. Eigenmann and Eigenmann have reviewed the Erythrininae and Curimatinae; Garman has published critical revisions of a few genera; Ulrey has reviewed parts of the Tetragonopterinae. Perugia reported on various collections re- ceived by the Museum of Genoa. Regan is describing species from the collec- tions of the British Museum and Holmberg, von Ihering, Senior and Junior, Berg, Lahille, Ribeiro, and Goeldi were the first of a group of resident naturalists who have made important observations on their own faunae, chiefly eastern Brazil and Argentina. While many of the authors suggest modifications in parts of the system proposed by Giinther, only Eigenmann and Regan con- cerned themselves with the broader questions of the classification of the Characins. In 1884 Sagemehl demonstrated the close relationship of the catfishes, electric eels (Gymnotidae), Cyprinidae, and Characidae, all of which he grouped, on account of the conamon possession of the complicated Weberian apparatus, in the super order Ostariophysi. Fm-ther studies on the anatomy of the Characins were published in 1903 by Rowntree. Gill, in 1893, (Families and subfamilies of fishes) admitted the two families of Heterognaths, Characinidae, and Erythrinidae. He had defined these in 1858 and redefined them in 1895 (Proc. U. S. N. M., 18) when he intimated the existence of a third family, the Citharinidae. Together with Miiller and Troschel and Kner he considers that the Characinidae, even after the exclusion of the Erythrinidae, "constitute a heterogenous group." Gill recognizes the sub- families Erythrininae, Pyrrhulininae, Lebiasininae, Tetragonopterinae, Ser- rasalmoninae, Hydrocyoninae, Myletinae, Distichodontinae, Anostominae, Curimatinae, and Citharininae. 30 THE AMERICAN CHARACIDAE. In a recent paper (Cambridge natural history. Fishes, 1904) Boulenger while recognizing that "The classification of the family is still in an unsatis- factory state "divides them into the following groups (hardly deserving the rank of subfamiHes) " : — A. Erythrininae American. B. Hydrocyoninae African and American. C. Serrasalmoninae American D. Ichthyoborinae African. E. Xiphostominae American. F. Anostominae American. G. Hemiodontinae American. H. Distichodontinae African. I. Citharininae African and American. Regan (Ann. mag. nat. hist., 1911, ser. 8, 8) comes to quite different con- elusions in regard to the "groups" of Boulenger. He divides his Characiformes, i. e. the Heterognathi of authors, into six families (a) the Characidae which equals the Erythrininae, Hydrocyoninae, and Serrasalmoninae of Boulenger, (b) the Xiphostomatidae = Xiphostominae of Boulenger, (c) the Anostomidae = Anostominae of Boulenger plus Curimatus and Prochilodus, (d) the Hemiodon- tidae = Hemiodontinae of Boulenger, (e) the Citharinidae = Ichthyoborinae, Distichodontinae, and Citharininae exclusive of Curimatus and Prochilodus of Boulenger, and (f) the Gastropelecidae proposed for the flying Chara- cins included in the Hydrocyoninae by Boulenger. Regan's paper offers some criticism of my classification published in the Reports of the Princeton University Expeditions to Patagonia, 1909, 3, p. 253- 256. In part I heartily agree with Regan, and long ago came to some of the conclusions reached by him. There are, however, many points in Regan's paper in which I think his conclusions are at variance with the facts. I reserve my criticism of Regan's strictures and a discussion of the general classification of the Characins until my study of all the subfamilies is completed. The keys of the subfamilies given in the Patagonian reports do not agree entirely with my present views, (c/. Rept. Princeton vmiv. exped. to Patagonia, 1909, 3, p. 253-256). Geographical Distribution. In the study of the Characins, as in that of all Tropical American fishes, the question of the distribution of the genera and species must in the future be among the first topics to be considered. In 1S91 and 1892 Eigenmann and Eigenmann enumerated all the known THE CHARACIDAE. 31 Tropical American fishes. They (Proc. U. S. N. M., 14, 15) considered and summarized the distribution of the genera and species so far as known. They recognized sixty-five genera and about four hundred and sixty species, as com- pared with the forty genera and one hundred and eighty-one species enumer- ated by Giinther (1864). It is not necessary to give here the details of the results of their inquiry into the geographical distribution of the species. It was found that nineteen of the genera recognized were distributed over the entire eastern slope of South America and that five of these had representatives in the La Plata and the Amazon, but not in the small rivers emptying into the Atlantic in southeastern Brazil. One genus, Saccodon, was confined to the Pacific slope. Twenty-seven genera were limited to the Amazons, or to the Amazons and the region north of it. The Guianas held two peculiar genera, the Rio Magdalena one, the southeastern coast streams one, while four genera had a wide but irregular distribution. Many modifications in these summaries of distribution are necessary, both on account of the changes in the boundaries of the genera, and owing to the increase in our knowledge of the distribution of the several species. The papers published since the enumeration of 1891 and 1892 have dealt largely with the fauna of Paraguay, Rio Grande do Sul, Guiana, Colombia, and Mexico at nearly opposite ends of the range of the family and with the fauna of the Pacific slope of Ecuador.^ A reconsideration of the entire problem of the dis- tribution of the fresh-water fishes of South America may be found iji The fresh water fishes of Patagonia and an examination of the Archiplata-Archhelenis theory. (Reports of the Princeton uliiversity expeditions to Patagonia, 1909, 3, p. 225-374. Catalogue of the fresh- water fishes of tropical and south tem- perate America. Ibid., 1910, 3, p. 375-512. ' While no attempt has been made to trace the details of the evolution of our knowledge of the African Characins the present account would be most inadequate and incomplete without reference to Boulenger's work on the Characins of tlie Congo and Nile Basins. The new genera and species from the Congo were descrilied for the most part in volume 1 and 2 of the Annales Musfe du Congo. A general account in which all the species were considered, formed part of his Les poissons du basin du Congo, 1901. The Nile representatives are described and figured in his superb volume, The fishes of the Nile, 1907. Among other recent authors on the African Characins is J. Pellegrin, who is describing the material of the Paris Museum. Finally Boulenger (Catalogue of the fresh-water fishes of Africa, 1909, 1, p. 17-1-298) redefines the subfamilies and genera and redescribes all of the African species. He recog- nizes twenty genera and one hundred species. 32 THE AMERICAN CHARACIDAE. Chronological List of Generic Names. Original name Date Current Name Charax Gnonow 1754 & 1777 Erythrinus Gronow 1756 & 1777 Gasteropelecus Gronow 1756 & 1777 Anostomus Gronow 1756 & 1777 Tetragonopterus Artedi 1758 Characinus Lacepede 1802 Serrasalmo Lacepede 1802 Curimates Cuvier 1815 Tetragonoptere Cuvier 1815 Les Curimates Cuvier 1817 Curimatus Les Anostomes Cuvier 1817 Anostomus Les Piabuques Cuvier 1817 Piabucus. Tetragonopterus Cuvier 1817 Myletes Cuvier 1817 Hydrocynus Cuvier 1817 Citharinus Cuvier 1817 Gasteropelecus Cuvier 1817 Erythrinus Cuvier 1817 Curimatus Oken 1817 Piabucus Oken 1817 Chalceus Cuvier 1818 Curimata Cloquet 1818 Hydrocyon Cuvier 1819 Hydrocynus Anodus Spix 1829 Prochilodus Agassiz 1829 Leporinus Spix 1829 Schizodon Agassiz 1829 Anostomus Rhaphiodon Agassiz 1829 Xiphostoma Spix 1829 Pacu Spix 1829 Prochilodus Cynodon Spix 1829 Xiphorhynchus Agassiz 1829 Acestrorhynchus Salminus Agassiz 1829 Macrodon MuUer 1842 HopHas Hemiodus Miiller 1842 Chilodus Miiller & Troschel 1844 Distichodus Muller & Troschel 1844 Alestes Muller & Troschel 1844 Brycon Miiller & Troschel 1844 Exodon Miiller & Troschel 1844 Epicyrtus Miiller & Troschel 1844 Charax Hydrolycus Miiller & Troschel 1844 Pygocentrus Miiller & Troschel 1844 Pygopristis Miiller & Troschel 1844 Catoprion Miiller & Troschel 1844 THE CHARACIDAE. 33 Original name Date Current Name Myleus Miiller & Troschel 1844 Xipliorhamphus Miiller & Troschel 1845 Acestrorhampluis Agoniates Miiller & Troschel 1845 Grundulus Cuv. & Valenciennes 1846 Lebiasina Cuv. & Valenciennes 1846 Pyn-hulina Cuv. & Valenciennes 1846 Parodon Cuv. & Valenciennes 1848 Brycinus Cuv. & Valenciennes 1848 Alestes? Piabucina Cuv. & Valenciennes 1848 Tometes Cuvier & Valenciennes 1848 Mylesinus Cuv. & Valenciennes 1848 Chalcinus Cuv. & Valenciennes 1848 Cynopotanius Cuv. & Valenciennes 1848 Hydropardus Reinhardt 1849 Raphiodon Cheirodon Girard 1854 Astyanax Baird & Girard 1854 Poecilurichthys Gill 1858 Astyanax Hemigrammus Gill 1858 Stevardia Gill 1858 Corynopoma Gill 1858 Stevardia Nematoponia Gill 1858 Stevardia Microdus Kner 1859 Caenotropus Rhytiodus Kner 1859 Bryconops Kner 1859 Ctenolucius Gill 1861 Luciocharax Hydrocyonoides Castelnau 1861 Sarcodaces Crenuchus Giinther 1863 Pseudochalceus Kner 1863 Chalcinopsis Kner 1863 Brycon Saccodon Kner & Steindachner 1863 Caenotropus Giinther 1884 Brachyalestes Giinther 1864 Creatochanes Giintlier 1864 Hemibry'con Giinther 1864 Scissor Giinther 1864 Creagrutus Giinther 1864 Anacyrtus Giinther 1864 Charax Roestes Giinther 1864 Roeboides Giinther 1864 Hystricodon Giinther 1864 Exodon Sarcodaces Giinther 1864 Oligosargus Giinther 1864 Ichthyborus Giinther 1865 Phago Giinther 1865 Piabina Reinhardt 1866 Characidium Reinhardt 1866 Nannocharax Giinther 1867 Xenocharax Giinther 1867 34 THE AMERICAN CHARACIDAE. Original name Date Current name Aphyocharax Giinther 1868 Megalobrycon Giinther 1869 Brycon Stethaprion Cope 1870 Holotaxis Cope 1870 Plethodectes Cope 1870 Chalceus Odontostilbe Cope 1870 . Laemolyta Cope 1871 Nannaethiops Giinther 1871 Iguanodectes Cope 1871 Nannostomus Giinther 1872 Bryconaethiops Giinther 1873 Leporellus Liitken 1874 Curimatopsis Steindachner 1876 Liitkenia Steindachner 1876 Stichanodon Paragoniates Steindachner 1876 Bramocharax Gill 1877 Luciocharax Steindachner 1878 Elopomorphiis Gill 1878 Potamorhina Cope 1878 Metynnis Cope 1878 Leptagoniates Boulenger 1887 Psectrogaster Eigenmann & Eigenmann 1889 Curimatella Eigenmann & Eigenmann 1889 Semitapicis Eigenmann & Eigenmann 1889 Henochilus Garman 1890 Schizodontopsis Garman 1890 Laemolyta Pseudocorynopoma Perugia 1891 Bergia Steindachner 1891 Pseudocorynopo] Chalcinopsis Holmberg 1891 Pseudocorynopoi Neolehias Stein«]achner 1894 Asiphonichthys Cope 1894 Chorimycterus Cope 1894 Diapoma Cope 1894 Petersius Hilgendorf 1894 Nanognathus Boulenger 1895 Hoplerythrinus Gill 1895 Eugnathichthys Boulenger 1898 Paraphago Boulenger 1899 Neoborus Boulenger 1899 Micralestes Boulenger 1899 Neolebias Boulenger 1899 Catabasis Eigenmann & Norris 1900 Mesoborus Pellegrin 1900 Hemistichodus Vaillant & Pellegrin 1900 Citharidium Boulenger 1902 Gymnocharacinus Steindachner 1903 Hoplias Gill 1903 Anisitsia Eigenmann & Kennedy 1903 THE CHAIUCIDAE. 35 Original name Datr Current Name Lahilliella Eigeninann & Kennedy 1903 Holoshesthes Eigenmann 1903 Holoesthes Holoprion Eigenmann 1903 Holopristis Eigenmann 1903 Pristella Markiana Eigenmann 1903 Moenkliausia Eigenmann 1903 Othonophanes Eigenmaim 1903 Bryconodon Eigenmann 1903 Stichanodon Eigenmann 1903 Evermannella Eigenmann 1903 Eucinopotamus Acestrorhynchus Eigenmann 1903 Acestrorhamphus Eigenmann 1903 Acestrocephalus Eigenmann 1903 Boulengerella Eigenmann 1903 Gilbertella Eigenmann 1903 Gibertohis Acnodon Eigenmann 1903 Myleoeollops Eigenmann 1903 Piaractus Eigenmann 1903 Orthomyleus Eigenmann 1903 Colosoma Eigenmann 1903 Mylosoma Eigenmann 1903 Eucinopotamus Fowler 1900 Ophiocephalops Fowler 1906 Hoplerythrinus Copeina Fowler 1906 Cyphocharax Fowler 1906 Steindachnerina Fowler 1906 Peltapleura Fowler 1906 Eigenmannina Fowler 1906 Chilomyzon Fowler 1906 Prochilodus Hemiodopsis Fowler 1906 Pitheeocharax Fowler 1906 Anostomus Poecilosomatops Fowler 1906 Characidium Garmanina Fowler 1906 Abramites Fowler 1906 Leporinus Pellegrinina Fowler 1906 Chalceus Coscinoxyron Fowler 1906 Thoracocharax Fowler 1906 Cyrtocharax Fowler 1906 Cynopotamus Cynocharax Fowler 1906 Sphyraenocharax Fowler 1906 Acestrorhamphus Belonocharax Fowler 1906 Luciocharax Waiteina Fowler 1906 Colosoma? Reganina Fowler 1906 Colosoma? Starksina Fowler 1906 Mylosoma Sealeina Fowler 1906 Evermannolus Eigenmann 1907 Eucinopotamus Gilbertolus Eigenmann 1907 Phenacogrammus Eigenmann 1907 36 THE AMERICAN CHARACIDAE. Original Name Date Current Navie Pogonocharax Regan ' 1907 Phoxinopsis Regan 1907 Mimagoniates Regan 1907 Ctenocharax Regan 1907 Grundulus Eobrycon Jordan 1907 Bryconaraericus Eigenmann 1907 Deuterodon Eigenmann 1907 Phenacogaster Eigenmann 1907 Astyanaeinus Eigenmann 1907 Fowlerina Eigenmann 1907 Ephippicharax Joinvillea Steindaehner 1908 Deuterodon Coelurichthys Ribeiro 1908 Gymnocorymbus Eigenmann 1908 Thayeria Eigenmann 1908 Ctenobrycon Eigenmann 1908 Pristella Eigenmann 1908 Psellogrammus Eigenmann 1908 Hyphessobrycon Durbin 1908 Brycochandus Eigenmann 1908 Poptella Eigenmann 1908 Champsoborus Boulenger 1909 Anostomoides Pellegrin 1909 Jobertina Pellegrin 1909 Pterodiscus Eigenmann 1909 Carnegiella Eigenmann 1909 Holobrycon Eigenmann 1909 Triurobrycon Eigenmann 1909 Brycon Poecilocharax Eigenmann 1909 Microcharax Eigenmann 1909 Poeeilobrycon Eigenmann 1909 Archicheir Eigenmann 1909 Hollandichthys 1910 Nematobrycon Eigenmann 1911 Knodus Eigenmann 1911 Bivibianchia Eigenmann 1911 Hasemania Ellis 1911 Probolodus Eigenmann 1911 Psalidodon Eigenmann 1911 Spintherobolus Eigenmann 1911 Glandulocauda Eigenmann 1911 Hysteronotus Eigenmann 1911 Vesicatrus Eigenmann 1911 Apodastyanax Fowler 1911 Ctenobrycon. Rhodsia Fowler 1911 Parastremma Eigenmann 1912 Genycharax Eigenmann 1912 ' This is probably a Cyprinoid from Ceylon not a Characin from South America. THE CHARACIDAE. 37 Original name Gephyrocharax Eigenmann Pterobrycon Eigenmann Argopleura Eigenmann Microgenys Eigenmann Zygogaster Eigenmann Ephippicharax Fowler Prionobrama Fowler Gnathocharax Fowler Tyttocharax Fowler Xenurocharax Regan Landonia Eigenmann Phenagoniates Eigenmann & Wilson Microbrycon Eigenmann & Wilson Ceratobranchia Eigenmann Bleptonema Eigenmann Parecbasis Eigenmann Myocharax Fowler Xiphocharax Fowler Leptobrycon Eigenmann Macropsobrycon Eigenmann Megalamphodus Eigenmann Microschemobrycon Eigenmann Oligobrycon Eigenmann Aphyocheirodon Eigenmann Compsoura Eigenmann Mixobrycon Eigenmann Date Current name 1912 1913 1913 Bryconamericus 1913 1913 Astyanax 1913 1913 1913 1913 1913 1914 1914 Phanagoniates 1914 1914 1914 Prionobrama 1914 1914 1914 1915 1915 1915 1915 1915 1915 1915 1915 38 THE AMERICAN CILVRACIDAE. THE AMERICAN TETRAGONOPTERINAE. Whether we regard the Heterognaths as a single family, or as several fanii- Ues, the fact remains that there are few groups of fishes within which the lines of evolution are so clearly portrayed by existing forms as in the Characidae. In order more satisfactorily to discuss their evolution I have divided the family into a large number of small groups of genera with undoubted affinity. Several facts, aside from the general structure point to the Tetragonop- terinae or perhaps the closely alhed Cheirodontinae as the group nearest to the ancestral Characins. 1. Tetragonopterid fishes have been found fossil in the Tertiary fresh-water deposits at Taubate. 2. They are found both in Africa and South America, the most nearly aUied genera on the two sides, Astyanax and Petersius, are scarcely generi- cally distinct. No other subfamily has representatives on both sides of the Atlantic. 3. They are the dominant groups both in Africa, where they form more than 36% of the Characins, and in America, where they form about 40%. 4. They are found over the entire area of distribution both in America and Africa. In America members of this subfamily form everywhere the vanguard in the distribution. 5. There are several lines of evolution diverging from the two subfamiUes, Cheirodontinae and Tetragonopterinae. Some of the fines of evolution radiating from different sections of the Tetra- gonopterinae, or the closely alfied Cheirodontinae are minor lines that have not diverged greatly. For instance Diapoma and Stevardia are Tetragonop- terids with modified opercles. The Gymnocharacinae are naked Tetragonop- terinae. The Crenuchinae are apparently an offshoot from the Cheirodontinae. The Stethaprioninae have developed a predorsal spine but are otherwise very close to Tetragonopterus and especially to Moenkhausia. The Myfinae and Serrasalmoninae have possibly diverged from the Stethaprioninae, increasing the number of the dorsal rays, increasing the depth, compressing the ventral surface, and adding spines and emphasizing the dentition without much alter- ing it. Another offshoot from the Cheirodontinae has given rise to a series of fishes with a decrease in the size of the mouth and the effectiveness of the THE AMERICAN TETRAGONOPTERINAE. 39 dentition. This series includes the Anastomatinae, Chilodinae, Prochilodinae, Hemiodontinae, Elopomorphinae, and finally, the toothless Curimatinae. Another line diverging from the Cheirodontinae has given rise to the Sal- niininae, Characinae, Acestrorhamphinae, Cynodontinae, and ultimately the Hydrocyninae. Another line of divergence from the Cheirodon or Tetragonopterus group led through the Bryconinae, Iguanodectinae, Pyrrhulininae, Piabuscininae, to the Lebiasininae. A side branch from this leads to the Chalcininae which points the way to the flying Gasteropelicinae. For the above reasons it seems best to begin the detailed examination of the species with the Tetragonopterinae. It may be left an open question whether the African and American genera owe their similarity to convergence or to community of origin. They are for the most part small or minute fishes ranging from 50-200 mm. in length. Myletes in Africa reaches 460 mm. The Tetragonopterinae are closely related to the Cheirodontinae, Diapominae, Glandulocaudinae, Stethaprioninae, Bry- coninae and to other subfamilies. In shape they vary from the fusiform Creatochanes to the deeply compressed Tetragonopterus. In the majority of the genera the mouth is small. The lower jaw is heavy, and on account of the obUquity of the mouth, when the mouth is opened it is thrown forward, so as to project beyond the snout; when the mouth is closed the teeth of the lower jaw usually fit in behind the innermost series of the premaxillary. In some genera from Africa there is an inner pair of conical teeth in the lower jaw. These appear in lieu of an inner series of teeth which in the American Brycon are still present on the sides of the lower jaw. Conical teeth like these appear in isolated (not closely related) genera of other subfamiUes both in Africa and America, and, since they cannot be genetically connected, appear to offer an example of homoplastic development. The skull is smooth in cross-section or slightly grooved in the smaller specimens; two fontanels are well developed in all the genera but Brycinus of Africa. In this genus there is no frontal fontanel. The bridge between the fontanels is either flush with the surface or sunk beneath it; the occipital process varies directly with the depth of the species, and serves to bridge the space between the skull and first interneurals, i. e. the space over the coalesced vertebrae which lack interneurals. In the deeper species the process is ciu-ved or bent upward and is long, reaching as much as one third of the distance of its base from the dorsal. In the slenderest species it is short and insignificant. It is always grooved to its tip, the groove leading to the parietal fontanel. 40 THE AMERICAN CHARACIDAE. The eyes are always large, the species depending largely on living and moving food. The cheeks may be narrow or deep and are one third or entirely covered by the second suborbitals. The premaxillary teeth are always in at least two series. The inner series consists of from 4-12 graduated; three to many-pointed (a few may be conical) teeth arranged in a regular series. The outer row is very variable both in the group and in individual species and ranges from one or more teeth near the middle of the premaxillary, to a complete, compact series of teeth, narrower than those of the inner series in Tetragonopterus. If the outer row consists of fovu- or more teeth the third tooth usually drops out of line and tends to form a third series of teeth. In Moenkhausia melanogramma the third tooth is entirely withdrawn from the line of the others. In the related Brycon in which the teeth are more numerous than in the genera of this subfamily, the fourth tooth and a few others also drop out of line and join the third tooth of the lower inner series which has moved forward. Other teeth of the inner series have also moved forward giving rise to a third series of teeth. A third series of teeth has independently arisen in Creagrutus and Bryconops, and in the Bryconinae. The maxillary may have no teeth, or a few may be crowded along the part of the margin nearest to the premaxillary, or it may have conical or tricuspid teeth along its entire margin. Usually the variation in any one species is very limited, but in Astyanax fasciatus nicaraguensis there is a variation of from one to nine teeth in the maxillary. The teeth in the lower jaw may all be alike and graduate, or the lateral teeth may be more or less abruptly smaller, the more abruptly the more proba- bly the lateral teeth will be conical. The larger teeth, 3-6 in number, may form a nearly transverse series or be arranged in a curve. The denticles of the individual teeth of the inner series of the premaxillary may be in a straight line, i. e., the teeth may be strictly incisors or, the line joining the cusps may be more and more curved so that it will be U-shaped, the open part of the U forward. The teeth in the lower jaw are usually the reverse of those of the inner series of the upper jaw. Gill-rakers are usually slender and not very long, they are all but absent in one genus. Scissor. The gill-membranes are free from each other, the nares close together. The breast is flat or rounded, never keeled. The scales are usually cycloid, rarely crenate or even ctenoid. They vary from 26 to about 60 in the lateral line, which is variously developed. The caudal may be naked, the scales of the sides passing on to its base and THE AMERICAN TETRAGONOPTERINAE. 41 ending in slightly enlarged scales, or the scales may become minute on the caudal adhering and covering the lobes to a greater or less extent. The anal may similarly have simply a basal sheath or may be covered to near its tip with minute scales. The scales may be regularly imbricate or in certain regions there may be interpolated rows. In some species there are but a few interpolated scales, i. e., a single series becomes divided into two above the anal. The point of division becomes more and more removed from the anal and the number of divided series increases. In long slender species, or larger-scaled, deep species, there are no interpolated rows. In deep, many-scaled species they sometimes become numerous. Closely allied species or even varieties may differ in this respect. In Asiyanax fasciatus individuals with interpolated series are rarely found except in the Rio Parahyba where the usual variety has been entirely replaced by one with interpolated series. The dorsal is short, of between 9 and 12 rays, counting everything, and its origin is usually in the middle of the body. The adipose dorsal is small but almost always well developed. The caudal is always forked, the lobes equal or subequal, rarely markedly different. The anal is another variable element. Its origin usually below or behind the last dorsal ray may (Psello- grammus and Phenacogaster) fall below its origin. Other things being equal the deeper species have the larger number of anal rays. The number of rays varies from 10 to 48. The reach of the pectorals and ventrals varies with the shape of the fish. In deep species with long anal they overlap, in slender species they ordinarily do not reach each other. The notable exception to this is found in Gephyro- charax. The alimentary canal varies but little from the entire length of the fish. There are a few pyloric coeca. The air-bladders are large, the posterior about twice the length of the anterior, curved down behind in the deeper species. The range of color is limited. In life the caudal of the male is frequently cherry-red, the dorsals and anal (and caudal in females) are frequently yellow. A shoulder-spot of varying shape is usually present. There is usually a silvery lateral band overlying a black band which becomes evident in formalin prepara- tions. There is frequently a dark spot on the base of the caudal which is often continued on the middle caudal rays. In Moenkhausia dichrourus and Brycona- mericus exodon the tips or bands across the caudal lobes are dark. In Moenk- 42 THE AMERICAN CHARACIDAE. hausia lepidurus and the species of Creatochanes the middle rays and upper caudal lobes are black. In Astyanax lineatus and A. steindachneri, in Moenk- hausia latissimus, and in Hollandichthys and Pseudochalceus dark lines follow the spaces between successive rows of scales. The dorsal and anal are marked with black in a few species. The points of greatest variability within a genus are : — 1. The depth and all that tliis carried with it, length of occipital process, rows of scales, number of rays of anal. 2. The size of the mouth and the dentition. 3. The degree of armature of the check. 4. The scaling of the caudal, anal, predorsal, and preventral areas. 5. The degree of development of the lateral line. Contrasted Generic Characters. The following mutually exclusive characters are found variously combined in different genera. The characters appearing in the largest number of genera are given in the first column. A. B. C. D. E. F. Caudal fin naked. Lateral line complete. Maxillary with few teeth or none. Premaxillary teeth in two series. Cheeks partly naked. Anterior edge of maxillary a simple curve. G. Predorsal line .scaled. H. Teeth of the sides of the dentary abruptly smaller. I. Premaxillary meeting the maxillary of an angle. J. Scales entire. K. Anal naked except at the base. L. Lateral line nearly straight. M. Preventral area with normal scales. N. Gill-rakers setiforni. 0. Adipose fin present. P. Origin of anal behind origin of dorsal. Q. Caudal without glandular scales. a. Caudal scaled. b. Lateral line incomplete. c. Maxillary with teeth along its entire edge. d. Premaxillary teeth in three series. e. Cheeks entirely coveretl by the third sub- orbital. /. Anterior edge of maxillary sigmoid. g. Predorsal line naked. h. Teeth of the sides of the dentary gradu- ated. i. Maxillary-premaxillary border a simple curv^ j. Scales ctenoid. j. Scales crenate. k. Anal scaled to near its tip. /. Lateral line sharply decurred in front. m. Preventral area with paired scales. n. Gill-rakers lanceolate. o. Adipose fin wanting. J). Origin of anal under or in front of origin of dorsal. q. Caudal with glandular scales. The characters appearing in the larger number of genera and given in the first column are all found in Astyanax, which may for that reason, be the central, THE AMERICAN TETRAGONOPTEIIINAE. 43 possibly the most primitive member, of the subfamily. Poecilm-ichthys, which differs from it only in having the character I instead of L, is scarcely distin- guishable.' The different characters of the first column are found in various combina- tions with most of the characters of the second column. Thus A is found in combination with all (not counting its contrasted character, a) except k and I. B is found in combination with all but o and p. C is found in combination with all the characters of the second column and the same is true of all the rest of the characters of the first column. POLYPHYLETXC CHARACTERS. It is quite certain that the characters of the right-hand column of the contrasted generic characters have sometimes, at least, been independently derived from the characters of the left-hand column, not only by the different genera possessing the character, but sometimes by different species of the same genus. That is, some of the genera possessing characters given in the second column are of polyphyletic origin. Begiiming at the bottom of the list, the character o is found in Hasemania from southeastern Brazil, and in the very different genus, Nematobrycon, from the west of the western Cordilleras of Colombia. The two genera are not related, and are widely separated geographically. There can be no doubt but that they have independently lost the adipose fin. The next undoubted case is that indicated by g. This character is found in Poecilurichthys, scarcely distinct from Astyanax, and in Gymnocorymbus, a very different fish. Here again the character has very probably been inde- pendently acquired by the two genera. ■ In Indiana Univ. Studies, 1914, No. 20, I have tried to carry the inference a step further. "The common possessions of all members of the Tetragonopterinse enumerated above, permit us to picture the ancestral type of the subfamily. In brief, it must have been a fish similar in most char- acters to Astyanax fasciatus Cuvier. This species, besides possessing all the characters common to all members of the subfamily, possesses also many of those positive (as contrasted with absent) characters enumerated for the family, and lacks some characters, like the highly specialized scaling of the ventral surface, ctenoid scales, extreme length of anal, extreme development of second suborbital, which are evidently highly specialized characters in a few of the genera. It is more widely distributed than any other species and has given rise to numerous variations. It represents an average in length of head (4.3), depth, (2.6-3); length of anal (about 30); scales (about 38); size of eye (2.5-3); general shape (compressed subf usif orm) ; position of dorsal (its base being in the space above the origins of the ventral and anal) ; size of mouth ; and the characters of the teeth. The fossil fishes found at TaubatiS — south of Rio de Janeiro — are similar to it in most char- acters. They are a little larger and may be members of the genus Brycon. In all but the teeth, they are very similar to Astyanax fasciatus." 44 THE AMERICAN CHARACIDAE. The next case is that of e, the enlarged third suborbital. This character is found in Knodus with a scaled caudal and in Creagrutus, Piabina, and Brycon- americus with a naked caudal. The last three genera are undoubtedly closely related and for the present purpose count as one. Knodus is a Bryconamericus in all but its scaled caudal. We are, therefore, compelled to assume either that if the ancestors of Knodus had the caudal scaled, that it has paralleled Brycon- americus in the character of its cheek or that if its ancestors were identical with those of the latter genus that its caudal has independently acquired scales. The case presented by the character d, three rows of premaxillary teeth, is much simpler and clearer. This character is found in Microgenys, Creagratus, Piabina, and Bryconops. It is quite certain that the latter genus has been derived from an ancestor Uke Creatochanes and that the others have been de- rived from an Astyanax-like ancestor. Microgenys, Creagratus, and Piabina, on the one hand, and Bryconops on the other, have independently acquired three series of teeth. In this case the steps by which this has been accomphshed are indicated in a variety of species of the subfamily. In many species alternate teeth of the front series of the premaxillary are withdrawn from the line of the rest, thus forming an incipient third series. In Moenkhausia melogramma and in Bryconamericus exodon this process has almost yielded additional genera with three series of teeth. The modification from one to the other condition is a perfectly progressive one, without notable breaks or saltations. The next character c, teeth along the entire maxillary, has again been inde- pendently derived from C several times. Here we have not only the evidence of several distinct, not closely related, genera which have the character but also the evidence from changing species. I have elsewhere called attention to some specimens referred to as Astyanas aeneus nicaraguensis . Of thirty-five specimens from Lake Nicaragua, there are nine with two teeth, two with three teeth, five with four teeth, five with five teeth, five with six teeth, five with seven teeth, three with eight teeth and one with nine teeth in the maxillary. The normal number is two. Phenacogaster is in a similar state of transition. Pris- tella, Hemibrycon, Nematobrycon, Hollandichthys, and Pseudochalceus have acquired complete dentition for their maxillaries in at least three independent groups. The most interesting and conclusive evidence of the independent origin of the same character in different genera is presented by the character b, the incompleteness of the lateral line. Not only have we the evidence from widely divergent genera with this character, but we again have species in a state of true mutation. Most remarkable of all is one species of which I have been able THE AMERICAN TETRAGONOPTERINAE. 45 to examine hundreds of specimens from the Amazon, not one of which showed signs of mutation, wlule the specimens coming from another region are in an evident state of mutation. Genera with a complete lateral line and the genera with an incomplete lateral line to which they have given rise are : — Genera with lateral line complete. Genera mth lateral line incomplete Tetragonopterus none. Entomolepis none. Moenkhausia Hemigrammus (further changed into Pristella and Thayeria) Astyanax Hyphessobrycon (into Hasemania). Knodus none. Markiana none. Gymnocorymbus none. Ctenobrycon Psellogramraus. Creatochanes Brycochandus. Bryconops none. Creagrutus none. Piabina none. Microgenys none. Bryconamericus none. Zygogaster none. Ceratobranchia none. Landonia none. Deuterodon none. Hemibrycon Hollandichthys, Pseudochalceus, and Nematobrycon. Phenacogaster Vesicatrus. Scissor none. Henochilus none. Psalidodon none. The deviation is so evident in a number of cases that the polyphyletic origin of the character b, an incomplete lateral line, is beyond the faintest shadow of a doubt. Without considering sporadic individuals of otherwise constant species, the species which are undoubtedly mutuating at the present time are Hemi- granmius inconstans, Moenkhausia cotinho, Phenacogaster beni, Astyanax mutator, and Hyphessobrycon poecilioides. The details for these species are given under their descriptions. Another species which has crossed the line but has not reached a state of equihbrium is Psellogrammus kennedyi. Many of the details of this species are again given under the proper caption. Of Ctenobrycon hauxwellianus I have been able to examine over fourteen hundred specimens from various places on the Amazon. In all of these the lateral line is complete. In at least six specimens out of nineteen from the Lagoa 46 THE AMERICAN CHARACIDAE. Parnagua, Paranahyba basin, the lateral line "stutters." This species, in other words, is mutuating at Parnagua. Moenkhausia cotinho is similarly locally mutating and the same seems to be true of Hyphessobrycon poecilioides and Astyanax fasciatus. The scaUng of the caudal I am not able to cope with satisfactorily. It is certain that it has several times been acquired independently by different members of the family, if not by different members of the subfamily under consideration. Selective Grouping op Characters. Another line of inquiry leads us to consider whether the contrasted unit char- acters are entirely combined as if by chance or whether there is a selected com- bination. The characters from g-q are found in but one or two genera and these may be omitted from the discussion since they would needlessly compUcate it. Taking only the first six pairs of contrasted characters, there are 62 or 64 possible combinations. Considering each combination a distinct genus we should have 64 genera, without considering the characters between g and q. In reality we have but about half as many. Taking only the first three characters we should have eight possible com- binations. An examination of the genera shows that six of these combinations are actually found, but not by any means in the same proportions. If we take the first four characters, each of which has probably several times, and inde- pendently, been modified into its contrasted character, these four pairs give us sixteen possible combinations. Of these only eight actually occur. A certain amount of selective grouping is thus found, if we take only three contrasted pairs of characters, a greater amount of selective omission if we take four pairs and this increases rapidly as we increase the number of contrasted forms. It is evident that either many of the possible combinations have never arisen, or, having appeared, they have not been preserved. POLYPHYLETIC GeNERA. In my paper, Indiana University Studies, 1914, no. 20, I wrote — "This independent origin of characters is responsible for the fact that some of the accepted genera of the Tetragonopterinse are of polyphyletic origin, i. e. our definitions of genera are in many cases enumerations of characters frequently independently acquired, not enumerations of the characters of the ancestral type of the genus from wliich the species have diverged. A result of this independent divergence is that frequently in a restricted, isolated area the species of different genera represented in the area are more nearly related to each other than to members of their own genera in remote regions. For instance Astyanax THE AMERICAN TETRAGONOPTERINAE. 47 festa; anrl Bnjconamcricus pcruanns of the Pacific slope of Ecuador are more intimately related than festw is to Astyanax anterior of the upper Amazon. And in this case, A!